You are viewing a plain text version of this content. The canonical link for it is here.
Posted to commits@hbase.apache.org by gi...@apache.org on 2021/10/08 20:17:22 UTC
[hbase-site] branch asf-site updated: Published site at
a384c239b9c5462c16ded4f74fc5dd7b3e1b1682.
This is an automated email from the ASF dual-hosted git repository.
git-site-role pushed a commit to branch asf-site
in repository https://gitbox.apache.org/repos/asf/hbase-site.git
The following commit(s) were added to refs/heads/asf-site by this push:
new 69a8fc7 Published site at a384c239b9c5462c16ded4f74fc5dd7b3e1b1682.
69a8fc7 is described below
commit 69a8fc768871af5dcb640fe37f8a86700ebf54a4
Author: jenkins <bu...@apache.org>
AuthorDate: Fri Oct 8 20:17:12 2021 +0000
Published site at a384c239b9c5462c16ded4f74fc5dd7b3e1b1682.
---
acid-semantics.html | 2 +-
apache_hbase_reference_guide.pdf | 4 +-
book.html | 2 +-
bulk-loads.html | 2 +-
checkstyle-aggregate.html | 2 +-
coc.html | 2 +-
dependencies.html | 2 +-
dependency-convergence.html | 2 +-
dependency-info.html | 2 +-
dependency-management.html | 2 +-
downloads.html | 2 +-
export_control.html | 2 +-
index.html | 2 +-
issue-tracking.html | 2 +-
mail-lists.html | 2 +-
metrics.html | 2 +-
old_news.html | 2 +-
plugin-management.html | 2 +-
plugins.html | 2 +-
poweredbyhbase.html | 2 +-
project-info.html | 2 +-
project-reports.html | 2 +-
project-summary.html | 2 +-
pseudo-distributed.html | 2 +-
replication.html | 2 +-
resources.html | 2 +-
source-repository.html | 2 +-
sponsors.html | 2 +-
supportingprojects.html | 2 +-
team-list.html | 2 +-
testdevapidocs/constant-values.html | 19 +
testdevapidocs/index-all.html | 6 +-
.../org/apache/hadoop/hbase/package-tree.html | 20 +-
.../hadoop/hbase/procedure2/package-tree.html | 4 +-
.../hadoop/hbase/regionserver/package-tree.html | 6 +-
.../IntegrationTestLoadCommonCrawl.Counts.html | 12 +-
...rationTestLoadCommonCrawl.HBaseKeyWritable.html | 76 +-
...ionTestLoadCommonCrawl.Loader.LoaderMapper.html | 26 +-
.../IntegrationTestLoadCommonCrawl.Loader.html | 16 +-
...onTestLoadCommonCrawl.OneFilePerMapperSFIF.html | 6 +-
...ionTestLoadCommonCrawl.Verify.VerifyMapper.html | 14 +-
.../IntegrationTestLoadCommonCrawl.Verify.html | 14 +-
.../hbase/test/IntegrationTestLoadCommonCrawl.html | 75 +-
.../org/apache/hadoop/hbase/test/package-tree.html | 6 +-
.../org/apache/hadoop/hbase/wal/package-tree.html | 4 +-
.../IntegrationTestLoadCommonCrawl.Counts.html | 1588 ++++++++++----------
...rationTestLoadCommonCrawl.HBaseKeyWritable.html | 1588 ++++++++++----------
...ionTestLoadCommonCrawl.Loader.LoaderMapper.html | 1588 ++++++++++----------
.../IntegrationTestLoadCommonCrawl.Loader.html | 1588 ++++++++++----------
...onTestLoadCommonCrawl.OneFilePerMapperSFIF.html | 1588 ++++++++++----------
...ionTestLoadCommonCrawl.Verify.VerifyMapper.html | 1588 ++++++++++----------
.../IntegrationTestLoadCommonCrawl.Verify.html | 1588 ++++++++++----------
.../hbase/test/IntegrationTestLoadCommonCrawl.html | 1588 ++++++++++----------
53 files changed, 6658 insertions(+), 6412 deletions(-)
diff --git a/acid-semantics.html b/acid-semantics.html
index 9079ff5..79ae56b 100644
--- a/acid-semantics.html
+++ b/acid-semantics.html
@@ -456,7 +456,7 @@
<div class="row">
<p>Copyright ©2007–2021
<a href="https://www.apache.org/">The Apache Software Foundation</a>.
-All rights reserved. <li id="publishDate" class="pull-right">Last Published: 2021-10-07</li>
+All rights reserved. <li id="publishDate" class="pull-right">Last Published: 2021-10-08</li>
</p>
</div>
<p id="poweredBy" class="pull-right"><a href="http://maven.apache.org/" title="Built by Maven" class="poweredBy"><img class="builtBy" alt="Built by Maven" src="./images/logos/maven-feather.png" /></a>
diff --git a/apache_hbase_reference_guide.pdf b/apache_hbase_reference_guide.pdf
index 763a2ea..bee8145 100644
--- a/apache_hbase_reference_guide.pdf
+++ b/apache_hbase_reference_guide.pdf
@@ -5,8 +5,8 @@
/Author (Apache HBase Team)
/Creator (Asciidoctor PDF 1.5.3, based on Prawn 2.2.2)
/Producer (Apache HBase Team)
-/ModDate (D:20211007200519+00'00')
-/CreationDate (D:20211007201523+00'00')
+/ModDate (D:20211008200506+00'00')
+/CreationDate (D:20211008201511+00'00')
>>
endobj
2 0 obj
diff --git a/book.html b/book.html
index 9eed4f0..b2aef74 100644
--- a/book.html
+++ b/book.html
@@ -46227,7 +46227,7 @@ org/apache/hadoop/hbase/security/access/AccessControlClient.revoke:(Lorg/apache/
<div id="footer">
<div id="footer-text">
Version 3.0.0-alpha-2-SNAPSHOT<br>
-Last updated 2021-10-07 20:05:19 UTC
+Last updated 2021-10-08 20:05:06 UTC
</div>
</div>
<link rel="stylesheet" href="./coderay-asciidoctor.css">
diff --git a/bulk-loads.html b/bulk-loads.html
index 90ba53b..9c0a137 100644
--- a/bulk-loads.html
+++ b/bulk-loads.html
@@ -161,7 +161,7 @@
<div class="row">
<p>Copyright ©2007–2021
<a href="https://www.apache.org/">The Apache Software Foundation</a>.
-All rights reserved. <li id="publishDate" class="pull-right">Last Published: 2021-10-07</li>
+All rights reserved. <li id="publishDate" class="pull-right">Last Published: 2021-10-08</li>
</p>
</div>
<p id="poweredBy" class="pull-right"><a href="http://maven.apache.org/" title="Built by Maven" class="poweredBy"><img class="builtBy" alt="Built by Maven" src="./images/logos/maven-feather.png" /></a>
diff --git a/checkstyle-aggregate.html b/checkstyle-aggregate.html
index 232ce2a..925c83f 100644
--- a/checkstyle-aggregate.html
+++ b/checkstyle-aggregate.html
@@ -73774,7 +73774,7 @@
<div class="row">
<p>Copyright ©2007–2021
<a href="https://www.apache.org/">The Apache Software Foundation</a>.
-All rights reserved. <li id="publishDate" class="pull-right">Last Published: 2021-10-07</li>
+All rights reserved. <li id="publishDate" class="pull-right">Last Published: 2021-10-08</li>
</p>
</div>
<p id="poweredBy" class="pull-right"><a href="http://maven.apache.org/" title="Built by Maven" class="poweredBy"><img class="builtBy" alt="Built by Maven" src="./images/logos/maven-feather.png" /></a>
diff --git a/coc.html b/coc.html
index 8b3a107..cf62722 100644
--- a/coc.html
+++ b/coc.html
@@ -230,7 +230,7 @@ email to <a class="externalLink" href="mailto:private@hbase.apache.org">the priv
<div class="row">
<p>Copyright ©2007–2021
<a href="https://www.apache.org/">The Apache Software Foundation</a>.
-All rights reserved. <li id="publishDate" class="pull-right">Last Published: 2021-10-07</li>
+All rights reserved. <li id="publishDate" class="pull-right">Last Published: 2021-10-08</li>
</p>
</div>
<p id="poweredBy" class="pull-right"><a href="http://maven.apache.org/" title="Built by Maven" class="poweredBy"><img class="builtBy" alt="Built by Maven" src="./images/logos/maven-feather.png" /></a>
diff --git a/dependencies.html b/dependencies.html
index d1ea454..6e02ac1 100644
--- a/dependencies.html
+++ b/dependencies.html
@@ -302,7 +302,7 @@
<div class="row">
<p>Copyright ©2007–2021
<a href="https://www.apache.org/">The Apache Software Foundation</a>.
-All rights reserved. <li id="publishDate" class="pull-right">Last Published: 2021-10-07</li>
+All rights reserved. <li id="publishDate" class="pull-right">Last Published: 2021-10-08</li>
</p>
</div>
<p id="poweredBy" class="pull-right"><a href="http://maven.apache.org/" title="Built by Maven" class="poweredBy"><img class="builtBy" alt="Built by Maven" src="./images/logos/maven-feather.png" /></a>
diff --git a/dependency-convergence.html b/dependency-convergence.html
index 4a4ed0f..2dbaae1 100644
--- a/dependency-convergence.html
+++ b/dependency-convergence.html
@@ -972,7 +972,7 @@
<div class="row">
<p>Copyright ©2007–2021
<a href="https://www.apache.org/">The Apache Software Foundation</a>.
-All rights reserved. <li id="publishDate" class="pull-right">Last Published: 2021-10-07</li>
+All rights reserved. <li id="publishDate" class="pull-right">Last Published: 2021-10-08</li>
</p>
</div>
<p id="poweredBy" class="pull-right"><a href="http://maven.apache.org/" title="Built by Maven" class="poweredBy"><img class="builtBy" alt="Built by Maven" src="./images/logos/maven-feather.png" /></a>
diff --git a/dependency-info.html b/dependency-info.html
index 156e330..5c6a0f0 100644
--- a/dependency-info.html
+++ b/dependency-info.html
@@ -183,7 +183,7 @@
<div class="row">
<p>Copyright ©2007–2021
<a href="https://www.apache.org/">The Apache Software Foundation</a>.
-All rights reserved. <li id="publishDate" class="pull-right">Last Published: 2021-10-07</li>
+All rights reserved. <li id="publishDate" class="pull-right">Last Published: 2021-10-08</li>
</p>
</div>
<p id="poweredBy" class="pull-right"><a href="http://maven.apache.org/" title="Built by Maven" class="poweredBy"><img class="builtBy" alt="Built by Maven" src="./images/logos/maven-feather.png" /></a>
diff --git a/dependency-management.html b/dependency-management.html
index 6710a279..b44c926 100644
--- a/dependency-management.html
+++ b/dependency-management.html
@@ -1260,7 +1260,7 @@
<div class="row">
<p>Copyright ©2007–2021
<a href="https://www.apache.org/">The Apache Software Foundation</a>.
-All rights reserved. <li id="publishDate" class="pull-right">Last Published: 2021-10-07</li>
+All rights reserved. <li id="publishDate" class="pull-right">Last Published: 2021-10-08</li>
</p>
</div>
<p id="poweredBy" class="pull-right"><a href="http://maven.apache.org/" title="Built by Maven" class="poweredBy"><img class="builtBy" alt="Built by Maven" src="./images/logos/maven-feather.png" /></a>
diff --git a/downloads.html b/downloads.html
index 0ab8879..68cdf77 100644
--- a/downloads.html
+++ b/downloads.html
@@ -457,7 +457,7 @@
<div class="row">
<p>Copyright ©2007–2021
<a href="https://www.apache.org/">The Apache Software Foundation</a>.
-All rights reserved. <li id="publishDate" class="pull-right">Last Published: 2021-10-07</li>
+All rights reserved. <li id="publishDate" class="pull-right">Last Published: 2021-10-08</li>
</p>
</div>
<p id="poweredBy" class="pull-right"><a href="http://maven.apache.org/" title="Built by Maven" class="poweredBy"><img class="builtBy" alt="Built by Maven" src="./images/logos/maven-feather.png" /></a>
diff --git a/export_control.html b/export_control.html
index 667900b..fd7de1d 100644
--- a/export_control.html
+++ b/export_control.html
@@ -186,7 +186,7 @@ for more details.</p>
<div class="row">
<p>Copyright ©2007–2021
<a href="https://www.apache.org/">The Apache Software Foundation</a>.
-All rights reserved. <li id="publishDate" class="pull-right">Last Published: 2021-10-07</li>
+All rights reserved. <li id="publishDate" class="pull-right">Last Published: 2021-10-08</li>
</p>
</div>
<p id="poweredBy" class="pull-right"><a href="http://maven.apache.org/" title="Built by Maven" class="poweredBy"><img class="builtBy" alt="Built by Maven" src="./images/logos/maven-feather.png" /></a>
diff --git a/index.html b/index.html
index 7489d4e..199ea53 100644
--- a/index.html
+++ b/index.html
@@ -264,7 +264,7 @@
<div class="row">
<p>Copyright ©2007–2021
<a href="https://www.apache.org/">The Apache Software Foundation</a>.
-All rights reserved. <li id="publishDate" class="pull-right">Last Published: 2021-10-07</li>
+All rights reserved. <li id="publishDate" class="pull-right">Last Published: 2021-10-08</li>
</p>
</div>
<p id="poweredBy" class="pull-right"><a href="http://maven.apache.org/" title="Built by Maven" class="poweredBy"><img class="builtBy" alt="Built by Maven" src="./images/logos/maven-feather.png" /></a>
diff --git a/issue-tracking.html b/issue-tracking.html
index 2eb7332..1548e32 100644
--- a/issue-tracking.html
+++ b/issue-tracking.html
@@ -158,7 +158,7 @@
<div class="row">
<p>Copyright ©2007–2021
<a href="https://www.apache.org/">The Apache Software Foundation</a>.
-All rights reserved. <li id="publishDate" class="pull-right">Last Published: 2021-10-07</li>
+All rights reserved. <li id="publishDate" class="pull-right">Last Published: 2021-10-08</li>
</p>
</div>
<p id="poweredBy" class="pull-right"><a href="http://maven.apache.org/" title="Built by Maven" class="poweredBy"><img class="builtBy" alt="Built by Maven" src="./images/logos/maven-feather.png" /></a>
diff --git a/mail-lists.html b/mail-lists.html
index 311fb7f..33921ec 100644
--- a/mail-lists.html
+++ b/mail-lists.html
@@ -204,7 +204,7 @@
<div class="row">
<p>Copyright ©2007–2021
<a href="https://www.apache.org/">The Apache Software Foundation</a>.
-All rights reserved. <li id="publishDate" class="pull-right">Last Published: 2021-10-07</li>
+All rights reserved. <li id="publishDate" class="pull-right">Last Published: 2021-10-08</li>
</p>
</div>
<p id="poweredBy" class="pull-right"><a href="http://maven.apache.org/" title="Built by Maven" class="poweredBy"><img class="builtBy" alt="Built by Maven" src="./images/logos/maven-feather.png" /></a>
diff --git a/metrics.html b/metrics.html
index 7f06510..3f827a5 100644
--- a/metrics.html
+++ b/metrics.html
@@ -314,7 +314,7 @@ export HBASE_REGIONSERVER_OPTS="$HBASE_JMX_OPTS -Dcom.sun.management.jmxrem
<div class="row">
<p>Copyright ©2007–2021
<a href="https://www.apache.org/">The Apache Software Foundation</a>.
-All rights reserved. <li id="publishDate" class="pull-right">Last Published: 2021-10-07</li>
+All rights reserved. <li id="publishDate" class="pull-right">Last Published: 2021-10-08</li>
</p>
</div>
<p id="poweredBy" class="pull-right"><a href="http://maven.apache.org/" title="Built by Maven" class="poweredBy"><img class="builtBy" alt="Built by Maven" src="./images/logos/maven-feather.png" /></a>
diff --git a/old_news.html b/old_news.html
index 75c974c..48228da 100644
--- a/old_news.html
+++ b/old_news.html
@@ -305,7 +305,7 @@
<div class="row">
<p>Copyright ©2007–2021
<a href="https://www.apache.org/">The Apache Software Foundation</a>.
-All rights reserved. <li id="publishDate" class="pull-right">Last Published: 2021-10-07</li>
+All rights reserved. <li id="publishDate" class="pull-right">Last Published: 2021-10-08</li>
</p>
</div>
<p id="poweredBy" class="pull-right"><a href="http://maven.apache.org/" title="Built by Maven" class="poweredBy"><img class="builtBy" alt="Built by Maven" src="./images/logos/maven-feather.png" /></a>
diff --git a/plugin-management.html b/plugin-management.html
index 9346dd5..450a384 100644
--- a/plugin-management.html
+++ b/plugin-management.html
@@ -310,7 +310,7 @@
<div class="row">
<p>Copyright ©2007–2021
<a href="https://www.apache.org/">The Apache Software Foundation</a>.
-All rights reserved. <li id="publishDate" class="pull-right">Last Published: 2021-10-07</li>
+All rights reserved. <li id="publishDate" class="pull-right">Last Published: 2021-10-08</li>
</p>
</div>
<p id="poweredBy" class="pull-right"><a href="http://maven.apache.org/" title="Built by Maven" class="poweredBy"><img class="builtBy" alt="Built by Maven" src="./images/logos/maven-feather.png" /></a>
diff --git a/plugins.html b/plugins.html
index dfd4c1a..fbe78e6 100644
--- a/plugins.html
+++ b/plugins.html
@@ -237,7 +237,7 @@
<div class="row">
<p>Copyright ©2007–2021
<a href="https://www.apache.org/">The Apache Software Foundation</a>.
-All rights reserved. <li id="publishDate" class="pull-right">Last Published: 2021-10-07</li>
+All rights reserved. <li id="publishDate" class="pull-right">Last Published: 2021-10-08</li>
</p>
</div>
<p id="poweredBy" class="pull-right"><a href="http://maven.apache.org/" title="Built by Maven" class="poweredBy"><img class="builtBy" alt="Built by Maven" src="./images/logos/maven-feather.png" /></a>
diff --git a/poweredbyhbase.html b/poweredbyhbase.html
index 4daecd9..d76ec24 100644
--- a/poweredbyhbase.html
+++ b/poweredbyhbase.html
@@ -639,7 +639,7 @@
<div class="row">
<p>Copyright ©2007–2021
<a href="https://www.apache.org/">The Apache Software Foundation</a>.
-All rights reserved. <li id="publishDate" class="pull-right">Last Published: 2021-10-07</li>
+All rights reserved. <li id="publishDate" class="pull-right">Last Published: 2021-10-08</li>
</p>
</div>
<p id="poweredBy" class="pull-right"><a href="http://maven.apache.org/" title="Built by Maven" class="poweredBy"><img class="builtBy" alt="Built by Maven" src="./images/logos/maven-feather.png" /></a>
diff --git a/project-info.html b/project-info.html
index 17a1e02..43aa7aa 100644
--- a/project-info.html
+++ b/project-info.html
@@ -199,7 +199,7 @@
<div class="row">
<p>Copyright ©2007–2021
<a href="https://www.apache.org/">The Apache Software Foundation</a>.
-All rights reserved. <li id="publishDate" class="pull-right">Last Published: 2021-10-07</li>
+All rights reserved. <li id="publishDate" class="pull-right">Last Published: 2021-10-08</li>
</p>
</div>
<p id="poweredBy" class="pull-right"><a href="http://maven.apache.org/" title="Built by Maven" class="poweredBy"><img class="builtBy" alt="Built by Maven" src="./images/logos/maven-feather.png" /></a>
diff --git a/project-reports.html b/project-reports.html
index 368f51b..861152a 100644
--- a/project-reports.html
+++ b/project-reports.html
@@ -175,7 +175,7 @@
<div class="row">
<p>Copyright ©2007–2021
<a href="https://www.apache.org/">The Apache Software Foundation</a>.
-All rights reserved. <li id="publishDate" class="pull-right">Last Published: 2021-10-07</li>
+All rights reserved. <li id="publishDate" class="pull-right">Last Published: 2021-10-08</li>
</p>
</div>
<p id="poweredBy" class="pull-right"><a href="http://maven.apache.org/" title="Built by Maven" class="poweredBy"><img class="builtBy" alt="Built by Maven" src="./images/logos/maven-feather.png" /></a>
diff --git a/project-summary.html b/project-summary.html
index 29514cd..73b5fe9 100644
--- a/project-summary.html
+++ b/project-summary.html
@@ -201,7 +201,7 @@
<div class="row">
<p>Copyright ©2007–2021
<a href="https://www.apache.org/">The Apache Software Foundation</a>.
-All rights reserved. <li id="publishDate" class="pull-right">Last Published: 2021-10-07</li>
+All rights reserved. <li id="publishDate" class="pull-right">Last Published: 2021-10-08</li>
</p>
</div>
<p id="poweredBy" class="pull-right"><a href="http://maven.apache.org/" title="Built by Maven" class="poweredBy"><img class="builtBy" alt="Built by Maven" src="./images/logos/maven-feather.png" /></a>
diff --git a/pseudo-distributed.html b/pseudo-distributed.html
index bf104cb..4e72141 100644
--- a/pseudo-distributed.html
+++ b/pseudo-distributed.html
@@ -163,7 +163,7 @@ Running Apache HBase (TM) in pseudo-distributed mode
<div class="row">
<p>Copyright ©2007–2021
<a href="https://www.apache.org/">The Apache Software Foundation</a>.
-All rights reserved. <li id="publishDate" class="pull-right">Last Published: 2021-10-07</li>
+All rights reserved. <li id="publishDate" class="pull-right">Last Published: 2021-10-08</li>
</p>
</div>
<p id="poweredBy" class="pull-right"><a href="http://maven.apache.org/" title="Built by Maven" class="poweredBy"><img class="builtBy" alt="Built by Maven" src="./images/logos/maven-feather.png" /></a>
diff --git a/replication.html b/replication.html
index e88eeab..7003a61 100644
--- a/replication.html
+++ b/replication.html
@@ -158,7 +158,7 @@
<div class="row">
<p>Copyright ©2007–2021
<a href="https://www.apache.org/">The Apache Software Foundation</a>.
-All rights reserved. <li id="publishDate" class="pull-right">Last Published: 2021-10-07</li>
+All rights reserved. <li id="publishDate" class="pull-right">Last Published: 2021-10-08</li>
</p>
</div>
<p id="poweredBy" class="pull-right"><a href="http://maven.apache.org/" title="Built by Maven" class="poweredBy"><img class="builtBy" alt="Built by Maven" src="./images/logos/maven-feather.png" /></a>
diff --git a/resources.html b/resources.html
index c255682..1cc7586 100644
--- a/resources.html
+++ b/resources.html
@@ -186,7 +186,7 @@
<div class="row">
<p>Copyright ©2007–2021
<a href="https://www.apache.org/">The Apache Software Foundation</a>.
-All rights reserved. <li id="publishDate" class="pull-right">Last Published: 2021-10-07</li>
+All rights reserved. <li id="publishDate" class="pull-right">Last Published: 2021-10-08</li>
</p>
</div>
<p id="poweredBy" class="pull-right"><a href="http://maven.apache.org/" title="Built by Maven" class="poweredBy"><img class="builtBy" alt="Built by Maven" src="./images/logos/maven-feather.png" /></a>
diff --git a/source-repository.html b/source-repository.html
index 078eed1..8237e92 100644
--- a/source-repository.html
+++ b/source-repository.html
@@ -169,7 +169,7 @@
<div class="row">
<p>Copyright ©2007–2021
<a href="https://www.apache.org/">The Apache Software Foundation</a>.
-All rights reserved. <li id="publishDate" class="pull-right">Last Published: 2021-10-07</li>
+All rights reserved. <li id="publishDate" class="pull-right">Last Published: 2021-10-08</li>
</p>
</div>
<p id="poweredBy" class="pull-right"><a href="http://maven.apache.org/" title="Built by Maven" class="poweredBy"><img class="builtBy" alt="Built by Maven" src="./images/logos/maven-feather.png" /></a>
diff --git a/sponsors.html b/sponsors.html
index 2446c70..196dd9f 100644
--- a/sponsors.html
+++ b/sponsors.html
@@ -188,7 +188,7 @@
<div class="row">
<p>Copyright ©2007–2021
<a href="https://www.apache.org/">The Apache Software Foundation</a>.
-All rights reserved. <li id="publishDate" class="pull-right">Last Published: 2021-10-07</li>
+All rights reserved. <li id="publishDate" class="pull-right">Last Published: 2021-10-08</li>
</p>
</div>
<p id="poweredBy" class="pull-right"><a href="http://maven.apache.org/" title="Built by Maven" class="poweredBy"><img class="builtBy" alt="Built by Maven" src="./images/logos/maven-feather.png" /></a>
diff --git a/supportingprojects.html b/supportingprojects.html
index 5d366d7..7101a31 100644
--- a/supportingprojects.html
+++ b/supportingprojects.html
@@ -379,7 +379,7 @@
<div class="row">
<p>Copyright ©2007–2021
<a href="https://www.apache.org/">The Apache Software Foundation</a>.
-All rights reserved. <li id="publishDate" class="pull-right">Last Published: 2021-10-07</li>
+All rights reserved. <li id="publishDate" class="pull-right">Last Published: 2021-10-08</li>
</p>
</div>
<p id="poweredBy" class="pull-right"><a href="http://maven.apache.org/" title="Built by Maven" class="poweredBy"><img class="builtBy" alt="Built by Maven" src="./images/logos/maven-feather.png" /></a>
diff --git a/team-list.html b/team-list.html
index cd516f5..fd4b71d 100644
--- a/team-list.html
+++ b/team-list.html
@@ -732,7 +732,7 @@
<div class="row">
<p>Copyright ©2007–2021
<a href="https://www.apache.org/">The Apache Software Foundation</a>.
-All rights reserved. <li id="publishDate" class="pull-right">Last Published: 2021-10-07</li>
+All rights reserved. <li id="publishDate" class="pull-right">Last Published: 2021-10-08</li>
</p>
</div>
<p id="poweredBy" class="pull-right"><a href="http://maven.apache.org/" title="Built by Maven" class="poweredBy"><img class="builtBy" alt="Built by Maven" src="./images/logos/maven-feather.png" /></a>
diff --git a/testdevapidocs/constant-values.html b/testdevapidocs/constant-values.html
index fb0aa07..973574f 100644
--- a/testdevapidocs/constant-values.html
+++ b/testdevapidocs/constant-values.html
@@ -15890,6 +15890,25 @@
</li>
<li class="blockList">
<table class="constantsSummary" border="0" cellpadding="3" cellspacing="0" summary="Constant Field Values table, listing constant fields, and values">
+<caption><span>org.apache.hadoop.hbase.test.<a href="org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.html" title="class in org.apache.hadoop.hbase.test">IntegrationTestLoadCommonCrawl</a></span><span class="tabEnd"> </span></caption>
+<tr>
+<th class="colFirst" scope="col">Modifier and Type</th>
+<th scope="col">Constant Field</th>
+<th class="colLast" scope="col">Value</th>
+</tr>
+<tbody>
+<tr class="altColor">
+<td class="colFirst"><a name="org.apache.hadoop.hbase.test.IntegrationTestLoadCommonCrawl.VERIFICATION_READ_RETRIES">
+<!-- -->
+</a><code>private static final int</code></td>
+<td><code><a href="org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.html#VERIFICATION_READ_RETRIES">VERIFICATION_READ_RETRIES</a></code></td>
+<td class="colLast"><code>10</code></td>
+</tr>
+</tbody>
+</table>
+</li>
+<li class="blockList">
+<table class="constantsSummary" border="0" cellpadding="3" cellspacing="0" summary="Constant Field Values table, listing constant fields, and values">
<caption><span>org.apache.hadoop.hbase.test.<a href="org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.Loader.html" title="class in org.apache.hadoop.hbase.test">IntegrationTestLoadCommonCrawl.Loader</a></span><span class="tabEnd"> </span></caption>
<tr>
<th class="colFirst" scope="col">Modifier and Type</th>
diff --git a/testdevapidocs/index-all.html b/testdevapidocs/index-all.html
index 99954a6..eb8990c 100644
--- a/testdevapidocs/index-all.html
+++ b/testdevapidocs/index-all.html
@@ -31827,6 +31827,8 @@
<dd> </dd>
<dt><span class="memberNameLink"><a href="org/apache/hadoop/hbase/test/IntegrationTestLoadAndVerify.LoadMapper.html#mutator">mutator</a></span> - Variable in class org.apache.hadoop.hbase.test.<a href="org/apache/hadoop/hbase/test/IntegrationTestLoadAndVerify.LoadMapper.html" title="class in org.apache.hadoop.hbase.test">IntegrationTestLoadAndVerify.LoadMapper</a></dt>
<dd> </dd>
+<dt><span class="memberNameLink"><a href="org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.Loader.LoaderMapper.html#mutator">mutator</a></span> - Variable in class org.apache.hadoop.hbase.test.<a href="org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.Loader.LoaderMapper.html" title="class in org.apache.hadoop.hbase.test">IntegrationTestLoadCommonCrawl.Loader.LoaderMapper</a></dt>
+<dd> </dd>
<dt><span class="memberNameLink"><a href="org/apache/hadoop/hbase/mapreduce/TestWALRecordReader.html#mvcc">mvcc</a></span> - Variable in class org.apache.hadoop.hbase.mapreduce.<a href="org/apache/hadoop/hbase/mapreduce/TestWALRecordReader.html" title="class in org.apache.hadoop.hbase.mapreduce">TestWALRecordReader</a></dt>
<dd> </dd>
<dt><span class="memberNameLink"><a href="org/apache/hadoop/hbase/regionserver/TestDefaultMemStore.html#mvcc">mvcc</a></span> - Variable in class org.apache.hadoop.hbase.regionserver.<a href="org/apache/hadoop/hbase/regionserver/TestDefaultMemStore.html" title="class in org.apache.hadoop.hbase.regionserver">TestDefaultMemStore</a></dt>
@@ -49293,8 +49295,6 @@
<dd> </dd>
<dt><span class="memberNameLink"><a href="org/apache/hadoop/hbase/snapshot/TestSnapshotWhenChoreCleaning.html#TABLE">TABLE</a></span> - Static variable in class org.apache.hadoop.hbase.snapshot.<a href="org/apache/hadoop/hbase/snapshot/TestSnapshotWhenChoreCleaning.html" title="class in org.apache.hadoop.hbase.snapshot">TestSnapshotWhenChoreCleaning</a></dt>
<dd> </dd>
-<dt><span class="memberNameLink"><a href="org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.Loader.LoaderMapper.html#table">table</a></span> - Variable in class org.apache.hadoop.hbase.test.<a href="org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.Loader.LoaderMapper.html" title="class in org.apache.hadoop.hbase.test">IntegrationTestLoadCommonCrawl.Loader.LoaderMapper</a></dt>
-<dd> </dd>
<dt><span class="memberNameLink"><a href="org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.Verify.VerifyMapper.html#table">table</a></span> - Variable in class org.apache.hadoop.hbase.test.<a href="org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.Verify.VerifyMapper.html" title="class in org.apache.hadoop.hbase.test">IntegrationTestLoadCommonCrawl.Verify.VerifyMapper</a></dt>
<dd> </dd>
<dt><span class="memberNameLink"><a href="org/apache/hadoop/hbase/test/util/CRC64.html#TABLE">TABLE</a></span> - Static variable in class org.apache.hadoop.hbase.test.util.<a href="org/apache/hadoop/hbase/test/util/CRC64.html" title="class in org.apache.hadoop.hbase.test.util">CRC64</a></dt>
@@ -82053,6 +82053,8 @@ the order they are declared.</div>
<dd> </dd>
<dt><span class="memberNameLink"><a href="org/apache/hadoop/hbase/util/MultiThreadedAction.html#verbose">verbose</a></span> - Variable in class org.apache.hadoop.hbase.util.<a href="org/apache/hadoop/hbase/util/MultiThreadedAction.html" title="class in org.apache.hadoop.hbase.util">MultiThreadedAction</a></dt>
<dd> </dd>
+<dt><span class="memberNameLink"><a href="org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.html#VERIFICATION_READ_RETRIES">VERIFICATION_READ_RETRIES</a></span> - Static variable in class org.apache.hadoop.hbase.test.<a href="org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.html" title="class in org.apache.hadoop.hbase.test">IntegrationTestLoadCommonCrawl</a></dt>
+<dd> </dd>
<dt><span class="memberNameLink"><a href="org/apache/hadoop/hbase/security/access/TestRpcAccessChecks.html#verifiedDeniedServiceException-org.apache.hadoop.hbase.security.User-org.apache.hadoop.hbase.security.access.TestRpcAccessChecks.Action-">verifiedDeniedServiceException(User, TestRpcAccessChecks.Action)</a></span> - Method in class org.apache.hadoop.hbase.security.access.<a href="org/apache/hadoop/hbase/security/access/TestRpcAccessChecks.html" title="class in org.apache.hadoop.hbas [...]
<dd> </dd>
<dt><span class="memberNameLink"><a href="org/apache/hadoop/hbase/IntegrationTestLazyCfLoading.DataGenerator.html#verify-byte:A-byte:A-byte:A-byte:A-">verify(byte[], byte[], byte[], byte[])</a></span> - Method in class org.apache.hadoop.hbase.<a href="org/apache/hadoop/hbase/IntegrationTestLazyCfLoading.DataGenerator.html" title="class in org.apache.hadoop.hbase">IntegrationTestLazyCfLoading.DataGenerator</a></dt>
diff --git a/testdevapidocs/org/apache/hadoop/hbase/package-tree.html b/testdevapidocs/org/apache/hadoop/hbase/package-tree.html
index dbfff9a..533411d 100644
--- a/testdevapidocs/org/apache/hadoop/hbase/package-tree.html
+++ b/testdevapidocs/org/apache/hadoop/hbase/package-tree.html
@@ -665,19 +665,19 @@
<ul>
<li type="circle">java.lang.<a href="https://docs.oracle.com/javase/8/docs/api/java/lang/Enum.html?is-external=true" title="class or interface in java.lang"><span class="typeNameLink">Enum</span></a><E> (implements java.lang.<a href="https://docs.oracle.com/javase/8/docs/api/java/lang/Comparable.html?is-external=true" title="class or interface in java.lang">Comparable</a><T>, java.io.<a href="https://docs.oracle.com/javase/8/docs/api/java/io/Serializable.html?is-external=true [...]
<ul>
-<li type="circle">org.apache.hadoop.hbase.<a href="../../../../org/apache/hadoop/hbase/HBaseClusterManager.Signal.html" title="enum in org.apache.hadoop.hbase"><span class="typeNameLink">HBaseClusterManager.Signal</span></a></li>
-<li type="circle">org.apache.hadoop.hbase.<a href="../../../../org/apache/hadoop/hbase/IntegrationTestRegionReplicaPerf.Stat.html" title="enum in org.apache.hadoop.hbase"><span class="typeNameLink">IntegrationTestRegionReplicaPerf.Stat</span></a></li>
-<li type="circle">org.apache.hadoop.hbase.<a href="../../../../org/apache/hadoop/hbase/PerformanceEvaluation.Counter.html" title="enum in org.apache.hadoop.hbase"><span class="typeNameLink">PerformanceEvaluation.Counter</span></a></li>
-<li type="circle">org.apache.hadoop.hbase.<a href="../../../../org/apache/hadoop/hbase/ResourceChecker.Phase.html" title="enum in org.apache.hadoop.hbase"><span class="typeNameLink">ResourceChecker.Phase</span></a></li>
-<li type="circle">org.apache.hadoop.hbase.<a href="../../../../org/apache/hadoop/hbase/ZNodeClusterManager.CmdType.html" title="enum in org.apache.hadoop.hbase"><span class="typeNameLink">ZNodeClusterManager.CmdType</span></a></li>
-<li type="circle">org.apache.hadoop.hbase.<a href="../../../../org/apache/hadoop/hbase/RESTApiClusterManager.RoleCommand.html" title="enum in org.apache.hadoop.hbase"><span class="typeNameLink">RESTApiClusterManager.RoleCommand</span></a></li>
-<li type="circle">org.apache.hadoop.hbase.<a href="../../../../org/apache/hadoop/hbase/RESTApiClusterManager.Service.html" title="enum in org.apache.hadoop.hbase"><span class="typeNameLink">RESTApiClusterManager.Service</span></a></li>
-<li type="circle">org.apache.hadoop.hbase.<a href="../../../../org/apache/hadoop/hbase/RESTApiClusterManager.HealthSummary.html" title="enum in org.apache.hadoop.hbase"><span class="typeNameLink">RESTApiClusterManager.HealthSummary</span></a></li>
-<li type="circle">org.apache.hadoop.hbase.<a href="../../../../org/apache/hadoop/hbase/IntegrationTestDDLMasterFailover.ACTION.html" title="enum in org.apache.hadoop.hbase"><span class="typeNameLink">IntegrationTestDDLMasterFailover.ACTION</span></a></li>
+<li type="circle">org.apache.hadoop.hbase.<a href="../../../../org/apache/hadoop/hbase/ClusterManager.ServiceType.html" title="enum in org.apache.hadoop.hbase"><span class="typeNameLink">ClusterManager.ServiceType</span></a></li>
<li type="circle">org.apache.hadoop.hbase.<a href="../../../../org/apache/hadoop/hbase/ScanPerformanceEvaluation.ScanCounter.html" title="enum in org.apache.hadoop.hbase"><span class="typeNameLink">ScanPerformanceEvaluation.ScanCounter</span></a></li>
+<li type="circle">org.apache.hadoop.hbase.<a href="../../../../org/apache/hadoop/hbase/IntegrationTestDDLMasterFailover.ACTION.html" title="enum in org.apache.hadoop.hbase"><span class="typeNameLink">IntegrationTestDDLMasterFailover.ACTION</span></a></li>
+<li type="circle">org.apache.hadoop.hbase.<a href="../../../../org/apache/hadoop/hbase/RESTApiClusterManager.HealthSummary.html" title="enum in org.apache.hadoop.hbase"><span class="typeNameLink">RESTApiClusterManager.HealthSummary</span></a></li>
<li type="circle">org.apache.hadoop.hbase.<a href="../../../../org/apache/hadoop/hbase/RESTApiClusterManager.RoleState.html" title="enum in org.apache.hadoop.hbase"><span class="typeNameLink">RESTApiClusterManager.RoleState</span></a></li>
+<li type="circle">org.apache.hadoop.hbase.<a href="../../../../org/apache/hadoop/hbase/RESTApiClusterManager.RoleCommand.html" title="enum in org.apache.hadoop.hbase"><span class="typeNameLink">RESTApiClusterManager.RoleCommand</span></a></li>
+<li type="circle">org.apache.hadoop.hbase.<a href="../../../../org/apache/hadoop/hbase/ZNodeClusterManager.CmdType.html" title="enum in org.apache.hadoop.hbase"><span class="typeNameLink">ZNodeClusterManager.CmdType</span></a></li>
<li type="circle">org.apache.hadoop.hbase.<a href="../../../../org/apache/hadoop/hbase/HBaseClusterManager.CommandProvider.Operation.html" title="enum in org.apache.hadoop.hbase"><span class="typeNameLink">HBaseClusterManager.CommandProvider.Operation</span></a></li>
-<li type="circle">org.apache.hadoop.hbase.<a href="../../../../org/apache/hadoop/hbase/ClusterManager.ServiceType.html" title="enum in org.apache.hadoop.hbase"><span class="typeNameLink">ClusterManager.ServiceType</span></a></li>
+<li type="circle">org.apache.hadoop.hbase.<a href="../../../../org/apache/hadoop/hbase/PerformanceEvaluation.Counter.html" title="enum in org.apache.hadoop.hbase"><span class="typeNameLink">PerformanceEvaluation.Counter</span></a></li>
+<li type="circle">org.apache.hadoop.hbase.<a href="../../../../org/apache/hadoop/hbase/IntegrationTestRegionReplicaPerf.Stat.html" title="enum in org.apache.hadoop.hbase"><span class="typeNameLink">IntegrationTestRegionReplicaPerf.Stat</span></a></li>
+<li type="circle">org.apache.hadoop.hbase.<a href="../../../../org/apache/hadoop/hbase/ResourceChecker.Phase.html" title="enum in org.apache.hadoop.hbase"><span class="typeNameLink">ResourceChecker.Phase</span></a></li>
+<li type="circle">org.apache.hadoop.hbase.<a href="../../../../org/apache/hadoop/hbase/HBaseClusterManager.Signal.html" title="enum in org.apache.hadoop.hbase"><span class="typeNameLink">HBaseClusterManager.Signal</span></a></li>
+<li type="circle">org.apache.hadoop.hbase.<a href="../../../../org/apache/hadoop/hbase/RESTApiClusterManager.Service.html" title="enum in org.apache.hadoop.hbase"><span class="typeNameLink">RESTApiClusterManager.Service</span></a></li>
</ul>
</li>
</ul>
diff --git a/testdevapidocs/org/apache/hadoop/hbase/procedure2/package-tree.html b/testdevapidocs/org/apache/hadoop/hbase/procedure2/package-tree.html
index aa2a449..382d1a5 100644
--- a/testdevapidocs/org/apache/hadoop/hbase/procedure2/package-tree.html
+++ b/testdevapidocs/org/apache/hadoop/hbase/procedure2/package-tree.html
@@ -241,9 +241,9 @@
<ul>
<li type="circle">java.lang.<a href="https://docs.oracle.com/javase/8/docs/api/java/lang/Enum.html?is-external=true" title="class or interface in java.lang"><span class="typeNameLink">Enum</span></a><E> (implements java.lang.<a href="https://docs.oracle.com/javase/8/docs/api/java/lang/Comparable.html?is-external=true" title="class or interface in java.lang">Comparable</a><T>, java.io.<a href="https://docs.oracle.com/javase/8/docs/api/java/io/Serializable.html?is-external=true [...]
<ul>
-<li type="circle">org.apache.hadoop.hbase.procedure2.<a href="../../../../../org/apache/hadoop/hbase/procedure2/TestProcedureRecovery.TestStateMachineProcedure.State.html" title="enum in org.apache.hadoop.hbase.procedure2"><span class="typeNameLink">TestProcedureRecovery.TestStateMachineProcedure.State</span></a></li>
-<li type="circle">org.apache.hadoop.hbase.procedure2.<a href="../../../../../org/apache/hadoop/hbase/procedure2/TestYieldProcedures.TestStateMachineProcedure.State.html" title="enum in org.apache.hadoop.hbase.procedure2"><span class="typeNameLink">TestYieldProcedures.TestStateMachineProcedure.State</span></a></li>
<li type="circle">org.apache.hadoop.hbase.procedure2.<a href="../../../../../org/apache/hadoop/hbase/procedure2/TestProcedureBypass.StuckStateMachineState.html" title="enum in org.apache.hadoop.hbase.procedure2"><span class="typeNameLink">TestProcedureBypass.StuckStateMachineState</span></a></li>
+<li type="circle">org.apache.hadoop.hbase.procedure2.<a href="../../../../../org/apache/hadoop/hbase/procedure2/TestYieldProcedures.TestStateMachineProcedure.State.html" title="enum in org.apache.hadoop.hbase.procedure2"><span class="typeNameLink">TestYieldProcedures.TestStateMachineProcedure.State</span></a></li>
+<li type="circle">org.apache.hadoop.hbase.procedure2.<a href="../../../../../org/apache/hadoop/hbase/procedure2/TestProcedureRecovery.TestStateMachineProcedure.State.html" title="enum in org.apache.hadoop.hbase.procedure2"><span class="typeNameLink">TestProcedureRecovery.TestStateMachineProcedure.State</span></a></li>
<li type="circle">org.apache.hadoop.hbase.procedure2.<a href="../../../../../org/apache/hadoop/hbase/procedure2/TestStateMachineProcedure.TestSMProcedureState.html" title="enum in org.apache.hadoop.hbase.procedure2"><span class="typeNameLink">TestStateMachineProcedure.TestSMProcedureState</span></a></li>
</ul>
</li>
diff --git a/testdevapidocs/org/apache/hadoop/hbase/regionserver/package-tree.html b/testdevapidocs/org/apache/hadoop/hbase/regionserver/package-tree.html
index 82328c4..8becaa2 100644
--- a/testdevapidocs/org/apache/hadoop/hbase/regionserver/package-tree.html
+++ b/testdevapidocs/org/apache/hadoop/hbase/regionserver/package-tree.html
@@ -736,12 +736,12 @@
<ul>
<li type="circle">java.lang.<a href="https://docs.oracle.com/javase/8/docs/api/java/lang/Enum.html?is-external=true" title="class or interface in java.lang"><span class="typeNameLink">Enum</span></a><E> (implements java.lang.<a href="https://docs.oracle.com/javase/8/docs/api/java/lang/Comparable.html?is-external=true" title="class or interface in java.lang">Comparable</a><T>, java.io.<a href="https://docs.oracle.com/javase/8/docs/api/java/io/Serializable.html?is-external=true [...]
<ul>
-<li type="circle">org.apache.hadoop.hbase.regionserver.<a href="../../../../../org/apache/hadoop/hbase/regionserver/DataBlockEncodingTool.Manipulation.html" title="enum in org.apache.hadoop.hbase.regionserver"><span class="typeNameLink">DataBlockEncodingTool.Manipulation</span></a></li>
-<li type="circle">org.apache.hadoop.hbase.regionserver.<a href="../../../../../org/apache/hadoop/hbase/regionserver/TestRegionServerReadRequestMetrics.Metric.html" title="enum in org.apache.hadoop.hbase.regionserver"><span class="typeNameLink">TestRegionServerReadRequestMetrics.Metric</span></a></li>
<li type="circle">org.apache.hadoop.hbase.regionserver.<a href="../../../../../org/apache/hadoop/hbase/regionserver/TestCompactionState.StateSource.html" title="enum in org.apache.hadoop.hbase.regionserver"><span class="typeNameLink">TestCompactionState.StateSource</span></a></li>
+<li type="circle">org.apache.hadoop.hbase.regionserver.<a href="../../../../../org/apache/hadoop/hbase/regionserver/DataBlockEncodingTool.Manipulation.html" title="enum in org.apache.hadoop.hbase.regionserver"><span class="typeNameLink">DataBlockEncodingTool.Manipulation</span></a></li>
+<li type="circle">org.apache.hadoop.hbase.regionserver.<a href="../../../../../org/apache/hadoop/hbase/regionserver/TestMultiLogThreshold.ActionType.html" title="enum in org.apache.hadoop.hbase.regionserver"><span class="typeNameLink">TestMultiLogThreshold.ActionType</span></a></li>
<li type="circle">org.apache.hadoop.hbase.regionserver.<a href="../../../../../org/apache/hadoop/hbase/regionserver/TestAtomicOperation.TestStep.html" title="enum in org.apache.hadoop.hbase.regionserver"><span class="typeNameLink">TestAtomicOperation.TestStep</span></a></li>
<li type="circle">org.apache.hadoop.hbase.regionserver.<a href="../../../../../org/apache/hadoop/hbase/regionserver/TestCacheOnWriteInSchema.CacheOnWriteType.html" title="enum in org.apache.hadoop.hbase.regionserver"><span class="typeNameLink">TestCacheOnWriteInSchema.CacheOnWriteType</span></a></li>
-<li type="circle">org.apache.hadoop.hbase.regionserver.<a href="../../../../../org/apache/hadoop/hbase/regionserver/TestMultiLogThreshold.ActionType.html" title="enum in org.apache.hadoop.hbase.regionserver"><span class="typeNameLink">TestMultiLogThreshold.ActionType</span></a></li>
+<li type="circle">org.apache.hadoop.hbase.regionserver.<a href="../../../../../org/apache/hadoop/hbase/regionserver/TestRegionServerReadRequestMetrics.Metric.html" title="enum in org.apache.hadoop.hbase.regionserver"><span class="typeNameLink">TestRegionServerReadRequestMetrics.Metric</span></a></li>
</ul>
</li>
</ul>
diff --git a/testdevapidocs/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.Counts.html b/testdevapidocs/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.Counts.html
index ce73c6d..80ea598 100644
--- a/testdevapidocs/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.Counts.html
+++ b/testdevapidocs/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.Counts.html
@@ -122,7 +122,7 @@ var activeTableTab = "activeTableTab";
</dl>
<hr>
<br>
-<pre>public static enum <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.html#line.159">IntegrationTestLoadCommonCrawl.Counts</a>
+<pre>public static enum <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.html#line.162">IntegrationTestLoadCommonCrawl.Counts</a>
extends <a href="https://docs.oracle.com/javase/8/docs/api/java/lang/Enum.html?is-external=true" title="class or interface in java.lang">Enum</a><<a href="../../../../../org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.Counts.html" title="enum in org.apache.hadoop.hbase.test">IntegrationTestLoadCommonCrawl.Counts</a>></pre>
</li>
</ul>
@@ -213,7 +213,7 @@ the order they are declared.</div>
<ul class="blockList">
<li class="blockList">
<h4>REFERENCED</h4>
-<pre>public static final <a href="../../../../../org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.Counts.html" title="enum in org.apache.hadoop.hbase.test">IntegrationTestLoadCommonCrawl.Counts</a> <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.Counts.html#line.160">REFERENCED</a></pre>
+<pre>public static final <a href="../../../../../org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.Counts.html" title="enum in org.apache.hadoop.hbase.test">IntegrationTestLoadCommonCrawl.Counts</a> <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.Counts.html#line.163">REFERENCED</a></pre>
</li>
</ul>
<a name="UNREFERENCED">
@@ -222,7 +222,7 @@ the order they are declared.</div>
<ul class="blockList">
<li class="blockList">
<h4>UNREFERENCED</h4>
-<pre>public static final <a href="../../../../../org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.Counts.html" title="enum in org.apache.hadoop.hbase.test">IntegrationTestLoadCommonCrawl.Counts</a> <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.Counts.html#line.160">UNREFERENCED</a></pre>
+<pre>public static final <a href="../../../../../org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.Counts.html" title="enum in org.apache.hadoop.hbase.test">IntegrationTestLoadCommonCrawl.Counts</a> <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.Counts.html#line.163">UNREFERENCED</a></pre>
</li>
</ul>
<a name="CORRUPT">
@@ -231,7 +231,7 @@ the order they are declared.</div>
<ul class="blockListLast">
<li class="blockList">
<h4>CORRUPT</h4>
-<pre>public static final <a href="../../../../../org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.Counts.html" title="enum in org.apache.hadoop.hbase.test">IntegrationTestLoadCommonCrawl.Counts</a> <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.Counts.html#line.160">CORRUPT</a></pre>
+<pre>public static final <a href="../../../../../org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.Counts.html" title="enum in org.apache.hadoop.hbase.test">IntegrationTestLoadCommonCrawl.Counts</a> <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.Counts.html#line.163">CORRUPT</a></pre>
</li>
</ul>
</li>
@@ -248,7 +248,7 @@ the order they are declared.</div>
<ul class="blockList">
<li class="blockList">
<h4>values</h4>
-<pre>public static <a href="../../../../../org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.Counts.html" title="enum in org.apache.hadoop.hbase.test">IntegrationTestLoadCommonCrawl.Counts</a>[] <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.Counts.html#line.159">values</a>()</pre>
+<pre>public static <a href="../../../../../org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.Counts.html" title="enum in org.apache.hadoop.hbase.test">IntegrationTestLoadCommonCrawl.Counts</a>[] <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.Counts.html#line.162">values</a>()</pre>
<div class="block">Returns an array containing the constants of this enum type, in
the order they are declared. This method may be used to iterate
over the constants as follows:
@@ -268,7 +268,7 @@ for (IntegrationTestLoadCommonCrawl.Counts c : IntegrationTestLoadCommonCrawl.Co
<ul class="blockListLast">
<li class="blockList">
<h4>valueOf</h4>
-<pre>public static <a href="../../../../../org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.Counts.html" title="enum in org.apache.hadoop.hbase.test">IntegrationTestLoadCommonCrawl.Counts</a> <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.Counts.html#line.159">valueOf</a>(<a href="https://docs.oracle.com/javase/8/docs/api/java/lang/String.html?is-external=true" title="class or interface in java.lang">String</a> name)</pre>
+<pre>public static <a href="../../../../../org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.Counts.html" title="enum in org.apache.hadoop.hbase.test">IntegrationTestLoadCommonCrawl.Counts</a> <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.Counts.html#line.162">valueOf</a>(<a href="https://docs.oracle.com/javase/8/docs/api/java/lang/String.html?is-external=true" title="class or interface in java.lang">String</a> name)</pre>
<div class="block">Returns the enum constant of this type with the specified name.
The string must match <i>exactly</i> an identifier used to declare an
enum constant in this type. (Extraneous whitespace characters are
diff --git a/testdevapidocs/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.HBaseKeyWritable.html b/testdevapidocs/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.HBaseKeyWritable.html
index ed002bd..249c902 100644
--- a/testdevapidocs/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.HBaseKeyWritable.html
+++ b/testdevapidocs/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.HBaseKeyWritable.html
@@ -117,7 +117,7 @@ var activeTableTab = "activeTableTab";
</dl>
<hr>
<br>
-<pre>public static class <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.html#line.261">IntegrationTestLoadCommonCrawl.HBaseKeyWritable</a>
+<pre>public static class <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.html#line.264">IntegrationTestLoadCommonCrawl.HBaseKeyWritable</a>
extends <a href="https://docs.oracle.com/javase/8/docs/api/java/lang/Object.html?is-external=true" title="class or interface in java.lang">Object</a>
implements org.apache.hadoop.io.Writable</pre>
</li>
@@ -352,7 +352,7 @@ implements org.apache.hadoop.io.Writable</pre>
<ul class="blockList">
<li class="blockList">
<h4>row</h4>
-<pre>private byte[] <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.HBaseKeyWritable.html#line.263">row</a></pre>
+<pre>private byte[] <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.HBaseKeyWritable.html#line.266">row</a></pre>
</li>
</ul>
<a name="rowOffset">
@@ -361,7 +361,7 @@ implements org.apache.hadoop.io.Writable</pre>
<ul class="blockList">
<li class="blockList">
<h4>rowOffset</h4>
-<pre>private int <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.HBaseKeyWritable.html#line.264">rowOffset</a></pre>
+<pre>private int <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.HBaseKeyWritable.html#line.267">rowOffset</a></pre>
</li>
</ul>
<a name="rowLength">
@@ -370,7 +370,7 @@ implements org.apache.hadoop.io.Writable</pre>
<ul class="blockList">
<li class="blockList">
<h4>rowLength</h4>
-<pre>private int <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.HBaseKeyWritable.html#line.265">rowLength</a></pre>
+<pre>private int <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.HBaseKeyWritable.html#line.268">rowLength</a></pre>
</li>
</ul>
<a name="family">
@@ -379,7 +379,7 @@ implements org.apache.hadoop.io.Writable</pre>
<ul class="blockList">
<li class="blockList">
<h4>family</h4>
-<pre>private byte[] <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.HBaseKeyWritable.html#line.266">family</a></pre>
+<pre>private byte[] <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.HBaseKeyWritable.html#line.269">family</a></pre>
</li>
</ul>
<a name="familyOffset">
@@ -388,7 +388,7 @@ implements org.apache.hadoop.io.Writable</pre>
<ul class="blockList">
<li class="blockList">
<h4>familyOffset</h4>
-<pre>private int <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.HBaseKeyWritable.html#line.267">familyOffset</a></pre>
+<pre>private int <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.HBaseKeyWritable.html#line.270">familyOffset</a></pre>
</li>
</ul>
<a name="familyLength">
@@ -397,7 +397,7 @@ implements org.apache.hadoop.io.Writable</pre>
<ul class="blockList">
<li class="blockList">
<h4>familyLength</h4>
-<pre>private int <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.HBaseKeyWritable.html#line.268">familyLength</a></pre>
+<pre>private int <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.HBaseKeyWritable.html#line.271">familyLength</a></pre>
</li>
</ul>
<a name="qualifier">
@@ -406,7 +406,7 @@ implements org.apache.hadoop.io.Writable</pre>
<ul class="blockList">
<li class="blockList">
<h4>qualifier</h4>
-<pre>private byte[] <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.HBaseKeyWritable.html#line.269">qualifier</a></pre>
+<pre>private byte[] <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.HBaseKeyWritable.html#line.272">qualifier</a></pre>
</li>
</ul>
<a name="qualifierOffset">
@@ -415,7 +415,7 @@ implements org.apache.hadoop.io.Writable</pre>
<ul class="blockList">
<li class="blockList">
<h4>qualifierOffset</h4>
-<pre>private int <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.HBaseKeyWritable.html#line.270">qualifierOffset</a></pre>
+<pre>private int <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.HBaseKeyWritable.html#line.273">qualifierOffset</a></pre>
</li>
</ul>
<a name="qualifierLength">
@@ -424,7 +424,7 @@ implements org.apache.hadoop.io.Writable</pre>
<ul class="blockList">
<li class="blockList">
<h4>qualifierLength</h4>
-<pre>private int <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.HBaseKeyWritable.html#line.271">qualifierLength</a></pre>
+<pre>private int <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.HBaseKeyWritable.html#line.274">qualifierLength</a></pre>
</li>
</ul>
<a name="ts">
@@ -433,7 +433,7 @@ implements org.apache.hadoop.io.Writable</pre>
<ul class="blockListLast">
<li class="blockList">
<h4>ts</h4>
-<pre>private long <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.HBaseKeyWritable.html#line.272">ts</a></pre>
+<pre>private long <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.HBaseKeyWritable.html#line.275">ts</a></pre>
</li>
</ul>
</li>
@@ -450,7 +450,7 @@ implements org.apache.hadoop.io.Writable</pre>
<ul class="blockList">
<li class="blockList">
<h4>HBaseKeyWritable</h4>
-<pre>public <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.HBaseKeyWritable.html#line.274">HBaseKeyWritable</a>()</pre>
+<pre>public <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.HBaseKeyWritable.html#line.277">HBaseKeyWritable</a>()</pre>
</li>
</ul>
<a name="HBaseKeyWritable-byte:A-int-int-byte:A-int-int-byte:A-int-int-long-">
@@ -459,7 +459,7 @@ implements org.apache.hadoop.io.Writable</pre>
<ul class="blockList">
<li class="blockList">
<h4>HBaseKeyWritable</h4>
-<pre>public <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.HBaseKeyWritable.html#line.276">HBaseKeyWritable</a>(byte[] row,
+<pre>public <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.HBaseKeyWritable.html#line.279">HBaseKeyWritable</a>(byte[] row,
int rowOffset,
int rowLength,
byte[] family,
@@ -477,7 +477,7 @@ implements org.apache.hadoop.io.Writable</pre>
<ul class="blockList">
<li class="blockList">
<h4>HBaseKeyWritable</h4>
-<pre>public <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.HBaseKeyWritable.html#line.291">HBaseKeyWritable</a>(byte[] row,
+<pre>public <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.HBaseKeyWritable.html#line.294">HBaseKeyWritable</a>(byte[] row,
byte[] family,
byte[] qualifier,
long ts)</pre>
@@ -489,7 +489,7 @@ implements org.apache.hadoop.io.Writable</pre>
<ul class="blockList">
<li class="blockList">
<h4>HBaseKeyWritable</h4>
-<pre>public <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.HBaseKeyWritable.html#line.297">HBaseKeyWritable</a>(byte[] row,
+<pre>public <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.HBaseKeyWritable.html#line.300">HBaseKeyWritable</a>(byte[] row,
byte[] family,
long ts)</pre>
</li>
@@ -500,7 +500,7 @@ implements org.apache.hadoop.io.Writable</pre>
<ul class="blockListLast">
<li class="blockList">
<h4>HBaseKeyWritable</h4>
-<pre>public <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.HBaseKeyWritable.html#line.301">HBaseKeyWritable</a>(org.apache.hadoop.hbase.Cell cell)</pre>
+<pre>public <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.HBaseKeyWritable.html#line.304">HBaseKeyWritable</a>(org.apache.hadoop.hbase.Cell cell)</pre>
</li>
</ul>
</li>
@@ -517,7 +517,7 @@ implements org.apache.hadoop.io.Writable</pre>
<ul class="blockList">
<li class="blockList">
<h4>readFields</h4>
-<pre>public void <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.HBaseKeyWritable.html#line.309">readFields</a>(<a href="https://docs.oracle.com/javase/8/docs/api/java/io/DataInput.html?is-external=true" title="class or interface in java.io">DataInput</a> in)
+<pre>public void <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.HBaseKeyWritable.html#line.312">readFields</a>(<a href="https://docs.oracle.com/javase/8/docs/api/java/io/DataInput.html?is-external=true" title="class or interface in java.io">DataInput</a> in)
throws <a href="https://docs.oracle.com/javase/8/docs/api/java/io/IOException.html?is-external=true" title="class or interface in java.io">IOException</a></pre>
<dl>
<dt><span class="overrideSpecifyLabel">Specified by:</span></dt>
@@ -533,7 +533,7 @@ implements org.apache.hadoop.io.Writable</pre>
<ul class="blockList">
<li class="blockList">
<h4>write</h4>
-<pre>public void <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.HBaseKeyWritable.html#line.323">write</a>(<a href="https://docs.oracle.com/javase/8/docs/api/java/io/DataOutput.html?is-external=true" title="class or interface in java.io">DataOutput</a> out)
+<pre>public void <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.HBaseKeyWritable.html#line.326">write</a>(<a href="https://docs.oracle.com/javase/8/docs/api/java/io/DataOutput.html?is-external=true" title="class or interface in java.io">DataOutput</a> out)
throws <a href="https://docs.oracle.com/javase/8/docs/api/java/io/IOException.html?is-external=true" title="class or interface in java.io">IOException</a></pre>
<dl>
<dt><span class="overrideSpecifyLabel">Specified by:</span></dt>
@@ -549,7 +549,7 @@ implements org.apache.hadoop.io.Writable</pre>
<ul class="blockList">
<li class="blockList">
<h4>getRowArray</h4>
-<pre>public byte[] <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.HBaseKeyWritable.html#line.335">getRowArray</a>()</pre>
+<pre>public byte[] <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.HBaseKeyWritable.html#line.338">getRowArray</a>()</pre>
</li>
</ul>
<a name="setRow-byte:A-">
@@ -558,7 +558,7 @@ implements org.apache.hadoop.io.Writable</pre>
<ul class="blockList">
<li class="blockList">
<h4>setRow</h4>
-<pre>public void <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.HBaseKeyWritable.html#line.339">setRow</a>(byte[] row)</pre>
+<pre>public void <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.HBaseKeyWritable.html#line.342">setRow</a>(byte[] row)</pre>
</li>
</ul>
<a name="getRowOffset--">
@@ -567,7 +567,7 @@ implements org.apache.hadoop.io.Writable</pre>
<ul class="blockList">
<li class="blockList">
<h4>getRowOffset</h4>
-<pre>public int <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.HBaseKeyWritable.html#line.343">getRowOffset</a>()</pre>
+<pre>public int <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.HBaseKeyWritable.html#line.346">getRowOffset</a>()</pre>
</li>
</ul>
<a name="setRowOffset-int-">
@@ -576,7 +576,7 @@ implements org.apache.hadoop.io.Writable</pre>
<ul class="blockList">
<li class="blockList">
<h4>setRowOffset</h4>
-<pre>public void <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.HBaseKeyWritable.html#line.347">setRowOffset</a>(int rowOffset)</pre>
+<pre>public void <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.HBaseKeyWritable.html#line.350">setRowOffset</a>(int rowOffset)</pre>
</li>
</ul>
<a name="getRowLength--">
@@ -585,7 +585,7 @@ implements org.apache.hadoop.io.Writable</pre>
<ul class="blockList">
<li class="blockList">
<h4>getRowLength</h4>
-<pre>public int <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.HBaseKeyWritable.html#line.351">getRowLength</a>()</pre>
+<pre>public int <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.HBaseKeyWritable.html#line.354">getRowLength</a>()</pre>
</li>
</ul>
<a name="setRowLength-int-">
@@ -594,7 +594,7 @@ implements org.apache.hadoop.io.Writable</pre>
<ul class="blockList">
<li class="blockList">
<h4>setRowLength</h4>
-<pre>public void <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.HBaseKeyWritable.html#line.355">setRowLength</a>(int rowLength)</pre>
+<pre>public void <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.HBaseKeyWritable.html#line.358">setRowLength</a>(int rowLength)</pre>
</li>
</ul>
<a name="getFamilyArray--">
@@ -603,7 +603,7 @@ implements org.apache.hadoop.io.Writable</pre>
<ul class="blockList">
<li class="blockList">
<h4>getFamilyArray</h4>
-<pre>public byte[] <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.HBaseKeyWritable.html#line.359">getFamilyArray</a>()</pre>
+<pre>public byte[] <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.HBaseKeyWritable.html#line.362">getFamilyArray</a>()</pre>
</li>
</ul>
<a name="setFamily-byte:A-">
@@ -612,7 +612,7 @@ implements org.apache.hadoop.io.Writable</pre>
<ul class="blockList">
<li class="blockList">
<h4>setFamily</h4>
-<pre>public void <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.HBaseKeyWritable.html#line.363">setFamily</a>(byte[] family)</pre>
+<pre>public void <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.HBaseKeyWritable.html#line.366">setFamily</a>(byte[] family)</pre>
</li>
</ul>
<a name="getFamilyOffset--">
@@ -621,7 +621,7 @@ implements org.apache.hadoop.io.Writable</pre>
<ul class="blockList">
<li class="blockList">
<h4>getFamilyOffset</h4>
-<pre>public int <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.HBaseKeyWritable.html#line.367">getFamilyOffset</a>()</pre>
+<pre>public int <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.HBaseKeyWritable.html#line.370">getFamilyOffset</a>()</pre>
</li>
</ul>
<a name="setFamilyOffset-int-">
@@ -630,7 +630,7 @@ implements org.apache.hadoop.io.Writable</pre>
<ul class="blockList">
<li class="blockList">
<h4>setFamilyOffset</h4>
-<pre>public void <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.HBaseKeyWritable.html#line.371">setFamilyOffset</a>(int familyOffset)</pre>
+<pre>public void <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.HBaseKeyWritable.html#line.374">setFamilyOffset</a>(int familyOffset)</pre>
</li>
</ul>
<a name="getFamilyLength--">
@@ -639,7 +639,7 @@ implements org.apache.hadoop.io.Writable</pre>
<ul class="blockList">
<li class="blockList">
<h4>getFamilyLength</h4>
-<pre>public int <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.HBaseKeyWritable.html#line.375">getFamilyLength</a>()</pre>
+<pre>public int <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.HBaseKeyWritable.html#line.378">getFamilyLength</a>()</pre>
</li>
</ul>
<a name="setFamilyLength-int-">
@@ -648,7 +648,7 @@ implements org.apache.hadoop.io.Writable</pre>
<ul class="blockList">
<li class="blockList">
<h4>setFamilyLength</h4>
-<pre>public void <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.HBaseKeyWritable.html#line.379">setFamilyLength</a>(int familyLength)</pre>
+<pre>public void <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.HBaseKeyWritable.html#line.382">setFamilyLength</a>(int familyLength)</pre>
</li>
</ul>
<a name="getQualifierArray--">
@@ -657,7 +657,7 @@ implements org.apache.hadoop.io.Writable</pre>
<ul class="blockList">
<li class="blockList">
<h4>getQualifierArray</h4>
-<pre>public byte[] <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.HBaseKeyWritable.html#line.383">getQualifierArray</a>()</pre>
+<pre>public byte[] <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.HBaseKeyWritable.html#line.386">getQualifierArray</a>()</pre>
</li>
</ul>
<a name="setQualifier-byte:A-">
@@ -666,7 +666,7 @@ implements org.apache.hadoop.io.Writable</pre>
<ul class="blockList">
<li class="blockList">
<h4>setQualifier</h4>
-<pre>public void <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.HBaseKeyWritable.html#line.387">setQualifier</a>(byte[] qualifier)</pre>
+<pre>public void <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.HBaseKeyWritable.html#line.390">setQualifier</a>(byte[] qualifier)</pre>
</li>
</ul>
<a name="getQualifierOffset--">
@@ -675,7 +675,7 @@ implements org.apache.hadoop.io.Writable</pre>
<ul class="blockList">
<li class="blockList">
<h4>getQualifierOffset</h4>
-<pre>public int <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.HBaseKeyWritable.html#line.391">getQualifierOffset</a>()</pre>
+<pre>public int <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.HBaseKeyWritable.html#line.394">getQualifierOffset</a>()</pre>
</li>
</ul>
<a name="setQualifierOffset-int-">
@@ -684,7 +684,7 @@ implements org.apache.hadoop.io.Writable</pre>
<ul class="blockList">
<li class="blockList">
<h4>setQualifierOffset</h4>
-<pre>public void <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.HBaseKeyWritable.html#line.395">setQualifierOffset</a>(int qualifierOffset)</pre>
+<pre>public void <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.HBaseKeyWritable.html#line.398">setQualifierOffset</a>(int qualifierOffset)</pre>
</li>
</ul>
<a name="getQualifierLength--">
@@ -693,7 +693,7 @@ implements org.apache.hadoop.io.Writable</pre>
<ul class="blockList">
<li class="blockList">
<h4>getQualifierLength</h4>
-<pre>public int <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.HBaseKeyWritable.html#line.399">getQualifierLength</a>()</pre>
+<pre>public int <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.HBaseKeyWritable.html#line.402">getQualifierLength</a>()</pre>
</li>
</ul>
<a name="setQualifierLength-int-">
@@ -702,7 +702,7 @@ implements org.apache.hadoop.io.Writable</pre>
<ul class="blockList">
<li class="blockList">
<h4>setQualifierLength</h4>
-<pre>public void <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.HBaseKeyWritable.html#line.403">setQualifierLength</a>(int qualifierLength)</pre>
+<pre>public void <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.HBaseKeyWritable.html#line.406">setQualifierLength</a>(int qualifierLength)</pre>
</li>
</ul>
<a name="getTimestamp--">
@@ -711,7 +711,7 @@ implements org.apache.hadoop.io.Writable</pre>
<ul class="blockList">
<li class="blockList">
<h4>getTimestamp</h4>
-<pre>public long <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.HBaseKeyWritable.html#line.407">getTimestamp</a>()</pre>
+<pre>public long <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.HBaseKeyWritable.html#line.410">getTimestamp</a>()</pre>
</li>
</ul>
<a name="setTimestamp-long-">
@@ -720,7 +720,7 @@ implements org.apache.hadoop.io.Writable</pre>
<ul class="blockListLast">
<li class="blockList">
<h4>setTimestamp</h4>
-<pre>public void <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.HBaseKeyWritable.html#line.411">setTimestamp</a>(long ts)</pre>
+<pre>public void <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.HBaseKeyWritable.html#line.414">setTimestamp</a>(long ts)</pre>
</li>
</ul>
</li>
diff --git a/testdevapidocs/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.Loader.LoaderMapper.html b/testdevapidocs/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.Loader.LoaderMapper.html
index dedfff3..86146b7 100644
--- a/testdevapidocs/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.Loader.LoaderMapper.html
+++ b/testdevapidocs/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.Loader.LoaderMapper.html
@@ -118,7 +118,7 @@ var activeTableTab = "activeTableTab";
</dl>
<hr>
<br>
-<pre>public static class <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.Loader.html#line.549">IntegrationTestLoadCommonCrawl.Loader.LoaderMapper</a>
+<pre>public static class <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.Loader.html#line.550">IntegrationTestLoadCommonCrawl.Loader.LoaderMapper</a>
extends org.apache.hadoop.mapreduce.Mapper<org.apache.hadoop.io.LongWritable,<a href="../../../../../org/apache/hadoop/hbase/test/util/warc/WARCWritable.html" title="class in org.apache.hadoop.hbase.test.util.warc">WARCWritable</a>,<a href="../../../../../org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.HBaseKeyWritable.html" title="class in org.apache.hadoop.hbase.test">IntegrationTestLoadCommonCrawl.HBaseKeyWritable</a>,org.apache.hadoop.io.BytesWritable></pre>
</li>
</ul>
@@ -162,8 +162,8 @@ extends org.apache.hadoop.mapreduce.Mapper<org.apache.hadoop.io.LongWritable,
<td class="colLast"><code><span class="memberNameLink"><a href="../../../../../org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.Loader.LoaderMapper.html#conn">conn</a></span></code> </td>
</tr>
<tr class="altColor">
-<td class="colFirst"><code>(package private) org.apache.hadoop.hbase.client.Table</code></td>
-<td class="colLast"><code><span class="memberNameLink"><a href="../../../../../org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.Loader.LoaderMapper.html#table">table</a></span></code> </td>
+<td class="colFirst"><code>(package private) org.apache.hadoop.hbase.client.BufferedMutator</code></td>
+<td class="colLast"><code><span class="memberNameLink"><a href="../../../../../org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.Loader.LoaderMapper.html#mutator">mutator</a></span></code> </td>
</tr>
</table>
</li>
@@ -250,7 +250,7 @@ extends org.apache.hadoop.mapreduce.Mapper<org.apache.hadoop.io.LongWritable,
<ul class="blockList">
<li class="blockList">
<h4>conf</h4>
-<pre>org.apache.hadoop.conf.Configuration <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.Loader.LoaderMapper.html#line.552">conf</a></pre>
+<pre>org.apache.hadoop.conf.Configuration <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.Loader.LoaderMapper.html#line.553">conf</a></pre>
</li>
</ul>
<a name="conn">
@@ -259,16 +259,16 @@ extends org.apache.hadoop.mapreduce.Mapper<org.apache.hadoop.io.LongWritable,
<ul class="blockList">
<li class="blockList">
<h4>conn</h4>
-<pre>org.apache.hadoop.hbase.client.Connection <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.Loader.LoaderMapper.html#line.553">conn</a></pre>
+<pre>org.apache.hadoop.hbase.client.Connection <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.Loader.LoaderMapper.html#line.554">conn</a></pre>
</li>
</ul>
-<a name="table">
+<a name="mutator">
<!-- -->
</a>
<ul class="blockListLast">
<li class="blockList">
-<h4>table</h4>
-<pre>org.apache.hadoop.hbase.client.Table <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.Loader.LoaderMapper.html#line.554">table</a></pre>
+<h4>mutator</h4>
+<pre>org.apache.hadoop.hbase.client.BufferedMutator <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.Loader.LoaderMapper.html#line.555">mutator</a></pre>
</li>
</ul>
</li>
@@ -285,7 +285,7 @@ extends org.apache.hadoop.mapreduce.Mapper<org.apache.hadoop.io.LongWritable,
<ul class="blockListLast">
<li class="blockList">
<h4>LoaderMapper</h4>
-<pre>public <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.Loader.LoaderMapper.html#line.549">LoaderMapper</a>()</pre>
+<pre>public <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.Loader.LoaderMapper.html#line.550">LoaderMapper</a>()</pre>
</li>
</ul>
</li>
@@ -302,7 +302,7 @@ extends org.apache.hadoop.mapreduce.Mapper<org.apache.hadoop.io.LongWritable,
<ul class="blockList">
<li class="blockList">
<h4>setup</h4>
-<pre>protected void <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.Loader.LoaderMapper.html#line.557">setup</a>(org.apache.hadoop.mapreduce.Mapper.Context context)
+<pre>protected void <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.Loader.LoaderMapper.html#line.558">setup</a>(org.apache.hadoop.mapreduce.Mapper.Context context)
throws <a href="https://docs.oracle.com/javase/8/docs/api/java/io/IOException.html?is-external=true" title="class or interface in java.io">IOException</a>,
<a href="https://docs.oracle.com/javase/8/docs/api/java/lang/InterruptedException.html?is-external=true" title="class or interface in java.lang">InterruptedException</a></pre>
<dl>
@@ -320,7 +320,7 @@ extends org.apache.hadoop.mapreduce.Mapper<org.apache.hadoop.io.LongWritable,
<ul class="blockList">
<li class="blockList">
<h4>cleanup</h4>
-<pre>protected void <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.Loader.LoaderMapper.html#line.563">cleanup</a>(org.apache.hadoop.mapreduce.Mapper.Context context)
+<pre>protected void <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.Loader.LoaderMapper.html#line.566">cleanup</a>(org.apache.hadoop.mapreduce.Mapper.Context context)
throws <a href="https://docs.oracle.com/javase/8/docs/api/java/io/IOException.html?is-external=true" title="class or interface in java.io">IOException</a>,
<a href="https://docs.oracle.com/javase/8/docs/api/java/lang/InterruptedException.html?is-external=true" title="class or interface in java.lang">InterruptedException</a></pre>
<dl>
@@ -338,7 +338,7 @@ extends org.apache.hadoop.mapreduce.Mapper<org.apache.hadoop.io.LongWritable,
<ul class="blockList">
<li class="blockList">
<h4>map</h4>
-<pre>protected void <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.Loader.LoaderMapper.html#line.577">map</a>(org.apache.hadoop.io.LongWritable key,
+<pre>protected void <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.Loader.LoaderMapper.html#line.580">map</a>(org.apache.hadoop.io.LongWritable key,
<a href="../../../../../org/apache/hadoop/hbase/test/util/warc/WARCWritable.html" title="class in org.apache.hadoop.hbase.test.util.warc">WARCWritable</a> value,
org.apache.hadoop.mapreduce.Mapper.Context output)
throws <a href="https://docs.oracle.com/javase/8/docs/api/java/io/IOException.html?is-external=true" title="class or interface in java.io">IOException</a>,
@@ -358,7 +358,7 @@ extends org.apache.hadoop.mapreduce.Mapper<org.apache.hadoop.io.LongWritable,
<ul class="blockListLast">
<li class="blockList">
<h4>rowKeyFromTargetURI</h4>
-<pre>private byte[] <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.Loader.LoaderMapper.html#line.651">rowKeyFromTargetURI</a>(<a href="https://docs.oracle.com/javase/8/docs/api/java/lang/String.html?is-external=true" title="class or interface in java.lang">String</a> targetUri)
+<pre>private byte[] <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.Loader.LoaderMapper.html#line.654">rowKeyFromTargetURI</a>(<a href="https://docs.oracle.com/javase/8/docs/api/java/lang/String.html?is-external=true" title="class or interface in java.lang">String</a> targetUri)
throws <a href="https://docs.oracle.com/javase/8/docs/api/java/net/URISyntaxException.html?is-external=true" title="class or interface in java.net">URISyntaxException</a>,
<a href="https://docs.oracle.com/javase/8/docs/api/java/lang/IllegalArgumentException.html?is-external=true" title="class or interface in java.lang">IllegalArgumentException</a></pre>
<dl>
diff --git a/testdevapidocs/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.Loader.html b/testdevapidocs/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.Loader.html
index bcb0090..ff89cc5 100644
--- a/testdevapidocs/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.Loader.html
+++ b/testdevapidocs/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.Loader.html
@@ -122,7 +122,7 @@ var activeTableTab = "activeTableTab";
</dl>
<hr>
<br>
-<pre>public static class <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.html#line.416">IntegrationTestLoadCommonCrawl.Loader</a>
+<pre>public static class <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.html#line.419">IntegrationTestLoadCommonCrawl.Loader</a>
extends org.apache.hadoop.conf.Configured
implements org.apache.hadoop.util.Tool</pre>
</li>
@@ -261,7 +261,7 @@ implements org.apache.hadoop.util.Tool</pre>
<ul class="blockList">
<li class="blockList">
<h4>LOG</h4>
-<pre>private static final org.slf4j.Logger <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.Loader.html#line.418">LOG</a></pre>
+<pre>private static final org.slf4j.Logger <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.Loader.html#line.421">LOG</a></pre>
</li>
</ul>
<a name="USAGE">
@@ -270,7 +270,7 @@ implements org.apache.hadoop.util.Tool</pre>
<ul class="blockListLast">
<li class="blockList">
<h4>USAGE</h4>
-<pre>private static final <a href="https://docs.oracle.com/javase/8/docs/api/java/lang/String.html?is-external=true" title="class or interface in java.lang">String</a> <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.Loader.html#line.419">USAGE</a></pre>
+<pre>private static final <a href="https://docs.oracle.com/javase/8/docs/api/java/lang/String.html?is-external=true" title="class or interface in java.lang">String</a> <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.Loader.html#line.422">USAGE</a></pre>
<dl>
<dt><span class="seeLabel">See Also:</span></dt>
<dd><a href="../../../../../constant-values.html#org.apache.hadoop.hbase.test.IntegrationTestLoadCommonCrawl.Loader.USAGE">Constant Field Values</a></dd>
@@ -291,7 +291,7 @@ implements org.apache.hadoop.util.Tool</pre>
<ul class="blockListLast">
<li class="blockList">
<h4>Loader</h4>
-<pre>public <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.Loader.html#line.416">Loader</a>()</pre>
+<pre>public <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.Loader.html#line.419">Loader</a>()</pre>
</li>
</ul>
</li>
@@ -308,7 +308,7 @@ implements org.apache.hadoop.util.Tool</pre>
<ul class="blockList">
<li class="blockList">
<h4>createSchema</h4>
-<pre>void <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.Loader.html#line.421">createSchema</a>(org.apache.hadoop.hbase.TableName tableName)
+<pre>void <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.Loader.html#line.424">createSchema</a>(org.apache.hadoop.hbase.TableName tableName)
throws <a href="https://docs.oracle.com/javase/8/docs/api/java/io/IOException.html?is-external=true" title="class or interface in java.io">IOException</a></pre>
<dl>
<dt><span class="throwsLabel">Throws:</span></dt>
@@ -322,7 +322,7 @@ implements org.apache.hadoop.util.Tool</pre>
<ul class="blockList">
<li class="blockList">
<h4>run</h4>
-<pre>int <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.Loader.html#line.477">run</a>(org.apache.hadoop.fs.Path warcFileInput,
+<pre>int <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.Loader.html#line.480">run</a>(org.apache.hadoop.fs.Path warcFileInput,
org.apache.hadoop.fs.Path outputDir)
throws <a href="https://docs.oracle.com/javase/8/docs/api/java/io/IOException.html?is-external=true" title="class or interface in java.io">IOException</a>,
<a href="https://docs.oracle.com/javase/8/docs/api/java/lang/ClassNotFoundException.html?is-external=true" title="class or interface in java.lang">ClassNotFoundException</a>,
@@ -341,7 +341,7 @@ implements org.apache.hadoop.util.Tool</pre>
<ul class="blockList">
<li class="blockList">
<h4>run</h4>
-<pre>public int <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.Loader.html#line.529">run</a>(<a href="https://docs.oracle.com/javase/8/docs/api/java/lang/String.html?is-external=true" title="class or interface in java.lang">String</a>[] args)
+<pre>public int <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.Loader.html#line.530">run</a>(<a href="https://docs.oracle.com/javase/8/docs/api/java/lang/String.html?is-external=true" title="class or interface in java.lang">String</a>[] args)
throws <a href="https://docs.oracle.com/javase/8/docs/api/java/lang/Exception.html?is-external=true" title="class or interface in java.lang">Exception</a></pre>
<dl>
<dt><span class="overrideSpecifyLabel">Specified by:</span></dt>
@@ -357,7 +357,7 @@ implements org.apache.hadoop.util.Tool</pre>
<ul class="blockListLast">
<li class="blockList">
<h4>main</h4>
-<pre>public static void <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.Loader.html#line.545">main</a>(<a href="https://docs.oracle.com/javase/8/docs/api/java/lang/String.html?is-external=true" title="class or interface in java.lang">String</a>[] args)
+<pre>public static void <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.Loader.html#line.546">main</a>(<a href="https://docs.oracle.com/javase/8/docs/api/java/lang/String.html?is-external=true" title="class or interface in java.lang">String</a>[] args)
throws <a href="https://docs.oracle.com/javase/8/docs/api/java/lang/Exception.html?is-external=true" title="class or interface in java.lang">Exception</a></pre>
<dl>
<dt><span class="throwsLabel">Throws:</span></dt>
diff --git a/testdevapidocs/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.OneFilePerMapperSFIF.html b/testdevapidocs/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.OneFilePerMapperSFIF.html
index 96cc413..062999a 100644
--- a/testdevapidocs/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.OneFilePerMapperSFIF.html
+++ b/testdevapidocs/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.OneFilePerMapperSFIF.html
@@ -128,7 +128,7 @@ var activeTableTab = "activeTableTab";
</dl>
<hr>
<br>
-<pre>public static class <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.html#line.694">IntegrationTestLoadCommonCrawl.OneFilePerMapperSFIF</a><K,V>
+<pre>public static class <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.html#line.701">IntegrationTestLoadCommonCrawl.OneFilePerMapperSFIF</a><K,V>
extends org.apache.hadoop.mapreduce.lib.input.SequenceFileInputFormat<K,V></pre>
</li>
</ul>
@@ -242,7 +242,7 @@ extends org.apache.hadoop.mapreduce.lib.input.SequenceFileInputFormat<K,V>
<ul class="blockListLast">
<li class="blockList">
<h4>OneFilePerMapperSFIF</h4>
-<pre>public <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.OneFilePerMapperSFIF.html#line.694">OneFilePerMapperSFIF</a>()</pre>
+<pre>public <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.OneFilePerMapperSFIF.html#line.701">OneFilePerMapperSFIF</a>()</pre>
</li>
</ul>
</li>
@@ -259,7 +259,7 @@ extends org.apache.hadoop.mapreduce.lib.input.SequenceFileInputFormat<K,V>
<ul class="blockListLast">
<li class="blockList">
<h4>isSplitable</h4>
-<pre>protected boolean <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.OneFilePerMapperSFIF.html#line.696">isSplitable</a>(org.apache.hadoop.mapreduce.JobContext context,
+<pre>protected boolean <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.OneFilePerMapperSFIF.html#line.703">isSplitable</a>(org.apache.hadoop.mapreduce.JobContext context,
org.apache.hadoop.fs.Path filename)</pre>
<dl>
<dt><span class="overrideSpecifyLabel">Overrides:</span></dt>
diff --git a/testdevapidocs/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.Verify.VerifyMapper.html b/testdevapidocs/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.Verify.VerifyMapper.html
index 6d21374..08063c6 100644
--- a/testdevapidocs/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.Verify.VerifyMapper.html
+++ b/testdevapidocs/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.Verify.VerifyMapper.html
@@ -118,7 +118,7 @@ var activeTableTab = "activeTableTab";
</dl>
<hr>
<br>
-<pre>public static class <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.Verify.html#line.742">IntegrationTestLoadCommonCrawl.Verify.VerifyMapper</a>
+<pre>public static class <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.Verify.html#line.749">IntegrationTestLoadCommonCrawl.Verify.VerifyMapper</a>
extends org.apache.hadoop.mapreduce.Mapper<<a href="../../../../../org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.HBaseKeyWritable.html" title="class in org.apache.hadoop.hbase.test">IntegrationTestLoadCommonCrawl.HBaseKeyWritable</a>,org.apache.hadoop.io.BytesWritable,org.apache.hadoop.io.NullWritable,org.apache.hadoop.io.NullWritable></pre>
</li>
</ul>
@@ -242,7 +242,7 @@ extends org.apache.hadoop.mapreduce.Mapper<<a href="../../../../../org/apache
<ul class="blockList">
<li class="blockList">
<h4>conn</h4>
-<pre>org.apache.hadoop.hbase.client.Connection <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.Verify.VerifyMapper.html#line.745">conn</a></pre>
+<pre>org.apache.hadoop.hbase.client.Connection <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.Verify.VerifyMapper.html#line.752">conn</a></pre>
</li>
</ul>
<a name="table">
@@ -251,7 +251,7 @@ extends org.apache.hadoop.mapreduce.Mapper<<a href="../../../../../org/apache
<ul class="blockListLast">
<li class="blockList">
<h4>table</h4>
-<pre>org.apache.hadoop.hbase.client.Table <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.Verify.VerifyMapper.html#line.746">table</a></pre>
+<pre>org.apache.hadoop.hbase.client.Table <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.Verify.VerifyMapper.html#line.753">table</a></pre>
</li>
</ul>
</li>
@@ -268,7 +268,7 @@ extends org.apache.hadoop.mapreduce.Mapper<<a href="../../../../../org/apache
<ul class="blockListLast">
<li class="blockList">
<h4>VerifyMapper</h4>
-<pre>public <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.Verify.VerifyMapper.html#line.742">VerifyMapper</a>()</pre>
+<pre>public <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.Verify.VerifyMapper.html#line.749">VerifyMapper</a>()</pre>
</li>
</ul>
</li>
@@ -285,7 +285,7 @@ extends org.apache.hadoop.mapreduce.Mapper<<a href="../../../../../org/apache
<ul class="blockList">
<li class="blockList">
<h4>setup</h4>
-<pre>protected void <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.Verify.VerifyMapper.html#line.749">setup</a>(org.apache.hadoop.mapreduce.Mapper.Context context)
+<pre>protected void <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.Verify.VerifyMapper.html#line.756">setup</a>(org.apache.hadoop.mapreduce.Mapper.Context context)
throws <a href="https://docs.oracle.com/javase/8/docs/api/java/io/IOException.html?is-external=true" title="class or interface in java.io">IOException</a>,
<a href="https://docs.oracle.com/javase/8/docs/api/java/lang/InterruptedException.html?is-external=true" title="class or interface in java.lang">InterruptedException</a></pre>
<dl>
@@ -303,7 +303,7 @@ extends org.apache.hadoop.mapreduce.Mapper<<a href="../../../../../org/apache
<ul class="blockList">
<li class="blockList">
<h4>cleanup</h4>
-<pre>protected void <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.Verify.VerifyMapper.html#line.755">cleanup</a>(org.apache.hadoop.mapreduce.Mapper.Context context)
+<pre>protected void <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.Verify.VerifyMapper.html#line.762">cleanup</a>(org.apache.hadoop.mapreduce.Mapper.Context context)
throws <a href="https://docs.oracle.com/javase/8/docs/api/java/io/IOException.html?is-external=true" title="class or interface in java.io">IOException</a>,
<a href="https://docs.oracle.com/javase/8/docs/api/java/lang/InterruptedException.html?is-external=true" title="class or interface in java.lang">InterruptedException</a></pre>
<dl>
@@ -321,7 +321,7 @@ extends org.apache.hadoop.mapreduce.Mapper<<a href="../../../../../org/apache
<ul class="blockListLast">
<li class="blockList">
<h4>map</h4>
-<pre>protected void <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.Verify.VerifyMapper.html#line.761">map</a>(<a href="../../../../../org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.HBaseKeyWritable.html" title="class in org.apache.hadoop.hbase.test">IntegrationTestLoadCommonCrawl.HBaseKeyWritable</a> key,
+<pre>protected void <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.Verify.VerifyMapper.html#line.768">map</a>(<a href="../../../../../org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.HBaseKeyWritable.html" title="class in org.apache.hadoop.hbase.test">IntegrationTestLoadCommonCrawl.HBaseKeyWritable</a> key,
org.apache.hadoop.io.BytesWritable value,
org.apache.hadoop.mapreduce.Mapper.Context output)
throws <a href="https://docs.oracle.com/javase/8/docs/api/java/io/IOException.html?is-external=true" title="class or interface in java.io">IOException</a>,
diff --git a/testdevapidocs/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.Verify.html b/testdevapidocs/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.Verify.html
index 289ca13..a4b81d8 100644
--- a/testdevapidocs/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.Verify.html
+++ b/testdevapidocs/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.Verify.html
@@ -122,7 +122,7 @@ var activeTableTab = "activeTableTab";
</dl>
<hr>
<br>
-<pre>public static class <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.html#line.701">IntegrationTestLoadCommonCrawl.Verify</a>
+<pre>public static class <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.html#line.708">IntegrationTestLoadCommonCrawl.Verify</a>
extends org.apache.hadoop.conf.Configured
implements org.apache.hadoop.util.Tool</pre>
</li>
@@ -256,7 +256,7 @@ implements org.apache.hadoop.util.Tool</pre>
<ul class="blockList">
<li class="blockList">
<h4>LOG</h4>
-<pre>public static final org.slf4j.Logger <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.Verify.html#line.703">LOG</a></pre>
+<pre>public static final org.slf4j.Logger <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.Verify.html#line.710">LOG</a></pre>
</li>
</ul>
<a name="USAGE">
@@ -265,7 +265,7 @@ implements org.apache.hadoop.util.Tool</pre>
<ul class="blockListLast">
<li class="blockList">
<h4>USAGE</h4>
-<pre>public static final <a href="https://docs.oracle.com/javase/8/docs/api/java/lang/String.html?is-external=true" title="class or interface in java.lang">String</a> <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.Verify.html#line.704">USAGE</a></pre>
+<pre>public static final <a href="https://docs.oracle.com/javase/8/docs/api/java/lang/String.html?is-external=true" title="class or interface in java.lang">String</a> <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.Verify.html#line.711">USAGE</a></pre>
<dl>
<dt><span class="seeLabel">See Also:</span></dt>
<dd><a href="../../../../../constant-values.html#org.apache.hadoop.hbase.test.IntegrationTestLoadCommonCrawl.Verify.USAGE">Constant Field Values</a></dd>
@@ -286,7 +286,7 @@ implements org.apache.hadoop.util.Tool</pre>
<ul class="blockListLast">
<li class="blockList">
<h4>Verify</h4>
-<pre>public <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.Verify.html#line.701">Verify</a>()</pre>
+<pre>public <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.Verify.html#line.708">Verify</a>()</pre>
</li>
</ul>
</li>
@@ -303,7 +303,7 @@ implements org.apache.hadoop.util.Tool</pre>
<ul class="blockList">
<li class="blockList">
<h4>run</h4>
-<pre>int <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.Verify.html#line.706">run</a>(org.apache.hadoop.fs.Path inputDir)
+<pre>int <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.Verify.html#line.713">run</a>(org.apache.hadoop.fs.Path inputDir)
throws <a href="https://docs.oracle.com/javase/8/docs/api/java/io/IOException.html?is-external=true" title="class or interface in java.io">IOException</a>,
<a href="https://docs.oracle.com/javase/8/docs/api/java/lang/ClassNotFoundException.html?is-external=true" title="class or interface in java.lang">ClassNotFoundException</a>,
<a href="https://docs.oracle.com/javase/8/docs/api/java/lang/InterruptedException.html?is-external=true" title="class or interface in java.lang">InterruptedException</a></pre>
@@ -321,7 +321,7 @@ implements org.apache.hadoop.util.Tool</pre>
<ul class="blockList">
<li class="blockList">
<h4>run</h4>
-<pre>public int <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.Verify.html#line.729">run</a>(<a href="https://docs.oracle.com/javase/8/docs/api/java/lang/String.html?is-external=true" title="class or interface in java.lang">String</a>[] args)
+<pre>public int <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.Verify.html#line.736">run</a>(<a href="https://docs.oracle.com/javase/8/docs/api/java/lang/String.html?is-external=true" title="class or interface in java.lang">String</a>[] args)
throws <a href="https://docs.oracle.com/javase/8/docs/api/java/lang/Exception.html?is-external=true" title="class or interface in java.lang">Exception</a></pre>
<dl>
<dt><span class="overrideSpecifyLabel">Specified by:</span></dt>
@@ -337,7 +337,7 @@ implements org.apache.hadoop.util.Tool</pre>
<ul class="blockListLast">
<li class="blockList">
<h4>main</h4>
-<pre>public static void <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.Verify.html#line.738">main</a>(<a href="https://docs.oracle.com/javase/8/docs/api/java/lang/String.html?is-external=true" title="class or interface in java.lang">String</a>[] args)
+<pre>public static void <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.Verify.html#line.745">main</a>(<a href="https://docs.oracle.com/javase/8/docs/api/java/lang/String.html?is-external=true" title="class or interface in java.lang">String</a>[] args)
throws <a href="https://docs.oracle.com/javase/8/docs/api/java/lang/Exception.html?is-external=true" title="class or interface in java.lang">Exception</a></pre>
<dl>
<dt><span class="throwsLabel">Throws:</span></dt>
diff --git a/testdevapidocs/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.html b/testdevapidocs/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.html
index 3dc7824..6fadc7e 100644
--- a/testdevapidocs/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.html
+++ b/testdevapidocs/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.html
@@ -123,7 +123,7 @@ var activeTableTab = "activeTableTab";
</dl>
<hr>
<br>
-<pre>public class <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.html#line.141">IntegrationTestLoadCommonCrawl</a>
+<pre>public class <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.html#line.142">IntegrationTestLoadCommonCrawl</a>
extends <a href="../../../../../org/apache/hadoop/hbase/IntegrationTestBase.html" title="class in org.apache.hadoop.hbase">IntegrationTestBase</a></pre>
<div class="block">This integration test loads successful resource retrieval records from the Common Crawl
(https://commoncrawl.org/) public dataset into an HBase table and writes records that can be
@@ -296,6 +296,10 @@ extends <a href="../../../../../org/apache/hadoop/hbase/IntegrationTestBase.html
<td class="colLast"><code><span class="memberNameLink"><a href="../../../../../org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.html#TARGET_URI_QUALIFIER">TARGET_URI_QUALIFIER</a></span></code> </td>
</tr>
<tr class="rowColor">
+<td class="colFirst"><code>private static int</code></td>
+<td class="colLast"><code><span class="memberNameLink"><a href="../../../../../org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.html#VERIFICATION_READ_RETRIES">VERIFICATION_READ_RETRIES</a></span></code> </td>
+</tr>
+<tr class="altColor">
<td class="colFirst"><code>(package private) org.apache.hadoop.fs.Path</code></td>
<td class="colLast"><code><span class="memberNameLink"><a href="../../../../../org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.html#warcFileInputDir">warcFileInputDir</a></span></code> </td>
</tr>
@@ -436,7 +440,7 @@ extends <a href="../../../../../org/apache/hadoop/hbase/IntegrationTestBase.html
<ul class="blockList">
<li class="blockList">
<h4>LOG</h4>
-<pre>private static final org.slf4j.Logger <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.html#line.143">LOG</a></pre>
+<pre>private static final org.slf4j.Logger <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.html#line.144">LOG</a></pre>
</li>
</ul>
<a name="TABLE_NAME_KEY">
@@ -445,7 +449,7 @@ extends <a href="../../../../../org/apache/hadoop/hbase/IntegrationTestBase.html
<ul class="blockList">
<li class="blockList">
<h4>TABLE_NAME_KEY</h4>
-<pre>protected static <a href="https://docs.oracle.com/javase/8/docs/api/java/lang/String.html?is-external=true" title="class or interface in java.lang">String</a> <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.html#line.145">TABLE_NAME_KEY</a></pre>
+<pre>protected static <a href="https://docs.oracle.com/javase/8/docs/api/java/lang/String.html?is-external=true" title="class or interface in java.lang">String</a> <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.html#line.146">TABLE_NAME_KEY</a></pre>
</li>
</ul>
<a name="DEFAULT_TABLE_NAME">
@@ -454,7 +458,7 @@ extends <a href="../../../../../org/apache/hadoop/hbase/IntegrationTestBase.html
<ul class="blockList">
<li class="blockList">
<h4>DEFAULT_TABLE_NAME</h4>
-<pre>protected static <a href="https://docs.oracle.com/javase/8/docs/api/java/lang/String.html?is-external=true" title="class or interface in java.lang">String</a> <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.html#line.146">DEFAULT_TABLE_NAME</a></pre>
+<pre>protected static <a href="https://docs.oracle.com/javase/8/docs/api/java/lang/String.html?is-external=true" title="class or interface in java.lang">String</a> <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.html#line.147">DEFAULT_TABLE_NAME</a></pre>
</li>
</ul>
<a name="CONTENT_FAMILY_NAME">
@@ -463,7 +467,7 @@ extends <a href="../../../../../org/apache/hadoop/hbase/IntegrationTestBase.html
<ul class="blockList">
<li class="blockList">
<h4>CONTENT_FAMILY_NAME</h4>
-<pre>protected static byte[] <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.html#line.148">CONTENT_FAMILY_NAME</a></pre>
+<pre>protected static byte[] <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.html#line.149">CONTENT_FAMILY_NAME</a></pre>
</li>
</ul>
<a name="INFO_FAMILY_NAME">
@@ -472,7 +476,7 @@ extends <a href="../../../../../org/apache/hadoop/hbase/IntegrationTestBase.html
<ul class="blockList">
<li class="blockList">
<h4>INFO_FAMILY_NAME</h4>
-<pre>protected static byte[] <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.html#line.149">INFO_FAMILY_NAME</a></pre>
+<pre>protected static byte[] <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.html#line.150">INFO_FAMILY_NAME</a></pre>
</li>
</ul>
<a name="CONTENT_QUALIFIER">
@@ -481,7 +485,7 @@ extends <a href="../../../../../org/apache/hadoop/hbase/IntegrationTestBase.html
<ul class="blockList">
<li class="blockList">
<h4>CONTENT_QUALIFIER</h4>
-<pre>protected static byte[] <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.html#line.150">CONTENT_QUALIFIER</a></pre>
+<pre>protected static byte[] <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.html#line.151">CONTENT_QUALIFIER</a></pre>
</li>
</ul>
<a name="CONTENT_LENGTH_QUALIFIER">
@@ -490,7 +494,7 @@ extends <a href="../../../../../org/apache/hadoop/hbase/IntegrationTestBase.html
<ul class="blockList">
<li class="blockList">
<h4>CONTENT_LENGTH_QUALIFIER</h4>
-<pre>protected static byte[] <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.html#line.151">CONTENT_LENGTH_QUALIFIER</a></pre>
+<pre>protected static byte[] <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.html#line.152">CONTENT_LENGTH_QUALIFIER</a></pre>
</li>
</ul>
<a name="CONTENT_TYPE_QUALIFIER">
@@ -499,7 +503,7 @@ extends <a href="../../../../../org/apache/hadoop/hbase/IntegrationTestBase.html
<ul class="blockList">
<li class="blockList">
<h4>CONTENT_TYPE_QUALIFIER</h4>
-<pre>protected static byte[] <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.html#line.152">CONTENT_TYPE_QUALIFIER</a></pre>
+<pre>protected static byte[] <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.html#line.153">CONTENT_TYPE_QUALIFIER</a></pre>
</li>
</ul>
<a name="CRC_QUALIFIER">
@@ -508,7 +512,7 @@ extends <a href="../../../../../org/apache/hadoop/hbase/IntegrationTestBase.html
<ul class="blockList">
<li class="blockList">
<h4>CRC_QUALIFIER</h4>
-<pre>protected static byte[] <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.html#line.153">CRC_QUALIFIER</a></pre>
+<pre>protected static byte[] <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.html#line.154">CRC_QUALIFIER</a></pre>
</li>
</ul>
<a name="DATE_QUALIFIER">
@@ -517,7 +521,7 @@ extends <a href="../../../../../org/apache/hadoop/hbase/IntegrationTestBase.html
<ul class="blockList">
<li class="blockList">
<h4>DATE_QUALIFIER</h4>
-<pre>protected static byte[] <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.html#line.154">DATE_QUALIFIER</a></pre>
+<pre>protected static byte[] <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.html#line.155">DATE_QUALIFIER</a></pre>
</li>
</ul>
<a name="IP_ADDRESS_QUALIFIER">
@@ -526,7 +530,7 @@ extends <a href="../../../../../org/apache/hadoop/hbase/IntegrationTestBase.html
<ul class="blockList">
<li class="blockList">
<h4>IP_ADDRESS_QUALIFIER</h4>
-<pre>protected static byte[] <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.html#line.155">IP_ADDRESS_QUALIFIER</a></pre>
+<pre>protected static byte[] <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.html#line.156">IP_ADDRESS_QUALIFIER</a></pre>
</li>
</ul>
<a name="RECORD_ID_QUALIFIER">
@@ -535,7 +539,7 @@ extends <a href="../../../../../org/apache/hadoop/hbase/IntegrationTestBase.html
<ul class="blockList">
<li class="blockList">
<h4>RECORD_ID_QUALIFIER</h4>
-<pre>protected static byte[] <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.html#line.156">RECORD_ID_QUALIFIER</a></pre>
+<pre>protected static byte[] <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.html#line.157">RECORD_ID_QUALIFIER</a></pre>
</li>
</ul>
<a name="TARGET_URI_QUALIFIER">
@@ -544,7 +548,20 @@ extends <a href="../../../../../org/apache/hadoop/hbase/IntegrationTestBase.html
<ul class="blockList">
<li class="blockList">
<h4>TARGET_URI_QUALIFIER</h4>
-<pre>protected static byte[] <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.html#line.157">TARGET_URI_QUALIFIER</a></pre>
+<pre>protected static byte[] <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.html#line.158">TARGET_URI_QUALIFIER</a></pre>
+</li>
+</ul>
+<a name="VERIFICATION_READ_RETRIES">
+<!-- -->
+</a>
+<ul class="blockList">
+<li class="blockList">
+<h4>VERIFICATION_READ_RETRIES</h4>
+<pre>private static final int <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.html#line.160">VERIFICATION_READ_RETRIES</a></pre>
+<dl>
+<dt><span class="seeLabel">See Also:</span></dt>
+<dd><a href="../../../../../constant-values.html#org.apache.hadoop.hbase.test.IntegrationTestLoadCommonCrawl.VERIFICATION_READ_RETRIES">Constant Field Values</a></dd>
+</dl>
</li>
</ul>
<a name="warcFileInputDir">
@@ -553,7 +570,7 @@ extends <a href="../../../../../org/apache/hadoop/hbase/IntegrationTestBase.html
<ul class="blockList">
<li class="blockList">
<h4>warcFileInputDir</h4>
-<pre>org.apache.hadoop.fs.Path <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.html#line.163">warcFileInputDir</a></pre>
+<pre>org.apache.hadoop.fs.Path <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.html#line.166">warcFileInputDir</a></pre>
</li>
</ul>
<a name="outputDir">
@@ -562,7 +579,7 @@ extends <a href="../../../../../org/apache/hadoop/hbase/IntegrationTestBase.html
<ul class="blockList">
<li class="blockList">
<h4>outputDir</h4>
-<pre>org.apache.hadoop.fs.Path <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.html#line.164">outputDir</a></pre>
+<pre>org.apache.hadoop.fs.Path <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.html#line.167">outputDir</a></pre>
</li>
</ul>
<a name="args">
@@ -571,7 +588,7 @@ extends <a href="../../../../../org/apache/hadoop/hbase/IntegrationTestBase.html
<ul class="blockListLast">
<li class="blockList">
<h4>args</h4>
-<pre><a href="https://docs.oracle.com/javase/8/docs/api/java/lang/String.html?is-external=true" title="class or interface in java.lang">String</a>[] <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.html#line.165">args</a></pre>
+<pre><a href="https://docs.oracle.com/javase/8/docs/api/java/lang/String.html?is-external=true" title="class or interface in java.lang">String</a>[] <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.html#line.168">args</a></pre>
</li>
</ul>
</li>
@@ -588,7 +605,7 @@ extends <a href="../../../../../org/apache/hadoop/hbase/IntegrationTestBase.html
<ul class="blockListLast">
<li class="blockList">
<h4>IntegrationTestLoadCommonCrawl</h4>
-<pre>public <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.html#line.141">IntegrationTestLoadCommonCrawl</a>()</pre>
+<pre>public <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.html#line.142">IntegrationTestLoadCommonCrawl</a>()</pre>
</li>
</ul>
</li>
@@ -605,7 +622,7 @@ extends <a href="../../../../../org/apache/hadoop/hbase/IntegrationTestBase.html
<ul class="blockList">
<li class="blockList">
<h4>runLoader</h4>
-<pre>protected int <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.html#line.167">runLoader</a>(org.apache.hadoop.fs.Path warcFileInputDir,
+<pre>protected int <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.html#line.170">runLoader</a>(org.apache.hadoop.fs.Path warcFileInputDir,
org.apache.hadoop.fs.Path outputDir)
throws <a href="https://docs.oracle.com/javase/8/docs/api/java/lang/Exception.html?is-external=true" title="class or interface in java.lang">Exception</a></pre>
<dl>
@@ -620,7 +637,7 @@ extends <a href="../../../../../org/apache/hadoop/hbase/IntegrationTestBase.html
<ul class="blockList">
<li class="blockList">
<h4>runVerify</h4>
-<pre>protected int <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.html#line.173">runVerify</a>(org.apache.hadoop.fs.Path inputDir)
+<pre>protected int <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.html#line.176">runVerify</a>(org.apache.hadoop.fs.Path inputDir)
throws <a href="https://docs.oracle.com/javase/8/docs/api/java/lang/Exception.html?is-external=true" title="class or interface in java.lang">Exception</a></pre>
<dl>
<dt><span class="throwsLabel">Throws:</span></dt>
@@ -634,7 +651,7 @@ extends <a href="../../../../../org/apache/hadoop/hbase/IntegrationTestBase.html
<ul class="blockList">
<li class="blockList">
<h4>run</h4>
-<pre>public int <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.html#line.180">run</a>(<a href="https://docs.oracle.com/javase/8/docs/api/java/lang/String.html?is-external=true" title="class or interface in java.lang">String</a>[] args)</pre>
+<pre>public int <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.html#line.183">run</a>(<a href="https://docs.oracle.com/javase/8/docs/api/java/lang/String.html?is-external=true" title="class or interface in java.lang">String</a>[] args)</pre>
<dl>
<dt><span class="overrideSpecifyLabel">Specified by:</span></dt>
<dd><code>run</code> in interface <code>org.apache.hadoop.util.Tool</code></dd>
@@ -649,7 +666,7 @@ extends <a href="../../../../../org/apache/hadoop/hbase/IntegrationTestBase.html
<ul class="blockList">
<li class="blockList">
<h4>processOptions</h4>
-<pre>protected void <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.html#line.208">processOptions</a>(org.apache.hbase.thirdparty.org.apache.commons.cli.CommandLine cmd)</pre>
+<pre>protected void <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.html#line.211">processOptions</a>(org.apache.hbase.thirdparty.org.apache.commons.cli.CommandLine cmd)</pre>
<dl>
<dt><span class="overrideSpecifyLabel">Overrides:</span></dt>
<dd><code><a href="../../../../../org/apache/hadoop/hbase/IntegrationTestBase.html#processOptions-org.apache.hbase.thirdparty.org.apache.commons.cli.CommandLine-">processOptions</a></code> in class <code><a href="../../../../../org/apache/hadoop/hbase/IntegrationTestBase.html" title="class in org.apache.hadoop.hbase">IntegrationTestBase</a></code></dd>
@@ -662,7 +679,7 @@ extends <a href="../../../../../org/apache/hadoop/hbase/IntegrationTestBase.html
<ul class="blockList">
<li class="blockList">
<h4>setUpCluster</h4>
-<pre>public void <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.html#line.214">setUpCluster</a>()
+<pre>public void <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.html#line.217">setUpCluster</a>()
throws <a href="https://docs.oracle.com/javase/8/docs/api/java/lang/Exception.html?is-external=true" title="class or interface in java.lang">Exception</a></pre>
<dl>
<dt><span class="overrideSpecifyLabel">Specified by:</span></dt>
@@ -678,7 +695,7 @@ extends <a href="../../../../../org/apache/hadoop/hbase/IntegrationTestBase.html
<ul class="blockList">
<li class="blockList">
<h4>cleanUpCluster</h4>
-<pre>public void <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.html#line.225">cleanUpCluster</a>()
+<pre>public void <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.html#line.228">cleanUpCluster</a>()
throws <a href="https://docs.oracle.com/javase/8/docs/api/java/lang/Exception.html?is-external=true" title="class or interface in java.lang">Exception</a></pre>
<dl>
<dt><span class="overrideSpecifyLabel">Overrides:</span></dt>
@@ -694,7 +711,7 @@ extends <a href="../../../../../org/apache/hadoop/hbase/IntegrationTestBase.html
<ul class="blockList">
<li class="blockList">
<h4>getTablename</h4>
-<pre>static org.apache.hadoop.hbase.TableName <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.html#line.232">getTablename</a>(org.apache.hadoop.conf.Configuration c)</pre>
+<pre>static org.apache.hadoop.hbase.TableName <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.html#line.235">getTablename</a>(org.apache.hadoop.conf.Configuration c)</pre>
</li>
</ul>
<a name="getTablename--">
@@ -703,7 +720,7 @@ extends <a href="../../../../../org/apache/hadoop/hbase/IntegrationTestBase.html
<ul class="blockList">
<li class="blockList">
<h4>getTablename</h4>
-<pre>public org.apache.hadoop.hbase.TableName <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.html#line.237">getTablename</a>()</pre>
+<pre>public org.apache.hadoop.hbase.TableName <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.html#line.240">getTablename</a>()</pre>
<div class="block"><span class="descfrmTypeLabel">Description copied from class: <code><a href="../../../../../org/apache/hadoop/hbase/IntegrationTestBase.html#getTablename--">IntegrationTestBase</a></code></span></div>
<div class="block">Provides the name of the table that is protected from random Chaos monkey activity</div>
<dl>
@@ -720,7 +737,7 @@ extends <a href="../../../../../org/apache/hadoop/hbase/IntegrationTestBase.html
<ul class="blockList">
<li class="blockList">
<h4>getColumnFamilies</h4>
-<pre>protected <a href="https://docs.oracle.com/javase/8/docs/api/java/util/Set.html?is-external=true" title="class or interface in java.util">Set</a><<a href="https://docs.oracle.com/javase/8/docs/api/java/lang/String.html?is-external=true" title="class or interface in java.lang">String</a>> <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.html#line.242">getColumnFamilies</a>()</pre>
+<pre>protected <a href="https://docs.oracle.com/javase/8/docs/api/java/util/Set.html?is-external=true" title="class or interface in java.util">Set</a><<a href="https://docs.oracle.com/javase/8/docs/api/java/lang/String.html?is-external=true" title="class or interface in java.lang">String</a>> <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.html#line.245">getColumnFamilies</a>()</pre>
<div class="block"><span class="descfrmTypeLabel">Description copied from class: <code><a href="../../../../../org/apache/hadoop/hbase/IntegrationTestBase.html#getColumnFamilies--">IntegrationTestBase</a></code></span></div>
<div class="block">Provides the name of the CFs that are protected from random Chaos monkey activity (alter)</div>
<dl>
@@ -737,7 +754,7 @@ extends <a href="../../../../../org/apache/hadoop/hbase/IntegrationTestBase.html
<ul class="blockList">
<li class="blockList">
<h4>runTestFromCommandLine</h4>
-<pre>public int <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.html#line.250">runTestFromCommandLine</a>()
+<pre>public int <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.html#line.253">runTestFromCommandLine</a>()
throws <a href="https://docs.oracle.com/javase/8/docs/api/java/lang/Exception.html?is-external=true" title="class or interface in java.lang">Exception</a></pre>
<dl>
<dt><span class="overrideSpecifyLabel">Specified by:</span></dt>
@@ -753,7 +770,7 @@ extends <a href="../../../../../org/apache/hadoop/hbase/IntegrationTestBase.html
<ul class="blockListLast">
<li class="blockList">
<h4>main</h4>
-<pre>public static void <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.html#line.254">main</a>(<a href="https://docs.oracle.com/javase/8/docs/api/java/lang/String.html?is-external=true" title="class or interface in java.lang">String</a>[] args)
+<pre>public static void <a href="../../../../../src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.html#line.257">main</a>(<a href="https://docs.oracle.com/javase/8/docs/api/java/lang/String.html?is-external=true" title="class or interface in java.lang">String</a>[] args)
throws <a href="https://docs.oracle.com/javase/8/docs/api/java/lang/Exception.html?is-external=true" title="class or interface in java.lang">Exception</a></pre>
<dl>
<dt><span class="throwsLabel">Throws:</span></dt>
diff --git a/testdevapidocs/org/apache/hadoop/hbase/test/package-tree.html b/testdevapidocs/org/apache/hadoop/hbase/test/package-tree.html
index de7cd45..5d8f4b5 100644
--- a/testdevapidocs/org/apache/hadoop/hbase/test/package-tree.html
+++ b/testdevapidocs/org/apache/hadoop/hbase/test/package-tree.html
@@ -260,11 +260,11 @@
<ul>
<li type="circle">java.lang.<a href="https://docs.oracle.com/javase/8/docs/api/java/lang/Enum.html?is-external=true" title="class or interface in java.lang"><span class="typeNameLink">Enum</span></a><E> (implements java.lang.<a href="https://docs.oracle.com/javase/8/docs/api/java/lang/Comparable.html?is-external=true" title="class or interface in java.lang">Comparable</a><T>, java.io.<a href="https://docs.oracle.com/javase/8/docs/api/java/io/Serializable.html?is-external=true [...]
<ul>
+<li type="circle">org.apache.hadoop.hbase.test.<a href="../../../../../org/apache/hadoop/hbase/test/IntegrationTestBigLinkedList.Verify.Counts.html" title="enum in org.apache.hadoop.hbase.test"><span class="typeNameLink">IntegrationTestBigLinkedList.Verify.Counts</span></a></li>
+<li type="circle">org.apache.hadoop.hbase.test.<a href="../../../../../org/apache/hadoop/hbase/test/IntegrationTestLoadAndVerify.Counters.html" title="enum in org.apache.hadoop.hbase.test"><span class="typeNameLink">IntegrationTestLoadAndVerify.Counters</span></a></li>
+<li type="circle">org.apache.hadoop.hbase.test.<a href="../../../../../org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.Counts.html" title="enum in org.apache.hadoop.hbase.test"><span class="typeNameLink">IntegrationTestLoadCommonCrawl.Counts</span></a></li>
<li type="circle">org.apache.hadoop.hbase.test.<a href="../../../../../org/apache/hadoop/hbase/test/IntegrationTestWithCellVisibilityLoadAndVerify.Counters.html" title="enum in org.apache.hadoop.hbase.test"><span class="typeNameLink">IntegrationTestWithCellVisibilityLoadAndVerify.Counters</span></a></li>
<li type="circle">org.apache.hadoop.hbase.test.<a href="../../../../../org/apache/hadoop/hbase/test/IntegrationTestBigLinkedList.Generator.Counts.html" title="enum in org.apache.hadoop.hbase.test"><span class="typeNameLink">IntegrationTestBigLinkedList.Generator.Counts</span></a></li>
-<li type="circle">org.apache.hadoop.hbase.test.<a href="../../../../../org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.Counts.html" title="enum in org.apache.hadoop.hbase.test"><span class="typeNameLink">IntegrationTestLoadCommonCrawl.Counts</span></a></li>
-<li type="circle">org.apache.hadoop.hbase.test.<a href="../../../../../org/apache/hadoop/hbase/test/IntegrationTestLoadAndVerify.Counters.html" title="enum in org.apache.hadoop.hbase.test"><span class="typeNameLink">IntegrationTestLoadAndVerify.Counters</span></a></li>
-<li type="circle">org.apache.hadoop.hbase.test.<a href="../../../../../org/apache/hadoop/hbase/test/IntegrationTestBigLinkedList.Verify.Counts.html" title="enum in org.apache.hadoop.hbase.test"><span class="typeNameLink">IntegrationTestBigLinkedList.Verify.Counts</span></a></li>
</ul>
</li>
</ul>
diff --git a/testdevapidocs/org/apache/hadoop/hbase/wal/package-tree.html b/testdevapidocs/org/apache/hadoop/hbase/wal/package-tree.html
index 14e3756..0f66641 100644
--- a/testdevapidocs/org/apache/hadoop/hbase/wal/package-tree.html
+++ b/testdevapidocs/org/apache/hadoop/hbase/wal/package-tree.html
@@ -206,9 +206,9 @@
<ul>
<li type="circle">java.lang.<a href="https://docs.oracle.com/javase/8/docs/api/java/lang/Enum.html?is-external=true" title="class or interface in java.lang"><span class="typeNameLink">Enum</span></a><E> (implements java.lang.<a href="https://docs.oracle.com/javase/8/docs/api/java/lang/Comparable.html?is-external=true" title="class or interface in java.lang">Comparable</a><T>, java.io.<a href="https://docs.oracle.com/javase/8/docs/api/java/io/Serializable.html?is-external=true [...]
<ul>
-<li type="circle">org.apache.hadoop.hbase.wal.<a href="../../../../../org/apache/hadoop/hbase/wal/FaultyFSLog.FailureType.html" title="enum in org.apache.hadoop.hbase.wal"><span class="typeNameLink">FaultyFSLog.FailureType</span></a></li>
-<li type="circle">org.apache.hadoop.hbase.wal.<a href="../../../../../org/apache/hadoop/hbase/wal/IOTestProvider.AllowedOperations.html" title="enum in org.apache.hadoop.hbase.wal"><span class="typeNameLink">IOTestProvider.AllowedOperations</span></a></li>
<li type="circle">org.apache.hadoop.hbase.wal.<a href="../../../../../org/apache/hadoop/hbase/wal/TestWALSplit.Corruptions.html" title="enum in org.apache.hadoop.hbase.wal"><span class="typeNameLink">TestWALSplit.Corruptions</span></a></li>
+<li type="circle">org.apache.hadoop.hbase.wal.<a href="../../../../../org/apache/hadoop/hbase/wal/IOTestProvider.AllowedOperations.html" title="enum in org.apache.hadoop.hbase.wal"><span class="typeNameLink">IOTestProvider.AllowedOperations</span></a></li>
+<li type="circle">org.apache.hadoop.hbase.wal.<a href="../../../../../org/apache/hadoop/hbase/wal/FaultyFSLog.FailureType.html" title="enum in org.apache.hadoop.hbase.wal"><span class="typeNameLink">FaultyFSLog.FailureType</span></a></li>
</ul>
</li>
</ul>
diff --git a/testdevapidocs/src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.Counts.html b/testdevapidocs/src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.Counts.html
index 4a40cbc..e89b63c 100644
--- a/testdevapidocs/src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.Counts.html
+++ b/testdevapidocs/src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.Counts.html
@@ -54,797 +54,823 @@
<span class="sourceLineNo">046</span>import org.apache.hadoop.hbase.MasterNotRunningException;<a name="line.46"></a>
<span class="sourceLineNo">047</span>import org.apache.hadoop.hbase.TableName;<a name="line.47"></a>
<span class="sourceLineNo">048</span>import org.apache.hadoop.hbase.client.Admin;<a name="line.48"></a>
-<span class="sourceLineNo">049</span>import org.apache.hadoop.hbase.client.ColumnFamilyDescriptor;<a name="line.49"></a>
-<span class="sourceLineNo">050</span>import org.apache.hadoop.hbase.client.ColumnFamilyDescriptorBuilder;<a name="line.50"></a>
-<span class="sourceLineNo">051</span>import org.apache.hadoop.hbase.client.Connection;<a name="line.51"></a>
-<span class="sourceLineNo">052</span>import org.apache.hadoop.hbase.client.ConnectionFactory;<a name="line.52"></a>
-<span class="sourceLineNo">053</span>import org.apache.hadoop.hbase.client.Get;<a name="line.53"></a>
-<span class="sourceLineNo">054</span>import org.apache.hadoop.hbase.client.Put;<a name="line.54"></a>
-<span class="sourceLineNo">055</span>import org.apache.hadoop.hbase.client.Result;<a name="line.55"></a>
-<span class="sourceLineNo">056</span>import org.apache.hadoop.hbase.client.Table;<a name="line.56"></a>
-<span class="sourceLineNo">057</span>import org.apache.hadoop.hbase.client.TableDescriptor;<a name="line.57"></a>
-<span class="sourceLineNo">058</span>import org.apache.hadoop.hbase.client.TableDescriptorBuilder;<a name="line.58"></a>
-<span class="sourceLineNo">059</span>import org.apache.hadoop.hbase.io.encoding.DataBlockEncoding;<a name="line.59"></a>
-<span class="sourceLineNo">060</span>import org.apache.hadoop.hbase.mapreduce.TableMapReduceUtil;<a name="line.60"></a>
-<span class="sourceLineNo">061</span>import org.apache.hadoop.hbase.regionserver.BloomType;<a name="line.61"></a>
-<span class="sourceLineNo">062</span>import org.apache.hadoop.hbase.test.util.CRC64;<a name="line.62"></a>
-<span class="sourceLineNo">063</span>import org.apache.hadoop.hbase.test.util.warc.WARCInputFormat;<a name="line.63"></a>
-<span class="sourceLineNo">064</span>import org.apache.hadoop.hbase.test.util.warc.WARCRecord;<a name="line.64"></a>
-<span class="sourceLineNo">065</span>import org.apache.hadoop.hbase.test.util.warc.WARCWritable;<a name="line.65"></a>
-<span class="sourceLineNo">066</span>import org.apache.hadoop.hbase.util.Bytes;<a name="line.66"></a>
-<span class="sourceLineNo">067</span>import org.apache.hadoop.hbase.util.EnvironmentEdgeManager;<a name="line.67"></a>
-<span class="sourceLineNo">068</span>import org.apache.hadoop.hbase.util.RegionSplitter;<a name="line.68"></a>
-<span class="sourceLineNo">069</span>import org.apache.hadoop.io.BytesWritable;<a name="line.69"></a>
-<span class="sourceLineNo">070</span>import org.apache.hadoop.io.LongWritable;<a name="line.70"></a>
-<span class="sourceLineNo">071</span>import org.apache.hadoop.io.NullWritable;<a name="line.71"></a>
-<span class="sourceLineNo">072</span>import org.apache.hadoop.io.SequenceFile.CompressionType;<a name="line.72"></a>
-<span class="sourceLineNo">073</span>import org.apache.hadoop.io.Writable;<a name="line.73"></a>
-<span class="sourceLineNo">074</span>import org.apache.hadoop.mapreduce.Counters;<a name="line.74"></a>
-<span class="sourceLineNo">075</span>import org.apache.hadoop.mapreduce.Job;<a name="line.75"></a>
-<span class="sourceLineNo">076</span>import org.apache.hadoop.mapreduce.JobContext;<a name="line.76"></a>
-<span class="sourceLineNo">077</span>import org.apache.hadoop.mapreduce.Mapper;<a name="line.77"></a>
-<span class="sourceLineNo">078</span>import org.apache.hadoop.mapreduce.lib.input.FileInputFormat;<a name="line.78"></a>
-<span class="sourceLineNo">079</span>import org.apache.hadoop.mapreduce.lib.input.SequenceFileInputFormat;<a name="line.79"></a>
-<span class="sourceLineNo">080</span>import org.apache.hadoop.mapreduce.lib.output.NullOutputFormat;<a name="line.80"></a>
-<span class="sourceLineNo">081</span>import org.apache.hadoop.mapreduce.lib.output.SequenceFileOutputFormat;<a name="line.81"></a>
-<span class="sourceLineNo">082</span>import org.apache.hadoop.util.Tool;<a name="line.82"></a>
-<span class="sourceLineNo">083</span>import org.apache.hadoop.util.ToolRunner;<a name="line.83"></a>
-<span class="sourceLineNo">084</span>import org.slf4j.Logger;<a name="line.84"></a>
-<span class="sourceLineNo">085</span>import org.slf4j.LoggerFactory;<a name="line.85"></a>
-<span class="sourceLineNo">086</span><a name="line.86"></a>
-<span class="sourceLineNo">087</span>import org.apache.hbase.thirdparty.org.apache.commons.cli.CommandLine;<a name="line.87"></a>
-<span class="sourceLineNo">088</span><a name="line.88"></a>
-<span class="sourceLineNo">089</span>/**<a name="line.89"></a>
-<span class="sourceLineNo">090</span> * This integration test loads successful resource retrieval records from the Common Crawl<a name="line.90"></a>
-<span class="sourceLineNo">091</span> * (https://commoncrawl.org/) public dataset into an HBase table and writes records that can be<a name="line.91"></a>
-<span class="sourceLineNo">092</span> * used to later verify the presence and integrity of those records.<a name="line.92"></a>
-<span class="sourceLineNo">093</span> * <p><a name="line.93"></a>
-<span class="sourceLineNo">094</span> * Run like:<a name="line.94"></a>
-<span class="sourceLineNo">095</span> * <blockquote><a name="line.95"></a>
-<span class="sourceLineNo">096</span> * ./bin/hbase org.apache.hadoop.hbase.test.IntegrationTestLoadCommonCrawl \<br><a name="line.96"></a>
-<span class="sourceLineNo">097</span> * &nbsp;&nbsp; -Dfs.s3n.awsAccessKeyId=&lt;AWS access key&gt; \<br><a name="line.97"></a>
-<span class="sourceLineNo">098</span> * &nbsp;&nbsp; -Dfs.s3n.awsSecretAccessKey=&lt;AWS secret key&gt; \<br><a name="line.98"></a>
-<span class="sourceLineNo">099</span> * &nbsp;&nbsp; /path/to/test-CC-MAIN-2021-10-warc.paths.gz \<br><a name="line.99"></a>
-<span class="sourceLineNo">100</span> * &nbsp;&nbsp; /path/to/tmp/warc-loader-output<a name="line.100"></a>
-<span class="sourceLineNo">101</span> * </blockquote><a name="line.101"></a>
-<span class="sourceLineNo">102</span> * <p><a name="line.102"></a>
-<span class="sourceLineNo">103</span> * Access to the Common Crawl dataset in S3 is made available to anyone by Amazon AWS, but<a name="line.103"></a>
-<span class="sourceLineNo">104</span> * Hadoop's S3N filesystem still requires valid access credentials to initialize.<a name="line.104"></a>
-<span class="sourceLineNo">105</span> * <p><a name="line.105"></a>
-<span class="sourceLineNo">106</span> * The input path can either specify a directory or a file. The file may optionally be<a name="line.106"></a>
-<span class="sourceLineNo">107</span> * compressed with gzip. If a directory, the loader expects the directory to contain one or more<a name="line.107"></a>
-<span class="sourceLineNo">108</span> * WARC files from the Common Crawl dataset. If a file, the loader expects a list of Hadoop S3N<a name="line.108"></a>
-<span class="sourceLineNo">109</span> * URIs which point to S3 locations for one or more WARC files from the Common Crawl dataset,<a name="line.109"></a>
-<span class="sourceLineNo">110</span> * one URI per line. Lines should be terminated with the UNIX line terminator.<a name="line.110"></a>
-<span class="sourceLineNo">111</span> * <p><a name="line.111"></a>
-<span class="sourceLineNo">112</span> * Included in hbase-it/src/test/resources/CC-MAIN-2021-10-warc.paths.gz is a list of all WARC<a name="line.112"></a>
-<span class="sourceLineNo">113</span> * files comprising the Q1 2021 crawl archive. There are 64,000 WARC files in this data set, each<a name="line.113"></a>
-<span class="sourceLineNo">114</span> * containing ~1GB of gzipped data. The WARC files contain several record types, such as metadata,<a name="line.114"></a>
-<span class="sourceLineNo">115</span> * request, and response, but we only load the response record types. If the HBase table schema<a name="line.115"></a>
-<span class="sourceLineNo">116</span> * does not specify compression (by default) there is roughly a 10x expansion. Loading the full<a name="line.116"></a>
-<span class="sourceLineNo">117</span> * crawl archive results in a table approximately 640 TB in size.<a name="line.117"></a>
-<span class="sourceLineNo">118</span> * <p><a name="line.118"></a>
-<span class="sourceLineNo">119</span> * You can also split the Loader and Verify stages:<a name="line.119"></a>
-<span class="sourceLineNo">120</span> * <p><a name="line.120"></a>
-<span class="sourceLineNo">121</span> * Load with:<a name="line.121"></a>
-<span class="sourceLineNo">122</span> * <blockquote><a name="line.122"></a>
-<span class="sourceLineNo">123</span> * ./bin/hbase 'org.apache.hadoop.hbase.test.IntegrationTestLoadCommonCrawl$Loader' \<br><a name="line.123"></a>
-<span class="sourceLineNo">124</span> * &nbsp;&nbsp; -files /path/to/hadoop-aws.jar \<br><a name="line.124"></a>
-<span class="sourceLineNo">125</span> * &nbsp;&nbsp; -Dfs.s3n.awsAccessKeyId=&lt;AWS access key&gt; \<br><a name="line.125"></a>
-<span class="sourceLineNo">126</span> * &nbsp;&nbsp; -Dfs.s3n.awsSecretAccessKey=&lt;AWS secret key&gt; \<br><a name="line.126"></a>
-<span class="sourceLineNo">127</span> * &nbsp;&nbsp; /path/to/test-CC-MAIN-2021-10-warc.paths.gz \<br><a name="line.127"></a>
-<span class="sourceLineNo">128</span> * &nbsp;&nbsp; /path/to/tmp/warc-loader-output<a name="line.128"></a>
-<span class="sourceLineNo">129</span> * </blockquote><a name="line.129"></a>
-<span class="sourceLineNo">130</span> * <p><a name="line.130"></a>
-<span class="sourceLineNo">131</span> * Note: The hadoop-aws jar will be needed at runtime to instantiate the S3N filesystem. Use<a name="line.131"></a>
-<span class="sourceLineNo">132</span> * the <tt>-files</tt> ToolRunner argument to add it.<a name="line.132"></a>
-<span class="sourceLineNo">133</span> * <p><a name="line.133"></a>
-<span class="sourceLineNo">134</span> * Verify with:<a name="line.134"></a>
-<span class="sourceLineNo">135</span> * <blockquote><a name="line.135"></a>
-<span class="sourceLineNo">136</span> * ./bin/hbase 'org.apache.hadoop.hbase.test.IntegrationTestLoadCommonCrawl$Verify' \<br><a name="line.136"></a>
-<span class="sourceLineNo">137</span> * &nbsp;&nbsp; /path/to/tmp/warc-loader-output<a name="line.137"></a>
-<span class="sourceLineNo">138</span> * </blockquote><a name="line.138"></a>
-<span class="sourceLineNo">139</span> * <p><a name="line.139"></a>
-<span class="sourceLineNo">140</span> */<a name="line.140"></a>
-<span class="sourceLineNo">141</span>public class IntegrationTestLoadCommonCrawl extends IntegrationTestBase {<a name="line.141"></a>
-<span class="sourceLineNo">142</span><a name="line.142"></a>
-<span class="sourceLineNo">143</span> private static final Logger LOG = LoggerFactory.getLogger(IntegrationTestLoadCommonCrawl.class);<a name="line.143"></a>
-<span class="sourceLineNo">144</span><a name="line.144"></a>
-<span class="sourceLineNo">145</span> protected static String TABLE_NAME_KEY = "IntegrationTestLoadCommonCrawl.table";<a name="line.145"></a>
-<span class="sourceLineNo">146</span> protected static String DEFAULT_TABLE_NAME = "IntegrationTestLoadCommonCrawl";<a name="line.146"></a>
-<span class="sourceLineNo">147</span><a name="line.147"></a>
-<span class="sourceLineNo">148</span> protected static byte[] CONTENT_FAMILY_NAME = Bytes.toBytes("c");<a name="line.148"></a>
-<span class="sourceLineNo">149</span> protected static byte[] INFO_FAMILY_NAME = Bytes.toBytes("i");<a name="line.149"></a>
-<span class="sourceLineNo">150</span> protected static byte[] CONTENT_QUALIFIER = HConstants.EMPTY_BYTE_ARRAY;<a name="line.150"></a>
-<span class="sourceLineNo">151</span> protected static byte[] CONTENT_LENGTH_QUALIFIER = Bytes.toBytes("l");<a name="line.151"></a>
-<span class="sourceLineNo">152</span> protected static byte[] CONTENT_TYPE_QUALIFIER = Bytes.toBytes("t");<a name="line.152"></a>
-<span class="sourceLineNo">153</span> protected static byte[] CRC_QUALIFIER = Bytes.toBytes("c");<a name="line.153"></a>
-<span class="sourceLineNo">154</span> protected static byte[] DATE_QUALIFIER = Bytes.toBytes("d");<a name="line.154"></a>
-<span class="sourceLineNo">155</span> protected static byte[] IP_ADDRESS_QUALIFIER = Bytes.toBytes("a");<a name="line.155"></a>
-<span class="sourceLineNo">156</span> protected static byte[] RECORD_ID_QUALIFIER = Bytes.toBytes("r");<a name="line.156"></a>
-<span class="sourceLineNo">157</span> protected static byte[] TARGET_URI_QUALIFIER = Bytes.toBytes("u");<a name="line.157"></a>
-<span class="sourceLineNo">158</span><a name="line.158"></a>
-<span class="sourceLineNo">159</span> public static enum Counts {<a name="line.159"></a>
-<span class="sourceLineNo">160</span> REFERENCED, UNREFERENCED, CORRUPT<a name="line.160"></a>
-<span class="sourceLineNo">161</span> }<a name="line.161"></a>
-<span class="sourceLineNo">162</span><a name="line.162"></a>
-<span class="sourceLineNo">163</span> Path warcFileInputDir = null;<a name="line.163"></a>
-<span class="sourceLineNo">164</span> Path outputDir = null;<a name="line.164"></a>
-<span class="sourceLineNo">165</span> String[] args;<a name="line.165"></a>
-<span class="sourceLineNo">166</span><a name="line.166"></a>
-<span class="sourceLineNo">167</span> protected int runLoader(Path warcFileInputDir, Path outputDir) throws Exception {<a name="line.167"></a>
-<span class="sourceLineNo">168</span> Loader loader = new Loader();<a name="line.168"></a>
-<span class="sourceLineNo">169</span> loader.setConf(conf);<a name="line.169"></a>
-<span class="sourceLineNo">170</span> return loader.run(warcFileInputDir, outputDir);<a name="line.170"></a>
-<span class="sourceLineNo">171</span> }<a name="line.171"></a>
-<span class="sourceLineNo">172</span><a name="line.172"></a>
-<span class="sourceLineNo">173</span> protected int runVerify(Path inputDir) throws Exception {<a name="line.173"></a>
-<span class="sourceLineNo">174</span> Verify verify = new Verify();<a name="line.174"></a>
-<span class="sourceLineNo">175</span> verify.setConf(conf);<a name="line.175"></a>
-<span class="sourceLineNo">176</span> return verify.run(inputDir);<a name="line.176"></a>
-<span class="sourceLineNo">177</span> }<a name="line.177"></a>
-<span class="sourceLineNo">178</span><a name="line.178"></a>
-<span class="sourceLineNo">179</span> @Override<a name="line.179"></a>
-<span class="sourceLineNo">180</span> public int run(String[] args) {<a name="line.180"></a>
-<span class="sourceLineNo">181</span> if (args.length > 0) {<a name="line.181"></a>
-<span class="sourceLineNo">182</span> warcFileInputDir = new Path(args[0]);<a name="line.182"></a>
-<span class="sourceLineNo">183</span> if (args.length > 1) {<a name="line.183"></a>
-<span class="sourceLineNo">184</span> outputDir = new Path(args[1]);<a name="line.184"></a>
-<span class="sourceLineNo">185</span> }<a name="line.185"></a>
-<span class="sourceLineNo">186</span> }<a name="line.186"></a>
-<span class="sourceLineNo">187</span> try {<a name="line.187"></a>
-<span class="sourceLineNo">188</span> if (warcFileInputDir == null) {<a name="line.188"></a>
-<span class="sourceLineNo">189</span> throw new IllegalArgumentException("WARC input file or directory not specified");<a name="line.189"></a>
-<span class="sourceLineNo">190</span> }<a name="line.190"></a>
-<span class="sourceLineNo">191</span> if (outputDir == null) {<a name="line.191"></a>
-<span class="sourceLineNo">192</span> throw new IllegalArgumentException("Output directory not specified");<a name="line.192"></a>
+<span class="sourceLineNo">049</span>import org.apache.hadoop.hbase.client.BufferedMutator;<a name="line.49"></a>
+<span class="sourceLineNo">050</span>import org.apache.hadoop.hbase.client.ColumnFamilyDescriptor;<a name="line.50"></a>
+<span class="sourceLineNo">051</span>import org.apache.hadoop.hbase.client.ColumnFamilyDescriptorBuilder;<a name="line.51"></a>
+<span class="sourceLineNo">052</span>import org.apache.hadoop.hbase.client.Connection;<a name="line.52"></a>
+<span class="sourceLineNo">053</span>import org.apache.hadoop.hbase.client.ConnectionFactory;<a name="line.53"></a>
+<span class="sourceLineNo">054</span>import org.apache.hadoop.hbase.client.Get;<a name="line.54"></a>
+<span class="sourceLineNo">055</span>import org.apache.hadoop.hbase.client.Put;<a name="line.55"></a>
+<span class="sourceLineNo">056</span>import org.apache.hadoop.hbase.client.Result;<a name="line.56"></a>
+<span class="sourceLineNo">057</span>import org.apache.hadoop.hbase.client.Table;<a name="line.57"></a>
+<span class="sourceLineNo">058</span>import org.apache.hadoop.hbase.client.TableDescriptor;<a name="line.58"></a>
+<span class="sourceLineNo">059</span>import org.apache.hadoop.hbase.client.TableDescriptorBuilder;<a name="line.59"></a>
+<span class="sourceLineNo">060</span>import org.apache.hadoop.hbase.io.encoding.DataBlockEncoding;<a name="line.60"></a>
+<span class="sourceLineNo">061</span>import org.apache.hadoop.hbase.mapreduce.TableMapReduceUtil;<a name="line.61"></a>
+<span class="sourceLineNo">062</span>import org.apache.hadoop.hbase.regionserver.BloomType;<a name="line.62"></a>
+<span class="sourceLineNo">063</span>import org.apache.hadoop.hbase.test.util.CRC64;<a name="line.63"></a>
+<span class="sourceLineNo">064</span>import org.apache.hadoop.hbase.test.util.warc.WARCInputFormat;<a name="line.64"></a>
+<span class="sourceLineNo">065</span>import org.apache.hadoop.hbase.test.util.warc.WARCRecord;<a name="line.65"></a>
+<span class="sourceLineNo">066</span>import org.apache.hadoop.hbase.test.util.warc.WARCWritable;<a name="line.66"></a>
+<span class="sourceLineNo">067</span>import org.apache.hadoop.hbase.util.Bytes;<a name="line.67"></a>
+<span class="sourceLineNo">068</span>import org.apache.hadoop.hbase.util.EnvironmentEdgeManager;<a name="line.68"></a>
+<span class="sourceLineNo">069</span>import org.apache.hadoop.hbase.util.RegionSplitter;<a name="line.69"></a>
+<span class="sourceLineNo">070</span>import org.apache.hadoop.io.BytesWritable;<a name="line.70"></a>
+<span class="sourceLineNo">071</span>import org.apache.hadoop.io.LongWritable;<a name="line.71"></a>
+<span class="sourceLineNo">072</span>import org.apache.hadoop.io.NullWritable;<a name="line.72"></a>
+<span class="sourceLineNo">073</span>import org.apache.hadoop.io.SequenceFile.CompressionType;<a name="line.73"></a>
+<span class="sourceLineNo">074</span>import org.apache.hadoop.io.Writable;<a name="line.74"></a>
+<span class="sourceLineNo">075</span>import org.apache.hadoop.mapreduce.Counters;<a name="line.75"></a>
+<span class="sourceLineNo">076</span>import org.apache.hadoop.mapreduce.Job;<a name="line.76"></a>
+<span class="sourceLineNo">077</span>import org.apache.hadoop.mapreduce.JobContext;<a name="line.77"></a>
+<span class="sourceLineNo">078</span>import org.apache.hadoop.mapreduce.Mapper;<a name="line.78"></a>
+<span class="sourceLineNo">079</span>import org.apache.hadoop.mapreduce.lib.input.FileInputFormat;<a name="line.79"></a>
+<span class="sourceLineNo">080</span>import org.apache.hadoop.mapreduce.lib.input.SequenceFileInputFormat;<a name="line.80"></a>
+<span class="sourceLineNo">081</span>import org.apache.hadoop.mapreduce.lib.output.NullOutputFormat;<a name="line.81"></a>
+<span class="sourceLineNo">082</span>import org.apache.hadoop.mapreduce.lib.output.SequenceFileOutputFormat;<a name="line.82"></a>
+<span class="sourceLineNo">083</span>import org.apache.hadoop.util.Tool;<a name="line.83"></a>
+<span class="sourceLineNo">084</span>import org.apache.hadoop.util.ToolRunner;<a name="line.84"></a>
+<span class="sourceLineNo">085</span>import org.slf4j.Logger;<a name="line.85"></a>
+<span class="sourceLineNo">086</span>import org.slf4j.LoggerFactory;<a name="line.86"></a>
+<span class="sourceLineNo">087</span><a name="line.87"></a>
+<span class="sourceLineNo">088</span>import org.apache.hbase.thirdparty.org.apache.commons.cli.CommandLine;<a name="line.88"></a>
+<span class="sourceLineNo">089</span><a name="line.89"></a>
+<span class="sourceLineNo">090</span>/**<a name="line.90"></a>
+<span class="sourceLineNo">091</span> * This integration test loads successful resource retrieval records from the Common Crawl<a name="line.91"></a>
+<span class="sourceLineNo">092</span> * (https://commoncrawl.org/) public dataset into an HBase table and writes records that can be<a name="line.92"></a>
+<span class="sourceLineNo">093</span> * used to later verify the presence and integrity of those records.<a name="line.93"></a>
+<span class="sourceLineNo">094</span> * <p><a name="line.94"></a>
+<span class="sourceLineNo">095</span> * Run like:<a name="line.95"></a>
+<span class="sourceLineNo">096</span> * <blockquote><a name="line.96"></a>
+<span class="sourceLineNo">097</span> * ./bin/hbase org.apache.hadoop.hbase.test.IntegrationTestLoadCommonCrawl \<br><a name="line.97"></a>
+<span class="sourceLineNo">098</span> * &nbsp;&nbsp; -Dfs.s3n.awsAccessKeyId=&lt;AWS access key&gt; \<br><a name="line.98"></a>
+<span class="sourceLineNo">099</span> * &nbsp;&nbsp; -Dfs.s3n.awsSecretAccessKey=&lt;AWS secret key&gt; \<br><a name="line.99"></a>
+<span class="sourceLineNo">100</span> * &nbsp;&nbsp; /path/to/test-CC-MAIN-2021-10-warc.paths.gz \<br><a name="line.100"></a>
+<span class="sourceLineNo">101</span> * &nbsp;&nbsp; /path/to/tmp/warc-loader-output<a name="line.101"></a>
+<span class="sourceLineNo">102</span> * </blockquote><a name="line.102"></a>
+<span class="sourceLineNo">103</span> * <p><a name="line.103"></a>
+<span class="sourceLineNo">104</span> * Access to the Common Crawl dataset in S3 is made available to anyone by Amazon AWS, but<a name="line.104"></a>
+<span class="sourceLineNo">105</span> * Hadoop's S3N filesystem still requires valid access credentials to initialize.<a name="line.105"></a>
+<span class="sourceLineNo">106</span> * <p><a name="line.106"></a>
+<span class="sourceLineNo">107</span> * The input path can either specify a directory or a file. The file may optionally be<a name="line.107"></a>
+<span class="sourceLineNo">108</span> * compressed with gzip. If a directory, the loader expects the directory to contain one or more<a name="line.108"></a>
+<span class="sourceLineNo">109</span> * WARC files from the Common Crawl dataset. If a file, the loader expects a list of Hadoop S3N<a name="line.109"></a>
+<span class="sourceLineNo">110</span> * URIs which point to S3 locations for one or more WARC files from the Common Crawl dataset,<a name="line.110"></a>
+<span class="sourceLineNo">111</span> * one URI per line. Lines should be terminated with the UNIX line terminator.<a name="line.111"></a>
+<span class="sourceLineNo">112</span> * <p><a name="line.112"></a>
+<span class="sourceLineNo">113</span> * Included in hbase-it/src/test/resources/CC-MAIN-2021-10-warc.paths.gz is a list of all WARC<a name="line.113"></a>
+<span class="sourceLineNo">114</span> * files comprising the Q1 2021 crawl archive. There are 64,000 WARC files in this data set, each<a name="line.114"></a>
+<span class="sourceLineNo">115</span> * containing ~1GB of gzipped data. The WARC files contain several record types, such as metadata,<a name="line.115"></a>
+<span class="sourceLineNo">116</span> * request, and response, but we only load the response record types. If the HBase table schema<a name="line.116"></a>
+<span class="sourceLineNo">117</span> * does not specify compression (by default) there is roughly a 10x expansion. Loading the full<a name="line.117"></a>
+<span class="sourceLineNo">118</span> * crawl archive results in a table approximately 640 TB in size.<a name="line.118"></a>
+<span class="sourceLineNo">119</span> * <p><a name="line.119"></a>
+<span class="sourceLineNo">120</span> * You can also split the Loader and Verify stages:<a name="line.120"></a>
+<span class="sourceLineNo">121</span> * <p><a name="line.121"></a>
+<span class="sourceLineNo">122</span> * Load with:<a name="line.122"></a>
+<span class="sourceLineNo">123</span> * <blockquote><a name="line.123"></a>
+<span class="sourceLineNo">124</span> * ./bin/hbase 'org.apache.hadoop.hbase.test.IntegrationTestLoadCommonCrawl$Loader' \<br><a name="line.124"></a>
+<span class="sourceLineNo">125</span> * &nbsp;&nbsp; -files /path/to/hadoop-aws.jar \<br><a name="line.125"></a>
+<span class="sourceLineNo">126</span> * &nbsp;&nbsp; -Dfs.s3n.awsAccessKeyId=&lt;AWS access key&gt; \<br><a name="line.126"></a>
+<span class="sourceLineNo">127</span> * &nbsp;&nbsp; -Dfs.s3n.awsSecretAccessKey=&lt;AWS secret key&gt; \<br><a name="line.127"></a>
+<span class="sourceLineNo">128</span> * &nbsp;&nbsp; /path/to/test-CC-MAIN-2021-10-warc.paths.gz \<br><a name="line.128"></a>
+<span class="sourceLineNo">129</span> * &nbsp;&nbsp; /path/to/tmp/warc-loader-output<a name="line.129"></a>
+<span class="sourceLineNo">130</span> * </blockquote><a name="line.130"></a>
+<span class="sourceLineNo">131</span> * <p><a name="line.131"></a>
+<span class="sourceLineNo">132</span> * Note: The hadoop-aws jar will be needed at runtime to instantiate the S3N filesystem. Use<a name="line.132"></a>
+<span class="sourceLineNo">133</span> * the <tt>-files</tt> ToolRunner argument to add it.<a name="line.133"></a>
+<span class="sourceLineNo">134</span> * <p><a name="line.134"></a>
+<span class="sourceLineNo">135</span> * Verify with:<a name="line.135"></a>
+<span class="sourceLineNo">136</span> * <blockquote><a name="line.136"></a>
+<span class="sourceLineNo">137</span> * ./bin/hbase 'org.apache.hadoop.hbase.test.IntegrationTestLoadCommonCrawl$Verify' \<br><a name="line.137"></a>
+<span class="sourceLineNo">138</span> * &nbsp;&nbsp; /path/to/tmp/warc-loader-output<a name="line.138"></a>
+<span class="sourceLineNo">139</span> * </blockquote><a name="line.139"></a>
+<span class="sourceLineNo">140</span> * <p><a name="line.140"></a>
+<span class="sourceLineNo">141</span> */<a name="line.141"></a>
+<span class="sourceLineNo">142</span>public class IntegrationTestLoadCommonCrawl extends IntegrationTestBase {<a name="line.142"></a>
+<span class="sourceLineNo">143</span><a name="line.143"></a>
+<span class="sourceLineNo">144</span> private static final Logger LOG = LoggerFactory.getLogger(IntegrationTestLoadCommonCrawl.class);<a name="line.144"></a>
+<span class="sourceLineNo">145</span><a name="line.145"></a>
+<span class="sourceLineNo">146</span> protected static String TABLE_NAME_KEY = "IntegrationTestLoadCommonCrawl.table";<a name="line.146"></a>
+<span class="sourceLineNo">147</span> protected static String DEFAULT_TABLE_NAME = "IntegrationTestLoadCommonCrawl";<a name="line.147"></a>
+<span class="sourceLineNo">148</span><a name="line.148"></a>
+<span class="sourceLineNo">149</span> protected static byte[] CONTENT_FAMILY_NAME = Bytes.toBytes("c");<a name="line.149"></a>
+<span class="sourceLineNo">150</span> protected static byte[] INFO_FAMILY_NAME = Bytes.toBytes("i");<a name="line.150"></a>
+<span class="sourceLineNo">151</span> protected static byte[] CONTENT_QUALIFIER = HConstants.EMPTY_BYTE_ARRAY;<a name="line.151"></a>
+<span class="sourceLineNo">152</span> protected static byte[] CONTENT_LENGTH_QUALIFIER = Bytes.toBytes("l");<a name="line.152"></a>
+<span class="sourceLineNo">153</span> protected static byte[] CONTENT_TYPE_QUALIFIER = Bytes.toBytes("t");<a name="line.153"></a>
+<span class="sourceLineNo">154</span> protected static byte[] CRC_QUALIFIER = Bytes.toBytes("c");<a name="line.154"></a>
+<span class="sourceLineNo">155</span> protected static byte[] DATE_QUALIFIER = Bytes.toBytes("d");<a name="line.155"></a>
+<span class="sourceLineNo">156</span> protected static byte[] IP_ADDRESS_QUALIFIER = Bytes.toBytes("a");<a name="line.156"></a>
+<span class="sourceLineNo">157</span> protected static byte[] RECORD_ID_QUALIFIER = Bytes.toBytes("r");<a name="line.157"></a>
+<span class="sourceLineNo">158</span> protected static byte[] TARGET_URI_QUALIFIER = Bytes.toBytes("u");<a name="line.158"></a>
+<span class="sourceLineNo">159</span><a name="line.159"></a>
+<span class="sourceLineNo">160</span> private static final int VERIFICATION_READ_RETRIES = 10;<a name="line.160"></a>
+<span class="sourceLineNo">161</span><a name="line.161"></a>
+<span class="sourceLineNo">162</span> public static enum Counts {<a name="line.162"></a>
+<span class="sourceLineNo">163</span> REFERENCED, UNREFERENCED, CORRUPT<a name="line.163"></a>
+<span class="sourceLineNo">164</span> }<a name="line.164"></a>
+<span class="sourceLineNo">165</span><a name="line.165"></a>
+<span class="sourceLineNo">166</span> Path warcFileInputDir = null;<a name="line.166"></a>
+<span class="sourceLineNo">167</span> Path outputDir = null;<a name="line.167"></a>
+<span class="sourceLineNo">168</span> String[] args;<a name="line.168"></a>
+<span class="sourceLineNo">169</span><a name="line.169"></a>
+<span class="sourceLineNo">170</span> protected int runLoader(Path warcFileInputDir, Path outputDir) throws Exception {<a name="line.170"></a>
+<span class="sourceLineNo">171</span> Loader loader = new Loader();<a name="line.171"></a>
+<span class="sourceLineNo">172</span> loader.setConf(conf);<a name="line.172"></a>
+<span class="sourceLineNo">173</span> return loader.run(warcFileInputDir, outputDir);<a name="line.173"></a>
+<span class="sourceLineNo">174</span> }<a name="line.174"></a>
+<span class="sourceLineNo">175</span><a name="line.175"></a>
+<span class="sourceLineNo">176</span> protected int runVerify(Path inputDir) throws Exception {<a name="line.176"></a>
+<span class="sourceLineNo">177</span> Verify verify = new Verify();<a name="line.177"></a>
+<span class="sourceLineNo">178</span> verify.setConf(conf);<a name="line.178"></a>
+<span class="sourceLineNo">179</span> return verify.run(inputDir);<a name="line.179"></a>
+<span class="sourceLineNo">180</span> }<a name="line.180"></a>
+<span class="sourceLineNo">181</span><a name="line.181"></a>
+<span class="sourceLineNo">182</span> @Override<a name="line.182"></a>
+<span class="sourceLineNo">183</span> public int run(String[] args) {<a name="line.183"></a>
+<span class="sourceLineNo">184</span> if (args.length > 0) {<a name="line.184"></a>
+<span class="sourceLineNo">185</span> warcFileInputDir = new Path(args[0]);<a name="line.185"></a>
+<span class="sourceLineNo">186</span> if (args.length > 1) {<a name="line.186"></a>
+<span class="sourceLineNo">187</span> outputDir = new Path(args[1]);<a name="line.187"></a>
+<span class="sourceLineNo">188</span> }<a name="line.188"></a>
+<span class="sourceLineNo">189</span> }<a name="line.189"></a>
+<span class="sourceLineNo">190</span> try {<a name="line.190"></a>
+<span class="sourceLineNo">191</span> if (warcFileInputDir == null) {<a name="line.191"></a>
+<span class="sourceLineNo">192</span> throw new IllegalArgumentException("WARC input file or directory not specified");<a name="line.192"></a>
<span class="sourceLineNo">193</span> }<a name="line.193"></a>
-<span class="sourceLineNo">194</span> int res = runLoader(warcFileInputDir, outputDir);<a name="line.194"></a>
-<span class="sourceLineNo">195</span> if (res != 0) {<a name="line.195"></a>
-<span class="sourceLineNo">196</span> LOG.error("Loader failed");<a name="line.196"></a>
-<span class="sourceLineNo">197</span> return -1;<a name="line.197"></a>
-<span class="sourceLineNo">198</span> }<a name="line.198"></a>
-<span class="sourceLineNo">199</span> res = runVerify(outputDir);<a name="line.199"></a>
-<span class="sourceLineNo">200</span> } catch (Exception e) {<a name="line.200"></a>
-<span class="sourceLineNo">201</span> LOG.error("Tool failed with exception", e);<a name="line.201"></a>
-<span class="sourceLineNo">202</span> return -1;<a name="line.202"></a>
-<span class="sourceLineNo">203</span> }<a name="line.203"></a>
-<span class="sourceLineNo">204</span> return 0;<a name="line.204"></a>
-<span class="sourceLineNo">205</span> }<a name="line.205"></a>
-<span class="sourceLineNo">206</span><a name="line.206"></a>
-<span class="sourceLineNo">207</span> @Override<a name="line.207"></a>
-<span class="sourceLineNo">208</span> protected void processOptions(CommandLine cmd) {<a name="line.208"></a>
-<span class="sourceLineNo">209</span> processBaseOptions(cmd);<a name="line.209"></a>
-<span class="sourceLineNo">210</span> args = cmd.getArgs();<a name="line.210"></a>
-<span class="sourceLineNo">211</span> }<a name="line.211"></a>
-<span class="sourceLineNo">212</span><a name="line.212"></a>
-<span class="sourceLineNo">213</span> @Override<a name="line.213"></a>
-<span class="sourceLineNo">214</span> public void setUpCluster() throws Exception {<a name="line.214"></a>
-<span class="sourceLineNo">215</span> util = getTestingUtil(getConf());<a name="line.215"></a>
-<span class="sourceLineNo">216</span> boolean isDistributed = util.isDistributedCluster();<a name="line.216"></a>
-<span class="sourceLineNo">217</span> util.initializeCluster(isDistributed ? 1 : 3);<a name="line.217"></a>
-<span class="sourceLineNo">218</span> if (!isDistributed) {<a name="line.218"></a>
-<span class="sourceLineNo">219</span> util.startMiniMapReduceCluster();<a name="line.219"></a>
-<span class="sourceLineNo">220</span> }<a name="line.220"></a>
-<span class="sourceLineNo">221</span> this.setConf(util.getConfiguration());<a name="line.221"></a>
-<span class="sourceLineNo">222</span> }<a name="line.222"></a>
-<span class="sourceLineNo">223</span><a name="line.223"></a>
-<span class="sourceLineNo">224</span> @Override<a name="line.224"></a>
-<span class="sourceLineNo">225</span> public void cleanUpCluster() throws Exception {<a name="line.225"></a>
-<span class="sourceLineNo">226</span> super.cleanUpCluster();<a name="line.226"></a>
-<span class="sourceLineNo">227</span> if (util.isDistributedCluster()) {<a name="line.227"></a>
-<span class="sourceLineNo">228</span> util.shutdownMiniMapReduceCluster();<a name="line.228"></a>
-<span class="sourceLineNo">229</span> }<a name="line.229"></a>
-<span class="sourceLineNo">230</span> }<a name="line.230"></a>
-<span class="sourceLineNo">231</span><a name="line.231"></a>
-<span class="sourceLineNo">232</span> static TableName getTablename(Configuration c) {<a name="line.232"></a>
-<span class="sourceLineNo">233</span> return TableName.valueOf(c.get(TABLE_NAME_KEY, DEFAULT_TABLE_NAME));<a name="line.233"></a>
-<span class="sourceLineNo">234</span> }<a name="line.234"></a>
-<span class="sourceLineNo">235</span><a name="line.235"></a>
-<span class="sourceLineNo">236</span> @Override<a name="line.236"></a>
-<span class="sourceLineNo">237</span> public TableName getTablename() {<a name="line.237"></a>
-<span class="sourceLineNo">238</span> return getTablename(getConf());<a name="line.238"></a>
-<span class="sourceLineNo">239</span> }<a name="line.239"></a>
-<span class="sourceLineNo">240</span><a name="line.240"></a>
-<span class="sourceLineNo">241</span> @Override<a name="line.241"></a>
-<span class="sourceLineNo">242</span> protected Set<String> getColumnFamilies() {<a name="line.242"></a>
-<span class="sourceLineNo">243</span> Set<String> families = new HashSet<>();<a name="line.243"></a>
-<span class="sourceLineNo">244</span> families.add(Bytes.toString(CONTENT_FAMILY_NAME));<a name="line.244"></a>
-<span class="sourceLineNo">245</span> families.add(Bytes.toString(INFO_FAMILY_NAME));<a name="line.245"></a>
-<span class="sourceLineNo">246</span> return families;<a name="line.246"></a>
-<span class="sourceLineNo">247</span> }<a name="line.247"></a>
-<span class="sourceLineNo">248</span><a name="line.248"></a>
-<span class="sourceLineNo">249</span> @Override<a name="line.249"></a>
-<span class="sourceLineNo">250</span> public int runTestFromCommandLine() throws Exception {<a name="line.250"></a>
-<span class="sourceLineNo">251</span> return ToolRunner.run(getConf(), this, args);<a name="line.251"></a>
-<span class="sourceLineNo">252</span> }<a name="line.252"></a>
-<span class="sourceLineNo">253</span><a name="line.253"></a>
-<span class="sourceLineNo">254</span> public static void main(String[] args) throws Exception {<a name="line.254"></a>
-<span class="sourceLineNo">255</span> Configuration conf = HBaseConfiguration.create();<a name="line.255"></a>
-<span class="sourceLineNo">256</span> IntegrationTestingUtility.setUseDistributedCluster(conf);<a name="line.256"></a>
-<span class="sourceLineNo">257</span> int ret = ToolRunner.run(conf, new IntegrationTestLoadCommonCrawl(), args);<a name="line.257"></a>
-<span class="sourceLineNo">258</span> System.exit(ret);<a name="line.258"></a>
-<span class="sourceLineNo">259</span> }<a name="line.259"></a>
-<span class="sourceLineNo">260</span><a name="line.260"></a>
-<span class="sourceLineNo">261</span> public static class HBaseKeyWritable implements Writable {<a name="line.261"></a>
-<span class="sourceLineNo">262</span><a name="line.262"></a>
-<span class="sourceLineNo">263</span> private byte[] row;<a name="line.263"></a>
-<span class="sourceLineNo">264</span> private int rowOffset;<a name="line.264"></a>
-<span class="sourceLineNo">265</span> private int rowLength;<a name="line.265"></a>
-<span class="sourceLineNo">266</span> private byte[] family;<a name="line.266"></a>
-<span class="sourceLineNo">267</span> private int familyOffset;<a name="line.267"></a>
-<span class="sourceLineNo">268</span> private int familyLength;<a name="line.268"></a>
-<span class="sourceLineNo">269</span> private byte[] qualifier;<a name="line.269"></a>
-<span class="sourceLineNo">270</span> private int qualifierOffset;<a name="line.270"></a>
-<span class="sourceLineNo">271</span> private int qualifierLength;<a name="line.271"></a>
-<span class="sourceLineNo">272</span> private long ts;<a name="line.272"></a>
-<span class="sourceLineNo">273</span><a name="line.273"></a>
-<span class="sourceLineNo">274</span> public HBaseKeyWritable() { }<a name="line.274"></a>
-<span class="sourceLineNo">275</span><a name="line.275"></a>
-<span class="sourceLineNo">276</span> public HBaseKeyWritable(byte[] row, int rowOffset, int rowLength,<a name="line.276"></a>
-<span class="sourceLineNo">277</span> byte[] family, int familyOffset, int familyLength,<a name="line.277"></a>
-<span class="sourceLineNo">278</span> byte[] qualifier, int qualifierOffset, int qualifierLength, long ts) {<a name="line.278"></a>
-<span class="sourceLineNo">279</span> this.row = row;<a name="line.279"></a>
-<span class="sourceLineNo">280</span> this.rowOffset = rowOffset;<a name="line.280"></a>
-<span class="sourceLineNo">281</span> this.rowLength = rowLength;<a name="line.281"></a>
-<span class="sourceLineNo">282</span> this.family = family;<a name="line.282"></a>
-<span class="sourceLineNo">283</span> this.familyOffset = familyOffset;<a name="line.283"></a>
-<span class="sourceLineNo">284</span> this.familyLength = familyLength;<a name="line.284"></a>
-<span class="sourceLineNo">285</span> this.qualifier = qualifier;<a name="line.285"></a>
-<span class="sourceLineNo">286</span> this.qualifierOffset = qualifierOffset;<a name="line.286"></a>
-<span class="sourceLineNo">287</span> this.qualifierLength = qualifierLength;<a name="line.287"></a>
-<span class="sourceLineNo">288</span> this.ts = ts;<a name="line.288"></a>
-<span class="sourceLineNo">289</span> }<a name="line.289"></a>
-<span class="sourceLineNo">290</span><a name="line.290"></a>
-<span class="sourceLineNo">291</span> public HBaseKeyWritable(byte[] row, byte[] family, byte[] qualifier, long ts) {<a name="line.291"></a>
-<span class="sourceLineNo">292</span> this(row, 0, row.length,<a name="line.292"></a>
-<span class="sourceLineNo">293</span> family, 0, family.length,<a name="line.293"></a>
-<span class="sourceLineNo">294</span> qualifier, 0, qualifier != null ? qualifier.length : 0, ts);<a name="line.294"></a>
-<span class="sourceLineNo">295</span> }<a name="line.295"></a>
-<span class="sourceLineNo">296</span><a name="line.296"></a>
-<span class="sourceLineNo">297</span> public HBaseKeyWritable(byte[] row, byte[] family, long ts) {<a name="line.297"></a>
-<span class="sourceLineNo">298</span> this(row, family, HConstants.EMPTY_BYTE_ARRAY, ts);<a name="line.298"></a>
-<span class="sourceLineNo">299</span> }<a name="line.299"></a>
-<span class="sourceLineNo">300</span><a name="line.300"></a>
-<span class="sourceLineNo">301</span> public HBaseKeyWritable(Cell cell) {<a name="line.301"></a>
-<span class="sourceLineNo">302</span> this(cell.getRowArray(), cell.getRowOffset(), cell.getRowLength(),<a name="line.302"></a>
-<span class="sourceLineNo">303</span> cell.getFamilyArray(), cell.getFamilyOffset(), cell.getFamilyLength(),<a name="line.303"></a>
-<span class="sourceLineNo">304</span> cell.getQualifierArray(), cell.getQualifierOffset(), cell.getQualifierLength(),<a name="line.304"></a>
-<span class="sourceLineNo">305</span> cell.getTimestamp());<a name="line.305"></a>
-<span class="sourceLineNo">306</span> }<a name="line.306"></a>
-<span class="sourceLineNo">307</span><a name="line.307"></a>
-<span class="sourceLineNo">308</span> @Override<a name="line.308"></a>
-<span class="sourceLineNo">309</span> public void readFields(DataInput in) throws IOException {<a name="line.309"></a>
-<span class="sourceLineNo">310</span> this.row = Bytes.toBytes(in.readUTF());<a name="line.310"></a>
-<span class="sourceLineNo">311</span> this.rowOffset = 0;<a name="line.311"></a>
-<span class="sourceLineNo">312</span> this.rowLength = row.length;<a name="line.312"></a>
-<span class="sourceLineNo">313</span> this.family = Bytes.toBytes(in.readUTF());<a name="line.313"></a>
-<span class="sourceLineNo">314</span> this.familyOffset = 0;<a name="line.314"></a>
-<span class="sourceLineNo">315</span> this.familyLength = family.length;<a name="line.315"></a>
-<span class="sourceLineNo">316</span> this.qualifier = Bytes.toBytes(in.readUTF());<a name="line.316"></a>
-<span class="sourceLineNo">317</span> this.qualifierOffset = 0;<a name="line.317"></a>
-<span class="sourceLineNo">318</span> this.qualifierLength = qualifier.length;<a name="line.318"></a>
-<span class="sourceLineNo">319</span> this.ts = in.readLong();<a name="line.319"></a>
-<span class="sourceLineNo">320</span> }<a name="line.320"></a>
-<span class="sourceLineNo">321</span><a name="line.321"></a>
-<span class="sourceLineNo">322</span> @Override<a name="line.322"></a>
-<span class="sourceLineNo">323</span> public void write(DataOutput out) throws IOException {<a name="line.323"></a>
-<span class="sourceLineNo">324</span> out.writeUTF(new String(row, rowOffset, rowLength, StandardCharsets.UTF_8));<a name="line.324"></a>
-<span class="sourceLineNo">325</span> out.writeUTF(new String(family, familyOffset, familyLength, StandardCharsets.UTF_8));<a name="line.325"></a>
-<span class="sourceLineNo">326</span> if (qualifier != null) {<a name="line.326"></a>
-<span class="sourceLineNo">327</span> out.writeUTF(new String(qualifier, qualifierOffset, qualifierLength,<a name="line.327"></a>
-<span class="sourceLineNo">328</span> StandardCharsets.UTF_8));<a name="line.328"></a>
-<span class="sourceLineNo">329</span> } else {<a name="line.329"></a>
-<span class="sourceLineNo">330</span> out.writeUTF("");<a name="line.330"></a>
-<span class="sourceLineNo">331</span> }<a name="line.331"></a>
-<span class="sourceLineNo">332</span> out.writeLong(ts);<a name="line.332"></a>
-<span class="sourceLineNo">333</span> }<a name="line.333"></a>
-<span class="sourceLineNo">334</span><a name="line.334"></a>
-<span class="sourceLineNo">335</span> public byte[] getRowArray() {<a name="line.335"></a>
-<span class="sourceLineNo">336</span> return row;<a name="line.336"></a>
-<span class="sourceLineNo">337</span> }<a name="line.337"></a>
-<span class="sourceLineNo">338</span><a name="line.338"></a>
-<span class="sourceLineNo">339</span> public void setRow(byte[] row) {<a name="line.339"></a>
-<span class="sourceLineNo">340</span> this.row = row;<a name="line.340"></a>
-<span class="sourceLineNo">341</span> }<a name="line.341"></a>
-<span class="sourceLineNo">342</span><a name="line.342"></a>
-<span class="sourceLineNo">343</span> public int getRowOffset() {<a name="line.343"></a>
-<span class="sourceLineNo">344</span> return rowOffset;<a name="line.344"></a>
-<span class="sourceLineNo">345</span> }<a name="line.345"></a>
-<span class="sourceLineNo">346</span><a name="line.346"></a>
-<span class="sourceLineNo">347</span> public void setRowOffset(int rowOffset) {<a name="line.347"></a>
-<span class="sourceLineNo">348</span> this.rowOffset = rowOffset;<a name="line.348"></a>
-<span class="sourceLineNo">349</span> }<a name="line.349"></a>
-<span class="sourceLineNo">350</span><a name="line.350"></a>
-<span class="sourceLineNo">351</span> public int getRowLength() {<a name="line.351"></a>
-<span class="sourceLineNo">352</span> return rowLength;<a name="line.352"></a>
-<span class="sourceLineNo">353</span> }<a name="line.353"></a>
-<span class="sourceLineNo">354</span><a name="line.354"></a>
-<span class="sourceLineNo">355</span> public void setRowLength(int rowLength) {<a name="line.355"></a>
-<span class="sourceLineNo">356</span> this.rowLength = rowLength;<a name="line.356"></a>
-<span class="sourceLineNo">357</span> }<a name="line.357"></a>
-<span class="sourceLineNo">358</span><a name="line.358"></a>
-<span class="sourceLineNo">359</span> public byte[] getFamilyArray() {<a name="line.359"></a>
-<span class="sourceLineNo">360</span> return family;<a name="line.360"></a>
-<span class="sourceLineNo">361</span> }<a name="line.361"></a>
-<span class="sourceLineNo">362</span><a name="line.362"></a>
-<span class="sourceLineNo">363</span> public void setFamily(byte[] family) {<a name="line.363"></a>
-<span class="sourceLineNo">364</span> this.family = family;<a name="line.364"></a>
-<span class="sourceLineNo">365</span> }<a name="line.365"></a>
-<span class="sourceLineNo">366</span><a name="line.366"></a>
-<span class="sourceLineNo">367</span> public int getFamilyOffset() {<a name="line.367"></a>
-<span class="sourceLineNo">368</span> return familyOffset;<a name="line.368"></a>
-<span class="sourceLineNo">369</span> }<a name="line.369"></a>
-<span class="sourceLineNo">370</span><a name="line.370"></a>
-<span class="sourceLineNo">371</span> public void setFamilyOffset(int familyOffset) {<a name="line.371"></a>
-<span class="sourceLineNo">372</span> this.familyOffset = familyOffset;<a name="line.372"></a>
-<span class="sourceLineNo">373</span> }<a name="line.373"></a>
-<span class="sourceLineNo">374</span><a name="line.374"></a>
-<span class="sourceLineNo">375</span> public int getFamilyLength() {<a name="line.375"></a>
-<span class="sourceLineNo">376</span> return familyLength;<a name="line.376"></a>
-<span class="sourceLineNo">377</span> }<a name="line.377"></a>
-<span class="sourceLineNo">378</span><a name="line.378"></a>
-<span class="sourceLineNo">379</span> public void setFamilyLength(int familyLength) {<a name="line.379"></a>
-<span class="sourceLineNo">380</span> this.familyLength = familyLength;<a name="line.380"></a>
-<span class="sourceLineNo">381</span> }<a name="line.381"></a>
-<span class="sourceLineNo">382</span><a name="line.382"></a>
-<span class="sourceLineNo">383</span> public byte[] getQualifierArray() {<a name="line.383"></a>
-<span class="sourceLineNo">384</span> return qualifier;<a name="line.384"></a>
-<span class="sourceLineNo">385</span> }<a name="line.385"></a>
-<span class="sourceLineNo">386</span><a name="line.386"></a>
-<span class="sourceLineNo">387</span> public void setQualifier(byte[] qualifier) {<a name="line.387"></a>
-<span class="sourceLineNo">388</span> this.qualifier = qualifier;<a name="line.388"></a>
-<span class="sourceLineNo">389</span> }<a name="line.389"></a>
-<span class="sourceLineNo">390</span><a name="line.390"></a>
-<span class="sourceLineNo">391</span> public int getQualifierOffset() {<a name="line.391"></a>
-<span class="sourceLineNo">392</span> return qualifierOffset;<a name="line.392"></a>
-<span class="sourceLineNo">393</span> }<a name="line.393"></a>
-<span class="sourceLineNo">394</span><a name="line.394"></a>
-<span class="sourceLineNo">395</span> public void setQualifierOffset(int qualifierOffset) {<a name="line.395"></a>
-<span class="sourceLineNo">396</span> this.qualifierOffset = qualifierOffset;<a name="line.396"></a>
-<span class="sourceLineNo">397</span> }<a name="line.397"></a>
-<span class="sourceLineNo">398</span><a name="line.398"></a>
-<span class="sourceLineNo">399</span> public int getQualifierLength() {<a name="line.399"></a>
-<span class="sourceLineNo">400</span> return qualifierLength;<a name="line.400"></a>
-<span class="sourceLineNo">401</span> }<a name="line.401"></a>
-<span class="sourceLineNo">402</span><a name="line.402"></a>
-<span class="sourceLineNo">403</span> public void setQualifierLength(int qualifierLength) {<a name="line.403"></a>
-<span class="sourceLineNo">404</span> this.qualifierLength = qualifierLength;<a name="line.404"></a>
-<span class="sourceLineNo">405</span> }<a name="line.405"></a>
-<span class="sourceLineNo">406</span><a name="line.406"></a>
-<span class="sourceLineNo">407</span> public long getTimestamp() {<a name="line.407"></a>
-<span class="sourceLineNo">408</span> return ts;<a name="line.408"></a>
-<span class="sourceLineNo">409</span> }<a name="line.409"></a>
-<span class="sourceLineNo">410</span><a name="line.410"></a>
-<span class="sourceLineNo">411</span> public void setTimestamp(long ts) {<a name="line.411"></a>
-<span class="sourceLineNo">412</span> this.ts = ts;<a name="line.412"></a>
-<span class="sourceLineNo">413</span> }<a name="line.413"></a>
-<span class="sourceLineNo">414</span> }<a name="line.414"></a>
-<span class="sourceLineNo">415</span><a name="line.415"></a>
-<span class="sourceLineNo">416</span> public static class Loader extends Configured implements Tool {<a name="line.416"></a>
-<span class="sourceLineNo">417</span><a name="line.417"></a>
-<span class="sourceLineNo">418</span> private static final Logger LOG = LoggerFactory.getLogger(Loader.class);<a name="line.418"></a>
-<span class="sourceLineNo">419</span> private static final String USAGE = "Loader <warInputDir | warFileList> <outputDir>";<a name="line.419"></a>
+<span class="sourceLineNo">194</span> if (outputDir == null) {<a name="line.194"></a>
+<span class="sourceLineNo">195</span> throw new IllegalArgumentException("Output directory not specified");<a name="line.195"></a>
+<span class="sourceLineNo">196</span> }<a name="line.196"></a>
+<span class="sourceLineNo">197</span> int res = runLoader(warcFileInputDir, outputDir);<a name="line.197"></a>
+<span class="sourceLineNo">198</span> if (res != 0) {<a name="line.198"></a>
+<span class="sourceLineNo">199</span> LOG.error("Loader failed");<a name="line.199"></a>
+<span class="sourceLineNo">200</span> return -1;<a name="line.200"></a>
+<span class="sourceLineNo">201</span> }<a name="line.201"></a>
+<span class="sourceLineNo">202</span> res = runVerify(outputDir);<a name="line.202"></a>
+<span class="sourceLineNo">203</span> } catch (Exception e) {<a name="line.203"></a>
+<span class="sourceLineNo">204</span> LOG.error("Tool failed with exception", e);<a name="line.204"></a>
+<span class="sourceLineNo">205</span> return -1;<a name="line.205"></a>
+<span class="sourceLineNo">206</span> }<a name="line.206"></a>
+<span class="sourceLineNo">207</span> return 0;<a name="line.207"></a>
+<span class="sourceLineNo">208</span> }<a name="line.208"></a>
+<span class="sourceLineNo">209</span><a name="line.209"></a>
+<span class="sourceLineNo">210</span> @Override<a name="line.210"></a>
+<span class="sourceLineNo">211</span> protected void processOptions(CommandLine cmd) {<a name="line.211"></a>
+<span class="sourceLineNo">212</span> processBaseOptions(cmd);<a name="line.212"></a>
+<span class="sourceLineNo">213</span> args = cmd.getArgs();<a name="line.213"></a>
+<span class="sourceLineNo">214</span> }<a name="line.214"></a>
+<span class="sourceLineNo">215</span><a name="line.215"></a>
+<span class="sourceLineNo">216</span> @Override<a name="line.216"></a>
+<span class="sourceLineNo">217</span> public void setUpCluster() throws Exception {<a name="line.217"></a>
+<span class="sourceLineNo">218</span> util = getTestingUtil(getConf());<a name="line.218"></a>
+<span class="sourceLineNo">219</span> boolean isDistributed = util.isDistributedCluster();<a name="line.219"></a>
+<span class="sourceLineNo">220</span> util.initializeCluster(isDistributed ? 1 : 3);<a name="line.220"></a>
+<span class="sourceLineNo">221</span> if (!isDistributed) {<a name="line.221"></a>
+<span class="sourceLineNo">222</span> util.startMiniMapReduceCluster();<a name="line.222"></a>
+<span class="sourceLineNo">223</span> }<a name="line.223"></a>
+<span class="sourceLineNo">224</span> this.setConf(util.getConfiguration());<a name="line.224"></a>
+<span class="sourceLineNo">225</span> }<a name="line.225"></a>
+<span class="sourceLineNo">226</span><a name="line.226"></a>
+<span class="sourceLineNo">227</span> @Override<a name="line.227"></a>
+<span class="sourceLineNo">228</span> public void cleanUpCluster() throws Exception {<a name="line.228"></a>
+<span class="sourceLineNo">229</span> super.cleanUpCluster();<a name="line.229"></a>
+<span class="sourceLineNo">230</span> if (util.isDistributedCluster()) {<a name="line.230"></a>
+<span class="sourceLineNo">231</span> util.shutdownMiniMapReduceCluster();<a name="line.231"></a>
+<span class="sourceLineNo">232</span> }<a name="line.232"></a>
+<span class="sourceLineNo">233</span> }<a name="line.233"></a>
+<span class="sourceLineNo">234</span><a name="line.234"></a>
+<span class="sourceLineNo">235</span> static TableName getTablename(Configuration c) {<a name="line.235"></a>
+<span class="sourceLineNo">236</span> return TableName.valueOf(c.get(TABLE_NAME_KEY, DEFAULT_TABLE_NAME));<a name="line.236"></a>
+<span class="sourceLineNo">237</span> }<a name="line.237"></a>
+<span class="sourceLineNo">238</span><a name="line.238"></a>
+<span class="sourceLineNo">239</span> @Override<a name="line.239"></a>
+<span class="sourceLineNo">240</span> public TableName getTablename() {<a name="line.240"></a>
+<span class="sourceLineNo">241</span> return getTablename(getConf());<a name="line.241"></a>
+<span class="sourceLineNo">242</span> }<a name="line.242"></a>
+<span class="sourceLineNo">243</span><a name="line.243"></a>
+<span class="sourceLineNo">244</span> @Override<a name="line.244"></a>
+<span class="sourceLineNo">245</span> protected Set<String> getColumnFamilies() {<a name="line.245"></a>
+<span class="sourceLineNo">246</span> Set<String> families = new HashSet<>();<a name="line.246"></a>
+<span class="sourceLineNo">247</span> families.add(Bytes.toString(CONTENT_FAMILY_NAME));<a name="line.247"></a>
+<span class="sourceLineNo">248</span> families.add(Bytes.toString(INFO_FAMILY_NAME));<a name="line.248"></a>
+<span class="sourceLineNo">249</span> return families;<a name="line.249"></a>
+<span class="sourceLineNo">250</span> }<a name="line.250"></a>
+<span class="sourceLineNo">251</span><a name="line.251"></a>
+<span class="sourceLineNo">252</span> @Override<a name="line.252"></a>
+<span class="sourceLineNo">253</span> public int runTestFromCommandLine() throws Exception {<a name="line.253"></a>
+<span class="sourceLineNo">254</span> return ToolRunner.run(getConf(), this, args);<a name="line.254"></a>
+<span class="sourceLineNo">255</span> }<a name="line.255"></a>
+<span class="sourceLineNo">256</span><a name="line.256"></a>
+<span class="sourceLineNo">257</span> public static void main(String[] args) throws Exception {<a name="line.257"></a>
+<span class="sourceLineNo">258</span> Configuration conf = HBaseConfiguration.create();<a name="line.258"></a>
+<span class="sourceLineNo">259</span> IntegrationTestingUtility.setUseDistributedCluster(conf);<a name="line.259"></a>
+<span class="sourceLineNo">260</span> int ret = ToolRunner.run(conf, new IntegrationTestLoadCommonCrawl(), args);<a name="line.260"></a>
+<span class="sourceLineNo">261</span> System.exit(ret);<a name="line.261"></a>
+<span class="sourceLineNo">262</span> }<a name="line.262"></a>
+<span class="sourceLineNo">263</span><a name="line.263"></a>
+<span class="sourceLineNo">264</span> public static class HBaseKeyWritable implements Writable {<a name="line.264"></a>
+<span class="sourceLineNo">265</span><a name="line.265"></a>
+<span class="sourceLineNo">266</span> private byte[] row;<a name="line.266"></a>
+<span class="sourceLineNo">267</span> private int rowOffset;<a name="line.267"></a>
+<span class="sourceLineNo">268</span> private int rowLength;<a name="line.268"></a>
+<span class="sourceLineNo">269</span> private byte[] family;<a name="line.269"></a>
+<span class="sourceLineNo">270</span> private int familyOffset;<a name="line.270"></a>
+<span class="sourceLineNo">271</span> private int familyLength;<a name="line.271"></a>
+<span class="sourceLineNo">272</span> private byte[] qualifier;<a name="line.272"></a>
+<span class="sourceLineNo">273</span> private int qualifierOffset;<a name="line.273"></a>
+<span class="sourceLineNo">274</span> private int qualifierLength;<a name="line.274"></a>
+<span class="sourceLineNo">275</span> private long ts;<a name="line.275"></a>
+<span class="sourceLineNo">276</span><a name="line.276"></a>
+<span class="sourceLineNo">277</span> public HBaseKeyWritable() { }<a name="line.277"></a>
+<span class="sourceLineNo">278</span><a name="line.278"></a>
+<span class="sourceLineNo">279</span> public HBaseKeyWritable(byte[] row, int rowOffset, int rowLength,<a name="line.279"></a>
+<span class="sourceLineNo">280</span> byte[] family, int familyOffset, int familyLength,<a name="line.280"></a>
+<span class="sourceLineNo">281</span> byte[] qualifier, int qualifierOffset, int qualifierLength, long ts) {<a name="line.281"></a>
+<span class="sourceLineNo">282</span> this.row = row;<a name="line.282"></a>
+<span class="sourceLineNo">283</span> this.rowOffset = rowOffset;<a name="line.283"></a>
+<span class="sourceLineNo">284</span> this.rowLength = rowLength;<a name="line.284"></a>
+<span class="sourceLineNo">285</span> this.family = family;<a name="line.285"></a>
+<span class="sourceLineNo">286</span> this.familyOffset = familyOffset;<a name="line.286"></a>
+<span class="sourceLineNo">287</span> this.familyLength = familyLength;<a name="line.287"></a>
+<span class="sourceLineNo">288</span> this.qualifier = qualifier;<a name="line.288"></a>
+<span class="sourceLineNo">289</span> this.qualifierOffset = qualifierOffset;<a name="line.289"></a>
+<span class="sourceLineNo">290</span> this.qualifierLength = qualifierLength;<a name="line.290"></a>
+<span class="sourceLineNo">291</span> this.ts = ts;<a name="line.291"></a>
+<span class="sourceLineNo">292</span> }<a name="line.292"></a>
+<span class="sourceLineNo">293</span><a name="line.293"></a>
+<span class="sourceLineNo">294</span> public HBaseKeyWritable(byte[] row, byte[] family, byte[] qualifier, long ts) {<a name="line.294"></a>
+<span class="sourceLineNo">295</span> this(row, 0, row.length,<a name="line.295"></a>
+<span class="sourceLineNo">296</span> family, 0, family.length,<a name="line.296"></a>
+<span class="sourceLineNo">297</span> qualifier, 0, qualifier != null ? qualifier.length : 0, ts);<a name="line.297"></a>
+<span class="sourceLineNo">298</span> }<a name="line.298"></a>
+<span class="sourceLineNo">299</span><a name="line.299"></a>
+<span class="sourceLineNo">300</span> public HBaseKeyWritable(byte[] row, byte[] family, long ts) {<a name="line.300"></a>
+<span class="sourceLineNo">301</span> this(row, family, HConstants.EMPTY_BYTE_ARRAY, ts);<a name="line.301"></a>
+<span class="sourceLineNo">302</span> }<a name="line.302"></a>
+<span class="sourceLineNo">303</span><a name="line.303"></a>
+<span class="sourceLineNo">304</span> public HBaseKeyWritable(Cell cell) {<a name="line.304"></a>
+<span class="sourceLineNo">305</span> this(cell.getRowArray(), cell.getRowOffset(), cell.getRowLength(),<a name="line.305"></a>
+<span class="sourceLineNo">306</span> cell.getFamilyArray(), cell.getFamilyOffset(), cell.getFamilyLength(),<a name="line.306"></a>
+<span class="sourceLineNo">307</span> cell.getQualifierArray(), cell.getQualifierOffset(), cell.getQualifierLength(),<a name="line.307"></a>
+<span class="sourceLineNo">308</span> cell.getTimestamp());<a name="line.308"></a>
+<span class="sourceLineNo">309</span> }<a name="line.309"></a>
+<span class="sourceLineNo">310</span><a name="line.310"></a>
+<span class="sourceLineNo">311</span> @Override<a name="line.311"></a>
+<span class="sourceLineNo">312</span> public void readFields(DataInput in) throws IOException {<a name="line.312"></a>
+<span class="sourceLineNo">313</span> this.row = Bytes.toBytes(in.readUTF());<a name="line.313"></a>
+<span class="sourceLineNo">314</span> this.rowOffset = 0;<a name="line.314"></a>
+<span class="sourceLineNo">315</span> this.rowLength = row.length;<a name="line.315"></a>
+<span class="sourceLineNo">316</span> this.family = Bytes.toBytes(in.readUTF());<a name="line.316"></a>
+<span class="sourceLineNo">317</span> this.familyOffset = 0;<a name="line.317"></a>
+<span class="sourceLineNo">318</span> this.familyLength = family.length;<a name="line.318"></a>
+<span class="sourceLineNo">319</span> this.qualifier = Bytes.toBytes(in.readUTF());<a name="line.319"></a>
+<span class="sourceLineNo">320</span> this.qualifierOffset = 0;<a name="line.320"></a>
+<span class="sourceLineNo">321</span> this.qualifierLength = qualifier.length;<a name="line.321"></a>
+<span class="sourceLineNo">322</span> this.ts = in.readLong();<a name="line.322"></a>
+<span class="sourceLineNo">323</span> }<a name="line.323"></a>
+<span class="sourceLineNo">324</span><a name="line.324"></a>
+<span class="sourceLineNo">325</span> @Override<a name="line.325"></a>
+<span class="sourceLineNo">326</span> public void write(DataOutput out) throws IOException {<a name="line.326"></a>
+<span class="sourceLineNo">327</span> out.writeUTF(new String(row, rowOffset, rowLength, StandardCharsets.UTF_8));<a name="line.327"></a>
+<span class="sourceLineNo">328</span> out.writeUTF(new String(family, familyOffset, familyLength, StandardCharsets.UTF_8));<a name="line.328"></a>
+<span class="sourceLineNo">329</span> if (qualifier != null) {<a name="line.329"></a>
+<span class="sourceLineNo">330</span> out.writeUTF(new String(qualifier, qualifierOffset, qualifierLength,<a name="line.330"></a>
+<span class="sourceLineNo">331</span> StandardCharsets.UTF_8));<a name="line.331"></a>
+<span class="sourceLineNo">332</span> } else {<a name="line.332"></a>
+<span class="sourceLineNo">333</span> out.writeUTF("");<a name="line.333"></a>
+<span class="sourceLineNo">334</span> }<a name="line.334"></a>
+<span class="sourceLineNo">335</span> out.writeLong(ts);<a name="line.335"></a>
+<span class="sourceLineNo">336</span> }<a name="line.336"></a>
+<span class="sourceLineNo">337</span><a name="line.337"></a>
+<span class="sourceLineNo">338</span> public byte[] getRowArray() {<a name="line.338"></a>
+<span class="sourceLineNo">339</span> return row;<a name="line.339"></a>
+<span class="sourceLineNo">340</span> }<a name="line.340"></a>
+<span class="sourceLineNo">341</span><a name="line.341"></a>
+<span class="sourceLineNo">342</span> public void setRow(byte[] row) {<a name="line.342"></a>
+<span class="sourceLineNo">343</span> this.row = row;<a name="line.343"></a>
+<span class="sourceLineNo">344</span> }<a name="line.344"></a>
+<span class="sourceLineNo">345</span><a name="line.345"></a>
+<span class="sourceLineNo">346</span> public int getRowOffset() {<a name="line.346"></a>
+<span class="sourceLineNo">347</span> return rowOffset;<a name="line.347"></a>
+<span class="sourceLineNo">348</span> }<a name="line.348"></a>
+<span class="sourceLineNo">349</span><a name="line.349"></a>
+<span class="sourceLineNo">350</span> public void setRowOffset(int rowOffset) {<a name="line.350"></a>
+<span class="sourceLineNo">351</span> this.rowOffset = rowOffset;<a name="line.351"></a>
+<span class="sourceLineNo">352</span> }<a name="line.352"></a>
+<span class="sourceLineNo">353</span><a name="line.353"></a>
+<span class="sourceLineNo">354</span> public int getRowLength() {<a name="line.354"></a>
+<span class="sourceLineNo">355</span> return rowLength;<a name="line.355"></a>
+<span class="sourceLineNo">356</span> }<a name="line.356"></a>
+<span class="sourceLineNo">357</span><a name="line.357"></a>
+<span class="sourceLineNo">358</span> public void setRowLength(int rowLength) {<a name="line.358"></a>
+<span class="sourceLineNo">359</span> this.rowLength = rowLength;<a name="line.359"></a>
+<span class="sourceLineNo">360</span> }<a name="line.360"></a>
+<span class="sourceLineNo">361</span><a name="line.361"></a>
+<span class="sourceLineNo">362</span> public byte[] getFamilyArray() {<a name="line.362"></a>
+<span class="sourceLineNo">363</span> return family;<a name="line.363"></a>
+<span class="sourceLineNo">364</span> }<a name="line.364"></a>
+<span class="sourceLineNo">365</span><a name="line.365"></a>
+<span class="sourceLineNo">366</span> public void setFamily(byte[] family) {<a name="line.366"></a>
+<span class="sourceLineNo">367</span> this.family = family;<a name="line.367"></a>
+<span class="sourceLineNo">368</span> }<a name="line.368"></a>
+<span class="sourceLineNo">369</span><a name="line.369"></a>
+<span class="sourceLineNo">370</span> public int getFamilyOffset() {<a name="line.370"></a>
+<span class="sourceLineNo">371</span> return familyOffset;<a name="line.371"></a>
+<span class="sourceLineNo">372</span> }<a name="line.372"></a>
+<span class="sourceLineNo">373</span><a name="line.373"></a>
+<span class="sourceLineNo">374</span> public void setFamilyOffset(int familyOffset) {<a name="line.374"></a>
+<span class="sourceLineNo">375</span> this.familyOffset = familyOffset;<a name="line.375"></a>
+<span class="sourceLineNo">376</span> }<a name="line.376"></a>
+<span class="sourceLineNo">377</span><a name="line.377"></a>
+<span class="sourceLineNo">378</span> public int getFamilyLength() {<a name="line.378"></a>
+<span class="sourceLineNo">379</span> return familyLength;<a name="line.379"></a>
+<span class="sourceLineNo">380</span> }<a name="line.380"></a>
+<span class="sourceLineNo">381</span><a name="line.381"></a>
+<span class="sourceLineNo">382</span> public void setFamilyLength(int familyLength) {<a name="line.382"></a>
+<span class="sourceLineNo">383</span> this.familyLength = familyLength;<a name="line.383"></a>
+<span class="sourceLineNo">384</span> }<a name="line.384"></a>
+<span class="sourceLineNo">385</span><a name="line.385"></a>
+<span class="sourceLineNo">386</span> public byte[] getQualifierArray() {<a name="line.386"></a>
+<span class="sourceLineNo">387</span> return qualifier;<a name="line.387"></a>
+<span class="sourceLineNo">388</span> }<a name="line.388"></a>
+<span class="sourceLineNo">389</span><a name="line.389"></a>
+<span class="sourceLineNo">390</span> public void setQualifier(byte[] qualifier) {<a name="line.390"></a>
+<span class="sourceLineNo">391</span> this.qualifier = qualifier;<a name="line.391"></a>
+<span class="sourceLineNo">392</span> }<a name="line.392"></a>
+<span class="sourceLineNo">393</span><a name="line.393"></a>
+<span class="sourceLineNo">394</span> public int getQualifierOffset() {<a name="line.394"></a>
+<span class="sourceLineNo">395</span> return qualifierOffset;<a name="line.395"></a>
+<span class="sourceLineNo">396</span> }<a name="line.396"></a>
+<span class="sourceLineNo">397</span><a name="line.397"></a>
+<span class="sourceLineNo">398</span> public void setQualifierOffset(int qualifierOffset) {<a name="line.398"></a>
+<span class="sourceLineNo">399</span> this.qualifierOffset = qualifierOffset;<a name="line.399"></a>
+<span class="sourceLineNo">400</span> }<a name="line.400"></a>
+<span class="sourceLineNo">401</span><a name="line.401"></a>
+<span class="sourceLineNo">402</span> public int getQualifierLength() {<a name="line.402"></a>
+<span class="sourceLineNo">403</span> return qualifierLength;<a name="line.403"></a>
+<span class="sourceLineNo">404</span> }<a name="line.404"></a>
+<span class="sourceLineNo">405</span><a name="line.405"></a>
+<span class="sourceLineNo">406</span> public void setQualifierLength(int qualifierLength) {<a name="line.406"></a>
+<span class="sourceLineNo">407</span> this.qualifierLength = qualifierLength;<a name="line.407"></a>
+<span class="sourceLineNo">408</span> }<a name="line.408"></a>
+<span class="sourceLineNo">409</span><a name="line.409"></a>
+<span class="sourceLineNo">410</span> public long getTimestamp() {<a name="line.410"></a>
+<span class="sourceLineNo">411</span> return ts;<a name="line.411"></a>
+<span class="sourceLineNo">412</span> }<a name="line.412"></a>
+<span class="sourceLineNo">413</span><a name="line.413"></a>
+<span class="sourceLineNo">414</span> public void setTimestamp(long ts) {<a name="line.414"></a>
+<span class="sourceLineNo">415</span> this.ts = ts;<a name="line.415"></a>
+<span class="sourceLineNo">416</span> }<a name="line.416"></a>
+<span class="sourceLineNo">417</span> }<a name="line.417"></a>
+<span class="sourceLineNo">418</span><a name="line.418"></a>
+<span class="sourceLineNo">419</span> public static class Loader extends Configured implements Tool {<a name="line.419"></a>
<span class="sourceLineNo">420</span><a name="line.420"></a>
-<span class="sourceLineNo">421</span> void createSchema(TableName tableName) throws IOException {<a name="line.421"></a>
-<span class="sourceLineNo">422</span><a name="line.422"></a>
-<span class="sourceLineNo">423</span> try (Connection conn = ConnectionFactory.createConnection(getConf());<a name="line.423"></a>
-<span class="sourceLineNo">424</span> Admin admin = conn.getAdmin()) {<a name="line.424"></a>
-<span class="sourceLineNo">425</span> if (!admin.tableExists(tableName)) {<a name="line.425"></a>
-<span class="sourceLineNo">426</span><a name="line.426"></a>
-<span class="sourceLineNo">427</span> ColumnFamilyDescriptorBuilder contentFamilyBuilder =<a name="line.427"></a>
-<span class="sourceLineNo">428</span> ColumnFamilyDescriptorBuilder.newBuilder(CONTENT_FAMILY_NAME)<a name="line.428"></a>
-<span class="sourceLineNo">429</span> .setDataBlockEncoding(DataBlockEncoding.NONE)<a name="line.429"></a>
-<span class="sourceLineNo">430</span> .setBloomFilterType(BloomType.ROW)<a name="line.430"></a>
-<span class="sourceLineNo">431</span> .setMaxVersions(1000)<a name="line.431"></a>
-<span class="sourceLineNo">432</span> .setBlocksize(256 * 1024)<a name="line.432"></a>
-<span class="sourceLineNo">433</span> ;<a name="line.433"></a>
-<span class="sourceLineNo">434</span><a name="line.434"></a>
-<span class="sourceLineNo">435</span> ColumnFamilyDescriptorBuilder infoFamilyBuilder =<a name="line.435"></a>
-<span class="sourceLineNo">436</span> ColumnFamilyDescriptorBuilder.newBuilder(INFO_FAMILY_NAME)<a name="line.436"></a>
-<span class="sourceLineNo">437</span> .setDataBlockEncoding(DataBlockEncoding.NONE)<a name="line.437"></a>
-<span class="sourceLineNo">438</span> .setBloomFilterType(BloomType.ROWCOL)<a name="line.438"></a>
-<span class="sourceLineNo">439</span> .setMaxVersions(1000)<a name="line.439"></a>
-<span class="sourceLineNo">440</span> .setBlocksize(8 * 1024)<a name="line.440"></a>
-<span class="sourceLineNo">441</span> ;<a name="line.441"></a>
-<span class="sourceLineNo">442</span><a name="line.442"></a>
-<span class="sourceLineNo">443</span> Set<ColumnFamilyDescriptor> families = new HashSet<>();<a name="line.443"></a>
-<span class="sourceLineNo">444</span> families.add(contentFamilyBuilder.build());<a name="line.444"></a>
-<span class="sourceLineNo">445</span> families.add(infoFamilyBuilder.build());<a name="line.445"></a>
-<span class="sourceLineNo">446</span><a name="line.446"></a>
-<span class="sourceLineNo">447</span> TableDescriptor tableDescriptor = TableDescriptorBuilder.newBuilder(tableName)<a name="line.447"></a>
-<span class="sourceLineNo">448</span> .setColumnFamilies(families)<a name="line.448"></a>
-<span class="sourceLineNo">449</span> .build();<a name="line.449"></a>
-<span class="sourceLineNo">450</span><a name="line.450"></a>
-<span class="sourceLineNo">451</span> if (getConf().getBoolean(HBaseTestingUtil.PRESPLIT_TEST_TABLE_KEY,<a name="line.451"></a>
-<span class="sourceLineNo">452</span> HBaseTestingUtil.PRESPLIT_TEST_TABLE)) {<a name="line.452"></a>
-<span class="sourceLineNo">453</span> int numberOfServers = admin.getRegionServers().size();<a name="line.453"></a>
-<span class="sourceLineNo">454</span> if (numberOfServers == 0) {<a name="line.454"></a>
-<span class="sourceLineNo">455</span> throw new IllegalStateException("No live regionservers");<a name="line.455"></a>
-<span class="sourceLineNo">456</span> }<a name="line.456"></a>
-<span class="sourceLineNo">457</span> int regionsPerServer = getConf().getInt(HBaseTestingUtil.REGIONS_PER_SERVER_KEY,<a name="line.457"></a>
-<span class="sourceLineNo">458</span> HBaseTestingUtil.DEFAULT_REGIONS_PER_SERVER);<a name="line.458"></a>
-<span class="sourceLineNo">459</span> int totalNumberOfRegions = numberOfServers * regionsPerServer;<a name="line.459"></a>
-<span class="sourceLineNo">460</span> LOG.info("Creating test table: " + tableDescriptor);<a name="line.460"></a>
-<span class="sourceLineNo">461</span> LOG.info("Number of live regionservers: " + numberOfServers + ", " +<a name="line.461"></a>
-<span class="sourceLineNo">462</span> "pre-splitting table into " + totalNumberOfRegions + " regions " +<a name="line.462"></a>
-<span class="sourceLineNo">463</span> "(default regions per server: " + regionsPerServer + ")");<a name="line.463"></a>
-<span class="sourceLineNo">464</span> byte[][] splits = new RegionSplitter.UniformSplit().split(totalNumberOfRegions);<a name="line.464"></a>
-<span class="sourceLineNo">465</span> admin.createTable(tableDescriptor, splits);<a name="line.465"></a>
-<span class="sourceLineNo">466</span> } else {<a name="line.466"></a>
-<span class="sourceLineNo">467</span> LOG.info("Creating test table: " + tableDescriptor);<a name="line.467"></a>
-<span class="sourceLineNo">468</span> admin.createTable(tableDescriptor);<a name="line.468"></a>
-<span class="sourceLineNo">469</span> }<a name="line.469"></a>
-<span class="sourceLineNo">470</span> }<a name="line.470"></a>
-<span class="sourceLineNo">471</span> } catch (MasterNotRunningException e) {<a name="line.471"></a>
-<span class="sourceLineNo">472</span> LOG.error("Master not running", e);<a name="line.472"></a>
-<span class="sourceLineNo">473</span> throw new IOException(e);<a name="line.473"></a>
-<span class="sourceLineNo">474</span> }<a name="line.474"></a>
-<span class="sourceLineNo">475</span> }<a name="line.475"></a>
-<span class="sourceLineNo">476</span><a name="line.476"></a>
-<span class="sourceLineNo">477</span> int run(Path warcFileInput, Path outputDir)<a name="line.477"></a>
-<span class="sourceLineNo">478</span> throws IOException, ClassNotFoundException, InterruptedException {<a name="line.478"></a>
+<span class="sourceLineNo">421</span> private static final Logger LOG = LoggerFactory.getLogger(Loader.class);<a name="line.421"></a>
+<span class="sourceLineNo">422</span> private static final String USAGE = "Loader <warInputDir | warFileList> <outputDir>";<a name="line.422"></a>
+<span class="sourceLineNo">423</span><a name="line.423"></a>
+<span class="sourceLineNo">424</span> void createSchema(TableName tableName) throws IOException {<a name="line.424"></a>
+<span class="sourceLineNo">425</span><a name="line.425"></a>
+<span class="sourceLineNo">426</span> try (Connection conn = ConnectionFactory.createConnection(getConf());<a name="line.426"></a>
+<span class="sourceLineNo">427</span> Admin admin = conn.getAdmin()) {<a name="line.427"></a>
+<span class="sourceLineNo">428</span> if (!admin.tableExists(tableName)) {<a name="line.428"></a>
+<span class="sourceLineNo">429</span><a name="line.429"></a>
+<span class="sourceLineNo">430</span> ColumnFamilyDescriptorBuilder contentFamilyBuilder =<a name="line.430"></a>
+<span class="sourceLineNo">431</span> ColumnFamilyDescriptorBuilder.newBuilder(CONTENT_FAMILY_NAME)<a name="line.431"></a>
+<span class="sourceLineNo">432</span> .setDataBlockEncoding(DataBlockEncoding.NONE)<a name="line.432"></a>
+<span class="sourceLineNo">433</span> .setBloomFilterType(BloomType.ROW)<a name="line.433"></a>
+<span class="sourceLineNo">434</span> .setMaxVersions(1000)<a name="line.434"></a>
+<span class="sourceLineNo">435</span> .setBlocksize(256 * 1024)<a name="line.435"></a>
+<span class="sourceLineNo">436</span> ;<a name="line.436"></a>
+<span class="sourceLineNo">437</span><a name="line.437"></a>
+<span class="sourceLineNo">438</span> ColumnFamilyDescriptorBuilder infoFamilyBuilder =<a name="line.438"></a>
+<span class="sourceLineNo">439</span> ColumnFamilyDescriptorBuilder.newBuilder(INFO_FAMILY_NAME)<a name="line.439"></a>
+<span class="sourceLineNo">440</span> .setDataBlockEncoding(DataBlockEncoding.NONE)<a name="line.440"></a>
+<span class="sourceLineNo">441</span> .setBloomFilterType(BloomType.ROWCOL)<a name="line.441"></a>
+<span class="sourceLineNo">442</span> .setMaxVersions(1000)<a name="line.442"></a>
+<span class="sourceLineNo">443</span> .setBlocksize(8 * 1024)<a name="line.443"></a>
+<span class="sourceLineNo">444</span> ;<a name="line.444"></a>
+<span class="sourceLineNo">445</span><a name="line.445"></a>
+<span class="sourceLineNo">446</span> Set<ColumnFamilyDescriptor> families = new HashSet<>();<a name="line.446"></a>
+<span class="sourceLineNo">447</span> families.add(contentFamilyBuilder.build());<a name="line.447"></a>
+<span class="sourceLineNo">448</span> families.add(infoFamilyBuilder.build());<a name="line.448"></a>
+<span class="sourceLineNo">449</span><a name="line.449"></a>
+<span class="sourceLineNo">450</span> TableDescriptor tableDescriptor = TableDescriptorBuilder.newBuilder(tableName)<a name="line.450"></a>
+<span class="sourceLineNo">451</span> .setColumnFamilies(families)<a name="line.451"></a>
+<span class="sourceLineNo">452</span> .build();<a name="line.452"></a>
+<span class="sourceLineNo">453</span><a name="line.453"></a>
+<span class="sourceLineNo">454</span> if (getConf().getBoolean(HBaseTestingUtil.PRESPLIT_TEST_TABLE_KEY,<a name="line.454"></a>
+<span class="sourceLineNo">455</span> HBaseTestingUtil.PRESPLIT_TEST_TABLE)) {<a name="line.455"></a>
+<span class="sourceLineNo">456</span> int numberOfServers = admin.getRegionServers().size();<a name="line.456"></a>
+<span class="sourceLineNo">457</span> if (numberOfServers == 0) {<a name="line.457"></a>
+<span class="sourceLineNo">458</span> throw new IllegalStateException("No live regionservers");<a name="line.458"></a>
+<span class="sourceLineNo">459</span> }<a name="line.459"></a>
+<span class="sourceLineNo">460</span> int regionsPerServer = getConf().getInt(HBaseTestingUtil.REGIONS_PER_SERVER_KEY,<a name="line.460"></a>
+<span class="sourceLineNo">461</span> HBaseTestingUtil.DEFAULT_REGIONS_PER_SERVER);<a name="line.461"></a>
+<span class="sourceLineNo">462</span> int totalNumberOfRegions = numberOfServers * regionsPerServer;<a name="line.462"></a>
+<span class="sourceLineNo">463</span> LOG.info("Creating test table: " + tableDescriptor);<a name="line.463"></a>
+<span class="sourceLineNo">464</span> LOG.info("Number of live regionservers: " + numberOfServers + ", " +<a name="line.464"></a>
+<span class="sourceLineNo">465</span> "pre-splitting table into " + totalNumberOfRegions + " regions " +<a name="line.465"></a>
+<span class="sourceLineNo">466</span> "(default regions per server: " + regionsPerServer + ")");<a name="line.466"></a>
+<span class="sourceLineNo">467</span> byte[][] splits = new RegionSplitter.UniformSplit().split(totalNumberOfRegions);<a name="line.467"></a>
+<span class="sourceLineNo">468</span> admin.createTable(tableDescriptor, splits);<a name="line.468"></a>
+<span class="sourceLineNo">469</span> } else {<a name="line.469"></a>
+<span class="sourceLineNo">470</span> LOG.info("Creating test table: " + tableDescriptor);<a name="line.470"></a>
+<span class="sourceLineNo">471</span> admin.createTable(tableDescriptor);<a name="line.471"></a>
+<span class="sourceLineNo">472</span> }<a name="line.472"></a>
+<span class="sourceLineNo">473</span> }<a name="line.473"></a>
+<span class="sourceLineNo">474</span> } catch (MasterNotRunningException e) {<a name="line.474"></a>
+<span class="sourceLineNo">475</span> LOG.error("Master not running", e);<a name="line.475"></a>
+<span class="sourceLineNo">476</span> throw new IOException(e);<a name="line.476"></a>
+<span class="sourceLineNo">477</span> }<a name="line.477"></a>
+<span class="sourceLineNo">478</span> }<a name="line.478"></a>
<span class="sourceLineNo">479</span><a name="line.479"></a>
-<span class="sourceLineNo">480</span> createSchema(getTablename(getConf()));<a name="line.480"></a>
-<span class="sourceLineNo">481</span><a name="line.481"></a>
-<span class="sourceLineNo">482</span> Job job = Job.getInstance(getConf());<a name="line.482"></a>
-<span class="sourceLineNo">483</span> job.setJobName(Loader.class.getName());<a name="line.483"></a>
-<span class="sourceLineNo">484</span> job.setNumReduceTasks(0);<a name="line.484"></a>
-<span class="sourceLineNo">485</span> job.setJarByClass(getClass());<a name="line.485"></a>
-<span class="sourceLineNo">486</span> job.setMapperClass(LoaderMapper.class);<a name="line.486"></a>
-<span class="sourceLineNo">487</span> job.setInputFormatClass(WARCInputFormat.class);<a name="line.487"></a>
-<span class="sourceLineNo">488</span> FileSystem fs = FileSystem.get(warcFileInput.toUri(), getConf());<a name="line.488"></a>
-<span class="sourceLineNo">489</span> if (fs.getFileStatus(warcFileInput).isDirectory()) {<a name="line.489"></a>
-<span class="sourceLineNo">490</span> LOG.info("Using directory as WARC input path: " + warcFileInput);<a name="line.490"></a>
-<span class="sourceLineNo">491</span> FileInputFormat.setInputPaths(job, warcFileInput);<a name="line.491"></a>
-<span class="sourceLineNo">492</span> } else {<a name="line.492"></a>
-<span class="sourceLineNo">493</span> LOG.info("Getting WARC input paths from file: " + warcFileInput);<a name="line.493"></a>
-<span class="sourceLineNo">494</span> List<Path> paths = new LinkedList<Path>();<a name="line.494"></a>
-<span class="sourceLineNo">495</span> try (FSDataInputStream is = fs.open(warcFileInput)) {<a name="line.495"></a>
-<span class="sourceLineNo">496</span> InputStreamReader reader;<a name="line.496"></a>
-<span class="sourceLineNo">497</span> if (warcFileInput.getName().toLowerCase().endsWith(".gz")) {<a name="line.497"></a>
-<span class="sourceLineNo">498</span> reader = new InputStreamReader(new GZIPInputStream(is));<a name="line.498"></a>
-<span class="sourceLineNo">499</span> } else {<a name="line.499"></a>
-<span class="sourceLineNo">500</span> reader = new InputStreamReader(is);<a name="line.500"></a>
-<span class="sourceLineNo">501</span> }<a name="line.501"></a>
-<span class="sourceLineNo">502</span> try (BufferedReader br = new BufferedReader(reader)) {<a name="line.502"></a>
-<span class="sourceLineNo">503</span> String line;<a name="line.503"></a>
-<span class="sourceLineNo">504</span> while ((line = br.readLine()) != null) {<a name="line.504"></a>
-<span class="sourceLineNo">505</span> paths.add(new Path(line));<a name="line.505"></a>
-<span class="sourceLineNo">506</span> }<a name="line.506"></a>
-<span class="sourceLineNo">507</span> }<a name="line.507"></a>
-<span class="sourceLineNo">508</span> }<a name="line.508"></a>
-<span class="sourceLineNo">509</span> LOG.info("Read " + paths.size() + " WARC input paths from " + warcFileInput);<a name="line.509"></a>
-<span class="sourceLineNo">510</span> FileInputFormat.setInputPaths(job, paths.toArray(new Path[paths.size()]));<a name="line.510"></a>
-<span class="sourceLineNo">511</span> }<a name="line.511"></a>
-<span class="sourceLineNo">512</span> job.setOutputFormatClass(SequenceFileOutputFormat.class);<a name="line.512"></a>
-<span class="sourceLineNo">513</span> SequenceFileOutputFormat.setOutputPath(job, outputDir);<a name="line.513"></a>
-<span class="sourceLineNo">514</span> SequenceFileOutputFormat.setOutputCompressionType(job, CompressionType.BLOCK);<a name="line.514"></a>
-<span class="sourceLineNo">515</span> job.setOutputKeyClass(HBaseKeyWritable.class);<a name="line.515"></a>
-<span class="sourceLineNo">516</span> job.setOutputValueClass(BytesWritable.class);<a name="line.516"></a>
-<span class="sourceLineNo">517</span> TableMapReduceUtil.addDependencyJars(job);<a name="line.517"></a>
-<span class="sourceLineNo">518</span><a name="line.518"></a>
-<span class="sourceLineNo">519</span> LOG.info("Submitting job." +<a name="line.519"></a>
-<span class="sourceLineNo">520</span> " This will take time proportional to the number of input files, please be patient.");<a name="line.520"></a>
-<span class="sourceLineNo">521</span> boolean success = job.waitForCompletion(true);<a name="line.521"></a>
-<span class="sourceLineNo">522</span> if (!success) {<a name="line.522"></a>
-<span class="sourceLineNo">523</span> LOG.error("Failure during job " + job.getJobID());<a name="line.523"></a>
-<span class="sourceLineNo">524</span> }<a name="line.524"></a>
-<span class="sourceLineNo">525</span> return success ? 0 : 1;<a name="line.525"></a>
-<span class="sourceLineNo">526</span> }<a name="line.526"></a>
-<span class="sourceLineNo">527</span><a name="line.527"></a>
-<span class="sourceLineNo">528</span> @Override<a name="line.528"></a>
-<span class="sourceLineNo">529</span> public int run(String[] args) throws Exception {<a name="line.529"></a>
-<span class="sourceLineNo">530</span> if (args.length < 2) {<a name="line.530"></a>
-<span class="sourceLineNo">531</span> System.err.println(USAGE);<a name="line.531"></a>
-<span class="sourceLineNo">532</span> return 1;<a name="line.532"></a>
-<span class="sourceLineNo">533</span> }<a name="line.533"></a>
-<span class="sourceLineNo">534</span> try {<a name="line.534"></a>
-<span class="sourceLineNo">535</span> Path warcFileInput = new Path(args[0]);<a name="line.535"></a>
-<span class="sourceLineNo">536</span> Path outputDir = new Path(args[1]);<a name="line.536"></a>
-<span class="sourceLineNo">537</span> return run(warcFileInput, outputDir);<a name="line.537"></a>
-<span class="sourceLineNo">538</span> } catch (NumberFormatException e) {<a name="line.538"></a>
-<span class="sourceLineNo">539</span> System.err.println("Parsing loader arguments failed: " + e.getMessage());<a name="line.539"></a>
-<span class="sourceLineNo">540</span> System.err.println(USAGE);<a name="line.540"></a>
-<span class="sourceLineNo">541</span> return 1;<a name="line.541"></a>
-<span class="sourceLineNo">542</span> }<a name="line.542"></a>
-<span class="sourceLineNo">543</span> }<a name="line.543"></a>
-<span class="sourceLineNo">544</span><a name="line.544"></a>
-<span class="sourceLineNo">545</span> public static void main(String[] args) throws Exception {<a name="line.545"></a>
-<span class="sourceLineNo">546</span> System.exit(ToolRunner.run(HBaseConfiguration.create(), new Loader(), args));<a name="line.546"></a>
-<span class="sourceLineNo">547</span> }<a name="line.547"></a>
-<span class="sourceLineNo">548</span><a name="line.548"></a>
-<span class="sourceLineNo">549</span> public static class LoaderMapper<a name="line.549"></a>
-<span class="sourceLineNo">550</span> extends Mapper<LongWritable, WARCWritable, HBaseKeyWritable, BytesWritable> {<a name="line.550"></a>
-<span class="sourceLineNo">551</span><a name="line.551"></a>
-<span class="sourceLineNo">552</span> Configuration conf;<a name="line.552"></a>
-<span class="sourceLineNo">553</span> Connection conn;<a name="line.553"></a>
-<span class="sourceLineNo">554</span> Table table;<a name="line.554"></a>
-<span class="sourceLineNo">555</span><a name="line.555"></a>
-<span class="sourceLineNo">556</span> @Override<a name="line.556"></a>
-<span class="sourceLineNo">557</span> protected void setup(Context context) throws IOException, InterruptedException {<a name="line.557"></a>
-<span class="sourceLineNo">558</span> conn = ConnectionFactory.createConnection(context.getConfiguration());<a name="line.558"></a>
-<span class="sourceLineNo">559</span> table = conn.getTable(getTablename(conn.getConfiguration()));<a name="line.559"></a>
-<span class="sourceLineNo">560</span> }<a name="line.560"></a>
-<span class="sourceLineNo">561</span><a name="line.561"></a>
-<span class="sourceLineNo">562</span> @Override<a name="line.562"></a>
-<span class="sourceLineNo">563</span> protected void cleanup(Context context) throws IOException, InterruptedException {<a name="line.563"></a>
-<span class="sourceLineNo">564</span> try {<a name="line.564"></a>
-<span class="sourceLineNo">565</span> table.close();<a name="line.565"></a>
-<span class="sourceLineNo">566</span> } catch (Exception e) {<a name="line.566"></a>
-<span class="sourceLineNo">567</span> LOG.warn("Exception closing Table", e);<a name="line.567"></a>
-<span class="sourceLineNo">568</span> }<a name="line.568"></a>
-<span class="sourceLineNo">569</span> try {<a name="line.569"></a>
-<span class="sourceLineNo">570</span> conn.close();<a name="line.570"></a>
-<span class="sourceLineNo">571</span> } catch (Exception e) {<a name="line.571"></a>
-<span class="sourceLineNo">572</span> LOG.warn("Exception closing Connection", e);<a name="line.572"></a>
-<span class="sourceLineNo">573</span> }<a name="line.573"></a>
-<span class="sourceLineNo">574</span> }<a name="line.574"></a>
-<span class="sourceLineNo">575</span><a name="line.575"></a>
-<span class="sourceLineNo">576</span> @Override<a name="line.576"></a>
-<span class="sourceLineNo">577</span> protected void map(LongWritable key, WARCWritable value, Context output)<a name="line.577"></a>
-<span class="sourceLineNo">578</span> throws IOException, InterruptedException {<a name="line.578"></a>
-<span class="sourceLineNo">579</span> WARCRecord.Header warcHeader = value.getRecord().getHeader();<a name="line.579"></a>
-<span class="sourceLineNo">580</span> String recordID = warcHeader.getRecordID();<a name="line.580"></a>
-<span class="sourceLineNo">581</span> String targetURI = warcHeader.getTargetURI();<a name="line.581"></a>
-<span class="sourceLineNo">582</span> if (warcHeader.getRecordType().equals("response") && targetURI != null) {<a name="line.582"></a>
-<span class="sourceLineNo">583</span> String contentType = warcHeader.getField("WARC-Identified-Payload-Type");<a name="line.583"></a>
-<span class="sourceLineNo">584</span> if (contentType != null) {<a name="line.584"></a>
-<span class="sourceLineNo">585</span> LOG.debug("Processing record id=" + recordID + ", targetURI=\"" + targetURI + "\"");<a name="line.585"></a>
-<span class="sourceLineNo">586</span> long now = EnvironmentEdgeManager.currentTime();<a name="line.586"></a>
-<span class="sourceLineNo">587</span><a name="line.587"></a>
-<span class="sourceLineNo">588</span> // Make row key<a name="line.588"></a>
-<span class="sourceLineNo">589</span><a name="line.589"></a>
-<span class="sourceLineNo">590</span> byte[] rowKey;<a name="line.590"></a>
-<span class="sourceLineNo">591</span> try {<a name="line.591"></a>
-<span class="sourceLineNo">592</span> rowKey = rowKeyFromTargetURI(targetURI);<a name="line.592"></a>
-<span class="sourceLineNo">593</span> } catch (IllegalArgumentException e) {<a name="line.593"></a>
-<span class="sourceLineNo">594</span> LOG.debug("Could not make a row key for record " + recordID + ", ignoring", e);<a name="line.594"></a>
-<span class="sourceLineNo">595</span> return;<a name="line.595"></a>
-<span class="sourceLineNo">596</span> } catch (URISyntaxException e) {<a name="line.596"></a>
-<span class="sourceLineNo">597</span> LOG.warn("Could not parse URI \"" + targetURI + "\" for record " + recordID +<a name="line.597"></a>
-<span class="sourceLineNo">598</span> ", ignoring");<a name="line.598"></a>
-<span class="sourceLineNo">599</span> return;<a name="line.599"></a>
-<span class="sourceLineNo">600</span> }<a name="line.600"></a>
-<span class="sourceLineNo">601</span><a name="line.601"></a>
-<span class="sourceLineNo">602</span> // Get the content and calculate the CRC64<a name="line.602"></a>
-<span class="sourceLineNo">603</span><a name="line.603"></a>
-<span class="sourceLineNo">604</span> byte[] content = value.getRecord().getContent();<a name="line.604"></a>
-<span class="sourceLineNo">605</span> CRC64 crc = new CRC64();<a name="line.605"></a>
-<span class="sourceLineNo">606</span> crc.update(content);<a name="line.606"></a>
-<span class="sourceLineNo">607</span> long crc64 = crc.getValue();<a name="line.607"></a>
-<span class="sourceLineNo">608</span><a name="line.608"></a>
-<span class="sourceLineNo">609</span> // Store to HBase<a name="line.609"></a>
-<span class="sourceLineNo">610</span><a name="line.610"></a>
-<span class="sourceLineNo">611</span> Put put = new Put(rowKey);<a name="line.611"></a>
-<span class="sourceLineNo">612</span> put.addColumn(CONTENT_FAMILY_NAME, CONTENT_QUALIFIER, now, content);<a name="line.612"></a>
-<span class="sourceLineNo">613</span> put.addColumn(INFO_FAMILY_NAME, CONTENT_LENGTH_QUALIFIER, now,<a name="line.613"></a>
-<span class="sourceLineNo">614</span> Bytes.toBytes(content.length));<a name="line.614"></a>
-<span class="sourceLineNo">615</span> put.addColumn(INFO_FAMILY_NAME, CONTENT_TYPE_QUALIFIER, now,<a name="line.615"></a>
-<span class="sourceLineNo">616</span> Bytes.toBytes(contentType));<a name="line.616"></a>
-<span class="sourceLineNo">617</span> put.addColumn(INFO_FAMILY_NAME, CRC_QUALIFIER, now, Bytes.toBytes(crc64));<a name="line.617"></a>
-<span class="sourceLineNo">618</span> put.addColumn(INFO_FAMILY_NAME, RECORD_ID_QUALIFIER, now, Bytes.toBytes(recordID));<a name="line.618"></a>
-<span class="sourceLineNo">619</span> put.addColumn(INFO_FAMILY_NAME, TARGET_URI_QUALIFIER, now, Bytes.toBytes(targetURI));<a name="line.619"></a>
-<span class="sourceLineNo">620</span> put.addColumn(INFO_FAMILY_NAME, DATE_QUALIFIER, now,<a name="line.620"></a>
-<span class="sourceLineNo">621</span> Bytes.toBytes(warcHeader.getDateString()));<a name="line.621"></a>
-<span class="sourceLineNo">622</span> String ipAddr = warcHeader.getField("WARC-IP-Address");<a name="line.622"></a>
-<span class="sourceLineNo">623</span> if (ipAddr != null) {<a name="line.623"></a>
-<span class="sourceLineNo">624</span> put.addColumn(INFO_FAMILY_NAME, IP_ADDRESS_QUALIFIER, now, Bytes.toBytes(ipAddr));<a name="line.624"></a>
-<span class="sourceLineNo">625</span> }<a name="line.625"></a>
-<span class="sourceLineNo">626</span> table.put(put);<a name="line.626"></a>
-<span class="sourceLineNo">627</span><a name="line.627"></a>
-<span class="sourceLineNo">628</span> // Write records out for later verification, one per HBase field except for the<a name="line.628"></a>
-<span class="sourceLineNo">629</span> // content record, which will be verified by CRC64.<a name="line.629"></a>
+<span class="sourceLineNo">480</span> int run(Path warcFileInput, Path outputDir)<a name="line.480"></a>
+<span class="sourceLineNo">481</span> throws IOException, ClassNotFoundException, InterruptedException {<a name="line.481"></a>
+<span class="sourceLineNo">482</span><a name="line.482"></a>
+<span class="sourceLineNo">483</span> createSchema(getTablename(getConf()));<a name="line.483"></a>
+<span class="sourceLineNo">484</span><a name="line.484"></a>
+<span class="sourceLineNo">485</span> Job job = Job.getInstance(getConf());<a name="line.485"></a>
+<span class="sourceLineNo">486</span> job.setJobName(Loader.class.getName());<a name="line.486"></a>
+<span class="sourceLineNo">487</span> job.setNumReduceTasks(0);<a name="line.487"></a>
+<span class="sourceLineNo">488</span> job.setJarByClass(getClass());<a name="line.488"></a>
+<span class="sourceLineNo">489</span> job.setMapperClass(LoaderMapper.class);<a name="line.489"></a>
+<span class="sourceLineNo">490</span> job.setInputFormatClass(WARCInputFormat.class);<a name="line.490"></a>
+<span class="sourceLineNo">491</span> FileSystem fs = FileSystem.get(warcFileInput.toUri(), getConf());<a name="line.491"></a>
+<span class="sourceLineNo">492</span> if (fs.getFileStatus(warcFileInput).isDirectory()) {<a name="line.492"></a>
+<span class="sourceLineNo">493</span> LOG.info("Using directory as WARC input path: " + warcFileInput);<a name="line.493"></a>
+<span class="sourceLineNo">494</span> FileInputFormat.setInputPaths(job, warcFileInput);<a name="line.494"></a>
+<span class="sourceLineNo">495</span> } else {<a name="line.495"></a>
+<span class="sourceLineNo">496</span> LOG.info("Getting WARC input paths from file: " + warcFileInput);<a name="line.496"></a>
+<span class="sourceLineNo">497</span> List<Path> paths = new LinkedList<Path>();<a name="line.497"></a>
+<span class="sourceLineNo">498</span> try (FSDataInputStream is = fs.open(warcFileInput)) {<a name="line.498"></a>
+<span class="sourceLineNo">499</span> InputStreamReader reader;<a name="line.499"></a>
+<span class="sourceLineNo">500</span> if (warcFileInput.getName().toLowerCase().endsWith(".gz")) {<a name="line.500"></a>
+<span class="sourceLineNo">501</span> reader = new InputStreamReader(new GZIPInputStream(is));<a name="line.501"></a>
+<span class="sourceLineNo">502</span> } else {<a name="line.502"></a>
+<span class="sourceLineNo">503</span> reader = new InputStreamReader(is);<a name="line.503"></a>
+<span class="sourceLineNo">504</span> }<a name="line.504"></a>
+<span class="sourceLineNo">505</span> try (BufferedReader br = new BufferedReader(reader)) {<a name="line.505"></a>
+<span class="sourceLineNo">506</span> String line;<a name="line.506"></a>
+<span class="sourceLineNo">507</span> while ((line = br.readLine()) != null) {<a name="line.507"></a>
+<span class="sourceLineNo">508</span> paths.add(new Path(line));<a name="line.508"></a>
+<span class="sourceLineNo">509</span> }<a name="line.509"></a>
+<span class="sourceLineNo">510</span> }<a name="line.510"></a>
+<span class="sourceLineNo">511</span> }<a name="line.511"></a>
+<span class="sourceLineNo">512</span> LOG.info("Read " + paths.size() + " WARC input paths from " + warcFileInput);<a name="line.512"></a>
+<span class="sourceLineNo">513</span> FileInputFormat.setInputPaths(job, paths.toArray(new Path[paths.size()]));<a name="line.513"></a>
+<span class="sourceLineNo">514</span> }<a name="line.514"></a>
+<span class="sourceLineNo">515</span> job.setOutputFormatClass(SequenceFileOutputFormat.class);<a name="line.515"></a>
+<span class="sourceLineNo">516</span> SequenceFileOutputFormat.setOutputPath(job, outputDir);<a name="line.516"></a>
+<span class="sourceLineNo">517</span> SequenceFileOutputFormat.setOutputCompressionType(job, CompressionType.BLOCK);<a name="line.517"></a>
+<span class="sourceLineNo">518</span> job.setOutputKeyClass(HBaseKeyWritable.class);<a name="line.518"></a>
+<span class="sourceLineNo">519</span> job.setOutputValueClass(BytesWritable.class);<a name="line.519"></a>
+<span class="sourceLineNo">520</span> TableMapReduceUtil.addDependencyJars(job);<a name="line.520"></a>
+<span class="sourceLineNo">521</span><a name="line.521"></a>
+<span class="sourceLineNo">522</span> boolean success = job.waitForCompletion(true);<a name="line.522"></a>
+<span class="sourceLineNo">523</span> if (!success) {<a name="line.523"></a>
+<span class="sourceLineNo">524</span> LOG.error("Failure during job " + job.getJobID());<a name="line.524"></a>
+<span class="sourceLineNo">525</span> }<a name="line.525"></a>
+<span class="sourceLineNo">526</span> return success ? 0 : 1;<a name="line.526"></a>
+<span class="sourceLineNo">527</span> }<a name="line.527"></a>
+<span class="sourceLineNo">528</span><a name="line.528"></a>
+<span class="sourceLineNo">529</span> @Override<a name="line.529"></a>
+<span class="sourceLineNo">530</span> public int run(String[] args) throws Exception {<a name="line.530"></a>
+<span class="sourceLineNo">531</span> if (args.length < 2) {<a name="line.531"></a>
+<span class="sourceLineNo">532</span> System.err.println(USAGE);<a name="line.532"></a>
+<span class="sourceLineNo">533</span> return 1;<a name="line.533"></a>
+<span class="sourceLineNo">534</span> }<a name="line.534"></a>
+<span class="sourceLineNo">535</span> try {<a name="line.535"></a>
+<span class="sourceLineNo">536</span> Path warcFileInput = new Path(args[0]);<a name="line.536"></a>
+<span class="sourceLineNo">537</span> Path outputDir = new Path(args[1]);<a name="line.537"></a>
+<span class="sourceLineNo">538</span> return run(warcFileInput, outputDir);<a name="line.538"></a>
+<span class="sourceLineNo">539</span> } catch (NumberFormatException e) {<a name="line.539"></a>
+<span class="sourceLineNo">540</span> System.err.println("Parsing loader arguments failed: " + e.getMessage());<a name="line.540"></a>
+<span class="sourceLineNo">541</span> System.err.println(USAGE);<a name="line.541"></a>
+<span class="sourceLineNo">542</span> return 1;<a name="line.542"></a>
+<span class="sourceLineNo">543</span> }<a name="line.543"></a>
+<span class="sourceLineNo">544</span> }<a name="line.544"></a>
+<span class="sourceLineNo">545</span><a name="line.545"></a>
+<span class="sourceLineNo">546</span> public static void main(String[] args) throws Exception {<a name="line.546"></a>
+<span class="sourceLineNo">547</span> System.exit(ToolRunner.run(HBaseConfiguration.create(), new Loader(), args));<a name="line.547"></a>
+<span class="sourceLineNo">548</span> }<a name="line.548"></a>
+<span class="sourceLineNo">549</span><a name="line.549"></a>
+<span class="sourceLineNo">550</span> public static class LoaderMapper<a name="line.550"></a>
+<span class="sourceLineNo">551</span> extends Mapper<LongWritable, WARCWritable, HBaseKeyWritable, BytesWritable> {<a name="line.551"></a>
+<span class="sourceLineNo">552</span><a name="line.552"></a>
+<span class="sourceLineNo">553</span> Configuration conf;<a name="line.553"></a>
+<span class="sourceLineNo">554</span> Connection conn;<a name="line.554"></a>
+<span class="sourceLineNo">555</span> BufferedMutator mutator;<a name="line.555"></a>
+<span class="sourceLineNo">556</span><a name="line.556"></a>
+<span class="sourceLineNo">557</span> @Override<a name="line.557"></a>
+<span class="sourceLineNo">558</span> protected void setup(Context context) throws IOException, InterruptedException {<a name="line.558"></a>
+<span class="sourceLineNo">559</span> conf = context.getConfiguration();<a name="line.559"></a>
+<span class="sourceLineNo">560</span> conn = ConnectionFactory.createConnection(conf);<a name="line.560"></a>
+<span class="sourceLineNo">561</span> mutator = conn.getBufferedMutator(getTablename(conf));<a name="line.561"></a>
+<span class="sourceLineNo">562</span> mutator.setWriteBufferPeriodicFlush(10 * 1000); // default is 1 sec, increase to 10<a name="line.562"></a>
+<span class="sourceLineNo">563</span> }<a name="line.563"></a>
+<span class="sourceLineNo">564</span><a name="line.564"></a>
+<span class="sourceLineNo">565</span> @Override<a name="line.565"></a>
+<span class="sourceLineNo">566</span> protected void cleanup(Context context) throws IOException, InterruptedException {<a name="line.566"></a>
+<span class="sourceLineNo">567</span> try {<a name="line.567"></a>
+<span class="sourceLineNo">568</span> mutator.close();<a name="line.568"></a>
+<span class="sourceLineNo">569</span> } catch (Exception e) {<a name="line.569"></a>
+<span class="sourceLineNo">570</span> LOG.warn("Exception closing Table", e);<a name="line.570"></a>
+<span class="sourceLineNo">571</span> }<a name="line.571"></a>
+<span class="sourceLineNo">572</span> try {<a name="line.572"></a>
+<span class="sourceLineNo">573</span> conn.close();<a name="line.573"></a>
+<span class="sourceLineNo">574</span> } catch (Exception e) {<a name="line.574"></a>
+<span class="sourceLineNo">575</span> LOG.warn("Exception closing Connection", e);<a name="line.575"></a>
+<span class="sourceLineNo">576</span> }<a name="line.576"></a>
+<span class="sourceLineNo">577</span> }<a name="line.577"></a>
+<span class="sourceLineNo">578</span><a name="line.578"></a>
+<span class="sourceLineNo">579</span> @Override<a name="line.579"></a>
+<span class="sourceLineNo">580</span> protected void map(LongWritable key, WARCWritable value, Context output)<a name="line.580"></a>
+<span class="sourceLineNo">581</span> throws IOException, InterruptedException {<a name="line.581"></a>
+<span class="sourceLineNo">582</span> WARCRecord.Header warcHeader = value.getRecord().getHeader();<a name="line.582"></a>
+<span class="sourceLineNo">583</span> String recordID = warcHeader.getRecordID();<a name="line.583"></a>
+<span class="sourceLineNo">584</span> String targetURI = warcHeader.getTargetURI();<a name="line.584"></a>
+<span class="sourceLineNo">585</span> if (warcHeader.getRecordType().equals("response") && targetURI != null) {<a name="line.585"></a>
+<span class="sourceLineNo">586</span> String contentType = warcHeader.getField("WARC-Identified-Payload-Type");<a name="line.586"></a>
+<span class="sourceLineNo">587</span> if (contentType != null) {<a name="line.587"></a>
+<span class="sourceLineNo">588</span> LOG.info("Processing uri=\"" + targetURI + "\", id=" + recordID);<a name="line.588"></a>
+<span class="sourceLineNo">589</span> long now = EnvironmentEdgeManager.currentTime();<a name="line.589"></a>
+<span class="sourceLineNo">590</span><a name="line.590"></a>
+<span class="sourceLineNo">591</span> // Make row key<a name="line.591"></a>
+<span class="sourceLineNo">592</span><a name="line.592"></a>
+<span class="sourceLineNo">593</span> byte[] rowKey;<a name="line.593"></a>
+<span class="sourceLineNo">594</span> try {<a name="line.594"></a>
+<span class="sourceLineNo">595</span> rowKey = rowKeyFromTargetURI(targetURI);<a name="line.595"></a>
+<span class="sourceLineNo">596</span> } catch (IllegalArgumentException e) {<a name="line.596"></a>
+<span class="sourceLineNo">597</span> LOG.debug("Could not make a row key for record " + recordID + ", ignoring", e);<a name="line.597"></a>
+<span class="sourceLineNo">598</span> return;<a name="line.598"></a>
+<span class="sourceLineNo">599</span> } catch (URISyntaxException e) {<a name="line.599"></a>
+<span class="sourceLineNo">600</span> LOG.warn("Could not parse URI \"" + targetURI + "\" for record " + recordID +<a name="line.600"></a>
+<span class="sourceLineNo">601</span> ", ignoring");<a name="line.601"></a>
+<span class="sourceLineNo">602</span> return;<a name="line.602"></a>
+<span class="sourceLineNo">603</span> }<a name="line.603"></a>
+<span class="sourceLineNo">604</span><a name="line.604"></a>
+<span class="sourceLineNo">605</span> // Get the content and calculate the CRC64<a name="line.605"></a>
+<span class="sourceLineNo">606</span><a name="line.606"></a>
+<span class="sourceLineNo">607</span> byte[] content = value.getRecord().getContent();<a name="line.607"></a>
+<span class="sourceLineNo">608</span> CRC64 crc = new CRC64();<a name="line.608"></a>
+<span class="sourceLineNo">609</span> crc.update(content);<a name="line.609"></a>
+<span class="sourceLineNo">610</span> long crc64 = crc.getValue();<a name="line.610"></a>
+<span class="sourceLineNo">611</span><a name="line.611"></a>
+<span class="sourceLineNo">612</span> // Store to HBase<a name="line.612"></a>
+<span class="sourceLineNo">613</span><a name="line.613"></a>
+<span class="sourceLineNo">614</span> Put put = new Put(rowKey);<a name="line.614"></a>
+<span class="sourceLineNo">615</span> put.addColumn(CONTENT_FAMILY_NAME, CONTENT_QUALIFIER, now, content);<a name="line.615"></a>
+<span class="sourceLineNo">616</span> put.addColumn(INFO_FAMILY_NAME, CONTENT_LENGTH_QUALIFIER, now,<a name="line.616"></a>
+<span class="sourceLineNo">617</span> Bytes.toBytes(content.length));<a name="line.617"></a>
+<span class="sourceLineNo">618</span> put.addColumn(INFO_FAMILY_NAME, CONTENT_TYPE_QUALIFIER, now,<a name="line.618"></a>
+<span class="sourceLineNo">619</span> Bytes.toBytes(contentType));<a name="line.619"></a>
+<span class="sourceLineNo">620</span> put.addColumn(INFO_FAMILY_NAME, CRC_QUALIFIER, now, Bytes.toBytes(crc64));<a name="line.620"></a>
+<span class="sourceLineNo">621</span> put.addColumn(INFO_FAMILY_NAME, RECORD_ID_QUALIFIER, now, Bytes.toBytes(recordID));<a name="line.621"></a>
+<span class="sourceLineNo">622</span> put.addColumn(INFO_FAMILY_NAME, TARGET_URI_QUALIFIER, now, Bytes.toBytes(targetURI));<a name="line.622"></a>
+<span class="sourceLineNo">623</span> put.addColumn(INFO_FAMILY_NAME, DATE_QUALIFIER, now,<a name="line.623"></a>
+<span class="sourceLineNo">624</span> Bytes.toBytes(warcHeader.getDateString()));<a name="line.624"></a>
+<span class="sourceLineNo">625</span> String ipAddr = warcHeader.getField("WARC-IP-Address");<a name="line.625"></a>
+<span class="sourceLineNo">626</span> if (ipAddr != null) {<a name="line.626"></a>
+<span class="sourceLineNo">627</span> put.addColumn(INFO_FAMILY_NAME, IP_ADDRESS_QUALIFIER, now, Bytes.toBytes(ipAddr));<a name="line.627"></a>
+<span class="sourceLineNo">628</span> }<a name="line.628"></a>
+<span class="sourceLineNo">629</span> mutator.mutate(put);<a name="line.629"></a>
<span class="sourceLineNo">630</span><a name="line.630"></a>
-<span class="sourceLineNo">631</span> output.write(new HBaseKeyWritable(rowKey, INFO_FAMILY_NAME, CRC_QUALIFIER, now),<a name="line.631"></a>
-<span class="sourceLineNo">632</span> new BytesWritable(Bytes.toBytes(crc64)));<a name="line.632"></a>
-<span class="sourceLineNo">633</span> output.write(new HBaseKeyWritable(rowKey, INFO_FAMILY_NAME, CONTENT_LENGTH_QUALIFIER,<a name="line.633"></a>
-<span class="sourceLineNo">634</span> now), new BytesWritable(Bytes.toBytes(content.length)));<a name="line.634"></a>
-<span class="sourceLineNo">635</span> output.write(new HBaseKeyWritable(rowKey, INFO_FAMILY_NAME, CONTENT_TYPE_QUALIFIER,<a name="line.635"></a>
-<span class="sourceLineNo">636</span> now), new BytesWritable(Bytes.toBytes(contentType)));<a name="line.636"></a>
-<span class="sourceLineNo">637</span> output.write(new HBaseKeyWritable(rowKey, INFO_FAMILY_NAME, RECORD_ID_QUALIFIER,<a name="line.637"></a>
-<span class="sourceLineNo">638</span> now), new BytesWritable(Bytes.toBytes(recordID)));<a name="line.638"></a>
-<span class="sourceLineNo">639</span> output.write(new HBaseKeyWritable(rowKey, INFO_FAMILY_NAME, TARGET_URI_QUALIFIER,<a name="line.639"></a>
-<span class="sourceLineNo">640</span> now), new BytesWritable(Bytes.toBytes(targetURI)));<a name="line.640"></a>
-<span class="sourceLineNo">641</span> output.write(new HBaseKeyWritable(rowKey, INFO_FAMILY_NAME, DATE_QUALIFIER, now),<a name="line.641"></a>
-<span class="sourceLineNo">642</span> new BytesWritable(Bytes.toBytes(warcHeader.getDateString())));<a name="line.642"></a>
-<span class="sourceLineNo">643</span> if (ipAddr != null) {<a name="line.643"></a>
-<span class="sourceLineNo">644</span> output.write(new HBaseKeyWritable(rowKey, INFO_FAMILY_NAME, IP_ADDRESS_QUALIFIER,<a name="line.644"></a>
-<span class="sourceLineNo">645</span> now), new BytesWritable(Bytes.toBytes(ipAddr)));<a name="line.645"></a>
-<span class="sourceLineNo">646</span> }<a name="line.646"></a>
-<span class="sourceLineNo">647</span> }<a name="line.647"></a>
-<span class="sourceLineNo">648</span> }<a name="line.648"></a>
-<span class="sourceLineNo">649</span> }<a name="line.649"></a>
-<span class="sourceLineNo">650</span><a name="line.650"></a>
-<span class="sourceLineNo">651</span> private byte[] rowKeyFromTargetURI(String targetUri)<a name="line.651"></a>
-<span class="sourceLineNo">652</span> throws URISyntaxException, IllegalArgumentException {<a name="line.652"></a>
-<span class="sourceLineNo">653</span> URI uri = new URI(targetUri);<a name="line.653"></a>
-<span class="sourceLineNo">654</span> StringBuffer sb = new StringBuffer();<a name="line.654"></a>
-<span class="sourceLineNo">655</span> // Ignore the scheme<a name="line.655"></a>
-<span class="sourceLineNo">656</span> // Reverse the components of the hostname<a name="line.656"></a>
-<span class="sourceLineNo">657</span> if (uri.getHost() != null) {<a name="line.657"></a>
-<span class="sourceLineNo">658</span> String[] hostComponents = uri.getHost().split("\\.");<a name="line.658"></a>
-<span class="sourceLineNo">659</span> for (int i = hostComponents.length - 1; i >= 0; i--) {<a name="line.659"></a>
-<span class="sourceLineNo">660</span> sb.append(hostComponents[i]);<a name="line.660"></a>
-<span class="sourceLineNo">661</span> if (i != 0) {<a name="line.661"></a>
-<span class="sourceLineNo">662</span> sb.append('.');<a name="line.662"></a>
-<span class="sourceLineNo">663</span> }<a name="line.663"></a>
-<span class="sourceLineNo">664</span> }<a name="line.664"></a>
-<span class="sourceLineNo">665</span> } else {<a name="line.665"></a>
-<span class="sourceLineNo">666</span> throw new IllegalArgumentException("URI is missing host component");<a name="line.666"></a>
-<span class="sourceLineNo">667</span> }<a name="line.667"></a>
-<span class="sourceLineNo">668</span> // Port<a name="line.668"></a>
-<span class="sourceLineNo">669</span> if (uri.getPort() != -1) {<a name="line.669"></a>
-<span class="sourceLineNo">670</span> sb.append(':');<a name="line.670"></a>
-<span class="sourceLineNo">671</span> sb.append(uri.getPort());<a name="line.671"></a>
+<span class="sourceLineNo">631</span> // Write records out for later verification, one per HBase field except for the<a name="line.631"></a>
+<span class="sourceLineNo">632</span> // content record, which will be verified by CRC64.<a name="line.632"></a>
+<span class="sourceLineNo">633</span><a name="line.633"></a>
+<span class="sourceLineNo">634</span> output.write(new HBaseKeyWritable(rowKey, INFO_FAMILY_NAME, CRC_QUALIFIER, now),<a name="line.634"></a>
+<span class="sourceLineNo">635</span> new BytesWritable(Bytes.toBytes(crc64)));<a name="line.635"></a>
+<span class="sourceLineNo">636</span> output.write(new HBaseKeyWritable(rowKey, INFO_FAMILY_NAME, CONTENT_LENGTH_QUALIFIER,<a name="line.636"></a>
+<span class="sourceLineNo">637</span> now), new BytesWritable(Bytes.toBytes(content.length)));<a name="line.637"></a>
+<span class="sourceLineNo">638</span> output.write(new HBaseKeyWritable(rowKey, INFO_FAMILY_NAME, CONTENT_TYPE_QUALIFIER,<a name="line.638"></a>
+<span class="sourceLineNo">639</span> now), new BytesWritable(Bytes.toBytes(contentType)));<a name="line.639"></a>
+<span class="sourceLineNo">640</span> output.write(new HBaseKeyWritable(rowKey, INFO_FAMILY_NAME, RECORD_ID_QUALIFIER,<a name="line.640"></a>
+<span class="sourceLineNo">641</span> now), new BytesWritable(Bytes.toBytes(recordID)));<a name="line.641"></a>
+<span class="sourceLineNo">642</span> output.write(new HBaseKeyWritable(rowKey, INFO_FAMILY_NAME, TARGET_URI_QUALIFIER,<a name="line.642"></a>
+<span class="sourceLineNo">643</span> now), new BytesWritable(Bytes.toBytes(targetURI)));<a name="line.643"></a>
+<span class="sourceLineNo">644</span> output.write(new HBaseKeyWritable(rowKey, INFO_FAMILY_NAME, DATE_QUALIFIER, now),<a name="line.644"></a>
+<span class="sourceLineNo">645</span> new BytesWritable(Bytes.toBytes(warcHeader.getDateString())));<a name="line.645"></a>
+<span class="sourceLineNo">646</span> if (ipAddr != null) {<a name="line.646"></a>
+<span class="sourceLineNo">647</span> output.write(new HBaseKeyWritable(rowKey, INFO_FAMILY_NAME, IP_ADDRESS_QUALIFIER,<a name="line.647"></a>
+<span class="sourceLineNo">648</span> now), new BytesWritable(Bytes.toBytes(ipAddr)));<a name="line.648"></a>
+<span class="sourceLineNo">649</span> }<a name="line.649"></a>
+<span class="sourceLineNo">650</span> }<a name="line.650"></a>
+<span class="sourceLineNo">651</span> }<a name="line.651"></a>
+<span class="sourceLineNo">652</span> }<a name="line.652"></a>
+<span class="sourceLineNo">653</span><a name="line.653"></a>
+<span class="sourceLineNo">654</span> private byte[] rowKeyFromTargetURI(String targetUri)<a name="line.654"></a>
+<span class="sourceLineNo">655</span> throws URISyntaxException, IllegalArgumentException {<a name="line.655"></a>
+<span class="sourceLineNo">656</span> URI uri = new URI(targetUri);<a name="line.656"></a>
+<span class="sourceLineNo">657</span> // Ignore the scheme<a name="line.657"></a>
+<span class="sourceLineNo">658</span> // Reverse the components of the hostname<a name="line.658"></a>
+<span class="sourceLineNo">659</span> String reversedHost;<a name="line.659"></a>
+<span class="sourceLineNo">660</span> if (uri.getHost() != null) {<a name="line.660"></a>
+<span class="sourceLineNo">661</span> StringBuffer sb = new StringBuffer();<a name="line.661"></a>
+<span class="sourceLineNo">662</span> String[] hostComponents = uri.getHost().split("\\.");<a name="line.662"></a>
+<span class="sourceLineNo">663</span> for (int i = hostComponents.length - 1; i >= 0; i--) {<a name="line.663"></a>
+<span class="sourceLineNo">664</span> sb.append(hostComponents[i]);<a name="line.664"></a>
+<span class="sourceLineNo">665</span> if (i != 0) {<a name="line.665"></a>
+<span class="sourceLineNo">666</span> sb.append('.');<a name="line.666"></a>
+<span class="sourceLineNo">667</span> }<a name="line.667"></a>
+<span class="sourceLineNo">668</span> }<a name="line.668"></a>
+<span class="sourceLineNo">669</span> reversedHost = sb.toString();<a name="line.669"></a>
+<span class="sourceLineNo">670</span> } else {<a name="line.670"></a>
+<span class="sourceLineNo">671</span> throw new IllegalArgumentException("URI is missing host component");<a name="line.671"></a>
<span class="sourceLineNo">672</span> }<a name="line.672"></a>
-<span class="sourceLineNo">673</span> if (uri.getRawPath() != null) {<a name="line.673"></a>
-<span class="sourceLineNo">674</span> sb.append(uri.getRawPath());<a name="line.674"></a>
-<span class="sourceLineNo">675</span> }<a name="line.675"></a>
-<span class="sourceLineNo">676</span> if (uri.getRawQuery() != null) {<a name="line.676"></a>
-<span class="sourceLineNo">677</span> sb.append('?');<a name="line.677"></a>
-<span class="sourceLineNo">678</span> sb.append(uri.getRawQuery());<a name="line.678"></a>
-<span class="sourceLineNo">679</span> }<a name="line.679"></a>
-<span class="sourceLineNo">680</span> if (uri.getRawFragment() != null) {<a name="line.680"></a>
-<span class="sourceLineNo">681</span> sb.append('#');<a name="line.681"></a>
-<span class="sourceLineNo">682</span> sb.append(uri.getRawFragment());<a name="line.682"></a>
-<span class="sourceLineNo">683</span> }<a name="line.683"></a>
-<span class="sourceLineNo">684</span> // Constrain the key size to the maximum allowed row key length<a name="line.684"></a>
-<span class="sourceLineNo">685</span> if (sb.length() > HConstants.MAX_ROW_LENGTH) {<a name="line.685"></a>
-<span class="sourceLineNo">686</span> sb.setLength(HConstants.MAX_ROW_LENGTH);<a name="line.686"></a>
-<span class="sourceLineNo">687</span> }<a name="line.687"></a>
-<span class="sourceLineNo">688</span> return Bytes.toBytes(sb.toString());<a name="line.688"></a>
-<span class="sourceLineNo">689</span> }<a name="line.689"></a>
-<span class="sourceLineNo">690</span><a name="line.690"></a>
-<span class="sourceLineNo">691</span> }<a name="line.691"></a>
-<span class="sourceLineNo">692</span> }<a name="line.692"></a>
-<span class="sourceLineNo">693</span><a name="line.693"></a>
-<span class="sourceLineNo">694</span> public static class OneFilePerMapperSFIF<K, V> extends SequenceFileInputFormat<K, V> {<a name="line.694"></a>
-<span class="sourceLineNo">695</span> @Override<a name="line.695"></a>
-<span class="sourceLineNo">696</span> protected boolean isSplitable(JobContext context, Path filename) {<a name="line.696"></a>
-<span class="sourceLineNo">697</span> return false;<a name="line.697"></a>
+<span class="sourceLineNo">673</span> StringBuffer sb = new StringBuffer();<a name="line.673"></a>
+<span class="sourceLineNo">674</span> sb.append(reversedHost);<a name="line.674"></a>
+<span class="sourceLineNo">675</span> if (uri.getPort() >= 0) {<a name="line.675"></a>
+<span class="sourceLineNo">676</span> sb.append(':');<a name="line.676"></a>
+<span class="sourceLineNo">677</span> sb.append(uri.getPort());<a name="line.677"></a>
+<span class="sourceLineNo">678</span> }<a name="line.678"></a>
+<span class="sourceLineNo">679</span> if (uri.getPath() != null) {<a name="line.679"></a>
+<span class="sourceLineNo">680</span> sb.append('/');<a name="line.680"></a>
+<span class="sourceLineNo">681</span> sb.append(uri.getPath());<a name="line.681"></a>
+<span class="sourceLineNo">682</span> }<a name="line.682"></a>
+<span class="sourceLineNo">683</span> if (uri.getQuery() != null) {<a name="line.683"></a>
+<span class="sourceLineNo">684</span> sb.append('?');<a name="line.684"></a>
+<span class="sourceLineNo">685</span> sb.append(uri.getQuery());<a name="line.685"></a>
+<span class="sourceLineNo">686</span> }<a name="line.686"></a>
+<span class="sourceLineNo">687</span> if (uri.getFragment() != null) {<a name="line.687"></a>
+<span class="sourceLineNo">688</span> sb.append('#');<a name="line.688"></a>
+<span class="sourceLineNo">689</span> sb.append(uri.getFragment());<a name="line.689"></a>
+<span class="sourceLineNo">690</span> }<a name="line.690"></a>
+<span class="sourceLineNo">691</span> if (sb.length() > HConstants.MAX_ROW_LENGTH) {<a name="line.691"></a>
+<span class="sourceLineNo">692</span> throw new IllegalArgumentException("Key would be too large (length=" + sb.length() +<a name="line.692"></a>
+<span class="sourceLineNo">693</span> ", limit=" + HConstants.MAX_ROW_LENGTH);<a name="line.693"></a>
+<span class="sourceLineNo">694</span> }<a name="line.694"></a>
+<span class="sourceLineNo">695</span> return Bytes.toBytes(sb.toString());<a name="line.695"></a>
+<span class="sourceLineNo">696</span> }<a name="line.696"></a>
+<span class="sourceLineNo">697</span><a name="line.697"></a>
<span class="sourceLineNo">698</span> }<a name="line.698"></a>
<span class="sourceLineNo">699</span> }<a name="line.699"></a>
<span class="sourceLineNo">700</span><a name="line.700"></a>
-<span class="sourceLineNo">701</span> public static class Verify extends Configured implements Tool {<a name="line.701"></a>
-<span class="sourceLineNo">702</span><a name="line.702"></a>
-<span class="sourceLineNo">703</span> public static final Logger LOG = LoggerFactory.getLogger(Verify.class);<a name="line.703"></a>
-<span class="sourceLineNo">704</span> public static final String USAGE = "Verify <inputDir>";<a name="line.704"></a>
-<span class="sourceLineNo">705</span><a name="line.705"></a>
-<span class="sourceLineNo">706</span> int run(Path inputDir) throws IOException, ClassNotFoundException, InterruptedException {<a name="line.706"></a>
-<span class="sourceLineNo">707</span> Job job = Job.getInstance(getConf());<a name="line.707"></a>
-<span class="sourceLineNo">708</span> job.setJobName(Verify.class.getName());<a name="line.708"></a>
-<span class="sourceLineNo">709</span> job.setJarByClass(getClass());<a name="line.709"></a>
-<span class="sourceLineNo">710</span> job.setMapperClass(VerifyMapper.class);<a name="line.710"></a>
-<span class="sourceLineNo">711</span> job.setInputFormatClass(OneFilePerMapperSFIF.class);<a name="line.711"></a>
-<span class="sourceLineNo">712</span> FileInputFormat.setInputPaths(job, inputDir);<a name="line.712"></a>
-<span class="sourceLineNo">713</span> job.setOutputFormatClass(NullOutputFormat.class);<a name="line.713"></a>
-<span class="sourceLineNo">714</span> job.setOutputKeyClass(NullWritable.class);<a name="line.714"></a>
-<span class="sourceLineNo">715</span> job.setOutputValueClass(NullWritable.class);<a name="line.715"></a>
-<span class="sourceLineNo">716</span> TableMapReduceUtil.addDependencyJars(job);<a name="line.716"></a>
-<span class="sourceLineNo">717</span> boolean success = job.waitForCompletion(true);<a name="line.717"></a>
-<span class="sourceLineNo">718</span> if (!success) {<a name="line.718"></a>
-<span class="sourceLineNo">719</span> LOG.error("Failure during job " + job.getJobID());<a name="line.719"></a>
-<span class="sourceLineNo">720</span> }<a name="line.720"></a>
-<span class="sourceLineNo">721</span> Counters counters = job.getCounters();<a name="line.721"></a>
-<span class="sourceLineNo">722</span> for (Counts c: Counts.values()) {<a name="line.722"></a>
-<span class="sourceLineNo">723</span> LOG.info(c + ": " + counters.findCounter(c).getValue());<a name="line.723"></a>
-<span class="sourceLineNo">724</span> }<a name="line.724"></a>
-<span class="sourceLineNo">725</span> return success ? 0 : 1;<a name="line.725"></a>
-<span class="sourceLineNo">726</span> }<a name="line.726"></a>
-<span class="sourceLineNo">727</span><a name="line.727"></a>
-<span class="sourceLineNo">728</span> @Override<a name="line.728"></a>
-<span class="sourceLineNo">729</span> public int run(String[] args) throws Exception {<a name="line.729"></a>
-<span class="sourceLineNo">730</span> if (args.length < 2) {<a name="line.730"></a>
-<span class="sourceLineNo">731</span> System.err.println(USAGE);<a name="line.731"></a>
-<span class="sourceLineNo">732</span> return 1;<a name="line.732"></a>
-<span class="sourceLineNo">733</span> }<a name="line.733"></a>
-<span class="sourceLineNo">734</span> Path loaderOutput = new Path(args[0]);<a name="line.734"></a>
-<span class="sourceLineNo">735</span> return run(loaderOutput);<a name="line.735"></a>
-<span class="sourceLineNo">736</span> }<a name="line.736"></a>
-<span class="sourceLineNo">737</span><a name="line.737"></a>
-<span class="sourceLineNo">738</span> public static void main(String[] args) throws Exception {<a name="line.738"></a>
-<span class="sourceLineNo">739</span> System.exit(ToolRunner.run(HBaseConfiguration.create(), new Verify(), args));<a name="line.739"></a>
-<span class="sourceLineNo">740</span> }<a name="line.740"></a>
-<span class="sourceLineNo">741</span><a name="line.741"></a>
-<span class="sourceLineNo">742</span> public static class VerifyMapper<a name="line.742"></a>
-<span class="sourceLineNo">743</span> extends Mapper<HBaseKeyWritable, BytesWritable, NullWritable, NullWritable> {<a name="line.743"></a>
+<span class="sourceLineNo">701</span> public static class OneFilePerMapperSFIF<K, V> extends SequenceFileInputFormat<K, V> {<a name="line.701"></a>
+<span class="sourceLineNo">702</span> @Override<a name="line.702"></a>
+<span class="sourceLineNo">703</span> protected boolean isSplitable(JobContext context, Path filename) {<a name="line.703"></a>
+<span class="sourceLineNo">704</span> return false;<a name="line.704"></a>
+<span class="sourceLineNo">705</span> }<a name="line.705"></a>
+<span class="sourceLineNo">706</span> }<a name="line.706"></a>
+<span class="sourceLineNo">707</span><a name="line.707"></a>
+<span class="sourceLineNo">708</span> public static class Verify extends Configured implements Tool {<a name="line.708"></a>
+<span class="sourceLineNo">709</span><a name="line.709"></a>
+<span class="sourceLineNo">710</span> public static final Logger LOG = LoggerFactory.getLogger(Verify.class);<a name="line.710"></a>
+<span class="sourceLineNo">711</span> public static final String USAGE = "Verify <inputDir>";<a name="line.711"></a>
+<span class="sourceLineNo">712</span><a name="line.712"></a>
+<span class="sourceLineNo">713</span> int run(Path inputDir) throws IOException, ClassNotFoundException, InterruptedException {<a name="line.713"></a>
+<span class="sourceLineNo">714</span> Job job = Job.getInstance(getConf());<a name="line.714"></a>
+<span class="sourceLineNo">715</span> job.setJobName(Verify.class.getName());<a name="line.715"></a>
+<span class="sourceLineNo">716</span> job.setJarByClass(getClass());<a name="line.716"></a>
+<span class="sourceLineNo">717</span> job.setMapperClass(VerifyMapper.class);<a name="line.717"></a>
+<span class="sourceLineNo">718</span> job.setInputFormatClass(OneFilePerMapperSFIF.class);<a name="line.718"></a>
+<span class="sourceLineNo">719</span> FileInputFormat.setInputPaths(job, inputDir);<a name="line.719"></a>
+<span class="sourceLineNo">720</span> job.setOutputFormatClass(NullOutputFormat.class);<a name="line.720"></a>
+<span class="sourceLineNo">721</span> job.setOutputKeyClass(NullWritable.class);<a name="line.721"></a>
+<span class="sourceLineNo">722</span> job.setOutputValueClass(NullWritable.class);<a name="line.722"></a>
+<span class="sourceLineNo">723</span> TableMapReduceUtil.addDependencyJars(job);<a name="line.723"></a>
+<span class="sourceLineNo">724</span> boolean success = job.waitForCompletion(true);<a name="line.724"></a>
+<span class="sourceLineNo">725</span> if (!success) {<a name="line.725"></a>
+<span class="sourceLineNo">726</span> LOG.error("Failure during job " + job.getJobID());<a name="line.726"></a>
+<span class="sourceLineNo">727</span> }<a name="line.727"></a>
+<span class="sourceLineNo">728</span> Counters counters = job.getCounters();<a name="line.728"></a>
+<span class="sourceLineNo">729</span> for (Counts c: Counts.values()) {<a name="line.729"></a>
+<span class="sourceLineNo">730</span> LOG.info(c + ": " + counters.findCounter(c).getValue());<a name="line.730"></a>
+<span class="sourceLineNo">731</span> }<a name="line.731"></a>
+<span class="sourceLineNo">732</span> return success ? 0 : 1;<a name="line.732"></a>
+<span class="sourceLineNo">733</span> }<a name="line.733"></a>
+<span class="sourceLineNo">734</span><a name="line.734"></a>
+<span class="sourceLineNo">735</span> @Override<a name="line.735"></a>
+<span class="sourceLineNo">736</span> public int run(String[] args) throws Exception {<a name="line.736"></a>
+<span class="sourceLineNo">737</span> if (args.length < 2) {<a name="line.737"></a>
+<span class="sourceLineNo">738</span> System.err.println(USAGE);<a name="line.738"></a>
+<span class="sourceLineNo">739</span> return 1;<a name="line.739"></a>
+<span class="sourceLineNo">740</span> }<a name="line.740"></a>
+<span class="sourceLineNo">741</span> Path loaderOutput = new Path(args[0]);<a name="line.741"></a>
+<span class="sourceLineNo">742</span> return run(loaderOutput);<a name="line.742"></a>
+<span class="sourceLineNo">743</span> }<a name="line.743"></a>
<span class="sourceLineNo">744</span><a name="line.744"></a>
-<span class="sourceLineNo">745</span> Connection conn;<a name="line.745"></a>
-<span class="sourceLineNo">746</span> Table table;<a name="line.746"></a>
-<span class="sourceLineNo">747</span><a name="line.747"></a>
-<span class="sourceLineNo">748</span> @Override<a name="line.748"></a>
-<span class="sourceLineNo">749</span> protected void setup(Context context) throws IOException, InterruptedException {<a name="line.749"></a>
-<span class="sourceLineNo">750</span> conn = ConnectionFactory.createConnection(context.getConfiguration());<a name="line.750"></a>
-<span class="sourceLineNo">751</span> table = conn.getTable(getTablename(conn.getConfiguration()));<a name="line.751"></a>
-<span class="sourceLineNo">752</span> }<a name="line.752"></a>
-<span class="sourceLineNo">753</span><a name="line.753"></a>
-<span class="sourceLineNo">754</span> @Override<a name="line.754"></a>
-<span class="sourceLineNo">755</span> protected void cleanup(Context context) throws IOException ,InterruptedException {<a name="line.755"></a>
-<span class="sourceLineNo">756</span> table.close();<a name="line.756"></a>
-<span class="sourceLineNo">757</span> conn.close();<a name="line.757"></a>
-<span class="sourceLineNo">758</span> }<a name="line.758"></a>
-<span class="sourceLineNo">759</span><a name="line.759"></a>
-<span class="sourceLineNo">760</span> @Override<a name="line.760"></a>
-<span class="sourceLineNo">761</span> protected void map(HBaseKeyWritable key, BytesWritable value, Context output)<a name="line.761"></a>
-<span class="sourceLineNo">762</span> throws IOException, InterruptedException {<a name="line.762"></a>
-<span class="sourceLineNo">763</span><a name="line.763"></a>
-<span class="sourceLineNo">764</span> byte[] row = Bytes.copy(key.getRowArray(), key.getRowOffset(), key.getRowLength());<a name="line.764"></a>
-<span class="sourceLineNo">765</span> byte[] family = Bytes.copy(key.getFamilyArray(), key.getFamilyOffset(),<a name="line.765"></a>
-<span class="sourceLineNo">766</span> key.getFamilyLength());<a name="line.766"></a>
-<span class="sourceLineNo">767</span> byte[] qualifier = Bytes.copy(key.getQualifierArray(), key.getQualifierOffset(),<a name="line.767"></a>
-<span class="sourceLineNo">768</span> key.getQualifierLength());<a name="line.768"></a>
-<span class="sourceLineNo">769</span> long ts = key.getTimestamp();<a name="line.769"></a>
+<span class="sourceLineNo">745</span> public static void main(String[] args) throws Exception {<a name="line.745"></a>
+<span class="sourceLineNo">746</span> System.exit(ToolRunner.run(HBaseConfiguration.create(), new Verify(), args));<a name="line.746"></a>
+<span class="sourceLineNo">747</span> }<a name="line.747"></a>
+<span class="sourceLineNo">748</span><a name="line.748"></a>
+<span class="sourceLineNo">749</span> public static class VerifyMapper<a name="line.749"></a>
+<span class="sourceLineNo">750</span> extends Mapper<HBaseKeyWritable, BytesWritable, NullWritable, NullWritable> {<a name="line.750"></a>
+<span class="sourceLineNo">751</span><a name="line.751"></a>
+<span class="sourceLineNo">752</span> Connection conn;<a name="line.752"></a>
+<span class="sourceLineNo">753</span> Table table;<a name="line.753"></a>
+<span class="sourceLineNo">754</span><a name="line.754"></a>
+<span class="sourceLineNo">755</span> @Override<a name="line.755"></a>
+<span class="sourceLineNo">756</span> protected void setup(Context context) throws IOException, InterruptedException {<a name="line.756"></a>
+<span class="sourceLineNo">757</span> conn = ConnectionFactory.createConnection(context.getConfiguration());<a name="line.757"></a>
+<span class="sourceLineNo">758</span> table = conn.getTable(getTablename(conn.getConfiguration()));<a name="line.758"></a>
+<span class="sourceLineNo">759</span> }<a name="line.759"></a>
+<span class="sourceLineNo">760</span><a name="line.760"></a>
+<span class="sourceLineNo">761</span> @Override<a name="line.761"></a>
+<span class="sourceLineNo">762</span> protected void cleanup(Context context) throws IOException ,InterruptedException {<a name="line.762"></a>
+<span class="sourceLineNo">763</span> table.close();<a name="line.763"></a>
+<span class="sourceLineNo">764</span> conn.close();<a name="line.764"></a>
+<span class="sourceLineNo">765</span> }<a name="line.765"></a>
+<span class="sourceLineNo">766</span><a name="line.766"></a>
+<span class="sourceLineNo">767</span> @Override<a name="line.767"></a>
+<span class="sourceLineNo">768</span> protected void map(HBaseKeyWritable key, BytesWritable value, Context output)<a name="line.768"></a>
+<span class="sourceLineNo">769</span> throws IOException, InterruptedException {<a name="line.769"></a>
<span class="sourceLineNo">770</span><a name="line.770"></a>
-<span class="sourceLineNo">771</span> if (Bytes.equals(INFO_FAMILY_NAME, family) &&<a name="line.771"></a>
-<span class="sourceLineNo">772</span> Bytes.equals(CRC_QUALIFIER, qualifier)) {<a name="line.772"></a>
-<span class="sourceLineNo">773</span><a name="line.773"></a>
-<span class="sourceLineNo">774</span> long expectedCRC64 = Bytes.toLong(value.getBytes(), 0, value.getLength());<a name="line.774"></a>
-<span class="sourceLineNo">775</span><a name="line.775"></a>
-<span class="sourceLineNo">776</span> Result result =<a name="line.776"></a>
-<span class="sourceLineNo">777</span> table.get(new Get(row)<a name="line.777"></a>
-<span class="sourceLineNo">778</span> .addColumn(CONTENT_FAMILY_NAME, CONTENT_QUALIFIER)<a name="line.778"></a>
-<span class="sourceLineNo">779</span> .addColumn(INFO_FAMILY_NAME, CRC_QUALIFIER)<a name="line.779"></a>
-<span class="sourceLineNo">780</span> .setTimestamp(ts));<a name="line.780"></a>
-<span class="sourceLineNo">781</span><a name="line.781"></a>
-<span class="sourceLineNo">782</span> byte[] content = result.getValue(CONTENT_FAMILY_NAME, CONTENT_QUALIFIER);<a name="line.782"></a>
-<span class="sourceLineNo">783</span> if (content == null) {<a name="line.783"></a>
-<span class="sourceLineNo">784</span> LOG.info("Row " + Bytes.toStringBinary(row) + ": missing content");<a name="line.784"></a>
-<span class="sourceLineNo">785</span> output.getCounter(Counts.UNREFERENCED).increment(1);<a name="line.785"></a>
-<span class="sourceLineNo">786</span> return;<a name="line.786"></a>
-<span class="sourceLineNo">787</span> } else {<a name="line.787"></a>
-<span class="sourceLineNo">788</span> CRC64 crc = new CRC64();<a name="line.788"></a>
-<span class="sourceLineNo">789</span> crc.update(content);<a name="line.789"></a>
-<span class="sourceLineNo">790</span> if (crc.getValue() != expectedCRC64) {<a name="line.790"></a>
-<span class="sourceLineNo">791</span> LOG.info("Row " + Bytes.toStringBinary(row) + ": corrupt content");<a name="line.791"></a>
-<span class="sourceLineNo">792</span> output.getCounter(Counts.CORRUPT).increment(1);<a name="line.792"></a>
-<span class="sourceLineNo">793</span> return;<a name="line.793"></a>
-<span class="sourceLineNo">794</span> }<a name="line.794"></a>
-<span class="sourceLineNo">795</span> }<a name="line.795"></a>
-<span class="sourceLineNo">796</span> byte[] crc = result.getValue(INFO_FAMILY_NAME, CRC_QUALIFIER);<a name="line.796"></a>
-<span class="sourceLineNo">797</span> if (crc == null) {<a name="line.797"></a>
-<span class="sourceLineNo">798</span> LOG.info("Row " + Bytes.toStringBinary(row) + ": missing i:c");<a name="line.798"></a>
-<span class="sourceLineNo">799</span> output.getCounter(Counts.UNREFERENCED).increment(1);<a name="line.799"></a>
-<span class="sourceLineNo">800</span> return;<a name="line.800"></a>
-<span class="sourceLineNo">801</span> }<a name="line.801"></a>
-<span class="sourceLineNo">802</span> if (Bytes.toLong(crc) != expectedCRC64) {<a name="line.802"></a>
-<span class="sourceLineNo">803</span> LOG.info("Row " + Bytes.toStringBinary(row) + ": i:c mismatch");<a name="line.803"></a>
-<span class="sourceLineNo">804</span> output.getCounter(Counts.CORRUPT).increment(1);<a name="line.804"></a>
-<span class="sourceLineNo">805</span> return;<a name="line.805"></a>
-<span class="sourceLineNo">806</span> }<a name="line.806"></a>
-<span class="sourceLineNo">807</span><a name="line.807"></a>
-<span class="sourceLineNo">808</span> } else {<a name="line.808"></a>
-<span class="sourceLineNo">809</span><a name="line.809"></a>
-<span class="sourceLineNo">810</span> Result result =<a name="line.810"></a>
-<span class="sourceLineNo">811</span> table.get(new Get(row)<a name="line.811"></a>
-<span class="sourceLineNo">812</span> .addColumn(family, qualifier)<a name="line.812"></a>
-<span class="sourceLineNo">813</span> .setTimestamp(ts));<a name="line.813"></a>
-<span class="sourceLineNo">814</span><a name="line.814"></a>
-<span class="sourceLineNo">815</span> byte[] bytes = result.getValue(family, qualifier);<a name="line.815"></a>
-<span class="sourceLineNo">816</span> if (bytes == null) {<a name="line.816"></a>
-<span class="sourceLineNo">817</span> LOG.info("Row " + Bytes.toStringBinary(row) + ": missing " +<a name="line.817"></a>
-<span class="sourceLineNo">818</span> Bytes.toStringBinary(family) + ":" + Bytes.toStringBinary(qualifier));<a name="line.818"></a>
-<span class="sourceLineNo">819</span> output.getCounter(Counts.UNREFERENCED).increment(1);<a name="line.819"></a>
-<span class="sourceLineNo">820</span> return;<a name="line.820"></a>
-<span class="sourceLineNo">821</span> }<a name="line.821"></a>
-<span class="sourceLineNo">822</span> if (!Bytes.equals(bytes, 0, bytes.length, value.getBytes(), 0, value.getLength())) {<a name="line.822"></a>
-<span class="sourceLineNo">823</span> LOG.info("Row " + Bytes.toStringBinary(row) + ": " +<a name="line.823"></a>
-<span class="sourceLineNo">824</span> Bytes.toStringBinary(family) + ":" + Bytes.toStringBinary(qualifier) +<a name="line.824"></a>
-<span class="sourceLineNo">825</span> " mismatch");<a name="line.825"></a>
-<span class="sourceLineNo">826</span> output.getCounter(Counts.CORRUPT).increment(1);<a name="line.826"></a>
-<span class="sourceLineNo">827</span> return;<a name="line.827"></a>
-<span class="sourceLineNo">828</span> }<a name="line.828"></a>
-<span class="sourceLineNo">829</span><a name="line.829"></a>
-<span class="sourceLineNo">830</span> }<a name="line.830"></a>
-<span class="sourceLineNo">831</span><a name="line.831"></a>
-<span class="sourceLineNo">832</span> output.getCounter(Counts.REFERENCED).increment(1);<a name="line.832"></a>
-<span class="sourceLineNo">833</span> }<a name="line.833"></a>
-<span class="sourceLineNo">834</span><a name="line.834"></a>
-<span class="sourceLineNo">835</span> }<a name="line.835"></a>
-<span class="sourceLineNo">836</span><a name="line.836"></a>
-<span class="sourceLineNo">837</span> }<a name="line.837"></a>
-<span class="sourceLineNo">838</span><a name="line.838"></a>
-<span class="sourceLineNo">839</span>}<a name="line.839"></a>
+<span class="sourceLineNo">771</span> byte[] row = Bytes.copy(key.getRowArray(), key.getRowOffset(), key.getRowLength());<a name="line.771"></a>
+<span class="sourceLineNo">772</span> byte[] family = Bytes.copy(key.getFamilyArray(), key.getFamilyOffset(),<a name="line.772"></a>
+<span class="sourceLineNo">773</span> key.getFamilyLength());<a name="line.773"></a>
+<span class="sourceLineNo">774</span> byte[] qualifier = Bytes.copy(key.getQualifierArray(), key.getQualifierOffset(),<a name="line.774"></a>
+<span class="sourceLineNo">775</span> key.getQualifierLength());<a name="line.775"></a>
+<span class="sourceLineNo">776</span> long ts = key.getTimestamp();<a name="line.776"></a>
+<span class="sourceLineNo">777</span> int retries = VERIFICATION_READ_RETRIES;<a name="line.777"></a>
+<span class="sourceLineNo">778</span><a name="line.778"></a>
+<span class="sourceLineNo">779</span> while (true) {<a name="line.779"></a>
+<span class="sourceLineNo">780</span><a name="line.780"></a>
+<span class="sourceLineNo">781</span> if (Bytes.equals(INFO_FAMILY_NAME, family) &&<a name="line.781"></a>
+<span class="sourceLineNo">782</span> Bytes.equals(CRC_QUALIFIER, qualifier)) {<a name="line.782"></a>
+<span class="sourceLineNo">783</span><a name="line.783"></a>
+<span class="sourceLineNo">784</span> long expectedCRC64 = Bytes.toLong(value.getBytes(), 0, value.getLength());<a name="line.784"></a>
+<span class="sourceLineNo">785</span> Result result = table.get(new Get(row)<a name="line.785"></a>
+<span class="sourceLineNo">786</span> .addColumn(CONTENT_FAMILY_NAME, CONTENT_QUALIFIER)<a name="line.786"></a>
+<span class="sourceLineNo">787</span> .addColumn(INFO_FAMILY_NAME, CRC_QUALIFIER)<a name="line.787"></a>
+<span class="sourceLineNo">788</span> .setTimestamp(ts));<a name="line.788"></a>
+<span class="sourceLineNo">789</span> byte[] content = result.getValue(CONTENT_FAMILY_NAME, CONTENT_QUALIFIER);<a name="line.789"></a>
+<span class="sourceLineNo">790</span> if (content == null) {<a name="line.790"></a>
+<span class="sourceLineNo">791</span> if (retries-- > 0) {<a name="line.791"></a>
+<span class="sourceLineNo">792</span> continue;<a name="line.792"></a>
+<span class="sourceLineNo">793</span> }<a name="line.793"></a>
+<span class="sourceLineNo">794</span> LOG.error("Row " + Bytes.toStringBinary(row) + ": missing content");<a name="line.794"></a>
+<span class="sourceLineNo">795</span> output.getCounter(Counts.UNREFERENCED).increment(1);<a name="line.795"></a>
+<span class="sourceLineNo">796</span> return;<a name="line.796"></a>
+<span class="sourceLineNo">797</span> } else {<a name="line.797"></a>
+<span class="sourceLineNo">798</span> CRC64 crc = new CRC64();<a name="line.798"></a>
+<span class="sourceLineNo">799</span> crc.update(content);<a name="line.799"></a>
+<span class="sourceLineNo">800</span> if (crc.getValue() != expectedCRC64) {<a name="line.800"></a>
+<span class="sourceLineNo">801</span> if (retries-- > 0) {<a name="line.801"></a>
+<span class="sourceLineNo">802</span> continue;<a name="line.802"></a>
+<span class="sourceLineNo">803</span> }<a name="line.803"></a>
+<span class="sourceLineNo">804</span> LOG.error("Row " + Bytes.toStringBinary(row) + ": corrupt content");<a name="line.804"></a>
+<span class="sourceLineNo">805</span> output.getCounter(Counts.CORRUPT).increment(1);<a name="line.805"></a>
+<span class="sourceLineNo">806</span> return;<a name="line.806"></a>
+<span class="sourceLineNo">807</span> }<a name="line.807"></a>
+<span class="sourceLineNo">808</span> }<a name="line.808"></a>
+<span class="sourceLineNo">809</span> byte[] crc = result.getValue(INFO_FAMILY_NAME, CRC_QUALIFIER);<a name="line.809"></a>
+<span class="sourceLineNo">810</span> if (crc == null) {<a name="line.810"></a>
+<span class="sourceLineNo">811</span> if (retries-- > 0) {<a name="line.811"></a>
+<span class="sourceLineNo">812</span> continue;<a name="line.812"></a>
+<span class="sourceLineNo">813</span> }<a name="line.813"></a>
+<span class="sourceLineNo">814</span> LOG.error("Row " + Bytes.toStringBinary(row) + ": missing i:c");<a name="line.814"></a>
+<span class="sourceLineNo">815</span> output.getCounter(Counts.UNREFERENCED).increment(1);<a name="line.815"></a>
+<span class="sourceLineNo">816</span> return;<a name="line.816"></a>
+<span class="sourceLineNo">817</span> }<a name="line.817"></a>
+<span class="sourceLineNo">818</span> if (Bytes.toLong(crc) != expectedCRC64) {<a name="line.818"></a>
+<span class="sourceLineNo">819</span> if (retries-- > 0) {<a name="line.819"></a>
+<span class="sourceLineNo">820</span> continue;<a name="line.820"></a>
+<span class="sourceLineNo">821</span> }<a name="line.821"></a>
+<span class="sourceLineNo">822</span> LOG.error("Row " + Bytes.toStringBinary(row) + ": i:c mismatch");<a name="line.822"></a>
+<span class="sourceLineNo">823</span> output.getCounter(Counts.CORRUPT).increment(1);<a name="line.823"></a>
+<span class="sourceLineNo">824</span> return;<a name="line.824"></a>
+<span class="sourceLineNo">825</span> }<a name="line.825"></a>
+<span class="sourceLineNo">826</span><a name="line.826"></a>
+<span class="sourceLineNo">827</span> } else {<a name="line.827"></a>
+<span class="sourceLineNo">828</span><a name="line.828"></a>
+<span class="sourceLineNo">829</span> Result result = table.get(new Get(row)<a name="line.829"></a>
+<span class="sourceLineNo">830</span> .addColumn(family, qualifier)<a name="line.830"></a>
+<span class="sourceLineNo">831</span> .setTimestamp(ts));<a name="line.831"></a>
+<span class="sourceLineNo">832</span> byte[] bytes = result.getValue(family, qualifier);<a name="line.832"></a>
+<span class="sourceLineNo">833</span> if (bytes == null) {<a name="line.833"></a>
+<span class="sourceLineNo">834</span> if (retries-- > 0) {<a name="line.834"></a>
+<span class="sourceLineNo">835</span> continue;<a name="line.835"></a>
+<span class="sourceLineNo">836</span> }<a name="line.836"></a>
+<span class="sourceLineNo">837</span> LOG.error("Row " + Bytes.toStringBinary(row) + ": missing " +<a name="line.837"></a>
+<span class="sourceLineNo">838</span> Bytes.toStringBinary(family) + ":" + Bytes.toStringBinary(qualifier));<a name="line.838"></a>
+<span class="sourceLineNo">839</span> output.getCounter(Counts.UNREFERENCED).increment(1);<a name="line.839"></a>
+<span class="sourceLineNo">840</span> return;<a name="line.840"></a>
+<span class="sourceLineNo">841</span> }<a name="line.841"></a>
+<span class="sourceLineNo">842</span> if (!Bytes.equals(bytes, 0, bytes.length, value.getBytes(), 0, value.getLength())) {<a name="line.842"></a>
+<span class="sourceLineNo">843</span> if (retries-- > 0) {<a name="line.843"></a>
+<span class="sourceLineNo">844</span> continue;<a name="line.844"></a>
+<span class="sourceLineNo">845</span> }<a name="line.845"></a>
+<span class="sourceLineNo">846</span> LOG.error("Row " + Bytes.toStringBinary(row) + ": " +<a name="line.846"></a>
+<span class="sourceLineNo">847</span> Bytes.toStringBinary(family) + ":" + Bytes.toStringBinary(qualifier) +<a name="line.847"></a>
+<span class="sourceLineNo">848</span> " mismatch");<a name="line.848"></a>
+<span class="sourceLineNo">849</span> output.getCounter(Counts.CORRUPT).increment(1);<a name="line.849"></a>
+<span class="sourceLineNo">850</span> return;<a name="line.850"></a>
+<span class="sourceLineNo">851</span> }<a name="line.851"></a>
+<span class="sourceLineNo">852</span><a name="line.852"></a>
+<span class="sourceLineNo">853</span> }<a name="line.853"></a>
+<span class="sourceLineNo">854</span><a name="line.854"></a>
+<span class="sourceLineNo">855</span> // If we fell through to here all verification checks have succeeded, potentially after<a name="line.855"></a>
+<span class="sourceLineNo">856</span> // retries, and we must exit the while loop.<a name="line.856"></a>
+<span class="sourceLineNo">857</span> output.getCounter(Counts.REFERENCED).increment(1);<a name="line.857"></a>
+<span class="sourceLineNo">858</span> break;<a name="line.858"></a>
+<span class="sourceLineNo">859</span><a name="line.859"></a>
+<span class="sourceLineNo">860</span> }<a name="line.860"></a>
+<span class="sourceLineNo">861</span> }<a name="line.861"></a>
+<span class="sourceLineNo">862</span> }<a name="line.862"></a>
+<span class="sourceLineNo">863</span> }<a name="line.863"></a>
+<span class="sourceLineNo">864</span><a name="line.864"></a>
+<span class="sourceLineNo">865</span>}<a name="line.865"></a>
diff --git a/testdevapidocs/src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.HBaseKeyWritable.html b/testdevapidocs/src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.HBaseKeyWritable.html
index 4a40cbc..e89b63c 100644
--- a/testdevapidocs/src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.HBaseKeyWritable.html
+++ b/testdevapidocs/src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.HBaseKeyWritable.html
@@ -54,797 +54,823 @@
<span class="sourceLineNo">046</span>import org.apache.hadoop.hbase.MasterNotRunningException;<a name="line.46"></a>
<span class="sourceLineNo">047</span>import org.apache.hadoop.hbase.TableName;<a name="line.47"></a>
<span class="sourceLineNo">048</span>import org.apache.hadoop.hbase.client.Admin;<a name="line.48"></a>
-<span class="sourceLineNo">049</span>import org.apache.hadoop.hbase.client.ColumnFamilyDescriptor;<a name="line.49"></a>
-<span class="sourceLineNo">050</span>import org.apache.hadoop.hbase.client.ColumnFamilyDescriptorBuilder;<a name="line.50"></a>
-<span class="sourceLineNo">051</span>import org.apache.hadoop.hbase.client.Connection;<a name="line.51"></a>
-<span class="sourceLineNo">052</span>import org.apache.hadoop.hbase.client.ConnectionFactory;<a name="line.52"></a>
-<span class="sourceLineNo">053</span>import org.apache.hadoop.hbase.client.Get;<a name="line.53"></a>
-<span class="sourceLineNo">054</span>import org.apache.hadoop.hbase.client.Put;<a name="line.54"></a>
-<span class="sourceLineNo">055</span>import org.apache.hadoop.hbase.client.Result;<a name="line.55"></a>
-<span class="sourceLineNo">056</span>import org.apache.hadoop.hbase.client.Table;<a name="line.56"></a>
-<span class="sourceLineNo">057</span>import org.apache.hadoop.hbase.client.TableDescriptor;<a name="line.57"></a>
-<span class="sourceLineNo">058</span>import org.apache.hadoop.hbase.client.TableDescriptorBuilder;<a name="line.58"></a>
-<span class="sourceLineNo">059</span>import org.apache.hadoop.hbase.io.encoding.DataBlockEncoding;<a name="line.59"></a>
-<span class="sourceLineNo">060</span>import org.apache.hadoop.hbase.mapreduce.TableMapReduceUtil;<a name="line.60"></a>
-<span class="sourceLineNo">061</span>import org.apache.hadoop.hbase.regionserver.BloomType;<a name="line.61"></a>
-<span class="sourceLineNo">062</span>import org.apache.hadoop.hbase.test.util.CRC64;<a name="line.62"></a>
-<span class="sourceLineNo">063</span>import org.apache.hadoop.hbase.test.util.warc.WARCInputFormat;<a name="line.63"></a>
-<span class="sourceLineNo">064</span>import org.apache.hadoop.hbase.test.util.warc.WARCRecord;<a name="line.64"></a>
-<span class="sourceLineNo">065</span>import org.apache.hadoop.hbase.test.util.warc.WARCWritable;<a name="line.65"></a>
-<span class="sourceLineNo">066</span>import org.apache.hadoop.hbase.util.Bytes;<a name="line.66"></a>
-<span class="sourceLineNo">067</span>import org.apache.hadoop.hbase.util.EnvironmentEdgeManager;<a name="line.67"></a>
-<span class="sourceLineNo">068</span>import org.apache.hadoop.hbase.util.RegionSplitter;<a name="line.68"></a>
-<span class="sourceLineNo">069</span>import org.apache.hadoop.io.BytesWritable;<a name="line.69"></a>
-<span class="sourceLineNo">070</span>import org.apache.hadoop.io.LongWritable;<a name="line.70"></a>
-<span class="sourceLineNo">071</span>import org.apache.hadoop.io.NullWritable;<a name="line.71"></a>
-<span class="sourceLineNo">072</span>import org.apache.hadoop.io.SequenceFile.CompressionType;<a name="line.72"></a>
-<span class="sourceLineNo">073</span>import org.apache.hadoop.io.Writable;<a name="line.73"></a>
-<span class="sourceLineNo">074</span>import org.apache.hadoop.mapreduce.Counters;<a name="line.74"></a>
-<span class="sourceLineNo">075</span>import org.apache.hadoop.mapreduce.Job;<a name="line.75"></a>
-<span class="sourceLineNo">076</span>import org.apache.hadoop.mapreduce.JobContext;<a name="line.76"></a>
-<span class="sourceLineNo">077</span>import org.apache.hadoop.mapreduce.Mapper;<a name="line.77"></a>
-<span class="sourceLineNo">078</span>import org.apache.hadoop.mapreduce.lib.input.FileInputFormat;<a name="line.78"></a>
-<span class="sourceLineNo">079</span>import org.apache.hadoop.mapreduce.lib.input.SequenceFileInputFormat;<a name="line.79"></a>
-<span class="sourceLineNo">080</span>import org.apache.hadoop.mapreduce.lib.output.NullOutputFormat;<a name="line.80"></a>
-<span class="sourceLineNo">081</span>import org.apache.hadoop.mapreduce.lib.output.SequenceFileOutputFormat;<a name="line.81"></a>
-<span class="sourceLineNo">082</span>import org.apache.hadoop.util.Tool;<a name="line.82"></a>
-<span class="sourceLineNo">083</span>import org.apache.hadoop.util.ToolRunner;<a name="line.83"></a>
-<span class="sourceLineNo">084</span>import org.slf4j.Logger;<a name="line.84"></a>
-<span class="sourceLineNo">085</span>import org.slf4j.LoggerFactory;<a name="line.85"></a>
-<span class="sourceLineNo">086</span><a name="line.86"></a>
-<span class="sourceLineNo">087</span>import org.apache.hbase.thirdparty.org.apache.commons.cli.CommandLine;<a name="line.87"></a>
-<span class="sourceLineNo">088</span><a name="line.88"></a>
-<span class="sourceLineNo">089</span>/**<a name="line.89"></a>
-<span class="sourceLineNo">090</span> * This integration test loads successful resource retrieval records from the Common Crawl<a name="line.90"></a>
-<span class="sourceLineNo">091</span> * (https://commoncrawl.org/) public dataset into an HBase table and writes records that can be<a name="line.91"></a>
-<span class="sourceLineNo">092</span> * used to later verify the presence and integrity of those records.<a name="line.92"></a>
-<span class="sourceLineNo">093</span> * <p><a name="line.93"></a>
-<span class="sourceLineNo">094</span> * Run like:<a name="line.94"></a>
-<span class="sourceLineNo">095</span> * <blockquote><a name="line.95"></a>
-<span class="sourceLineNo">096</span> * ./bin/hbase org.apache.hadoop.hbase.test.IntegrationTestLoadCommonCrawl \<br><a name="line.96"></a>
-<span class="sourceLineNo">097</span> * &nbsp;&nbsp; -Dfs.s3n.awsAccessKeyId=&lt;AWS access key&gt; \<br><a name="line.97"></a>
-<span class="sourceLineNo">098</span> * &nbsp;&nbsp; -Dfs.s3n.awsSecretAccessKey=&lt;AWS secret key&gt; \<br><a name="line.98"></a>
-<span class="sourceLineNo">099</span> * &nbsp;&nbsp; /path/to/test-CC-MAIN-2021-10-warc.paths.gz \<br><a name="line.99"></a>
-<span class="sourceLineNo">100</span> * &nbsp;&nbsp; /path/to/tmp/warc-loader-output<a name="line.100"></a>
-<span class="sourceLineNo">101</span> * </blockquote><a name="line.101"></a>
-<span class="sourceLineNo">102</span> * <p><a name="line.102"></a>
-<span class="sourceLineNo">103</span> * Access to the Common Crawl dataset in S3 is made available to anyone by Amazon AWS, but<a name="line.103"></a>
-<span class="sourceLineNo">104</span> * Hadoop's S3N filesystem still requires valid access credentials to initialize.<a name="line.104"></a>
-<span class="sourceLineNo">105</span> * <p><a name="line.105"></a>
-<span class="sourceLineNo">106</span> * The input path can either specify a directory or a file. The file may optionally be<a name="line.106"></a>
-<span class="sourceLineNo">107</span> * compressed with gzip. If a directory, the loader expects the directory to contain one or more<a name="line.107"></a>
-<span class="sourceLineNo">108</span> * WARC files from the Common Crawl dataset. If a file, the loader expects a list of Hadoop S3N<a name="line.108"></a>
-<span class="sourceLineNo">109</span> * URIs which point to S3 locations for one or more WARC files from the Common Crawl dataset,<a name="line.109"></a>
-<span class="sourceLineNo">110</span> * one URI per line. Lines should be terminated with the UNIX line terminator.<a name="line.110"></a>
-<span class="sourceLineNo">111</span> * <p><a name="line.111"></a>
-<span class="sourceLineNo">112</span> * Included in hbase-it/src/test/resources/CC-MAIN-2021-10-warc.paths.gz is a list of all WARC<a name="line.112"></a>
-<span class="sourceLineNo">113</span> * files comprising the Q1 2021 crawl archive. There are 64,000 WARC files in this data set, each<a name="line.113"></a>
-<span class="sourceLineNo">114</span> * containing ~1GB of gzipped data. The WARC files contain several record types, such as metadata,<a name="line.114"></a>
-<span class="sourceLineNo">115</span> * request, and response, but we only load the response record types. If the HBase table schema<a name="line.115"></a>
-<span class="sourceLineNo">116</span> * does not specify compression (by default) there is roughly a 10x expansion. Loading the full<a name="line.116"></a>
-<span class="sourceLineNo">117</span> * crawl archive results in a table approximately 640 TB in size.<a name="line.117"></a>
-<span class="sourceLineNo">118</span> * <p><a name="line.118"></a>
-<span class="sourceLineNo">119</span> * You can also split the Loader and Verify stages:<a name="line.119"></a>
-<span class="sourceLineNo">120</span> * <p><a name="line.120"></a>
-<span class="sourceLineNo">121</span> * Load with:<a name="line.121"></a>
-<span class="sourceLineNo">122</span> * <blockquote><a name="line.122"></a>
-<span class="sourceLineNo">123</span> * ./bin/hbase 'org.apache.hadoop.hbase.test.IntegrationTestLoadCommonCrawl$Loader' \<br><a name="line.123"></a>
-<span class="sourceLineNo">124</span> * &nbsp;&nbsp; -files /path/to/hadoop-aws.jar \<br><a name="line.124"></a>
-<span class="sourceLineNo">125</span> * &nbsp;&nbsp; -Dfs.s3n.awsAccessKeyId=&lt;AWS access key&gt; \<br><a name="line.125"></a>
-<span class="sourceLineNo">126</span> * &nbsp;&nbsp; -Dfs.s3n.awsSecretAccessKey=&lt;AWS secret key&gt; \<br><a name="line.126"></a>
-<span class="sourceLineNo">127</span> * &nbsp;&nbsp; /path/to/test-CC-MAIN-2021-10-warc.paths.gz \<br><a name="line.127"></a>
-<span class="sourceLineNo">128</span> * &nbsp;&nbsp; /path/to/tmp/warc-loader-output<a name="line.128"></a>
-<span class="sourceLineNo">129</span> * </blockquote><a name="line.129"></a>
-<span class="sourceLineNo">130</span> * <p><a name="line.130"></a>
-<span class="sourceLineNo">131</span> * Note: The hadoop-aws jar will be needed at runtime to instantiate the S3N filesystem. Use<a name="line.131"></a>
-<span class="sourceLineNo">132</span> * the <tt>-files</tt> ToolRunner argument to add it.<a name="line.132"></a>
-<span class="sourceLineNo">133</span> * <p><a name="line.133"></a>
-<span class="sourceLineNo">134</span> * Verify with:<a name="line.134"></a>
-<span class="sourceLineNo">135</span> * <blockquote><a name="line.135"></a>
-<span class="sourceLineNo">136</span> * ./bin/hbase 'org.apache.hadoop.hbase.test.IntegrationTestLoadCommonCrawl$Verify' \<br><a name="line.136"></a>
-<span class="sourceLineNo">137</span> * &nbsp;&nbsp; /path/to/tmp/warc-loader-output<a name="line.137"></a>
-<span class="sourceLineNo">138</span> * </blockquote><a name="line.138"></a>
-<span class="sourceLineNo">139</span> * <p><a name="line.139"></a>
-<span class="sourceLineNo">140</span> */<a name="line.140"></a>
-<span class="sourceLineNo">141</span>public class IntegrationTestLoadCommonCrawl extends IntegrationTestBase {<a name="line.141"></a>
-<span class="sourceLineNo">142</span><a name="line.142"></a>
-<span class="sourceLineNo">143</span> private static final Logger LOG = LoggerFactory.getLogger(IntegrationTestLoadCommonCrawl.class);<a name="line.143"></a>
-<span class="sourceLineNo">144</span><a name="line.144"></a>
-<span class="sourceLineNo">145</span> protected static String TABLE_NAME_KEY = "IntegrationTestLoadCommonCrawl.table";<a name="line.145"></a>
-<span class="sourceLineNo">146</span> protected static String DEFAULT_TABLE_NAME = "IntegrationTestLoadCommonCrawl";<a name="line.146"></a>
-<span class="sourceLineNo">147</span><a name="line.147"></a>
-<span class="sourceLineNo">148</span> protected static byte[] CONTENT_FAMILY_NAME = Bytes.toBytes("c");<a name="line.148"></a>
-<span class="sourceLineNo">149</span> protected static byte[] INFO_FAMILY_NAME = Bytes.toBytes("i");<a name="line.149"></a>
-<span class="sourceLineNo">150</span> protected static byte[] CONTENT_QUALIFIER = HConstants.EMPTY_BYTE_ARRAY;<a name="line.150"></a>
-<span class="sourceLineNo">151</span> protected static byte[] CONTENT_LENGTH_QUALIFIER = Bytes.toBytes("l");<a name="line.151"></a>
-<span class="sourceLineNo">152</span> protected static byte[] CONTENT_TYPE_QUALIFIER = Bytes.toBytes("t");<a name="line.152"></a>
-<span class="sourceLineNo">153</span> protected static byte[] CRC_QUALIFIER = Bytes.toBytes("c");<a name="line.153"></a>
-<span class="sourceLineNo">154</span> protected static byte[] DATE_QUALIFIER = Bytes.toBytes("d");<a name="line.154"></a>
-<span class="sourceLineNo">155</span> protected static byte[] IP_ADDRESS_QUALIFIER = Bytes.toBytes("a");<a name="line.155"></a>
-<span class="sourceLineNo">156</span> protected static byte[] RECORD_ID_QUALIFIER = Bytes.toBytes("r");<a name="line.156"></a>
-<span class="sourceLineNo">157</span> protected static byte[] TARGET_URI_QUALIFIER = Bytes.toBytes("u");<a name="line.157"></a>
-<span class="sourceLineNo">158</span><a name="line.158"></a>
-<span class="sourceLineNo">159</span> public static enum Counts {<a name="line.159"></a>
-<span class="sourceLineNo">160</span> REFERENCED, UNREFERENCED, CORRUPT<a name="line.160"></a>
-<span class="sourceLineNo">161</span> }<a name="line.161"></a>
-<span class="sourceLineNo">162</span><a name="line.162"></a>
-<span class="sourceLineNo">163</span> Path warcFileInputDir = null;<a name="line.163"></a>
-<span class="sourceLineNo">164</span> Path outputDir = null;<a name="line.164"></a>
-<span class="sourceLineNo">165</span> String[] args;<a name="line.165"></a>
-<span class="sourceLineNo">166</span><a name="line.166"></a>
-<span class="sourceLineNo">167</span> protected int runLoader(Path warcFileInputDir, Path outputDir) throws Exception {<a name="line.167"></a>
-<span class="sourceLineNo">168</span> Loader loader = new Loader();<a name="line.168"></a>
-<span class="sourceLineNo">169</span> loader.setConf(conf);<a name="line.169"></a>
-<span class="sourceLineNo">170</span> return loader.run(warcFileInputDir, outputDir);<a name="line.170"></a>
-<span class="sourceLineNo">171</span> }<a name="line.171"></a>
-<span class="sourceLineNo">172</span><a name="line.172"></a>
-<span class="sourceLineNo">173</span> protected int runVerify(Path inputDir) throws Exception {<a name="line.173"></a>
-<span class="sourceLineNo">174</span> Verify verify = new Verify();<a name="line.174"></a>
-<span class="sourceLineNo">175</span> verify.setConf(conf);<a name="line.175"></a>
-<span class="sourceLineNo">176</span> return verify.run(inputDir);<a name="line.176"></a>
-<span class="sourceLineNo">177</span> }<a name="line.177"></a>
-<span class="sourceLineNo">178</span><a name="line.178"></a>
-<span class="sourceLineNo">179</span> @Override<a name="line.179"></a>
-<span class="sourceLineNo">180</span> public int run(String[] args) {<a name="line.180"></a>
-<span class="sourceLineNo">181</span> if (args.length > 0) {<a name="line.181"></a>
-<span class="sourceLineNo">182</span> warcFileInputDir = new Path(args[0]);<a name="line.182"></a>
-<span class="sourceLineNo">183</span> if (args.length > 1) {<a name="line.183"></a>
-<span class="sourceLineNo">184</span> outputDir = new Path(args[1]);<a name="line.184"></a>
-<span class="sourceLineNo">185</span> }<a name="line.185"></a>
-<span class="sourceLineNo">186</span> }<a name="line.186"></a>
-<span class="sourceLineNo">187</span> try {<a name="line.187"></a>
-<span class="sourceLineNo">188</span> if (warcFileInputDir == null) {<a name="line.188"></a>
-<span class="sourceLineNo">189</span> throw new IllegalArgumentException("WARC input file or directory not specified");<a name="line.189"></a>
-<span class="sourceLineNo">190</span> }<a name="line.190"></a>
-<span class="sourceLineNo">191</span> if (outputDir == null) {<a name="line.191"></a>
-<span class="sourceLineNo">192</span> throw new IllegalArgumentException("Output directory not specified");<a name="line.192"></a>
+<span class="sourceLineNo">049</span>import org.apache.hadoop.hbase.client.BufferedMutator;<a name="line.49"></a>
+<span class="sourceLineNo">050</span>import org.apache.hadoop.hbase.client.ColumnFamilyDescriptor;<a name="line.50"></a>
+<span class="sourceLineNo">051</span>import org.apache.hadoop.hbase.client.ColumnFamilyDescriptorBuilder;<a name="line.51"></a>
+<span class="sourceLineNo">052</span>import org.apache.hadoop.hbase.client.Connection;<a name="line.52"></a>
+<span class="sourceLineNo">053</span>import org.apache.hadoop.hbase.client.ConnectionFactory;<a name="line.53"></a>
+<span class="sourceLineNo">054</span>import org.apache.hadoop.hbase.client.Get;<a name="line.54"></a>
+<span class="sourceLineNo">055</span>import org.apache.hadoop.hbase.client.Put;<a name="line.55"></a>
+<span class="sourceLineNo">056</span>import org.apache.hadoop.hbase.client.Result;<a name="line.56"></a>
+<span class="sourceLineNo">057</span>import org.apache.hadoop.hbase.client.Table;<a name="line.57"></a>
+<span class="sourceLineNo">058</span>import org.apache.hadoop.hbase.client.TableDescriptor;<a name="line.58"></a>
+<span class="sourceLineNo">059</span>import org.apache.hadoop.hbase.client.TableDescriptorBuilder;<a name="line.59"></a>
+<span class="sourceLineNo">060</span>import org.apache.hadoop.hbase.io.encoding.DataBlockEncoding;<a name="line.60"></a>
+<span class="sourceLineNo">061</span>import org.apache.hadoop.hbase.mapreduce.TableMapReduceUtil;<a name="line.61"></a>
+<span class="sourceLineNo">062</span>import org.apache.hadoop.hbase.regionserver.BloomType;<a name="line.62"></a>
+<span class="sourceLineNo">063</span>import org.apache.hadoop.hbase.test.util.CRC64;<a name="line.63"></a>
+<span class="sourceLineNo">064</span>import org.apache.hadoop.hbase.test.util.warc.WARCInputFormat;<a name="line.64"></a>
+<span class="sourceLineNo">065</span>import org.apache.hadoop.hbase.test.util.warc.WARCRecord;<a name="line.65"></a>
+<span class="sourceLineNo">066</span>import org.apache.hadoop.hbase.test.util.warc.WARCWritable;<a name="line.66"></a>
+<span class="sourceLineNo">067</span>import org.apache.hadoop.hbase.util.Bytes;<a name="line.67"></a>
+<span class="sourceLineNo">068</span>import org.apache.hadoop.hbase.util.EnvironmentEdgeManager;<a name="line.68"></a>
+<span class="sourceLineNo">069</span>import org.apache.hadoop.hbase.util.RegionSplitter;<a name="line.69"></a>
+<span class="sourceLineNo">070</span>import org.apache.hadoop.io.BytesWritable;<a name="line.70"></a>
+<span class="sourceLineNo">071</span>import org.apache.hadoop.io.LongWritable;<a name="line.71"></a>
+<span class="sourceLineNo">072</span>import org.apache.hadoop.io.NullWritable;<a name="line.72"></a>
+<span class="sourceLineNo">073</span>import org.apache.hadoop.io.SequenceFile.CompressionType;<a name="line.73"></a>
+<span class="sourceLineNo">074</span>import org.apache.hadoop.io.Writable;<a name="line.74"></a>
+<span class="sourceLineNo">075</span>import org.apache.hadoop.mapreduce.Counters;<a name="line.75"></a>
+<span class="sourceLineNo">076</span>import org.apache.hadoop.mapreduce.Job;<a name="line.76"></a>
+<span class="sourceLineNo">077</span>import org.apache.hadoop.mapreduce.JobContext;<a name="line.77"></a>
+<span class="sourceLineNo">078</span>import org.apache.hadoop.mapreduce.Mapper;<a name="line.78"></a>
+<span class="sourceLineNo">079</span>import org.apache.hadoop.mapreduce.lib.input.FileInputFormat;<a name="line.79"></a>
+<span class="sourceLineNo">080</span>import org.apache.hadoop.mapreduce.lib.input.SequenceFileInputFormat;<a name="line.80"></a>
+<span class="sourceLineNo">081</span>import org.apache.hadoop.mapreduce.lib.output.NullOutputFormat;<a name="line.81"></a>
+<span class="sourceLineNo">082</span>import org.apache.hadoop.mapreduce.lib.output.SequenceFileOutputFormat;<a name="line.82"></a>
+<span class="sourceLineNo">083</span>import org.apache.hadoop.util.Tool;<a name="line.83"></a>
+<span class="sourceLineNo">084</span>import org.apache.hadoop.util.ToolRunner;<a name="line.84"></a>
+<span class="sourceLineNo">085</span>import org.slf4j.Logger;<a name="line.85"></a>
+<span class="sourceLineNo">086</span>import org.slf4j.LoggerFactory;<a name="line.86"></a>
+<span class="sourceLineNo">087</span><a name="line.87"></a>
+<span class="sourceLineNo">088</span>import org.apache.hbase.thirdparty.org.apache.commons.cli.CommandLine;<a name="line.88"></a>
+<span class="sourceLineNo">089</span><a name="line.89"></a>
+<span class="sourceLineNo">090</span>/**<a name="line.90"></a>
+<span class="sourceLineNo">091</span> * This integration test loads successful resource retrieval records from the Common Crawl<a name="line.91"></a>
+<span class="sourceLineNo">092</span> * (https://commoncrawl.org/) public dataset into an HBase table and writes records that can be<a name="line.92"></a>
+<span class="sourceLineNo">093</span> * used to later verify the presence and integrity of those records.<a name="line.93"></a>
+<span class="sourceLineNo">094</span> * <p><a name="line.94"></a>
+<span class="sourceLineNo">095</span> * Run like:<a name="line.95"></a>
+<span class="sourceLineNo">096</span> * <blockquote><a name="line.96"></a>
+<span class="sourceLineNo">097</span> * ./bin/hbase org.apache.hadoop.hbase.test.IntegrationTestLoadCommonCrawl \<br><a name="line.97"></a>
+<span class="sourceLineNo">098</span> * &nbsp;&nbsp; -Dfs.s3n.awsAccessKeyId=&lt;AWS access key&gt; \<br><a name="line.98"></a>
+<span class="sourceLineNo">099</span> * &nbsp;&nbsp; -Dfs.s3n.awsSecretAccessKey=&lt;AWS secret key&gt; \<br><a name="line.99"></a>
+<span class="sourceLineNo">100</span> * &nbsp;&nbsp; /path/to/test-CC-MAIN-2021-10-warc.paths.gz \<br><a name="line.100"></a>
+<span class="sourceLineNo">101</span> * &nbsp;&nbsp; /path/to/tmp/warc-loader-output<a name="line.101"></a>
+<span class="sourceLineNo">102</span> * </blockquote><a name="line.102"></a>
+<span class="sourceLineNo">103</span> * <p><a name="line.103"></a>
+<span class="sourceLineNo">104</span> * Access to the Common Crawl dataset in S3 is made available to anyone by Amazon AWS, but<a name="line.104"></a>
+<span class="sourceLineNo">105</span> * Hadoop's S3N filesystem still requires valid access credentials to initialize.<a name="line.105"></a>
+<span class="sourceLineNo">106</span> * <p><a name="line.106"></a>
+<span class="sourceLineNo">107</span> * The input path can either specify a directory or a file. The file may optionally be<a name="line.107"></a>
+<span class="sourceLineNo">108</span> * compressed with gzip. If a directory, the loader expects the directory to contain one or more<a name="line.108"></a>
+<span class="sourceLineNo">109</span> * WARC files from the Common Crawl dataset. If a file, the loader expects a list of Hadoop S3N<a name="line.109"></a>
+<span class="sourceLineNo">110</span> * URIs which point to S3 locations for one or more WARC files from the Common Crawl dataset,<a name="line.110"></a>
+<span class="sourceLineNo">111</span> * one URI per line. Lines should be terminated with the UNIX line terminator.<a name="line.111"></a>
+<span class="sourceLineNo">112</span> * <p><a name="line.112"></a>
+<span class="sourceLineNo">113</span> * Included in hbase-it/src/test/resources/CC-MAIN-2021-10-warc.paths.gz is a list of all WARC<a name="line.113"></a>
+<span class="sourceLineNo">114</span> * files comprising the Q1 2021 crawl archive. There are 64,000 WARC files in this data set, each<a name="line.114"></a>
+<span class="sourceLineNo">115</span> * containing ~1GB of gzipped data. The WARC files contain several record types, such as metadata,<a name="line.115"></a>
+<span class="sourceLineNo">116</span> * request, and response, but we only load the response record types. If the HBase table schema<a name="line.116"></a>
+<span class="sourceLineNo">117</span> * does not specify compression (by default) there is roughly a 10x expansion. Loading the full<a name="line.117"></a>
+<span class="sourceLineNo">118</span> * crawl archive results in a table approximately 640 TB in size.<a name="line.118"></a>
+<span class="sourceLineNo">119</span> * <p><a name="line.119"></a>
+<span class="sourceLineNo">120</span> * You can also split the Loader and Verify stages:<a name="line.120"></a>
+<span class="sourceLineNo">121</span> * <p><a name="line.121"></a>
+<span class="sourceLineNo">122</span> * Load with:<a name="line.122"></a>
+<span class="sourceLineNo">123</span> * <blockquote><a name="line.123"></a>
+<span class="sourceLineNo">124</span> * ./bin/hbase 'org.apache.hadoop.hbase.test.IntegrationTestLoadCommonCrawl$Loader' \<br><a name="line.124"></a>
+<span class="sourceLineNo">125</span> * &nbsp;&nbsp; -files /path/to/hadoop-aws.jar \<br><a name="line.125"></a>
+<span class="sourceLineNo">126</span> * &nbsp;&nbsp; -Dfs.s3n.awsAccessKeyId=&lt;AWS access key&gt; \<br><a name="line.126"></a>
+<span class="sourceLineNo">127</span> * &nbsp;&nbsp; -Dfs.s3n.awsSecretAccessKey=&lt;AWS secret key&gt; \<br><a name="line.127"></a>
+<span class="sourceLineNo">128</span> * &nbsp;&nbsp; /path/to/test-CC-MAIN-2021-10-warc.paths.gz \<br><a name="line.128"></a>
+<span class="sourceLineNo">129</span> * &nbsp;&nbsp; /path/to/tmp/warc-loader-output<a name="line.129"></a>
+<span class="sourceLineNo">130</span> * </blockquote><a name="line.130"></a>
+<span class="sourceLineNo">131</span> * <p><a name="line.131"></a>
+<span class="sourceLineNo">132</span> * Note: The hadoop-aws jar will be needed at runtime to instantiate the S3N filesystem. Use<a name="line.132"></a>
+<span class="sourceLineNo">133</span> * the <tt>-files</tt> ToolRunner argument to add it.<a name="line.133"></a>
+<span class="sourceLineNo">134</span> * <p><a name="line.134"></a>
+<span class="sourceLineNo">135</span> * Verify with:<a name="line.135"></a>
+<span class="sourceLineNo">136</span> * <blockquote><a name="line.136"></a>
+<span class="sourceLineNo">137</span> * ./bin/hbase 'org.apache.hadoop.hbase.test.IntegrationTestLoadCommonCrawl$Verify' \<br><a name="line.137"></a>
+<span class="sourceLineNo">138</span> * &nbsp;&nbsp; /path/to/tmp/warc-loader-output<a name="line.138"></a>
+<span class="sourceLineNo">139</span> * </blockquote><a name="line.139"></a>
+<span class="sourceLineNo">140</span> * <p><a name="line.140"></a>
+<span class="sourceLineNo">141</span> */<a name="line.141"></a>
+<span class="sourceLineNo">142</span>public class IntegrationTestLoadCommonCrawl extends IntegrationTestBase {<a name="line.142"></a>
+<span class="sourceLineNo">143</span><a name="line.143"></a>
+<span class="sourceLineNo">144</span> private static final Logger LOG = LoggerFactory.getLogger(IntegrationTestLoadCommonCrawl.class);<a name="line.144"></a>
+<span class="sourceLineNo">145</span><a name="line.145"></a>
+<span class="sourceLineNo">146</span> protected static String TABLE_NAME_KEY = "IntegrationTestLoadCommonCrawl.table";<a name="line.146"></a>
+<span class="sourceLineNo">147</span> protected static String DEFAULT_TABLE_NAME = "IntegrationTestLoadCommonCrawl";<a name="line.147"></a>
+<span class="sourceLineNo">148</span><a name="line.148"></a>
+<span class="sourceLineNo">149</span> protected static byte[] CONTENT_FAMILY_NAME = Bytes.toBytes("c");<a name="line.149"></a>
+<span class="sourceLineNo">150</span> protected static byte[] INFO_FAMILY_NAME = Bytes.toBytes("i");<a name="line.150"></a>
+<span class="sourceLineNo">151</span> protected static byte[] CONTENT_QUALIFIER = HConstants.EMPTY_BYTE_ARRAY;<a name="line.151"></a>
+<span class="sourceLineNo">152</span> protected static byte[] CONTENT_LENGTH_QUALIFIER = Bytes.toBytes("l");<a name="line.152"></a>
+<span class="sourceLineNo">153</span> protected static byte[] CONTENT_TYPE_QUALIFIER = Bytes.toBytes("t");<a name="line.153"></a>
+<span class="sourceLineNo">154</span> protected static byte[] CRC_QUALIFIER = Bytes.toBytes("c");<a name="line.154"></a>
+<span class="sourceLineNo">155</span> protected static byte[] DATE_QUALIFIER = Bytes.toBytes("d");<a name="line.155"></a>
+<span class="sourceLineNo">156</span> protected static byte[] IP_ADDRESS_QUALIFIER = Bytes.toBytes("a");<a name="line.156"></a>
+<span class="sourceLineNo">157</span> protected static byte[] RECORD_ID_QUALIFIER = Bytes.toBytes("r");<a name="line.157"></a>
+<span class="sourceLineNo">158</span> protected static byte[] TARGET_URI_QUALIFIER = Bytes.toBytes("u");<a name="line.158"></a>
+<span class="sourceLineNo">159</span><a name="line.159"></a>
+<span class="sourceLineNo">160</span> private static final int VERIFICATION_READ_RETRIES = 10;<a name="line.160"></a>
+<span class="sourceLineNo">161</span><a name="line.161"></a>
+<span class="sourceLineNo">162</span> public static enum Counts {<a name="line.162"></a>
+<span class="sourceLineNo">163</span> REFERENCED, UNREFERENCED, CORRUPT<a name="line.163"></a>
+<span class="sourceLineNo">164</span> }<a name="line.164"></a>
+<span class="sourceLineNo">165</span><a name="line.165"></a>
+<span class="sourceLineNo">166</span> Path warcFileInputDir = null;<a name="line.166"></a>
+<span class="sourceLineNo">167</span> Path outputDir = null;<a name="line.167"></a>
+<span class="sourceLineNo">168</span> String[] args;<a name="line.168"></a>
+<span class="sourceLineNo">169</span><a name="line.169"></a>
+<span class="sourceLineNo">170</span> protected int runLoader(Path warcFileInputDir, Path outputDir) throws Exception {<a name="line.170"></a>
+<span class="sourceLineNo">171</span> Loader loader = new Loader();<a name="line.171"></a>
+<span class="sourceLineNo">172</span> loader.setConf(conf);<a name="line.172"></a>
+<span class="sourceLineNo">173</span> return loader.run(warcFileInputDir, outputDir);<a name="line.173"></a>
+<span class="sourceLineNo">174</span> }<a name="line.174"></a>
+<span class="sourceLineNo">175</span><a name="line.175"></a>
+<span class="sourceLineNo">176</span> protected int runVerify(Path inputDir) throws Exception {<a name="line.176"></a>
+<span class="sourceLineNo">177</span> Verify verify = new Verify();<a name="line.177"></a>
+<span class="sourceLineNo">178</span> verify.setConf(conf);<a name="line.178"></a>
+<span class="sourceLineNo">179</span> return verify.run(inputDir);<a name="line.179"></a>
+<span class="sourceLineNo">180</span> }<a name="line.180"></a>
+<span class="sourceLineNo">181</span><a name="line.181"></a>
+<span class="sourceLineNo">182</span> @Override<a name="line.182"></a>
+<span class="sourceLineNo">183</span> public int run(String[] args) {<a name="line.183"></a>
+<span class="sourceLineNo">184</span> if (args.length > 0) {<a name="line.184"></a>
+<span class="sourceLineNo">185</span> warcFileInputDir = new Path(args[0]);<a name="line.185"></a>
+<span class="sourceLineNo">186</span> if (args.length > 1) {<a name="line.186"></a>
+<span class="sourceLineNo">187</span> outputDir = new Path(args[1]);<a name="line.187"></a>
+<span class="sourceLineNo">188</span> }<a name="line.188"></a>
+<span class="sourceLineNo">189</span> }<a name="line.189"></a>
+<span class="sourceLineNo">190</span> try {<a name="line.190"></a>
+<span class="sourceLineNo">191</span> if (warcFileInputDir == null) {<a name="line.191"></a>
+<span class="sourceLineNo">192</span> throw new IllegalArgumentException("WARC input file or directory not specified");<a name="line.192"></a>
<span class="sourceLineNo">193</span> }<a name="line.193"></a>
-<span class="sourceLineNo">194</span> int res = runLoader(warcFileInputDir, outputDir);<a name="line.194"></a>
-<span class="sourceLineNo">195</span> if (res != 0) {<a name="line.195"></a>
-<span class="sourceLineNo">196</span> LOG.error("Loader failed");<a name="line.196"></a>
-<span class="sourceLineNo">197</span> return -1;<a name="line.197"></a>
-<span class="sourceLineNo">198</span> }<a name="line.198"></a>
-<span class="sourceLineNo">199</span> res = runVerify(outputDir);<a name="line.199"></a>
-<span class="sourceLineNo">200</span> } catch (Exception e) {<a name="line.200"></a>
-<span class="sourceLineNo">201</span> LOG.error("Tool failed with exception", e);<a name="line.201"></a>
-<span class="sourceLineNo">202</span> return -1;<a name="line.202"></a>
-<span class="sourceLineNo">203</span> }<a name="line.203"></a>
-<span class="sourceLineNo">204</span> return 0;<a name="line.204"></a>
-<span class="sourceLineNo">205</span> }<a name="line.205"></a>
-<span class="sourceLineNo">206</span><a name="line.206"></a>
-<span class="sourceLineNo">207</span> @Override<a name="line.207"></a>
-<span class="sourceLineNo">208</span> protected void processOptions(CommandLine cmd) {<a name="line.208"></a>
-<span class="sourceLineNo">209</span> processBaseOptions(cmd);<a name="line.209"></a>
-<span class="sourceLineNo">210</span> args = cmd.getArgs();<a name="line.210"></a>
-<span class="sourceLineNo">211</span> }<a name="line.211"></a>
-<span class="sourceLineNo">212</span><a name="line.212"></a>
-<span class="sourceLineNo">213</span> @Override<a name="line.213"></a>
-<span class="sourceLineNo">214</span> public void setUpCluster() throws Exception {<a name="line.214"></a>
-<span class="sourceLineNo">215</span> util = getTestingUtil(getConf());<a name="line.215"></a>
-<span class="sourceLineNo">216</span> boolean isDistributed = util.isDistributedCluster();<a name="line.216"></a>
-<span class="sourceLineNo">217</span> util.initializeCluster(isDistributed ? 1 : 3);<a name="line.217"></a>
-<span class="sourceLineNo">218</span> if (!isDistributed) {<a name="line.218"></a>
-<span class="sourceLineNo">219</span> util.startMiniMapReduceCluster();<a name="line.219"></a>
-<span class="sourceLineNo">220</span> }<a name="line.220"></a>
-<span class="sourceLineNo">221</span> this.setConf(util.getConfiguration());<a name="line.221"></a>
-<span class="sourceLineNo">222</span> }<a name="line.222"></a>
-<span class="sourceLineNo">223</span><a name="line.223"></a>
-<span class="sourceLineNo">224</span> @Override<a name="line.224"></a>
-<span class="sourceLineNo">225</span> public void cleanUpCluster() throws Exception {<a name="line.225"></a>
-<span class="sourceLineNo">226</span> super.cleanUpCluster();<a name="line.226"></a>
-<span class="sourceLineNo">227</span> if (util.isDistributedCluster()) {<a name="line.227"></a>
-<span class="sourceLineNo">228</span> util.shutdownMiniMapReduceCluster();<a name="line.228"></a>
-<span class="sourceLineNo">229</span> }<a name="line.229"></a>
-<span class="sourceLineNo">230</span> }<a name="line.230"></a>
-<span class="sourceLineNo">231</span><a name="line.231"></a>
-<span class="sourceLineNo">232</span> static TableName getTablename(Configuration c) {<a name="line.232"></a>
-<span class="sourceLineNo">233</span> return TableName.valueOf(c.get(TABLE_NAME_KEY, DEFAULT_TABLE_NAME));<a name="line.233"></a>
-<span class="sourceLineNo">234</span> }<a name="line.234"></a>
-<span class="sourceLineNo">235</span><a name="line.235"></a>
-<span class="sourceLineNo">236</span> @Override<a name="line.236"></a>
-<span class="sourceLineNo">237</span> public TableName getTablename() {<a name="line.237"></a>
-<span class="sourceLineNo">238</span> return getTablename(getConf());<a name="line.238"></a>
-<span class="sourceLineNo">239</span> }<a name="line.239"></a>
-<span class="sourceLineNo">240</span><a name="line.240"></a>
-<span class="sourceLineNo">241</span> @Override<a name="line.241"></a>
-<span class="sourceLineNo">242</span> protected Set<String> getColumnFamilies() {<a name="line.242"></a>
-<span class="sourceLineNo">243</span> Set<String> families = new HashSet<>();<a name="line.243"></a>
-<span class="sourceLineNo">244</span> families.add(Bytes.toString(CONTENT_FAMILY_NAME));<a name="line.244"></a>
-<span class="sourceLineNo">245</span> families.add(Bytes.toString(INFO_FAMILY_NAME));<a name="line.245"></a>
-<span class="sourceLineNo">246</span> return families;<a name="line.246"></a>
-<span class="sourceLineNo">247</span> }<a name="line.247"></a>
-<span class="sourceLineNo">248</span><a name="line.248"></a>
-<span class="sourceLineNo">249</span> @Override<a name="line.249"></a>
-<span class="sourceLineNo">250</span> public int runTestFromCommandLine() throws Exception {<a name="line.250"></a>
-<span class="sourceLineNo">251</span> return ToolRunner.run(getConf(), this, args);<a name="line.251"></a>
-<span class="sourceLineNo">252</span> }<a name="line.252"></a>
-<span class="sourceLineNo">253</span><a name="line.253"></a>
-<span class="sourceLineNo">254</span> public static void main(String[] args) throws Exception {<a name="line.254"></a>
-<span class="sourceLineNo">255</span> Configuration conf = HBaseConfiguration.create();<a name="line.255"></a>
-<span class="sourceLineNo">256</span> IntegrationTestingUtility.setUseDistributedCluster(conf);<a name="line.256"></a>
-<span class="sourceLineNo">257</span> int ret = ToolRunner.run(conf, new IntegrationTestLoadCommonCrawl(), args);<a name="line.257"></a>
-<span class="sourceLineNo">258</span> System.exit(ret);<a name="line.258"></a>
-<span class="sourceLineNo">259</span> }<a name="line.259"></a>
-<span class="sourceLineNo">260</span><a name="line.260"></a>
-<span class="sourceLineNo">261</span> public static class HBaseKeyWritable implements Writable {<a name="line.261"></a>
-<span class="sourceLineNo">262</span><a name="line.262"></a>
-<span class="sourceLineNo">263</span> private byte[] row;<a name="line.263"></a>
-<span class="sourceLineNo">264</span> private int rowOffset;<a name="line.264"></a>
-<span class="sourceLineNo">265</span> private int rowLength;<a name="line.265"></a>
-<span class="sourceLineNo">266</span> private byte[] family;<a name="line.266"></a>
-<span class="sourceLineNo">267</span> private int familyOffset;<a name="line.267"></a>
-<span class="sourceLineNo">268</span> private int familyLength;<a name="line.268"></a>
-<span class="sourceLineNo">269</span> private byte[] qualifier;<a name="line.269"></a>
-<span class="sourceLineNo">270</span> private int qualifierOffset;<a name="line.270"></a>
-<span class="sourceLineNo">271</span> private int qualifierLength;<a name="line.271"></a>
-<span class="sourceLineNo">272</span> private long ts;<a name="line.272"></a>
-<span class="sourceLineNo">273</span><a name="line.273"></a>
-<span class="sourceLineNo">274</span> public HBaseKeyWritable() { }<a name="line.274"></a>
-<span class="sourceLineNo">275</span><a name="line.275"></a>
-<span class="sourceLineNo">276</span> public HBaseKeyWritable(byte[] row, int rowOffset, int rowLength,<a name="line.276"></a>
-<span class="sourceLineNo">277</span> byte[] family, int familyOffset, int familyLength,<a name="line.277"></a>
-<span class="sourceLineNo">278</span> byte[] qualifier, int qualifierOffset, int qualifierLength, long ts) {<a name="line.278"></a>
-<span class="sourceLineNo">279</span> this.row = row;<a name="line.279"></a>
-<span class="sourceLineNo">280</span> this.rowOffset = rowOffset;<a name="line.280"></a>
-<span class="sourceLineNo">281</span> this.rowLength = rowLength;<a name="line.281"></a>
-<span class="sourceLineNo">282</span> this.family = family;<a name="line.282"></a>
-<span class="sourceLineNo">283</span> this.familyOffset = familyOffset;<a name="line.283"></a>
-<span class="sourceLineNo">284</span> this.familyLength = familyLength;<a name="line.284"></a>
-<span class="sourceLineNo">285</span> this.qualifier = qualifier;<a name="line.285"></a>
-<span class="sourceLineNo">286</span> this.qualifierOffset = qualifierOffset;<a name="line.286"></a>
-<span class="sourceLineNo">287</span> this.qualifierLength = qualifierLength;<a name="line.287"></a>
-<span class="sourceLineNo">288</span> this.ts = ts;<a name="line.288"></a>
-<span class="sourceLineNo">289</span> }<a name="line.289"></a>
-<span class="sourceLineNo">290</span><a name="line.290"></a>
-<span class="sourceLineNo">291</span> public HBaseKeyWritable(byte[] row, byte[] family, byte[] qualifier, long ts) {<a name="line.291"></a>
-<span class="sourceLineNo">292</span> this(row, 0, row.length,<a name="line.292"></a>
-<span class="sourceLineNo">293</span> family, 0, family.length,<a name="line.293"></a>
-<span class="sourceLineNo">294</span> qualifier, 0, qualifier != null ? qualifier.length : 0, ts);<a name="line.294"></a>
-<span class="sourceLineNo">295</span> }<a name="line.295"></a>
-<span class="sourceLineNo">296</span><a name="line.296"></a>
-<span class="sourceLineNo">297</span> public HBaseKeyWritable(byte[] row, byte[] family, long ts) {<a name="line.297"></a>
-<span class="sourceLineNo">298</span> this(row, family, HConstants.EMPTY_BYTE_ARRAY, ts);<a name="line.298"></a>
-<span class="sourceLineNo">299</span> }<a name="line.299"></a>
-<span class="sourceLineNo">300</span><a name="line.300"></a>
-<span class="sourceLineNo">301</span> public HBaseKeyWritable(Cell cell) {<a name="line.301"></a>
-<span class="sourceLineNo">302</span> this(cell.getRowArray(), cell.getRowOffset(), cell.getRowLength(),<a name="line.302"></a>
-<span class="sourceLineNo">303</span> cell.getFamilyArray(), cell.getFamilyOffset(), cell.getFamilyLength(),<a name="line.303"></a>
-<span class="sourceLineNo">304</span> cell.getQualifierArray(), cell.getQualifierOffset(), cell.getQualifierLength(),<a name="line.304"></a>
-<span class="sourceLineNo">305</span> cell.getTimestamp());<a name="line.305"></a>
-<span class="sourceLineNo">306</span> }<a name="line.306"></a>
-<span class="sourceLineNo">307</span><a name="line.307"></a>
-<span class="sourceLineNo">308</span> @Override<a name="line.308"></a>
-<span class="sourceLineNo">309</span> public void readFields(DataInput in) throws IOException {<a name="line.309"></a>
-<span class="sourceLineNo">310</span> this.row = Bytes.toBytes(in.readUTF());<a name="line.310"></a>
-<span class="sourceLineNo">311</span> this.rowOffset = 0;<a name="line.311"></a>
-<span class="sourceLineNo">312</span> this.rowLength = row.length;<a name="line.312"></a>
-<span class="sourceLineNo">313</span> this.family = Bytes.toBytes(in.readUTF());<a name="line.313"></a>
-<span class="sourceLineNo">314</span> this.familyOffset = 0;<a name="line.314"></a>
-<span class="sourceLineNo">315</span> this.familyLength = family.length;<a name="line.315"></a>
-<span class="sourceLineNo">316</span> this.qualifier = Bytes.toBytes(in.readUTF());<a name="line.316"></a>
-<span class="sourceLineNo">317</span> this.qualifierOffset = 0;<a name="line.317"></a>
-<span class="sourceLineNo">318</span> this.qualifierLength = qualifier.length;<a name="line.318"></a>
-<span class="sourceLineNo">319</span> this.ts = in.readLong();<a name="line.319"></a>
-<span class="sourceLineNo">320</span> }<a name="line.320"></a>
-<span class="sourceLineNo">321</span><a name="line.321"></a>
-<span class="sourceLineNo">322</span> @Override<a name="line.322"></a>
-<span class="sourceLineNo">323</span> public void write(DataOutput out) throws IOException {<a name="line.323"></a>
-<span class="sourceLineNo">324</span> out.writeUTF(new String(row, rowOffset, rowLength, StandardCharsets.UTF_8));<a name="line.324"></a>
-<span class="sourceLineNo">325</span> out.writeUTF(new String(family, familyOffset, familyLength, StandardCharsets.UTF_8));<a name="line.325"></a>
-<span class="sourceLineNo">326</span> if (qualifier != null) {<a name="line.326"></a>
-<span class="sourceLineNo">327</span> out.writeUTF(new String(qualifier, qualifierOffset, qualifierLength,<a name="line.327"></a>
-<span class="sourceLineNo">328</span> StandardCharsets.UTF_8));<a name="line.328"></a>
-<span class="sourceLineNo">329</span> } else {<a name="line.329"></a>
-<span class="sourceLineNo">330</span> out.writeUTF("");<a name="line.330"></a>
-<span class="sourceLineNo">331</span> }<a name="line.331"></a>
-<span class="sourceLineNo">332</span> out.writeLong(ts);<a name="line.332"></a>
-<span class="sourceLineNo">333</span> }<a name="line.333"></a>
-<span class="sourceLineNo">334</span><a name="line.334"></a>
-<span class="sourceLineNo">335</span> public byte[] getRowArray() {<a name="line.335"></a>
-<span class="sourceLineNo">336</span> return row;<a name="line.336"></a>
-<span class="sourceLineNo">337</span> }<a name="line.337"></a>
-<span class="sourceLineNo">338</span><a name="line.338"></a>
-<span class="sourceLineNo">339</span> public void setRow(byte[] row) {<a name="line.339"></a>
-<span class="sourceLineNo">340</span> this.row = row;<a name="line.340"></a>
-<span class="sourceLineNo">341</span> }<a name="line.341"></a>
-<span class="sourceLineNo">342</span><a name="line.342"></a>
-<span class="sourceLineNo">343</span> public int getRowOffset() {<a name="line.343"></a>
-<span class="sourceLineNo">344</span> return rowOffset;<a name="line.344"></a>
-<span class="sourceLineNo">345</span> }<a name="line.345"></a>
-<span class="sourceLineNo">346</span><a name="line.346"></a>
-<span class="sourceLineNo">347</span> public void setRowOffset(int rowOffset) {<a name="line.347"></a>
-<span class="sourceLineNo">348</span> this.rowOffset = rowOffset;<a name="line.348"></a>
-<span class="sourceLineNo">349</span> }<a name="line.349"></a>
-<span class="sourceLineNo">350</span><a name="line.350"></a>
-<span class="sourceLineNo">351</span> public int getRowLength() {<a name="line.351"></a>
-<span class="sourceLineNo">352</span> return rowLength;<a name="line.352"></a>
-<span class="sourceLineNo">353</span> }<a name="line.353"></a>
-<span class="sourceLineNo">354</span><a name="line.354"></a>
-<span class="sourceLineNo">355</span> public void setRowLength(int rowLength) {<a name="line.355"></a>
-<span class="sourceLineNo">356</span> this.rowLength = rowLength;<a name="line.356"></a>
-<span class="sourceLineNo">357</span> }<a name="line.357"></a>
-<span class="sourceLineNo">358</span><a name="line.358"></a>
-<span class="sourceLineNo">359</span> public byte[] getFamilyArray() {<a name="line.359"></a>
-<span class="sourceLineNo">360</span> return family;<a name="line.360"></a>
-<span class="sourceLineNo">361</span> }<a name="line.361"></a>
-<span class="sourceLineNo">362</span><a name="line.362"></a>
-<span class="sourceLineNo">363</span> public void setFamily(byte[] family) {<a name="line.363"></a>
-<span class="sourceLineNo">364</span> this.family = family;<a name="line.364"></a>
-<span class="sourceLineNo">365</span> }<a name="line.365"></a>
-<span class="sourceLineNo">366</span><a name="line.366"></a>
-<span class="sourceLineNo">367</span> public int getFamilyOffset() {<a name="line.367"></a>
-<span class="sourceLineNo">368</span> return familyOffset;<a name="line.368"></a>
-<span class="sourceLineNo">369</span> }<a name="line.369"></a>
-<span class="sourceLineNo">370</span><a name="line.370"></a>
-<span class="sourceLineNo">371</span> public void setFamilyOffset(int familyOffset) {<a name="line.371"></a>
-<span class="sourceLineNo">372</span> this.familyOffset = familyOffset;<a name="line.372"></a>
-<span class="sourceLineNo">373</span> }<a name="line.373"></a>
-<span class="sourceLineNo">374</span><a name="line.374"></a>
-<span class="sourceLineNo">375</span> public int getFamilyLength() {<a name="line.375"></a>
-<span class="sourceLineNo">376</span> return familyLength;<a name="line.376"></a>
-<span class="sourceLineNo">377</span> }<a name="line.377"></a>
-<span class="sourceLineNo">378</span><a name="line.378"></a>
-<span class="sourceLineNo">379</span> public void setFamilyLength(int familyLength) {<a name="line.379"></a>
-<span class="sourceLineNo">380</span> this.familyLength = familyLength;<a name="line.380"></a>
-<span class="sourceLineNo">381</span> }<a name="line.381"></a>
-<span class="sourceLineNo">382</span><a name="line.382"></a>
-<span class="sourceLineNo">383</span> public byte[] getQualifierArray() {<a name="line.383"></a>
-<span class="sourceLineNo">384</span> return qualifier;<a name="line.384"></a>
-<span class="sourceLineNo">385</span> }<a name="line.385"></a>
-<span class="sourceLineNo">386</span><a name="line.386"></a>
-<span class="sourceLineNo">387</span> public void setQualifier(byte[] qualifier) {<a name="line.387"></a>
-<span class="sourceLineNo">388</span> this.qualifier = qualifier;<a name="line.388"></a>
-<span class="sourceLineNo">389</span> }<a name="line.389"></a>
-<span class="sourceLineNo">390</span><a name="line.390"></a>
-<span class="sourceLineNo">391</span> public int getQualifierOffset() {<a name="line.391"></a>
-<span class="sourceLineNo">392</span> return qualifierOffset;<a name="line.392"></a>
-<span class="sourceLineNo">393</span> }<a name="line.393"></a>
-<span class="sourceLineNo">394</span><a name="line.394"></a>
-<span class="sourceLineNo">395</span> public void setQualifierOffset(int qualifierOffset) {<a name="line.395"></a>
-<span class="sourceLineNo">396</span> this.qualifierOffset = qualifierOffset;<a name="line.396"></a>
-<span class="sourceLineNo">397</span> }<a name="line.397"></a>
-<span class="sourceLineNo">398</span><a name="line.398"></a>
-<span class="sourceLineNo">399</span> public int getQualifierLength() {<a name="line.399"></a>
-<span class="sourceLineNo">400</span> return qualifierLength;<a name="line.400"></a>
-<span class="sourceLineNo">401</span> }<a name="line.401"></a>
-<span class="sourceLineNo">402</span><a name="line.402"></a>
-<span class="sourceLineNo">403</span> public void setQualifierLength(int qualifierLength) {<a name="line.403"></a>
-<span class="sourceLineNo">404</span> this.qualifierLength = qualifierLength;<a name="line.404"></a>
-<span class="sourceLineNo">405</span> }<a name="line.405"></a>
-<span class="sourceLineNo">406</span><a name="line.406"></a>
-<span class="sourceLineNo">407</span> public long getTimestamp() {<a name="line.407"></a>
-<span class="sourceLineNo">408</span> return ts;<a name="line.408"></a>
-<span class="sourceLineNo">409</span> }<a name="line.409"></a>
-<span class="sourceLineNo">410</span><a name="line.410"></a>
-<span class="sourceLineNo">411</span> public void setTimestamp(long ts) {<a name="line.411"></a>
-<span class="sourceLineNo">412</span> this.ts = ts;<a name="line.412"></a>
-<span class="sourceLineNo">413</span> }<a name="line.413"></a>
-<span class="sourceLineNo">414</span> }<a name="line.414"></a>
-<span class="sourceLineNo">415</span><a name="line.415"></a>
-<span class="sourceLineNo">416</span> public static class Loader extends Configured implements Tool {<a name="line.416"></a>
-<span class="sourceLineNo">417</span><a name="line.417"></a>
-<span class="sourceLineNo">418</span> private static final Logger LOG = LoggerFactory.getLogger(Loader.class);<a name="line.418"></a>
-<span class="sourceLineNo">419</span> private static final String USAGE = "Loader <warInputDir | warFileList> <outputDir>";<a name="line.419"></a>
+<span class="sourceLineNo">194</span> if (outputDir == null) {<a name="line.194"></a>
+<span class="sourceLineNo">195</span> throw new IllegalArgumentException("Output directory not specified");<a name="line.195"></a>
+<span class="sourceLineNo">196</span> }<a name="line.196"></a>
+<span class="sourceLineNo">197</span> int res = runLoader(warcFileInputDir, outputDir);<a name="line.197"></a>
+<span class="sourceLineNo">198</span> if (res != 0) {<a name="line.198"></a>
+<span class="sourceLineNo">199</span> LOG.error("Loader failed");<a name="line.199"></a>
+<span class="sourceLineNo">200</span> return -1;<a name="line.200"></a>
+<span class="sourceLineNo">201</span> }<a name="line.201"></a>
+<span class="sourceLineNo">202</span> res = runVerify(outputDir);<a name="line.202"></a>
+<span class="sourceLineNo">203</span> } catch (Exception e) {<a name="line.203"></a>
+<span class="sourceLineNo">204</span> LOG.error("Tool failed with exception", e);<a name="line.204"></a>
+<span class="sourceLineNo">205</span> return -1;<a name="line.205"></a>
+<span class="sourceLineNo">206</span> }<a name="line.206"></a>
+<span class="sourceLineNo">207</span> return 0;<a name="line.207"></a>
+<span class="sourceLineNo">208</span> }<a name="line.208"></a>
+<span class="sourceLineNo">209</span><a name="line.209"></a>
+<span class="sourceLineNo">210</span> @Override<a name="line.210"></a>
+<span class="sourceLineNo">211</span> protected void processOptions(CommandLine cmd) {<a name="line.211"></a>
+<span class="sourceLineNo">212</span> processBaseOptions(cmd);<a name="line.212"></a>
+<span class="sourceLineNo">213</span> args = cmd.getArgs();<a name="line.213"></a>
+<span class="sourceLineNo">214</span> }<a name="line.214"></a>
+<span class="sourceLineNo">215</span><a name="line.215"></a>
+<span class="sourceLineNo">216</span> @Override<a name="line.216"></a>
+<span class="sourceLineNo">217</span> public void setUpCluster() throws Exception {<a name="line.217"></a>
+<span class="sourceLineNo">218</span> util = getTestingUtil(getConf());<a name="line.218"></a>
+<span class="sourceLineNo">219</span> boolean isDistributed = util.isDistributedCluster();<a name="line.219"></a>
+<span class="sourceLineNo">220</span> util.initializeCluster(isDistributed ? 1 : 3);<a name="line.220"></a>
+<span class="sourceLineNo">221</span> if (!isDistributed) {<a name="line.221"></a>
+<span class="sourceLineNo">222</span> util.startMiniMapReduceCluster();<a name="line.222"></a>
+<span class="sourceLineNo">223</span> }<a name="line.223"></a>
+<span class="sourceLineNo">224</span> this.setConf(util.getConfiguration());<a name="line.224"></a>
+<span class="sourceLineNo">225</span> }<a name="line.225"></a>
+<span class="sourceLineNo">226</span><a name="line.226"></a>
+<span class="sourceLineNo">227</span> @Override<a name="line.227"></a>
+<span class="sourceLineNo">228</span> public void cleanUpCluster() throws Exception {<a name="line.228"></a>
+<span class="sourceLineNo">229</span> super.cleanUpCluster();<a name="line.229"></a>
+<span class="sourceLineNo">230</span> if (util.isDistributedCluster()) {<a name="line.230"></a>
+<span class="sourceLineNo">231</span> util.shutdownMiniMapReduceCluster();<a name="line.231"></a>
+<span class="sourceLineNo">232</span> }<a name="line.232"></a>
+<span class="sourceLineNo">233</span> }<a name="line.233"></a>
+<span class="sourceLineNo">234</span><a name="line.234"></a>
+<span class="sourceLineNo">235</span> static TableName getTablename(Configuration c) {<a name="line.235"></a>
+<span class="sourceLineNo">236</span> return TableName.valueOf(c.get(TABLE_NAME_KEY, DEFAULT_TABLE_NAME));<a name="line.236"></a>
+<span class="sourceLineNo">237</span> }<a name="line.237"></a>
+<span class="sourceLineNo">238</span><a name="line.238"></a>
+<span class="sourceLineNo">239</span> @Override<a name="line.239"></a>
+<span class="sourceLineNo">240</span> public TableName getTablename() {<a name="line.240"></a>
+<span class="sourceLineNo">241</span> return getTablename(getConf());<a name="line.241"></a>
+<span class="sourceLineNo">242</span> }<a name="line.242"></a>
+<span class="sourceLineNo">243</span><a name="line.243"></a>
+<span class="sourceLineNo">244</span> @Override<a name="line.244"></a>
+<span class="sourceLineNo">245</span> protected Set<String> getColumnFamilies() {<a name="line.245"></a>
+<span class="sourceLineNo">246</span> Set<String> families = new HashSet<>();<a name="line.246"></a>
+<span class="sourceLineNo">247</span> families.add(Bytes.toString(CONTENT_FAMILY_NAME));<a name="line.247"></a>
+<span class="sourceLineNo">248</span> families.add(Bytes.toString(INFO_FAMILY_NAME));<a name="line.248"></a>
+<span class="sourceLineNo">249</span> return families;<a name="line.249"></a>
+<span class="sourceLineNo">250</span> }<a name="line.250"></a>
+<span class="sourceLineNo">251</span><a name="line.251"></a>
+<span class="sourceLineNo">252</span> @Override<a name="line.252"></a>
+<span class="sourceLineNo">253</span> public int runTestFromCommandLine() throws Exception {<a name="line.253"></a>
+<span class="sourceLineNo">254</span> return ToolRunner.run(getConf(), this, args);<a name="line.254"></a>
+<span class="sourceLineNo">255</span> }<a name="line.255"></a>
+<span class="sourceLineNo">256</span><a name="line.256"></a>
+<span class="sourceLineNo">257</span> public static void main(String[] args) throws Exception {<a name="line.257"></a>
+<span class="sourceLineNo">258</span> Configuration conf = HBaseConfiguration.create();<a name="line.258"></a>
+<span class="sourceLineNo">259</span> IntegrationTestingUtility.setUseDistributedCluster(conf);<a name="line.259"></a>
+<span class="sourceLineNo">260</span> int ret = ToolRunner.run(conf, new IntegrationTestLoadCommonCrawl(), args);<a name="line.260"></a>
+<span class="sourceLineNo">261</span> System.exit(ret);<a name="line.261"></a>
+<span class="sourceLineNo">262</span> }<a name="line.262"></a>
+<span class="sourceLineNo">263</span><a name="line.263"></a>
+<span class="sourceLineNo">264</span> public static class HBaseKeyWritable implements Writable {<a name="line.264"></a>
+<span class="sourceLineNo">265</span><a name="line.265"></a>
+<span class="sourceLineNo">266</span> private byte[] row;<a name="line.266"></a>
+<span class="sourceLineNo">267</span> private int rowOffset;<a name="line.267"></a>
+<span class="sourceLineNo">268</span> private int rowLength;<a name="line.268"></a>
+<span class="sourceLineNo">269</span> private byte[] family;<a name="line.269"></a>
+<span class="sourceLineNo">270</span> private int familyOffset;<a name="line.270"></a>
+<span class="sourceLineNo">271</span> private int familyLength;<a name="line.271"></a>
+<span class="sourceLineNo">272</span> private byte[] qualifier;<a name="line.272"></a>
+<span class="sourceLineNo">273</span> private int qualifierOffset;<a name="line.273"></a>
+<span class="sourceLineNo">274</span> private int qualifierLength;<a name="line.274"></a>
+<span class="sourceLineNo">275</span> private long ts;<a name="line.275"></a>
+<span class="sourceLineNo">276</span><a name="line.276"></a>
+<span class="sourceLineNo">277</span> public HBaseKeyWritable() { }<a name="line.277"></a>
+<span class="sourceLineNo">278</span><a name="line.278"></a>
+<span class="sourceLineNo">279</span> public HBaseKeyWritable(byte[] row, int rowOffset, int rowLength,<a name="line.279"></a>
+<span class="sourceLineNo">280</span> byte[] family, int familyOffset, int familyLength,<a name="line.280"></a>
+<span class="sourceLineNo">281</span> byte[] qualifier, int qualifierOffset, int qualifierLength, long ts) {<a name="line.281"></a>
+<span class="sourceLineNo">282</span> this.row = row;<a name="line.282"></a>
+<span class="sourceLineNo">283</span> this.rowOffset = rowOffset;<a name="line.283"></a>
+<span class="sourceLineNo">284</span> this.rowLength = rowLength;<a name="line.284"></a>
+<span class="sourceLineNo">285</span> this.family = family;<a name="line.285"></a>
+<span class="sourceLineNo">286</span> this.familyOffset = familyOffset;<a name="line.286"></a>
+<span class="sourceLineNo">287</span> this.familyLength = familyLength;<a name="line.287"></a>
+<span class="sourceLineNo">288</span> this.qualifier = qualifier;<a name="line.288"></a>
+<span class="sourceLineNo">289</span> this.qualifierOffset = qualifierOffset;<a name="line.289"></a>
+<span class="sourceLineNo">290</span> this.qualifierLength = qualifierLength;<a name="line.290"></a>
+<span class="sourceLineNo">291</span> this.ts = ts;<a name="line.291"></a>
+<span class="sourceLineNo">292</span> }<a name="line.292"></a>
+<span class="sourceLineNo">293</span><a name="line.293"></a>
+<span class="sourceLineNo">294</span> public HBaseKeyWritable(byte[] row, byte[] family, byte[] qualifier, long ts) {<a name="line.294"></a>
+<span class="sourceLineNo">295</span> this(row, 0, row.length,<a name="line.295"></a>
+<span class="sourceLineNo">296</span> family, 0, family.length,<a name="line.296"></a>
+<span class="sourceLineNo">297</span> qualifier, 0, qualifier != null ? qualifier.length : 0, ts);<a name="line.297"></a>
+<span class="sourceLineNo">298</span> }<a name="line.298"></a>
+<span class="sourceLineNo">299</span><a name="line.299"></a>
+<span class="sourceLineNo">300</span> public HBaseKeyWritable(byte[] row, byte[] family, long ts) {<a name="line.300"></a>
+<span class="sourceLineNo">301</span> this(row, family, HConstants.EMPTY_BYTE_ARRAY, ts);<a name="line.301"></a>
+<span class="sourceLineNo">302</span> }<a name="line.302"></a>
+<span class="sourceLineNo">303</span><a name="line.303"></a>
+<span class="sourceLineNo">304</span> public HBaseKeyWritable(Cell cell) {<a name="line.304"></a>
+<span class="sourceLineNo">305</span> this(cell.getRowArray(), cell.getRowOffset(), cell.getRowLength(),<a name="line.305"></a>
+<span class="sourceLineNo">306</span> cell.getFamilyArray(), cell.getFamilyOffset(), cell.getFamilyLength(),<a name="line.306"></a>
+<span class="sourceLineNo">307</span> cell.getQualifierArray(), cell.getQualifierOffset(), cell.getQualifierLength(),<a name="line.307"></a>
+<span class="sourceLineNo">308</span> cell.getTimestamp());<a name="line.308"></a>
+<span class="sourceLineNo">309</span> }<a name="line.309"></a>
+<span class="sourceLineNo">310</span><a name="line.310"></a>
+<span class="sourceLineNo">311</span> @Override<a name="line.311"></a>
+<span class="sourceLineNo">312</span> public void readFields(DataInput in) throws IOException {<a name="line.312"></a>
+<span class="sourceLineNo">313</span> this.row = Bytes.toBytes(in.readUTF());<a name="line.313"></a>
+<span class="sourceLineNo">314</span> this.rowOffset = 0;<a name="line.314"></a>
+<span class="sourceLineNo">315</span> this.rowLength = row.length;<a name="line.315"></a>
+<span class="sourceLineNo">316</span> this.family = Bytes.toBytes(in.readUTF());<a name="line.316"></a>
+<span class="sourceLineNo">317</span> this.familyOffset = 0;<a name="line.317"></a>
+<span class="sourceLineNo">318</span> this.familyLength = family.length;<a name="line.318"></a>
+<span class="sourceLineNo">319</span> this.qualifier = Bytes.toBytes(in.readUTF());<a name="line.319"></a>
+<span class="sourceLineNo">320</span> this.qualifierOffset = 0;<a name="line.320"></a>
+<span class="sourceLineNo">321</span> this.qualifierLength = qualifier.length;<a name="line.321"></a>
+<span class="sourceLineNo">322</span> this.ts = in.readLong();<a name="line.322"></a>
+<span class="sourceLineNo">323</span> }<a name="line.323"></a>
+<span class="sourceLineNo">324</span><a name="line.324"></a>
+<span class="sourceLineNo">325</span> @Override<a name="line.325"></a>
+<span class="sourceLineNo">326</span> public void write(DataOutput out) throws IOException {<a name="line.326"></a>
+<span class="sourceLineNo">327</span> out.writeUTF(new String(row, rowOffset, rowLength, StandardCharsets.UTF_8));<a name="line.327"></a>
+<span class="sourceLineNo">328</span> out.writeUTF(new String(family, familyOffset, familyLength, StandardCharsets.UTF_8));<a name="line.328"></a>
+<span class="sourceLineNo">329</span> if (qualifier != null) {<a name="line.329"></a>
+<span class="sourceLineNo">330</span> out.writeUTF(new String(qualifier, qualifierOffset, qualifierLength,<a name="line.330"></a>
+<span class="sourceLineNo">331</span> StandardCharsets.UTF_8));<a name="line.331"></a>
+<span class="sourceLineNo">332</span> } else {<a name="line.332"></a>
+<span class="sourceLineNo">333</span> out.writeUTF("");<a name="line.333"></a>
+<span class="sourceLineNo">334</span> }<a name="line.334"></a>
+<span class="sourceLineNo">335</span> out.writeLong(ts);<a name="line.335"></a>
+<span class="sourceLineNo">336</span> }<a name="line.336"></a>
+<span class="sourceLineNo">337</span><a name="line.337"></a>
+<span class="sourceLineNo">338</span> public byte[] getRowArray() {<a name="line.338"></a>
+<span class="sourceLineNo">339</span> return row;<a name="line.339"></a>
+<span class="sourceLineNo">340</span> }<a name="line.340"></a>
+<span class="sourceLineNo">341</span><a name="line.341"></a>
+<span class="sourceLineNo">342</span> public void setRow(byte[] row) {<a name="line.342"></a>
+<span class="sourceLineNo">343</span> this.row = row;<a name="line.343"></a>
+<span class="sourceLineNo">344</span> }<a name="line.344"></a>
+<span class="sourceLineNo">345</span><a name="line.345"></a>
+<span class="sourceLineNo">346</span> public int getRowOffset() {<a name="line.346"></a>
+<span class="sourceLineNo">347</span> return rowOffset;<a name="line.347"></a>
+<span class="sourceLineNo">348</span> }<a name="line.348"></a>
+<span class="sourceLineNo">349</span><a name="line.349"></a>
+<span class="sourceLineNo">350</span> public void setRowOffset(int rowOffset) {<a name="line.350"></a>
+<span class="sourceLineNo">351</span> this.rowOffset = rowOffset;<a name="line.351"></a>
+<span class="sourceLineNo">352</span> }<a name="line.352"></a>
+<span class="sourceLineNo">353</span><a name="line.353"></a>
+<span class="sourceLineNo">354</span> public int getRowLength() {<a name="line.354"></a>
+<span class="sourceLineNo">355</span> return rowLength;<a name="line.355"></a>
+<span class="sourceLineNo">356</span> }<a name="line.356"></a>
+<span class="sourceLineNo">357</span><a name="line.357"></a>
+<span class="sourceLineNo">358</span> public void setRowLength(int rowLength) {<a name="line.358"></a>
+<span class="sourceLineNo">359</span> this.rowLength = rowLength;<a name="line.359"></a>
+<span class="sourceLineNo">360</span> }<a name="line.360"></a>
+<span class="sourceLineNo">361</span><a name="line.361"></a>
+<span class="sourceLineNo">362</span> public byte[] getFamilyArray() {<a name="line.362"></a>
+<span class="sourceLineNo">363</span> return family;<a name="line.363"></a>
+<span class="sourceLineNo">364</span> }<a name="line.364"></a>
+<span class="sourceLineNo">365</span><a name="line.365"></a>
+<span class="sourceLineNo">366</span> public void setFamily(byte[] family) {<a name="line.366"></a>
+<span class="sourceLineNo">367</span> this.family = family;<a name="line.367"></a>
+<span class="sourceLineNo">368</span> }<a name="line.368"></a>
+<span class="sourceLineNo">369</span><a name="line.369"></a>
+<span class="sourceLineNo">370</span> public int getFamilyOffset() {<a name="line.370"></a>
+<span class="sourceLineNo">371</span> return familyOffset;<a name="line.371"></a>
+<span class="sourceLineNo">372</span> }<a name="line.372"></a>
+<span class="sourceLineNo">373</span><a name="line.373"></a>
+<span class="sourceLineNo">374</span> public void setFamilyOffset(int familyOffset) {<a name="line.374"></a>
+<span class="sourceLineNo">375</span> this.familyOffset = familyOffset;<a name="line.375"></a>
+<span class="sourceLineNo">376</span> }<a name="line.376"></a>
+<span class="sourceLineNo">377</span><a name="line.377"></a>
+<span class="sourceLineNo">378</span> public int getFamilyLength() {<a name="line.378"></a>
+<span class="sourceLineNo">379</span> return familyLength;<a name="line.379"></a>
+<span class="sourceLineNo">380</span> }<a name="line.380"></a>
+<span class="sourceLineNo">381</span><a name="line.381"></a>
+<span class="sourceLineNo">382</span> public void setFamilyLength(int familyLength) {<a name="line.382"></a>
+<span class="sourceLineNo">383</span> this.familyLength = familyLength;<a name="line.383"></a>
+<span class="sourceLineNo">384</span> }<a name="line.384"></a>
+<span class="sourceLineNo">385</span><a name="line.385"></a>
+<span class="sourceLineNo">386</span> public byte[] getQualifierArray() {<a name="line.386"></a>
+<span class="sourceLineNo">387</span> return qualifier;<a name="line.387"></a>
+<span class="sourceLineNo">388</span> }<a name="line.388"></a>
+<span class="sourceLineNo">389</span><a name="line.389"></a>
+<span class="sourceLineNo">390</span> public void setQualifier(byte[] qualifier) {<a name="line.390"></a>
+<span class="sourceLineNo">391</span> this.qualifier = qualifier;<a name="line.391"></a>
+<span class="sourceLineNo">392</span> }<a name="line.392"></a>
+<span class="sourceLineNo">393</span><a name="line.393"></a>
+<span class="sourceLineNo">394</span> public int getQualifierOffset() {<a name="line.394"></a>
+<span class="sourceLineNo">395</span> return qualifierOffset;<a name="line.395"></a>
+<span class="sourceLineNo">396</span> }<a name="line.396"></a>
+<span class="sourceLineNo">397</span><a name="line.397"></a>
+<span class="sourceLineNo">398</span> public void setQualifierOffset(int qualifierOffset) {<a name="line.398"></a>
+<span class="sourceLineNo">399</span> this.qualifierOffset = qualifierOffset;<a name="line.399"></a>
+<span class="sourceLineNo">400</span> }<a name="line.400"></a>
+<span class="sourceLineNo">401</span><a name="line.401"></a>
+<span class="sourceLineNo">402</span> public int getQualifierLength() {<a name="line.402"></a>
+<span class="sourceLineNo">403</span> return qualifierLength;<a name="line.403"></a>
+<span class="sourceLineNo">404</span> }<a name="line.404"></a>
+<span class="sourceLineNo">405</span><a name="line.405"></a>
+<span class="sourceLineNo">406</span> public void setQualifierLength(int qualifierLength) {<a name="line.406"></a>
+<span class="sourceLineNo">407</span> this.qualifierLength = qualifierLength;<a name="line.407"></a>
+<span class="sourceLineNo">408</span> }<a name="line.408"></a>
+<span class="sourceLineNo">409</span><a name="line.409"></a>
+<span class="sourceLineNo">410</span> public long getTimestamp() {<a name="line.410"></a>
+<span class="sourceLineNo">411</span> return ts;<a name="line.411"></a>
+<span class="sourceLineNo">412</span> }<a name="line.412"></a>
+<span class="sourceLineNo">413</span><a name="line.413"></a>
+<span class="sourceLineNo">414</span> public void setTimestamp(long ts) {<a name="line.414"></a>
+<span class="sourceLineNo">415</span> this.ts = ts;<a name="line.415"></a>
+<span class="sourceLineNo">416</span> }<a name="line.416"></a>
+<span class="sourceLineNo">417</span> }<a name="line.417"></a>
+<span class="sourceLineNo">418</span><a name="line.418"></a>
+<span class="sourceLineNo">419</span> public static class Loader extends Configured implements Tool {<a name="line.419"></a>
<span class="sourceLineNo">420</span><a name="line.420"></a>
-<span class="sourceLineNo">421</span> void createSchema(TableName tableName) throws IOException {<a name="line.421"></a>
-<span class="sourceLineNo">422</span><a name="line.422"></a>
-<span class="sourceLineNo">423</span> try (Connection conn = ConnectionFactory.createConnection(getConf());<a name="line.423"></a>
-<span class="sourceLineNo">424</span> Admin admin = conn.getAdmin()) {<a name="line.424"></a>
-<span class="sourceLineNo">425</span> if (!admin.tableExists(tableName)) {<a name="line.425"></a>
-<span class="sourceLineNo">426</span><a name="line.426"></a>
-<span class="sourceLineNo">427</span> ColumnFamilyDescriptorBuilder contentFamilyBuilder =<a name="line.427"></a>
-<span class="sourceLineNo">428</span> ColumnFamilyDescriptorBuilder.newBuilder(CONTENT_FAMILY_NAME)<a name="line.428"></a>
-<span class="sourceLineNo">429</span> .setDataBlockEncoding(DataBlockEncoding.NONE)<a name="line.429"></a>
-<span class="sourceLineNo">430</span> .setBloomFilterType(BloomType.ROW)<a name="line.430"></a>
-<span class="sourceLineNo">431</span> .setMaxVersions(1000)<a name="line.431"></a>
-<span class="sourceLineNo">432</span> .setBlocksize(256 * 1024)<a name="line.432"></a>
-<span class="sourceLineNo">433</span> ;<a name="line.433"></a>
-<span class="sourceLineNo">434</span><a name="line.434"></a>
-<span class="sourceLineNo">435</span> ColumnFamilyDescriptorBuilder infoFamilyBuilder =<a name="line.435"></a>
-<span class="sourceLineNo">436</span> ColumnFamilyDescriptorBuilder.newBuilder(INFO_FAMILY_NAME)<a name="line.436"></a>
-<span class="sourceLineNo">437</span> .setDataBlockEncoding(DataBlockEncoding.NONE)<a name="line.437"></a>
-<span class="sourceLineNo">438</span> .setBloomFilterType(BloomType.ROWCOL)<a name="line.438"></a>
-<span class="sourceLineNo">439</span> .setMaxVersions(1000)<a name="line.439"></a>
-<span class="sourceLineNo">440</span> .setBlocksize(8 * 1024)<a name="line.440"></a>
-<span class="sourceLineNo">441</span> ;<a name="line.441"></a>
-<span class="sourceLineNo">442</span><a name="line.442"></a>
-<span class="sourceLineNo">443</span> Set<ColumnFamilyDescriptor> families = new HashSet<>();<a name="line.443"></a>
-<span class="sourceLineNo">444</span> families.add(contentFamilyBuilder.build());<a name="line.444"></a>
-<span class="sourceLineNo">445</span> families.add(infoFamilyBuilder.build());<a name="line.445"></a>
-<span class="sourceLineNo">446</span><a name="line.446"></a>
-<span class="sourceLineNo">447</span> TableDescriptor tableDescriptor = TableDescriptorBuilder.newBuilder(tableName)<a name="line.447"></a>
-<span class="sourceLineNo">448</span> .setColumnFamilies(families)<a name="line.448"></a>
-<span class="sourceLineNo">449</span> .build();<a name="line.449"></a>
-<span class="sourceLineNo">450</span><a name="line.450"></a>
-<span class="sourceLineNo">451</span> if (getConf().getBoolean(HBaseTestingUtil.PRESPLIT_TEST_TABLE_KEY,<a name="line.451"></a>
-<span class="sourceLineNo">452</span> HBaseTestingUtil.PRESPLIT_TEST_TABLE)) {<a name="line.452"></a>
-<span class="sourceLineNo">453</span> int numberOfServers = admin.getRegionServers().size();<a name="line.453"></a>
-<span class="sourceLineNo">454</span> if (numberOfServers == 0) {<a name="line.454"></a>
-<span class="sourceLineNo">455</span> throw new IllegalStateException("No live regionservers");<a name="line.455"></a>
-<span class="sourceLineNo">456</span> }<a name="line.456"></a>
-<span class="sourceLineNo">457</span> int regionsPerServer = getConf().getInt(HBaseTestingUtil.REGIONS_PER_SERVER_KEY,<a name="line.457"></a>
-<span class="sourceLineNo">458</span> HBaseTestingUtil.DEFAULT_REGIONS_PER_SERVER);<a name="line.458"></a>
-<span class="sourceLineNo">459</span> int totalNumberOfRegions = numberOfServers * regionsPerServer;<a name="line.459"></a>
-<span class="sourceLineNo">460</span> LOG.info("Creating test table: " + tableDescriptor);<a name="line.460"></a>
-<span class="sourceLineNo">461</span> LOG.info("Number of live regionservers: " + numberOfServers + ", " +<a name="line.461"></a>
-<span class="sourceLineNo">462</span> "pre-splitting table into " + totalNumberOfRegions + " regions " +<a name="line.462"></a>
-<span class="sourceLineNo">463</span> "(default regions per server: " + regionsPerServer + ")");<a name="line.463"></a>
-<span class="sourceLineNo">464</span> byte[][] splits = new RegionSplitter.UniformSplit().split(totalNumberOfRegions);<a name="line.464"></a>
-<span class="sourceLineNo">465</span> admin.createTable(tableDescriptor, splits);<a name="line.465"></a>
-<span class="sourceLineNo">466</span> } else {<a name="line.466"></a>
-<span class="sourceLineNo">467</span> LOG.info("Creating test table: " + tableDescriptor);<a name="line.467"></a>
-<span class="sourceLineNo">468</span> admin.createTable(tableDescriptor);<a name="line.468"></a>
-<span class="sourceLineNo">469</span> }<a name="line.469"></a>
-<span class="sourceLineNo">470</span> }<a name="line.470"></a>
-<span class="sourceLineNo">471</span> } catch (MasterNotRunningException e) {<a name="line.471"></a>
-<span class="sourceLineNo">472</span> LOG.error("Master not running", e);<a name="line.472"></a>
-<span class="sourceLineNo">473</span> throw new IOException(e);<a name="line.473"></a>
-<span class="sourceLineNo">474</span> }<a name="line.474"></a>
-<span class="sourceLineNo">475</span> }<a name="line.475"></a>
-<span class="sourceLineNo">476</span><a name="line.476"></a>
-<span class="sourceLineNo">477</span> int run(Path warcFileInput, Path outputDir)<a name="line.477"></a>
-<span class="sourceLineNo">478</span> throws IOException, ClassNotFoundException, InterruptedException {<a name="line.478"></a>
+<span class="sourceLineNo">421</span> private static final Logger LOG = LoggerFactory.getLogger(Loader.class);<a name="line.421"></a>
+<span class="sourceLineNo">422</span> private static final String USAGE = "Loader <warInputDir | warFileList> <outputDir>";<a name="line.422"></a>
+<span class="sourceLineNo">423</span><a name="line.423"></a>
+<span class="sourceLineNo">424</span> void createSchema(TableName tableName) throws IOException {<a name="line.424"></a>
+<span class="sourceLineNo">425</span><a name="line.425"></a>
+<span class="sourceLineNo">426</span> try (Connection conn = ConnectionFactory.createConnection(getConf());<a name="line.426"></a>
+<span class="sourceLineNo">427</span> Admin admin = conn.getAdmin()) {<a name="line.427"></a>
+<span class="sourceLineNo">428</span> if (!admin.tableExists(tableName)) {<a name="line.428"></a>
+<span class="sourceLineNo">429</span><a name="line.429"></a>
+<span class="sourceLineNo">430</span> ColumnFamilyDescriptorBuilder contentFamilyBuilder =<a name="line.430"></a>
+<span class="sourceLineNo">431</span> ColumnFamilyDescriptorBuilder.newBuilder(CONTENT_FAMILY_NAME)<a name="line.431"></a>
+<span class="sourceLineNo">432</span> .setDataBlockEncoding(DataBlockEncoding.NONE)<a name="line.432"></a>
+<span class="sourceLineNo">433</span> .setBloomFilterType(BloomType.ROW)<a name="line.433"></a>
+<span class="sourceLineNo">434</span> .setMaxVersions(1000)<a name="line.434"></a>
+<span class="sourceLineNo">435</span> .setBlocksize(256 * 1024)<a name="line.435"></a>
+<span class="sourceLineNo">436</span> ;<a name="line.436"></a>
+<span class="sourceLineNo">437</span><a name="line.437"></a>
+<span class="sourceLineNo">438</span> ColumnFamilyDescriptorBuilder infoFamilyBuilder =<a name="line.438"></a>
+<span class="sourceLineNo">439</span> ColumnFamilyDescriptorBuilder.newBuilder(INFO_FAMILY_NAME)<a name="line.439"></a>
+<span class="sourceLineNo">440</span> .setDataBlockEncoding(DataBlockEncoding.NONE)<a name="line.440"></a>
+<span class="sourceLineNo">441</span> .setBloomFilterType(BloomType.ROWCOL)<a name="line.441"></a>
+<span class="sourceLineNo">442</span> .setMaxVersions(1000)<a name="line.442"></a>
+<span class="sourceLineNo">443</span> .setBlocksize(8 * 1024)<a name="line.443"></a>
+<span class="sourceLineNo">444</span> ;<a name="line.444"></a>
+<span class="sourceLineNo">445</span><a name="line.445"></a>
+<span class="sourceLineNo">446</span> Set<ColumnFamilyDescriptor> families = new HashSet<>();<a name="line.446"></a>
+<span class="sourceLineNo">447</span> families.add(contentFamilyBuilder.build());<a name="line.447"></a>
+<span class="sourceLineNo">448</span> families.add(infoFamilyBuilder.build());<a name="line.448"></a>
+<span class="sourceLineNo">449</span><a name="line.449"></a>
+<span class="sourceLineNo">450</span> TableDescriptor tableDescriptor = TableDescriptorBuilder.newBuilder(tableName)<a name="line.450"></a>
+<span class="sourceLineNo">451</span> .setColumnFamilies(families)<a name="line.451"></a>
+<span class="sourceLineNo">452</span> .build();<a name="line.452"></a>
+<span class="sourceLineNo">453</span><a name="line.453"></a>
+<span class="sourceLineNo">454</span> if (getConf().getBoolean(HBaseTestingUtil.PRESPLIT_TEST_TABLE_KEY,<a name="line.454"></a>
+<span class="sourceLineNo">455</span> HBaseTestingUtil.PRESPLIT_TEST_TABLE)) {<a name="line.455"></a>
+<span class="sourceLineNo">456</span> int numberOfServers = admin.getRegionServers().size();<a name="line.456"></a>
+<span class="sourceLineNo">457</span> if (numberOfServers == 0) {<a name="line.457"></a>
+<span class="sourceLineNo">458</span> throw new IllegalStateException("No live regionservers");<a name="line.458"></a>
+<span class="sourceLineNo">459</span> }<a name="line.459"></a>
+<span class="sourceLineNo">460</span> int regionsPerServer = getConf().getInt(HBaseTestingUtil.REGIONS_PER_SERVER_KEY,<a name="line.460"></a>
+<span class="sourceLineNo">461</span> HBaseTestingUtil.DEFAULT_REGIONS_PER_SERVER);<a name="line.461"></a>
+<span class="sourceLineNo">462</span> int totalNumberOfRegions = numberOfServers * regionsPerServer;<a name="line.462"></a>
+<span class="sourceLineNo">463</span> LOG.info("Creating test table: " + tableDescriptor);<a name="line.463"></a>
+<span class="sourceLineNo">464</span> LOG.info("Number of live regionservers: " + numberOfServers + ", " +<a name="line.464"></a>
+<span class="sourceLineNo">465</span> "pre-splitting table into " + totalNumberOfRegions + " regions " +<a name="line.465"></a>
+<span class="sourceLineNo">466</span> "(default regions per server: " + regionsPerServer + ")");<a name="line.466"></a>
+<span class="sourceLineNo">467</span> byte[][] splits = new RegionSplitter.UniformSplit().split(totalNumberOfRegions);<a name="line.467"></a>
+<span class="sourceLineNo">468</span> admin.createTable(tableDescriptor, splits);<a name="line.468"></a>
+<span class="sourceLineNo">469</span> } else {<a name="line.469"></a>
+<span class="sourceLineNo">470</span> LOG.info("Creating test table: " + tableDescriptor);<a name="line.470"></a>
+<span class="sourceLineNo">471</span> admin.createTable(tableDescriptor);<a name="line.471"></a>
+<span class="sourceLineNo">472</span> }<a name="line.472"></a>
+<span class="sourceLineNo">473</span> }<a name="line.473"></a>
+<span class="sourceLineNo">474</span> } catch (MasterNotRunningException e) {<a name="line.474"></a>
+<span class="sourceLineNo">475</span> LOG.error("Master not running", e);<a name="line.475"></a>
+<span class="sourceLineNo">476</span> throw new IOException(e);<a name="line.476"></a>
+<span class="sourceLineNo">477</span> }<a name="line.477"></a>
+<span class="sourceLineNo">478</span> }<a name="line.478"></a>
<span class="sourceLineNo">479</span><a name="line.479"></a>
-<span class="sourceLineNo">480</span> createSchema(getTablename(getConf()));<a name="line.480"></a>
-<span class="sourceLineNo">481</span><a name="line.481"></a>
-<span class="sourceLineNo">482</span> Job job = Job.getInstance(getConf());<a name="line.482"></a>
-<span class="sourceLineNo">483</span> job.setJobName(Loader.class.getName());<a name="line.483"></a>
-<span class="sourceLineNo">484</span> job.setNumReduceTasks(0);<a name="line.484"></a>
-<span class="sourceLineNo">485</span> job.setJarByClass(getClass());<a name="line.485"></a>
-<span class="sourceLineNo">486</span> job.setMapperClass(LoaderMapper.class);<a name="line.486"></a>
-<span class="sourceLineNo">487</span> job.setInputFormatClass(WARCInputFormat.class);<a name="line.487"></a>
-<span class="sourceLineNo">488</span> FileSystem fs = FileSystem.get(warcFileInput.toUri(), getConf());<a name="line.488"></a>
-<span class="sourceLineNo">489</span> if (fs.getFileStatus(warcFileInput).isDirectory()) {<a name="line.489"></a>
-<span class="sourceLineNo">490</span> LOG.info("Using directory as WARC input path: " + warcFileInput);<a name="line.490"></a>
-<span class="sourceLineNo">491</span> FileInputFormat.setInputPaths(job, warcFileInput);<a name="line.491"></a>
-<span class="sourceLineNo">492</span> } else {<a name="line.492"></a>
-<span class="sourceLineNo">493</span> LOG.info("Getting WARC input paths from file: " + warcFileInput);<a name="line.493"></a>
-<span class="sourceLineNo">494</span> List<Path> paths = new LinkedList<Path>();<a name="line.494"></a>
-<span class="sourceLineNo">495</span> try (FSDataInputStream is = fs.open(warcFileInput)) {<a name="line.495"></a>
-<span class="sourceLineNo">496</span> InputStreamReader reader;<a name="line.496"></a>
-<span class="sourceLineNo">497</span> if (warcFileInput.getName().toLowerCase().endsWith(".gz")) {<a name="line.497"></a>
-<span class="sourceLineNo">498</span> reader = new InputStreamReader(new GZIPInputStream(is));<a name="line.498"></a>
-<span class="sourceLineNo">499</span> } else {<a name="line.499"></a>
-<span class="sourceLineNo">500</span> reader = new InputStreamReader(is);<a name="line.500"></a>
-<span class="sourceLineNo">501</span> }<a name="line.501"></a>
-<span class="sourceLineNo">502</span> try (BufferedReader br = new BufferedReader(reader)) {<a name="line.502"></a>
-<span class="sourceLineNo">503</span> String line;<a name="line.503"></a>
-<span class="sourceLineNo">504</span> while ((line = br.readLine()) != null) {<a name="line.504"></a>
-<span class="sourceLineNo">505</span> paths.add(new Path(line));<a name="line.505"></a>
-<span class="sourceLineNo">506</span> }<a name="line.506"></a>
-<span class="sourceLineNo">507</span> }<a name="line.507"></a>
-<span class="sourceLineNo">508</span> }<a name="line.508"></a>
-<span class="sourceLineNo">509</span> LOG.info("Read " + paths.size() + " WARC input paths from " + warcFileInput);<a name="line.509"></a>
-<span class="sourceLineNo">510</span> FileInputFormat.setInputPaths(job, paths.toArray(new Path[paths.size()]));<a name="line.510"></a>
-<span class="sourceLineNo">511</span> }<a name="line.511"></a>
-<span class="sourceLineNo">512</span> job.setOutputFormatClass(SequenceFileOutputFormat.class);<a name="line.512"></a>
-<span class="sourceLineNo">513</span> SequenceFileOutputFormat.setOutputPath(job, outputDir);<a name="line.513"></a>
-<span class="sourceLineNo">514</span> SequenceFileOutputFormat.setOutputCompressionType(job, CompressionType.BLOCK);<a name="line.514"></a>
-<span class="sourceLineNo">515</span> job.setOutputKeyClass(HBaseKeyWritable.class);<a name="line.515"></a>
-<span class="sourceLineNo">516</span> job.setOutputValueClass(BytesWritable.class);<a name="line.516"></a>
-<span class="sourceLineNo">517</span> TableMapReduceUtil.addDependencyJars(job);<a name="line.517"></a>
-<span class="sourceLineNo">518</span><a name="line.518"></a>
-<span class="sourceLineNo">519</span> LOG.info("Submitting job." +<a name="line.519"></a>
-<span class="sourceLineNo">520</span> " This will take time proportional to the number of input files, please be patient.");<a name="line.520"></a>
-<span class="sourceLineNo">521</span> boolean success = job.waitForCompletion(true);<a name="line.521"></a>
-<span class="sourceLineNo">522</span> if (!success) {<a name="line.522"></a>
-<span class="sourceLineNo">523</span> LOG.error("Failure during job " + job.getJobID());<a name="line.523"></a>
-<span class="sourceLineNo">524</span> }<a name="line.524"></a>
-<span class="sourceLineNo">525</span> return success ? 0 : 1;<a name="line.525"></a>
-<span class="sourceLineNo">526</span> }<a name="line.526"></a>
-<span class="sourceLineNo">527</span><a name="line.527"></a>
-<span class="sourceLineNo">528</span> @Override<a name="line.528"></a>
-<span class="sourceLineNo">529</span> public int run(String[] args) throws Exception {<a name="line.529"></a>
-<span class="sourceLineNo">530</span> if (args.length < 2) {<a name="line.530"></a>
-<span class="sourceLineNo">531</span> System.err.println(USAGE);<a name="line.531"></a>
-<span class="sourceLineNo">532</span> return 1;<a name="line.532"></a>
-<span class="sourceLineNo">533</span> }<a name="line.533"></a>
-<span class="sourceLineNo">534</span> try {<a name="line.534"></a>
-<span class="sourceLineNo">535</span> Path warcFileInput = new Path(args[0]);<a name="line.535"></a>
-<span class="sourceLineNo">536</span> Path outputDir = new Path(args[1]);<a name="line.536"></a>
-<span class="sourceLineNo">537</span> return run(warcFileInput, outputDir);<a name="line.537"></a>
-<span class="sourceLineNo">538</span> } catch (NumberFormatException e) {<a name="line.538"></a>
-<span class="sourceLineNo">539</span> System.err.println("Parsing loader arguments failed: " + e.getMessage());<a name="line.539"></a>
-<span class="sourceLineNo">540</span> System.err.println(USAGE);<a name="line.540"></a>
-<span class="sourceLineNo">541</span> return 1;<a name="line.541"></a>
-<span class="sourceLineNo">542</span> }<a name="line.542"></a>
-<span class="sourceLineNo">543</span> }<a name="line.543"></a>
-<span class="sourceLineNo">544</span><a name="line.544"></a>
-<span class="sourceLineNo">545</span> public static void main(String[] args) throws Exception {<a name="line.545"></a>
-<span class="sourceLineNo">546</span> System.exit(ToolRunner.run(HBaseConfiguration.create(), new Loader(), args));<a name="line.546"></a>
-<span class="sourceLineNo">547</span> }<a name="line.547"></a>
-<span class="sourceLineNo">548</span><a name="line.548"></a>
-<span class="sourceLineNo">549</span> public static class LoaderMapper<a name="line.549"></a>
-<span class="sourceLineNo">550</span> extends Mapper<LongWritable, WARCWritable, HBaseKeyWritable, BytesWritable> {<a name="line.550"></a>
-<span class="sourceLineNo">551</span><a name="line.551"></a>
-<span class="sourceLineNo">552</span> Configuration conf;<a name="line.552"></a>
-<span class="sourceLineNo">553</span> Connection conn;<a name="line.553"></a>
-<span class="sourceLineNo">554</span> Table table;<a name="line.554"></a>
-<span class="sourceLineNo">555</span><a name="line.555"></a>
-<span class="sourceLineNo">556</span> @Override<a name="line.556"></a>
-<span class="sourceLineNo">557</span> protected void setup(Context context) throws IOException, InterruptedException {<a name="line.557"></a>
-<span class="sourceLineNo">558</span> conn = ConnectionFactory.createConnection(context.getConfiguration());<a name="line.558"></a>
-<span class="sourceLineNo">559</span> table = conn.getTable(getTablename(conn.getConfiguration()));<a name="line.559"></a>
-<span class="sourceLineNo">560</span> }<a name="line.560"></a>
-<span class="sourceLineNo">561</span><a name="line.561"></a>
-<span class="sourceLineNo">562</span> @Override<a name="line.562"></a>
-<span class="sourceLineNo">563</span> protected void cleanup(Context context) throws IOException, InterruptedException {<a name="line.563"></a>
-<span class="sourceLineNo">564</span> try {<a name="line.564"></a>
-<span class="sourceLineNo">565</span> table.close();<a name="line.565"></a>
-<span class="sourceLineNo">566</span> } catch (Exception e) {<a name="line.566"></a>
-<span class="sourceLineNo">567</span> LOG.warn("Exception closing Table", e);<a name="line.567"></a>
-<span class="sourceLineNo">568</span> }<a name="line.568"></a>
-<span class="sourceLineNo">569</span> try {<a name="line.569"></a>
-<span class="sourceLineNo">570</span> conn.close();<a name="line.570"></a>
-<span class="sourceLineNo">571</span> } catch (Exception e) {<a name="line.571"></a>
-<span class="sourceLineNo">572</span> LOG.warn("Exception closing Connection", e);<a name="line.572"></a>
-<span class="sourceLineNo">573</span> }<a name="line.573"></a>
-<span class="sourceLineNo">574</span> }<a name="line.574"></a>
-<span class="sourceLineNo">575</span><a name="line.575"></a>
-<span class="sourceLineNo">576</span> @Override<a name="line.576"></a>
-<span class="sourceLineNo">577</span> protected void map(LongWritable key, WARCWritable value, Context output)<a name="line.577"></a>
-<span class="sourceLineNo">578</span> throws IOException, InterruptedException {<a name="line.578"></a>
-<span class="sourceLineNo">579</span> WARCRecord.Header warcHeader = value.getRecord().getHeader();<a name="line.579"></a>
-<span class="sourceLineNo">580</span> String recordID = warcHeader.getRecordID();<a name="line.580"></a>
-<span class="sourceLineNo">581</span> String targetURI = warcHeader.getTargetURI();<a name="line.581"></a>
-<span class="sourceLineNo">582</span> if (warcHeader.getRecordType().equals("response") && targetURI != null) {<a name="line.582"></a>
-<span class="sourceLineNo">583</span> String contentType = warcHeader.getField("WARC-Identified-Payload-Type");<a name="line.583"></a>
-<span class="sourceLineNo">584</span> if (contentType != null) {<a name="line.584"></a>
-<span class="sourceLineNo">585</span> LOG.debug("Processing record id=" + recordID + ", targetURI=\"" + targetURI + "\"");<a name="line.585"></a>
-<span class="sourceLineNo">586</span> long now = EnvironmentEdgeManager.currentTime();<a name="line.586"></a>
-<span class="sourceLineNo">587</span><a name="line.587"></a>
-<span class="sourceLineNo">588</span> // Make row key<a name="line.588"></a>
-<span class="sourceLineNo">589</span><a name="line.589"></a>
-<span class="sourceLineNo">590</span> byte[] rowKey;<a name="line.590"></a>
-<span class="sourceLineNo">591</span> try {<a name="line.591"></a>
-<span class="sourceLineNo">592</span> rowKey = rowKeyFromTargetURI(targetURI);<a name="line.592"></a>
-<span class="sourceLineNo">593</span> } catch (IllegalArgumentException e) {<a name="line.593"></a>
-<span class="sourceLineNo">594</span> LOG.debug("Could not make a row key for record " + recordID + ", ignoring", e);<a name="line.594"></a>
-<span class="sourceLineNo">595</span> return;<a name="line.595"></a>
-<span class="sourceLineNo">596</span> } catch (URISyntaxException e) {<a name="line.596"></a>
-<span class="sourceLineNo">597</span> LOG.warn("Could not parse URI \"" + targetURI + "\" for record " + recordID +<a name="line.597"></a>
-<span class="sourceLineNo">598</span> ", ignoring");<a name="line.598"></a>
-<span class="sourceLineNo">599</span> return;<a name="line.599"></a>
-<span class="sourceLineNo">600</span> }<a name="line.600"></a>
-<span class="sourceLineNo">601</span><a name="line.601"></a>
-<span class="sourceLineNo">602</span> // Get the content and calculate the CRC64<a name="line.602"></a>
-<span class="sourceLineNo">603</span><a name="line.603"></a>
-<span class="sourceLineNo">604</span> byte[] content = value.getRecord().getContent();<a name="line.604"></a>
-<span class="sourceLineNo">605</span> CRC64 crc = new CRC64();<a name="line.605"></a>
-<span class="sourceLineNo">606</span> crc.update(content);<a name="line.606"></a>
-<span class="sourceLineNo">607</span> long crc64 = crc.getValue();<a name="line.607"></a>
-<span class="sourceLineNo">608</span><a name="line.608"></a>
-<span class="sourceLineNo">609</span> // Store to HBase<a name="line.609"></a>
-<span class="sourceLineNo">610</span><a name="line.610"></a>
-<span class="sourceLineNo">611</span> Put put = new Put(rowKey);<a name="line.611"></a>
-<span class="sourceLineNo">612</span> put.addColumn(CONTENT_FAMILY_NAME, CONTENT_QUALIFIER, now, content);<a name="line.612"></a>
-<span class="sourceLineNo">613</span> put.addColumn(INFO_FAMILY_NAME, CONTENT_LENGTH_QUALIFIER, now,<a name="line.613"></a>
-<span class="sourceLineNo">614</span> Bytes.toBytes(content.length));<a name="line.614"></a>
-<span class="sourceLineNo">615</span> put.addColumn(INFO_FAMILY_NAME, CONTENT_TYPE_QUALIFIER, now,<a name="line.615"></a>
-<span class="sourceLineNo">616</span> Bytes.toBytes(contentType));<a name="line.616"></a>
-<span class="sourceLineNo">617</span> put.addColumn(INFO_FAMILY_NAME, CRC_QUALIFIER, now, Bytes.toBytes(crc64));<a name="line.617"></a>
-<span class="sourceLineNo">618</span> put.addColumn(INFO_FAMILY_NAME, RECORD_ID_QUALIFIER, now, Bytes.toBytes(recordID));<a name="line.618"></a>
-<span class="sourceLineNo">619</span> put.addColumn(INFO_FAMILY_NAME, TARGET_URI_QUALIFIER, now, Bytes.toBytes(targetURI));<a name="line.619"></a>
-<span class="sourceLineNo">620</span> put.addColumn(INFO_FAMILY_NAME, DATE_QUALIFIER, now,<a name="line.620"></a>
-<span class="sourceLineNo">621</span> Bytes.toBytes(warcHeader.getDateString()));<a name="line.621"></a>
-<span class="sourceLineNo">622</span> String ipAddr = warcHeader.getField("WARC-IP-Address");<a name="line.622"></a>
-<span class="sourceLineNo">623</span> if (ipAddr != null) {<a name="line.623"></a>
-<span class="sourceLineNo">624</span> put.addColumn(INFO_FAMILY_NAME, IP_ADDRESS_QUALIFIER, now, Bytes.toBytes(ipAddr));<a name="line.624"></a>
-<span class="sourceLineNo">625</span> }<a name="line.625"></a>
-<span class="sourceLineNo">626</span> table.put(put);<a name="line.626"></a>
-<span class="sourceLineNo">627</span><a name="line.627"></a>
-<span class="sourceLineNo">628</span> // Write records out for later verification, one per HBase field except for the<a name="line.628"></a>
-<span class="sourceLineNo">629</span> // content record, which will be verified by CRC64.<a name="line.629"></a>
+<span class="sourceLineNo">480</span> int run(Path warcFileInput, Path outputDir)<a name="line.480"></a>
+<span class="sourceLineNo">481</span> throws IOException, ClassNotFoundException, InterruptedException {<a name="line.481"></a>
+<span class="sourceLineNo">482</span><a name="line.482"></a>
+<span class="sourceLineNo">483</span> createSchema(getTablename(getConf()));<a name="line.483"></a>
+<span class="sourceLineNo">484</span><a name="line.484"></a>
+<span class="sourceLineNo">485</span> Job job = Job.getInstance(getConf());<a name="line.485"></a>
+<span class="sourceLineNo">486</span> job.setJobName(Loader.class.getName());<a name="line.486"></a>
+<span class="sourceLineNo">487</span> job.setNumReduceTasks(0);<a name="line.487"></a>
+<span class="sourceLineNo">488</span> job.setJarByClass(getClass());<a name="line.488"></a>
+<span class="sourceLineNo">489</span> job.setMapperClass(LoaderMapper.class);<a name="line.489"></a>
+<span class="sourceLineNo">490</span> job.setInputFormatClass(WARCInputFormat.class);<a name="line.490"></a>
+<span class="sourceLineNo">491</span> FileSystem fs = FileSystem.get(warcFileInput.toUri(), getConf());<a name="line.491"></a>
+<span class="sourceLineNo">492</span> if (fs.getFileStatus(warcFileInput).isDirectory()) {<a name="line.492"></a>
+<span class="sourceLineNo">493</span> LOG.info("Using directory as WARC input path: " + warcFileInput);<a name="line.493"></a>
+<span class="sourceLineNo">494</span> FileInputFormat.setInputPaths(job, warcFileInput);<a name="line.494"></a>
+<span class="sourceLineNo">495</span> } else {<a name="line.495"></a>
+<span class="sourceLineNo">496</span> LOG.info("Getting WARC input paths from file: " + warcFileInput);<a name="line.496"></a>
+<span class="sourceLineNo">497</span> List<Path> paths = new LinkedList<Path>();<a name="line.497"></a>
+<span class="sourceLineNo">498</span> try (FSDataInputStream is = fs.open(warcFileInput)) {<a name="line.498"></a>
+<span class="sourceLineNo">499</span> InputStreamReader reader;<a name="line.499"></a>
+<span class="sourceLineNo">500</span> if (warcFileInput.getName().toLowerCase().endsWith(".gz")) {<a name="line.500"></a>
+<span class="sourceLineNo">501</span> reader = new InputStreamReader(new GZIPInputStream(is));<a name="line.501"></a>
+<span class="sourceLineNo">502</span> } else {<a name="line.502"></a>
+<span class="sourceLineNo">503</span> reader = new InputStreamReader(is);<a name="line.503"></a>
+<span class="sourceLineNo">504</span> }<a name="line.504"></a>
+<span class="sourceLineNo">505</span> try (BufferedReader br = new BufferedReader(reader)) {<a name="line.505"></a>
+<span class="sourceLineNo">506</span> String line;<a name="line.506"></a>
+<span class="sourceLineNo">507</span> while ((line = br.readLine()) != null) {<a name="line.507"></a>
+<span class="sourceLineNo">508</span> paths.add(new Path(line));<a name="line.508"></a>
+<span class="sourceLineNo">509</span> }<a name="line.509"></a>
+<span class="sourceLineNo">510</span> }<a name="line.510"></a>
+<span class="sourceLineNo">511</span> }<a name="line.511"></a>
+<span class="sourceLineNo">512</span> LOG.info("Read " + paths.size() + " WARC input paths from " + warcFileInput);<a name="line.512"></a>
+<span class="sourceLineNo">513</span> FileInputFormat.setInputPaths(job, paths.toArray(new Path[paths.size()]));<a name="line.513"></a>
+<span class="sourceLineNo">514</span> }<a name="line.514"></a>
+<span class="sourceLineNo">515</span> job.setOutputFormatClass(SequenceFileOutputFormat.class);<a name="line.515"></a>
+<span class="sourceLineNo">516</span> SequenceFileOutputFormat.setOutputPath(job, outputDir);<a name="line.516"></a>
+<span class="sourceLineNo">517</span> SequenceFileOutputFormat.setOutputCompressionType(job, CompressionType.BLOCK);<a name="line.517"></a>
+<span class="sourceLineNo">518</span> job.setOutputKeyClass(HBaseKeyWritable.class);<a name="line.518"></a>
+<span class="sourceLineNo">519</span> job.setOutputValueClass(BytesWritable.class);<a name="line.519"></a>
+<span class="sourceLineNo">520</span> TableMapReduceUtil.addDependencyJars(job);<a name="line.520"></a>
+<span class="sourceLineNo">521</span><a name="line.521"></a>
+<span class="sourceLineNo">522</span> boolean success = job.waitForCompletion(true);<a name="line.522"></a>
+<span class="sourceLineNo">523</span> if (!success) {<a name="line.523"></a>
+<span class="sourceLineNo">524</span> LOG.error("Failure during job " + job.getJobID());<a name="line.524"></a>
+<span class="sourceLineNo">525</span> }<a name="line.525"></a>
+<span class="sourceLineNo">526</span> return success ? 0 : 1;<a name="line.526"></a>
+<span class="sourceLineNo">527</span> }<a name="line.527"></a>
+<span class="sourceLineNo">528</span><a name="line.528"></a>
+<span class="sourceLineNo">529</span> @Override<a name="line.529"></a>
+<span class="sourceLineNo">530</span> public int run(String[] args) throws Exception {<a name="line.530"></a>
+<span class="sourceLineNo">531</span> if (args.length < 2) {<a name="line.531"></a>
+<span class="sourceLineNo">532</span> System.err.println(USAGE);<a name="line.532"></a>
+<span class="sourceLineNo">533</span> return 1;<a name="line.533"></a>
+<span class="sourceLineNo">534</span> }<a name="line.534"></a>
+<span class="sourceLineNo">535</span> try {<a name="line.535"></a>
+<span class="sourceLineNo">536</span> Path warcFileInput = new Path(args[0]);<a name="line.536"></a>
+<span class="sourceLineNo">537</span> Path outputDir = new Path(args[1]);<a name="line.537"></a>
+<span class="sourceLineNo">538</span> return run(warcFileInput, outputDir);<a name="line.538"></a>
+<span class="sourceLineNo">539</span> } catch (NumberFormatException e) {<a name="line.539"></a>
+<span class="sourceLineNo">540</span> System.err.println("Parsing loader arguments failed: " + e.getMessage());<a name="line.540"></a>
+<span class="sourceLineNo">541</span> System.err.println(USAGE);<a name="line.541"></a>
+<span class="sourceLineNo">542</span> return 1;<a name="line.542"></a>
+<span class="sourceLineNo">543</span> }<a name="line.543"></a>
+<span class="sourceLineNo">544</span> }<a name="line.544"></a>
+<span class="sourceLineNo">545</span><a name="line.545"></a>
+<span class="sourceLineNo">546</span> public static void main(String[] args) throws Exception {<a name="line.546"></a>
+<span class="sourceLineNo">547</span> System.exit(ToolRunner.run(HBaseConfiguration.create(), new Loader(), args));<a name="line.547"></a>
+<span class="sourceLineNo">548</span> }<a name="line.548"></a>
+<span class="sourceLineNo">549</span><a name="line.549"></a>
+<span class="sourceLineNo">550</span> public static class LoaderMapper<a name="line.550"></a>
+<span class="sourceLineNo">551</span> extends Mapper<LongWritable, WARCWritable, HBaseKeyWritable, BytesWritable> {<a name="line.551"></a>
+<span class="sourceLineNo">552</span><a name="line.552"></a>
+<span class="sourceLineNo">553</span> Configuration conf;<a name="line.553"></a>
+<span class="sourceLineNo">554</span> Connection conn;<a name="line.554"></a>
+<span class="sourceLineNo">555</span> BufferedMutator mutator;<a name="line.555"></a>
+<span class="sourceLineNo">556</span><a name="line.556"></a>
+<span class="sourceLineNo">557</span> @Override<a name="line.557"></a>
+<span class="sourceLineNo">558</span> protected void setup(Context context) throws IOException, InterruptedException {<a name="line.558"></a>
+<span class="sourceLineNo">559</span> conf = context.getConfiguration();<a name="line.559"></a>
+<span class="sourceLineNo">560</span> conn = ConnectionFactory.createConnection(conf);<a name="line.560"></a>
+<span class="sourceLineNo">561</span> mutator = conn.getBufferedMutator(getTablename(conf));<a name="line.561"></a>
+<span class="sourceLineNo">562</span> mutator.setWriteBufferPeriodicFlush(10 * 1000); // default is 1 sec, increase to 10<a name="line.562"></a>
+<span class="sourceLineNo">563</span> }<a name="line.563"></a>
+<span class="sourceLineNo">564</span><a name="line.564"></a>
+<span class="sourceLineNo">565</span> @Override<a name="line.565"></a>
+<span class="sourceLineNo">566</span> protected void cleanup(Context context) throws IOException, InterruptedException {<a name="line.566"></a>
+<span class="sourceLineNo">567</span> try {<a name="line.567"></a>
+<span class="sourceLineNo">568</span> mutator.close();<a name="line.568"></a>
+<span class="sourceLineNo">569</span> } catch (Exception e) {<a name="line.569"></a>
+<span class="sourceLineNo">570</span> LOG.warn("Exception closing Table", e);<a name="line.570"></a>
+<span class="sourceLineNo">571</span> }<a name="line.571"></a>
+<span class="sourceLineNo">572</span> try {<a name="line.572"></a>
+<span class="sourceLineNo">573</span> conn.close();<a name="line.573"></a>
+<span class="sourceLineNo">574</span> } catch (Exception e) {<a name="line.574"></a>
+<span class="sourceLineNo">575</span> LOG.warn("Exception closing Connection", e);<a name="line.575"></a>
+<span class="sourceLineNo">576</span> }<a name="line.576"></a>
+<span class="sourceLineNo">577</span> }<a name="line.577"></a>
+<span class="sourceLineNo">578</span><a name="line.578"></a>
+<span class="sourceLineNo">579</span> @Override<a name="line.579"></a>
+<span class="sourceLineNo">580</span> protected void map(LongWritable key, WARCWritable value, Context output)<a name="line.580"></a>
+<span class="sourceLineNo">581</span> throws IOException, InterruptedException {<a name="line.581"></a>
+<span class="sourceLineNo">582</span> WARCRecord.Header warcHeader = value.getRecord().getHeader();<a name="line.582"></a>
+<span class="sourceLineNo">583</span> String recordID = warcHeader.getRecordID();<a name="line.583"></a>
+<span class="sourceLineNo">584</span> String targetURI = warcHeader.getTargetURI();<a name="line.584"></a>
+<span class="sourceLineNo">585</span> if (warcHeader.getRecordType().equals("response") && targetURI != null) {<a name="line.585"></a>
+<span class="sourceLineNo">586</span> String contentType = warcHeader.getField("WARC-Identified-Payload-Type");<a name="line.586"></a>
+<span class="sourceLineNo">587</span> if (contentType != null) {<a name="line.587"></a>
+<span class="sourceLineNo">588</span> LOG.info("Processing uri=\"" + targetURI + "\", id=" + recordID);<a name="line.588"></a>
+<span class="sourceLineNo">589</span> long now = EnvironmentEdgeManager.currentTime();<a name="line.589"></a>
+<span class="sourceLineNo">590</span><a name="line.590"></a>
+<span class="sourceLineNo">591</span> // Make row key<a name="line.591"></a>
+<span class="sourceLineNo">592</span><a name="line.592"></a>
+<span class="sourceLineNo">593</span> byte[] rowKey;<a name="line.593"></a>
+<span class="sourceLineNo">594</span> try {<a name="line.594"></a>
+<span class="sourceLineNo">595</span> rowKey = rowKeyFromTargetURI(targetURI);<a name="line.595"></a>
+<span class="sourceLineNo">596</span> } catch (IllegalArgumentException e) {<a name="line.596"></a>
+<span class="sourceLineNo">597</span> LOG.debug("Could not make a row key for record " + recordID + ", ignoring", e);<a name="line.597"></a>
+<span class="sourceLineNo">598</span> return;<a name="line.598"></a>
+<span class="sourceLineNo">599</span> } catch (URISyntaxException e) {<a name="line.599"></a>
+<span class="sourceLineNo">600</span> LOG.warn("Could not parse URI \"" + targetURI + "\" for record " + recordID +<a name="line.600"></a>
+<span class="sourceLineNo">601</span> ", ignoring");<a name="line.601"></a>
+<span class="sourceLineNo">602</span> return;<a name="line.602"></a>
+<span class="sourceLineNo">603</span> }<a name="line.603"></a>
+<span class="sourceLineNo">604</span><a name="line.604"></a>
+<span class="sourceLineNo">605</span> // Get the content and calculate the CRC64<a name="line.605"></a>
+<span class="sourceLineNo">606</span><a name="line.606"></a>
+<span class="sourceLineNo">607</span> byte[] content = value.getRecord().getContent();<a name="line.607"></a>
+<span class="sourceLineNo">608</span> CRC64 crc = new CRC64();<a name="line.608"></a>
+<span class="sourceLineNo">609</span> crc.update(content);<a name="line.609"></a>
+<span class="sourceLineNo">610</span> long crc64 = crc.getValue();<a name="line.610"></a>
+<span class="sourceLineNo">611</span><a name="line.611"></a>
+<span class="sourceLineNo">612</span> // Store to HBase<a name="line.612"></a>
+<span class="sourceLineNo">613</span><a name="line.613"></a>
+<span class="sourceLineNo">614</span> Put put = new Put(rowKey);<a name="line.614"></a>
+<span class="sourceLineNo">615</span> put.addColumn(CONTENT_FAMILY_NAME, CONTENT_QUALIFIER, now, content);<a name="line.615"></a>
+<span class="sourceLineNo">616</span> put.addColumn(INFO_FAMILY_NAME, CONTENT_LENGTH_QUALIFIER, now,<a name="line.616"></a>
+<span class="sourceLineNo">617</span> Bytes.toBytes(content.length));<a name="line.617"></a>
+<span class="sourceLineNo">618</span> put.addColumn(INFO_FAMILY_NAME, CONTENT_TYPE_QUALIFIER, now,<a name="line.618"></a>
+<span class="sourceLineNo">619</span> Bytes.toBytes(contentType));<a name="line.619"></a>
+<span class="sourceLineNo">620</span> put.addColumn(INFO_FAMILY_NAME, CRC_QUALIFIER, now, Bytes.toBytes(crc64));<a name="line.620"></a>
+<span class="sourceLineNo">621</span> put.addColumn(INFO_FAMILY_NAME, RECORD_ID_QUALIFIER, now, Bytes.toBytes(recordID));<a name="line.621"></a>
+<span class="sourceLineNo">622</span> put.addColumn(INFO_FAMILY_NAME, TARGET_URI_QUALIFIER, now, Bytes.toBytes(targetURI));<a name="line.622"></a>
+<span class="sourceLineNo">623</span> put.addColumn(INFO_FAMILY_NAME, DATE_QUALIFIER, now,<a name="line.623"></a>
+<span class="sourceLineNo">624</span> Bytes.toBytes(warcHeader.getDateString()));<a name="line.624"></a>
+<span class="sourceLineNo">625</span> String ipAddr = warcHeader.getField("WARC-IP-Address");<a name="line.625"></a>
+<span class="sourceLineNo">626</span> if (ipAddr != null) {<a name="line.626"></a>
+<span class="sourceLineNo">627</span> put.addColumn(INFO_FAMILY_NAME, IP_ADDRESS_QUALIFIER, now, Bytes.toBytes(ipAddr));<a name="line.627"></a>
+<span class="sourceLineNo">628</span> }<a name="line.628"></a>
+<span class="sourceLineNo">629</span> mutator.mutate(put);<a name="line.629"></a>
<span class="sourceLineNo">630</span><a name="line.630"></a>
-<span class="sourceLineNo">631</span> output.write(new HBaseKeyWritable(rowKey, INFO_FAMILY_NAME, CRC_QUALIFIER, now),<a name="line.631"></a>
-<span class="sourceLineNo">632</span> new BytesWritable(Bytes.toBytes(crc64)));<a name="line.632"></a>
-<span class="sourceLineNo">633</span> output.write(new HBaseKeyWritable(rowKey, INFO_FAMILY_NAME, CONTENT_LENGTH_QUALIFIER,<a name="line.633"></a>
-<span class="sourceLineNo">634</span> now), new BytesWritable(Bytes.toBytes(content.length)));<a name="line.634"></a>
-<span class="sourceLineNo">635</span> output.write(new HBaseKeyWritable(rowKey, INFO_FAMILY_NAME, CONTENT_TYPE_QUALIFIER,<a name="line.635"></a>
-<span class="sourceLineNo">636</span> now), new BytesWritable(Bytes.toBytes(contentType)));<a name="line.636"></a>
-<span class="sourceLineNo">637</span> output.write(new HBaseKeyWritable(rowKey, INFO_FAMILY_NAME, RECORD_ID_QUALIFIER,<a name="line.637"></a>
-<span class="sourceLineNo">638</span> now), new BytesWritable(Bytes.toBytes(recordID)));<a name="line.638"></a>
-<span class="sourceLineNo">639</span> output.write(new HBaseKeyWritable(rowKey, INFO_FAMILY_NAME, TARGET_URI_QUALIFIER,<a name="line.639"></a>
-<span class="sourceLineNo">640</span> now), new BytesWritable(Bytes.toBytes(targetURI)));<a name="line.640"></a>
-<span class="sourceLineNo">641</span> output.write(new HBaseKeyWritable(rowKey, INFO_FAMILY_NAME, DATE_QUALIFIER, now),<a name="line.641"></a>
-<span class="sourceLineNo">642</span> new BytesWritable(Bytes.toBytes(warcHeader.getDateString())));<a name="line.642"></a>
-<span class="sourceLineNo">643</span> if (ipAddr != null) {<a name="line.643"></a>
-<span class="sourceLineNo">644</span> output.write(new HBaseKeyWritable(rowKey, INFO_FAMILY_NAME, IP_ADDRESS_QUALIFIER,<a name="line.644"></a>
-<span class="sourceLineNo">645</span> now), new BytesWritable(Bytes.toBytes(ipAddr)));<a name="line.645"></a>
-<span class="sourceLineNo">646</span> }<a name="line.646"></a>
-<span class="sourceLineNo">647</span> }<a name="line.647"></a>
-<span class="sourceLineNo">648</span> }<a name="line.648"></a>
-<span class="sourceLineNo">649</span> }<a name="line.649"></a>
-<span class="sourceLineNo">650</span><a name="line.650"></a>
-<span class="sourceLineNo">651</span> private byte[] rowKeyFromTargetURI(String targetUri)<a name="line.651"></a>
-<span class="sourceLineNo">652</span> throws URISyntaxException, IllegalArgumentException {<a name="line.652"></a>
-<span class="sourceLineNo">653</span> URI uri = new URI(targetUri);<a name="line.653"></a>
-<span class="sourceLineNo">654</span> StringBuffer sb = new StringBuffer();<a name="line.654"></a>
-<span class="sourceLineNo">655</span> // Ignore the scheme<a name="line.655"></a>
-<span class="sourceLineNo">656</span> // Reverse the components of the hostname<a name="line.656"></a>
-<span class="sourceLineNo">657</span> if (uri.getHost() != null) {<a name="line.657"></a>
-<span class="sourceLineNo">658</span> String[] hostComponents = uri.getHost().split("\\.");<a name="line.658"></a>
-<span class="sourceLineNo">659</span> for (int i = hostComponents.length - 1; i >= 0; i--) {<a name="line.659"></a>
-<span class="sourceLineNo">660</span> sb.append(hostComponents[i]);<a name="line.660"></a>
-<span class="sourceLineNo">661</span> if (i != 0) {<a name="line.661"></a>
-<span class="sourceLineNo">662</span> sb.append('.');<a name="line.662"></a>
-<span class="sourceLineNo">663</span> }<a name="line.663"></a>
-<span class="sourceLineNo">664</span> }<a name="line.664"></a>
-<span class="sourceLineNo">665</span> } else {<a name="line.665"></a>
-<span class="sourceLineNo">666</span> throw new IllegalArgumentException("URI is missing host component");<a name="line.666"></a>
-<span class="sourceLineNo">667</span> }<a name="line.667"></a>
-<span class="sourceLineNo">668</span> // Port<a name="line.668"></a>
-<span class="sourceLineNo">669</span> if (uri.getPort() != -1) {<a name="line.669"></a>
-<span class="sourceLineNo">670</span> sb.append(':');<a name="line.670"></a>
-<span class="sourceLineNo">671</span> sb.append(uri.getPort());<a name="line.671"></a>
+<span class="sourceLineNo">631</span> // Write records out for later verification, one per HBase field except for the<a name="line.631"></a>
+<span class="sourceLineNo">632</span> // content record, which will be verified by CRC64.<a name="line.632"></a>
+<span class="sourceLineNo">633</span><a name="line.633"></a>
+<span class="sourceLineNo">634</span> output.write(new HBaseKeyWritable(rowKey, INFO_FAMILY_NAME, CRC_QUALIFIER, now),<a name="line.634"></a>
+<span class="sourceLineNo">635</span> new BytesWritable(Bytes.toBytes(crc64)));<a name="line.635"></a>
+<span class="sourceLineNo">636</span> output.write(new HBaseKeyWritable(rowKey, INFO_FAMILY_NAME, CONTENT_LENGTH_QUALIFIER,<a name="line.636"></a>
+<span class="sourceLineNo">637</span> now), new BytesWritable(Bytes.toBytes(content.length)));<a name="line.637"></a>
+<span class="sourceLineNo">638</span> output.write(new HBaseKeyWritable(rowKey, INFO_FAMILY_NAME, CONTENT_TYPE_QUALIFIER,<a name="line.638"></a>
+<span class="sourceLineNo">639</span> now), new BytesWritable(Bytes.toBytes(contentType)));<a name="line.639"></a>
+<span class="sourceLineNo">640</span> output.write(new HBaseKeyWritable(rowKey, INFO_FAMILY_NAME, RECORD_ID_QUALIFIER,<a name="line.640"></a>
+<span class="sourceLineNo">641</span> now), new BytesWritable(Bytes.toBytes(recordID)));<a name="line.641"></a>
+<span class="sourceLineNo">642</span> output.write(new HBaseKeyWritable(rowKey, INFO_FAMILY_NAME, TARGET_URI_QUALIFIER,<a name="line.642"></a>
+<span class="sourceLineNo">643</span> now), new BytesWritable(Bytes.toBytes(targetURI)));<a name="line.643"></a>
+<span class="sourceLineNo">644</span> output.write(new HBaseKeyWritable(rowKey, INFO_FAMILY_NAME, DATE_QUALIFIER, now),<a name="line.644"></a>
+<span class="sourceLineNo">645</span> new BytesWritable(Bytes.toBytes(warcHeader.getDateString())));<a name="line.645"></a>
+<span class="sourceLineNo">646</span> if (ipAddr != null) {<a name="line.646"></a>
+<span class="sourceLineNo">647</span> output.write(new HBaseKeyWritable(rowKey, INFO_FAMILY_NAME, IP_ADDRESS_QUALIFIER,<a name="line.647"></a>
+<span class="sourceLineNo">648</span> now), new BytesWritable(Bytes.toBytes(ipAddr)));<a name="line.648"></a>
+<span class="sourceLineNo">649</span> }<a name="line.649"></a>
+<span class="sourceLineNo">650</span> }<a name="line.650"></a>
+<span class="sourceLineNo">651</span> }<a name="line.651"></a>
+<span class="sourceLineNo">652</span> }<a name="line.652"></a>
+<span class="sourceLineNo">653</span><a name="line.653"></a>
+<span class="sourceLineNo">654</span> private byte[] rowKeyFromTargetURI(String targetUri)<a name="line.654"></a>
+<span class="sourceLineNo">655</span> throws URISyntaxException, IllegalArgumentException {<a name="line.655"></a>
+<span class="sourceLineNo">656</span> URI uri = new URI(targetUri);<a name="line.656"></a>
+<span class="sourceLineNo">657</span> // Ignore the scheme<a name="line.657"></a>
+<span class="sourceLineNo">658</span> // Reverse the components of the hostname<a name="line.658"></a>
+<span class="sourceLineNo">659</span> String reversedHost;<a name="line.659"></a>
+<span class="sourceLineNo">660</span> if (uri.getHost() != null) {<a name="line.660"></a>
+<span class="sourceLineNo">661</span> StringBuffer sb = new StringBuffer();<a name="line.661"></a>
+<span class="sourceLineNo">662</span> String[] hostComponents = uri.getHost().split("\\.");<a name="line.662"></a>
+<span class="sourceLineNo">663</span> for (int i = hostComponents.length - 1; i >= 0; i--) {<a name="line.663"></a>
+<span class="sourceLineNo">664</span> sb.append(hostComponents[i]);<a name="line.664"></a>
+<span class="sourceLineNo">665</span> if (i != 0) {<a name="line.665"></a>
+<span class="sourceLineNo">666</span> sb.append('.');<a name="line.666"></a>
+<span class="sourceLineNo">667</span> }<a name="line.667"></a>
+<span class="sourceLineNo">668</span> }<a name="line.668"></a>
+<span class="sourceLineNo">669</span> reversedHost = sb.toString();<a name="line.669"></a>
+<span class="sourceLineNo">670</span> } else {<a name="line.670"></a>
+<span class="sourceLineNo">671</span> throw new IllegalArgumentException("URI is missing host component");<a name="line.671"></a>
<span class="sourceLineNo">672</span> }<a name="line.672"></a>
-<span class="sourceLineNo">673</span> if (uri.getRawPath() != null) {<a name="line.673"></a>
-<span class="sourceLineNo">674</span> sb.append(uri.getRawPath());<a name="line.674"></a>
-<span class="sourceLineNo">675</span> }<a name="line.675"></a>
-<span class="sourceLineNo">676</span> if (uri.getRawQuery() != null) {<a name="line.676"></a>
-<span class="sourceLineNo">677</span> sb.append('?');<a name="line.677"></a>
-<span class="sourceLineNo">678</span> sb.append(uri.getRawQuery());<a name="line.678"></a>
-<span class="sourceLineNo">679</span> }<a name="line.679"></a>
-<span class="sourceLineNo">680</span> if (uri.getRawFragment() != null) {<a name="line.680"></a>
-<span class="sourceLineNo">681</span> sb.append('#');<a name="line.681"></a>
-<span class="sourceLineNo">682</span> sb.append(uri.getRawFragment());<a name="line.682"></a>
-<span class="sourceLineNo">683</span> }<a name="line.683"></a>
-<span class="sourceLineNo">684</span> // Constrain the key size to the maximum allowed row key length<a name="line.684"></a>
-<span class="sourceLineNo">685</span> if (sb.length() > HConstants.MAX_ROW_LENGTH) {<a name="line.685"></a>
-<span class="sourceLineNo">686</span> sb.setLength(HConstants.MAX_ROW_LENGTH);<a name="line.686"></a>
-<span class="sourceLineNo">687</span> }<a name="line.687"></a>
-<span class="sourceLineNo">688</span> return Bytes.toBytes(sb.toString());<a name="line.688"></a>
-<span class="sourceLineNo">689</span> }<a name="line.689"></a>
-<span class="sourceLineNo">690</span><a name="line.690"></a>
-<span class="sourceLineNo">691</span> }<a name="line.691"></a>
-<span class="sourceLineNo">692</span> }<a name="line.692"></a>
-<span class="sourceLineNo">693</span><a name="line.693"></a>
-<span class="sourceLineNo">694</span> public static class OneFilePerMapperSFIF<K, V> extends SequenceFileInputFormat<K, V> {<a name="line.694"></a>
-<span class="sourceLineNo">695</span> @Override<a name="line.695"></a>
-<span class="sourceLineNo">696</span> protected boolean isSplitable(JobContext context, Path filename) {<a name="line.696"></a>
-<span class="sourceLineNo">697</span> return false;<a name="line.697"></a>
+<span class="sourceLineNo">673</span> StringBuffer sb = new StringBuffer();<a name="line.673"></a>
+<span class="sourceLineNo">674</span> sb.append(reversedHost);<a name="line.674"></a>
+<span class="sourceLineNo">675</span> if (uri.getPort() >= 0) {<a name="line.675"></a>
+<span class="sourceLineNo">676</span> sb.append(':');<a name="line.676"></a>
+<span class="sourceLineNo">677</span> sb.append(uri.getPort());<a name="line.677"></a>
+<span class="sourceLineNo">678</span> }<a name="line.678"></a>
+<span class="sourceLineNo">679</span> if (uri.getPath() != null) {<a name="line.679"></a>
+<span class="sourceLineNo">680</span> sb.append('/');<a name="line.680"></a>
+<span class="sourceLineNo">681</span> sb.append(uri.getPath());<a name="line.681"></a>
+<span class="sourceLineNo">682</span> }<a name="line.682"></a>
+<span class="sourceLineNo">683</span> if (uri.getQuery() != null) {<a name="line.683"></a>
+<span class="sourceLineNo">684</span> sb.append('?');<a name="line.684"></a>
+<span class="sourceLineNo">685</span> sb.append(uri.getQuery());<a name="line.685"></a>
+<span class="sourceLineNo">686</span> }<a name="line.686"></a>
+<span class="sourceLineNo">687</span> if (uri.getFragment() != null) {<a name="line.687"></a>
+<span class="sourceLineNo">688</span> sb.append('#');<a name="line.688"></a>
+<span class="sourceLineNo">689</span> sb.append(uri.getFragment());<a name="line.689"></a>
+<span class="sourceLineNo">690</span> }<a name="line.690"></a>
+<span class="sourceLineNo">691</span> if (sb.length() > HConstants.MAX_ROW_LENGTH) {<a name="line.691"></a>
+<span class="sourceLineNo">692</span> throw new IllegalArgumentException("Key would be too large (length=" + sb.length() +<a name="line.692"></a>
+<span class="sourceLineNo">693</span> ", limit=" + HConstants.MAX_ROW_LENGTH);<a name="line.693"></a>
+<span class="sourceLineNo">694</span> }<a name="line.694"></a>
+<span class="sourceLineNo">695</span> return Bytes.toBytes(sb.toString());<a name="line.695"></a>
+<span class="sourceLineNo">696</span> }<a name="line.696"></a>
+<span class="sourceLineNo">697</span><a name="line.697"></a>
<span class="sourceLineNo">698</span> }<a name="line.698"></a>
<span class="sourceLineNo">699</span> }<a name="line.699"></a>
<span class="sourceLineNo">700</span><a name="line.700"></a>
-<span class="sourceLineNo">701</span> public static class Verify extends Configured implements Tool {<a name="line.701"></a>
-<span class="sourceLineNo">702</span><a name="line.702"></a>
-<span class="sourceLineNo">703</span> public static final Logger LOG = LoggerFactory.getLogger(Verify.class);<a name="line.703"></a>
-<span class="sourceLineNo">704</span> public static final String USAGE = "Verify <inputDir>";<a name="line.704"></a>
-<span class="sourceLineNo">705</span><a name="line.705"></a>
-<span class="sourceLineNo">706</span> int run(Path inputDir) throws IOException, ClassNotFoundException, InterruptedException {<a name="line.706"></a>
-<span class="sourceLineNo">707</span> Job job = Job.getInstance(getConf());<a name="line.707"></a>
-<span class="sourceLineNo">708</span> job.setJobName(Verify.class.getName());<a name="line.708"></a>
-<span class="sourceLineNo">709</span> job.setJarByClass(getClass());<a name="line.709"></a>
-<span class="sourceLineNo">710</span> job.setMapperClass(VerifyMapper.class);<a name="line.710"></a>
-<span class="sourceLineNo">711</span> job.setInputFormatClass(OneFilePerMapperSFIF.class);<a name="line.711"></a>
-<span class="sourceLineNo">712</span> FileInputFormat.setInputPaths(job, inputDir);<a name="line.712"></a>
-<span class="sourceLineNo">713</span> job.setOutputFormatClass(NullOutputFormat.class);<a name="line.713"></a>
-<span class="sourceLineNo">714</span> job.setOutputKeyClass(NullWritable.class);<a name="line.714"></a>
-<span class="sourceLineNo">715</span> job.setOutputValueClass(NullWritable.class);<a name="line.715"></a>
-<span class="sourceLineNo">716</span> TableMapReduceUtil.addDependencyJars(job);<a name="line.716"></a>
-<span class="sourceLineNo">717</span> boolean success = job.waitForCompletion(true);<a name="line.717"></a>
-<span class="sourceLineNo">718</span> if (!success) {<a name="line.718"></a>
-<span class="sourceLineNo">719</span> LOG.error("Failure during job " + job.getJobID());<a name="line.719"></a>
-<span class="sourceLineNo">720</span> }<a name="line.720"></a>
-<span class="sourceLineNo">721</span> Counters counters = job.getCounters();<a name="line.721"></a>
-<span class="sourceLineNo">722</span> for (Counts c: Counts.values()) {<a name="line.722"></a>
-<span class="sourceLineNo">723</span> LOG.info(c + ": " + counters.findCounter(c).getValue());<a name="line.723"></a>
-<span class="sourceLineNo">724</span> }<a name="line.724"></a>
-<span class="sourceLineNo">725</span> return success ? 0 : 1;<a name="line.725"></a>
-<span class="sourceLineNo">726</span> }<a name="line.726"></a>
-<span class="sourceLineNo">727</span><a name="line.727"></a>
-<span class="sourceLineNo">728</span> @Override<a name="line.728"></a>
-<span class="sourceLineNo">729</span> public int run(String[] args) throws Exception {<a name="line.729"></a>
-<span class="sourceLineNo">730</span> if (args.length < 2) {<a name="line.730"></a>
-<span class="sourceLineNo">731</span> System.err.println(USAGE);<a name="line.731"></a>
-<span class="sourceLineNo">732</span> return 1;<a name="line.732"></a>
-<span class="sourceLineNo">733</span> }<a name="line.733"></a>
-<span class="sourceLineNo">734</span> Path loaderOutput = new Path(args[0]);<a name="line.734"></a>
-<span class="sourceLineNo">735</span> return run(loaderOutput);<a name="line.735"></a>
-<span class="sourceLineNo">736</span> }<a name="line.736"></a>
-<span class="sourceLineNo">737</span><a name="line.737"></a>
-<span class="sourceLineNo">738</span> public static void main(String[] args) throws Exception {<a name="line.738"></a>
-<span class="sourceLineNo">739</span> System.exit(ToolRunner.run(HBaseConfiguration.create(), new Verify(), args));<a name="line.739"></a>
-<span class="sourceLineNo">740</span> }<a name="line.740"></a>
-<span class="sourceLineNo">741</span><a name="line.741"></a>
-<span class="sourceLineNo">742</span> public static class VerifyMapper<a name="line.742"></a>
-<span class="sourceLineNo">743</span> extends Mapper<HBaseKeyWritable, BytesWritable, NullWritable, NullWritable> {<a name="line.743"></a>
+<span class="sourceLineNo">701</span> public static class OneFilePerMapperSFIF<K, V> extends SequenceFileInputFormat<K, V> {<a name="line.701"></a>
+<span class="sourceLineNo">702</span> @Override<a name="line.702"></a>
+<span class="sourceLineNo">703</span> protected boolean isSplitable(JobContext context, Path filename) {<a name="line.703"></a>
+<span class="sourceLineNo">704</span> return false;<a name="line.704"></a>
+<span class="sourceLineNo">705</span> }<a name="line.705"></a>
+<span class="sourceLineNo">706</span> }<a name="line.706"></a>
+<span class="sourceLineNo">707</span><a name="line.707"></a>
+<span class="sourceLineNo">708</span> public static class Verify extends Configured implements Tool {<a name="line.708"></a>
+<span class="sourceLineNo">709</span><a name="line.709"></a>
+<span class="sourceLineNo">710</span> public static final Logger LOG = LoggerFactory.getLogger(Verify.class);<a name="line.710"></a>
+<span class="sourceLineNo">711</span> public static final String USAGE = "Verify <inputDir>";<a name="line.711"></a>
+<span class="sourceLineNo">712</span><a name="line.712"></a>
+<span class="sourceLineNo">713</span> int run(Path inputDir) throws IOException, ClassNotFoundException, InterruptedException {<a name="line.713"></a>
+<span class="sourceLineNo">714</span> Job job = Job.getInstance(getConf());<a name="line.714"></a>
+<span class="sourceLineNo">715</span> job.setJobName(Verify.class.getName());<a name="line.715"></a>
+<span class="sourceLineNo">716</span> job.setJarByClass(getClass());<a name="line.716"></a>
+<span class="sourceLineNo">717</span> job.setMapperClass(VerifyMapper.class);<a name="line.717"></a>
+<span class="sourceLineNo">718</span> job.setInputFormatClass(OneFilePerMapperSFIF.class);<a name="line.718"></a>
+<span class="sourceLineNo">719</span> FileInputFormat.setInputPaths(job, inputDir);<a name="line.719"></a>
+<span class="sourceLineNo">720</span> job.setOutputFormatClass(NullOutputFormat.class);<a name="line.720"></a>
+<span class="sourceLineNo">721</span> job.setOutputKeyClass(NullWritable.class);<a name="line.721"></a>
+<span class="sourceLineNo">722</span> job.setOutputValueClass(NullWritable.class);<a name="line.722"></a>
+<span class="sourceLineNo">723</span> TableMapReduceUtil.addDependencyJars(job);<a name="line.723"></a>
+<span class="sourceLineNo">724</span> boolean success = job.waitForCompletion(true);<a name="line.724"></a>
+<span class="sourceLineNo">725</span> if (!success) {<a name="line.725"></a>
+<span class="sourceLineNo">726</span> LOG.error("Failure during job " + job.getJobID());<a name="line.726"></a>
+<span class="sourceLineNo">727</span> }<a name="line.727"></a>
+<span class="sourceLineNo">728</span> Counters counters = job.getCounters();<a name="line.728"></a>
+<span class="sourceLineNo">729</span> for (Counts c: Counts.values()) {<a name="line.729"></a>
+<span class="sourceLineNo">730</span> LOG.info(c + ": " + counters.findCounter(c).getValue());<a name="line.730"></a>
+<span class="sourceLineNo">731</span> }<a name="line.731"></a>
+<span class="sourceLineNo">732</span> return success ? 0 : 1;<a name="line.732"></a>
+<span class="sourceLineNo">733</span> }<a name="line.733"></a>
+<span class="sourceLineNo">734</span><a name="line.734"></a>
+<span class="sourceLineNo">735</span> @Override<a name="line.735"></a>
+<span class="sourceLineNo">736</span> public int run(String[] args) throws Exception {<a name="line.736"></a>
+<span class="sourceLineNo">737</span> if (args.length < 2) {<a name="line.737"></a>
+<span class="sourceLineNo">738</span> System.err.println(USAGE);<a name="line.738"></a>
+<span class="sourceLineNo">739</span> return 1;<a name="line.739"></a>
+<span class="sourceLineNo">740</span> }<a name="line.740"></a>
+<span class="sourceLineNo">741</span> Path loaderOutput = new Path(args[0]);<a name="line.741"></a>
+<span class="sourceLineNo">742</span> return run(loaderOutput);<a name="line.742"></a>
+<span class="sourceLineNo">743</span> }<a name="line.743"></a>
<span class="sourceLineNo">744</span><a name="line.744"></a>
-<span class="sourceLineNo">745</span> Connection conn;<a name="line.745"></a>
-<span class="sourceLineNo">746</span> Table table;<a name="line.746"></a>
-<span class="sourceLineNo">747</span><a name="line.747"></a>
-<span class="sourceLineNo">748</span> @Override<a name="line.748"></a>
-<span class="sourceLineNo">749</span> protected void setup(Context context) throws IOException, InterruptedException {<a name="line.749"></a>
-<span class="sourceLineNo">750</span> conn = ConnectionFactory.createConnection(context.getConfiguration());<a name="line.750"></a>
-<span class="sourceLineNo">751</span> table = conn.getTable(getTablename(conn.getConfiguration()));<a name="line.751"></a>
-<span class="sourceLineNo">752</span> }<a name="line.752"></a>
-<span class="sourceLineNo">753</span><a name="line.753"></a>
-<span class="sourceLineNo">754</span> @Override<a name="line.754"></a>
-<span class="sourceLineNo">755</span> protected void cleanup(Context context) throws IOException ,InterruptedException {<a name="line.755"></a>
-<span class="sourceLineNo">756</span> table.close();<a name="line.756"></a>
-<span class="sourceLineNo">757</span> conn.close();<a name="line.757"></a>
-<span class="sourceLineNo">758</span> }<a name="line.758"></a>
-<span class="sourceLineNo">759</span><a name="line.759"></a>
-<span class="sourceLineNo">760</span> @Override<a name="line.760"></a>
-<span class="sourceLineNo">761</span> protected void map(HBaseKeyWritable key, BytesWritable value, Context output)<a name="line.761"></a>
-<span class="sourceLineNo">762</span> throws IOException, InterruptedException {<a name="line.762"></a>
-<span class="sourceLineNo">763</span><a name="line.763"></a>
-<span class="sourceLineNo">764</span> byte[] row = Bytes.copy(key.getRowArray(), key.getRowOffset(), key.getRowLength());<a name="line.764"></a>
-<span class="sourceLineNo">765</span> byte[] family = Bytes.copy(key.getFamilyArray(), key.getFamilyOffset(),<a name="line.765"></a>
-<span class="sourceLineNo">766</span> key.getFamilyLength());<a name="line.766"></a>
-<span class="sourceLineNo">767</span> byte[] qualifier = Bytes.copy(key.getQualifierArray(), key.getQualifierOffset(),<a name="line.767"></a>
-<span class="sourceLineNo">768</span> key.getQualifierLength());<a name="line.768"></a>
-<span class="sourceLineNo">769</span> long ts = key.getTimestamp();<a name="line.769"></a>
+<span class="sourceLineNo">745</span> public static void main(String[] args) throws Exception {<a name="line.745"></a>
+<span class="sourceLineNo">746</span> System.exit(ToolRunner.run(HBaseConfiguration.create(), new Verify(), args));<a name="line.746"></a>
+<span class="sourceLineNo">747</span> }<a name="line.747"></a>
+<span class="sourceLineNo">748</span><a name="line.748"></a>
+<span class="sourceLineNo">749</span> public static class VerifyMapper<a name="line.749"></a>
+<span class="sourceLineNo">750</span> extends Mapper<HBaseKeyWritable, BytesWritable, NullWritable, NullWritable> {<a name="line.750"></a>
+<span class="sourceLineNo">751</span><a name="line.751"></a>
+<span class="sourceLineNo">752</span> Connection conn;<a name="line.752"></a>
+<span class="sourceLineNo">753</span> Table table;<a name="line.753"></a>
+<span class="sourceLineNo">754</span><a name="line.754"></a>
+<span class="sourceLineNo">755</span> @Override<a name="line.755"></a>
+<span class="sourceLineNo">756</span> protected void setup(Context context) throws IOException, InterruptedException {<a name="line.756"></a>
+<span class="sourceLineNo">757</span> conn = ConnectionFactory.createConnection(context.getConfiguration());<a name="line.757"></a>
+<span class="sourceLineNo">758</span> table = conn.getTable(getTablename(conn.getConfiguration()));<a name="line.758"></a>
+<span class="sourceLineNo">759</span> }<a name="line.759"></a>
+<span class="sourceLineNo">760</span><a name="line.760"></a>
+<span class="sourceLineNo">761</span> @Override<a name="line.761"></a>
+<span class="sourceLineNo">762</span> protected void cleanup(Context context) throws IOException ,InterruptedException {<a name="line.762"></a>
+<span class="sourceLineNo">763</span> table.close();<a name="line.763"></a>
+<span class="sourceLineNo">764</span> conn.close();<a name="line.764"></a>
+<span class="sourceLineNo">765</span> }<a name="line.765"></a>
+<span class="sourceLineNo">766</span><a name="line.766"></a>
+<span class="sourceLineNo">767</span> @Override<a name="line.767"></a>
+<span class="sourceLineNo">768</span> protected void map(HBaseKeyWritable key, BytesWritable value, Context output)<a name="line.768"></a>
+<span class="sourceLineNo">769</span> throws IOException, InterruptedException {<a name="line.769"></a>
<span class="sourceLineNo">770</span><a name="line.770"></a>
-<span class="sourceLineNo">771</span> if (Bytes.equals(INFO_FAMILY_NAME, family) &&<a name="line.771"></a>
-<span class="sourceLineNo">772</span> Bytes.equals(CRC_QUALIFIER, qualifier)) {<a name="line.772"></a>
-<span class="sourceLineNo">773</span><a name="line.773"></a>
-<span class="sourceLineNo">774</span> long expectedCRC64 = Bytes.toLong(value.getBytes(), 0, value.getLength());<a name="line.774"></a>
-<span class="sourceLineNo">775</span><a name="line.775"></a>
-<span class="sourceLineNo">776</span> Result result =<a name="line.776"></a>
-<span class="sourceLineNo">777</span> table.get(new Get(row)<a name="line.777"></a>
-<span class="sourceLineNo">778</span> .addColumn(CONTENT_FAMILY_NAME, CONTENT_QUALIFIER)<a name="line.778"></a>
-<span class="sourceLineNo">779</span> .addColumn(INFO_FAMILY_NAME, CRC_QUALIFIER)<a name="line.779"></a>
-<span class="sourceLineNo">780</span> .setTimestamp(ts));<a name="line.780"></a>
-<span class="sourceLineNo">781</span><a name="line.781"></a>
-<span class="sourceLineNo">782</span> byte[] content = result.getValue(CONTENT_FAMILY_NAME, CONTENT_QUALIFIER);<a name="line.782"></a>
-<span class="sourceLineNo">783</span> if (content == null) {<a name="line.783"></a>
-<span class="sourceLineNo">784</span> LOG.info("Row " + Bytes.toStringBinary(row) + ": missing content");<a name="line.784"></a>
-<span class="sourceLineNo">785</span> output.getCounter(Counts.UNREFERENCED).increment(1);<a name="line.785"></a>
-<span class="sourceLineNo">786</span> return;<a name="line.786"></a>
-<span class="sourceLineNo">787</span> } else {<a name="line.787"></a>
-<span class="sourceLineNo">788</span> CRC64 crc = new CRC64();<a name="line.788"></a>
-<span class="sourceLineNo">789</span> crc.update(content);<a name="line.789"></a>
-<span class="sourceLineNo">790</span> if (crc.getValue() != expectedCRC64) {<a name="line.790"></a>
-<span class="sourceLineNo">791</span> LOG.info("Row " + Bytes.toStringBinary(row) + ": corrupt content");<a name="line.791"></a>
-<span class="sourceLineNo">792</span> output.getCounter(Counts.CORRUPT).increment(1);<a name="line.792"></a>
-<span class="sourceLineNo">793</span> return;<a name="line.793"></a>
-<span class="sourceLineNo">794</span> }<a name="line.794"></a>
-<span class="sourceLineNo">795</span> }<a name="line.795"></a>
-<span class="sourceLineNo">796</span> byte[] crc = result.getValue(INFO_FAMILY_NAME, CRC_QUALIFIER);<a name="line.796"></a>
-<span class="sourceLineNo">797</span> if (crc == null) {<a name="line.797"></a>
-<span class="sourceLineNo">798</span> LOG.info("Row " + Bytes.toStringBinary(row) + ": missing i:c");<a name="line.798"></a>
-<span class="sourceLineNo">799</span> output.getCounter(Counts.UNREFERENCED).increment(1);<a name="line.799"></a>
-<span class="sourceLineNo">800</span> return;<a name="line.800"></a>
-<span class="sourceLineNo">801</span> }<a name="line.801"></a>
-<span class="sourceLineNo">802</span> if (Bytes.toLong(crc) != expectedCRC64) {<a name="line.802"></a>
-<span class="sourceLineNo">803</span> LOG.info("Row " + Bytes.toStringBinary(row) + ": i:c mismatch");<a name="line.803"></a>
-<span class="sourceLineNo">804</span> output.getCounter(Counts.CORRUPT).increment(1);<a name="line.804"></a>
-<span class="sourceLineNo">805</span> return;<a name="line.805"></a>
-<span class="sourceLineNo">806</span> }<a name="line.806"></a>
-<span class="sourceLineNo">807</span><a name="line.807"></a>
-<span class="sourceLineNo">808</span> } else {<a name="line.808"></a>
-<span class="sourceLineNo">809</span><a name="line.809"></a>
-<span class="sourceLineNo">810</span> Result result =<a name="line.810"></a>
-<span class="sourceLineNo">811</span> table.get(new Get(row)<a name="line.811"></a>
-<span class="sourceLineNo">812</span> .addColumn(family, qualifier)<a name="line.812"></a>
-<span class="sourceLineNo">813</span> .setTimestamp(ts));<a name="line.813"></a>
-<span class="sourceLineNo">814</span><a name="line.814"></a>
-<span class="sourceLineNo">815</span> byte[] bytes = result.getValue(family, qualifier);<a name="line.815"></a>
-<span class="sourceLineNo">816</span> if (bytes == null) {<a name="line.816"></a>
-<span class="sourceLineNo">817</span> LOG.info("Row " + Bytes.toStringBinary(row) + ": missing " +<a name="line.817"></a>
-<span class="sourceLineNo">818</span> Bytes.toStringBinary(family) + ":" + Bytes.toStringBinary(qualifier));<a name="line.818"></a>
-<span class="sourceLineNo">819</span> output.getCounter(Counts.UNREFERENCED).increment(1);<a name="line.819"></a>
-<span class="sourceLineNo">820</span> return;<a name="line.820"></a>
-<span class="sourceLineNo">821</span> }<a name="line.821"></a>
-<span class="sourceLineNo">822</span> if (!Bytes.equals(bytes, 0, bytes.length, value.getBytes(), 0, value.getLength())) {<a name="line.822"></a>
-<span class="sourceLineNo">823</span> LOG.info("Row " + Bytes.toStringBinary(row) + ": " +<a name="line.823"></a>
-<span class="sourceLineNo">824</span> Bytes.toStringBinary(family) + ":" + Bytes.toStringBinary(qualifier) +<a name="line.824"></a>
-<span class="sourceLineNo">825</span> " mismatch");<a name="line.825"></a>
-<span class="sourceLineNo">826</span> output.getCounter(Counts.CORRUPT).increment(1);<a name="line.826"></a>
-<span class="sourceLineNo">827</span> return;<a name="line.827"></a>
-<span class="sourceLineNo">828</span> }<a name="line.828"></a>
-<span class="sourceLineNo">829</span><a name="line.829"></a>
-<span class="sourceLineNo">830</span> }<a name="line.830"></a>
-<span class="sourceLineNo">831</span><a name="line.831"></a>
-<span class="sourceLineNo">832</span> output.getCounter(Counts.REFERENCED).increment(1);<a name="line.832"></a>
-<span class="sourceLineNo">833</span> }<a name="line.833"></a>
-<span class="sourceLineNo">834</span><a name="line.834"></a>
-<span class="sourceLineNo">835</span> }<a name="line.835"></a>
-<span class="sourceLineNo">836</span><a name="line.836"></a>
-<span class="sourceLineNo">837</span> }<a name="line.837"></a>
-<span class="sourceLineNo">838</span><a name="line.838"></a>
-<span class="sourceLineNo">839</span>}<a name="line.839"></a>
+<span class="sourceLineNo">771</span> byte[] row = Bytes.copy(key.getRowArray(), key.getRowOffset(), key.getRowLength());<a name="line.771"></a>
+<span class="sourceLineNo">772</span> byte[] family = Bytes.copy(key.getFamilyArray(), key.getFamilyOffset(),<a name="line.772"></a>
+<span class="sourceLineNo">773</span> key.getFamilyLength());<a name="line.773"></a>
+<span class="sourceLineNo">774</span> byte[] qualifier = Bytes.copy(key.getQualifierArray(), key.getQualifierOffset(),<a name="line.774"></a>
+<span class="sourceLineNo">775</span> key.getQualifierLength());<a name="line.775"></a>
+<span class="sourceLineNo">776</span> long ts = key.getTimestamp();<a name="line.776"></a>
+<span class="sourceLineNo">777</span> int retries = VERIFICATION_READ_RETRIES;<a name="line.777"></a>
+<span class="sourceLineNo">778</span><a name="line.778"></a>
+<span class="sourceLineNo">779</span> while (true) {<a name="line.779"></a>
+<span class="sourceLineNo">780</span><a name="line.780"></a>
+<span class="sourceLineNo">781</span> if (Bytes.equals(INFO_FAMILY_NAME, family) &&<a name="line.781"></a>
+<span class="sourceLineNo">782</span> Bytes.equals(CRC_QUALIFIER, qualifier)) {<a name="line.782"></a>
+<span class="sourceLineNo">783</span><a name="line.783"></a>
+<span class="sourceLineNo">784</span> long expectedCRC64 = Bytes.toLong(value.getBytes(), 0, value.getLength());<a name="line.784"></a>
+<span class="sourceLineNo">785</span> Result result = table.get(new Get(row)<a name="line.785"></a>
+<span class="sourceLineNo">786</span> .addColumn(CONTENT_FAMILY_NAME, CONTENT_QUALIFIER)<a name="line.786"></a>
+<span class="sourceLineNo">787</span> .addColumn(INFO_FAMILY_NAME, CRC_QUALIFIER)<a name="line.787"></a>
+<span class="sourceLineNo">788</span> .setTimestamp(ts));<a name="line.788"></a>
+<span class="sourceLineNo">789</span> byte[] content = result.getValue(CONTENT_FAMILY_NAME, CONTENT_QUALIFIER);<a name="line.789"></a>
+<span class="sourceLineNo">790</span> if (content == null) {<a name="line.790"></a>
+<span class="sourceLineNo">791</span> if (retries-- > 0) {<a name="line.791"></a>
+<span class="sourceLineNo">792</span> continue;<a name="line.792"></a>
+<span class="sourceLineNo">793</span> }<a name="line.793"></a>
+<span class="sourceLineNo">794</span> LOG.error("Row " + Bytes.toStringBinary(row) + ": missing content");<a name="line.794"></a>
+<span class="sourceLineNo">795</span> output.getCounter(Counts.UNREFERENCED).increment(1);<a name="line.795"></a>
+<span class="sourceLineNo">796</span> return;<a name="line.796"></a>
+<span class="sourceLineNo">797</span> } else {<a name="line.797"></a>
+<span class="sourceLineNo">798</span> CRC64 crc = new CRC64();<a name="line.798"></a>
+<span class="sourceLineNo">799</span> crc.update(content);<a name="line.799"></a>
+<span class="sourceLineNo">800</span> if (crc.getValue() != expectedCRC64) {<a name="line.800"></a>
+<span class="sourceLineNo">801</span> if (retries-- > 0) {<a name="line.801"></a>
+<span class="sourceLineNo">802</span> continue;<a name="line.802"></a>
+<span class="sourceLineNo">803</span> }<a name="line.803"></a>
+<span class="sourceLineNo">804</span> LOG.error("Row " + Bytes.toStringBinary(row) + ": corrupt content");<a name="line.804"></a>
+<span class="sourceLineNo">805</span> output.getCounter(Counts.CORRUPT).increment(1);<a name="line.805"></a>
+<span class="sourceLineNo">806</span> return;<a name="line.806"></a>
+<span class="sourceLineNo">807</span> }<a name="line.807"></a>
+<span class="sourceLineNo">808</span> }<a name="line.808"></a>
+<span class="sourceLineNo">809</span> byte[] crc = result.getValue(INFO_FAMILY_NAME, CRC_QUALIFIER);<a name="line.809"></a>
+<span class="sourceLineNo">810</span> if (crc == null) {<a name="line.810"></a>
+<span class="sourceLineNo">811</span> if (retries-- > 0) {<a name="line.811"></a>
+<span class="sourceLineNo">812</span> continue;<a name="line.812"></a>
+<span class="sourceLineNo">813</span> }<a name="line.813"></a>
+<span class="sourceLineNo">814</span> LOG.error("Row " + Bytes.toStringBinary(row) + ": missing i:c");<a name="line.814"></a>
+<span class="sourceLineNo">815</span> output.getCounter(Counts.UNREFERENCED).increment(1);<a name="line.815"></a>
+<span class="sourceLineNo">816</span> return;<a name="line.816"></a>
+<span class="sourceLineNo">817</span> }<a name="line.817"></a>
+<span class="sourceLineNo">818</span> if (Bytes.toLong(crc) != expectedCRC64) {<a name="line.818"></a>
+<span class="sourceLineNo">819</span> if (retries-- > 0) {<a name="line.819"></a>
+<span class="sourceLineNo">820</span> continue;<a name="line.820"></a>
+<span class="sourceLineNo">821</span> }<a name="line.821"></a>
+<span class="sourceLineNo">822</span> LOG.error("Row " + Bytes.toStringBinary(row) + ": i:c mismatch");<a name="line.822"></a>
+<span class="sourceLineNo">823</span> output.getCounter(Counts.CORRUPT).increment(1);<a name="line.823"></a>
+<span class="sourceLineNo">824</span> return;<a name="line.824"></a>
+<span class="sourceLineNo">825</span> }<a name="line.825"></a>
+<span class="sourceLineNo">826</span><a name="line.826"></a>
+<span class="sourceLineNo">827</span> } else {<a name="line.827"></a>
+<span class="sourceLineNo">828</span><a name="line.828"></a>
+<span class="sourceLineNo">829</span> Result result = table.get(new Get(row)<a name="line.829"></a>
+<span class="sourceLineNo">830</span> .addColumn(family, qualifier)<a name="line.830"></a>
+<span class="sourceLineNo">831</span> .setTimestamp(ts));<a name="line.831"></a>
+<span class="sourceLineNo">832</span> byte[] bytes = result.getValue(family, qualifier);<a name="line.832"></a>
+<span class="sourceLineNo">833</span> if (bytes == null) {<a name="line.833"></a>
+<span class="sourceLineNo">834</span> if (retries-- > 0) {<a name="line.834"></a>
+<span class="sourceLineNo">835</span> continue;<a name="line.835"></a>
+<span class="sourceLineNo">836</span> }<a name="line.836"></a>
+<span class="sourceLineNo">837</span> LOG.error("Row " + Bytes.toStringBinary(row) + ": missing " +<a name="line.837"></a>
+<span class="sourceLineNo">838</span> Bytes.toStringBinary(family) + ":" + Bytes.toStringBinary(qualifier));<a name="line.838"></a>
+<span class="sourceLineNo">839</span> output.getCounter(Counts.UNREFERENCED).increment(1);<a name="line.839"></a>
+<span class="sourceLineNo">840</span> return;<a name="line.840"></a>
+<span class="sourceLineNo">841</span> }<a name="line.841"></a>
+<span class="sourceLineNo">842</span> if (!Bytes.equals(bytes, 0, bytes.length, value.getBytes(), 0, value.getLength())) {<a name="line.842"></a>
+<span class="sourceLineNo">843</span> if (retries-- > 0) {<a name="line.843"></a>
+<span class="sourceLineNo">844</span> continue;<a name="line.844"></a>
+<span class="sourceLineNo">845</span> }<a name="line.845"></a>
+<span class="sourceLineNo">846</span> LOG.error("Row " + Bytes.toStringBinary(row) + ": " +<a name="line.846"></a>
+<span class="sourceLineNo">847</span> Bytes.toStringBinary(family) + ":" + Bytes.toStringBinary(qualifier) +<a name="line.847"></a>
+<span class="sourceLineNo">848</span> " mismatch");<a name="line.848"></a>
+<span class="sourceLineNo">849</span> output.getCounter(Counts.CORRUPT).increment(1);<a name="line.849"></a>
+<span class="sourceLineNo">850</span> return;<a name="line.850"></a>
+<span class="sourceLineNo">851</span> }<a name="line.851"></a>
+<span class="sourceLineNo">852</span><a name="line.852"></a>
+<span class="sourceLineNo">853</span> }<a name="line.853"></a>
+<span class="sourceLineNo">854</span><a name="line.854"></a>
+<span class="sourceLineNo">855</span> // If we fell through to here all verification checks have succeeded, potentially after<a name="line.855"></a>
+<span class="sourceLineNo">856</span> // retries, and we must exit the while loop.<a name="line.856"></a>
+<span class="sourceLineNo">857</span> output.getCounter(Counts.REFERENCED).increment(1);<a name="line.857"></a>
+<span class="sourceLineNo">858</span> break;<a name="line.858"></a>
+<span class="sourceLineNo">859</span><a name="line.859"></a>
+<span class="sourceLineNo">860</span> }<a name="line.860"></a>
+<span class="sourceLineNo">861</span> }<a name="line.861"></a>
+<span class="sourceLineNo">862</span> }<a name="line.862"></a>
+<span class="sourceLineNo">863</span> }<a name="line.863"></a>
+<span class="sourceLineNo">864</span><a name="line.864"></a>
+<span class="sourceLineNo">865</span>}<a name="line.865"></a>
diff --git a/testdevapidocs/src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.Loader.LoaderMapper.html b/testdevapidocs/src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.Loader.LoaderMapper.html
index 4a40cbc..e89b63c 100644
--- a/testdevapidocs/src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.Loader.LoaderMapper.html
+++ b/testdevapidocs/src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.Loader.LoaderMapper.html
@@ -54,797 +54,823 @@
<span class="sourceLineNo">046</span>import org.apache.hadoop.hbase.MasterNotRunningException;<a name="line.46"></a>
<span class="sourceLineNo">047</span>import org.apache.hadoop.hbase.TableName;<a name="line.47"></a>
<span class="sourceLineNo">048</span>import org.apache.hadoop.hbase.client.Admin;<a name="line.48"></a>
-<span class="sourceLineNo">049</span>import org.apache.hadoop.hbase.client.ColumnFamilyDescriptor;<a name="line.49"></a>
-<span class="sourceLineNo">050</span>import org.apache.hadoop.hbase.client.ColumnFamilyDescriptorBuilder;<a name="line.50"></a>
-<span class="sourceLineNo">051</span>import org.apache.hadoop.hbase.client.Connection;<a name="line.51"></a>
-<span class="sourceLineNo">052</span>import org.apache.hadoop.hbase.client.ConnectionFactory;<a name="line.52"></a>
-<span class="sourceLineNo">053</span>import org.apache.hadoop.hbase.client.Get;<a name="line.53"></a>
-<span class="sourceLineNo">054</span>import org.apache.hadoop.hbase.client.Put;<a name="line.54"></a>
-<span class="sourceLineNo">055</span>import org.apache.hadoop.hbase.client.Result;<a name="line.55"></a>
-<span class="sourceLineNo">056</span>import org.apache.hadoop.hbase.client.Table;<a name="line.56"></a>
-<span class="sourceLineNo">057</span>import org.apache.hadoop.hbase.client.TableDescriptor;<a name="line.57"></a>
-<span class="sourceLineNo">058</span>import org.apache.hadoop.hbase.client.TableDescriptorBuilder;<a name="line.58"></a>
-<span class="sourceLineNo">059</span>import org.apache.hadoop.hbase.io.encoding.DataBlockEncoding;<a name="line.59"></a>
-<span class="sourceLineNo">060</span>import org.apache.hadoop.hbase.mapreduce.TableMapReduceUtil;<a name="line.60"></a>
-<span class="sourceLineNo">061</span>import org.apache.hadoop.hbase.regionserver.BloomType;<a name="line.61"></a>
-<span class="sourceLineNo">062</span>import org.apache.hadoop.hbase.test.util.CRC64;<a name="line.62"></a>
-<span class="sourceLineNo">063</span>import org.apache.hadoop.hbase.test.util.warc.WARCInputFormat;<a name="line.63"></a>
-<span class="sourceLineNo">064</span>import org.apache.hadoop.hbase.test.util.warc.WARCRecord;<a name="line.64"></a>
-<span class="sourceLineNo">065</span>import org.apache.hadoop.hbase.test.util.warc.WARCWritable;<a name="line.65"></a>
-<span class="sourceLineNo">066</span>import org.apache.hadoop.hbase.util.Bytes;<a name="line.66"></a>
-<span class="sourceLineNo">067</span>import org.apache.hadoop.hbase.util.EnvironmentEdgeManager;<a name="line.67"></a>
-<span class="sourceLineNo">068</span>import org.apache.hadoop.hbase.util.RegionSplitter;<a name="line.68"></a>
-<span class="sourceLineNo">069</span>import org.apache.hadoop.io.BytesWritable;<a name="line.69"></a>
-<span class="sourceLineNo">070</span>import org.apache.hadoop.io.LongWritable;<a name="line.70"></a>
-<span class="sourceLineNo">071</span>import org.apache.hadoop.io.NullWritable;<a name="line.71"></a>
-<span class="sourceLineNo">072</span>import org.apache.hadoop.io.SequenceFile.CompressionType;<a name="line.72"></a>
-<span class="sourceLineNo">073</span>import org.apache.hadoop.io.Writable;<a name="line.73"></a>
-<span class="sourceLineNo">074</span>import org.apache.hadoop.mapreduce.Counters;<a name="line.74"></a>
-<span class="sourceLineNo">075</span>import org.apache.hadoop.mapreduce.Job;<a name="line.75"></a>
-<span class="sourceLineNo">076</span>import org.apache.hadoop.mapreduce.JobContext;<a name="line.76"></a>
-<span class="sourceLineNo">077</span>import org.apache.hadoop.mapreduce.Mapper;<a name="line.77"></a>
-<span class="sourceLineNo">078</span>import org.apache.hadoop.mapreduce.lib.input.FileInputFormat;<a name="line.78"></a>
-<span class="sourceLineNo">079</span>import org.apache.hadoop.mapreduce.lib.input.SequenceFileInputFormat;<a name="line.79"></a>
-<span class="sourceLineNo">080</span>import org.apache.hadoop.mapreduce.lib.output.NullOutputFormat;<a name="line.80"></a>
-<span class="sourceLineNo">081</span>import org.apache.hadoop.mapreduce.lib.output.SequenceFileOutputFormat;<a name="line.81"></a>
-<span class="sourceLineNo">082</span>import org.apache.hadoop.util.Tool;<a name="line.82"></a>
-<span class="sourceLineNo">083</span>import org.apache.hadoop.util.ToolRunner;<a name="line.83"></a>
-<span class="sourceLineNo">084</span>import org.slf4j.Logger;<a name="line.84"></a>
-<span class="sourceLineNo">085</span>import org.slf4j.LoggerFactory;<a name="line.85"></a>
-<span class="sourceLineNo">086</span><a name="line.86"></a>
-<span class="sourceLineNo">087</span>import org.apache.hbase.thirdparty.org.apache.commons.cli.CommandLine;<a name="line.87"></a>
-<span class="sourceLineNo">088</span><a name="line.88"></a>
-<span class="sourceLineNo">089</span>/**<a name="line.89"></a>
-<span class="sourceLineNo">090</span> * This integration test loads successful resource retrieval records from the Common Crawl<a name="line.90"></a>
-<span class="sourceLineNo">091</span> * (https://commoncrawl.org/) public dataset into an HBase table and writes records that can be<a name="line.91"></a>
-<span class="sourceLineNo">092</span> * used to later verify the presence and integrity of those records.<a name="line.92"></a>
-<span class="sourceLineNo">093</span> * <p><a name="line.93"></a>
-<span class="sourceLineNo">094</span> * Run like:<a name="line.94"></a>
-<span class="sourceLineNo">095</span> * <blockquote><a name="line.95"></a>
-<span class="sourceLineNo">096</span> * ./bin/hbase org.apache.hadoop.hbase.test.IntegrationTestLoadCommonCrawl \<br><a name="line.96"></a>
-<span class="sourceLineNo">097</span> * &nbsp;&nbsp; -Dfs.s3n.awsAccessKeyId=&lt;AWS access key&gt; \<br><a name="line.97"></a>
-<span class="sourceLineNo">098</span> * &nbsp;&nbsp; -Dfs.s3n.awsSecretAccessKey=&lt;AWS secret key&gt; \<br><a name="line.98"></a>
-<span class="sourceLineNo">099</span> * &nbsp;&nbsp; /path/to/test-CC-MAIN-2021-10-warc.paths.gz \<br><a name="line.99"></a>
-<span class="sourceLineNo">100</span> * &nbsp;&nbsp; /path/to/tmp/warc-loader-output<a name="line.100"></a>
-<span class="sourceLineNo">101</span> * </blockquote><a name="line.101"></a>
-<span class="sourceLineNo">102</span> * <p><a name="line.102"></a>
-<span class="sourceLineNo">103</span> * Access to the Common Crawl dataset in S3 is made available to anyone by Amazon AWS, but<a name="line.103"></a>
-<span class="sourceLineNo">104</span> * Hadoop's S3N filesystem still requires valid access credentials to initialize.<a name="line.104"></a>
-<span class="sourceLineNo">105</span> * <p><a name="line.105"></a>
-<span class="sourceLineNo">106</span> * The input path can either specify a directory or a file. The file may optionally be<a name="line.106"></a>
-<span class="sourceLineNo">107</span> * compressed with gzip. If a directory, the loader expects the directory to contain one or more<a name="line.107"></a>
-<span class="sourceLineNo">108</span> * WARC files from the Common Crawl dataset. If a file, the loader expects a list of Hadoop S3N<a name="line.108"></a>
-<span class="sourceLineNo">109</span> * URIs which point to S3 locations for one or more WARC files from the Common Crawl dataset,<a name="line.109"></a>
-<span class="sourceLineNo">110</span> * one URI per line. Lines should be terminated with the UNIX line terminator.<a name="line.110"></a>
-<span class="sourceLineNo">111</span> * <p><a name="line.111"></a>
-<span class="sourceLineNo">112</span> * Included in hbase-it/src/test/resources/CC-MAIN-2021-10-warc.paths.gz is a list of all WARC<a name="line.112"></a>
-<span class="sourceLineNo">113</span> * files comprising the Q1 2021 crawl archive. There are 64,000 WARC files in this data set, each<a name="line.113"></a>
-<span class="sourceLineNo">114</span> * containing ~1GB of gzipped data. The WARC files contain several record types, such as metadata,<a name="line.114"></a>
-<span class="sourceLineNo">115</span> * request, and response, but we only load the response record types. If the HBase table schema<a name="line.115"></a>
-<span class="sourceLineNo">116</span> * does not specify compression (by default) there is roughly a 10x expansion. Loading the full<a name="line.116"></a>
-<span class="sourceLineNo">117</span> * crawl archive results in a table approximately 640 TB in size.<a name="line.117"></a>
-<span class="sourceLineNo">118</span> * <p><a name="line.118"></a>
-<span class="sourceLineNo">119</span> * You can also split the Loader and Verify stages:<a name="line.119"></a>
-<span class="sourceLineNo">120</span> * <p><a name="line.120"></a>
-<span class="sourceLineNo">121</span> * Load with:<a name="line.121"></a>
-<span class="sourceLineNo">122</span> * <blockquote><a name="line.122"></a>
-<span class="sourceLineNo">123</span> * ./bin/hbase 'org.apache.hadoop.hbase.test.IntegrationTestLoadCommonCrawl$Loader' \<br><a name="line.123"></a>
-<span class="sourceLineNo">124</span> * &nbsp;&nbsp; -files /path/to/hadoop-aws.jar \<br><a name="line.124"></a>
-<span class="sourceLineNo">125</span> * &nbsp;&nbsp; -Dfs.s3n.awsAccessKeyId=&lt;AWS access key&gt; \<br><a name="line.125"></a>
-<span class="sourceLineNo">126</span> * &nbsp;&nbsp; -Dfs.s3n.awsSecretAccessKey=&lt;AWS secret key&gt; \<br><a name="line.126"></a>
-<span class="sourceLineNo">127</span> * &nbsp;&nbsp; /path/to/test-CC-MAIN-2021-10-warc.paths.gz \<br><a name="line.127"></a>
-<span class="sourceLineNo">128</span> * &nbsp;&nbsp; /path/to/tmp/warc-loader-output<a name="line.128"></a>
-<span class="sourceLineNo">129</span> * </blockquote><a name="line.129"></a>
-<span class="sourceLineNo">130</span> * <p><a name="line.130"></a>
-<span class="sourceLineNo">131</span> * Note: The hadoop-aws jar will be needed at runtime to instantiate the S3N filesystem. Use<a name="line.131"></a>
-<span class="sourceLineNo">132</span> * the <tt>-files</tt> ToolRunner argument to add it.<a name="line.132"></a>
-<span class="sourceLineNo">133</span> * <p><a name="line.133"></a>
-<span class="sourceLineNo">134</span> * Verify with:<a name="line.134"></a>
-<span class="sourceLineNo">135</span> * <blockquote><a name="line.135"></a>
-<span class="sourceLineNo">136</span> * ./bin/hbase 'org.apache.hadoop.hbase.test.IntegrationTestLoadCommonCrawl$Verify' \<br><a name="line.136"></a>
-<span class="sourceLineNo">137</span> * &nbsp;&nbsp; /path/to/tmp/warc-loader-output<a name="line.137"></a>
-<span class="sourceLineNo">138</span> * </blockquote><a name="line.138"></a>
-<span class="sourceLineNo">139</span> * <p><a name="line.139"></a>
-<span class="sourceLineNo">140</span> */<a name="line.140"></a>
-<span class="sourceLineNo">141</span>public class IntegrationTestLoadCommonCrawl extends IntegrationTestBase {<a name="line.141"></a>
-<span class="sourceLineNo">142</span><a name="line.142"></a>
-<span class="sourceLineNo">143</span> private static final Logger LOG = LoggerFactory.getLogger(IntegrationTestLoadCommonCrawl.class);<a name="line.143"></a>
-<span class="sourceLineNo">144</span><a name="line.144"></a>
-<span class="sourceLineNo">145</span> protected static String TABLE_NAME_KEY = "IntegrationTestLoadCommonCrawl.table";<a name="line.145"></a>
-<span class="sourceLineNo">146</span> protected static String DEFAULT_TABLE_NAME = "IntegrationTestLoadCommonCrawl";<a name="line.146"></a>
-<span class="sourceLineNo">147</span><a name="line.147"></a>
-<span class="sourceLineNo">148</span> protected static byte[] CONTENT_FAMILY_NAME = Bytes.toBytes("c");<a name="line.148"></a>
-<span class="sourceLineNo">149</span> protected static byte[] INFO_FAMILY_NAME = Bytes.toBytes("i");<a name="line.149"></a>
-<span class="sourceLineNo">150</span> protected static byte[] CONTENT_QUALIFIER = HConstants.EMPTY_BYTE_ARRAY;<a name="line.150"></a>
-<span class="sourceLineNo">151</span> protected static byte[] CONTENT_LENGTH_QUALIFIER = Bytes.toBytes("l");<a name="line.151"></a>
-<span class="sourceLineNo">152</span> protected static byte[] CONTENT_TYPE_QUALIFIER = Bytes.toBytes("t");<a name="line.152"></a>
-<span class="sourceLineNo">153</span> protected static byte[] CRC_QUALIFIER = Bytes.toBytes("c");<a name="line.153"></a>
-<span class="sourceLineNo">154</span> protected static byte[] DATE_QUALIFIER = Bytes.toBytes("d");<a name="line.154"></a>
-<span class="sourceLineNo">155</span> protected static byte[] IP_ADDRESS_QUALIFIER = Bytes.toBytes("a");<a name="line.155"></a>
-<span class="sourceLineNo">156</span> protected static byte[] RECORD_ID_QUALIFIER = Bytes.toBytes("r");<a name="line.156"></a>
-<span class="sourceLineNo">157</span> protected static byte[] TARGET_URI_QUALIFIER = Bytes.toBytes("u");<a name="line.157"></a>
-<span class="sourceLineNo">158</span><a name="line.158"></a>
-<span class="sourceLineNo">159</span> public static enum Counts {<a name="line.159"></a>
-<span class="sourceLineNo">160</span> REFERENCED, UNREFERENCED, CORRUPT<a name="line.160"></a>
-<span class="sourceLineNo">161</span> }<a name="line.161"></a>
-<span class="sourceLineNo">162</span><a name="line.162"></a>
-<span class="sourceLineNo">163</span> Path warcFileInputDir = null;<a name="line.163"></a>
-<span class="sourceLineNo">164</span> Path outputDir = null;<a name="line.164"></a>
-<span class="sourceLineNo">165</span> String[] args;<a name="line.165"></a>
-<span class="sourceLineNo">166</span><a name="line.166"></a>
-<span class="sourceLineNo">167</span> protected int runLoader(Path warcFileInputDir, Path outputDir) throws Exception {<a name="line.167"></a>
-<span class="sourceLineNo">168</span> Loader loader = new Loader();<a name="line.168"></a>
-<span class="sourceLineNo">169</span> loader.setConf(conf);<a name="line.169"></a>
-<span class="sourceLineNo">170</span> return loader.run(warcFileInputDir, outputDir);<a name="line.170"></a>
-<span class="sourceLineNo">171</span> }<a name="line.171"></a>
-<span class="sourceLineNo">172</span><a name="line.172"></a>
-<span class="sourceLineNo">173</span> protected int runVerify(Path inputDir) throws Exception {<a name="line.173"></a>
-<span class="sourceLineNo">174</span> Verify verify = new Verify();<a name="line.174"></a>
-<span class="sourceLineNo">175</span> verify.setConf(conf);<a name="line.175"></a>
-<span class="sourceLineNo">176</span> return verify.run(inputDir);<a name="line.176"></a>
-<span class="sourceLineNo">177</span> }<a name="line.177"></a>
-<span class="sourceLineNo">178</span><a name="line.178"></a>
-<span class="sourceLineNo">179</span> @Override<a name="line.179"></a>
-<span class="sourceLineNo">180</span> public int run(String[] args) {<a name="line.180"></a>
-<span class="sourceLineNo">181</span> if (args.length > 0) {<a name="line.181"></a>
-<span class="sourceLineNo">182</span> warcFileInputDir = new Path(args[0]);<a name="line.182"></a>
-<span class="sourceLineNo">183</span> if (args.length > 1) {<a name="line.183"></a>
-<span class="sourceLineNo">184</span> outputDir = new Path(args[1]);<a name="line.184"></a>
-<span class="sourceLineNo">185</span> }<a name="line.185"></a>
-<span class="sourceLineNo">186</span> }<a name="line.186"></a>
-<span class="sourceLineNo">187</span> try {<a name="line.187"></a>
-<span class="sourceLineNo">188</span> if (warcFileInputDir == null) {<a name="line.188"></a>
-<span class="sourceLineNo">189</span> throw new IllegalArgumentException("WARC input file or directory not specified");<a name="line.189"></a>
-<span class="sourceLineNo">190</span> }<a name="line.190"></a>
-<span class="sourceLineNo">191</span> if (outputDir == null) {<a name="line.191"></a>
-<span class="sourceLineNo">192</span> throw new IllegalArgumentException("Output directory not specified");<a name="line.192"></a>
+<span class="sourceLineNo">049</span>import org.apache.hadoop.hbase.client.BufferedMutator;<a name="line.49"></a>
+<span class="sourceLineNo">050</span>import org.apache.hadoop.hbase.client.ColumnFamilyDescriptor;<a name="line.50"></a>
+<span class="sourceLineNo">051</span>import org.apache.hadoop.hbase.client.ColumnFamilyDescriptorBuilder;<a name="line.51"></a>
+<span class="sourceLineNo">052</span>import org.apache.hadoop.hbase.client.Connection;<a name="line.52"></a>
+<span class="sourceLineNo">053</span>import org.apache.hadoop.hbase.client.ConnectionFactory;<a name="line.53"></a>
+<span class="sourceLineNo">054</span>import org.apache.hadoop.hbase.client.Get;<a name="line.54"></a>
+<span class="sourceLineNo">055</span>import org.apache.hadoop.hbase.client.Put;<a name="line.55"></a>
+<span class="sourceLineNo">056</span>import org.apache.hadoop.hbase.client.Result;<a name="line.56"></a>
+<span class="sourceLineNo">057</span>import org.apache.hadoop.hbase.client.Table;<a name="line.57"></a>
+<span class="sourceLineNo">058</span>import org.apache.hadoop.hbase.client.TableDescriptor;<a name="line.58"></a>
+<span class="sourceLineNo">059</span>import org.apache.hadoop.hbase.client.TableDescriptorBuilder;<a name="line.59"></a>
+<span class="sourceLineNo">060</span>import org.apache.hadoop.hbase.io.encoding.DataBlockEncoding;<a name="line.60"></a>
+<span class="sourceLineNo">061</span>import org.apache.hadoop.hbase.mapreduce.TableMapReduceUtil;<a name="line.61"></a>
+<span class="sourceLineNo">062</span>import org.apache.hadoop.hbase.regionserver.BloomType;<a name="line.62"></a>
+<span class="sourceLineNo">063</span>import org.apache.hadoop.hbase.test.util.CRC64;<a name="line.63"></a>
+<span class="sourceLineNo">064</span>import org.apache.hadoop.hbase.test.util.warc.WARCInputFormat;<a name="line.64"></a>
+<span class="sourceLineNo">065</span>import org.apache.hadoop.hbase.test.util.warc.WARCRecord;<a name="line.65"></a>
+<span class="sourceLineNo">066</span>import org.apache.hadoop.hbase.test.util.warc.WARCWritable;<a name="line.66"></a>
+<span class="sourceLineNo">067</span>import org.apache.hadoop.hbase.util.Bytes;<a name="line.67"></a>
+<span class="sourceLineNo">068</span>import org.apache.hadoop.hbase.util.EnvironmentEdgeManager;<a name="line.68"></a>
+<span class="sourceLineNo">069</span>import org.apache.hadoop.hbase.util.RegionSplitter;<a name="line.69"></a>
+<span class="sourceLineNo">070</span>import org.apache.hadoop.io.BytesWritable;<a name="line.70"></a>
+<span class="sourceLineNo">071</span>import org.apache.hadoop.io.LongWritable;<a name="line.71"></a>
+<span class="sourceLineNo">072</span>import org.apache.hadoop.io.NullWritable;<a name="line.72"></a>
+<span class="sourceLineNo">073</span>import org.apache.hadoop.io.SequenceFile.CompressionType;<a name="line.73"></a>
+<span class="sourceLineNo">074</span>import org.apache.hadoop.io.Writable;<a name="line.74"></a>
+<span class="sourceLineNo">075</span>import org.apache.hadoop.mapreduce.Counters;<a name="line.75"></a>
+<span class="sourceLineNo">076</span>import org.apache.hadoop.mapreduce.Job;<a name="line.76"></a>
+<span class="sourceLineNo">077</span>import org.apache.hadoop.mapreduce.JobContext;<a name="line.77"></a>
+<span class="sourceLineNo">078</span>import org.apache.hadoop.mapreduce.Mapper;<a name="line.78"></a>
+<span class="sourceLineNo">079</span>import org.apache.hadoop.mapreduce.lib.input.FileInputFormat;<a name="line.79"></a>
+<span class="sourceLineNo">080</span>import org.apache.hadoop.mapreduce.lib.input.SequenceFileInputFormat;<a name="line.80"></a>
+<span class="sourceLineNo">081</span>import org.apache.hadoop.mapreduce.lib.output.NullOutputFormat;<a name="line.81"></a>
+<span class="sourceLineNo">082</span>import org.apache.hadoop.mapreduce.lib.output.SequenceFileOutputFormat;<a name="line.82"></a>
+<span class="sourceLineNo">083</span>import org.apache.hadoop.util.Tool;<a name="line.83"></a>
+<span class="sourceLineNo">084</span>import org.apache.hadoop.util.ToolRunner;<a name="line.84"></a>
+<span class="sourceLineNo">085</span>import org.slf4j.Logger;<a name="line.85"></a>
+<span class="sourceLineNo">086</span>import org.slf4j.LoggerFactory;<a name="line.86"></a>
+<span class="sourceLineNo">087</span><a name="line.87"></a>
+<span class="sourceLineNo">088</span>import org.apache.hbase.thirdparty.org.apache.commons.cli.CommandLine;<a name="line.88"></a>
+<span class="sourceLineNo">089</span><a name="line.89"></a>
+<span class="sourceLineNo">090</span>/**<a name="line.90"></a>
+<span class="sourceLineNo">091</span> * This integration test loads successful resource retrieval records from the Common Crawl<a name="line.91"></a>
+<span class="sourceLineNo">092</span> * (https://commoncrawl.org/) public dataset into an HBase table and writes records that can be<a name="line.92"></a>
+<span class="sourceLineNo">093</span> * used to later verify the presence and integrity of those records.<a name="line.93"></a>
+<span class="sourceLineNo">094</span> * <p><a name="line.94"></a>
+<span class="sourceLineNo">095</span> * Run like:<a name="line.95"></a>
+<span class="sourceLineNo">096</span> * <blockquote><a name="line.96"></a>
+<span class="sourceLineNo">097</span> * ./bin/hbase org.apache.hadoop.hbase.test.IntegrationTestLoadCommonCrawl \<br><a name="line.97"></a>
+<span class="sourceLineNo">098</span> * &nbsp;&nbsp; -Dfs.s3n.awsAccessKeyId=&lt;AWS access key&gt; \<br><a name="line.98"></a>
+<span class="sourceLineNo">099</span> * &nbsp;&nbsp; -Dfs.s3n.awsSecretAccessKey=&lt;AWS secret key&gt; \<br><a name="line.99"></a>
+<span class="sourceLineNo">100</span> * &nbsp;&nbsp; /path/to/test-CC-MAIN-2021-10-warc.paths.gz \<br><a name="line.100"></a>
+<span class="sourceLineNo">101</span> * &nbsp;&nbsp; /path/to/tmp/warc-loader-output<a name="line.101"></a>
+<span class="sourceLineNo">102</span> * </blockquote><a name="line.102"></a>
+<span class="sourceLineNo">103</span> * <p><a name="line.103"></a>
+<span class="sourceLineNo">104</span> * Access to the Common Crawl dataset in S3 is made available to anyone by Amazon AWS, but<a name="line.104"></a>
+<span class="sourceLineNo">105</span> * Hadoop's S3N filesystem still requires valid access credentials to initialize.<a name="line.105"></a>
+<span class="sourceLineNo">106</span> * <p><a name="line.106"></a>
+<span class="sourceLineNo">107</span> * The input path can either specify a directory or a file. The file may optionally be<a name="line.107"></a>
+<span class="sourceLineNo">108</span> * compressed with gzip. If a directory, the loader expects the directory to contain one or more<a name="line.108"></a>
+<span class="sourceLineNo">109</span> * WARC files from the Common Crawl dataset. If a file, the loader expects a list of Hadoop S3N<a name="line.109"></a>
+<span class="sourceLineNo">110</span> * URIs which point to S3 locations for one or more WARC files from the Common Crawl dataset,<a name="line.110"></a>
+<span class="sourceLineNo">111</span> * one URI per line. Lines should be terminated with the UNIX line terminator.<a name="line.111"></a>
+<span class="sourceLineNo">112</span> * <p><a name="line.112"></a>
+<span class="sourceLineNo">113</span> * Included in hbase-it/src/test/resources/CC-MAIN-2021-10-warc.paths.gz is a list of all WARC<a name="line.113"></a>
+<span class="sourceLineNo">114</span> * files comprising the Q1 2021 crawl archive. There are 64,000 WARC files in this data set, each<a name="line.114"></a>
+<span class="sourceLineNo">115</span> * containing ~1GB of gzipped data. The WARC files contain several record types, such as metadata,<a name="line.115"></a>
+<span class="sourceLineNo">116</span> * request, and response, but we only load the response record types. If the HBase table schema<a name="line.116"></a>
+<span class="sourceLineNo">117</span> * does not specify compression (by default) there is roughly a 10x expansion. Loading the full<a name="line.117"></a>
+<span class="sourceLineNo">118</span> * crawl archive results in a table approximately 640 TB in size.<a name="line.118"></a>
+<span class="sourceLineNo">119</span> * <p><a name="line.119"></a>
+<span class="sourceLineNo">120</span> * You can also split the Loader and Verify stages:<a name="line.120"></a>
+<span class="sourceLineNo">121</span> * <p><a name="line.121"></a>
+<span class="sourceLineNo">122</span> * Load with:<a name="line.122"></a>
+<span class="sourceLineNo">123</span> * <blockquote><a name="line.123"></a>
+<span class="sourceLineNo">124</span> * ./bin/hbase 'org.apache.hadoop.hbase.test.IntegrationTestLoadCommonCrawl$Loader' \<br><a name="line.124"></a>
+<span class="sourceLineNo">125</span> * &nbsp;&nbsp; -files /path/to/hadoop-aws.jar \<br><a name="line.125"></a>
+<span class="sourceLineNo">126</span> * &nbsp;&nbsp; -Dfs.s3n.awsAccessKeyId=&lt;AWS access key&gt; \<br><a name="line.126"></a>
+<span class="sourceLineNo">127</span> * &nbsp;&nbsp; -Dfs.s3n.awsSecretAccessKey=&lt;AWS secret key&gt; \<br><a name="line.127"></a>
+<span class="sourceLineNo">128</span> * &nbsp;&nbsp; /path/to/test-CC-MAIN-2021-10-warc.paths.gz \<br><a name="line.128"></a>
+<span class="sourceLineNo">129</span> * &nbsp;&nbsp; /path/to/tmp/warc-loader-output<a name="line.129"></a>
+<span class="sourceLineNo">130</span> * </blockquote><a name="line.130"></a>
+<span class="sourceLineNo">131</span> * <p><a name="line.131"></a>
+<span class="sourceLineNo">132</span> * Note: The hadoop-aws jar will be needed at runtime to instantiate the S3N filesystem. Use<a name="line.132"></a>
+<span class="sourceLineNo">133</span> * the <tt>-files</tt> ToolRunner argument to add it.<a name="line.133"></a>
+<span class="sourceLineNo">134</span> * <p><a name="line.134"></a>
+<span class="sourceLineNo">135</span> * Verify with:<a name="line.135"></a>
+<span class="sourceLineNo">136</span> * <blockquote><a name="line.136"></a>
+<span class="sourceLineNo">137</span> * ./bin/hbase 'org.apache.hadoop.hbase.test.IntegrationTestLoadCommonCrawl$Verify' \<br><a name="line.137"></a>
+<span class="sourceLineNo">138</span> * &nbsp;&nbsp; /path/to/tmp/warc-loader-output<a name="line.138"></a>
+<span class="sourceLineNo">139</span> * </blockquote><a name="line.139"></a>
+<span class="sourceLineNo">140</span> * <p><a name="line.140"></a>
+<span class="sourceLineNo">141</span> */<a name="line.141"></a>
+<span class="sourceLineNo">142</span>public class IntegrationTestLoadCommonCrawl extends IntegrationTestBase {<a name="line.142"></a>
+<span class="sourceLineNo">143</span><a name="line.143"></a>
+<span class="sourceLineNo">144</span> private static final Logger LOG = LoggerFactory.getLogger(IntegrationTestLoadCommonCrawl.class);<a name="line.144"></a>
+<span class="sourceLineNo">145</span><a name="line.145"></a>
+<span class="sourceLineNo">146</span> protected static String TABLE_NAME_KEY = "IntegrationTestLoadCommonCrawl.table";<a name="line.146"></a>
+<span class="sourceLineNo">147</span> protected static String DEFAULT_TABLE_NAME = "IntegrationTestLoadCommonCrawl";<a name="line.147"></a>
+<span class="sourceLineNo">148</span><a name="line.148"></a>
+<span class="sourceLineNo">149</span> protected static byte[] CONTENT_FAMILY_NAME = Bytes.toBytes("c");<a name="line.149"></a>
+<span class="sourceLineNo">150</span> protected static byte[] INFO_FAMILY_NAME = Bytes.toBytes("i");<a name="line.150"></a>
+<span class="sourceLineNo">151</span> protected static byte[] CONTENT_QUALIFIER = HConstants.EMPTY_BYTE_ARRAY;<a name="line.151"></a>
+<span class="sourceLineNo">152</span> protected static byte[] CONTENT_LENGTH_QUALIFIER = Bytes.toBytes("l");<a name="line.152"></a>
+<span class="sourceLineNo">153</span> protected static byte[] CONTENT_TYPE_QUALIFIER = Bytes.toBytes("t");<a name="line.153"></a>
+<span class="sourceLineNo">154</span> protected static byte[] CRC_QUALIFIER = Bytes.toBytes("c");<a name="line.154"></a>
+<span class="sourceLineNo">155</span> protected static byte[] DATE_QUALIFIER = Bytes.toBytes("d");<a name="line.155"></a>
+<span class="sourceLineNo">156</span> protected static byte[] IP_ADDRESS_QUALIFIER = Bytes.toBytes("a");<a name="line.156"></a>
+<span class="sourceLineNo">157</span> protected static byte[] RECORD_ID_QUALIFIER = Bytes.toBytes("r");<a name="line.157"></a>
+<span class="sourceLineNo">158</span> protected static byte[] TARGET_URI_QUALIFIER = Bytes.toBytes("u");<a name="line.158"></a>
+<span class="sourceLineNo">159</span><a name="line.159"></a>
+<span class="sourceLineNo">160</span> private static final int VERIFICATION_READ_RETRIES = 10;<a name="line.160"></a>
+<span class="sourceLineNo">161</span><a name="line.161"></a>
+<span class="sourceLineNo">162</span> public static enum Counts {<a name="line.162"></a>
+<span class="sourceLineNo">163</span> REFERENCED, UNREFERENCED, CORRUPT<a name="line.163"></a>
+<span class="sourceLineNo">164</span> }<a name="line.164"></a>
+<span class="sourceLineNo">165</span><a name="line.165"></a>
+<span class="sourceLineNo">166</span> Path warcFileInputDir = null;<a name="line.166"></a>
+<span class="sourceLineNo">167</span> Path outputDir = null;<a name="line.167"></a>
+<span class="sourceLineNo">168</span> String[] args;<a name="line.168"></a>
+<span class="sourceLineNo">169</span><a name="line.169"></a>
+<span class="sourceLineNo">170</span> protected int runLoader(Path warcFileInputDir, Path outputDir) throws Exception {<a name="line.170"></a>
+<span class="sourceLineNo">171</span> Loader loader = new Loader();<a name="line.171"></a>
+<span class="sourceLineNo">172</span> loader.setConf(conf);<a name="line.172"></a>
+<span class="sourceLineNo">173</span> return loader.run(warcFileInputDir, outputDir);<a name="line.173"></a>
+<span class="sourceLineNo">174</span> }<a name="line.174"></a>
+<span class="sourceLineNo">175</span><a name="line.175"></a>
+<span class="sourceLineNo">176</span> protected int runVerify(Path inputDir) throws Exception {<a name="line.176"></a>
+<span class="sourceLineNo">177</span> Verify verify = new Verify();<a name="line.177"></a>
+<span class="sourceLineNo">178</span> verify.setConf(conf);<a name="line.178"></a>
+<span class="sourceLineNo">179</span> return verify.run(inputDir);<a name="line.179"></a>
+<span class="sourceLineNo">180</span> }<a name="line.180"></a>
+<span class="sourceLineNo">181</span><a name="line.181"></a>
+<span class="sourceLineNo">182</span> @Override<a name="line.182"></a>
+<span class="sourceLineNo">183</span> public int run(String[] args) {<a name="line.183"></a>
+<span class="sourceLineNo">184</span> if (args.length > 0) {<a name="line.184"></a>
+<span class="sourceLineNo">185</span> warcFileInputDir = new Path(args[0]);<a name="line.185"></a>
+<span class="sourceLineNo">186</span> if (args.length > 1) {<a name="line.186"></a>
+<span class="sourceLineNo">187</span> outputDir = new Path(args[1]);<a name="line.187"></a>
+<span class="sourceLineNo">188</span> }<a name="line.188"></a>
+<span class="sourceLineNo">189</span> }<a name="line.189"></a>
+<span class="sourceLineNo">190</span> try {<a name="line.190"></a>
+<span class="sourceLineNo">191</span> if (warcFileInputDir == null) {<a name="line.191"></a>
+<span class="sourceLineNo">192</span> throw new IllegalArgumentException("WARC input file or directory not specified");<a name="line.192"></a>
<span class="sourceLineNo">193</span> }<a name="line.193"></a>
-<span class="sourceLineNo">194</span> int res = runLoader(warcFileInputDir, outputDir);<a name="line.194"></a>
-<span class="sourceLineNo">195</span> if (res != 0) {<a name="line.195"></a>
-<span class="sourceLineNo">196</span> LOG.error("Loader failed");<a name="line.196"></a>
-<span class="sourceLineNo">197</span> return -1;<a name="line.197"></a>
-<span class="sourceLineNo">198</span> }<a name="line.198"></a>
-<span class="sourceLineNo">199</span> res = runVerify(outputDir);<a name="line.199"></a>
-<span class="sourceLineNo">200</span> } catch (Exception e) {<a name="line.200"></a>
-<span class="sourceLineNo">201</span> LOG.error("Tool failed with exception", e);<a name="line.201"></a>
-<span class="sourceLineNo">202</span> return -1;<a name="line.202"></a>
-<span class="sourceLineNo">203</span> }<a name="line.203"></a>
-<span class="sourceLineNo">204</span> return 0;<a name="line.204"></a>
-<span class="sourceLineNo">205</span> }<a name="line.205"></a>
-<span class="sourceLineNo">206</span><a name="line.206"></a>
-<span class="sourceLineNo">207</span> @Override<a name="line.207"></a>
-<span class="sourceLineNo">208</span> protected void processOptions(CommandLine cmd) {<a name="line.208"></a>
-<span class="sourceLineNo">209</span> processBaseOptions(cmd);<a name="line.209"></a>
-<span class="sourceLineNo">210</span> args = cmd.getArgs();<a name="line.210"></a>
-<span class="sourceLineNo">211</span> }<a name="line.211"></a>
-<span class="sourceLineNo">212</span><a name="line.212"></a>
-<span class="sourceLineNo">213</span> @Override<a name="line.213"></a>
-<span class="sourceLineNo">214</span> public void setUpCluster() throws Exception {<a name="line.214"></a>
-<span class="sourceLineNo">215</span> util = getTestingUtil(getConf());<a name="line.215"></a>
-<span class="sourceLineNo">216</span> boolean isDistributed = util.isDistributedCluster();<a name="line.216"></a>
-<span class="sourceLineNo">217</span> util.initializeCluster(isDistributed ? 1 : 3);<a name="line.217"></a>
-<span class="sourceLineNo">218</span> if (!isDistributed) {<a name="line.218"></a>
-<span class="sourceLineNo">219</span> util.startMiniMapReduceCluster();<a name="line.219"></a>
-<span class="sourceLineNo">220</span> }<a name="line.220"></a>
-<span class="sourceLineNo">221</span> this.setConf(util.getConfiguration());<a name="line.221"></a>
-<span class="sourceLineNo">222</span> }<a name="line.222"></a>
-<span class="sourceLineNo">223</span><a name="line.223"></a>
-<span class="sourceLineNo">224</span> @Override<a name="line.224"></a>
-<span class="sourceLineNo">225</span> public void cleanUpCluster() throws Exception {<a name="line.225"></a>
-<span class="sourceLineNo">226</span> super.cleanUpCluster();<a name="line.226"></a>
-<span class="sourceLineNo">227</span> if (util.isDistributedCluster()) {<a name="line.227"></a>
-<span class="sourceLineNo">228</span> util.shutdownMiniMapReduceCluster();<a name="line.228"></a>
-<span class="sourceLineNo">229</span> }<a name="line.229"></a>
-<span class="sourceLineNo">230</span> }<a name="line.230"></a>
-<span class="sourceLineNo">231</span><a name="line.231"></a>
-<span class="sourceLineNo">232</span> static TableName getTablename(Configuration c) {<a name="line.232"></a>
-<span class="sourceLineNo">233</span> return TableName.valueOf(c.get(TABLE_NAME_KEY, DEFAULT_TABLE_NAME));<a name="line.233"></a>
-<span class="sourceLineNo">234</span> }<a name="line.234"></a>
-<span class="sourceLineNo">235</span><a name="line.235"></a>
-<span class="sourceLineNo">236</span> @Override<a name="line.236"></a>
-<span class="sourceLineNo">237</span> public TableName getTablename() {<a name="line.237"></a>
-<span class="sourceLineNo">238</span> return getTablename(getConf());<a name="line.238"></a>
-<span class="sourceLineNo">239</span> }<a name="line.239"></a>
-<span class="sourceLineNo">240</span><a name="line.240"></a>
-<span class="sourceLineNo">241</span> @Override<a name="line.241"></a>
-<span class="sourceLineNo">242</span> protected Set<String> getColumnFamilies() {<a name="line.242"></a>
-<span class="sourceLineNo">243</span> Set<String> families = new HashSet<>();<a name="line.243"></a>
-<span class="sourceLineNo">244</span> families.add(Bytes.toString(CONTENT_FAMILY_NAME));<a name="line.244"></a>
-<span class="sourceLineNo">245</span> families.add(Bytes.toString(INFO_FAMILY_NAME));<a name="line.245"></a>
-<span class="sourceLineNo">246</span> return families;<a name="line.246"></a>
-<span class="sourceLineNo">247</span> }<a name="line.247"></a>
-<span class="sourceLineNo">248</span><a name="line.248"></a>
-<span class="sourceLineNo">249</span> @Override<a name="line.249"></a>
-<span class="sourceLineNo">250</span> public int runTestFromCommandLine() throws Exception {<a name="line.250"></a>
-<span class="sourceLineNo">251</span> return ToolRunner.run(getConf(), this, args);<a name="line.251"></a>
-<span class="sourceLineNo">252</span> }<a name="line.252"></a>
-<span class="sourceLineNo">253</span><a name="line.253"></a>
-<span class="sourceLineNo">254</span> public static void main(String[] args) throws Exception {<a name="line.254"></a>
-<span class="sourceLineNo">255</span> Configuration conf = HBaseConfiguration.create();<a name="line.255"></a>
-<span class="sourceLineNo">256</span> IntegrationTestingUtility.setUseDistributedCluster(conf);<a name="line.256"></a>
-<span class="sourceLineNo">257</span> int ret = ToolRunner.run(conf, new IntegrationTestLoadCommonCrawl(), args);<a name="line.257"></a>
-<span class="sourceLineNo">258</span> System.exit(ret);<a name="line.258"></a>
-<span class="sourceLineNo">259</span> }<a name="line.259"></a>
-<span class="sourceLineNo">260</span><a name="line.260"></a>
-<span class="sourceLineNo">261</span> public static class HBaseKeyWritable implements Writable {<a name="line.261"></a>
-<span class="sourceLineNo">262</span><a name="line.262"></a>
-<span class="sourceLineNo">263</span> private byte[] row;<a name="line.263"></a>
-<span class="sourceLineNo">264</span> private int rowOffset;<a name="line.264"></a>
-<span class="sourceLineNo">265</span> private int rowLength;<a name="line.265"></a>
-<span class="sourceLineNo">266</span> private byte[] family;<a name="line.266"></a>
-<span class="sourceLineNo">267</span> private int familyOffset;<a name="line.267"></a>
-<span class="sourceLineNo">268</span> private int familyLength;<a name="line.268"></a>
-<span class="sourceLineNo">269</span> private byte[] qualifier;<a name="line.269"></a>
-<span class="sourceLineNo">270</span> private int qualifierOffset;<a name="line.270"></a>
-<span class="sourceLineNo">271</span> private int qualifierLength;<a name="line.271"></a>
-<span class="sourceLineNo">272</span> private long ts;<a name="line.272"></a>
-<span class="sourceLineNo">273</span><a name="line.273"></a>
-<span class="sourceLineNo">274</span> public HBaseKeyWritable() { }<a name="line.274"></a>
-<span class="sourceLineNo">275</span><a name="line.275"></a>
-<span class="sourceLineNo">276</span> public HBaseKeyWritable(byte[] row, int rowOffset, int rowLength,<a name="line.276"></a>
-<span class="sourceLineNo">277</span> byte[] family, int familyOffset, int familyLength,<a name="line.277"></a>
-<span class="sourceLineNo">278</span> byte[] qualifier, int qualifierOffset, int qualifierLength, long ts) {<a name="line.278"></a>
-<span class="sourceLineNo">279</span> this.row = row;<a name="line.279"></a>
-<span class="sourceLineNo">280</span> this.rowOffset = rowOffset;<a name="line.280"></a>
-<span class="sourceLineNo">281</span> this.rowLength = rowLength;<a name="line.281"></a>
-<span class="sourceLineNo">282</span> this.family = family;<a name="line.282"></a>
-<span class="sourceLineNo">283</span> this.familyOffset = familyOffset;<a name="line.283"></a>
-<span class="sourceLineNo">284</span> this.familyLength = familyLength;<a name="line.284"></a>
-<span class="sourceLineNo">285</span> this.qualifier = qualifier;<a name="line.285"></a>
-<span class="sourceLineNo">286</span> this.qualifierOffset = qualifierOffset;<a name="line.286"></a>
-<span class="sourceLineNo">287</span> this.qualifierLength = qualifierLength;<a name="line.287"></a>
-<span class="sourceLineNo">288</span> this.ts = ts;<a name="line.288"></a>
-<span class="sourceLineNo">289</span> }<a name="line.289"></a>
-<span class="sourceLineNo">290</span><a name="line.290"></a>
-<span class="sourceLineNo">291</span> public HBaseKeyWritable(byte[] row, byte[] family, byte[] qualifier, long ts) {<a name="line.291"></a>
-<span class="sourceLineNo">292</span> this(row, 0, row.length,<a name="line.292"></a>
-<span class="sourceLineNo">293</span> family, 0, family.length,<a name="line.293"></a>
-<span class="sourceLineNo">294</span> qualifier, 0, qualifier != null ? qualifier.length : 0, ts);<a name="line.294"></a>
-<span class="sourceLineNo">295</span> }<a name="line.295"></a>
-<span class="sourceLineNo">296</span><a name="line.296"></a>
-<span class="sourceLineNo">297</span> public HBaseKeyWritable(byte[] row, byte[] family, long ts) {<a name="line.297"></a>
-<span class="sourceLineNo">298</span> this(row, family, HConstants.EMPTY_BYTE_ARRAY, ts);<a name="line.298"></a>
-<span class="sourceLineNo">299</span> }<a name="line.299"></a>
-<span class="sourceLineNo">300</span><a name="line.300"></a>
-<span class="sourceLineNo">301</span> public HBaseKeyWritable(Cell cell) {<a name="line.301"></a>
-<span class="sourceLineNo">302</span> this(cell.getRowArray(), cell.getRowOffset(), cell.getRowLength(),<a name="line.302"></a>
-<span class="sourceLineNo">303</span> cell.getFamilyArray(), cell.getFamilyOffset(), cell.getFamilyLength(),<a name="line.303"></a>
-<span class="sourceLineNo">304</span> cell.getQualifierArray(), cell.getQualifierOffset(), cell.getQualifierLength(),<a name="line.304"></a>
-<span class="sourceLineNo">305</span> cell.getTimestamp());<a name="line.305"></a>
-<span class="sourceLineNo">306</span> }<a name="line.306"></a>
-<span class="sourceLineNo">307</span><a name="line.307"></a>
-<span class="sourceLineNo">308</span> @Override<a name="line.308"></a>
-<span class="sourceLineNo">309</span> public void readFields(DataInput in) throws IOException {<a name="line.309"></a>
-<span class="sourceLineNo">310</span> this.row = Bytes.toBytes(in.readUTF());<a name="line.310"></a>
-<span class="sourceLineNo">311</span> this.rowOffset = 0;<a name="line.311"></a>
-<span class="sourceLineNo">312</span> this.rowLength = row.length;<a name="line.312"></a>
-<span class="sourceLineNo">313</span> this.family = Bytes.toBytes(in.readUTF());<a name="line.313"></a>
-<span class="sourceLineNo">314</span> this.familyOffset = 0;<a name="line.314"></a>
-<span class="sourceLineNo">315</span> this.familyLength = family.length;<a name="line.315"></a>
-<span class="sourceLineNo">316</span> this.qualifier = Bytes.toBytes(in.readUTF());<a name="line.316"></a>
-<span class="sourceLineNo">317</span> this.qualifierOffset = 0;<a name="line.317"></a>
-<span class="sourceLineNo">318</span> this.qualifierLength = qualifier.length;<a name="line.318"></a>
-<span class="sourceLineNo">319</span> this.ts = in.readLong();<a name="line.319"></a>
-<span class="sourceLineNo">320</span> }<a name="line.320"></a>
-<span class="sourceLineNo">321</span><a name="line.321"></a>
-<span class="sourceLineNo">322</span> @Override<a name="line.322"></a>
-<span class="sourceLineNo">323</span> public void write(DataOutput out) throws IOException {<a name="line.323"></a>
-<span class="sourceLineNo">324</span> out.writeUTF(new String(row, rowOffset, rowLength, StandardCharsets.UTF_8));<a name="line.324"></a>
-<span class="sourceLineNo">325</span> out.writeUTF(new String(family, familyOffset, familyLength, StandardCharsets.UTF_8));<a name="line.325"></a>
-<span class="sourceLineNo">326</span> if (qualifier != null) {<a name="line.326"></a>
-<span class="sourceLineNo">327</span> out.writeUTF(new String(qualifier, qualifierOffset, qualifierLength,<a name="line.327"></a>
-<span class="sourceLineNo">328</span> StandardCharsets.UTF_8));<a name="line.328"></a>
-<span class="sourceLineNo">329</span> } else {<a name="line.329"></a>
-<span class="sourceLineNo">330</span> out.writeUTF("");<a name="line.330"></a>
-<span class="sourceLineNo">331</span> }<a name="line.331"></a>
-<span class="sourceLineNo">332</span> out.writeLong(ts);<a name="line.332"></a>
-<span class="sourceLineNo">333</span> }<a name="line.333"></a>
-<span class="sourceLineNo">334</span><a name="line.334"></a>
-<span class="sourceLineNo">335</span> public byte[] getRowArray() {<a name="line.335"></a>
-<span class="sourceLineNo">336</span> return row;<a name="line.336"></a>
-<span class="sourceLineNo">337</span> }<a name="line.337"></a>
-<span class="sourceLineNo">338</span><a name="line.338"></a>
-<span class="sourceLineNo">339</span> public void setRow(byte[] row) {<a name="line.339"></a>
-<span class="sourceLineNo">340</span> this.row = row;<a name="line.340"></a>
-<span class="sourceLineNo">341</span> }<a name="line.341"></a>
-<span class="sourceLineNo">342</span><a name="line.342"></a>
-<span class="sourceLineNo">343</span> public int getRowOffset() {<a name="line.343"></a>
-<span class="sourceLineNo">344</span> return rowOffset;<a name="line.344"></a>
-<span class="sourceLineNo">345</span> }<a name="line.345"></a>
-<span class="sourceLineNo">346</span><a name="line.346"></a>
-<span class="sourceLineNo">347</span> public void setRowOffset(int rowOffset) {<a name="line.347"></a>
-<span class="sourceLineNo">348</span> this.rowOffset = rowOffset;<a name="line.348"></a>
-<span class="sourceLineNo">349</span> }<a name="line.349"></a>
-<span class="sourceLineNo">350</span><a name="line.350"></a>
-<span class="sourceLineNo">351</span> public int getRowLength() {<a name="line.351"></a>
-<span class="sourceLineNo">352</span> return rowLength;<a name="line.352"></a>
-<span class="sourceLineNo">353</span> }<a name="line.353"></a>
-<span class="sourceLineNo">354</span><a name="line.354"></a>
-<span class="sourceLineNo">355</span> public void setRowLength(int rowLength) {<a name="line.355"></a>
-<span class="sourceLineNo">356</span> this.rowLength = rowLength;<a name="line.356"></a>
-<span class="sourceLineNo">357</span> }<a name="line.357"></a>
-<span class="sourceLineNo">358</span><a name="line.358"></a>
-<span class="sourceLineNo">359</span> public byte[] getFamilyArray() {<a name="line.359"></a>
-<span class="sourceLineNo">360</span> return family;<a name="line.360"></a>
-<span class="sourceLineNo">361</span> }<a name="line.361"></a>
-<span class="sourceLineNo">362</span><a name="line.362"></a>
-<span class="sourceLineNo">363</span> public void setFamily(byte[] family) {<a name="line.363"></a>
-<span class="sourceLineNo">364</span> this.family = family;<a name="line.364"></a>
-<span class="sourceLineNo">365</span> }<a name="line.365"></a>
-<span class="sourceLineNo">366</span><a name="line.366"></a>
-<span class="sourceLineNo">367</span> public int getFamilyOffset() {<a name="line.367"></a>
-<span class="sourceLineNo">368</span> return familyOffset;<a name="line.368"></a>
-<span class="sourceLineNo">369</span> }<a name="line.369"></a>
-<span class="sourceLineNo">370</span><a name="line.370"></a>
-<span class="sourceLineNo">371</span> public void setFamilyOffset(int familyOffset) {<a name="line.371"></a>
-<span class="sourceLineNo">372</span> this.familyOffset = familyOffset;<a name="line.372"></a>
-<span class="sourceLineNo">373</span> }<a name="line.373"></a>
-<span class="sourceLineNo">374</span><a name="line.374"></a>
-<span class="sourceLineNo">375</span> public int getFamilyLength() {<a name="line.375"></a>
-<span class="sourceLineNo">376</span> return familyLength;<a name="line.376"></a>
-<span class="sourceLineNo">377</span> }<a name="line.377"></a>
-<span class="sourceLineNo">378</span><a name="line.378"></a>
-<span class="sourceLineNo">379</span> public void setFamilyLength(int familyLength) {<a name="line.379"></a>
-<span class="sourceLineNo">380</span> this.familyLength = familyLength;<a name="line.380"></a>
-<span class="sourceLineNo">381</span> }<a name="line.381"></a>
-<span class="sourceLineNo">382</span><a name="line.382"></a>
-<span class="sourceLineNo">383</span> public byte[] getQualifierArray() {<a name="line.383"></a>
-<span class="sourceLineNo">384</span> return qualifier;<a name="line.384"></a>
-<span class="sourceLineNo">385</span> }<a name="line.385"></a>
-<span class="sourceLineNo">386</span><a name="line.386"></a>
-<span class="sourceLineNo">387</span> public void setQualifier(byte[] qualifier) {<a name="line.387"></a>
-<span class="sourceLineNo">388</span> this.qualifier = qualifier;<a name="line.388"></a>
-<span class="sourceLineNo">389</span> }<a name="line.389"></a>
-<span class="sourceLineNo">390</span><a name="line.390"></a>
-<span class="sourceLineNo">391</span> public int getQualifierOffset() {<a name="line.391"></a>
-<span class="sourceLineNo">392</span> return qualifierOffset;<a name="line.392"></a>
-<span class="sourceLineNo">393</span> }<a name="line.393"></a>
-<span class="sourceLineNo">394</span><a name="line.394"></a>
-<span class="sourceLineNo">395</span> public void setQualifierOffset(int qualifierOffset) {<a name="line.395"></a>
-<span class="sourceLineNo">396</span> this.qualifierOffset = qualifierOffset;<a name="line.396"></a>
-<span class="sourceLineNo">397</span> }<a name="line.397"></a>
-<span class="sourceLineNo">398</span><a name="line.398"></a>
-<span class="sourceLineNo">399</span> public int getQualifierLength() {<a name="line.399"></a>
-<span class="sourceLineNo">400</span> return qualifierLength;<a name="line.400"></a>
-<span class="sourceLineNo">401</span> }<a name="line.401"></a>
-<span class="sourceLineNo">402</span><a name="line.402"></a>
-<span class="sourceLineNo">403</span> public void setQualifierLength(int qualifierLength) {<a name="line.403"></a>
-<span class="sourceLineNo">404</span> this.qualifierLength = qualifierLength;<a name="line.404"></a>
-<span class="sourceLineNo">405</span> }<a name="line.405"></a>
-<span class="sourceLineNo">406</span><a name="line.406"></a>
-<span class="sourceLineNo">407</span> public long getTimestamp() {<a name="line.407"></a>
-<span class="sourceLineNo">408</span> return ts;<a name="line.408"></a>
-<span class="sourceLineNo">409</span> }<a name="line.409"></a>
-<span class="sourceLineNo">410</span><a name="line.410"></a>
-<span class="sourceLineNo">411</span> public void setTimestamp(long ts) {<a name="line.411"></a>
-<span class="sourceLineNo">412</span> this.ts = ts;<a name="line.412"></a>
-<span class="sourceLineNo">413</span> }<a name="line.413"></a>
-<span class="sourceLineNo">414</span> }<a name="line.414"></a>
-<span class="sourceLineNo">415</span><a name="line.415"></a>
-<span class="sourceLineNo">416</span> public static class Loader extends Configured implements Tool {<a name="line.416"></a>
-<span class="sourceLineNo">417</span><a name="line.417"></a>
-<span class="sourceLineNo">418</span> private static final Logger LOG = LoggerFactory.getLogger(Loader.class);<a name="line.418"></a>
-<span class="sourceLineNo">419</span> private static final String USAGE = "Loader <warInputDir | warFileList> <outputDir>";<a name="line.419"></a>
+<span class="sourceLineNo">194</span> if (outputDir == null) {<a name="line.194"></a>
+<span class="sourceLineNo">195</span> throw new IllegalArgumentException("Output directory not specified");<a name="line.195"></a>
+<span class="sourceLineNo">196</span> }<a name="line.196"></a>
+<span class="sourceLineNo">197</span> int res = runLoader(warcFileInputDir, outputDir);<a name="line.197"></a>
+<span class="sourceLineNo">198</span> if (res != 0) {<a name="line.198"></a>
+<span class="sourceLineNo">199</span> LOG.error("Loader failed");<a name="line.199"></a>
+<span class="sourceLineNo">200</span> return -1;<a name="line.200"></a>
+<span class="sourceLineNo">201</span> }<a name="line.201"></a>
+<span class="sourceLineNo">202</span> res = runVerify(outputDir);<a name="line.202"></a>
+<span class="sourceLineNo">203</span> } catch (Exception e) {<a name="line.203"></a>
+<span class="sourceLineNo">204</span> LOG.error("Tool failed with exception", e);<a name="line.204"></a>
+<span class="sourceLineNo">205</span> return -1;<a name="line.205"></a>
+<span class="sourceLineNo">206</span> }<a name="line.206"></a>
+<span class="sourceLineNo">207</span> return 0;<a name="line.207"></a>
+<span class="sourceLineNo">208</span> }<a name="line.208"></a>
+<span class="sourceLineNo">209</span><a name="line.209"></a>
+<span class="sourceLineNo">210</span> @Override<a name="line.210"></a>
+<span class="sourceLineNo">211</span> protected void processOptions(CommandLine cmd) {<a name="line.211"></a>
+<span class="sourceLineNo">212</span> processBaseOptions(cmd);<a name="line.212"></a>
+<span class="sourceLineNo">213</span> args = cmd.getArgs();<a name="line.213"></a>
+<span class="sourceLineNo">214</span> }<a name="line.214"></a>
+<span class="sourceLineNo">215</span><a name="line.215"></a>
+<span class="sourceLineNo">216</span> @Override<a name="line.216"></a>
+<span class="sourceLineNo">217</span> public void setUpCluster() throws Exception {<a name="line.217"></a>
+<span class="sourceLineNo">218</span> util = getTestingUtil(getConf());<a name="line.218"></a>
+<span class="sourceLineNo">219</span> boolean isDistributed = util.isDistributedCluster();<a name="line.219"></a>
+<span class="sourceLineNo">220</span> util.initializeCluster(isDistributed ? 1 : 3);<a name="line.220"></a>
+<span class="sourceLineNo">221</span> if (!isDistributed) {<a name="line.221"></a>
+<span class="sourceLineNo">222</span> util.startMiniMapReduceCluster();<a name="line.222"></a>
+<span class="sourceLineNo">223</span> }<a name="line.223"></a>
+<span class="sourceLineNo">224</span> this.setConf(util.getConfiguration());<a name="line.224"></a>
+<span class="sourceLineNo">225</span> }<a name="line.225"></a>
+<span class="sourceLineNo">226</span><a name="line.226"></a>
+<span class="sourceLineNo">227</span> @Override<a name="line.227"></a>
+<span class="sourceLineNo">228</span> public void cleanUpCluster() throws Exception {<a name="line.228"></a>
+<span class="sourceLineNo">229</span> super.cleanUpCluster();<a name="line.229"></a>
+<span class="sourceLineNo">230</span> if (util.isDistributedCluster()) {<a name="line.230"></a>
+<span class="sourceLineNo">231</span> util.shutdownMiniMapReduceCluster();<a name="line.231"></a>
+<span class="sourceLineNo">232</span> }<a name="line.232"></a>
+<span class="sourceLineNo">233</span> }<a name="line.233"></a>
+<span class="sourceLineNo">234</span><a name="line.234"></a>
+<span class="sourceLineNo">235</span> static TableName getTablename(Configuration c) {<a name="line.235"></a>
+<span class="sourceLineNo">236</span> return TableName.valueOf(c.get(TABLE_NAME_KEY, DEFAULT_TABLE_NAME));<a name="line.236"></a>
+<span class="sourceLineNo">237</span> }<a name="line.237"></a>
+<span class="sourceLineNo">238</span><a name="line.238"></a>
+<span class="sourceLineNo">239</span> @Override<a name="line.239"></a>
+<span class="sourceLineNo">240</span> public TableName getTablename() {<a name="line.240"></a>
+<span class="sourceLineNo">241</span> return getTablename(getConf());<a name="line.241"></a>
+<span class="sourceLineNo">242</span> }<a name="line.242"></a>
+<span class="sourceLineNo">243</span><a name="line.243"></a>
+<span class="sourceLineNo">244</span> @Override<a name="line.244"></a>
+<span class="sourceLineNo">245</span> protected Set<String> getColumnFamilies() {<a name="line.245"></a>
+<span class="sourceLineNo">246</span> Set<String> families = new HashSet<>();<a name="line.246"></a>
+<span class="sourceLineNo">247</span> families.add(Bytes.toString(CONTENT_FAMILY_NAME));<a name="line.247"></a>
+<span class="sourceLineNo">248</span> families.add(Bytes.toString(INFO_FAMILY_NAME));<a name="line.248"></a>
+<span class="sourceLineNo">249</span> return families;<a name="line.249"></a>
+<span class="sourceLineNo">250</span> }<a name="line.250"></a>
+<span class="sourceLineNo">251</span><a name="line.251"></a>
+<span class="sourceLineNo">252</span> @Override<a name="line.252"></a>
+<span class="sourceLineNo">253</span> public int runTestFromCommandLine() throws Exception {<a name="line.253"></a>
+<span class="sourceLineNo">254</span> return ToolRunner.run(getConf(), this, args);<a name="line.254"></a>
+<span class="sourceLineNo">255</span> }<a name="line.255"></a>
+<span class="sourceLineNo">256</span><a name="line.256"></a>
+<span class="sourceLineNo">257</span> public static void main(String[] args) throws Exception {<a name="line.257"></a>
+<span class="sourceLineNo">258</span> Configuration conf = HBaseConfiguration.create();<a name="line.258"></a>
+<span class="sourceLineNo">259</span> IntegrationTestingUtility.setUseDistributedCluster(conf);<a name="line.259"></a>
+<span class="sourceLineNo">260</span> int ret = ToolRunner.run(conf, new IntegrationTestLoadCommonCrawl(), args);<a name="line.260"></a>
+<span class="sourceLineNo">261</span> System.exit(ret);<a name="line.261"></a>
+<span class="sourceLineNo">262</span> }<a name="line.262"></a>
+<span class="sourceLineNo">263</span><a name="line.263"></a>
+<span class="sourceLineNo">264</span> public static class HBaseKeyWritable implements Writable {<a name="line.264"></a>
+<span class="sourceLineNo">265</span><a name="line.265"></a>
+<span class="sourceLineNo">266</span> private byte[] row;<a name="line.266"></a>
+<span class="sourceLineNo">267</span> private int rowOffset;<a name="line.267"></a>
+<span class="sourceLineNo">268</span> private int rowLength;<a name="line.268"></a>
+<span class="sourceLineNo">269</span> private byte[] family;<a name="line.269"></a>
+<span class="sourceLineNo">270</span> private int familyOffset;<a name="line.270"></a>
+<span class="sourceLineNo">271</span> private int familyLength;<a name="line.271"></a>
+<span class="sourceLineNo">272</span> private byte[] qualifier;<a name="line.272"></a>
+<span class="sourceLineNo">273</span> private int qualifierOffset;<a name="line.273"></a>
+<span class="sourceLineNo">274</span> private int qualifierLength;<a name="line.274"></a>
+<span class="sourceLineNo">275</span> private long ts;<a name="line.275"></a>
+<span class="sourceLineNo">276</span><a name="line.276"></a>
+<span class="sourceLineNo">277</span> public HBaseKeyWritable() { }<a name="line.277"></a>
+<span class="sourceLineNo">278</span><a name="line.278"></a>
+<span class="sourceLineNo">279</span> public HBaseKeyWritable(byte[] row, int rowOffset, int rowLength,<a name="line.279"></a>
+<span class="sourceLineNo">280</span> byte[] family, int familyOffset, int familyLength,<a name="line.280"></a>
+<span class="sourceLineNo">281</span> byte[] qualifier, int qualifierOffset, int qualifierLength, long ts) {<a name="line.281"></a>
+<span class="sourceLineNo">282</span> this.row = row;<a name="line.282"></a>
+<span class="sourceLineNo">283</span> this.rowOffset = rowOffset;<a name="line.283"></a>
+<span class="sourceLineNo">284</span> this.rowLength = rowLength;<a name="line.284"></a>
+<span class="sourceLineNo">285</span> this.family = family;<a name="line.285"></a>
+<span class="sourceLineNo">286</span> this.familyOffset = familyOffset;<a name="line.286"></a>
+<span class="sourceLineNo">287</span> this.familyLength = familyLength;<a name="line.287"></a>
+<span class="sourceLineNo">288</span> this.qualifier = qualifier;<a name="line.288"></a>
+<span class="sourceLineNo">289</span> this.qualifierOffset = qualifierOffset;<a name="line.289"></a>
+<span class="sourceLineNo">290</span> this.qualifierLength = qualifierLength;<a name="line.290"></a>
+<span class="sourceLineNo">291</span> this.ts = ts;<a name="line.291"></a>
+<span class="sourceLineNo">292</span> }<a name="line.292"></a>
+<span class="sourceLineNo">293</span><a name="line.293"></a>
+<span class="sourceLineNo">294</span> public HBaseKeyWritable(byte[] row, byte[] family, byte[] qualifier, long ts) {<a name="line.294"></a>
+<span class="sourceLineNo">295</span> this(row, 0, row.length,<a name="line.295"></a>
+<span class="sourceLineNo">296</span> family, 0, family.length,<a name="line.296"></a>
+<span class="sourceLineNo">297</span> qualifier, 0, qualifier != null ? qualifier.length : 0, ts);<a name="line.297"></a>
+<span class="sourceLineNo">298</span> }<a name="line.298"></a>
+<span class="sourceLineNo">299</span><a name="line.299"></a>
+<span class="sourceLineNo">300</span> public HBaseKeyWritable(byte[] row, byte[] family, long ts) {<a name="line.300"></a>
+<span class="sourceLineNo">301</span> this(row, family, HConstants.EMPTY_BYTE_ARRAY, ts);<a name="line.301"></a>
+<span class="sourceLineNo">302</span> }<a name="line.302"></a>
+<span class="sourceLineNo">303</span><a name="line.303"></a>
+<span class="sourceLineNo">304</span> public HBaseKeyWritable(Cell cell) {<a name="line.304"></a>
+<span class="sourceLineNo">305</span> this(cell.getRowArray(), cell.getRowOffset(), cell.getRowLength(),<a name="line.305"></a>
+<span class="sourceLineNo">306</span> cell.getFamilyArray(), cell.getFamilyOffset(), cell.getFamilyLength(),<a name="line.306"></a>
+<span class="sourceLineNo">307</span> cell.getQualifierArray(), cell.getQualifierOffset(), cell.getQualifierLength(),<a name="line.307"></a>
+<span class="sourceLineNo">308</span> cell.getTimestamp());<a name="line.308"></a>
+<span class="sourceLineNo">309</span> }<a name="line.309"></a>
+<span class="sourceLineNo">310</span><a name="line.310"></a>
+<span class="sourceLineNo">311</span> @Override<a name="line.311"></a>
+<span class="sourceLineNo">312</span> public void readFields(DataInput in) throws IOException {<a name="line.312"></a>
+<span class="sourceLineNo">313</span> this.row = Bytes.toBytes(in.readUTF());<a name="line.313"></a>
+<span class="sourceLineNo">314</span> this.rowOffset = 0;<a name="line.314"></a>
+<span class="sourceLineNo">315</span> this.rowLength = row.length;<a name="line.315"></a>
+<span class="sourceLineNo">316</span> this.family = Bytes.toBytes(in.readUTF());<a name="line.316"></a>
+<span class="sourceLineNo">317</span> this.familyOffset = 0;<a name="line.317"></a>
+<span class="sourceLineNo">318</span> this.familyLength = family.length;<a name="line.318"></a>
+<span class="sourceLineNo">319</span> this.qualifier = Bytes.toBytes(in.readUTF());<a name="line.319"></a>
+<span class="sourceLineNo">320</span> this.qualifierOffset = 0;<a name="line.320"></a>
+<span class="sourceLineNo">321</span> this.qualifierLength = qualifier.length;<a name="line.321"></a>
+<span class="sourceLineNo">322</span> this.ts = in.readLong();<a name="line.322"></a>
+<span class="sourceLineNo">323</span> }<a name="line.323"></a>
+<span class="sourceLineNo">324</span><a name="line.324"></a>
+<span class="sourceLineNo">325</span> @Override<a name="line.325"></a>
+<span class="sourceLineNo">326</span> public void write(DataOutput out) throws IOException {<a name="line.326"></a>
+<span class="sourceLineNo">327</span> out.writeUTF(new String(row, rowOffset, rowLength, StandardCharsets.UTF_8));<a name="line.327"></a>
+<span class="sourceLineNo">328</span> out.writeUTF(new String(family, familyOffset, familyLength, StandardCharsets.UTF_8));<a name="line.328"></a>
+<span class="sourceLineNo">329</span> if (qualifier != null) {<a name="line.329"></a>
+<span class="sourceLineNo">330</span> out.writeUTF(new String(qualifier, qualifierOffset, qualifierLength,<a name="line.330"></a>
+<span class="sourceLineNo">331</span> StandardCharsets.UTF_8));<a name="line.331"></a>
+<span class="sourceLineNo">332</span> } else {<a name="line.332"></a>
+<span class="sourceLineNo">333</span> out.writeUTF("");<a name="line.333"></a>
+<span class="sourceLineNo">334</span> }<a name="line.334"></a>
+<span class="sourceLineNo">335</span> out.writeLong(ts);<a name="line.335"></a>
+<span class="sourceLineNo">336</span> }<a name="line.336"></a>
+<span class="sourceLineNo">337</span><a name="line.337"></a>
+<span class="sourceLineNo">338</span> public byte[] getRowArray() {<a name="line.338"></a>
+<span class="sourceLineNo">339</span> return row;<a name="line.339"></a>
+<span class="sourceLineNo">340</span> }<a name="line.340"></a>
+<span class="sourceLineNo">341</span><a name="line.341"></a>
+<span class="sourceLineNo">342</span> public void setRow(byte[] row) {<a name="line.342"></a>
+<span class="sourceLineNo">343</span> this.row = row;<a name="line.343"></a>
+<span class="sourceLineNo">344</span> }<a name="line.344"></a>
+<span class="sourceLineNo">345</span><a name="line.345"></a>
+<span class="sourceLineNo">346</span> public int getRowOffset() {<a name="line.346"></a>
+<span class="sourceLineNo">347</span> return rowOffset;<a name="line.347"></a>
+<span class="sourceLineNo">348</span> }<a name="line.348"></a>
+<span class="sourceLineNo">349</span><a name="line.349"></a>
+<span class="sourceLineNo">350</span> public void setRowOffset(int rowOffset) {<a name="line.350"></a>
+<span class="sourceLineNo">351</span> this.rowOffset = rowOffset;<a name="line.351"></a>
+<span class="sourceLineNo">352</span> }<a name="line.352"></a>
+<span class="sourceLineNo">353</span><a name="line.353"></a>
+<span class="sourceLineNo">354</span> public int getRowLength() {<a name="line.354"></a>
+<span class="sourceLineNo">355</span> return rowLength;<a name="line.355"></a>
+<span class="sourceLineNo">356</span> }<a name="line.356"></a>
+<span class="sourceLineNo">357</span><a name="line.357"></a>
+<span class="sourceLineNo">358</span> public void setRowLength(int rowLength) {<a name="line.358"></a>
+<span class="sourceLineNo">359</span> this.rowLength = rowLength;<a name="line.359"></a>
+<span class="sourceLineNo">360</span> }<a name="line.360"></a>
+<span class="sourceLineNo">361</span><a name="line.361"></a>
+<span class="sourceLineNo">362</span> public byte[] getFamilyArray() {<a name="line.362"></a>
+<span class="sourceLineNo">363</span> return family;<a name="line.363"></a>
+<span class="sourceLineNo">364</span> }<a name="line.364"></a>
+<span class="sourceLineNo">365</span><a name="line.365"></a>
+<span class="sourceLineNo">366</span> public void setFamily(byte[] family) {<a name="line.366"></a>
+<span class="sourceLineNo">367</span> this.family = family;<a name="line.367"></a>
+<span class="sourceLineNo">368</span> }<a name="line.368"></a>
+<span class="sourceLineNo">369</span><a name="line.369"></a>
+<span class="sourceLineNo">370</span> public int getFamilyOffset() {<a name="line.370"></a>
+<span class="sourceLineNo">371</span> return familyOffset;<a name="line.371"></a>
+<span class="sourceLineNo">372</span> }<a name="line.372"></a>
+<span class="sourceLineNo">373</span><a name="line.373"></a>
+<span class="sourceLineNo">374</span> public void setFamilyOffset(int familyOffset) {<a name="line.374"></a>
+<span class="sourceLineNo">375</span> this.familyOffset = familyOffset;<a name="line.375"></a>
+<span class="sourceLineNo">376</span> }<a name="line.376"></a>
+<span class="sourceLineNo">377</span><a name="line.377"></a>
+<span class="sourceLineNo">378</span> public int getFamilyLength() {<a name="line.378"></a>
+<span class="sourceLineNo">379</span> return familyLength;<a name="line.379"></a>
+<span class="sourceLineNo">380</span> }<a name="line.380"></a>
+<span class="sourceLineNo">381</span><a name="line.381"></a>
+<span class="sourceLineNo">382</span> public void setFamilyLength(int familyLength) {<a name="line.382"></a>
+<span class="sourceLineNo">383</span> this.familyLength = familyLength;<a name="line.383"></a>
+<span class="sourceLineNo">384</span> }<a name="line.384"></a>
+<span class="sourceLineNo">385</span><a name="line.385"></a>
+<span class="sourceLineNo">386</span> public byte[] getQualifierArray() {<a name="line.386"></a>
+<span class="sourceLineNo">387</span> return qualifier;<a name="line.387"></a>
+<span class="sourceLineNo">388</span> }<a name="line.388"></a>
+<span class="sourceLineNo">389</span><a name="line.389"></a>
+<span class="sourceLineNo">390</span> public void setQualifier(byte[] qualifier) {<a name="line.390"></a>
+<span class="sourceLineNo">391</span> this.qualifier = qualifier;<a name="line.391"></a>
+<span class="sourceLineNo">392</span> }<a name="line.392"></a>
+<span class="sourceLineNo">393</span><a name="line.393"></a>
+<span class="sourceLineNo">394</span> public int getQualifierOffset() {<a name="line.394"></a>
+<span class="sourceLineNo">395</span> return qualifierOffset;<a name="line.395"></a>
+<span class="sourceLineNo">396</span> }<a name="line.396"></a>
+<span class="sourceLineNo">397</span><a name="line.397"></a>
+<span class="sourceLineNo">398</span> public void setQualifierOffset(int qualifierOffset) {<a name="line.398"></a>
+<span class="sourceLineNo">399</span> this.qualifierOffset = qualifierOffset;<a name="line.399"></a>
+<span class="sourceLineNo">400</span> }<a name="line.400"></a>
+<span class="sourceLineNo">401</span><a name="line.401"></a>
+<span class="sourceLineNo">402</span> public int getQualifierLength() {<a name="line.402"></a>
+<span class="sourceLineNo">403</span> return qualifierLength;<a name="line.403"></a>
+<span class="sourceLineNo">404</span> }<a name="line.404"></a>
+<span class="sourceLineNo">405</span><a name="line.405"></a>
+<span class="sourceLineNo">406</span> public void setQualifierLength(int qualifierLength) {<a name="line.406"></a>
+<span class="sourceLineNo">407</span> this.qualifierLength = qualifierLength;<a name="line.407"></a>
+<span class="sourceLineNo">408</span> }<a name="line.408"></a>
+<span class="sourceLineNo">409</span><a name="line.409"></a>
+<span class="sourceLineNo">410</span> public long getTimestamp() {<a name="line.410"></a>
+<span class="sourceLineNo">411</span> return ts;<a name="line.411"></a>
+<span class="sourceLineNo">412</span> }<a name="line.412"></a>
+<span class="sourceLineNo">413</span><a name="line.413"></a>
+<span class="sourceLineNo">414</span> public void setTimestamp(long ts) {<a name="line.414"></a>
+<span class="sourceLineNo">415</span> this.ts = ts;<a name="line.415"></a>
+<span class="sourceLineNo">416</span> }<a name="line.416"></a>
+<span class="sourceLineNo">417</span> }<a name="line.417"></a>
+<span class="sourceLineNo">418</span><a name="line.418"></a>
+<span class="sourceLineNo">419</span> public static class Loader extends Configured implements Tool {<a name="line.419"></a>
<span class="sourceLineNo">420</span><a name="line.420"></a>
-<span class="sourceLineNo">421</span> void createSchema(TableName tableName) throws IOException {<a name="line.421"></a>
-<span class="sourceLineNo">422</span><a name="line.422"></a>
-<span class="sourceLineNo">423</span> try (Connection conn = ConnectionFactory.createConnection(getConf());<a name="line.423"></a>
-<span class="sourceLineNo">424</span> Admin admin = conn.getAdmin()) {<a name="line.424"></a>
-<span class="sourceLineNo">425</span> if (!admin.tableExists(tableName)) {<a name="line.425"></a>
-<span class="sourceLineNo">426</span><a name="line.426"></a>
-<span class="sourceLineNo">427</span> ColumnFamilyDescriptorBuilder contentFamilyBuilder =<a name="line.427"></a>
-<span class="sourceLineNo">428</span> ColumnFamilyDescriptorBuilder.newBuilder(CONTENT_FAMILY_NAME)<a name="line.428"></a>
-<span class="sourceLineNo">429</span> .setDataBlockEncoding(DataBlockEncoding.NONE)<a name="line.429"></a>
-<span class="sourceLineNo">430</span> .setBloomFilterType(BloomType.ROW)<a name="line.430"></a>
-<span class="sourceLineNo">431</span> .setMaxVersions(1000)<a name="line.431"></a>
-<span class="sourceLineNo">432</span> .setBlocksize(256 * 1024)<a name="line.432"></a>
-<span class="sourceLineNo">433</span> ;<a name="line.433"></a>
-<span class="sourceLineNo">434</span><a name="line.434"></a>
-<span class="sourceLineNo">435</span> ColumnFamilyDescriptorBuilder infoFamilyBuilder =<a name="line.435"></a>
-<span class="sourceLineNo">436</span> ColumnFamilyDescriptorBuilder.newBuilder(INFO_FAMILY_NAME)<a name="line.436"></a>
-<span class="sourceLineNo">437</span> .setDataBlockEncoding(DataBlockEncoding.NONE)<a name="line.437"></a>
-<span class="sourceLineNo">438</span> .setBloomFilterType(BloomType.ROWCOL)<a name="line.438"></a>
-<span class="sourceLineNo">439</span> .setMaxVersions(1000)<a name="line.439"></a>
-<span class="sourceLineNo">440</span> .setBlocksize(8 * 1024)<a name="line.440"></a>
-<span class="sourceLineNo">441</span> ;<a name="line.441"></a>
-<span class="sourceLineNo">442</span><a name="line.442"></a>
-<span class="sourceLineNo">443</span> Set<ColumnFamilyDescriptor> families = new HashSet<>();<a name="line.443"></a>
-<span class="sourceLineNo">444</span> families.add(contentFamilyBuilder.build());<a name="line.444"></a>
-<span class="sourceLineNo">445</span> families.add(infoFamilyBuilder.build());<a name="line.445"></a>
-<span class="sourceLineNo">446</span><a name="line.446"></a>
-<span class="sourceLineNo">447</span> TableDescriptor tableDescriptor = TableDescriptorBuilder.newBuilder(tableName)<a name="line.447"></a>
-<span class="sourceLineNo">448</span> .setColumnFamilies(families)<a name="line.448"></a>
-<span class="sourceLineNo">449</span> .build();<a name="line.449"></a>
-<span class="sourceLineNo">450</span><a name="line.450"></a>
-<span class="sourceLineNo">451</span> if (getConf().getBoolean(HBaseTestingUtil.PRESPLIT_TEST_TABLE_KEY,<a name="line.451"></a>
-<span class="sourceLineNo">452</span> HBaseTestingUtil.PRESPLIT_TEST_TABLE)) {<a name="line.452"></a>
-<span class="sourceLineNo">453</span> int numberOfServers = admin.getRegionServers().size();<a name="line.453"></a>
-<span class="sourceLineNo">454</span> if (numberOfServers == 0) {<a name="line.454"></a>
-<span class="sourceLineNo">455</span> throw new IllegalStateException("No live regionservers");<a name="line.455"></a>
-<span class="sourceLineNo">456</span> }<a name="line.456"></a>
-<span class="sourceLineNo">457</span> int regionsPerServer = getConf().getInt(HBaseTestingUtil.REGIONS_PER_SERVER_KEY,<a name="line.457"></a>
-<span class="sourceLineNo">458</span> HBaseTestingUtil.DEFAULT_REGIONS_PER_SERVER);<a name="line.458"></a>
-<span class="sourceLineNo">459</span> int totalNumberOfRegions = numberOfServers * regionsPerServer;<a name="line.459"></a>
-<span class="sourceLineNo">460</span> LOG.info("Creating test table: " + tableDescriptor);<a name="line.460"></a>
-<span class="sourceLineNo">461</span> LOG.info("Number of live regionservers: " + numberOfServers + ", " +<a name="line.461"></a>
-<span class="sourceLineNo">462</span> "pre-splitting table into " + totalNumberOfRegions + " regions " +<a name="line.462"></a>
-<span class="sourceLineNo">463</span> "(default regions per server: " + regionsPerServer + ")");<a name="line.463"></a>
-<span class="sourceLineNo">464</span> byte[][] splits = new RegionSplitter.UniformSplit().split(totalNumberOfRegions);<a name="line.464"></a>
-<span class="sourceLineNo">465</span> admin.createTable(tableDescriptor, splits);<a name="line.465"></a>
-<span class="sourceLineNo">466</span> } else {<a name="line.466"></a>
-<span class="sourceLineNo">467</span> LOG.info("Creating test table: " + tableDescriptor);<a name="line.467"></a>
-<span class="sourceLineNo">468</span> admin.createTable(tableDescriptor);<a name="line.468"></a>
-<span class="sourceLineNo">469</span> }<a name="line.469"></a>
-<span class="sourceLineNo">470</span> }<a name="line.470"></a>
-<span class="sourceLineNo">471</span> } catch (MasterNotRunningException e) {<a name="line.471"></a>
-<span class="sourceLineNo">472</span> LOG.error("Master not running", e);<a name="line.472"></a>
-<span class="sourceLineNo">473</span> throw new IOException(e);<a name="line.473"></a>
-<span class="sourceLineNo">474</span> }<a name="line.474"></a>
-<span class="sourceLineNo">475</span> }<a name="line.475"></a>
-<span class="sourceLineNo">476</span><a name="line.476"></a>
-<span class="sourceLineNo">477</span> int run(Path warcFileInput, Path outputDir)<a name="line.477"></a>
-<span class="sourceLineNo">478</span> throws IOException, ClassNotFoundException, InterruptedException {<a name="line.478"></a>
+<span class="sourceLineNo">421</span> private static final Logger LOG = LoggerFactory.getLogger(Loader.class);<a name="line.421"></a>
+<span class="sourceLineNo">422</span> private static final String USAGE = "Loader <warInputDir | warFileList> <outputDir>";<a name="line.422"></a>
+<span class="sourceLineNo">423</span><a name="line.423"></a>
+<span class="sourceLineNo">424</span> void createSchema(TableName tableName) throws IOException {<a name="line.424"></a>
+<span class="sourceLineNo">425</span><a name="line.425"></a>
+<span class="sourceLineNo">426</span> try (Connection conn = ConnectionFactory.createConnection(getConf());<a name="line.426"></a>
+<span class="sourceLineNo">427</span> Admin admin = conn.getAdmin()) {<a name="line.427"></a>
+<span class="sourceLineNo">428</span> if (!admin.tableExists(tableName)) {<a name="line.428"></a>
+<span class="sourceLineNo">429</span><a name="line.429"></a>
+<span class="sourceLineNo">430</span> ColumnFamilyDescriptorBuilder contentFamilyBuilder =<a name="line.430"></a>
+<span class="sourceLineNo">431</span> ColumnFamilyDescriptorBuilder.newBuilder(CONTENT_FAMILY_NAME)<a name="line.431"></a>
+<span class="sourceLineNo">432</span> .setDataBlockEncoding(DataBlockEncoding.NONE)<a name="line.432"></a>
+<span class="sourceLineNo">433</span> .setBloomFilterType(BloomType.ROW)<a name="line.433"></a>
+<span class="sourceLineNo">434</span> .setMaxVersions(1000)<a name="line.434"></a>
+<span class="sourceLineNo">435</span> .setBlocksize(256 * 1024)<a name="line.435"></a>
+<span class="sourceLineNo">436</span> ;<a name="line.436"></a>
+<span class="sourceLineNo">437</span><a name="line.437"></a>
+<span class="sourceLineNo">438</span> ColumnFamilyDescriptorBuilder infoFamilyBuilder =<a name="line.438"></a>
+<span class="sourceLineNo">439</span> ColumnFamilyDescriptorBuilder.newBuilder(INFO_FAMILY_NAME)<a name="line.439"></a>
+<span class="sourceLineNo">440</span> .setDataBlockEncoding(DataBlockEncoding.NONE)<a name="line.440"></a>
+<span class="sourceLineNo">441</span> .setBloomFilterType(BloomType.ROWCOL)<a name="line.441"></a>
+<span class="sourceLineNo">442</span> .setMaxVersions(1000)<a name="line.442"></a>
+<span class="sourceLineNo">443</span> .setBlocksize(8 * 1024)<a name="line.443"></a>
+<span class="sourceLineNo">444</span> ;<a name="line.444"></a>
+<span class="sourceLineNo">445</span><a name="line.445"></a>
+<span class="sourceLineNo">446</span> Set<ColumnFamilyDescriptor> families = new HashSet<>();<a name="line.446"></a>
+<span class="sourceLineNo">447</span> families.add(contentFamilyBuilder.build());<a name="line.447"></a>
+<span class="sourceLineNo">448</span> families.add(infoFamilyBuilder.build());<a name="line.448"></a>
+<span class="sourceLineNo">449</span><a name="line.449"></a>
+<span class="sourceLineNo">450</span> TableDescriptor tableDescriptor = TableDescriptorBuilder.newBuilder(tableName)<a name="line.450"></a>
+<span class="sourceLineNo">451</span> .setColumnFamilies(families)<a name="line.451"></a>
+<span class="sourceLineNo">452</span> .build();<a name="line.452"></a>
+<span class="sourceLineNo">453</span><a name="line.453"></a>
+<span class="sourceLineNo">454</span> if (getConf().getBoolean(HBaseTestingUtil.PRESPLIT_TEST_TABLE_KEY,<a name="line.454"></a>
+<span class="sourceLineNo">455</span> HBaseTestingUtil.PRESPLIT_TEST_TABLE)) {<a name="line.455"></a>
+<span class="sourceLineNo">456</span> int numberOfServers = admin.getRegionServers().size();<a name="line.456"></a>
+<span class="sourceLineNo">457</span> if (numberOfServers == 0) {<a name="line.457"></a>
+<span class="sourceLineNo">458</span> throw new IllegalStateException("No live regionservers");<a name="line.458"></a>
+<span class="sourceLineNo">459</span> }<a name="line.459"></a>
+<span class="sourceLineNo">460</span> int regionsPerServer = getConf().getInt(HBaseTestingUtil.REGIONS_PER_SERVER_KEY,<a name="line.460"></a>
+<span class="sourceLineNo">461</span> HBaseTestingUtil.DEFAULT_REGIONS_PER_SERVER);<a name="line.461"></a>
+<span class="sourceLineNo">462</span> int totalNumberOfRegions = numberOfServers * regionsPerServer;<a name="line.462"></a>
+<span class="sourceLineNo">463</span> LOG.info("Creating test table: " + tableDescriptor);<a name="line.463"></a>
+<span class="sourceLineNo">464</span> LOG.info("Number of live regionservers: " + numberOfServers + ", " +<a name="line.464"></a>
+<span class="sourceLineNo">465</span> "pre-splitting table into " + totalNumberOfRegions + " regions " +<a name="line.465"></a>
+<span class="sourceLineNo">466</span> "(default regions per server: " + regionsPerServer + ")");<a name="line.466"></a>
+<span class="sourceLineNo">467</span> byte[][] splits = new RegionSplitter.UniformSplit().split(totalNumberOfRegions);<a name="line.467"></a>
+<span class="sourceLineNo">468</span> admin.createTable(tableDescriptor, splits);<a name="line.468"></a>
+<span class="sourceLineNo">469</span> } else {<a name="line.469"></a>
+<span class="sourceLineNo">470</span> LOG.info("Creating test table: " + tableDescriptor);<a name="line.470"></a>
+<span class="sourceLineNo">471</span> admin.createTable(tableDescriptor);<a name="line.471"></a>
+<span class="sourceLineNo">472</span> }<a name="line.472"></a>
+<span class="sourceLineNo">473</span> }<a name="line.473"></a>
+<span class="sourceLineNo">474</span> } catch (MasterNotRunningException e) {<a name="line.474"></a>
+<span class="sourceLineNo">475</span> LOG.error("Master not running", e);<a name="line.475"></a>
+<span class="sourceLineNo">476</span> throw new IOException(e);<a name="line.476"></a>
+<span class="sourceLineNo">477</span> }<a name="line.477"></a>
+<span class="sourceLineNo">478</span> }<a name="line.478"></a>
<span class="sourceLineNo">479</span><a name="line.479"></a>
-<span class="sourceLineNo">480</span> createSchema(getTablename(getConf()));<a name="line.480"></a>
-<span class="sourceLineNo">481</span><a name="line.481"></a>
-<span class="sourceLineNo">482</span> Job job = Job.getInstance(getConf());<a name="line.482"></a>
-<span class="sourceLineNo">483</span> job.setJobName(Loader.class.getName());<a name="line.483"></a>
-<span class="sourceLineNo">484</span> job.setNumReduceTasks(0);<a name="line.484"></a>
-<span class="sourceLineNo">485</span> job.setJarByClass(getClass());<a name="line.485"></a>
-<span class="sourceLineNo">486</span> job.setMapperClass(LoaderMapper.class);<a name="line.486"></a>
-<span class="sourceLineNo">487</span> job.setInputFormatClass(WARCInputFormat.class);<a name="line.487"></a>
-<span class="sourceLineNo">488</span> FileSystem fs = FileSystem.get(warcFileInput.toUri(), getConf());<a name="line.488"></a>
-<span class="sourceLineNo">489</span> if (fs.getFileStatus(warcFileInput).isDirectory()) {<a name="line.489"></a>
-<span class="sourceLineNo">490</span> LOG.info("Using directory as WARC input path: " + warcFileInput);<a name="line.490"></a>
-<span class="sourceLineNo">491</span> FileInputFormat.setInputPaths(job, warcFileInput);<a name="line.491"></a>
-<span class="sourceLineNo">492</span> } else {<a name="line.492"></a>
-<span class="sourceLineNo">493</span> LOG.info("Getting WARC input paths from file: " + warcFileInput);<a name="line.493"></a>
-<span class="sourceLineNo">494</span> List<Path> paths = new LinkedList<Path>();<a name="line.494"></a>
-<span class="sourceLineNo">495</span> try (FSDataInputStream is = fs.open(warcFileInput)) {<a name="line.495"></a>
-<span class="sourceLineNo">496</span> InputStreamReader reader;<a name="line.496"></a>
-<span class="sourceLineNo">497</span> if (warcFileInput.getName().toLowerCase().endsWith(".gz")) {<a name="line.497"></a>
-<span class="sourceLineNo">498</span> reader = new InputStreamReader(new GZIPInputStream(is));<a name="line.498"></a>
-<span class="sourceLineNo">499</span> } else {<a name="line.499"></a>
-<span class="sourceLineNo">500</span> reader = new InputStreamReader(is);<a name="line.500"></a>
-<span class="sourceLineNo">501</span> }<a name="line.501"></a>
-<span class="sourceLineNo">502</span> try (BufferedReader br = new BufferedReader(reader)) {<a name="line.502"></a>
-<span class="sourceLineNo">503</span> String line;<a name="line.503"></a>
-<span class="sourceLineNo">504</span> while ((line = br.readLine()) != null) {<a name="line.504"></a>
-<span class="sourceLineNo">505</span> paths.add(new Path(line));<a name="line.505"></a>
-<span class="sourceLineNo">506</span> }<a name="line.506"></a>
-<span class="sourceLineNo">507</span> }<a name="line.507"></a>
-<span class="sourceLineNo">508</span> }<a name="line.508"></a>
-<span class="sourceLineNo">509</span> LOG.info("Read " + paths.size() + " WARC input paths from " + warcFileInput);<a name="line.509"></a>
-<span class="sourceLineNo">510</span> FileInputFormat.setInputPaths(job, paths.toArray(new Path[paths.size()]));<a name="line.510"></a>
-<span class="sourceLineNo">511</span> }<a name="line.511"></a>
-<span class="sourceLineNo">512</span> job.setOutputFormatClass(SequenceFileOutputFormat.class);<a name="line.512"></a>
-<span class="sourceLineNo">513</span> SequenceFileOutputFormat.setOutputPath(job, outputDir);<a name="line.513"></a>
-<span class="sourceLineNo">514</span> SequenceFileOutputFormat.setOutputCompressionType(job, CompressionType.BLOCK);<a name="line.514"></a>
-<span class="sourceLineNo">515</span> job.setOutputKeyClass(HBaseKeyWritable.class);<a name="line.515"></a>
-<span class="sourceLineNo">516</span> job.setOutputValueClass(BytesWritable.class);<a name="line.516"></a>
-<span class="sourceLineNo">517</span> TableMapReduceUtil.addDependencyJars(job);<a name="line.517"></a>
-<span class="sourceLineNo">518</span><a name="line.518"></a>
-<span class="sourceLineNo">519</span> LOG.info("Submitting job." +<a name="line.519"></a>
-<span class="sourceLineNo">520</span> " This will take time proportional to the number of input files, please be patient.");<a name="line.520"></a>
-<span class="sourceLineNo">521</span> boolean success = job.waitForCompletion(true);<a name="line.521"></a>
-<span class="sourceLineNo">522</span> if (!success) {<a name="line.522"></a>
-<span class="sourceLineNo">523</span> LOG.error("Failure during job " + job.getJobID());<a name="line.523"></a>
-<span class="sourceLineNo">524</span> }<a name="line.524"></a>
-<span class="sourceLineNo">525</span> return success ? 0 : 1;<a name="line.525"></a>
-<span class="sourceLineNo">526</span> }<a name="line.526"></a>
-<span class="sourceLineNo">527</span><a name="line.527"></a>
-<span class="sourceLineNo">528</span> @Override<a name="line.528"></a>
-<span class="sourceLineNo">529</span> public int run(String[] args) throws Exception {<a name="line.529"></a>
-<span class="sourceLineNo">530</span> if (args.length < 2) {<a name="line.530"></a>
-<span class="sourceLineNo">531</span> System.err.println(USAGE);<a name="line.531"></a>
-<span class="sourceLineNo">532</span> return 1;<a name="line.532"></a>
-<span class="sourceLineNo">533</span> }<a name="line.533"></a>
-<span class="sourceLineNo">534</span> try {<a name="line.534"></a>
-<span class="sourceLineNo">535</span> Path warcFileInput = new Path(args[0]);<a name="line.535"></a>
-<span class="sourceLineNo">536</span> Path outputDir = new Path(args[1]);<a name="line.536"></a>
-<span class="sourceLineNo">537</span> return run(warcFileInput, outputDir);<a name="line.537"></a>
-<span class="sourceLineNo">538</span> } catch (NumberFormatException e) {<a name="line.538"></a>
-<span class="sourceLineNo">539</span> System.err.println("Parsing loader arguments failed: " + e.getMessage());<a name="line.539"></a>
-<span class="sourceLineNo">540</span> System.err.println(USAGE);<a name="line.540"></a>
-<span class="sourceLineNo">541</span> return 1;<a name="line.541"></a>
-<span class="sourceLineNo">542</span> }<a name="line.542"></a>
-<span class="sourceLineNo">543</span> }<a name="line.543"></a>
-<span class="sourceLineNo">544</span><a name="line.544"></a>
-<span class="sourceLineNo">545</span> public static void main(String[] args) throws Exception {<a name="line.545"></a>
-<span class="sourceLineNo">546</span> System.exit(ToolRunner.run(HBaseConfiguration.create(), new Loader(), args));<a name="line.546"></a>
-<span class="sourceLineNo">547</span> }<a name="line.547"></a>
-<span class="sourceLineNo">548</span><a name="line.548"></a>
-<span class="sourceLineNo">549</span> public static class LoaderMapper<a name="line.549"></a>
-<span class="sourceLineNo">550</span> extends Mapper<LongWritable, WARCWritable, HBaseKeyWritable, BytesWritable> {<a name="line.550"></a>
-<span class="sourceLineNo">551</span><a name="line.551"></a>
-<span class="sourceLineNo">552</span> Configuration conf;<a name="line.552"></a>
-<span class="sourceLineNo">553</span> Connection conn;<a name="line.553"></a>
-<span class="sourceLineNo">554</span> Table table;<a name="line.554"></a>
-<span class="sourceLineNo">555</span><a name="line.555"></a>
-<span class="sourceLineNo">556</span> @Override<a name="line.556"></a>
-<span class="sourceLineNo">557</span> protected void setup(Context context) throws IOException, InterruptedException {<a name="line.557"></a>
-<span class="sourceLineNo">558</span> conn = ConnectionFactory.createConnection(context.getConfiguration());<a name="line.558"></a>
-<span class="sourceLineNo">559</span> table = conn.getTable(getTablename(conn.getConfiguration()));<a name="line.559"></a>
-<span class="sourceLineNo">560</span> }<a name="line.560"></a>
-<span class="sourceLineNo">561</span><a name="line.561"></a>
-<span class="sourceLineNo">562</span> @Override<a name="line.562"></a>
-<span class="sourceLineNo">563</span> protected void cleanup(Context context) throws IOException, InterruptedException {<a name="line.563"></a>
-<span class="sourceLineNo">564</span> try {<a name="line.564"></a>
-<span class="sourceLineNo">565</span> table.close();<a name="line.565"></a>
-<span class="sourceLineNo">566</span> } catch (Exception e) {<a name="line.566"></a>
-<span class="sourceLineNo">567</span> LOG.warn("Exception closing Table", e);<a name="line.567"></a>
-<span class="sourceLineNo">568</span> }<a name="line.568"></a>
-<span class="sourceLineNo">569</span> try {<a name="line.569"></a>
-<span class="sourceLineNo">570</span> conn.close();<a name="line.570"></a>
-<span class="sourceLineNo">571</span> } catch (Exception e) {<a name="line.571"></a>
-<span class="sourceLineNo">572</span> LOG.warn("Exception closing Connection", e);<a name="line.572"></a>
-<span class="sourceLineNo">573</span> }<a name="line.573"></a>
-<span class="sourceLineNo">574</span> }<a name="line.574"></a>
-<span class="sourceLineNo">575</span><a name="line.575"></a>
-<span class="sourceLineNo">576</span> @Override<a name="line.576"></a>
-<span class="sourceLineNo">577</span> protected void map(LongWritable key, WARCWritable value, Context output)<a name="line.577"></a>
-<span class="sourceLineNo">578</span> throws IOException, InterruptedException {<a name="line.578"></a>
-<span class="sourceLineNo">579</span> WARCRecord.Header warcHeader = value.getRecord().getHeader();<a name="line.579"></a>
-<span class="sourceLineNo">580</span> String recordID = warcHeader.getRecordID();<a name="line.580"></a>
-<span class="sourceLineNo">581</span> String targetURI = warcHeader.getTargetURI();<a name="line.581"></a>
-<span class="sourceLineNo">582</span> if (warcHeader.getRecordType().equals("response") && targetURI != null) {<a name="line.582"></a>
-<span class="sourceLineNo">583</span> String contentType = warcHeader.getField("WARC-Identified-Payload-Type");<a name="line.583"></a>
-<span class="sourceLineNo">584</span> if (contentType != null) {<a name="line.584"></a>
-<span class="sourceLineNo">585</span> LOG.debug("Processing record id=" + recordID + ", targetURI=\"" + targetURI + "\"");<a name="line.585"></a>
-<span class="sourceLineNo">586</span> long now = EnvironmentEdgeManager.currentTime();<a name="line.586"></a>
-<span class="sourceLineNo">587</span><a name="line.587"></a>
-<span class="sourceLineNo">588</span> // Make row key<a name="line.588"></a>
-<span class="sourceLineNo">589</span><a name="line.589"></a>
-<span class="sourceLineNo">590</span> byte[] rowKey;<a name="line.590"></a>
-<span class="sourceLineNo">591</span> try {<a name="line.591"></a>
-<span class="sourceLineNo">592</span> rowKey = rowKeyFromTargetURI(targetURI);<a name="line.592"></a>
-<span class="sourceLineNo">593</span> } catch (IllegalArgumentException e) {<a name="line.593"></a>
-<span class="sourceLineNo">594</span> LOG.debug("Could not make a row key for record " + recordID + ", ignoring", e);<a name="line.594"></a>
-<span class="sourceLineNo">595</span> return;<a name="line.595"></a>
-<span class="sourceLineNo">596</span> } catch (URISyntaxException e) {<a name="line.596"></a>
-<span class="sourceLineNo">597</span> LOG.warn("Could not parse URI \"" + targetURI + "\" for record " + recordID +<a name="line.597"></a>
-<span class="sourceLineNo">598</span> ", ignoring");<a name="line.598"></a>
-<span class="sourceLineNo">599</span> return;<a name="line.599"></a>
-<span class="sourceLineNo">600</span> }<a name="line.600"></a>
-<span class="sourceLineNo">601</span><a name="line.601"></a>
-<span class="sourceLineNo">602</span> // Get the content and calculate the CRC64<a name="line.602"></a>
-<span class="sourceLineNo">603</span><a name="line.603"></a>
-<span class="sourceLineNo">604</span> byte[] content = value.getRecord().getContent();<a name="line.604"></a>
-<span class="sourceLineNo">605</span> CRC64 crc = new CRC64();<a name="line.605"></a>
-<span class="sourceLineNo">606</span> crc.update(content);<a name="line.606"></a>
-<span class="sourceLineNo">607</span> long crc64 = crc.getValue();<a name="line.607"></a>
-<span class="sourceLineNo">608</span><a name="line.608"></a>
-<span class="sourceLineNo">609</span> // Store to HBase<a name="line.609"></a>
-<span class="sourceLineNo">610</span><a name="line.610"></a>
-<span class="sourceLineNo">611</span> Put put = new Put(rowKey);<a name="line.611"></a>
-<span class="sourceLineNo">612</span> put.addColumn(CONTENT_FAMILY_NAME, CONTENT_QUALIFIER, now, content);<a name="line.612"></a>
-<span class="sourceLineNo">613</span> put.addColumn(INFO_FAMILY_NAME, CONTENT_LENGTH_QUALIFIER, now,<a name="line.613"></a>
-<span class="sourceLineNo">614</span> Bytes.toBytes(content.length));<a name="line.614"></a>
-<span class="sourceLineNo">615</span> put.addColumn(INFO_FAMILY_NAME, CONTENT_TYPE_QUALIFIER, now,<a name="line.615"></a>
-<span class="sourceLineNo">616</span> Bytes.toBytes(contentType));<a name="line.616"></a>
-<span class="sourceLineNo">617</span> put.addColumn(INFO_FAMILY_NAME, CRC_QUALIFIER, now, Bytes.toBytes(crc64));<a name="line.617"></a>
-<span class="sourceLineNo">618</span> put.addColumn(INFO_FAMILY_NAME, RECORD_ID_QUALIFIER, now, Bytes.toBytes(recordID));<a name="line.618"></a>
-<span class="sourceLineNo">619</span> put.addColumn(INFO_FAMILY_NAME, TARGET_URI_QUALIFIER, now, Bytes.toBytes(targetURI));<a name="line.619"></a>
-<span class="sourceLineNo">620</span> put.addColumn(INFO_FAMILY_NAME, DATE_QUALIFIER, now,<a name="line.620"></a>
-<span class="sourceLineNo">621</span> Bytes.toBytes(warcHeader.getDateString()));<a name="line.621"></a>
-<span class="sourceLineNo">622</span> String ipAddr = warcHeader.getField("WARC-IP-Address");<a name="line.622"></a>
-<span class="sourceLineNo">623</span> if (ipAddr != null) {<a name="line.623"></a>
-<span class="sourceLineNo">624</span> put.addColumn(INFO_FAMILY_NAME, IP_ADDRESS_QUALIFIER, now, Bytes.toBytes(ipAddr));<a name="line.624"></a>
-<span class="sourceLineNo">625</span> }<a name="line.625"></a>
-<span class="sourceLineNo">626</span> table.put(put);<a name="line.626"></a>
-<span class="sourceLineNo">627</span><a name="line.627"></a>
-<span class="sourceLineNo">628</span> // Write records out for later verification, one per HBase field except for the<a name="line.628"></a>
-<span class="sourceLineNo">629</span> // content record, which will be verified by CRC64.<a name="line.629"></a>
+<span class="sourceLineNo">480</span> int run(Path warcFileInput, Path outputDir)<a name="line.480"></a>
+<span class="sourceLineNo">481</span> throws IOException, ClassNotFoundException, InterruptedException {<a name="line.481"></a>
+<span class="sourceLineNo">482</span><a name="line.482"></a>
+<span class="sourceLineNo">483</span> createSchema(getTablename(getConf()));<a name="line.483"></a>
+<span class="sourceLineNo">484</span><a name="line.484"></a>
+<span class="sourceLineNo">485</span> Job job = Job.getInstance(getConf());<a name="line.485"></a>
+<span class="sourceLineNo">486</span> job.setJobName(Loader.class.getName());<a name="line.486"></a>
+<span class="sourceLineNo">487</span> job.setNumReduceTasks(0);<a name="line.487"></a>
+<span class="sourceLineNo">488</span> job.setJarByClass(getClass());<a name="line.488"></a>
+<span class="sourceLineNo">489</span> job.setMapperClass(LoaderMapper.class);<a name="line.489"></a>
+<span class="sourceLineNo">490</span> job.setInputFormatClass(WARCInputFormat.class);<a name="line.490"></a>
+<span class="sourceLineNo">491</span> FileSystem fs = FileSystem.get(warcFileInput.toUri(), getConf());<a name="line.491"></a>
+<span class="sourceLineNo">492</span> if (fs.getFileStatus(warcFileInput).isDirectory()) {<a name="line.492"></a>
+<span class="sourceLineNo">493</span> LOG.info("Using directory as WARC input path: " + warcFileInput);<a name="line.493"></a>
+<span class="sourceLineNo">494</span> FileInputFormat.setInputPaths(job, warcFileInput);<a name="line.494"></a>
+<span class="sourceLineNo">495</span> } else {<a name="line.495"></a>
+<span class="sourceLineNo">496</span> LOG.info("Getting WARC input paths from file: " + warcFileInput);<a name="line.496"></a>
+<span class="sourceLineNo">497</span> List<Path> paths = new LinkedList<Path>();<a name="line.497"></a>
+<span class="sourceLineNo">498</span> try (FSDataInputStream is = fs.open(warcFileInput)) {<a name="line.498"></a>
+<span class="sourceLineNo">499</span> InputStreamReader reader;<a name="line.499"></a>
+<span class="sourceLineNo">500</span> if (warcFileInput.getName().toLowerCase().endsWith(".gz")) {<a name="line.500"></a>
+<span class="sourceLineNo">501</span> reader = new InputStreamReader(new GZIPInputStream(is));<a name="line.501"></a>
+<span class="sourceLineNo">502</span> } else {<a name="line.502"></a>
+<span class="sourceLineNo">503</span> reader = new InputStreamReader(is);<a name="line.503"></a>
+<span class="sourceLineNo">504</span> }<a name="line.504"></a>
+<span class="sourceLineNo">505</span> try (BufferedReader br = new BufferedReader(reader)) {<a name="line.505"></a>
+<span class="sourceLineNo">506</span> String line;<a name="line.506"></a>
+<span class="sourceLineNo">507</span> while ((line = br.readLine()) != null) {<a name="line.507"></a>
+<span class="sourceLineNo">508</span> paths.add(new Path(line));<a name="line.508"></a>
+<span class="sourceLineNo">509</span> }<a name="line.509"></a>
+<span class="sourceLineNo">510</span> }<a name="line.510"></a>
+<span class="sourceLineNo">511</span> }<a name="line.511"></a>
+<span class="sourceLineNo">512</span> LOG.info("Read " + paths.size() + " WARC input paths from " + warcFileInput);<a name="line.512"></a>
+<span class="sourceLineNo">513</span> FileInputFormat.setInputPaths(job, paths.toArray(new Path[paths.size()]));<a name="line.513"></a>
+<span class="sourceLineNo">514</span> }<a name="line.514"></a>
+<span class="sourceLineNo">515</span> job.setOutputFormatClass(SequenceFileOutputFormat.class);<a name="line.515"></a>
+<span class="sourceLineNo">516</span> SequenceFileOutputFormat.setOutputPath(job, outputDir);<a name="line.516"></a>
+<span class="sourceLineNo">517</span> SequenceFileOutputFormat.setOutputCompressionType(job, CompressionType.BLOCK);<a name="line.517"></a>
+<span class="sourceLineNo">518</span> job.setOutputKeyClass(HBaseKeyWritable.class);<a name="line.518"></a>
+<span class="sourceLineNo">519</span> job.setOutputValueClass(BytesWritable.class);<a name="line.519"></a>
+<span class="sourceLineNo">520</span> TableMapReduceUtil.addDependencyJars(job);<a name="line.520"></a>
+<span class="sourceLineNo">521</span><a name="line.521"></a>
+<span class="sourceLineNo">522</span> boolean success = job.waitForCompletion(true);<a name="line.522"></a>
+<span class="sourceLineNo">523</span> if (!success) {<a name="line.523"></a>
+<span class="sourceLineNo">524</span> LOG.error("Failure during job " + job.getJobID());<a name="line.524"></a>
+<span class="sourceLineNo">525</span> }<a name="line.525"></a>
+<span class="sourceLineNo">526</span> return success ? 0 : 1;<a name="line.526"></a>
+<span class="sourceLineNo">527</span> }<a name="line.527"></a>
+<span class="sourceLineNo">528</span><a name="line.528"></a>
+<span class="sourceLineNo">529</span> @Override<a name="line.529"></a>
+<span class="sourceLineNo">530</span> public int run(String[] args) throws Exception {<a name="line.530"></a>
+<span class="sourceLineNo">531</span> if (args.length < 2) {<a name="line.531"></a>
+<span class="sourceLineNo">532</span> System.err.println(USAGE);<a name="line.532"></a>
+<span class="sourceLineNo">533</span> return 1;<a name="line.533"></a>
+<span class="sourceLineNo">534</span> }<a name="line.534"></a>
+<span class="sourceLineNo">535</span> try {<a name="line.535"></a>
+<span class="sourceLineNo">536</span> Path warcFileInput = new Path(args[0]);<a name="line.536"></a>
+<span class="sourceLineNo">537</span> Path outputDir = new Path(args[1]);<a name="line.537"></a>
+<span class="sourceLineNo">538</span> return run(warcFileInput, outputDir);<a name="line.538"></a>
+<span class="sourceLineNo">539</span> } catch (NumberFormatException e) {<a name="line.539"></a>
+<span class="sourceLineNo">540</span> System.err.println("Parsing loader arguments failed: " + e.getMessage());<a name="line.540"></a>
+<span class="sourceLineNo">541</span> System.err.println(USAGE);<a name="line.541"></a>
+<span class="sourceLineNo">542</span> return 1;<a name="line.542"></a>
+<span class="sourceLineNo">543</span> }<a name="line.543"></a>
+<span class="sourceLineNo">544</span> }<a name="line.544"></a>
+<span class="sourceLineNo">545</span><a name="line.545"></a>
+<span class="sourceLineNo">546</span> public static void main(String[] args) throws Exception {<a name="line.546"></a>
+<span class="sourceLineNo">547</span> System.exit(ToolRunner.run(HBaseConfiguration.create(), new Loader(), args));<a name="line.547"></a>
+<span class="sourceLineNo">548</span> }<a name="line.548"></a>
+<span class="sourceLineNo">549</span><a name="line.549"></a>
+<span class="sourceLineNo">550</span> public static class LoaderMapper<a name="line.550"></a>
+<span class="sourceLineNo">551</span> extends Mapper<LongWritable, WARCWritable, HBaseKeyWritable, BytesWritable> {<a name="line.551"></a>
+<span class="sourceLineNo">552</span><a name="line.552"></a>
+<span class="sourceLineNo">553</span> Configuration conf;<a name="line.553"></a>
+<span class="sourceLineNo">554</span> Connection conn;<a name="line.554"></a>
+<span class="sourceLineNo">555</span> BufferedMutator mutator;<a name="line.555"></a>
+<span class="sourceLineNo">556</span><a name="line.556"></a>
+<span class="sourceLineNo">557</span> @Override<a name="line.557"></a>
+<span class="sourceLineNo">558</span> protected void setup(Context context) throws IOException, InterruptedException {<a name="line.558"></a>
+<span class="sourceLineNo">559</span> conf = context.getConfiguration();<a name="line.559"></a>
+<span class="sourceLineNo">560</span> conn = ConnectionFactory.createConnection(conf);<a name="line.560"></a>
+<span class="sourceLineNo">561</span> mutator = conn.getBufferedMutator(getTablename(conf));<a name="line.561"></a>
+<span class="sourceLineNo">562</span> mutator.setWriteBufferPeriodicFlush(10 * 1000); // default is 1 sec, increase to 10<a name="line.562"></a>
+<span class="sourceLineNo">563</span> }<a name="line.563"></a>
+<span class="sourceLineNo">564</span><a name="line.564"></a>
+<span class="sourceLineNo">565</span> @Override<a name="line.565"></a>
+<span class="sourceLineNo">566</span> protected void cleanup(Context context) throws IOException, InterruptedException {<a name="line.566"></a>
+<span class="sourceLineNo">567</span> try {<a name="line.567"></a>
+<span class="sourceLineNo">568</span> mutator.close();<a name="line.568"></a>
+<span class="sourceLineNo">569</span> } catch (Exception e) {<a name="line.569"></a>
+<span class="sourceLineNo">570</span> LOG.warn("Exception closing Table", e);<a name="line.570"></a>
+<span class="sourceLineNo">571</span> }<a name="line.571"></a>
+<span class="sourceLineNo">572</span> try {<a name="line.572"></a>
+<span class="sourceLineNo">573</span> conn.close();<a name="line.573"></a>
+<span class="sourceLineNo">574</span> } catch (Exception e) {<a name="line.574"></a>
+<span class="sourceLineNo">575</span> LOG.warn("Exception closing Connection", e);<a name="line.575"></a>
+<span class="sourceLineNo">576</span> }<a name="line.576"></a>
+<span class="sourceLineNo">577</span> }<a name="line.577"></a>
+<span class="sourceLineNo">578</span><a name="line.578"></a>
+<span class="sourceLineNo">579</span> @Override<a name="line.579"></a>
+<span class="sourceLineNo">580</span> protected void map(LongWritable key, WARCWritable value, Context output)<a name="line.580"></a>
+<span class="sourceLineNo">581</span> throws IOException, InterruptedException {<a name="line.581"></a>
+<span class="sourceLineNo">582</span> WARCRecord.Header warcHeader = value.getRecord().getHeader();<a name="line.582"></a>
+<span class="sourceLineNo">583</span> String recordID = warcHeader.getRecordID();<a name="line.583"></a>
+<span class="sourceLineNo">584</span> String targetURI = warcHeader.getTargetURI();<a name="line.584"></a>
+<span class="sourceLineNo">585</span> if (warcHeader.getRecordType().equals("response") && targetURI != null) {<a name="line.585"></a>
+<span class="sourceLineNo">586</span> String contentType = warcHeader.getField("WARC-Identified-Payload-Type");<a name="line.586"></a>
+<span class="sourceLineNo">587</span> if (contentType != null) {<a name="line.587"></a>
+<span class="sourceLineNo">588</span> LOG.info("Processing uri=\"" + targetURI + "\", id=" + recordID);<a name="line.588"></a>
+<span class="sourceLineNo">589</span> long now = EnvironmentEdgeManager.currentTime();<a name="line.589"></a>
+<span class="sourceLineNo">590</span><a name="line.590"></a>
+<span class="sourceLineNo">591</span> // Make row key<a name="line.591"></a>
+<span class="sourceLineNo">592</span><a name="line.592"></a>
+<span class="sourceLineNo">593</span> byte[] rowKey;<a name="line.593"></a>
+<span class="sourceLineNo">594</span> try {<a name="line.594"></a>
+<span class="sourceLineNo">595</span> rowKey = rowKeyFromTargetURI(targetURI);<a name="line.595"></a>
+<span class="sourceLineNo">596</span> } catch (IllegalArgumentException e) {<a name="line.596"></a>
+<span class="sourceLineNo">597</span> LOG.debug("Could not make a row key for record " + recordID + ", ignoring", e);<a name="line.597"></a>
+<span class="sourceLineNo">598</span> return;<a name="line.598"></a>
+<span class="sourceLineNo">599</span> } catch (URISyntaxException e) {<a name="line.599"></a>
+<span class="sourceLineNo">600</span> LOG.warn("Could not parse URI \"" + targetURI + "\" for record " + recordID +<a name="line.600"></a>
+<span class="sourceLineNo">601</span> ", ignoring");<a name="line.601"></a>
+<span class="sourceLineNo">602</span> return;<a name="line.602"></a>
+<span class="sourceLineNo">603</span> }<a name="line.603"></a>
+<span class="sourceLineNo">604</span><a name="line.604"></a>
+<span class="sourceLineNo">605</span> // Get the content and calculate the CRC64<a name="line.605"></a>
+<span class="sourceLineNo">606</span><a name="line.606"></a>
+<span class="sourceLineNo">607</span> byte[] content = value.getRecord().getContent();<a name="line.607"></a>
+<span class="sourceLineNo">608</span> CRC64 crc = new CRC64();<a name="line.608"></a>
+<span class="sourceLineNo">609</span> crc.update(content);<a name="line.609"></a>
+<span class="sourceLineNo">610</span> long crc64 = crc.getValue();<a name="line.610"></a>
+<span class="sourceLineNo">611</span><a name="line.611"></a>
+<span class="sourceLineNo">612</span> // Store to HBase<a name="line.612"></a>
+<span class="sourceLineNo">613</span><a name="line.613"></a>
+<span class="sourceLineNo">614</span> Put put = new Put(rowKey);<a name="line.614"></a>
+<span class="sourceLineNo">615</span> put.addColumn(CONTENT_FAMILY_NAME, CONTENT_QUALIFIER, now, content);<a name="line.615"></a>
+<span class="sourceLineNo">616</span> put.addColumn(INFO_FAMILY_NAME, CONTENT_LENGTH_QUALIFIER, now,<a name="line.616"></a>
+<span class="sourceLineNo">617</span> Bytes.toBytes(content.length));<a name="line.617"></a>
+<span class="sourceLineNo">618</span> put.addColumn(INFO_FAMILY_NAME, CONTENT_TYPE_QUALIFIER, now,<a name="line.618"></a>
+<span class="sourceLineNo">619</span> Bytes.toBytes(contentType));<a name="line.619"></a>
+<span class="sourceLineNo">620</span> put.addColumn(INFO_FAMILY_NAME, CRC_QUALIFIER, now, Bytes.toBytes(crc64));<a name="line.620"></a>
+<span class="sourceLineNo">621</span> put.addColumn(INFO_FAMILY_NAME, RECORD_ID_QUALIFIER, now, Bytes.toBytes(recordID));<a name="line.621"></a>
+<span class="sourceLineNo">622</span> put.addColumn(INFO_FAMILY_NAME, TARGET_URI_QUALIFIER, now, Bytes.toBytes(targetURI));<a name="line.622"></a>
+<span class="sourceLineNo">623</span> put.addColumn(INFO_FAMILY_NAME, DATE_QUALIFIER, now,<a name="line.623"></a>
+<span class="sourceLineNo">624</span> Bytes.toBytes(warcHeader.getDateString()));<a name="line.624"></a>
+<span class="sourceLineNo">625</span> String ipAddr = warcHeader.getField("WARC-IP-Address");<a name="line.625"></a>
+<span class="sourceLineNo">626</span> if (ipAddr != null) {<a name="line.626"></a>
+<span class="sourceLineNo">627</span> put.addColumn(INFO_FAMILY_NAME, IP_ADDRESS_QUALIFIER, now, Bytes.toBytes(ipAddr));<a name="line.627"></a>
+<span class="sourceLineNo">628</span> }<a name="line.628"></a>
+<span class="sourceLineNo">629</span> mutator.mutate(put);<a name="line.629"></a>
<span class="sourceLineNo">630</span><a name="line.630"></a>
-<span class="sourceLineNo">631</span> output.write(new HBaseKeyWritable(rowKey, INFO_FAMILY_NAME, CRC_QUALIFIER, now),<a name="line.631"></a>
-<span class="sourceLineNo">632</span> new BytesWritable(Bytes.toBytes(crc64)));<a name="line.632"></a>
-<span class="sourceLineNo">633</span> output.write(new HBaseKeyWritable(rowKey, INFO_FAMILY_NAME, CONTENT_LENGTH_QUALIFIER,<a name="line.633"></a>
-<span class="sourceLineNo">634</span> now), new BytesWritable(Bytes.toBytes(content.length)));<a name="line.634"></a>
-<span class="sourceLineNo">635</span> output.write(new HBaseKeyWritable(rowKey, INFO_FAMILY_NAME, CONTENT_TYPE_QUALIFIER,<a name="line.635"></a>
-<span class="sourceLineNo">636</span> now), new BytesWritable(Bytes.toBytes(contentType)));<a name="line.636"></a>
-<span class="sourceLineNo">637</span> output.write(new HBaseKeyWritable(rowKey, INFO_FAMILY_NAME, RECORD_ID_QUALIFIER,<a name="line.637"></a>
-<span class="sourceLineNo">638</span> now), new BytesWritable(Bytes.toBytes(recordID)));<a name="line.638"></a>
-<span class="sourceLineNo">639</span> output.write(new HBaseKeyWritable(rowKey, INFO_FAMILY_NAME, TARGET_URI_QUALIFIER,<a name="line.639"></a>
-<span class="sourceLineNo">640</span> now), new BytesWritable(Bytes.toBytes(targetURI)));<a name="line.640"></a>
-<span class="sourceLineNo">641</span> output.write(new HBaseKeyWritable(rowKey, INFO_FAMILY_NAME, DATE_QUALIFIER, now),<a name="line.641"></a>
-<span class="sourceLineNo">642</span> new BytesWritable(Bytes.toBytes(warcHeader.getDateString())));<a name="line.642"></a>
-<span class="sourceLineNo">643</span> if (ipAddr != null) {<a name="line.643"></a>
-<span class="sourceLineNo">644</span> output.write(new HBaseKeyWritable(rowKey, INFO_FAMILY_NAME, IP_ADDRESS_QUALIFIER,<a name="line.644"></a>
-<span class="sourceLineNo">645</span> now), new BytesWritable(Bytes.toBytes(ipAddr)));<a name="line.645"></a>
-<span class="sourceLineNo">646</span> }<a name="line.646"></a>
-<span class="sourceLineNo">647</span> }<a name="line.647"></a>
-<span class="sourceLineNo">648</span> }<a name="line.648"></a>
-<span class="sourceLineNo">649</span> }<a name="line.649"></a>
-<span class="sourceLineNo">650</span><a name="line.650"></a>
-<span class="sourceLineNo">651</span> private byte[] rowKeyFromTargetURI(String targetUri)<a name="line.651"></a>
-<span class="sourceLineNo">652</span> throws URISyntaxException, IllegalArgumentException {<a name="line.652"></a>
-<span class="sourceLineNo">653</span> URI uri = new URI(targetUri);<a name="line.653"></a>
-<span class="sourceLineNo">654</span> StringBuffer sb = new StringBuffer();<a name="line.654"></a>
-<span class="sourceLineNo">655</span> // Ignore the scheme<a name="line.655"></a>
-<span class="sourceLineNo">656</span> // Reverse the components of the hostname<a name="line.656"></a>
-<span class="sourceLineNo">657</span> if (uri.getHost() != null) {<a name="line.657"></a>
-<span class="sourceLineNo">658</span> String[] hostComponents = uri.getHost().split("\\.");<a name="line.658"></a>
-<span class="sourceLineNo">659</span> for (int i = hostComponents.length - 1; i >= 0; i--) {<a name="line.659"></a>
-<span class="sourceLineNo">660</span> sb.append(hostComponents[i]);<a name="line.660"></a>
-<span class="sourceLineNo">661</span> if (i != 0) {<a name="line.661"></a>
-<span class="sourceLineNo">662</span> sb.append('.');<a name="line.662"></a>
-<span class="sourceLineNo">663</span> }<a name="line.663"></a>
-<span class="sourceLineNo">664</span> }<a name="line.664"></a>
-<span class="sourceLineNo">665</span> } else {<a name="line.665"></a>
-<span class="sourceLineNo">666</span> throw new IllegalArgumentException("URI is missing host component");<a name="line.666"></a>
-<span class="sourceLineNo">667</span> }<a name="line.667"></a>
-<span class="sourceLineNo">668</span> // Port<a name="line.668"></a>
-<span class="sourceLineNo">669</span> if (uri.getPort() != -1) {<a name="line.669"></a>
-<span class="sourceLineNo">670</span> sb.append(':');<a name="line.670"></a>
-<span class="sourceLineNo">671</span> sb.append(uri.getPort());<a name="line.671"></a>
+<span class="sourceLineNo">631</span> // Write records out for later verification, one per HBase field except for the<a name="line.631"></a>
+<span class="sourceLineNo">632</span> // content record, which will be verified by CRC64.<a name="line.632"></a>
+<span class="sourceLineNo">633</span><a name="line.633"></a>
+<span class="sourceLineNo">634</span> output.write(new HBaseKeyWritable(rowKey, INFO_FAMILY_NAME, CRC_QUALIFIER, now),<a name="line.634"></a>
+<span class="sourceLineNo">635</span> new BytesWritable(Bytes.toBytes(crc64)));<a name="line.635"></a>
+<span class="sourceLineNo">636</span> output.write(new HBaseKeyWritable(rowKey, INFO_FAMILY_NAME, CONTENT_LENGTH_QUALIFIER,<a name="line.636"></a>
+<span class="sourceLineNo">637</span> now), new BytesWritable(Bytes.toBytes(content.length)));<a name="line.637"></a>
+<span class="sourceLineNo">638</span> output.write(new HBaseKeyWritable(rowKey, INFO_FAMILY_NAME, CONTENT_TYPE_QUALIFIER,<a name="line.638"></a>
+<span class="sourceLineNo">639</span> now), new BytesWritable(Bytes.toBytes(contentType)));<a name="line.639"></a>
+<span class="sourceLineNo">640</span> output.write(new HBaseKeyWritable(rowKey, INFO_FAMILY_NAME, RECORD_ID_QUALIFIER,<a name="line.640"></a>
+<span class="sourceLineNo">641</span> now), new BytesWritable(Bytes.toBytes(recordID)));<a name="line.641"></a>
+<span class="sourceLineNo">642</span> output.write(new HBaseKeyWritable(rowKey, INFO_FAMILY_NAME, TARGET_URI_QUALIFIER,<a name="line.642"></a>
+<span class="sourceLineNo">643</span> now), new BytesWritable(Bytes.toBytes(targetURI)));<a name="line.643"></a>
+<span class="sourceLineNo">644</span> output.write(new HBaseKeyWritable(rowKey, INFO_FAMILY_NAME, DATE_QUALIFIER, now),<a name="line.644"></a>
+<span class="sourceLineNo">645</span> new BytesWritable(Bytes.toBytes(warcHeader.getDateString())));<a name="line.645"></a>
+<span class="sourceLineNo">646</span> if (ipAddr != null) {<a name="line.646"></a>
+<span class="sourceLineNo">647</span> output.write(new HBaseKeyWritable(rowKey, INFO_FAMILY_NAME, IP_ADDRESS_QUALIFIER,<a name="line.647"></a>
+<span class="sourceLineNo">648</span> now), new BytesWritable(Bytes.toBytes(ipAddr)));<a name="line.648"></a>
+<span class="sourceLineNo">649</span> }<a name="line.649"></a>
+<span class="sourceLineNo">650</span> }<a name="line.650"></a>
+<span class="sourceLineNo">651</span> }<a name="line.651"></a>
+<span class="sourceLineNo">652</span> }<a name="line.652"></a>
+<span class="sourceLineNo">653</span><a name="line.653"></a>
+<span class="sourceLineNo">654</span> private byte[] rowKeyFromTargetURI(String targetUri)<a name="line.654"></a>
+<span class="sourceLineNo">655</span> throws URISyntaxException, IllegalArgumentException {<a name="line.655"></a>
+<span class="sourceLineNo">656</span> URI uri = new URI(targetUri);<a name="line.656"></a>
+<span class="sourceLineNo">657</span> // Ignore the scheme<a name="line.657"></a>
+<span class="sourceLineNo">658</span> // Reverse the components of the hostname<a name="line.658"></a>
+<span class="sourceLineNo">659</span> String reversedHost;<a name="line.659"></a>
+<span class="sourceLineNo">660</span> if (uri.getHost() != null) {<a name="line.660"></a>
+<span class="sourceLineNo">661</span> StringBuffer sb = new StringBuffer();<a name="line.661"></a>
+<span class="sourceLineNo">662</span> String[] hostComponents = uri.getHost().split("\\.");<a name="line.662"></a>
+<span class="sourceLineNo">663</span> for (int i = hostComponents.length - 1; i >= 0; i--) {<a name="line.663"></a>
+<span class="sourceLineNo">664</span> sb.append(hostComponents[i]);<a name="line.664"></a>
+<span class="sourceLineNo">665</span> if (i != 0) {<a name="line.665"></a>
+<span class="sourceLineNo">666</span> sb.append('.');<a name="line.666"></a>
+<span class="sourceLineNo">667</span> }<a name="line.667"></a>
+<span class="sourceLineNo">668</span> }<a name="line.668"></a>
+<span class="sourceLineNo">669</span> reversedHost = sb.toString();<a name="line.669"></a>
+<span class="sourceLineNo">670</span> } else {<a name="line.670"></a>
+<span class="sourceLineNo">671</span> throw new IllegalArgumentException("URI is missing host component");<a name="line.671"></a>
<span class="sourceLineNo">672</span> }<a name="line.672"></a>
-<span class="sourceLineNo">673</span> if (uri.getRawPath() != null) {<a name="line.673"></a>
-<span class="sourceLineNo">674</span> sb.append(uri.getRawPath());<a name="line.674"></a>
-<span class="sourceLineNo">675</span> }<a name="line.675"></a>
-<span class="sourceLineNo">676</span> if (uri.getRawQuery() != null) {<a name="line.676"></a>
-<span class="sourceLineNo">677</span> sb.append('?');<a name="line.677"></a>
-<span class="sourceLineNo">678</span> sb.append(uri.getRawQuery());<a name="line.678"></a>
-<span class="sourceLineNo">679</span> }<a name="line.679"></a>
-<span class="sourceLineNo">680</span> if (uri.getRawFragment() != null) {<a name="line.680"></a>
-<span class="sourceLineNo">681</span> sb.append('#');<a name="line.681"></a>
-<span class="sourceLineNo">682</span> sb.append(uri.getRawFragment());<a name="line.682"></a>
-<span class="sourceLineNo">683</span> }<a name="line.683"></a>
-<span class="sourceLineNo">684</span> // Constrain the key size to the maximum allowed row key length<a name="line.684"></a>
-<span class="sourceLineNo">685</span> if (sb.length() > HConstants.MAX_ROW_LENGTH) {<a name="line.685"></a>
-<span class="sourceLineNo">686</span> sb.setLength(HConstants.MAX_ROW_LENGTH);<a name="line.686"></a>
-<span class="sourceLineNo">687</span> }<a name="line.687"></a>
-<span class="sourceLineNo">688</span> return Bytes.toBytes(sb.toString());<a name="line.688"></a>
-<span class="sourceLineNo">689</span> }<a name="line.689"></a>
-<span class="sourceLineNo">690</span><a name="line.690"></a>
-<span class="sourceLineNo">691</span> }<a name="line.691"></a>
-<span class="sourceLineNo">692</span> }<a name="line.692"></a>
-<span class="sourceLineNo">693</span><a name="line.693"></a>
-<span class="sourceLineNo">694</span> public static class OneFilePerMapperSFIF<K, V> extends SequenceFileInputFormat<K, V> {<a name="line.694"></a>
-<span class="sourceLineNo">695</span> @Override<a name="line.695"></a>
-<span class="sourceLineNo">696</span> protected boolean isSplitable(JobContext context, Path filename) {<a name="line.696"></a>
-<span class="sourceLineNo">697</span> return false;<a name="line.697"></a>
+<span class="sourceLineNo">673</span> StringBuffer sb = new StringBuffer();<a name="line.673"></a>
+<span class="sourceLineNo">674</span> sb.append(reversedHost);<a name="line.674"></a>
+<span class="sourceLineNo">675</span> if (uri.getPort() >= 0) {<a name="line.675"></a>
+<span class="sourceLineNo">676</span> sb.append(':');<a name="line.676"></a>
+<span class="sourceLineNo">677</span> sb.append(uri.getPort());<a name="line.677"></a>
+<span class="sourceLineNo">678</span> }<a name="line.678"></a>
+<span class="sourceLineNo">679</span> if (uri.getPath() != null) {<a name="line.679"></a>
+<span class="sourceLineNo">680</span> sb.append('/');<a name="line.680"></a>
+<span class="sourceLineNo">681</span> sb.append(uri.getPath());<a name="line.681"></a>
+<span class="sourceLineNo">682</span> }<a name="line.682"></a>
+<span class="sourceLineNo">683</span> if (uri.getQuery() != null) {<a name="line.683"></a>
+<span class="sourceLineNo">684</span> sb.append('?');<a name="line.684"></a>
+<span class="sourceLineNo">685</span> sb.append(uri.getQuery());<a name="line.685"></a>
+<span class="sourceLineNo">686</span> }<a name="line.686"></a>
+<span class="sourceLineNo">687</span> if (uri.getFragment() != null) {<a name="line.687"></a>
+<span class="sourceLineNo">688</span> sb.append('#');<a name="line.688"></a>
+<span class="sourceLineNo">689</span> sb.append(uri.getFragment());<a name="line.689"></a>
+<span class="sourceLineNo">690</span> }<a name="line.690"></a>
+<span class="sourceLineNo">691</span> if (sb.length() > HConstants.MAX_ROW_LENGTH) {<a name="line.691"></a>
+<span class="sourceLineNo">692</span> throw new IllegalArgumentException("Key would be too large (length=" + sb.length() +<a name="line.692"></a>
+<span class="sourceLineNo">693</span> ", limit=" + HConstants.MAX_ROW_LENGTH);<a name="line.693"></a>
+<span class="sourceLineNo">694</span> }<a name="line.694"></a>
+<span class="sourceLineNo">695</span> return Bytes.toBytes(sb.toString());<a name="line.695"></a>
+<span class="sourceLineNo">696</span> }<a name="line.696"></a>
+<span class="sourceLineNo">697</span><a name="line.697"></a>
<span class="sourceLineNo">698</span> }<a name="line.698"></a>
<span class="sourceLineNo">699</span> }<a name="line.699"></a>
<span class="sourceLineNo">700</span><a name="line.700"></a>
-<span class="sourceLineNo">701</span> public static class Verify extends Configured implements Tool {<a name="line.701"></a>
-<span class="sourceLineNo">702</span><a name="line.702"></a>
-<span class="sourceLineNo">703</span> public static final Logger LOG = LoggerFactory.getLogger(Verify.class);<a name="line.703"></a>
-<span class="sourceLineNo">704</span> public static final String USAGE = "Verify <inputDir>";<a name="line.704"></a>
-<span class="sourceLineNo">705</span><a name="line.705"></a>
-<span class="sourceLineNo">706</span> int run(Path inputDir) throws IOException, ClassNotFoundException, InterruptedException {<a name="line.706"></a>
-<span class="sourceLineNo">707</span> Job job = Job.getInstance(getConf());<a name="line.707"></a>
-<span class="sourceLineNo">708</span> job.setJobName(Verify.class.getName());<a name="line.708"></a>
-<span class="sourceLineNo">709</span> job.setJarByClass(getClass());<a name="line.709"></a>
-<span class="sourceLineNo">710</span> job.setMapperClass(VerifyMapper.class);<a name="line.710"></a>
-<span class="sourceLineNo">711</span> job.setInputFormatClass(OneFilePerMapperSFIF.class);<a name="line.711"></a>
-<span class="sourceLineNo">712</span> FileInputFormat.setInputPaths(job, inputDir);<a name="line.712"></a>
-<span class="sourceLineNo">713</span> job.setOutputFormatClass(NullOutputFormat.class);<a name="line.713"></a>
-<span class="sourceLineNo">714</span> job.setOutputKeyClass(NullWritable.class);<a name="line.714"></a>
-<span class="sourceLineNo">715</span> job.setOutputValueClass(NullWritable.class);<a name="line.715"></a>
-<span class="sourceLineNo">716</span> TableMapReduceUtil.addDependencyJars(job);<a name="line.716"></a>
-<span class="sourceLineNo">717</span> boolean success = job.waitForCompletion(true);<a name="line.717"></a>
-<span class="sourceLineNo">718</span> if (!success) {<a name="line.718"></a>
-<span class="sourceLineNo">719</span> LOG.error("Failure during job " + job.getJobID());<a name="line.719"></a>
-<span class="sourceLineNo">720</span> }<a name="line.720"></a>
-<span class="sourceLineNo">721</span> Counters counters = job.getCounters();<a name="line.721"></a>
-<span class="sourceLineNo">722</span> for (Counts c: Counts.values()) {<a name="line.722"></a>
-<span class="sourceLineNo">723</span> LOG.info(c + ": " + counters.findCounter(c).getValue());<a name="line.723"></a>
-<span class="sourceLineNo">724</span> }<a name="line.724"></a>
-<span class="sourceLineNo">725</span> return success ? 0 : 1;<a name="line.725"></a>
-<span class="sourceLineNo">726</span> }<a name="line.726"></a>
-<span class="sourceLineNo">727</span><a name="line.727"></a>
-<span class="sourceLineNo">728</span> @Override<a name="line.728"></a>
-<span class="sourceLineNo">729</span> public int run(String[] args) throws Exception {<a name="line.729"></a>
-<span class="sourceLineNo">730</span> if (args.length < 2) {<a name="line.730"></a>
-<span class="sourceLineNo">731</span> System.err.println(USAGE);<a name="line.731"></a>
-<span class="sourceLineNo">732</span> return 1;<a name="line.732"></a>
-<span class="sourceLineNo">733</span> }<a name="line.733"></a>
-<span class="sourceLineNo">734</span> Path loaderOutput = new Path(args[0]);<a name="line.734"></a>
-<span class="sourceLineNo">735</span> return run(loaderOutput);<a name="line.735"></a>
-<span class="sourceLineNo">736</span> }<a name="line.736"></a>
-<span class="sourceLineNo">737</span><a name="line.737"></a>
-<span class="sourceLineNo">738</span> public static void main(String[] args) throws Exception {<a name="line.738"></a>
-<span class="sourceLineNo">739</span> System.exit(ToolRunner.run(HBaseConfiguration.create(), new Verify(), args));<a name="line.739"></a>
-<span class="sourceLineNo">740</span> }<a name="line.740"></a>
-<span class="sourceLineNo">741</span><a name="line.741"></a>
-<span class="sourceLineNo">742</span> public static class VerifyMapper<a name="line.742"></a>
-<span class="sourceLineNo">743</span> extends Mapper<HBaseKeyWritable, BytesWritable, NullWritable, NullWritable> {<a name="line.743"></a>
+<span class="sourceLineNo">701</span> public static class OneFilePerMapperSFIF<K, V> extends SequenceFileInputFormat<K, V> {<a name="line.701"></a>
+<span class="sourceLineNo">702</span> @Override<a name="line.702"></a>
+<span class="sourceLineNo">703</span> protected boolean isSplitable(JobContext context, Path filename) {<a name="line.703"></a>
+<span class="sourceLineNo">704</span> return false;<a name="line.704"></a>
+<span class="sourceLineNo">705</span> }<a name="line.705"></a>
+<span class="sourceLineNo">706</span> }<a name="line.706"></a>
+<span class="sourceLineNo">707</span><a name="line.707"></a>
+<span class="sourceLineNo">708</span> public static class Verify extends Configured implements Tool {<a name="line.708"></a>
+<span class="sourceLineNo">709</span><a name="line.709"></a>
+<span class="sourceLineNo">710</span> public static final Logger LOG = LoggerFactory.getLogger(Verify.class);<a name="line.710"></a>
+<span class="sourceLineNo">711</span> public static final String USAGE = "Verify <inputDir>";<a name="line.711"></a>
+<span class="sourceLineNo">712</span><a name="line.712"></a>
+<span class="sourceLineNo">713</span> int run(Path inputDir) throws IOException, ClassNotFoundException, InterruptedException {<a name="line.713"></a>
+<span class="sourceLineNo">714</span> Job job = Job.getInstance(getConf());<a name="line.714"></a>
+<span class="sourceLineNo">715</span> job.setJobName(Verify.class.getName());<a name="line.715"></a>
+<span class="sourceLineNo">716</span> job.setJarByClass(getClass());<a name="line.716"></a>
+<span class="sourceLineNo">717</span> job.setMapperClass(VerifyMapper.class);<a name="line.717"></a>
+<span class="sourceLineNo">718</span> job.setInputFormatClass(OneFilePerMapperSFIF.class);<a name="line.718"></a>
+<span class="sourceLineNo">719</span> FileInputFormat.setInputPaths(job, inputDir);<a name="line.719"></a>
+<span class="sourceLineNo">720</span> job.setOutputFormatClass(NullOutputFormat.class);<a name="line.720"></a>
+<span class="sourceLineNo">721</span> job.setOutputKeyClass(NullWritable.class);<a name="line.721"></a>
+<span class="sourceLineNo">722</span> job.setOutputValueClass(NullWritable.class);<a name="line.722"></a>
+<span class="sourceLineNo">723</span> TableMapReduceUtil.addDependencyJars(job);<a name="line.723"></a>
+<span class="sourceLineNo">724</span> boolean success = job.waitForCompletion(true);<a name="line.724"></a>
+<span class="sourceLineNo">725</span> if (!success) {<a name="line.725"></a>
+<span class="sourceLineNo">726</span> LOG.error("Failure during job " + job.getJobID());<a name="line.726"></a>
+<span class="sourceLineNo">727</span> }<a name="line.727"></a>
+<span class="sourceLineNo">728</span> Counters counters = job.getCounters();<a name="line.728"></a>
+<span class="sourceLineNo">729</span> for (Counts c: Counts.values()) {<a name="line.729"></a>
+<span class="sourceLineNo">730</span> LOG.info(c + ": " + counters.findCounter(c).getValue());<a name="line.730"></a>
+<span class="sourceLineNo">731</span> }<a name="line.731"></a>
+<span class="sourceLineNo">732</span> return success ? 0 : 1;<a name="line.732"></a>
+<span class="sourceLineNo">733</span> }<a name="line.733"></a>
+<span class="sourceLineNo">734</span><a name="line.734"></a>
+<span class="sourceLineNo">735</span> @Override<a name="line.735"></a>
+<span class="sourceLineNo">736</span> public int run(String[] args) throws Exception {<a name="line.736"></a>
+<span class="sourceLineNo">737</span> if (args.length < 2) {<a name="line.737"></a>
+<span class="sourceLineNo">738</span> System.err.println(USAGE);<a name="line.738"></a>
+<span class="sourceLineNo">739</span> return 1;<a name="line.739"></a>
+<span class="sourceLineNo">740</span> }<a name="line.740"></a>
+<span class="sourceLineNo">741</span> Path loaderOutput = new Path(args[0]);<a name="line.741"></a>
+<span class="sourceLineNo">742</span> return run(loaderOutput);<a name="line.742"></a>
+<span class="sourceLineNo">743</span> }<a name="line.743"></a>
<span class="sourceLineNo">744</span><a name="line.744"></a>
-<span class="sourceLineNo">745</span> Connection conn;<a name="line.745"></a>
-<span class="sourceLineNo">746</span> Table table;<a name="line.746"></a>
-<span class="sourceLineNo">747</span><a name="line.747"></a>
-<span class="sourceLineNo">748</span> @Override<a name="line.748"></a>
-<span class="sourceLineNo">749</span> protected void setup(Context context) throws IOException, InterruptedException {<a name="line.749"></a>
-<span class="sourceLineNo">750</span> conn = ConnectionFactory.createConnection(context.getConfiguration());<a name="line.750"></a>
-<span class="sourceLineNo">751</span> table = conn.getTable(getTablename(conn.getConfiguration()));<a name="line.751"></a>
-<span class="sourceLineNo">752</span> }<a name="line.752"></a>
-<span class="sourceLineNo">753</span><a name="line.753"></a>
-<span class="sourceLineNo">754</span> @Override<a name="line.754"></a>
-<span class="sourceLineNo">755</span> protected void cleanup(Context context) throws IOException ,InterruptedException {<a name="line.755"></a>
-<span class="sourceLineNo">756</span> table.close();<a name="line.756"></a>
-<span class="sourceLineNo">757</span> conn.close();<a name="line.757"></a>
-<span class="sourceLineNo">758</span> }<a name="line.758"></a>
-<span class="sourceLineNo">759</span><a name="line.759"></a>
-<span class="sourceLineNo">760</span> @Override<a name="line.760"></a>
-<span class="sourceLineNo">761</span> protected void map(HBaseKeyWritable key, BytesWritable value, Context output)<a name="line.761"></a>
-<span class="sourceLineNo">762</span> throws IOException, InterruptedException {<a name="line.762"></a>
-<span class="sourceLineNo">763</span><a name="line.763"></a>
-<span class="sourceLineNo">764</span> byte[] row = Bytes.copy(key.getRowArray(), key.getRowOffset(), key.getRowLength());<a name="line.764"></a>
-<span class="sourceLineNo">765</span> byte[] family = Bytes.copy(key.getFamilyArray(), key.getFamilyOffset(),<a name="line.765"></a>
-<span class="sourceLineNo">766</span> key.getFamilyLength());<a name="line.766"></a>
-<span class="sourceLineNo">767</span> byte[] qualifier = Bytes.copy(key.getQualifierArray(), key.getQualifierOffset(),<a name="line.767"></a>
-<span class="sourceLineNo">768</span> key.getQualifierLength());<a name="line.768"></a>
-<span class="sourceLineNo">769</span> long ts = key.getTimestamp();<a name="line.769"></a>
+<span class="sourceLineNo">745</span> public static void main(String[] args) throws Exception {<a name="line.745"></a>
+<span class="sourceLineNo">746</span> System.exit(ToolRunner.run(HBaseConfiguration.create(), new Verify(), args));<a name="line.746"></a>
+<span class="sourceLineNo">747</span> }<a name="line.747"></a>
+<span class="sourceLineNo">748</span><a name="line.748"></a>
+<span class="sourceLineNo">749</span> public static class VerifyMapper<a name="line.749"></a>
+<span class="sourceLineNo">750</span> extends Mapper<HBaseKeyWritable, BytesWritable, NullWritable, NullWritable> {<a name="line.750"></a>
+<span class="sourceLineNo">751</span><a name="line.751"></a>
+<span class="sourceLineNo">752</span> Connection conn;<a name="line.752"></a>
+<span class="sourceLineNo">753</span> Table table;<a name="line.753"></a>
+<span class="sourceLineNo">754</span><a name="line.754"></a>
+<span class="sourceLineNo">755</span> @Override<a name="line.755"></a>
+<span class="sourceLineNo">756</span> protected void setup(Context context) throws IOException, InterruptedException {<a name="line.756"></a>
+<span class="sourceLineNo">757</span> conn = ConnectionFactory.createConnection(context.getConfiguration());<a name="line.757"></a>
+<span class="sourceLineNo">758</span> table = conn.getTable(getTablename(conn.getConfiguration()));<a name="line.758"></a>
+<span class="sourceLineNo">759</span> }<a name="line.759"></a>
+<span class="sourceLineNo">760</span><a name="line.760"></a>
+<span class="sourceLineNo">761</span> @Override<a name="line.761"></a>
+<span class="sourceLineNo">762</span> protected void cleanup(Context context) throws IOException ,InterruptedException {<a name="line.762"></a>
+<span class="sourceLineNo">763</span> table.close();<a name="line.763"></a>
+<span class="sourceLineNo">764</span> conn.close();<a name="line.764"></a>
+<span class="sourceLineNo">765</span> }<a name="line.765"></a>
+<span class="sourceLineNo">766</span><a name="line.766"></a>
+<span class="sourceLineNo">767</span> @Override<a name="line.767"></a>
+<span class="sourceLineNo">768</span> protected void map(HBaseKeyWritable key, BytesWritable value, Context output)<a name="line.768"></a>
+<span class="sourceLineNo">769</span> throws IOException, InterruptedException {<a name="line.769"></a>
<span class="sourceLineNo">770</span><a name="line.770"></a>
-<span class="sourceLineNo">771</span> if (Bytes.equals(INFO_FAMILY_NAME, family) &&<a name="line.771"></a>
-<span class="sourceLineNo">772</span> Bytes.equals(CRC_QUALIFIER, qualifier)) {<a name="line.772"></a>
-<span class="sourceLineNo">773</span><a name="line.773"></a>
-<span class="sourceLineNo">774</span> long expectedCRC64 = Bytes.toLong(value.getBytes(), 0, value.getLength());<a name="line.774"></a>
-<span class="sourceLineNo">775</span><a name="line.775"></a>
-<span class="sourceLineNo">776</span> Result result =<a name="line.776"></a>
-<span class="sourceLineNo">777</span> table.get(new Get(row)<a name="line.777"></a>
-<span class="sourceLineNo">778</span> .addColumn(CONTENT_FAMILY_NAME, CONTENT_QUALIFIER)<a name="line.778"></a>
-<span class="sourceLineNo">779</span> .addColumn(INFO_FAMILY_NAME, CRC_QUALIFIER)<a name="line.779"></a>
-<span class="sourceLineNo">780</span> .setTimestamp(ts));<a name="line.780"></a>
-<span class="sourceLineNo">781</span><a name="line.781"></a>
-<span class="sourceLineNo">782</span> byte[] content = result.getValue(CONTENT_FAMILY_NAME, CONTENT_QUALIFIER);<a name="line.782"></a>
-<span class="sourceLineNo">783</span> if (content == null) {<a name="line.783"></a>
-<span class="sourceLineNo">784</span> LOG.info("Row " + Bytes.toStringBinary(row) + ": missing content");<a name="line.784"></a>
-<span class="sourceLineNo">785</span> output.getCounter(Counts.UNREFERENCED).increment(1);<a name="line.785"></a>
-<span class="sourceLineNo">786</span> return;<a name="line.786"></a>
-<span class="sourceLineNo">787</span> } else {<a name="line.787"></a>
-<span class="sourceLineNo">788</span> CRC64 crc = new CRC64();<a name="line.788"></a>
-<span class="sourceLineNo">789</span> crc.update(content);<a name="line.789"></a>
-<span class="sourceLineNo">790</span> if (crc.getValue() != expectedCRC64) {<a name="line.790"></a>
-<span class="sourceLineNo">791</span> LOG.info("Row " + Bytes.toStringBinary(row) + ": corrupt content");<a name="line.791"></a>
-<span class="sourceLineNo">792</span> output.getCounter(Counts.CORRUPT).increment(1);<a name="line.792"></a>
-<span class="sourceLineNo">793</span> return;<a name="line.793"></a>
-<span class="sourceLineNo">794</span> }<a name="line.794"></a>
-<span class="sourceLineNo">795</span> }<a name="line.795"></a>
-<span class="sourceLineNo">796</span> byte[] crc = result.getValue(INFO_FAMILY_NAME, CRC_QUALIFIER);<a name="line.796"></a>
-<span class="sourceLineNo">797</span> if (crc == null) {<a name="line.797"></a>
-<span class="sourceLineNo">798</span> LOG.info("Row " + Bytes.toStringBinary(row) + ": missing i:c");<a name="line.798"></a>
-<span class="sourceLineNo">799</span> output.getCounter(Counts.UNREFERENCED).increment(1);<a name="line.799"></a>
-<span class="sourceLineNo">800</span> return;<a name="line.800"></a>
-<span class="sourceLineNo">801</span> }<a name="line.801"></a>
-<span class="sourceLineNo">802</span> if (Bytes.toLong(crc) != expectedCRC64) {<a name="line.802"></a>
-<span class="sourceLineNo">803</span> LOG.info("Row " + Bytes.toStringBinary(row) + ": i:c mismatch");<a name="line.803"></a>
-<span class="sourceLineNo">804</span> output.getCounter(Counts.CORRUPT).increment(1);<a name="line.804"></a>
-<span class="sourceLineNo">805</span> return;<a name="line.805"></a>
-<span class="sourceLineNo">806</span> }<a name="line.806"></a>
-<span class="sourceLineNo">807</span><a name="line.807"></a>
-<span class="sourceLineNo">808</span> } else {<a name="line.808"></a>
-<span class="sourceLineNo">809</span><a name="line.809"></a>
-<span class="sourceLineNo">810</span> Result result =<a name="line.810"></a>
-<span class="sourceLineNo">811</span> table.get(new Get(row)<a name="line.811"></a>
-<span class="sourceLineNo">812</span> .addColumn(family, qualifier)<a name="line.812"></a>
-<span class="sourceLineNo">813</span> .setTimestamp(ts));<a name="line.813"></a>
-<span class="sourceLineNo">814</span><a name="line.814"></a>
-<span class="sourceLineNo">815</span> byte[] bytes = result.getValue(family, qualifier);<a name="line.815"></a>
-<span class="sourceLineNo">816</span> if (bytes == null) {<a name="line.816"></a>
-<span class="sourceLineNo">817</span> LOG.info("Row " + Bytes.toStringBinary(row) + ": missing " +<a name="line.817"></a>
-<span class="sourceLineNo">818</span> Bytes.toStringBinary(family) + ":" + Bytes.toStringBinary(qualifier));<a name="line.818"></a>
-<span class="sourceLineNo">819</span> output.getCounter(Counts.UNREFERENCED).increment(1);<a name="line.819"></a>
-<span class="sourceLineNo">820</span> return;<a name="line.820"></a>
-<span class="sourceLineNo">821</span> }<a name="line.821"></a>
-<span class="sourceLineNo">822</span> if (!Bytes.equals(bytes, 0, bytes.length, value.getBytes(), 0, value.getLength())) {<a name="line.822"></a>
-<span class="sourceLineNo">823</span> LOG.info("Row " + Bytes.toStringBinary(row) + ": " +<a name="line.823"></a>
-<span class="sourceLineNo">824</span> Bytes.toStringBinary(family) + ":" + Bytes.toStringBinary(qualifier) +<a name="line.824"></a>
-<span class="sourceLineNo">825</span> " mismatch");<a name="line.825"></a>
-<span class="sourceLineNo">826</span> output.getCounter(Counts.CORRUPT).increment(1);<a name="line.826"></a>
-<span class="sourceLineNo">827</span> return;<a name="line.827"></a>
-<span class="sourceLineNo">828</span> }<a name="line.828"></a>
-<span class="sourceLineNo">829</span><a name="line.829"></a>
-<span class="sourceLineNo">830</span> }<a name="line.830"></a>
-<span class="sourceLineNo">831</span><a name="line.831"></a>
-<span class="sourceLineNo">832</span> output.getCounter(Counts.REFERENCED).increment(1);<a name="line.832"></a>
-<span class="sourceLineNo">833</span> }<a name="line.833"></a>
-<span class="sourceLineNo">834</span><a name="line.834"></a>
-<span class="sourceLineNo">835</span> }<a name="line.835"></a>
-<span class="sourceLineNo">836</span><a name="line.836"></a>
-<span class="sourceLineNo">837</span> }<a name="line.837"></a>
-<span class="sourceLineNo">838</span><a name="line.838"></a>
-<span class="sourceLineNo">839</span>}<a name="line.839"></a>
+<span class="sourceLineNo">771</span> byte[] row = Bytes.copy(key.getRowArray(), key.getRowOffset(), key.getRowLength());<a name="line.771"></a>
+<span class="sourceLineNo">772</span> byte[] family = Bytes.copy(key.getFamilyArray(), key.getFamilyOffset(),<a name="line.772"></a>
+<span class="sourceLineNo">773</span> key.getFamilyLength());<a name="line.773"></a>
+<span class="sourceLineNo">774</span> byte[] qualifier = Bytes.copy(key.getQualifierArray(), key.getQualifierOffset(),<a name="line.774"></a>
+<span class="sourceLineNo">775</span> key.getQualifierLength());<a name="line.775"></a>
+<span class="sourceLineNo">776</span> long ts = key.getTimestamp();<a name="line.776"></a>
+<span class="sourceLineNo">777</span> int retries = VERIFICATION_READ_RETRIES;<a name="line.777"></a>
+<span class="sourceLineNo">778</span><a name="line.778"></a>
+<span class="sourceLineNo">779</span> while (true) {<a name="line.779"></a>
+<span class="sourceLineNo">780</span><a name="line.780"></a>
+<span class="sourceLineNo">781</span> if (Bytes.equals(INFO_FAMILY_NAME, family) &&<a name="line.781"></a>
+<span class="sourceLineNo">782</span> Bytes.equals(CRC_QUALIFIER, qualifier)) {<a name="line.782"></a>
+<span class="sourceLineNo">783</span><a name="line.783"></a>
+<span class="sourceLineNo">784</span> long expectedCRC64 = Bytes.toLong(value.getBytes(), 0, value.getLength());<a name="line.784"></a>
+<span class="sourceLineNo">785</span> Result result = table.get(new Get(row)<a name="line.785"></a>
+<span class="sourceLineNo">786</span> .addColumn(CONTENT_FAMILY_NAME, CONTENT_QUALIFIER)<a name="line.786"></a>
+<span class="sourceLineNo">787</span> .addColumn(INFO_FAMILY_NAME, CRC_QUALIFIER)<a name="line.787"></a>
+<span class="sourceLineNo">788</span> .setTimestamp(ts));<a name="line.788"></a>
+<span class="sourceLineNo">789</span> byte[] content = result.getValue(CONTENT_FAMILY_NAME, CONTENT_QUALIFIER);<a name="line.789"></a>
+<span class="sourceLineNo">790</span> if (content == null) {<a name="line.790"></a>
+<span class="sourceLineNo">791</span> if (retries-- > 0) {<a name="line.791"></a>
+<span class="sourceLineNo">792</span> continue;<a name="line.792"></a>
+<span class="sourceLineNo">793</span> }<a name="line.793"></a>
+<span class="sourceLineNo">794</span> LOG.error("Row " + Bytes.toStringBinary(row) + ": missing content");<a name="line.794"></a>
+<span class="sourceLineNo">795</span> output.getCounter(Counts.UNREFERENCED).increment(1);<a name="line.795"></a>
+<span class="sourceLineNo">796</span> return;<a name="line.796"></a>
+<span class="sourceLineNo">797</span> } else {<a name="line.797"></a>
+<span class="sourceLineNo">798</span> CRC64 crc = new CRC64();<a name="line.798"></a>
+<span class="sourceLineNo">799</span> crc.update(content);<a name="line.799"></a>
+<span class="sourceLineNo">800</span> if (crc.getValue() != expectedCRC64) {<a name="line.800"></a>
+<span class="sourceLineNo">801</span> if (retries-- > 0) {<a name="line.801"></a>
+<span class="sourceLineNo">802</span> continue;<a name="line.802"></a>
+<span class="sourceLineNo">803</span> }<a name="line.803"></a>
+<span class="sourceLineNo">804</span> LOG.error("Row " + Bytes.toStringBinary(row) + ": corrupt content");<a name="line.804"></a>
+<span class="sourceLineNo">805</span> output.getCounter(Counts.CORRUPT).increment(1);<a name="line.805"></a>
+<span class="sourceLineNo">806</span> return;<a name="line.806"></a>
+<span class="sourceLineNo">807</span> }<a name="line.807"></a>
+<span class="sourceLineNo">808</span> }<a name="line.808"></a>
+<span class="sourceLineNo">809</span> byte[] crc = result.getValue(INFO_FAMILY_NAME, CRC_QUALIFIER);<a name="line.809"></a>
+<span class="sourceLineNo">810</span> if (crc == null) {<a name="line.810"></a>
+<span class="sourceLineNo">811</span> if (retries-- > 0) {<a name="line.811"></a>
+<span class="sourceLineNo">812</span> continue;<a name="line.812"></a>
+<span class="sourceLineNo">813</span> }<a name="line.813"></a>
+<span class="sourceLineNo">814</span> LOG.error("Row " + Bytes.toStringBinary(row) + ": missing i:c");<a name="line.814"></a>
+<span class="sourceLineNo">815</span> output.getCounter(Counts.UNREFERENCED).increment(1);<a name="line.815"></a>
+<span class="sourceLineNo">816</span> return;<a name="line.816"></a>
+<span class="sourceLineNo">817</span> }<a name="line.817"></a>
+<span class="sourceLineNo">818</span> if (Bytes.toLong(crc) != expectedCRC64) {<a name="line.818"></a>
+<span class="sourceLineNo">819</span> if (retries-- > 0) {<a name="line.819"></a>
+<span class="sourceLineNo">820</span> continue;<a name="line.820"></a>
+<span class="sourceLineNo">821</span> }<a name="line.821"></a>
+<span class="sourceLineNo">822</span> LOG.error("Row " + Bytes.toStringBinary(row) + ": i:c mismatch");<a name="line.822"></a>
+<span class="sourceLineNo">823</span> output.getCounter(Counts.CORRUPT).increment(1);<a name="line.823"></a>
+<span class="sourceLineNo">824</span> return;<a name="line.824"></a>
+<span class="sourceLineNo">825</span> }<a name="line.825"></a>
+<span class="sourceLineNo">826</span><a name="line.826"></a>
+<span class="sourceLineNo">827</span> } else {<a name="line.827"></a>
+<span class="sourceLineNo">828</span><a name="line.828"></a>
+<span class="sourceLineNo">829</span> Result result = table.get(new Get(row)<a name="line.829"></a>
+<span class="sourceLineNo">830</span> .addColumn(family, qualifier)<a name="line.830"></a>
+<span class="sourceLineNo">831</span> .setTimestamp(ts));<a name="line.831"></a>
+<span class="sourceLineNo">832</span> byte[] bytes = result.getValue(family, qualifier);<a name="line.832"></a>
+<span class="sourceLineNo">833</span> if (bytes == null) {<a name="line.833"></a>
+<span class="sourceLineNo">834</span> if (retries-- > 0) {<a name="line.834"></a>
+<span class="sourceLineNo">835</span> continue;<a name="line.835"></a>
+<span class="sourceLineNo">836</span> }<a name="line.836"></a>
+<span class="sourceLineNo">837</span> LOG.error("Row " + Bytes.toStringBinary(row) + ": missing " +<a name="line.837"></a>
+<span class="sourceLineNo">838</span> Bytes.toStringBinary(family) + ":" + Bytes.toStringBinary(qualifier));<a name="line.838"></a>
+<span class="sourceLineNo">839</span> output.getCounter(Counts.UNREFERENCED).increment(1);<a name="line.839"></a>
+<span class="sourceLineNo">840</span> return;<a name="line.840"></a>
+<span class="sourceLineNo">841</span> }<a name="line.841"></a>
+<span class="sourceLineNo">842</span> if (!Bytes.equals(bytes, 0, bytes.length, value.getBytes(), 0, value.getLength())) {<a name="line.842"></a>
+<span class="sourceLineNo">843</span> if (retries-- > 0) {<a name="line.843"></a>
+<span class="sourceLineNo">844</span> continue;<a name="line.844"></a>
+<span class="sourceLineNo">845</span> }<a name="line.845"></a>
+<span class="sourceLineNo">846</span> LOG.error("Row " + Bytes.toStringBinary(row) + ": " +<a name="line.846"></a>
+<span class="sourceLineNo">847</span> Bytes.toStringBinary(family) + ":" + Bytes.toStringBinary(qualifier) +<a name="line.847"></a>
+<span class="sourceLineNo">848</span> " mismatch");<a name="line.848"></a>
+<span class="sourceLineNo">849</span> output.getCounter(Counts.CORRUPT).increment(1);<a name="line.849"></a>
+<span class="sourceLineNo">850</span> return;<a name="line.850"></a>
+<span class="sourceLineNo">851</span> }<a name="line.851"></a>
+<span class="sourceLineNo">852</span><a name="line.852"></a>
+<span class="sourceLineNo">853</span> }<a name="line.853"></a>
+<span class="sourceLineNo">854</span><a name="line.854"></a>
+<span class="sourceLineNo">855</span> // If we fell through to here all verification checks have succeeded, potentially after<a name="line.855"></a>
+<span class="sourceLineNo">856</span> // retries, and we must exit the while loop.<a name="line.856"></a>
+<span class="sourceLineNo">857</span> output.getCounter(Counts.REFERENCED).increment(1);<a name="line.857"></a>
+<span class="sourceLineNo">858</span> break;<a name="line.858"></a>
+<span class="sourceLineNo">859</span><a name="line.859"></a>
+<span class="sourceLineNo">860</span> }<a name="line.860"></a>
+<span class="sourceLineNo">861</span> }<a name="line.861"></a>
+<span class="sourceLineNo">862</span> }<a name="line.862"></a>
+<span class="sourceLineNo">863</span> }<a name="line.863"></a>
+<span class="sourceLineNo">864</span><a name="line.864"></a>
+<span class="sourceLineNo">865</span>}<a name="line.865"></a>
diff --git a/testdevapidocs/src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.Loader.html b/testdevapidocs/src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.Loader.html
index 4a40cbc..e89b63c 100644
--- a/testdevapidocs/src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.Loader.html
+++ b/testdevapidocs/src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.Loader.html
@@ -54,797 +54,823 @@
<span class="sourceLineNo">046</span>import org.apache.hadoop.hbase.MasterNotRunningException;<a name="line.46"></a>
<span class="sourceLineNo">047</span>import org.apache.hadoop.hbase.TableName;<a name="line.47"></a>
<span class="sourceLineNo">048</span>import org.apache.hadoop.hbase.client.Admin;<a name="line.48"></a>
-<span class="sourceLineNo">049</span>import org.apache.hadoop.hbase.client.ColumnFamilyDescriptor;<a name="line.49"></a>
-<span class="sourceLineNo">050</span>import org.apache.hadoop.hbase.client.ColumnFamilyDescriptorBuilder;<a name="line.50"></a>
-<span class="sourceLineNo">051</span>import org.apache.hadoop.hbase.client.Connection;<a name="line.51"></a>
-<span class="sourceLineNo">052</span>import org.apache.hadoop.hbase.client.ConnectionFactory;<a name="line.52"></a>
-<span class="sourceLineNo">053</span>import org.apache.hadoop.hbase.client.Get;<a name="line.53"></a>
-<span class="sourceLineNo">054</span>import org.apache.hadoop.hbase.client.Put;<a name="line.54"></a>
-<span class="sourceLineNo">055</span>import org.apache.hadoop.hbase.client.Result;<a name="line.55"></a>
-<span class="sourceLineNo">056</span>import org.apache.hadoop.hbase.client.Table;<a name="line.56"></a>
-<span class="sourceLineNo">057</span>import org.apache.hadoop.hbase.client.TableDescriptor;<a name="line.57"></a>
-<span class="sourceLineNo">058</span>import org.apache.hadoop.hbase.client.TableDescriptorBuilder;<a name="line.58"></a>
-<span class="sourceLineNo">059</span>import org.apache.hadoop.hbase.io.encoding.DataBlockEncoding;<a name="line.59"></a>
-<span class="sourceLineNo">060</span>import org.apache.hadoop.hbase.mapreduce.TableMapReduceUtil;<a name="line.60"></a>
-<span class="sourceLineNo">061</span>import org.apache.hadoop.hbase.regionserver.BloomType;<a name="line.61"></a>
-<span class="sourceLineNo">062</span>import org.apache.hadoop.hbase.test.util.CRC64;<a name="line.62"></a>
-<span class="sourceLineNo">063</span>import org.apache.hadoop.hbase.test.util.warc.WARCInputFormat;<a name="line.63"></a>
-<span class="sourceLineNo">064</span>import org.apache.hadoop.hbase.test.util.warc.WARCRecord;<a name="line.64"></a>
-<span class="sourceLineNo">065</span>import org.apache.hadoop.hbase.test.util.warc.WARCWritable;<a name="line.65"></a>
-<span class="sourceLineNo">066</span>import org.apache.hadoop.hbase.util.Bytes;<a name="line.66"></a>
-<span class="sourceLineNo">067</span>import org.apache.hadoop.hbase.util.EnvironmentEdgeManager;<a name="line.67"></a>
-<span class="sourceLineNo">068</span>import org.apache.hadoop.hbase.util.RegionSplitter;<a name="line.68"></a>
-<span class="sourceLineNo">069</span>import org.apache.hadoop.io.BytesWritable;<a name="line.69"></a>
-<span class="sourceLineNo">070</span>import org.apache.hadoop.io.LongWritable;<a name="line.70"></a>
-<span class="sourceLineNo">071</span>import org.apache.hadoop.io.NullWritable;<a name="line.71"></a>
-<span class="sourceLineNo">072</span>import org.apache.hadoop.io.SequenceFile.CompressionType;<a name="line.72"></a>
-<span class="sourceLineNo">073</span>import org.apache.hadoop.io.Writable;<a name="line.73"></a>
-<span class="sourceLineNo">074</span>import org.apache.hadoop.mapreduce.Counters;<a name="line.74"></a>
-<span class="sourceLineNo">075</span>import org.apache.hadoop.mapreduce.Job;<a name="line.75"></a>
-<span class="sourceLineNo">076</span>import org.apache.hadoop.mapreduce.JobContext;<a name="line.76"></a>
-<span class="sourceLineNo">077</span>import org.apache.hadoop.mapreduce.Mapper;<a name="line.77"></a>
-<span class="sourceLineNo">078</span>import org.apache.hadoop.mapreduce.lib.input.FileInputFormat;<a name="line.78"></a>
-<span class="sourceLineNo">079</span>import org.apache.hadoop.mapreduce.lib.input.SequenceFileInputFormat;<a name="line.79"></a>
-<span class="sourceLineNo">080</span>import org.apache.hadoop.mapreduce.lib.output.NullOutputFormat;<a name="line.80"></a>
-<span class="sourceLineNo">081</span>import org.apache.hadoop.mapreduce.lib.output.SequenceFileOutputFormat;<a name="line.81"></a>
-<span class="sourceLineNo">082</span>import org.apache.hadoop.util.Tool;<a name="line.82"></a>
-<span class="sourceLineNo">083</span>import org.apache.hadoop.util.ToolRunner;<a name="line.83"></a>
-<span class="sourceLineNo">084</span>import org.slf4j.Logger;<a name="line.84"></a>
-<span class="sourceLineNo">085</span>import org.slf4j.LoggerFactory;<a name="line.85"></a>
-<span class="sourceLineNo">086</span><a name="line.86"></a>
-<span class="sourceLineNo">087</span>import org.apache.hbase.thirdparty.org.apache.commons.cli.CommandLine;<a name="line.87"></a>
-<span class="sourceLineNo">088</span><a name="line.88"></a>
-<span class="sourceLineNo">089</span>/**<a name="line.89"></a>
-<span class="sourceLineNo">090</span> * This integration test loads successful resource retrieval records from the Common Crawl<a name="line.90"></a>
-<span class="sourceLineNo">091</span> * (https://commoncrawl.org/) public dataset into an HBase table and writes records that can be<a name="line.91"></a>
-<span class="sourceLineNo">092</span> * used to later verify the presence and integrity of those records.<a name="line.92"></a>
-<span class="sourceLineNo">093</span> * <p><a name="line.93"></a>
-<span class="sourceLineNo">094</span> * Run like:<a name="line.94"></a>
-<span class="sourceLineNo">095</span> * <blockquote><a name="line.95"></a>
-<span class="sourceLineNo">096</span> * ./bin/hbase org.apache.hadoop.hbase.test.IntegrationTestLoadCommonCrawl \<br><a name="line.96"></a>
-<span class="sourceLineNo">097</span> * &nbsp;&nbsp; -Dfs.s3n.awsAccessKeyId=&lt;AWS access key&gt; \<br><a name="line.97"></a>
-<span class="sourceLineNo">098</span> * &nbsp;&nbsp; -Dfs.s3n.awsSecretAccessKey=&lt;AWS secret key&gt; \<br><a name="line.98"></a>
-<span class="sourceLineNo">099</span> * &nbsp;&nbsp; /path/to/test-CC-MAIN-2021-10-warc.paths.gz \<br><a name="line.99"></a>
-<span class="sourceLineNo">100</span> * &nbsp;&nbsp; /path/to/tmp/warc-loader-output<a name="line.100"></a>
-<span class="sourceLineNo">101</span> * </blockquote><a name="line.101"></a>
-<span class="sourceLineNo">102</span> * <p><a name="line.102"></a>
-<span class="sourceLineNo">103</span> * Access to the Common Crawl dataset in S3 is made available to anyone by Amazon AWS, but<a name="line.103"></a>
-<span class="sourceLineNo">104</span> * Hadoop's S3N filesystem still requires valid access credentials to initialize.<a name="line.104"></a>
-<span class="sourceLineNo">105</span> * <p><a name="line.105"></a>
-<span class="sourceLineNo">106</span> * The input path can either specify a directory or a file. The file may optionally be<a name="line.106"></a>
-<span class="sourceLineNo">107</span> * compressed with gzip. If a directory, the loader expects the directory to contain one or more<a name="line.107"></a>
-<span class="sourceLineNo">108</span> * WARC files from the Common Crawl dataset. If a file, the loader expects a list of Hadoop S3N<a name="line.108"></a>
-<span class="sourceLineNo">109</span> * URIs which point to S3 locations for one or more WARC files from the Common Crawl dataset,<a name="line.109"></a>
-<span class="sourceLineNo">110</span> * one URI per line. Lines should be terminated with the UNIX line terminator.<a name="line.110"></a>
-<span class="sourceLineNo">111</span> * <p><a name="line.111"></a>
-<span class="sourceLineNo">112</span> * Included in hbase-it/src/test/resources/CC-MAIN-2021-10-warc.paths.gz is a list of all WARC<a name="line.112"></a>
-<span class="sourceLineNo">113</span> * files comprising the Q1 2021 crawl archive. There are 64,000 WARC files in this data set, each<a name="line.113"></a>
-<span class="sourceLineNo">114</span> * containing ~1GB of gzipped data. The WARC files contain several record types, such as metadata,<a name="line.114"></a>
-<span class="sourceLineNo">115</span> * request, and response, but we only load the response record types. If the HBase table schema<a name="line.115"></a>
-<span class="sourceLineNo">116</span> * does not specify compression (by default) there is roughly a 10x expansion. Loading the full<a name="line.116"></a>
-<span class="sourceLineNo">117</span> * crawl archive results in a table approximately 640 TB in size.<a name="line.117"></a>
-<span class="sourceLineNo">118</span> * <p><a name="line.118"></a>
-<span class="sourceLineNo">119</span> * You can also split the Loader and Verify stages:<a name="line.119"></a>
-<span class="sourceLineNo">120</span> * <p><a name="line.120"></a>
-<span class="sourceLineNo">121</span> * Load with:<a name="line.121"></a>
-<span class="sourceLineNo">122</span> * <blockquote><a name="line.122"></a>
-<span class="sourceLineNo">123</span> * ./bin/hbase 'org.apache.hadoop.hbase.test.IntegrationTestLoadCommonCrawl$Loader' \<br><a name="line.123"></a>
-<span class="sourceLineNo">124</span> * &nbsp;&nbsp; -files /path/to/hadoop-aws.jar \<br><a name="line.124"></a>
-<span class="sourceLineNo">125</span> * &nbsp;&nbsp; -Dfs.s3n.awsAccessKeyId=&lt;AWS access key&gt; \<br><a name="line.125"></a>
-<span class="sourceLineNo">126</span> * &nbsp;&nbsp; -Dfs.s3n.awsSecretAccessKey=&lt;AWS secret key&gt; \<br><a name="line.126"></a>
-<span class="sourceLineNo">127</span> * &nbsp;&nbsp; /path/to/test-CC-MAIN-2021-10-warc.paths.gz \<br><a name="line.127"></a>
-<span class="sourceLineNo">128</span> * &nbsp;&nbsp; /path/to/tmp/warc-loader-output<a name="line.128"></a>
-<span class="sourceLineNo">129</span> * </blockquote><a name="line.129"></a>
-<span class="sourceLineNo">130</span> * <p><a name="line.130"></a>
-<span class="sourceLineNo">131</span> * Note: The hadoop-aws jar will be needed at runtime to instantiate the S3N filesystem. Use<a name="line.131"></a>
-<span class="sourceLineNo">132</span> * the <tt>-files</tt> ToolRunner argument to add it.<a name="line.132"></a>
-<span class="sourceLineNo">133</span> * <p><a name="line.133"></a>
-<span class="sourceLineNo">134</span> * Verify with:<a name="line.134"></a>
-<span class="sourceLineNo">135</span> * <blockquote><a name="line.135"></a>
-<span class="sourceLineNo">136</span> * ./bin/hbase 'org.apache.hadoop.hbase.test.IntegrationTestLoadCommonCrawl$Verify' \<br><a name="line.136"></a>
-<span class="sourceLineNo">137</span> * &nbsp;&nbsp; /path/to/tmp/warc-loader-output<a name="line.137"></a>
-<span class="sourceLineNo">138</span> * </blockquote><a name="line.138"></a>
-<span class="sourceLineNo">139</span> * <p><a name="line.139"></a>
-<span class="sourceLineNo">140</span> */<a name="line.140"></a>
-<span class="sourceLineNo">141</span>public class IntegrationTestLoadCommonCrawl extends IntegrationTestBase {<a name="line.141"></a>
-<span class="sourceLineNo">142</span><a name="line.142"></a>
-<span class="sourceLineNo">143</span> private static final Logger LOG = LoggerFactory.getLogger(IntegrationTestLoadCommonCrawl.class);<a name="line.143"></a>
-<span class="sourceLineNo">144</span><a name="line.144"></a>
-<span class="sourceLineNo">145</span> protected static String TABLE_NAME_KEY = "IntegrationTestLoadCommonCrawl.table";<a name="line.145"></a>
-<span class="sourceLineNo">146</span> protected static String DEFAULT_TABLE_NAME = "IntegrationTestLoadCommonCrawl";<a name="line.146"></a>
-<span class="sourceLineNo">147</span><a name="line.147"></a>
-<span class="sourceLineNo">148</span> protected static byte[] CONTENT_FAMILY_NAME = Bytes.toBytes("c");<a name="line.148"></a>
-<span class="sourceLineNo">149</span> protected static byte[] INFO_FAMILY_NAME = Bytes.toBytes("i");<a name="line.149"></a>
-<span class="sourceLineNo">150</span> protected static byte[] CONTENT_QUALIFIER = HConstants.EMPTY_BYTE_ARRAY;<a name="line.150"></a>
-<span class="sourceLineNo">151</span> protected static byte[] CONTENT_LENGTH_QUALIFIER = Bytes.toBytes("l");<a name="line.151"></a>
-<span class="sourceLineNo">152</span> protected static byte[] CONTENT_TYPE_QUALIFIER = Bytes.toBytes("t");<a name="line.152"></a>
-<span class="sourceLineNo">153</span> protected static byte[] CRC_QUALIFIER = Bytes.toBytes("c");<a name="line.153"></a>
-<span class="sourceLineNo">154</span> protected static byte[] DATE_QUALIFIER = Bytes.toBytes("d");<a name="line.154"></a>
-<span class="sourceLineNo">155</span> protected static byte[] IP_ADDRESS_QUALIFIER = Bytes.toBytes("a");<a name="line.155"></a>
-<span class="sourceLineNo">156</span> protected static byte[] RECORD_ID_QUALIFIER = Bytes.toBytes("r");<a name="line.156"></a>
-<span class="sourceLineNo">157</span> protected static byte[] TARGET_URI_QUALIFIER = Bytes.toBytes("u");<a name="line.157"></a>
-<span class="sourceLineNo">158</span><a name="line.158"></a>
-<span class="sourceLineNo">159</span> public static enum Counts {<a name="line.159"></a>
-<span class="sourceLineNo">160</span> REFERENCED, UNREFERENCED, CORRUPT<a name="line.160"></a>
-<span class="sourceLineNo">161</span> }<a name="line.161"></a>
-<span class="sourceLineNo">162</span><a name="line.162"></a>
-<span class="sourceLineNo">163</span> Path warcFileInputDir = null;<a name="line.163"></a>
-<span class="sourceLineNo">164</span> Path outputDir = null;<a name="line.164"></a>
-<span class="sourceLineNo">165</span> String[] args;<a name="line.165"></a>
-<span class="sourceLineNo">166</span><a name="line.166"></a>
-<span class="sourceLineNo">167</span> protected int runLoader(Path warcFileInputDir, Path outputDir) throws Exception {<a name="line.167"></a>
-<span class="sourceLineNo">168</span> Loader loader = new Loader();<a name="line.168"></a>
-<span class="sourceLineNo">169</span> loader.setConf(conf);<a name="line.169"></a>
-<span class="sourceLineNo">170</span> return loader.run(warcFileInputDir, outputDir);<a name="line.170"></a>
-<span class="sourceLineNo">171</span> }<a name="line.171"></a>
-<span class="sourceLineNo">172</span><a name="line.172"></a>
-<span class="sourceLineNo">173</span> protected int runVerify(Path inputDir) throws Exception {<a name="line.173"></a>
-<span class="sourceLineNo">174</span> Verify verify = new Verify();<a name="line.174"></a>
-<span class="sourceLineNo">175</span> verify.setConf(conf);<a name="line.175"></a>
-<span class="sourceLineNo">176</span> return verify.run(inputDir);<a name="line.176"></a>
-<span class="sourceLineNo">177</span> }<a name="line.177"></a>
-<span class="sourceLineNo">178</span><a name="line.178"></a>
-<span class="sourceLineNo">179</span> @Override<a name="line.179"></a>
-<span class="sourceLineNo">180</span> public int run(String[] args) {<a name="line.180"></a>
-<span class="sourceLineNo">181</span> if (args.length > 0) {<a name="line.181"></a>
-<span class="sourceLineNo">182</span> warcFileInputDir = new Path(args[0]);<a name="line.182"></a>
-<span class="sourceLineNo">183</span> if (args.length > 1) {<a name="line.183"></a>
-<span class="sourceLineNo">184</span> outputDir = new Path(args[1]);<a name="line.184"></a>
-<span class="sourceLineNo">185</span> }<a name="line.185"></a>
-<span class="sourceLineNo">186</span> }<a name="line.186"></a>
-<span class="sourceLineNo">187</span> try {<a name="line.187"></a>
-<span class="sourceLineNo">188</span> if (warcFileInputDir == null) {<a name="line.188"></a>
-<span class="sourceLineNo">189</span> throw new IllegalArgumentException("WARC input file or directory not specified");<a name="line.189"></a>
-<span class="sourceLineNo">190</span> }<a name="line.190"></a>
-<span class="sourceLineNo">191</span> if (outputDir == null) {<a name="line.191"></a>
-<span class="sourceLineNo">192</span> throw new IllegalArgumentException("Output directory not specified");<a name="line.192"></a>
+<span class="sourceLineNo">049</span>import org.apache.hadoop.hbase.client.BufferedMutator;<a name="line.49"></a>
+<span class="sourceLineNo">050</span>import org.apache.hadoop.hbase.client.ColumnFamilyDescriptor;<a name="line.50"></a>
+<span class="sourceLineNo">051</span>import org.apache.hadoop.hbase.client.ColumnFamilyDescriptorBuilder;<a name="line.51"></a>
+<span class="sourceLineNo">052</span>import org.apache.hadoop.hbase.client.Connection;<a name="line.52"></a>
+<span class="sourceLineNo">053</span>import org.apache.hadoop.hbase.client.ConnectionFactory;<a name="line.53"></a>
+<span class="sourceLineNo">054</span>import org.apache.hadoop.hbase.client.Get;<a name="line.54"></a>
+<span class="sourceLineNo">055</span>import org.apache.hadoop.hbase.client.Put;<a name="line.55"></a>
+<span class="sourceLineNo">056</span>import org.apache.hadoop.hbase.client.Result;<a name="line.56"></a>
+<span class="sourceLineNo">057</span>import org.apache.hadoop.hbase.client.Table;<a name="line.57"></a>
+<span class="sourceLineNo">058</span>import org.apache.hadoop.hbase.client.TableDescriptor;<a name="line.58"></a>
+<span class="sourceLineNo">059</span>import org.apache.hadoop.hbase.client.TableDescriptorBuilder;<a name="line.59"></a>
+<span class="sourceLineNo">060</span>import org.apache.hadoop.hbase.io.encoding.DataBlockEncoding;<a name="line.60"></a>
+<span class="sourceLineNo">061</span>import org.apache.hadoop.hbase.mapreduce.TableMapReduceUtil;<a name="line.61"></a>
+<span class="sourceLineNo">062</span>import org.apache.hadoop.hbase.regionserver.BloomType;<a name="line.62"></a>
+<span class="sourceLineNo">063</span>import org.apache.hadoop.hbase.test.util.CRC64;<a name="line.63"></a>
+<span class="sourceLineNo">064</span>import org.apache.hadoop.hbase.test.util.warc.WARCInputFormat;<a name="line.64"></a>
+<span class="sourceLineNo">065</span>import org.apache.hadoop.hbase.test.util.warc.WARCRecord;<a name="line.65"></a>
+<span class="sourceLineNo">066</span>import org.apache.hadoop.hbase.test.util.warc.WARCWritable;<a name="line.66"></a>
+<span class="sourceLineNo">067</span>import org.apache.hadoop.hbase.util.Bytes;<a name="line.67"></a>
+<span class="sourceLineNo">068</span>import org.apache.hadoop.hbase.util.EnvironmentEdgeManager;<a name="line.68"></a>
+<span class="sourceLineNo">069</span>import org.apache.hadoop.hbase.util.RegionSplitter;<a name="line.69"></a>
+<span class="sourceLineNo">070</span>import org.apache.hadoop.io.BytesWritable;<a name="line.70"></a>
+<span class="sourceLineNo">071</span>import org.apache.hadoop.io.LongWritable;<a name="line.71"></a>
+<span class="sourceLineNo">072</span>import org.apache.hadoop.io.NullWritable;<a name="line.72"></a>
+<span class="sourceLineNo">073</span>import org.apache.hadoop.io.SequenceFile.CompressionType;<a name="line.73"></a>
+<span class="sourceLineNo">074</span>import org.apache.hadoop.io.Writable;<a name="line.74"></a>
+<span class="sourceLineNo">075</span>import org.apache.hadoop.mapreduce.Counters;<a name="line.75"></a>
+<span class="sourceLineNo">076</span>import org.apache.hadoop.mapreduce.Job;<a name="line.76"></a>
+<span class="sourceLineNo">077</span>import org.apache.hadoop.mapreduce.JobContext;<a name="line.77"></a>
+<span class="sourceLineNo">078</span>import org.apache.hadoop.mapreduce.Mapper;<a name="line.78"></a>
+<span class="sourceLineNo">079</span>import org.apache.hadoop.mapreduce.lib.input.FileInputFormat;<a name="line.79"></a>
+<span class="sourceLineNo">080</span>import org.apache.hadoop.mapreduce.lib.input.SequenceFileInputFormat;<a name="line.80"></a>
+<span class="sourceLineNo">081</span>import org.apache.hadoop.mapreduce.lib.output.NullOutputFormat;<a name="line.81"></a>
+<span class="sourceLineNo">082</span>import org.apache.hadoop.mapreduce.lib.output.SequenceFileOutputFormat;<a name="line.82"></a>
+<span class="sourceLineNo">083</span>import org.apache.hadoop.util.Tool;<a name="line.83"></a>
+<span class="sourceLineNo">084</span>import org.apache.hadoop.util.ToolRunner;<a name="line.84"></a>
+<span class="sourceLineNo">085</span>import org.slf4j.Logger;<a name="line.85"></a>
+<span class="sourceLineNo">086</span>import org.slf4j.LoggerFactory;<a name="line.86"></a>
+<span class="sourceLineNo">087</span><a name="line.87"></a>
+<span class="sourceLineNo">088</span>import org.apache.hbase.thirdparty.org.apache.commons.cli.CommandLine;<a name="line.88"></a>
+<span class="sourceLineNo">089</span><a name="line.89"></a>
+<span class="sourceLineNo">090</span>/**<a name="line.90"></a>
+<span class="sourceLineNo">091</span> * This integration test loads successful resource retrieval records from the Common Crawl<a name="line.91"></a>
+<span class="sourceLineNo">092</span> * (https://commoncrawl.org/) public dataset into an HBase table and writes records that can be<a name="line.92"></a>
+<span class="sourceLineNo">093</span> * used to later verify the presence and integrity of those records.<a name="line.93"></a>
+<span class="sourceLineNo">094</span> * <p><a name="line.94"></a>
+<span class="sourceLineNo">095</span> * Run like:<a name="line.95"></a>
+<span class="sourceLineNo">096</span> * <blockquote><a name="line.96"></a>
+<span class="sourceLineNo">097</span> * ./bin/hbase org.apache.hadoop.hbase.test.IntegrationTestLoadCommonCrawl \<br><a name="line.97"></a>
+<span class="sourceLineNo">098</span> * &nbsp;&nbsp; -Dfs.s3n.awsAccessKeyId=&lt;AWS access key&gt; \<br><a name="line.98"></a>
+<span class="sourceLineNo">099</span> * &nbsp;&nbsp; -Dfs.s3n.awsSecretAccessKey=&lt;AWS secret key&gt; \<br><a name="line.99"></a>
+<span class="sourceLineNo">100</span> * &nbsp;&nbsp; /path/to/test-CC-MAIN-2021-10-warc.paths.gz \<br><a name="line.100"></a>
+<span class="sourceLineNo">101</span> * &nbsp;&nbsp; /path/to/tmp/warc-loader-output<a name="line.101"></a>
+<span class="sourceLineNo">102</span> * </blockquote><a name="line.102"></a>
+<span class="sourceLineNo">103</span> * <p><a name="line.103"></a>
+<span class="sourceLineNo">104</span> * Access to the Common Crawl dataset in S3 is made available to anyone by Amazon AWS, but<a name="line.104"></a>
+<span class="sourceLineNo">105</span> * Hadoop's S3N filesystem still requires valid access credentials to initialize.<a name="line.105"></a>
+<span class="sourceLineNo">106</span> * <p><a name="line.106"></a>
+<span class="sourceLineNo">107</span> * The input path can either specify a directory or a file. The file may optionally be<a name="line.107"></a>
+<span class="sourceLineNo">108</span> * compressed with gzip. If a directory, the loader expects the directory to contain one or more<a name="line.108"></a>
+<span class="sourceLineNo">109</span> * WARC files from the Common Crawl dataset. If a file, the loader expects a list of Hadoop S3N<a name="line.109"></a>
+<span class="sourceLineNo">110</span> * URIs which point to S3 locations for one or more WARC files from the Common Crawl dataset,<a name="line.110"></a>
+<span class="sourceLineNo">111</span> * one URI per line. Lines should be terminated with the UNIX line terminator.<a name="line.111"></a>
+<span class="sourceLineNo">112</span> * <p><a name="line.112"></a>
+<span class="sourceLineNo">113</span> * Included in hbase-it/src/test/resources/CC-MAIN-2021-10-warc.paths.gz is a list of all WARC<a name="line.113"></a>
+<span class="sourceLineNo">114</span> * files comprising the Q1 2021 crawl archive. There are 64,000 WARC files in this data set, each<a name="line.114"></a>
+<span class="sourceLineNo">115</span> * containing ~1GB of gzipped data. The WARC files contain several record types, such as metadata,<a name="line.115"></a>
+<span class="sourceLineNo">116</span> * request, and response, but we only load the response record types. If the HBase table schema<a name="line.116"></a>
+<span class="sourceLineNo">117</span> * does not specify compression (by default) there is roughly a 10x expansion. Loading the full<a name="line.117"></a>
+<span class="sourceLineNo">118</span> * crawl archive results in a table approximately 640 TB in size.<a name="line.118"></a>
+<span class="sourceLineNo">119</span> * <p><a name="line.119"></a>
+<span class="sourceLineNo">120</span> * You can also split the Loader and Verify stages:<a name="line.120"></a>
+<span class="sourceLineNo">121</span> * <p><a name="line.121"></a>
+<span class="sourceLineNo">122</span> * Load with:<a name="line.122"></a>
+<span class="sourceLineNo">123</span> * <blockquote><a name="line.123"></a>
+<span class="sourceLineNo">124</span> * ./bin/hbase 'org.apache.hadoop.hbase.test.IntegrationTestLoadCommonCrawl$Loader' \<br><a name="line.124"></a>
+<span class="sourceLineNo">125</span> * &nbsp;&nbsp; -files /path/to/hadoop-aws.jar \<br><a name="line.125"></a>
+<span class="sourceLineNo">126</span> * &nbsp;&nbsp; -Dfs.s3n.awsAccessKeyId=&lt;AWS access key&gt; \<br><a name="line.126"></a>
+<span class="sourceLineNo">127</span> * &nbsp;&nbsp; -Dfs.s3n.awsSecretAccessKey=&lt;AWS secret key&gt; \<br><a name="line.127"></a>
+<span class="sourceLineNo">128</span> * &nbsp;&nbsp; /path/to/test-CC-MAIN-2021-10-warc.paths.gz \<br><a name="line.128"></a>
+<span class="sourceLineNo">129</span> * &nbsp;&nbsp; /path/to/tmp/warc-loader-output<a name="line.129"></a>
+<span class="sourceLineNo">130</span> * </blockquote><a name="line.130"></a>
+<span class="sourceLineNo">131</span> * <p><a name="line.131"></a>
+<span class="sourceLineNo">132</span> * Note: The hadoop-aws jar will be needed at runtime to instantiate the S3N filesystem. Use<a name="line.132"></a>
+<span class="sourceLineNo">133</span> * the <tt>-files</tt> ToolRunner argument to add it.<a name="line.133"></a>
+<span class="sourceLineNo">134</span> * <p><a name="line.134"></a>
+<span class="sourceLineNo">135</span> * Verify with:<a name="line.135"></a>
+<span class="sourceLineNo">136</span> * <blockquote><a name="line.136"></a>
+<span class="sourceLineNo">137</span> * ./bin/hbase 'org.apache.hadoop.hbase.test.IntegrationTestLoadCommonCrawl$Verify' \<br><a name="line.137"></a>
+<span class="sourceLineNo">138</span> * &nbsp;&nbsp; /path/to/tmp/warc-loader-output<a name="line.138"></a>
+<span class="sourceLineNo">139</span> * </blockquote><a name="line.139"></a>
+<span class="sourceLineNo">140</span> * <p><a name="line.140"></a>
+<span class="sourceLineNo">141</span> */<a name="line.141"></a>
+<span class="sourceLineNo">142</span>public class IntegrationTestLoadCommonCrawl extends IntegrationTestBase {<a name="line.142"></a>
+<span class="sourceLineNo">143</span><a name="line.143"></a>
+<span class="sourceLineNo">144</span> private static final Logger LOG = LoggerFactory.getLogger(IntegrationTestLoadCommonCrawl.class);<a name="line.144"></a>
+<span class="sourceLineNo">145</span><a name="line.145"></a>
+<span class="sourceLineNo">146</span> protected static String TABLE_NAME_KEY = "IntegrationTestLoadCommonCrawl.table";<a name="line.146"></a>
+<span class="sourceLineNo">147</span> protected static String DEFAULT_TABLE_NAME = "IntegrationTestLoadCommonCrawl";<a name="line.147"></a>
+<span class="sourceLineNo">148</span><a name="line.148"></a>
+<span class="sourceLineNo">149</span> protected static byte[] CONTENT_FAMILY_NAME = Bytes.toBytes("c");<a name="line.149"></a>
+<span class="sourceLineNo">150</span> protected static byte[] INFO_FAMILY_NAME = Bytes.toBytes("i");<a name="line.150"></a>
+<span class="sourceLineNo">151</span> protected static byte[] CONTENT_QUALIFIER = HConstants.EMPTY_BYTE_ARRAY;<a name="line.151"></a>
+<span class="sourceLineNo">152</span> protected static byte[] CONTENT_LENGTH_QUALIFIER = Bytes.toBytes("l");<a name="line.152"></a>
+<span class="sourceLineNo">153</span> protected static byte[] CONTENT_TYPE_QUALIFIER = Bytes.toBytes("t");<a name="line.153"></a>
+<span class="sourceLineNo">154</span> protected static byte[] CRC_QUALIFIER = Bytes.toBytes("c");<a name="line.154"></a>
+<span class="sourceLineNo">155</span> protected static byte[] DATE_QUALIFIER = Bytes.toBytes("d");<a name="line.155"></a>
+<span class="sourceLineNo">156</span> protected static byte[] IP_ADDRESS_QUALIFIER = Bytes.toBytes("a");<a name="line.156"></a>
+<span class="sourceLineNo">157</span> protected static byte[] RECORD_ID_QUALIFIER = Bytes.toBytes("r");<a name="line.157"></a>
+<span class="sourceLineNo">158</span> protected static byte[] TARGET_URI_QUALIFIER = Bytes.toBytes("u");<a name="line.158"></a>
+<span class="sourceLineNo">159</span><a name="line.159"></a>
+<span class="sourceLineNo">160</span> private static final int VERIFICATION_READ_RETRIES = 10;<a name="line.160"></a>
+<span class="sourceLineNo">161</span><a name="line.161"></a>
+<span class="sourceLineNo">162</span> public static enum Counts {<a name="line.162"></a>
+<span class="sourceLineNo">163</span> REFERENCED, UNREFERENCED, CORRUPT<a name="line.163"></a>
+<span class="sourceLineNo">164</span> }<a name="line.164"></a>
+<span class="sourceLineNo">165</span><a name="line.165"></a>
+<span class="sourceLineNo">166</span> Path warcFileInputDir = null;<a name="line.166"></a>
+<span class="sourceLineNo">167</span> Path outputDir = null;<a name="line.167"></a>
+<span class="sourceLineNo">168</span> String[] args;<a name="line.168"></a>
+<span class="sourceLineNo">169</span><a name="line.169"></a>
+<span class="sourceLineNo">170</span> protected int runLoader(Path warcFileInputDir, Path outputDir) throws Exception {<a name="line.170"></a>
+<span class="sourceLineNo">171</span> Loader loader = new Loader();<a name="line.171"></a>
+<span class="sourceLineNo">172</span> loader.setConf(conf);<a name="line.172"></a>
+<span class="sourceLineNo">173</span> return loader.run(warcFileInputDir, outputDir);<a name="line.173"></a>
+<span class="sourceLineNo">174</span> }<a name="line.174"></a>
+<span class="sourceLineNo">175</span><a name="line.175"></a>
+<span class="sourceLineNo">176</span> protected int runVerify(Path inputDir) throws Exception {<a name="line.176"></a>
+<span class="sourceLineNo">177</span> Verify verify = new Verify();<a name="line.177"></a>
+<span class="sourceLineNo">178</span> verify.setConf(conf);<a name="line.178"></a>
+<span class="sourceLineNo">179</span> return verify.run(inputDir);<a name="line.179"></a>
+<span class="sourceLineNo">180</span> }<a name="line.180"></a>
+<span class="sourceLineNo">181</span><a name="line.181"></a>
+<span class="sourceLineNo">182</span> @Override<a name="line.182"></a>
+<span class="sourceLineNo">183</span> public int run(String[] args) {<a name="line.183"></a>
+<span class="sourceLineNo">184</span> if (args.length > 0) {<a name="line.184"></a>
+<span class="sourceLineNo">185</span> warcFileInputDir = new Path(args[0]);<a name="line.185"></a>
+<span class="sourceLineNo">186</span> if (args.length > 1) {<a name="line.186"></a>
+<span class="sourceLineNo">187</span> outputDir = new Path(args[1]);<a name="line.187"></a>
+<span class="sourceLineNo">188</span> }<a name="line.188"></a>
+<span class="sourceLineNo">189</span> }<a name="line.189"></a>
+<span class="sourceLineNo">190</span> try {<a name="line.190"></a>
+<span class="sourceLineNo">191</span> if (warcFileInputDir == null) {<a name="line.191"></a>
+<span class="sourceLineNo">192</span> throw new IllegalArgumentException("WARC input file or directory not specified");<a name="line.192"></a>
<span class="sourceLineNo">193</span> }<a name="line.193"></a>
-<span class="sourceLineNo">194</span> int res = runLoader(warcFileInputDir, outputDir);<a name="line.194"></a>
-<span class="sourceLineNo">195</span> if (res != 0) {<a name="line.195"></a>
-<span class="sourceLineNo">196</span> LOG.error("Loader failed");<a name="line.196"></a>
-<span class="sourceLineNo">197</span> return -1;<a name="line.197"></a>
-<span class="sourceLineNo">198</span> }<a name="line.198"></a>
-<span class="sourceLineNo">199</span> res = runVerify(outputDir);<a name="line.199"></a>
-<span class="sourceLineNo">200</span> } catch (Exception e) {<a name="line.200"></a>
-<span class="sourceLineNo">201</span> LOG.error("Tool failed with exception", e);<a name="line.201"></a>
-<span class="sourceLineNo">202</span> return -1;<a name="line.202"></a>
-<span class="sourceLineNo">203</span> }<a name="line.203"></a>
-<span class="sourceLineNo">204</span> return 0;<a name="line.204"></a>
-<span class="sourceLineNo">205</span> }<a name="line.205"></a>
-<span class="sourceLineNo">206</span><a name="line.206"></a>
-<span class="sourceLineNo">207</span> @Override<a name="line.207"></a>
-<span class="sourceLineNo">208</span> protected void processOptions(CommandLine cmd) {<a name="line.208"></a>
-<span class="sourceLineNo">209</span> processBaseOptions(cmd);<a name="line.209"></a>
-<span class="sourceLineNo">210</span> args = cmd.getArgs();<a name="line.210"></a>
-<span class="sourceLineNo">211</span> }<a name="line.211"></a>
-<span class="sourceLineNo">212</span><a name="line.212"></a>
-<span class="sourceLineNo">213</span> @Override<a name="line.213"></a>
-<span class="sourceLineNo">214</span> public void setUpCluster() throws Exception {<a name="line.214"></a>
-<span class="sourceLineNo">215</span> util = getTestingUtil(getConf());<a name="line.215"></a>
-<span class="sourceLineNo">216</span> boolean isDistributed = util.isDistributedCluster();<a name="line.216"></a>
-<span class="sourceLineNo">217</span> util.initializeCluster(isDistributed ? 1 : 3);<a name="line.217"></a>
-<span class="sourceLineNo">218</span> if (!isDistributed) {<a name="line.218"></a>
-<span class="sourceLineNo">219</span> util.startMiniMapReduceCluster();<a name="line.219"></a>
-<span class="sourceLineNo">220</span> }<a name="line.220"></a>
-<span class="sourceLineNo">221</span> this.setConf(util.getConfiguration());<a name="line.221"></a>
-<span class="sourceLineNo">222</span> }<a name="line.222"></a>
-<span class="sourceLineNo">223</span><a name="line.223"></a>
-<span class="sourceLineNo">224</span> @Override<a name="line.224"></a>
-<span class="sourceLineNo">225</span> public void cleanUpCluster() throws Exception {<a name="line.225"></a>
-<span class="sourceLineNo">226</span> super.cleanUpCluster();<a name="line.226"></a>
-<span class="sourceLineNo">227</span> if (util.isDistributedCluster()) {<a name="line.227"></a>
-<span class="sourceLineNo">228</span> util.shutdownMiniMapReduceCluster();<a name="line.228"></a>
-<span class="sourceLineNo">229</span> }<a name="line.229"></a>
-<span class="sourceLineNo">230</span> }<a name="line.230"></a>
-<span class="sourceLineNo">231</span><a name="line.231"></a>
-<span class="sourceLineNo">232</span> static TableName getTablename(Configuration c) {<a name="line.232"></a>
-<span class="sourceLineNo">233</span> return TableName.valueOf(c.get(TABLE_NAME_KEY, DEFAULT_TABLE_NAME));<a name="line.233"></a>
-<span class="sourceLineNo">234</span> }<a name="line.234"></a>
-<span class="sourceLineNo">235</span><a name="line.235"></a>
-<span class="sourceLineNo">236</span> @Override<a name="line.236"></a>
-<span class="sourceLineNo">237</span> public TableName getTablename() {<a name="line.237"></a>
-<span class="sourceLineNo">238</span> return getTablename(getConf());<a name="line.238"></a>
-<span class="sourceLineNo">239</span> }<a name="line.239"></a>
-<span class="sourceLineNo">240</span><a name="line.240"></a>
-<span class="sourceLineNo">241</span> @Override<a name="line.241"></a>
-<span class="sourceLineNo">242</span> protected Set<String> getColumnFamilies() {<a name="line.242"></a>
-<span class="sourceLineNo">243</span> Set<String> families = new HashSet<>();<a name="line.243"></a>
-<span class="sourceLineNo">244</span> families.add(Bytes.toString(CONTENT_FAMILY_NAME));<a name="line.244"></a>
-<span class="sourceLineNo">245</span> families.add(Bytes.toString(INFO_FAMILY_NAME));<a name="line.245"></a>
-<span class="sourceLineNo">246</span> return families;<a name="line.246"></a>
-<span class="sourceLineNo">247</span> }<a name="line.247"></a>
-<span class="sourceLineNo">248</span><a name="line.248"></a>
-<span class="sourceLineNo">249</span> @Override<a name="line.249"></a>
-<span class="sourceLineNo">250</span> public int runTestFromCommandLine() throws Exception {<a name="line.250"></a>
-<span class="sourceLineNo">251</span> return ToolRunner.run(getConf(), this, args);<a name="line.251"></a>
-<span class="sourceLineNo">252</span> }<a name="line.252"></a>
-<span class="sourceLineNo">253</span><a name="line.253"></a>
-<span class="sourceLineNo">254</span> public static void main(String[] args) throws Exception {<a name="line.254"></a>
-<span class="sourceLineNo">255</span> Configuration conf = HBaseConfiguration.create();<a name="line.255"></a>
-<span class="sourceLineNo">256</span> IntegrationTestingUtility.setUseDistributedCluster(conf);<a name="line.256"></a>
-<span class="sourceLineNo">257</span> int ret = ToolRunner.run(conf, new IntegrationTestLoadCommonCrawl(), args);<a name="line.257"></a>
-<span class="sourceLineNo">258</span> System.exit(ret);<a name="line.258"></a>
-<span class="sourceLineNo">259</span> }<a name="line.259"></a>
-<span class="sourceLineNo">260</span><a name="line.260"></a>
-<span class="sourceLineNo">261</span> public static class HBaseKeyWritable implements Writable {<a name="line.261"></a>
-<span class="sourceLineNo">262</span><a name="line.262"></a>
-<span class="sourceLineNo">263</span> private byte[] row;<a name="line.263"></a>
-<span class="sourceLineNo">264</span> private int rowOffset;<a name="line.264"></a>
-<span class="sourceLineNo">265</span> private int rowLength;<a name="line.265"></a>
-<span class="sourceLineNo">266</span> private byte[] family;<a name="line.266"></a>
-<span class="sourceLineNo">267</span> private int familyOffset;<a name="line.267"></a>
-<span class="sourceLineNo">268</span> private int familyLength;<a name="line.268"></a>
-<span class="sourceLineNo">269</span> private byte[] qualifier;<a name="line.269"></a>
-<span class="sourceLineNo">270</span> private int qualifierOffset;<a name="line.270"></a>
-<span class="sourceLineNo">271</span> private int qualifierLength;<a name="line.271"></a>
-<span class="sourceLineNo">272</span> private long ts;<a name="line.272"></a>
-<span class="sourceLineNo">273</span><a name="line.273"></a>
-<span class="sourceLineNo">274</span> public HBaseKeyWritable() { }<a name="line.274"></a>
-<span class="sourceLineNo">275</span><a name="line.275"></a>
-<span class="sourceLineNo">276</span> public HBaseKeyWritable(byte[] row, int rowOffset, int rowLength,<a name="line.276"></a>
-<span class="sourceLineNo">277</span> byte[] family, int familyOffset, int familyLength,<a name="line.277"></a>
-<span class="sourceLineNo">278</span> byte[] qualifier, int qualifierOffset, int qualifierLength, long ts) {<a name="line.278"></a>
-<span class="sourceLineNo">279</span> this.row = row;<a name="line.279"></a>
-<span class="sourceLineNo">280</span> this.rowOffset = rowOffset;<a name="line.280"></a>
-<span class="sourceLineNo">281</span> this.rowLength = rowLength;<a name="line.281"></a>
-<span class="sourceLineNo">282</span> this.family = family;<a name="line.282"></a>
-<span class="sourceLineNo">283</span> this.familyOffset = familyOffset;<a name="line.283"></a>
-<span class="sourceLineNo">284</span> this.familyLength = familyLength;<a name="line.284"></a>
-<span class="sourceLineNo">285</span> this.qualifier = qualifier;<a name="line.285"></a>
-<span class="sourceLineNo">286</span> this.qualifierOffset = qualifierOffset;<a name="line.286"></a>
-<span class="sourceLineNo">287</span> this.qualifierLength = qualifierLength;<a name="line.287"></a>
-<span class="sourceLineNo">288</span> this.ts = ts;<a name="line.288"></a>
-<span class="sourceLineNo">289</span> }<a name="line.289"></a>
-<span class="sourceLineNo">290</span><a name="line.290"></a>
-<span class="sourceLineNo">291</span> public HBaseKeyWritable(byte[] row, byte[] family, byte[] qualifier, long ts) {<a name="line.291"></a>
-<span class="sourceLineNo">292</span> this(row, 0, row.length,<a name="line.292"></a>
-<span class="sourceLineNo">293</span> family, 0, family.length,<a name="line.293"></a>
-<span class="sourceLineNo">294</span> qualifier, 0, qualifier != null ? qualifier.length : 0, ts);<a name="line.294"></a>
-<span class="sourceLineNo">295</span> }<a name="line.295"></a>
-<span class="sourceLineNo">296</span><a name="line.296"></a>
-<span class="sourceLineNo">297</span> public HBaseKeyWritable(byte[] row, byte[] family, long ts) {<a name="line.297"></a>
-<span class="sourceLineNo">298</span> this(row, family, HConstants.EMPTY_BYTE_ARRAY, ts);<a name="line.298"></a>
-<span class="sourceLineNo">299</span> }<a name="line.299"></a>
-<span class="sourceLineNo">300</span><a name="line.300"></a>
-<span class="sourceLineNo">301</span> public HBaseKeyWritable(Cell cell) {<a name="line.301"></a>
-<span class="sourceLineNo">302</span> this(cell.getRowArray(), cell.getRowOffset(), cell.getRowLength(),<a name="line.302"></a>
-<span class="sourceLineNo">303</span> cell.getFamilyArray(), cell.getFamilyOffset(), cell.getFamilyLength(),<a name="line.303"></a>
-<span class="sourceLineNo">304</span> cell.getQualifierArray(), cell.getQualifierOffset(), cell.getQualifierLength(),<a name="line.304"></a>
-<span class="sourceLineNo">305</span> cell.getTimestamp());<a name="line.305"></a>
-<span class="sourceLineNo">306</span> }<a name="line.306"></a>
-<span class="sourceLineNo">307</span><a name="line.307"></a>
-<span class="sourceLineNo">308</span> @Override<a name="line.308"></a>
-<span class="sourceLineNo">309</span> public void readFields(DataInput in) throws IOException {<a name="line.309"></a>
-<span class="sourceLineNo">310</span> this.row = Bytes.toBytes(in.readUTF());<a name="line.310"></a>
-<span class="sourceLineNo">311</span> this.rowOffset = 0;<a name="line.311"></a>
-<span class="sourceLineNo">312</span> this.rowLength = row.length;<a name="line.312"></a>
-<span class="sourceLineNo">313</span> this.family = Bytes.toBytes(in.readUTF());<a name="line.313"></a>
-<span class="sourceLineNo">314</span> this.familyOffset = 0;<a name="line.314"></a>
-<span class="sourceLineNo">315</span> this.familyLength = family.length;<a name="line.315"></a>
-<span class="sourceLineNo">316</span> this.qualifier = Bytes.toBytes(in.readUTF());<a name="line.316"></a>
-<span class="sourceLineNo">317</span> this.qualifierOffset = 0;<a name="line.317"></a>
-<span class="sourceLineNo">318</span> this.qualifierLength = qualifier.length;<a name="line.318"></a>
-<span class="sourceLineNo">319</span> this.ts = in.readLong();<a name="line.319"></a>
-<span class="sourceLineNo">320</span> }<a name="line.320"></a>
-<span class="sourceLineNo">321</span><a name="line.321"></a>
-<span class="sourceLineNo">322</span> @Override<a name="line.322"></a>
-<span class="sourceLineNo">323</span> public void write(DataOutput out) throws IOException {<a name="line.323"></a>
-<span class="sourceLineNo">324</span> out.writeUTF(new String(row, rowOffset, rowLength, StandardCharsets.UTF_8));<a name="line.324"></a>
-<span class="sourceLineNo">325</span> out.writeUTF(new String(family, familyOffset, familyLength, StandardCharsets.UTF_8));<a name="line.325"></a>
-<span class="sourceLineNo">326</span> if (qualifier != null) {<a name="line.326"></a>
-<span class="sourceLineNo">327</span> out.writeUTF(new String(qualifier, qualifierOffset, qualifierLength,<a name="line.327"></a>
-<span class="sourceLineNo">328</span> StandardCharsets.UTF_8));<a name="line.328"></a>
-<span class="sourceLineNo">329</span> } else {<a name="line.329"></a>
-<span class="sourceLineNo">330</span> out.writeUTF("");<a name="line.330"></a>
-<span class="sourceLineNo">331</span> }<a name="line.331"></a>
-<span class="sourceLineNo">332</span> out.writeLong(ts);<a name="line.332"></a>
-<span class="sourceLineNo">333</span> }<a name="line.333"></a>
-<span class="sourceLineNo">334</span><a name="line.334"></a>
-<span class="sourceLineNo">335</span> public byte[] getRowArray() {<a name="line.335"></a>
-<span class="sourceLineNo">336</span> return row;<a name="line.336"></a>
-<span class="sourceLineNo">337</span> }<a name="line.337"></a>
-<span class="sourceLineNo">338</span><a name="line.338"></a>
-<span class="sourceLineNo">339</span> public void setRow(byte[] row) {<a name="line.339"></a>
-<span class="sourceLineNo">340</span> this.row = row;<a name="line.340"></a>
-<span class="sourceLineNo">341</span> }<a name="line.341"></a>
-<span class="sourceLineNo">342</span><a name="line.342"></a>
-<span class="sourceLineNo">343</span> public int getRowOffset() {<a name="line.343"></a>
-<span class="sourceLineNo">344</span> return rowOffset;<a name="line.344"></a>
-<span class="sourceLineNo">345</span> }<a name="line.345"></a>
-<span class="sourceLineNo">346</span><a name="line.346"></a>
-<span class="sourceLineNo">347</span> public void setRowOffset(int rowOffset) {<a name="line.347"></a>
-<span class="sourceLineNo">348</span> this.rowOffset = rowOffset;<a name="line.348"></a>
-<span class="sourceLineNo">349</span> }<a name="line.349"></a>
-<span class="sourceLineNo">350</span><a name="line.350"></a>
-<span class="sourceLineNo">351</span> public int getRowLength() {<a name="line.351"></a>
-<span class="sourceLineNo">352</span> return rowLength;<a name="line.352"></a>
-<span class="sourceLineNo">353</span> }<a name="line.353"></a>
-<span class="sourceLineNo">354</span><a name="line.354"></a>
-<span class="sourceLineNo">355</span> public void setRowLength(int rowLength) {<a name="line.355"></a>
-<span class="sourceLineNo">356</span> this.rowLength = rowLength;<a name="line.356"></a>
-<span class="sourceLineNo">357</span> }<a name="line.357"></a>
-<span class="sourceLineNo">358</span><a name="line.358"></a>
-<span class="sourceLineNo">359</span> public byte[] getFamilyArray() {<a name="line.359"></a>
-<span class="sourceLineNo">360</span> return family;<a name="line.360"></a>
-<span class="sourceLineNo">361</span> }<a name="line.361"></a>
-<span class="sourceLineNo">362</span><a name="line.362"></a>
-<span class="sourceLineNo">363</span> public void setFamily(byte[] family) {<a name="line.363"></a>
-<span class="sourceLineNo">364</span> this.family = family;<a name="line.364"></a>
-<span class="sourceLineNo">365</span> }<a name="line.365"></a>
-<span class="sourceLineNo">366</span><a name="line.366"></a>
-<span class="sourceLineNo">367</span> public int getFamilyOffset() {<a name="line.367"></a>
-<span class="sourceLineNo">368</span> return familyOffset;<a name="line.368"></a>
-<span class="sourceLineNo">369</span> }<a name="line.369"></a>
-<span class="sourceLineNo">370</span><a name="line.370"></a>
-<span class="sourceLineNo">371</span> public void setFamilyOffset(int familyOffset) {<a name="line.371"></a>
-<span class="sourceLineNo">372</span> this.familyOffset = familyOffset;<a name="line.372"></a>
-<span class="sourceLineNo">373</span> }<a name="line.373"></a>
-<span class="sourceLineNo">374</span><a name="line.374"></a>
-<span class="sourceLineNo">375</span> public int getFamilyLength() {<a name="line.375"></a>
-<span class="sourceLineNo">376</span> return familyLength;<a name="line.376"></a>
-<span class="sourceLineNo">377</span> }<a name="line.377"></a>
-<span class="sourceLineNo">378</span><a name="line.378"></a>
-<span class="sourceLineNo">379</span> public void setFamilyLength(int familyLength) {<a name="line.379"></a>
-<span class="sourceLineNo">380</span> this.familyLength = familyLength;<a name="line.380"></a>
-<span class="sourceLineNo">381</span> }<a name="line.381"></a>
-<span class="sourceLineNo">382</span><a name="line.382"></a>
-<span class="sourceLineNo">383</span> public byte[] getQualifierArray() {<a name="line.383"></a>
-<span class="sourceLineNo">384</span> return qualifier;<a name="line.384"></a>
-<span class="sourceLineNo">385</span> }<a name="line.385"></a>
-<span class="sourceLineNo">386</span><a name="line.386"></a>
-<span class="sourceLineNo">387</span> public void setQualifier(byte[] qualifier) {<a name="line.387"></a>
-<span class="sourceLineNo">388</span> this.qualifier = qualifier;<a name="line.388"></a>
-<span class="sourceLineNo">389</span> }<a name="line.389"></a>
-<span class="sourceLineNo">390</span><a name="line.390"></a>
-<span class="sourceLineNo">391</span> public int getQualifierOffset() {<a name="line.391"></a>
-<span class="sourceLineNo">392</span> return qualifierOffset;<a name="line.392"></a>
-<span class="sourceLineNo">393</span> }<a name="line.393"></a>
-<span class="sourceLineNo">394</span><a name="line.394"></a>
-<span class="sourceLineNo">395</span> public void setQualifierOffset(int qualifierOffset) {<a name="line.395"></a>
-<span class="sourceLineNo">396</span> this.qualifierOffset = qualifierOffset;<a name="line.396"></a>
-<span class="sourceLineNo">397</span> }<a name="line.397"></a>
-<span class="sourceLineNo">398</span><a name="line.398"></a>
-<span class="sourceLineNo">399</span> public int getQualifierLength() {<a name="line.399"></a>
-<span class="sourceLineNo">400</span> return qualifierLength;<a name="line.400"></a>
-<span class="sourceLineNo">401</span> }<a name="line.401"></a>
-<span class="sourceLineNo">402</span><a name="line.402"></a>
-<span class="sourceLineNo">403</span> public void setQualifierLength(int qualifierLength) {<a name="line.403"></a>
-<span class="sourceLineNo">404</span> this.qualifierLength = qualifierLength;<a name="line.404"></a>
-<span class="sourceLineNo">405</span> }<a name="line.405"></a>
-<span class="sourceLineNo">406</span><a name="line.406"></a>
-<span class="sourceLineNo">407</span> public long getTimestamp() {<a name="line.407"></a>
-<span class="sourceLineNo">408</span> return ts;<a name="line.408"></a>
-<span class="sourceLineNo">409</span> }<a name="line.409"></a>
-<span class="sourceLineNo">410</span><a name="line.410"></a>
-<span class="sourceLineNo">411</span> public void setTimestamp(long ts) {<a name="line.411"></a>
-<span class="sourceLineNo">412</span> this.ts = ts;<a name="line.412"></a>
-<span class="sourceLineNo">413</span> }<a name="line.413"></a>
-<span class="sourceLineNo">414</span> }<a name="line.414"></a>
-<span class="sourceLineNo">415</span><a name="line.415"></a>
-<span class="sourceLineNo">416</span> public static class Loader extends Configured implements Tool {<a name="line.416"></a>
-<span class="sourceLineNo">417</span><a name="line.417"></a>
-<span class="sourceLineNo">418</span> private static final Logger LOG = LoggerFactory.getLogger(Loader.class);<a name="line.418"></a>
-<span class="sourceLineNo">419</span> private static final String USAGE = "Loader <warInputDir | warFileList> <outputDir>";<a name="line.419"></a>
+<span class="sourceLineNo">194</span> if (outputDir == null) {<a name="line.194"></a>
+<span class="sourceLineNo">195</span> throw new IllegalArgumentException("Output directory not specified");<a name="line.195"></a>
+<span class="sourceLineNo">196</span> }<a name="line.196"></a>
+<span class="sourceLineNo">197</span> int res = runLoader(warcFileInputDir, outputDir);<a name="line.197"></a>
+<span class="sourceLineNo">198</span> if (res != 0) {<a name="line.198"></a>
+<span class="sourceLineNo">199</span> LOG.error("Loader failed");<a name="line.199"></a>
+<span class="sourceLineNo">200</span> return -1;<a name="line.200"></a>
+<span class="sourceLineNo">201</span> }<a name="line.201"></a>
+<span class="sourceLineNo">202</span> res = runVerify(outputDir);<a name="line.202"></a>
+<span class="sourceLineNo">203</span> } catch (Exception e) {<a name="line.203"></a>
+<span class="sourceLineNo">204</span> LOG.error("Tool failed with exception", e);<a name="line.204"></a>
+<span class="sourceLineNo">205</span> return -1;<a name="line.205"></a>
+<span class="sourceLineNo">206</span> }<a name="line.206"></a>
+<span class="sourceLineNo">207</span> return 0;<a name="line.207"></a>
+<span class="sourceLineNo">208</span> }<a name="line.208"></a>
+<span class="sourceLineNo">209</span><a name="line.209"></a>
+<span class="sourceLineNo">210</span> @Override<a name="line.210"></a>
+<span class="sourceLineNo">211</span> protected void processOptions(CommandLine cmd) {<a name="line.211"></a>
+<span class="sourceLineNo">212</span> processBaseOptions(cmd);<a name="line.212"></a>
+<span class="sourceLineNo">213</span> args = cmd.getArgs();<a name="line.213"></a>
+<span class="sourceLineNo">214</span> }<a name="line.214"></a>
+<span class="sourceLineNo">215</span><a name="line.215"></a>
+<span class="sourceLineNo">216</span> @Override<a name="line.216"></a>
+<span class="sourceLineNo">217</span> public void setUpCluster() throws Exception {<a name="line.217"></a>
+<span class="sourceLineNo">218</span> util = getTestingUtil(getConf());<a name="line.218"></a>
+<span class="sourceLineNo">219</span> boolean isDistributed = util.isDistributedCluster();<a name="line.219"></a>
+<span class="sourceLineNo">220</span> util.initializeCluster(isDistributed ? 1 : 3);<a name="line.220"></a>
+<span class="sourceLineNo">221</span> if (!isDistributed) {<a name="line.221"></a>
+<span class="sourceLineNo">222</span> util.startMiniMapReduceCluster();<a name="line.222"></a>
+<span class="sourceLineNo">223</span> }<a name="line.223"></a>
+<span class="sourceLineNo">224</span> this.setConf(util.getConfiguration());<a name="line.224"></a>
+<span class="sourceLineNo">225</span> }<a name="line.225"></a>
+<span class="sourceLineNo">226</span><a name="line.226"></a>
+<span class="sourceLineNo">227</span> @Override<a name="line.227"></a>
+<span class="sourceLineNo">228</span> public void cleanUpCluster() throws Exception {<a name="line.228"></a>
+<span class="sourceLineNo">229</span> super.cleanUpCluster();<a name="line.229"></a>
+<span class="sourceLineNo">230</span> if (util.isDistributedCluster()) {<a name="line.230"></a>
+<span class="sourceLineNo">231</span> util.shutdownMiniMapReduceCluster();<a name="line.231"></a>
+<span class="sourceLineNo">232</span> }<a name="line.232"></a>
+<span class="sourceLineNo">233</span> }<a name="line.233"></a>
+<span class="sourceLineNo">234</span><a name="line.234"></a>
+<span class="sourceLineNo">235</span> static TableName getTablename(Configuration c) {<a name="line.235"></a>
+<span class="sourceLineNo">236</span> return TableName.valueOf(c.get(TABLE_NAME_KEY, DEFAULT_TABLE_NAME));<a name="line.236"></a>
+<span class="sourceLineNo">237</span> }<a name="line.237"></a>
+<span class="sourceLineNo">238</span><a name="line.238"></a>
+<span class="sourceLineNo">239</span> @Override<a name="line.239"></a>
+<span class="sourceLineNo">240</span> public TableName getTablename() {<a name="line.240"></a>
+<span class="sourceLineNo">241</span> return getTablename(getConf());<a name="line.241"></a>
+<span class="sourceLineNo">242</span> }<a name="line.242"></a>
+<span class="sourceLineNo">243</span><a name="line.243"></a>
+<span class="sourceLineNo">244</span> @Override<a name="line.244"></a>
+<span class="sourceLineNo">245</span> protected Set<String> getColumnFamilies() {<a name="line.245"></a>
+<span class="sourceLineNo">246</span> Set<String> families = new HashSet<>();<a name="line.246"></a>
+<span class="sourceLineNo">247</span> families.add(Bytes.toString(CONTENT_FAMILY_NAME));<a name="line.247"></a>
+<span class="sourceLineNo">248</span> families.add(Bytes.toString(INFO_FAMILY_NAME));<a name="line.248"></a>
+<span class="sourceLineNo">249</span> return families;<a name="line.249"></a>
+<span class="sourceLineNo">250</span> }<a name="line.250"></a>
+<span class="sourceLineNo">251</span><a name="line.251"></a>
+<span class="sourceLineNo">252</span> @Override<a name="line.252"></a>
+<span class="sourceLineNo">253</span> public int runTestFromCommandLine() throws Exception {<a name="line.253"></a>
+<span class="sourceLineNo">254</span> return ToolRunner.run(getConf(), this, args);<a name="line.254"></a>
+<span class="sourceLineNo">255</span> }<a name="line.255"></a>
+<span class="sourceLineNo">256</span><a name="line.256"></a>
+<span class="sourceLineNo">257</span> public static void main(String[] args) throws Exception {<a name="line.257"></a>
+<span class="sourceLineNo">258</span> Configuration conf = HBaseConfiguration.create();<a name="line.258"></a>
+<span class="sourceLineNo">259</span> IntegrationTestingUtility.setUseDistributedCluster(conf);<a name="line.259"></a>
+<span class="sourceLineNo">260</span> int ret = ToolRunner.run(conf, new IntegrationTestLoadCommonCrawl(), args);<a name="line.260"></a>
+<span class="sourceLineNo">261</span> System.exit(ret);<a name="line.261"></a>
+<span class="sourceLineNo">262</span> }<a name="line.262"></a>
+<span class="sourceLineNo">263</span><a name="line.263"></a>
+<span class="sourceLineNo">264</span> public static class HBaseKeyWritable implements Writable {<a name="line.264"></a>
+<span class="sourceLineNo">265</span><a name="line.265"></a>
+<span class="sourceLineNo">266</span> private byte[] row;<a name="line.266"></a>
+<span class="sourceLineNo">267</span> private int rowOffset;<a name="line.267"></a>
+<span class="sourceLineNo">268</span> private int rowLength;<a name="line.268"></a>
+<span class="sourceLineNo">269</span> private byte[] family;<a name="line.269"></a>
+<span class="sourceLineNo">270</span> private int familyOffset;<a name="line.270"></a>
+<span class="sourceLineNo">271</span> private int familyLength;<a name="line.271"></a>
+<span class="sourceLineNo">272</span> private byte[] qualifier;<a name="line.272"></a>
+<span class="sourceLineNo">273</span> private int qualifierOffset;<a name="line.273"></a>
+<span class="sourceLineNo">274</span> private int qualifierLength;<a name="line.274"></a>
+<span class="sourceLineNo">275</span> private long ts;<a name="line.275"></a>
+<span class="sourceLineNo">276</span><a name="line.276"></a>
+<span class="sourceLineNo">277</span> public HBaseKeyWritable() { }<a name="line.277"></a>
+<span class="sourceLineNo">278</span><a name="line.278"></a>
+<span class="sourceLineNo">279</span> public HBaseKeyWritable(byte[] row, int rowOffset, int rowLength,<a name="line.279"></a>
+<span class="sourceLineNo">280</span> byte[] family, int familyOffset, int familyLength,<a name="line.280"></a>
+<span class="sourceLineNo">281</span> byte[] qualifier, int qualifierOffset, int qualifierLength, long ts) {<a name="line.281"></a>
+<span class="sourceLineNo">282</span> this.row = row;<a name="line.282"></a>
+<span class="sourceLineNo">283</span> this.rowOffset = rowOffset;<a name="line.283"></a>
+<span class="sourceLineNo">284</span> this.rowLength = rowLength;<a name="line.284"></a>
+<span class="sourceLineNo">285</span> this.family = family;<a name="line.285"></a>
+<span class="sourceLineNo">286</span> this.familyOffset = familyOffset;<a name="line.286"></a>
+<span class="sourceLineNo">287</span> this.familyLength = familyLength;<a name="line.287"></a>
+<span class="sourceLineNo">288</span> this.qualifier = qualifier;<a name="line.288"></a>
+<span class="sourceLineNo">289</span> this.qualifierOffset = qualifierOffset;<a name="line.289"></a>
+<span class="sourceLineNo">290</span> this.qualifierLength = qualifierLength;<a name="line.290"></a>
+<span class="sourceLineNo">291</span> this.ts = ts;<a name="line.291"></a>
+<span class="sourceLineNo">292</span> }<a name="line.292"></a>
+<span class="sourceLineNo">293</span><a name="line.293"></a>
+<span class="sourceLineNo">294</span> public HBaseKeyWritable(byte[] row, byte[] family, byte[] qualifier, long ts) {<a name="line.294"></a>
+<span class="sourceLineNo">295</span> this(row, 0, row.length,<a name="line.295"></a>
+<span class="sourceLineNo">296</span> family, 0, family.length,<a name="line.296"></a>
+<span class="sourceLineNo">297</span> qualifier, 0, qualifier != null ? qualifier.length : 0, ts);<a name="line.297"></a>
+<span class="sourceLineNo">298</span> }<a name="line.298"></a>
+<span class="sourceLineNo">299</span><a name="line.299"></a>
+<span class="sourceLineNo">300</span> public HBaseKeyWritable(byte[] row, byte[] family, long ts) {<a name="line.300"></a>
+<span class="sourceLineNo">301</span> this(row, family, HConstants.EMPTY_BYTE_ARRAY, ts);<a name="line.301"></a>
+<span class="sourceLineNo">302</span> }<a name="line.302"></a>
+<span class="sourceLineNo">303</span><a name="line.303"></a>
+<span class="sourceLineNo">304</span> public HBaseKeyWritable(Cell cell) {<a name="line.304"></a>
+<span class="sourceLineNo">305</span> this(cell.getRowArray(), cell.getRowOffset(), cell.getRowLength(),<a name="line.305"></a>
+<span class="sourceLineNo">306</span> cell.getFamilyArray(), cell.getFamilyOffset(), cell.getFamilyLength(),<a name="line.306"></a>
+<span class="sourceLineNo">307</span> cell.getQualifierArray(), cell.getQualifierOffset(), cell.getQualifierLength(),<a name="line.307"></a>
+<span class="sourceLineNo">308</span> cell.getTimestamp());<a name="line.308"></a>
+<span class="sourceLineNo">309</span> }<a name="line.309"></a>
+<span class="sourceLineNo">310</span><a name="line.310"></a>
+<span class="sourceLineNo">311</span> @Override<a name="line.311"></a>
+<span class="sourceLineNo">312</span> public void readFields(DataInput in) throws IOException {<a name="line.312"></a>
+<span class="sourceLineNo">313</span> this.row = Bytes.toBytes(in.readUTF());<a name="line.313"></a>
+<span class="sourceLineNo">314</span> this.rowOffset = 0;<a name="line.314"></a>
+<span class="sourceLineNo">315</span> this.rowLength = row.length;<a name="line.315"></a>
+<span class="sourceLineNo">316</span> this.family = Bytes.toBytes(in.readUTF());<a name="line.316"></a>
+<span class="sourceLineNo">317</span> this.familyOffset = 0;<a name="line.317"></a>
+<span class="sourceLineNo">318</span> this.familyLength = family.length;<a name="line.318"></a>
+<span class="sourceLineNo">319</span> this.qualifier = Bytes.toBytes(in.readUTF());<a name="line.319"></a>
+<span class="sourceLineNo">320</span> this.qualifierOffset = 0;<a name="line.320"></a>
+<span class="sourceLineNo">321</span> this.qualifierLength = qualifier.length;<a name="line.321"></a>
+<span class="sourceLineNo">322</span> this.ts = in.readLong();<a name="line.322"></a>
+<span class="sourceLineNo">323</span> }<a name="line.323"></a>
+<span class="sourceLineNo">324</span><a name="line.324"></a>
+<span class="sourceLineNo">325</span> @Override<a name="line.325"></a>
+<span class="sourceLineNo">326</span> public void write(DataOutput out) throws IOException {<a name="line.326"></a>
+<span class="sourceLineNo">327</span> out.writeUTF(new String(row, rowOffset, rowLength, StandardCharsets.UTF_8));<a name="line.327"></a>
+<span class="sourceLineNo">328</span> out.writeUTF(new String(family, familyOffset, familyLength, StandardCharsets.UTF_8));<a name="line.328"></a>
+<span class="sourceLineNo">329</span> if (qualifier != null) {<a name="line.329"></a>
+<span class="sourceLineNo">330</span> out.writeUTF(new String(qualifier, qualifierOffset, qualifierLength,<a name="line.330"></a>
+<span class="sourceLineNo">331</span> StandardCharsets.UTF_8));<a name="line.331"></a>
+<span class="sourceLineNo">332</span> } else {<a name="line.332"></a>
+<span class="sourceLineNo">333</span> out.writeUTF("");<a name="line.333"></a>
+<span class="sourceLineNo">334</span> }<a name="line.334"></a>
+<span class="sourceLineNo">335</span> out.writeLong(ts);<a name="line.335"></a>
+<span class="sourceLineNo">336</span> }<a name="line.336"></a>
+<span class="sourceLineNo">337</span><a name="line.337"></a>
+<span class="sourceLineNo">338</span> public byte[] getRowArray() {<a name="line.338"></a>
+<span class="sourceLineNo">339</span> return row;<a name="line.339"></a>
+<span class="sourceLineNo">340</span> }<a name="line.340"></a>
+<span class="sourceLineNo">341</span><a name="line.341"></a>
+<span class="sourceLineNo">342</span> public void setRow(byte[] row) {<a name="line.342"></a>
+<span class="sourceLineNo">343</span> this.row = row;<a name="line.343"></a>
+<span class="sourceLineNo">344</span> }<a name="line.344"></a>
+<span class="sourceLineNo">345</span><a name="line.345"></a>
+<span class="sourceLineNo">346</span> public int getRowOffset() {<a name="line.346"></a>
+<span class="sourceLineNo">347</span> return rowOffset;<a name="line.347"></a>
+<span class="sourceLineNo">348</span> }<a name="line.348"></a>
+<span class="sourceLineNo">349</span><a name="line.349"></a>
+<span class="sourceLineNo">350</span> public void setRowOffset(int rowOffset) {<a name="line.350"></a>
+<span class="sourceLineNo">351</span> this.rowOffset = rowOffset;<a name="line.351"></a>
+<span class="sourceLineNo">352</span> }<a name="line.352"></a>
+<span class="sourceLineNo">353</span><a name="line.353"></a>
+<span class="sourceLineNo">354</span> public int getRowLength() {<a name="line.354"></a>
+<span class="sourceLineNo">355</span> return rowLength;<a name="line.355"></a>
+<span class="sourceLineNo">356</span> }<a name="line.356"></a>
+<span class="sourceLineNo">357</span><a name="line.357"></a>
+<span class="sourceLineNo">358</span> public void setRowLength(int rowLength) {<a name="line.358"></a>
+<span class="sourceLineNo">359</span> this.rowLength = rowLength;<a name="line.359"></a>
+<span class="sourceLineNo">360</span> }<a name="line.360"></a>
+<span class="sourceLineNo">361</span><a name="line.361"></a>
+<span class="sourceLineNo">362</span> public byte[] getFamilyArray() {<a name="line.362"></a>
+<span class="sourceLineNo">363</span> return family;<a name="line.363"></a>
+<span class="sourceLineNo">364</span> }<a name="line.364"></a>
+<span class="sourceLineNo">365</span><a name="line.365"></a>
+<span class="sourceLineNo">366</span> public void setFamily(byte[] family) {<a name="line.366"></a>
+<span class="sourceLineNo">367</span> this.family = family;<a name="line.367"></a>
+<span class="sourceLineNo">368</span> }<a name="line.368"></a>
+<span class="sourceLineNo">369</span><a name="line.369"></a>
+<span class="sourceLineNo">370</span> public int getFamilyOffset() {<a name="line.370"></a>
+<span class="sourceLineNo">371</span> return familyOffset;<a name="line.371"></a>
+<span class="sourceLineNo">372</span> }<a name="line.372"></a>
+<span class="sourceLineNo">373</span><a name="line.373"></a>
+<span class="sourceLineNo">374</span> public void setFamilyOffset(int familyOffset) {<a name="line.374"></a>
+<span class="sourceLineNo">375</span> this.familyOffset = familyOffset;<a name="line.375"></a>
+<span class="sourceLineNo">376</span> }<a name="line.376"></a>
+<span class="sourceLineNo">377</span><a name="line.377"></a>
+<span class="sourceLineNo">378</span> public int getFamilyLength() {<a name="line.378"></a>
+<span class="sourceLineNo">379</span> return familyLength;<a name="line.379"></a>
+<span class="sourceLineNo">380</span> }<a name="line.380"></a>
+<span class="sourceLineNo">381</span><a name="line.381"></a>
+<span class="sourceLineNo">382</span> public void setFamilyLength(int familyLength) {<a name="line.382"></a>
+<span class="sourceLineNo">383</span> this.familyLength = familyLength;<a name="line.383"></a>
+<span class="sourceLineNo">384</span> }<a name="line.384"></a>
+<span class="sourceLineNo">385</span><a name="line.385"></a>
+<span class="sourceLineNo">386</span> public byte[] getQualifierArray() {<a name="line.386"></a>
+<span class="sourceLineNo">387</span> return qualifier;<a name="line.387"></a>
+<span class="sourceLineNo">388</span> }<a name="line.388"></a>
+<span class="sourceLineNo">389</span><a name="line.389"></a>
+<span class="sourceLineNo">390</span> public void setQualifier(byte[] qualifier) {<a name="line.390"></a>
+<span class="sourceLineNo">391</span> this.qualifier = qualifier;<a name="line.391"></a>
+<span class="sourceLineNo">392</span> }<a name="line.392"></a>
+<span class="sourceLineNo">393</span><a name="line.393"></a>
+<span class="sourceLineNo">394</span> public int getQualifierOffset() {<a name="line.394"></a>
+<span class="sourceLineNo">395</span> return qualifierOffset;<a name="line.395"></a>
+<span class="sourceLineNo">396</span> }<a name="line.396"></a>
+<span class="sourceLineNo">397</span><a name="line.397"></a>
+<span class="sourceLineNo">398</span> public void setQualifierOffset(int qualifierOffset) {<a name="line.398"></a>
+<span class="sourceLineNo">399</span> this.qualifierOffset = qualifierOffset;<a name="line.399"></a>
+<span class="sourceLineNo">400</span> }<a name="line.400"></a>
+<span class="sourceLineNo">401</span><a name="line.401"></a>
+<span class="sourceLineNo">402</span> public int getQualifierLength() {<a name="line.402"></a>
+<span class="sourceLineNo">403</span> return qualifierLength;<a name="line.403"></a>
+<span class="sourceLineNo">404</span> }<a name="line.404"></a>
+<span class="sourceLineNo">405</span><a name="line.405"></a>
+<span class="sourceLineNo">406</span> public void setQualifierLength(int qualifierLength) {<a name="line.406"></a>
+<span class="sourceLineNo">407</span> this.qualifierLength = qualifierLength;<a name="line.407"></a>
+<span class="sourceLineNo">408</span> }<a name="line.408"></a>
+<span class="sourceLineNo">409</span><a name="line.409"></a>
+<span class="sourceLineNo">410</span> public long getTimestamp() {<a name="line.410"></a>
+<span class="sourceLineNo">411</span> return ts;<a name="line.411"></a>
+<span class="sourceLineNo">412</span> }<a name="line.412"></a>
+<span class="sourceLineNo">413</span><a name="line.413"></a>
+<span class="sourceLineNo">414</span> public void setTimestamp(long ts) {<a name="line.414"></a>
+<span class="sourceLineNo">415</span> this.ts = ts;<a name="line.415"></a>
+<span class="sourceLineNo">416</span> }<a name="line.416"></a>
+<span class="sourceLineNo">417</span> }<a name="line.417"></a>
+<span class="sourceLineNo">418</span><a name="line.418"></a>
+<span class="sourceLineNo">419</span> public static class Loader extends Configured implements Tool {<a name="line.419"></a>
<span class="sourceLineNo">420</span><a name="line.420"></a>
-<span class="sourceLineNo">421</span> void createSchema(TableName tableName) throws IOException {<a name="line.421"></a>
-<span class="sourceLineNo">422</span><a name="line.422"></a>
-<span class="sourceLineNo">423</span> try (Connection conn = ConnectionFactory.createConnection(getConf());<a name="line.423"></a>
-<span class="sourceLineNo">424</span> Admin admin = conn.getAdmin()) {<a name="line.424"></a>
-<span class="sourceLineNo">425</span> if (!admin.tableExists(tableName)) {<a name="line.425"></a>
-<span class="sourceLineNo">426</span><a name="line.426"></a>
-<span class="sourceLineNo">427</span> ColumnFamilyDescriptorBuilder contentFamilyBuilder =<a name="line.427"></a>
-<span class="sourceLineNo">428</span> ColumnFamilyDescriptorBuilder.newBuilder(CONTENT_FAMILY_NAME)<a name="line.428"></a>
-<span class="sourceLineNo">429</span> .setDataBlockEncoding(DataBlockEncoding.NONE)<a name="line.429"></a>
-<span class="sourceLineNo">430</span> .setBloomFilterType(BloomType.ROW)<a name="line.430"></a>
-<span class="sourceLineNo">431</span> .setMaxVersions(1000)<a name="line.431"></a>
-<span class="sourceLineNo">432</span> .setBlocksize(256 * 1024)<a name="line.432"></a>
-<span class="sourceLineNo">433</span> ;<a name="line.433"></a>
-<span class="sourceLineNo">434</span><a name="line.434"></a>
-<span class="sourceLineNo">435</span> ColumnFamilyDescriptorBuilder infoFamilyBuilder =<a name="line.435"></a>
-<span class="sourceLineNo">436</span> ColumnFamilyDescriptorBuilder.newBuilder(INFO_FAMILY_NAME)<a name="line.436"></a>
-<span class="sourceLineNo">437</span> .setDataBlockEncoding(DataBlockEncoding.NONE)<a name="line.437"></a>
-<span class="sourceLineNo">438</span> .setBloomFilterType(BloomType.ROWCOL)<a name="line.438"></a>
-<span class="sourceLineNo">439</span> .setMaxVersions(1000)<a name="line.439"></a>
-<span class="sourceLineNo">440</span> .setBlocksize(8 * 1024)<a name="line.440"></a>
-<span class="sourceLineNo">441</span> ;<a name="line.441"></a>
-<span class="sourceLineNo">442</span><a name="line.442"></a>
-<span class="sourceLineNo">443</span> Set<ColumnFamilyDescriptor> families = new HashSet<>();<a name="line.443"></a>
-<span class="sourceLineNo">444</span> families.add(contentFamilyBuilder.build());<a name="line.444"></a>
-<span class="sourceLineNo">445</span> families.add(infoFamilyBuilder.build());<a name="line.445"></a>
-<span class="sourceLineNo">446</span><a name="line.446"></a>
-<span class="sourceLineNo">447</span> TableDescriptor tableDescriptor = TableDescriptorBuilder.newBuilder(tableName)<a name="line.447"></a>
-<span class="sourceLineNo">448</span> .setColumnFamilies(families)<a name="line.448"></a>
-<span class="sourceLineNo">449</span> .build();<a name="line.449"></a>
-<span class="sourceLineNo">450</span><a name="line.450"></a>
-<span class="sourceLineNo">451</span> if (getConf().getBoolean(HBaseTestingUtil.PRESPLIT_TEST_TABLE_KEY,<a name="line.451"></a>
-<span class="sourceLineNo">452</span> HBaseTestingUtil.PRESPLIT_TEST_TABLE)) {<a name="line.452"></a>
-<span class="sourceLineNo">453</span> int numberOfServers = admin.getRegionServers().size();<a name="line.453"></a>
-<span class="sourceLineNo">454</span> if (numberOfServers == 0) {<a name="line.454"></a>
-<span class="sourceLineNo">455</span> throw new IllegalStateException("No live regionservers");<a name="line.455"></a>
-<span class="sourceLineNo">456</span> }<a name="line.456"></a>
-<span class="sourceLineNo">457</span> int regionsPerServer = getConf().getInt(HBaseTestingUtil.REGIONS_PER_SERVER_KEY,<a name="line.457"></a>
-<span class="sourceLineNo">458</span> HBaseTestingUtil.DEFAULT_REGIONS_PER_SERVER);<a name="line.458"></a>
-<span class="sourceLineNo">459</span> int totalNumberOfRegions = numberOfServers * regionsPerServer;<a name="line.459"></a>
-<span class="sourceLineNo">460</span> LOG.info("Creating test table: " + tableDescriptor);<a name="line.460"></a>
-<span class="sourceLineNo">461</span> LOG.info("Number of live regionservers: " + numberOfServers + ", " +<a name="line.461"></a>
-<span class="sourceLineNo">462</span> "pre-splitting table into " + totalNumberOfRegions + " regions " +<a name="line.462"></a>
-<span class="sourceLineNo">463</span> "(default regions per server: " + regionsPerServer + ")");<a name="line.463"></a>
-<span class="sourceLineNo">464</span> byte[][] splits = new RegionSplitter.UniformSplit().split(totalNumberOfRegions);<a name="line.464"></a>
-<span class="sourceLineNo">465</span> admin.createTable(tableDescriptor, splits);<a name="line.465"></a>
-<span class="sourceLineNo">466</span> } else {<a name="line.466"></a>
-<span class="sourceLineNo">467</span> LOG.info("Creating test table: " + tableDescriptor);<a name="line.467"></a>
-<span class="sourceLineNo">468</span> admin.createTable(tableDescriptor);<a name="line.468"></a>
-<span class="sourceLineNo">469</span> }<a name="line.469"></a>
-<span class="sourceLineNo">470</span> }<a name="line.470"></a>
-<span class="sourceLineNo">471</span> } catch (MasterNotRunningException e) {<a name="line.471"></a>
-<span class="sourceLineNo">472</span> LOG.error("Master not running", e);<a name="line.472"></a>
-<span class="sourceLineNo">473</span> throw new IOException(e);<a name="line.473"></a>
-<span class="sourceLineNo">474</span> }<a name="line.474"></a>
-<span class="sourceLineNo">475</span> }<a name="line.475"></a>
-<span class="sourceLineNo">476</span><a name="line.476"></a>
-<span class="sourceLineNo">477</span> int run(Path warcFileInput, Path outputDir)<a name="line.477"></a>
-<span class="sourceLineNo">478</span> throws IOException, ClassNotFoundException, InterruptedException {<a name="line.478"></a>
+<span class="sourceLineNo">421</span> private static final Logger LOG = LoggerFactory.getLogger(Loader.class);<a name="line.421"></a>
+<span class="sourceLineNo">422</span> private static final String USAGE = "Loader <warInputDir | warFileList> <outputDir>";<a name="line.422"></a>
+<span class="sourceLineNo">423</span><a name="line.423"></a>
+<span class="sourceLineNo">424</span> void createSchema(TableName tableName) throws IOException {<a name="line.424"></a>
+<span class="sourceLineNo">425</span><a name="line.425"></a>
+<span class="sourceLineNo">426</span> try (Connection conn = ConnectionFactory.createConnection(getConf());<a name="line.426"></a>
+<span class="sourceLineNo">427</span> Admin admin = conn.getAdmin()) {<a name="line.427"></a>
+<span class="sourceLineNo">428</span> if (!admin.tableExists(tableName)) {<a name="line.428"></a>
+<span class="sourceLineNo">429</span><a name="line.429"></a>
+<span class="sourceLineNo">430</span> ColumnFamilyDescriptorBuilder contentFamilyBuilder =<a name="line.430"></a>
+<span class="sourceLineNo">431</span> ColumnFamilyDescriptorBuilder.newBuilder(CONTENT_FAMILY_NAME)<a name="line.431"></a>
+<span class="sourceLineNo">432</span> .setDataBlockEncoding(DataBlockEncoding.NONE)<a name="line.432"></a>
+<span class="sourceLineNo">433</span> .setBloomFilterType(BloomType.ROW)<a name="line.433"></a>
+<span class="sourceLineNo">434</span> .setMaxVersions(1000)<a name="line.434"></a>
+<span class="sourceLineNo">435</span> .setBlocksize(256 * 1024)<a name="line.435"></a>
+<span class="sourceLineNo">436</span> ;<a name="line.436"></a>
+<span class="sourceLineNo">437</span><a name="line.437"></a>
+<span class="sourceLineNo">438</span> ColumnFamilyDescriptorBuilder infoFamilyBuilder =<a name="line.438"></a>
+<span class="sourceLineNo">439</span> ColumnFamilyDescriptorBuilder.newBuilder(INFO_FAMILY_NAME)<a name="line.439"></a>
+<span class="sourceLineNo">440</span> .setDataBlockEncoding(DataBlockEncoding.NONE)<a name="line.440"></a>
+<span class="sourceLineNo">441</span> .setBloomFilterType(BloomType.ROWCOL)<a name="line.441"></a>
+<span class="sourceLineNo">442</span> .setMaxVersions(1000)<a name="line.442"></a>
+<span class="sourceLineNo">443</span> .setBlocksize(8 * 1024)<a name="line.443"></a>
+<span class="sourceLineNo">444</span> ;<a name="line.444"></a>
+<span class="sourceLineNo">445</span><a name="line.445"></a>
+<span class="sourceLineNo">446</span> Set<ColumnFamilyDescriptor> families = new HashSet<>();<a name="line.446"></a>
+<span class="sourceLineNo">447</span> families.add(contentFamilyBuilder.build());<a name="line.447"></a>
+<span class="sourceLineNo">448</span> families.add(infoFamilyBuilder.build());<a name="line.448"></a>
+<span class="sourceLineNo">449</span><a name="line.449"></a>
+<span class="sourceLineNo">450</span> TableDescriptor tableDescriptor = TableDescriptorBuilder.newBuilder(tableName)<a name="line.450"></a>
+<span class="sourceLineNo">451</span> .setColumnFamilies(families)<a name="line.451"></a>
+<span class="sourceLineNo">452</span> .build();<a name="line.452"></a>
+<span class="sourceLineNo">453</span><a name="line.453"></a>
+<span class="sourceLineNo">454</span> if (getConf().getBoolean(HBaseTestingUtil.PRESPLIT_TEST_TABLE_KEY,<a name="line.454"></a>
+<span class="sourceLineNo">455</span> HBaseTestingUtil.PRESPLIT_TEST_TABLE)) {<a name="line.455"></a>
+<span class="sourceLineNo">456</span> int numberOfServers = admin.getRegionServers().size();<a name="line.456"></a>
+<span class="sourceLineNo">457</span> if (numberOfServers == 0) {<a name="line.457"></a>
+<span class="sourceLineNo">458</span> throw new IllegalStateException("No live regionservers");<a name="line.458"></a>
+<span class="sourceLineNo">459</span> }<a name="line.459"></a>
+<span class="sourceLineNo">460</span> int regionsPerServer = getConf().getInt(HBaseTestingUtil.REGIONS_PER_SERVER_KEY,<a name="line.460"></a>
+<span class="sourceLineNo">461</span> HBaseTestingUtil.DEFAULT_REGIONS_PER_SERVER);<a name="line.461"></a>
+<span class="sourceLineNo">462</span> int totalNumberOfRegions = numberOfServers * regionsPerServer;<a name="line.462"></a>
+<span class="sourceLineNo">463</span> LOG.info("Creating test table: " + tableDescriptor);<a name="line.463"></a>
+<span class="sourceLineNo">464</span> LOG.info("Number of live regionservers: " + numberOfServers + ", " +<a name="line.464"></a>
+<span class="sourceLineNo">465</span> "pre-splitting table into " + totalNumberOfRegions + " regions " +<a name="line.465"></a>
+<span class="sourceLineNo">466</span> "(default regions per server: " + regionsPerServer + ")");<a name="line.466"></a>
+<span class="sourceLineNo">467</span> byte[][] splits = new RegionSplitter.UniformSplit().split(totalNumberOfRegions);<a name="line.467"></a>
+<span class="sourceLineNo">468</span> admin.createTable(tableDescriptor, splits);<a name="line.468"></a>
+<span class="sourceLineNo">469</span> } else {<a name="line.469"></a>
+<span class="sourceLineNo">470</span> LOG.info("Creating test table: " + tableDescriptor);<a name="line.470"></a>
+<span class="sourceLineNo">471</span> admin.createTable(tableDescriptor);<a name="line.471"></a>
+<span class="sourceLineNo">472</span> }<a name="line.472"></a>
+<span class="sourceLineNo">473</span> }<a name="line.473"></a>
+<span class="sourceLineNo">474</span> } catch (MasterNotRunningException e) {<a name="line.474"></a>
+<span class="sourceLineNo">475</span> LOG.error("Master not running", e);<a name="line.475"></a>
+<span class="sourceLineNo">476</span> throw new IOException(e);<a name="line.476"></a>
+<span class="sourceLineNo">477</span> }<a name="line.477"></a>
+<span class="sourceLineNo">478</span> }<a name="line.478"></a>
<span class="sourceLineNo">479</span><a name="line.479"></a>
-<span class="sourceLineNo">480</span> createSchema(getTablename(getConf()));<a name="line.480"></a>
-<span class="sourceLineNo">481</span><a name="line.481"></a>
-<span class="sourceLineNo">482</span> Job job = Job.getInstance(getConf());<a name="line.482"></a>
-<span class="sourceLineNo">483</span> job.setJobName(Loader.class.getName());<a name="line.483"></a>
-<span class="sourceLineNo">484</span> job.setNumReduceTasks(0);<a name="line.484"></a>
-<span class="sourceLineNo">485</span> job.setJarByClass(getClass());<a name="line.485"></a>
-<span class="sourceLineNo">486</span> job.setMapperClass(LoaderMapper.class);<a name="line.486"></a>
-<span class="sourceLineNo">487</span> job.setInputFormatClass(WARCInputFormat.class);<a name="line.487"></a>
-<span class="sourceLineNo">488</span> FileSystem fs = FileSystem.get(warcFileInput.toUri(), getConf());<a name="line.488"></a>
-<span class="sourceLineNo">489</span> if (fs.getFileStatus(warcFileInput).isDirectory()) {<a name="line.489"></a>
-<span class="sourceLineNo">490</span> LOG.info("Using directory as WARC input path: " + warcFileInput);<a name="line.490"></a>
-<span class="sourceLineNo">491</span> FileInputFormat.setInputPaths(job, warcFileInput);<a name="line.491"></a>
-<span class="sourceLineNo">492</span> } else {<a name="line.492"></a>
-<span class="sourceLineNo">493</span> LOG.info("Getting WARC input paths from file: " + warcFileInput);<a name="line.493"></a>
-<span class="sourceLineNo">494</span> List<Path> paths = new LinkedList<Path>();<a name="line.494"></a>
-<span class="sourceLineNo">495</span> try (FSDataInputStream is = fs.open(warcFileInput)) {<a name="line.495"></a>
-<span class="sourceLineNo">496</span> InputStreamReader reader;<a name="line.496"></a>
-<span class="sourceLineNo">497</span> if (warcFileInput.getName().toLowerCase().endsWith(".gz")) {<a name="line.497"></a>
-<span class="sourceLineNo">498</span> reader = new InputStreamReader(new GZIPInputStream(is));<a name="line.498"></a>
-<span class="sourceLineNo">499</span> } else {<a name="line.499"></a>
-<span class="sourceLineNo">500</span> reader = new InputStreamReader(is);<a name="line.500"></a>
-<span class="sourceLineNo">501</span> }<a name="line.501"></a>
-<span class="sourceLineNo">502</span> try (BufferedReader br = new BufferedReader(reader)) {<a name="line.502"></a>
-<span class="sourceLineNo">503</span> String line;<a name="line.503"></a>
-<span class="sourceLineNo">504</span> while ((line = br.readLine()) != null) {<a name="line.504"></a>
-<span class="sourceLineNo">505</span> paths.add(new Path(line));<a name="line.505"></a>
-<span class="sourceLineNo">506</span> }<a name="line.506"></a>
-<span class="sourceLineNo">507</span> }<a name="line.507"></a>
-<span class="sourceLineNo">508</span> }<a name="line.508"></a>
-<span class="sourceLineNo">509</span> LOG.info("Read " + paths.size() + " WARC input paths from " + warcFileInput);<a name="line.509"></a>
-<span class="sourceLineNo">510</span> FileInputFormat.setInputPaths(job, paths.toArray(new Path[paths.size()]));<a name="line.510"></a>
-<span class="sourceLineNo">511</span> }<a name="line.511"></a>
-<span class="sourceLineNo">512</span> job.setOutputFormatClass(SequenceFileOutputFormat.class);<a name="line.512"></a>
-<span class="sourceLineNo">513</span> SequenceFileOutputFormat.setOutputPath(job, outputDir);<a name="line.513"></a>
-<span class="sourceLineNo">514</span> SequenceFileOutputFormat.setOutputCompressionType(job, CompressionType.BLOCK);<a name="line.514"></a>
-<span class="sourceLineNo">515</span> job.setOutputKeyClass(HBaseKeyWritable.class);<a name="line.515"></a>
-<span class="sourceLineNo">516</span> job.setOutputValueClass(BytesWritable.class);<a name="line.516"></a>
-<span class="sourceLineNo">517</span> TableMapReduceUtil.addDependencyJars(job);<a name="line.517"></a>
-<span class="sourceLineNo">518</span><a name="line.518"></a>
-<span class="sourceLineNo">519</span> LOG.info("Submitting job." +<a name="line.519"></a>
-<span class="sourceLineNo">520</span> " This will take time proportional to the number of input files, please be patient.");<a name="line.520"></a>
-<span class="sourceLineNo">521</span> boolean success = job.waitForCompletion(true);<a name="line.521"></a>
-<span class="sourceLineNo">522</span> if (!success) {<a name="line.522"></a>
-<span class="sourceLineNo">523</span> LOG.error("Failure during job " + job.getJobID());<a name="line.523"></a>
-<span class="sourceLineNo">524</span> }<a name="line.524"></a>
-<span class="sourceLineNo">525</span> return success ? 0 : 1;<a name="line.525"></a>
-<span class="sourceLineNo">526</span> }<a name="line.526"></a>
-<span class="sourceLineNo">527</span><a name="line.527"></a>
-<span class="sourceLineNo">528</span> @Override<a name="line.528"></a>
-<span class="sourceLineNo">529</span> public int run(String[] args) throws Exception {<a name="line.529"></a>
-<span class="sourceLineNo">530</span> if (args.length < 2) {<a name="line.530"></a>
-<span class="sourceLineNo">531</span> System.err.println(USAGE);<a name="line.531"></a>
-<span class="sourceLineNo">532</span> return 1;<a name="line.532"></a>
-<span class="sourceLineNo">533</span> }<a name="line.533"></a>
-<span class="sourceLineNo">534</span> try {<a name="line.534"></a>
-<span class="sourceLineNo">535</span> Path warcFileInput = new Path(args[0]);<a name="line.535"></a>
-<span class="sourceLineNo">536</span> Path outputDir = new Path(args[1]);<a name="line.536"></a>
-<span class="sourceLineNo">537</span> return run(warcFileInput, outputDir);<a name="line.537"></a>
-<span class="sourceLineNo">538</span> } catch (NumberFormatException e) {<a name="line.538"></a>
-<span class="sourceLineNo">539</span> System.err.println("Parsing loader arguments failed: " + e.getMessage());<a name="line.539"></a>
-<span class="sourceLineNo">540</span> System.err.println(USAGE);<a name="line.540"></a>
-<span class="sourceLineNo">541</span> return 1;<a name="line.541"></a>
-<span class="sourceLineNo">542</span> }<a name="line.542"></a>
-<span class="sourceLineNo">543</span> }<a name="line.543"></a>
-<span class="sourceLineNo">544</span><a name="line.544"></a>
-<span class="sourceLineNo">545</span> public static void main(String[] args) throws Exception {<a name="line.545"></a>
-<span class="sourceLineNo">546</span> System.exit(ToolRunner.run(HBaseConfiguration.create(), new Loader(), args));<a name="line.546"></a>
-<span class="sourceLineNo">547</span> }<a name="line.547"></a>
-<span class="sourceLineNo">548</span><a name="line.548"></a>
-<span class="sourceLineNo">549</span> public static class LoaderMapper<a name="line.549"></a>
-<span class="sourceLineNo">550</span> extends Mapper<LongWritable, WARCWritable, HBaseKeyWritable, BytesWritable> {<a name="line.550"></a>
-<span class="sourceLineNo">551</span><a name="line.551"></a>
-<span class="sourceLineNo">552</span> Configuration conf;<a name="line.552"></a>
-<span class="sourceLineNo">553</span> Connection conn;<a name="line.553"></a>
-<span class="sourceLineNo">554</span> Table table;<a name="line.554"></a>
-<span class="sourceLineNo">555</span><a name="line.555"></a>
-<span class="sourceLineNo">556</span> @Override<a name="line.556"></a>
-<span class="sourceLineNo">557</span> protected void setup(Context context) throws IOException, InterruptedException {<a name="line.557"></a>
-<span class="sourceLineNo">558</span> conn = ConnectionFactory.createConnection(context.getConfiguration());<a name="line.558"></a>
-<span class="sourceLineNo">559</span> table = conn.getTable(getTablename(conn.getConfiguration()));<a name="line.559"></a>
-<span class="sourceLineNo">560</span> }<a name="line.560"></a>
-<span class="sourceLineNo">561</span><a name="line.561"></a>
-<span class="sourceLineNo">562</span> @Override<a name="line.562"></a>
-<span class="sourceLineNo">563</span> protected void cleanup(Context context) throws IOException, InterruptedException {<a name="line.563"></a>
-<span class="sourceLineNo">564</span> try {<a name="line.564"></a>
-<span class="sourceLineNo">565</span> table.close();<a name="line.565"></a>
-<span class="sourceLineNo">566</span> } catch (Exception e) {<a name="line.566"></a>
-<span class="sourceLineNo">567</span> LOG.warn("Exception closing Table", e);<a name="line.567"></a>
-<span class="sourceLineNo">568</span> }<a name="line.568"></a>
-<span class="sourceLineNo">569</span> try {<a name="line.569"></a>
-<span class="sourceLineNo">570</span> conn.close();<a name="line.570"></a>
-<span class="sourceLineNo">571</span> } catch (Exception e) {<a name="line.571"></a>
-<span class="sourceLineNo">572</span> LOG.warn("Exception closing Connection", e);<a name="line.572"></a>
-<span class="sourceLineNo">573</span> }<a name="line.573"></a>
-<span class="sourceLineNo">574</span> }<a name="line.574"></a>
-<span class="sourceLineNo">575</span><a name="line.575"></a>
-<span class="sourceLineNo">576</span> @Override<a name="line.576"></a>
-<span class="sourceLineNo">577</span> protected void map(LongWritable key, WARCWritable value, Context output)<a name="line.577"></a>
-<span class="sourceLineNo">578</span> throws IOException, InterruptedException {<a name="line.578"></a>
-<span class="sourceLineNo">579</span> WARCRecord.Header warcHeader = value.getRecord().getHeader();<a name="line.579"></a>
-<span class="sourceLineNo">580</span> String recordID = warcHeader.getRecordID();<a name="line.580"></a>
-<span class="sourceLineNo">581</span> String targetURI = warcHeader.getTargetURI();<a name="line.581"></a>
-<span class="sourceLineNo">582</span> if (warcHeader.getRecordType().equals("response") && targetURI != null) {<a name="line.582"></a>
-<span class="sourceLineNo">583</span> String contentType = warcHeader.getField("WARC-Identified-Payload-Type");<a name="line.583"></a>
-<span class="sourceLineNo">584</span> if (contentType != null) {<a name="line.584"></a>
-<span class="sourceLineNo">585</span> LOG.debug("Processing record id=" + recordID + ", targetURI=\"" + targetURI + "\"");<a name="line.585"></a>
-<span class="sourceLineNo">586</span> long now = EnvironmentEdgeManager.currentTime();<a name="line.586"></a>
-<span class="sourceLineNo">587</span><a name="line.587"></a>
-<span class="sourceLineNo">588</span> // Make row key<a name="line.588"></a>
-<span class="sourceLineNo">589</span><a name="line.589"></a>
-<span class="sourceLineNo">590</span> byte[] rowKey;<a name="line.590"></a>
-<span class="sourceLineNo">591</span> try {<a name="line.591"></a>
-<span class="sourceLineNo">592</span> rowKey = rowKeyFromTargetURI(targetURI);<a name="line.592"></a>
-<span class="sourceLineNo">593</span> } catch (IllegalArgumentException e) {<a name="line.593"></a>
-<span class="sourceLineNo">594</span> LOG.debug("Could not make a row key for record " + recordID + ", ignoring", e);<a name="line.594"></a>
-<span class="sourceLineNo">595</span> return;<a name="line.595"></a>
-<span class="sourceLineNo">596</span> } catch (URISyntaxException e) {<a name="line.596"></a>
-<span class="sourceLineNo">597</span> LOG.warn("Could not parse URI \"" + targetURI + "\" for record " + recordID +<a name="line.597"></a>
-<span class="sourceLineNo">598</span> ", ignoring");<a name="line.598"></a>
-<span class="sourceLineNo">599</span> return;<a name="line.599"></a>
-<span class="sourceLineNo">600</span> }<a name="line.600"></a>
-<span class="sourceLineNo">601</span><a name="line.601"></a>
-<span class="sourceLineNo">602</span> // Get the content and calculate the CRC64<a name="line.602"></a>
-<span class="sourceLineNo">603</span><a name="line.603"></a>
-<span class="sourceLineNo">604</span> byte[] content = value.getRecord().getContent();<a name="line.604"></a>
-<span class="sourceLineNo">605</span> CRC64 crc = new CRC64();<a name="line.605"></a>
-<span class="sourceLineNo">606</span> crc.update(content);<a name="line.606"></a>
-<span class="sourceLineNo">607</span> long crc64 = crc.getValue();<a name="line.607"></a>
-<span class="sourceLineNo">608</span><a name="line.608"></a>
-<span class="sourceLineNo">609</span> // Store to HBase<a name="line.609"></a>
-<span class="sourceLineNo">610</span><a name="line.610"></a>
-<span class="sourceLineNo">611</span> Put put = new Put(rowKey);<a name="line.611"></a>
-<span class="sourceLineNo">612</span> put.addColumn(CONTENT_FAMILY_NAME, CONTENT_QUALIFIER, now, content);<a name="line.612"></a>
-<span class="sourceLineNo">613</span> put.addColumn(INFO_FAMILY_NAME, CONTENT_LENGTH_QUALIFIER, now,<a name="line.613"></a>
-<span class="sourceLineNo">614</span> Bytes.toBytes(content.length));<a name="line.614"></a>
-<span class="sourceLineNo">615</span> put.addColumn(INFO_FAMILY_NAME, CONTENT_TYPE_QUALIFIER, now,<a name="line.615"></a>
-<span class="sourceLineNo">616</span> Bytes.toBytes(contentType));<a name="line.616"></a>
-<span class="sourceLineNo">617</span> put.addColumn(INFO_FAMILY_NAME, CRC_QUALIFIER, now, Bytes.toBytes(crc64));<a name="line.617"></a>
-<span class="sourceLineNo">618</span> put.addColumn(INFO_FAMILY_NAME, RECORD_ID_QUALIFIER, now, Bytes.toBytes(recordID));<a name="line.618"></a>
-<span class="sourceLineNo">619</span> put.addColumn(INFO_FAMILY_NAME, TARGET_URI_QUALIFIER, now, Bytes.toBytes(targetURI));<a name="line.619"></a>
-<span class="sourceLineNo">620</span> put.addColumn(INFO_FAMILY_NAME, DATE_QUALIFIER, now,<a name="line.620"></a>
-<span class="sourceLineNo">621</span> Bytes.toBytes(warcHeader.getDateString()));<a name="line.621"></a>
-<span class="sourceLineNo">622</span> String ipAddr = warcHeader.getField("WARC-IP-Address");<a name="line.622"></a>
-<span class="sourceLineNo">623</span> if (ipAddr != null) {<a name="line.623"></a>
-<span class="sourceLineNo">624</span> put.addColumn(INFO_FAMILY_NAME, IP_ADDRESS_QUALIFIER, now, Bytes.toBytes(ipAddr));<a name="line.624"></a>
-<span class="sourceLineNo">625</span> }<a name="line.625"></a>
-<span class="sourceLineNo">626</span> table.put(put);<a name="line.626"></a>
-<span class="sourceLineNo">627</span><a name="line.627"></a>
-<span class="sourceLineNo">628</span> // Write records out for later verification, one per HBase field except for the<a name="line.628"></a>
-<span class="sourceLineNo">629</span> // content record, which will be verified by CRC64.<a name="line.629"></a>
+<span class="sourceLineNo">480</span> int run(Path warcFileInput, Path outputDir)<a name="line.480"></a>
+<span class="sourceLineNo">481</span> throws IOException, ClassNotFoundException, InterruptedException {<a name="line.481"></a>
+<span class="sourceLineNo">482</span><a name="line.482"></a>
+<span class="sourceLineNo">483</span> createSchema(getTablename(getConf()));<a name="line.483"></a>
+<span class="sourceLineNo">484</span><a name="line.484"></a>
+<span class="sourceLineNo">485</span> Job job = Job.getInstance(getConf());<a name="line.485"></a>
+<span class="sourceLineNo">486</span> job.setJobName(Loader.class.getName());<a name="line.486"></a>
+<span class="sourceLineNo">487</span> job.setNumReduceTasks(0);<a name="line.487"></a>
+<span class="sourceLineNo">488</span> job.setJarByClass(getClass());<a name="line.488"></a>
+<span class="sourceLineNo">489</span> job.setMapperClass(LoaderMapper.class);<a name="line.489"></a>
+<span class="sourceLineNo">490</span> job.setInputFormatClass(WARCInputFormat.class);<a name="line.490"></a>
+<span class="sourceLineNo">491</span> FileSystem fs = FileSystem.get(warcFileInput.toUri(), getConf());<a name="line.491"></a>
+<span class="sourceLineNo">492</span> if (fs.getFileStatus(warcFileInput).isDirectory()) {<a name="line.492"></a>
+<span class="sourceLineNo">493</span> LOG.info("Using directory as WARC input path: " + warcFileInput);<a name="line.493"></a>
+<span class="sourceLineNo">494</span> FileInputFormat.setInputPaths(job, warcFileInput);<a name="line.494"></a>
+<span class="sourceLineNo">495</span> } else {<a name="line.495"></a>
+<span class="sourceLineNo">496</span> LOG.info("Getting WARC input paths from file: " + warcFileInput);<a name="line.496"></a>
+<span class="sourceLineNo">497</span> List<Path> paths = new LinkedList<Path>();<a name="line.497"></a>
+<span class="sourceLineNo">498</span> try (FSDataInputStream is = fs.open(warcFileInput)) {<a name="line.498"></a>
+<span class="sourceLineNo">499</span> InputStreamReader reader;<a name="line.499"></a>
+<span class="sourceLineNo">500</span> if (warcFileInput.getName().toLowerCase().endsWith(".gz")) {<a name="line.500"></a>
+<span class="sourceLineNo">501</span> reader = new InputStreamReader(new GZIPInputStream(is));<a name="line.501"></a>
+<span class="sourceLineNo">502</span> } else {<a name="line.502"></a>
+<span class="sourceLineNo">503</span> reader = new InputStreamReader(is);<a name="line.503"></a>
+<span class="sourceLineNo">504</span> }<a name="line.504"></a>
+<span class="sourceLineNo">505</span> try (BufferedReader br = new BufferedReader(reader)) {<a name="line.505"></a>
+<span class="sourceLineNo">506</span> String line;<a name="line.506"></a>
+<span class="sourceLineNo">507</span> while ((line = br.readLine()) != null) {<a name="line.507"></a>
+<span class="sourceLineNo">508</span> paths.add(new Path(line));<a name="line.508"></a>
+<span class="sourceLineNo">509</span> }<a name="line.509"></a>
+<span class="sourceLineNo">510</span> }<a name="line.510"></a>
+<span class="sourceLineNo">511</span> }<a name="line.511"></a>
+<span class="sourceLineNo">512</span> LOG.info("Read " + paths.size() + " WARC input paths from " + warcFileInput);<a name="line.512"></a>
+<span class="sourceLineNo">513</span> FileInputFormat.setInputPaths(job, paths.toArray(new Path[paths.size()]));<a name="line.513"></a>
+<span class="sourceLineNo">514</span> }<a name="line.514"></a>
+<span class="sourceLineNo">515</span> job.setOutputFormatClass(SequenceFileOutputFormat.class);<a name="line.515"></a>
+<span class="sourceLineNo">516</span> SequenceFileOutputFormat.setOutputPath(job, outputDir);<a name="line.516"></a>
+<span class="sourceLineNo">517</span> SequenceFileOutputFormat.setOutputCompressionType(job, CompressionType.BLOCK);<a name="line.517"></a>
+<span class="sourceLineNo">518</span> job.setOutputKeyClass(HBaseKeyWritable.class);<a name="line.518"></a>
+<span class="sourceLineNo">519</span> job.setOutputValueClass(BytesWritable.class);<a name="line.519"></a>
+<span class="sourceLineNo">520</span> TableMapReduceUtil.addDependencyJars(job);<a name="line.520"></a>
+<span class="sourceLineNo">521</span><a name="line.521"></a>
+<span class="sourceLineNo">522</span> boolean success = job.waitForCompletion(true);<a name="line.522"></a>
+<span class="sourceLineNo">523</span> if (!success) {<a name="line.523"></a>
+<span class="sourceLineNo">524</span> LOG.error("Failure during job " + job.getJobID());<a name="line.524"></a>
+<span class="sourceLineNo">525</span> }<a name="line.525"></a>
+<span class="sourceLineNo">526</span> return success ? 0 : 1;<a name="line.526"></a>
+<span class="sourceLineNo">527</span> }<a name="line.527"></a>
+<span class="sourceLineNo">528</span><a name="line.528"></a>
+<span class="sourceLineNo">529</span> @Override<a name="line.529"></a>
+<span class="sourceLineNo">530</span> public int run(String[] args) throws Exception {<a name="line.530"></a>
+<span class="sourceLineNo">531</span> if (args.length < 2) {<a name="line.531"></a>
+<span class="sourceLineNo">532</span> System.err.println(USAGE);<a name="line.532"></a>
+<span class="sourceLineNo">533</span> return 1;<a name="line.533"></a>
+<span class="sourceLineNo">534</span> }<a name="line.534"></a>
+<span class="sourceLineNo">535</span> try {<a name="line.535"></a>
+<span class="sourceLineNo">536</span> Path warcFileInput = new Path(args[0]);<a name="line.536"></a>
+<span class="sourceLineNo">537</span> Path outputDir = new Path(args[1]);<a name="line.537"></a>
+<span class="sourceLineNo">538</span> return run(warcFileInput, outputDir);<a name="line.538"></a>
+<span class="sourceLineNo">539</span> } catch (NumberFormatException e) {<a name="line.539"></a>
+<span class="sourceLineNo">540</span> System.err.println("Parsing loader arguments failed: " + e.getMessage());<a name="line.540"></a>
+<span class="sourceLineNo">541</span> System.err.println(USAGE);<a name="line.541"></a>
+<span class="sourceLineNo">542</span> return 1;<a name="line.542"></a>
+<span class="sourceLineNo">543</span> }<a name="line.543"></a>
+<span class="sourceLineNo">544</span> }<a name="line.544"></a>
+<span class="sourceLineNo">545</span><a name="line.545"></a>
+<span class="sourceLineNo">546</span> public static void main(String[] args) throws Exception {<a name="line.546"></a>
+<span class="sourceLineNo">547</span> System.exit(ToolRunner.run(HBaseConfiguration.create(), new Loader(), args));<a name="line.547"></a>
+<span class="sourceLineNo">548</span> }<a name="line.548"></a>
+<span class="sourceLineNo">549</span><a name="line.549"></a>
+<span class="sourceLineNo">550</span> public static class LoaderMapper<a name="line.550"></a>
+<span class="sourceLineNo">551</span> extends Mapper<LongWritable, WARCWritable, HBaseKeyWritable, BytesWritable> {<a name="line.551"></a>
+<span class="sourceLineNo">552</span><a name="line.552"></a>
+<span class="sourceLineNo">553</span> Configuration conf;<a name="line.553"></a>
+<span class="sourceLineNo">554</span> Connection conn;<a name="line.554"></a>
+<span class="sourceLineNo">555</span> BufferedMutator mutator;<a name="line.555"></a>
+<span class="sourceLineNo">556</span><a name="line.556"></a>
+<span class="sourceLineNo">557</span> @Override<a name="line.557"></a>
+<span class="sourceLineNo">558</span> protected void setup(Context context) throws IOException, InterruptedException {<a name="line.558"></a>
+<span class="sourceLineNo">559</span> conf = context.getConfiguration();<a name="line.559"></a>
+<span class="sourceLineNo">560</span> conn = ConnectionFactory.createConnection(conf);<a name="line.560"></a>
+<span class="sourceLineNo">561</span> mutator = conn.getBufferedMutator(getTablename(conf));<a name="line.561"></a>
+<span class="sourceLineNo">562</span> mutator.setWriteBufferPeriodicFlush(10 * 1000); // default is 1 sec, increase to 10<a name="line.562"></a>
+<span class="sourceLineNo">563</span> }<a name="line.563"></a>
+<span class="sourceLineNo">564</span><a name="line.564"></a>
+<span class="sourceLineNo">565</span> @Override<a name="line.565"></a>
+<span class="sourceLineNo">566</span> protected void cleanup(Context context) throws IOException, InterruptedException {<a name="line.566"></a>
+<span class="sourceLineNo">567</span> try {<a name="line.567"></a>
+<span class="sourceLineNo">568</span> mutator.close();<a name="line.568"></a>
+<span class="sourceLineNo">569</span> } catch (Exception e) {<a name="line.569"></a>
+<span class="sourceLineNo">570</span> LOG.warn("Exception closing Table", e);<a name="line.570"></a>
+<span class="sourceLineNo">571</span> }<a name="line.571"></a>
+<span class="sourceLineNo">572</span> try {<a name="line.572"></a>
+<span class="sourceLineNo">573</span> conn.close();<a name="line.573"></a>
+<span class="sourceLineNo">574</span> } catch (Exception e) {<a name="line.574"></a>
+<span class="sourceLineNo">575</span> LOG.warn("Exception closing Connection", e);<a name="line.575"></a>
+<span class="sourceLineNo">576</span> }<a name="line.576"></a>
+<span class="sourceLineNo">577</span> }<a name="line.577"></a>
+<span class="sourceLineNo">578</span><a name="line.578"></a>
+<span class="sourceLineNo">579</span> @Override<a name="line.579"></a>
+<span class="sourceLineNo">580</span> protected void map(LongWritable key, WARCWritable value, Context output)<a name="line.580"></a>
+<span class="sourceLineNo">581</span> throws IOException, InterruptedException {<a name="line.581"></a>
+<span class="sourceLineNo">582</span> WARCRecord.Header warcHeader = value.getRecord().getHeader();<a name="line.582"></a>
+<span class="sourceLineNo">583</span> String recordID = warcHeader.getRecordID();<a name="line.583"></a>
+<span class="sourceLineNo">584</span> String targetURI = warcHeader.getTargetURI();<a name="line.584"></a>
+<span class="sourceLineNo">585</span> if (warcHeader.getRecordType().equals("response") && targetURI != null) {<a name="line.585"></a>
+<span class="sourceLineNo">586</span> String contentType = warcHeader.getField("WARC-Identified-Payload-Type");<a name="line.586"></a>
+<span class="sourceLineNo">587</span> if (contentType != null) {<a name="line.587"></a>
+<span class="sourceLineNo">588</span> LOG.info("Processing uri=\"" + targetURI + "\", id=" + recordID);<a name="line.588"></a>
+<span class="sourceLineNo">589</span> long now = EnvironmentEdgeManager.currentTime();<a name="line.589"></a>
+<span class="sourceLineNo">590</span><a name="line.590"></a>
+<span class="sourceLineNo">591</span> // Make row key<a name="line.591"></a>
+<span class="sourceLineNo">592</span><a name="line.592"></a>
+<span class="sourceLineNo">593</span> byte[] rowKey;<a name="line.593"></a>
+<span class="sourceLineNo">594</span> try {<a name="line.594"></a>
+<span class="sourceLineNo">595</span> rowKey = rowKeyFromTargetURI(targetURI);<a name="line.595"></a>
+<span class="sourceLineNo">596</span> } catch (IllegalArgumentException e) {<a name="line.596"></a>
+<span class="sourceLineNo">597</span> LOG.debug("Could not make a row key for record " + recordID + ", ignoring", e);<a name="line.597"></a>
+<span class="sourceLineNo">598</span> return;<a name="line.598"></a>
+<span class="sourceLineNo">599</span> } catch (URISyntaxException e) {<a name="line.599"></a>
+<span class="sourceLineNo">600</span> LOG.warn("Could not parse URI \"" + targetURI + "\" for record " + recordID +<a name="line.600"></a>
+<span class="sourceLineNo">601</span> ", ignoring");<a name="line.601"></a>
+<span class="sourceLineNo">602</span> return;<a name="line.602"></a>
+<span class="sourceLineNo">603</span> }<a name="line.603"></a>
+<span class="sourceLineNo">604</span><a name="line.604"></a>
+<span class="sourceLineNo">605</span> // Get the content and calculate the CRC64<a name="line.605"></a>
+<span class="sourceLineNo">606</span><a name="line.606"></a>
+<span class="sourceLineNo">607</span> byte[] content = value.getRecord().getContent();<a name="line.607"></a>
+<span class="sourceLineNo">608</span> CRC64 crc = new CRC64();<a name="line.608"></a>
+<span class="sourceLineNo">609</span> crc.update(content);<a name="line.609"></a>
+<span class="sourceLineNo">610</span> long crc64 = crc.getValue();<a name="line.610"></a>
+<span class="sourceLineNo">611</span><a name="line.611"></a>
+<span class="sourceLineNo">612</span> // Store to HBase<a name="line.612"></a>
+<span class="sourceLineNo">613</span><a name="line.613"></a>
+<span class="sourceLineNo">614</span> Put put = new Put(rowKey);<a name="line.614"></a>
+<span class="sourceLineNo">615</span> put.addColumn(CONTENT_FAMILY_NAME, CONTENT_QUALIFIER, now, content);<a name="line.615"></a>
+<span class="sourceLineNo">616</span> put.addColumn(INFO_FAMILY_NAME, CONTENT_LENGTH_QUALIFIER, now,<a name="line.616"></a>
+<span class="sourceLineNo">617</span> Bytes.toBytes(content.length));<a name="line.617"></a>
+<span class="sourceLineNo">618</span> put.addColumn(INFO_FAMILY_NAME, CONTENT_TYPE_QUALIFIER, now,<a name="line.618"></a>
+<span class="sourceLineNo">619</span> Bytes.toBytes(contentType));<a name="line.619"></a>
+<span class="sourceLineNo">620</span> put.addColumn(INFO_FAMILY_NAME, CRC_QUALIFIER, now, Bytes.toBytes(crc64));<a name="line.620"></a>
+<span class="sourceLineNo">621</span> put.addColumn(INFO_FAMILY_NAME, RECORD_ID_QUALIFIER, now, Bytes.toBytes(recordID));<a name="line.621"></a>
+<span class="sourceLineNo">622</span> put.addColumn(INFO_FAMILY_NAME, TARGET_URI_QUALIFIER, now, Bytes.toBytes(targetURI));<a name="line.622"></a>
+<span class="sourceLineNo">623</span> put.addColumn(INFO_FAMILY_NAME, DATE_QUALIFIER, now,<a name="line.623"></a>
+<span class="sourceLineNo">624</span> Bytes.toBytes(warcHeader.getDateString()));<a name="line.624"></a>
+<span class="sourceLineNo">625</span> String ipAddr = warcHeader.getField("WARC-IP-Address");<a name="line.625"></a>
+<span class="sourceLineNo">626</span> if (ipAddr != null) {<a name="line.626"></a>
+<span class="sourceLineNo">627</span> put.addColumn(INFO_FAMILY_NAME, IP_ADDRESS_QUALIFIER, now, Bytes.toBytes(ipAddr));<a name="line.627"></a>
+<span class="sourceLineNo">628</span> }<a name="line.628"></a>
+<span class="sourceLineNo">629</span> mutator.mutate(put);<a name="line.629"></a>
<span class="sourceLineNo">630</span><a name="line.630"></a>
-<span class="sourceLineNo">631</span> output.write(new HBaseKeyWritable(rowKey, INFO_FAMILY_NAME, CRC_QUALIFIER, now),<a name="line.631"></a>
-<span class="sourceLineNo">632</span> new BytesWritable(Bytes.toBytes(crc64)));<a name="line.632"></a>
-<span class="sourceLineNo">633</span> output.write(new HBaseKeyWritable(rowKey, INFO_FAMILY_NAME, CONTENT_LENGTH_QUALIFIER,<a name="line.633"></a>
-<span class="sourceLineNo">634</span> now), new BytesWritable(Bytes.toBytes(content.length)));<a name="line.634"></a>
-<span class="sourceLineNo">635</span> output.write(new HBaseKeyWritable(rowKey, INFO_FAMILY_NAME, CONTENT_TYPE_QUALIFIER,<a name="line.635"></a>
-<span class="sourceLineNo">636</span> now), new BytesWritable(Bytes.toBytes(contentType)));<a name="line.636"></a>
-<span class="sourceLineNo">637</span> output.write(new HBaseKeyWritable(rowKey, INFO_FAMILY_NAME, RECORD_ID_QUALIFIER,<a name="line.637"></a>
-<span class="sourceLineNo">638</span> now), new BytesWritable(Bytes.toBytes(recordID)));<a name="line.638"></a>
-<span class="sourceLineNo">639</span> output.write(new HBaseKeyWritable(rowKey, INFO_FAMILY_NAME, TARGET_URI_QUALIFIER,<a name="line.639"></a>
-<span class="sourceLineNo">640</span> now), new BytesWritable(Bytes.toBytes(targetURI)));<a name="line.640"></a>
-<span class="sourceLineNo">641</span> output.write(new HBaseKeyWritable(rowKey, INFO_FAMILY_NAME, DATE_QUALIFIER, now),<a name="line.641"></a>
-<span class="sourceLineNo">642</span> new BytesWritable(Bytes.toBytes(warcHeader.getDateString())));<a name="line.642"></a>
-<span class="sourceLineNo">643</span> if (ipAddr != null) {<a name="line.643"></a>
-<span class="sourceLineNo">644</span> output.write(new HBaseKeyWritable(rowKey, INFO_FAMILY_NAME, IP_ADDRESS_QUALIFIER,<a name="line.644"></a>
-<span class="sourceLineNo">645</span> now), new BytesWritable(Bytes.toBytes(ipAddr)));<a name="line.645"></a>
-<span class="sourceLineNo">646</span> }<a name="line.646"></a>
-<span class="sourceLineNo">647</span> }<a name="line.647"></a>
-<span class="sourceLineNo">648</span> }<a name="line.648"></a>
-<span class="sourceLineNo">649</span> }<a name="line.649"></a>
-<span class="sourceLineNo">650</span><a name="line.650"></a>
-<span class="sourceLineNo">651</span> private byte[] rowKeyFromTargetURI(String targetUri)<a name="line.651"></a>
-<span class="sourceLineNo">652</span> throws URISyntaxException, IllegalArgumentException {<a name="line.652"></a>
-<span class="sourceLineNo">653</span> URI uri = new URI(targetUri);<a name="line.653"></a>
-<span class="sourceLineNo">654</span> StringBuffer sb = new StringBuffer();<a name="line.654"></a>
-<span class="sourceLineNo">655</span> // Ignore the scheme<a name="line.655"></a>
-<span class="sourceLineNo">656</span> // Reverse the components of the hostname<a name="line.656"></a>
-<span class="sourceLineNo">657</span> if (uri.getHost() != null) {<a name="line.657"></a>
-<span class="sourceLineNo">658</span> String[] hostComponents = uri.getHost().split("\\.");<a name="line.658"></a>
-<span class="sourceLineNo">659</span> for (int i = hostComponents.length - 1; i >= 0; i--) {<a name="line.659"></a>
-<span class="sourceLineNo">660</span> sb.append(hostComponents[i]);<a name="line.660"></a>
-<span class="sourceLineNo">661</span> if (i != 0) {<a name="line.661"></a>
-<span class="sourceLineNo">662</span> sb.append('.');<a name="line.662"></a>
-<span class="sourceLineNo">663</span> }<a name="line.663"></a>
-<span class="sourceLineNo">664</span> }<a name="line.664"></a>
-<span class="sourceLineNo">665</span> } else {<a name="line.665"></a>
-<span class="sourceLineNo">666</span> throw new IllegalArgumentException("URI is missing host component");<a name="line.666"></a>
-<span class="sourceLineNo">667</span> }<a name="line.667"></a>
-<span class="sourceLineNo">668</span> // Port<a name="line.668"></a>
-<span class="sourceLineNo">669</span> if (uri.getPort() != -1) {<a name="line.669"></a>
-<span class="sourceLineNo">670</span> sb.append(':');<a name="line.670"></a>
-<span class="sourceLineNo">671</span> sb.append(uri.getPort());<a name="line.671"></a>
+<span class="sourceLineNo">631</span> // Write records out for later verification, one per HBase field except for the<a name="line.631"></a>
+<span class="sourceLineNo">632</span> // content record, which will be verified by CRC64.<a name="line.632"></a>
+<span class="sourceLineNo">633</span><a name="line.633"></a>
+<span class="sourceLineNo">634</span> output.write(new HBaseKeyWritable(rowKey, INFO_FAMILY_NAME, CRC_QUALIFIER, now),<a name="line.634"></a>
+<span class="sourceLineNo">635</span> new BytesWritable(Bytes.toBytes(crc64)));<a name="line.635"></a>
+<span class="sourceLineNo">636</span> output.write(new HBaseKeyWritable(rowKey, INFO_FAMILY_NAME, CONTENT_LENGTH_QUALIFIER,<a name="line.636"></a>
+<span class="sourceLineNo">637</span> now), new BytesWritable(Bytes.toBytes(content.length)));<a name="line.637"></a>
+<span class="sourceLineNo">638</span> output.write(new HBaseKeyWritable(rowKey, INFO_FAMILY_NAME, CONTENT_TYPE_QUALIFIER,<a name="line.638"></a>
+<span class="sourceLineNo">639</span> now), new BytesWritable(Bytes.toBytes(contentType)));<a name="line.639"></a>
+<span class="sourceLineNo">640</span> output.write(new HBaseKeyWritable(rowKey, INFO_FAMILY_NAME, RECORD_ID_QUALIFIER,<a name="line.640"></a>
+<span class="sourceLineNo">641</span> now), new BytesWritable(Bytes.toBytes(recordID)));<a name="line.641"></a>
+<span class="sourceLineNo">642</span> output.write(new HBaseKeyWritable(rowKey, INFO_FAMILY_NAME, TARGET_URI_QUALIFIER,<a name="line.642"></a>
+<span class="sourceLineNo">643</span> now), new BytesWritable(Bytes.toBytes(targetURI)));<a name="line.643"></a>
+<span class="sourceLineNo">644</span> output.write(new HBaseKeyWritable(rowKey, INFO_FAMILY_NAME, DATE_QUALIFIER, now),<a name="line.644"></a>
+<span class="sourceLineNo">645</span> new BytesWritable(Bytes.toBytes(warcHeader.getDateString())));<a name="line.645"></a>
+<span class="sourceLineNo">646</span> if (ipAddr != null) {<a name="line.646"></a>
+<span class="sourceLineNo">647</span> output.write(new HBaseKeyWritable(rowKey, INFO_FAMILY_NAME, IP_ADDRESS_QUALIFIER,<a name="line.647"></a>
+<span class="sourceLineNo">648</span> now), new BytesWritable(Bytes.toBytes(ipAddr)));<a name="line.648"></a>
+<span class="sourceLineNo">649</span> }<a name="line.649"></a>
+<span class="sourceLineNo">650</span> }<a name="line.650"></a>
+<span class="sourceLineNo">651</span> }<a name="line.651"></a>
+<span class="sourceLineNo">652</span> }<a name="line.652"></a>
+<span class="sourceLineNo">653</span><a name="line.653"></a>
+<span class="sourceLineNo">654</span> private byte[] rowKeyFromTargetURI(String targetUri)<a name="line.654"></a>
+<span class="sourceLineNo">655</span> throws URISyntaxException, IllegalArgumentException {<a name="line.655"></a>
+<span class="sourceLineNo">656</span> URI uri = new URI(targetUri);<a name="line.656"></a>
+<span class="sourceLineNo">657</span> // Ignore the scheme<a name="line.657"></a>
+<span class="sourceLineNo">658</span> // Reverse the components of the hostname<a name="line.658"></a>
+<span class="sourceLineNo">659</span> String reversedHost;<a name="line.659"></a>
+<span class="sourceLineNo">660</span> if (uri.getHost() != null) {<a name="line.660"></a>
+<span class="sourceLineNo">661</span> StringBuffer sb = new StringBuffer();<a name="line.661"></a>
+<span class="sourceLineNo">662</span> String[] hostComponents = uri.getHost().split("\\.");<a name="line.662"></a>
+<span class="sourceLineNo">663</span> for (int i = hostComponents.length - 1; i >= 0; i--) {<a name="line.663"></a>
+<span class="sourceLineNo">664</span> sb.append(hostComponents[i]);<a name="line.664"></a>
+<span class="sourceLineNo">665</span> if (i != 0) {<a name="line.665"></a>
+<span class="sourceLineNo">666</span> sb.append('.');<a name="line.666"></a>
+<span class="sourceLineNo">667</span> }<a name="line.667"></a>
+<span class="sourceLineNo">668</span> }<a name="line.668"></a>
+<span class="sourceLineNo">669</span> reversedHost = sb.toString();<a name="line.669"></a>
+<span class="sourceLineNo">670</span> } else {<a name="line.670"></a>
+<span class="sourceLineNo">671</span> throw new IllegalArgumentException("URI is missing host component");<a name="line.671"></a>
<span class="sourceLineNo">672</span> }<a name="line.672"></a>
-<span class="sourceLineNo">673</span> if (uri.getRawPath() != null) {<a name="line.673"></a>
-<span class="sourceLineNo">674</span> sb.append(uri.getRawPath());<a name="line.674"></a>
-<span class="sourceLineNo">675</span> }<a name="line.675"></a>
-<span class="sourceLineNo">676</span> if (uri.getRawQuery() != null) {<a name="line.676"></a>
-<span class="sourceLineNo">677</span> sb.append('?');<a name="line.677"></a>
-<span class="sourceLineNo">678</span> sb.append(uri.getRawQuery());<a name="line.678"></a>
-<span class="sourceLineNo">679</span> }<a name="line.679"></a>
-<span class="sourceLineNo">680</span> if (uri.getRawFragment() != null) {<a name="line.680"></a>
-<span class="sourceLineNo">681</span> sb.append('#');<a name="line.681"></a>
-<span class="sourceLineNo">682</span> sb.append(uri.getRawFragment());<a name="line.682"></a>
-<span class="sourceLineNo">683</span> }<a name="line.683"></a>
-<span class="sourceLineNo">684</span> // Constrain the key size to the maximum allowed row key length<a name="line.684"></a>
-<span class="sourceLineNo">685</span> if (sb.length() > HConstants.MAX_ROW_LENGTH) {<a name="line.685"></a>
-<span class="sourceLineNo">686</span> sb.setLength(HConstants.MAX_ROW_LENGTH);<a name="line.686"></a>
-<span class="sourceLineNo">687</span> }<a name="line.687"></a>
-<span class="sourceLineNo">688</span> return Bytes.toBytes(sb.toString());<a name="line.688"></a>
-<span class="sourceLineNo">689</span> }<a name="line.689"></a>
-<span class="sourceLineNo">690</span><a name="line.690"></a>
-<span class="sourceLineNo">691</span> }<a name="line.691"></a>
-<span class="sourceLineNo">692</span> }<a name="line.692"></a>
-<span class="sourceLineNo">693</span><a name="line.693"></a>
-<span class="sourceLineNo">694</span> public static class OneFilePerMapperSFIF<K, V> extends SequenceFileInputFormat<K, V> {<a name="line.694"></a>
-<span class="sourceLineNo">695</span> @Override<a name="line.695"></a>
-<span class="sourceLineNo">696</span> protected boolean isSplitable(JobContext context, Path filename) {<a name="line.696"></a>
-<span class="sourceLineNo">697</span> return false;<a name="line.697"></a>
+<span class="sourceLineNo">673</span> StringBuffer sb = new StringBuffer();<a name="line.673"></a>
+<span class="sourceLineNo">674</span> sb.append(reversedHost);<a name="line.674"></a>
+<span class="sourceLineNo">675</span> if (uri.getPort() >= 0) {<a name="line.675"></a>
+<span class="sourceLineNo">676</span> sb.append(':');<a name="line.676"></a>
+<span class="sourceLineNo">677</span> sb.append(uri.getPort());<a name="line.677"></a>
+<span class="sourceLineNo">678</span> }<a name="line.678"></a>
+<span class="sourceLineNo">679</span> if (uri.getPath() != null) {<a name="line.679"></a>
+<span class="sourceLineNo">680</span> sb.append('/');<a name="line.680"></a>
+<span class="sourceLineNo">681</span> sb.append(uri.getPath());<a name="line.681"></a>
+<span class="sourceLineNo">682</span> }<a name="line.682"></a>
+<span class="sourceLineNo">683</span> if (uri.getQuery() != null) {<a name="line.683"></a>
+<span class="sourceLineNo">684</span> sb.append('?');<a name="line.684"></a>
+<span class="sourceLineNo">685</span> sb.append(uri.getQuery());<a name="line.685"></a>
+<span class="sourceLineNo">686</span> }<a name="line.686"></a>
+<span class="sourceLineNo">687</span> if (uri.getFragment() != null) {<a name="line.687"></a>
+<span class="sourceLineNo">688</span> sb.append('#');<a name="line.688"></a>
+<span class="sourceLineNo">689</span> sb.append(uri.getFragment());<a name="line.689"></a>
+<span class="sourceLineNo">690</span> }<a name="line.690"></a>
+<span class="sourceLineNo">691</span> if (sb.length() > HConstants.MAX_ROW_LENGTH) {<a name="line.691"></a>
+<span class="sourceLineNo">692</span> throw new IllegalArgumentException("Key would be too large (length=" + sb.length() +<a name="line.692"></a>
+<span class="sourceLineNo">693</span> ", limit=" + HConstants.MAX_ROW_LENGTH);<a name="line.693"></a>
+<span class="sourceLineNo">694</span> }<a name="line.694"></a>
+<span class="sourceLineNo">695</span> return Bytes.toBytes(sb.toString());<a name="line.695"></a>
+<span class="sourceLineNo">696</span> }<a name="line.696"></a>
+<span class="sourceLineNo">697</span><a name="line.697"></a>
<span class="sourceLineNo">698</span> }<a name="line.698"></a>
<span class="sourceLineNo">699</span> }<a name="line.699"></a>
<span class="sourceLineNo">700</span><a name="line.700"></a>
-<span class="sourceLineNo">701</span> public static class Verify extends Configured implements Tool {<a name="line.701"></a>
-<span class="sourceLineNo">702</span><a name="line.702"></a>
-<span class="sourceLineNo">703</span> public static final Logger LOG = LoggerFactory.getLogger(Verify.class);<a name="line.703"></a>
-<span class="sourceLineNo">704</span> public static final String USAGE = "Verify <inputDir>";<a name="line.704"></a>
-<span class="sourceLineNo">705</span><a name="line.705"></a>
-<span class="sourceLineNo">706</span> int run(Path inputDir) throws IOException, ClassNotFoundException, InterruptedException {<a name="line.706"></a>
-<span class="sourceLineNo">707</span> Job job = Job.getInstance(getConf());<a name="line.707"></a>
-<span class="sourceLineNo">708</span> job.setJobName(Verify.class.getName());<a name="line.708"></a>
-<span class="sourceLineNo">709</span> job.setJarByClass(getClass());<a name="line.709"></a>
-<span class="sourceLineNo">710</span> job.setMapperClass(VerifyMapper.class);<a name="line.710"></a>
-<span class="sourceLineNo">711</span> job.setInputFormatClass(OneFilePerMapperSFIF.class);<a name="line.711"></a>
-<span class="sourceLineNo">712</span> FileInputFormat.setInputPaths(job, inputDir);<a name="line.712"></a>
-<span class="sourceLineNo">713</span> job.setOutputFormatClass(NullOutputFormat.class);<a name="line.713"></a>
-<span class="sourceLineNo">714</span> job.setOutputKeyClass(NullWritable.class);<a name="line.714"></a>
-<span class="sourceLineNo">715</span> job.setOutputValueClass(NullWritable.class);<a name="line.715"></a>
-<span class="sourceLineNo">716</span> TableMapReduceUtil.addDependencyJars(job);<a name="line.716"></a>
-<span class="sourceLineNo">717</span> boolean success = job.waitForCompletion(true);<a name="line.717"></a>
-<span class="sourceLineNo">718</span> if (!success) {<a name="line.718"></a>
-<span class="sourceLineNo">719</span> LOG.error("Failure during job " + job.getJobID());<a name="line.719"></a>
-<span class="sourceLineNo">720</span> }<a name="line.720"></a>
-<span class="sourceLineNo">721</span> Counters counters = job.getCounters();<a name="line.721"></a>
-<span class="sourceLineNo">722</span> for (Counts c: Counts.values()) {<a name="line.722"></a>
-<span class="sourceLineNo">723</span> LOG.info(c + ": " + counters.findCounter(c).getValue());<a name="line.723"></a>
-<span class="sourceLineNo">724</span> }<a name="line.724"></a>
-<span class="sourceLineNo">725</span> return success ? 0 : 1;<a name="line.725"></a>
-<span class="sourceLineNo">726</span> }<a name="line.726"></a>
-<span class="sourceLineNo">727</span><a name="line.727"></a>
-<span class="sourceLineNo">728</span> @Override<a name="line.728"></a>
-<span class="sourceLineNo">729</span> public int run(String[] args) throws Exception {<a name="line.729"></a>
-<span class="sourceLineNo">730</span> if (args.length < 2) {<a name="line.730"></a>
-<span class="sourceLineNo">731</span> System.err.println(USAGE);<a name="line.731"></a>
-<span class="sourceLineNo">732</span> return 1;<a name="line.732"></a>
-<span class="sourceLineNo">733</span> }<a name="line.733"></a>
-<span class="sourceLineNo">734</span> Path loaderOutput = new Path(args[0]);<a name="line.734"></a>
-<span class="sourceLineNo">735</span> return run(loaderOutput);<a name="line.735"></a>
-<span class="sourceLineNo">736</span> }<a name="line.736"></a>
-<span class="sourceLineNo">737</span><a name="line.737"></a>
-<span class="sourceLineNo">738</span> public static void main(String[] args) throws Exception {<a name="line.738"></a>
-<span class="sourceLineNo">739</span> System.exit(ToolRunner.run(HBaseConfiguration.create(), new Verify(), args));<a name="line.739"></a>
-<span class="sourceLineNo">740</span> }<a name="line.740"></a>
-<span class="sourceLineNo">741</span><a name="line.741"></a>
-<span class="sourceLineNo">742</span> public static class VerifyMapper<a name="line.742"></a>
-<span class="sourceLineNo">743</span> extends Mapper<HBaseKeyWritable, BytesWritable, NullWritable, NullWritable> {<a name="line.743"></a>
+<span class="sourceLineNo">701</span> public static class OneFilePerMapperSFIF<K, V> extends SequenceFileInputFormat<K, V> {<a name="line.701"></a>
+<span class="sourceLineNo">702</span> @Override<a name="line.702"></a>
+<span class="sourceLineNo">703</span> protected boolean isSplitable(JobContext context, Path filename) {<a name="line.703"></a>
+<span class="sourceLineNo">704</span> return false;<a name="line.704"></a>
+<span class="sourceLineNo">705</span> }<a name="line.705"></a>
+<span class="sourceLineNo">706</span> }<a name="line.706"></a>
+<span class="sourceLineNo">707</span><a name="line.707"></a>
+<span class="sourceLineNo">708</span> public static class Verify extends Configured implements Tool {<a name="line.708"></a>
+<span class="sourceLineNo">709</span><a name="line.709"></a>
+<span class="sourceLineNo">710</span> public static final Logger LOG = LoggerFactory.getLogger(Verify.class);<a name="line.710"></a>
+<span class="sourceLineNo">711</span> public static final String USAGE = "Verify <inputDir>";<a name="line.711"></a>
+<span class="sourceLineNo">712</span><a name="line.712"></a>
+<span class="sourceLineNo">713</span> int run(Path inputDir) throws IOException, ClassNotFoundException, InterruptedException {<a name="line.713"></a>
+<span class="sourceLineNo">714</span> Job job = Job.getInstance(getConf());<a name="line.714"></a>
+<span class="sourceLineNo">715</span> job.setJobName(Verify.class.getName());<a name="line.715"></a>
+<span class="sourceLineNo">716</span> job.setJarByClass(getClass());<a name="line.716"></a>
+<span class="sourceLineNo">717</span> job.setMapperClass(VerifyMapper.class);<a name="line.717"></a>
+<span class="sourceLineNo">718</span> job.setInputFormatClass(OneFilePerMapperSFIF.class);<a name="line.718"></a>
+<span class="sourceLineNo">719</span> FileInputFormat.setInputPaths(job, inputDir);<a name="line.719"></a>
+<span class="sourceLineNo">720</span> job.setOutputFormatClass(NullOutputFormat.class);<a name="line.720"></a>
+<span class="sourceLineNo">721</span> job.setOutputKeyClass(NullWritable.class);<a name="line.721"></a>
+<span class="sourceLineNo">722</span> job.setOutputValueClass(NullWritable.class);<a name="line.722"></a>
+<span class="sourceLineNo">723</span> TableMapReduceUtil.addDependencyJars(job);<a name="line.723"></a>
+<span class="sourceLineNo">724</span> boolean success = job.waitForCompletion(true);<a name="line.724"></a>
+<span class="sourceLineNo">725</span> if (!success) {<a name="line.725"></a>
+<span class="sourceLineNo">726</span> LOG.error("Failure during job " + job.getJobID());<a name="line.726"></a>
+<span class="sourceLineNo">727</span> }<a name="line.727"></a>
+<span class="sourceLineNo">728</span> Counters counters = job.getCounters();<a name="line.728"></a>
+<span class="sourceLineNo">729</span> for (Counts c: Counts.values()) {<a name="line.729"></a>
+<span class="sourceLineNo">730</span> LOG.info(c + ": " + counters.findCounter(c).getValue());<a name="line.730"></a>
+<span class="sourceLineNo">731</span> }<a name="line.731"></a>
+<span class="sourceLineNo">732</span> return success ? 0 : 1;<a name="line.732"></a>
+<span class="sourceLineNo">733</span> }<a name="line.733"></a>
+<span class="sourceLineNo">734</span><a name="line.734"></a>
+<span class="sourceLineNo">735</span> @Override<a name="line.735"></a>
+<span class="sourceLineNo">736</span> public int run(String[] args) throws Exception {<a name="line.736"></a>
+<span class="sourceLineNo">737</span> if (args.length < 2) {<a name="line.737"></a>
+<span class="sourceLineNo">738</span> System.err.println(USAGE);<a name="line.738"></a>
+<span class="sourceLineNo">739</span> return 1;<a name="line.739"></a>
+<span class="sourceLineNo">740</span> }<a name="line.740"></a>
+<span class="sourceLineNo">741</span> Path loaderOutput = new Path(args[0]);<a name="line.741"></a>
+<span class="sourceLineNo">742</span> return run(loaderOutput);<a name="line.742"></a>
+<span class="sourceLineNo">743</span> }<a name="line.743"></a>
<span class="sourceLineNo">744</span><a name="line.744"></a>
-<span class="sourceLineNo">745</span> Connection conn;<a name="line.745"></a>
-<span class="sourceLineNo">746</span> Table table;<a name="line.746"></a>
-<span class="sourceLineNo">747</span><a name="line.747"></a>
-<span class="sourceLineNo">748</span> @Override<a name="line.748"></a>
-<span class="sourceLineNo">749</span> protected void setup(Context context) throws IOException, InterruptedException {<a name="line.749"></a>
-<span class="sourceLineNo">750</span> conn = ConnectionFactory.createConnection(context.getConfiguration());<a name="line.750"></a>
-<span class="sourceLineNo">751</span> table = conn.getTable(getTablename(conn.getConfiguration()));<a name="line.751"></a>
-<span class="sourceLineNo">752</span> }<a name="line.752"></a>
-<span class="sourceLineNo">753</span><a name="line.753"></a>
-<span class="sourceLineNo">754</span> @Override<a name="line.754"></a>
-<span class="sourceLineNo">755</span> protected void cleanup(Context context) throws IOException ,InterruptedException {<a name="line.755"></a>
-<span class="sourceLineNo">756</span> table.close();<a name="line.756"></a>
-<span class="sourceLineNo">757</span> conn.close();<a name="line.757"></a>
-<span class="sourceLineNo">758</span> }<a name="line.758"></a>
-<span class="sourceLineNo">759</span><a name="line.759"></a>
-<span class="sourceLineNo">760</span> @Override<a name="line.760"></a>
-<span class="sourceLineNo">761</span> protected void map(HBaseKeyWritable key, BytesWritable value, Context output)<a name="line.761"></a>
-<span class="sourceLineNo">762</span> throws IOException, InterruptedException {<a name="line.762"></a>
-<span class="sourceLineNo">763</span><a name="line.763"></a>
-<span class="sourceLineNo">764</span> byte[] row = Bytes.copy(key.getRowArray(), key.getRowOffset(), key.getRowLength());<a name="line.764"></a>
-<span class="sourceLineNo">765</span> byte[] family = Bytes.copy(key.getFamilyArray(), key.getFamilyOffset(),<a name="line.765"></a>
-<span class="sourceLineNo">766</span> key.getFamilyLength());<a name="line.766"></a>
-<span class="sourceLineNo">767</span> byte[] qualifier = Bytes.copy(key.getQualifierArray(), key.getQualifierOffset(),<a name="line.767"></a>
-<span class="sourceLineNo">768</span> key.getQualifierLength());<a name="line.768"></a>
-<span class="sourceLineNo">769</span> long ts = key.getTimestamp();<a name="line.769"></a>
+<span class="sourceLineNo">745</span> public static void main(String[] args) throws Exception {<a name="line.745"></a>
+<span class="sourceLineNo">746</span> System.exit(ToolRunner.run(HBaseConfiguration.create(), new Verify(), args));<a name="line.746"></a>
+<span class="sourceLineNo">747</span> }<a name="line.747"></a>
+<span class="sourceLineNo">748</span><a name="line.748"></a>
+<span class="sourceLineNo">749</span> public static class VerifyMapper<a name="line.749"></a>
+<span class="sourceLineNo">750</span> extends Mapper<HBaseKeyWritable, BytesWritable, NullWritable, NullWritable> {<a name="line.750"></a>
+<span class="sourceLineNo">751</span><a name="line.751"></a>
+<span class="sourceLineNo">752</span> Connection conn;<a name="line.752"></a>
+<span class="sourceLineNo">753</span> Table table;<a name="line.753"></a>
+<span class="sourceLineNo">754</span><a name="line.754"></a>
+<span class="sourceLineNo">755</span> @Override<a name="line.755"></a>
+<span class="sourceLineNo">756</span> protected void setup(Context context) throws IOException, InterruptedException {<a name="line.756"></a>
+<span class="sourceLineNo">757</span> conn = ConnectionFactory.createConnection(context.getConfiguration());<a name="line.757"></a>
+<span class="sourceLineNo">758</span> table = conn.getTable(getTablename(conn.getConfiguration()));<a name="line.758"></a>
+<span class="sourceLineNo">759</span> }<a name="line.759"></a>
+<span class="sourceLineNo">760</span><a name="line.760"></a>
+<span class="sourceLineNo">761</span> @Override<a name="line.761"></a>
+<span class="sourceLineNo">762</span> protected void cleanup(Context context) throws IOException ,InterruptedException {<a name="line.762"></a>
+<span class="sourceLineNo">763</span> table.close();<a name="line.763"></a>
+<span class="sourceLineNo">764</span> conn.close();<a name="line.764"></a>
+<span class="sourceLineNo">765</span> }<a name="line.765"></a>
+<span class="sourceLineNo">766</span><a name="line.766"></a>
+<span class="sourceLineNo">767</span> @Override<a name="line.767"></a>
+<span class="sourceLineNo">768</span> protected void map(HBaseKeyWritable key, BytesWritable value, Context output)<a name="line.768"></a>
+<span class="sourceLineNo">769</span> throws IOException, InterruptedException {<a name="line.769"></a>
<span class="sourceLineNo">770</span><a name="line.770"></a>
-<span class="sourceLineNo">771</span> if (Bytes.equals(INFO_FAMILY_NAME, family) &&<a name="line.771"></a>
-<span class="sourceLineNo">772</span> Bytes.equals(CRC_QUALIFIER, qualifier)) {<a name="line.772"></a>
-<span class="sourceLineNo">773</span><a name="line.773"></a>
-<span class="sourceLineNo">774</span> long expectedCRC64 = Bytes.toLong(value.getBytes(), 0, value.getLength());<a name="line.774"></a>
-<span class="sourceLineNo">775</span><a name="line.775"></a>
-<span class="sourceLineNo">776</span> Result result =<a name="line.776"></a>
-<span class="sourceLineNo">777</span> table.get(new Get(row)<a name="line.777"></a>
-<span class="sourceLineNo">778</span> .addColumn(CONTENT_FAMILY_NAME, CONTENT_QUALIFIER)<a name="line.778"></a>
-<span class="sourceLineNo">779</span> .addColumn(INFO_FAMILY_NAME, CRC_QUALIFIER)<a name="line.779"></a>
-<span class="sourceLineNo">780</span> .setTimestamp(ts));<a name="line.780"></a>
-<span class="sourceLineNo">781</span><a name="line.781"></a>
-<span class="sourceLineNo">782</span> byte[] content = result.getValue(CONTENT_FAMILY_NAME, CONTENT_QUALIFIER);<a name="line.782"></a>
-<span class="sourceLineNo">783</span> if (content == null) {<a name="line.783"></a>
-<span class="sourceLineNo">784</span> LOG.info("Row " + Bytes.toStringBinary(row) + ": missing content");<a name="line.784"></a>
-<span class="sourceLineNo">785</span> output.getCounter(Counts.UNREFERENCED).increment(1);<a name="line.785"></a>
-<span class="sourceLineNo">786</span> return;<a name="line.786"></a>
-<span class="sourceLineNo">787</span> } else {<a name="line.787"></a>
-<span class="sourceLineNo">788</span> CRC64 crc = new CRC64();<a name="line.788"></a>
-<span class="sourceLineNo">789</span> crc.update(content);<a name="line.789"></a>
-<span class="sourceLineNo">790</span> if (crc.getValue() != expectedCRC64) {<a name="line.790"></a>
-<span class="sourceLineNo">791</span> LOG.info("Row " + Bytes.toStringBinary(row) + ": corrupt content");<a name="line.791"></a>
-<span class="sourceLineNo">792</span> output.getCounter(Counts.CORRUPT).increment(1);<a name="line.792"></a>
-<span class="sourceLineNo">793</span> return;<a name="line.793"></a>
-<span class="sourceLineNo">794</span> }<a name="line.794"></a>
-<span class="sourceLineNo">795</span> }<a name="line.795"></a>
-<span class="sourceLineNo">796</span> byte[] crc = result.getValue(INFO_FAMILY_NAME, CRC_QUALIFIER);<a name="line.796"></a>
-<span class="sourceLineNo">797</span> if (crc == null) {<a name="line.797"></a>
-<span class="sourceLineNo">798</span> LOG.info("Row " + Bytes.toStringBinary(row) + ": missing i:c");<a name="line.798"></a>
-<span class="sourceLineNo">799</span> output.getCounter(Counts.UNREFERENCED).increment(1);<a name="line.799"></a>
-<span class="sourceLineNo">800</span> return;<a name="line.800"></a>
-<span class="sourceLineNo">801</span> }<a name="line.801"></a>
-<span class="sourceLineNo">802</span> if (Bytes.toLong(crc) != expectedCRC64) {<a name="line.802"></a>
-<span class="sourceLineNo">803</span> LOG.info("Row " + Bytes.toStringBinary(row) + ": i:c mismatch");<a name="line.803"></a>
-<span class="sourceLineNo">804</span> output.getCounter(Counts.CORRUPT).increment(1);<a name="line.804"></a>
-<span class="sourceLineNo">805</span> return;<a name="line.805"></a>
-<span class="sourceLineNo">806</span> }<a name="line.806"></a>
-<span class="sourceLineNo">807</span><a name="line.807"></a>
-<span class="sourceLineNo">808</span> } else {<a name="line.808"></a>
-<span class="sourceLineNo">809</span><a name="line.809"></a>
-<span class="sourceLineNo">810</span> Result result =<a name="line.810"></a>
-<span class="sourceLineNo">811</span> table.get(new Get(row)<a name="line.811"></a>
-<span class="sourceLineNo">812</span> .addColumn(family, qualifier)<a name="line.812"></a>
-<span class="sourceLineNo">813</span> .setTimestamp(ts));<a name="line.813"></a>
-<span class="sourceLineNo">814</span><a name="line.814"></a>
-<span class="sourceLineNo">815</span> byte[] bytes = result.getValue(family, qualifier);<a name="line.815"></a>
-<span class="sourceLineNo">816</span> if (bytes == null) {<a name="line.816"></a>
-<span class="sourceLineNo">817</span> LOG.info("Row " + Bytes.toStringBinary(row) + ": missing " +<a name="line.817"></a>
-<span class="sourceLineNo">818</span> Bytes.toStringBinary(family) + ":" + Bytes.toStringBinary(qualifier));<a name="line.818"></a>
-<span class="sourceLineNo">819</span> output.getCounter(Counts.UNREFERENCED).increment(1);<a name="line.819"></a>
-<span class="sourceLineNo">820</span> return;<a name="line.820"></a>
-<span class="sourceLineNo">821</span> }<a name="line.821"></a>
-<span class="sourceLineNo">822</span> if (!Bytes.equals(bytes, 0, bytes.length, value.getBytes(), 0, value.getLength())) {<a name="line.822"></a>
-<span class="sourceLineNo">823</span> LOG.info("Row " + Bytes.toStringBinary(row) + ": " +<a name="line.823"></a>
-<span class="sourceLineNo">824</span> Bytes.toStringBinary(family) + ":" + Bytes.toStringBinary(qualifier) +<a name="line.824"></a>
-<span class="sourceLineNo">825</span> " mismatch");<a name="line.825"></a>
-<span class="sourceLineNo">826</span> output.getCounter(Counts.CORRUPT).increment(1);<a name="line.826"></a>
-<span class="sourceLineNo">827</span> return;<a name="line.827"></a>
-<span class="sourceLineNo">828</span> }<a name="line.828"></a>
-<span class="sourceLineNo">829</span><a name="line.829"></a>
-<span class="sourceLineNo">830</span> }<a name="line.830"></a>
-<span class="sourceLineNo">831</span><a name="line.831"></a>
-<span class="sourceLineNo">832</span> output.getCounter(Counts.REFERENCED).increment(1);<a name="line.832"></a>
-<span class="sourceLineNo">833</span> }<a name="line.833"></a>
-<span class="sourceLineNo">834</span><a name="line.834"></a>
-<span class="sourceLineNo">835</span> }<a name="line.835"></a>
-<span class="sourceLineNo">836</span><a name="line.836"></a>
-<span class="sourceLineNo">837</span> }<a name="line.837"></a>
-<span class="sourceLineNo">838</span><a name="line.838"></a>
-<span class="sourceLineNo">839</span>}<a name="line.839"></a>
+<span class="sourceLineNo">771</span> byte[] row = Bytes.copy(key.getRowArray(), key.getRowOffset(), key.getRowLength());<a name="line.771"></a>
+<span class="sourceLineNo">772</span> byte[] family = Bytes.copy(key.getFamilyArray(), key.getFamilyOffset(),<a name="line.772"></a>
+<span class="sourceLineNo">773</span> key.getFamilyLength());<a name="line.773"></a>
+<span class="sourceLineNo">774</span> byte[] qualifier = Bytes.copy(key.getQualifierArray(), key.getQualifierOffset(),<a name="line.774"></a>
+<span class="sourceLineNo">775</span> key.getQualifierLength());<a name="line.775"></a>
+<span class="sourceLineNo">776</span> long ts = key.getTimestamp();<a name="line.776"></a>
+<span class="sourceLineNo">777</span> int retries = VERIFICATION_READ_RETRIES;<a name="line.777"></a>
+<span class="sourceLineNo">778</span><a name="line.778"></a>
+<span class="sourceLineNo">779</span> while (true) {<a name="line.779"></a>
+<span class="sourceLineNo">780</span><a name="line.780"></a>
+<span class="sourceLineNo">781</span> if (Bytes.equals(INFO_FAMILY_NAME, family) &&<a name="line.781"></a>
+<span class="sourceLineNo">782</span> Bytes.equals(CRC_QUALIFIER, qualifier)) {<a name="line.782"></a>
+<span class="sourceLineNo">783</span><a name="line.783"></a>
+<span class="sourceLineNo">784</span> long expectedCRC64 = Bytes.toLong(value.getBytes(), 0, value.getLength());<a name="line.784"></a>
+<span class="sourceLineNo">785</span> Result result = table.get(new Get(row)<a name="line.785"></a>
+<span class="sourceLineNo">786</span> .addColumn(CONTENT_FAMILY_NAME, CONTENT_QUALIFIER)<a name="line.786"></a>
+<span class="sourceLineNo">787</span> .addColumn(INFO_FAMILY_NAME, CRC_QUALIFIER)<a name="line.787"></a>
+<span class="sourceLineNo">788</span> .setTimestamp(ts));<a name="line.788"></a>
+<span class="sourceLineNo">789</span> byte[] content = result.getValue(CONTENT_FAMILY_NAME, CONTENT_QUALIFIER);<a name="line.789"></a>
+<span class="sourceLineNo">790</span> if (content == null) {<a name="line.790"></a>
+<span class="sourceLineNo">791</span> if (retries-- > 0) {<a name="line.791"></a>
+<span class="sourceLineNo">792</span> continue;<a name="line.792"></a>
+<span class="sourceLineNo">793</span> }<a name="line.793"></a>
+<span class="sourceLineNo">794</span> LOG.error("Row " + Bytes.toStringBinary(row) + ": missing content");<a name="line.794"></a>
+<span class="sourceLineNo">795</span> output.getCounter(Counts.UNREFERENCED).increment(1);<a name="line.795"></a>
+<span class="sourceLineNo">796</span> return;<a name="line.796"></a>
+<span class="sourceLineNo">797</span> } else {<a name="line.797"></a>
+<span class="sourceLineNo">798</span> CRC64 crc = new CRC64();<a name="line.798"></a>
+<span class="sourceLineNo">799</span> crc.update(content);<a name="line.799"></a>
+<span class="sourceLineNo">800</span> if (crc.getValue() != expectedCRC64) {<a name="line.800"></a>
+<span class="sourceLineNo">801</span> if (retries-- > 0) {<a name="line.801"></a>
+<span class="sourceLineNo">802</span> continue;<a name="line.802"></a>
+<span class="sourceLineNo">803</span> }<a name="line.803"></a>
+<span class="sourceLineNo">804</span> LOG.error("Row " + Bytes.toStringBinary(row) + ": corrupt content");<a name="line.804"></a>
+<span class="sourceLineNo">805</span> output.getCounter(Counts.CORRUPT).increment(1);<a name="line.805"></a>
+<span class="sourceLineNo">806</span> return;<a name="line.806"></a>
+<span class="sourceLineNo">807</span> }<a name="line.807"></a>
+<span class="sourceLineNo">808</span> }<a name="line.808"></a>
+<span class="sourceLineNo">809</span> byte[] crc = result.getValue(INFO_FAMILY_NAME, CRC_QUALIFIER);<a name="line.809"></a>
+<span class="sourceLineNo">810</span> if (crc == null) {<a name="line.810"></a>
+<span class="sourceLineNo">811</span> if (retries-- > 0) {<a name="line.811"></a>
+<span class="sourceLineNo">812</span> continue;<a name="line.812"></a>
+<span class="sourceLineNo">813</span> }<a name="line.813"></a>
+<span class="sourceLineNo">814</span> LOG.error("Row " + Bytes.toStringBinary(row) + ": missing i:c");<a name="line.814"></a>
+<span class="sourceLineNo">815</span> output.getCounter(Counts.UNREFERENCED).increment(1);<a name="line.815"></a>
+<span class="sourceLineNo">816</span> return;<a name="line.816"></a>
+<span class="sourceLineNo">817</span> }<a name="line.817"></a>
+<span class="sourceLineNo">818</span> if (Bytes.toLong(crc) != expectedCRC64) {<a name="line.818"></a>
+<span class="sourceLineNo">819</span> if (retries-- > 0) {<a name="line.819"></a>
+<span class="sourceLineNo">820</span> continue;<a name="line.820"></a>
+<span class="sourceLineNo">821</span> }<a name="line.821"></a>
+<span class="sourceLineNo">822</span> LOG.error("Row " + Bytes.toStringBinary(row) + ": i:c mismatch");<a name="line.822"></a>
+<span class="sourceLineNo">823</span> output.getCounter(Counts.CORRUPT).increment(1);<a name="line.823"></a>
+<span class="sourceLineNo">824</span> return;<a name="line.824"></a>
+<span class="sourceLineNo">825</span> }<a name="line.825"></a>
+<span class="sourceLineNo">826</span><a name="line.826"></a>
+<span class="sourceLineNo">827</span> } else {<a name="line.827"></a>
+<span class="sourceLineNo">828</span><a name="line.828"></a>
+<span class="sourceLineNo">829</span> Result result = table.get(new Get(row)<a name="line.829"></a>
+<span class="sourceLineNo">830</span> .addColumn(family, qualifier)<a name="line.830"></a>
+<span class="sourceLineNo">831</span> .setTimestamp(ts));<a name="line.831"></a>
+<span class="sourceLineNo">832</span> byte[] bytes = result.getValue(family, qualifier);<a name="line.832"></a>
+<span class="sourceLineNo">833</span> if (bytes == null) {<a name="line.833"></a>
+<span class="sourceLineNo">834</span> if (retries-- > 0) {<a name="line.834"></a>
+<span class="sourceLineNo">835</span> continue;<a name="line.835"></a>
+<span class="sourceLineNo">836</span> }<a name="line.836"></a>
+<span class="sourceLineNo">837</span> LOG.error("Row " + Bytes.toStringBinary(row) + ": missing " +<a name="line.837"></a>
+<span class="sourceLineNo">838</span> Bytes.toStringBinary(family) + ":" + Bytes.toStringBinary(qualifier));<a name="line.838"></a>
+<span class="sourceLineNo">839</span> output.getCounter(Counts.UNREFERENCED).increment(1);<a name="line.839"></a>
+<span class="sourceLineNo">840</span> return;<a name="line.840"></a>
+<span class="sourceLineNo">841</span> }<a name="line.841"></a>
+<span class="sourceLineNo">842</span> if (!Bytes.equals(bytes, 0, bytes.length, value.getBytes(), 0, value.getLength())) {<a name="line.842"></a>
+<span class="sourceLineNo">843</span> if (retries-- > 0) {<a name="line.843"></a>
+<span class="sourceLineNo">844</span> continue;<a name="line.844"></a>
+<span class="sourceLineNo">845</span> }<a name="line.845"></a>
+<span class="sourceLineNo">846</span> LOG.error("Row " + Bytes.toStringBinary(row) + ": " +<a name="line.846"></a>
+<span class="sourceLineNo">847</span> Bytes.toStringBinary(family) + ":" + Bytes.toStringBinary(qualifier) +<a name="line.847"></a>
+<span class="sourceLineNo">848</span> " mismatch");<a name="line.848"></a>
+<span class="sourceLineNo">849</span> output.getCounter(Counts.CORRUPT).increment(1);<a name="line.849"></a>
+<span class="sourceLineNo">850</span> return;<a name="line.850"></a>
+<span class="sourceLineNo">851</span> }<a name="line.851"></a>
+<span class="sourceLineNo">852</span><a name="line.852"></a>
+<span class="sourceLineNo">853</span> }<a name="line.853"></a>
+<span class="sourceLineNo">854</span><a name="line.854"></a>
+<span class="sourceLineNo">855</span> // If we fell through to here all verification checks have succeeded, potentially after<a name="line.855"></a>
+<span class="sourceLineNo">856</span> // retries, and we must exit the while loop.<a name="line.856"></a>
+<span class="sourceLineNo">857</span> output.getCounter(Counts.REFERENCED).increment(1);<a name="line.857"></a>
+<span class="sourceLineNo">858</span> break;<a name="line.858"></a>
+<span class="sourceLineNo">859</span><a name="line.859"></a>
+<span class="sourceLineNo">860</span> }<a name="line.860"></a>
+<span class="sourceLineNo">861</span> }<a name="line.861"></a>
+<span class="sourceLineNo">862</span> }<a name="line.862"></a>
+<span class="sourceLineNo">863</span> }<a name="line.863"></a>
+<span class="sourceLineNo">864</span><a name="line.864"></a>
+<span class="sourceLineNo">865</span>}<a name="line.865"></a>
diff --git a/testdevapidocs/src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.OneFilePerMapperSFIF.html b/testdevapidocs/src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.OneFilePerMapperSFIF.html
index 4a40cbc..e89b63c 100644
--- a/testdevapidocs/src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.OneFilePerMapperSFIF.html
+++ b/testdevapidocs/src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.OneFilePerMapperSFIF.html
@@ -54,797 +54,823 @@
<span class="sourceLineNo">046</span>import org.apache.hadoop.hbase.MasterNotRunningException;<a name="line.46"></a>
<span class="sourceLineNo">047</span>import org.apache.hadoop.hbase.TableName;<a name="line.47"></a>
<span class="sourceLineNo">048</span>import org.apache.hadoop.hbase.client.Admin;<a name="line.48"></a>
-<span class="sourceLineNo">049</span>import org.apache.hadoop.hbase.client.ColumnFamilyDescriptor;<a name="line.49"></a>
-<span class="sourceLineNo">050</span>import org.apache.hadoop.hbase.client.ColumnFamilyDescriptorBuilder;<a name="line.50"></a>
-<span class="sourceLineNo">051</span>import org.apache.hadoop.hbase.client.Connection;<a name="line.51"></a>
-<span class="sourceLineNo">052</span>import org.apache.hadoop.hbase.client.ConnectionFactory;<a name="line.52"></a>
-<span class="sourceLineNo">053</span>import org.apache.hadoop.hbase.client.Get;<a name="line.53"></a>
-<span class="sourceLineNo">054</span>import org.apache.hadoop.hbase.client.Put;<a name="line.54"></a>
-<span class="sourceLineNo">055</span>import org.apache.hadoop.hbase.client.Result;<a name="line.55"></a>
-<span class="sourceLineNo">056</span>import org.apache.hadoop.hbase.client.Table;<a name="line.56"></a>
-<span class="sourceLineNo">057</span>import org.apache.hadoop.hbase.client.TableDescriptor;<a name="line.57"></a>
-<span class="sourceLineNo">058</span>import org.apache.hadoop.hbase.client.TableDescriptorBuilder;<a name="line.58"></a>
-<span class="sourceLineNo">059</span>import org.apache.hadoop.hbase.io.encoding.DataBlockEncoding;<a name="line.59"></a>
-<span class="sourceLineNo">060</span>import org.apache.hadoop.hbase.mapreduce.TableMapReduceUtil;<a name="line.60"></a>
-<span class="sourceLineNo">061</span>import org.apache.hadoop.hbase.regionserver.BloomType;<a name="line.61"></a>
-<span class="sourceLineNo">062</span>import org.apache.hadoop.hbase.test.util.CRC64;<a name="line.62"></a>
-<span class="sourceLineNo">063</span>import org.apache.hadoop.hbase.test.util.warc.WARCInputFormat;<a name="line.63"></a>
-<span class="sourceLineNo">064</span>import org.apache.hadoop.hbase.test.util.warc.WARCRecord;<a name="line.64"></a>
-<span class="sourceLineNo">065</span>import org.apache.hadoop.hbase.test.util.warc.WARCWritable;<a name="line.65"></a>
-<span class="sourceLineNo">066</span>import org.apache.hadoop.hbase.util.Bytes;<a name="line.66"></a>
-<span class="sourceLineNo">067</span>import org.apache.hadoop.hbase.util.EnvironmentEdgeManager;<a name="line.67"></a>
-<span class="sourceLineNo">068</span>import org.apache.hadoop.hbase.util.RegionSplitter;<a name="line.68"></a>
-<span class="sourceLineNo">069</span>import org.apache.hadoop.io.BytesWritable;<a name="line.69"></a>
-<span class="sourceLineNo">070</span>import org.apache.hadoop.io.LongWritable;<a name="line.70"></a>
-<span class="sourceLineNo">071</span>import org.apache.hadoop.io.NullWritable;<a name="line.71"></a>
-<span class="sourceLineNo">072</span>import org.apache.hadoop.io.SequenceFile.CompressionType;<a name="line.72"></a>
-<span class="sourceLineNo">073</span>import org.apache.hadoop.io.Writable;<a name="line.73"></a>
-<span class="sourceLineNo">074</span>import org.apache.hadoop.mapreduce.Counters;<a name="line.74"></a>
-<span class="sourceLineNo">075</span>import org.apache.hadoop.mapreduce.Job;<a name="line.75"></a>
-<span class="sourceLineNo">076</span>import org.apache.hadoop.mapreduce.JobContext;<a name="line.76"></a>
-<span class="sourceLineNo">077</span>import org.apache.hadoop.mapreduce.Mapper;<a name="line.77"></a>
-<span class="sourceLineNo">078</span>import org.apache.hadoop.mapreduce.lib.input.FileInputFormat;<a name="line.78"></a>
-<span class="sourceLineNo">079</span>import org.apache.hadoop.mapreduce.lib.input.SequenceFileInputFormat;<a name="line.79"></a>
-<span class="sourceLineNo">080</span>import org.apache.hadoop.mapreduce.lib.output.NullOutputFormat;<a name="line.80"></a>
-<span class="sourceLineNo">081</span>import org.apache.hadoop.mapreduce.lib.output.SequenceFileOutputFormat;<a name="line.81"></a>
-<span class="sourceLineNo">082</span>import org.apache.hadoop.util.Tool;<a name="line.82"></a>
-<span class="sourceLineNo">083</span>import org.apache.hadoop.util.ToolRunner;<a name="line.83"></a>
-<span class="sourceLineNo">084</span>import org.slf4j.Logger;<a name="line.84"></a>
-<span class="sourceLineNo">085</span>import org.slf4j.LoggerFactory;<a name="line.85"></a>
-<span class="sourceLineNo">086</span><a name="line.86"></a>
-<span class="sourceLineNo">087</span>import org.apache.hbase.thirdparty.org.apache.commons.cli.CommandLine;<a name="line.87"></a>
-<span class="sourceLineNo">088</span><a name="line.88"></a>
-<span class="sourceLineNo">089</span>/**<a name="line.89"></a>
-<span class="sourceLineNo">090</span> * This integration test loads successful resource retrieval records from the Common Crawl<a name="line.90"></a>
-<span class="sourceLineNo">091</span> * (https://commoncrawl.org/) public dataset into an HBase table and writes records that can be<a name="line.91"></a>
-<span class="sourceLineNo">092</span> * used to later verify the presence and integrity of those records.<a name="line.92"></a>
-<span class="sourceLineNo">093</span> * <p><a name="line.93"></a>
-<span class="sourceLineNo">094</span> * Run like:<a name="line.94"></a>
-<span class="sourceLineNo">095</span> * <blockquote><a name="line.95"></a>
-<span class="sourceLineNo">096</span> * ./bin/hbase org.apache.hadoop.hbase.test.IntegrationTestLoadCommonCrawl \<br><a name="line.96"></a>
-<span class="sourceLineNo">097</span> * &nbsp;&nbsp; -Dfs.s3n.awsAccessKeyId=&lt;AWS access key&gt; \<br><a name="line.97"></a>
-<span class="sourceLineNo">098</span> * &nbsp;&nbsp; -Dfs.s3n.awsSecretAccessKey=&lt;AWS secret key&gt; \<br><a name="line.98"></a>
-<span class="sourceLineNo">099</span> * &nbsp;&nbsp; /path/to/test-CC-MAIN-2021-10-warc.paths.gz \<br><a name="line.99"></a>
-<span class="sourceLineNo">100</span> * &nbsp;&nbsp; /path/to/tmp/warc-loader-output<a name="line.100"></a>
-<span class="sourceLineNo">101</span> * </blockquote><a name="line.101"></a>
-<span class="sourceLineNo">102</span> * <p><a name="line.102"></a>
-<span class="sourceLineNo">103</span> * Access to the Common Crawl dataset in S3 is made available to anyone by Amazon AWS, but<a name="line.103"></a>
-<span class="sourceLineNo">104</span> * Hadoop's S3N filesystem still requires valid access credentials to initialize.<a name="line.104"></a>
-<span class="sourceLineNo">105</span> * <p><a name="line.105"></a>
-<span class="sourceLineNo">106</span> * The input path can either specify a directory or a file. The file may optionally be<a name="line.106"></a>
-<span class="sourceLineNo">107</span> * compressed with gzip. If a directory, the loader expects the directory to contain one or more<a name="line.107"></a>
-<span class="sourceLineNo">108</span> * WARC files from the Common Crawl dataset. If a file, the loader expects a list of Hadoop S3N<a name="line.108"></a>
-<span class="sourceLineNo">109</span> * URIs which point to S3 locations for one or more WARC files from the Common Crawl dataset,<a name="line.109"></a>
-<span class="sourceLineNo">110</span> * one URI per line. Lines should be terminated with the UNIX line terminator.<a name="line.110"></a>
-<span class="sourceLineNo">111</span> * <p><a name="line.111"></a>
-<span class="sourceLineNo">112</span> * Included in hbase-it/src/test/resources/CC-MAIN-2021-10-warc.paths.gz is a list of all WARC<a name="line.112"></a>
-<span class="sourceLineNo">113</span> * files comprising the Q1 2021 crawl archive. There are 64,000 WARC files in this data set, each<a name="line.113"></a>
-<span class="sourceLineNo">114</span> * containing ~1GB of gzipped data. The WARC files contain several record types, such as metadata,<a name="line.114"></a>
-<span class="sourceLineNo">115</span> * request, and response, but we only load the response record types. If the HBase table schema<a name="line.115"></a>
-<span class="sourceLineNo">116</span> * does not specify compression (by default) there is roughly a 10x expansion. Loading the full<a name="line.116"></a>
-<span class="sourceLineNo">117</span> * crawl archive results in a table approximately 640 TB in size.<a name="line.117"></a>
-<span class="sourceLineNo">118</span> * <p><a name="line.118"></a>
-<span class="sourceLineNo">119</span> * You can also split the Loader and Verify stages:<a name="line.119"></a>
-<span class="sourceLineNo">120</span> * <p><a name="line.120"></a>
-<span class="sourceLineNo">121</span> * Load with:<a name="line.121"></a>
-<span class="sourceLineNo">122</span> * <blockquote><a name="line.122"></a>
-<span class="sourceLineNo">123</span> * ./bin/hbase 'org.apache.hadoop.hbase.test.IntegrationTestLoadCommonCrawl$Loader' \<br><a name="line.123"></a>
-<span class="sourceLineNo">124</span> * &nbsp;&nbsp; -files /path/to/hadoop-aws.jar \<br><a name="line.124"></a>
-<span class="sourceLineNo">125</span> * &nbsp;&nbsp; -Dfs.s3n.awsAccessKeyId=&lt;AWS access key&gt; \<br><a name="line.125"></a>
-<span class="sourceLineNo">126</span> * &nbsp;&nbsp; -Dfs.s3n.awsSecretAccessKey=&lt;AWS secret key&gt; \<br><a name="line.126"></a>
-<span class="sourceLineNo">127</span> * &nbsp;&nbsp; /path/to/test-CC-MAIN-2021-10-warc.paths.gz \<br><a name="line.127"></a>
-<span class="sourceLineNo">128</span> * &nbsp;&nbsp; /path/to/tmp/warc-loader-output<a name="line.128"></a>
-<span class="sourceLineNo">129</span> * </blockquote><a name="line.129"></a>
-<span class="sourceLineNo">130</span> * <p><a name="line.130"></a>
-<span class="sourceLineNo">131</span> * Note: The hadoop-aws jar will be needed at runtime to instantiate the S3N filesystem. Use<a name="line.131"></a>
-<span class="sourceLineNo">132</span> * the <tt>-files</tt> ToolRunner argument to add it.<a name="line.132"></a>
-<span class="sourceLineNo">133</span> * <p><a name="line.133"></a>
-<span class="sourceLineNo">134</span> * Verify with:<a name="line.134"></a>
-<span class="sourceLineNo">135</span> * <blockquote><a name="line.135"></a>
-<span class="sourceLineNo">136</span> * ./bin/hbase 'org.apache.hadoop.hbase.test.IntegrationTestLoadCommonCrawl$Verify' \<br><a name="line.136"></a>
-<span class="sourceLineNo">137</span> * &nbsp;&nbsp; /path/to/tmp/warc-loader-output<a name="line.137"></a>
-<span class="sourceLineNo">138</span> * </blockquote><a name="line.138"></a>
-<span class="sourceLineNo">139</span> * <p><a name="line.139"></a>
-<span class="sourceLineNo">140</span> */<a name="line.140"></a>
-<span class="sourceLineNo">141</span>public class IntegrationTestLoadCommonCrawl extends IntegrationTestBase {<a name="line.141"></a>
-<span class="sourceLineNo">142</span><a name="line.142"></a>
-<span class="sourceLineNo">143</span> private static final Logger LOG = LoggerFactory.getLogger(IntegrationTestLoadCommonCrawl.class);<a name="line.143"></a>
-<span class="sourceLineNo">144</span><a name="line.144"></a>
-<span class="sourceLineNo">145</span> protected static String TABLE_NAME_KEY = "IntegrationTestLoadCommonCrawl.table";<a name="line.145"></a>
-<span class="sourceLineNo">146</span> protected static String DEFAULT_TABLE_NAME = "IntegrationTestLoadCommonCrawl";<a name="line.146"></a>
-<span class="sourceLineNo">147</span><a name="line.147"></a>
-<span class="sourceLineNo">148</span> protected static byte[] CONTENT_FAMILY_NAME = Bytes.toBytes("c");<a name="line.148"></a>
-<span class="sourceLineNo">149</span> protected static byte[] INFO_FAMILY_NAME = Bytes.toBytes("i");<a name="line.149"></a>
-<span class="sourceLineNo">150</span> protected static byte[] CONTENT_QUALIFIER = HConstants.EMPTY_BYTE_ARRAY;<a name="line.150"></a>
-<span class="sourceLineNo">151</span> protected static byte[] CONTENT_LENGTH_QUALIFIER = Bytes.toBytes("l");<a name="line.151"></a>
-<span class="sourceLineNo">152</span> protected static byte[] CONTENT_TYPE_QUALIFIER = Bytes.toBytes("t");<a name="line.152"></a>
-<span class="sourceLineNo">153</span> protected static byte[] CRC_QUALIFIER = Bytes.toBytes("c");<a name="line.153"></a>
-<span class="sourceLineNo">154</span> protected static byte[] DATE_QUALIFIER = Bytes.toBytes("d");<a name="line.154"></a>
-<span class="sourceLineNo">155</span> protected static byte[] IP_ADDRESS_QUALIFIER = Bytes.toBytes("a");<a name="line.155"></a>
-<span class="sourceLineNo">156</span> protected static byte[] RECORD_ID_QUALIFIER = Bytes.toBytes("r");<a name="line.156"></a>
-<span class="sourceLineNo">157</span> protected static byte[] TARGET_URI_QUALIFIER = Bytes.toBytes("u");<a name="line.157"></a>
-<span class="sourceLineNo">158</span><a name="line.158"></a>
-<span class="sourceLineNo">159</span> public static enum Counts {<a name="line.159"></a>
-<span class="sourceLineNo">160</span> REFERENCED, UNREFERENCED, CORRUPT<a name="line.160"></a>
-<span class="sourceLineNo">161</span> }<a name="line.161"></a>
-<span class="sourceLineNo">162</span><a name="line.162"></a>
-<span class="sourceLineNo">163</span> Path warcFileInputDir = null;<a name="line.163"></a>
-<span class="sourceLineNo">164</span> Path outputDir = null;<a name="line.164"></a>
-<span class="sourceLineNo">165</span> String[] args;<a name="line.165"></a>
-<span class="sourceLineNo">166</span><a name="line.166"></a>
-<span class="sourceLineNo">167</span> protected int runLoader(Path warcFileInputDir, Path outputDir) throws Exception {<a name="line.167"></a>
-<span class="sourceLineNo">168</span> Loader loader = new Loader();<a name="line.168"></a>
-<span class="sourceLineNo">169</span> loader.setConf(conf);<a name="line.169"></a>
-<span class="sourceLineNo">170</span> return loader.run(warcFileInputDir, outputDir);<a name="line.170"></a>
-<span class="sourceLineNo">171</span> }<a name="line.171"></a>
-<span class="sourceLineNo">172</span><a name="line.172"></a>
-<span class="sourceLineNo">173</span> protected int runVerify(Path inputDir) throws Exception {<a name="line.173"></a>
-<span class="sourceLineNo">174</span> Verify verify = new Verify();<a name="line.174"></a>
-<span class="sourceLineNo">175</span> verify.setConf(conf);<a name="line.175"></a>
-<span class="sourceLineNo">176</span> return verify.run(inputDir);<a name="line.176"></a>
-<span class="sourceLineNo">177</span> }<a name="line.177"></a>
-<span class="sourceLineNo">178</span><a name="line.178"></a>
-<span class="sourceLineNo">179</span> @Override<a name="line.179"></a>
-<span class="sourceLineNo">180</span> public int run(String[] args) {<a name="line.180"></a>
-<span class="sourceLineNo">181</span> if (args.length > 0) {<a name="line.181"></a>
-<span class="sourceLineNo">182</span> warcFileInputDir = new Path(args[0]);<a name="line.182"></a>
-<span class="sourceLineNo">183</span> if (args.length > 1) {<a name="line.183"></a>
-<span class="sourceLineNo">184</span> outputDir = new Path(args[1]);<a name="line.184"></a>
-<span class="sourceLineNo">185</span> }<a name="line.185"></a>
-<span class="sourceLineNo">186</span> }<a name="line.186"></a>
-<span class="sourceLineNo">187</span> try {<a name="line.187"></a>
-<span class="sourceLineNo">188</span> if (warcFileInputDir == null) {<a name="line.188"></a>
-<span class="sourceLineNo">189</span> throw new IllegalArgumentException("WARC input file or directory not specified");<a name="line.189"></a>
-<span class="sourceLineNo">190</span> }<a name="line.190"></a>
-<span class="sourceLineNo">191</span> if (outputDir == null) {<a name="line.191"></a>
-<span class="sourceLineNo">192</span> throw new IllegalArgumentException("Output directory not specified");<a name="line.192"></a>
+<span class="sourceLineNo">049</span>import org.apache.hadoop.hbase.client.BufferedMutator;<a name="line.49"></a>
+<span class="sourceLineNo">050</span>import org.apache.hadoop.hbase.client.ColumnFamilyDescriptor;<a name="line.50"></a>
+<span class="sourceLineNo">051</span>import org.apache.hadoop.hbase.client.ColumnFamilyDescriptorBuilder;<a name="line.51"></a>
+<span class="sourceLineNo">052</span>import org.apache.hadoop.hbase.client.Connection;<a name="line.52"></a>
+<span class="sourceLineNo">053</span>import org.apache.hadoop.hbase.client.ConnectionFactory;<a name="line.53"></a>
+<span class="sourceLineNo">054</span>import org.apache.hadoop.hbase.client.Get;<a name="line.54"></a>
+<span class="sourceLineNo">055</span>import org.apache.hadoop.hbase.client.Put;<a name="line.55"></a>
+<span class="sourceLineNo">056</span>import org.apache.hadoop.hbase.client.Result;<a name="line.56"></a>
+<span class="sourceLineNo">057</span>import org.apache.hadoop.hbase.client.Table;<a name="line.57"></a>
+<span class="sourceLineNo">058</span>import org.apache.hadoop.hbase.client.TableDescriptor;<a name="line.58"></a>
+<span class="sourceLineNo">059</span>import org.apache.hadoop.hbase.client.TableDescriptorBuilder;<a name="line.59"></a>
+<span class="sourceLineNo">060</span>import org.apache.hadoop.hbase.io.encoding.DataBlockEncoding;<a name="line.60"></a>
+<span class="sourceLineNo">061</span>import org.apache.hadoop.hbase.mapreduce.TableMapReduceUtil;<a name="line.61"></a>
+<span class="sourceLineNo">062</span>import org.apache.hadoop.hbase.regionserver.BloomType;<a name="line.62"></a>
+<span class="sourceLineNo">063</span>import org.apache.hadoop.hbase.test.util.CRC64;<a name="line.63"></a>
+<span class="sourceLineNo">064</span>import org.apache.hadoop.hbase.test.util.warc.WARCInputFormat;<a name="line.64"></a>
+<span class="sourceLineNo">065</span>import org.apache.hadoop.hbase.test.util.warc.WARCRecord;<a name="line.65"></a>
+<span class="sourceLineNo">066</span>import org.apache.hadoop.hbase.test.util.warc.WARCWritable;<a name="line.66"></a>
+<span class="sourceLineNo">067</span>import org.apache.hadoop.hbase.util.Bytes;<a name="line.67"></a>
+<span class="sourceLineNo">068</span>import org.apache.hadoop.hbase.util.EnvironmentEdgeManager;<a name="line.68"></a>
+<span class="sourceLineNo">069</span>import org.apache.hadoop.hbase.util.RegionSplitter;<a name="line.69"></a>
+<span class="sourceLineNo">070</span>import org.apache.hadoop.io.BytesWritable;<a name="line.70"></a>
+<span class="sourceLineNo">071</span>import org.apache.hadoop.io.LongWritable;<a name="line.71"></a>
+<span class="sourceLineNo">072</span>import org.apache.hadoop.io.NullWritable;<a name="line.72"></a>
+<span class="sourceLineNo">073</span>import org.apache.hadoop.io.SequenceFile.CompressionType;<a name="line.73"></a>
+<span class="sourceLineNo">074</span>import org.apache.hadoop.io.Writable;<a name="line.74"></a>
+<span class="sourceLineNo">075</span>import org.apache.hadoop.mapreduce.Counters;<a name="line.75"></a>
+<span class="sourceLineNo">076</span>import org.apache.hadoop.mapreduce.Job;<a name="line.76"></a>
+<span class="sourceLineNo">077</span>import org.apache.hadoop.mapreduce.JobContext;<a name="line.77"></a>
+<span class="sourceLineNo">078</span>import org.apache.hadoop.mapreduce.Mapper;<a name="line.78"></a>
+<span class="sourceLineNo">079</span>import org.apache.hadoop.mapreduce.lib.input.FileInputFormat;<a name="line.79"></a>
+<span class="sourceLineNo">080</span>import org.apache.hadoop.mapreduce.lib.input.SequenceFileInputFormat;<a name="line.80"></a>
+<span class="sourceLineNo">081</span>import org.apache.hadoop.mapreduce.lib.output.NullOutputFormat;<a name="line.81"></a>
+<span class="sourceLineNo">082</span>import org.apache.hadoop.mapreduce.lib.output.SequenceFileOutputFormat;<a name="line.82"></a>
+<span class="sourceLineNo">083</span>import org.apache.hadoop.util.Tool;<a name="line.83"></a>
+<span class="sourceLineNo">084</span>import org.apache.hadoop.util.ToolRunner;<a name="line.84"></a>
+<span class="sourceLineNo">085</span>import org.slf4j.Logger;<a name="line.85"></a>
+<span class="sourceLineNo">086</span>import org.slf4j.LoggerFactory;<a name="line.86"></a>
+<span class="sourceLineNo">087</span><a name="line.87"></a>
+<span class="sourceLineNo">088</span>import org.apache.hbase.thirdparty.org.apache.commons.cli.CommandLine;<a name="line.88"></a>
+<span class="sourceLineNo">089</span><a name="line.89"></a>
+<span class="sourceLineNo">090</span>/**<a name="line.90"></a>
+<span class="sourceLineNo">091</span> * This integration test loads successful resource retrieval records from the Common Crawl<a name="line.91"></a>
+<span class="sourceLineNo">092</span> * (https://commoncrawl.org/) public dataset into an HBase table and writes records that can be<a name="line.92"></a>
+<span class="sourceLineNo">093</span> * used to later verify the presence and integrity of those records.<a name="line.93"></a>
+<span class="sourceLineNo">094</span> * <p><a name="line.94"></a>
+<span class="sourceLineNo">095</span> * Run like:<a name="line.95"></a>
+<span class="sourceLineNo">096</span> * <blockquote><a name="line.96"></a>
+<span class="sourceLineNo">097</span> * ./bin/hbase org.apache.hadoop.hbase.test.IntegrationTestLoadCommonCrawl \<br><a name="line.97"></a>
+<span class="sourceLineNo">098</span> * &nbsp;&nbsp; -Dfs.s3n.awsAccessKeyId=&lt;AWS access key&gt; \<br><a name="line.98"></a>
+<span class="sourceLineNo">099</span> * &nbsp;&nbsp; -Dfs.s3n.awsSecretAccessKey=&lt;AWS secret key&gt; \<br><a name="line.99"></a>
+<span class="sourceLineNo">100</span> * &nbsp;&nbsp; /path/to/test-CC-MAIN-2021-10-warc.paths.gz \<br><a name="line.100"></a>
+<span class="sourceLineNo">101</span> * &nbsp;&nbsp; /path/to/tmp/warc-loader-output<a name="line.101"></a>
+<span class="sourceLineNo">102</span> * </blockquote><a name="line.102"></a>
+<span class="sourceLineNo">103</span> * <p><a name="line.103"></a>
+<span class="sourceLineNo">104</span> * Access to the Common Crawl dataset in S3 is made available to anyone by Amazon AWS, but<a name="line.104"></a>
+<span class="sourceLineNo">105</span> * Hadoop's S3N filesystem still requires valid access credentials to initialize.<a name="line.105"></a>
+<span class="sourceLineNo">106</span> * <p><a name="line.106"></a>
+<span class="sourceLineNo">107</span> * The input path can either specify a directory or a file. The file may optionally be<a name="line.107"></a>
+<span class="sourceLineNo">108</span> * compressed with gzip. If a directory, the loader expects the directory to contain one or more<a name="line.108"></a>
+<span class="sourceLineNo">109</span> * WARC files from the Common Crawl dataset. If a file, the loader expects a list of Hadoop S3N<a name="line.109"></a>
+<span class="sourceLineNo">110</span> * URIs which point to S3 locations for one or more WARC files from the Common Crawl dataset,<a name="line.110"></a>
+<span class="sourceLineNo">111</span> * one URI per line. Lines should be terminated with the UNIX line terminator.<a name="line.111"></a>
+<span class="sourceLineNo">112</span> * <p><a name="line.112"></a>
+<span class="sourceLineNo">113</span> * Included in hbase-it/src/test/resources/CC-MAIN-2021-10-warc.paths.gz is a list of all WARC<a name="line.113"></a>
+<span class="sourceLineNo">114</span> * files comprising the Q1 2021 crawl archive. There are 64,000 WARC files in this data set, each<a name="line.114"></a>
+<span class="sourceLineNo">115</span> * containing ~1GB of gzipped data. The WARC files contain several record types, such as metadata,<a name="line.115"></a>
+<span class="sourceLineNo">116</span> * request, and response, but we only load the response record types. If the HBase table schema<a name="line.116"></a>
+<span class="sourceLineNo">117</span> * does not specify compression (by default) there is roughly a 10x expansion. Loading the full<a name="line.117"></a>
+<span class="sourceLineNo">118</span> * crawl archive results in a table approximately 640 TB in size.<a name="line.118"></a>
+<span class="sourceLineNo">119</span> * <p><a name="line.119"></a>
+<span class="sourceLineNo">120</span> * You can also split the Loader and Verify stages:<a name="line.120"></a>
+<span class="sourceLineNo">121</span> * <p><a name="line.121"></a>
+<span class="sourceLineNo">122</span> * Load with:<a name="line.122"></a>
+<span class="sourceLineNo">123</span> * <blockquote><a name="line.123"></a>
+<span class="sourceLineNo">124</span> * ./bin/hbase 'org.apache.hadoop.hbase.test.IntegrationTestLoadCommonCrawl$Loader' \<br><a name="line.124"></a>
+<span class="sourceLineNo">125</span> * &nbsp;&nbsp; -files /path/to/hadoop-aws.jar \<br><a name="line.125"></a>
+<span class="sourceLineNo">126</span> * &nbsp;&nbsp; -Dfs.s3n.awsAccessKeyId=&lt;AWS access key&gt; \<br><a name="line.126"></a>
+<span class="sourceLineNo">127</span> * &nbsp;&nbsp; -Dfs.s3n.awsSecretAccessKey=&lt;AWS secret key&gt; \<br><a name="line.127"></a>
+<span class="sourceLineNo">128</span> * &nbsp;&nbsp; /path/to/test-CC-MAIN-2021-10-warc.paths.gz \<br><a name="line.128"></a>
+<span class="sourceLineNo">129</span> * &nbsp;&nbsp; /path/to/tmp/warc-loader-output<a name="line.129"></a>
+<span class="sourceLineNo">130</span> * </blockquote><a name="line.130"></a>
+<span class="sourceLineNo">131</span> * <p><a name="line.131"></a>
+<span class="sourceLineNo">132</span> * Note: The hadoop-aws jar will be needed at runtime to instantiate the S3N filesystem. Use<a name="line.132"></a>
+<span class="sourceLineNo">133</span> * the <tt>-files</tt> ToolRunner argument to add it.<a name="line.133"></a>
+<span class="sourceLineNo">134</span> * <p><a name="line.134"></a>
+<span class="sourceLineNo">135</span> * Verify with:<a name="line.135"></a>
+<span class="sourceLineNo">136</span> * <blockquote><a name="line.136"></a>
+<span class="sourceLineNo">137</span> * ./bin/hbase 'org.apache.hadoop.hbase.test.IntegrationTestLoadCommonCrawl$Verify' \<br><a name="line.137"></a>
+<span class="sourceLineNo">138</span> * &nbsp;&nbsp; /path/to/tmp/warc-loader-output<a name="line.138"></a>
+<span class="sourceLineNo">139</span> * </blockquote><a name="line.139"></a>
+<span class="sourceLineNo">140</span> * <p><a name="line.140"></a>
+<span class="sourceLineNo">141</span> */<a name="line.141"></a>
+<span class="sourceLineNo">142</span>public class IntegrationTestLoadCommonCrawl extends IntegrationTestBase {<a name="line.142"></a>
+<span class="sourceLineNo">143</span><a name="line.143"></a>
+<span class="sourceLineNo">144</span> private static final Logger LOG = LoggerFactory.getLogger(IntegrationTestLoadCommonCrawl.class);<a name="line.144"></a>
+<span class="sourceLineNo">145</span><a name="line.145"></a>
+<span class="sourceLineNo">146</span> protected static String TABLE_NAME_KEY = "IntegrationTestLoadCommonCrawl.table";<a name="line.146"></a>
+<span class="sourceLineNo">147</span> protected static String DEFAULT_TABLE_NAME = "IntegrationTestLoadCommonCrawl";<a name="line.147"></a>
+<span class="sourceLineNo">148</span><a name="line.148"></a>
+<span class="sourceLineNo">149</span> protected static byte[] CONTENT_FAMILY_NAME = Bytes.toBytes("c");<a name="line.149"></a>
+<span class="sourceLineNo">150</span> protected static byte[] INFO_FAMILY_NAME = Bytes.toBytes("i");<a name="line.150"></a>
+<span class="sourceLineNo">151</span> protected static byte[] CONTENT_QUALIFIER = HConstants.EMPTY_BYTE_ARRAY;<a name="line.151"></a>
+<span class="sourceLineNo">152</span> protected static byte[] CONTENT_LENGTH_QUALIFIER = Bytes.toBytes("l");<a name="line.152"></a>
+<span class="sourceLineNo">153</span> protected static byte[] CONTENT_TYPE_QUALIFIER = Bytes.toBytes("t");<a name="line.153"></a>
+<span class="sourceLineNo">154</span> protected static byte[] CRC_QUALIFIER = Bytes.toBytes("c");<a name="line.154"></a>
+<span class="sourceLineNo">155</span> protected static byte[] DATE_QUALIFIER = Bytes.toBytes("d");<a name="line.155"></a>
+<span class="sourceLineNo">156</span> protected static byte[] IP_ADDRESS_QUALIFIER = Bytes.toBytes("a");<a name="line.156"></a>
+<span class="sourceLineNo">157</span> protected static byte[] RECORD_ID_QUALIFIER = Bytes.toBytes("r");<a name="line.157"></a>
+<span class="sourceLineNo">158</span> protected static byte[] TARGET_URI_QUALIFIER = Bytes.toBytes("u");<a name="line.158"></a>
+<span class="sourceLineNo">159</span><a name="line.159"></a>
+<span class="sourceLineNo">160</span> private static final int VERIFICATION_READ_RETRIES = 10;<a name="line.160"></a>
+<span class="sourceLineNo">161</span><a name="line.161"></a>
+<span class="sourceLineNo">162</span> public static enum Counts {<a name="line.162"></a>
+<span class="sourceLineNo">163</span> REFERENCED, UNREFERENCED, CORRUPT<a name="line.163"></a>
+<span class="sourceLineNo">164</span> }<a name="line.164"></a>
+<span class="sourceLineNo">165</span><a name="line.165"></a>
+<span class="sourceLineNo">166</span> Path warcFileInputDir = null;<a name="line.166"></a>
+<span class="sourceLineNo">167</span> Path outputDir = null;<a name="line.167"></a>
+<span class="sourceLineNo">168</span> String[] args;<a name="line.168"></a>
+<span class="sourceLineNo">169</span><a name="line.169"></a>
+<span class="sourceLineNo">170</span> protected int runLoader(Path warcFileInputDir, Path outputDir) throws Exception {<a name="line.170"></a>
+<span class="sourceLineNo">171</span> Loader loader = new Loader();<a name="line.171"></a>
+<span class="sourceLineNo">172</span> loader.setConf(conf);<a name="line.172"></a>
+<span class="sourceLineNo">173</span> return loader.run(warcFileInputDir, outputDir);<a name="line.173"></a>
+<span class="sourceLineNo">174</span> }<a name="line.174"></a>
+<span class="sourceLineNo">175</span><a name="line.175"></a>
+<span class="sourceLineNo">176</span> protected int runVerify(Path inputDir) throws Exception {<a name="line.176"></a>
+<span class="sourceLineNo">177</span> Verify verify = new Verify();<a name="line.177"></a>
+<span class="sourceLineNo">178</span> verify.setConf(conf);<a name="line.178"></a>
+<span class="sourceLineNo">179</span> return verify.run(inputDir);<a name="line.179"></a>
+<span class="sourceLineNo">180</span> }<a name="line.180"></a>
+<span class="sourceLineNo">181</span><a name="line.181"></a>
+<span class="sourceLineNo">182</span> @Override<a name="line.182"></a>
+<span class="sourceLineNo">183</span> public int run(String[] args) {<a name="line.183"></a>
+<span class="sourceLineNo">184</span> if (args.length > 0) {<a name="line.184"></a>
+<span class="sourceLineNo">185</span> warcFileInputDir = new Path(args[0]);<a name="line.185"></a>
+<span class="sourceLineNo">186</span> if (args.length > 1) {<a name="line.186"></a>
+<span class="sourceLineNo">187</span> outputDir = new Path(args[1]);<a name="line.187"></a>
+<span class="sourceLineNo">188</span> }<a name="line.188"></a>
+<span class="sourceLineNo">189</span> }<a name="line.189"></a>
+<span class="sourceLineNo">190</span> try {<a name="line.190"></a>
+<span class="sourceLineNo">191</span> if (warcFileInputDir == null) {<a name="line.191"></a>
+<span class="sourceLineNo">192</span> throw new IllegalArgumentException("WARC input file or directory not specified");<a name="line.192"></a>
<span class="sourceLineNo">193</span> }<a name="line.193"></a>
-<span class="sourceLineNo">194</span> int res = runLoader(warcFileInputDir, outputDir);<a name="line.194"></a>
-<span class="sourceLineNo">195</span> if (res != 0) {<a name="line.195"></a>
-<span class="sourceLineNo">196</span> LOG.error("Loader failed");<a name="line.196"></a>
-<span class="sourceLineNo">197</span> return -1;<a name="line.197"></a>
-<span class="sourceLineNo">198</span> }<a name="line.198"></a>
-<span class="sourceLineNo">199</span> res = runVerify(outputDir);<a name="line.199"></a>
-<span class="sourceLineNo">200</span> } catch (Exception e) {<a name="line.200"></a>
-<span class="sourceLineNo">201</span> LOG.error("Tool failed with exception", e);<a name="line.201"></a>
-<span class="sourceLineNo">202</span> return -1;<a name="line.202"></a>
-<span class="sourceLineNo">203</span> }<a name="line.203"></a>
-<span class="sourceLineNo">204</span> return 0;<a name="line.204"></a>
-<span class="sourceLineNo">205</span> }<a name="line.205"></a>
-<span class="sourceLineNo">206</span><a name="line.206"></a>
-<span class="sourceLineNo">207</span> @Override<a name="line.207"></a>
-<span class="sourceLineNo">208</span> protected void processOptions(CommandLine cmd) {<a name="line.208"></a>
-<span class="sourceLineNo">209</span> processBaseOptions(cmd);<a name="line.209"></a>
-<span class="sourceLineNo">210</span> args = cmd.getArgs();<a name="line.210"></a>
-<span class="sourceLineNo">211</span> }<a name="line.211"></a>
-<span class="sourceLineNo">212</span><a name="line.212"></a>
-<span class="sourceLineNo">213</span> @Override<a name="line.213"></a>
-<span class="sourceLineNo">214</span> public void setUpCluster() throws Exception {<a name="line.214"></a>
-<span class="sourceLineNo">215</span> util = getTestingUtil(getConf());<a name="line.215"></a>
-<span class="sourceLineNo">216</span> boolean isDistributed = util.isDistributedCluster();<a name="line.216"></a>
-<span class="sourceLineNo">217</span> util.initializeCluster(isDistributed ? 1 : 3);<a name="line.217"></a>
-<span class="sourceLineNo">218</span> if (!isDistributed) {<a name="line.218"></a>
-<span class="sourceLineNo">219</span> util.startMiniMapReduceCluster();<a name="line.219"></a>
-<span class="sourceLineNo">220</span> }<a name="line.220"></a>
-<span class="sourceLineNo">221</span> this.setConf(util.getConfiguration());<a name="line.221"></a>
-<span class="sourceLineNo">222</span> }<a name="line.222"></a>
-<span class="sourceLineNo">223</span><a name="line.223"></a>
-<span class="sourceLineNo">224</span> @Override<a name="line.224"></a>
-<span class="sourceLineNo">225</span> public void cleanUpCluster() throws Exception {<a name="line.225"></a>
-<span class="sourceLineNo">226</span> super.cleanUpCluster();<a name="line.226"></a>
-<span class="sourceLineNo">227</span> if (util.isDistributedCluster()) {<a name="line.227"></a>
-<span class="sourceLineNo">228</span> util.shutdownMiniMapReduceCluster();<a name="line.228"></a>
-<span class="sourceLineNo">229</span> }<a name="line.229"></a>
-<span class="sourceLineNo">230</span> }<a name="line.230"></a>
-<span class="sourceLineNo">231</span><a name="line.231"></a>
-<span class="sourceLineNo">232</span> static TableName getTablename(Configuration c) {<a name="line.232"></a>
-<span class="sourceLineNo">233</span> return TableName.valueOf(c.get(TABLE_NAME_KEY, DEFAULT_TABLE_NAME));<a name="line.233"></a>
-<span class="sourceLineNo">234</span> }<a name="line.234"></a>
-<span class="sourceLineNo">235</span><a name="line.235"></a>
-<span class="sourceLineNo">236</span> @Override<a name="line.236"></a>
-<span class="sourceLineNo">237</span> public TableName getTablename() {<a name="line.237"></a>
-<span class="sourceLineNo">238</span> return getTablename(getConf());<a name="line.238"></a>
-<span class="sourceLineNo">239</span> }<a name="line.239"></a>
-<span class="sourceLineNo">240</span><a name="line.240"></a>
-<span class="sourceLineNo">241</span> @Override<a name="line.241"></a>
-<span class="sourceLineNo">242</span> protected Set<String> getColumnFamilies() {<a name="line.242"></a>
-<span class="sourceLineNo">243</span> Set<String> families = new HashSet<>();<a name="line.243"></a>
-<span class="sourceLineNo">244</span> families.add(Bytes.toString(CONTENT_FAMILY_NAME));<a name="line.244"></a>
-<span class="sourceLineNo">245</span> families.add(Bytes.toString(INFO_FAMILY_NAME));<a name="line.245"></a>
-<span class="sourceLineNo">246</span> return families;<a name="line.246"></a>
-<span class="sourceLineNo">247</span> }<a name="line.247"></a>
-<span class="sourceLineNo">248</span><a name="line.248"></a>
-<span class="sourceLineNo">249</span> @Override<a name="line.249"></a>
-<span class="sourceLineNo">250</span> public int runTestFromCommandLine() throws Exception {<a name="line.250"></a>
-<span class="sourceLineNo">251</span> return ToolRunner.run(getConf(), this, args);<a name="line.251"></a>
-<span class="sourceLineNo">252</span> }<a name="line.252"></a>
-<span class="sourceLineNo">253</span><a name="line.253"></a>
-<span class="sourceLineNo">254</span> public static void main(String[] args) throws Exception {<a name="line.254"></a>
-<span class="sourceLineNo">255</span> Configuration conf = HBaseConfiguration.create();<a name="line.255"></a>
-<span class="sourceLineNo">256</span> IntegrationTestingUtility.setUseDistributedCluster(conf);<a name="line.256"></a>
-<span class="sourceLineNo">257</span> int ret = ToolRunner.run(conf, new IntegrationTestLoadCommonCrawl(), args);<a name="line.257"></a>
-<span class="sourceLineNo">258</span> System.exit(ret);<a name="line.258"></a>
-<span class="sourceLineNo">259</span> }<a name="line.259"></a>
-<span class="sourceLineNo">260</span><a name="line.260"></a>
-<span class="sourceLineNo">261</span> public static class HBaseKeyWritable implements Writable {<a name="line.261"></a>
-<span class="sourceLineNo">262</span><a name="line.262"></a>
-<span class="sourceLineNo">263</span> private byte[] row;<a name="line.263"></a>
-<span class="sourceLineNo">264</span> private int rowOffset;<a name="line.264"></a>
-<span class="sourceLineNo">265</span> private int rowLength;<a name="line.265"></a>
-<span class="sourceLineNo">266</span> private byte[] family;<a name="line.266"></a>
-<span class="sourceLineNo">267</span> private int familyOffset;<a name="line.267"></a>
-<span class="sourceLineNo">268</span> private int familyLength;<a name="line.268"></a>
-<span class="sourceLineNo">269</span> private byte[] qualifier;<a name="line.269"></a>
-<span class="sourceLineNo">270</span> private int qualifierOffset;<a name="line.270"></a>
-<span class="sourceLineNo">271</span> private int qualifierLength;<a name="line.271"></a>
-<span class="sourceLineNo">272</span> private long ts;<a name="line.272"></a>
-<span class="sourceLineNo">273</span><a name="line.273"></a>
-<span class="sourceLineNo">274</span> public HBaseKeyWritable() { }<a name="line.274"></a>
-<span class="sourceLineNo">275</span><a name="line.275"></a>
-<span class="sourceLineNo">276</span> public HBaseKeyWritable(byte[] row, int rowOffset, int rowLength,<a name="line.276"></a>
-<span class="sourceLineNo">277</span> byte[] family, int familyOffset, int familyLength,<a name="line.277"></a>
-<span class="sourceLineNo">278</span> byte[] qualifier, int qualifierOffset, int qualifierLength, long ts) {<a name="line.278"></a>
-<span class="sourceLineNo">279</span> this.row = row;<a name="line.279"></a>
-<span class="sourceLineNo">280</span> this.rowOffset = rowOffset;<a name="line.280"></a>
-<span class="sourceLineNo">281</span> this.rowLength = rowLength;<a name="line.281"></a>
-<span class="sourceLineNo">282</span> this.family = family;<a name="line.282"></a>
-<span class="sourceLineNo">283</span> this.familyOffset = familyOffset;<a name="line.283"></a>
-<span class="sourceLineNo">284</span> this.familyLength = familyLength;<a name="line.284"></a>
-<span class="sourceLineNo">285</span> this.qualifier = qualifier;<a name="line.285"></a>
-<span class="sourceLineNo">286</span> this.qualifierOffset = qualifierOffset;<a name="line.286"></a>
-<span class="sourceLineNo">287</span> this.qualifierLength = qualifierLength;<a name="line.287"></a>
-<span class="sourceLineNo">288</span> this.ts = ts;<a name="line.288"></a>
-<span class="sourceLineNo">289</span> }<a name="line.289"></a>
-<span class="sourceLineNo">290</span><a name="line.290"></a>
-<span class="sourceLineNo">291</span> public HBaseKeyWritable(byte[] row, byte[] family, byte[] qualifier, long ts) {<a name="line.291"></a>
-<span class="sourceLineNo">292</span> this(row, 0, row.length,<a name="line.292"></a>
-<span class="sourceLineNo">293</span> family, 0, family.length,<a name="line.293"></a>
-<span class="sourceLineNo">294</span> qualifier, 0, qualifier != null ? qualifier.length : 0, ts);<a name="line.294"></a>
-<span class="sourceLineNo">295</span> }<a name="line.295"></a>
-<span class="sourceLineNo">296</span><a name="line.296"></a>
-<span class="sourceLineNo">297</span> public HBaseKeyWritable(byte[] row, byte[] family, long ts) {<a name="line.297"></a>
-<span class="sourceLineNo">298</span> this(row, family, HConstants.EMPTY_BYTE_ARRAY, ts);<a name="line.298"></a>
-<span class="sourceLineNo">299</span> }<a name="line.299"></a>
-<span class="sourceLineNo">300</span><a name="line.300"></a>
-<span class="sourceLineNo">301</span> public HBaseKeyWritable(Cell cell) {<a name="line.301"></a>
-<span class="sourceLineNo">302</span> this(cell.getRowArray(), cell.getRowOffset(), cell.getRowLength(),<a name="line.302"></a>
-<span class="sourceLineNo">303</span> cell.getFamilyArray(), cell.getFamilyOffset(), cell.getFamilyLength(),<a name="line.303"></a>
-<span class="sourceLineNo">304</span> cell.getQualifierArray(), cell.getQualifierOffset(), cell.getQualifierLength(),<a name="line.304"></a>
-<span class="sourceLineNo">305</span> cell.getTimestamp());<a name="line.305"></a>
-<span class="sourceLineNo">306</span> }<a name="line.306"></a>
-<span class="sourceLineNo">307</span><a name="line.307"></a>
-<span class="sourceLineNo">308</span> @Override<a name="line.308"></a>
-<span class="sourceLineNo">309</span> public void readFields(DataInput in) throws IOException {<a name="line.309"></a>
-<span class="sourceLineNo">310</span> this.row = Bytes.toBytes(in.readUTF());<a name="line.310"></a>
-<span class="sourceLineNo">311</span> this.rowOffset = 0;<a name="line.311"></a>
-<span class="sourceLineNo">312</span> this.rowLength = row.length;<a name="line.312"></a>
-<span class="sourceLineNo">313</span> this.family = Bytes.toBytes(in.readUTF());<a name="line.313"></a>
-<span class="sourceLineNo">314</span> this.familyOffset = 0;<a name="line.314"></a>
-<span class="sourceLineNo">315</span> this.familyLength = family.length;<a name="line.315"></a>
-<span class="sourceLineNo">316</span> this.qualifier = Bytes.toBytes(in.readUTF());<a name="line.316"></a>
-<span class="sourceLineNo">317</span> this.qualifierOffset = 0;<a name="line.317"></a>
-<span class="sourceLineNo">318</span> this.qualifierLength = qualifier.length;<a name="line.318"></a>
-<span class="sourceLineNo">319</span> this.ts = in.readLong();<a name="line.319"></a>
-<span class="sourceLineNo">320</span> }<a name="line.320"></a>
-<span class="sourceLineNo">321</span><a name="line.321"></a>
-<span class="sourceLineNo">322</span> @Override<a name="line.322"></a>
-<span class="sourceLineNo">323</span> public void write(DataOutput out) throws IOException {<a name="line.323"></a>
-<span class="sourceLineNo">324</span> out.writeUTF(new String(row, rowOffset, rowLength, StandardCharsets.UTF_8));<a name="line.324"></a>
-<span class="sourceLineNo">325</span> out.writeUTF(new String(family, familyOffset, familyLength, StandardCharsets.UTF_8));<a name="line.325"></a>
-<span class="sourceLineNo">326</span> if (qualifier != null) {<a name="line.326"></a>
-<span class="sourceLineNo">327</span> out.writeUTF(new String(qualifier, qualifierOffset, qualifierLength,<a name="line.327"></a>
-<span class="sourceLineNo">328</span> StandardCharsets.UTF_8));<a name="line.328"></a>
-<span class="sourceLineNo">329</span> } else {<a name="line.329"></a>
-<span class="sourceLineNo">330</span> out.writeUTF("");<a name="line.330"></a>
-<span class="sourceLineNo">331</span> }<a name="line.331"></a>
-<span class="sourceLineNo">332</span> out.writeLong(ts);<a name="line.332"></a>
-<span class="sourceLineNo">333</span> }<a name="line.333"></a>
-<span class="sourceLineNo">334</span><a name="line.334"></a>
-<span class="sourceLineNo">335</span> public byte[] getRowArray() {<a name="line.335"></a>
-<span class="sourceLineNo">336</span> return row;<a name="line.336"></a>
-<span class="sourceLineNo">337</span> }<a name="line.337"></a>
-<span class="sourceLineNo">338</span><a name="line.338"></a>
-<span class="sourceLineNo">339</span> public void setRow(byte[] row) {<a name="line.339"></a>
-<span class="sourceLineNo">340</span> this.row = row;<a name="line.340"></a>
-<span class="sourceLineNo">341</span> }<a name="line.341"></a>
-<span class="sourceLineNo">342</span><a name="line.342"></a>
-<span class="sourceLineNo">343</span> public int getRowOffset() {<a name="line.343"></a>
-<span class="sourceLineNo">344</span> return rowOffset;<a name="line.344"></a>
-<span class="sourceLineNo">345</span> }<a name="line.345"></a>
-<span class="sourceLineNo">346</span><a name="line.346"></a>
-<span class="sourceLineNo">347</span> public void setRowOffset(int rowOffset) {<a name="line.347"></a>
-<span class="sourceLineNo">348</span> this.rowOffset = rowOffset;<a name="line.348"></a>
-<span class="sourceLineNo">349</span> }<a name="line.349"></a>
-<span class="sourceLineNo">350</span><a name="line.350"></a>
-<span class="sourceLineNo">351</span> public int getRowLength() {<a name="line.351"></a>
-<span class="sourceLineNo">352</span> return rowLength;<a name="line.352"></a>
-<span class="sourceLineNo">353</span> }<a name="line.353"></a>
-<span class="sourceLineNo">354</span><a name="line.354"></a>
-<span class="sourceLineNo">355</span> public void setRowLength(int rowLength) {<a name="line.355"></a>
-<span class="sourceLineNo">356</span> this.rowLength = rowLength;<a name="line.356"></a>
-<span class="sourceLineNo">357</span> }<a name="line.357"></a>
-<span class="sourceLineNo">358</span><a name="line.358"></a>
-<span class="sourceLineNo">359</span> public byte[] getFamilyArray() {<a name="line.359"></a>
-<span class="sourceLineNo">360</span> return family;<a name="line.360"></a>
-<span class="sourceLineNo">361</span> }<a name="line.361"></a>
-<span class="sourceLineNo">362</span><a name="line.362"></a>
-<span class="sourceLineNo">363</span> public void setFamily(byte[] family) {<a name="line.363"></a>
-<span class="sourceLineNo">364</span> this.family = family;<a name="line.364"></a>
-<span class="sourceLineNo">365</span> }<a name="line.365"></a>
-<span class="sourceLineNo">366</span><a name="line.366"></a>
-<span class="sourceLineNo">367</span> public int getFamilyOffset() {<a name="line.367"></a>
-<span class="sourceLineNo">368</span> return familyOffset;<a name="line.368"></a>
-<span class="sourceLineNo">369</span> }<a name="line.369"></a>
-<span class="sourceLineNo">370</span><a name="line.370"></a>
-<span class="sourceLineNo">371</span> public void setFamilyOffset(int familyOffset) {<a name="line.371"></a>
-<span class="sourceLineNo">372</span> this.familyOffset = familyOffset;<a name="line.372"></a>
-<span class="sourceLineNo">373</span> }<a name="line.373"></a>
-<span class="sourceLineNo">374</span><a name="line.374"></a>
-<span class="sourceLineNo">375</span> public int getFamilyLength() {<a name="line.375"></a>
-<span class="sourceLineNo">376</span> return familyLength;<a name="line.376"></a>
-<span class="sourceLineNo">377</span> }<a name="line.377"></a>
-<span class="sourceLineNo">378</span><a name="line.378"></a>
-<span class="sourceLineNo">379</span> public void setFamilyLength(int familyLength) {<a name="line.379"></a>
-<span class="sourceLineNo">380</span> this.familyLength = familyLength;<a name="line.380"></a>
-<span class="sourceLineNo">381</span> }<a name="line.381"></a>
-<span class="sourceLineNo">382</span><a name="line.382"></a>
-<span class="sourceLineNo">383</span> public byte[] getQualifierArray() {<a name="line.383"></a>
-<span class="sourceLineNo">384</span> return qualifier;<a name="line.384"></a>
-<span class="sourceLineNo">385</span> }<a name="line.385"></a>
-<span class="sourceLineNo">386</span><a name="line.386"></a>
-<span class="sourceLineNo">387</span> public void setQualifier(byte[] qualifier) {<a name="line.387"></a>
-<span class="sourceLineNo">388</span> this.qualifier = qualifier;<a name="line.388"></a>
-<span class="sourceLineNo">389</span> }<a name="line.389"></a>
-<span class="sourceLineNo">390</span><a name="line.390"></a>
-<span class="sourceLineNo">391</span> public int getQualifierOffset() {<a name="line.391"></a>
-<span class="sourceLineNo">392</span> return qualifierOffset;<a name="line.392"></a>
-<span class="sourceLineNo">393</span> }<a name="line.393"></a>
-<span class="sourceLineNo">394</span><a name="line.394"></a>
-<span class="sourceLineNo">395</span> public void setQualifierOffset(int qualifierOffset) {<a name="line.395"></a>
-<span class="sourceLineNo">396</span> this.qualifierOffset = qualifierOffset;<a name="line.396"></a>
-<span class="sourceLineNo">397</span> }<a name="line.397"></a>
-<span class="sourceLineNo">398</span><a name="line.398"></a>
-<span class="sourceLineNo">399</span> public int getQualifierLength() {<a name="line.399"></a>
-<span class="sourceLineNo">400</span> return qualifierLength;<a name="line.400"></a>
-<span class="sourceLineNo">401</span> }<a name="line.401"></a>
-<span class="sourceLineNo">402</span><a name="line.402"></a>
-<span class="sourceLineNo">403</span> public void setQualifierLength(int qualifierLength) {<a name="line.403"></a>
-<span class="sourceLineNo">404</span> this.qualifierLength = qualifierLength;<a name="line.404"></a>
-<span class="sourceLineNo">405</span> }<a name="line.405"></a>
-<span class="sourceLineNo">406</span><a name="line.406"></a>
-<span class="sourceLineNo">407</span> public long getTimestamp() {<a name="line.407"></a>
-<span class="sourceLineNo">408</span> return ts;<a name="line.408"></a>
-<span class="sourceLineNo">409</span> }<a name="line.409"></a>
-<span class="sourceLineNo">410</span><a name="line.410"></a>
-<span class="sourceLineNo">411</span> public void setTimestamp(long ts) {<a name="line.411"></a>
-<span class="sourceLineNo">412</span> this.ts = ts;<a name="line.412"></a>
-<span class="sourceLineNo">413</span> }<a name="line.413"></a>
-<span class="sourceLineNo">414</span> }<a name="line.414"></a>
-<span class="sourceLineNo">415</span><a name="line.415"></a>
-<span class="sourceLineNo">416</span> public static class Loader extends Configured implements Tool {<a name="line.416"></a>
-<span class="sourceLineNo">417</span><a name="line.417"></a>
-<span class="sourceLineNo">418</span> private static final Logger LOG = LoggerFactory.getLogger(Loader.class);<a name="line.418"></a>
-<span class="sourceLineNo">419</span> private static final String USAGE = "Loader <warInputDir | warFileList> <outputDir>";<a name="line.419"></a>
+<span class="sourceLineNo">194</span> if (outputDir == null) {<a name="line.194"></a>
+<span class="sourceLineNo">195</span> throw new IllegalArgumentException("Output directory not specified");<a name="line.195"></a>
+<span class="sourceLineNo">196</span> }<a name="line.196"></a>
+<span class="sourceLineNo">197</span> int res = runLoader(warcFileInputDir, outputDir);<a name="line.197"></a>
+<span class="sourceLineNo">198</span> if (res != 0) {<a name="line.198"></a>
+<span class="sourceLineNo">199</span> LOG.error("Loader failed");<a name="line.199"></a>
+<span class="sourceLineNo">200</span> return -1;<a name="line.200"></a>
+<span class="sourceLineNo">201</span> }<a name="line.201"></a>
+<span class="sourceLineNo">202</span> res = runVerify(outputDir);<a name="line.202"></a>
+<span class="sourceLineNo">203</span> } catch (Exception e) {<a name="line.203"></a>
+<span class="sourceLineNo">204</span> LOG.error("Tool failed with exception", e);<a name="line.204"></a>
+<span class="sourceLineNo">205</span> return -1;<a name="line.205"></a>
+<span class="sourceLineNo">206</span> }<a name="line.206"></a>
+<span class="sourceLineNo">207</span> return 0;<a name="line.207"></a>
+<span class="sourceLineNo">208</span> }<a name="line.208"></a>
+<span class="sourceLineNo">209</span><a name="line.209"></a>
+<span class="sourceLineNo">210</span> @Override<a name="line.210"></a>
+<span class="sourceLineNo">211</span> protected void processOptions(CommandLine cmd) {<a name="line.211"></a>
+<span class="sourceLineNo">212</span> processBaseOptions(cmd);<a name="line.212"></a>
+<span class="sourceLineNo">213</span> args = cmd.getArgs();<a name="line.213"></a>
+<span class="sourceLineNo">214</span> }<a name="line.214"></a>
+<span class="sourceLineNo">215</span><a name="line.215"></a>
+<span class="sourceLineNo">216</span> @Override<a name="line.216"></a>
+<span class="sourceLineNo">217</span> public void setUpCluster() throws Exception {<a name="line.217"></a>
+<span class="sourceLineNo">218</span> util = getTestingUtil(getConf());<a name="line.218"></a>
+<span class="sourceLineNo">219</span> boolean isDistributed = util.isDistributedCluster();<a name="line.219"></a>
+<span class="sourceLineNo">220</span> util.initializeCluster(isDistributed ? 1 : 3);<a name="line.220"></a>
+<span class="sourceLineNo">221</span> if (!isDistributed) {<a name="line.221"></a>
+<span class="sourceLineNo">222</span> util.startMiniMapReduceCluster();<a name="line.222"></a>
+<span class="sourceLineNo">223</span> }<a name="line.223"></a>
+<span class="sourceLineNo">224</span> this.setConf(util.getConfiguration());<a name="line.224"></a>
+<span class="sourceLineNo">225</span> }<a name="line.225"></a>
+<span class="sourceLineNo">226</span><a name="line.226"></a>
+<span class="sourceLineNo">227</span> @Override<a name="line.227"></a>
+<span class="sourceLineNo">228</span> public void cleanUpCluster() throws Exception {<a name="line.228"></a>
+<span class="sourceLineNo">229</span> super.cleanUpCluster();<a name="line.229"></a>
+<span class="sourceLineNo">230</span> if (util.isDistributedCluster()) {<a name="line.230"></a>
+<span class="sourceLineNo">231</span> util.shutdownMiniMapReduceCluster();<a name="line.231"></a>
+<span class="sourceLineNo">232</span> }<a name="line.232"></a>
+<span class="sourceLineNo">233</span> }<a name="line.233"></a>
+<span class="sourceLineNo">234</span><a name="line.234"></a>
+<span class="sourceLineNo">235</span> static TableName getTablename(Configuration c) {<a name="line.235"></a>
+<span class="sourceLineNo">236</span> return TableName.valueOf(c.get(TABLE_NAME_KEY, DEFAULT_TABLE_NAME));<a name="line.236"></a>
+<span class="sourceLineNo">237</span> }<a name="line.237"></a>
+<span class="sourceLineNo">238</span><a name="line.238"></a>
+<span class="sourceLineNo">239</span> @Override<a name="line.239"></a>
+<span class="sourceLineNo">240</span> public TableName getTablename() {<a name="line.240"></a>
+<span class="sourceLineNo">241</span> return getTablename(getConf());<a name="line.241"></a>
+<span class="sourceLineNo">242</span> }<a name="line.242"></a>
+<span class="sourceLineNo">243</span><a name="line.243"></a>
+<span class="sourceLineNo">244</span> @Override<a name="line.244"></a>
+<span class="sourceLineNo">245</span> protected Set<String> getColumnFamilies() {<a name="line.245"></a>
+<span class="sourceLineNo">246</span> Set<String> families = new HashSet<>();<a name="line.246"></a>
+<span class="sourceLineNo">247</span> families.add(Bytes.toString(CONTENT_FAMILY_NAME));<a name="line.247"></a>
+<span class="sourceLineNo">248</span> families.add(Bytes.toString(INFO_FAMILY_NAME));<a name="line.248"></a>
+<span class="sourceLineNo">249</span> return families;<a name="line.249"></a>
+<span class="sourceLineNo">250</span> }<a name="line.250"></a>
+<span class="sourceLineNo">251</span><a name="line.251"></a>
+<span class="sourceLineNo">252</span> @Override<a name="line.252"></a>
+<span class="sourceLineNo">253</span> public int runTestFromCommandLine() throws Exception {<a name="line.253"></a>
+<span class="sourceLineNo">254</span> return ToolRunner.run(getConf(), this, args);<a name="line.254"></a>
+<span class="sourceLineNo">255</span> }<a name="line.255"></a>
+<span class="sourceLineNo">256</span><a name="line.256"></a>
+<span class="sourceLineNo">257</span> public static void main(String[] args) throws Exception {<a name="line.257"></a>
+<span class="sourceLineNo">258</span> Configuration conf = HBaseConfiguration.create();<a name="line.258"></a>
+<span class="sourceLineNo">259</span> IntegrationTestingUtility.setUseDistributedCluster(conf);<a name="line.259"></a>
+<span class="sourceLineNo">260</span> int ret = ToolRunner.run(conf, new IntegrationTestLoadCommonCrawl(), args);<a name="line.260"></a>
+<span class="sourceLineNo">261</span> System.exit(ret);<a name="line.261"></a>
+<span class="sourceLineNo">262</span> }<a name="line.262"></a>
+<span class="sourceLineNo">263</span><a name="line.263"></a>
+<span class="sourceLineNo">264</span> public static class HBaseKeyWritable implements Writable {<a name="line.264"></a>
+<span class="sourceLineNo">265</span><a name="line.265"></a>
+<span class="sourceLineNo">266</span> private byte[] row;<a name="line.266"></a>
+<span class="sourceLineNo">267</span> private int rowOffset;<a name="line.267"></a>
+<span class="sourceLineNo">268</span> private int rowLength;<a name="line.268"></a>
+<span class="sourceLineNo">269</span> private byte[] family;<a name="line.269"></a>
+<span class="sourceLineNo">270</span> private int familyOffset;<a name="line.270"></a>
+<span class="sourceLineNo">271</span> private int familyLength;<a name="line.271"></a>
+<span class="sourceLineNo">272</span> private byte[] qualifier;<a name="line.272"></a>
+<span class="sourceLineNo">273</span> private int qualifierOffset;<a name="line.273"></a>
+<span class="sourceLineNo">274</span> private int qualifierLength;<a name="line.274"></a>
+<span class="sourceLineNo">275</span> private long ts;<a name="line.275"></a>
+<span class="sourceLineNo">276</span><a name="line.276"></a>
+<span class="sourceLineNo">277</span> public HBaseKeyWritable() { }<a name="line.277"></a>
+<span class="sourceLineNo">278</span><a name="line.278"></a>
+<span class="sourceLineNo">279</span> public HBaseKeyWritable(byte[] row, int rowOffset, int rowLength,<a name="line.279"></a>
+<span class="sourceLineNo">280</span> byte[] family, int familyOffset, int familyLength,<a name="line.280"></a>
+<span class="sourceLineNo">281</span> byte[] qualifier, int qualifierOffset, int qualifierLength, long ts) {<a name="line.281"></a>
+<span class="sourceLineNo">282</span> this.row = row;<a name="line.282"></a>
+<span class="sourceLineNo">283</span> this.rowOffset = rowOffset;<a name="line.283"></a>
+<span class="sourceLineNo">284</span> this.rowLength = rowLength;<a name="line.284"></a>
+<span class="sourceLineNo">285</span> this.family = family;<a name="line.285"></a>
+<span class="sourceLineNo">286</span> this.familyOffset = familyOffset;<a name="line.286"></a>
+<span class="sourceLineNo">287</span> this.familyLength = familyLength;<a name="line.287"></a>
+<span class="sourceLineNo">288</span> this.qualifier = qualifier;<a name="line.288"></a>
+<span class="sourceLineNo">289</span> this.qualifierOffset = qualifierOffset;<a name="line.289"></a>
+<span class="sourceLineNo">290</span> this.qualifierLength = qualifierLength;<a name="line.290"></a>
+<span class="sourceLineNo">291</span> this.ts = ts;<a name="line.291"></a>
+<span class="sourceLineNo">292</span> }<a name="line.292"></a>
+<span class="sourceLineNo">293</span><a name="line.293"></a>
+<span class="sourceLineNo">294</span> public HBaseKeyWritable(byte[] row, byte[] family, byte[] qualifier, long ts) {<a name="line.294"></a>
+<span class="sourceLineNo">295</span> this(row, 0, row.length,<a name="line.295"></a>
+<span class="sourceLineNo">296</span> family, 0, family.length,<a name="line.296"></a>
+<span class="sourceLineNo">297</span> qualifier, 0, qualifier != null ? qualifier.length : 0, ts);<a name="line.297"></a>
+<span class="sourceLineNo">298</span> }<a name="line.298"></a>
+<span class="sourceLineNo">299</span><a name="line.299"></a>
+<span class="sourceLineNo">300</span> public HBaseKeyWritable(byte[] row, byte[] family, long ts) {<a name="line.300"></a>
+<span class="sourceLineNo">301</span> this(row, family, HConstants.EMPTY_BYTE_ARRAY, ts);<a name="line.301"></a>
+<span class="sourceLineNo">302</span> }<a name="line.302"></a>
+<span class="sourceLineNo">303</span><a name="line.303"></a>
+<span class="sourceLineNo">304</span> public HBaseKeyWritable(Cell cell) {<a name="line.304"></a>
+<span class="sourceLineNo">305</span> this(cell.getRowArray(), cell.getRowOffset(), cell.getRowLength(),<a name="line.305"></a>
+<span class="sourceLineNo">306</span> cell.getFamilyArray(), cell.getFamilyOffset(), cell.getFamilyLength(),<a name="line.306"></a>
+<span class="sourceLineNo">307</span> cell.getQualifierArray(), cell.getQualifierOffset(), cell.getQualifierLength(),<a name="line.307"></a>
+<span class="sourceLineNo">308</span> cell.getTimestamp());<a name="line.308"></a>
+<span class="sourceLineNo">309</span> }<a name="line.309"></a>
+<span class="sourceLineNo">310</span><a name="line.310"></a>
+<span class="sourceLineNo">311</span> @Override<a name="line.311"></a>
+<span class="sourceLineNo">312</span> public void readFields(DataInput in) throws IOException {<a name="line.312"></a>
+<span class="sourceLineNo">313</span> this.row = Bytes.toBytes(in.readUTF());<a name="line.313"></a>
+<span class="sourceLineNo">314</span> this.rowOffset = 0;<a name="line.314"></a>
+<span class="sourceLineNo">315</span> this.rowLength = row.length;<a name="line.315"></a>
+<span class="sourceLineNo">316</span> this.family = Bytes.toBytes(in.readUTF());<a name="line.316"></a>
+<span class="sourceLineNo">317</span> this.familyOffset = 0;<a name="line.317"></a>
+<span class="sourceLineNo">318</span> this.familyLength = family.length;<a name="line.318"></a>
+<span class="sourceLineNo">319</span> this.qualifier = Bytes.toBytes(in.readUTF());<a name="line.319"></a>
+<span class="sourceLineNo">320</span> this.qualifierOffset = 0;<a name="line.320"></a>
+<span class="sourceLineNo">321</span> this.qualifierLength = qualifier.length;<a name="line.321"></a>
+<span class="sourceLineNo">322</span> this.ts = in.readLong();<a name="line.322"></a>
+<span class="sourceLineNo">323</span> }<a name="line.323"></a>
+<span class="sourceLineNo">324</span><a name="line.324"></a>
+<span class="sourceLineNo">325</span> @Override<a name="line.325"></a>
+<span class="sourceLineNo">326</span> public void write(DataOutput out) throws IOException {<a name="line.326"></a>
+<span class="sourceLineNo">327</span> out.writeUTF(new String(row, rowOffset, rowLength, StandardCharsets.UTF_8));<a name="line.327"></a>
+<span class="sourceLineNo">328</span> out.writeUTF(new String(family, familyOffset, familyLength, StandardCharsets.UTF_8));<a name="line.328"></a>
+<span class="sourceLineNo">329</span> if (qualifier != null) {<a name="line.329"></a>
+<span class="sourceLineNo">330</span> out.writeUTF(new String(qualifier, qualifierOffset, qualifierLength,<a name="line.330"></a>
+<span class="sourceLineNo">331</span> StandardCharsets.UTF_8));<a name="line.331"></a>
+<span class="sourceLineNo">332</span> } else {<a name="line.332"></a>
+<span class="sourceLineNo">333</span> out.writeUTF("");<a name="line.333"></a>
+<span class="sourceLineNo">334</span> }<a name="line.334"></a>
+<span class="sourceLineNo">335</span> out.writeLong(ts);<a name="line.335"></a>
+<span class="sourceLineNo">336</span> }<a name="line.336"></a>
+<span class="sourceLineNo">337</span><a name="line.337"></a>
+<span class="sourceLineNo">338</span> public byte[] getRowArray() {<a name="line.338"></a>
+<span class="sourceLineNo">339</span> return row;<a name="line.339"></a>
+<span class="sourceLineNo">340</span> }<a name="line.340"></a>
+<span class="sourceLineNo">341</span><a name="line.341"></a>
+<span class="sourceLineNo">342</span> public void setRow(byte[] row) {<a name="line.342"></a>
+<span class="sourceLineNo">343</span> this.row = row;<a name="line.343"></a>
+<span class="sourceLineNo">344</span> }<a name="line.344"></a>
+<span class="sourceLineNo">345</span><a name="line.345"></a>
+<span class="sourceLineNo">346</span> public int getRowOffset() {<a name="line.346"></a>
+<span class="sourceLineNo">347</span> return rowOffset;<a name="line.347"></a>
+<span class="sourceLineNo">348</span> }<a name="line.348"></a>
+<span class="sourceLineNo">349</span><a name="line.349"></a>
+<span class="sourceLineNo">350</span> public void setRowOffset(int rowOffset) {<a name="line.350"></a>
+<span class="sourceLineNo">351</span> this.rowOffset = rowOffset;<a name="line.351"></a>
+<span class="sourceLineNo">352</span> }<a name="line.352"></a>
+<span class="sourceLineNo">353</span><a name="line.353"></a>
+<span class="sourceLineNo">354</span> public int getRowLength() {<a name="line.354"></a>
+<span class="sourceLineNo">355</span> return rowLength;<a name="line.355"></a>
+<span class="sourceLineNo">356</span> }<a name="line.356"></a>
+<span class="sourceLineNo">357</span><a name="line.357"></a>
+<span class="sourceLineNo">358</span> public void setRowLength(int rowLength) {<a name="line.358"></a>
+<span class="sourceLineNo">359</span> this.rowLength = rowLength;<a name="line.359"></a>
+<span class="sourceLineNo">360</span> }<a name="line.360"></a>
+<span class="sourceLineNo">361</span><a name="line.361"></a>
+<span class="sourceLineNo">362</span> public byte[] getFamilyArray() {<a name="line.362"></a>
+<span class="sourceLineNo">363</span> return family;<a name="line.363"></a>
+<span class="sourceLineNo">364</span> }<a name="line.364"></a>
+<span class="sourceLineNo">365</span><a name="line.365"></a>
+<span class="sourceLineNo">366</span> public void setFamily(byte[] family) {<a name="line.366"></a>
+<span class="sourceLineNo">367</span> this.family = family;<a name="line.367"></a>
+<span class="sourceLineNo">368</span> }<a name="line.368"></a>
+<span class="sourceLineNo">369</span><a name="line.369"></a>
+<span class="sourceLineNo">370</span> public int getFamilyOffset() {<a name="line.370"></a>
+<span class="sourceLineNo">371</span> return familyOffset;<a name="line.371"></a>
+<span class="sourceLineNo">372</span> }<a name="line.372"></a>
+<span class="sourceLineNo">373</span><a name="line.373"></a>
+<span class="sourceLineNo">374</span> public void setFamilyOffset(int familyOffset) {<a name="line.374"></a>
+<span class="sourceLineNo">375</span> this.familyOffset = familyOffset;<a name="line.375"></a>
+<span class="sourceLineNo">376</span> }<a name="line.376"></a>
+<span class="sourceLineNo">377</span><a name="line.377"></a>
+<span class="sourceLineNo">378</span> public int getFamilyLength() {<a name="line.378"></a>
+<span class="sourceLineNo">379</span> return familyLength;<a name="line.379"></a>
+<span class="sourceLineNo">380</span> }<a name="line.380"></a>
+<span class="sourceLineNo">381</span><a name="line.381"></a>
+<span class="sourceLineNo">382</span> public void setFamilyLength(int familyLength) {<a name="line.382"></a>
+<span class="sourceLineNo">383</span> this.familyLength = familyLength;<a name="line.383"></a>
+<span class="sourceLineNo">384</span> }<a name="line.384"></a>
+<span class="sourceLineNo">385</span><a name="line.385"></a>
+<span class="sourceLineNo">386</span> public byte[] getQualifierArray() {<a name="line.386"></a>
+<span class="sourceLineNo">387</span> return qualifier;<a name="line.387"></a>
+<span class="sourceLineNo">388</span> }<a name="line.388"></a>
+<span class="sourceLineNo">389</span><a name="line.389"></a>
+<span class="sourceLineNo">390</span> public void setQualifier(byte[] qualifier) {<a name="line.390"></a>
+<span class="sourceLineNo">391</span> this.qualifier = qualifier;<a name="line.391"></a>
+<span class="sourceLineNo">392</span> }<a name="line.392"></a>
+<span class="sourceLineNo">393</span><a name="line.393"></a>
+<span class="sourceLineNo">394</span> public int getQualifierOffset() {<a name="line.394"></a>
+<span class="sourceLineNo">395</span> return qualifierOffset;<a name="line.395"></a>
+<span class="sourceLineNo">396</span> }<a name="line.396"></a>
+<span class="sourceLineNo">397</span><a name="line.397"></a>
+<span class="sourceLineNo">398</span> public void setQualifierOffset(int qualifierOffset) {<a name="line.398"></a>
+<span class="sourceLineNo">399</span> this.qualifierOffset = qualifierOffset;<a name="line.399"></a>
+<span class="sourceLineNo">400</span> }<a name="line.400"></a>
+<span class="sourceLineNo">401</span><a name="line.401"></a>
+<span class="sourceLineNo">402</span> public int getQualifierLength() {<a name="line.402"></a>
+<span class="sourceLineNo">403</span> return qualifierLength;<a name="line.403"></a>
+<span class="sourceLineNo">404</span> }<a name="line.404"></a>
+<span class="sourceLineNo">405</span><a name="line.405"></a>
+<span class="sourceLineNo">406</span> public void setQualifierLength(int qualifierLength) {<a name="line.406"></a>
+<span class="sourceLineNo">407</span> this.qualifierLength = qualifierLength;<a name="line.407"></a>
+<span class="sourceLineNo">408</span> }<a name="line.408"></a>
+<span class="sourceLineNo">409</span><a name="line.409"></a>
+<span class="sourceLineNo">410</span> public long getTimestamp() {<a name="line.410"></a>
+<span class="sourceLineNo">411</span> return ts;<a name="line.411"></a>
+<span class="sourceLineNo">412</span> }<a name="line.412"></a>
+<span class="sourceLineNo">413</span><a name="line.413"></a>
+<span class="sourceLineNo">414</span> public void setTimestamp(long ts) {<a name="line.414"></a>
+<span class="sourceLineNo">415</span> this.ts = ts;<a name="line.415"></a>
+<span class="sourceLineNo">416</span> }<a name="line.416"></a>
+<span class="sourceLineNo">417</span> }<a name="line.417"></a>
+<span class="sourceLineNo">418</span><a name="line.418"></a>
+<span class="sourceLineNo">419</span> public static class Loader extends Configured implements Tool {<a name="line.419"></a>
<span class="sourceLineNo">420</span><a name="line.420"></a>
-<span class="sourceLineNo">421</span> void createSchema(TableName tableName) throws IOException {<a name="line.421"></a>
-<span class="sourceLineNo">422</span><a name="line.422"></a>
-<span class="sourceLineNo">423</span> try (Connection conn = ConnectionFactory.createConnection(getConf());<a name="line.423"></a>
-<span class="sourceLineNo">424</span> Admin admin = conn.getAdmin()) {<a name="line.424"></a>
-<span class="sourceLineNo">425</span> if (!admin.tableExists(tableName)) {<a name="line.425"></a>
-<span class="sourceLineNo">426</span><a name="line.426"></a>
-<span class="sourceLineNo">427</span> ColumnFamilyDescriptorBuilder contentFamilyBuilder =<a name="line.427"></a>
-<span class="sourceLineNo">428</span> ColumnFamilyDescriptorBuilder.newBuilder(CONTENT_FAMILY_NAME)<a name="line.428"></a>
-<span class="sourceLineNo">429</span> .setDataBlockEncoding(DataBlockEncoding.NONE)<a name="line.429"></a>
-<span class="sourceLineNo">430</span> .setBloomFilterType(BloomType.ROW)<a name="line.430"></a>
-<span class="sourceLineNo">431</span> .setMaxVersions(1000)<a name="line.431"></a>
-<span class="sourceLineNo">432</span> .setBlocksize(256 * 1024)<a name="line.432"></a>
-<span class="sourceLineNo">433</span> ;<a name="line.433"></a>
-<span class="sourceLineNo">434</span><a name="line.434"></a>
-<span class="sourceLineNo">435</span> ColumnFamilyDescriptorBuilder infoFamilyBuilder =<a name="line.435"></a>
-<span class="sourceLineNo">436</span> ColumnFamilyDescriptorBuilder.newBuilder(INFO_FAMILY_NAME)<a name="line.436"></a>
-<span class="sourceLineNo">437</span> .setDataBlockEncoding(DataBlockEncoding.NONE)<a name="line.437"></a>
-<span class="sourceLineNo">438</span> .setBloomFilterType(BloomType.ROWCOL)<a name="line.438"></a>
-<span class="sourceLineNo">439</span> .setMaxVersions(1000)<a name="line.439"></a>
-<span class="sourceLineNo">440</span> .setBlocksize(8 * 1024)<a name="line.440"></a>
-<span class="sourceLineNo">441</span> ;<a name="line.441"></a>
-<span class="sourceLineNo">442</span><a name="line.442"></a>
-<span class="sourceLineNo">443</span> Set<ColumnFamilyDescriptor> families = new HashSet<>();<a name="line.443"></a>
-<span class="sourceLineNo">444</span> families.add(contentFamilyBuilder.build());<a name="line.444"></a>
-<span class="sourceLineNo">445</span> families.add(infoFamilyBuilder.build());<a name="line.445"></a>
-<span class="sourceLineNo">446</span><a name="line.446"></a>
-<span class="sourceLineNo">447</span> TableDescriptor tableDescriptor = TableDescriptorBuilder.newBuilder(tableName)<a name="line.447"></a>
-<span class="sourceLineNo">448</span> .setColumnFamilies(families)<a name="line.448"></a>
-<span class="sourceLineNo">449</span> .build();<a name="line.449"></a>
-<span class="sourceLineNo">450</span><a name="line.450"></a>
-<span class="sourceLineNo">451</span> if (getConf().getBoolean(HBaseTestingUtil.PRESPLIT_TEST_TABLE_KEY,<a name="line.451"></a>
-<span class="sourceLineNo">452</span> HBaseTestingUtil.PRESPLIT_TEST_TABLE)) {<a name="line.452"></a>
-<span class="sourceLineNo">453</span> int numberOfServers = admin.getRegionServers().size();<a name="line.453"></a>
-<span class="sourceLineNo">454</span> if (numberOfServers == 0) {<a name="line.454"></a>
-<span class="sourceLineNo">455</span> throw new IllegalStateException("No live regionservers");<a name="line.455"></a>
-<span class="sourceLineNo">456</span> }<a name="line.456"></a>
-<span class="sourceLineNo">457</span> int regionsPerServer = getConf().getInt(HBaseTestingUtil.REGIONS_PER_SERVER_KEY,<a name="line.457"></a>
-<span class="sourceLineNo">458</span> HBaseTestingUtil.DEFAULT_REGIONS_PER_SERVER);<a name="line.458"></a>
-<span class="sourceLineNo">459</span> int totalNumberOfRegions = numberOfServers * regionsPerServer;<a name="line.459"></a>
-<span class="sourceLineNo">460</span> LOG.info("Creating test table: " + tableDescriptor);<a name="line.460"></a>
-<span class="sourceLineNo">461</span> LOG.info("Number of live regionservers: " + numberOfServers + ", " +<a name="line.461"></a>
-<span class="sourceLineNo">462</span> "pre-splitting table into " + totalNumberOfRegions + " regions " +<a name="line.462"></a>
-<span class="sourceLineNo">463</span> "(default regions per server: " + regionsPerServer + ")");<a name="line.463"></a>
-<span class="sourceLineNo">464</span> byte[][] splits = new RegionSplitter.UniformSplit().split(totalNumberOfRegions);<a name="line.464"></a>
-<span class="sourceLineNo">465</span> admin.createTable(tableDescriptor, splits);<a name="line.465"></a>
-<span class="sourceLineNo">466</span> } else {<a name="line.466"></a>
-<span class="sourceLineNo">467</span> LOG.info("Creating test table: " + tableDescriptor);<a name="line.467"></a>
-<span class="sourceLineNo">468</span> admin.createTable(tableDescriptor);<a name="line.468"></a>
-<span class="sourceLineNo">469</span> }<a name="line.469"></a>
-<span class="sourceLineNo">470</span> }<a name="line.470"></a>
-<span class="sourceLineNo">471</span> } catch (MasterNotRunningException e) {<a name="line.471"></a>
-<span class="sourceLineNo">472</span> LOG.error("Master not running", e);<a name="line.472"></a>
-<span class="sourceLineNo">473</span> throw new IOException(e);<a name="line.473"></a>
-<span class="sourceLineNo">474</span> }<a name="line.474"></a>
-<span class="sourceLineNo">475</span> }<a name="line.475"></a>
-<span class="sourceLineNo">476</span><a name="line.476"></a>
-<span class="sourceLineNo">477</span> int run(Path warcFileInput, Path outputDir)<a name="line.477"></a>
-<span class="sourceLineNo">478</span> throws IOException, ClassNotFoundException, InterruptedException {<a name="line.478"></a>
+<span class="sourceLineNo">421</span> private static final Logger LOG = LoggerFactory.getLogger(Loader.class);<a name="line.421"></a>
+<span class="sourceLineNo">422</span> private static final String USAGE = "Loader <warInputDir | warFileList> <outputDir>";<a name="line.422"></a>
+<span class="sourceLineNo">423</span><a name="line.423"></a>
+<span class="sourceLineNo">424</span> void createSchema(TableName tableName) throws IOException {<a name="line.424"></a>
+<span class="sourceLineNo">425</span><a name="line.425"></a>
+<span class="sourceLineNo">426</span> try (Connection conn = ConnectionFactory.createConnection(getConf());<a name="line.426"></a>
+<span class="sourceLineNo">427</span> Admin admin = conn.getAdmin()) {<a name="line.427"></a>
+<span class="sourceLineNo">428</span> if (!admin.tableExists(tableName)) {<a name="line.428"></a>
+<span class="sourceLineNo">429</span><a name="line.429"></a>
+<span class="sourceLineNo">430</span> ColumnFamilyDescriptorBuilder contentFamilyBuilder =<a name="line.430"></a>
+<span class="sourceLineNo">431</span> ColumnFamilyDescriptorBuilder.newBuilder(CONTENT_FAMILY_NAME)<a name="line.431"></a>
+<span class="sourceLineNo">432</span> .setDataBlockEncoding(DataBlockEncoding.NONE)<a name="line.432"></a>
+<span class="sourceLineNo">433</span> .setBloomFilterType(BloomType.ROW)<a name="line.433"></a>
+<span class="sourceLineNo">434</span> .setMaxVersions(1000)<a name="line.434"></a>
+<span class="sourceLineNo">435</span> .setBlocksize(256 * 1024)<a name="line.435"></a>
+<span class="sourceLineNo">436</span> ;<a name="line.436"></a>
+<span class="sourceLineNo">437</span><a name="line.437"></a>
+<span class="sourceLineNo">438</span> ColumnFamilyDescriptorBuilder infoFamilyBuilder =<a name="line.438"></a>
+<span class="sourceLineNo">439</span> ColumnFamilyDescriptorBuilder.newBuilder(INFO_FAMILY_NAME)<a name="line.439"></a>
+<span class="sourceLineNo">440</span> .setDataBlockEncoding(DataBlockEncoding.NONE)<a name="line.440"></a>
+<span class="sourceLineNo">441</span> .setBloomFilterType(BloomType.ROWCOL)<a name="line.441"></a>
+<span class="sourceLineNo">442</span> .setMaxVersions(1000)<a name="line.442"></a>
+<span class="sourceLineNo">443</span> .setBlocksize(8 * 1024)<a name="line.443"></a>
+<span class="sourceLineNo">444</span> ;<a name="line.444"></a>
+<span class="sourceLineNo">445</span><a name="line.445"></a>
+<span class="sourceLineNo">446</span> Set<ColumnFamilyDescriptor> families = new HashSet<>();<a name="line.446"></a>
+<span class="sourceLineNo">447</span> families.add(contentFamilyBuilder.build());<a name="line.447"></a>
+<span class="sourceLineNo">448</span> families.add(infoFamilyBuilder.build());<a name="line.448"></a>
+<span class="sourceLineNo">449</span><a name="line.449"></a>
+<span class="sourceLineNo">450</span> TableDescriptor tableDescriptor = TableDescriptorBuilder.newBuilder(tableName)<a name="line.450"></a>
+<span class="sourceLineNo">451</span> .setColumnFamilies(families)<a name="line.451"></a>
+<span class="sourceLineNo">452</span> .build();<a name="line.452"></a>
+<span class="sourceLineNo">453</span><a name="line.453"></a>
+<span class="sourceLineNo">454</span> if (getConf().getBoolean(HBaseTestingUtil.PRESPLIT_TEST_TABLE_KEY,<a name="line.454"></a>
+<span class="sourceLineNo">455</span> HBaseTestingUtil.PRESPLIT_TEST_TABLE)) {<a name="line.455"></a>
+<span class="sourceLineNo">456</span> int numberOfServers = admin.getRegionServers().size();<a name="line.456"></a>
+<span class="sourceLineNo">457</span> if (numberOfServers == 0) {<a name="line.457"></a>
+<span class="sourceLineNo">458</span> throw new IllegalStateException("No live regionservers");<a name="line.458"></a>
+<span class="sourceLineNo">459</span> }<a name="line.459"></a>
+<span class="sourceLineNo">460</span> int regionsPerServer = getConf().getInt(HBaseTestingUtil.REGIONS_PER_SERVER_KEY,<a name="line.460"></a>
+<span class="sourceLineNo">461</span> HBaseTestingUtil.DEFAULT_REGIONS_PER_SERVER);<a name="line.461"></a>
+<span class="sourceLineNo">462</span> int totalNumberOfRegions = numberOfServers * regionsPerServer;<a name="line.462"></a>
+<span class="sourceLineNo">463</span> LOG.info("Creating test table: " + tableDescriptor);<a name="line.463"></a>
+<span class="sourceLineNo">464</span> LOG.info("Number of live regionservers: " + numberOfServers + ", " +<a name="line.464"></a>
+<span class="sourceLineNo">465</span> "pre-splitting table into " + totalNumberOfRegions + " regions " +<a name="line.465"></a>
+<span class="sourceLineNo">466</span> "(default regions per server: " + regionsPerServer + ")");<a name="line.466"></a>
+<span class="sourceLineNo">467</span> byte[][] splits = new RegionSplitter.UniformSplit().split(totalNumberOfRegions);<a name="line.467"></a>
+<span class="sourceLineNo">468</span> admin.createTable(tableDescriptor, splits);<a name="line.468"></a>
+<span class="sourceLineNo">469</span> } else {<a name="line.469"></a>
+<span class="sourceLineNo">470</span> LOG.info("Creating test table: " + tableDescriptor);<a name="line.470"></a>
+<span class="sourceLineNo">471</span> admin.createTable(tableDescriptor);<a name="line.471"></a>
+<span class="sourceLineNo">472</span> }<a name="line.472"></a>
+<span class="sourceLineNo">473</span> }<a name="line.473"></a>
+<span class="sourceLineNo">474</span> } catch (MasterNotRunningException e) {<a name="line.474"></a>
+<span class="sourceLineNo">475</span> LOG.error("Master not running", e);<a name="line.475"></a>
+<span class="sourceLineNo">476</span> throw new IOException(e);<a name="line.476"></a>
+<span class="sourceLineNo">477</span> }<a name="line.477"></a>
+<span class="sourceLineNo">478</span> }<a name="line.478"></a>
<span class="sourceLineNo">479</span><a name="line.479"></a>
-<span class="sourceLineNo">480</span> createSchema(getTablename(getConf()));<a name="line.480"></a>
-<span class="sourceLineNo">481</span><a name="line.481"></a>
-<span class="sourceLineNo">482</span> Job job = Job.getInstance(getConf());<a name="line.482"></a>
-<span class="sourceLineNo">483</span> job.setJobName(Loader.class.getName());<a name="line.483"></a>
-<span class="sourceLineNo">484</span> job.setNumReduceTasks(0);<a name="line.484"></a>
-<span class="sourceLineNo">485</span> job.setJarByClass(getClass());<a name="line.485"></a>
-<span class="sourceLineNo">486</span> job.setMapperClass(LoaderMapper.class);<a name="line.486"></a>
-<span class="sourceLineNo">487</span> job.setInputFormatClass(WARCInputFormat.class);<a name="line.487"></a>
-<span class="sourceLineNo">488</span> FileSystem fs = FileSystem.get(warcFileInput.toUri(), getConf());<a name="line.488"></a>
-<span class="sourceLineNo">489</span> if (fs.getFileStatus(warcFileInput).isDirectory()) {<a name="line.489"></a>
-<span class="sourceLineNo">490</span> LOG.info("Using directory as WARC input path: " + warcFileInput);<a name="line.490"></a>
-<span class="sourceLineNo">491</span> FileInputFormat.setInputPaths(job, warcFileInput);<a name="line.491"></a>
-<span class="sourceLineNo">492</span> } else {<a name="line.492"></a>
-<span class="sourceLineNo">493</span> LOG.info("Getting WARC input paths from file: " + warcFileInput);<a name="line.493"></a>
-<span class="sourceLineNo">494</span> List<Path> paths = new LinkedList<Path>();<a name="line.494"></a>
-<span class="sourceLineNo">495</span> try (FSDataInputStream is = fs.open(warcFileInput)) {<a name="line.495"></a>
-<span class="sourceLineNo">496</span> InputStreamReader reader;<a name="line.496"></a>
-<span class="sourceLineNo">497</span> if (warcFileInput.getName().toLowerCase().endsWith(".gz")) {<a name="line.497"></a>
-<span class="sourceLineNo">498</span> reader = new InputStreamReader(new GZIPInputStream(is));<a name="line.498"></a>
-<span class="sourceLineNo">499</span> } else {<a name="line.499"></a>
-<span class="sourceLineNo">500</span> reader = new InputStreamReader(is);<a name="line.500"></a>
-<span class="sourceLineNo">501</span> }<a name="line.501"></a>
-<span class="sourceLineNo">502</span> try (BufferedReader br = new BufferedReader(reader)) {<a name="line.502"></a>
-<span class="sourceLineNo">503</span> String line;<a name="line.503"></a>
-<span class="sourceLineNo">504</span> while ((line = br.readLine()) != null) {<a name="line.504"></a>
-<span class="sourceLineNo">505</span> paths.add(new Path(line));<a name="line.505"></a>
-<span class="sourceLineNo">506</span> }<a name="line.506"></a>
-<span class="sourceLineNo">507</span> }<a name="line.507"></a>
-<span class="sourceLineNo">508</span> }<a name="line.508"></a>
-<span class="sourceLineNo">509</span> LOG.info("Read " + paths.size() + " WARC input paths from " + warcFileInput);<a name="line.509"></a>
-<span class="sourceLineNo">510</span> FileInputFormat.setInputPaths(job, paths.toArray(new Path[paths.size()]));<a name="line.510"></a>
-<span class="sourceLineNo">511</span> }<a name="line.511"></a>
-<span class="sourceLineNo">512</span> job.setOutputFormatClass(SequenceFileOutputFormat.class);<a name="line.512"></a>
-<span class="sourceLineNo">513</span> SequenceFileOutputFormat.setOutputPath(job, outputDir);<a name="line.513"></a>
-<span class="sourceLineNo">514</span> SequenceFileOutputFormat.setOutputCompressionType(job, CompressionType.BLOCK);<a name="line.514"></a>
-<span class="sourceLineNo">515</span> job.setOutputKeyClass(HBaseKeyWritable.class);<a name="line.515"></a>
-<span class="sourceLineNo">516</span> job.setOutputValueClass(BytesWritable.class);<a name="line.516"></a>
-<span class="sourceLineNo">517</span> TableMapReduceUtil.addDependencyJars(job);<a name="line.517"></a>
-<span class="sourceLineNo">518</span><a name="line.518"></a>
-<span class="sourceLineNo">519</span> LOG.info("Submitting job." +<a name="line.519"></a>
-<span class="sourceLineNo">520</span> " This will take time proportional to the number of input files, please be patient.");<a name="line.520"></a>
-<span class="sourceLineNo">521</span> boolean success = job.waitForCompletion(true);<a name="line.521"></a>
-<span class="sourceLineNo">522</span> if (!success) {<a name="line.522"></a>
-<span class="sourceLineNo">523</span> LOG.error("Failure during job " + job.getJobID());<a name="line.523"></a>
-<span class="sourceLineNo">524</span> }<a name="line.524"></a>
-<span class="sourceLineNo">525</span> return success ? 0 : 1;<a name="line.525"></a>
-<span class="sourceLineNo">526</span> }<a name="line.526"></a>
-<span class="sourceLineNo">527</span><a name="line.527"></a>
-<span class="sourceLineNo">528</span> @Override<a name="line.528"></a>
-<span class="sourceLineNo">529</span> public int run(String[] args) throws Exception {<a name="line.529"></a>
-<span class="sourceLineNo">530</span> if (args.length < 2) {<a name="line.530"></a>
-<span class="sourceLineNo">531</span> System.err.println(USAGE);<a name="line.531"></a>
-<span class="sourceLineNo">532</span> return 1;<a name="line.532"></a>
-<span class="sourceLineNo">533</span> }<a name="line.533"></a>
-<span class="sourceLineNo">534</span> try {<a name="line.534"></a>
-<span class="sourceLineNo">535</span> Path warcFileInput = new Path(args[0]);<a name="line.535"></a>
-<span class="sourceLineNo">536</span> Path outputDir = new Path(args[1]);<a name="line.536"></a>
-<span class="sourceLineNo">537</span> return run(warcFileInput, outputDir);<a name="line.537"></a>
-<span class="sourceLineNo">538</span> } catch (NumberFormatException e) {<a name="line.538"></a>
-<span class="sourceLineNo">539</span> System.err.println("Parsing loader arguments failed: " + e.getMessage());<a name="line.539"></a>
-<span class="sourceLineNo">540</span> System.err.println(USAGE);<a name="line.540"></a>
-<span class="sourceLineNo">541</span> return 1;<a name="line.541"></a>
-<span class="sourceLineNo">542</span> }<a name="line.542"></a>
-<span class="sourceLineNo">543</span> }<a name="line.543"></a>
-<span class="sourceLineNo">544</span><a name="line.544"></a>
-<span class="sourceLineNo">545</span> public static void main(String[] args) throws Exception {<a name="line.545"></a>
-<span class="sourceLineNo">546</span> System.exit(ToolRunner.run(HBaseConfiguration.create(), new Loader(), args));<a name="line.546"></a>
-<span class="sourceLineNo">547</span> }<a name="line.547"></a>
-<span class="sourceLineNo">548</span><a name="line.548"></a>
-<span class="sourceLineNo">549</span> public static class LoaderMapper<a name="line.549"></a>
-<span class="sourceLineNo">550</span> extends Mapper<LongWritable, WARCWritable, HBaseKeyWritable, BytesWritable> {<a name="line.550"></a>
-<span class="sourceLineNo">551</span><a name="line.551"></a>
-<span class="sourceLineNo">552</span> Configuration conf;<a name="line.552"></a>
-<span class="sourceLineNo">553</span> Connection conn;<a name="line.553"></a>
-<span class="sourceLineNo">554</span> Table table;<a name="line.554"></a>
-<span class="sourceLineNo">555</span><a name="line.555"></a>
-<span class="sourceLineNo">556</span> @Override<a name="line.556"></a>
-<span class="sourceLineNo">557</span> protected void setup(Context context) throws IOException, InterruptedException {<a name="line.557"></a>
-<span class="sourceLineNo">558</span> conn = ConnectionFactory.createConnection(context.getConfiguration());<a name="line.558"></a>
-<span class="sourceLineNo">559</span> table = conn.getTable(getTablename(conn.getConfiguration()));<a name="line.559"></a>
-<span class="sourceLineNo">560</span> }<a name="line.560"></a>
-<span class="sourceLineNo">561</span><a name="line.561"></a>
-<span class="sourceLineNo">562</span> @Override<a name="line.562"></a>
-<span class="sourceLineNo">563</span> protected void cleanup(Context context) throws IOException, InterruptedException {<a name="line.563"></a>
-<span class="sourceLineNo">564</span> try {<a name="line.564"></a>
-<span class="sourceLineNo">565</span> table.close();<a name="line.565"></a>
-<span class="sourceLineNo">566</span> } catch (Exception e) {<a name="line.566"></a>
-<span class="sourceLineNo">567</span> LOG.warn("Exception closing Table", e);<a name="line.567"></a>
-<span class="sourceLineNo">568</span> }<a name="line.568"></a>
-<span class="sourceLineNo">569</span> try {<a name="line.569"></a>
-<span class="sourceLineNo">570</span> conn.close();<a name="line.570"></a>
-<span class="sourceLineNo">571</span> } catch (Exception e) {<a name="line.571"></a>
-<span class="sourceLineNo">572</span> LOG.warn("Exception closing Connection", e);<a name="line.572"></a>
-<span class="sourceLineNo">573</span> }<a name="line.573"></a>
-<span class="sourceLineNo">574</span> }<a name="line.574"></a>
-<span class="sourceLineNo">575</span><a name="line.575"></a>
-<span class="sourceLineNo">576</span> @Override<a name="line.576"></a>
-<span class="sourceLineNo">577</span> protected void map(LongWritable key, WARCWritable value, Context output)<a name="line.577"></a>
-<span class="sourceLineNo">578</span> throws IOException, InterruptedException {<a name="line.578"></a>
-<span class="sourceLineNo">579</span> WARCRecord.Header warcHeader = value.getRecord().getHeader();<a name="line.579"></a>
-<span class="sourceLineNo">580</span> String recordID = warcHeader.getRecordID();<a name="line.580"></a>
-<span class="sourceLineNo">581</span> String targetURI = warcHeader.getTargetURI();<a name="line.581"></a>
-<span class="sourceLineNo">582</span> if (warcHeader.getRecordType().equals("response") && targetURI != null) {<a name="line.582"></a>
-<span class="sourceLineNo">583</span> String contentType = warcHeader.getField("WARC-Identified-Payload-Type");<a name="line.583"></a>
-<span class="sourceLineNo">584</span> if (contentType != null) {<a name="line.584"></a>
-<span class="sourceLineNo">585</span> LOG.debug("Processing record id=" + recordID + ", targetURI=\"" + targetURI + "\"");<a name="line.585"></a>
-<span class="sourceLineNo">586</span> long now = EnvironmentEdgeManager.currentTime();<a name="line.586"></a>
-<span class="sourceLineNo">587</span><a name="line.587"></a>
-<span class="sourceLineNo">588</span> // Make row key<a name="line.588"></a>
-<span class="sourceLineNo">589</span><a name="line.589"></a>
-<span class="sourceLineNo">590</span> byte[] rowKey;<a name="line.590"></a>
-<span class="sourceLineNo">591</span> try {<a name="line.591"></a>
-<span class="sourceLineNo">592</span> rowKey = rowKeyFromTargetURI(targetURI);<a name="line.592"></a>
-<span class="sourceLineNo">593</span> } catch (IllegalArgumentException e) {<a name="line.593"></a>
-<span class="sourceLineNo">594</span> LOG.debug("Could not make a row key for record " + recordID + ", ignoring", e);<a name="line.594"></a>
-<span class="sourceLineNo">595</span> return;<a name="line.595"></a>
-<span class="sourceLineNo">596</span> } catch (URISyntaxException e) {<a name="line.596"></a>
-<span class="sourceLineNo">597</span> LOG.warn("Could not parse URI \"" + targetURI + "\" for record " + recordID +<a name="line.597"></a>
-<span class="sourceLineNo">598</span> ", ignoring");<a name="line.598"></a>
-<span class="sourceLineNo">599</span> return;<a name="line.599"></a>
-<span class="sourceLineNo">600</span> }<a name="line.600"></a>
-<span class="sourceLineNo">601</span><a name="line.601"></a>
-<span class="sourceLineNo">602</span> // Get the content and calculate the CRC64<a name="line.602"></a>
-<span class="sourceLineNo">603</span><a name="line.603"></a>
-<span class="sourceLineNo">604</span> byte[] content = value.getRecord().getContent();<a name="line.604"></a>
-<span class="sourceLineNo">605</span> CRC64 crc = new CRC64();<a name="line.605"></a>
-<span class="sourceLineNo">606</span> crc.update(content);<a name="line.606"></a>
-<span class="sourceLineNo">607</span> long crc64 = crc.getValue();<a name="line.607"></a>
-<span class="sourceLineNo">608</span><a name="line.608"></a>
-<span class="sourceLineNo">609</span> // Store to HBase<a name="line.609"></a>
-<span class="sourceLineNo">610</span><a name="line.610"></a>
-<span class="sourceLineNo">611</span> Put put = new Put(rowKey);<a name="line.611"></a>
-<span class="sourceLineNo">612</span> put.addColumn(CONTENT_FAMILY_NAME, CONTENT_QUALIFIER, now, content);<a name="line.612"></a>
-<span class="sourceLineNo">613</span> put.addColumn(INFO_FAMILY_NAME, CONTENT_LENGTH_QUALIFIER, now,<a name="line.613"></a>
-<span class="sourceLineNo">614</span> Bytes.toBytes(content.length));<a name="line.614"></a>
-<span class="sourceLineNo">615</span> put.addColumn(INFO_FAMILY_NAME, CONTENT_TYPE_QUALIFIER, now,<a name="line.615"></a>
-<span class="sourceLineNo">616</span> Bytes.toBytes(contentType));<a name="line.616"></a>
-<span class="sourceLineNo">617</span> put.addColumn(INFO_FAMILY_NAME, CRC_QUALIFIER, now, Bytes.toBytes(crc64));<a name="line.617"></a>
-<span class="sourceLineNo">618</span> put.addColumn(INFO_FAMILY_NAME, RECORD_ID_QUALIFIER, now, Bytes.toBytes(recordID));<a name="line.618"></a>
-<span class="sourceLineNo">619</span> put.addColumn(INFO_FAMILY_NAME, TARGET_URI_QUALIFIER, now, Bytes.toBytes(targetURI));<a name="line.619"></a>
-<span class="sourceLineNo">620</span> put.addColumn(INFO_FAMILY_NAME, DATE_QUALIFIER, now,<a name="line.620"></a>
-<span class="sourceLineNo">621</span> Bytes.toBytes(warcHeader.getDateString()));<a name="line.621"></a>
-<span class="sourceLineNo">622</span> String ipAddr = warcHeader.getField("WARC-IP-Address");<a name="line.622"></a>
-<span class="sourceLineNo">623</span> if (ipAddr != null) {<a name="line.623"></a>
-<span class="sourceLineNo">624</span> put.addColumn(INFO_FAMILY_NAME, IP_ADDRESS_QUALIFIER, now, Bytes.toBytes(ipAddr));<a name="line.624"></a>
-<span class="sourceLineNo">625</span> }<a name="line.625"></a>
-<span class="sourceLineNo">626</span> table.put(put);<a name="line.626"></a>
-<span class="sourceLineNo">627</span><a name="line.627"></a>
-<span class="sourceLineNo">628</span> // Write records out for later verification, one per HBase field except for the<a name="line.628"></a>
-<span class="sourceLineNo">629</span> // content record, which will be verified by CRC64.<a name="line.629"></a>
+<span class="sourceLineNo">480</span> int run(Path warcFileInput, Path outputDir)<a name="line.480"></a>
+<span class="sourceLineNo">481</span> throws IOException, ClassNotFoundException, InterruptedException {<a name="line.481"></a>
+<span class="sourceLineNo">482</span><a name="line.482"></a>
+<span class="sourceLineNo">483</span> createSchema(getTablename(getConf()));<a name="line.483"></a>
+<span class="sourceLineNo">484</span><a name="line.484"></a>
+<span class="sourceLineNo">485</span> Job job = Job.getInstance(getConf());<a name="line.485"></a>
+<span class="sourceLineNo">486</span> job.setJobName(Loader.class.getName());<a name="line.486"></a>
+<span class="sourceLineNo">487</span> job.setNumReduceTasks(0);<a name="line.487"></a>
+<span class="sourceLineNo">488</span> job.setJarByClass(getClass());<a name="line.488"></a>
+<span class="sourceLineNo">489</span> job.setMapperClass(LoaderMapper.class);<a name="line.489"></a>
+<span class="sourceLineNo">490</span> job.setInputFormatClass(WARCInputFormat.class);<a name="line.490"></a>
+<span class="sourceLineNo">491</span> FileSystem fs = FileSystem.get(warcFileInput.toUri(), getConf());<a name="line.491"></a>
+<span class="sourceLineNo">492</span> if (fs.getFileStatus(warcFileInput).isDirectory()) {<a name="line.492"></a>
+<span class="sourceLineNo">493</span> LOG.info("Using directory as WARC input path: " + warcFileInput);<a name="line.493"></a>
+<span class="sourceLineNo">494</span> FileInputFormat.setInputPaths(job, warcFileInput);<a name="line.494"></a>
+<span class="sourceLineNo">495</span> } else {<a name="line.495"></a>
+<span class="sourceLineNo">496</span> LOG.info("Getting WARC input paths from file: " + warcFileInput);<a name="line.496"></a>
+<span class="sourceLineNo">497</span> List<Path> paths = new LinkedList<Path>();<a name="line.497"></a>
+<span class="sourceLineNo">498</span> try (FSDataInputStream is = fs.open(warcFileInput)) {<a name="line.498"></a>
+<span class="sourceLineNo">499</span> InputStreamReader reader;<a name="line.499"></a>
+<span class="sourceLineNo">500</span> if (warcFileInput.getName().toLowerCase().endsWith(".gz")) {<a name="line.500"></a>
+<span class="sourceLineNo">501</span> reader = new InputStreamReader(new GZIPInputStream(is));<a name="line.501"></a>
+<span class="sourceLineNo">502</span> } else {<a name="line.502"></a>
+<span class="sourceLineNo">503</span> reader = new InputStreamReader(is);<a name="line.503"></a>
+<span class="sourceLineNo">504</span> }<a name="line.504"></a>
+<span class="sourceLineNo">505</span> try (BufferedReader br = new BufferedReader(reader)) {<a name="line.505"></a>
+<span class="sourceLineNo">506</span> String line;<a name="line.506"></a>
+<span class="sourceLineNo">507</span> while ((line = br.readLine()) != null) {<a name="line.507"></a>
+<span class="sourceLineNo">508</span> paths.add(new Path(line));<a name="line.508"></a>
+<span class="sourceLineNo">509</span> }<a name="line.509"></a>
+<span class="sourceLineNo">510</span> }<a name="line.510"></a>
+<span class="sourceLineNo">511</span> }<a name="line.511"></a>
+<span class="sourceLineNo">512</span> LOG.info("Read " + paths.size() + " WARC input paths from " + warcFileInput);<a name="line.512"></a>
+<span class="sourceLineNo">513</span> FileInputFormat.setInputPaths(job, paths.toArray(new Path[paths.size()]));<a name="line.513"></a>
+<span class="sourceLineNo">514</span> }<a name="line.514"></a>
+<span class="sourceLineNo">515</span> job.setOutputFormatClass(SequenceFileOutputFormat.class);<a name="line.515"></a>
+<span class="sourceLineNo">516</span> SequenceFileOutputFormat.setOutputPath(job, outputDir);<a name="line.516"></a>
+<span class="sourceLineNo">517</span> SequenceFileOutputFormat.setOutputCompressionType(job, CompressionType.BLOCK);<a name="line.517"></a>
+<span class="sourceLineNo">518</span> job.setOutputKeyClass(HBaseKeyWritable.class);<a name="line.518"></a>
+<span class="sourceLineNo">519</span> job.setOutputValueClass(BytesWritable.class);<a name="line.519"></a>
+<span class="sourceLineNo">520</span> TableMapReduceUtil.addDependencyJars(job);<a name="line.520"></a>
+<span class="sourceLineNo">521</span><a name="line.521"></a>
+<span class="sourceLineNo">522</span> boolean success = job.waitForCompletion(true);<a name="line.522"></a>
+<span class="sourceLineNo">523</span> if (!success) {<a name="line.523"></a>
+<span class="sourceLineNo">524</span> LOG.error("Failure during job " + job.getJobID());<a name="line.524"></a>
+<span class="sourceLineNo">525</span> }<a name="line.525"></a>
+<span class="sourceLineNo">526</span> return success ? 0 : 1;<a name="line.526"></a>
+<span class="sourceLineNo">527</span> }<a name="line.527"></a>
+<span class="sourceLineNo">528</span><a name="line.528"></a>
+<span class="sourceLineNo">529</span> @Override<a name="line.529"></a>
+<span class="sourceLineNo">530</span> public int run(String[] args) throws Exception {<a name="line.530"></a>
+<span class="sourceLineNo">531</span> if (args.length < 2) {<a name="line.531"></a>
+<span class="sourceLineNo">532</span> System.err.println(USAGE);<a name="line.532"></a>
+<span class="sourceLineNo">533</span> return 1;<a name="line.533"></a>
+<span class="sourceLineNo">534</span> }<a name="line.534"></a>
+<span class="sourceLineNo">535</span> try {<a name="line.535"></a>
+<span class="sourceLineNo">536</span> Path warcFileInput = new Path(args[0]);<a name="line.536"></a>
+<span class="sourceLineNo">537</span> Path outputDir = new Path(args[1]);<a name="line.537"></a>
+<span class="sourceLineNo">538</span> return run(warcFileInput, outputDir);<a name="line.538"></a>
+<span class="sourceLineNo">539</span> } catch (NumberFormatException e) {<a name="line.539"></a>
+<span class="sourceLineNo">540</span> System.err.println("Parsing loader arguments failed: " + e.getMessage());<a name="line.540"></a>
+<span class="sourceLineNo">541</span> System.err.println(USAGE);<a name="line.541"></a>
+<span class="sourceLineNo">542</span> return 1;<a name="line.542"></a>
+<span class="sourceLineNo">543</span> }<a name="line.543"></a>
+<span class="sourceLineNo">544</span> }<a name="line.544"></a>
+<span class="sourceLineNo">545</span><a name="line.545"></a>
+<span class="sourceLineNo">546</span> public static void main(String[] args) throws Exception {<a name="line.546"></a>
+<span class="sourceLineNo">547</span> System.exit(ToolRunner.run(HBaseConfiguration.create(), new Loader(), args));<a name="line.547"></a>
+<span class="sourceLineNo">548</span> }<a name="line.548"></a>
+<span class="sourceLineNo">549</span><a name="line.549"></a>
+<span class="sourceLineNo">550</span> public static class LoaderMapper<a name="line.550"></a>
+<span class="sourceLineNo">551</span> extends Mapper<LongWritable, WARCWritable, HBaseKeyWritable, BytesWritable> {<a name="line.551"></a>
+<span class="sourceLineNo">552</span><a name="line.552"></a>
+<span class="sourceLineNo">553</span> Configuration conf;<a name="line.553"></a>
+<span class="sourceLineNo">554</span> Connection conn;<a name="line.554"></a>
+<span class="sourceLineNo">555</span> BufferedMutator mutator;<a name="line.555"></a>
+<span class="sourceLineNo">556</span><a name="line.556"></a>
+<span class="sourceLineNo">557</span> @Override<a name="line.557"></a>
+<span class="sourceLineNo">558</span> protected void setup(Context context) throws IOException, InterruptedException {<a name="line.558"></a>
+<span class="sourceLineNo">559</span> conf = context.getConfiguration();<a name="line.559"></a>
+<span class="sourceLineNo">560</span> conn = ConnectionFactory.createConnection(conf);<a name="line.560"></a>
+<span class="sourceLineNo">561</span> mutator = conn.getBufferedMutator(getTablename(conf));<a name="line.561"></a>
+<span class="sourceLineNo">562</span> mutator.setWriteBufferPeriodicFlush(10 * 1000); // default is 1 sec, increase to 10<a name="line.562"></a>
+<span class="sourceLineNo">563</span> }<a name="line.563"></a>
+<span class="sourceLineNo">564</span><a name="line.564"></a>
+<span class="sourceLineNo">565</span> @Override<a name="line.565"></a>
+<span class="sourceLineNo">566</span> protected void cleanup(Context context) throws IOException, InterruptedException {<a name="line.566"></a>
+<span class="sourceLineNo">567</span> try {<a name="line.567"></a>
+<span class="sourceLineNo">568</span> mutator.close();<a name="line.568"></a>
+<span class="sourceLineNo">569</span> } catch (Exception e) {<a name="line.569"></a>
+<span class="sourceLineNo">570</span> LOG.warn("Exception closing Table", e);<a name="line.570"></a>
+<span class="sourceLineNo">571</span> }<a name="line.571"></a>
+<span class="sourceLineNo">572</span> try {<a name="line.572"></a>
+<span class="sourceLineNo">573</span> conn.close();<a name="line.573"></a>
+<span class="sourceLineNo">574</span> } catch (Exception e) {<a name="line.574"></a>
+<span class="sourceLineNo">575</span> LOG.warn("Exception closing Connection", e);<a name="line.575"></a>
+<span class="sourceLineNo">576</span> }<a name="line.576"></a>
+<span class="sourceLineNo">577</span> }<a name="line.577"></a>
+<span class="sourceLineNo">578</span><a name="line.578"></a>
+<span class="sourceLineNo">579</span> @Override<a name="line.579"></a>
+<span class="sourceLineNo">580</span> protected void map(LongWritable key, WARCWritable value, Context output)<a name="line.580"></a>
+<span class="sourceLineNo">581</span> throws IOException, InterruptedException {<a name="line.581"></a>
+<span class="sourceLineNo">582</span> WARCRecord.Header warcHeader = value.getRecord().getHeader();<a name="line.582"></a>
+<span class="sourceLineNo">583</span> String recordID = warcHeader.getRecordID();<a name="line.583"></a>
+<span class="sourceLineNo">584</span> String targetURI = warcHeader.getTargetURI();<a name="line.584"></a>
+<span class="sourceLineNo">585</span> if (warcHeader.getRecordType().equals("response") && targetURI != null) {<a name="line.585"></a>
+<span class="sourceLineNo">586</span> String contentType = warcHeader.getField("WARC-Identified-Payload-Type");<a name="line.586"></a>
+<span class="sourceLineNo">587</span> if (contentType != null) {<a name="line.587"></a>
+<span class="sourceLineNo">588</span> LOG.info("Processing uri=\"" + targetURI + "\", id=" + recordID);<a name="line.588"></a>
+<span class="sourceLineNo">589</span> long now = EnvironmentEdgeManager.currentTime();<a name="line.589"></a>
+<span class="sourceLineNo">590</span><a name="line.590"></a>
+<span class="sourceLineNo">591</span> // Make row key<a name="line.591"></a>
+<span class="sourceLineNo">592</span><a name="line.592"></a>
+<span class="sourceLineNo">593</span> byte[] rowKey;<a name="line.593"></a>
+<span class="sourceLineNo">594</span> try {<a name="line.594"></a>
+<span class="sourceLineNo">595</span> rowKey = rowKeyFromTargetURI(targetURI);<a name="line.595"></a>
+<span class="sourceLineNo">596</span> } catch (IllegalArgumentException e) {<a name="line.596"></a>
+<span class="sourceLineNo">597</span> LOG.debug("Could not make a row key for record " + recordID + ", ignoring", e);<a name="line.597"></a>
+<span class="sourceLineNo">598</span> return;<a name="line.598"></a>
+<span class="sourceLineNo">599</span> } catch (URISyntaxException e) {<a name="line.599"></a>
+<span class="sourceLineNo">600</span> LOG.warn("Could not parse URI \"" + targetURI + "\" for record " + recordID +<a name="line.600"></a>
+<span class="sourceLineNo">601</span> ", ignoring");<a name="line.601"></a>
+<span class="sourceLineNo">602</span> return;<a name="line.602"></a>
+<span class="sourceLineNo">603</span> }<a name="line.603"></a>
+<span class="sourceLineNo">604</span><a name="line.604"></a>
+<span class="sourceLineNo">605</span> // Get the content and calculate the CRC64<a name="line.605"></a>
+<span class="sourceLineNo">606</span><a name="line.606"></a>
+<span class="sourceLineNo">607</span> byte[] content = value.getRecord().getContent();<a name="line.607"></a>
+<span class="sourceLineNo">608</span> CRC64 crc = new CRC64();<a name="line.608"></a>
+<span class="sourceLineNo">609</span> crc.update(content);<a name="line.609"></a>
+<span class="sourceLineNo">610</span> long crc64 = crc.getValue();<a name="line.610"></a>
+<span class="sourceLineNo">611</span><a name="line.611"></a>
+<span class="sourceLineNo">612</span> // Store to HBase<a name="line.612"></a>
+<span class="sourceLineNo">613</span><a name="line.613"></a>
+<span class="sourceLineNo">614</span> Put put = new Put(rowKey);<a name="line.614"></a>
+<span class="sourceLineNo">615</span> put.addColumn(CONTENT_FAMILY_NAME, CONTENT_QUALIFIER, now, content);<a name="line.615"></a>
+<span class="sourceLineNo">616</span> put.addColumn(INFO_FAMILY_NAME, CONTENT_LENGTH_QUALIFIER, now,<a name="line.616"></a>
+<span class="sourceLineNo">617</span> Bytes.toBytes(content.length));<a name="line.617"></a>
+<span class="sourceLineNo">618</span> put.addColumn(INFO_FAMILY_NAME, CONTENT_TYPE_QUALIFIER, now,<a name="line.618"></a>
+<span class="sourceLineNo">619</span> Bytes.toBytes(contentType));<a name="line.619"></a>
+<span class="sourceLineNo">620</span> put.addColumn(INFO_FAMILY_NAME, CRC_QUALIFIER, now, Bytes.toBytes(crc64));<a name="line.620"></a>
+<span class="sourceLineNo">621</span> put.addColumn(INFO_FAMILY_NAME, RECORD_ID_QUALIFIER, now, Bytes.toBytes(recordID));<a name="line.621"></a>
+<span class="sourceLineNo">622</span> put.addColumn(INFO_FAMILY_NAME, TARGET_URI_QUALIFIER, now, Bytes.toBytes(targetURI));<a name="line.622"></a>
+<span class="sourceLineNo">623</span> put.addColumn(INFO_FAMILY_NAME, DATE_QUALIFIER, now,<a name="line.623"></a>
+<span class="sourceLineNo">624</span> Bytes.toBytes(warcHeader.getDateString()));<a name="line.624"></a>
+<span class="sourceLineNo">625</span> String ipAddr = warcHeader.getField("WARC-IP-Address");<a name="line.625"></a>
+<span class="sourceLineNo">626</span> if (ipAddr != null) {<a name="line.626"></a>
+<span class="sourceLineNo">627</span> put.addColumn(INFO_FAMILY_NAME, IP_ADDRESS_QUALIFIER, now, Bytes.toBytes(ipAddr));<a name="line.627"></a>
+<span class="sourceLineNo">628</span> }<a name="line.628"></a>
+<span class="sourceLineNo">629</span> mutator.mutate(put);<a name="line.629"></a>
<span class="sourceLineNo">630</span><a name="line.630"></a>
-<span class="sourceLineNo">631</span> output.write(new HBaseKeyWritable(rowKey, INFO_FAMILY_NAME, CRC_QUALIFIER, now),<a name="line.631"></a>
-<span class="sourceLineNo">632</span> new BytesWritable(Bytes.toBytes(crc64)));<a name="line.632"></a>
-<span class="sourceLineNo">633</span> output.write(new HBaseKeyWritable(rowKey, INFO_FAMILY_NAME, CONTENT_LENGTH_QUALIFIER,<a name="line.633"></a>
-<span class="sourceLineNo">634</span> now), new BytesWritable(Bytes.toBytes(content.length)));<a name="line.634"></a>
-<span class="sourceLineNo">635</span> output.write(new HBaseKeyWritable(rowKey, INFO_FAMILY_NAME, CONTENT_TYPE_QUALIFIER,<a name="line.635"></a>
-<span class="sourceLineNo">636</span> now), new BytesWritable(Bytes.toBytes(contentType)));<a name="line.636"></a>
-<span class="sourceLineNo">637</span> output.write(new HBaseKeyWritable(rowKey, INFO_FAMILY_NAME, RECORD_ID_QUALIFIER,<a name="line.637"></a>
-<span class="sourceLineNo">638</span> now), new BytesWritable(Bytes.toBytes(recordID)));<a name="line.638"></a>
-<span class="sourceLineNo">639</span> output.write(new HBaseKeyWritable(rowKey, INFO_FAMILY_NAME, TARGET_URI_QUALIFIER,<a name="line.639"></a>
-<span class="sourceLineNo">640</span> now), new BytesWritable(Bytes.toBytes(targetURI)));<a name="line.640"></a>
-<span class="sourceLineNo">641</span> output.write(new HBaseKeyWritable(rowKey, INFO_FAMILY_NAME, DATE_QUALIFIER, now),<a name="line.641"></a>
-<span class="sourceLineNo">642</span> new BytesWritable(Bytes.toBytes(warcHeader.getDateString())));<a name="line.642"></a>
-<span class="sourceLineNo">643</span> if (ipAddr != null) {<a name="line.643"></a>
-<span class="sourceLineNo">644</span> output.write(new HBaseKeyWritable(rowKey, INFO_FAMILY_NAME, IP_ADDRESS_QUALIFIER,<a name="line.644"></a>
-<span class="sourceLineNo">645</span> now), new BytesWritable(Bytes.toBytes(ipAddr)));<a name="line.645"></a>
-<span class="sourceLineNo">646</span> }<a name="line.646"></a>
-<span class="sourceLineNo">647</span> }<a name="line.647"></a>
-<span class="sourceLineNo">648</span> }<a name="line.648"></a>
-<span class="sourceLineNo">649</span> }<a name="line.649"></a>
-<span class="sourceLineNo">650</span><a name="line.650"></a>
-<span class="sourceLineNo">651</span> private byte[] rowKeyFromTargetURI(String targetUri)<a name="line.651"></a>
-<span class="sourceLineNo">652</span> throws URISyntaxException, IllegalArgumentException {<a name="line.652"></a>
-<span class="sourceLineNo">653</span> URI uri = new URI(targetUri);<a name="line.653"></a>
-<span class="sourceLineNo">654</span> StringBuffer sb = new StringBuffer();<a name="line.654"></a>
-<span class="sourceLineNo">655</span> // Ignore the scheme<a name="line.655"></a>
-<span class="sourceLineNo">656</span> // Reverse the components of the hostname<a name="line.656"></a>
-<span class="sourceLineNo">657</span> if (uri.getHost() != null) {<a name="line.657"></a>
-<span class="sourceLineNo">658</span> String[] hostComponents = uri.getHost().split("\\.");<a name="line.658"></a>
-<span class="sourceLineNo">659</span> for (int i = hostComponents.length - 1; i >= 0; i--) {<a name="line.659"></a>
-<span class="sourceLineNo">660</span> sb.append(hostComponents[i]);<a name="line.660"></a>
-<span class="sourceLineNo">661</span> if (i != 0) {<a name="line.661"></a>
-<span class="sourceLineNo">662</span> sb.append('.');<a name="line.662"></a>
-<span class="sourceLineNo">663</span> }<a name="line.663"></a>
-<span class="sourceLineNo">664</span> }<a name="line.664"></a>
-<span class="sourceLineNo">665</span> } else {<a name="line.665"></a>
-<span class="sourceLineNo">666</span> throw new IllegalArgumentException("URI is missing host component");<a name="line.666"></a>
-<span class="sourceLineNo">667</span> }<a name="line.667"></a>
-<span class="sourceLineNo">668</span> // Port<a name="line.668"></a>
-<span class="sourceLineNo">669</span> if (uri.getPort() != -1) {<a name="line.669"></a>
-<span class="sourceLineNo">670</span> sb.append(':');<a name="line.670"></a>
-<span class="sourceLineNo">671</span> sb.append(uri.getPort());<a name="line.671"></a>
+<span class="sourceLineNo">631</span> // Write records out for later verification, one per HBase field except for the<a name="line.631"></a>
+<span class="sourceLineNo">632</span> // content record, which will be verified by CRC64.<a name="line.632"></a>
+<span class="sourceLineNo">633</span><a name="line.633"></a>
+<span class="sourceLineNo">634</span> output.write(new HBaseKeyWritable(rowKey, INFO_FAMILY_NAME, CRC_QUALIFIER, now),<a name="line.634"></a>
+<span class="sourceLineNo">635</span> new BytesWritable(Bytes.toBytes(crc64)));<a name="line.635"></a>
+<span class="sourceLineNo">636</span> output.write(new HBaseKeyWritable(rowKey, INFO_FAMILY_NAME, CONTENT_LENGTH_QUALIFIER,<a name="line.636"></a>
+<span class="sourceLineNo">637</span> now), new BytesWritable(Bytes.toBytes(content.length)));<a name="line.637"></a>
+<span class="sourceLineNo">638</span> output.write(new HBaseKeyWritable(rowKey, INFO_FAMILY_NAME, CONTENT_TYPE_QUALIFIER,<a name="line.638"></a>
+<span class="sourceLineNo">639</span> now), new BytesWritable(Bytes.toBytes(contentType)));<a name="line.639"></a>
+<span class="sourceLineNo">640</span> output.write(new HBaseKeyWritable(rowKey, INFO_FAMILY_NAME, RECORD_ID_QUALIFIER,<a name="line.640"></a>
+<span class="sourceLineNo">641</span> now), new BytesWritable(Bytes.toBytes(recordID)));<a name="line.641"></a>
+<span class="sourceLineNo">642</span> output.write(new HBaseKeyWritable(rowKey, INFO_FAMILY_NAME, TARGET_URI_QUALIFIER,<a name="line.642"></a>
+<span class="sourceLineNo">643</span> now), new BytesWritable(Bytes.toBytes(targetURI)));<a name="line.643"></a>
+<span class="sourceLineNo">644</span> output.write(new HBaseKeyWritable(rowKey, INFO_FAMILY_NAME, DATE_QUALIFIER, now),<a name="line.644"></a>
+<span class="sourceLineNo">645</span> new BytesWritable(Bytes.toBytes(warcHeader.getDateString())));<a name="line.645"></a>
+<span class="sourceLineNo">646</span> if (ipAddr != null) {<a name="line.646"></a>
+<span class="sourceLineNo">647</span> output.write(new HBaseKeyWritable(rowKey, INFO_FAMILY_NAME, IP_ADDRESS_QUALIFIER,<a name="line.647"></a>
+<span class="sourceLineNo">648</span> now), new BytesWritable(Bytes.toBytes(ipAddr)));<a name="line.648"></a>
+<span class="sourceLineNo">649</span> }<a name="line.649"></a>
+<span class="sourceLineNo">650</span> }<a name="line.650"></a>
+<span class="sourceLineNo">651</span> }<a name="line.651"></a>
+<span class="sourceLineNo">652</span> }<a name="line.652"></a>
+<span class="sourceLineNo">653</span><a name="line.653"></a>
+<span class="sourceLineNo">654</span> private byte[] rowKeyFromTargetURI(String targetUri)<a name="line.654"></a>
+<span class="sourceLineNo">655</span> throws URISyntaxException, IllegalArgumentException {<a name="line.655"></a>
+<span class="sourceLineNo">656</span> URI uri = new URI(targetUri);<a name="line.656"></a>
+<span class="sourceLineNo">657</span> // Ignore the scheme<a name="line.657"></a>
+<span class="sourceLineNo">658</span> // Reverse the components of the hostname<a name="line.658"></a>
+<span class="sourceLineNo">659</span> String reversedHost;<a name="line.659"></a>
+<span class="sourceLineNo">660</span> if (uri.getHost() != null) {<a name="line.660"></a>
+<span class="sourceLineNo">661</span> StringBuffer sb = new StringBuffer();<a name="line.661"></a>
+<span class="sourceLineNo">662</span> String[] hostComponents = uri.getHost().split("\\.");<a name="line.662"></a>
+<span class="sourceLineNo">663</span> for (int i = hostComponents.length - 1; i >= 0; i--) {<a name="line.663"></a>
+<span class="sourceLineNo">664</span> sb.append(hostComponents[i]);<a name="line.664"></a>
+<span class="sourceLineNo">665</span> if (i != 0) {<a name="line.665"></a>
+<span class="sourceLineNo">666</span> sb.append('.');<a name="line.666"></a>
+<span class="sourceLineNo">667</span> }<a name="line.667"></a>
+<span class="sourceLineNo">668</span> }<a name="line.668"></a>
+<span class="sourceLineNo">669</span> reversedHost = sb.toString();<a name="line.669"></a>
+<span class="sourceLineNo">670</span> } else {<a name="line.670"></a>
+<span class="sourceLineNo">671</span> throw new IllegalArgumentException("URI is missing host component");<a name="line.671"></a>
<span class="sourceLineNo">672</span> }<a name="line.672"></a>
-<span class="sourceLineNo">673</span> if (uri.getRawPath() != null) {<a name="line.673"></a>
-<span class="sourceLineNo">674</span> sb.append(uri.getRawPath());<a name="line.674"></a>
-<span class="sourceLineNo">675</span> }<a name="line.675"></a>
-<span class="sourceLineNo">676</span> if (uri.getRawQuery() != null) {<a name="line.676"></a>
-<span class="sourceLineNo">677</span> sb.append('?');<a name="line.677"></a>
-<span class="sourceLineNo">678</span> sb.append(uri.getRawQuery());<a name="line.678"></a>
-<span class="sourceLineNo">679</span> }<a name="line.679"></a>
-<span class="sourceLineNo">680</span> if (uri.getRawFragment() != null) {<a name="line.680"></a>
-<span class="sourceLineNo">681</span> sb.append('#');<a name="line.681"></a>
-<span class="sourceLineNo">682</span> sb.append(uri.getRawFragment());<a name="line.682"></a>
-<span class="sourceLineNo">683</span> }<a name="line.683"></a>
-<span class="sourceLineNo">684</span> // Constrain the key size to the maximum allowed row key length<a name="line.684"></a>
-<span class="sourceLineNo">685</span> if (sb.length() > HConstants.MAX_ROW_LENGTH) {<a name="line.685"></a>
-<span class="sourceLineNo">686</span> sb.setLength(HConstants.MAX_ROW_LENGTH);<a name="line.686"></a>
-<span class="sourceLineNo">687</span> }<a name="line.687"></a>
-<span class="sourceLineNo">688</span> return Bytes.toBytes(sb.toString());<a name="line.688"></a>
-<span class="sourceLineNo">689</span> }<a name="line.689"></a>
-<span class="sourceLineNo">690</span><a name="line.690"></a>
-<span class="sourceLineNo">691</span> }<a name="line.691"></a>
-<span class="sourceLineNo">692</span> }<a name="line.692"></a>
-<span class="sourceLineNo">693</span><a name="line.693"></a>
-<span class="sourceLineNo">694</span> public static class OneFilePerMapperSFIF<K, V> extends SequenceFileInputFormat<K, V> {<a name="line.694"></a>
-<span class="sourceLineNo">695</span> @Override<a name="line.695"></a>
-<span class="sourceLineNo">696</span> protected boolean isSplitable(JobContext context, Path filename) {<a name="line.696"></a>
-<span class="sourceLineNo">697</span> return false;<a name="line.697"></a>
+<span class="sourceLineNo">673</span> StringBuffer sb = new StringBuffer();<a name="line.673"></a>
+<span class="sourceLineNo">674</span> sb.append(reversedHost);<a name="line.674"></a>
+<span class="sourceLineNo">675</span> if (uri.getPort() >= 0) {<a name="line.675"></a>
+<span class="sourceLineNo">676</span> sb.append(':');<a name="line.676"></a>
+<span class="sourceLineNo">677</span> sb.append(uri.getPort());<a name="line.677"></a>
+<span class="sourceLineNo">678</span> }<a name="line.678"></a>
+<span class="sourceLineNo">679</span> if (uri.getPath() != null) {<a name="line.679"></a>
+<span class="sourceLineNo">680</span> sb.append('/');<a name="line.680"></a>
+<span class="sourceLineNo">681</span> sb.append(uri.getPath());<a name="line.681"></a>
+<span class="sourceLineNo">682</span> }<a name="line.682"></a>
+<span class="sourceLineNo">683</span> if (uri.getQuery() != null) {<a name="line.683"></a>
+<span class="sourceLineNo">684</span> sb.append('?');<a name="line.684"></a>
+<span class="sourceLineNo">685</span> sb.append(uri.getQuery());<a name="line.685"></a>
+<span class="sourceLineNo">686</span> }<a name="line.686"></a>
+<span class="sourceLineNo">687</span> if (uri.getFragment() != null) {<a name="line.687"></a>
+<span class="sourceLineNo">688</span> sb.append('#');<a name="line.688"></a>
+<span class="sourceLineNo">689</span> sb.append(uri.getFragment());<a name="line.689"></a>
+<span class="sourceLineNo">690</span> }<a name="line.690"></a>
+<span class="sourceLineNo">691</span> if (sb.length() > HConstants.MAX_ROW_LENGTH) {<a name="line.691"></a>
+<span class="sourceLineNo">692</span> throw new IllegalArgumentException("Key would be too large (length=" + sb.length() +<a name="line.692"></a>
+<span class="sourceLineNo">693</span> ", limit=" + HConstants.MAX_ROW_LENGTH);<a name="line.693"></a>
+<span class="sourceLineNo">694</span> }<a name="line.694"></a>
+<span class="sourceLineNo">695</span> return Bytes.toBytes(sb.toString());<a name="line.695"></a>
+<span class="sourceLineNo">696</span> }<a name="line.696"></a>
+<span class="sourceLineNo">697</span><a name="line.697"></a>
<span class="sourceLineNo">698</span> }<a name="line.698"></a>
<span class="sourceLineNo">699</span> }<a name="line.699"></a>
<span class="sourceLineNo">700</span><a name="line.700"></a>
-<span class="sourceLineNo">701</span> public static class Verify extends Configured implements Tool {<a name="line.701"></a>
-<span class="sourceLineNo">702</span><a name="line.702"></a>
-<span class="sourceLineNo">703</span> public static final Logger LOG = LoggerFactory.getLogger(Verify.class);<a name="line.703"></a>
-<span class="sourceLineNo">704</span> public static final String USAGE = "Verify <inputDir>";<a name="line.704"></a>
-<span class="sourceLineNo">705</span><a name="line.705"></a>
-<span class="sourceLineNo">706</span> int run(Path inputDir) throws IOException, ClassNotFoundException, InterruptedException {<a name="line.706"></a>
-<span class="sourceLineNo">707</span> Job job = Job.getInstance(getConf());<a name="line.707"></a>
-<span class="sourceLineNo">708</span> job.setJobName(Verify.class.getName());<a name="line.708"></a>
-<span class="sourceLineNo">709</span> job.setJarByClass(getClass());<a name="line.709"></a>
-<span class="sourceLineNo">710</span> job.setMapperClass(VerifyMapper.class);<a name="line.710"></a>
-<span class="sourceLineNo">711</span> job.setInputFormatClass(OneFilePerMapperSFIF.class);<a name="line.711"></a>
-<span class="sourceLineNo">712</span> FileInputFormat.setInputPaths(job, inputDir);<a name="line.712"></a>
-<span class="sourceLineNo">713</span> job.setOutputFormatClass(NullOutputFormat.class);<a name="line.713"></a>
-<span class="sourceLineNo">714</span> job.setOutputKeyClass(NullWritable.class);<a name="line.714"></a>
-<span class="sourceLineNo">715</span> job.setOutputValueClass(NullWritable.class);<a name="line.715"></a>
-<span class="sourceLineNo">716</span> TableMapReduceUtil.addDependencyJars(job);<a name="line.716"></a>
-<span class="sourceLineNo">717</span> boolean success = job.waitForCompletion(true);<a name="line.717"></a>
-<span class="sourceLineNo">718</span> if (!success) {<a name="line.718"></a>
-<span class="sourceLineNo">719</span> LOG.error("Failure during job " + job.getJobID());<a name="line.719"></a>
-<span class="sourceLineNo">720</span> }<a name="line.720"></a>
-<span class="sourceLineNo">721</span> Counters counters = job.getCounters();<a name="line.721"></a>
-<span class="sourceLineNo">722</span> for (Counts c: Counts.values()) {<a name="line.722"></a>
-<span class="sourceLineNo">723</span> LOG.info(c + ": " + counters.findCounter(c).getValue());<a name="line.723"></a>
-<span class="sourceLineNo">724</span> }<a name="line.724"></a>
-<span class="sourceLineNo">725</span> return success ? 0 : 1;<a name="line.725"></a>
-<span class="sourceLineNo">726</span> }<a name="line.726"></a>
-<span class="sourceLineNo">727</span><a name="line.727"></a>
-<span class="sourceLineNo">728</span> @Override<a name="line.728"></a>
-<span class="sourceLineNo">729</span> public int run(String[] args) throws Exception {<a name="line.729"></a>
-<span class="sourceLineNo">730</span> if (args.length < 2) {<a name="line.730"></a>
-<span class="sourceLineNo">731</span> System.err.println(USAGE);<a name="line.731"></a>
-<span class="sourceLineNo">732</span> return 1;<a name="line.732"></a>
-<span class="sourceLineNo">733</span> }<a name="line.733"></a>
-<span class="sourceLineNo">734</span> Path loaderOutput = new Path(args[0]);<a name="line.734"></a>
-<span class="sourceLineNo">735</span> return run(loaderOutput);<a name="line.735"></a>
-<span class="sourceLineNo">736</span> }<a name="line.736"></a>
-<span class="sourceLineNo">737</span><a name="line.737"></a>
-<span class="sourceLineNo">738</span> public static void main(String[] args) throws Exception {<a name="line.738"></a>
-<span class="sourceLineNo">739</span> System.exit(ToolRunner.run(HBaseConfiguration.create(), new Verify(), args));<a name="line.739"></a>
-<span class="sourceLineNo">740</span> }<a name="line.740"></a>
-<span class="sourceLineNo">741</span><a name="line.741"></a>
-<span class="sourceLineNo">742</span> public static class VerifyMapper<a name="line.742"></a>
-<span class="sourceLineNo">743</span> extends Mapper<HBaseKeyWritable, BytesWritable, NullWritable, NullWritable> {<a name="line.743"></a>
+<span class="sourceLineNo">701</span> public static class OneFilePerMapperSFIF<K, V> extends SequenceFileInputFormat<K, V> {<a name="line.701"></a>
+<span class="sourceLineNo">702</span> @Override<a name="line.702"></a>
+<span class="sourceLineNo">703</span> protected boolean isSplitable(JobContext context, Path filename) {<a name="line.703"></a>
+<span class="sourceLineNo">704</span> return false;<a name="line.704"></a>
+<span class="sourceLineNo">705</span> }<a name="line.705"></a>
+<span class="sourceLineNo">706</span> }<a name="line.706"></a>
+<span class="sourceLineNo">707</span><a name="line.707"></a>
+<span class="sourceLineNo">708</span> public static class Verify extends Configured implements Tool {<a name="line.708"></a>
+<span class="sourceLineNo">709</span><a name="line.709"></a>
+<span class="sourceLineNo">710</span> public static final Logger LOG = LoggerFactory.getLogger(Verify.class);<a name="line.710"></a>
+<span class="sourceLineNo">711</span> public static final String USAGE = "Verify <inputDir>";<a name="line.711"></a>
+<span class="sourceLineNo">712</span><a name="line.712"></a>
+<span class="sourceLineNo">713</span> int run(Path inputDir) throws IOException, ClassNotFoundException, InterruptedException {<a name="line.713"></a>
+<span class="sourceLineNo">714</span> Job job = Job.getInstance(getConf());<a name="line.714"></a>
+<span class="sourceLineNo">715</span> job.setJobName(Verify.class.getName());<a name="line.715"></a>
+<span class="sourceLineNo">716</span> job.setJarByClass(getClass());<a name="line.716"></a>
+<span class="sourceLineNo">717</span> job.setMapperClass(VerifyMapper.class);<a name="line.717"></a>
+<span class="sourceLineNo">718</span> job.setInputFormatClass(OneFilePerMapperSFIF.class);<a name="line.718"></a>
+<span class="sourceLineNo">719</span> FileInputFormat.setInputPaths(job, inputDir);<a name="line.719"></a>
+<span class="sourceLineNo">720</span> job.setOutputFormatClass(NullOutputFormat.class);<a name="line.720"></a>
+<span class="sourceLineNo">721</span> job.setOutputKeyClass(NullWritable.class);<a name="line.721"></a>
+<span class="sourceLineNo">722</span> job.setOutputValueClass(NullWritable.class);<a name="line.722"></a>
+<span class="sourceLineNo">723</span> TableMapReduceUtil.addDependencyJars(job);<a name="line.723"></a>
+<span class="sourceLineNo">724</span> boolean success = job.waitForCompletion(true);<a name="line.724"></a>
+<span class="sourceLineNo">725</span> if (!success) {<a name="line.725"></a>
+<span class="sourceLineNo">726</span> LOG.error("Failure during job " + job.getJobID());<a name="line.726"></a>
+<span class="sourceLineNo">727</span> }<a name="line.727"></a>
+<span class="sourceLineNo">728</span> Counters counters = job.getCounters();<a name="line.728"></a>
+<span class="sourceLineNo">729</span> for (Counts c: Counts.values()) {<a name="line.729"></a>
+<span class="sourceLineNo">730</span> LOG.info(c + ": " + counters.findCounter(c).getValue());<a name="line.730"></a>
+<span class="sourceLineNo">731</span> }<a name="line.731"></a>
+<span class="sourceLineNo">732</span> return success ? 0 : 1;<a name="line.732"></a>
+<span class="sourceLineNo">733</span> }<a name="line.733"></a>
+<span class="sourceLineNo">734</span><a name="line.734"></a>
+<span class="sourceLineNo">735</span> @Override<a name="line.735"></a>
+<span class="sourceLineNo">736</span> public int run(String[] args) throws Exception {<a name="line.736"></a>
+<span class="sourceLineNo">737</span> if (args.length < 2) {<a name="line.737"></a>
+<span class="sourceLineNo">738</span> System.err.println(USAGE);<a name="line.738"></a>
+<span class="sourceLineNo">739</span> return 1;<a name="line.739"></a>
+<span class="sourceLineNo">740</span> }<a name="line.740"></a>
+<span class="sourceLineNo">741</span> Path loaderOutput = new Path(args[0]);<a name="line.741"></a>
+<span class="sourceLineNo">742</span> return run(loaderOutput);<a name="line.742"></a>
+<span class="sourceLineNo">743</span> }<a name="line.743"></a>
<span class="sourceLineNo">744</span><a name="line.744"></a>
-<span class="sourceLineNo">745</span> Connection conn;<a name="line.745"></a>
-<span class="sourceLineNo">746</span> Table table;<a name="line.746"></a>
-<span class="sourceLineNo">747</span><a name="line.747"></a>
-<span class="sourceLineNo">748</span> @Override<a name="line.748"></a>
-<span class="sourceLineNo">749</span> protected void setup(Context context) throws IOException, InterruptedException {<a name="line.749"></a>
-<span class="sourceLineNo">750</span> conn = ConnectionFactory.createConnection(context.getConfiguration());<a name="line.750"></a>
-<span class="sourceLineNo">751</span> table = conn.getTable(getTablename(conn.getConfiguration()));<a name="line.751"></a>
-<span class="sourceLineNo">752</span> }<a name="line.752"></a>
-<span class="sourceLineNo">753</span><a name="line.753"></a>
-<span class="sourceLineNo">754</span> @Override<a name="line.754"></a>
-<span class="sourceLineNo">755</span> protected void cleanup(Context context) throws IOException ,InterruptedException {<a name="line.755"></a>
-<span class="sourceLineNo">756</span> table.close();<a name="line.756"></a>
-<span class="sourceLineNo">757</span> conn.close();<a name="line.757"></a>
-<span class="sourceLineNo">758</span> }<a name="line.758"></a>
-<span class="sourceLineNo">759</span><a name="line.759"></a>
-<span class="sourceLineNo">760</span> @Override<a name="line.760"></a>
-<span class="sourceLineNo">761</span> protected void map(HBaseKeyWritable key, BytesWritable value, Context output)<a name="line.761"></a>
-<span class="sourceLineNo">762</span> throws IOException, InterruptedException {<a name="line.762"></a>
-<span class="sourceLineNo">763</span><a name="line.763"></a>
-<span class="sourceLineNo">764</span> byte[] row = Bytes.copy(key.getRowArray(), key.getRowOffset(), key.getRowLength());<a name="line.764"></a>
-<span class="sourceLineNo">765</span> byte[] family = Bytes.copy(key.getFamilyArray(), key.getFamilyOffset(),<a name="line.765"></a>
-<span class="sourceLineNo">766</span> key.getFamilyLength());<a name="line.766"></a>
-<span class="sourceLineNo">767</span> byte[] qualifier = Bytes.copy(key.getQualifierArray(), key.getQualifierOffset(),<a name="line.767"></a>
-<span class="sourceLineNo">768</span> key.getQualifierLength());<a name="line.768"></a>
-<span class="sourceLineNo">769</span> long ts = key.getTimestamp();<a name="line.769"></a>
+<span class="sourceLineNo">745</span> public static void main(String[] args) throws Exception {<a name="line.745"></a>
+<span class="sourceLineNo">746</span> System.exit(ToolRunner.run(HBaseConfiguration.create(), new Verify(), args));<a name="line.746"></a>
+<span class="sourceLineNo">747</span> }<a name="line.747"></a>
+<span class="sourceLineNo">748</span><a name="line.748"></a>
+<span class="sourceLineNo">749</span> public static class VerifyMapper<a name="line.749"></a>
+<span class="sourceLineNo">750</span> extends Mapper<HBaseKeyWritable, BytesWritable, NullWritable, NullWritable> {<a name="line.750"></a>
+<span class="sourceLineNo">751</span><a name="line.751"></a>
+<span class="sourceLineNo">752</span> Connection conn;<a name="line.752"></a>
+<span class="sourceLineNo">753</span> Table table;<a name="line.753"></a>
+<span class="sourceLineNo">754</span><a name="line.754"></a>
+<span class="sourceLineNo">755</span> @Override<a name="line.755"></a>
+<span class="sourceLineNo">756</span> protected void setup(Context context) throws IOException, InterruptedException {<a name="line.756"></a>
+<span class="sourceLineNo">757</span> conn = ConnectionFactory.createConnection(context.getConfiguration());<a name="line.757"></a>
+<span class="sourceLineNo">758</span> table = conn.getTable(getTablename(conn.getConfiguration()));<a name="line.758"></a>
+<span class="sourceLineNo">759</span> }<a name="line.759"></a>
+<span class="sourceLineNo">760</span><a name="line.760"></a>
+<span class="sourceLineNo">761</span> @Override<a name="line.761"></a>
+<span class="sourceLineNo">762</span> protected void cleanup(Context context) throws IOException ,InterruptedException {<a name="line.762"></a>
+<span class="sourceLineNo">763</span> table.close();<a name="line.763"></a>
+<span class="sourceLineNo">764</span> conn.close();<a name="line.764"></a>
+<span class="sourceLineNo">765</span> }<a name="line.765"></a>
+<span class="sourceLineNo">766</span><a name="line.766"></a>
+<span class="sourceLineNo">767</span> @Override<a name="line.767"></a>
+<span class="sourceLineNo">768</span> protected void map(HBaseKeyWritable key, BytesWritable value, Context output)<a name="line.768"></a>
+<span class="sourceLineNo">769</span> throws IOException, InterruptedException {<a name="line.769"></a>
<span class="sourceLineNo">770</span><a name="line.770"></a>
-<span class="sourceLineNo">771</span> if (Bytes.equals(INFO_FAMILY_NAME, family) &&<a name="line.771"></a>
-<span class="sourceLineNo">772</span> Bytes.equals(CRC_QUALIFIER, qualifier)) {<a name="line.772"></a>
-<span class="sourceLineNo">773</span><a name="line.773"></a>
-<span class="sourceLineNo">774</span> long expectedCRC64 = Bytes.toLong(value.getBytes(), 0, value.getLength());<a name="line.774"></a>
-<span class="sourceLineNo">775</span><a name="line.775"></a>
-<span class="sourceLineNo">776</span> Result result =<a name="line.776"></a>
-<span class="sourceLineNo">777</span> table.get(new Get(row)<a name="line.777"></a>
-<span class="sourceLineNo">778</span> .addColumn(CONTENT_FAMILY_NAME, CONTENT_QUALIFIER)<a name="line.778"></a>
-<span class="sourceLineNo">779</span> .addColumn(INFO_FAMILY_NAME, CRC_QUALIFIER)<a name="line.779"></a>
-<span class="sourceLineNo">780</span> .setTimestamp(ts));<a name="line.780"></a>
-<span class="sourceLineNo">781</span><a name="line.781"></a>
-<span class="sourceLineNo">782</span> byte[] content = result.getValue(CONTENT_FAMILY_NAME, CONTENT_QUALIFIER);<a name="line.782"></a>
-<span class="sourceLineNo">783</span> if (content == null) {<a name="line.783"></a>
-<span class="sourceLineNo">784</span> LOG.info("Row " + Bytes.toStringBinary(row) + ": missing content");<a name="line.784"></a>
-<span class="sourceLineNo">785</span> output.getCounter(Counts.UNREFERENCED).increment(1);<a name="line.785"></a>
-<span class="sourceLineNo">786</span> return;<a name="line.786"></a>
-<span class="sourceLineNo">787</span> } else {<a name="line.787"></a>
-<span class="sourceLineNo">788</span> CRC64 crc = new CRC64();<a name="line.788"></a>
-<span class="sourceLineNo">789</span> crc.update(content);<a name="line.789"></a>
-<span class="sourceLineNo">790</span> if (crc.getValue() != expectedCRC64) {<a name="line.790"></a>
-<span class="sourceLineNo">791</span> LOG.info("Row " + Bytes.toStringBinary(row) + ": corrupt content");<a name="line.791"></a>
-<span class="sourceLineNo">792</span> output.getCounter(Counts.CORRUPT).increment(1);<a name="line.792"></a>
-<span class="sourceLineNo">793</span> return;<a name="line.793"></a>
-<span class="sourceLineNo">794</span> }<a name="line.794"></a>
-<span class="sourceLineNo">795</span> }<a name="line.795"></a>
-<span class="sourceLineNo">796</span> byte[] crc = result.getValue(INFO_FAMILY_NAME, CRC_QUALIFIER);<a name="line.796"></a>
-<span class="sourceLineNo">797</span> if (crc == null) {<a name="line.797"></a>
-<span class="sourceLineNo">798</span> LOG.info("Row " + Bytes.toStringBinary(row) + ": missing i:c");<a name="line.798"></a>
-<span class="sourceLineNo">799</span> output.getCounter(Counts.UNREFERENCED).increment(1);<a name="line.799"></a>
-<span class="sourceLineNo">800</span> return;<a name="line.800"></a>
-<span class="sourceLineNo">801</span> }<a name="line.801"></a>
-<span class="sourceLineNo">802</span> if (Bytes.toLong(crc) != expectedCRC64) {<a name="line.802"></a>
-<span class="sourceLineNo">803</span> LOG.info("Row " + Bytes.toStringBinary(row) + ": i:c mismatch");<a name="line.803"></a>
-<span class="sourceLineNo">804</span> output.getCounter(Counts.CORRUPT).increment(1);<a name="line.804"></a>
-<span class="sourceLineNo">805</span> return;<a name="line.805"></a>
-<span class="sourceLineNo">806</span> }<a name="line.806"></a>
-<span class="sourceLineNo">807</span><a name="line.807"></a>
-<span class="sourceLineNo">808</span> } else {<a name="line.808"></a>
-<span class="sourceLineNo">809</span><a name="line.809"></a>
-<span class="sourceLineNo">810</span> Result result =<a name="line.810"></a>
-<span class="sourceLineNo">811</span> table.get(new Get(row)<a name="line.811"></a>
-<span class="sourceLineNo">812</span> .addColumn(family, qualifier)<a name="line.812"></a>
-<span class="sourceLineNo">813</span> .setTimestamp(ts));<a name="line.813"></a>
-<span class="sourceLineNo">814</span><a name="line.814"></a>
-<span class="sourceLineNo">815</span> byte[] bytes = result.getValue(family, qualifier);<a name="line.815"></a>
-<span class="sourceLineNo">816</span> if (bytes == null) {<a name="line.816"></a>
-<span class="sourceLineNo">817</span> LOG.info("Row " + Bytes.toStringBinary(row) + ": missing " +<a name="line.817"></a>
-<span class="sourceLineNo">818</span> Bytes.toStringBinary(family) + ":" + Bytes.toStringBinary(qualifier));<a name="line.818"></a>
-<span class="sourceLineNo">819</span> output.getCounter(Counts.UNREFERENCED).increment(1);<a name="line.819"></a>
-<span class="sourceLineNo">820</span> return;<a name="line.820"></a>
-<span class="sourceLineNo">821</span> }<a name="line.821"></a>
-<span class="sourceLineNo">822</span> if (!Bytes.equals(bytes, 0, bytes.length, value.getBytes(), 0, value.getLength())) {<a name="line.822"></a>
-<span class="sourceLineNo">823</span> LOG.info("Row " + Bytes.toStringBinary(row) + ": " +<a name="line.823"></a>
-<span class="sourceLineNo">824</span> Bytes.toStringBinary(family) + ":" + Bytes.toStringBinary(qualifier) +<a name="line.824"></a>
-<span class="sourceLineNo">825</span> " mismatch");<a name="line.825"></a>
-<span class="sourceLineNo">826</span> output.getCounter(Counts.CORRUPT).increment(1);<a name="line.826"></a>
-<span class="sourceLineNo">827</span> return;<a name="line.827"></a>
-<span class="sourceLineNo">828</span> }<a name="line.828"></a>
-<span class="sourceLineNo">829</span><a name="line.829"></a>
-<span class="sourceLineNo">830</span> }<a name="line.830"></a>
-<span class="sourceLineNo">831</span><a name="line.831"></a>
-<span class="sourceLineNo">832</span> output.getCounter(Counts.REFERENCED).increment(1);<a name="line.832"></a>
-<span class="sourceLineNo">833</span> }<a name="line.833"></a>
-<span class="sourceLineNo">834</span><a name="line.834"></a>
-<span class="sourceLineNo">835</span> }<a name="line.835"></a>
-<span class="sourceLineNo">836</span><a name="line.836"></a>
-<span class="sourceLineNo">837</span> }<a name="line.837"></a>
-<span class="sourceLineNo">838</span><a name="line.838"></a>
-<span class="sourceLineNo">839</span>}<a name="line.839"></a>
+<span class="sourceLineNo">771</span> byte[] row = Bytes.copy(key.getRowArray(), key.getRowOffset(), key.getRowLength());<a name="line.771"></a>
+<span class="sourceLineNo">772</span> byte[] family = Bytes.copy(key.getFamilyArray(), key.getFamilyOffset(),<a name="line.772"></a>
+<span class="sourceLineNo">773</span> key.getFamilyLength());<a name="line.773"></a>
+<span class="sourceLineNo">774</span> byte[] qualifier = Bytes.copy(key.getQualifierArray(), key.getQualifierOffset(),<a name="line.774"></a>
+<span class="sourceLineNo">775</span> key.getQualifierLength());<a name="line.775"></a>
+<span class="sourceLineNo">776</span> long ts = key.getTimestamp();<a name="line.776"></a>
+<span class="sourceLineNo">777</span> int retries = VERIFICATION_READ_RETRIES;<a name="line.777"></a>
+<span class="sourceLineNo">778</span><a name="line.778"></a>
+<span class="sourceLineNo">779</span> while (true) {<a name="line.779"></a>
+<span class="sourceLineNo">780</span><a name="line.780"></a>
+<span class="sourceLineNo">781</span> if (Bytes.equals(INFO_FAMILY_NAME, family) &&<a name="line.781"></a>
+<span class="sourceLineNo">782</span> Bytes.equals(CRC_QUALIFIER, qualifier)) {<a name="line.782"></a>
+<span class="sourceLineNo">783</span><a name="line.783"></a>
+<span class="sourceLineNo">784</span> long expectedCRC64 = Bytes.toLong(value.getBytes(), 0, value.getLength());<a name="line.784"></a>
+<span class="sourceLineNo">785</span> Result result = table.get(new Get(row)<a name="line.785"></a>
+<span class="sourceLineNo">786</span> .addColumn(CONTENT_FAMILY_NAME, CONTENT_QUALIFIER)<a name="line.786"></a>
+<span class="sourceLineNo">787</span> .addColumn(INFO_FAMILY_NAME, CRC_QUALIFIER)<a name="line.787"></a>
+<span class="sourceLineNo">788</span> .setTimestamp(ts));<a name="line.788"></a>
+<span class="sourceLineNo">789</span> byte[] content = result.getValue(CONTENT_FAMILY_NAME, CONTENT_QUALIFIER);<a name="line.789"></a>
+<span class="sourceLineNo">790</span> if (content == null) {<a name="line.790"></a>
+<span class="sourceLineNo">791</span> if (retries-- > 0) {<a name="line.791"></a>
+<span class="sourceLineNo">792</span> continue;<a name="line.792"></a>
+<span class="sourceLineNo">793</span> }<a name="line.793"></a>
+<span class="sourceLineNo">794</span> LOG.error("Row " + Bytes.toStringBinary(row) + ": missing content");<a name="line.794"></a>
+<span class="sourceLineNo">795</span> output.getCounter(Counts.UNREFERENCED).increment(1);<a name="line.795"></a>
+<span class="sourceLineNo">796</span> return;<a name="line.796"></a>
+<span class="sourceLineNo">797</span> } else {<a name="line.797"></a>
+<span class="sourceLineNo">798</span> CRC64 crc = new CRC64();<a name="line.798"></a>
+<span class="sourceLineNo">799</span> crc.update(content);<a name="line.799"></a>
+<span class="sourceLineNo">800</span> if (crc.getValue() != expectedCRC64) {<a name="line.800"></a>
+<span class="sourceLineNo">801</span> if (retries-- > 0) {<a name="line.801"></a>
+<span class="sourceLineNo">802</span> continue;<a name="line.802"></a>
+<span class="sourceLineNo">803</span> }<a name="line.803"></a>
+<span class="sourceLineNo">804</span> LOG.error("Row " + Bytes.toStringBinary(row) + ": corrupt content");<a name="line.804"></a>
+<span class="sourceLineNo">805</span> output.getCounter(Counts.CORRUPT).increment(1);<a name="line.805"></a>
+<span class="sourceLineNo">806</span> return;<a name="line.806"></a>
+<span class="sourceLineNo">807</span> }<a name="line.807"></a>
+<span class="sourceLineNo">808</span> }<a name="line.808"></a>
+<span class="sourceLineNo">809</span> byte[] crc = result.getValue(INFO_FAMILY_NAME, CRC_QUALIFIER);<a name="line.809"></a>
+<span class="sourceLineNo">810</span> if (crc == null) {<a name="line.810"></a>
+<span class="sourceLineNo">811</span> if (retries-- > 0) {<a name="line.811"></a>
+<span class="sourceLineNo">812</span> continue;<a name="line.812"></a>
+<span class="sourceLineNo">813</span> }<a name="line.813"></a>
+<span class="sourceLineNo">814</span> LOG.error("Row " + Bytes.toStringBinary(row) + ": missing i:c");<a name="line.814"></a>
+<span class="sourceLineNo">815</span> output.getCounter(Counts.UNREFERENCED).increment(1);<a name="line.815"></a>
+<span class="sourceLineNo">816</span> return;<a name="line.816"></a>
+<span class="sourceLineNo">817</span> }<a name="line.817"></a>
+<span class="sourceLineNo">818</span> if (Bytes.toLong(crc) != expectedCRC64) {<a name="line.818"></a>
+<span class="sourceLineNo">819</span> if (retries-- > 0) {<a name="line.819"></a>
+<span class="sourceLineNo">820</span> continue;<a name="line.820"></a>
+<span class="sourceLineNo">821</span> }<a name="line.821"></a>
+<span class="sourceLineNo">822</span> LOG.error("Row " + Bytes.toStringBinary(row) + ": i:c mismatch");<a name="line.822"></a>
+<span class="sourceLineNo">823</span> output.getCounter(Counts.CORRUPT).increment(1);<a name="line.823"></a>
+<span class="sourceLineNo">824</span> return;<a name="line.824"></a>
+<span class="sourceLineNo">825</span> }<a name="line.825"></a>
+<span class="sourceLineNo">826</span><a name="line.826"></a>
+<span class="sourceLineNo">827</span> } else {<a name="line.827"></a>
+<span class="sourceLineNo">828</span><a name="line.828"></a>
+<span class="sourceLineNo">829</span> Result result = table.get(new Get(row)<a name="line.829"></a>
+<span class="sourceLineNo">830</span> .addColumn(family, qualifier)<a name="line.830"></a>
+<span class="sourceLineNo">831</span> .setTimestamp(ts));<a name="line.831"></a>
+<span class="sourceLineNo">832</span> byte[] bytes = result.getValue(family, qualifier);<a name="line.832"></a>
+<span class="sourceLineNo">833</span> if (bytes == null) {<a name="line.833"></a>
+<span class="sourceLineNo">834</span> if (retries-- > 0) {<a name="line.834"></a>
+<span class="sourceLineNo">835</span> continue;<a name="line.835"></a>
+<span class="sourceLineNo">836</span> }<a name="line.836"></a>
+<span class="sourceLineNo">837</span> LOG.error("Row " + Bytes.toStringBinary(row) + ": missing " +<a name="line.837"></a>
+<span class="sourceLineNo">838</span> Bytes.toStringBinary(family) + ":" + Bytes.toStringBinary(qualifier));<a name="line.838"></a>
+<span class="sourceLineNo">839</span> output.getCounter(Counts.UNREFERENCED).increment(1);<a name="line.839"></a>
+<span class="sourceLineNo">840</span> return;<a name="line.840"></a>
+<span class="sourceLineNo">841</span> }<a name="line.841"></a>
+<span class="sourceLineNo">842</span> if (!Bytes.equals(bytes, 0, bytes.length, value.getBytes(), 0, value.getLength())) {<a name="line.842"></a>
+<span class="sourceLineNo">843</span> if (retries-- > 0) {<a name="line.843"></a>
+<span class="sourceLineNo">844</span> continue;<a name="line.844"></a>
+<span class="sourceLineNo">845</span> }<a name="line.845"></a>
+<span class="sourceLineNo">846</span> LOG.error("Row " + Bytes.toStringBinary(row) + ": " +<a name="line.846"></a>
+<span class="sourceLineNo">847</span> Bytes.toStringBinary(family) + ":" + Bytes.toStringBinary(qualifier) +<a name="line.847"></a>
+<span class="sourceLineNo">848</span> " mismatch");<a name="line.848"></a>
+<span class="sourceLineNo">849</span> output.getCounter(Counts.CORRUPT).increment(1);<a name="line.849"></a>
+<span class="sourceLineNo">850</span> return;<a name="line.850"></a>
+<span class="sourceLineNo">851</span> }<a name="line.851"></a>
+<span class="sourceLineNo">852</span><a name="line.852"></a>
+<span class="sourceLineNo">853</span> }<a name="line.853"></a>
+<span class="sourceLineNo">854</span><a name="line.854"></a>
+<span class="sourceLineNo">855</span> // If we fell through to here all verification checks have succeeded, potentially after<a name="line.855"></a>
+<span class="sourceLineNo">856</span> // retries, and we must exit the while loop.<a name="line.856"></a>
+<span class="sourceLineNo">857</span> output.getCounter(Counts.REFERENCED).increment(1);<a name="line.857"></a>
+<span class="sourceLineNo">858</span> break;<a name="line.858"></a>
+<span class="sourceLineNo">859</span><a name="line.859"></a>
+<span class="sourceLineNo">860</span> }<a name="line.860"></a>
+<span class="sourceLineNo">861</span> }<a name="line.861"></a>
+<span class="sourceLineNo">862</span> }<a name="line.862"></a>
+<span class="sourceLineNo">863</span> }<a name="line.863"></a>
+<span class="sourceLineNo">864</span><a name="line.864"></a>
+<span class="sourceLineNo">865</span>}<a name="line.865"></a>
diff --git a/testdevapidocs/src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.Verify.VerifyMapper.html b/testdevapidocs/src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.Verify.VerifyMapper.html
index 4a40cbc..e89b63c 100644
--- a/testdevapidocs/src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.Verify.VerifyMapper.html
+++ b/testdevapidocs/src-html/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.Verify.VerifyMapper.html
@@ -54,797 +54,823 @@
<span class="sourceLineNo">046</span>import org.apache.hadoop.hbase.MasterNotRunningException;<a name="line.46"></a>
<span class="sourceLineNo">047</span>import org.apache.hadoop.hbase.TableName;<a name="line.47"></a>
<span class="sourceLineNo">048</span>import org.apache.hadoop.hbase.client.Admin;<a name="line.48"></a>
-<span class="sourceLineNo">049</span>import org.apache.hadoop.hbase.client.ColumnFamilyDescriptor;<a name="line.49"></a>
-<span class="sourceLineNo">050</span>import org.apache.hadoop.hbase.client.ColumnFamilyDescriptorBuilder;<a name="line.50"></a>
-<span class="sourceLineNo">051</span>import org.apache.hadoop.hbase.client.Connection;<a name="line.51"></a>
-<span class="sourceLineNo">052</span>import org.apache.hadoop.hbase.client.ConnectionFactory;<a name="line.52"></a>
-<span class="sourceLineNo">053</span>import org.apache.hadoop.hbase.client.Get;<a name="line.53"></a>
-<span class="sourceLineNo">054</span>import org.apache.hadoop.hbase.client.Put;<a name="line.54"></a>
-<span class="sourceLineNo">055</span>import org.apache.hadoop.hbase.client.Result;<a name="line.55"></a>
-<span class="sourceLineNo">056</span>import org.apache.hadoop.hbase.client.Table;<a name="line.56"></a>
-<span class="sourceLineNo">057</span>import org.apache.hadoop.hbase.client.TableDescriptor;<a name="line.57"></a>
-<span class="sourceLineNo">058</span>import org.apache.hadoop.hbase.client.TableDescriptorBuilder;<a name="line.58"></a>
-<span class="sourceLineNo">059</span>import org.apache.hadoop.hbase.io.encoding.DataBlockEncoding;<a name="line.59"></a>
-<span class="sourceLineNo">060</span>import org.apache.hadoop.hbase.mapreduce.TableMapReduceUtil;<a name="line.60"></a>
-<span class="sourceLineNo">061</span>import org.apache.hadoop.hbase.regionserver.BloomType;<a name="line.61"></a>
-<span class="sourceLineNo">062</span>import org.apache.hadoop.hbase.test.util.CRC64;<a name="line.62"></a>
-<span class="sourceLineNo">063</span>import org.apache.hadoop.hbase.test.util.warc.WARCInputFormat;<a name="line.63"></a>
-<span class="sourceLineNo">064</span>import org.apache.hadoop.hbase.test.util.warc.WARCRecord;<a name="line.64"></a>
-<span class="sourceLineNo">065</span>import org.apache.hadoop.hbase.test.util.warc.WARCWritable;<a name="line.65"></a>
-<span class="sourceLineNo">066</span>import org.apache.hadoop.hbase.util.Bytes;<a name="line.66"></a>
-<span class="sourceLineNo">067</span>import org.apache.hadoop.hbase.util.EnvironmentEdgeManager;<a name="line.67"></a>
-<span class="sourceLineNo">068</span>import org.apache.hadoop.hbase.util.RegionSplitter;<a name="line.68"></a>
-<span class="sourceLineNo">069</span>import org.apache.hadoop.io.BytesWritable;<a name="line.69"></a>
-<span class="sourceLineNo">070</span>import org.apache.hadoop.io.LongWritable;<a name="line.70"></a>
-<span class="sourceLineNo">071</span>import org.apache.hadoop.io.NullWritable;<a name="line.71"></a>
-<span class="sourceLineNo">072</span>import org.apache.hadoop.io.SequenceFile.CompressionType;<a name="line.72"></a>
-<span class="sourceLineNo">073</span>import org.apache.hadoop.io.Writable;<a name="line.73"></a>
-<span class="sourceLineNo">074</span>import org.apache.hadoop.mapreduce.Counters;<a name="line.74"></a>
-<span class="sourceLineNo">075</span>import org.apache.hadoop.mapreduce.Job;<a name="line.75"></a>
-<span class="sourceLineNo">076</span>import org.apache.hadoop.mapreduce.JobContext;<a name="line.76"></a>
-<span class="sourceLineNo">077</span>import org.apache.hadoop.mapreduce.Mapper;<a name="line.77"></a>
-<span class="sourceLineNo">078</span>import org.apache.hadoop.mapreduce.lib.input.FileInputFormat;<a name="line.78"></a>
-<span class="sourceLineNo">079</span>import org.apache.hadoop.mapreduce.lib.input.SequenceFileInputFormat;<a name="line.79"></a>
-<span class="sourceLineNo">080</span>import org.apache.hadoop.mapreduce.lib.output.NullOutputFormat;<a name="line.80"></a>
-<span class="sourceLineNo">081</span>import org.apache.hadoop.mapreduce.lib.output.SequenceFileOutputFormat;<a name="line.81"></a>
-<span class="sourceLineNo">082</span>import org.apache.hadoop.util.Tool;<a name="line.82"></a>
-<span class="sourceLineNo">083</span>import org.apache.hadoop.util.ToolRunner;<a name="line.83"></a>
-<span class="sourceLineNo">084</span>import org.slf4j.Logger;<a name="line.84"></a>
-<span class="sourceLineNo">085</span>import org.slf4j.LoggerFactory;<a name="line.85"></a>
-<span class="sourceLineNo">086</span><a name="line.86"></a>
-<span class="sourceLineNo">087</span>import org.apache.hbase.thirdparty.org.apache.commons.cli.CommandLine;<a name="line.87"></a>
-<span class="sourceLineNo">088</span><a name="line.88"></a>
-<span class="sourceLineNo">089</span>/**<a name="line.89"></a>
-<span class="sourceLineNo">090</span> * This integration test loads successful resource retrieval records from the Common Crawl<a name="line.90"></a>
-<span class="sourceLineNo">091</span> * (https://commoncrawl.org/) public dataset into an HBase table and writes records that can be<a name="line.91"></a>
-<span class="sourceLineNo">092</span> * used to later verify the presence and integrity of those records.<a name="line.92"></a>
-<span class="sourceLineNo">093</span> * <p><a name="line.93"></a>
-<span class="sourceLineNo">094</span> * Run like:<a name="line.94"></a>
-<span class="sourceLineNo">095</span> * <blockquote><a name="line.95"></a>
-<span class="sourceLineNo">096</span> * ./bin/hbase org.apache.hadoop.hbase.test.IntegrationTestLoadCommonCrawl \<br><a name="line.96"></a>
-<span class="sourceLineNo">097</span> * &nbsp;&nbsp; -Dfs.s3n.awsAccessKeyId=&lt;AWS access key&gt; \<br><a name="line.97"></a>
-<span class="sourceLineNo">098</span> * &nbsp;&nbsp; -Dfs.s3n.awsSecretAccessKey=&lt;AWS secret key&gt; \<br><a name="line.98"></a>
-<span class="sourceLineNo">099</span> * &nbsp;&nbsp; /path/to/test-CC-MAIN-2021-10-warc.paths.gz \<br><a name="line.99"></a>
-<span class="sourceLineNo">100</span> * &nbsp;&nbsp; /path/to/tmp/warc-loader-output<a name="line.100"></a>
-<span class="sourceLineNo">101</span> * </blockquote><a name="line.101"></a>
-<span class="sourceLineNo">102</span> * <p><a name="line.102"></a>
-<span class="sourceLineNo">103</span> * Access to the Common Crawl dataset in S3 is made available to anyone by Amazon AWS, but<a name="line.103"></a>
-<span class="sourceLineNo">104</span> * Hadoop's S3N filesystem still requires valid access credentials to initialize.<a name="line.104"></a>
-<span class="sourceLineNo">105</span> * <p><a name="line.105"></a>
-<span class="sourceLineNo">106</span> * The input path can either specify a directory or a file. The file may optionally be<a name="line.106"></a>
-<span class="sourceLineNo">107</span> * compressed with gzip. If a directory, the loader expects the directory to contain one or more<a name="line.107"></a>
-<span class="sourceLineNo">108</span> * WARC files from the Common Crawl dataset. If a file, the loader expects a list of Hadoop S3N<a name="line.108"></a>
-<span class="sourceLineNo">109</span> * URIs which point to S3 locations for one or more WARC files from the Common Crawl dataset,<a name="line.109"></a>
-<span class="sourceLineNo">110</span> * one URI per line. Lines should be terminated with the UNIX line terminator.<a name="line.110"></a>
-<span class="sourceLineNo">111</span> * <p><a name="line.111"></a>
-<span class="sourceLineNo">112</span> * Included in hbase-it/src/test/resources/CC-MAIN-2021-10-warc.paths.gz is a list of all WARC<a name="line.112"></a>
-<span class="sourceLineNo">113</span> * files comprising the Q1 2021 crawl archive. There are 64,000 WARC files in this data set, each<a name="line.113"></a>
-<span class="sourceLineNo">114</span> * containing ~1GB of gzipped data. The WARC files contain several record types, such as metadata,<a name="line.114"></a>
-<span class="sourceLineNo">115</span> * request, and response, but we only load the response record types. If the HBase table schema<a name="line.115"></a>
-<span class="sourceLineNo">116</span> * does not specify compression (by default) there is roughly a 10x expansion. Loading the full<a name="line.116"></a>
-<span class="sourceLineNo">117</span> * crawl archive results in a table approximately 640 TB in size.<a name="line.117"></a>
-<span class="sourceLineNo">118</span> * <p><a name="line.118"></a>
-<span class="sourceLineNo">119</span> * You can also split the Loader and Verify stages:<a name="line.119"></a>
-<span class="sourceLineNo">120</span> * <p><a name="line.120"></a>
-<span class="sourceLineNo">121</span> * Load with:<a name="line.121"></a>
-<span class="sourceLineNo">122</span> * <blockquote><a name="line.122"></a>
-<span class="sourceLineNo">123</span> * ./bin/hbase 'org.apache.hadoop.hbase.test.IntegrationTestLoadCommonCrawl$Loader' \<br><a name="line.123"></a>
-<span class="sourceLineNo">124</span> * &nbsp;&nbsp; -files /path/to/hadoop-aws.jar \<br><a name="line.124"></a>
-<span class="sourceLineNo">125</span> * &nbsp;&nbsp; -Dfs.s3n.awsAccessKeyId=&lt;AWS access key&gt; \<br><a name="line.125"></a>
-<span class="sourceLineNo">126</span> * &nbsp;&nbsp; -Dfs.s3n.awsSecretAccessKey=&lt;AWS secret key&gt; \<br><a name="line.126"></a>
-<span class="sourceLineNo">127</span> * &nbsp;&nbsp; /path/to/test-CC-MAIN-2021-10-warc.paths.gz \<br><a name="line.127"></a>
-<span class="sourceLineNo">128</span> * &nbsp;&nbsp; /path/to/tmp/warc-loader-output<a name="line.128"></a>
-<span class="sourceLineNo">129</span> * </blockquote><a name="line.129"></a>
-<span class="sourceLineNo">130</span> * <p><a name="line.130"></a>
-<span class="sourceLineNo">131</span> * Note: The hadoop-aws jar will be needed at runtime to instantiate the S3N filesystem. Use<a name="line.131"></a>
-<span class="sourceLineNo">132</span> * the <tt>-files</tt> ToolRunner argument to add it.<a name="line.132"></a>
-<span class="sourceLineNo">133</span> * <p><a name="line.133"></a>
-<span class="sourceLineNo">134</span> * Verify with:<a name="line.134"></a>
-<span class="sourceLineNo">135</span> * <blockquote><a name="line.135"></a>
-<span class="sourceLineNo">136</span> * ./bin/hbase 'org.apache.hadoop.hbase.test.IntegrationTestLoadCommonCrawl$Verify' \<br><a name="line.136"></a>
-<span class="sourceLineNo">137</span> * &nbsp;&nbsp; /path/to/tmp/warc-loader-output<a name="line.137"></a>
-<span class="sourceLineNo">138</span> * </blockquote><a name="line.138"></a>
-<span class="sourceLineNo">139</span> * <p><a name="line.139"></a>
-<span class="sourceLineNo">140</span> */<a name="line.140"></a>
-<span class="sourceLineNo">141</span>public class IntegrationTestLoadCommonCrawl extends IntegrationTestBase {<a name="line.141"></a>
-<span class="sourceLineNo">142</span><a name="line.142"></a>
-<span class="sourceLineNo">143</span> private static final Logger LOG = LoggerFactory.getLogger(IntegrationTestLoadCommonCrawl.class);<a name="line.143"></a>
-<span class="sourceLineNo">144</span><a name="line.144"></a>
-<span class="sourceLineNo">145</span> protected static String TABLE_NAME_KEY = "IntegrationTestLoadCommonCrawl.table";<a name="line.145"></a>
-<span class="sourceLineNo">146</span> protected static String DEFAULT_TABLE_NAME = "IntegrationTestLoadCommonCrawl";<a name="line.146"></a>
-<span class="sourceLineNo">147</span><a name="line.147"></a>
-<span class="sourceLineNo">148</span> protected static byte[] CONTENT_FAMILY_NAME = Bytes.toBytes("c");<a name="line.148"></a>
-<span class="sourceLineNo">149</span> protected static byte[] INFO_FAMILY_NAME = Bytes.toBytes("i");<a name="line.149"></a>
-<span class="sourceLineNo">150</span> protected static byte[] CONTENT_QUALIFIER = HConstants.EMPTY_BYTE_ARRAY;<a name="line.150"></a>
-<span class="sourceLineNo">151</span> protected static byte[] CONTENT_LENGTH_QUALIFIER = Bytes.toBytes("l");<a name="line.151"></a>
-<span class="sourceLineNo">152</span> protected static byte[] CONTENT_TYPE_QUALIFIER = Bytes.toBytes("t");<a name="line.152"></a>
-<span class="sourceLineNo">153</span> protected static byte[] CRC_QUALIFIER = Bytes.toBytes("c");<a name="line.153"></a>
-<span class="sourceLineNo">154</span> protected static byte[] DATE_QUALIFIER = Bytes.toBytes("d");<a name="line.154"></a>
-<span class="sourceLineNo">155</span> protected static byte[] IP_ADDRESS_QUALIFIER = Bytes.toBytes("a");<a name="line.155"></a>
-<span class="sourceLineNo">156</span> protected static byte[] RECORD_ID_QUALIFIER = Bytes.toBytes("r");<a name="line.156"></a>
-<span class="sourceLineNo">157</span> protected static byte[] TARGET_URI_QUALIFIER = Bytes.toBytes("u");<a name="line.157"></a>
-<span class="sourceLineNo">158</span><a name="line.158"></a>
-<span class="sourceLineNo">159</span> public static enum Counts {<a name="line.159"></a>
-<span class="sourceLineNo">160</span> REFERENCED, UNREFERENCED, CORRUPT<a name="line.160"></a>
-<span class="sourceLineNo">161</span> }<a name="line.161"></a>
-<span class="sourceLineNo">162</span><a name="line.162"></a>
-<span class="sourceLineNo">163</span> Path warcFileInputDir = null;<a name="line.163"></a>
-<span class="sourceLineNo">164</span> Path outputDir = null;<a name="line.164"></a>
-<span class="sourceLineNo">165</span> String[] args;<a name="line.165"></a>
-<span class="sourceLineNo">166</span><a name="line.166"></a>
-<span class="sourceLineNo">167</span> protected int runLoader(Path warcFileInputDir, Path outputDir) throws Exception {<a name="line.167"></a>
-<span class="sourceLineNo">168</span> Loader loader = new Loader();<a name="line.168"></a>
-<span class="sourceLineNo">169</span> loader.setConf(conf);<a name="line.169"></a>
-<span class="sourceLineNo">170</span> return loader.run(warcFileInputDir, outputDir);<a name="line.170"></a>
-<span class="sourceLineNo">171</span> }<a name="line.171"></a>
-<span class="sourceLineNo">172</span><a name="line.172"></a>
-<span class="sourceLineNo">173</span> protected int runVerify(Path inputDir) throws Exception {<a name="line.173"></a>
-<span class="sourceLineNo">174</span> Verify verify = new Verify();<a name="line.174"></a>
-<span class="sourceLineNo">175</span> verify.setConf(conf);<a name="line.175"></a>
-<span class="sourceLineNo">176</span> return verify.run(inputDir);<a name="line.176"></a>
-<span class="sourceLineNo">177</span> }<a name="line.177"></a>
-<span class="sourceLineNo">178</span><a name="line.178"></a>
-<span class="sourceLineNo">179</span> @Override<a name="line.179"></a>
-<span class="sourceLineNo">180</span> public int run(String[] args) {<a name="line.180"></a>
-<span class="sourceLineNo">181</span> if (args.length > 0) {<a name="line.181"></a>
-<span class="sourceLineNo">182</span> warcFileInputDir = new Path(args[0]);<a name="line.182"></a>
-<span class="sourceLineNo">183</span> if (args.length > 1) {<a name="line.183"></a>
-<span class="sourceLineNo">184</span> outputDir = new Path(args[1]);<a name="line.184"></a>
-<span class="sourceLineNo">185</span> }<a name="line.185"></a>
-<span class="sourceLineNo">186</span> }<a name="line.186"></a>
-<span class="sourceLineNo">187</span> try {<a name="line.187"></a>
-<span class="sourceLineNo">188</span> if (warcFileInputDir == null) {<a name="line.188"></a>
-<span class="sourceLineNo">189</span> throw new IllegalArgumentException("WARC input file or directory not specified");<a name="line.189"></a>
-<span class="sourceLineNo">190</span> }<a name="line.190"></a>
-<span class="sourceLineNo">191</span> if (outputDir == null) {<a name="line.191"></a>
-<span class="sourceLineNo">192</span> throw new IllegalArgumentException("Output directory not specified");<a name="line.192"></a>
+<span class="sourceLineNo">049</span>import org.apache.hadoop.hbase.client.BufferedMutator;<a name="line.49"></a>
+<span class="sourceLineNo">050</span>import org.apache.hadoop.hbase.client.ColumnFamilyDescriptor;<a name="line.50"></a>
+<span class="sourceLineNo">051</span>import org.apache.hadoop.hbase.client.ColumnFamilyDescriptorBuilder;<a name="line.51"></a>
+<span class="sourceLineNo">052</span>import org.apache.hadoop.hbase.client.Connection;<a name="line.52"></a>
+<span class="sourceLineNo">053</span>import org.apache.hadoop.hbase.client.ConnectionFactory;<a name="line.53"></a>
+<span class="sourceLineNo">054</span>import org.apache.hadoop.hbase.client.Get;<a name="line.54"></a>
+<span class="sourceLineNo">055</span>import org.apache.hadoop.hbase.client.Put;<a name="line.55"></a>
+<span class="sourceLineNo">056</span>import org.apache.hadoop.hbase.client.Result;<a name="line.56"></a>
+<span class="sourceLineNo">057</span>import org.apache.hadoop.hbase.client.Table;<a name="line.57"></a>
+<span class="sourceLineNo">058</span>import org.apache.hadoop.hbase.client.TableDescriptor;<a name="line.58"></a>
+<span class="sourceLineNo">059</span>import org.apache.hadoop.hbase.client.TableDescriptorBuilder;<a name="line.59"></a>
+<span class="sourceLineNo">060</span>import org.apache.hadoop.hbase.io.encoding.DataBlockEncoding;<a name="line.60"></a>
+<span class="sourceLineNo">061</span>import org.apache.hadoop.hbase.mapreduce.TableMapReduceUtil;<a name="line.61"></a>
+<span class="sourceLineNo">062</span>import org.apache.hadoop.hbase.regionserver.BloomType;<a name="line.62"></a>
+<span class="sourceLineNo">063</span>import org.apache.hadoop.hbase.test.util.CRC64;<a name="line.63"></a>
+<span class="sourceLineNo">064</span>import org.apache.hadoop.hbase.test.util.warc.WARCInputFormat;<a name="line.64"></a>
+<span class="sourceLineNo">065</span>import org.apache.hadoop.hbase.test.util.warc.WARCRecord;<a name="line.65"></a>
+<span class="sourceLineNo">066</span>import org.apache.hadoop.hbase.test.util.warc.WARCWritable;<a name="line.66"></a>
+<span class="sourceLineNo">067</span>import org.apache.hadoop.hbase.util.Bytes;<a name="line.67"></a>
+<span class="sourceLineNo">068</span>import org.apache.hadoop.hbase.util.EnvironmentEdgeManager;<a name="line.68"></a>
+<span class="sourceLineNo">069</span>import org.apache.hadoop.hbase.util.RegionSplitter;<a name="line.69"></a>
+<span class="sourceLineNo">070</span>import org.apache.hadoop.io.BytesWritable;<a name="line.70"></a>
+<span class="sourceLineNo">071</span>import org.apache.hadoop.io.LongWritable;<a name="line.71"></a>
+<span class="sourceLineNo">072</span>import org.apache.hadoop.io.NullWritable;<a name="line.72"></a>
+<span class="sourceLineNo">073</span>import org.apache.hadoop.io.SequenceFile.CompressionType;<a name="line.73"></a>
+<span class="sourceLineNo">074</span>import org.apache.hadoop.io.Writable;<a name="line.74"></a>
+<span class="sourceLineNo">075</span>import org.apache.hadoop.mapreduce.Counters;<a name="line.75"></a>
+<span class="sourceLineNo">076</span>import org.apache.hadoop.mapreduce.Job;<a name="line.76"></a>
+<span class="sourceLineNo">077</span>import org.apache.hadoop.mapreduce.JobContext;<a name="line.77"></a>
+<span class="sourceLineNo">078</span>import org.apache.hadoop.mapreduce.Mapper;<a name="line.78"></a>
+<span class="sourceLineNo">079</span>import org.apache.hadoop.mapreduce.lib.input.FileInputFormat;<a name="line.79"></a>
+<span class="sourceLineNo">080</span>import org.apache.hadoop.mapreduce.lib.input.SequenceFileInputFormat;<a name="line.80"></a>
+<span class="sourceLineNo">081</span>import org.apache.hadoop.mapreduce.lib.output.NullOutputFormat;<a name="line.81"></a>
+<span class="sourceLineNo">082</span>import org.apache.hadoop.mapreduce.lib.output.SequenceFileOutputFormat;<a name="line.82"></a>
+<span class="sourceLineNo">083</span>import org.apache.hadoop.util.Tool;<a name="line.83"></a>
+<span class="sourceLineNo">084</span>import org.apache.hadoop.util.ToolRunner;<a name="line.84"></a>
+<span class="sourceLineNo">085</span>import org.slf4j.Logger;<a name="line.85"></a>
+<span class="sourceLineNo">086</span>import org.slf4j.LoggerFactory;<a name="line.86"></a>
+<span class="sourceLineNo">087</span><a name="line.87"></a>
+<span class="sourceLineNo">088</span>import org.apache.hbase.thirdparty.org.apache.commons.cli.CommandLine;<a name="line.88"></a>
+<span class="sourceLineNo">089</span><a name="line.89"></a>
+<span class="sourceLineNo">090</span>/**<a name="line.90"></a>
+<span class="sourceLineNo">091</span> * This integration test loads successful resource retrieval records from the Common Crawl<a name="line.91"></a>
+<span class="sourceLineNo">092</span> * (https://commoncrawl.org/) public dataset into an HBase table and writes records that can be<a name="line.92"></a>
+<span class="sourceLineNo">093</span> * used to later verify the presence and integrity of those records.<a name="line.93"></a>
+<span class="sourceLineNo">094</span> * <p><a name="line.94"></a>
+<span class="sourceLineNo">095</span> * Run like:<a name="line.95"></a>
+<span class="sourceLineNo">096</span> * <blockquote><a name="line.96"></a>
+<span class="sourceLineNo">097</span> * ./bin/hbase org.apache.hadoop.hbase.test.IntegrationTestLoadCommonCrawl \<br><a name="line.97"></a>
+<span class="sourceLineNo">098</span> * &nbsp;&nbsp; -Dfs.s3n.awsAccessKeyId=&lt;AWS access key&gt; \<br><a name="line.98"></a>
+<span class="sourceLineNo">099</span> * &nbsp;&nbsp; -Dfs.s3n.awsSecretAccessKey=&lt;AWS secret key&gt; \<br><a name="line.99"></a>
+<span class="sourceLineNo">100</span> * &nbsp;&nbsp; /path/to/test-CC-MAIN-2021-10-warc.paths.gz \<br><a name="line.100"></a>
+<span class="sourceLineNo">101</span> * &nbsp;&nbsp; /path/to/tmp/warc-loader-output<a name="line.101"></a>
+<span class="sourceLineNo">102</span> * </blockquote><a name="line.102"></a>
+<span class="sourceLineNo">103</span> * <p><a name="line.103"></a>
+<span class="sourceLineNo">104</span> * Access to the Common Crawl dataset in S3 is made available to anyone by Amazon AWS, but<a name="line.104"></a>
+<span class="sourceLineNo">105</span> * Hadoop's S3N filesystem still requires valid access credentials to initialize.<a name="line.105"></a>
+<span class="sourceLineNo">106</span> * <p><a name="line.106"></a>
+<span class="sourceLineNo">107</span> * The input path can either specify a directory or a file. The file may optionally be<a name="line.107"></a>
+<span class="sourceLineNo">108</span> * compressed with gzip. If a directory, the loader expects the directory to contain one or more<a name="line.108"></a>
+<span class="sourceLineNo">109</span> * WARC files from the Common Crawl dataset. If a file, the loader expects a list of Hadoop S3N<a name="line.109"></a>
+<span class="sourceLineNo">110</span> * URIs which point to S3 locations for one or more WARC files from the Common Crawl dataset,<a name="line.110"></a>
+<span class="sourceLineNo">111</span> * one URI per line. Lines should be terminated with the UNIX line terminator.<a name="line.111"></a>
+<span class="sourceLineNo">112</span> * <p><a name="line.112"></a>
+<span class="sourceLineNo">113</span> * Included in hbase-it/src/test/resources/CC-MAIN-2021-10-warc.paths.gz is a list of all WARC<a name="line.113"></a>
+<span class="sourceLineNo">114</span> * files comprising the Q1 2021 crawl archive. There are 64,000 WARC files in this data set, each<a name="line.114"></a>
+<span class="sourceLineNo">115</span> * containing ~1GB of gzipped data. The WARC files contain several record types, such as metadata,<a name="line.115"></a>
+<span class="sourceLineNo">116</span> * request, and response, but we only load the response record types. If the HBase table schema<a name="line.116"></a>
+<span class="sourceLineNo">117</span> * does not specify compression (by default) there is roughly a 10x expansion. Loading the full<a name="line.117"></a>
+<span class="sourceLineNo">118</span> * crawl archive results in a table approximately 640 TB in size.<a name="line.118"></a>
+<span class="sourceLineNo">119</span> * <p><a name="line.119"></a>
+<span class="sourceLineNo">120</span> * You can also split the Loader and Verify stages:<a name="line.120"></a>
+<span class="sourceLineNo">121</span> * <p><a name="line.121"></a>
+<span class="sourceLineNo">122</span> * Load with:<a name="line.122"></a>
+<span class="sourceLineNo">123</span> * <blockquote><a name="line.123"></a>
+<span class="sourceLineNo">124</span> * ./bin/hbase 'org.apache.hadoop.hbase.test.IntegrationTestLoadCommonCrawl$Loader' \<br><a name="line.124"></a>
+<span class="sourceLineNo">125</span> * &nbsp;&nbsp; -files /path/to/hadoop-aws.jar \<br><a name="line.125"></a>
+<span class="sourceLineNo">126</span> * &nbsp;&nbsp; -Dfs.s3n.awsAccessKeyId=&lt;AWS access key&gt; \<br><a name="line.126"></a>
+<span class="sourceLineNo">127</span> * &nbsp;&nbsp; -Dfs.s3n.awsSecretAccessKey=&lt;AWS secret key&gt; \<br><a name="line.127"></a>
+<span class="sourceLineNo">128</span> * &nbsp;&nbsp; /path/to/test-CC-MAIN-2021-10-warc.paths.gz \<br><a name="line.128"></a>
+<span class="sourceLineNo">129</span> * &nbsp;&nbsp; /path/to/tmp/warc-loader-output<a name="line.129"></a>
+<span class="sourceLineNo">130</span> * </blockquote><a name="line.130"></a>
+<span class="sourceLineNo">131</span> * <p><a name="line.131"></a>
+<span class="sourceLineNo">132</span> * Note: The hadoop-aws jar will be needed at runtime to instantiate the S3N filesystem. Use<a name="line.132"></a>
+<span class="sourceLineNo">133</span> * the <tt>-files</tt> ToolRunner argument to add it.<a name="line.133"></a>
+<span class="sourceLineNo">134</span> * <p><a name="line.134"></a>
+<span class="sourceLineNo">135</span> * Verify with:<a name="line.135"></a>
+<span class="sourceLineNo">136</span> * <blockquote><a name="line.136"></a>
+<span class="sourceLineNo">137</span> * ./bin/hbase 'org.apache.hadoop.hbase.test.IntegrationTestLoadCommonCrawl$Verify' \<br><a name="line.137"></a>
+<span class="sourceLineNo">138</span> * &nbsp;&nbsp; /path/to/tmp/warc-loader-output<a name="line.138"></a>
+<span class="sourceLineNo">139</span> * </blockquote><a name="line.139"></a>
+<span class="sourceLineNo">140</span> * <p><a name="line.140"></a>
+<span class="sourceLineNo">141</span> */<a name="line.141"></a>
+<span class="sourceLineNo">142</span>public class IntegrationTestLoadCommonCrawl extends IntegrationTestBase {<a name="line.142"></a>
+<span class="sourceLineNo">143</span><a name="line.143"></a>
+<span class="sourceLineNo">144</span> private static final Logger LOG = LoggerFactory.getLogger(IntegrationTestLoadCommonCrawl.class);<a name="line.144"></a>
+<span class="sourceLineNo">145</span><a name="line.145"></a>
+<span class="sourceLineNo">146</span> protected static String TABLE_NAME_KEY = "IntegrationTestLoadCommonCrawl.table";<a name="line.146"></a>
+<span class="sourceLineNo">147</span> protected static String DEFAULT_TABLE_NAME = "IntegrationTestLoadCommonCrawl";<a name="line.147"></a>
+<span class="sourceLineNo">148</span><a name="line.148"></a>
+<span class="sourceLineNo">149</span> protected static byte[] CONTENT_FAMILY_NAME = Bytes.toBytes("c");<a name="line.149"></a>
+<span class="sourceLineNo">150</span> protected static byte[] INFO_FAMILY_NAME = Bytes.toBytes("i");<a name="line.150"></a>
+<span class="sourceLineNo">151</span> protected static byte[] CONTENT_QUALIFIER = HConstants.EMPTY_BYTE_ARRAY;<a name="line.151"></a>
+<span class="sourceLineNo">152</span> protected static byte[] CONTENT_LENGTH_QUALIFIER = Bytes.toBytes("l");<a name="line.152"></a>
+<span class="sourceLineNo">153</span> protected static byte[] CONTENT_TYPE_QUALIFIER = Bytes.toBytes("t");<a name="line.153"></a>
+<span class="sourceLineNo">154</span> protected static byte[] CRC_QUALIFIER = Bytes.toBytes("c");<a name="line.154"></a>
+<span class="sourceLineNo">155</span> protected static byte[] DATE_QUALIFIER = Bytes.toBytes("d");<a name="line.155"></a>
+<span class="sourceLineNo">156</span> protected static byte[] IP_ADDRESS_QUALIFIER = Bytes.toBytes("a");<a name="line.156"></a>
+<span class="sourceLineNo">157</span> protected static byte[] RECORD_ID_QUALIFIER = Bytes.toBytes("r");<a name="line.157"></a>
+<span class="sourceLineNo">158</span> protected static byte[] TARGET_URI_QUALIFIER = Bytes.toBytes("u");<a name="line.158"></a>
+<span class="sourceLineNo">159</span><a name="line.159"></a>
+<span class="sourceLineNo">160</span> private static final int VERIFICATION_READ_RETRIES = 10;<a name="line.160"></a>
+<span class="sourceLineNo">161</span><a name="line.161"></a>
+<span class="sourceLineNo">162</span> public static enum Counts {<a name="line.162"></a>
+<span class="sourceLineNo">163</span> REFERENCED, UNREFERENCED, CORRUPT<a name="line.163"></a>
+<span class="sourceLineNo">164</span> }<a name="line.164"></a>
+<span class="sourceLineNo">165</span><a name="line.165"></a>
+<span class="sourceLineNo">166</span> Path warcFileInputDir = null;<a name="line.166"></a>
+<span class="sourceLineNo">167</span> Path outputDir = null;<a name="line.167"></a>
+<span class="sourceLineNo">168</span> String[] args;<a name="line.168"></a>
+<span class="sourceLineNo">169</span><a name="line.169"></a>
+<span class="sourceLineNo">170</span> protected int runLoader(Path warcFileInputDir, Path outputDir) throws Exception {<a name="line.170"></a>
+<span class="sourceLineNo">171</span> Loader loader = new Loader();<a name="line.171"></a>
+<span class="sourceLineNo">172</span> loader.setConf(conf);<a name="line.172"></a>
+<span class="sourceLineNo">173</span> return loader.run(warcFileInputDir, outputDir);<a name="line.173"></a>
+<span class="sourceLineNo">174</span> }<a name="line.174"></a>
+<span class="sourceLineNo">175</span><a name="line.175"></a>
+<span class="sourceLineNo">176</span> protected int runVerify(Path inputDir) throws Exception {<a name="line.176"></a>
+<span class="sourceLineNo">177</span> Verify verify = new Verify();<a name="line.177"></a>
+<span class="sourceLineNo">178</span> verify.setConf(conf);<a name="line.178"></a>
+<span class="sourceLineNo">179</span> return verify.run(inputDir);<a name="line.179"></a>
+<span class="sourceLineNo">180</span> }<a name="line.180"></a>
+<span class="sourceLineNo">181</span><a name="line.181"></a>
+<span class="sourceLineNo">182</span> @Override<a name="line.182"></a>
+<span class="sourceLineNo">183</span> public int run(String[] args) {<a name="line.183"></a>
+<span class="sourceLineNo">184</span> if (args.length > 0) {<a name="line.184"></a>
+<span class="sourceLineNo">185</span> warcFileInputDir = new Path(args[0]);<a name="line.185"></a>
+<span class="sourceLineNo">186</span> if (args.length > 1) {<a name="line.186"></a>
+<span class="sourceLineNo">187</span> outputDir = new Path(args[1]);<a name="line.187"></a>
+<span class="sourceLineNo">188</span> }<a name="line.188"></a>
+<span class="sourceLineNo">189</span> }<a name="line.189"></a>
+<span class="sourceLineNo">190</span> try {<a name="line.190"></a>
+<span class="sourceLineNo">191</span> if (warcFileInputDir == null) {<a name="line.191"></a>
+<span class="sourceLineNo">192</span> throw new IllegalArgumentException("WARC input file or directory not specified");<a name="line.192"></a>
<span class="sourceLineNo">193</span> }<a name="line.193"></a>
-<span class="sourceLineNo">194</span> int res = runLoader(warcFileInputDir, outputDir);<a name="line.194"></a>
-<span class="sourceLineNo">195</span> if (res != 0) {<a name="line.195"></a>
... 4528 lines suppressed ...