You are viewing a plain text version of this content. The canonical link for it is here.
Posted to commits@manifoldcf.apache.org by kw...@apache.org on 2015/05/31 13:22:13 UTC
svn commit: r1682722 - in /manifoldcf/branches/dev_1x: ./
connectors/alfresco-webscript/connector/src/main/java/org/apache/manifoldcf/crawler/connectors/alfrescowebscript/AlfrescoConnector.java
Author: kwright
Date: Sun May 31 11:22:13 2015
New Revision: 1682722
URL: http://svn.apache.org/r1682722
Log:
Pull up fix for CONNECTORS-1200 from trunk
Modified:
manifoldcf/branches/dev_1x/ (props changed)
manifoldcf/branches/dev_1x/connectors/alfresco-webscript/connector/src/main/java/org/apache/manifoldcf/crawler/connectors/alfrescowebscript/AlfrescoConnector.java
Propchange: manifoldcf/branches/dev_1x/
------------------------------------------------------------------------------
--- svn:mergeinfo (original)
+++ svn:mergeinfo Sun May 31 11:22:13 2015
@@ -124,4 +124,4 @@
/manifoldcf/branches/CONNECTORS-981:1605049-1605773
/manifoldcf/branches/CONNECTORS-989:1611600-1612101
/manifoldcf/branches/CONNECTORS-990:1610284-1610707
-/manifoldcf/trunk:1620703,1620748,1620812,1620862,1621449,1621613,1621855,1622213,1622740,1622850,1622853-1622854,1623249,1623251,1623314,1623599,1623951,1623953-1623954,1623956,1623972,1624058,1624085,1624174,1624236,1624377,1624384,1624399,1624449,1624464,1624504,1624729-1624731,1624906,1624909-1624910,1624982,1625023,1625095,1625103,1625108,1625264,1625270,1625394,1625400,1625910,1626090,1626097,1626102,1626638-1626639,1626973,1627687,1627690,1627959,1628046,1628066,1628106,1628168,1628188,1628699,1628798,1628808,1628845,1628905,1629122,1629374-1629375,1629379,1629541,1629994,1630188,1630535,1630623,1630671,1630812,1630885,1631039,1631162,1631164,1631252,1631750,1631953,1632013,1632225,1632289,1632562,1632844,1632847,1632854,1633062-1633063,1633108,1633193,1633202,1633282,1633284,1633295,1633336,1633339,1633345,1633348,1633364,1633378,1633383,1633432,1633546,1633590,1633634,1633668,1633727,1633760,1633764,1633786,1633910,1633923,1634021,1634028,1634067,1634132,1634145,1634148,163
4155,1634188,1634202,1634264,1634373,1634530,1634688,1634850,1634857,1635103,1635116,1635421,1635438,1635478,1635481,1635484,1635490,1635809,1635939,1636146,1636167,1636180,1636207,1636215,1636232,1636334,1636519,1636570,1636684,1636940,1637011,1637310,1637350,1637364,1637373,1637378,1639259,1639593,1639600,1640018,1640101,1640199,1640204,1640208,1640314,1640319,1640749,1640772,1640805,1640888,1640925,1640941-1640942,1641222,1641328,1641557,1641559,1641629,1641633,1641724,1641754,1641911,1642163,1642255,1642318,1642531,1642650,1642658,1642673,1642716,1644197,1644399,1644538,1644920,1644931,1646317,1646397,1646403,1646408,1646640,1646947,1647574,1647585,1647608,1648686,1648976,1649201,1649203,1649529,1649605,1649628,1649794,1650351,1650722,1650741-1650742,1650745,1650747,1650911,1650954,1651332,1651539,1651907,1651921,1652071,1652974,1653175,1653899,1654651,1655205,1655261,1655264,1655377,1655411,1655618,1655914,1657346,1657443,1658004,1658036,1658121,1658155,1658188,1658463,1658476,
1660258,1660276,1661454,1665848,1666160,1666781,1666820,1668312,1669100,1669238,1669487,1669523,1669586,1669660,1670614,1670625,1670715,1671496,1672169,1672301,1672616,1672737,1673559,1673573,1673579,1673722,1675781,1675898,1676094,1676882,1676910,1678300,1678329,1678471,1678551,1679730,1679826,1681390,1681735,1682232,1682252,1682410,1682602,1682719
+/manifoldcf/trunk:1620703,1620748,1620812,1620862,1621449,1621613,1621855,1622213,1622740,1622850,1622853-1622854,1623249,1623251,1623314,1623599,1623951,1623953-1623954,1623956,1623972,1624058,1624085,1624174,1624236,1624377,1624384,1624399,1624449,1624464,1624504,1624729-1624731,1624906,1624909-1624910,1624982,1625023,1625095,1625103,1625108,1625264,1625270,1625394,1625400,1625910,1626090,1626097,1626102,1626638-1626639,1626973,1627687,1627690,1627959,1628046,1628066,1628106,1628168,1628188,1628699,1628798,1628808,1628845,1628905,1629122,1629374-1629375,1629379,1629541,1629994,1630188,1630535,1630623,1630671,1630812,1630885,1631039,1631162,1631164,1631252,1631750,1631953,1632013,1632225,1632289,1632562,1632844,1632847,1632854,1633062-1633063,1633108,1633193,1633202,1633282,1633284,1633295,1633336,1633339,1633345,1633348,1633364,1633378,1633383,1633432,1633546,1633590,1633634,1633668,1633727,1633760,1633764,1633786,1633910,1633923,1634021,1634028,1634067,1634132,1634145,1634148,163
4155,1634188,1634202,1634264,1634373,1634530,1634688,1634850,1634857,1635103,1635116,1635421,1635438,1635478,1635481,1635484,1635490,1635809,1635939,1636146,1636167,1636180,1636207,1636215,1636232,1636334,1636519,1636570,1636684,1636940,1637011,1637310,1637350,1637364,1637373,1637378,1639259,1639593,1639600,1640018,1640101,1640199,1640204,1640208,1640314,1640319,1640749,1640772,1640805,1640888,1640925,1640941-1640942,1641222,1641328,1641557,1641559,1641629,1641633,1641724,1641754,1641911,1642163,1642255,1642318,1642531,1642650,1642658,1642673,1642716,1644197,1644399,1644538,1644920,1644931,1646317,1646397,1646403,1646408,1646640,1646947,1647574,1647585,1647608,1648686,1648976,1649201,1649203,1649529,1649605,1649628,1649794,1650351,1650722,1650741-1650742,1650745,1650747,1650911,1650954,1651332,1651539,1651907,1651921,1652071,1652974,1653175,1653899,1654651,1655205,1655261,1655264,1655377,1655411,1655618,1655914,1657346,1657443,1658004,1658036,1658121,1658155,1658188,1658463,1658476,
1660258,1660276,1661454,1665848,1666160,1666781,1666820,1668312,1669100,1669238,1669487,1669523,1669586,1669660,1670614,1670625,1670715,1671496,1672169,1672301,1672616,1672737,1673559,1673573,1673579,1673722,1675781,1675898,1676094,1676882,1676910,1678300,1678329,1678471,1678551,1679730,1679826,1681390,1681735,1682232,1682252,1682410,1682602,1682622,1682719
Modified: manifoldcf/branches/dev_1x/connectors/alfresco-webscript/connector/src/main/java/org/apache/manifoldcf/crawler/connectors/alfrescowebscript/AlfrescoConnector.java
URL: http://svn.apache.org/viewvc/manifoldcf/branches/dev_1x/connectors/alfresco-webscript/connector/src/main/java/org/apache/manifoldcf/crawler/connectors/alfrescowebscript/AlfrescoConnector.java?rev=1682722&r1=1682721&r2=1682722&view=diff
==============================================================================
--- manifoldcf/branches/dev_1x/connectors/alfresco-webscript/connector/src/main/java/org/apache/manifoldcf/crawler/connectors/alfrescowebscript/AlfrescoConnector.java (original)
+++ manifoldcf/branches/dev_1x/connectors/alfresco-webscript/connector/src/main/java/org/apache/manifoldcf/crawler/connectors/alfrescowebscript/AlfrescoConnector.java Sun May 31 11:22:13 2015
@@ -41,10 +41,10 @@ import java.util.*;
public class AlfrescoConnector extends BaseRepositoryConnector {
-
+
private static final String ACTIVITY_FETCH = "fetch document";
private static final String[] activitiesList = new String[]{ACTIVITY_FETCH};
-
+
private AlfrescoClient alfrescoClient;
private static final String CONTENT_URL_PROPERTY = "contentUrlPath";
@@ -127,8 +127,13 @@ public class AlfrescoConnector extends B
if (Logging.connectors != null) {
Logging.connectors.warn(e.getMessage(), e);
}
- return "Connection failed: " + e.getMessage();
- }
+ return "Alfresco connection check failed: " + e.getMessage();
+ } catch (Exception e) {
+ if (Logging.connectors != null) {
+ Logging.connectors.error(e.getMessage(), e);
+ }
+ throw new ManifoldCFException("Alfresco connection check failed",e);
+ }
}
@Override
@@ -153,7 +158,7 @@ public class AlfrescoConnector extends B
try {
long lastTransactionId = 0;
long lastAclChangesetId = 0;
-
+
if(lastSeedVersion != null && !lastSeedVersion.isEmpty()) {
StringTokenizer tokenizer = new StringTokenizer(lastSeedVersion,"|");
@@ -162,15 +167,15 @@ public class AlfrescoConnector extends B
lastAclChangesetId = new Long(tokenizer.nextToken());
}
}
-
+
if (Logging.connectors != null && Logging.connectors.isDebugEnabled())
Logging.connectors.debug(MessageFormat.format("Starting from transaction id: {0} and acl changeset id: {1}", new Object[]{lastTransactionId, lastAclChangesetId}));
-
+
long transactionIdsProcessed;
long aclChangesetsProcessed;
do {
final AlfrescoResponse response = alfrescoClient.
- fetchNodes(lastTransactionId,
+ fetchNodes(lastTransactionId,
lastAclChangesetId,
ConfigurationHandler.getFilters(spec));
int count = 0;
@@ -210,7 +215,7 @@ public class AlfrescoConnector extends B
boolean enableDocumentProcessing = ConfigurationHandler.getEnableDocumentProcessing(spec);
for (String doc : documentIdentifiers) {
-
+
String errorCode = null;
String errorDesc = null;
Long fileLengthLong = null;
@@ -218,8 +223,8 @@ public class AlfrescoConnector extends B
try {
- String nextVersion = statuses.getIndexedVersionString(doc);
-
+ String nextVersion = statuses.getIndexedVersionString(doc);
+
// Calling again Alfresco API because Document's actions are lost from seeding method
AlfrescoResponse response = alfrescoClient.fetchNode(doc);
if(response.getDocumentList().isEmpty()){ // Not found seeded document. Could reflect an error in Alfresco
@@ -253,7 +258,7 @@ public class AlfrescoConnector extends B
String size = mdObject.toString();
lSize = new Long(size);
}
-
+
// Modified Date
Date modifiedDate = null;
mdObject = properties.get(MODIFIED_DATE_PROPERTY);
@@ -274,7 +279,7 @@ public class AlfrescoConnector extends B
activities.deleteDocument(doc);
continue;
}
-
+
String documentVersion = (enableDocumentProcessing?"+":"-") + new Long(modifiedDate.getTime()).toString();
if(!activities.checkDocumentNeedsReindexing(doc, documentVersion))
@@ -292,7 +297,7 @@ public class AlfrescoConnector extends B
errorDesc = "Excluding document because of length ("+lSize+")";
continue;
}
-
+
if (!activities.checkMimeTypeIndexable(mimeType)) {
activities.noDocument(doc, documentVersion);
errorCode = activities.EXCLUDED_MIMETYPE;
@@ -306,7 +311,7 @@ public class AlfrescoConnector extends B
errorDesc = "Excluding document because of date ("+modifiedDate+")";
continue;
}
-
+
String contentUrlPath = (String) properties.get(CONTENT_URL_PROPERTY);
if (contentUrlPath == null || contentUrlPath.isEmpty()) {
activities.noDocument(doc, documentVersion);
@@ -314,7 +319,7 @@ public class AlfrescoConnector extends B
errorDesc = "Excluding document because no URL found";
continue;
}
-
+
if (!activities.checkURLIndexable(contentUrlPath)) {
activities.noDocument(doc, documentVersion);
errorCode = activities.EXCLUDED_URL;
@@ -326,7 +331,7 @@ public class AlfrescoConnector extends B
rd.addField(FIELD_NODEREF, nodeRef);
rd.addField(FIELD_TYPE, type);
rd.setFileName(name);
-
+
if (modifiedDate != null)
rd.setModifiedDate(modifiedDate);
@@ -353,7 +358,7 @@ public class AlfrescoConnector extends B
InputStream stream;
long length;
byte[] empty = new byte[0];
-
+
if (enableDocumentProcessing) {
if (lSize != null) {
stream = alfrescoClient.fetchContent(contentUrlPath);
@@ -372,7 +377,7 @@ public class AlfrescoConnector extends B
stream = new ByteArrayInputStream(empty);
length = 0L;
}
-
+
try {
rd.setBinary(stream, length);
if (Logging.connectors != null && Logging.connectors.isDebugEnabled())
@@ -405,7 +410,7 @@ public class AlfrescoConnector extends B
}
protected final static long interruptionRetryTime = 5L*60L*1000L;
-
+
protected static void handleAlfrescoDownException(AlfrescoDownException e, String context)
throws ManifoldCFException, ServiceInterruption {
long currentTime = System.currentTimeMillis();
@@ -420,7 +425,7 @@ public class AlfrescoConnector extends B
3,
true);
}
-
+
protected static void handleIOException(IOException e, String context)
throws ManifoldCFException, ServiceInterruption
{
@@ -428,7 +433,7 @@ public class AlfrescoConnector extends B
throw new ManifoldCFException(e.getMessage(), ManifoldCFException.INTERRUPTED);
long currentTime = System.currentTimeMillis();
-
+
if (e instanceof java.net.ConnectException)
{
// Server isn't up at all. Try for a brief time then give up.
@@ -441,7 +446,7 @@ public class AlfrescoConnector extends B
3,
true);
}
-
+
if (e instanceof java.net.SocketTimeoutException)
{
String message2 = "Socket timeout exception during "+context+": "+e.getMessage();
@@ -453,7 +458,7 @@ public class AlfrescoConnector extends B
-1,
false);
}
-
+
if (e.getClass().getName().equals("java.net.SocketException"))
{
// In the past we would have treated this as a straight document rejection, and
@@ -480,8 +485,8 @@ public class AlfrescoConnector extends B
3,
false);
}
-
- // Other socket exceptions are service interruptions - but if we keep getting them, it means
+
+ // Other socket exceptions are service interruptions - but if we keep getting them, it means
// that a socket timeout is probably set too low to accept this particular document. So
// we retry for a while, then skip the document.
String message2 = "Socket exception during "+context+": "+e.getMessage();
@@ -504,7 +509,7 @@ public class AlfrescoConnector extends B
-1,
true);
}
-
+
@Override
public void outputConfigurationHeader(IThreadContext threadContext,
IHTTPOutput out, Locale locale, ConfigParams parameters,