You are viewing a plain text version of this content. The canonical link for it is here.
Posted to github@beam.apache.org by GitBox <gi...@apache.org> on 2020/08/20 04:10:28 UTC
[GitHub] [beam] abhiy13 opened a new pull request #12645: [BEAM-10124] Add ContextualTextIO
abhiy13 opened a new pull request #12645:
URL: https://github.com/apache/beam/pull/12645
[BEAM-10124] Added ContextualTextIO
------------------------
Thank you for your contribution! Follow this checklist to help us incorporate your contribution quickly and easily:
- [x] [**Choose reviewer(s)**](https://beam.apache.org/contribute/#make-your-change) and mention them in a comment (`R: @username`).
- [x] Format the pull request title like `[BEAM-XXX] Fixes bug in ApproximateQuantiles`, where you replace `BEAM-XXX` with the appropriate JIRA issue, if applicable. This will automatically link the pull request to the issue.
- [ ] Update `CHANGES.md` with noteworthy changes.
- [ ] If this contribution is large, please file an Apache [Individual Contributor License Agreement](https://www.apache.org/licenses/icla.pdf).
See the [Contributor Guide](https://beam.apache.org/contribute) for more tips on [how to make review process smoother](https://beam.apache.org/contribute/#make-reviewers-job-easier).
Post-Commit Tests Status (on master branch)
------------------------------------------------------------------------------------------------
Lang | SDK | Dataflow | Flink | Samza | Spark | Twister2
--- | --- | --- | --- | --- | --- | ---
Go | [](https://ci-beam.apache.org/job/beam_PostCommit_Go/lastCompletedBuild/) | --- | [](https://ci-beam.apache.org/job/beam_PostCommit_Go_VR_Flink/lastCompletedBuild/) | --- | [](https://ci-beam.apache.org/job/beam_PostCommit_Go_VR_Spark/lastCompletedBuild/) | ---
Java | [](https://ci-beam.apache.org/job/beam_PostCommit_Java/lastCompletedBuild/) | [](https://ci-beam.apache.org/job/beam_PostCommit_Java_ValidatesRunner_Dataflow/lastCompletedBuild/)<br>[](https://ci-beam.apache.org/job/beam_PostCommit_Java_ValidatesRunner_Dataflow_Java11/lastCompletedBuild/) | [](https://ci-beam.apache.org/job/beam_PostCommit_Java_ValidatesRunner_Flink/lastCompletedBuild/)<br>[](https://ci-beam.apache.org/job/beam_PostCommit_Java_ValidatesRunner_Flink_Java11/lastCompletedBuild/)<br>[](https://ci-beam.apache.org/job/beam_PostCommit_Java_PVR_Flink_Batch/lastCompletedBuild/)<br>[](https://ci-beam.apache.org/job/beam_PostCommit_Java_PVR_Flink_Streaming/lastCompletedBuild/) | [](https://ci-beam.apache.org/job/beam_PostCommit_Java_ValidatesRunner_Samza/lastCompletedBuild/) | [](https://ci-beam.apache.org/job/beam_PostCommit_Java_ValidatesRunner_Spark/lastCompletedBuild/)<br>[](https://ci-beam.apache.org/job/beam_PostCommit_Java_PVR_Spark_Batch/lastCompletedBuild/)<br>[](https://ci-beam.apache.org/job/beam_PostCommit_Java_ValidatesRunner_SparkStructuredStreaming/lastCompletedBuild/) | [](https://builds.apache.org/job/beam_PostCommit_Java_ValidatesRunner_Twister2/lastCompletedBuild/)
Python | [](https://ci-beam.apache.org/job/beam_PostCommit_Python2/lastCompletedBuild/)<br>[](https://ci-beam.apache.org/job/beam_PostCommit_Python35/lastCompletedBuild/)<br>[](https://ci-beam.apache.org/job/beam_PostCommit_Python36/lastCompletedBuild/)<br>[](https://ci-beam.apache.org/job/beam_PostCommit_Python37/lastCompletedBuild/)<br>[](https://ci-beam.apache.org/job/beam_PostCommit_Python38/lastCompletedBuild/) | [](https://ci-beam.apache.org/job/beam_PostCommit_Py_VR_Dataflow/lastCompletedBuild/)<br>[](https://ci-beam.apache.org/job/beam_PostCommit_Py_VR_Dataflow_V2/lastCompletedBuild/)<br>[](https://ci-beam.apache.org/job/beam_PostCommit_Py_ValCont/lastCompletedBuild/) | [](https://ci-beam.apache.org/job/beam_PreCommit_Python2_PVR_Flink_Cron/lastCompletedBuild/)<br>[](https://ci-beam.apache.org/job/beam_PostCommit_Python35_VR_Flink/lastCompletedBuild/) | --- | [](https://ci-beam.apache.org/job/beam_PostCommit_Python_VR_Spark/lastCompletedBuild/) | ---
XLang | [](https://ci-beam.apache.org/job/beam_PostCommit_XVR_Direct/lastCompletedBuild/) | --- | [](https://ci-beam.apache.org/job/beam_PostCommit_XVR_Flink/lastCompletedBuild/) | --- | [](https://ci-beam.apache.org/job/beam_PostCommit_XVR_Spark/lastCompletedBuild/) | ---
Pre-Commit Tests Status (on master branch)
------------------------------------------------------------------------------------------------
--- |Java | Python | Go | Website
--- | --- | --- | --- | ---
Non-portable | [](https://ci-beam.apache.org/job/beam_PreCommit_Java_Cron/lastCompletedBuild/) | [](https://ci-beam.apache.org/job/beam_PreCommit_Python_Cron/lastCompletedBuild/)<br>[](https://ci-beam.apache.org/job/beam_PreCommit_PythonLint_Cron/lastCompletedBuild/)<br>[](https://ci-beam.apache.org/job/beam_PreCommit_PythonDocker_Cron/lastCompletedBuild/) <br>[](https://ci-beam.apache.org/job/beam_PreCommit_PythonDocs_Cron/lastCompletedBuild/) | [](https://ci-beam.apache.org/job/beam_PreCommit_Go_Cron/lastCompletedBuild/) | [](https://ci-beam.apache.org/job/beam_PreCommit_Website_Cron/lastCompletedBuild/)
Portable | --- | [](https://ci-beam.apache.org/job/beam_PreCommit_Portable_Python_Cron/lastCompletedBuild/) | --- | ---
See [.test-infra/jenkins/README](https://github.com/apache/beam/blob/master/.test-infra/jenkins/README.md) for trigger phrase, status and link of all Jenkins jobs.
GitHub Actions Tests Status (on master branch)
------------------------------------------------------------------------------------------------
See [CI.md](https://github.com/apache/beam/blob/master/CI.md) for more information about GitHub Actions CI.
R: @tvalentyn @rezarokni
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
[GitHub] [beam] codecov[bot] edited a comment on pull request #12645: [BEAM-10124] Add ContextualTextIO
Posted by GitBox <gi...@apache.org>.
codecov[bot] edited a comment on pull request #12645:
URL: https://github.com/apache/beam/pull/12645#issuecomment-688630083
# [Codecov](https://codecov.io/gh/apache/beam/pull/12645?src=pr&el=h1) Report
> :exclamation: No coverage uploaded for pull request base (`master@2b2b8e7`). [Click here to learn what that means](https://docs.codecov.io/docs/error-reference#section-missing-base-commit).
> The diff coverage is `n/a`.
[](https://codecov.io/gh/apache/beam/pull/12645?src=pr&el=tree)
```diff
@@ Coverage Diff @@
## master #12645 +/- ##
=========================================
Coverage ? 82.33%
=========================================
Files ? 453
Lines ? 54054
Branches ? 0
=========================================
Hits ? 44507
Misses ? 9547
Partials ? 0
```
| [Impacted Files](https://codecov.io/gh/apache/beam/pull/12645?src=pr&el=tree) | Coverage Δ | |
|---|---|---|
| [...nners/direct/consumer\_tracking\_pipeline\_visitor.py](https://codecov.io/gh/apache/beam/pull/12645/diff?src=pr&el=tree#diff-c2Rrcy9weXRob24vYXBhY2hlX2JlYW0vcnVubmVycy9kaXJlY3QvY29uc3VtZXJfdHJhY2tpbmdfcGlwZWxpbmVfdmlzaXRvci5weQ==) | `96.96% <0.00%> (ø)` | |
| [sdks/python/apache\_beam/internal/pickler.py](https://codecov.io/gh/apache/beam/pull/12645/diff?src=pr&el=tree#diff-c2Rrcy9weXRob24vYXBhY2hlX2JlYW0vaW50ZXJuYWwvcGlja2xlci5weQ==) | `85.82% <0.00%> (ø)` | |
| [...beam/portability/api/beam\_artifact\_api\_pb2\_urns.py](https://codecov.io/gh/apache/beam/pull/12645/diff?src=pr&el=tree#diff-c2Rrcy9weXRob24vYXBhY2hlX2JlYW0vcG9ydGFiaWxpdHkvYXBpL2JlYW1fYXJ0aWZhY3RfYXBpX3BiMl91cm5zLnB5) | `0.00% <0.00%> (ø)` | |
| [sdks/python/apache\_beam/examples/\_\_init\_\_.py](https://codecov.io/gh/apache/beam/pull/12645/diff?src=pr&el=tree#diff-c2Rrcy9weXRob24vYXBhY2hlX2JlYW0vZXhhbXBsZXMvX19pbml0X18ucHk=) | `100.00% <0.00%> (ø)` | |
| [...nners/portability/fn\_api\_runner/worker\_handlers.py](https://codecov.io/gh/apache/beam/pull/12645/diff?src=pr&el=tree#diff-c2Rrcy9weXRob24vYXBhY2hlX2JlYW0vcnVubmVycy9wb3J0YWJpbGl0eS9mbl9hcGlfcnVubmVyL3dvcmtlcl9oYW5kbGVycy5weQ==) | `80.75% <0.00%> (ø)` | |
| [...eam/testing/benchmarks/nexmark/nexmark\_launcher.py](https://codecov.io/gh/apache/beam/pull/12645/diff?src=pr&el=tree#diff-c2Rrcy9weXRob24vYXBhY2hlX2JlYW0vdGVzdGluZy9iZW5jaG1hcmtzL25leG1hcmsvbmV4bWFya19sYXVuY2hlci5weQ==) | `0.00% <0.00%> (ø)` | |
| [sdks/python/apache\_beam/io/gcp/bigtableio.py](https://codecov.io/gh/apache/beam/pull/12645/diff?src=pr&el=tree#diff-c2Rrcy9weXRob24vYXBhY2hlX2JlYW0vaW8vZ2NwL2JpZ3RhYmxlaW8ucHk=) | `42.22% <0.00%> (ø)` | |
| [...ache\_beam/io/gcp/datastore/v1new/query\_splitter.py](https://codecov.io/gh/apache/beam/pull/12645/diff?src=pr&el=tree#diff-c2Rrcy9weXRob24vYXBhY2hlX2JlYW0vaW8vZ2NwL2RhdGFzdG9yZS92MW5ldy9xdWVyeV9zcGxpdHRlci5weQ==) | `94.11% <0.00%> (ø)` | |
| [sdks/python/apache\_beam/transforms/\_\_init\_\_.py](https://codecov.io/gh/apache/beam/pull/12645/diff?src=pr&el=tree#diff-c2Rrcy9weXRob24vYXBhY2hlX2JlYW0vdHJhbnNmb3Jtcy9fX2luaXRfXy5weQ==) | `100.00% <0.00%> (ø)` | |
| [...am/testing/benchmarks/chicago\_taxi/trainer/taxi.py](https://codecov.io/gh/apache/beam/pull/12645/diff?src=pr&el=tree#diff-c2Rrcy9weXRob24vYXBhY2hlX2JlYW0vdGVzdGluZy9iZW5jaG1hcmtzL2NoaWNhZ29fdGF4aS90cmFpbmVyL3RheGkucHk=) | `0.00% <0.00%> (ø)` | |
| ... and [443 more](https://codecov.io/gh/apache/beam/pull/12645/diff?src=pr&el=tree-more) | |
------
[Continue to review full report at Codecov](https://codecov.io/gh/apache/beam/pull/12645?src=pr&el=continue).
> **Legend** - [Click here to learn more](https://docs.codecov.io/docs/codecov-delta)
> `Δ = absolute <relative> (impact)`, `ø = not affected`, `? = missing data`
> Powered by [Codecov](https://codecov.io/gh/apache/beam/pull/12645?src=pr&el=footer). Last update [2b2b8e7...85c579f](https://codecov.io/gh/apache/beam/pull/12645?src=pr&el=lastupdated). Read the [comment docs](https://docs.codecov.io/docs/pull-request-comments).
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
[GitHub] [beam] abhiy13 commented on a change in pull request #12645: [BEAM-10124] Add ContextualTextIO
Posted by GitBox <gi...@apache.org>.
abhiy13 commented on a change in pull request #12645:
URL: https://github.com/apache/beam/pull/12645#discussion_r474445924
##########
File path: sdks/java/io/contextual-text-io/src/main/java/org/apache/beam/sdk/io/contextualtextio/ContextualTextIO.java
##########
@@ -319,31 +326,169 @@ static boolean isSelfOverlapping(byte[] s) {
}
@Override
- public PCollection<String> expand(PBegin input) {
- checkNotNull(getFilepattern(), "need to set the filepattern of a TextIO.Read transform");
+ public PCollection<RecordWithMetadata> expand(PBegin input) {
+ checkNotNull(
+ getFilepattern(), "need to set the filepattern of a ContextualTextIO.Read transform");
+ PCollection<RecordWithMetadata> lines = null;
if (getMatchConfiguration().getWatchInterval() == null && !getHintMatchesManyFiles()) {
- return input.apply("Read", org.apache.beam.sdk.io.Read.from(getSource()));
+ lines = input.apply("Read", org.apache.beam.sdk.io.Read.from(getSource()));
+ } else {
+ // All other cases go through FileIO + ReadFiles
+ lines =
+ input
+ .apply(
+ "Create filepattern", Create.ofProvider(getFilepattern(), StringUtf8Coder.of()))
+ .apply("Match All", FileIO.matchAll().withConfiguration(getMatchConfiguration()))
+ .apply(
+ "Read Matches",
+ FileIO.readMatches()
+ .withCompression(getCompression())
+ .withDirectoryTreatment(DirectoryTreatment.PROHIBIT))
+ .apply("Via ReadFiles", readFiles().withDelimiter(getDelimiter()));
}
- // All other cases go through FileIO + ReadFiles
- return input
- .apply("Create filepattern", Create.ofProvider(getFilepattern(), StringUtf8Coder.of()))
- .apply("Match All", FileIO.matchAll().withConfiguration(getMatchConfiguration()))
- .apply(
- "Read Matches",
- FileIO.readMatches()
- .withCompression(getCompression())
- .withDirectoryTreatment(DirectoryTreatment.PROHIBIT))
- .apply("Via ReadFiles", readFiles().withDelimiter(getDelimiter()));
+ // Check if the user decided to opt out of recordNums associated with records
+ if (getWithoutLineNumMetadata()) {
+ return lines;
+ }
+
+ // At this point the line number in RecordWithMetadata contains the relative line offset from
+ // the
+ // beginning of the read range.
+
+ // To compute the absolute position from the beginning of the input,
+ // we group the lines within the same ranges, and evaluate the size of each range.
+
+ PCollection<KV<KV<String, Long>, RecordWithMetadata>> linesGroupedByFileAndRange =
+ lines.apply("AddFileNameAndRange", ParDo.of(new AddFileNameAndRange()));
+
+ PCollectionView<Map<KV<String, Long>, Long>> rangeSizes =
+ linesGroupedByFileAndRange
+ .apply("CountLinesForEachFileRange", Count.perKey())
+ .apply("SizesAsView", View.asMap());
+
+ // Get Pipeline to create a dummy PCollection with one element to help compute the lines
+ // before each Range
+ PCollection<Integer> singletonPcoll =
+ input.getPipeline().apply("CreateSingletonPcoll", Create.of(Arrays.asList(1)));
+
+ // For each (File, Offset) pair, calculate the number of lines occurring before the Range for
Review comment:
Ack.
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
[GitHub] [beam] rezarokni commented on a change in pull request #12645: [BEAM-10124] Add ContextualTextIO
Posted by GitBox <gi...@apache.org>.
rezarokni commented on a change in pull request #12645:
URL: https://github.com/apache/beam/pull/12645#discussion_r474359911
##########
File path: sdks/java/io/contextual-text-io/src/main/java/org/apache/beam/sdk/io/contextualtextio/RecordWithMetadata.java
##########
@@ -0,0 +1,57 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one
+ * or more contributor license agreements. See the NOTICE file
+ * distributed with this work for additional information
+ * regarding copyright ownership. The ASF licenses this file
+ * to you under the Apache License, Version 2.0 (the
+ * "License"); you may not use this file except in compliance
+ * with the License. You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+package org.apache.beam.sdk.io.contextualtextio;
+
+import com.google.auto.value.AutoValue;
+import org.apache.beam.sdk.annotations.Experimental;
+import org.apache.beam.sdk.annotations.Internal;
+import org.apache.beam.sdk.schemas.AutoValueSchema;
+import org.apache.beam.sdk.schemas.annotations.DefaultSchema;
+
+@Internal
+@Experimental(Experimental.Kind.SCHEMAS)
+@DefaultSchema(AutoValueSchema.class)
+@AutoValue
+public abstract class RecordWithMetadata {
+ public abstract Range getRange();
+
+ public abstract Long getRecordNum();
Review comment:
Not sure I can think of a usecase where the offset is useful outside of when a user wants to rebuild the actual linenumber location in the file.
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
[GitHub] [beam] tvalentyn commented on a change in pull request #12645: [BEAM-10124] Add ContextualTextIO
Posted by GitBox <gi...@apache.org>.
tvalentyn commented on a change in pull request #12645:
URL: https://github.com/apache/beam/pull/12645#discussion_r473585601
##########
File path: sdks/java/io/contextual-text-io/src/main/java/org/apache/beam/sdk/io/contextualtextio/ContextualTextIO.java
##########
@@ -319,31 +326,169 @@ static boolean isSelfOverlapping(byte[] s) {
}
@Override
- public PCollection<String> expand(PBegin input) {
- checkNotNull(getFilepattern(), "need to set the filepattern of a TextIO.Read transform");
+ public PCollection<RecordWithMetadata> expand(PBegin input) {
+ checkNotNull(
+ getFilepattern(), "need to set the filepattern of a ContextualTextIO.Read transform");
+ PCollection<RecordWithMetadata> lines = null;
if (getMatchConfiguration().getWatchInterval() == null && !getHintMatchesManyFiles()) {
- return input.apply("Read", org.apache.beam.sdk.io.Read.from(getSource()));
+ lines = input.apply("Read", org.apache.beam.sdk.io.Read.from(getSource()));
+ } else {
+ // All other cases go through FileIO + ReadFiles
+ lines =
+ input
+ .apply(
+ "Create filepattern", Create.ofProvider(getFilepattern(), StringUtf8Coder.of()))
+ .apply("Match All", FileIO.matchAll().withConfiguration(getMatchConfiguration()))
+ .apply(
+ "Read Matches",
+ FileIO.readMatches()
+ .withCompression(getCompression())
+ .withDirectoryTreatment(DirectoryTreatment.PROHIBIT))
+ .apply("Via ReadFiles", readFiles().withDelimiter(getDelimiter()));
}
- // All other cases go through FileIO + ReadFiles
- return input
- .apply("Create filepattern", Create.ofProvider(getFilepattern(), StringUtf8Coder.of()))
- .apply("Match All", FileIO.matchAll().withConfiguration(getMatchConfiguration()))
- .apply(
- "Read Matches",
- FileIO.readMatches()
- .withCompression(getCompression())
- .withDirectoryTreatment(DirectoryTreatment.PROHIBIT))
- .apply("Via ReadFiles", readFiles().withDelimiter(getDelimiter()));
+ // Check if the user decided to opt out of recordNums associated with records
+ if (getWithoutLineNumMetadata()) {
+ return lines;
+ }
+
+ // At this point the line number in RecordWithMetadata contains the relative line offset from
Review comment:
nit: fix the indentation
##########
File path: sdks/java/io/contextual-text-io/src/main/java/org/apache/beam/sdk/io/contextualtextio/ContextualTextIO.java
##########
@@ -319,31 +326,169 @@ static boolean isSelfOverlapping(byte[] s) {
}
@Override
- public PCollection<String> expand(PBegin input) {
- checkNotNull(getFilepattern(), "need to set the filepattern of a TextIO.Read transform");
+ public PCollection<RecordWithMetadata> expand(PBegin input) {
+ checkNotNull(
+ getFilepattern(), "need to set the filepattern of a ContextualTextIO.Read transform");
+ PCollection<RecordWithMetadata> lines = null;
if (getMatchConfiguration().getWatchInterval() == null && !getHintMatchesManyFiles()) {
- return input.apply("Read", org.apache.beam.sdk.io.Read.from(getSource()));
+ lines = input.apply("Read", org.apache.beam.sdk.io.Read.from(getSource()));
+ } else {
+ // All other cases go through FileIO + ReadFiles
+ lines =
Review comment:
records?
##########
File path: sdks/java/io/contextual-text-io/src/main/java/org/apache/beam/sdk/io/contextualtextio/Range.java
##########
@@ -0,0 +1,42 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one
+ * or more contributor license agreements. See the NOTICE file
+ * distributed with this work for additional information
+ * regarding copyright ownership. The ASF licenses this file
+ * to you under the Apache License, Version 2.0 (the
+ * "License"); you may not use this file except in compliance
+ * with the License. You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+package org.apache.beam.sdk.io.contextualtextio;
+
+import com.google.auto.value.AutoValue;
+
+@AutoValue
+public abstract class Range {
+ public abstract Long getRangeNum();
Review comment:
(Please add docstrings and consider a different name).
##########
File path: sdks/java/io/contextual-text-io/src/main/java/org/apache/beam/sdk/io/contextualtextio/ContextualTextIO.java
##########
@@ -319,31 +326,169 @@ static boolean isSelfOverlapping(byte[] s) {
}
@Override
- public PCollection<String> expand(PBegin input) {
- checkNotNull(getFilepattern(), "need to set the filepattern of a TextIO.Read transform");
+ public PCollection<RecordWithMetadata> expand(PBegin input) {
+ checkNotNull(
+ getFilepattern(), "need to set the filepattern of a ContextualTextIO.Read transform");
+ PCollection<RecordWithMetadata> lines = null;
if (getMatchConfiguration().getWatchInterval() == null && !getHintMatchesManyFiles()) {
- return input.apply("Read", org.apache.beam.sdk.io.Read.from(getSource()));
+ lines = input.apply("Read", org.apache.beam.sdk.io.Read.from(getSource()));
+ } else {
+ // All other cases go through FileIO + ReadFiles
+ lines =
+ input
+ .apply(
+ "Create filepattern", Create.ofProvider(getFilepattern(), StringUtf8Coder.of()))
+ .apply("Match All", FileIO.matchAll().withConfiguration(getMatchConfiguration()))
+ .apply(
+ "Read Matches",
+ FileIO.readMatches()
+ .withCompression(getCompression())
+ .withDirectoryTreatment(DirectoryTreatment.PROHIBIT))
+ .apply("Via ReadFiles", readFiles().withDelimiter(getDelimiter()));
}
- // All other cases go through FileIO + ReadFiles
- return input
- .apply("Create filepattern", Create.ofProvider(getFilepattern(), StringUtf8Coder.of()))
- .apply("Match All", FileIO.matchAll().withConfiguration(getMatchConfiguration()))
- .apply(
- "Read Matches",
- FileIO.readMatches()
- .withCompression(getCompression())
- .withDirectoryTreatment(DirectoryTreatment.PROHIBIT))
- .apply("Via ReadFiles", readFiles().withDelimiter(getDelimiter()));
+ // Check if the user decided to opt out of recordNums associated with records
+ if (getWithoutLineNumMetadata()) {
+ return lines;
+ }
+
+ // At this point the line number in RecordWithMetadata contains the relative line offset from
+ // the
+ // beginning of the read range.
+
+ // To compute the absolute position from the beginning of the input,
+ // we group the lines within the same ranges, and evaluate the size of each range.
+
+ PCollection<KV<KV<String, Long>, RecordWithMetadata>> linesGroupedByFileAndRange =
+ lines.apply("AddFileNameAndRange", ParDo.of(new AddFileNameAndRange()));
+
+ PCollectionView<Map<KV<String, Long>, Long>> rangeSizes =
+ linesGroupedByFileAndRange
+ .apply("CountLinesForEachFileRange", Count.perKey())
+ .apply("SizesAsView", View.asMap());
+
+ // Get Pipeline to create a dummy PCollection with one element to help compute the lines
+ // before each Range
+ PCollection<Integer> singletonPcoll =
+ input.getPipeline().apply("CreateSingletonPcoll", Create.of(Arrays.asList(1)));
+
+ // For each (File, Offset) pair, calculate the number of lines occurring before the Range for
+ // each File
+
+ // After computing the number of lines before each range, we can find the line number in
+ // original file as numLiesBeforeOffset + lineNumInCurrentOffset
+ PCollectionView<Map<KV<String, Long>, Long>> numLinesBeforeEachRange =
+ singletonPcoll
+ .apply(
+ "ComputeLinesBeforeRange",
Review comment:
ComputeRecordsBeforeRange
##########
File path: sdks/java/io/contextual-text-io/src/main/java/org/apache/beam/sdk/io/contextualtextio/ContextualTextIO.java
##########
@@ -296,6 +290,19 @@ public Read withHintMatchesManyFiles() {
return toBuilder().setHintMatchesManyFiles(true).build();
}
+ /**
+ * Allows the user the opt out of getting recordNums associated with each record.
Review comment:
Allows the user to opt out of ...
##########
File path: sdks/java/io/contextual-text-io/src/main/java/org/apache/beam/sdk/io/contextualtextio/ContextualTextIO.java
##########
@@ -296,6 +290,19 @@ public Read withHintMatchesManyFiles() {
return toBuilder().setHintMatchesManyFiles(true).build();
}
+ /**
+ * Allows the user the opt out of getting recordNums associated with each record.
+ *
+ * <p>ContextualTextIO uses a shuffle step to assemble the recordNums for each record which may
+ * result in some performance loss.
+ *
+ * <p>Use this when metadata like fileNames are required and their position/order can be
+ * ignored.
+ */
+ public Read withoutLineNumMetadata() {
Review comment:
s/withoutLineNumMetadata/withoutRecordNumMetadata ?
##########
File path: sdks/java/io/contextual-text-io/src/main/java/org/apache/beam/sdk/io/contextualtextio/ContextualTextIO.java
##########
@@ -319,31 +326,169 @@ static boolean isSelfOverlapping(byte[] s) {
}
@Override
- public PCollection<String> expand(PBegin input) {
- checkNotNull(getFilepattern(), "need to set the filepattern of a TextIO.Read transform");
+ public PCollection<RecordWithMetadata> expand(PBegin input) {
+ checkNotNull(
+ getFilepattern(), "need to set the filepattern of a ContextualTextIO.Read transform");
+ PCollection<RecordWithMetadata> lines = null;
if (getMatchConfiguration().getWatchInterval() == null && !getHintMatchesManyFiles()) {
- return input.apply("Read", org.apache.beam.sdk.io.Read.from(getSource()));
+ lines = input.apply("Read", org.apache.beam.sdk.io.Read.from(getSource()));
+ } else {
+ // All other cases go through FileIO + ReadFiles
+ lines =
+ input
+ .apply(
+ "Create filepattern", Create.ofProvider(getFilepattern(), StringUtf8Coder.of()))
+ .apply("Match All", FileIO.matchAll().withConfiguration(getMatchConfiguration()))
+ .apply(
+ "Read Matches",
+ FileIO.readMatches()
+ .withCompression(getCompression())
+ .withDirectoryTreatment(DirectoryTreatment.PROHIBIT))
+ .apply("Via ReadFiles", readFiles().withDelimiter(getDelimiter()));
}
- // All other cases go through FileIO + ReadFiles
- return input
- .apply("Create filepattern", Create.ofProvider(getFilepattern(), StringUtf8Coder.of()))
- .apply("Match All", FileIO.matchAll().withConfiguration(getMatchConfiguration()))
- .apply(
- "Read Matches",
- FileIO.readMatches()
- .withCompression(getCompression())
- .withDirectoryTreatment(DirectoryTreatment.PROHIBIT))
- .apply("Via ReadFiles", readFiles().withDelimiter(getDelimiter()));
+ // Check if the user decided to opt out of recordNums associated with records
+ if (getWithoutLineNumMetadata()) {
+ return lines;
+ }
+
+ // At this point the line number in RecordWithMetadata contains the relative line offset from
+ // the
+ // beginning of the read range.
+
+ // To compute the absolute position from the beginning of the input,
+ // we group the lines within the same ranges, and evaluate the size of each range.
+
+ PCollection<KV<KV<String, Long>, RecordWithMetadata>> linesGroupedByFileAndRange =
+ lines.apply("AddFileNameAndRange", ParDo.of(new AddFileNameAndRange()));
+
+ PCollectionView<Map<KV<String, Long>, Long>> rangeSizes =
+ linesGroupedByFileAndRange
+ .apply("CountLinesForEachFileRange", Count.perKey())
+ .apply("SizesAsView", View.asMap());
+
+ // Get Pipeline to create a dummy PCollection with one element to help compute the lines
+ // before each Range
+ PCollection<Integer> singletonPcoll =
+ input.getPipeline().apply("CreateSingletonPcoll", Create.of(Arrays.asList(1)));
+
+ // For each (File, Offset) pair, calculate the number of lines occurring before the Range for
+ // each File
+
+ // After computing the number of lines before each range, we can find the line number in
Review comment:
similarly here line numbers -> records
##########
File path: sdks/java/io/contextual-text-io/src/main/java/org/apache/beam/sdk/io/contextualtextio/ContextualTextIO.java
##########
@@ -319,31 +326,169 @@ static boolean isSelfOverlapping(byte[] s) {
}
@Override
- public PCollection<String> expand(PBegin input) {
- checkNotNull(getFilepattern(), "need to set the filepattern of a TextIO.Read transform");
+ public PCollection<RecordWithMetadata> expand(PBegin input) {
+ checkNotNull(
+ getFilepattern(), "need to set the filepattern of a ContextualTextIO.Read transform");
+ PCollection<RecordWithMetadata> lines = null;
if (getMatchConfiguration().getWatchInterval() == null && !getHintMatchesManyFiles()) {
- return input.apply("Read", org.apache.beam.sdk.io.Read.from(getSource()));
+ lines = input.apply("Read", org.apache.beam.sdk.io.Read.from(getSource()));
+ } else {
+ // All other cases go through FileIO + ReadFiles
+ lines =
+ input
+ .apply(
+ "Create filepattern", Create.ofProvider(getFilepattern(), StringUtf8Coder.of()))
+ .apply("Match All", FileIO.matchAll().withConfiguration(getMatchConfiguration()))
+ .apply(
+ "Read Matches",
+ FileIO.readMatches()
+ .withCompression(getCompression())
+ .withDirectoryTreatment(DirectoryTreatment.PROHIBIT))
+ .apply("Via ReadFiles", readFiles().withDelimiter(getDelimiter()));
}
- // All other cases go through FileIO + ReadFiles
- return input
- .apply("Create filepattern", Create.ofProvider(getFilepattern(), StringUtf8Coder.of()))
- .apply("Match All", FileIO.matchAll().withConfiguration(getMatchConfiguration()))
- .apply(
- "Read Matches",
- FileIO.readMatches()
- .withCompression(getCompression())
- .withDirectoryTreatment(DirectoryTreatment.PROHIBIT))
- .apply("Via ReadFiles", readFiles().withDelimiter(getDelimiter()));
+ // Check if the user decided to opt out of recordNums associated with records
+ if (getWithoutLineNumMetadata()) {
+ return lines;
+ }
+
+ // At this point the line number in RecordWithMetadata contains the relative line offset from
+ // the
+ // beginning of the read range.
+
+ // To compute the absolute position from the beginning of the input,
+ // we group the lines within the same ranges, and evaluate the size of each range.
+
+ PCollection<KV<KV<String, Long>, RecordWithMetadata>> linesGroupedByFileAndRange =
+ lines.apply("AddFileNameAndRange", ParDo.of(new AddFileNameAndRange()));
+
+ PCollectionView<Map<KV<String, Long>, Long>> rangeSizes =
+ linesGroupedByFileAndRange
Review comment:
recordsGroupedByFileAndRange
##########
File path: sdks/java/io/contextual-text-io/src/main/java/org/apache/beam/sdk/io/contextualtextio/ContextualTextIO.java
##########
@@ -319,31 +326,169 @@ static boolean isSelfOverlapping(byte[] s) {
}
@Override
- public PCollection<String> expand(PBegin input) {
- checkNotNull(getFilepattern(), "need to set the filepattern of a TextIO.Read transform");
+ public PCollection<RecordWithMetadata> expand(PBegin input) {
+ checkNotNull(
+ getFilepattern(), "need to set the filepattern of a ContextualTextIO.Read transform");
+ PCollection<RecordWithMetadata> lines = null;
if (getMatchConfiguration().getWatchInterval() == null && !getHintMatchesManyFiles()) {
- return input.apply("Read", org.apache.beam.sdk.io.Read.from(getSource()));
+ lines = input.apply("Read", org.apache.beam.sdk.io.Read.from(getSource()));
+ } else {
+ // All other cases go through FileIO + ReadFiles
+ lines =
+ input
+ .apply(
+ "Create filepattern", Create.ofProvider(getFilepattern(), StringUtf8Coder.of()))
+ .apply("Match All", FileIO.matchAll().withConfiguration(getMatchConfiguration()))
+ .apply(
+ "Read Matches",
+ FileIO.readMatches()
+ .withCompression(getCompression())
+ .withDirectoryTreatment(DirectoryTreatment.PROHIBIT))
+ .apply("Via ReadFiles", readFiles().withDelimiter(getDelimiter()));
}
- // All other cases go through FileIO + ReadFiles
- return input
- .apply("Create filepattern", Create.ofProvider(getFilepattern(), StringUtf8Coder.of()))
- .apply("Match All", FileIO.matchAll().withConfiguration(getMatchConfiguration()))
- .apply(
- "Read Matches",
- FileIO.readMatches()
- .withCompression(getCompression())
- .withDirectoryTreatment(DirectoryTreatment.PROHIBIT))
- .apply("Via ReadFiles", readFiles().withDelimiter(getDelimiter()));
+ // Check if the user decided to opt out of recordNums associated with records
+ if (getWithoutLineNumMetadata()) {
+ return lines;
+ }
+
+ // At this point the line number in RecordWithMetadata contains the relative line offset from
+ // the
+ // beginning of the read range.
+
+ // To compute the absolute position from the beginning of the input,
+ // we group the lines within the same ranges, and evaluate the size of each range.
+
+ PCollection<KV<KV<String, Long>, RecordWithMetadata>> linesGroupedByFileAndRange =
+ lines.apply("AddFileNameAndRange", ParDo.of(new AddFileNameAndRange()));
+
+ PCollectionView<Map<KV<String, Long>, Long>> rangeSizes =
+ linesGroupedByFileAndRange
+ .apply("CountLinesForEachFileRange", Count.perKey())
+ .apply("SizesAsView", View.asMap());
+
+ // Get Pipeline to create a dummy PCollection with one element to help compute the lines
+ // before each Range
+ PCollection<Integer> singletonPcoll =
+ input.getPipeline().apply("CreateSingletonPcoll", Create.of(Arrays.asList(1)));
+
+ // For each (File, Offset) pair, calculate the number of lines occurring before the Range for
Review comment:
... calculate the number of records occurring before the beginning of the Range for each file?
##########
File path: sdks/java/io/contextual-text-io/src/main/java/org/apache/beam/sdk/io/contextualtextio/ContextualTextIOSource.java
##########
@@ -0,0 +1,363 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one
+ * or more contributor license agreements. See the NOTICE file
+ * distributed with this work for additional information
+ * regarding copyright ownership. The ASF licenses this file
+ * to you under the Apache License, Version 2.0 (the
+ * "License"); you may not use this file except in compliance
+ * with the License. You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+package org.apache.beam.sdk.io.contextualtextio;
+
+import java.io.IOException;
+import java.nio.ByteBuffer;
+import java.nio.channels.ReadableByteChannel;
+import java.nio.channels.SeekableByteChannel;
+import java.util.NoSuchElementException;
+import org.apache.beam.sdk.coders.Coder;
+import org.apache.beam.sdk.io.FileBasedSource;
+import org.apache.beam.sdk.io.fs.EmptyMatchTreatment;
+import org.apache.beam.sdk.io.fs.MatchResult;
+import org.apache.beam.sdk.options.PipelineOptions;
+import org.apache.beam.sdk.options.ValueProvider;
+import org.apache.beam.sdk.schemas.NoSuchSchemaException;
+import org.apache.beam.sdk.schemas.SchemaCoder;
+import org.apache.beam.sdk.schemas.SchemaRegistry;
+import org.apache.beam.vendor.grpc.v1p26p0.com.google.protobuf.ByteString;
+import org.apache.beam.vendor.guava.v26_0_jre.com.google.common.annotations.VisibleForTesting;
+import org.apache.beam.vendor.guava.v26_0_jre.com.google.common.base.Preconditions;
+import org.checkerframework.checker.nullness.qual.Nullable;
+
+/**
+ * Implementation detail of {@link ContextualTextIO.Read}.
+ *
+ * <p>A {@link FileBasedSource} which can decode records delimited by newline characters.
+ *
+ * <p>This source splits the data into records using {@code UTF-8} {@code \n}, {@code \r}, or {@code
+ * \r\n} as the delimiter. This source is not strict and supports decoding the last record even if
+ * it is not delimited. Finally, no records are decoded if the stream is empty.
+ *
+ * <p>This source supports reading from any arbitrary byte position within the stream. If the
+ * starting position is not {@code 0}, then bytes are skipped until the first delimiter is found
+ * representing the beginning of the first record to be decoded.
+ */
+@VisibleForTesting
+class ContextualTextIOSource extends FileBasedSource<RecordWithMetadata> {
+ byte[] delimiter;
+
+ // Used to Override isSplittable
+ private boolean hasMultilineCSVRecords;
+
+ @Override
+ protected boolean isSplittable() throws Exception {
+ if (hasMultilineCSVRecords) {
+ return false;
+ }
+ return super.isSplittable();
+ }
+
+ ContextualTextIOSource(
+ ValueProvider<String> fileSpec,
+ EmptyMatchTreatment emptyMatchTreatment,
+ byte[] delimiter,
+ boolean hasMultilineCSVRecords) {
+ super(fileSpec, emptyMatchTreatment, 1L);
+ this.delimiter = delimiter;
+ this.hasMultilineCSVRecords = hasMultilineCSVRecords;
+ }
+
+ private ContextualTextIOSource(
+ MatchResult.Metadata metadata,
+ long start,
+ long end,
+ byte[] delimiter,
+ boolean hasMultilineCSVRecords) {
+ super(metadata, 1L, start, end);
+ this.delimiter = delimiter;
+ this.hasMultilineCSVRecords = hasMultilineCSVRecords;
+ }
+
+ @Override
+ protected FileBasedSource<RecordWithMetadata> createForSubrangeOfFile(
+ MatchResult.Metadata metadata, long start, long end) {
+ return new ContextualTextIOSource(metadata, start, end, delimiter, hasMultilineCSVRecords);
+ }
+
+ @Override
+ protected FileBasedReader<RecordWithMetadata> createSingleFileReader(PipelineOptions options) {
+ return new MultiLineTextBasedReader(this, delimiter, hasMultilineCSVRecords);
+ }
+
+ @Override
+ public Coder<RecordWithMetadata> getOutputCoder() {
+ SchemaCoder<RecordWithMetadata> coder = null;
+ try {
+ coder = SchemaRegistry.createDefault().getSchemaCoder(RecordWithMetadata.class);
+ } catch (NoSuchSchemaException e) {
+ System.out.println("No Coder!");
Review comment:
Logger here as well, see: https://github.com/apache/beam/blob/71c7760f4b5c5bf0d91e2c8403fae99216308a3e/sdks/java/io/google-cloud-platform/src/main/java/org/apache/beam/sdk/io/gcp/datastore/DatastoreV1.java#L897
##########
File path: sdks/java/io/contextual-text-io/src/main/java/org/apache/beam/sdk/io/contextualtextio/ContextualTextIO.java
##########
@@ -319,31 +326,169 @@ static boolean isSelfOverlapping(byte[] s) {
}
@Override
- public PCollection<String> expand(PBegin input) {
- checkNotNull(getFilepattern(), "need to set the filepattern of a TextIO.Read transform");
+ public PCollection<RecordWithMetadata> expand(PBegin input) {
+ checkNotNull(
+ getFilepattern(), "need to set the filepattern of a ContextualTextIO.Read transform");
+ PCollection<RecordWithMetadata> lines = null;
if (getMatchConfiguration().getWatchInterval() == null && !getHintMatchesManyFiles()) {
- return input.apply("Read", org.apache.beam.sdk.io.Read.from(getSource()));
+ lines = input.apply("Read", org.apache.beam.sdk.io.Read.from(getSource()));
+ } else {
+ // All other cases go through FileIO + ReadFiles
+ lines =
+ input
+ .apply(
+ "Create filepattern", Create.ofProvider(getFilepattern(), StringUtf8Coder.of()))
+ .apply("Match All", FileIO.matchAll().withConfiguration(getMatchConfiguration()))
+ .apply(
+ "Read Matches",
+ FileIO.readMatches()
+ .withCompression(getCompression())
+ .withDirectoryTreatment(DirectoryTreatment.PROHIBIT))
+ .apply("Via ReadFiles", readFiles().withDelimiter(getDelimiter()));
}
- // All other cases go through FileIO + ReadFiles
- return input
- .apply("Create filepattern", Create.ofProvider(getFilepattern(), StringUtf8Coder.of()))
- .apply("Match All", FileIO.matchAll().withConfiguration(getMatchConfiguration()))
- .apply(
- "Read Matches",
- FileIO.readMatches()
- .withCompression(getCompression())
- .withDirectoryTreatment(DirectoryTreatment.PROHIBIT))
- .apply("Via ReadFiles", readFiles().withDelimiter(getDelimiter()));
+ // Check if the user decided to opt out of recordNums associated with records
+ if (getWithoutLineNumMetadata()) {
+ return lines;
+ }
+
+ // At this point the line number in RecordWithMetadata contains the relative line offset from
+ // the
+ // beginning of the read range.
+
+ // To compute the absolute position from the beginning of the input,
+ // we group the lines within the same ranges, and evaluate the size of each range.
+
+ PCollection<KV<KV<String, Long>, RecordWithMetadata>> linesGroupedByFileAndRange =
+ lines.apply("AddFileNameAndRange", ParDo.of(new AddFileNameAndRange()));
+
+ PCollectionView<Map<KV<String, Long>, Long>> rangeSizes =
+ linesGroupedByFileAndRange
+ .apply("CountLinesForEachFileRange", Count.perKey())
+ .apply("SizesAsView", View.asMap());
+
+ // Get Pipeline to create a dummy PCollection with one element to help compute the lines
+ // before each Range
+ PCollection<Integer> singletonPcoll =
+ input.getPipeline().apply("CreateSingletonPcoll", Create.of(Arrays.asList(1)));
+
+ // For each (File, Offset) pair, calculate the number of lines occurring before the Range for
+ // each File
+
+ // After computing the number of lines before each range, we can find the line number in
+ // original file as numLiesBeforeOffset + lineNumInCurrentOffset
+ PCollectionView<Map<KV<String, Long>, Long>> numLinesBeforeEachRange =
+ singletonPcoll
+ .apply(
+ "ComputeLinesBeforeRange",
+ ParDo.of(new ComputeLinesBeforeEachRange(rangeSizes)).withSideInputs(rangeSizes))
+ .apply("NumLinesBeforeEachRangeAsView", View.asMap());
+
+ return linesGroupedByFileAndRange.apply(
+ "AssignLineNums",
+ ParDo.of(new AssignLineNums(numLinesBeforeEachRange))
+ .withSideInputs(numLinesBeforeEachRange));
+ }
+
+ @VisibleForTesting
+ static class AddFileNameAndRange
+ extends DoFn<RecordWithMetadata, KV<KV<String, Long>, RecordWithMetadata>> {
+ @ProcessElement
+ public void processElement(
+ @Element RecordWithMetadata line,
+ OutputReceiver<KV<KV<String, Long>, RecordWithMetadata>> out) {
+ out.output(KV.of(KV.of(line.getFileName(), line.getRange().getRangeNum()), line));
+ }
+ }
+
+ /**
+ * Helper class for computing number of lines in the File preceding the beginning of the Range
+ * in this file.
+ */
+ @VisibleForTesting
+ static class ComputeLinesBeforeEachRange extends DoFn<Integer, KV<KV<String, Long>, Long>> {
+ private final PCollectionView<Map<KV<String, Long>, Long>> rangeSizes;
+
+ public ComputeLinesBeforeEachRange(PCollectionView<Map<KV<String, Long>, Long>> rangeSizes) {
+ this.rangeSizes = rangeSizes;
+ }
+
+ // Add custom comparator as KV<K, V> is not comparable by default
+ private static class FileRangeComparator<K extends Comparable<K>, V extends Comparable<V>>
+ implements Comparator<KV<K, V>> {
+ @Override
+ public int compare(KV<K, V> a, KV<K, V> b) {
+ if (a.getKey().compareTo(b.getKey()) == 0) {
+ return a.getValue().compareTo(b.getValue());
+ }
+ return a.getKey().compareTo(b.getKey());
+ }
+ }
+
+ @ProcessElement
+ public void processElement(ProcessContext p) {
+ // Get the Map Containing the size from side-input
+ Map<KV<String, Long>, Long> rangeSizesMap = p.sideInput(rangeSizes);
+
+ // The FileRange Pair must be sorted
+ SortedMap<KV<String, Long>, Long> sorted = new TreeMap<>(new FileRangeComparator<>());
+
+ // Initialize sorted map with values
+ for (Map.Entry<KV<String, Long>, Long> entry : rangeSizesMap.entrySet()) {
+ sorted.put(entry.getKey(), entry.getValue());
+ }
+
+ // HashMap that tracks lines passed for each file
+ Map<String, Long> pastLines = new HashMap<>();
+
+ // For each (File, Range) Pair, compute the number of lines before it
+ for (Map.Entry entry : sorted.entrySet()) {
+ Long lines = (long) entry.getValue();
Review comment:
lines/records?
##########
File path: sdks/java/io/contextual-text-io/src/main/java/org/apache/beam/sdk/io/contextualtextio/ContextualTextIOSource.java
##########
@@ -259,7 +317,31 @@ private void decodeCurrentElement() throws IOException {
if (startOfRecord == 0 && dataToDecode.startsWith(UTF8_BOM)) {
dataToDecode = dataToDecode.substring(UTF8_BOM.size());
}
- currentValue = dataToDecode.toStringUtf8();
+
+ /////////////////////////////////////////////
+
+ // Data of the Current Line
+ // dataToDecode.toStringUtf8();
+
+ // The line num is:
Review comment:
redundant comment
##########
File path: sdks/java/io/contextual-text-io/src/main/java/org/apache/beam/sdk/io/contextualtextio/ContextualTextIOSource.java
##########
@@ -259,7 +317,31 @@ private void decodeCurrentElement() throws IOException {
if (startOfRecord == 0 && dataToDecode.startsWith(UTF8_BOM)) {
dataToDecode = dataToDecode.substring(UTF8_BOM.size());
}
- currentValue = dataToDecode.toStringUtf8();
+
+ /////////////////////////////////////////////
+
+ // Data of the Current Line
Review comment:
Do we need this commented-out code?
##########
File path: sdks/java/io/contextual-text-io/src/main/java/org/apache/beam/sdk/io/contextualtextio/ContextualTextIOSource.java
##########
@@ -99,14 +126,23 @@ private ContextualTextIOSource(MatchResult.Metadata metadata, long start, long e
private volatile long startOfNextRecord;
private volatile boolean eof;
private volatile boolean elementIsPresent;
- private @Nullable String currentValue;
+ private @Nullable RecordWithMetadata currentValue;
private @Nullable ReadableByteChannel inChannel;
private byte @Nullable [] delimiter;
- private TextBasedReader(ContextualTextIOSource source, byte[] delimiter) {
+ // Add to override the isSplittable
+ private boolean hasRFC4180MultiLineColumn;
+
+ private long startingOffset;
+ private long readerlineNum;
+
+ private MultiLineTextBasedReader(
+ ContextualTextIOSource source, byte[] delimiter, boolean hasRFC4180MultiLineColumn) {
Review comment:
hasMultilineCSVRecords
##########
File path: sdks/java/io/contextual-text-io/src/main/java/org/apache/beam/sdk/io/contextualtextio/ContextualTextIOSource.java
##########
@@ -152,18 +188,27 @@ protected void startReading(ReadableByteChannel channel) throws IOException {
requiredPosition = startOffset - delimiter.length;
}
((SeekableByteChannel) channel).position(requiredPosition);
- findDelimiterBounds();
+ findDelimiterBoundsWithMultiLineCheck();
buffer = buffer.substring(endOfDelimiterInBuffer);
startOfNextRecord = requiredPosition + endOfDelimiterInBuffer;
endOfDelimiterInBuffer = 0;
startOfDelimiterInBuffer = 0;
}
}
+ private void findDelimiterBoundsWithMultiLineCheck() throws IOException {
Review comment:
Do we need this helper?
##########
File path: sdks/java/io/contextual-text-io/src/main/java/org/apache/beam/sdk/io/contextualtextio/Range.java
##########
@@ -0,0 +1,42 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one
+ * or more contributor license agreements. See the NOTICE file
+ * distributed with this work for additional information
+ * regarding copyright ownership. The ASF licenses this file
+ * to you under the Apache License, Version 2.0 (the
+ * "License"); you may not use this file except in compliance
+ * with the License. You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+package org.apache.beam.sdk.io.contextualtextio;
+
+import com.google.auto.value.AutoValue;
+
+@AutoValue
+public abstract class Range {
+ public abstract Long getRangeNum();
+
+ public abstract Long getRangeLineNum();
Review comment:
What is this? Number of records in the range?
(Please add docstrings and consider a different name).
##########
File path: sdks/java/io/contextual-text-io/src/main/java/org/apache/beam/sdk/io/contextualtextio/RecordWithMetadata.java
##########
@@ -0,0 +1,57 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one
+ * or more contributor license agreements. See the NOTICE file
+ * distributed with this work for additional information
+ * regarding copyright ownership. The ASF licenses this file
+ * to you under the Apache License, Version 2.0 (the
+ * "License"); you may not use this file except in compliance
+ * with the License. You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+package org.apache.beam.sdk.io.contextualtextio;
+
+import com.google.auto.value.AutoValue;
+import org.apache.beam.sdk.annotations.Experimental;
+import org.apache.beam.sdk.annotations.Internal;
+import org.apache.beam.sdk.schemas.AutoValueSchema;
+import org.apache.beam.sdk.schemas.annotations.DefaultSchema;
+
+@Internal
+@Experimental(Experimental.Kind.SCHEMAS)
+@DefaultSchema(AutoValueSchema.class)
+@AutoValue
+public abstract class RecordWithMetadata {
+ public abstract Range getRange();
+
+ public abstract Long getRecordNum();
Review comment:
@rezarokni @abhiy13 would it make sense to include the offset of the record in the input (if it's easy to compute, also can be computed even when user disables withoutRecordNumMetadata (without shuffles/sideinput) )?
Can also be done later..
##########
File path: sdks/java/io/contextual-text-io/src/main/java/org/apache/beam/sdk/io/contextualtextio/ContextualTextIO.java
##########
@@ -319,31 +326,169 @@ static boolean isSelfOverlapping(byte[] s) {
}
@Override
- public PCollection<String> expand(PBegin input) {
- checkNotNull(getFilepattern(), "need to set the filepattern of a TextIO.Read transform");
+ public PCollection<RecordWithMetadata> expand(PBegin input) {
+ checkNotNull(
+ getFilepattern(), "need to set the filepattern of a ContextualTextIO.Read transform");
+ PCollection<RecordWithMetadata> lines = null;
if (getMatchConfiguration().getWatchInterval() == null && !getHintMatchesManyFiles()) {
- return input.apply("Read", org.apache.beam.sdk.io.Read.from(getSource()));
+ lines = input.apply("Read", org.apache.beam.sdk.io.Read.from(getSource()));
+ } else {
+ // All other cases go through FileIO + ReadFiles
+ lines =
+ input
+ .apply(
+ "Create filepattern", Create.ofProvider(getFilepattern(), StringUtf8Coder.of()))
+ .apply("Match All", FileIO.matchAll().withConfiguration(getMatchConfiguration()))
+ .apply(
+ "Read Matches",
+ FileIO.readMatches()
+ .withCompression(getCompression())
+ .withDirectoryTreatment(DirectoryTreatment.PROHIBIT))
+ .apply("Via ReadFiles", readFiles().withDelimiter(getDelimiter()));
}
- // All other cases go through FileIO + ReadFiles
- return input
- .apply("Create filepattern", Create.ofProvider(getFilepattern(), StringUtf8Coder.of()))
- .apply("Match All", FileIO.matchAll().withConfiguration(getMatchConfiguration()))
- .apply(
- "Read Matches",
- FileIO.readMatches()
- .withCompression(getCompression())
- .withDirectoryTreatment(DirectoryTreatment.PROHIBIT))
- .apply("Via ReadFiles", readFiles().withDelimiter(getDelimiter()));
+ // Check if the user decided to opt out of recordNums associated with records
+ if (getWithoutLineNumMetadata()) {
+ return lines;
+ }
+
+ // At this point the line number in RecordWithMetadata contains the relative line offset from
+ // the
+ // beginning of the read range.
+
+ // To compute the absolute position from the beginning of the input,
+ // we group the lines within the same ranges, and evaluate the size of each range.
+
+ PCollection<KV<KV<String, Long>, RecordWithMetadata>> linesGroupedByFileAndRange =
+ lines.apply("AddFileNameAndRange", ParDo.of(new AddFileNameAndRange()));
+
+ PCollectionView<Map<KV<String, Long>, Long>> rangeSizes =
+ linesGroupedByFileAndRange
+ .apply("CountLinesForEachFileRange", Count.perKey())
+ .apply("SizesAsView", View.asMap());
+
+ // Get Pipeline to create a dummy PCollection with one element to help compute the lines
+ // before each Range
+ PCollection<Integer> singletonPcoll =
+ input.getPipeline().apply("CreateSingletonPcoll", Create.of(Arrays.asList(1)));
+
+ // For each (File, Offset) pair, calculate the number of lines occurring before the Range for
+ // each File
+
+ // After computing the number of lines before each range, we can find the line number in
+ // original file as numLiesBeforeOffset + lineNumInCurrentOffset
+ PCollectionView<Map<KV<String, Long>, Long>> numLinesBeforeEachRange =
+ singletonPcoll
+ .apply(
+ "ComputeLinesBeforeRange",
+ ParDo.of(new ComputeLinesBeforeEachRange(rangeSizes)).withSideInputs(rangeSizes))
+ .apply("NumLinesBeforeEachRangeAsView", View.asMap());
+
+ return linesGroupedByFileAndRange.apply(
+ "AssignLineNums",
+ ParDo.of(new AssignLineNums(numLinesBeforeEachRange))
+ .withSideInputs(numLinesBeforeEachRange));
+ }
+
+ @VisibleForTesting
+ static class AddFileNameAndRange
+ extends DoFn<RecordWithMetadata, KV<KV<String, Long>, RecordWithMetadata>> {
+ @ProcessElement
+ public void processElement(
+ @Element RecordWithMetadata line,
+ OutputReceiver<KV<KV<String, Long>, RecordWithMetadata>> out) {
+ out.output(KV.of(KV.of(line.getFileName(), line.getRange().getRangeNum()), line));
+ }
+ }
+
+ /**
+ * Helper class for computing number of lines in the File preceding the beginning of the Range
+ * in this file.
+ */
+ @VisibleForTesting
+ static class ComputeLinesBeforeEachRange extends DoFn<Integer, KV<KV<String, Long>, Long>> {
Review comment:
ComputeRecordsBeforeEachRange or ComputeNumRecordsBeforeEachRange ?
##########
File path: sdks/java/io/contextual-text-io/src/main/java/org/apache/beam/sdk/io/contextualtextio/ContextualTextIO.java
##########
@@ -319,31 +326,169 @@ static boolean isSelfOverlapping(byte[] s) {
}
@Override
- public PCollection<String> expand(PBegin input) {
- checkNotNull(getFilepattern(), "need to set the filepattern of a TextIO.Read transform");
+ public PCollection<RecordWithMetadata> expand(PBegin input) {
+ checkNotNull(
+ getFilepattern(), "need to set the filepattern of a ContextualTextIO.Read transform");
+ PCollection<RecordWithMetadata> lines = null;
if (getMatchConfiguration().getWatchInterval() == null && !getHintMatchesManyFiles()) {
- return input.apply("Read", org.apache.beam.sdk.io.Read.from(getSource()));
+ lines = input.apply("Read", org.apache.beam.sdk.io.Read.from(getSource()));
+ } else {
+ // All other cases go through FileIO + ReadFiles
+ lines =
+ input
+ .apply(
+ "Create filepattern", Create.ofProvider(getFilepattern(), StringUtf8Coder.of()))
+ .apply("Match All", FileIO.matchAll().withConfiguration(getMatchConfiguration()))
+ .apply(
+ "Read Matches",
+ FileIO.readMatches()
+ .withCompression(getCompression())
+ .withDirectoryTreatment(DirectoryTreatment.PROHIBIT))
+ .apply("Via ReadFiles", readFiles().withDelimiter(getDelimiter()));
}
- // All other cases go through FileIO + ReadFiles
- return input
- .apply("Create filepattern", Create.ofProvider(getFilepattern(), StringUtf8Coder.of()))
- .apply("Match All", FileIO.matchAll().withConfiguration(getMatchConfiguration()))
- .apply(
- "Read Matches",
- FileIO.readMatches()
- .withCompression(getCompression())
- .withDirectoryTreatment(DirectoryTreatment.PROHIBIT))
- .apply("Via ReadFiles", readFiles().withDelimiter(getDelimiter()));
+ // Check if the user decided to opt out of recordNums associated with records
+ if (getWithoutLineNumMetadata()) {
+ return lines;
+ }
+
+ // At this point the line number in RecordWithMetadata contains the relative line offset from
+ // the
+ // beginning of the read range.
+
+ // To compute the absolute position from the beginning of the input,
+ // we group the lines within the same ranges, and evaluate the size of each range.
+
+ PCollection<KV<KV<String, Long>, RecordWithMetadata>> linesGroupedByFileAndRange =
+ lines.apply("AddFileNameAndRange", ParDo.of(new AddFileNameAndRange()));
+
+ PCollectionView<Map<KV<String, Long>, Long>> rangeSizes =
+ linesGroupedByFileAndRange
+ .apply("CountLinesForEachFileRange", Count.perKey())
+ .apply("SizesAsView", View.asMap());
+
+ // Get Pipeline to create a dummy PCollection with one element to help compute the lines
+ // before each Range
+ PCollection<Integer> singletonPcoll =
+ input.getPipeline().apply("CreateSingletonPcoll", Create.of(Arrays.asList(1)));
+
+ // For each (File, Offset) pair, calculate the number of lines occurring before the Range for
+ // each File
+
+ // After computing the number of lines before each range, we can find the line number in
+ // original file as numLiesBeforeOffset + lineNumInCurrentOffset
+ PCollectionView<Map<KV<String, Long>, Long>> numLinesBeforeEachRange =
+ singletonPcoll
+ .apply(
+ "ComputeLinesBeforeRange",
+ ParDo.of(new ComputeLinesBeforeEachRange(rangeSizes)).withSideInputs(rangeSizes))
+ .apply("NumLinesBeforeEachRangeAsView", View.asMap());
+
+ return linesGroupedByFileAndRange.apply(
+ "AssignLineNums",
+ ParDo.of(new AssignLineNums(numLinesBeforeEachRange))
+ .withSideInputs(numLinesBeforeEachRange));
+ }
+
+ @VisibleForTesting
+ static class AddFileNameAndRange
+ extends DoFn<RecordWithMetadata, KV<KV<String, Long>, RecordWithMetadata>> {
+ @ProcessElement
+ public void processElement(
+ @Element RecordWithMetadata line,
+ OutputReceiver<KV<KV<String, Long>, RecordWithMetadata>> out) {
+ out.output(KV.of(KV.of(line.getFileName(), line.getRange().getRangeNum()), line));
+ }
+ }
+
+ /**
+ * Helper class for computing number of lines in the File preceding the beginning of the Range
+ * in this file.
+ */
+ @VisibleForTesting
+ static class ComputeLinesBeforeEachRange extends DoFn<Integer, KV<KV<String, Long>, Long>> {
+ private final PCollectionView<Map<KV<String, Long>, Long>> rangeSizes;
+
+ public ComputeLinesBeforeEachRange(PCollectionView<Map<KV<String, Long>, Long>> rangeSizes) {
+ this.rangeSizes = rangeSizes;
+ }
+
+ // Add custom comparator as KV<K, V> is not comparable by default
+ private static class FileRangeComparator<K extends Comparable<K>, V extends Comparable<V>>
+ implements Comparator<KV<K, V>> {
+ @Override
+ public int compare(KV<K, V> a, KV<K, V> b) {
+ if (a.getKey().compareTo(b.getKey()) == 0) {
+ return a.getValue().compareTo(b.getValue());
+ }
+ return a.getKey().compareTo(b.getKey());
+ }
+ }
+
+ @ProcessElement
+ public void processElement(ProcessContext p) {
+ // Get the Map Containing the size from side-input
+ Map<KV<String, Long>, Long> rangeSizesMap = p.sideInput(rangeSizes);
+
+ // The FileRange Pair must be sorted
+ SortedMap<KV<String, Long>, Long> sorted = new TreeMap<>(new FileRangeComparator<>());
+
+ // Initialize sorted map with values
+ for (Map.Entry<KV<String, Long>, Long> entry : rangeSizesMap.entrySet()) {
+ sorted.put(entry.getKey(), entry.getValue());
+ }
+
+ // HashMap that tracks lines passed for each file
+ Map<String, Long> pastLines = new HashMap<>();
+
+ // For each (File, Range) Pair, compute the number of lines before it
+ for (Map.Entry entry : sorted.entrySet()) {
+ Long lines = (long) entry.getValue();
Review comment:
(see also other `line`, `lines` mentions in this PR.)
##########
File path: sdks/java/io/contextual-text-io/src/main/java/org/apache/beam/sdk/io/contextualtextio/ContextualTextIOSource.java
##########
@@ -0,0 +1,363 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one
+ * or more contributor license agreements. See the NOTICE file
+ * distributed with this work for additional information
+ * regarding copyright ownership. The ASF licenses this file
+ * to you under the Apache License, Version 2.0 (the
+ * "License"); you may not use this file except in compliance
+ * with the License. You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+package org.apache.beam.sdk.io.contextualtextio;
+
+import java.io.IOException;
+import java.nio.ByteBuffer;
+import java.nio.channels.ReadableByteChannel;
+import java.nio.channels.SeekableByteChannel;
+import java.util.NoSuchElementException;
+import org.apache.beam.sdk.coders.Coder;
+import org.apache.beam.sdk.io.FileBasedSource;
+import org.apache.beam.sdk.io.fs.EmptyMatchTreatment;
+import org.apache.beam.sdk.io.fs.MatchResult;
+import org.apache.beam.sdk.options.PipelineOptions;
+import org.apache.beam.sdk.options.ValueProvider;
+import org.apache.beam.sdk.schemas.NoSuchSchemaException;
+import org.apache.beam.sdk.schemas.SchemaCoder;
+import org.apache.beam.sdk.schemas.SchemaRegistry;
+import org.apache.beam.vendor.grpc.v1p26p0.com.google.protobuf.ByteString;
+import org.apache.beam.vendor.guava.v26_0_jre.com.google.common.annotations.VisibleForTesting;
+import org.apache.beam.vendor.guava.v26_0_jre.com.google.common.base.Preconditions;
+import org.checkerframework.checker.nullness.qual.Nullable;
+
+/**
+ * Implementation detail of {@link ContextualTextIO.Read}.
+ *
+ * <p>A {@link FileBasedSource} which can decode records delimited by newline characters.
+ *
+ * <p>This source splits the data into records using {@code UTF-8} {@code \n}, {@code \r}, or {@code
+ * \r\n} as the delimiter. This source is not strict and supports decoding the last record even if
+ * it is not delimited. Finally, no records are decoded if the stream is empty.
+ *
+ * <p>This source supports reading from any arbitrary byte position within the stream. If the
+ * starting position is not {@code 0}, then bytes are skipped until the first delimiter is found
+ * representing the beginning of the first record to be decoded.
+ */
+@VisibleForTesting
+class ContextualTextIOSource extends FileBasedSource<RecordWithMetadata> {
+ byte[] delimiter;
+
+ // Used to Override isSplittable
+ private boolean hasMultilineCSVRecords;
+
+ @Override
+ protected boolean isSplittable() throws Exception {
+ if (hasMultilineCSVRecords) {
+ return false;
+ }
+ return super.isSplittable();
+ }
+
+ ContextualTextIOSource(
+ ValueProvider<String> fileSpec,
+ EmptyMatchTreatment emptyMatchTreatment,
+ byte[] delimiter,
+ boolean hasMultilineCSVRecords) {
+ super(fileSpec, emptyMatchTreatment, 1L);
+ this.delimiter = delimiter;
+ this.hasMultilineCSVRecords = hasMultilineCSVRecords;
+ }
+
+ private ContextualTextIOSource(
+ MatchResult.Metadata metadata,
+ long start,
+ long end,
+ byte[] delimiter,
+ boolean hasMultilineCSVRecords) {
+ super(metadata, 1L, start, end);
+ this.delimiter = delimiter;
+ this.hasMultilineCSVRecords = hasMultilineCSVRecords;
+ }
+
+ @Override
+ protected FileBasedSource<RecordWithMetadata> createForSubrangeOfFile(
+ MatchResult.Metadata metadata, long start, long end) {
+ return new ContextualTextIOSource(metadata, start, end, delimiter, hasMultilineCSVRecords);
+ }
+
+ @Override
+ protected FileBasedReader<RecordWithMetadata> createSingleFileReader(PipelineOptions options) {
+ return new MultiLineTextBasedReader(this, delimiter, hasMultilineCSVRecords);
+ }
+
+ @Override
+ public Coder<RecordWithMetadata> getOutputCoder() {
+ SchemaCoder<RecordWithMetadata> coder = null;
+ try {
+ coder = SchemaRegistry.createDefault().getSchemaCoder(RecordWithMetadata.class);
+ } catch (NoSuchSchemaException e) {
+ System.out.println("No Coder!");
+ }
+ return coder;
+ }
+
+ /**
+ * A {@link FileBasedReader FileBasedReader} which can decode records delimited by delimiter
+ * characters.
+ *
+ * <p>See {@link ContextualTextIOSource } for further details.
+ */
+ @VisibleForTesting
+ static class MultiLineTextBasedReader extends FileBasedReader<RecordWithMetadata> {
+ public static final int READ_BUFFER_SIZE = 8192;
+ private static final ByteString UTF8_BOM =
+ ByteString.copyFrom(new byte[] {(byte) 0xEF, (byte) 0xBB, (byte) 0xBF});
+ private final ByteBuffer readBuffer = ByteBuffer.allocate(READ_BUFFER_SIZE);
+ private ByteString buffer;
+ private int startOfDelimiterInBuffer;
+ private int endOfDelimiterInBuffer;
+ private long startOfRecord;
+ private volatile long startOfNextRecord;
+ private volatile boolean eof;
+ private volatile boolean elementIsPresent;
+ private @Nullable RecordWithMetadata currentValue;
+ private @Nullable ReadableByteChannel inChannel;
+ private byte @Nullable [] delimiter;
+
+ // Add to override the isSplittable
+ private boolean hasRFC4180MultiLineColumn;
+
+ private long startingOffset;
+ private long readerlineNum;
+
+ private MultiLineTextBasedReader(
+ ContextualTextIOSource source, byte[] delimiter, boolean hasRFC4180MultiLineColumn) {
+ super(source);
+ buffer = ByteString.EMPTY;
+ this.delimiter = delimiter;
+ this.hasRFC4180MultiLineColumn = hasRFC4180MultiLineColumn;
+ startingOffset = getCurrentSource().getStartOffset(); // Start offset;
+ }
+
+ @Override
+ protected long getCurrentOffset() throws NoSuchElementException {
+ if (!elementIsPresent) {
+ throw new NoSuchElementException();
+ }
+ return startOfRecord;
+ }
+
+ @Override
+ public long getSplitPointsRemaining() {
+ if (isStarted() && startOfNextRecord >= getCurrentSource().getEndOffset()) {
+ return isDone() ? 0 : 1;
+ }
+ return super.getSplitPointsRemaining();
+ }
+
+ @Override
+ public RecordWithMetadata getCurrent() throws NoSuchElementException {
+ if (!elementIsPresent) {
+ throw new NoSuchElementException();
+ }
+ return currentValue;
+ }
+
+ @Override
+ protected void startReading(ReadableByteChannel channel) throws IOException {
+ this.inChannel = channel;
+ // If the first offset is greater than zero, we need to skip bytes until we see our
+ // first delimiter.
+ long startOffset = getCurrentSource().getStartOffset();
+ if (startOffset > 0) {
+ Preconditions.checkState(
+ channel instanceof SeekableByteChannel,
+ "%s only supports reading from a SeekableByteChannel when given a start offset"
+ + " greater than 0.",
+ ContextualTextIOSource.class.getSimpleName());
+ long requiredPosition = startOffset - 1;
+ if (delimiter != null && startOffset >= delimiter.length) {
+ // we need to move back the offset of at worse delimiter.size to be sure to see
+ // all the bytes of the delimiter in the call to findDelimiterBounds() below
+ requiredPosition = startOffset - delimiter.length;
+ }
+ ((SeekableByteChannel) channel).position(requiredPosition);
+ findDelimiterBoundsWithMultiLineCheck();
+ buffer = buffer.substring(endOfDelimiterInBuffer);
+ startOfNextRecord = requiredPosition + endOfDelimiterInBuffer;
+ endOfDelimiterInBuffer = 0;
+ startOfDelimiterInBuffer = 0;
+ }
+ }
+
+ private void findDelimiterBoundsWithMultiLineCheck() throws IOException {
+ findDelimiterBounds();
+ }
+
+ /**
+ * Locates the start position and end position of the next delimiter. Will consume the channel
+ * till either EOF or the delimiter bounds are found.
+ *
+ * <p>If {@link ContextualTextIOSource#hasMultilineCSVRecords} is set then the behaviour will
+ * change from the standard read seen in {@link org.apache.beam.sdk.io.TextIO}. The assumption
+ * when {@link ContextualTextIOSource#hasMultilineCSVRecords} is set is that the file is being
+ * read with a single thread.
+ *
+ * <p>This fills the buffer and updates the positions as follows:
+ *
+ * <pre>{@code
+ * ------------------------------------------------------
+ * | element bytes | delimiter bytes | unconsumed bytes |
+ * ------------------------------------------------------
+ * 0 start of end of buffer
+ * delimiter delimiter size
+ * in buffer in buffer
+ * }</pre>
+ */
+ private void findDelimiterBounds() throws IOException {
+ int bytePositionInBuffer = 0;
+ boolean doubleQuoteClosed = true;
+
+ while (true) {
+ if (!tryToEnsureNumberOfBytesInBuffer(bytePositionInBuffer + 1)) {
+ startOfDelimiterInBuffer = endOfDelimiterInBuffer = bytePositionInBuffer;
+ break;
+ }
+
+ byte currentByte = buffer.byteAt(bytePositionInBuffer);
+ if (hasRFC4180MultiLineColumn) {
+ // Check if we are inside an open Quote
+ if (currentByte == '"') {
+ doubleQuoteClosed = !doubleQuoteClosed;
+ }
+ } else {
+ doubleQuoteClosed = true;
+ }
+
+ if (delimiter == null) {
+ // default delimiter
+ if (currentByte == '\n') {
+ startOfDelimiterInBuffer = bytePositionInBuffer;
+ endOfDelimiterInBuffer = startOfDelimiterInBuffer + 1;
+ if (doubleQuoteClosed) {
+ break;
+ }
+ } else if (currentByte == '\r') {
+ startOfDelimiterInBuffer = bytePositionInBuffer;
+ endOfDelimiterInBuffer = startOfDelimiterInBuffer + 1;
+ if (tryToEnsureNumberOfBytesInBuffer(bytePositionInBuffer + 2)) {
+ currentByte = buffer.byteAt(bytePositionInBuffer + 1);
+ if (currentByte == '\n') {
+ endOfDelimiterInBuffer += 1;
+ }
+ }
+ if (doubleQuoteClosed) {
+ break;
+ }
+ }
+ } else {
+ // when the user defines a delimiter
+ int i = 0;
+ startOfDelimiterInBuffer = endOfDelimiterInBuffer = bytePositionInBuffer;
+ while ((i < delimiter.length) && (currentByte == delimiter[i])) {
+ // read next byte;
+ i++;
+ if (tryToEnsureNumberOfBytesInBuffer(bytePositionInBuffer + i + 1)) {
+ currentByte = buffer.byteAt(bytePositionInBuffer + i);
+ } else {
+ // corner case: delimiter truncate at the end of file
+ startOfDelimiterInBuffer = endOfDelimiterInBuffer = bytePositionInBuffer;
+ break;
+ }
+ }
+ if (i == delimiter.length) {
+ endOfDelimiterInBuffer = bytePositionInBuffer + i;
+ if (doubleQuoteClosed) {
+ break;
+ }
+ }
+ }
+ bytePositionInBuffer += 1;
+ }
+ }
+
+ @Override
+ protected boolean readNextRecord() throws IOException {
+ startOfRecord = startOfNextRecord;
+
+ findDelimiterBoundsWithMultiLineCheck();
+
+ // If we have reached EOF file and consumed all of the buffer then we know
+ // that there are no more records.
+ if (eof && buffer.isEmpty()) {
+ elementIsPresent = false;
+ return false;
+ }
+
+ decodeCurrentElement();
+ startOfNextRecord = startOfRecord + endOfDelimiterInBuffer;
+ return true;
+ }
+
+ /**
+ * Decodes the current element updating the buffer to only contain the unconsumed bytes.
+ *
+ * <p>This invalidates the currently stored {@code startOfDelimiterInBuffer} and {@code
+ * endOfDelimiterInBuffer}.
+ */
+ private void decodeCurrentElement() throws IOException {
+ ByteString dataToDecode = buffer.substring(0, startOfDelimiterInBuffer);
+ // If present, the UTF8 Byte Order Mark (BOM) will be removed.
+ if (startOfRecord == 0 && dataToDecode.startsWith(UTF8_BOM)) {
+ dataToDecode = dataToDecode.substring(UTF8_BOM.size());
+ }
+
+ /////////////////////////////////////////////
+
+ // Data of the Current Line
+ // dataToDecode.toStringUtf8();
+
+ // The line num is:
+ Long lineUniqueLineNum = readerlineNum++;
+ // The Complete FileName (with uri if this is a web url eg: temp/abc.txt) is:
+ String fileName = getCurrentSource().getSingleFileMetadata().resourceId().toString();
Review comment:
Hm, interesting point. I don't know to which extent size is a concern.
Would the user need to join Uri + FileName and have to worry to know the correct separator for the two?
Perhaps worth considering something along the lines of Filename and (optionally) FullFilename?
##########
File path: sdks/java/io/contextual-text-io/src/main/java/org/apache/beam/sdk/io/contextualtextio/Range.java
##########
@@ -0,0 +1,42 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one
+ * or more contributor license agreements. See the NOTICE file
+ * distributed with this work for additional information
+ * regarding copyright ownership. The ASF licenses this file
+ * to you under the Apache License, Version 2.0 (the
+ * "License"); you may not use this file except in compliance
+ * with the License. You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+package org.apache.beam.sdk.io.contextualtextio;
+
+import com.google.auto.value.AutoValue;
+
+@AutoValue
+public abstract class Range {
+ public abstract Long getRangeNum();
Review comment:
What is this? starting offset?
##########
File path: sdks/java/io/contextual-text-io/src/main/java/org/apache/beam/sdk/io/contextualtextio/ContextualTextIOSource.java
##########
@@ -259,7 +317,31 @@ private void decodeCurrentElement() throws IOException {
if (startOfRecord == 0 && dataToDecode.startsWith(UTF8_BOM)) {
dataToDecode = dataToDecode.substring(UTF8_BOM.size());
}
- currentValue = dataToDecode.toStringUtf8();
+
+ /////////////////////////////////////////////
+
+ // Data of the Current Line
+ // dataToDecode.toStringUtf8();
+
+ // The line num is:
+ Long lineUniqueLineNum = readerlineNum++;
Review comment:
the naming is confusing here.
`recordNum = totalRecordCount++` perhaps?
##########
File path: sdks/java/core/src/main/java/org/apache/beam/sdk/io/FileBasedSource.java
##########
@@ -130,7 +130,7 @@ protected FileBasedSource(
*
* @throws IllegalArgumentException if this source is in {@link Mode#FILEPATTERN} mode.
*/
- protected final MatchResult.Metadata getSingleFileMetadata() {
+ public final MatchResult.Metadata getSingleFileMetadata() {
Review comment:
SGTM, different commit should be sufficient.
##########
File path: sdks/java/io/contextual-text-io/src/main/java/org/apache/beam/sdk/io/contextualtextio/RecordWithMetadata.java
##########
@@ -0,0 +1,57 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one
+ * or more contributor license agreements. See the NOTICE file
+ * distributed with this work for additional information
+ * regarding copyright ownership. The ASF licenses this file
+ * to you under the Apache License, Version 2.0 (the
+ * "License"); you may not use this file except in compliance
+ * with the License. You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+package org.apache.beam.sdk.io.contextualtextio;
+
+import com.google.auto.value.AutoValue;
+import org.apache.beam.sdk.annotations.Experimental;
+import org.apache.beam.sdk.annotations.Internal;
+import org.apache.beam.sdk.schemas.AutoValueSchema;
+import org.apache.beam.sdk.schemas.annotations.DefaultSchema;
+
+@Internal
+@Experimental(Experimental.Kind.SCHEMAS)
+@DefaultSchema(AutoValueSchema.class)
+@AutoValue
+public abstract class RecordWithMetadata {
+ public abstract Range getRange();
Review comment:
Should this be user-visible (will it be useful?)
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
[GitHub] [beam] abhiy13 commented on a change in pull request #12645: [BEAM-10124] Add ContextualTextIO
Posted by GitBox <gi...@apache.org>.
abhiy13 commented on a change in pull request #12645:
URL: https://github.com/apache/beam/pull/12645#discussion_r474444178
##########
File path: sdks/java/io/contextual-text-io/src/main/java/org/apache/beam/sdk/io/contextualtextio/ContextualTextIOSource.java
##########
@@ -0,0 +1,363 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one
+ * or more contributor license agreements. See the NOTICE file
+ * distributed with this work for additional information
+ * regarding copyright ownership. The ASF licenses this file
+ * to you under the Apache License, Version 2.0 (the
+ * "License"); you may not use this file except in compliance
+ * with the License. You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+package org.apache.beam.sdk.io.contextualtextio;
+
+import java.io.IOException;
+import java.nio.ByteBuffer;
+import java.nio.channels.ReadableByteChannel;
+import java.nio.channels.SeekableByteChannel;
+import java.util.NoSuchElementException;
+import org.apache.beam.sdk.coders.Coder;
+import org.apache.beam.sdk.io.FileBasedSource;
+import org.apache.beam.sdk.io.fs.EmptyMatchTreatment;
+import org.apache.beam.sdk.io.fs.MatchResult;
+import org.apache.beam.sdk.options.PipelineOptions;
+import org.apache.beam.sdk.options.ValueProvider;
+import org.apache.beam.sdk.schemas.NoSuchSchemaException;
+import org.apache.beam.sdk.schemas.SchemaCoder;
+import org.apache.beam.sdk.schemas.SchemaRegistry;
+import org.apache.beam.vendor.grpc.v1p26p0.com.google.protobuf.ByteString;
+import org.apache.beam.vendor.guava.v26_0_jre.com.google.common.annotations.VisibleForTesting;
+import org.apache.beam.vendor.guava.v26_0_jre.com.google.common.base.Preconditions;
+import org.checkerframework.checker.nullness.qual.Nullable;
+
+/**
+ * Implementation detail of {@link ContextualTextIO.Read}.
+ *
+ * <p>A {@link FileBasedSource} which can decode records delimited by newline characters.
+ *
+ * <p>This source splits the data into records using {@code UTF-8} {@code \n}, {@code \r}, or {@code
+ * \r\n} as the delimiter. This source is not strict and supports decoding the last record even if
+ * it is not delimited. Finally, no records are decoded if the stream is empty.
+ *
+ * <p>This source supports reading from any arbitrary byte position within the stream. If the
+ * starting position is not {@code 0}, then bytes are skipped until the first delimiter is found
+ * representing the beginning of the first record to be decoded.
+ */
+@VisibleForTesting
+class ContextualTextIOSource extends FileBasedSource<RecordWithMetadata> {
+ byte[] delimiter;
+
+ // Used to Override isSplittable
+ private boolean hasMultilineCSVRecords;
+
+ @Override
+ protected boolean isSplittable() throws Exception {
+ if (hasMultilineCSVRecords) {
Review comment:
Ack.
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
[GitHub] [beam] rezarokni commented on a change in pull request #12645: [BEAM-10124] Add ContextualTextIO
Posted by GitBox <gi...@apache.org>.
rezarokni commented on a change in pull request #12645:
URL: https://github.com/apache/beam/pull/12645#discussion_r474359464
##########
File path: sdks/java/io/contextual-text-io/src/main/java/org/apache/beam/sdk/io/contextualtextio/ContextualTextIOSource.java
##########
@@ -0,0 +1,363 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one
+ * or more contributor license agreements. See the NOTICE file
+ * distributed with this work for additional information
+ * regarding copyright ownership. The ASF licenses this file
+ * to you under the Apache License, Version 2.0 (the
+ * "License"); you may not use this file except in compliance
+ * with the License. You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+package org.apache.beam.sdk.io.contextualtextio;
+
+import java.io.IOException;
+import java.nio.ByteBuffer;
+import java.nio.channels.ReadableByteChannel;
+import java.nio.channels.SeekableByteChannel;
+import java.util.NoSuchElementException;
+import org.apache.beam.sdk.coders.Coder;
+import org.apache.beam.sdk.io.FileBasedSource;
+import org.apache.beam.sdk.io.fs.EmptyMatchTreatment;
+import org.apache.beam.sdk.io.fs.MatchResult;
+import org.apache.beam.sdk.options.PipelineOptions;
+import org.apache.beam.sdk.options.ValueProvider;
+import org.apache.beam.sdk.schemas.NoSuchSchemaException;
+import org.apache.beam.sdk.schemas.SchemaCoder;
+import org.apache.beam.sdk.schemas.SchemaRegistry;
+import org.apache.beam.vendor.grpc.v1p26p0.com.google.protobuf.ByteString;
+import org.apache.beam.vendor.guava.v26_0_jre.com.google.common.annotations.VisibleForTesting;
+import org.apache.beam.vendor.guava.v26_0_jre.com.google.common.base.Preconditions;
+import org.checkerframework.checker.nullness.qual.Nullable;
+
+/**
+ * Implementation detail of {@link ContextualTextIO.Read}.
+ *
+ * <p>A {@link FileBasedSource} which can decode records delimited by newline characters.
+ *
+ * <p>This source splits the data into records using {@code UTF-8} {@code \n}, {@code \r}, or {@code
+ * \r\n} as the delimiter. This source is not strict and supports decoding the last record even if
+ * it is not delimited. Finally, no records are decoded if the stream is empty.
+ *
+ * <p>This source supports reading from any arbitrary byte position within the stream. If the
+ * starting position is not {@code 0}, then bytes are skipped until the first delimiter is found
+ * representing the beginning of the first record to be decoded.
+ */
+@VisibleForTesting
+class ContextualTextIOSource extends FileBasedSource<RecordWithMetadata> {
+ byte[] delimiter;
+
+ // Used to Override isSplittable
+ private boolean hasMultilineCSVRecords;
+
+ @Override
+ protected boolean isSplittable() throws Exception {
+ if (hasMultilineCSVRecords) {
+ return false;
+ }
+ return super.isSplittable();
+ }
+
+ ContextualTextIOSource(
+ ValueProvider<String> fileSpec,
+ EmptyMatchTreatment emptyMatchTreatment,
+ byte[] delimiter,
+ boolean hasMultilineCSVRecords) {
+ super(fileSpec, emptyMatchTreatment, 1L);
+ this.delimiter = delimiter;
+ this.hasMultilineCSVRecords = hasMultilineCSVRecords;
+ }
+
+ private ContextualTextIOSource(
+ MatchResult.Metadata metadata,
+ long start,
+ long end,
+ byte[] delimiter,
+ boolean hasMultilineCSVRecords) {
+ super(metadata, 1L, start, end);
+ this.delimiter = delimiter;
+ this.hasMultilineCSVRecords = hasMultilineCSVRecords;
+ }
+
+ @Override
+ protected FileBasedSource<RecordWithMetadata> createForSubrangeOfFile(
+ MatchResult.Metadata metadata, long start, long end) {
+ return new ContextualTextIOSource(metadata, start, end, delimiter, hasMultilineCSVRecords);
+ }
+
+ @Override
+ protected FileBasedReader<RecordWithMetadata> createSingleFileReader(PipelineOptions options) {
+ return new MultiLineTextBasedReader(this, delimiter, hasMultilineCSVRecords);
+ }
+
+ @Override
+ public Coder<RecordWithMetadata> getOutputCoder() {
+ SchemaCoder<RecordWithMetadata> coder = null;
+ try {
+ coder = SchemaRegistry.createDefault().getSchemaCoder(RecordWithMetadata.class);
+ } catch (NoSuchSchemaException e) {
+ System.out.println("No Coder!");
+ }
+ return coder;
+ }
+
+ /**
+ * A {@link FileBasedReader FileBasedReader} which can decode records delimited by delimiter
+ * characters.
+ *
+ * <p>See {@link ContextualTextIOSource } for further details.
+ */
+ @VisibleForTesting
+ static class MultiLineTextBasedReader extends FileBasedReader<RecordWithMetadata> {
+ public static final int READ_BUFFER_SIZE = 8192;
+ private static final ByteString UTF8_BOM =
+ ByteString.copyFrom(new byte[] {(byte) 0xEF, (byte) 0xBB, (byte) 0xBF});
+ private final ByteBuffer readBuffer = ByteBuffer.allocate(READ_BUFFER_SIZE);
+ private ByteString buffer;
+ private int startOfDelimiterInBuffer;
+ private int endOfDelimiterInBuffer;
+ private long startOfRecord;
+ private volatile long startOfNextRecord;
+ private volatile boolean eof;
+ private volatile boolean elementIsPresent;
+ private @Nullable RecordWithMetadata currentValue;
+ private @Nullable ReadableByteChannel inChannel;
+ private byte @Nullable [] delimiter;
+
+ // Add to override the isSplittable
+ private boolean hasRFC4180MultiLineColumn;
+
+ private long startingOffset;
+ private long readerlineNum;
+
+ private MultiLineTextBasedReader(
+ ContextualTextIOSource source, byte[] delimiter, boolean hasRFC4180MultiLineColumn) {
+ super(source);
+ buffer = ByteString.EMPTY;
+ this.delimiter = delimiter;
+ this.hasRFC4180MultiLineColumn = hasRFC4180MultiLineColumn;
+ startingOffset = getCurrentSource().getStartOffset(); // Start offset;
+ }
+
+ @Override
+ protected long getCurrentOffset() throws NoSuchElementException {
+ if (!elementIsPresent) {
+ throw new NoSuchElementException();
+ }
+ return startOfRecord;
+ }
+
+ @Override
+ public long getSplitPointsRemaining() {
+ if (isStarted() && startOfNextRecord >= getCurrentSource().getEndOffset()) {
+ return isDone() ? 0 : 1;
+ }
+ return super.getSplitPointsRemaining();
+ }
+
+ @Override
+ public RecordWithMetadata getCurrent() throws NoSuchElementException {
+ if (!elementIsPresent) {
+ throw new NoSuchElementException();
+ }
+ return currentValue;
+ }
+
+ @Override
+ protected void startReading(ReadableByteChannel channel) throws IOException {
+ this.inChannel = channel;
+ // If the first offset is greater than zero, we need to skip bytes until we see our
+ // first delimiter.
+ long startOffset = getCurrentSource().getStartOffset();
+ if (startOffset > 0) {
+ Preconditions.checkState(
+ channel instanceof SeekableByteChannel,
+ "%s only supports reading from a SeekableByteChannel when given a start offset"
+ + " greater than 0.",
+ ContextualTextIOSource.class.getSimpleName());
+ long requiredPosition = startOffset - 1;
+ if (delimiter != null && startOffset >= delimiter.length) {
+ // we need to move back the offset of at worse delimiter.size to be sure to see
+ // all the bytes of the delimiter in the call to findDelimiterBounds() below
+ requiredPosition = startOffset - delimiter.length;
+ }
+ ((SeekableByteChannel) channel).position(requiredPosition);
+ findDelimiterBoundsWithMultiLineCheck();
+ buffer = buffer.substring(endOfDelimiterInBuffer);
+ startOfNextRecord = requiredPosition + endOfDelimiterInBuffer;
+ endOfDelimiterInBuffer = 0;
+ startOfDelimiterInBuffer = 0;
+ }
+ }
+
+ private void findDelimiterBoundsWithMultiLineCheck() throws IOException {
+ findDelimiterBounds();
+ }
+
+ /**
+ * Locates the start position and end position of the next delimiter. Will consume the channel
+ * till either EOF or the delimiter bounds are found.
+ *
+ * <p>If {@link ContextualTextIOSource#hasMultilineCSVRecords} is set then the behaviour will
+ * change from the standard read seen in {@link org.apache.beam.sdk.io.TextIO}. The assumption
+ * when {@link ContextualTextIOSource#hasMultilineCSVRecords} is set is that the file is being
+ * read with a single thread.
+ *
+ * <p>This fills the buffer and updates the positions as follows:
+ *
+ * <pre>{@code
+ * ------------------------------------------------------
+ * | element bytes | delimiter bytes | unconsumed bytes |
+ * ------------------------------------------------------
+ * 0 start of end of buffer
+ * delimiter delimiter size
+ * in buffer in buffer
+ * }</pre>
+ */
+ private void findDelimiterBounds() throws IOException {
+ int bytePositionInBuffer = 0;
+ boolean doubleQuoteClosed = true;
+
+ while (true) {
+ if (!tryToEnsureNumberOfBytesInBuffer(bytePositionInBuffer + 1)) {
+ startOfDelimiterInBuffer = endOfDelimiterInBuffer = bytePositionInBuffer;
+ break;
+ }
+
+ byte currentByte = buffer.byteAt(bytePositionInBuffer);
+ if (hasRFC4180MultiLineColumn) {
+ // Check if we are inside an open Quote
+ if (currentByte == '"') {
+ doubleQuoteClosed = !doubleQuoteClosed;
+ }
+ } else {
+ doubleQuoteClosed = true;
+ }
+
+ if (delimiter == null) {
+ // default delimiter
+ if (currentByte == '\n') {
+ startOfDelimiterInBuffer = bytePositionInBuffer;
+ endOfDelimiterInBuffer = startOfDelimiterInBuffer + 1;
+ if (doubleQuoteClosed) {
+ break;
+ }
+ } else if (currentByte == '\r') {
+ startOfDelimiterInBuffer = bytePositionInBuffer;
+ endOfDelimiterInBuffer = startOfDelimiterInBuffer + 1;
+ if (tryToEnsureNumberOfBytesInBuffer(bytePositionInBuffer + 2)) {
+ currentByte = buffer.byteAt(bytePositionInBuffer + 1);
+ if (currentByte == '\n') {
+ endOfDelimiterInBuffer += 1;
+ }
+ }
+ if (doubleQuoteClosed) {
+ break;
+ }
+ }
+ } else {
+ // when the user defines a delimiter
+ int i = 0;
+ startOfDelimiterInBuffer = endOfDelimiterInBuffer = bytePositionInBuffer;
+ while ((i < delimiter.length) && (currentByte == delimiter[i])) {
+ // read next byte;
+ i++;
+ if (tryToEnsureNumberOfBytesInBuffer(bytePositionInBuffer + i + 1)) {
+ currentByte = buffer.byteAt(bytePositionInBuffer + i);
+ } else {
+ // corner case: delimiter truncate at the end of file
+ startOfDelimiterInBuffer = endOfDelimiterInBuffer = bytePositionInBuffer;
+ break;
+ }
+ }
+ if (i == delimiter.length) {
+ endOfDelimiterInBuffer = bytePositionInBuffer + i;
+ if (doubleQuoteClosed) {
+ break;
+ }
+ }
+ }
+ bytePositionInBuffer += 1;
+ }
+ }
+
+ @Override
+ protected boolean readNextRecord() throws IOException {
+ startOfRecord = startOfNextRecord;
+
+ findDelimiterBoundsWithMultiLineCheck();
+
+ // If we have reached EOF file and consumed all of the buffer then we know
+ // that there are no more records.
+ if (eof && buffer.isEmpty()) {
+ elementIsPresent = false;
+ return false;
+ }
+
+ decodeCurrentElement();
+ startOfNextRecord = startOfRecord + endOfDelimiterInBuffer;
+ return true;
+ }
+
+ /**
+ * Decodes the current element updating the buffer to only contain the unconsumed bytes.
+ *
+ * <p>This invalidates the currently stored {@code startOfDelimiterInBuffer} and {@code
+ * endOfDelimiterInBuffer}.
+ */
+ private void decodeCurrentElement() throws IOException {
+ ByteString dataToDecode = buffer.substring(0, startOfDelimiterInBuffer);
+ // If present, the UTF8 Byte Order Mark (BOM) will be removed.
+ if (startOfRecord == 0 && dataToDecode.startsWith(UTF8_BOM)) {
+ dataToDecode = dataToDecode.substring(UTF8_BOM.size());
+ }
+
+ /////////////////////////////////////////////
+
+ // Data of the Current Line
+ // dataToDecode.toStringUtf8();
+
+ // The line num is:
+ Long lineUniqueLineNum = readerlineNum++;
+ // The Complete FileName (with uri if this is a web url eg: temp/abc.txt) is:
+ String fileName = getCurrentSource().getSingleFileMetadata().resourceId().toString();
Review comment:
+1 on (optionally) Full URI.
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
[GitHub] [beam] codecov[bot] edited a comment on pull request #12645: [BEAM-10124] Add ContextualTextIO
Posted by GitBox <gi...@apache.org>.
codecov[bot] edited a comment on pull request #12645:
URL: https://github.com/apache/beam/pull/12645#issuecomment-688630083
# [Codecov](https://codecov.io/gh/apache/beam/pull/12645?src=pr&el=h1) Report
> :exclamation: No coverage uploaded for pull request base (`master@2b2b8e7`). [Click here to learn what that means](https://docs.codecov.io/docs/error-reference#section-missing-base-commit).
> The diff coverage is `n/a`.
[](https://codecov.io/gh/apache/beam/pull/12645?src=pr&el=tree)
```diff
@@ Coverage Diff @@
## master #12645 +/- ##
=========================================
Coverage ? 82.33%
=========================================
Files ? 453
Lines ? 54054
Branches ? 0
=========================================
Hits ? 44506
Misses ? 9548
Partials ? 0
```
| [Impacted Files](https://codecov.io/gh/apache/beam/pull/12645?src=pr&el=tree) | Coverage Δ | |
|---|---|---|
| [...hon/apache\_beam/examples/wordcount\_with\_metrics.py](https://codecov.io/gh/apache/beam/pull/12645/diff?src=pr&el=tree#diff-c2Rrcy9weXRob24vYXBhY2hlX2JlYW0vZXhhbXBsZXMvd29yZGNvdW50X3dpdGhfbWV0cmljcy5weQ==) | `28.12% <0.00%> (ø)` | |
| [...dks/python/apache\_beam/transforms/create\_source.py](https://codecov.io/gh/apache/beam/pull/12645/diff?src=pr&el=tree#diff-c2Rrcy9weXRob24vYXBhY2hlX2JlYW0vdHJhbnNmb3Jtcy9jcmVhdGVfc291cmNlLnB5) | `98.18% <0.00%> (ø)` | |
| [sdks/python/apache\_beam/internal/pickler.py](https://codecov.io/gh/apache/beam/pull/12645/diff?src=pr&el=tree#diff-c2Rrcy9weXRob24vYXBhY2hlX2JlYW0vaW50ZXJuYWwvcGlja2xlci5weQ==) | `85.82% <0.00%> (ø)` | |
| [.../apache\_beam/examples/cookbook/group\_with\_coder.py](https://codecov.io/gh/apache/beam/pull/12645/diff?src=pr&el=tree#diff-c2Rrcy9weXRob24vYXBhY2hlX2JlYW0vZXhhbXBsZXMvY29va2Jvb2svZ3JvdXBfd2l0aF9jb2Rlci5weQ==) | `84.44% <0.00%> (ø)` | |
| [...s/python/apache\_beam/testing/pipeline\_verifiers.py](https://codecov.io/gh/apache/beam/pull/12645/diff?src=pr&el=tree#diff-c2Rrcy9weXRob24vYXBhY2hlX2JlYW0vdGVzdGluZy9waXBlbGluZV92ZXJpZmllcnMucHk=) | `91.80% <0.00%> (ø)` | |
| [sdks/python/apache\_beam/examples/snippets/util.py](https://codecov.io/gh/apache/beam/pull/12645/diff?src=pr&el=tree#diff-c2Rrcy9weXRob24vYXBhY2hlX2JlYW0vZXhhbXBsZXMvc25pcHBldHMvdXRpbC5weQ==) | `100.00% <0.00%> (ø)` | |
| [...ython/apache\_beam/runners/direct/direct\_metrics.py](https://codecov.io/gh/apache/beam/pull/12645/diff?src=pr&el=tree#diff-c2Rrcy9weXRob24vYXBhY2hlX2JlYW0vcnVubmVycy9kaXJlY3QvZGlyZWN0X21ldHJpY3MucHk=) | `98.38% <0.00%> (ø)` | |
| [...ks/python/apache\_beam/io/gcp/datastore/\_\_init\_\_.py](https://codecov.io/gh/apache/beam/pull/12645/diff?src=pr&el=tree#diff-c2Rrcy9weXRob24vYXBhY2hlX2JlYW0vaW8vZ2NwL2RhdGFzdG9yZS9fX2luaXRfXy5weQ==) | `100.00% <0.00%> (ø)` | |
| [sdks/python/apache\_beam/io/external/kafka.py](https://codecov.io/gh/apache/beam/pull/12645/diff?src=pr&el=tree#diff-c2Rrcy9weXRob24vYXBhY2hlX2JlYW0vaW8vZXh0ZXJuYWwva2Fma2EucHk=) | `0.00% <0.00%> (ø)` | |
| [...apache\_beam/examples/cookbook/custom\_ptransform.py](https://codecov.io/gh/apache/beam/pull/12645/diff?src=pr&el=tree#diff-c2Rrcy9weXRob24vYXBhY2hlX2JlYW0vZXhhbXBsZXMvY29va2Jvb2svY3VzdG9tX3B0cmFuc2Zvcm0ucHk=) | `53.48% <0.00%> (ø)` | |
| ... and [443 more](https://codecov.io/gh/apache/beam/pull/12645/diff?src=pr&el=tree-more) | |
------
[Continue to review full report at Codecov](https://codecov.io/gh/apache/beam/pull/12645?src=pr&el=continue).
> **Legend** - [Click here to learn more](https://docs.codecov.io/docs/codecov-delta)
> `Δ = absolute <relative> (impact)`, `ø = not affected`, `? = missing data`
> Powered by [Codecov](https://codecov.io/gh/apache/beam/pull/12645?src=pr&el=footer). Last update [2b2b8e7...85c579f](https://codecov.io/gh/apache/beam/pull/12645?src=pr&el=lastupdated). Read the [comment docs](https://docs.codecov.io/docs/pull-request-comments).
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
[GitHub] [beam] abhiy13 commented on a change in pull request #12645: [BEAM-10124] Add ContextualTextIO
Posted by GitBox <gi...@apache.org>.
abhiy13 commented on a change in pull request #12645:
URL: https://github.com/apache/beam/pull/12645#discussion_r492475183
##########
File path: sdks/java/io/contextual-text-io/src/main/java/org/apache/beam/sdk/io/contextualtextio/RecordWithMetadata.java
##########
@@ -0,0 +1,84 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one
+ * or more contributor license agreements. See the NOTICE file
+ * distributed with this work for additional information
+ * regarding copyright ownership. The ASF licenses this file
+ * to you under the Apache License, Version 2.0 (the
+ * "License"); you may not use this file except in compliance
+ * with the License. You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+package org.apache.beam.sdk.io.contextualtextio;
+
+import com.google.auto.value.AutoValue;
+import org.apache.beam.sdk.annotations.Experimental;
+import org.apache.beam.sdk.schemas.AutoValueSchema;
+import org.apache.beam.sdk.schemas.annotations.DefaultSchema;
+
+/**
+ * Helper Class based on {@link AutoValueSchema}, it provides Metadata associated with each Record
+ * when reading from file(s) using {@link ContextualTextIO}.
+ *
+ * <h3>Fields:</h3>
+ *
+ * <ul>
+ * <li>recordOffset: The offset of a record (the byte at which the record begins) in a file. This
+ * information can be useful if you wish to reconstruct the file. {@link
+ * RecordWithMetadata#getRecordOffset()}
+ * <li>recordNum: The ordinal number of the record in its file. {@link
+ * RecordWithMetadata#getRecordNum()}
+ * <li>recordValue: The value / contents of the record {@link RecordWithMetadata#getRecordValue()}
+ * <li>rangeOffset: The starting offset of the range (split), which contained the record, when the
+ * record was read. {@link RecordWithMetadata#getRangeOffset()}
+ * <li>recordNumInOffset: The record number relative to the Range. (line number within the range)
+ * {@link RecordWithMetadata#getRecordNumInOffset()}
+ * <li>fileName: Name of the file to which the record belongs (this is the full filename,
+ * eg:path/to/file.txt) {@link RecordWithMetadata#getFileName()}
+ * </ul>
+ */
+@Experimental(Experimental.Kind.SCHEMAS)
+@DefaultSchema(AutoValueSchema.class)
+@AutoValue
+public abstract class RecordWithMetadata {
+ public abstract Long getRecordOffset();
Review comment:
Done
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
[GitHub] [beam] rezarokni commented on a change in pull request #12645: [BEAM-10124] Add ContextualTextIO
Posted by GitBox <gi...@apache.org>.
rezarokni commented on a change in pull request #12645:
URL: https://github.com/apache/beam/pull/12645#discussion_r487934071
##########
File path: sdks/java/io/contextual-text-io/src/main/java/org/apache/beam/sdk/io/contextualtextio/ContextualTextIO.java
##########
@@ -0,0 +1,631 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one
+ * or more contributor license agreements. See the NOTICE file
+ * distributed with this work for additional information
+ * regarding copyright ownership. The ASF licenses this file
+ * to you under the Apache License, Version 2.0 (the
+ * "License"); you may not use this file except in compliance
+ * with the License. You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+package org.apache.beam.sdk.io.contextualtextio;
+
+import static org.apache.beam.sdk.io.FileIO.ReadMatches.DirectoryTreatment;
+import static org.apache.beam.vendor.guava.v26_0_jre.com.google.common.base.Preconditions.checkArgument;
+import static org.apache.beam.vendor.guava.v26_0_jre.com.google.common.base.Preconditions.checkNotNull;
+
+import com.google.auto.value.AutoValue;
+import java.nio.ByteBuffer;
+import java.util.Arrays;
+import java.util.Comparator;
+import java.util.HashMap;
+import java.util.Map;
+import java.util.SortedMap;
+import java.util.TreeMap;
+import org.apache.beam.sdk.coders.StringUtf8Coder;
+import org.apache.beam.sdk.io.CompressedSource;
+import org.apache.beam.sdk.io.Compression;
+import org.apache.beam.sdk.io.FileBasedSource;
+import org.apache.beam.sdk.io.FileIO;
+import org.apache.beam.sdk.io.FileIO.MatchConfiguration;
+import org.apache.beam.sdk.io.ReadAllViaFileBasedSource;
+import org.apache.beam.sdk.io.TextIO;
+import org.apache.beam.sdk.io.fs.EmptyMatchTreatment;
+import org.apache.beam.sdk.options.ValueProvider;
+import org.apache.beam.sdk.options.ValueProvider.StaticValueProvider;
+import org.apache.beam.sdk.schemas.NoSuchSchemaException;
+import org.apache.beam.sdk.schemas.SchemaCoder;
+import org.apache.beam.sdk.transforms.Count;
+import org.apache.beam.sdk.transforms.Create;
+import org.apache.beam.sdk.transforms.DoFn;
+import org.apache.beam.sdk.transforms.PTransform;
+import org.apache.beam.sdk.transforms.ParDo;
+import org.apache.beam.sdk.transforms.SerializableFunction;
+import org.apache.beam.sdk.transforms.View;
+import org.apache.beam.sdk.transforms.Watch.Growth.TerminationCondition;
+import org.apache.beam.sdk.transforms.display.DisplayData;
+import org.apache.beam.sdk.values.KV;
+import org.apache.beam.sdk.values.PBegin;
+import org.apache.beam.sdk.values.PCollection;
+import org.apache.beam.sdk.values.PCollectionView;
+import org.apache.beam.vendor.guava.v26_0_jre.com.google.common.annotations.VisibleForTesting;
+import org.checkerframework.checker.nullness.qual.Nullable;
+import org.joda.time.Duration;
+import org.slf4j.Logger;
+import org.slf4j.LoggerFactory;
+
+/**
+ * {@link PTransform}s that read text files and collect contextual information of the elements in
+ * the input.
+ *
+ * <p>Use {@link TextIO} when not reading file with Multiline Records or additional metadata is not
+ * required.
+ *
+ * <h2>Reading from text files</h2>
+ *
+ * <p>To read a {@link PCollection} from one or more text files, use {@code
+ * ContextualTextIO.read()}. To instantiate a transform use {@link
+ * ContextualTextIO.Read#from(String)} and specify the path of the file(s) to be read.
+ * Alternatively, if the filenames to be read are themselves in a {@link PCollection} you can use
+ * {@link FileIO} to match them and {@link ContextualTextIO#readFiles()} to read them.
+ *
+ * <p>{@link #read} returns a {@link PCollection} of {@link RecordWithMetadata RecordWithMetadata},
+ * each corresponding to one line of an input UTF-8 text file (split into lines delimited by '\n',
+ * '\r', '\r\n', or specified delimiter see {@link ContextualTextIO.Read#withDelimiter})
+ *
+ * <h3>Filepattern expansion and watching</h3>
+ *
+ * <p>By default, the filepatterns are expanded only once. The combination of {@link
+ * FileIO.Match#continuously(Duration, TerminationCondition)} and {@link #readFiles()} allow
+ * streaming of new files matching the filepattern(s).
+ *
+ * <p>By default, {@link #read} prohibits filepatterns that match no files, and {@link #readFiles()}
+ * allows them in case the filepattern contains a glob wildcard character. Use {@link
+ * ContextualTextIO.Read#withEmptyMatchTreatment} or {@link
+ * FileIO.Match#withEmptyMatchTreatment(EmptyMatchTreatment)} plus {@link #readFiles()} to configure
+ * this behavior.
+ *
+ * <p>Example 1: reading a file or filepattern.
+ *
+ * <pre>{@code
+ * Pipeline p = ...;
+ *
+ * // A simple Read of a file:
+ * PCollection<RecordWithMetadata> records = p.apply(ContextualTextIO.read().from("/local/path/to/file.txt"));
+ * }</pre>
+ *
+ * <p>Example 2: reading a PCollection of filenames.
+ *
+ * <pre>{@code
+ * Pipeline p = ...;
+ *
+ * // E.g. the filenames might be computed from other data in the pipeline, or
+ * // read from a data source.
+ * PCollection<String> filenames = ...;
+ *
+ * // Read all files in the collection.
+ * PCollection<RecordWithMetadata> records =
+ * filenames
+ * .apply(FileIO.matchAll())
+ * .apply(FileIO.readMatches())
+ * .apply(ContextualTextIO.readFiles());
+ * }</pre>
+ *
+ * <p>Example 3: streaming new files matching a filepattern.
+ *
+ * <pre>{@code
+ * Pipeline p = ...;
+ *
+ * PCollection<RecordWithMetadata> records = p.apply(ContextualTextIO.read()
+ * .from("/local/path/to/files/*")
+ * .watchForNewFiles(
+ * // Check for new files every minute
+ * Duration.standardMinutes(1),
+ * // Stop watching the filepattern if no new files appear within an hour
+ * afterTimeSinceNewOutput(Duration.standardHours(1))));
+ * }</pre>
+ *
+ * <p>Example 4: reading a file or file pattern of RFC4180-compliant CSV files with fields that may
+ * contain line breaks.
+ *
+ * Example of such a file could be:
+ *
+ * "aaa","b CRLF
+ * bb","ccc" CRLF
+ * zzz,yyy,xxx
+ *
+ * <pre>{@code
+ * Pipeline p = ...;
+ *
+ * PCollection<RecordWithMetadata> records = p.apply(ContextualTextIO.read()
+ * .from("/local/path/to/files/*.csv")
+ * .withHasMultilineCSVRecords(true));
+ * }</pre>
+ *
+ * <p>Example 5: reading while watching for new files
+ *
+ * <pre>{@code
+ * Pipeline p = ...;
+ *
+ * PCollection<RecordWithMetadata> records = p.apply(FileIO.match()
+ * .filepattern("filepattern")
+ * .continuously(
+ * Duration.millis(100),
+ * Watch.Growth.afterTimeSinceNewOutput(Duration.standardSeconds(3))))
+ * .apply(FileIO.readMatches())
+ * .apply(ContextualTextIO.readFiles());
+ * }</pre>
+ *
+ * <p>Example 6: reading without recordNum metadata, or only fileName associated Metadata. (the
+ * Objects would still contain recordNums, but these recordNums would correspond to their positions
+ * in their respective offsets rather than their positions within the entire file).
+ *
+ * <pre>{@code
+ * Pipeline p = ...;
+ *
+ * PCollection<RecordWithMetadata> records = p.apply(ContextualTextIO.read()
+ * .from("/local/path/to/files/*.csv")
+ * .setWithoutRecordNumMetadata(true));
+ * }</pre>
+ *
+ * <p>NOTE: When using {@link ContextualTextIO.Read#withHasMultilineCSVRecords(Boolean)} this
+ * option, a single reader will be used to process the file, rather than multiple readers which can
+ * read from different offsets. For a large file this can result in lower performance.
+ *
+ * <p>NOTE: Use {@link Read#withoutRecordNumMetadata()} when recordNum metadata is not required, for
+ * example, when when only filename metadata is required. Computing record positions currently
+ * introduces a shuffle step, which increases the resources used by the pipeline. <b> By default
+ * withoutRecordNumMetadata is set to false, so the shuffle step is performed.</b>
+ *
+ * <h3>Reading a very large number of files</h3>
+ *
+ * <p>If it is known that the filepattern will match a very large number of files (e.g. tens of
+ * thousands or more), use {@link ContextualTextIO.Read#withHintMatchesManyFiles} for better
+ * performance and scalability. Note that it may decrease performance if the filepattern matches
+ * only a small number of files.
+ */
+public class ContextualTextIO {
+ private static final long DEFAULT_BUNDLE_SIZE_BYTES = 64 * 1024 * 1024L;
+ private static final Logger LOG = LoggerFactory.getLogger(ContextualTextIO.class);
+
+ /**
+ * A {@link PTransform} that reads from one or more text files and returns a bounded {@link
+ * PCollection} containing one {@link RecordWithMetadata}element for each line of the input files.
+ */
+ public static Read read() {
+ return new AutoValue_ContextualTextIO_Read.Builder()
+ .setCompression(Compression.AUTO)
+ .setHintMatchesManyFiles(false)
+ .setWithoutRecordNumMetadata(false)
+ .setMatchConfiguration(MatchConfiguration.create(EmptyMatchTreatment.DISALLOW))
+ .setHasMultilineCSVRecords(false)
+ .build();
+ }
+
+ /**
+ * Like {@link #read}, but reads each file in a {@link PCollection} of {@link
+ * FileIO.ReadableFile}, returned by {@link FileIO#readMatches}.
+ */
+ public static ReadFiles readFiles() {
+ return new AutoValue_ContextualTextIO_ReadFiles.Builder()
+ // 64MB is a reasonable value that allows to amortize the cost of opening files,
+ // but is not so large as to exhaust a typical runner's maximum amount of output per
+ // ProcessElement call.
+ .setDesiredBundleSizeBytes(DEFAULT_BUNDLE_SIZE_BYTES)
+ .setHasMultilineCSVRecords(false)
+ .build();
+ }
+
+ /** Implementation of {@link #read}. */
+ @AutoValue
+ public abstract static class Read extends PTransform<PBegin, PCollection<RecordWithMetadata>> {
+ abstract @Nullable ValueProvider<String> getFilepattern();
+
+ abstract MatchConfiguration getMatchConfiguration();
+
+ abstract boolean getHintMatchesManyFiles();
+
+ abstract boolean getWithoutRecordNumMetadata();
+
+ abstract Compression getCompression();
+
+ abstract @Nullable Boolean getHasMultilineCSVRecords();
+
+ @SuppressWarnings("mutable") // this returns an array that can be mutated by the caller
+ abstract byte @Nullable [] getDelimiter();
+
+ abstract Builder toBuilder();
+
+ @AutoValue.Builder
+ abstract static class Builder {
+ abstract Builder setFilepattern(ValueProvider<String> filepattern);
+
+ abstract Builder setMatchConfiguration(MatchConfiguration matchConfiguration);
+
+ abstract Builder setHintMatchesManyFiles(boolean hintManyFiles);
+
+ abstract Builder setWithoutRecordNumMetadata(boolean withoutLineNumMetadata);
+
+ abstract Builder setCompression(Compression compression);
+
+ abstract Builder setDelimiter(byte @Nullable [] delimiter);
+
+ abstract Builder setHasMultilineCSVRecords(Boolean hasMultilineCSVRecords);
+
+ abstract Read build();
+ }
+
+ /**
+ * Reads text from the file(s) with the given filename or filename pattern.
+ *
+ * <p>This can be a local path (if running locally), or a Google Cloud Storage filename or
+ * filename pattern of the form {@code "gs://<bucket>/<filepath>"} (if running locally or using
+ * remote execution service).
+ *
+ * <p>Standard <a href="http://docs.oracle.com/javase/tutorial/essential/io/find.html" >Java
+ * Filesystem glob patterns</a> ("*", "?", "[..]") are supported.
+ *
+ * <p>If it is known that the filepattern will match a very large number of files (at least tens
+ * of thousands), use {@link #withHintMatchesManyFiles} for better performance and scalability.
+ */
+ public Read from(String filepattern) {
+ checkArgument(filepattern != null, "filepattern can not be null");
+ return from(StaticValueProvider.of(filepattern));
+ }
+
+ /** Same as {@code from(filepattern)}, but accepting a {@link ValueProvider}. */
+ public Read from(ValueProvider<String> filepattern) {
+ checkArgument(filepattern != null, "filepattern can not be null");
+ return toBuilder().setFilepattern(filepattern).build();
+ }
+
+ /** Sets the {@link MatchConfiguration}. */
+ public Read withMatchConfiguration(MatchConfiguration matchConfiguration) {
+ return toBuilder().setMatchConfiguration(matchConfiguration).build();
+ }
+
+ /**
+ * When reading RFC4180 CSV files that have values that span multiple lines, set this to true.
+ * Note: this reduces the read performance (see: {@link ContextualTextIO}).
+ */
+ public Read withHasMultilineCSVRecords(Boolean hasMultilineCSVRecords) {
+ return toBuilder().setHasMultilineCSVRecords(hasMultilineCSVRecords).build();
+ }
+
+ /**
+ * Reads from input sources using the specified compression type.
+ *
+ * <p>If no compression type is specified, the default is {@link Compression#AUTO}.
+ */
+ public Read withCompression(Compression compression) {
+ return toBuilder().setCompression(compression).build();
+ }
+
+ /**
+ * Hints that the filepattern specified in {@link #from(String)} matches a very large number of
+ * files.
+ *
+ * <p>This hint may cause a runner to execute the transform differently, in a way that improves
+ * performance for this case, but it may worsen performance if the filepattern matches only a
+ * small number of files (e.g., in a runner that supports dynamic work rebalancing, it will
+ * happen less efficiently within individual files).
+ */
+ public Read withHintMatchesManyFiles() {
+ return toBuilder().setHintMatchesManyFiles(true).build();
+ }
+
+ /**
+ * Allows the user to opt out of getting recordNums associated with each record.
+ *
+ * <p>When set to true, it will introduce a shuffle step to assemble the recordNums for each
+ * record, which will increase the resources used by the pipeline.
+ *
+ * <p>Use this when metadata like fileNames are required and their position/order can be
+ * ignored.
+ */
+ public Read withoutRecordNumMetadata() {
+ return toBuilder().setWithoutRecordNumMetadata(true).build();
+ }
+
+ /** See {@link MatchConfiguration#withEmptyMatchTreatment}. */
+ public Read withEmptyMatchTreatment(EmptyMatchTreatment treatment) {
+ return withMatchConfiguration(getMatchConfiguration().withEmptyMatchTreatment(treatment));
+ }
+
+ /** Set the custom delimiter to be used in place of the default ones ('\r', '\n' or '\r\n'). */
+ public Read withDelimiter(byte[] delimiter) {
+ checkArgument(delimiter != null, "delimiter can not be null");
+ checkArgument(!isSelfOverlapping(delimiter), "delimiter must not self-overlap");
+ return toBuilder().setDelimiter(delimiter).build();
+ }
+
+ static boolean isSelfOverlapping(byte[] s) {
+ // s self-overlaps if v exists such as s = vu = wv with u and w non empty
+ for (int i = 1; i < s.length - 1; ++i) {
+ if (ByteBuffer.wrap(s, 0, i).equals(ByteBuffer.wrap(s, s.length - i, i))) {
+ return true;
+ }
+ }
+ return false;
+ }
+
+ @Override
+ public PCollection<RecordWithMetadata> expand(PBegin input) {
+ checkNotNull(
+ getFilepattern(), "need to set the filepattern of a ContextualTextIO.Read transform");
+ PCollection<RecordWithMetadata> records = null;
+ if (getMatchConfiguration().getWatchInterval() == null && !getHintMatchesManyFiles()) {
+ records = input.apply("Read", org.apache.beam.sdk.io.Read.from(getSource()));
+ } else {
+ // All other cases go through FileIO + ReadFiles
+ records =
+ input
+ .apply(
+ "Create filepattern", Create.ofProvider(getFilepattern(), StringUtf8Coder.of()))
+ .apply("Match All", FileIO.matchAll().withConfiguration(getMatchConfiguration()))
+ .apply(
+ "Read Matches",
+ FileIO.readMatches()
+ .withCompression(getCompression())
+ .withDirectoryTreatment(DirectoryTreatment.PROHIBIT))
+ .apply("Via ReadFiles", readFiles().withDelimiter(getDelimiter()));
+ }
+
+ // Check if the user decided to opt out of recordNums associated with records
+ if (getWithoutRecordNumMetadata()) {
+ return records;
+ }
+
+ /*
+ * At this point the line number in RecordWithMetadata contains the relative line offset from the beginning of the read range.
+ *
+ * To compute the absolute position from the beginning of the input we group the lines within the same ranges, and evaluate the size of each range.
+ */
+
+ PCollection<KV<KV<String, Long>, RecordWithMetadata>> recordsGroupedByFileAndRange =
+ records.apply("AddFileNameAndRange", ParDo.of(new AddFileNameAndRange()));
+
+ PCollectionView<Map<KV<String, Long>, Long>> rangeSizes =
+ recordsGroupedByFileAndRange
+ .apply("CountRecordsForEachFileRange", Count.perKey())
+ .apply("SizesAsView", View.asMap());
+
+ // Get Pipeline to create a dummy PCollection with one element to help compute the lines
+ // before each Range
+ PCollection<Integer> singletonPcoll =
+ input.getPipeline().apply("CreateSingletonPcoll", Create.of(Arrays.asList(1)));
+
+ /*
+ * For each (File, Offset) pair, calculate the number of lines occurring before the Range for each file
+ *
+ * After computing the number of lines before each range, we can find the line number in original file as numLiesBeforeOffset + lineNumInCurrentOffset
+ */
+
+ PCollectionView<Map<KV<String, Long>, Long>> numRecordsBeforeEachRange =
+ singletonPcoll
+ .apply(
+ "ComputeRecordsBeforeRange",
+ ParDo.of(new ComputeRecordsBeforeEachRange(rangeSizes))
+ .withSideInputs(rangeSizes))
+ .apply("NumRecordsBeforeEachRangeAsView", View.asMap());
+
+ return recordsGroupedByFileAndRange.apply(
+ "AssignLineNums",
+ ParDo.of(new AssignRecordNums(numRecordsBeforeEachRange))
+ .withSideInputs(numRecordsBeforeEachRange));
+ }
+
+ @VisibleForTesting
+ static class AddFileNameAndRange
+ extends DoFn<RecordWithMetadata, KV<KV<String, Long>, RecordWithMetadata>> {
+ @ProcessElement
+ public void processElement(
+ @Element RecordWithMetadata record,
+ OutputReceiver<KV<KV<String, Long>, RecordWithMetadata>> out) {
+ out.output(KV.of(KV.of(record.getFileName().toString(), record.getRangeOffset()), record));
+ }
+ }
+
+ /**
+ * Helper class for computing number of record in the File preceding the beginning of the Range
+ * in this file.
+ */
+ @VisibleForTesting
+ static class ComputeRecordsBeforeEachRange extends DoFn<Integer, KV<KV<String, Long>, Long>> {
+ private final PCollectionView<Map<KV<String, Long>, Long>> rangeSizes;
+
+ public ComputeRecordsBeforeEachRange(
+ PCollectionView<Map<KV<String, Long>, Long>> rangeSizes) {
+ this.rangeSizes = rangeSizes;
+ }
+
+ // Add custom comparator as KV<K, V> is not comparable by default
+ private static class FileRangeComparator<K extends Comparable<K>, V extends Comparable<V>>
+ implements Comparator<KV<K, V>> {
+ @Override
+ public int compare(KV<K, V> a, KV<K, V> b) {
+ if (a.getKey().compareTo(b.getKey()) == 0) {
+ return a.getValue().compareTo(b.getValue());
+ }
+ return a.getKey().compareTo(b.getKey());
+ }
+ }
+
+ @ProcessElement
+ public void processElement(ProcessContext p) {
+ // Get the Map Containing the size from side-input
+ Map<KV<String, Long>, Long> rangeSizesMap = p.sideInput(rangeSizes);
+
+ // The FileRange Pair must be sorted
+ SortedMap<KV<String, Long>, Long> sorted = new TreeMap<>(new FileRangeComparator<>());
+
+ // Initialize sorted map with values
+ for (Map.Entry<KV<String, Long>, Long> entry : rangeSizesMap.entrySet()) {
+ sorted.put(entry.getKey(), entry.getValue());
+ }
+
+ // HashMap that tracks number of records passed for each file
+ Map<String, Long> pastRecords = new HashMap<>();
+
+ // For each (File, Range) Pair, compute the number of records before it
+ for (Map.Entry entry : sorted.entrySet()) {
Review comment:
`Make Map.Entry be Map.Entry<<KV<String, Long>, Long> >` ?
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
[GitHub] [beam] abhiy13 commented on a change in pull request #12645: [BEAM-10124] Add ContextualTextIO
Posted by GitBox <gi...@apache.org>.
abhiy13 commented on a change in pull request #12645:
URL: https://github.com/apache/beam/pull/12645#discussion_r492474752
##########
File path: sdks/java/io/contextual-text-io/src/main/java/org/apache/beam/sdk/io/contextualtextio/RecordWithMetadata.java
##########
@@ -0,0 +1,85 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one
+ * or more contributor license agreements. See the NOTICE file
+ * distributed with this work for additional information
+ * regarding copyright ownership. The ASF licenses this file
+ * to you under the Apache License, Version 2.0 (the
+ * "License"); you may not use this file except in compliance
+ * with the License. You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+package org.apache.beam.sdk.io.contextualtextio;
+
+import com.google.auto.value.AutoValue;
+import org.apache.beam.sdk.annotations.Experimental;
+import org.apache.beam.sdk.io.fs.ResourceId;
+import org.apache.beam.sdk.schemas.AutoValueSchema;
+import org.apache.beam.sdk.schemas.annotations.DefaultSchema;
+
+/**
+ * Helper Class based on {@link AutoValueSchema}, it provides Metadata associated with each Record
+ * when reading from file(s) using {@link ContextualTextIO}.
+ *
+ * <h3>Fields:</h3>
+ *
+ * <ul>
+ * <li>recordOffset: The offset of a record (the byte at which the record begins) in a file. This
+ * information can be useful if you wish to reconstruct the file. {@link
+ * RecordWithMetadata#getRecordOffset()}
+ * <li>recordNum: The ordinal number of the record in its file. {@link
+ * RecordWithMetadata#getRecordNum()}
+ * <li>recordValue: The value / contents of the record {@link RecordWithMetadata#getValue()}
+ * <li>rangeOffset: The starting offset of the range (split), which contained the record, when the
+ * record was read. {@link RecordWithMetadata#getRangeOffset()}
+ * <li>recordNumInOffset: The record number relative to the Range. (line number within the range)
+ * {@link RecordWithMetadata#getRecordNumInOffset()}
+ * <li>fileName: Name of the file to which the record belongs (this is the full filename,
+ * eg:path/to/file.txt) {@link RecordWithMetadata#getFileName()}
+ * </ul>
+ */
+@Experimental(Experimental.Kind.SCHEMAS)
+@DefaultSchema(AutoValueSchema.class)
+@AutoValue
+public abstract class RecordWithMetadata {
+ public abstract long getRecordOffset();
+
+ public abstract long getRecordNum();
+
+ public abstract String getValue();
+
+ public abstract long getRangeOffset();
+
+ public abstract long getRecordNumInOffset();
+
+ public abstract Builder toBuilder();
+
+ public abstract String getFileName();
+
+ public static Builder newBuilder() {
+ return new AutoValue_RecordWithMetadata.Builder();
+ }
+
+ @AutoValue.Builder
+ public abstract static class Builder {
+ public abstract Builder setRecordNum(long lineNum);
+
+ public abstract Builder setRecordOffset(long recordOffset);
+
+ public abstract Builder setValue(String Value);
Review comment:
Done.
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
[GitHub] [beam] rezarokni commented on a change in pull request #12645: [BEAM-10124] Add ContextualTextIO
Posted by GitBox <gi...@apache.org>.
rezarokni commented on a change in pull request #12645:
URL: https://github.com/apache/beam/pull/12645#discussion_r487899783
##########
File path: sdks/java/io/contextual-text-io/src/main/java/org/apache/beam/sdk/io/contextualtextio/RecordWithMetadata.java
##########
@@ -0,0 +1,84 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one
+ * or more contributor license agreements. See the NOTICE file
+ * distributed with this work for additional information
+ * regarding copyright ownership. The ASF licenses this file
+ * to you under the Apache License, Version 2.0 (the
+ * "License"); you may not use this file except in compliance
+ * with the License. You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+package org.apache.beam.sdk.io.contextualtextio;
+
+import com.google.auto.value.AutoValue;
+import org.apache.beam.sdk.annotations.Experimental;
+import org.apache.beam.sdk.schemas.AutoValueSchema;
+import org.apache.beam.sdk.schemas.annotations.DefaultSchema;
+
+/**
+ * Helper Class based on {@link AutoValueSchema}, it provides Metadata associated with each Record
+ * when reading from file(s) using {@link ContextualTextIO}.
+ *
+ * <h3>Fields:</h3>
+ *
+ * <ul>
+ * <li>recordOffset: The offset of a record (the byte at which the record begins) in a file. This
+ * information can be useful if you wish to reconstruct the file. {@link
+ * RecordWithMetadata#getRecordOffset()}
+ * <li>recordNum: The ordinal number of the record in its file. {@link
+ * RecordWithMetadata#getRecordNum()}
+ * <li>recordValue: The value / contents of the record {@link RecordWithMetadata#getRecordValue()}
+ * <li>rangeOffset: The starting offset of the range (split), which contained the record, when the
+ * record was read. {@link RecordWithMetadata#getRangeOffset()}
+ * <li>recordNumInOffset: The record number relative to the Range. (line number within the range)
+ * {@link RecordWithMetadata#getRecordNumInOffset()}
+ * <li>fileName: Name of the file to which the record belongs (this is the full filename,
+ * eg:path/to/file.txt) {@link RecordWithMetadata#getFileName()}
+ * </ul>
+ */
+@Experimental(Experimental.Kind.SCHEMAS)
+@DefaultSchema(AutoValueSchema.class)
+@AutoValue
+public abstract class RecordWithMetadata {
+ public abstract Long getRecordOffset();
+
+ public abstract Long getRecordNum();
+
+ public abstract String getRecordValue();
+
+ public abstract Long getRangeOffset();
+
+ public abstract Long getRecordNumInOffset();
+
+ public abstract Builder toBuilder();
+
+ public abstract String getFileName();
Review comment:
See note above as to why we cant use ResourceID
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
[GitHub] [beam] abhiy13 commented on a change in pull request #12645: [BEAM-10124] Add ContextualTextIO
Posted by GitBox <gi...@apache.org>.
abhiy13 commented on a change in pull request #12645:
URL: https://github.com/apache/beam/pull/12645#discussion_r474445761
##########
File path: sdks/java/io/contextual-text-io/src/main/java/org/apache/beam/sdk/io/contextualtextio/ContextualTextIO.java
##########
@@ -296,6 +290,19 @@ public Read withHintMatchesManyFiles() {
return toBuilder().setHintMatchesManyFiles(true).build();
}
+ /**
+ * Allows the user the opt out of getting recordNums associated with each record.
Review comment:
Ack.
##########
File path: sdks/java/io/contextual-text-io/src/main/java/org/apache/beam/sdk/io/contextualtextio/ContextualTextIO.java
##########
@@ -296,6 +290,19 @@ public Read withHintMatchesManyFiles() {
return toBuilder().setHintMatchesManyFiles(true).build();
}
+ /**
+ * Allows the user the opt out of getting recordNums associated with each record.
+ *
+ * <p>ContextualTextIO uses a shuffle step to assemble the recordNums for each record which may
+ * result in some performance loss.
+ *
+ * <p>Use this when metadata like fileNames are required and their position/order can be
+ * ignored.
+ */
+ public Read withoutLineNumMetadata() {
Review comment:
Ack.
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
[GitHub] [beam] tvalentyn commented on a change in pull request #12645: [BEAM-10124] Add ContextualTextIO
Posted by GitBox <gi...@apache.org>.
tvalentyn commented on a change in pull request #12645:
URL: https://github.com/apache/beam/pull/12645#discussion_r474808754
##########
File path: sdks/java/io/contextual-text-io/src/main/java/org/apache/beam/sdk/io/contextualtextio/ContextualTextIOSource.java
##########
@@ -259,7 +317,31 @@ private void decodeCurrentElement() throws IOException {
if (startOfRecord == 0 && dataToDecode.startsWith(UTF8_BOM)) {
dataToDecode = dataToDecode.substring(UTF8_BOM.size());
}
- currentValue = dataToDecode.toStringUtf8();
+
+ /////////////////////////////////////////////
+
+ // Data of the Current Line
+ // dataToDecode.toStringUtf8();
+
+ // The line num is:
Review comment:
Looks like it's still here :) Pehaps you didn't push the changes?
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
[GitHub] [beam] abhiy13 commented on a change in pull request #12645: [BEAM-10124] Add ContextualTextIO
Posted by GitBox <gi...@apache.org>.
abhiy13 commented on a change in pull request #12645:
URL: https://github.com/apache/beam/pull/12645#discussion_r474442666
##########
File path: sdks/java/io/contextual-text-io/src/main/java/org/apache/beam/sdk/io/contextualtextio/ContextualTextIO.java
##########
@@ -0,0 +1,594 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one
+ * or more contributor license agreements. See the NOTICE file
+ * distributed with this work for additional information
+ * regarding copyright ownership. The ASF licenses this file
+ * to you under the Apache License, Version 2.0 (the
+ * "License"); you may not use this file except in compliance
+ * with the License. You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+package org.apache.beam.sdk.io.contextualtextio;
+
+import static org.apache.beam.sdk.io.FileIO.ReadMatches.DirectoryTreatment;
+import static org.apache.beam.vendor.guava.v26_0_jre.com.google.common.base.Preconditions.checkArgument;
+import static org.apache.beam.vendor.guava.v26_0_jre.com.google.common.base.Preconditions.checkNotNull;
+
+import com.google.auto.value.AutoValue;
+import java.nio.ByteBuffer;
+import java.util.Arrays;
+import java.util.Comparator;
+import java.util.HashMap;
+import java.util.Map;
+import java.util.SortedMap;
+import java.util.TreeMap;
+import org.apache.beam.sdk.coders.StringUtf8Coder;
+import org.apache.beam.sdk.io.CompressedSource;
+import org.apache.beam.sdk.io.Compression;
+import org.apache.beam.sdk.io.FileBasedSource;
+import org.apache.beam.sdk.io.FileIO;
+import org.apache.beam.sdk.io.FileIO.MatchConfiguration;
+import org.apache.beam.sdk.io.ReadAllViaFileBasedSource;
+import org.apache.beam.sdk.io.fs.EmptyMatchTreatment;
+import org.apache.beam.sdk.options.ValueProvider;
+import org.apache.beam.sdk.options.ValueProvider.StaticValueProvider;
+import org.apache.beam.sdk.schemas.NoSuchSchemaException;
+import org.apache.beam.sdk.schemas.SchemaCoder;
+import org.apache.beam.sdk.transforms.Count;
+import org.apache.beam.sdk.transforms.Create;
+import org.apache.beam.sdk.transforms.DoFn;
+import org.apache.beam.sdk.transforms.PTransform;
+import org.apache.beam.sdk.transforms.ParDo;
+import org.apache.beam.sdk.transforms.SerializableFunction;
+import org.apache.beam.sdk.transforms.View;
+import org.apache.beam.sdk.transforms.Watch.Growth.TerminationCondition;
+import org.apache.beam.sdk.transforms.display.DisplayData;
+import org.apache.beam.sdk.values.KV;
+import org.apache.beam.sdk.values.PBegin;
+import org.apache.beam.sdk.values.PCollection;
+import org.apache.beam.sdk.values.PCollectionView;
+import org.apache.beam.vendor.guava.v26_0_jre.com.google.common.annotations.VisibleForTesting;
+import org.checkerframework.checker.nullness.qual.Nullable;
+import org.joda.time.Duration;
+
+/**
+ * {@link PTransform}s that read text files and collect contextual information of the elements in
+ * the input.
+ *
+ * <h2>Reading from text files</h2>
+ *
+ * <p>To read a {@link PCollection} from one or more text files, use {@code
+ * ContextualTextIO.read()}. To instantiate a transform use {@link
+ * ContextualTextIO.Read#from(String)} and specify the path of the file(s) to be read.
+ * Alternatively, if the filenames to be read are themselves in a {@link PCollection} you can use
+ * {@link FileIO} to match them and {@link ContextualTextIO#readFiles()} to read them.
+ *
+ * <p>{@link #read} returns a {@link PCollection} of {@link RecordWithMetadata RecordWithMetadata},
+ * each corresponding to one line of an inout UTF-8 text file (split into lines delimited by '\n',
+ * '\r', '\r\n', or specified delimiter see {@link ContextualTextIO.Read#withDelimiter})
+ *
+ * <h3>Filepattern expansion and watching</h3>
+ *
+ * <p>By default, the filepatterns are expanded only once. The combination of {@link
+ * FileIO.Match#continuously(Duration, TerminationCondition)} and {@link #readFiles()} allow
+ * streaming of new files matching the filepattern(s).
+ *
+ * <p>By default, {@link #read} prohibits filepatterns that match no files, and {@link #readFiles()}
+ * allows them in case the filepattern contains a glob wildcard character. Use {@link
+ * ContextualTextIO.Read#withEmptyMatchTreatment} or {@link
+ * FileIO.Match#withEmptyMatchTreatment(EmptyMatchTreatment)} plus {@link #readFiles()} to configure
+ * this behavior.
+ *
+ * <p>Example 1: reading a file or filepattern.
+ *
+ * <pre>{@code
+ * Pipeline p = ...;
+ *
+ * // A simple Read of a file:
+ * PCollection<RecordWithMetadata> lines = p.apply(ContextualTextIO.read().from("/local/path/to/file.txt"));
+ * }</pre>
+ *
+ * <p>Example 2: reading a PCollection of filenames.
+ *
+ * <pre>{@code
+ * Pipeline p = ...;
+ *
+ * // E.g. the filenames might be computed from other data in the pipeline, or
+ * // read from a data source.
+ * PCollection<String> filenames = ...;
+ *
+ * // Read all files in the collection.
+ * PCollection<RecordWithMetadata> lines =
+ * filenames
+ * .apply(FileIO.matchAll())
+ * .apply(FileIO.readMatches())
+ * .apply(ContextualTextIO.readFiles());
+ * }</pre>
+ *
+ * <p>Example 3: streaming new files matching a filepattern.
+ *
+ * <pre>{@code
+ * Pipeline p = ...;
+ *
+ * PCollection<RecordWithMetadata> lines = p.apply(ContextualTextIO.read()
+ * .from("/local/path/to/files/*")
+ * .watchForNewFiles(
+ * // Check for new files every minute
+ * Duration.standardMinutes(1),
+ * // Stop watching the filepattern if no new files appear within an hour
+ * afterTimeSinceNewOutput(Duration.standardHours(1))));
+ * }</pre>
+ *
+ * <p>Example 4: reading a file or file pattern of RFC4180-compliant CSV files with fields that may
+ * contain line breaks.
+ *
+ * <pre>{@code
+ * Pipeline p = ...;
+ *
+ * PCollection<RecordWithMetadata> lines = p.apply(ContextualTextIO.read()
+ * .from("/local/path/to/files/*.csv")
+ * .withHasMultilineCSVRecords(true));
+ * }</pre>
+ *
+ * <p>Example 5: reading while watching for new files
+ *
+ * <pre>{@code
+ * Pipeline p = ...;
+ *
+ * PCollection<RecordWithMetadata> lines = p.apply(FileIO.match()
+ * .filepattern("filepattern")
+ * .continuously(
+ * Duration.millis(100),
+ * Watch.Growth.afterTimeSinceNewOutput(Duration.standardSeconds(3))))
+ * .apply(FileIO.readMatches())
+ * .apply(ContextualTextIO.readFiles());
+ * }</pre>
+ *
+ * NOTE: Using {@link ContextualTextIO.Read#withHasMultilineCSVRecords(Boolean)} introduces a
+ * performance penalty: when this option is enabled, the input cannot be split and read in parallel.
+ *
+ * <h3>Reading a very large number of files</h3>
+ *
+ * <p>If it is known that the filepattern will match a very large number of files (e.g. tens of
+ * thousands or more), use {@link ContextualTextIO.Read#withHintMatchesManyFiles} for better
+ * performance and scalability. Note that it may decrease performance if the filepattern matches
+ * only a small number of files.
+ */
+public class ContextualTextIO {
+ private static final long DEFAULT_BUNDLE_SIZE_BYTES = 64 * 1024 * 1024L;
+
+ /**
+ * A {@link PTransform} that reads from one or more text files and returns a bounded {@link
+ * PCollection} containing one {@link RecordWithMetadata}element for each line of the input files.
+ */
+ public static Read read() {
+ return new AutoValue_ContextualTextIO_Read.Builder()
+ .setCompression(Compression.AUTO)
+ .setHintMatchesManyFiles(false)
+ .setWithoutLineNumMetadata(false)
+ .setMatchConfiguration(MatchConfiguration.create(EmptyMatchTreatment.DISALLOW))
+ .setHasMultilineCSVRecords(false)
+ .build();
+ }
+
+ /**
+ * Like {@link #read}, but reads each file in a {@link PCollection} of {@link
+ * FileIO.ReadableFile}, returned by {@link FileIO#readMatches}.
+ */
+ public static ReadFiles readFiles() {
+ return new AutoValue_ContextualTextIO_ReadFiles.Builder()
+ // 64MB is a reasonable value that allows to amortize the cost of opening files,
+ // but is not so large as to exhaust a typical runner's maximum amount of output per
+ // ProcessElement call.
+ .setDesiredBundleSizeBytes(DEFAULT_BUNDLE_SIZE_BYTES)
+ .setHasMultilineCSVRecords(false)
+ .build();
+ }
+
+ /** Implementation of {@link #read}. */
+ @AutoValue
+ public abstract static class Read extends PTransform<PBegin, PCollection<RecordWithMetadata>> {
+ abstract @Nullable ValueProvider<String> getFilepattern();
+
+ abstract MatchConfiguration getMatchConfiguration();
+
+ abstract boolean getHintMatchesManyFiles();
+
+ abstract boolean getWithoutLineNumMetadata();
+
+ abstract Compression getCompression();
+
+ abstract @Nullable Boolean getHasMultilineCSVRecords();
+
+ @SuppressWarnings("mutable") // this returns an array that can be mutated by the caller
+ abstract byte @Nullable [] getDelimiter();
+
+ abstract Builder toBuilder();
+
+ @AutoValue.Builder
+ abstract static class Builder {
+ abstract Builder setFilepattern(ValueProvider<String> filepattern);
+
+ abstract Builder setMatchConfiguration(MatchConfiguration matchConfiguration);
+
+ abstract Builder setHintMatchesManyFiles(boolean hintManyFiles);
+
+ abstract Builder setWithoutLineNumMetadata(boolean withoutLineNumMetadata);
+
+ abstract Builder setCompression(Compression compression);
+
+ abstract Builder setDelimiter(byte @Nullable [] delimiter);
+
+ abstract Builder setHasMultilineCSVRecords(Boolean hasMultilineCSVRecords);
+
+ abstract Read build();
+ }
+
+ /**
+ * Reads text from the file(s) with the given filename or filename pattern.
+ *
+ * <p>This can be a local path (if running locally), or a Google Cloud Storage filename or
+ * filename pattern of the form {@code "gs://<bucket>/<filepath>"} (if running locally or using
+ * remote execution service).
+ *
+ * <p>Standard <a href="http://docs.oracle.com/javase/tutorial/essential/io/find.html" >Java
+ * Filesystem glob patterns</a> ("*", "?", "[..]") are supported.
+ *
+ * <p>If it is known that the filepattern will match a very large number of files (at least tens
+ * of thousands), use {@link #withHintMatchesManyFiles} for better performance and scalability.
+ */
+ public Read from(String filepattern) {
+ checkArgument(filepattern != null, "filepattern can not be null");
+ return from(StaticValueProvider.of(filepattern));
+ }
+
+ /** Same as {@code from(filepattern)}, but accepting a {@link ValueProvider}. */
+ public Read from(ValueProvider<String> filepattern) {
+ checkArgument(filepattern != null, "filepattern can not be null");
+ return toBuilder().setFilepattern(filepattern).build();
+ }
+
+ /** Sets the {@link MatchConfiguration}. */
+ public Read withMatchConfiguration(MatchConfiguration matchConfiguration) {
+ return toBuilder().setMatchConfiguration(matchConfiguration).build();
+ }
+
+ /**
+ * When reading RFC4180 CSV files that have values that span multiple lines, set this to true.
+ * Note: this reduces the read performance (see: {@link ContextualTextIO}).
+ */
+ public Read withHasMultilineCSVRecords(Boolean hasMultilineCSVRecords) {
+ return toBuilder().setHasMultilineCSVRecords(hasMultilineCSVRecords).build();
+ }
+
+ /**
+ * Reads from input sources using the specified compression type.
+ *
+ * <p>If no compression type is specified, the default is {@link Compression#AUTO}.
+ */
+ public Read withCompression(Compression compression) {
+ return toBuilder().setCompression(compression).build();
+ }
+
+ /**
+ * Hints that the filepattern specified in {@link #from(String)} matches a very large number of
+ * files.
+ *
+ * <p>This hint may cause a runner to execute the transform differently, in a way that improves
+ * performance for this case, but it may worsen performance if the filepattern matches only a
+ * small number of files (e.g., in a runner that supports dynamic work rebalancing, it will
+ * happen less efficiently within individual files).
+ */
+ public Read withHintMatchesManyFiles() {
+ return toBuilder().setHintMatchesManyFiles(true).build();
+ }
+
+ /**
+ * Allows the user the opt out of getting recordNums associated with each record.
+ *
+ * <p>ContextualTextIO uses a shuffle step to assemble the recordNums for each record which may
+ * result in some performance loss.
+ *
+ * <p>Use this when metadata like fileNames are required and their position/order can be
+ * ignored.
+ */
+ public Read withoutLineNumMetadata() {
+ return toBuilder().setWithoutLineNumMetadata(true).build();
+ }
+
+ /** See {@link MatchConfiguration#withEmptyMatchTreatment}. */
+ public Read withEmptyMatchTreatment(EmptyMatchTreatment treatment) {
+ return withMatchConfiguration(getMatchConfiguration().withEmptyMatchTreatment(treatment));
+ }
+
+ /** Set the custom delimiter to be used in place of the default ones ('\r', '\n' or '\r\n'). */
+ public Read withDelimiter(byte[] delimiter) {
+ checkArgument(delimiter != null, "delimiter can not be null");
+ checkArgument(!isSelfOverlapping(delimiter), "delimiter must not self-overlap");
+ return toBuilder().setDelimiter(delimiter).build();
+ }
+
+ static boolean isSelfOverlapping(byte[] s) {
+ // s self-overlaps if v exists such as s = vu = wv with u and w non empty
+ for (int i = 1; i < s.length - 1; ++i) {
+ if (ByteBuffer.wrap(s, 0, i).equals(ByteBuffer.wrap(s, s.length - i, i))) {
+ return true;
+ }
+ }
+ return false;
+ }
+
+ @Override
+ public PCollection<RecordWithMetadata> expand(PBegin input) {
+ checkNotNull(
+ getFilepattern(), "need to set the filepattern of a ContextualTextIO.Read transform");
+ PCollection<RecordWithMetadata> lines = null;
+ if (getMatchConfiguration().getWatchInterval() == null && !getHintMatchesManyFiles()) {
+ lines = input.apply("Read", org.apache.beam.sdk.io.Read.from(getSource()));
+ } else {
+ // All other cases go through FileIO + ReadFiles
+ lines =
+ input
+ .apply(
+ "Create filepattern", Create.ofProvider(getFilepattern(), StringUtf8Coder.of()))
+ .apply("Match All", FileIO.matchAll().withConfiguration(getMatchConfiguration()))
+ .apply(
+ "Read Matches",
+ FileIO.readMatches()
+ .withCompression(getCompression())
+ .withDirectoryTreatment(DirectoryTreatment.PROHIBIT))
+ .apply("Via ReadFiles", readFiles().withDelimiter(getDelimiter()));
+ }
+
+ // Check if the user decided to opt out of recordNums associated with records
+ if (getWithoutLineNumMetadata()) {
+ return lines;
+ }
+
+ // At this point the line number in RecordWithMetadata contains the relative line offset from
+ // the
+ // beginning of the read range.
+
+ // To compute the absolute position from the beginning of the input,
+ // we group the lines within the same ranges, and evaluate the size of each range.
+
+ PCollection<KV<KV<String, Long>, RecordWithMetadata>> linesGroupedByFileAndRange =
+ lines.apply("AddFileNameAndRange", ParDo.of(new AddFileNameAndRange()));
+
+ PCollectionView<Map<KV<String, Long>, Long>> rangeSizes =
+ linesGroupedByFileAndRange
+ .apply("CountLinesForEachFileRange", Count.perKey())
+ .apply("SizesAsView", View.asMap());
+
+ // Get Pipeline to create a dummy PCollection with one element to help compute the lines
+ // before each Range
+ PCollection<Integer> singletonPcoll =
+ input.getPipeline().apply("CreateSingletonPcoll", Create.of(Arrays.asList(1)));
+
+ // For each (File, Offset) pair, calculate the number of lines occurring before the Range for
+ // each File
+
+ // After computing the number of lines before each range, we can find the line number in
+ // original file as numLiesBeforeOffset + lineNumInCurrentOffset
+ PCollectionView<Map<KV<String, Long>, Long>> numLinesBeforeEachRange =
+ singletonPcoll
+ .apply(
+ "ComputeLinesBeforeRange",
+ ParDo.of(new ComputeLinesBeforeEachRange(rangeSizes)).withSideInputs(rangeSizes))
+ .apply("NumLinesBeforeEachRangeAsView", View.asMap());
+
+ return linesGroupedByFileAndRange.apply(
+ "AssignLineNums",
+ ParDo.of(new AssignLineNums(numLinesBeforeEachRange))
+ .withSideInputs(numLinesBeforeEachRange));
+ }
+
+ @VisibleForTesting
+ static class AddFileNameAndRange
+ extends DoFn<RecordWithMetadata, KV<KV<String, Long>, RecordWithMetadata>> {
+ @ProcessElement
+ public void processElement(
+ @Element RecordWithMetadata line,
+ OutputReceiver<KV<KV<String, Long>, RecordWithMetadata>> out) {
+ out.output(KV.of(KV.of(line.getFileName(), line.getRange().getRangeNum()), line));
+ }
+ }
+
+ /**
+ * Helper class for computing number of lines in the File preceding the beginning of the Range
+ * in this file.
+ */
+ @VisibleForTesting
+ static class ComputeLinesBeforeEachRange extends DoFn<Integer, KV<KV<String, Long>, Long>> {
+ private final PCollectionView<Map<KV<String, Long>, Long>> rangeSizes;
+
+ public ComputeLinesBeforeEachRange(PCollectionView<Map<KV<String, Long>, Long>> rangeSizes) {
+ this.rangeSizes = rangeSizes;
+ }
+
+ // Add custom comparator as KV<K, V> is not comparable by default
+ private static class FileRangeComparator<K extends Comparable<K>, V extends Comparable<V>>
+ implements Comparator<KV<K, V>> {
+ @Override
+ public int compare(KV<K, V> a, KV<K, V> b) {
+ if (a.getKey().compareTo(b.getKey()) == 0) {
+ return a.getValue().compareTo(b.getValue());
+ }
+ return a.getKey().compareTo(b.getKey());
+ }
+ }
+
+ @ProcessElement
+ public void processElement(ProcessContext p) {
+ // Get the Map Containing the size from side-input
+ Map<KV<String, Long>, Long> rangeSizesMap = p.sideInput(rangeSizes);
+
+ // The FileRange Pair must be sorted
+ SortedMap<KV<String, Long>, Long> sorted = new TreeMap<>(new FileRangeComparator<>());
+
+ // Initialize sorted map with values
+ for (Map.Entry<KV<String, Long>, Long> entry : rangeSizesMap.entrySet()) {
+ sorted.put(entry.getKey(), entry.getValue());
+ }
+
+ // HashMap that tracks lines passed for each file
+ Map<String, Long> pastLines = new HashMap<>();
+
+ // For each (File, Range) Pair, compute the number of lines before it
+ for (Map.Entry entry : sorted.entrySet()) {
+ Long lines = (long) entry.getValue();
+ KV<String, Long> fileRange = (KV<String, Long>) entry.getKey();
+ String file = fileRange.getKey();
+ Long linesBefore = 0L;
+ if (pastLines.containsKey(file)) {
+ linesBefore = pastLines.get(file);
+ }
+ p.output(KV.of(fileRange, linesBefore));
+ pastLines.put(file, linesBefore + lines);
+ }
+ }
+ }
+
+ static class AssignLineNums
+ extends DoFn<KV<KV<String, Long>, RecordWithMetadata>, RecordWithMetadata> {
+ PCollectionView<Map<KV<String, Long>, Long>> numLinesBeforeEachRange;
+
+ public AssignLineNums(PCollectionView<Map<KV<String, Long>, Long>> numLinesBeforeEachRange) {
+ this.numLinesBeforeEachRange = numLinesBeforeEachRange;
+ }
+
+ @ProcessElement
+ public void processElement(ProcessContext p) {
+ Long range = p.element().getKey().getValue();
+ String file = p.element().getKey().getKey();
+ RecordWithMetadata line = p.element().getValue();
+ Long linesLessThanThisRange = p.sideInput(numLinesBeforeEachRange).get(KV.of(file, range));
+ RecordWithMetadata newLine =
+ RecordWithMetadata.newBuilder()
+ .setRecordValue(line.getRecordValue())
+ .setRecordNum(line.getRange().getRangeLineNum() + linesLessThanThisRange)
+ .setFileName(line.getFileName())
+ .setRange(line.getRange())
+ .build();
+ p.output(newLine);
+ }
+ }
+
+ // Helper to create a source specific to the requested compression type.
+ protected FileBasedSource<RecordWithMetadata> getSource() {
+ return CompressedSource.from(
+ new ContextualTextIOSource(
+ getFilepattern(),
+ getMatchConfiguration().getEmptyMatchTreatment(),
+ getDelimiter(),
+ getHasMultilineCSVRecords()))
+ .withCompression(getCompression());
+ }
+
+ @Override
+ public void populateDisplayData(DisplayData.Builder builder) {
+ super.populateDisplayData(builder);
+ builder
+ .add(
+ DisplayData.item("compressionType", getCompression().toString())
+ .withLabel("Compression Type"))
+ .addIfNotNull(DisplayData.item("filePattern", getFilepattern()).withLabel("File Pattern"))
+ .include("matchConfiguration", getMatchConfiguration())
+ .addIfNotNull(
+ DisplayData.item("delimiter", Arrays.toString(getDelimiter()))
+ .withLabel("Custom delimiter to split records"))
+ .addIfNotNull(
+ DisplayData.item("hasMultilineCSVRecords", getHasMultilineCSVRecords())
+ .withLabel("Has RFC4180 MultiLineCSV Records"));
+ }
+ }
+
+ /** Implementation of {@link #readFiles}. */
+ @AutoValue
+ public abstract static class ReadFiles
+ extends PTransform<PCollection<FileIO.ReadableFile>, PCollection<RecordWithMetadata>> {
+ abstract long getDesiredBundleSizeBytes();
+
+ @SuppressWarnings("mutable") // this returns an array that can be mutated by the caller
+ abstract byte @Nullable [] getDelimiter();
+
+ abstract boolean getHasMultilineCSVRecords();
+
+ abstract Builder toBuilder();
+
+ @AutoValue.Builder
+ abstract static class Builder {
+ abstract Builder setDesiredBundleSizeBytes(long desiredBundleSizeBytes);
+
+ abstract Builder setHasMultilineCSVRecords(boolean hasMultilineCSVRecords);
+
+ abstract Builder setDelimiter(byte @Nullable [] delimiter);
+
+ abstract ReadFiles build();
+ }
+
+ @VisibleForTesting
+ ReadFiles withDesiredBundleSizeBytes(long desiredBundleSizeBytes) {
+ return toBuilder().setDesiredBundleSizeBytes(desiredBundleSizeBytes).build();
+ }
+
+ /** Like {@link Read#withDelimiter}. */
+ public ReadFiles withDelimiter(byte[] delimiter) {
+ return toBuilder().setDelimiter(delimiter).build();
+ }
+
+ @Override
+ public PCollection<RecordWithMetadata> expand(PCollection<FileIO.ReadableFile> input) {
+ SchemaCoder<RecordWithMetadata> coder = null;
+ try {
+ coder = input.getPipeline().getSchemaRegistry().getSchemaCoder(RecordWithMetadata.class);
+ } catch (NoSuchSchemaException e) {
+ System.out.println("No Coder!");
Review comment:
Ack.
##########
File path: sdks/java/io/contextual-text-io/src/main/java/org/apache/beam/sdk/io/contextualtextio/ContextualTextIO.java
##########
@@ -0,0 +1,594 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one
+ * or more contributor license agreements. See the NOTICE file
+ * distributed with this work for additional information
+ * regarding copyright ownership. The ASF licenses this file
+ * to you under the Apache License, Version 2.0 (the
+ * "License"); you may not use this file except in compliance
+ * with the License. You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+package org.apache.beam.sdk.io.contextualtextio;
+
+import static org.apache.beam.sdk.io.FileIO.ReadMatches.DirectoryTreatment;
+import static org.apache.beam.vendor.guava.v26_0_jre.com.google.common.base.Preconditions.checkArgument;
+import static org.apache.beam.vendor.guava.v26_0_jre.com.google.common.base.Preconditions.checkNotNull;
+
+import com.google.auto.value.AutoValue;
+import java.nio.ByteBuffer;
+import java.util.Arrays;
+import java.util.Comparator;
+import java.util.HashMap;
+import java.util.Map;
+import java.util.SortedMap;
+import java.util.TreeMap;
+import org.apache.beam.sdk.coders.StringUtf8Coder;
+import org.apache.beam.sdk.io.CompressedSource;
+import org.apache.beam.sdk.io.Compression;
+import org.apache.beam.sdk.io.FileBasedSource;
+import org.apache.beam.sdk.io.FileIO;
+import org.apache.beam.sdk.io.FileIO.MatchConfiguration;
+import org.apache.beam.sdk.io.ReadAllViaFileBasedSource;
+import org.apache.beam.sdk.io.fs.EmptyMatchTreatment;
+import org.apache.beam.sdk.options.ValueProvider;
+import org.apache.beam.sdk.options.ValueProvider.StaticValueProvider;
+import org.apache.beam.sdk.schemas.NoSuchSchemaException;
+import org.apache.beam.sdk.schemas.SchemaCoder;
+import org.apache.beam.sdk.transforms.Count;
+import org.apache.beam.sdk.transforms.Create;
+import org.apache.beam.sdk.transforms.DoFn;
+import org.apache.beam.sdk.transforms.PTransform;
+import org.apache.beam.sdk.transforms.ParDo;
+import org.apache.beam.sdk.transforms.SerializableFunction;
+import org.apache.beam.sdk.transforms.View;
+import org.apache.beam.sdk.transforms.Watch.Growth.TerminationCondition;
+import org.apache.beam.sdk.transforms.display.DisplayData;
+import org.apache.beam.sdk.values.KV;
+import org.apache.beam.sdk.values.PBegin;
+import org.apache.beam.sdk.values.PCollection;
+import org.apache.beam.sdk.values.PCollectionView;
+import org.apache.beam.vendor.guava.v26_0_jre.com.google.common.annotations.VisibleForTesting;
+import org.checkerframework.checker.nullness.qual.Nullable;
+import org.joda.time.Duration;
+
+/**
+ * {@link PTransform}s that read text files and collect contextual information of the elements in
+ * the input.
+ *
+ * <h2>Reading from text files</h2>
+ *
+ * <p>To read a {@link PCollection} from one or more text files, use {@code
+ * ContextualTextIO.read()}. To instantiate a transform use {@link
+ * ContextualTextIO.Read#from(String)} and specify the path of the file(s) to be read.
+ * Alternatively, if the filenames to be read are themselves in a {@link PCollection} you can use
+ * {@link FileIO} to match them and {@link ContextualTextIO#readFiles()} to read them.
+ *
+ * <p>{@link #read} returns a {@link PCollection} of {@link RecordWithMetadata RecordWithMetadata},
+ * each corresponding to one line of an inout UTF-8 text file (split into lines delimited by '\n',
+ * '\r', '\r\n', or specified delimiter see {@link ContextualTextIO.Read#withDelimiter})
+ *
+ * <h3>Filepattern expansion and watching</h3>
+ *
+ * <p>By default, the filepatterns are expanded only once. The combination of {@link
+ * FileIO.Match#continuously(Duration, TerminationCondition)} and {@link #readFiles()} allow
+ * streaming of new files matching the filepattern(s).
+ *
+ * <p>By default, {@link #read} prohibits filepatterns that match no files, and {@link #readFiles()}
+ * allows them in case the filepattern contains a glob wildcard character. Use {@link
+ * ContextualTextIO.Read#withEmptyMatchTreatment} or {@link
+ * FileIO.Match#withEmptyMatchTreatment(EmptyMatchTreatment)} plus {@link #readFiles()} to configure
+ * this behavior.
+ *
+ * <p>Example 1: reading a file or filepattern.
+ *
+ * <pre>{@code
+ * Pipeline p = ...;
+ *
+ * // A simple Read of a file:
+ * PCollection<RecordWithMetadata> lines = p.apply(ContextualTextIO.read().from("/local/path/to/file.txt"));
+ * }</pre>
+ *
+ * <p>Example 2: reading a PCollection of filenames.
+ *
+ * <pre>{@code
+ * Pipeline p = ...;
+ *
+ * // E.g. the filenames might be computed from other data in the pipeline, or
+ * // read from a data source.
+ * PCollection<String> filenames = ...;
+ *
+ * // Read all files in the collection.
+ * PCollection<RecordWithMetadata> lines =
+ * filenames
+ * .apply(FileIO.matchAll())
+ * .apply(FileIO.readMatches())
+ * .apply(ContextualTextIO.readFiles());
+ * }</pre>
+ *
+ * <p>Example 3: streaming new files matching a filepattern.
+ *
+ * <pre>{@code
+ * Pipeline p = ...;
+ *
+ * PCollection<RecordWithMetadata> lines = p.apply(ContextualTextIO.read()
+ * .from("/local/path/to/files/*")
+ * .watchForNewFiles(
+ * // Check for new files every minute
+ * Duration.standardMinutes(1),
+ * // Stop watching the filepattern if no new files appear within an hour
+ * afterTimeSinceNewOutput(Duration.standardHours(1))));
+ * }</pre>
+ *
+ * <p>Example 4: reading a file or file pattern of RFC4180-compliant CSV files with fields that may
+ * contain line breaks.
+ *
+ * <pre>{@code
+ * Pipeline p = ...;
+ *
+ * PCollection<RecordWithMetadata> lines = p.apply(ContextualTextIO.read()
+ * .from("/local/path/to/files/*.csv")
+ * .withHasMultilineCSVRecords(true));
+ * }</pre>
+ *
+ * <p>Example 5: reading while watching for new files
+ *
+ * <pre>{@code
+ * Pipeline p = ...;
+ *
+ * PCollection<RecordWithMetadata> lines = p.apply(FileIO.match()
+ * .filepattern("filepattern")
+ * .continuously(
+ * Duration.millis(100),
+ * Watch.Growth.afterTimeSinceNewOutput(Duration.standardSeconds(3))))
+ * .apply(FileIO.readMatches())
+ * .apply(ContextualTextIO.readFiles());
+ * }</pre>
+ *
+ * NOTE: Using {@link ContextualTextIO.Read#withHasMultilineCSVRecords(Boolean)} introduces a
+ * performance penalty: when this option is enabled, the input cannot be split and read in parallel.
+ *
+ * <h3>Reading a very large number of files</h3>
+ *
+ * <p>If it is known that the filepattern will match a very large number of files (e.g. tens of
+ * thousands or more), use {@link ContextualTextIO.Read#withHintMatchesManyFiles} for better
+ * performance and scalability. Note that it may decrease performance if the filepattern matches
+ * only a small number of files.
+ */
+public class ContextualTextIO {
+ private static final long DEFAULT_BUNDLE_SIZE_BYTES = 64 * 1024 * 1024L;
+
+ /**
+ * A {@link PTransform} that reads from one or more text files and returns a bounded {@link
+ * PCollection} containing one {@link RecordWithMetadata}element for each line of the input files.
+ */
+ public static Read read() {
+ return new AutoValue_ContextualTextIO_Read.Builder()
+ .setCompression(Compression.AUTO)
+ .setHintMatchesManyFiles(false)
+ .setWithoutLineNumMetadata(false)
+ .setMatchConfiguration(MatchConfiguration.create(EmptyMatchTreatment.DISALLOW))
+ .setHasMultilineCSVRecords(false)
+ .build();
+ }
+
+ /**
+ * Like {@link #read}, but reads each file in a {@link PCollection} of {@link
+ * FileIO.ReadableFile}, returned by {@link FileIO#readMatches}.
+ */
+ public static ReadFiles readFiles() {
+ return new AutoValue_ContextualTextIO_ReadFiles.Builder()
+ // 64MB is a reasonable value that allows to amortize the cost of opening files,
+ // but is not so large as to exhaust a typical runner's maximum amount of output per
+ // ProcessElement call.
+ .setDesiredBundleSizeBytes(DEFAULT_BUNDLE_SIZE_BYTES)
+ .setHasMultilineCSVRecords(false)
+ .build();
+ }
+
+ /** Implementation of {@link #read}. */
+ @AutoValue
+ public abstract static class Read extends PTransform<PBegin, PCollection<RecordWithMetadata>> {
+ abstract @Nullable ValueProvider<String> getFilepattern();
+
+ abstract MatchConfiguration getMatchConfiguration();
+
+ abstract boolean getHintMatchesManyFiles();
+
+ abstract boolean getWithoutLineNumMetadata();
+
+ abstract Compression getCompression();
+
+ abstract @Nullable Boolean getHasMultilineCSVRecords();
+
+ @SuppressWarnings("mutable") // this returns an array that can be mutated by the caller
+ abstract byte @Nullable [] getDelimiter();
+
+ abstract Builder toBuilder();
+
+ @AutoValue.Builder
+ abstract static class Builder {
+ abstract Builder setFilepattern(ValueProvider<String> filepattern);
+
+ abstract Builder setMatchConfiguration(MatchConfiguration matchConfiguration);
+
+ abstract Builder setHintMatchesManyFiles(boolean hintManyFiles);
+
+ abstract Builder setWithoutLineNumMetadata(boolean withoutLineNumMetadata);
+
+ abstract Builder setCompression(Compression compression);
+
+ abstract Builder setDelimiter(byte @Nullable [] delimiter);
+
+ abstract Builder setHasMultilineCSVRecords(Boolean hasMultilineCSVRecords);
+
+ abstract Read build();
+ }
+
+ /**
+ * Reads text from the file(s) with the given filename or filename pattern.
+ *
+ * <p>This can be a local path (if running locally), or a Google Cloud Storage filename or
+ * filename pattern of the form {@code "gs://<bucket>/<filepath>"} (if running locally or using
+ * remote execution service).
+ *
+ * <p>Standard <a href="http://docs.oracle.com/javase/tutorial/essential/io/find.html" >Java
+ * Filesystem glob patterns</a> ("*", "?", "[..]") are supported.
+ *
+ * <p>If it is known that the filepattern will match a very large number of files (at least tens
+ * of thousands), use {@link #withHintMatchesManyFiles} for better performance and scalability.
+ */
+ public Read from(String filepattern) {
+ checkArgument(filepattern != null, "filepattern can not be null");
+ return from(StaticValueProvider.of(filepattern));
+ }
+
+ /** Same as {@code from(filepattern)}, but accepting a {@link ValueProvider}. */
+ public Read from(ValueProvider<String> filepattern) {
+ checkArgument(filepattern != null, "filepattern can not be null");
+ return toBuilder().setFilepattern(filepattern).build();
+ }
+
+ /** Sets the {@link MatchConfiguration}. */
+ public Read withMatchConfiguration(MatchConfiguration matchConfiguration) {
+ return toBuilder().setMatchConfiguration(matchConfiguration).build();
+ }
+
+ /**
+ * When reading RFC4180 CSV files that have values that span multiple lines, set this to true.
+ * Note: this reduces the read performance (see: {@link ContextualTextIO}).
+ */
+ public Read withHasMultilineCSVRecords(Boolean hasMultilineCSVRecords) {
+ return toBuilder().setHasMultilineCSVRecords(hasMultilineCSVRecords).build();
+ }
+
+ /**
+ * Reads from input sources using the specified compression type.
+ *
+ * <p>If no compression type is specified, the default is {@link Compression#AUTO}.
+ */
+ public Read withCompression(Compression compression) {
+ return toBuilder().setCompression(compression).build();
+ }
+
+ /**
+ * Hints that the filepattern specified in {@link #from(String)} matches a very large number of
+ * files.
+ *
+ * <p>This hint may cause a runner to execute the transform differently, in a way that improves
+ * performance for this case, but it may worsen performance if the filepattern matches only a
+ * small number of files (e.g., in a runner that supports dynamic work rebalancing, it will
+ * happen less efficiently within individual files).
+ */
+ public Read withHintMatchesManyFiles() {
+ return toBuilder().setHintMatchesManyFiles(true).build();
+ }
+
+ /**
+ * Allows the user the opt out of getting recordNums associated with each record.
+ *
+ * <p>ContextualTextIO uses a shuffle step to assemble the recordNums for each record which may
Review comment:
Ack.
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
[GitHub] [beam] abhiy13 commented on a change in pull request #12645: [BEAM-10124] Add ContextualTextIO
Posted by GitBox <gi...@apache.org>.
abhiy13 commented on a change in pull request #12645:
URL: https://github.com/apache/beam/pull/12645#discussion_r474442918
##########
File path: sdks/java/io/contextual-text-io/src/main/java/org/apache/beam/sdk/io/contextualtextio/ContextualTextIO.java
##########
@@ -0,0 +1,594 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one
+ * or more contributor license agreements. See the NOTICE file
+ * distributed with this work for additional information
+ * regarding copyright ownership. The ASF licenses this file
+ * to you under the Apache License, Version 2.0 (the
+ * "License"); you may not use this file except in compliance
+ * with the License. You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+package org.apache.beam.sdk.io.contextualtextio;
+
+import static org.apache.beam.sdk.io.FileIO.ReadMatches.DirectoryTreatment;
+import static org.apache.beam.vendor.guava.v26_0_jre.com.google.common.base.Preconditions.checkArgument;
+import static org.apache.beam.vendor.guava.v26_0_jre.com.google.common.base.Preconditions.checkNotNull;
+
+import com.google.auto.value.AutoValue;
+import java.nio.ByteBuffer;
+import java.util.Arrays;
+import java.util.Comparator;
+import java.util.HashMap;
+import java.util.Map;
+import java.util.SortedMap;
+import java.util.TreeMap;
+import org.apache.beam.sdk.coders.StringUtf8Coder;
+import org.apache.beam.sdk.io.CompressedSource;
+import org.apache.beam.sdk.io.Compression;
+import org.apache.beam.sdk.io.FileBasedSource;
+import org.apache.beam.sdk.io.FileIO;
+import org.apache.beam.sdk.io.FileIO.MatchConfiguration;
+import org.apache.beam.sdk.io.ReadAllViaFileBasedSource;
+import org.apache.beam.sdk.io.fs.EmptyMatchTreatment;
+import org.apache.beam.sdk.options.ValueProvider;
+import org.apache.beam.sdk.options.ValueProvider.StaticValueProvider;
+import org.apache.beam.sdk.schemas.NoSuchSchemaException;
+import org.apache.beam.sdk.schemas.SchemaCoder;
+import org.apache.beam.sdk.transforms.Count;
+import org.apache.beam.sdk.transforms.Create;
+import org.apache.beam.sdk.transforms.DoFn;
+import org.apache.beam.sdk.transforms.PTransform;
+import org.apache.beam.sdk.transforms.ParDo;
+import org.apache.beam.sdk.transforms.SerializableFunction;
+import org.apache.beam.sdk.transforms.View;
+import org.apache.beam.sdk.transforms.Watch.Growth.TerminationCondition;
+import org.apache.beam.sdk.transforms.display.DisplayData;
+import org.apache.beam.sdk.values.KV;
+import org.apache.beam.sdk.values.PBegin;
+import org.apache.beam.sdk.values.PCollection;
+import org.apache.beam.sdk.values.PCollectionView;
+import org.apache.beam.vendor.guava.v26_0_jre.com.google.common.annotations.VisibleForTesting;
+import org.checkerframework.checker.nullness.qual.Nullable;
+import org.joda.time.Duration;
+
+/**
+ * {@link PTransform}s that read text files and collect contextual information of the elements in
+ * the input.
+ *
+ * <h2>Reading from text files</h2>
+ *
+ * <p>To read a {@link PCollection} from one or more text files, use {@code
+ * ContextualTextIO.read()}. To instantiate a transform use {@link
+ * ContextualTextIO.Read#from(String)} and specify the path of the file(s) to be read.
+ * Alternatively, if the filenames to be read are themselves in a {@link PCollection} you can use
+ * {@link FileIO} to match them and {@link ContextualTextIO#readFiles()} to read them.
+ *
+ * <p>{@link #read} returns a {@link PCollection} of {@link RecordWithMetadata RecordWithMetadata},
+ * each corresponding to one line of an inout UTF-8 text file (split into lines delimited by '\n',
Review comment:
ack.
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
[GitHub] [beam] codecov[bot] edited a comment on pull request #12645: [BEAM-10124] Add ContextualTextIO
Posted by GitBox <gi...@apache.org>.
codecov[bot] edited a comment on pull request #12645:
URL: https://github.com/apache/beam/pull/12645#issuecomment-688630083
# [Codecov](https://codecov.io/gh/apache/beam/pull/12645?src=pr&el=h1) Report
> :exclamation: No coverage uploaded for pull request base (`master@2b2b8e7`). [Click here to learn what that means](https://docs.codecov.io/docs/error-reference#section-missing-base-commit).
> The diff coverage is `n/a`.
[](https://codecov.io/gh/apache/beam/pull/12645?src=pr&el=tree)
```diff
@@ Coverage Diff @@
## master #12645 +/- ##
=========================================
Coverage ? 34.47%
=========================================
Files ? 684
Lines ? 81483
Branches ? 9180
=========================================
Hits ? 28090
Misses ? 52972
Partials ? 421
```
| [Impacted Files](https://codecov.io/gh/apache/beam/pull/12645?src=pr&el=tree) | Coverage Δ | |
|---|---|---|
| [io/textio\_test.py](https://codecov.io/gh/apache/beam/pull/12645/diff?src=pr&el=tree#diff-aW8vdGV4dGlvX3Rlc3QucHk=) | `16.96% <0.00%> (ø)` | |
| [io/gcp/bigquery\_io\_read\_it\_test.py](https://codecov.io/gh/apache/beam/pull/12645/diff?src=pr&el=tree#diff-aW8vZ2NwL2JpZ3F1ZXJ5X2lvX3JlYWRfaXRfdGVzdC5weQ==) | `75.00% <0.00%> (ø)` | |
| [...nners/portability/fn\_api\_runner/worker\_handlers.py](https://codecov.io/gh/apache/beam/pull/12645/diff?src=pr&el=tree#diff-cnVubmVycy9wb3J0YWJpbGl0eS9mbl9hcGlfcnVubmVyL3dvcmtlcl9oYW5kbGVycy5weQ==) | `34.90% <0.00%> (ø)` | |
| [examples/avro\_bitcoin.py](https://codecov.io/gh/apache/beam/pull/12645/diff?src=pr&el=tree#diff-ZXhhbXBsZXMvYXZyb19iaXRjb2luLnB5) | `0.00% <0.00%> (ø)` | |
| [.../snippets/transforms/elementwise/partition\_test.py](https://codecov.io/gh/apache/beam/pull/12645/diff?src=pr&el=tree#diff-ZXhhbXBsZXMvc25pcHBldHMvdHJhbnNmb3Jtcy9lbGVtZW50d2lzZS9wYXJ0aXRpb25fdGVzdC5weQ==) | `46.66% <0.00%> (ø)` | |
| [io/gcp/bigquery\_file\_loads.py](https://codecov.io/gh/apache/beam/pull/12645/diff?src=pr&el=tree#diff-aW8vZ2NwL2JpZ3F1ZXJ5X2ZpbGVfbG9hZHMucHk=) | `23.36% <0.00%> (ø)` | |
| [ml/gcp/visionml.py](https://codecov.io/gh/apache/beam/pull/12645/diff?src=pr&el=tree#diff-bWwvZ2NwL3Zpc2lvbm1sLnB5) | `47.61% <0.00%> (ø)` | |
| [runners/interactive/pipeline\_analyzer.py](https://codecov.io/gh/apache/beam/pull/12645/diff?src=pr&el=tree#diff-cnVubmVycy9pbnRlcmFjdGl2ZS9waXBlbGluZV9hbmFseXplci5weQ==) | `20.00% <0.00%> (ø)` | |
| [transforms/trigger.py](https://codecov.io/gh/apache/beam/pull/12645/diff?src=pr&el=tree#diff-dHJhbnNmb3Jtcy90cmlnZ2VyLnB5) | `37.84% <0.00%> (ø)` | |
| [runners/common\_test.py](https://codecov.io/gh/apache/beam/pull/12645/diff?src=pr&el=tree#diff-cnVubmVycy9jb21tb25fdGVzdC5weQ==) | `24.29% <0.00%> (ø)` | |
| ... and [674 more](https://codecov.io/gh/apache/beam/pull/12645/diff?src=pr&el=tree-more) | |
------
[Continue to review full report at Codecov](https://codecov.io/gh/apache/beam/pull/12645?src=pr&el=continue).
> **Legend** - [Click here to learn more](https://docs.codecov.io/docs/codecov-delta)
> `Δ = absolute <relative> (impact)`, `ø = not affected`, `? = missing data`
> Powered by [Codecov](https://codecov.io/gh/apache/beam/pull/12645?src=pr&el=footer). Last update [2b2b8e7...3bed6b7](https://codecov.io/gh/apache/beam/pull/12645?src=pr&el=lastupdated). Read the [comment docs](https://docs.codecov.io/docs/pull-request-comments).
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
[GitHub] [beam] abhiy13 commented on a change in pull request #12645: [BEAM-10124] Add ContextualTextIO
Posted by GitBox <gi...@apache.org>.
abhiy13 commented on a change in pull request #12645:
URL: https://github.com/apache/beam/pull/12645#discussion_r478654077
##########
File path: sdks/java/io/contextual-text-io/src/main/java/org/apache/beam/sdk/io/contextualtextio/ContextualTextIOSource.java
##########
@@ -0,0 +1,364 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one
+ * or more contributor license agreements. See the NOTICE file
+ * distributed with this work for additional information
+ * regarding copyright ownership. The ASF licenses this file
+ * to you under the Apache License, Version 2.0 (the
+ * "License"); you may not use this file except in compliance
+ * with the License. You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+package org.apache.beam.sdk.io.contextualtextio;
+
+import java.io.IOException;
+import java.nio.ByteBuffer;
+import java.nio.channels.ReadableByteChannel;
+import java.nio.channels.SeekableByteChannel;
+import java.util.NoSuchElementException;
+import org.apache.beam.sdk.coders.Coder;
+import org.apache.beam.sdk.io.FileBasedSource;
+import org.apache.beam.sdk.io.fs.EmptyMatchTreatment;
+import org.apache.beam.sdk.io.fs.MatchResult;
+import org.apache.beam.sdk.options.PipelineOptions;
+import org.apache.beam.sdk.options.ValueProvider;
+import org.apache.beam.sdk.schemas.NoSuchSchemaException;
+import org.apache.beam.sdk.schemas.SchemaCoder;
+import org.apache.beam.sdk.schemas.SchemaRegistry;
+import org.apache.beam.vendor.grpc.v1p26p0.com.google.protobuf.ByteString;
+import org.apache.beam.vendor.guava.v26_0_jre.com.google.common.annotations.VisibleForTesting;
+import org.apache.beam.vendor.guava.v26_0_jre.com.google.common.base.Preconditions;
+import org.checkerframework.checker.nullness.qual.Nullable;
+import org.slf4j.Logger;
+import org.slf4j.LoggerFactory;
+
+/**
+ * Implementation detail of {@link ContextualTextIO.Read}.
+ *
+ * <p>A {@link FileBasedSource} which can decode records delimited by newline characters.
+ *
+ * <p>This source splits the data into records using {@code UTF-8} {@code \n}, {@code \r}, or {@code
+ * \r\n} as the delimiter. This source is not strict and supports decoding the last record even if
+ * it is not delimited. Finally, no records are decoded if the stream is empty.
+ *
+ * <p>This source supports reading from any arbitrary byte position within the stream. If the
+ * starting position is not {@code 0}, then bytes are skipped until the first delimiter is found
+ * representing the beginning of the first record to be decoded.
+ */
+@VisibleForTesting
+class ContextualTextIOSource extends FileBasedSource<RecordWithMetadata> {
Review comment:
Thanks @lukecwik .
@tvalentyn should we add an SDF implementation here or could this be made as a follow up PR ?
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
[GitHub] [beam] abhiy13 commented on a change in pull request #12645: [BEAM-10124] Add ContextualTextIO
Posted by GitBox <gi...@apache.org>.
abhiy13 commented on a change in pull request #12645:
URL: https://github.com/apache/beam/pull/12645#discussion_r474446599
##########
File path: sdks/java/io/contextual-text-io/src/main/java/org/apache/beam/sdk/io/contextualtextio/ContextualTextIOSource.java
##########
@@ -259,7 +317,31 @@ private void decodeCurrentElement() throws IOException {
if (startOfRecord == 0 && dataToDecode.startsWith(UTF8_BOM)) {
dataToDecode = dataToDecode.substring(UTF8_BOM.size());
}
- currentValue = dataToDecode.toStringUtf8();
+
+ /////////////////////////////////////////////
+
+ // Data of the Current Line
+ // dataToDecode.toStringUtf8();
+
+ // The line num is:
+ Long lineUniqueLineNum = readerlineNum++;
Review comment:
Ack.
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
[GitHub] [beam] tvalentyn commented on a change in pull request #12645: [BEAM-10124] Add ContextualTextIO
Posted by GitBox <gi...@apache.org>.
tvalentyn commented on a change in pull request #12645:
URL: https://github.com/apache/beam/pull/12645#discussion_r477565866
##########
File path: sdks/java/io/contextual-text-io/src/main/java/org/apache/beam/sdk/io/contextualtextio/ContextualTextIOSource.java
##########
@@ -0,0 +1,364 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one
+ * or more contributor license agreements. See the NOTICE file
+ * distributed with this work for additional information
+ * regarding copyright ownership. The ASF licenses this file
+ * to you under the Apache License, Version 2.0 (the
+ * "License"); you may not use this file except in compliance
+ * with the License. You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+package org.apache.beam.sdk.io.contextualtextio;
+
+import java.io.IOException;
+import java.nio.ByteBuffer;
+import java.nio.channels.ReadableByteChannel;
+import java.nio.channels.SeekableByteChannel;
+import java.util.NoSuchElementException;
+import org.apache.beam.sdk.coders.Coder;
+import org.apache.beam.sdk.io.FileBasedSource;
+import org.apache.beam.sdk.io.fs.EmptyMatchTreatment;
+import org.apache.beam.sdk.io.fs.MatchResult;
+import org.apache.beam.sdk.options.PipelineOptions;
+import org.apache.beam.sdk.options.ValueProvider;
+import org.apache.beam.sdk.schemas.NoSuchSchemaException;
+import org.apache.beam.sdk.schemas.SchemaCoder;
+import org.apache.beam.sdk.schemas.SchemaRegistry;
+import org.apache.beam.vendor.grpc.v1p26p0.com.google.protobuf.ByteString;
+import org.apache.beam.vendor.guava.v26_0_jre.com.google.common.annotations.VisibleForTesting;
+import org.apache.beam.vendor.guava.v26_0_jre.com.google.common.base.Preconditions;
+import org.checkerframework.checker.nullness.qual.Nullable;
+import org.slf4j.Logger;
+import org.slf4j.LoggerFactory;
+
+/**
+ * Implementation detail of {@link ContextualTextIO.Read}.
+ *
+ * <p>A {@link FileBasedSource} which can decode records delimited by newline characters.
+ *
+ * <p>This source splits the data into records using {@code UTF-8} {@code \n}, {@code \r}, or {@code
+ * \r\n} as the delimiter. This source is not strict and supports decoding the last record even if
+ * it is not delimited. Finally, no records are decoded if the stream is empty.
+ *
+ * <p>This source supports reading from any arbitrary byte position within the stream. If the
+ * starting position is not {@code 0}, then bytes are skipped until the first delimiter is found
+ * representing the beginning of the first record to be decoded.
+ */
+@VisibleForTesting
+class ContextualTextIOSource extends FileBasedSource<RecordWithMetadata> {
Review comment:
Thanks, good to know. We can look into adding a commit here to use SDF. cc: @rezarokni FYI.
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
[GitHub] [beam] codecov[bot] edited a comment on pull request #12645: [BEAM-10124] Add ContextualTextIO
Posted by GitBox <gi...@apache.org>.
codecov[bot] edited a comment on pull request #12645:
URL: https://github.com/apache/beam/pull/12645#issuecomment-688630083
# [Codecov](https://codecov.io/gh/apache/beam/pull/12645?src=pr&el=h1) Report
> :exclamation: No coverage uploaded for pull request base (`master@2b2b8e7`). [Click here to learn what that means](https://docs.codecov.io/docs/error-reference#section-missing-base-commit).
> The diff coverage is `n/a`.
[](https://codecov.io/gh/apache/beam/pull/12645?src=pr&el=tree)
```diff
@@ Coverage Diff @@
## master #12645 +/- ##
=========================================
Coverage ? 82.33%
=========================================
Files ? 453
Lines ? 54054
Branches ? 0
=========================================
Hits ? 44506
Misses ? 9548
Partials ? 0
```
| [Impacted Files](https://codecov.io/gh/apache/beam/pull/12645?src=pr&el=tree) | Coverage Δ | |
|---|---|---|
| [sdks/python/apache\_beam/coders/\_\_init\_\_.py](https://codecov.io/gh/apache/beam/pull/12645/diff?src=pr&el=tree#diff-c2Rrcy9weXRob24vYXBhY2hlX2JlYW0vY29kZXJzL19faW5pdF9fLnB5) | `100.00% <0.00%> (ø)` | |
| [...m/runners/portability/spark\_uber\_jar\_job\_server.py](https://codecov.io/gh/apache/beam/pull/12645/diff?src=pr&el=tree#diff-c2Rrcy9weXRob24vYXBhY2hlX2JlYW0vcnVubmVycy9wb3J0YWJpbGl0eS9zcGFya191YmVyX2phcl9qb2Jfc2VydmVyLnB5) | `85.60% <0.00%> (ø)` | |
| [...am/examples/snippets/transforms/aggregation/sum.py](https://codecov.io/gh/apache/beam/pull/12645/diff?src=pr&el=tree#diff-c2Rrcy9weXRob24vYXBhY2hlX2JlYW0vZXhhbXBsZXMvc25pcHBldHMvdHJhbnNmb3Jtcy9hZ2dyZWdhdGlvbi9zdW0ucHk=) | `100.00% <0.00%> (ø)` | |
| [...eam/testing/benchmarks/nexmark/nexmark\_launcher.py](https://codecov.io/gh/apache/beam/pull/12645/diff?src=pr&el=tree#diff-c2Rrcy9weXRob24vYXBhY2hlX2JlYW0vdGVzdGluZy9iZW5jaG1hcmtzL25leG1hcmsvbmV4bWFya19sYXVuY2hlci5weQ==) | `0.00% <0.00%> (ø)` | |
| [sdks/python/apache\_beam/io/gcp/bigquery.py](https://codecov.io/gh/apache/beam/pull/12645/diff?src=pr&el=tree#diff-c2Rrcy9weXRob24vYXBhY2hlX2JlYW0vaW8vZ2NwL2JpZ3F1ZXJ5LnB5) | `79.78% <0.00%> (ø)` | |
| [...ache\_beam/io/gcp/datastore/v1new/query\_splitter.py](https://codecov.io/gh/apache/beam/pull/12645/diff?src=pr&el=tree#diff-c2Rrcy9weXRob24vYXBhY2hlX2JlYW0vaW8vZ2NwL2RhdGFzdG9yZS92MW5ldy9xdWVyeV9zcGxpdHRlci5weQ==) | `94.11% <0.00%> (ø)` | |
| [...eam/portability/api/beam\_expansion\_api\_pb2\_grpc.py](https://codecov.io/gh/apache/beam/pull/12645/diff?src=pr&el=tree#diff-c2Rrcy9weXRob24vYXBhY2hlX2JlYW0vcG9ydGFiaWxpdHkvYXBpL2JlYW1fZXhwYW5zaW9uX2FwaV9wYjJfZ3JwYy5weQ==) | `61.90% <0.00%> (ø)` | |
| [...n/apache\_beam/typehints/typed\_pipeline\_test\_py3.py](https://codecov.io/gh/apache/beam/pull/12645/diff?src=pr&el=tree#diff-c2Rrcy9weXRob24vYXBhY2hlX2JlYW0vdHlwZWhpbnRzL3R5cGVkX3BpcGVsaW5lX3Rlc3RfcHkzLnB5) | `90.30% <0.00%> (ø)` | |
| [...s/snippets/transforms/aggregation/combinevalues.py](https://codecov.io/gh/apache/beam/pull/12645/diff?src=pr&el=tree#diff-c2Rrcy9weXRob24vYXBhY2hlX2JlYW0vZXhhbXBsZXMvc25pcHBldHMvdHJhbnNmb3Jtcy9hZ2dyZWdhdGlvbi9jb21iaW5ldmFsdWVzLnB5) | `94.73% <0.00%> (ø)` | |
| [...examples/snippets/transforms/elementwise/values.py](https://codecov.io/gh/apache/beam/pull/12645/diff?src=pr&el=tree#diff-c2Rrcy9weXRob24vYXBhY2hlX2JlYW0vZXhhbXBsZXMvc25pcHBldHMvdHJhbnNmb3Jtcy9lbGVtZW50d2lzZS92YWx1ZXMucHk=) | `100.00% <0.00%> (ø)` | |
| ... and [443 more](https://codecov.io/gh/apache/beam/pull/12645/diff?src=pr&el=tree-more) | |
------
[Continue to review full report at Codecov](https://codecov.io/gh/apache/beam/pull/12645?src=pr&el=continue).
> **Legend** - [Click here to learn more](https://docs.codecov.io/docs/codecov-delta)
> `Δ = absolute <relative> (impact)`, `ø = not affected`, `? = missing data`
> Powered by [Codecov](https://codecov.io/gh/apache/beam/pull/12645?src=pr&el=footer). Last update [2b2b8e7...3bed6b7](https://codecov.io/gh/apache/beam/pull/12645?src=pr&el=lastupdated). Read the [comment docs](https://docs.codecov.io/docs/pull-request-comments).
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
[GitHub] [beam] codecov[bot] edited a comment on pull request #12645: [BEAM-10124] Add ContextualTextIO
Posted by GitBox <gi...@apache.org>.
codecov[bot] edited a comment on pull request #12645:
URL: https://github.com/apache/beam/pull/12645#issuecomment-688630083
# [Codecov](https://codecov.io/gh/apache/beam/pull/12645?src=pr&el=h1) Report
> :exclamation: No coverage uploaded for pull request base (`master@2b2b8e7`). [Click here to learn what that means](https://docs.codecov.io/docs/error-reference#section-missing-base-commit).
> The diff coverage is `n/a`.
[](https://codecov.io/gh/apache/beam/pull/12645?src=pr&el=tree)
```diff
@@ Coverage Diff @@
## master #12645 +/- ##
=========================================
Coverage ? 82.33%
=========================================
Files ? 453
Lines ? 54054
Branches ? 0
=========================================
Hits ? 44507
Misses ? 9547
Partials ? 0
```
| [Impacted Files](https://codecov.io/gh/apache/beam/pull/12645?src=pr&el=tree) | Coverage Δ | |
|---|---|---|
| [sdks/python/apache\_beam/version.py](https://codecov.io/gh/apache/beam/pull/12645/diff?src=pr&el=tree#diff-c2Rrcy9weXRob24vYXBhY2hlX2JlYW0vdmVyc2lvbi5weQ==) | `100.00% <0.00%> (ø)` | |
| [sdks/python/apache\_beam/transforms/\_\_init\_\_.py](https://codecov.io/gh/apache/beam/pull/12645/diff?src=pr&el=tree#diff-c2Rrcy9weXRob24vYXBhY2hlX2JlYW0vdHJhbnNmb3Jtcy9fX2luaXRfXy5weQ==) | `100.00% <0.00%> (ø)` | |
| [...s/python/apache\_beam/examples/wordcount\_minimal.py](https://codecov.io/gh/apache/beam/pull/12645/diff?src=pr&el=tree#diff-c2Rrcy9weXRob24vYXBhY2hlX2JlYW0vZXhhbXBsZXMvd29yZGNvdW50X21pbmltYWwucHk=) | `93.33% <0.00%> (ø)` | |
| [...on/apache\_beam/portability/api/metrics\_pb2\_grpc.py](https://codecov.io/gh/apache/beam/pull/12645/diff?src=pr&el=tree#diff-c2Rrcy9weXRob24vYXBhY2hlX2JlYW0vcG9ydGFiaWxpdHkvYXBpL21ldHJpY3NfcGIyX2dycGMucHk=) | `0.00% <0.00%> (ø)` | |
| [...ache\_beam/examples/cookbook/datastore\_wordcount.py](https://codecov.io/gh/apache/beam/pull/12645/diff?src=pr&el=tree#diff-c2Rrcy9weXRob24vYXBhY2hlX2JlYW0vZXhhbXBsZXMvY29va2Jvb2svZGF0YXN0b3JlX3dvcmRjb3VudC5weQ==) | `32.72% <0.00%> (ø)` | |
| [...dks/python/apache\_beam/transforms/external\_java.py](https://codecov.io/gh/apache/beam/pull/12645/diff?src=pr&el=tree#diff-c2Rrcy9weXRob24vYXBhY2hlX2JlYW0vdHJhbnNmb3Jtcy9leHRlcm5hbF9qYXZhLnB5) | `0.00% <0.00%> (ø)` | |
| [...pache\_beam/typehints/trivial\_inference\_test\_py3.py](https://codecov.io/gh/apache/beam/pull/12645/diff?src=pr&el=tree#diff-c2Rrcy9weXRob24vYXBhY2hlX2JlYW0vdHlwZWhpbnRzL3RyaXZpYWxfaW5mZXJlbmNlX3Rlc3RfcHkzLnB5) | `92.85% <0.00%> (ø)` | |
| [...on/apache\_beam/runners/direct/watermark\_manager.py](https://codecov.io/gh/apache/beam/pull/12645/diff?src=pr&el=tree#diff-c2Rrcy9weXRob24vYXBhY2hlX2JlYW0vcnVubmVycy9kaXJlY3Qvd2F0ZXJtYXJrX21hbmFnZXIucHk=) | `97.17% <0.00%> (ø)` | |
| [sdks/python/apache\_beam/internal/pickler.py](https://codecov.io/gh/apache/beam/pull/12645/diff?src=pr&el=tree#diff-c2Rrcy9weXRob24vYXBhY2hlX2JlYW0vaW50ZXJuYWwvcGlja2xlci5weQ==) | `85.82% <0.00%> (ø)` | |
| [sdks/python/apache\_beam/internal/\_\_init\_\_.py](https://codecov.io/gh/apache/beam/pull/12645/diff?src=pr&el=tree#diff-c2Rrcy9weXRob24vYXBhY2hlX2JlYW0vaW50ZXJuYWwvX19pbml0X18ucHk=) | `100.00% <0.00%> (ø)` | |
| ... and [443 more](https://codecov.io/gh/apache/beam/pull/12645/diff?src=pr&el=tree-more) | |
------
[Continue to review full report at Codecov](https://codecov.io/gh/apache/beam/pull/12645?src=pr&el=continue).
> **Legend** - [Click here to learn more](https://docs.codecov.io/docs/codecov-delta)
> `Δ = absolute <relative> (impact)`, `ø = not affected`, `? = missing data`
> Powered by [Codecov](https://codecov.io/gh/apache/beam/pull/12645?src=pr&el=footer). Last update [2b2b8e7...85c579f](https://codecov.io/gh/apache/beam/pull/12645?src=pr&el=lastupdated). Read the [comment docs](https://docs.codecov.io/docs/pull-request-comments).
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
[GitHub] [beam] abhiy13 commented on a change in pull request #12645: [BEAM-10124] Add ContextualTextIO
Posted by GitBox <gi...@apache.org>.
abhiy13 commented on a change in pull request #12645:
URL: https://github.com/apache/beam/pull/12645#discussion_r492475231
##########
File path: sdks/java/io/contextual-text-io/build.gradle
##########
@@ -0,0 +1,41 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one
+ * or more contributor license agreements. See the NOTICE file
+ * distributed with this work for additional information
+ * regarding copyright ownership. The ASF licenses this file
+ * to you under the Apache License, Version 2.0 (the
+ * License); you may not use this file except in compliance
+ * with the License. You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an AS IS BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+plugins { id 'org.apache.beam.module' }
+applyJavaNature(
+ automaticModuleName: 'org.apache.beam.sdk.io.contextual-text-io',
+ enableChecker: false,
+ ignoreRawtypeErrors: true)
+
+description = "Apache Beam :: SDKs :: Java :: Contextual-Text-IO"
+ext.summary = "Context-aware Text IO."
+
+dependencies {
+
+ compile library.java.vendored_guava_26_0_jre
+ compile library.java.protobuf_java
+ compile project(path: ":sdks:java:core", configuration: "shadow")
+ testCompile project(path: ":sdks:java:core", configuration: "shadowTest")
+
+ testCompile library.java.guava_testlib
+ testCompile library.java.junit
+ testCompile library.java.hamcrest_core
+ testRuntimeOnly library.java.slf4j_jdk14
+ testCompile project(path: ":runners:direct-java", configuration: "shadow")
+
+}
Review comment:
Done
##########
File path: sdks/java/io/contextual-text-io/build.gradle
##########
@@ -0,0 +1,41 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one
+ * or more contributor license agreements. See the NOTICE file
+ * distributed with this work for additional information
+ * regarding copyright ownership. The ASF licenses this file
+ * to you under the Apache License, Version 2.0 (the
+ * License); you may not use this file except in compliance
+ * with the License. You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an AS IS BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+plugins { id 'org.apache.beam.module' }
+applyJavaNature(
+ automaticModuleName: 'org.apache.beam.sdk.io.contextual-text-io',
+ enableChecker: false,
+ ignoreRawtypeErrors: true)
+
+description = "Apache Beam :: SDKs :: Java :: Contextual-Text-IO"
+ext.summary = "Context-aware Text IO."
+
+dependencies {
+
Review comment:
Done
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
[GitHub] [beam] lukecwik commented on a change in pull request #12645: [BEAM-10124] Add ContextualTextIO
Posted by GitBox <gi...@apache.org>.
lukecwik commented on a change in pull request #12645:
URL: https://github.com/apache/beam/pull/12645#discussion_r477495350
##########
File path: sdks/java/io/contextual-text-io/src/main/java/org/apache/beam/sdk/io/contextualtextio/ContextualTextIOSource.java
##########
@@ -0,0 +1,364 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one
+ * or more contributor license agreements. See the NOTICE file
+ * distributed with this work for additional information
+ * regarding copyright ownership. The ASF licenses this file
+ * to you under the Apache License, Version 2.0 (the
+ * "License"); you may not use this file except in compliance
+ * with the License. You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+package org.apache.beam.sdk.io.contextualtextio;
+
+import java.io.IOException;
+import java.nio.ByteBuffer;
+import java.nio.channels.ReadableByteChannel;
+import java.nio.channels.SeekableByteChannel;
+import java.util.NoSuchElementException;
+import org.apache.beam.sdk.coders.Coder;
+import org.apache.beam.sdk.io.FileBasedSource;
+import org.apache.beam.sdk.io.fs.EmptyMatchTreatment;
+import org.apache.beam.sdk.io.fs.MatchResult;
+import org.apache.beam.sdk.options.PipelineOptions;
+import org.apache.beam.sdk.options.ValueProvider;
+import org.apache.beam.sdk.schemas.NoSuchSchemaException;
+import org.apache.beam.sdk.schemas.SchemaCoder;
+import org.apache.beam.sdk.schemas.SchemaRegistry;
+import org.apache.beam.vendor.grpc.v1p26p0.com.google.protobuf.ByteString;
+import org.apache.beam.vendor.guava.v26_0_jre.com.google.common.annotations.VisibleForTesting;
+import org.apache.beam.vendor.guava.v26_0_jre.com.google.common.base.Preconditions;
+import org.checkerframework.checker.nullness.qual.Nullable;
+import org.slf4j.Logger;
+import org.slf4j.LoggerFactory;
+
+/**
+ * Implementation detail of {@link ContextualTextIO.Read}.
+ *
+ * <p>A {@link FileBasedSource} which can decode records delimited by newline characters.
+ *
+ * <p>This source splits the data into records using {@code UTF-8} {@code \n}, {@code \r}, or {@code
+ * \r\n} as the delimiter. This source is not strict and supports decoding the last record even if
+ * it is not delimited. Finally, no records are decoded if the stream is empty.
+ *
+ * <p>This source supports reading from any arbitrary byte position within the stream. If the
+ * starting position is not {@code 0}, then bytes are skipped until the first delimiter is found
+ * representing the beginning of the first record to be decoded.
+ */
+@VisibleForTesting
+class ContextualTextIOSource extends FileBasedSource<RecordWithMetadata> {
Review comment:
Swapping will likely lead to be update incompatible for runners supporting update so we will have to live with both versions .
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
[GitHub] [beam] lukecwik commented on a change in pull request #12645: [BEAM-10124] Add ContextualTextIO
Posted by GitBox <gi...@apache.org>.
lukecwik commented on a change in pull request #12645:
URL: https://github.com/apache/beam/pull/12645#discussion_r477463132
##########
File path: sdks/java/io/contextual-text-io/src/main/java/org/apache/beam/sdk/io/contextualtextio/RecordWithMetadata.java
##########
@@ -0,0 +1,84 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one
+ * or more contributor license agreements. See the NOTICE file
+ * distributed with this work for additional information
+ * regarding copyright ownership. The ASF licenses this file
+ * to you under the Apache License, Version 2.0 (the
+ * "License"); you may not use this file except in compliance
+ * with the License. You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+package org.apache.beam.sdk.io.contextualtextio;
+
+import com.google.auto.value.AutoValue;
+import org.apache.beam.sdk.annotations.Experimental;
+import org.apache.beam.sdk.schemas.AutoValueSchema;
+import org.apache.beam.sdk.schemas.annotations.DefaultSchema;
+
+/**
+ * Helper Class based on {@link AutoValueSchema}, it provides Metadata associated with each Record
+ * when reading from file(s) using {@link ContextualTextIO}.
+ *
+ * <h3>Fields:</h3>
+ *
+ * <ul>
+ * <li>recordOffset: The offset of a record (the byte at which the record begins) in a file. This
+ * information can be useful if you wish to reconstruct the file. {@link
+ * RecordWithMetadata#getRecordOffset()}
+ * <li>recordNum: The ordinal number of the record in its file. {@link
+ * RecordWithMetadata#getRecordNum()}
+ * <li>recordValue: The value / contents of the record {@link RecordWithMetadata#getRecordValue()}
+ * <li>rangeOffset: The starting offset of the range (split), which contained the record, when the
+ * record was read. {@link RecordWithMetadata#getRangeOffset()}
+ * <li>recordNumInOffset: The record number relative to the Range. (line number within the range)
+ * {@link RecordWithMetadata#getRecordNumInOffset()}
+ * <li>fileName: Name of the file to which the record belongs (this is the full filename,
+ * eg:path/to/file.txt) {@link RecordWithMetadata#getFileName()}
+ * </ul>
+ */
+@Experimental(Experimental.Kind.SCHEMAS)
+@DefaultSchema(AutoValueSchema.class)
+@AutoValue
+public abstract class RecordWithMetadata {
Review comment:
Should this be `RecordWithMetadata<T>`?
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
[GitHub] [beam] lukecwik commented on a change in pull request #12645: [BEAM-10124] Add ContextualTextIO
Posted by GitBox <gi...@apache.org>.
lukecwik commented on a change in pull request #12645:
URL: https://github.com/apache/beam/pull/12645#discussion_r477495350
##########
File path: sdks/java/io/contextual-text-io/src/main/java/org/apache/beam/sdk/io/contextualtextio/ContextualTextIOSource.java
##########
@@ -0,0 +1,364 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one
+ * or more contributor license agreements. See the NOTICE file
+ * distributed with this work for additional information
+ * regarding copyright ownership. The ASF licenses this file
+ * to you under the Apache License, Version 2.0 (the
+ * "License"); you may not use this file except in compliance
+ * with the License. You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+package org.apache.beam.sdk.io.contextualtextio;
+
+import java.io.IOException;
+import java.nio.ByteBuffer;
+import java.nio.channels.ReadableByteChannel;
+import java.nio.channels.SeekableByteChannel;
+import java.util.NoSuchElementException;
+import org.apache.beam.sdk.coders.Coder;
+import org.apache.beam.sdk.io.FileBasedSource;
+import org.apache.beam.sdk.io.fs.EmptyMatchTreatment;
+import org.apache.beam.sdk.io.fs.MatchResult;
+import org.apache.beam.sdk.options.PipelineOptions;
+import org.apache.beam.sdk.options.ValueProvider;
+import org.apache.beam.sdk.schemas.NoSuchSchemaException;
+import org.apache.beam.sdk.schemas.SchemaCoder;
+import org.apache.beam.sdk.schemas.SchemaRegistry;
+import org.apache.beam.vendor.grpc.v1p26p0.com.google.protobuf.ByteString;
+import org.apache.beam.vendor.guava.v26_0_jre.com.google.common.annotations.VisibleForTesting;
+import org.apache.beam.vendor.guava.v26_0_jre.com.google.common.base.Preconditions;
+import org.checkerframework.checker.nullness.qual.Nullable;
+import org.slf4j.Logger;
+import org.slf4j.LoggerFactory;
+
+/**
+ * Implementation detail of {@link ContextualTextIO.Read}.
+ *
+ * <p>A {@link FileBasedSource} which can decode records delimited by newline characters.
+ *
+ * <p>This source splits the data into records using {@code UTF-8} {@code \n}, {@code \r}, or {@code
+ * \r\n} as the delimiter. This source is not strict and supports decoding the last record even if
+ * it is not delimited. Finally, no records are decoded if the stream is empty.
+ *
+ * <p>This source supports reading from any arbitrary byte position within the stream. If the
+ * starting position is not {@code 0}, then bytes are skipped until the first delimiter is found
+ * representing the beginning of the first record to be decoded.
+ */
+@VisibleForTesting
+class ContextualTextIOSource extends FileBasedSource<RecordWithMetadata> {
Review comment:
Swapping will be update incompatible for runners supporting update so we will have to live with both versions .
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
[GitHub] [beam] abhiy13 commented on a change in pull request #12645: [BEAM-10124] Add ContextualTextIO
Posted by GitBox <gi...@apache.org>.
abhiy13 commented on a change in pull request #12645:
URL: https://github.com/apache/beam/pull/12645#discussion_r492474778
##########
File path: sdks/java/io/contextual-text-io/src/main/java/org/apache/beam/sdk/io/contextualtextio/ContextualTextIO.java
##########
@@ -0,0 +1,631 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one
+ * or more contributor license agreements. See the NOTICE file
+ * distributed with this work for additional information
+ * regarding copyright ownership. The ASF licenses this file
+ * to you under the Apache License, Version 2.0 (the
+ * "License"); you may not use this file except in compliance
+ * with the License. You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+package org.apache.beam.sdk.io.contextualtextio;
+
+import static org.apache.beam.sdk.io.FileIO.ReadMatches.DirectoryTreatment;
+import static org.apache.beam.vendor.guava.v26_0_jre.com.google.common.base.Preconditions.checkArgument;
+import static org.apache.beam.vendor.guava.v26_0_jre.com.google.common.base.Preconditions.checkNotNull;
+
+import com.google.auto.value.AutoValue;
+import java.nio.ByteBuffer;
+import java.util.Arrays;
+import java.util.Comparator;
+import java.util.HashMap;
+import java.util.Map;
+import java.util.SortedMap;
+import java.util.TreeMap;
+import org.apache.beam.sdk.coders.StringUtf8Coder;
+import org.apache.beam.sdk.io.CompressedSource;
+import org.apache.beam.sdk.io.Compression;
+import org.apache.beam.sdk.io.FileBasedSource;
+import org.apache.beam.sdk.io.FileIO;
+import org.apache.beam.sdk.io.FileIO.MatchConfiguration;
+import org.apache.beam.sdk.io.ReadAllViaFileBasedSource;
+import org.apache.beam.sdk.io.TextIO;
+import org.apache.beam.sdk.io.fs.EmptyMatchTreatment;
+import org.apache.beam.sdk.options.ValueProvider;
+import org.apache.beam.sdk.options.ValueProvider.StaticValueProvider;
+import org.apache.beam.sdk.schemas.NoSuchSchemaException;
+import org.apache.beam.sdk.schemas.SchemaCoder;
+import org.apache.beam.sdk.transforms.Count;
+import org.apache.beam.sdk.transforms.Create;
+import org.apache.beam.sdk.transforms.DoFn;
+import org.apache.beam.sdk.transforms.PTransform;
+import org.apache.beam.sdk.transforms.ParDo;
+import org.apache.beam.sdk.transforms.SerializableFunction;
+import org.apache.beam.sdk.transforms.View;
+import org.apache.beam.sdk.transforms.Watch.Growth.TerminationCondition;
+import org.apache.beam.sdk.transforms.display.DisplayData;
+import org.apache.beam.sdk.values.KV;
+import org.apache.beam.sdk.values.PBegin;
+import org.apache.beam.sdk.values.PCollection;
+import org.apache.beam.sdk.values.PCollectionView;
+import org.apache.beam.vendor.guava.v26_0_jre.com.google.common.annotations.VisibleForTesting;
+import org.checkerframework.checker.nullness.qual.Nullable;
+import org.joda.time.Duration;
+import org.slf4j.Logger;
+import org.slf4j.LoggerFactory;
+
+/**
+ * {@link PTransform}s that read text files and collect contextual information of the elements in
+ * the input.
+ *
+ * <p>Use {@link TextIO} when not reading file with Multiline Records or additional metadata is not
+ * required.
+ *
+ * <h2>Reading from text files</h2>
+ *
+ * <p>To read a {@link PCollection} from one or more text files, use {@code
+ * ContextualTextIO.read()}. To instantiate a transform use {@link
+ * ContextualTextIO.Read#from(String)} and specify the path of the file(s) to be read.
+ * Alternatively, if the filenames to be read are themselves in a {@link PCollection} you can use
+ * {@link FileIO} to match them and {@link ContextualTextIO#readFiles()} to read them.
+ *
+ * <p>{@link #read} returns a {@link PCollection} of {@link RecordWithMetadata RecordWithMetadata},
+ * each corresponding to one line of an input UTF-8 text file (split into lines delimited by '\n',
+ * '\r', '\r\n', or specified delimiter see {@link ContextualTextIO.Read#withDelimiter})
+ *
+ * <h3>Filepattern expansion and watching</h3>
+ *
+ * <p>By default, the filepatterns are expanded only once. The combination of {@link
+ * FileIO.Match#continuously(Duration, TerminationCondition)} and {@link #readFiles()} allow
+ * streaming of new files matching the filepattern(s).
+ *
+ * <p>By default, {@link #read} prohibits filepatterns that match no files, and {@link #readFiles()}
+ * allows them in case the filepattern contains a glob wildcard character. Use {@link
+ * ContextualTextIO.Read#withEmptyMatchTreatment} or {@link
+ * FileIO.Match#withEmptyMatchTreatment(EmptyMatchTreatment)} plus {@link #readFiles()} to configure
+ * this behavior.
+ *
+ * <p>Example 1: reading a file or filepattern.
+ *
+ * <pre>{@code
+ * Pipeline p = ...;
+ *
+ * // A simple Read of a file:
+ * PCollection<RecordWithMetadata> records = p.apply(ContextualTextIO.read().from("/local/path/to/file.txt"));
+ * }</pre>
+ *
+ * <p>Example 2: reading a PCollection of filenames.
+ *
+ * <pre>{@code
+ * Pipeline p = ...;
+ *
+ * // E.g. the filenames might be computed from other data in the pipeline, or
+ * // read from a data source.
+ * PCollection<String> filenames = ...;
+ *
+ * // Read all files in the collection.
+ * PCollection<RecordWithMetadata> records =
+ * filenames
+ * .apply(FileIO.matchAll())
+ * .apply(FileIO.readMatches())
+ * .apply(ContextualTextIO.readFiles());
+ * }</pre>
+ *
+ * <p>Example 3: streaming new files matching a filepattern.
+ *
+ * <pre>{@code
+ * Pipeline p = ...;
+ *
+ * PCollection<RecordWithMetadata> records = p.apply(ContextualTextIO.read()
+ * .from("/local/path/to/files/*")
+ * .watchForNewFiles(
+ * // Check for new files every minute
+ * Duration.standardMinutes(1),
+ * // Stop watching the filepattern if no new files appear within an hour
+ * afterTimeSinceNewOutput(Duration.standardHours(1))));
+ * }</pre>
+ *
+ * <p>Example 4: reading a file or file pattern of RFC4180-compliant CSV files with fields that may
+ * contain line breaks.
+ *
+ * Example of such a file could be:
+ *
+ * "aaa","b CRLF
+ * bb","ccc" CRLF
+ * zzz,yyy,xxx
+ *
+ * <pre>{@code
+ * Pipeline p = ...;
+ *
+ * PCollection<RecordWithMetadata> records = p.apply(ContextualTextIO.read()
+ * .from("/local/path/to/files/*.csv")
+ * .withHasMultilineCSVRecords(true));
+ * }</pre>
+ *
+ * <p>Example 5: reading while watching for new files
+ *
+ * <pre>{@code
+ * Pipeline p = ...;
+ *
+ * PCollection<RecordWithMetadata> records = p.apply(FileIO.match()
+ * .filepattern("filepattern")
+ * .continuously(
+ * Duration.millis(100),
+ * Watch.Growth.afterTimeSinceNewOutput(Duration.standardSeconds(3))))
+ * .apply(FileIO.readMatches())
+ * .apply(ContextualTextIO.readFiles());
+ * }</pre>
+ *
+ * <p>Example 6: reading without recordNum metadata, or only fileName associated Metadata. (the
+ * Objects would still contain recordNums, but these recordNums would correspond to their positions
+ * in their respective offsets rather than their positions within the entire file).
+ *
+ * <pre>{@code
+ * Pipeline p = ...;
+ *
+ * PCollection<RecordWithMetadata> records = p.apply(ContextualTextIO.read()
+ * .from("/local/path/to/files/*.csv")
+ * .setWithoutRecordNumMetadata(true));
+ * }</pre>
+ *
+ * <p>NOTE: When using {@link ContextualTextIO.Read#withHasMultilineCSVRecords(Boolean)} this
+ * option, a single reader will be used to process the file, rather than multiple readers which can
+ * read from different offsets. For a large file this can result in lower performance.
+ *
+ * <p>NOTE: Use {@link Read#withoutRecordNumMetadata()} when recordNum metadata is not required, for
+ * example, when when only filename metadata is required. Computing record positions currently
+ * introduces a shuffle step, which increases the resources used by the pipeline. <b> By default
+ * withoutRecordNumMetadata is set to false, so the shuffle step is performed.</b>
+ *
+ * <h3>Reading a very large number of files</h3>
+ *
+ * <p>If it is known that the filepattern will match a very large number of files (e.g. tens of
+ * thousands or more), use {@link ContextualTextIO.Read#withHintMatchesManyFiles} for better
+ * performance and scalability. Note that it may decrease performance if the filepattern matches
+ * only a small number of files.
+ */
+public class ContextualTextIO {
+ private static final long DEFAULT_BUNDLE_SIZE_BYTES = 64 * 1024 * 1024L;
+ private static final Logger LOG = LoggerFactory.getLogger(ContextualTextIO.class);
+
+ /**
+ * A {@link PTransform} that reads from one or more text files and returns a bounded {@link
+ * PCollection} containing one {@link RecordWithMetadata}element for each line of the input files.
+ */
+ public static Read read() {
+ return new AutoValue_ContextualTextIO_Read.Builder()
+ .setCompression(Compression.AUTO)
+ .setHintMatchesManyFiles(false)
+ .setWithoutRecordNumMetadata(false)
+ .setMatchConfiguration(MatchConfiguration.create(EmptyMatchTreatment.DISALLOW))
+ .setHasMultilineCSVRecords(false)
+ .build();
+ }
+
+ /**
+ * Like {@link #read}, but reads each file in a {@link PCollection} of {@link
+ * FileIO.ReadableFile}, returned by {@link FileIO#readMatches}.
+ */
+ public static ReadFiles readFiles() {
+ return new AutoValue_ContextualTextIO_ReadFiles.Builder()
+ // 64MB is a reasonable value that allows to amortize the cost of opening files,
+ // but is not so large as to exhaust a typical runner's maximum amount of output per
+ // ProcessElement call.
+ .setDesiredBundleSizeBytes(DEFAULT_BUNDLE_SIZE_BYTES)
+ .setHasMultilineCSVRecords(false)
+ .build();
+ }
+
+ /** Implementation of {@link #read}. */
+ @AutoValue
+ public abstract static class Read extends PTransform<PBegin, PCollection<RecordWithMetadata>> {
+ abstract @Nullable ValueProvider<String> getFilepattern();
+
+ abstract MatchConfiguration getMatchConfiguration();
+
+ abstract boolean getHintMatchesManyFiles();
+
+ abstract boolean getWithoutRecordNumMetadata();
+
+ abstract Compression getCompression();
+
+ abstract @Nullable Boolean getHasMultilineCSVRecords();
+
+ @SuppressWarnings("mutable") // this returns an array that can be mutated by the caller
+ abstract byte @Nullable [] getDelimiter();
+
+ abstract Builder toBuilder();
+
+ @AutoValue.Builder
+ abstract static class Builder {
+ abstract Builder setFilepattern(ValueProvider<String> filepattern);
+
+ abstract Builder setMatchConfiguration(MatchConfiguration matchConfiguration);
+
+ abstract Builder setHintMatchesManyFiles(boolean hintManyFiles);
+
+ abstract Builder setWithoutRecordNumMetadata(boolean withoutLineNumMetadata);
+
+ abstract Builder setCompression(Compression compression);
+
+ abstract Builder setDelimiter(byte @Nullable [] delimiter);
+
+ abstract Builder setHasMultilineCSVRecords(Boolean hasMultilineCSVRecords);
+
+ abstract Read build();
+ }
+
+ /**
+ * Reads text from the file(s) with the given filename or filename pattern.
+ *
+ * <p>This can be a local path (if running locally), or a Google Cloud Storage filename or
+ * filename pattern of the form {@code "gs://<bucket>/<filepath>"} (if running locally or using
+ * remote execution service).
+ *
+ * <p>Standard <a href="http://docs.oracle.com/javase/tutorial/essential/io/find.html" >Java
+ * Filesystem glob patterns</a> ("*", "?", "[..]") are supported.
+ *
+ * <p>If it is known that the filepattern will match a very large number of files (at least tens
+ * of thousands), use {@link #withHintMatchesManyFiles} for better performance and scalability.
+ */
+ public Read from(String filepattern) {
+ checkArgument(filepattern != null, "filepattern can not be null");
+ return from(StaticValueProvider.of(filepattern));
+ }
+
+ /** Same as {@code from(filepattern)}, but accepting a {@link ValueProvider}. */
+ public Read from(ValueProvider<String> filepattern) {
+ checkArgument(filepattern != null, "filepattern can not be null");
+ return toBuilder().setFilepattern(filepattern).build();
+ }
+
+ /** Sets the {@link MatchConfiguration}. */
+ public Read withMatchConfiguration(MatchConfiguration matchConfiguration) {
+ return toBuilder().setMatchConfiguration(matchConfiguration).build();
+ }
+
+ /**
+ * When reading RFC4180 CSV files that have values that span multiple lines, set this to true.
+ * Note: this reduces the read performance (see: {@link ContextualTextIO}).
+ */
+ public Read withHasMultilineCSVRecords(Boolean hasMultilineCSVRecords) {
+ return toBuilder().setHasMultilineCSVRecords(hasMultilineCSVRecords).build();
+ }
+
+ /**
+ * Reads from input sources using the specified compression type.
+ *
+ * <p>If no compression type is specified, the default is {@link Compression#AUTO}.
+ */
+ public Read withCompression(Compression compression) {
+ return toBuilder().setCompression(compression).build();
+ }
+
+ /**
+ * Hints that the filepattern specified in {@link #from(String)} matches a very large number of
+ * files.
+ *
+ * <p>This hint may cause a runner to execute the transform differently, in a way that improves
+ * performance for this case, but it may worsen performance if the filepattern matches only a
+ * small number of files (e.g., in a runner that supports dynamic work rebalancing, it will
+ * happen less efficiently within individual files).
+ */
+ public Read withHintMatchesManyFiles() {
+ return toBuilder().setHintMatchesManyFiles(true).build();
+ }
+
+ /**
+ * Allows the user to opt out of getting recordNums associated with each record.
+ *
+ * <p>When set to true, it will introduce a shuffle step to assemble the recordNums for each
+ * record, which will increase the resources used by the pipeline.
+ *
+ * <p>Use this when metadata like fileNames are required and their position/order can be
+ * ignored.
+ */
+ public Read withoutRecordNumMetadata() {
+ return toBuilder().setWithoutRecordNumMetadata(true).build();
+ }
+
+ /** See {@link MatchConfiguration#withEmptyMatchTreatment}. */
+ public Read withEmptyMatchTreatment(EmptyMatchTreatment treatment) {
+ return withMatchConfiguration(getMatchConfiguration().withEmptyMatchTreatment(treatment));
+ }
+
+ /** Set the custom delimiter to be used in place of the default ones ('\r', '\n' or '\r\n'). */
+ public Read withDelimiter(byte[] delimiter) {
+ checkArgument(delimiter != null, "delimiter can not be null");
+ checkArgument(!isSelfOverlapping(delimiter), "delimiter must not self-overlap");
+ return toBuilder().setDelimiter(delimiter).build();
+ }
+
+ static boolean isSelfOverlapping(byte[] s) {
+ // s self-overlaps if v exists such as s = vu = wv with u and w non empty
+ for (int i = 1; i < s.length - 1; ++i) {
+ if (ByteBuffer.wrap(s, 0, i).equals(ByteBuffer.wrap(s, s.length - i, i))) {
+ return true;
+ }
+ }
+ return false;
+ }
+
+ @Override
+ public PCollection<RecordWithMetadata> expand(PBegin input) {
+ checkNotNull(
+ getFilepattern(), "need to set the filepattern of a ContextualTextIO.Read transform");
+ PCollection<RecordWithMetadata> records = null;
+ if (getMatchConfiguration().getWatchInterval() == null && !getHintMatchesManyFiles()) {
+ records = input.apply("Read", org.apache.beam.sdk.io.Read.from(getSource()));
+ } else {
+ // All other cases go through FileIO + ReadFiles
+ records =
+ input
+ .apply(
+ "Create filepattern", Create.ofProvider(getFilepattern(), StringUtf8Coder.of()))
+ .apply("Match All", FileIO.matchAll().withConfiguration(getMatchConfiguration()))
+ .apply(
+ "Read Matches",
+ FileIO.readMatches()
+ .withCompression(getCompression())
+ .withDirectoryTreatment(DirectoryTreatment.PROHIBIT))
+ .apply("Via ReadFiles", readFiles().withDelimiter(getDelimiter()));
+ }
+
+ // Check if the user decided to opt out of recordNums associated with records
+ if (getWithoutRecordNumMetadata()) {
+ return records;
+ }
+
+ /*
+ * At this point the line number in RecordWithMetadata contains the relative line offset from the beginning of the read range.
+ *
+ * To compute the absolute position from the beginning of the input we group the lines within the same ranges, and evaluate the size of each range.
+ */
+
+ PCollection<KV<KV<String, Long>, RecordWithMetadata>> recordsGroupedByFileAndRange =
+ records.apply("AddFileNameAndRange", ParDo.of(new AddFileNameAndRange()));
+
+ PCollectionView<Map<KV<String, Long>, Long>> rangeSizes =
+ recordsGroupedByFileAndRange
+ .apply("CountRecordsForEachFileRange", Count.perKey())
+ .apply("SizesAsView", View.asMap());
+
+ // Get Pipeline to create a dummy PCollection with one element to help compute the lines
+ // before each Range
+ PCollection<Integer> singletonPcoll =
+ input.getPipeline().apply("CreateSingletonPcoll", Create.of(Arrays.asList(1)));
+
+ /*
+ * For each (File, Offset) pair, calculate the number of lines occurring before the Range for each file
+ *
+ * After computing the number of lines before each range, we can find the line number in original file as numLiesBeforeOffset + lineNumInCurrentOffset
+ */
+
+ PCollectionView<Map<KV<String, Long>, Long>> numRecordsBeforeEachRange =
+ singletonPcoll
+ .apply(
+ "ComputeRecordsBeforeRange",
+ ParDo.of(new ComputeRecordsBeforeEachRange(rangeSizes))
+ .withSideInputs(rangeSizes))
+ .apply("NumRecordsBeforeEachRangeAsView", View.asMap());
+
+ return recordsGroupedByFileAndRange.apply(
+ "AssignLineNums",
+ ParDo.of(new AssignRecordNums(numRecordsBeforeEachRange))
+ .withSideInputs(numRecordsBeforeEachRange));
+ }
+
+ @VisibleForTesting
+ static class AddFileNameAndRange
+ extends DoFn<RecordWithMetadata, KV<KV<String, Long>, RecordWithMetadata>> {
+ @ProcessElement
+ public void processElement(
+ @Element RecordWithMetadata record,
+ OutputReceiver<KV<KV<String, Long>, RecordWithMetadata>> out) {
+ out.output(KV.of(KV.of(record.getFileName().toString(), record.getRangeOffset()), record));
+ }
+ }
+
+ /**
+ * Helper class for computing number of record in the File preceding the beginning of the Range
+ * in this file.
+ */
+ @VisibleForTesting
+ static class ComputeRecordsBeforeEachRange extends DoFn<Integer, KV<KV<String, Long>, Long>> {
+ private final PCollectionView<Map<KV<String, Long>, Long>> rangeSizes;
+
+ public ComputeRecordsBeforeEachRange(
+ PCollectionView<Map<KV<String, Long>, Long>> rangeSizes) {
+ this.rangeSizes = rangeSizes;
+ }
+
+ // Add custom comparator as KV<K, V> is not comparable by default
+ private static class FileRangeComparator<K extends Comparable<K>, V extends Comparable<V>>
+ implements Comparator<KV<K, V>> {
+ @Override
+ public int compare(KV<K, V> a, KV<K, V> b) {
+ if (a.getKey().compareTo(b.getKey()) == 0) {
+ return a.getValue().compareTo(b.getValue());
+ }
+ return a.getKey().compareTo(b.getKey());
+ }
+ }
+
+ @ProcessElement
+ public void processElement(ProcessContext p) {
+ // Get the Map Containing the size from side-input
+ Map<KV<String, Long>, Long> rangeSizesMap = p.sideInput(rangeSizes);
+
+ // The FileRange Pair must be sorted
+ SortedMap<KV<String, Long>, Long> sorted = new TreeMap<>(new FileRangeComparator<>());
+
+ // Initialize sorted map with values
+ for (Map.Entry<KV<String, Long>, Long> entry : rangeSizesMap.entrySet()) {
+ sorted.put(entry.getKey(), entry.getValue());
+ }
+
+ // HashMap that tracks number of records passed for each file
+ Map<String, Long> pastRecords = new HashMap<>();
+
+ // For each (File, Range) Pair, compute the number of records before it
+ for (Map.Entry entry : sorted.entrySet()) {
Review comment:
Done
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
[GitHub] [beam] rezarokni commented on a change in pull request #12645: [BEAM-10124] Add ContextualTextIO
Posted by GitBox <gi...@apache.org>.
rezarokni commented on a change in pull request #12645:
URL: https://github.com/apache/beam/pull/12645#discussion_r487927710
##########
File path: sdks/java/io/contextual-text-io/src/main/java/org/apache/beam/sdk/io/contextualtextio/RecordWithMetadata.java
##########
@@ -0,0 +1,84 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one
+ * or more contributor license agreements. See the NOTICE file
+ * distributed with this work for additional information
+ * regarding copyright ownership. The ASF licenses this file
+ * to you under the Apache License, Version 2.0 (the
+ * "License"); you may not use this file except in compliance
+ * with the License. You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+package org.apache.beam.sdk.io.contextualtextio;
+
+import com.google.auto.value.AutoValue;
+import org.apache.beam.sdk.annotations.Experimental;
+import org.apache.beam.sdk.schemas.AutoValueSchema;
+import org.apache.beam.sdk.schemas.annotations.DefaultSchema;
+
+/**
+ * Helper Class based on {@link AutoValueSchema}, it provides Metadata associated with each Record
+ * when reading from file(s) using {@link ContextualTextIO}.
+ *
+ * <h3>Fields:</h3>
+ *
+ * <ul>
+ * <li>recordOffset: The offset of a record (the byte at which the record begins) in a file. This
+ * information can be useful if you wish to reconstruct the file. {@link
+ * RecordWithMetadata#getRecordOffset()}
+ * <li>recordNum: The ordinal number of the record in its file. {@link
+ * RecordWithMetadata#getRecordNum()}
+ * <li>recordValue: The value / contents of the record {@link RecordWithMetadata#getRecordValue()}
+ * <li>rangeOffset: The starting offset of the range (split), which contained the record, when the
+ * record was read. {@link RecordWithMetadata#getRangeOffset()}
+ * <li>recordNumInOffset: The record number relative to the Range. (line number within the range)
+ * {@link RecordWithMetadata#getRecordNumInOffset()}
+ * <li>fileName: Name of the file to which the record belongs (this is the full filename,
+ * eg:path/to/file.txt) {@link RecordWithMetadata#getFileName()}
+ * </ul>
+ */
+@Experimental(Experimental.Kind.SCHEMAS)
+@DefaultSchema(AutoValueSchema.class)
+@AutoValue
+public abstract class RecordWithMetadata {
Review comment:
@lukecwik This class is a Schema class, looking into this we cant make it generic, until schemas supports generics. But for ContextualTextIO we would be setting it to ContextualTextIO<String> anyway ... So while it will be useful in the future, I am not sure it adds much for now? WDYT?
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
[GitHub] [beam] abhiy13 commented on a change in pull request #12645: [BEAM-10124] Add ContextualTextIO
Posted by GitBox <gi...@apache.org>.
abhiy13 commented on a change in pull request #12645:
URL: https://github.com/apache/beam/pull/12645#discussion_r492480890
##########
File path: sdks/java/io/contextual-text-io/src/main/java/org/apache/beam/sdk/io/contextualtextio/RecordWithMetadata.java
##########
@@ -0,0 +1,85 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one
+ * or more contributor license agreements. See the NOTICE file
+ * distributed with this work for additional information
+ * regarding copyright ownership. The ASF licenses this file
+ * to you under the Apache License, Version 2.0 (the
+ * "License"); you may not use this file except in compliance
+ * with the License. You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+package org.apache.beam.sdk.io.contextualtextio;
+
+import com.google.auto.value.AutoValue;
+import org.apache.beam.sdk.annotations.Experimental;
+import org.apache.beam.sdk.io.fs.ResourceId;
+import org.apache.beam.sdk.schemas.AutoValueSchema;
+import org.apache.beam.sdk.schemas.annotations.DefaultSchema;
+
+/**
+ * Helper Class based on {@link AutoValueSchema}, it provides Metadata associated with each Record
+ * when reading from file(s) using {@link ContextualTextIO}.
+ *
+ * <h3>Fields:</h3>
+ *
+ * <ul>
+ * <li>recordOffset: The offset of a record (the byte at which the record begins) in a file. This
+ * information can be useful if you wish to reconstruct the file. {@link
+ * RecordWithMetadata#getRecordOffset()}
+ * <li>recordNum: The ordinal number of the record in its file. {@link
+ * RecordWithMetadata#getRecordNum()}
+ * <li>recordValue: The value / contents of the record {@link RecordWithMetadata#getValue()}
+ * <li>rangeOffset: The starting offset of the range (split), which contained the record, when the
+ * record was read. {@link RecordWithMetadata#getRangeOffset()}
+ * <li>recordNumInOffset: The record number relative to the Range. (line number within the range)
+ * {@link RecordWithMetadata#getRecordNumInOffset()}
+ * <li>fileName: Name of the file to which the record belongs (this is the full filename,
+ * eg:path/to/file.txt) {@link RecordWithMetadata#getFileName()}
+ * </ul>
+ */
+@Experimental(Experimental.Kind.SCHEMAS)
+@DefaultSchema(AutoValueSchema.class)
+@AutoValue
+public abstract class RecordWithMetadata {
+ public abstract long getRecordOffset();
Review comment:
Done
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
[GitHub] [beam] tvalentyn commented on a change in pull request #12645: [BEAM-10124] Add ContextualTextIO
Posted by GitBox <gi...@apache.org>.
tvalentyn commented on a change in pull request #12645:
URL: https://github.com/apache/beam/pull/12645#discussion_r479536159
##########
File path: sdks/java/io/contextual-text-io/src/main/java/org/apache/beam/sdk/io/contextualtextio/ContextualTextIOSource.java
##########
@@ -0,0 +1,364 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one
+ * or more contributor license agreements. See the NOTICE file
+ * distributed with this work for additional information
+ * regarding copyright ownership. The ASF licenses this file
+ * to you under the Apache License, Version 2.0 (the
+ * "License"); you may not use this file except in compliance
+ * with the License. You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+package org.apache.beam.sdk.io.contextualtextio;
+
+import java.io.IOException;
+import java.nio.ByteBuffer;
+import java.nio.channels.ReadableByteChannel;
+import java.nio.channels.SeekableByteChannel;
+import java.util.NoSuchElementException;
+import org.apache.beam.sdk.coders.Coder;
+import org.apache.beam.sdk.io.FileBasedSource;
+import org.apache.beam.sdk.io.fs.EmptyMatchTreatment;
+import org.apache.beam.sdk.io.fs.MatchResult;
+import org.apache.beam.sdk.options.PipelineOptions;
+import org.apache.beam.sdk.options.ValueProvider;
+import org.apache.beam.sdk.schemas.NoSuchSchemaException;
+import org.apache.beam.sdk.schemas.SchemaCoder;
+import org.apache.beam.sdk.schemas.SchemaRegistry;
+import org.apache.beam.vendor.grpc.v1p26p0.com.google.protobuf.ByteString;
+import org.apache.beam.vendor.guava.v26_0_jre.com.google.common.annotations.VisibleForTesting;
+import org.apache.beam.vendor.guava.v26_0_jre.com.google.common.base.Preconditions;
+import org.checkerframework.checker.nullness.qual.Nullable;
+import org.slf4j.Logger;
+import org.slf4j.LoggerFactory;
+
+/**
+ * Implementation detail of {@link ContextualTextIO.Read}.
+ *
+ * <p>A {@link FileBasedSource} which can decode records delimited by newline characters.
+ *
+ * <p>This source splits the data into records using {@code UTF-8} {@code \n}, {@code \r}, or {@code
+ * \r\n} as the delimiter. This source is not strict and supports decoding the last record even if
+ * it is not delimited. Finally, no records are decoded if the stream is empty.
+ *
+ * <p>This source supports reading from any arbitrary byte position within the stream. If the
+ * starting position is not {@code 0}, then bytes are skipped until the first delimiter is found
+ * representing the beginning of the first record to be decoded.
+ */
+@VisibleForTesting
+class ContextualTextIOSource extends FileBasedSource<RecordWithMetadata> {
Review comment:
Hey @abhiy13 - given your internship and school timeline @rezarokni and I don't expect you to be available to act on the feedback on this PR, and we can make necessary cleanups for the merge. We probably won't get to it for at least a week though. Long term we should use SDF here. If you are interested to give it a try, I would recommend you create a separate PR for this purpose, and add commits already available on this PR into your new branch.
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
[GitHub] [beam] tvalentyn closed pull request #12645: [BEAM-10124] Add ContextualTextIO
Posted by GitBox <gi...@apache.org>.
tvalentyn closed pull request #12645:
URL: https://github.com/apache/beam/pull/12645
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
[GitHub] [beam] abhiy13 commented on a change in pull request #12645: [BEAM-10124] Add ContextualTextIO
Posted by GitBox <gi...@apache.org>.
abhiy13 commented on a change in pull request #12645:
URL: https://github.com/apache/beam/pull/12645#discussion_r475174519
##########
File path: sdks/java/io/contextual-text-io/src/main/java/org/apache/beam/sdk/io/contextualtextio/ContextualTextIO.java
##########
@@ -151,8 +157,27 @@
* .apply(ContextualTextIO.readFiles());
* }</pre>
*
- * NOTE: Using {@link ContextualTextIO.Read#withHasMultilineCSVRecords(Boolean)} introduces a
- * performance penalty: when this option is enabled, the input cannot be split and read in parallel.
+ * <p>Example 6: reading without recordNum metadata, or only fileName associated Metadata. (the
+ * Objects would still contain recordNums, but these recordNums would correspond to their positions
+ * in their respective offsets rather than their positions within the entire file).
+ *
+ * <pre>{@code
+ * Pipeline p = ...;
+ *
+ * PCollection<RecordWithMetadata> records = p.apply(ContextualTextIO.read()
+ * .from("/local/path/to/files/*.csv")
+ * .setWithoutRecordNumMetadata(true));
+ * }</pre>
+ *
+ * <p>NOTE: When using {@link ContextualTextIO.Read#withHasMultilineCSVRecords(Boolean)} this
+ * option, a single reader will be used to process the file, rather than multiple readers which can
+ * read from different offsets. For a large file this can result in lower performance.
+ *
+ * <p>NOTE: Use {@link Read#withoutRecordNumMetadata()} when recordNum metadata is not required or
+ * when only metadata associated with filenames is required, Not using this option introduces a
+ * shuffle step which increases the resources used by the pipeline. <b>This option is set to false
+ * by default. Meaning that the shuffle step will be performed, set it to false to avoid the shuffle
Review comment:
Done.
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
[GitHub] [beam] abhiy13 commented on a change in pull request #12645: [BEAM-10124] Add ContextualTextIO
Posted by GitBox <gi...@apache.org>.
abhiy13 commented on a change in pull request #12645:
URL: https://github.com/apache/beam/pull/12645#discussion_r474446145
##########
File path: sdks/java/io/contextual-text-io/src/main/java/org/apache/beam/sdk/io/contextualtextio/ContextualTextIO.java
##########
@@ -319,31 +326,169 @@ static boolean isSelfOverlapping(byte[] s) {
}
@Override
- public PCollection<String> expand(PBegin input) {
- checkNotNull(getFilepattern(), "need to set the filepattern of a TextIO.Read transform");
+ public PCollection<RecordWithMetadata> expand(PBegin input) {
+ checkNotNull(
+ getFilepattern(), "need to set the filepattern of a ContextualTextIO.Read transform");
+ PCollection<RecordWithMetadata> lines = null;
if (getMatchConfiguration().getWatchInterval() == null && !getHintMatchesManyFiles()) {
- return input.apply("Read", org.apache.beam.sdk.io.Read.from(getSource()));
+ lines = input.apply("Read", org.apache.beam.sdk.io.Read.from(getSource()));
+ } else {
+ // All other cases go through FileIO + ReadFiles
+ lines =
+ input
+ .apply(
+ "Create filepattern", Create.ofProvider(getFilepattern(), StringUtf8Coder.of()))
+ .apply("Match All", FileIO.matchAll().withConfiguration(getMatchConfiguration()))
+ .apply(
+ "Read Matches",
+ FileIO.readMatches()
+ .withCompression(getCompression())
+ .withDirectoryTreatment(DirectoryTreatment.PROHIBIT))
+ .apply("Via ReadFiles", readFiles().withDelimiter(getDelimiter()));
}
- // All other cases go through FileIO + ReadFiles
- return input
- .apply("Create filepattern", Create.ofProvider(getFilepattern(), StringUtf8Coder.of()))
- .apply("Match All", FileIO.matchAll().withConfiguration(getMatchConfiguration()))
- .apply(
- "Read Matches",
- FileIO.readMatches()
- .withCompression(getCompression())
- .withDirectoryTreatment(DirectoryTreatment.PROHIBIT))
- .apply("Via ReadFiles", readFiles().withDelimiter(getDelimiter()));
+ // Check if the user decided to opt out of recordNums associated with records
+ if (getWithoutLineNumMetadata()) {
+ return lines;
+ }
+
+ // At this point the line number in RecordWithMetadata contains the relative line offset from
+ // the
+ // beginning of the read range.
+
+ // To compute the absolute position from the beginning of the input,
+ // we group the lines within the same ranges, and evaluate the size of each range.
+
+ PCollection<KV<KV<String, Long>, RecordWithMetadata>> linesGroupedByFileAndRange =
+ lines.apply("AddFileNameAndRange", ParDo.of(new AddFileNameAndRange()));
+
+ PCollectionView<Map<KV<String, Long>, Long>> rangeSizes =
+ linesGroupedByFileAndRange
+ .apply("CountLinesForEachFileRange", Count.perKey())
+ .apply("SizesAsView", View.asMap());
+
+ // Get Pipeline to create a dummy PCollection with one element to help compute the lines
+ // before each Range
+ PCollection<Integer> singletonPcoll =
+ input.getPipeline().apply("CreateSingletonPcoll", Create.of(Arrays.asList(1)));
+
+ // For each (File, Offset) pair, calculate the number of lines occurring before the Range for
+ // each File
+
+ // After computing the number of lines before each range, we can find the line number in
+ // original file as numLiesBeforeOffset + lineNumInCurrentOffset
+ PCollectionView<Map<KV<String, Long>, Long>> numLinesBeforeEachRange =
+ singletonPcoll
+ .apply(
+ "ComputeLinesBeforeRange",
+ ParDo.of(new ComputeLinesBeforeEachRange(rangeSizes)).withSideInputs(rangeSizes))
+ .apply("NumLinesBeforeEachRangeAsView", View.asMap());
+
+ return linesGroupedByFileAndRange.apply(
+ "AssignLineNums",
+ ParDo.of(new AssignLineNums(numLinesBeforeEachRange))
+ .withSideInputs(numLinesBeforeEachRange));
+ }
+
+ @VisibleForTesting
+ static class AddFileNameAndRange
+ extends DoFn<RecordWithMetadata, KV<KV<String, Long>, RecordWithMetadata>> {
+ @ProcessElement
+ public void processElement(
+ @Element RecordWithMetadata line,
+ OutputReceiver<KV<KV<String, Long>, RecordWithMetadata>> out) {
+ out.output(KV.of(KV.of(line.getFileName(), line.getRange().getRangeNum()), line));
+ }
+ }
+
+ /**
+ * Helper class for computing number of lines in the File preceding the beginning of the Range
+ * in this file.
+ */
+ @VisibleForTesting
+ static class ComputeLinesBeforeEachRange extends DoFn<Integer, KV<KV<String, Long>, Long>> {
+ private final PCollectionView<Map<KV<String, Long>, Long>> rangeSizes;
+
+ public ComputeLinesBeforeEachRange(PCollectionView<Map<KV<String, Long>, Long>> rangeSizes) {
+ this.rangeSizes = rangeSizes;
+ }
+
+ // Add custom comparator as KV<K, V> is not comparable by default
+ private static class FileRangeComparator<K extends Comparable<K>, V extends Comparable<V>>
+ implements Comparator<KV<K, V>> {
+ @Override
+ public int compare(KV<K, V> a, KV<K, V> b) {
+ if (a.getKey().compareTo(b.getKey()) == 0) {
+ return a.getValue().compareTo(b.getValue());
+ }
+ return a.getKey().compareTo(b.getKey());
+ }
+ }
+
+ @ProcessElement
+ public void processElement(ProcessContext p) {
+ // Get the Map Containing the size from side-input
+ Map<KV<String, Long>, Long> rangeSizesMap = p.sideInput(rangeSizes);
+
+ // The FileRange Pair must be sorted
+ SortedMap<KV<String, Long>, Long> sorted = new TreeMap<>(new FileRangeComparator<>());
+
+ // Initialize sorted map with values
+ for (Map.Entry<KV<String, Long>, Long> entry : rangeSizesMap.entrySet()) {
+ sorted.put(entry.getKey(), entry.getValue());
+ }
+
+ // HashMap that tracks lines passed for each file
+ Map<String, Long> pastLines = new HashMap<>();
+
+ // For each (File, Range) Pair, compute the number of lines before it
+ for (Map.Entry entry : sorted.entrySet()) {
+ Long lines = (long) entry.getValue();
Review comment:
Ack.
##########
File path: sdks/java/io/contextual-text-io/src/main/java/org/apache/beam/sdk/io/contextualtextio/ContextualTextIOSource.java
##########
@@ -0,0 +1,363 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one
+ * or more contributor license agreements. See the NOTICE file
+ * distributed with this work for additional information
+ * regarding copyright ownership. The ASF licenses this file
+ * to you under the Apache License, Version 2.0 (the
+ * "License"); you may not use this file except in compliance
+ * with the License. You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+package org.apache.beam.sdk.io.contextualtextio;
+
+import java.io.IOException;
+import java.nio.ByteBuffer;
+import java.nio.channels.ReadableByteChannel;
+import java.nio.channels.SeekableByteChannel;
+import java.util.NoSuchElementException;
+import org.apache.beam.sdk.coders.Coder;
+import org.apache.beam.sdk.io.FileBasedSource;
+import org.apache.beam.sdk.io.fs.EmptyMatchTreatment;
+import org.apache.beam.sdk.io.fs.MatchResult;
+import org.apache.beam.sdk.options.PipelineOptions;
+import org.apache.beam.sdk.options.ValueProvider;
+import org.apache.beam.sdk.schemas.NoSuchSchemaException;
+import org.apache.beam.sdk.schemas.SchemaCoder;
+import org.apache.beam.sdk.schemas.SchemaRegistry;
+import org.apache.beam.vendor.grpc.v1p26p0.com.google.protobuf.ByteString;
+import org.apache.beam.vendor.guava.v26_0_jre.com.google.common.annotations.VisibleForTesting;
+import org.apache.beam.vendor.guava.v26_0_jre.com.google.common.base.Preconditions;
+import org.checkerframework.checker.nullness.qual.Nullable;
+
+/**
+ * Implementation detail of {@link ContextualTextIO.Read}.
+ *
+ * <p>A {@link FileBasedSource} which can decode records delimited by newline characters.
+ *
+ * <p>This source splits the data into records using {@code UTF-8} {@code \n}, {@code \r}, or {@code
+ * \r\n} as the delimiter. This source is not strict and supports decoding the last record even if
+ * it is not delimited. Finally, no records are decoded if the stream is empty.
+ *
+ * <p>This source supports reading from any arbitrary byte position within the stream. If the
+ * starting position is not {@code 0}, then bytes are skipped until the first delimiter is found
+ * representing the beginning of the first record to be decoded.
+ */
+@VisibleForTesting
+class ContextualTextIOSource extends FileBasedSource<RecordWithMetadata> {
+ byte[] delimiter;
+
+ // Used to Override isSplittable
+ private boolean hasMultilineCSVRecords;
+
+ @Override
+ protected boolean isSplittable() throws Exception {
+ if (hasMultilineCSVRecords) {
+ return false;
+ }
+ return super.isSplittable();
+ }
+
+ ContextualTextIOSource(
+ ValueProvider<String> fileSpec,
+ EmptyMatchTreatment emptyMatchTreatment,
+ byte[] delimiter,
+ boolean hasMultilineCSVRecords) {
+ super(fileSpec, emptyMatchTreatment, 1L);
+ this.delimiter = delimiter;
+ this.hasMultilineCSVRecords = hasMultilineCSVRecords;
+ }
+
+ private ContextualTextIOSource(
+ MatchResult.Metadata metadata,
+ long start,
+ long end,
+ byte[] delimiter,
+ boolean hasMultilineCSVRecords) {
+ super(metadata, 1L, start, end);
+ this.delimiter = delimiter;
+ this.hasMultilineCSVRecords = hasMultilineCSVRecords;
+ }
+
+ @Override
+ protected FileBasedSource<RecordWithMetadata> createForSubrangeOfFile(
+ MatchResult.Metadata metadata, long start, long end) {
+ return new ContextualTextIOSource(metadata, start, end, delimiter, hasMultilineCSVRecords);
+ }
+
+ @Override
+ protected FileBasedReader<RecordWithMetadata> createSingleFileReader(PipelineOptions options) {
+ return new MultiLineTextBasedReader(this, delimiter, hasMultilineCSVRecords);
+ }
+
+ @Override
+ public Coder<RecordWithMetadata> getOutputCoder() {
+ SchemaCoder<RecordWithMetadata> coder = null;
+ try {
+ coder = SchemaRegistry.createDefault().getSchemaCoder(RecordWithMetadata.class);
+ } catch (NoSuchSchemaException e) {
+ System.out.println("No Coder!");
Review comment:
Ack.
##########
File path: sdks/java/io/contextual-text-io/src/main/java/org/apache/beam/sdk/io/contextualtextio/ContextualTextIOSource.java
##########
@@ -99,14 +126,23 @@ private ContextualTextIOSource(MatchResult.Metadata metadata, long start, long e
private volatile long startOfNextRecord;
private volatile boolean eof;
private volatile boolean elementIsPresent;
- private @Nullable String currentValue;
+ private @Nullable RecordWithMetadata currentValue;
private @Nullable ReadableByteChannel inChannel;
private byte @Nullable [] delimiter;
- private TextBasedReader(ContextualTextIOSource source, byte[] delimiter) {
+ // Add to override the isSplittable
+ private boolean hasRFC4180MultiLineColumn;
+
+ private long startingOffset;
+ private long readerlineNum;
+
+ private MultiLineTextBasedReader(
+ ContextualTextIOSource source, byte[] delimiter, boolean hasRFC4180MultiLineColumn) {
Review comment:
Ack.
##########
File path: sdks/java/io/contextual-text-io/src/main/java/org/apache/beam/sdk/io/contextualtextio/ContextualTextIOSource.java
##########
@@ -259,7 +317,31 @@ private void decodeCurrentElement() throws IOException {
if (startOfRecord == 0 && dataToDecode.startsWith(UTF8_BOM)) {
dataToDecode = dataToDecode.substring(UTF8_BOM.size());
}
- currentValue = dataToDecode.toStringUtf8();
+
+ /////////////////////////////////////////////
+
+ // Data of the Current Line
+ // dataToDecode.toStringUtf8();
+
+ // The line num is:
Review comment:
Removed.
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
[GitHub] [beam] lukecwik commented on a change in pull request #12645: [BEAM-10124] Add ContextualTextIO
Posted by GitBox <gi...@apache.org>.
lukecwik commented on a change in pull request #12645:
URL: https://github.com/apache/beam/pull/12645#discussion_r477478222
##########
File path: sdks/java/io/contextual-text-io/src/main/java/org/apache/beam/sdk/io/contextualtextio/RecordWithMetadata.java
##########
@@ -0,0 +1,84 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one
+ * or more contributor license agreements. See the NOTICE file
+ * distributed with this work for additional information
+ * regarding copyright ownership. The ASF licenses this file
+ * to you under the Apache License, Version 2.0 (the
+ * "License"); you may not use this file except in compliance
+ * with the License. You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+package org.apache.beam.sdk.io.contextualtextio;
+
+import com.google.auto.value.AutoValue;
+import org.apache.beam.sdk.annotations.Experimental;
+import org.apache.beam.sdk.schemas.AutoValueSchema;
+import org.apache.beam.sdk.schemas.annotations.DefaultSchema;
+
+/**
+ * Helper Class based on {@link AutoValueSchema}, it provides Metadata associated with each Record
+ * when reading from file(s) using {@link ContextualTextIO}.
+ *
+ * <h3>Fields:</h3>
+ *
+ * <ul>
+ * <li>recordOffset: The offset of a record (the byte at which the record begins) in a file. This
+ * information can be useful if you wish to reconstruct the file. {@link
+ * RecordWithMetadata#getRecordOffset()}
+ * <li>recordNum: The ordinal number of the record in its file. {@link
+ * RecordWithMetadata#getRecordNum()}
+ * <li>recordValue: The value / contents of the record {@link RecordWithMetadata#getRecordValue()}
+ * <li>rangeOffset: The starting offset of the range (split), which contained the record, when the
+ * record was read. {@link RecordWithMetadata#getRangeOffset()}
+ * <li>recordNumInOffset: The record number relative to the Range. (line number within the range)
+ * {@link RecordWithMetadata#getRecordNumInOffset()}
+ * <li>fileName: Name of the file to which the record belongs (this is the full filename,
+ * eg:path/to/file.txt) {@link RecordWithMetadata#getFileName()}
+ * </ul>
+ */
+@Experimental(Experimental.Kind.SCHEMAS)
+@DefaultSchema(AutoValueSchema.class)
+@AutoValue
+public abstract class RecordWithMetadata {
Review comment:
Missed the '`', so the `<T>` was lost.
##########
File path: sdks/java/io/contextual-text-io/src/main/java/org/apache/beam/sdk/io/contextualtextio/RecordWithMetadata.java
##########
@@ -0,0 +1,84 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one
+ * or more contributor license agreements. See the NOTICE file
+ * distributed with this work for additional information
+ * regarding copyright ownership. The ASF licenses this file
+ * to you under the Apache License, Version 2.0 (the
+ * "License"); you may not use this file except in compliance
+ * with the License. You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+package org.apache.beam.sdk.io.contextualtextio;
+
+import com.google.auto.value.AutoValue;
+import org.apache.beam.sdk.annotations.Experimental;
+import org.apache.beam.sdk.schemas.AutoValueSchema;
+import org.apache.beam.sdk.schemas.annotations.DefaultSchema;
+
+/**
+ * Helper Class based on {@link AutoValueSchema}, it provides Metadata associated with each Record
+ * when reading from file(s) using {@link ContextualTextIO}.
+ *
+ * <h3>Fields:</h3>
+ *
+ * <ul>
+ * <li>recordOffset: The offset of a record (the byte at which the record begins) in a file. This
+ * information can be useful if you wish to reconstruct the file. {@link
+ * RecordWithMetadata#getRecordOffset()}
+ * <li>recordNum: The ordinal number of the record in its file. {@link
+ * RecordWithMetadata#getRecordNum()}
+ * <li>recordValue: The value / contents of the record {@link RecordWithMetadata#getRecordValue()}
+ * <li>rangeOffset: The starting offset of the range (split), which contained the record, when the
+ * record was read. {@link RecordWithMetadata#getRangeOffset()}
+ * <li>recordNumInOffset: The record number relative to the Range. (line number within the range)
+ * {@link RecordWithMetadata#getRecordNumInOffset()}
+ * <li>fileName: Name of the file to which the record belongs (this is the full filename,
+ * eg:path/to/file.txt) {@link RecordWithMetadata#getFileName()}
+ * </ul>
+ */
+@Experimental(Experimental.Kind.SCHEMAS)
+@DefaultSchema(AutoValueSchema.class)
+@AutoValue
+public abstract class RecordWithMetadata {
Review comment:
Missed the '\`', so the `<T>` was lost.
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
[GitHub] [beam] abhiy13 commented on a change in pull request #12645: [BEAM-10124] Add ContextualTextIO
Posted by GitBox <gi...@apache.org>.
abhiy13 commented on a change in pull request #12645:
URL: https://github.com/apache/beam/pull/12645#discussion_r474041941
##########
File path: sdks/java/io/contextual-text-io/src/main/java/org/apache/beam/sdk/io/contextualtextio/ContextualTextIO.java
##########
@@ -0,0 +1,594 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one
+ * or more contributor license agreements. See the NOTICE file
+ * distributed with this work for additional information
+ * regarding copyright ownership. The ASF licenses this file
+ * to you under the Apache License, Version 2.0 (the
+ * "License"); you may not use this file except in compliance
+ * with the License. You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+package org.apache.beam.sdk.io.contextualtextio;
+
+import static org.apache.beam.sdk.io.FileIO.ReadMatches.DirectoryTreatment;
+import static org.apache.beam.vendor.guava.v26_0_jre.com.google.common.base.Preconditions.checkArgument;
+import static org.apache.beam.vendor.guava.v26_0_jre.com.google.common.base.Preconditions.checkNotNull;
+
+import com.google.auto.value.AutoValue;
+import java.nio.ByteBuffer;
+import java.util.Arrays;
+import java.util.Comparator;
+import java.util.HashMap;
+import java.util.Map;
+import java.util.SortedMap;
+import java.util.TreeMap;
+import org.apache.beam.sdk.coders.StringUtf8Coder;
+import org.apache.beam.sdk.io.CompressedSource;
+import org.apache.beam.sdk.io.Compression;
+import org.apache.beam.sdk.io.FileBasedSource;
+import org.apache.beam.sdk.io.FileIO;
+import org.apache.beam.sdk.io.FileIO.MatchConfiguration;
+import org.apache.beam.sdk.io.ReadAllViaFileBasedSource;
+import org.apache.beam.sdk.io.fs.EmptyMatchTreatment;
+import org.apache.beam.sdk.options.ValueProvider;
+import org.apache.beam.sdk.options.ValueProvider.StaticValueProvider;
+import org.apache.beam.sdk.schemas.NoSuchSchemaException;
+import org.apache.beam.sdk.schemas.SchemaCoder;
+import org.apache.beam.sdk.transforms.Count;
+import org.apache.beam.sdk.transforms.Create;
+import org.apache.beam.sdk.transforms.DoFn;
+import org.apache.beam.sdk.transforms.PTransform;
+import org.apache.beam.sdk.transforms.ParDo;
+import org.apache.beam.sdk.transforms.SerializableFunction;
+import org.apache.beam.sdk.transforms.View;
+import org.apache.beam.sdk.transforms.Watch.Growth.TerminationCondition;
+import org.apache.beam.sdk.transforms.display.DisplayData;
+import org.apache.beam.sdk.values.KV;
+import org.apache.beam.sdk.values.PBegin;
+import org.apache.beam.sdk.values.PCollection;
+import org.apache.beam.sdk.values.PCollectionView;
+import org.apache.beam.vendor.guava.v26_0_jre.com.google.common.annotations.VisibleForTesting;
+import org.checkerframework.checker.nullness.qual.Nullable;
+import org.joda.time.Duration;
+
+/**
+ * {@link PTransform}s that read text files and collect contextual information of the elements in
Review comment:
Ack
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
[GitHub] [beam] tvalentyn commented on a change in pull request #12645: [BEAM-10124] Add ContextualTextIO
Posted by GitBox <gi...@apache.org>.
tvalentyn commented on a change in pull request #12645:
URL: https://github.com/apache/beam/pull/12645#discussion_r477483186
##########
File path: sdks/java/io/contextual-text-io/src/main/java/org/apache/beam/sdk/io/contextualtextio/ContextualTextIOSource.java
##########
@@ -0,0 +1,364 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one
+ * or more contributor license agreements. See the NOTICE file
+ * distributed with this work for additional information
+ * regarding copyright ownership. The ASF licenses this file
+ * to you under the Apache License, Version 2.0 (the
+ * "License"); you may not use this file except in compliance
+ * with the License. You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+package org.apache.beam.sdk.io.contextualtextio;
+
+import java.io.IOException;
+import java.nio.ByteBuffer;
+import java.nio.channels.ReadableByteChannel;
+import java.nio.channels.SeekableByteChannel;
+import java.util.NoSuchElementException;
+import org.apache.beam.sdk.coders.Coder;
+import org.apache.beam.sdk.io.FileBasedSource;
+import org.apache.beam.sdk.io.fs.EmptyMatchTreatment;
+import org.apache.beam.sdk.io.fs.MatchResult;
+import org.apache.beam.sdk.options.PipelineOptions;
+import org.apache.beam.sdk.options.ValueProvider;
+import org.apache.beam.sdk.schemas.NoSuchSchemaException;
+import org.apache.beam.sdk.schemas.SchemaCoder;
+import org.apache.beam.sdk.schemas.SchemaRegistry;
+import org.apache.beam.vendor.grpc.v1p26p0.com.google.protobuf.ByteString;
+import org.apache.beam.vendor.guava.v26_0_jre.com.google.common.annotations.VisibleForTesting;
+import org.apache.beam.vendor.guava.v26_0_jre.com.google.common.base.Preconditions;
+import org.checkerframework.checker.nullness.qual.Nullable;
+import org.slf4j.Logger;
+import org.slf4j.LoggerFactory;
+
+/**
+ * Implementation detail of {@link ContextualTextIO.Read}.
+ *
+ * <p>A {@link FileBasedSource} which can decode records delimited by newline characters.
+ *
+ * <p>This source splits the data into records using {@code UTF-8} {@code \n}, {@code \r}, or {@code
+ * \r\n} as the delimiter. This source is not strict and supports decoding the last record even if
+ * it is not delimited. Finally, no records are decoded if the stream is empty.
+ *
+ * <p>This source supports reading from any arbitrary byte position within the stream. If the
+ * starting position is not {@code 0}, then bytes are skipped until the first delimiter is found
+ * representing the beginning of the first record to be decoded.
+ */
+@VisibleForTesting
+class ContextualTextIOSource extends FileBasedSource<RecordWithMetadata> {
Review comment:
Would it be acceptable to have this as FileBasedSource in initial implementation, and convert to SDF as a follow up change?
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
[GitHub] [beam] abhiy13 commented on a change in pull request #12645: [BEAM-10124] Add ContextualTextIO
Posted by GitBox <gi...@apache.org>.
abhiy13 commented on a change in pull request #12645:
URL: https://github.com/apache/beam/pull/12645#discussion_r492474704
##########
File path: sdks/java/io/contextual-text-io/src/main/java/org/apache/beam/sdk/io/contextualtextio/ContextualTextIOSource.java
##########
@@ -0,0 +1,364 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one
+ * or more contributor license agreements. See the NOTICE file
+ * distributed with this work for additional information
+ * regarding copyright ownership. The ASF licenses this file
+ * to you under the Apache License, Version 2.0 (the
+ * "License"); you may not use this file except in compliance
+ * with the License. You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+package org.apache.beam.sdk.io.contextualtextio;
+
+import java.io.IOException;
+import java.nio.ByteBuffer;
+import java.nio.channels.ReadableByteChannel;
+import java.nio.channels.SeekableByteChannel;
+import java.util.NoSuchElementException;
+import org.apache.beam.sdk.coders.Coder;
+import org.apache.beam.sdk.io.FileBasedSource;
+import org.apache.beam.sdk.io.fs.EmptyMatchTreatment;
+import org.apache.beam.sdk.io.fs.MatchResult;
+import org.apache.beam.sdk.options.PipelineOptions;
+import org.apache.beam.sdk.options.ValueProvider;
+import org.apache.beam.sdk.schemas.NoSuchSchemaException;
+import org.apache.beam.sdk.schemas.SchemaCoder;
+import org.apache.beam.sdk.schemas.SchemaRegistry;
+import org.apache.beam.vendor.grpc.v1p26p0.com.google.protobuf.ByteString;
+import org.apache.beam.vendor.guava.v26_0_jre.com.google.common.annotations.VisibleForTesting;
+import org.apache.beam.vendor.guava.v26_0_jre.com.google.common.base.Preconditions;
+import org.checkerframework.checker.nullness.qual.Nullable;
+import org.slf4j.Logger;
+import org.slf4j.LoggerFactory;
+
+/**
+ * Implementation detail of {@link ContextualTextIO.Read}.
+ *
+ * <p>A {@link FileBasedSource} which can decode records delimited by newline characters.
+ *
+ * <p>This source splits the data into records using {@code UTF-8} {@code \n}, {@code \r}, or {@code
+ * \r\n} as the delimiter. This source is not strict and supports decoding the last record even if
+ * it is not delimited. Finally, no records are decoded if the stream is empty.
+ *
+ * <p>This source supports reading from any arbitrary byte position within the stream. If the
+ * starting position is not {@code 0}, then bytes are skipped until the first delimiter is found
+ * representing the beginning of the first record to be decoded.
+ */
+@VisibleForTesting
+class ContextualTextIOSource extends FileBasedSource<RecordWithMetadata> {
+ byte[] delimiter;
+
+ private static final Logger LOG = LoggerFactory.getLogger(ContextualTextIOSource.class);
+
+ // Used to Override isSplittable
+ private boolean hasMultilineCSVRecords;
+
+ @Override
+ protected boolean isSplittable() throws Exception {
+ if (hasMultilineCSVRecords) {
+ // When Having Multiline CSV Records,
+ // Splitting the file may cause a split to be within a record,
+ // Disabling split prevents this from happening
+ return false;
+ }
+ return super.isSplittable();
+ }
+
+ ContextualTextIOSource(
+ ValueProvider<String> fileSpec,
+ EmptyMatchTreatment emptyMatchTreatment,
+ byte[] delimiter,
+ boolean hasMultilineCSVRecords) {
+ super(fileSpec, emptyMatchTreatment, 1L);
+ this.delimiter = delimiter;
+ this.hasMultilineCSVRecords = hasMultilineCSVRecords;
+ }
+
+ private ContextualTextIOSource(
+ MatchResult.Metadata metadata,
+ long start,
+ long end,
+ byte[] delimiter,
+ boolean hasMultilineCSVRecords) {
+ super(metadata, 1L, start, end);
+ this.delimiter = delimiter;
+ this.hasMultilineCSVRecords = hasMultilineCSVRecords;
+ }
+
+ @Override
+ protected FileBasedSource<RecordWithMetadata> createForSubrangeOfFile(
+ MatchResult.Metadata metadata, long start, long end) {
+ return new ContextualTextIOSource(metadata, start, end, delimiter, hasMultilineCSVRecords);
+ }
+
+ @Override
+ protected FileBasedReader<RecordWithMetadata> createSingleFileReader(PipelineOptions options) {
+ return new MultiLineTextBasedReader(this, delimiter, hasMultilineCSVRecords);
+ }
+
+ @Override
+ public Coder<RecordWithMetadata> getOutputCoder() {
+ SchemaCoder<RecordWithMetadata> coder = null;
+ try {
+ coder = SchemaRegistry.createDefault().getSchemaCoder(RecordWithMetadata.class);
+ } catch (NoSuchSchemaException e) {
+ LOG.error("No Coder Found for RecordWithMetadata");
+ }
+ return coder;
+ }
+
+ /**
+ * A {@link FileBasedReader FileBasedReader} which can decode records delimited by delimiter
+ * characters.
+ *
+ * <p>See {@link ContextualTextIOSource } for further details.
+ */
+ @VisibleForTesting
+ static class MultiLineTextBasedReader extends FileBasedReader<RecordWithMetadata> {
+ public static final int READ_BUFFER_SIZE = 8192;
+ private static final ByteString UTF8_BOM =
+ ByteString.copyFrom(new byte[] {(byte) 0xEF, (byte) 0xBB, (byte) 0xBF});
+ private final ByteBuffer readBuffer = ByteBuffer.allocate(READ_BUFFER_SIZE);
+ private ByteString buffer;
+ private int startOfDelimiterInBuffer;
+ private int endOfDelimiterInBuffer;
+ private long startOfRecord;
+ private volatile long startOfNextRecord;
+ private volatile boolean eof;
+ private volatile boolean elementIsPresent;
+ private @Nullable RecordWithMetadata currentValue;
+ private @Nullable ReadableByteChannel inChannel;
+ private byte @Nullable [] delimiter;
+
+ // Add to override the isSplittable
+ private boolean hasMultilineCSVRecords;
+
+ private long startingOffset;
+ private long totalRecordCount;
+
+ private MultiLineTextBasedReader(
+ ContextualTextIOSource source, byte[] delimiter, boolean hasMultilineCSVRecords) {
+ super(source);
+ buffer = ByteString.EMPTY;
+ this.delimiter = delimiter;
+ this.hasMultilineCSVRecords = hasMultilineCSVRecords;
+ startingOffset = getCurrentSource().getStartOffset(); // Start offset;
+ }
+
+ @Override
+ protected long getCurrentOffset() throws NoSuchElementException {
+ if (!elementIsPresent) {
+ throw new NoSuchElementException();
+ }
+ return startOfRecord;
+ }
+
+ @Override
+ public long getSplitPointsRemaining() {
+ if (isStarted() && startOfNextRecord >= getCurrentSource().getEndOffset()) {
+ return isDone() ? 0 : 1;
+ }
+ return super.getSplitPointsRemaining();
+ }
+
+ @Override
+ public RecordWithMetadata getCurrent() throws NoSuchElementException {
+ if (!elementIsPresent) {
+ throw new NoSuchElementException();
+ }
+ return currentValue;
+ }
+
+ @Override
+ protected void startReading(ReadableByteChannel channel) throws IOException {
+ this.inChannel = channel;
+ // If the first offset is greater than zero, we need to skip bytes until we see our
+ // first delimiter.
+ long startOffset = getCurrentSource().getStartOffset();
+ if (startOffset > 0) {
+ Preconditions.checkState(
+ channel instanceof SeekableByteChannel,
+ "%s only supports reading from a SeekableByteChannel when given a start offset"
+ + " greater than 0.",
+ ContextualTextIOSource.class.getSimpleName());
+ long requiredPosition = startOffset - 1;
+ if (delimiter != null && startOffset >= delimiter.length) {
+ // we need to move back the offset of at worse delimiter.size to be sure to see
+ // all the bytes of the delimiter in the call to findDelimiterBounds() below
+ requiredPosition = startOffset - delimiter.length;
+ }
+ ((SeekableByteChannel) channel).position(requiredPosition);
+ findDelimiterBoundsWithMultiLineCheck();
+ buffer = buffer.substring(endOfDelimiterInBuffer);
+ startOfNextRecord = requiredPosition + endOfDelimiterInBuffer;
+ endOfDelimiterInBuffer = 0;
+ startOfDelimiterInBuffer = 0;
+ }
+ }
+
+ private void findDelimiterBoundsWithMultiLineCheck() throws IOException {
+ findDelimiterBounds();
+ }
+
+ /**
+ * Locates the start position and end position of the next delimiter. Will consume the channel
+ * till either EOF or the delimiter bounds are found.
+ *
+ * <p>If {@link ContextualTextIOSource#hasMultilineCSVRecords} is set then the behaviour will
+ * change from the standard read seen in {@link org.apache.beam.sdk.io.TextIO}. The assumption
+ * when {@link ContextualTextIOSource#hasMultilineCSVRecords} is set is that the file is being
+ * read with a single thread.
+ *
+ * <p>This fills the buffer and updates the positions as follows:
+ *
+ * <pre>{@code
+ * ------------------------------------------------------
+ * | element bytes | delimiter bytes | unconsumed bytes |
+ * ------------------------------------------------------
+ * 0 start of end of buffer
+ * delimiter delimiter size
+ * in buffer in buffer
+ * }</pre>
+ */
+ private void findDelimiterBounds() throws IOException {
+ int bytePositionInBuffer = 0;
+ boolean doubleQuoteClosed = true;
+
+ while (true) {
+ if (!tryToEnsureNumberOfBytesInBuffer(bytePositionInBuffer + 1)) {
+ startOfDelimiterInBuffer = endOfDelimiterInBuffer = bytePositionInBuffer;
+ break;
+ }
+
+ byte currentByte = buffer.byteAt(bytePositionInBuffer);
+ if (hasMultilineCSVRecords) {
+ // Check if we are inside an open Quote
+ if (currentByte == '"') {
+ doubleQuoteClosed = !doubleQuoteClosed;
+ }
+ } else {
+ doubleQuoteClosed = true;
+ }
+
+ if (delimiter == null) {
+ // default delimiter
+ if (currentByte == '\n') {
+ startOfDelimiterInBuffer = bytePositionInBuffer;
+ endOfDelimiterInBuffer = startOfDelimiterInBuffer + 1;
+ if (doubleQuoteClosed) {
+ break;
+ }
+ } else if (currentByte == '\r') {
+ startOfDelimiterInBuffer = bytePositionInBuffer;
+ endOfDelimiterInBuffer = startOfDelimiterInBuffer + 1;
+ if (tryToEnsureNumberOfBytesInBuffer(bytePositionInBuffer + 2)) {
+ currentByte = buffer.byteAt(bytePositionInBuffer + 1);
+ if (currentByte == '\n') {
+ endOfDelimiterInBuffer += 1;
+ }
+ }
+ if (doubleQuoteClosed) {
+ break;
+ }
+ }
+ } else {
+ // when the user defines a delimiter
+ int i = 0;
+ startOfDelimiterInBuffer = endOfDelimiterInBuffer = bytePositionInBuffer;
+ while ((i < delimiter.length) && (currentByte == delimiter[i])) {
+ // read next byte;
+ i++;
+ if (tryToEnsureNumberOfBytesInBuffer(bytePositionInBuffer + i + 1)) {
+ currentByte = buffer.byteAt(bytePositionInBuffer + i);
+ } else {
+ // corner case: delimiter truncate at the end of file
+ startOfDelimiterInBuffer = endOfDelimiterInBuffer = bytePositionInBuffer;
+ break;
+ }
+ }
+ if (i == delimiter.length) {
+ endOfDelimiterInBuffer = bytePositionInBuffer + i;
+ if (doubleQuoteClosed) {
+ break;
+ }
+ }
+ }
+ bytePositionInBuffer += 1;
+ }
+ }
+
+ @Override
+ protected boolean readNextRecord() throws IOException {
+ startOfRecord = startOfNextRecord;
+
+ findDelimiterBoundsWithMultiLineCheck();
+
+ // If we have reached EOF file and consumed all of the buffer then we know
+ // that there are no more records.
+ if (eof && buffer.isEmpty()) {
+ elementIsPresent = false;
+ return false;
+ }
+
+ decodeCurrentElement();
+ startOfNextRecord = startOfRecord + endOfDelimiterInBuffer;
+ return true;
+ }
+
+ /**
+ * Decodes the current element updating the buffer to only contain the unconsumed bytes.
+ *
+ * <p>This invalidates the currently stored {@code startOfDelimiterInBuffer} and {@code
+ * endOfDelimiterInBuffer}.
+ */
+ private void decodeCurrentElement() throws IOException {
+ ByteString dataToDecode = buffer.substring(0, startOfDelimiterInBuffer);
+ // If present, the UTF8 Byte Order Mark (BOM) will be removed.
+ if (startOfRecord == 0 && dataToDecode.startsWith(UTF8_BOM)) {
+ dataToDecode = dataToDecode.substring(UTF8_BOM.size());
+ }
+
+ // The line num is:
+ Long recordUniqueNum = totalRecordCount++;
+ // The Complete FileName (with uri if this is a web url eg: temp/abc.txt) is:
+ String fileName = getCurrentSource().getSingleFileMetadata().resourceId().toString();
Review comment:
@rezarokni thanks !
I was facing this same issue that is why could not make it part of the previous commit.
FileName does seem to be the best option for now.
##########
File path: sdks/java/io/contextual-text-io/src/main/java/org/apache/beam/sdk/io/contextualtextio/RecordWithMetadata.java
##########
@@ -0,0 +1,85 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one
+ * or more contributor license agreements. See the NOTICE file
+ * distributed with this work for additional information
+ * regarding copyright ownership. The ASF licenses this file
+ * to you under the Apache License, Version 2.0 (the
+ * "License"); you may not use this file except in compliance
+ * with the License. You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+package org.apache.beam.sdk.io.contextualtextio;
+
+import com.google.auto.value.AutoValue;
+import org.apache.beam.sdk.annotations.Experimental;
+import org.apache.beam.sdk.io.fs.ResourceId;
+import org.apache.beam.sdk.schemas.AutoValueSchema;
+import org.apache.beam.sdk.schemas.annotations.DefaultSchema;
+
+/**
+ * Helper Class based on {@link AutoValueSchema}, it provides Metadata associated with each Record
+ * when reading from file(s) using {@link ContextualTextIO}.
+ *
+ * <h3>Fields:</h3>
+ *
+ * <ul>
+ * <li>recordOffset: The offset of a record (the byte at which the record begins) in a file. This
+ * information can be useful if you wish to reconstruct the file. {@link
+ * RecordWithMetadata#getRecordOffset()}
+ * <li>recordNum: The ordinal number of the record in its file. {@link
+ * RecordWithMetadata#getRecordNum()}
+ * <li>recordValue: The value / contents of the record {@link RecordWithMetadata#getValue()}
+ * <li>rangeOffset: The starting offset of the range (split), which contained the record, when the
+ * record was read. {@link RecordWithMetadata#getRangeOffset()}
+ * <li>recordNumInOffset: The record number relative to the Range. (line number within the range)
+ * {@link RecordWithMetadata#getRecordNumInOffset()}
+ * <li>fileName: Name of the file to which the record belongs (this is the full filename,
+ * eg:path/to/file.txt) {@link RecordWithMetadata#getFileName()}
+ * </ul>
+ */
+@Experimental(Experimental.Kind.SCHEMAS)
+@DefaultSchema(AutoValueSchema.class)
+@AutoValue
+public abstract class RecordWithMetadata {
+ public abstract long getRecordOffset();
+
+ public abstract long getRecordNum();
+
+ public abstract String getValue();
+
+ public abstract long getRangeOffset();
+
+ public abstract long getRecordNumInOffset();
+
+ public abstract Builder toBuilder();
+
+ public abstract String getFileName();
+
+ public static Builder newBuilder() {
+ return new AutoValue_RecordWithMetadata.Builder();
+ }
+
+ @AutoValue.Builder
+ public abstract static class Builder {
+ public abstract Builder setRecordNum(long lineNum);
+
+ public abstract Builder setRecordOffset(long recordOffset);
+
+ public abstract Builder setValue(String Value);
Review comment:
Done.
##########
File path: sdks/java/io/contextual-text-io/src/main/java/org/apache/beam/sdk/io/contextualtextio/ContextualTextIO.java
##########
@@ -0,0 +1,631 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one
+ * or more contributor license agreements. See the NOTICE file
+ * distributed with this work for additional information
+ * regarding copyright ownership. The ASF licenses this file
+ * to you under the Apache License, Version 2.0 (the
+ * "License"); you may not use this file except in compliance
+ * with the License. You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+package org.apache.beam.sdk.io.contextualtextio;
+
+import static org.apache.beam.sdk.io.FileIO.ReadMatches.DirectoryTreatment;
+import static org.apache.beam.vendor.guava.v26_0_jre.com.google.common.base.Preconditions.checkArgument;
+import static org.apache.beam.vendor.guava.v26_0_jre.com.google.common.base.Preconditions.checkNotNull;
+
+import com.google.auto.value.AutoValue;
+import java.nio.ByteBuffer;
+import java.util.Arrays;
+import java.util.Comparator;
+import java.util.HashMap;
+import java.util.Map;
+import java.util.SortedMap;
+import java.util.TreeMap;
+import org.apache.beam.sdk.coders.StringUtf8Coder;
+import org.apache.beam.sdk.io.CompressedSource;
+import org.apache.beam.sdk.io.Compression;
+import org.apache.beam.sdk.io.FileBasedSource;
+import org.apache.beam.sdk.io.FileIO;
+import org.apache.beam.sdk.io.FileIO.MatchConfiguration;
+import org.apache.beam.sdk.io.ReadAllViaFileBasedSource;
+import org.apache.beam.sdk.io.TextIO;
+import org.apache.beam.sdk.io.fs.EmptyMatchTreatment;
+import org.apache.beam.sdk.options.ValueProvider;
+import org.apache.beam.sdk.options.ValueProvider.StaticValueProvider;
+import org.apache.beam.sdk.schemas.NoSuchSchemaException;
+import org.apache.beam.sdk.schemas.SchemaCoder;
+import org.apache.beam.sdk.transforms.Count;
+import org.apache.beam.sdk.transforms.Create;
+import org.apache.beam.sdk.transforms.DoFn;
+import org.apache.beam.sdk.transforms.PTransform;
+import org.apache.beam.sdk.transforms.ParDo;
+import org.apache.beam.sdk.transforms.SerializableFunction;
+import org.apache.beam.sdk.transforms.View;
+import org.apache.beam.sdk.transforms.Watch.Growth.TerminationCondition;
+import org.apache.beam.sdk.transforms.display.DisplayData;
+import org.apache.beam.sdk.values.KV;
+import org.apache.beam.sdk.values.PBegin;
+import org.apache.beam.sdk.values.PCollection;
+import org.apache.beam.sdk.values.PCollectionView;
+import org.apache.beam.vendor.guava.v26_0_jre.com.google.common.annotations.VisibleForTesting;
+import org.checkerframework.checker.nullness.qual.Nullable;
+import org.joda.time.Duration;
+import org.slf4j.Logger;
+import org.slf4j.LoggerFactory;
+
+/**
+ * {@link PTransform}s that read text files and collect contextual information of the elements in
+ * the input.
+ *
+ * <p>Use {@link TextIO} when not reading file with Multiline Records or additional metadata is not
+ * required.
+ *
+ * <h2>Reading from text files</h2>
+ *
+ * <p>To read a {@link PCollection} from one or more text files, use {@code
+ * ContextualTextIO.read()}. To instantiate a transform use {@link
+ * ContextualTextIO.Read#from(String)} and specify the path of the file(s) to be read.
+ * Alternatively, if the filenames to be read are themselves in a {@link PCollection} you can use
+ * {@link FileIO} to match them and {@link ContextualTextIO#readFiles()} to read them.
+ *
+ * <p>{@link #read} returns a {@link PCollection} of {@link RecordWithMetadata RecordWithMetadata},
+ * each corresponding to one line of an input UTF-8 text file (split into lines delimited by '\n',
+ * '\r', '\r\n', or specified delimiter see {@link ContextualTextIO.Read#withDelimiter})
+ *
+ * <h3>Filepattern expansion and watching</h3>
+ *
+ * <p>By default, the filepatterns are expanded only once. The combination of {@link
+ * FileIO.Match#continuously(Duration, TerminationCondition)} and {@link #readFiles()} allow
+ * streaming of new files matching the filepattern(s).
+ *
+ * <p>By default, {@link #read} prohibits filepatterns that match no files, and {@link #readFiles()}
+ * allows them in case the filepattern contains a glob wildcard character. Use {@link
+ * ContextualTextIO.Read#withEmptyMatchTreatment} or {@link
+ * FileIO.Match#withEmptyMatchTreatment(EmptyMatchTreatment)} plus {@link #readFiles()} to configure
+ * this behavior.
+ *
+ * <p>Example 1: reading a file or filepattern.
+ *
+ * <pre>{@code
+ * Pipeline p = ...;
+ *
+ * // A simple Read of a file:
+ * PCollection<RecordWithMetadata> records = p.apply(ContextualTextIO.read().from("/local/path/to/file.txt"));
+ * }</pre>
+ *
+ * <p>Example 2: reading a PCollection of filenames.
+ *
+ * <pre>{@code
+ * Pipeline p = ...;
+ *
+ * // E.g. the filenames might be computed from other data in the pipeline, or
+ * // read from a data source.
+ * PCollection<String> filenames = ...;
+ *
+ * // Read all files in the collection.
+ * PCollection<RecordWithMetadata> records =
+ * filenames
+ * .apply(FileIO.matchAll())
+ * .apply(FileIO.readMatches())
+ * .apply(ContextualTextIO.readFiles());
+ * }</pre>
+ *
+ * <p>Example 3: streaming new files matching a filepattern.
+ *
+ * <pre>{@code
+ * Pipeline p = ...;
+ *
+ * PCollection<RecordWithMetadata> records = p.apply(ContextualTextIO.read()
+ * .from("/local/path/to/files/*")
+ * .watchForNewFiles(
+ * // Check for new files every minute
+ * Duration.standardMinutes(1),
+ * // Stop watching the filepattern if no new files appear within an hour
+ * afterTimeSinceNewOutput(Duration.standardHours(1))));
+ * }</pre>
+ *
+ * <p>Example 4: reading a file or file pattern of RFC4180-compliant CSV files with fields that may
+ * contain line breaks.
+ *
+ * Example of such a file could be:
+ *
+ * "aaa","b CRLF
+ * bb","ccc" CRLF
+ * zzz,yyy,xxx
+ *
+ * <pre>{@code
+ * Pipeline p = ...;
+ *
+ * PCollection<RecordWithMetadata> records = p.apply(ContextualTextIO.read()
+ * .from("/local/path/to/files/*.csv")
+ * .withHasMultilineCSVRecords(true));
+ * }</pre>
+ *
+ * <p>Example 5: reading while watching for new files
+ *
+ * <pre>{@code
+ * Pipeline p = ...;
+ *
+ * PCollection<RecordWithMetadata> records = p.apply(FileIO.match()
+ * .filepattern("filepattern")
+ * .continuously(
+ * Duration.millis(100),
+ * Watch.Growth.afterTimeSinceNewOutput(Duration.standardSeconds(3))))
+ * .apply(FileIO.readMatches())
+ * .apply(ContextualTextIO.readFiles());
+ * }</pre>
+ *
+ * <p>Example 6: reading without recordNum metadata, or only fileName associated Metadata. (the
+ * Objects would still contain recordNums, but these recordNums would correspond to their positions
+ * in their respective offsets rather than their positions within the entire file).
+ *
+ * <pre>{@code
+ * Pipeline p = ...;
+ *
+ * PCollection<RecordWithMetadata> records = p.apply(ContextualTextIO.read()
+ * .from("/local/path/to/files/*.csv")
+ * .setWithoutRecordNumMetadata(true));
+ * }</pre>
+ *
+ * <p>NOTE: When using {@link ContextualTextIO.Read#withHasMultilineCSVRecords(Boolean)} this
+ * option, a single reader will be used to process the file, rather than multiple readers which can
+ * read from different offsets. For a large file this can result in lower performance.
+ *
+ * <p>NOTE: Use {@link Read#withoutRecordNumMetadata()} when recordNum metadata is not required, for
+ * example, when when only filename metadata is required. Computing record positions currently
+ * introduces a shuffle step, which increases the resources used by the pipeline. <b> By default
+ * withoutRecordNumMetadata is set to false, so the shuffle step is performed.</b>
+ *
+ * <h3>Reading a very large number of files</h3>
+ *
+ * <p>If it is known that the filepattern will match a very large number of files (e.g. tens of
+ * thousands or more), use {@link ContextualTextIO.Read#withHintMatchesManyFiles} for better
+ * performance and scalability. Note that it may decrease performance if the filepattern matches
+ * only a small number of files.
+ */
+public class ContextualTextIO {
+ private static final long DEFAULT_BUNDLE_SIZE_BYTES = 64 * 1024 * 1024L;
+ private static final Logger LOG = LoggerFactory.getLogger(ContextualTextIO.class);
+
+ /**
+ * A {@link PTransform} that reads from one or more text files and returns a bounded {@link
+ * PCollection} containing one {@link RecordWithMetadata}element for each line of the input files.
+ */
+ public static Read read() {
+ return new AutoValue_ContextualTextIO_Read.Builder()
+ .setCompression(Compression.AUTO)
+ .setHintMatchesManyFiles(false)
+ .setWithoutRecordNumMetadata(false)
+ .setMatchConfiguration(MatchConfiguration.create(EmptyMatchTreatment.DISALLOW))
+ .setHasMultilineCSVRecords(false)
+ .build();
+ }
+
+ /**
+ * Like {@link #read}, but reads each file in a {@link PCollection} of {@link
+ * FileIO.ReadableFile}, returned by {@link FileIO#readMatches}.
+ */
+ public static ReadFiles readFiles() {
+ return new AutoValue_ContextualTextIO_ReadFiles.Builder()
+ // 64MB is a reasonable value that allows to amortize the cost of opening files,
+ // but is not so large as to exhaust a typical runner's maximum amount of output per
+ // ProcessElement call.
+ .setDesiredBundleSizeBytes(DEFAULT_BUNDLE_SIZE_BYTES)
+ .setHasMultilineCSVRecords(false)
+ .build();
+ }
+
+ /** Implementation of {@link #read}. */
+ @AutoValue
+ public abstract static class Read extends PTransform<PBegin, PCollection<RecordWithMetadata>> {
+ abstract @Nullable ValueProvider<String> getFilepattern();
+
+ abstract MatchConfiguration getMatchConfiguration();
+
+ abstract boolean getHintMatchesManyFiles();
+
+ abstract boolean getWithoutRecordNumMetadata();
+
+ abstract Compression getCompression();
+
+ abstract @Nullable Boolean getHasMultilineCSVRecords();
+
+ @SuppressWarnings("mutable") // this returns an array that can be mutated by the caller
+ abstract byte @Nullable [] getDelimiter();
+
+ abstract Builder toBuilder();
+
+ @AutoValue.Builder
+ abstract static class Builder {
+ abstract Builder setFilepattern(ValueProvider<String> filepattern);
+
+ abstract Builder setMatchConfiguration(MatchConfiguration matchConfiguration);
+
+ abstract Builder setHintMatchesManyFiles(boolean hintManyFiles);
+
+ abstract Builder setWithoutRecordNumMetadata(boolean withoutLineNumMetadata);
+
+ abstract Builder setCompression(Compression compression);
+
+ abstract Builder setDelimiter(byte @Nullable [] delimiter);
+
+ abstract Builder setHasMultilineCSVRecords(Boolean hasMultilineCSVRecords);
+
+ abstract Read build();
+ }
+
+ /**
+ * Reads text from the file(s) with the given filename or filename pattern.
+ *
+ * <p>This can be a local path (if running locally), or a Google Cloud Storage filename or
+ * filename pattern of the form {@code "gs://<bucket>/<filepath>"} (if running locally or using
+ * remote execution service).
+ *
+ * <p>Standard <a href="http://docs.oracle.com/javase/tutorial/essential/io/find.html" >Java
+ * Filesystem glob patterns</a> ("*", "?", "[..]") are supported.
+ *
+ * <p>If it is known that the filepattern will match a very large number of files (at least tens
+ * of thousands), use {@link #withHintMatchesManyFiles} for better performance and scalability.
+ */
+ public Read from(String filepattern) {
+ checkArgument(filepattern != null, "filepattern can not be null");
+ return from(StaticValueProvider.of(filepattern));
+ }
+
+ /** Same as {@code from(filepattern)}, but accepting a {@link ValueProvider}. */
+ public Read from(ValueProvider<String> filepattern) {
+ checkArgument(filepattern != null, "filepattern can not be null");
+ return toBuilder().setFilepattern(filepattern).build();
+ }
+
+ /** Sets the {@link MatchConfiguration}. */
+ public Read withMatchConfiguration(MatchConfiguration matchConfiguration) {
+ return toBuilder().setMatchConfiguration(matchConfiguration).build();
+ }
+
+ /**
+ * When reading RFC4180 CSV files that have values that span multiple lines, set this to true.
+ * Note: this reduces the read performance (see: {@link ContextualTextIO}).
+ */
+ public Read withHasMultilineCSVRecords(Boolean hasMultilineCSVRecords) {
+ return toBuilder().setHasMultilineCSVRecords(hasMultilineCSVRecords).build();
+ }
+
+ /**
+ * Reads from input sources using the specified compression type.
+ *
+ * <p>If no compression type is specified, the default is {@link Compression#AUTO}.
+ */
+ public Read withCompression(Compression compression) {
+ return toBuilder().setCompression(compression).build();
+ }
+
+ /**
+ * Hints that the filepattern specified in {@link #from(String)} matches a very large number of
+ * files.
+ *
+ * <p>This hint may cause a runner to execute the transform differently, in a way that improves
+ * performance for this case, but it may worsen performance if the filepattern matches only a
+ * small number of files (e.g., in a runner that supports dynamic work rebalancing, it will
+ * happen less efficiently within individual files).
+ */
+ public Read withHintMatchesManyFiles() {
+ return toBuilder().setHintMatchesManyFiles(true).build();
+ }
+
+ /**
+ * Allows the user to opt out of getting recordNums associated with each record.
+ *
+ * <p>When set to true, it will introduce a shuffle step to assemble the recordNums for each
+ * record, which will increase the resources used by the pipeline.
+ *
+ * <p>Use this when metadata like fileNames are required and their position/order can be
+ * ignored.
+ */
+ public Read withoutRecordNumMetadata() {
+ return toBuilder().setWithoutRecordNumMetadata(true).build();
+ }
+
+ /** See {@link MatchConfiguration#withEmptyMatchTreatment}. */
+ public Read withEmptyMatchTreatment(EmptyMatchTreatment treatment) {
+ return withMatchConfiguration(getMatchConfiguration().withEmptyMatchTreatment(treatment));
+ }
+
+ /** Set the custom delimiter to be used in place of the default ones ('\r', '\n' or '\r\n'). */
+ public Read withDelimiter(byte[] delimiter) {
+ checkArgument(delimiter != null, "delimiter can not be null");
+ checkArgument(!isSelfOverlapping(delimiter), "delimiter must not self-overlap");
+ return toBuilder().setDelimiter(delimiter).build();
+ }
+
+ static boolean isSelfOverlapping(byte[] s) {
+ // s self-overlaps if v exists such as s = vu = wv with u and w non empty
+ for (int i = 1; i < s.length - 1; ++i) {
+ if (ByteBuffer.wrap(s, 0, i).equals(ByteBuffer.wrap(s, s.length - i, i))) {
+ return true;
+ }
+ }
+ return false;
+ }
+
+ @Override
+ public PCollection<RecordWithMetadata> expand(PBegin input) {
+ checkNotNull(
+ getFilepattern(), "need to set the filepattern of a ContextualTextIO.Read transform");
+ PCollection<RecordWithMetadata> records = null;
+ if (getMatchConfiguration().getWatchInterval() == null && !getHintMatchesManyFiles()) {
+ records = input.apply("Read", org.apache.beam.sdk.io.Read.from(getSource()));
+ } else {
+ // All other cases go through FileIO + ReadFiles
+ records =
+ input
+ .apply(
+ "Create filepattern", Create.ofProvider(getFilepattern(), StringUtf8Coder.of()))
+ .apply("Match All", FileIO.matchAll().withConfiguration(getMatchConfiguration()))
+ .apply(
+ "Read Matches",
+ FileIO.readMatches()
+ .withCompression(getCompression())
+ .withDirectoryTreatment(DirectoryTreatment.PROHIBIT))
+ .apply("Via ReadFiles", readFiles().withDelimiter(getDelimiter()));
+ }
+
+ // Check if the user decided to opt out of recordNums associated with records
+ if (getWithoutRecordNumMetadata()) {
+ return records;
+ }
+
+ /*
+ * At this point the line number in RecordWithMetadata contains the relative line offset from the beginning of the read range.
+ *
+ * To compute the absolute position from the beginning of the input we group the lines within the same ranges, and evaluate the size of each range.
+ */
+
+ PCollection<KV<KV<String, Long>, RecordWithMetadata>> recordsGroupedByFileAndRange =
+ records.apply("AddFileNameAndRange", ParDo.of(new AddFileNameAndRange()));
+
+ PCollectionView<Map<KV<String, Long>, Long>> rangeSizes =
+ recordsGroupedByFileAndRange
+ .apply("CountRecordsForEachFileRange", Count.perKey())
+ .apply("SizesAsView", View.asMap());
+
+ // Get Pipeline to create a dummy PCollection with one element to help compute the lines
+ // before each Range
+ PCollection<Integer> singletonPcoll =
+ input.getPipeline().apply("CreateSingletonPcoll", Create.of(Arrays.asList(1)));
+
+ /*
+ * For each (File, Offset) pair, calculate the number of lines occurring before the Range for each file
+ *
+ * After computing the number of lines before each range, we can find the line number in original file as numLiesBeforeOffset + lineNumInCurrentOffset
+ */
+
+ PCollectionView<Map<KV<String, Long>, Long>> numRecordsBeforeEachRange =
+ singletonPcoll
+ .apply(
+ "ComputeRecordsBeforeRange",
+ ParDo.of(new ComputeRecordsBeforeEachRange(rangeSizes))
+ .withSideInputs(rangeSizes))
+ .apply("NumRecordsBeforeEachRangeAsView", View.asMap());
+
+ return recordsGroupedByFileAndRange.apply(
+ "AssignLineNums",
+ ParDo.of(new AssignRecordNums(numRecordsBeforeEachRange))
+ .withSideInputs(numRecordsBeforeEachRange));
+ }
+
+ @VisibleForTesting
+ static class AddFileNameAndRange
+ extends DoFn<RecordWithMetadata, KV<KV<String, Long>, RecordWithMetadata>> {
+ @ProcessElement
+ public void processElement(
+ @Element RecordWithMetadata record,
+ OutputReceiver<KV<KV<String, Long>, RecordWithMetadata>> out) {
+ out.output(KV.of(KV.of(record.getFileName().toString(), record.getRangeOffset()), record));
+ }
+ }
+
+ /**
+ * Helper class for computing number of record in the File preceding the beginning of the Range
+ * in this file.
+ */
+ @VisibleForTesting
+ static class ComputeRecordsBeforeEachRange extends DoFn<Integer, KV<KV<String, Long>, Long>> {
+ private final PCollectionView<Map<KV<String, Long>, Long>> rangeSizes;
+
+ public ComputeRecordsBeforeEachRange(
+ PCollectionView<Map<KV<String, Long>, Long>> rangeSizes) {
+ this.rangeSizes = rangeSizes;
+ }
+
+ // Add custom comparator as KV<K, V> is not comparable by default
+ private static class FileRangeComparator<K extends Comparable<K>, V extends Comparable<V>>
+ implements Comparator<KV<K, V>> {
+ @Override
+ public int compare(KV<K, V> a, KV<K, V> b) {
+ if (a.getKey().compareTo(b.getKey()) == 0) {
+ return a.getValue().compareTo(b.getValue());
+ }
+ return a.getKey().compareTo(b.getKey());
+ }
+ }
+
+ @ProcessElement
+ public void processElement(ProcessContext p) {
+ // Get the Map Containing the size from side-input
+ Map<KV<String, Long>, Long> rangeSizesMap = p.sideInput(rangeSizes);
+
+ // The FileRange Pair must be sorted
+ SortedMap<KV<String, Long>, Long> sorted = new TreeMap<>(new FileRangeComparator<>());
+
+ // Initialize sorted map with values
+ for (Map.Entry<KV<String, Long>, Long> entry : rangeSizesMap.entrySet()) {
+ sorted.put(entry.getKey(), entry.getValue());
+ }
+
+ // HashMap that tracks number of records passed for each file
+ Map<String, Long> pastRecords = new HashMap<>();
+
+ // For each (File, Range) Pair, compute the number of records before it
+ for (Map.Entry entry : sorted.entrySet()) {
Review comment:
Done
##########
File path: sdks/java/io/contextual-text-io/src/main/java/org/apache/beam/sdk/io/contextualtextio/ContextualTextIO.java
##########
@@ -0,0 +1,631 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one
+ * or more contributor license agreements. See the NOTICE file
+ * distributed with this work for additional information
+ * regarding copyright ownership. The ASF licenses this file
+ * to you under the Apache License, Version 2.0 (the
+ * "License"); you may not use this file except in compliance
+ * with the License. You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+package org.apache.beam.sdk.io.contextualtextio;
+
+import static org.apache.beam.sdk.io.FileIO.ReadMatches.DirectoryTreatment;
+import static org.apache.beam.vendor.guava.v26_0_jre.com.google.common.base.Preconditions.checkArgument;
+import static org.apache.beam.vendor.guava.v26_0_jre.com.google.common.base.Preconditions.checkNotNull;
+
+import com.google.auto.value.AutoValue;
+import java.nio.ByteBuffer;
+import java.util.Arrays;
+import java.util.Comparator;
+import java.util.HashMap;
+import java.util.Map;
+import java.util.SortedMap;
+import java.util.TreeMap;
+import org.apache.beam.sdk.coders.StringUtf8Coder;
+import org.apache.beam.sdk.io.CompressedSource;
+import org.apache.beam.sdk.io.Compression;
+import org.apache.beam.sdk.io.FileBasedSource;
+import org.apache.beam.sdk.io.FileIO;
+import org.apache.beam.sdk.io.FileIO.MatchConfiguration;
+import org.apache.beam.sdk.io.ReadAllViaFileBasedSource;
+import org.apache.beam.sdk.io.TextIO;
+import org.apache.beam.sdk.io.fs.EmptyMatchTreatment;
+import org.apache.beam.sdk.options.ValueProvider;
+import org.apache.beam.sdk.options.ValueProvider.StaticValueProvider;
+import org.apache.beam.sdk.schemas.NoSuchSchemaException;
+import org.apache.beam.sdk.schemas.SchemaCoder;
+import org.apache.beam.sdk.transforms.Count;
+import org.apache.beam.sdk.transforms.Create;
+import org.apache.beam.sdk.transforms.DoFn;
+import org.apache.beam.sdk.transforms.PTransform;
+import org.apache.beam.sdk.transforms.ParDo;
+import org.apache.beam.sdk.transforms.SerializableFunction;
+import org.apache.beam.sdk.transforms.View;
+import org.apache.beam.sdk.transforms.Watch.Growth.TerminationCondition;
+import org.apache.beam.sdk.transforms.display.DisplayData;
+import org.apache.beam.sdk.values.KV;
+import org.apache.beam.sdk.values.PBegin;
+import org.apache.beam.sdk.values.PCollection;
+import org.apache.beam.sdk.values.PCollectionView;
+import org.apache.beam.vendor.guava.v26_0_jre.com.google.common.annotations.VisibleForTesting;
+import org.checkerframework.checker.nullness.qual.Nullable;
+import org.joda.time.Duration;
+import org.slf4j.Logger;
+import org.slf4j.LoggerFactory;
+
+/**
+ * {@link PTransform}s that read text files and collect contextual information of the elements in
+ * the input.
+ *
+ * <p>Use {@link TextIO} when not reading file with Multiline Records or additional metadata is not
+ * required.
+ *
+ * <h2>Reading from text files</h2>
+ *
+ * <p>To read a {@link PCollection} from one or more text files, use {@code
+ * ContextualTextIO.read()}. To instantiate a transform use {@link
+ * ContextualTextIO.Read#from(String)} and specify the path of the file(s) to be read.
+ * Alternatively, if the filenames to be read are themselves in a {@link PCollection} you can use
+ * {@link FileIO} to match them and {@link ContextualTextIO#readFiles()} to read them.
+ *
+ * <p>{@link #read} returns a {@link PCollection} of {@link RecordWithMetadata RecordWithMetadata},
+ * each corresponding to one line of an input UTF-8 text file (split into lines delimited by '\n',
+ * '\r', '\r\n', or specified delimiter see {@link ContextualTextIO.Read#withDelimiter})
+ *
+ * <h3>Filepattern expansion and watching</h3>
+ *
+ * <p>By default, the filepatterns are expanded only once. The combination of {@link
+ * FileIO.Match#continuously(Duration, TerminationCondition)} and {@link #readFiles()} allow
+ * streaming of new files matching the filepattern(s).
+ *
+ * <p>By default, {@link #read} prohibits filepatterns that match no files, and {@link #readFiles()}
+ * allows them in case the filepattern contains a glob wildcard character. Use {@link
+ * ContextualTextIO.Read#withEmptyMatchTreatment} or {@link
+ * FileIO.Match#withEmptyMatchTreatment(EmptyMatchTreatment)} plus {@link #readFiles()} to configure
+ * this behavior.
+ *
+ * <p>Example 1: reading a file or filepattern.
+ *
+ * <pre>{@code
+ * Pipeline p = ...;
+ *
+ * // A simple Read of a file:
+ * PCollection<RecordWithMetadata> records = p.apply(ContextualTextIO.read().from("/local/path/to/file.txt"));
+ * }</pre>
+ *
+ * <p>Example 2: reading a PCollection of filenames.
+ *
+ * <pre>{@code
+ * Pipeline p = ...;
+ *
+ * // E.g. the filenames might be computed from other data in the pipeline, or
+ * // read from a data source.
+ * PCollection<String> filenames = ...;
+ *
+ * // Read all files in the collection.
+ * PCollection<RecordWithMetadata> records =
+ * filenames
+ * .apply(FileIO.matchAll())
+ * .apply(FileIO.readMatches())
+ * .apply(ContextualTextIO.readFiles());
+ * }</pre>
+ *
+ * <p>Example 3: streaming new files matching a filepattern.
+ *
+ * <pre>{@code
+ * Pipeline p = ...;
+ *
+ * PCollection<RecordWithMetadata> records = p.apply(ContextualTextIO.read()
+ * .from("/local/path/to/files/*")
+ * .watchForNewFiles(
+ * // Check for new files every minute
+ * Duration.standardMinutes(1),
+ * // Stop watching the filepattern if no new files appear within an hour
+ * afterTimeSinceNewOutput(Duration.standardHours(1))));
+ * }</pre>
+ *
+ * <p>Example 4: reading a file or file pattern of RFC4180-compliant CSV files with fields that may
+ * contain line breaks.
+ *
+ * Example of such a file could be:
+ *
+ * "aaa","b CRLF
+ * bb","ccc" CRLF
+ * zzz,yyy,xxx
+ *
+ * <pre>{@code
+ * Pipeline p = ...;
+ *
+ * PCollection<RecordWithMetadata> records = p.apply(ContextualTextIO.read()
+ * .from("/local/path/to/files/*.csv")
+ * .withHasMultilineCSVRecords(true));
+ * }</pre>
+ *
+ * <p>Example 5: reading while watching for new files
+ *
+ * <pre>{@code
+ * Pipeline p = ...;
+ *
+ * PCollection<RecordWithMetadata> records = p.apply(FileIO.match()
+ * .filepattern("filepattern")
+ * .continuously(
+ * Duration.millis(100),
+ * Watch.Growth.afterTimeSinceNewOutput(Duration.standardSeconds(3))))
+ * .apply(FileIO.readMatches())
+ * .apply(ContextualTextIO.readFiles());
+ * }</pre>
+ *
+ * <p>Example 6: reading without recordNum metadata, or only fileName associated Metadata. (the
+ * Objects would still contain recordNums, but these recordNums would correspond to their positions
+ * in their respective offsets rather than their positions within the entire file).
+ *
+ * <pre>{@code
+ * Pipeline p = ...;
+ *
+ * PCollection<RecordWithMetadata> records = p.apply(ContextualTextIO.read()
+ * .from("/local/path/to/files/*.csv")
+ * .setWithoutRecordNumMetadata(true));
+ * }</pre>
+ *
+ * <p>NOTE: When using {@link ContextualTextIO.Read#withHasMultilineCSVRecords(Boolean)} this
+ * option, a single reader will be used to process the file, rather than multiple readers which can
+ * read from different offsets. For a large file this can result in lower performance.
+ *
+ * <p>NOTE: Use {@link Read#withoutRecordNumMetadata()} when recordNum metadata is not required, for
+ * example, when when only filename metadata is required. Computing record positions currently
+ * introduces a shuffle step, which increases the resources used by the pipeline. <b> By default
+ * withoutRecordNumMetadata is set to false, so the shuffle step is performed.</b>
+ *
+ * <h3>Reading a very large number of files</h3>
+ *
+ * <p>If it is known that the filepattern will match a very large number of files (e.g. tens of
+ * thousands or more), use {@link ContextualTextIO.Read#withHintMatchesManyFiles} for better
+ * performance and scalability. Note that it may decrease performance if the filepattern matches
+ * only a small number of files.
+ */
+public class ContextualTextIO {
+ private static final long DEFAULT_BUNDLE_SIZE_BYTES = 64 * 1024 * 1024L;
+ private static final Logger LOG = LoggerFactory.getLogger(ContextualTextIO.class);
+
+ /**
+ * A {@link PTransform} that reads from one or more text files and returns a bounded {@link
+ * PCollection} containing one {@link RecordWithMetadata}element for each line of the input files.
+ */
+ public static Read read() {
+ return new AutoValue_ContextualTextIO_Read.Builder()
+ .setCompression(Compression.AUTO)
+ .setHintMatchesManyFiles(false)
+ .setWithoutRecordNumMetadata(false)
+ .setMatchConfiguration(MatchConfiguration.create(EmptyMatchTreatment.DISALLOW))
+ .setHasMultilineCSVRecords(false)
+ .build();
+ }
+
+ /**
+ * Like {@link #read}, but reads each file in a {@link PCollection} of {@link
+ * FileIO.ReadableFile}, returned by {@link FileIO#readMatches}.
+ */
+ public static ReadFiles readFiles() {
+ return new AutoValue_ContextualTextIO_ReadFiles.Builder()
+ // 64MB is a reasonable value that allows to amortize the cost of opening files,
+ // but is not so large as to exhaust a typical runner's maximum amount of output per
+ // ProcessElement call.
+ .setDesiredBundleSizeBytes(DEFAULT_BUNDLE_SIZE_BYTES)
+ .setHasMultilineCSVRecords(false)
+ .build();
+ }
+
+ /** Implementation of {@link #read}. */
+ @AutoValue
+ public abstract static class Read extends PTransform<PBegin, PCollection<RecordWithMetadata>> {
+ abstract @Nullable ValueProvider<String> getFilepattern();
+
+ abstract MatchConfiguration getMatchConfiguration();
+
+ abstract boolean getHintMatchesManyFiles();
+
+ abstract boolean getWithoutRecordNumMetadata();
+
+ abstract Compression getCompression();
+
+ abstract @Nullable Boolean getHasMultilineCSVRecords();
+
+ @SuppressWarnings("mutable") // this returns an array that can be mutated by the caller
+ abstract byte @Nullable [] getDelimiter();
+
+ abstract Builder toBuilder();
+
+ @AutoValue.Builder
+ abstract static class Builder {
+ abstract Builder setFilepattern(ValueProvider<String> filepattern);
+
+ abstract Builder setMatchConfiguration(MatchConfiguration matchConfiguration);
+
+ abstract Builder setHintMatchesManyFiles(boolean hintManyFiles);
+
+ abstract Builder setWithoutRecordNumMetadata(boolean withoutLineNumMetadata);
+
+ abstract Builder setCompression(Compression compression);
+
+ abstract Builder setDelimiter(byte @Nullable [] delimiter);
+
+ abstract Builder setHasMultilineCSVRecords(Boolean hasMultilineCSVRecords);
+
+ abstract Read build();
+ }
+
+ /**
+ * Reads text from the file(s) with the given filename or filename pattern.
+ *
+ * <p>This can be a local path (if running locally), or a Google Cloud Storage filename or
+ * filename pattern of the form {@code "gs://<bucket>/<filepath>"} (if running locally or using
+ * remote execution service).
+ *
+ * <p>Standard <a href="http://docs.oracle.com/javase/tutorial/essential/io/find.html" >Java
+ * Filesystem glob patterns</a> ("*", "?", "[..]") are supported.
+ *
+ * <p>If it is known that the filepattern will match a very large number of files (at least tens
+ * of thousands), use {@link #withHintMatchesManyFiles} for better performance and scalability.
+ */
+ public Read from(String filepattern) {
+ checkArgument(filepattern != null, "filepattern can not be null");
+ return from(StaticValueProvider.of(filepattern));
+ }
+
+ /** Same as {@code from(filepattern)}, but accepting a {@link ValueProvider}. */
+ public Read from(ValueProvider<String> filepattern) {
+ checkArgument(filepattern != null, "filepattern can not be null");
+ return toBuilder().setFilepattern(filepattern).build();
+ }
+
+ /** Sets the {@link MatchConfiguration}. */
+ public Read withMatchConfiguration(MatchConfiguration matchConfiguration) {
+ return toBuilder().setMatchConfiguration(matchConfiguration).build();
+ }
+
+ /**
+ * When reading RFC4180 CSV files that have values that span multiple lines, set this to true.
+ * Note: this reduces the read performance (see: {@link ContextualTextIO}).
+ */
+ public Read withHasMultilineCSVRecords(Boolean hasMultilineCSVRecords) {
+ return toBuilder().setHasMultilineCSVRecords(hasMultilineCSVRecords).build();
+ }
+
+ /**
+ * Reads from input sources using the specified compression type.
+ *
+ * <p>If no compression type is specified, the default is {@link Compression#AUTO}.
+ */
+ public Read withCompression(Compression compression) {
+ return toBuilder().setCompression(compression).build();
+ }
+
+ /**
+ * Hints that the filepattern specified in {@link #from(String)} matches a very large number of
+ * files.
+ *
+ * <p>This hint may cause a runner to execute the transform differently, in a way that improves
+ * performance for this case, but it may worsen performance if the filepattern matches only a
+ * small number of files (e.g., in a runner that supports dynamic work rebalancing, it will
+ * happen less efficiently within individual files).
+ */
+ public Read withHintMatchesManyFiles() {
+ return toBuilder().setHintMatchesManyFiles(true).build();
+ }
+
+ /**
+ * Allows the user to opt out of getting recordNums associated with each record.
+ *
+ * <p>When set to true, it will introduce a shuffle step to assemble the recordNums for each
+ * record, which will increase the resources used by the pipeline.
+ *
+ * <p>Use this when metadata like fileNames are required and their position/order can be
+ * ignored.
+ */
+ public Read withoutRecordNumMetadata() {
+ return toBuilder().setWithoutRecordNumMetadata(true).build();
+ }
+
+ /** See {@link MatchConfiguration#withEmptyMatchTreatment}. */
+ public Read withEmptyMatchTreatment(EmptyMatchTreatment treatment) {
+ return withMatchConfiguration(getMatchConfiguration().withEmptyMatchTreatment(treatment));
+ }
+
+ /** Set the custom delimiter to be used in place of the default ones ('\r', '\n' or '\r\n'). */
+ public Read withDelimiter(byte[] delimiter) {
+ checkArgument(delimiter != null, "delimiter can not be null");
+ checkArgument(!isSelfOverlapping(delimiter), "delimiter must not self-overlap");
+ return toBuilder().setDelimiter(delimiter).build();
+ }
+
+ static boolean isSelfOverlapping(byte[] s) {
+ // s self-overlaps if v exists such as s = vu = wv with u and w non empty
+ for (int i = 1; i < s.length - 1; ++i) {
+ if (ByteBuffer.wrap(s, 0, i).equals(ByteBuffer.wrap(s, s.length - i, i))) {
+ return true;
+ }
+ }
+ return false;
+ }
+
+ @Override
+ public PCollection<RecordWithMetadata> expand(PBegin input) {
+ checkNotNull(
+ getFilepattern(), "need to set the filepattern of a ContextualTextIO.Read transform");
+ PCollection<RecordWithMetadata> records = null;
+ if (getMatchConfiguration().getWatchInterval() == null && !getHintMatchesManyFiles()) {
+ records = input.apply("Read", org.apache.beam.sdk.io.Read.from(getSource()));
+ } else {
+ // All other cases go through FileIO + ReadFiles
+ records =
+ input
+ .apply(
+ "Create filepattern", Create.ofProvider(getFilepattern(), StringUtf8Coder.of()))
+ .apply("Match All", FileIO.matchAll().withConfiguration(getMatchConfiguration()))
+ .apply(
+ "Read Matches",
+ FileIO.readMatches()
+ .withCompression(getCompression())
+ .withDirectoryTreatment(DirectoryTreatment.PROHIBIT))
+ .apply("Via ReadFiles", readFiles().withDelimiter(getDelimiter()));
+ }
+
+ // Check if the user decided to opt out of recordNums associated with records
+ if (getWithoutRecordNumMetadata()) {
+ return records;
+ }
+
+ /*
+ * At this point the line number in RecordWithMetadata contains the relative line offset from the beginning of the read range.
+ *
+ * To compute the absolute position from the beginning of the input we group the lines within the same ranges, and evaluate the size of each range.
+ */
+
+ PCollection<KV<KV<String, Long>, RecordWithMetadata>> recordsGroupedByFileAndRange =
+ records.apply("AddFileNameAndRange", ParDo.of(new AddFileNameAndRange()));
+
+ PCollectionView<Map<KV<String, Long>, Long>> rangeSizes =
+ recordsGroupedByFileAndRange
+ .apply("CountRecordsForEachFileRange", Count.perKey())
+ .apply("SizesAsView", View.asMap());
+
+ // Get Pipeline to create a dummy PCollection with one element to help compute the lines
+ // before each Range
+ PCollection<Integer> singletonPcoll =
+ input.getPipeline().apply("CreateSingletonPcoll", Create.of(Arrays.asList(1)));
+
+ /*
+ * For each (File, Offset) pair, calculate the number of lines occurring before the Range for each file
+ *
+ * After computing the number of lines before each range, we can find the line number in original file as numLiesBeforeOffset + lineNumInCurrentOffset
+ */
+
+ PCollectionView<Map<KV<String, Long>, Long>> numRecordsBeforeEachRange =
+ singletonPcoll
+ .apply(
+ "ComputeRecordsBeforeRange",
+ ParDo.of(new ComputeRecordsBeforeEachRange(rangeSizes))
+ .withSideInputs(rangeSizes))
+ .apply("NumRecordsBeforeEachRangeAsView", View.asMap());
+
+ return recordsGroupedByFileAndRange.apply(
+ "AssignLineNums",
+ ParDo.of(new AssignRecordNums(numRecordsBeforeEachRange))
+ .withSideInputs(numRecordsBeforeEachRange));
+ }
+
+ @VisibleForTesting
+ static class AddFileNameAndRange
+ extends DoFn<RecordWithMetadata, KV<KV<String, Long>, RecordWithMetadata>> {
+ @ProcessElement
+ public void processElement(
+ @Element RecordWithMetadata record,
+ OutputReceiver<KV<KV<String, Long>, RecordWithMetadata>> out) {
+ out.output(KV.of(KV.of(record.getFileName().toString(), record.getRangeOffset()), record));
+ }
+ }
+
+ /**
+ * Helper class for computing number of record in the File preceding the beginning of the Range
+ * in this file.
+ */
+ @VisibleForTesting
+ static class ComputeRecordsBeforeEachRange extends DoFn<Integer, KV<KV<String, Long>, Long>> {
+ private final PCollectionView<Map<KV<String, Long>, Long>> rangeSizes;
+
+ public ComputeRecordsBeforeEachRange(
+ PCollectionView<Map<KV<String, Long>, Long>> rangeSizes) {
+ this.rangeSizes = rangeSizes;
+ }
+
+ // Add custom comparator as KV<K, V> is not comparable by default
+ private static class FileRangeComparator<K extends Comparable<K>, V extends Comparable<V>>
+ implements Comparator<KV<K, V>> {
+ @Override
+ public int compare(KV<K, V> a, KV<K, V> b) {
+ if (a.getKey().compareTo(b.getKey()) == 0) {
+ return a.getValue().compareTo(b.getValue());
+ }
+ return a.getKey().compareTo(b.getKey());
+ }
+ }
+
+ @ProcessElement
+ public void processElement(ProcessContext p) {
+ // Get the Map Containing the size from side-input
+ Map<KV<String, Long>, Long> rangeSizesMap = p.sideInput(rangeSizes);
+
+ // The FileRange Pair must be sorted
+ SortedMap<KV<String, Long>, Long> sorted = new TreeMap<>(new FileRangeComparator<>());
+
+ // Initialize sorted map with values
+ for (Map.Entry<KV<String, Long>, Long> entry : rangeSizesMap.entrySet()) {
+ sorted.put(entry.getKey(), entry.getValue());
+ }
+
+ // HashMap that tracks number of records passed for each file
+ Map<String, Long> pastRecords = new HashMap<>();
+
+ // For each (File, Range) Pair, compute the number of records before it
+ for (Map.Entry entry : sorted.entrySet()) {
+ Long numRecords = (long) entry.getValue();
+ KV<String, Long> fileRange = (KV<String, Long>) entry.getKey();
+ String file = fileRange.getKey();
+ Long numRecordsBefore = 0L;
+ if (pastRecords.containsKey(file)) {
+ numRecordsBefore = pastRecords.get(file);
+ }
+ p.output(KV.of(fileRange, numRecordsBefore));
+ pastRecords.put(file, numRecordsBefore + numRecords);
+ }
+ }
+ }
+
+ static class AssignRecordNums
+ extends DoFn<KV<KV<String, Long>, RecordWithMetadata>, RecordWithMetadata> {
+ PCollectionView<Map<KV<String, Long>, Long>> numRecordsBeforeEachRange;
+
+ public AssignRecordNums(
+ PCollectionView<Map<KV<String, Long>, Long>> numRecordsBeforeEachRange) {
+ this.numRecordsBeforeEachRange = numRecordsBeforeEachRange;
+ }
+
+ @ProcessElement
+ public void processElement(ProcessContext p) {
+ Long range = p.element().getKey().getValue();
+ String file = p.element().getKey().getKey();
+ RecordWithMetadata record = p.element().getValue();
+ Long numRecordsLessThanThisRange =
+ p.sideInput(numRecordsBeforeEachRange).get(KV.of(file, range));
+ RecordWithMetadata newLine =
+ RecordWithMetadata.newBuilder()
Review comment:
Thanks !
Done.
##########
File path: sdks/java/io/contextual-text-io/src/main/java/org/apache/beam/sdk/io/contextualtextio/RecordWithMetadata.java
##########
@@ -0,0 +1,84 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one
+ * or more contributor license agreements. See the NOTICE file
+ * distributed with this work for additional information
+ * regarding copyright ownership. The ASF licenses this file
+ * to you under the Apache License, Version 2.0 (the
+ * "License"); you may not use this file except in compliance
+ * with the License. You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+package org.apache.beam.sdk.io.contextualtextio;
+
+import com.google.auto.value.AutoValue;
+import org.apache.beam.sdk.annotations.Experimental;
+import org.apache.beam.sdk.schemas.AutoValueSchema;
+import org.apache.beam.sdk.schemas.annotations.DefaultSchema;
+
+/**
+ * Helper Class based on {@link AutoValueSchema}, it provides Metadata associated with each Record
+ * when reading from file(s) using {@link ContextualTextIO}.
+ *
+ * <h3>Fields:</h3>
+ *
+ * <ul>
+ * <li>recordOffset: The offset of a record (the byte at which the record begins) in a file. This
+ * information can be useful if you wish to reconstruct the file. {@link
+ * RecordWithMetadata#getRecordOffset()}
+ * <li>recordNum: The ordinal number of the record in its file. {@link
+ * RecordWithMetadata#getRecordNum()}
+ * <li>recordValue: The value / contents of the record {@link RecordWithMetadata#getRecordValue()}
+ * <li>rangeOffset: The starting offset of the range (split), which contained the record, when the
+ * record was read. {@link RecordWithMetadata#getRangeOffset()}
+ * <li>recordNumInOffset: The record number relative to the Range. (line number within the range)
+ * {@link RecordWithMetadata#getRecordNumInOffset()}
+ * <li>fileName: Name of the file to which the record belongs (this is the full filename,
+ * eg:path/to/file.txt) {@link RecordWithMetadata#getFileName()}
+ * </ul>
+ */
+@Experimental(Experimental.Kind.SCHEMAS)
+@DefaultSchema(AutoValueSchema.class)
+@AutoValue
+public abstract class RecordWithMetadata {
+ public abstract Long getRecordOffset();
Review comment:
Done
##########
File path: sdks/java/io/contextual-text-io/build.gradle
##########
@@ -0,0 +1,41 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one
+ * or more contributor license agreements. See the NOTICE file
+ * distributed with this work for additional information
+ * regarding copyright ownership. The ASF licenses this file
+ * to you under the Apache License, Version 2.0 (the
+ * License); you may not use this file except in compliance
+ * with the License. You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an AS IS BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+plugins { id 'org.apache.beam.module' }
+applyJavaNature(
+ automaticModuleName: 'org.apache.beam.sdk.io.contextual-text-io',
+ enableChecker: false,
+ ignoreRawtypeErrors: true)
+
+description = "Apache Beam :: SDKs :: Java :: Contextual-Text-IO"
+ext.summary = "Context-aware Text IO."
+
+dependencies {
+
+ compile library.java.vendored_guava_26_0_jre
+ compile library.java.protobuf_java
+ compile project(path: ":sdks:java:core", configuration: "shadow")
+ testCompile project(path: ":sdks:java:core", configuration: "shadowTest")
+
+ testCompile library.java.guava_testlib
+ testCompile library.java.junit
+ testCompile library.java.hamcrest_core
+ testRuntimeOnly library.java.slf4j_jdk14
+ testCompile project(path: ":runners:direct-java", configuration: "shadow")
+
+}
Review comment:
Done
##########
File path: sdks/java/io/contextual-text-io/build.gradle
##########
@@ -0,0 +1,41 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one
+ * or more contributor license agreements. See the NOTICE file
+ * distributed with this work for additional information
+ * regarding copyright ownership. The ASF licenses this file
+ * to you under the Apache License, Version 2.0 (the
+ * License); you may not use this file except in compliance
+ * with the License. You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an AS IS BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+plugins { id 'org.apache.beam.module' }
+applyJavaNature(
+ automaticModuleName: 'org.apache.beam.sdk.io.contextual-text-io',
+ enableChecker: false,
+ ignoreRawtypeErrors: true)
+
+description = "Apache Beam :: SDKs :: Java :: Contextual-Text-IO"
+ext.summary = "Context-aware Text IO."
+
+dependencies {
+
Review comment:
Done
##########
File path: sdks/java/io/contextual-text-io/src/main/java/org/apache/beam/sdk/io/contextualtextio/RecordWithMetadata.java
##########
@@ -0,0 +1,85 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one
+ * or more contributor license agreements. See the NOTICE file
+ * distributed with this work for additional information
+ * regarding copyright ownership. The ASF licenses this file
+ * to you under the Apache License, Version 2.0 (the
+ * "License"); you may not use this file except in compliance
+ * with the License. You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+package org.apache.beam.sdk.io.contextualtextio;
+
+import com.google.auto.value.AutoValue;
+import org.apache.beam.sdk.annotations.Experimental;
+import org.apache.beam.sdk.io.fs.ResourceId;
+import org.apache.beam.sdk.schemas.AutoValueSchema;
+import org.apache.beam.sdk.schemas.annotations.DefaultSchema;
+
+/**
+ * Helper Class based on {@link AutoValueSchema}, it provides Metadata associated with each Record
+ * when reading from file(s) using {@link ContextualTextIO}.
+ *
+ * <h3>Fields:</h3>
+ *
+ * <ul>
+ * <li>recordOffset: The offset of a record (the byte at which the record begins) in a file. This
+ * information can be useful if you wish to reconstruct the file. {@link
+ * RecordWithMetadata#getRecordOffset()}
+ * <li>recordNum: The ordinal number of the record in its file. {@link
+ * RecordWithMetadata#getRecordNum()}
+ * <li>recordValue: The value / contents of the record {@link RecordWithMetadata#getValue()}
+ * <li>rangeOffset: The starting offset of the range (split), which contained the record, when the
+ * record was read. {@link RecordWithMetadata#getRangeOffset()}
+ * <li>recordNumInOffset: The record number relative to the Range. (line number within the range)
+ * {@link RecordWithMetadata#getRecordNumInOffset()}
+ * <li>fileName: Name of the file to which the record belongs (this is the full filename,
+ * eg:path/to/file.txt) {@link RecordWithMetadata#getFileName()}
+ * </ul>
+ */
+@Experimental(Experimental.Kind.SCHEMAS)
+@DefaultSchema(AutoValueSchema.class)
+@AutoValue
+public abstract class RecordWithMetadata {
+ public abstract long getRecordOffset();
Review comment:
Done
##########
File path: sdks/java/io/contextual-text-io/src/main/java/org/apache/beam/sdk/io/contextualtextio/ContextualTextIOSource.java
##########
@@ -0,0 +1,364 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one
+ * or more contributor license agreements. See the NOTICE file
+ * distributed with this work for additional information
+ * regarding copyright ownership. The ASF licenses this file
+ * to you under the Apache License, Version 2.0 (the
+ * "License"); you may not use this file except in compliance
+ * with the License. You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+package org.apache.beam.sdk.io.contextualtextio;
+
+import java.io.IOException;
+import java.nio.ByteBuffer;
+import java.nio.channels.ReadableByteChannel;
+import java.nio.channels.SeekableByteChannel;
+import java.util.NoSuchElementException;
+import org.apache.beam.sdk.coders.Coder;
+import org.apache.beam.sdk.io.FileBasedSource;
+import org.apache.beam.sdk.io.fs.EmptyMatchTreatment;
+import org.apache.beam.sdk.io.fs.MatchResult;
+import org.apache.beam.sdk.options.PipelineOptions;
+import org.apache.beam.sdk.options.ValueProvider;
+import org.apache.beam.sdk.schemas.NoSuchSchemaException;
+import org.apache.beam.sdk.schemas.SchemaCoder;
+import org.apache.beam.sdk.schemas.SchemaRegistry;
+import org.apache.beam.vendor.grpc.v1p26p0.com.google.protobuf.ByteString;
+import org.apache.beam.vendor.guava.v26_0_jre.com.google.common.annotations.VisibleForTesting;
+import org.apache.beam.vendor.guava.v26_0_jre.com.google.common.base.Preconditions;
+import org.checkerframework.checker.nullness.qual.Nullable;
+import org.slf4j.Logger;
+import org.slf4j.LoggerFactory;
+
+/**
+ * Implementation detail of {@link ContextualTextIO.Read}.
+ *
+ * <p>A {@link FileBasedSource} which can decode records delimited by newline characters.
+ *
+ * <p>This source splits the data into records using {@code UTF-8} {@code \n}, {@code \r}, or {@code
+ * \r\n} as the delimiter. This source is not strict and supports decoding the last record even if
+ * it is not delimited. Finally, no records are decoded if the stream is empty.
+ *
+ * <p>This source supports reading from any arbitrary byte position within the stream. If the
+ * starting position is not {@code 0}, then bytes are skipped until the first delimiter is found
+ * representing the beginning of the first record to be decoded.
+ */
+@VisibleForTesting
+class ContextualTextIOSource extends FileBasedSource<RecordWithMetadata> {
+ byte[] delimiter;
+
+ private static final Logger LOG = LoggerFactory.getLogger(ContextualTextIOSource.class);
+
+ // Used to Override isSplittable
+ private boolean hasMultilineCSVRecords;
+
+ @Override
+ protected boolean isSplittable() throws Exception {
+ if (hasMultilineCSVRecords) {
+ // When Having Multiline CSV Records,
+ // Splitting the file may cause a split to be within a record,
+ // Disabling split prevents this from happening
+ return false;
+ }
+ return super.isSplittable();
+ }
+
+ ContextualTextIOSource(
+ ValueProvider<String> fileSpec,
+ EmptyMatchTreatment emptyMatchTreatment,
+ byte[] delimiter,
+ boolean hasMultilineCSVRecords) {
+ super(fileSpec, emptyMatchTreatment, 1L);
+ this.delimiter = delimiter;
+ this.hasMultilineCSVRecords = hasMultilineCSVRecords;
+ }
+
+ private ContextualTextIOSource(
+ MatchResult.Metadata metadata,
+ long start,
+ long end,
+ byte[] delimiter,
+ boolean hasMultilineCSVRecords) {
+ super(metadata, 1L, start, end);
+ this.delimiter = delimiter;
+ this.hasMultilineCSVRecords = hasMultilineCSVRecords;
+ }
+
+ @Override
+ protected FileBasedSource<RecordWithMetadata> createForSubrangeOfFile(
+ MatchResult.Metadata metadata, long start, long end) {
+ return new ContextualTextIOSource(metadata, start, end, delimiter, hasMultilineCSVRecords);
+ }
+
+ @Override
+ protected FileBasedReader<RecordWithMetadata> createSingleFileReader(PipelineOptions options) {
+ return new MultiLineTextBasedReader(this, delimiter, hasMultilineCSVRecords);
+ }
+
+ @Override
+ public Coder<RecordWithMetadata> getOutputCoder() {
Review comment:
Done
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
[GitHub] [beam] abhiy13 commented on a change in pull request #12645: [BEAM-10124] Add ContextualTextIO
Posted by GitBox <gi...@apache.org>.
abhiy13 commented on a change in pull request #12645:
URL: https://github.com/apache/beam/pull/12645#discussion_r492474704
##########
File path: sdks/java/io/contextual-text-io/src/main/java/org/apache/beam/sdk/io/contextualtextio/ContextualTextIOSource.java
##########
@@ -0,0 +1,364 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one
+ * or more contributor license agreements. See the NOTICE file
+ * distributed with this work for additional information
+ * regarding copyright ownership. The ASF licenses this file
+ * to you under the Apache License, Version 2.0 (the
+ * "License"); you may not use this file except in compliance
+ * with the License. You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+package org.apache.beam.sdk.io.contextualtextio;
+
+import java.io.IOException;
+import java.nio.ByteBuffer;
+import java.nio.channels.ReadableByteChannel;
+import java.nio.channels.SeekableByteChannel;
+import java.util.NoSuchElementException;
+import org.apache.beam.sdk.coders.Coder;
+import org.apache.beam.sdk.io.FileBasedSource;
+import org.apache.beam.sdk.io.fs.EmptyMatchTreatment;
+import org.apache.beam.sdk.io.fs.MatchResult;
+import org.apache.beam.sdk.options.PipelineOptions;
+import org.apache.beam.sdk.options.ValueProvider;
+import org.apache.beam.sdk.schemas.NoSuchSchemaException;
+import org.apache.beam.sdk.schemas.SchemaCoder;
+import org.apache.beam.sdk.schemas.SchemaRegistry;
+import org.apache.beam.vendor.grpc.v1p26p0.com.google.protobuf.ByteString;
+import org.apache.beam.vendor.guava.v26_0_jre.com.google.common.annotations.VisibleForTesting;
+import org.apache.beam.vendor.guava.v26_0_jre.com.google.common.base.Preconditions;
+import org.checkerframework.checker.nullness.qual.Nullable;
+import org.slf4j.Logger;
+import org.slf4j.LoggerFactory;
+
+/**
+ * Implementation detail of {@link ContextualTextIO.Read}.
+ *
+ * <p>A {@link FileBasedSource} which can decode records delimited by newline characters.
+ *
+ * <p>This source splits the data into records using {@code UTF-8} {@code \n}, {@code \r}, or {@code
+ * \r\n} as the delimiter. This source is not strict and supports decoding the last record even if
+ * it is not delimited. Finally, no records are decoded if the stream is empty.
+ *
+ * <p>This source supports reading from any arbitrary byte position within the stream. If the
+ * starting position is not {@code 0}, then bytes are skipped until the first delimiter is found
+ * representing the beginning of the first record to be decoded.
+ */
+@VisibleForTesting
+class ContextualTextIOSource extends FileBasedSource<RecordWithMetadata> {
+ byte[] delimiter;
+
+ private static final Logger LOG = LoggerFactory.getLogger(ContextualTextIOSource.class);
+
+ // Used to Override isSplittable
+ private boolean hasMultilineCSVRecords;
+
+ @Override
+ protected boolean isSplittable() throws Exception {
+ if (hasMultilineCSVRecords) {
+ // When Having Multiline CSV Records,
+ // Splitting the file may cause a split to be within a record,
+ // Disabling split prevents this from happening
+ return false;
+ }
+ return super.isSplittable();
+ }
+
+ ContextualTextIOSource(
+ ValueProvider<String> fileSpec,
+ EmptyMatchTreatment emptyMatchTreatment,
+ byte[] delimiter,
+ boolean hasMultilineCSVRecords) {
+ super(fileSpec, emptyMatchTreatment, 1L);
+ this.delimiter = delimiter;
+ this.hasMultilineCSVRecords = hasMultilineCSVRecords;
+ }
+
+ private ContextualTextIOSource(
+ MatchResult.Metadata metadata,
+ long start,
+ long end,
+ byte[] delimiter,
+ boolean hasMultilineCSVRecords) {
+ super(metadata, 1L, start, end);
+ this.delimiter = delimiter;
+ this.hasMultilineCSVRecords = hasMultilineCSVRecords;
+ }
+
+ @Override
+ protected FileBasedSource<RecordWithMetadata> createForSubrangeOfFile(
+ MatchResult.Metadata metadata, long start, long end) {
+ return new ContextualTextIOSource(metadata, start, end, delimiter, hasMultilineCSVRecords);
+ }
+
+ @Override
+ protected FileBasedReader<RecordWithMetadata> createSingleFileReader(PipelineOptions options) {
+ return new MultiLineTextBasedReader(this, delimiter, hasMultilineCSVRecords);
+ }
+
+ @Override
+ public Coder<RecordWithMetadata> getOutputCoder() {
+ SchemaCoder<RecordWithMetadata> coder = null;
+ try {
+ coder = SchemaRegistry.createDefault().getSchemaCoder(RecordWithMetadata.class);
+ } catch (NoSuchSchemaException e) {
+ LOG.error("No Coder Found for RecordWithMetadata");
+ }
+ return coder;
+ }
+
+ /**
+ * A {@link FileBasedReader FileBasedReader} which can decode records delimited by delimiter
+ * characters.
+ *
+ * <p>See {@link ContextualTextIOSource } for further details.
+ */
+ @VisibleForTesting
+ static class MultiLineTextBasedReader extends FileBasedReader<RecordWithMetadata> {
+ public static final int READ_BUFFER_SIZE = 8192;
+ private static final ByteString UTF8_BOM =
+ ByteString.copyFrom(new byte[] {(byte) 0xEF, (byte) 0xBB, (byte) 0xBF});
+ private final ByteBuffer readBuffer = ByteBuffer.allocate(READ_BUFFER_SIZE);
+ private ByteString buffer;
+ private int startOfDelimiterInBuffer;
+ private int endOfDelimiterInBuffer;
+ private long startOfRecord;
+ private volatile long startOfNextRecord;
+ private volatile boolean eof;
+ private volatile boolean elementIsPresent;
+ private @Nullable RecordWithMetadata currentValue;
+ private @Nullable ReadableByteChannel inChannel;
+ private byte @Nullable [] delimiter;
+
+ // Add to override the isSplittable
+ private boolean hasMultilineCSVRecords;
+
+ private long startingOffset;
+ private long totalRecordCount;
+
+ private MultiLineTextBasedReader(
+ ContextualTextIOSource source, byte[] delimiter, boolean hasMultilineCSVRecords) {
+ super(source);
+ buffer = ByteString.EMPTY;
+ this.delimiter = delimiter;
+ this.hasMultilineCSVRecords = hasMultilineCSVRecords;
+ startingOffset = getCurrentSource().getStartOffset(); // Start offset;
+ }
+
+ @Override
+ protected long getCurrentOffset() throws NoSuchElementException {
+ if (!elementIsPresent) {
+ throw new NoSuchElementException();
+ }
+ return startOfRecord;
+ }
+
+ @Override
+ public long getSplitPointsRemaining() {
+ if (isStarted() && startOfNextRecord >= getCurrentSource().getEndOffset()) {
+ return isDone() ? 0 : 1;
+ }
+ return super.getSplitPointsRemaining();
+ }
+
+ @Override
+ public RecordWithMetadata getCurrent() throws NoSuchElementException {
+ if (!elementIsPresent) {
+ throw new NoSuchElementException();
+ }
+ return currentValue;
+ }
+
+ @Override
+ protected void startReading(ReadableByteChannel channel) throws IOException {
+ this.inChannel = channel;
+ // If the first offset is greater than zero, we need to skip bytes until we see our
+ // first delimiter.
+ long startOffset = getCurrentSource().getStartOffset();
+ if (startOffset > 0) {
+ Preconditions.checkState(
+ channel instanceof SeekableByteChannel,
+ "%s only supports reading from a SeekableByteChannel when given a start offset"
+ + " greater than 0.",
+ ContextualTextIOSource.class.getSimpleName());
+ long requiredPosition = startOffset - 1;
+ if (delimiter != null && startOffset >= delimiter.length) {
+ // we need to move back the offset of at worse delimiter.size to be sure to see
+ // all the bytes of the delimiter in the call to findDelimiterBounds() below
+ requiredPosition = startOffset - delimiter.length;
+ }
+ ((SeekableByteChannel) channel).position(requiredPosition);
+ findDelimiterBoundsWithMultiLineCheck();
+ buffer = buffer.substring(endOfDelimiterInBuffer);
+ startOfNextRecord = requiredPosition + endOfDelimiterInBuffer;
+ endOfDelimiterInBuffer = 0;
+ startOfDelimiterInBuffer = 0;
+ }
+ }
+
+ private void findDelimiterBoundsWithMultiLineCheck() throws IOException {
+ findDelimiterBounds();
+ }
+
+ /**
+ * Locates the start position and end position of the next delimiter. Will consume the channel
+ * till either EOF or the delimiter bounds are found.
+ *
+ * <p>If {@link ContextualTextIOSource#hasMultilineCSVRecords} is set then the behaviour will
+ * change from the standard read seen in {@link org.apache.beam.sdk.io.TextIO}. The assumption
+ * when {@link ContextualTextIOSource#hasMultilineCSVRecords} is set is that the file is being
+ * read with a single thread.
+ *
+ * <p>This fills the buffer and updates the positions as follows:
+ *
+ * <pre>{@code
+ * ------------------------------------------------------
+ * | element bytes | delimiter bytes | unconsumed bytes |
+ * ------------------------------------------------------
+ * 0 start of end of buffer
+ * delimiter delimiter size
+ * in buffer in buffer
+ * }</pre>
+ */
+ private void findDelimiterBounds() throws IOException {
+ int bytePositionInBuffer = 0;
+ boolean doubleQuoteClosed = true;
+
+ while (true) {
+ if (!tryToEnsureNumberOfBytesInBuffer(bytePositionInBuffer + 1)) {
+ startOfDelimiterInBuffer = endOfDelimiterInBuffer = bytePositionInBuffer;
+ break;
+ }
+
+ byte currentByte = buffer.byteAt(bytePositionInBuffer);
+ if (hasMultilineCSVRecords) {
+ // Check if we are inside an open Quote
+ if (currentByte == '"') {
+ doubleQuoteClosed = !doubleQuoteClosed;
+ }
+ } else {
+ doubleQuoteClosed = true;
+ }
+
+ if (delimiter == null) {
+ // default delimiter
+ if (currentByte == '\n') {
+ startOfDelimiterInBuffer = bytePositionInBuffer;
+ endOfDelimiterInBuffer = startOfDelimiterInBuffer + 1;
+ if (doubleQuoteClosed) {
+ break;
+ }
+ } else if (currentByte == '\r') {
+ startOfDelimiterInBuffer = bytePositionInBuffer;
+ endOfDelimiterInBuffer = startOfDelimiterInBuffer + 1;
+ if (tryToEnsureNumberOfBytesInBuffer(bytePositionInBuffer + 2)) {
+ currentByte = buffer.byteAt(bytePositionInBuffer + 1);
+ if (currentByte == '\n') {
+ endOfDelimiterInBuffer += 1;
+ }
+ }
+ if (doubleQuoteClosed) {
+ break;
+ }
+ }
+ } else {
+ // when the user defines a delimiter
+ int i = 0;
+ startOfDelimiterInBuffer = endOfDelimiterInBuffer = bytePositionInBuffer;
+ while ((i < delimiter.length) && (currentByte == delimiter[i])) {
+ // read next byte;
+ i++;
+ if (tryToEnsureNumberOfBytesInBuffer(bytePositionInBuffer + i + 1)) {
+ currentByte = buffer.byteAt(bytePositionInBuffer + i);
+ } else {
+ // corner case: delimiter truncate at the end of file
+ startOfDelimiterInBuffer = endOfDelimiterInBuffer = bytePositionInBuffer;
+ break;
+ }
+ }
+ if (i == delimiter.length) {
+ endOfDelimiterInBuffer = bytePositionInBuffer + i;
+ if (doubleQuoteClosed) {
+ break;
+ }
+ }
+ }
+ bytePositionInBuffer += 1;
+ }
+ }
+
+ @Override
+ protected boolean readNextRecord() throws IOException {
+ startOfRecord = startOfNextRecord;
+
+ findDelimiterBoundsWithMultiLineCheck();
+
+ // If we have reached EOF file and consumed all of the buffer then we know
+ // that there are no more records.
+ if (eof && buffer.isEmpty()) {
+ elementIsPresent = false;
+ return false;
+ }
+
+ decodeCurrentElement();
+ startOfNextRecord = startOfRecord + endOfDelimiterInBuffer;
+ return true;
+ }
+
+ /**
+ * Decodes the current element updating the buffer to only contain the unconsumed bytes.
+ *
+ * <p>This invalidates the currently stored {@code startOfDelimiterInBuffer} and {@code
+ * endOfDelimiterInBuffer}.
+ */
+ private void decodeCurrentElement() throws IOException {
+ ByteString dataToDecode = buffer.substring(0, startOfDelimiterInBuffer);
+ // If present, the UTF8 Byte Order Mark (BOM) will be removed.
+ if (startOfRecord == 0 && dataToDecode.startsWith(UTF8_BOM)) {
+ dataToDecode = dataToDecode.substring(UTF8_BOM.size());
+ }
+
+ // The line num is:
+ Long recordUniqueNum = totalRecordCount++;
+ // The Complete FileName (with uri if this is a web url eg: temp/abc.txt) is:
+ String fileName = getCurrentSource().getSingleFileMetadata().resourceId().toString();
Review comment:
@rezarokni thanks !
I was facing this same issue that is why could not make it part of the previous commit.
FileName does seem to be the best option for now.
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
[GitHub] [beam] abhiy13 commented on a change in pull request #12645: [BEAM-10124] Add ContextualTextIO
Posted by GitBox <gi...@apache.org>.
abhiy13 commented on a change in pull request #12645:
URL: https://github.com/apache/beam/pull/12645#discussion_r475174496
##########
File path: sdks/java/io/contextual-text-io/src/main/java/org/apache/beam/sdk/io/contextualtextio/RecordWithMetadata.java
##########
@@ -19,21 +19,45 @@
import com.google.auto.value.AutoValue;
import org.apache.beam.sdk.annotations.Experimental;
-import org.apache.beam.sdk.annotations.Internal;
import org.apache.beam.sdk.schemas.AutoValueSchema;
import org.apache.beam.sdk.schemas.annotations.DefaultSchema;
-@Internal
+/**
+ * Helper Class based on {@link AutoValueSchema}, it provides Metadata associated with each Record
+ * when reading from file(s) using {@link ContextualTextIO}.
+ *
+ * <h3>Fields:</h3>
+ *
+ * <ul>
+ * <li>recordOffset: The offset of a record (the byte at which the record begins) in a file. This
+ * information can be useful if you wish to reconstruct the file. {@link
+ * RecordWithMetadata#getRecordOffset()}
+ * <li>recordNum: The record number of the record from its ordinal file. {@link
+ * RecordWithMetadata#getRecordNum()}
+ * <li>recordValue: The value / contents of the records {@link
Review comment:
Done.
##########
File path: sdks/java/io/contextual-text-io/src/main/java/org/apache/beam/sdk/io/contextualtextio/RecordWithMetadata.java
##########
@@ -19,21 +19,45 @@
import com.google.auto.value.AutoValue;
import org.apache.beam.sdk.annotations.Experimental;
-import org.apache.beam.sdk.annotations.Internal;
import org.apache.beam.sdk.schemas.AutoValueSchema;
import org.apache.beam.sdk.schemas.annotations.DefaultSchema;
-@Internal
+/**
+ * Helper Class based on {@link AutoValueSchema}, it provides Metadata associated with each Record
+ * when reading from file(s) using {@link ContextualTextIO}.
+ *
+ * <h3>Fields:</h3>
+ *
+ * <ul>
+ * <li>recordOffset: The offset of a record (the byte at which the record begins) in a file. This
+ * information can be useful if you wish to reconstruct the file. {@link
+ * RecordWithMetadata#getRecordOffset()}
+ * <li>recordNum: The record number of the record from its ordinal file. {@link
Review comment:
Done.
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
[GitHub] [beam] abhiy13 commented on a change in pull request #12645: [BEAM-10124] Add ContextualTextIO
Posted by GitBox <gi...@apache.org>.
abhiy13 commented on a change in pull request #12645:
URL: https://github.com/apache/beam/pull/12645#discussion_r474442986
##########
File path: sdks/java/io/contextual-text-io/src/main/java/org/apache/beam/sdk/io/contextualtextio/ContextualTextIO.java
##########
@@ -0,0 +1,594 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one
+ * or more contributor license agreements. See the NOTICE file
+ * distributed with this work for additional information
+ * regarding copyright ownership. The ASF licenses this file
+ * to you under the Apache License, Version 2.0 (the
+ * "License"); you may not use this file except in compliance
+ * with the License. You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+package org.apache.beam.sdk.io.contextualtextio;
+
+import static org.apache.beam.sdk.io.FileIO.ReadMatches.DirectoryTreatment;
+import static org.apache.beam.vendor.guava.v26_0_jre.com.google.common.base.Preconditions.checkArgument;
+import static org.apache.beam.vendor.guava.v26_0_jre.com.google.common.base.Preconditions.checkNotNull;
+
+import com.google.auto.value.AutoValue;
+import java.nio.ByteBuffer;
+import java.util.Arrays;
+import java.util.Comparator;
+import java.util.HashMap;
+import java.util.Map;
+import java.util.SortedMap;
+import java.util.TreeMap;
+import org.apache.beam.sdk.coders.StringUtf8Coder;
+import org.apache.beam.sdk.io.CompressedSource;
+import org.apache.beam.sdk.io.Compression;
+import org.apache.beam.sdk.io.FileBasedSource;
+import org.apache.beam.sdk.io.FileIO;
+import org.apache.beam.sdk.io.FileIO.MatchConfiguration;
+import org.apache.beam.sdk.io.ReadAllViaFileBasedSource;
+import org.apache.beam.sdk.io.fs.EmptyMatchTreatment;
+import org.apache.beam.sdk.options.ValueProvider;
+import org.apache.beam.sdk.options.ValueProvider.StaticValueProvider;
+import org.apache.beam.sdk.schemas.NoSuchSchemaException;
+import org.apache.beam.sdk.schemas.SchemaCoder;
+import org.apache.beam.sdk.transforms.Count;
+import org.apache.beam.sdk.transforms.Create;
+import org.apache.beam.sdk.transforms.DoFn;
+import org.apache.beam.sdk.transforms.PTransform;
+import org.apache.beam.sdk.transforms.ParDo;
+import org.apache.beam.sdk.transforms.SerializableFunction;
+import org.apache.beam.sdk.transforms.View;
+import org.apache.beam.sdk.transforms.Watch.Growth.TerminationCondition;
+import org.apache.beam.sdk.transforms.display.DisplayData;
+import org.apache.beam.sdk.values.KV;
+import org.apache.beam.sdk.values.PBegin;
+import org.apache.beam.sdk.values.PCollection;
+import org.apache.beam.sdk.values.PCollectionView;
+import org.apache.beam.vendor.guava.v26_0_jre.com.google.common.annotations.VisibleForTesting;
+import org.checkerframework.checker.nullness.qual.Nullable;
+import org.joda.time.Duration;
+
+/**
+ * {@link PTransform}s that read text files and collect contextual information of the elements in
+ * the input.
+ *
+ * <h2>Reading from text files</h2>
+ *
+ * <p>To read a {@link PCollection} from one or more text files, use {@code
+ * ContextualTextIO.read()}. To instantiate a transform use {@link
+ * ContextualTextIO.Read#from(String)} and specify the path of the file(s) to be read.
+ * Alternatively, if the filenames to be read are themselves in a {@link PCollection} you can use
+ * {@link FileIO} to match them and {@link ContextualTextIO#readFiles()} to read them.
+ *
+ * <p>{@link #read} returns a {@link PCollection} of {@link RecordWithMetadata RecordWithMetadata},
+ * each corresponding to one line of an inout UTF-8 text file (split into lines delimited by '\n',
+ * '\r', '\r\n', or specified delimiter see {@link ContextualTextIO.Read#withDelimiter})
+ *
+ * <h3>Filepattern expansion and watching</h3>
+ *
+ * <p>By default, the filepatterns are expanded only once. The combination of {@link
+ * FileIO.Match#continuously(Duration, TerminationCondition)} and {@link #readFiles()} allow
+ * streaming of new files matching the filepattern(s).
+ *
+ * <p>By default, {@link #read} prohibits filepatterns that match no files, and {@link #readFiles()}
+ * allows them in case the filepattern contains a glob wildcard character. Use {@link
+ * ContextualTextIO.Read#withEmptyMatchTreatment} or {@link
+ * FileIO.Match#withEmptyMatchTreatment(EmptyMatchTreatment)} plus {@link #readFiles()} to configure
+ * this behavior.
+ *
+ * <p>Example 1: reading a file or filepattern.
+ *
+ * <pre>{@code
+ * Pipeline p = ...;
+ *
+ * // A simple Read of a file:
+ * PCollection<RecordWithMetadata> lines = p.apply(ContextualTextIO.read().from("/local/path/to/file.txt"));
+ * }</pre>
+ *
+ * <p>Example 2: reading a PCollection of filenames.
+ *
+ * <pre>{@code
+ * Pipeline p = ...;
+ *
+ * // E.g. the filenames might be computed from other data in the pipeline, or
+ * // read from a data source.
+ * PCollection<String> filenames = ...;
+ *
+ * // Read all files in the collection.
+ * PCollection<RecordWithMetadata> lines =
+ * filenames
+ * .apply(FileIO.matchAll())
+ * .apply(FileIO.readMatches())
+ * .apply(ContextualTextIO.readFiles());
+ * }</pre>
+ *
+ * <p>Example 3: streaming new files matching a filepattern.
+ *
+ * <pre>{@code
+ * Pipeline p = ...;
+ *
+ * PCollection<RecordWithMetadata> lines = p.apply(ContextualTextIO.read()
+ * .from("/local/path/to/files/*")
+ * .watchForNewFiles(
+ * // Check for new files every minute
+ * Duration.standardMinutes(1),
+ * // Stop watching the filepattern if no new files appear within an hour
+ * afterTimeSinceNewOutput(Duration.standardHours(1))));
+ * }</pre>
+ *
+ * <p>Example 4: reading a file or file pattern of RFC4180-compliant CSV files with fields that may
+ * contain line breaks.
+ *
+ * <pre>{@code
+ * Pipeline p = ...;
+ *
+ * PCollection<RecordWithMetadata> lines = p.apply(ContextualTextIO.read()
+ * .from("/local/path/to/files/*.csv")
+ * .withHasMultilineCSVRecords(true));
+ * }</pre>
+ *
+ * <p>Example 5: reading while watching for new files
+ *
+ * <pre>{@code
+ * Pipeline p = ...;
+ *
+ * PCollection<RecordWithMetadata> lines = p.apply(FileIO.match()
+ * .filepattern("filepattern")
+ * .continuously(
+ * Duration.millis(100),
+ * Watch.Growth.afterTimeSinceNewOutput(Duration.standardSeconds(3))))
+ * .apply(FileIO.readMatches())
+ * .apply(ContextualTextIO.readFiles());
+ * }</pre>
+ *
+ * NOTE: Using {@link ContextualTextIO.Read#withHasMultilineCSVRecords(Boolean)} introduces a
Review comment:
Ack.
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
[GitHub] [beam] codecov[bot] edited a comment on pull request #12645: [BEAM-10124] Add ContextualTextIO
Posted by GitBox <gi...@apache.org>.
codecov[bot] edited a comment on pull request #12645:
URL: https://github.com/apache/beam/pull/12645#issuecomment-688630083
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
[GitHub] [beam] rezarokni commented on a change in pull request #12645: [BEAM-10124] Add ContextualTextIO
Posted by GitBox <gi...@apache.org>.
rezarokni commented on a change in pull request #12645:
URL: https://github.com/apache/beam/pull/12645#discussion_r492513409
##########
File path: sdks/java/io/contextual-text-io/src/main/java/org/apache/beam/sdk/io/contextualtextio/ContextualTextIOSource.java
##########
@@ -0,0 +1,364 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one
+ * or more contributor license agreements. See the NOTICE file
+ * distributed with this work for additional information
+ * regarding copyright ownership. The ASF licenses this file
+ * to you under the Apache License, Version 2.0 (the
+ * "License"); you may not use this file except in compliance
+ * with the License. You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+package org.apache.beam.sdk.io.contextualtextio;
+
+import java.io.IOException;
+import java.nio.ByteBuffer;
+import java.nio.channels.ReadableByteChannel;
+import java.nio.channels.SeekableByteChannel;
+import java.util.NoSuchElementException;
+import org.apache.beam.sdk.coders.Coder;
+import org.apache.beam.sdk.io.FileBasedSource;
+import org.apache.beam.sdk.io.fs.EmptyMatchTreatment;
+import org.apache.beam.sdk.io.fs.MatchResult;
+import org.apache.beam.sdk.options.PipelineOptions;
+import org.apache.beam.sdk.options.ValueProvider;
+import org.apache.beam.sdk.schemas.NoSuchSchemaException;
+import org.apache.beam.sdk.schemas.SchemaCoder;
+import org.apache.beam.sdk.schemas.SchemaRegistry;
+import org.apache.beam.vendor.grpc.v1p26p0.com.google.protobuf.ByteString;
+import org.apache.beam.vendor.guava.v26_0_jre.com.google.common.annotations.VisibleForTesting;
+import org.apache.beam.vendor.guava.v26_0_jre.com.google.common.base.Preconditions;
+import org.checkerframework.checker.nullness.qual.Nullable;
+import org.slf4j.Logger;
+import org.slf4j.LoggerFactory;
+
+/**
+ * Implementation detail of {@link ContextualTextIO.Read}.
+ *
+ * <p>A {@link FileBasedSource} which can decode records delimited by newline characters.
+ *
+ * <p>This source splits the data into records using {@code UTF-8} {@code \n}, {@code \r}, or {@code
+ * \r\n} as the delimiter. This source is not strict and supports decoding the last record even if
+ * it is not delimited. Finally, no records are decoded if the stream is empty.
+ *
+ * <p>This source supports reading from any arbitrary byte position within the stream. If the
+ * starting position is not {@code 0}, then bytes are skipped until the first delimiter is found
+ * representing the beginning of the first record to be decoded.
+ */
+@VisibleForTesting
+class ContextualTextIOSource extends FileBasedSource<RecordWithMetadata> {
+ byte[] delimiter;
+
+ private static final Logger LOG = LoggerFactory.getLogger(ContextualTextIOSource.class);
+
+ // Used to Override isSplittable
+ private boolean hasMultilineCSVRecords;
+
+ @Override
+ protected boolean isSplittable() throws Exception {
+ if (hasMultilineCSVRecords) {
+ // When Having Multiline CSV Records,
+ // Splitting the file may cause a split to be within a record,
+ // Disabling split prevents this from happening
+ return false;
+ }
+ return super.isSplittable();
+ }
+
+ ContextualTextIOSource(
+ ValueProvider<String> fileSpec,
+ EmptyMatchTreatment emptyMatchTreatment,
+ byte[] delimiter,
+ boolean hasMultilineCSVRecords) {
+ super(fileSpec, emptyMatchTreatment, 1L);
+ this.delimiter = delimiter;
+ this.hasMultilineCSVRecords = hasMultilineCSVRecords;
+ }
+
+ private ContextualTextIOSource(
+ MatchResult.Metadata metadata,
+ long start,
+ long end,
+ byte[] delimiter,
+ boolean hasMultilineCSVRecords) {
+ super(metadata, 1L, start, end);
+ this.delimiter = delimiter;
+ this.hasMultilineCSVRecords = hasMultilineCSVRecords;
+ }
+
+ @Override
+ protected FileBasedSource<RecordWithMetadata> createForSubrangeOfFile(
+ MatchResult.Metadata metadata, long start, long end) {
+ return new ContextualTextIOSource(metadata, start, end, delimiter, hasMultilineCSVRecords);
+ }
+
+ @Override
+ protected FileBasedReader<RecordWithMetadata> createSingleFileReader(PipelineOptions options) {
+ return new MultiLineTextBasedReader(this, delimiter, hasMultilineCSVRecords);
+ }
+
+ @Override
+ public Coder<RecordWithMetadata> getOutputCoder() {
+ SchemaCoder<RecordWithMetadata> coder = null;
+ try {
+ coder = SchemaRegistry.createDefault().getSchemaCoder(RecordWithMetadata.class);
+ } catch (NoSuchSchemaException e) {
+ LOG.error("No Coder Found for RecordWithMetadata");
+ }
+ return coder;
+ }
+
+ /**
+ * A {@link FileBasedReader FileBasedReader} which can decode records delimited by delimiter
+ * characters.
+ *
+ * <p>See {@link ContextualTextIOSource } for further details.
+ */
+ @VisibleForTesting
+ static class MultiLineTextBasedReader extends FileBasedReader<RecordWithMetadata> {
+ public static final int READ_BUFFER_SIZE = 8192;
+ private static final ByteString UTF8_BOM =
+ ByteString.copyFrom(new byte[] {(byte) 0xEF, (byte) 0xBB, (byte) 0xBF});
+ private final ByteBuffer readBuffer = ByteBuffer.allocate(READ_BUFFER_SIZE);
+ private ByteString buffer;
+ private int startOfDelimiterInBuffer;
+ private int endOfDelimiterInBuffer;
+ private long startOfRecord;
+ private volatile long startOfNextRecord;
+ private volatile boolean eof;
+ private volatile boolean elementIsPresent;
+ private @Nullable RecordWithMetadata currentValue;
+ private @Nullable ReadableByteChannel inChannel;
+ private byte @Nullable [] delimiter;
+
+ // Add to override the isSplittable
+ private boolean hasMultilineCSVRecords;
+
+ private long startingOffset;
+ private long totalRecordCount;
+
+ private MultiLineTextBasedReader(
+ ContextualTextIOSource source, byte[] delimiter, boolean hasMultilineCSVRecords) {
+ super(source);
+ buffer = ByteString.EMPTY;
+ this.delimiter = delimiter;
+ this.hasMultilineCSVRecords = hasMultilineCSVRecords;
+ startingOffset = getCurrentSource().getStartOffset(); // Start offset;
+ }
+
+ @Override
+ protected long getCurrentOffset() throws NoSuchElementException {
+ if (!elementIsPresent) {
+ throw new NoSuchElementException();
+ }
+ return startOfRecord;
+ }
+
+ @Override
+ public long getSplitPointsRemaining() {
+ if (isStarted() && startOfNextRecord >= getCurrentSource().getEndOffset()) {
+ return isDone() ? 0 : 1;
+ }
+ return super.getSplitPointsRemaining();
+ }
+
+ @Override
+ public RecordWithMetadata getCurrent() throws NoSuchElementException {
+ if (!elementIsPresent) {
+ throw new NoSuchElementException();
+ }
+ return currentValue;
+ }
+
+ @Override
+ protected void startReading(ReadableByteChannel channel) throws IOException {
+ this.inChannel = channel;
+ // If the first offset is greater than zero, we need to skip bytes until we see our
+ // first delimiter.
+ long startOffset = getCurrentSource().getStartOffset();
+ if (startOffset > 0) {
+ Preconditions.checkState(
+ channel instanceof SeekableByteChannel,
+ "%s only supports reading from a SeekableByteChannel when given a start offset"
+ + " greater than 0.",
+ ContextualTextIOSource.class.getSimpleName());
+ long requiredPosition = startOffset - 1;
+ if (delimiter != null && startOffset >= delimiter.length) {
+ // we need to move back the offset of at worse delimiter.size to be sure to see
+ // all the bytes of the delimiter in the call to findDelimiterBounds() below
+ requiredPosition = startOffset - delimiter.length;
+ }
+ ((SeekableByteChannel) channel).position(requiredPosition);
+ findDelimiterBoundsWithMultiLineCheck();
+ buffer = buffer.substring(endOfDelimiterInBuffer);
+ startOfNextRecord = requiredPosition + endOfDelimiterInBuffer;
+ endOfDelimiterInBuffer = 0;
+ startOfDelimiterInBuffer = 0;
+ }
+ }
+
+ private void findDelimiterBoundsWithMultiLineCheck() throws IOException {
+ findDelimiterBounds();
+ }
+
+ /**
+ * Locates the start position and end position of the next delimiter. Will consume the channel
+ * till either EOF or the delimiter bounds are found.
+ *
+ * <p>If {@link ContextualTextIOSource#hasMultilineCSVRecords} is set then the behaviour will
+ * change from the standard read seen in {@link org.apache.beam.sdk.io.TextIO}. The assumption
+ * when {@link ContextualTextIOSource#hasMultilineCSVRecords} is set is that the file is being
+ * read with a single thread.
+ *
+ * <p>This fills the buffer and updates the positions as follows:
+ *
+ * <pre>{@code
+ * ------------------------------------------------------
+ * | element bytes | delimiter bytes | unconsumed bytes |
+ * ------------------------------------------------------
+ * 0 start of end of buffer
+ * delimiter delimiter size
+ * in buffer in buffer
+ * }</pre>
+ */
+ private void findDelimiterBounds() throws IOException {
+ int bytePositionInBuffer = 0;
+ boolean doubleQuoteClosed = true;
+
+ while (true) {
+ if (!tryToEnsureNumberOfBytesInBuffer(bytePositionInBuffer + 1)) {
+ startOfDelimiterInBuffer = endOfDelimiterInBuffer = bytePositionInBuffer;
+ break;
+ }
+
+ byte currentByte = buffer.byteAt(bytePositionInBuffer);
+ if (hasMultilineCSVRecords) {
+ // Check if we are inside an open Quote
+ if (currentByte == '"') {
+ doubleQuoteClosed = !doubleQuoteClosed;
+ }
+ } else {
+ doubleQuoteClosed = true;
+ }
+
+ if (delimiter == null) {
+ // default delimiter
+ if (currentByte == '\n') {
+ startOfDelimiterInBuffer = bytePositionInBuffer;
+ endOfDelimiterInBuffer = startOfDelimiterInBuffer + 1;
+ if (doubleQuoteClosed) {
+ break;
+ }
+ } else if (currentByte == '\r') {
+ startOfDelimiterInBuffer = bytePositionInBuffer;
+ endOfDelimiterInBuffer = startOfDelimiterInBuffer + 1;
+ if (tryToEnsureNumberOfBytesInBuffer(bytePositionInBuffer + 2)) {
+ currentByte = buffer.byteAt(bytePositionInBuffer + 1);
+ if (currentByte == '\n') {
+ endOfDelimiterInBuffer += 1;
+ }
+ }
+ if (doubleQuoteClosed) {
+ break;
+ }
+ }
+ } else {
+ // when the user defines a delimiter
+ int i = 0;
+ startOfDelimiterInBuffer = endOfDelimiterInBuffer = bytePositionInBuffer;
+ while ((i < delimiter.length) && (currentByte == delimiter[i])) {
+ // read next byte;
+ i++;
+ if (tryToEnsureNumberOfBytesInBuffer(bytePositionInBuffer + i + 1)) {
+ currentByte = buffer.byteAt(bytePositionInBuffer + i);
+ } else {
+ // corner case: delimiter truncate at the end of file
+ startOfDelimiterInBuffer = endOfDelimiterInBuffer = bytePositionInBuffer;
+ break;
+ }
+ }
+ if (i == delimiter.length) {
+ endOfDelimiterInBuffer = bytePositionInBuffer + i;
+ if (doubleQuoteClosed) {
+ break;
+ }
+ }
+ }
+ bytePositionInBuffer += 1;
+ }
+ }
+
+ @Override
+ protected boolean readNextRecord() throws IOException {
+ startOfRecord = startOfNextRecord;
+
+ findDelimiterBoundsWithMultiLineCheck();
+
+ // If we have reached EOF file and consumed all of the buffer then we know
+ // that there are no more records.
+ if (eof && buffer.isEmpty()) {
+ elementIsPresent = false;
+ return false;
+ }
+
+ decodeCurrentElement();
+ startOfNextRecord = startOfRecord + endOfDelimiterInBuffer;
+ return true;
+ }
+
+ /**
+ * Decodes the current element updating the buffer to only contain the unconsumed bytes.
+ *
+ * <p>This invalidates the currently stored {@code startOfDelimiterInBuffer} and {@code
+ * endOfDelimiterInBuffer}.
+ */
+ private void decodeCurrentElement() throws IOException {
+ ByteString dataToDecode = buffer.substring(0, startOfDelimiterInBuffer);
+ // If present, the UTF8 Byte Order Mark (BOM) will be removed.
+ if (startOfRecord == 0 && dataToDecode.startsWith(UTF8_BOM)) {
+ dataToDecode = dataToDecode.substring(UTF8_BOM.size());
+ }
+
+ // The line num is:
+ Long recordUniqueNum = totalRecordCount++;
+ // The Complete FileName (with uri if this is a web url eg: temp/abc.txt) is:
+ String fileName = getCurrentSource().getSingleFileMetadata().resourceId().toString();
Review comment:
@abhiy13 I raised a PR in your repo with some fixes that includes changing from <RecordWithMetadata> to <Row>, do you have time to merge that PR into yours? The Java doc still needs to be updated.
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
[GitHub] [beam] rezarokni commented on a change in pull request #12645: [BEAM-10124] Add ContextualTextIO
Posted by GitBox <gi...@apache.org>.
rezarokni commented on a change in pull request #12645:
URL: https://github.com/apache/beam/pull/12645#discussion_r487927710
##########
File path: sdks/java/io/contextual-text-io/src/main/java/org/apache/beam/sdk/io/contextualtextio/RecordWithMetadata.java
##########
@@ -0,0 +1,84 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one
+ * or more contributor license agreements. See the NOTICE file
+ * distributed with this work for additional information
+ * regarding copyright ownership. The ASF licenses this file
+ * to you under the Apache License, Version 2.0 (the
+ * "License"); you may not use this file except in compliance
+ * with the License. You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+package org.apache.beam.sdk.io.contextualtextio;
+
+import com.google.auto.value.AutoValue;
+import org.apache.beam.sdk.annotations.Experimental;
+import org.apache.beam.sdk.schemas.AutoValueSchema;
+import org.apache.beam.sdk.schemas.annotations.DefaultSchema;
+
+/**
+ * Helper Class based on {@link AutoValueSchema}, it provides Metadata associated with each Record
+ * when reading from file(s) using {@link ContextualTextIO}.
+ *
+ * <h3>Fields:</h3>
+ *
+ * <ul>
+ * <li>recordOffset: The offset of a record (the byte at which the record begins) in a file. This
+ * information can be useful if you wish to reconstruct the file. {@link
+ * RecordWithMetadata#getRecordOffset()}
+ * <li>recordNum: The ordinal number of the record in its file. {@link
+ * RecordWithMetadata#getRecordNum()}
+ * <li>recordValue: The value / contents of the record {@link RecordWithMetadata#getRecordValue()}
+ * <li>rangeOffset: The starting offset of the range (split), which contained the record, when the
+ * record was read. {@link RecordWithMetadata#getRangeOffset()}
+ * <li>recordNumInOffset: The record number relative to the Range. (line number within the range)
+ * {@link RecordWithMetadata#getRecordNumInOffset()}
+ * <li>fileName: Name of the file to which the record belongs (this is the full filename,
+ * eg:path/to/file.txt) {@link RecordWithMetadata#getFileName()}
+ * </ul>
+ */
+@Experimental(Experimental.Kind.SCHEMAS)
+@DefaultSchema(AutoValueSchema.class)
+@AutoValue
+public abstract class RecordWithMetadata {
Review comment:
@lukecwik This class is a Schema class, looking into this we cant make it generic, until schemas supports generics. But for ContextualTextIO we would be setting it to ContextualTextIO`<String>` anyway ... So while it will be useful in the future, I am not sure it adds much for now? WDYT?
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
[GitHub] [beam] abhiy13 commented on a change in pull request #12645: [BEAM-10124] Add ContextualTextIO
Posted by GitBox <gi...@apache.org>.
abhiy13 commented on a change in pull request #12645:
URL: https://github.com/apache/beam/pull/12645#discussion_r484661943
##########
File path: sdks/java/io/contextual-text-io/src/main/java/org/apache/beam/sdk/io/contextualtextio/ContextualTextIOSource.java
##########
@@ -0,0 +1,364 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one
+ * or more contributor license agreements. See the NOTICE file
+ * distributed with this work for additional information
+ * regarding copyright ownership. The ASF licenses this file
+ * to you under the Apache License, Version 2.0 (the
+ * "License"); you may not use this file except in compliance
+ * with the License. You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+package org.apache.beam.sdk.io.contextualtextio;
+
+import java.io.IOException;
+import java.nio.ByteBuffer;
+import java.nio.channels.ReadableByteChannel;
+import java.nio.channels.SeekableByteChannel;
+import java.util.NoSuchElementException;
+import org.apache.beam.sdk.coders.Coder;
+import org.apache.beam.sdk.io.FileBasedSource;
+import org.apache.beam.sdk.io.fs.EmptyMatchTreatment;
+import org.apache.beam.sdk.io.fs.MatchResult;
+import org.apache.beam.sdk.options.PipelineOptions;
+import org.apache.beam.sdk.options.ValueProvider;
+import org.apache.beam.sdk.schemas.NoSuchSchemaException;
+import org.apache.beam.sdk.schemas.SchemaCoder;
+import org.apache.beam.sdk.schemas.SchemaRegistry;
+import org.apache.beam.vendor.grpc.v1p26p0.com.google.protobuf.ByteString;
+import org.apache.beam.vendor.guava.v26_0_jre.com.google.common.annotations.VisibleForTesting;
+import org.apache.beam.vendor.guava.v26_0_jre.com.google.common.base.Preconditions;
+import org.checkerframework.checker.nullness.qual.Nullable;
+import org.slf4j.Logger;
+import org.slf4j.LoggerFactory;
+
+/**
+ * Implementation detail of {@link ContextualTextIO.Read}.
+ *
+ * <p>A {@link FileBasedSource} which can decode records delimited by newline characters.
+ *
+ * <p>This source splits the data into records using {@code UTF-8} {@code \n}, {@code \r}, or {@code
+ * \r\n} as the delimiter. This source is not strict and supports decoding the last record even if
+ * it is not delimited. Finally, no records are decoded if the stream is empty.
+ *
+ * <p>This source supports reading from any arbitrary byte position within the stream. If the
+ * starting position is not {@code 0}, then bytes are skipped until the first delimiter is found
+ * representing the beginning of the first record to be decoded.
+ */
+@VisibleForTesting
+class ContextualTextIOSource extends FileBasedSource<RecordWithMetadata> {
Review comment:
Hey @tvalentyn sorry for the delay, got caught up with some work from school.
Made the requested changes, PTAL.
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
[GitHub] [beam] codecov[bot] edited a comment on pull request #12645: [BEAM-10124] Add ContextualTextIO
Posted by GitBox <gi...@apache.org>.
codecov[bot] edited a comment on pull request #12645:
URL: https://github.com/apache/beam/pull/12645#issuecomment-688630083
# [Codecov](https://codecov.io/gh/apache/beam/pull/12645?src=pr&el=h1) Report
> :exclamation: No coverage uploaded for pull request base (`master@2b2b8e7`). [Click here to learn what that means](https://docs.codecov.io/docs/error-reference#section-missing-base-commit).
> The diff coverage is `n/a`.
[](https://codecov.io/gh/apache/beam/pull/12645?src=pr&el=tree)
```diff
@@ Coverage Diff @@
## master #12645 +/- ##
=========================================
Coverage ? 82.33%
=========================================
Files ? 453
Lines ? 54054
Branches ? 0
=========================================
Hits ? 44506
Misses ? 9548
Partials ? 0
```
| [Impacted Files](https://codecov.io/gh/apache/beam/pull/12645?src=pr&el=tree) | Coverage Δ | |
|---|---|---|
| [...s/python/apache\_beam/testing/pipeline\_verifiers.py](https://codecov.io/gh/apache/beam/pull/12645/diff?src=pr&el=tree#diff-c2Rrcy9weXRob24vYXBhY2hlX2JlYW0vdGVzdGluZy9waXBlbGluZV92ZXJpZmllcnMucHk=) | `91.80% <0.00%> (ø)` | |
| [sdks/python/apache\_beam/io/gcp/tests/\_\_init\_\_.py](https://codecov.io/gh/apache/beam/pull/12645/diff?src=pr&el=tree#diff-c2Rrcy9weXRob24vYXBhY2hlX2JlYW0vaW8vZ2NwL3Rlc3RzL19faW5pdF9fLnB5) | `100.00% <0.00%> (ø)` | |
| [...dks/python/apache\_beam/io/gcp/gce\_metadata\_util.py](https://codecov.io/gh/apache/beam/pull/12645/diff?src=pr&el=tree#diff-c2Rrcy9weXRob24vYXBhY2hlX2JlYW0vaW8vZ2NwL2djZV9tZXRhZGF0YV91dGlsLnB5) | `83.33% <0.00%> (ø)` | |
| [...hon/apache\_beam/examples/wordcount\_with\_metrics.py](https://codecov.io/gh/apache/beam/pull/12645/diff?src=pr&el=tree#diff-c2Rrcy9weXRob24vYXBhY2hlX2JlYW0vZXhhbXBsZXMvd29yZGNvdW50X3dpdGhfbWV0cmljcy5weQ==) | `28.12% <0.00%> (ø)` | |
| [...ache\_beam/runners/portability/local\_job\_service.py](https://codecov.io/gh/apache/beam/pull/12645/diff?src=pr&el=tree#diff-c2Rrcy9weXRob24vYXBhY2hlX2JlYW0vcnVubmVycy9wb3J0YWJpbGl0eS9sb2NhbF9qb2Jfc2VydmljZS5weQ==) | `81.16% <0.00%> (ø)` | |
| [sdks/python/apache\_beam/examples/flink/\_\_init\_\_.py](https://codecov.io/gh/apache/beam/pull/12645/diff?src=pr&el=tree#diff-c2Rrcy9weXRob24vYXBhY2hlX2JlYW0vZXhhbXBsZXMvZmxpbmsvX19pbml0X18ucHk=) | `0.00% <0.00%> (ø)` | |
| [sdks/python/apache\_beam/internal/\_\_init\_\_.py](https://codecov.io/gh/apache/beam/pull/12645/diff?src=pr&el=tree#diff-c2Rrcy9weXRob24vYXBhY2hlX2JlYW0vaW50ZXJuYWwvX19pbml0X18ucHk=) | `100.00% <0.00%> (ø)` | |
| [...thon/apache\_beam/io/azure/blobstoragefilesystem.py](https://codecov.io/gh/apache/beam/pull/12645/diff?src=pr&el=tree#diff-c2Rrcy9weXRob24vYXBhY2hlX2JlYW0vaW8vYXp1cmUvYmxvYnN0b3JhZ2VmaWxlc3lzdGVtLnB5) | `77.31% <0.00%> (ø)` | |
| [...\_beam/testing/benchmarks/nexmark/queries/query6.py](https://codecov.io/gh/apache/beam/pull/12645/diff?src=pr&el=tree#diff-c2Rrcy9weXRob24vYXBhY2hlX2JlYW0vdGVzdGluZy9iZW5jaG1hcmtzL25leG1hcmsvcXVlcmllcy9xdWVyeTYucHk=) | `0.00% <0.00%> (ø)` | |
| [...s/snippets/transforms/aggregation/combineperkey.py](https://codecov.io/gh/apache/beam/pull/12645/diff?src=pr&el=tree#diff-c2Rrcy9weXRob24vYXBhY2hlX2JlYW0vZXhhbXBsZXMvc25pcHBldHMvdHJhbnNmb3Jtcy9hZ2dyZWdhdGlvbi9jb21iaW5lcGVya2V5LnB5) | `98.91% <0.00%> (ø)` | |
| ... and [443 more](https://codecov.io/gh/apache/beam/pull/12645/diff?src=pr&el=tree-more) | |
------
[Continue to review full report at Codecov](https://codecov.io/gh/apache/beam/pull/12645?src=pr&el=continue).
> **Legend** - [Click here to learn more](https://docs.codecov.io/docs/codecov-delta)
> `Δ = absolute <relative> (impact)`, `ø = not affected`, `? = missing data`
> Powered by [Codecov](https://codecov.io/gh/apache/beam/pull/12645?src=pr&el=footer). Last update [2b2b8e7...3bed6b7](https://codecov.io/gh/apache/beam/pull/12645?src=pr&el=lastupdated). Read the [comment docs](https://docs.codecov.io/docs/pull-request-comments).
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
[GitHub] [beam] codecov[bot] edited a comment on pull request #12645: [BEAM-10124] Add ContextualTextIO
Posted by GitBox <gi...@apache.org>.
codecov[bot] edited a comment on pull request #12645:
URL: https://github.com/apache/beam/pull/12645#issuecomment-688630083
# [Codecov](https://codecov.io/gh/apache/beam/pull/12645?src=pr&el=h1) Report
> :exclamation: No coverage uploaded for pull request base (`master@2b2b8e7`). [Click here to learn what that means](https://docs.codecov.io/docs/error-reference#section-missing-base-commit).
> The diff coverage is `n/a`.
[](https://codecov.io/gh/apache/beam/pull/12645?src=pr&el=tree)
```diff
@@ Coverage Diff @@
## master #12645 +/- ##
=========================================
Coverage ? 82.33%
=========================================
Files ? 453
Lines ? 54054
Branches ? 0
=========================================
Hits ? 44507
Misses ? 9547
Partials ? 0
```
| [Impacted Files](https://codecov.io/gh/apache/beam/pull/12645?src=pr&el=tree) | Coverage Δ | |
|---|---|---|
| [sdks/python/apache\_beam/version.py](https://codecov.io/gh/apache/beam/pull/12645/diff?src=pr&el=tree#diff-c2Rrcy9weXRob24vYXBhY2hlX2JlYW0vdmVyc2lvbi5weQ==) | `100.00% <0.00%> (ø)` | |
| [sdks/python/apache\_beam/transforms/\_\_init\_\_.py](https://codecov.io/gh/apache/beam/pull/12645/diff?src=pr&el=tree#diff-c2Rrcy9weXRob24vYXBhY2hlX2JlYW0vdHJhbnNmb3Jtcy9fX2luaXRfXy5weQ==) | `100.00% <0.00%> (ø)` | |
| [...s/python/apache\_beam/examples/wordcount\_minimal.py](https://codecov.io/gh/apache/beam/pull/12645/diff?src=pr&el=tree#diff-c2Rrcy9weXRob24vYXBhY2hlX2JlYW0vZXhhbXBsZXMvd29yZGNvdW50X21pbmltYWwucHk=) | `93.33% <0.00%> (ø)` | |
| [...on/apache\_beam/portability/api/metrics\_pb2\_grpc.py](https://codecov.io/gh/apache/beam/pull/12645/diff?src=pr&el=tree#diff-c2Rrcy9weXRob24vYXBhY2hlX2JlYW0vcG9ydGFiaWxpdHkvYXBpL21ldHJpY3NfcGIyX2dycGMucHk=) | `0.00% <0.00%> (ø)` | |
| [...ache\_beam/examples/cookbook/datastore\_wordcount.py](https://codecov.io/gh/apache/beam/pull/12645/diff?src=pr&el=tree#diff-c2Rrcy9weXRob24vYXBhY2hlX2JlYW0vZXhhbXBsZXMvY29va2Jvb2svZGF0YXN0b3JlX3dvcmRjb3VudC5weQ==) | `32.72% <0.00%> (ø)` | |
| [...dks/python/apache\_beam/transforms/external\_java.py](https://codecov.io/gh/apache/beam/pull/12645/diff?src=pr&el=tree#diff-c2Rrcy9weXRob24vYXBhY2hlX2JlYW0vdHJhbnNmb3Jtcy9leHRlcm5hbF9qYXZhLnB5) | `0.00% <0.00%> (ø)` | |
| [...pache\_beam/typehints/trivial\_inference\_test\_py3.py](https://codecov.io/gh/apache/beam/pull/12645/diff?src=pr&el=tree#diff-c2Rrcy9weXRob24vYXBhY2hlX2JlYW0vdHlwZWhpbnRzL3RyaXZpYWxfaW5mZXJlbmNlX3Rlc3RfcHkzLnB5) | `92.85% <0.00%> (ø)` | |
| [...on/apache\_beam/runners/direct/watermark\_manager.py](https://codecov.io/gh/apache/beam/pull/12645/diff?src=pr&el=tree#diff-c2Rrcy9weXRob24vYXBhY2hlX2JlYW0vcnVubmVycy9kaXJlY3Qvd2F0ZXJtYXJrX21hbmFnZXIucHk=) | `97.17% <0.00%> (ø)` | |
| [sdks/python/apache\_beam/internal/pickler.py](https://codecov.io/gh/apache/beam/pull/12645/diff?src=pr&el=tree#diff-c2Rrcy9weXRob24vYXBhY2hlX2JlYW0vaW50ZXJuYWwvcGlja2xlci5weQ==) | `85.82% <0.00%> (ø)` | |
| [sdks/python/apache\_beam/internal/\_\_init\_\_.py](https://codecov.io/gh/apache/beam/pull/12645/diff?src=pr&el=tree#diff-c2Rrcy9weXRob24vYXBhY2hlX2JlYW0vaW50ZXJuYWwvX19pbml0X18ucHk=) | `100.00% <0.00%> (ø)` | |
| ... and [443 more](https://codecov.io/gh/apache/beam/pull/12645/diff?src=pr&el=tree-more) | |
------
[Continue to review full report at Codecov](https://codecov.io/gh/apache/beam/pull/12645?src=pr&el=continue).
> **Legend** - [Click here to learn more](https://docs.codecov.io/docs/codecov-delta)
> `Δ = absolute <relative> (impact)`, `ø = not affected`, `? = missing data`
> Powered by [Codecov](https://codecov.io/gh/apache/beam/pull/12645?src=pr&el=footer). Last update [2b2b8e7...85c579f](https://codecov.io/gh/apache/beam/pull/12645?src=pr&el=lastupdated). Read the [comment docs](https://docs.codecov.io/docs/pull-request-comments).
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
[GitHub] [beam] abhiy13 commented on a change in pull request #12645: [BEAM-10124] Add ContextualTextIO
Posted by GitBox <gi...@apache.org>.
abhiy13 commented on a change in pull request #12645:
URL: https://github.com/apache/beam/pull/12645#discussion_r474516753
##########
File path: sdks/java/io/contextual-text-io/src/main/java/org/apache/beam/sdk/io/contextualtextio/RecordWithMetadata.java
##########
@@ -19,21 +19,45 @@
import com.google.auto.value.AutoValue;
import org.apache.beam.sdk.annotations.Experimental;
-import org.apache.beam.sdk.annotations.Internal;
import org.apache.beam.sdk.schemas.AutoValueSchema;
import org.apache.beam.sdk.schemas.annotations.DefaultSchema;
-@Internal
+/**
+ * Helper Class based on {@link AutoValueSchema}, it provides Metadata associated with each Record
+ * when reading from file(s) using {@link ContextualTextIO}.
+ *
+ * <h3>Fields:</h3>
+ *
+ * <ul>
+ * <li>recordOffset: The offset of a record (the byte at which the record begins) in a file. This
+ * information can be useful if you wish to reconstruct the file. {@link
+ * RecordWithMetadata#getRecordOffset()}
+ * <li>recordNum: The record number of the record from its ordinal file. {@link
+ * RecordWithMetadata#getRecordNum()}
+ * <li>recordValue: The value / contents of the records {@link
+ * RecordWithMetadata#getRecordValue()}
+ * <li>startingOffset: The starting offset of the range to which the record belongs. {@link
Review comment:
Thanks.
`rangeOffset` seems like a good name.
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
[GitHub] [beam] rezarokni commented on a change in pull request #12645: [BEAM-10124] Add ContextualTextIO
Posted by GitBox <gi...@apache.org>.
rezarokni commented on a change in pull request #12645:
URL: https://github.com/apache/beam/pull/12645#discussion_r473578722
##########
File path: sdks/java/core/src/main/java/org/apache/beam/sdk/io/FileBasedSource.java
##########
@@ -130,7 +130,7 @@ protected FileBasedSource(
*
* @throws IllegalArgumentException if this source is in {@link Mode#FILEPATTERN} mode.
*/
- protected final MatchResult.Metadata getSingleFileMetadata() {
+ public final MatchResult.Metadata getSingleFileMetadata() {
Review comment:
@tvalentyn what do you think about this and the FileIO changes being done in a different commit / pr . As its a change that is potentially useful to other sources api dev and maybe easier to reason about then if tied to this larger commit?
##########
File path: sdks/java/io/contextual-text-io/src/main/java/org/apache/beam/sdk/io/contextualtextio/ContextualTextIO.java
##########
@@ -0,0 +1,594 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one
+ * or more contributor license agreements. See the NOTICE file
+ * distributed with this work for additional information
+ * regarding copyright ownership. The ASF licenses this file
+ * to you under the Apache License, Version 2.0 (the
+ * "License"); you may not use this file except in compliance
+ * with the License. You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+package org.apache.beam.sdk.io.contextualtextio;
+
+import static org.apache.beam.sdk.io.FileIO.ReadMatches.DirectoryTreatment;
+import static org.apache.beam.vendor.guava.v26_0_jre.com.google.common.base.Preconditions.checkArgument;
+import static org.apache.beam.vendor.guava.v26_0_jre.com.google.common.base.Preconditions.checkNotNull;
+
+import com.google.auto.value.AutoValue;
+import java.nio.ByteBuffer;
+import java.util.Arrays;
+import java.util.Comparator;
+import java.util.HashMap;
+import java.util.Map;
+import java.util.SortedMap;
+import java.util.TreeMap;
+import org.apache.beam.sdk.coders.StringUtf8Coder;
+import org.apache.beam.sdk.io.CompressedSource;
+import org.apache.beam.sdk.io.Compression;
+import org.apache.beam.sdk.io.FileBasedSource;
+import org.apache.beam.sdk.io.FileIO;
+import org.apache.beam.sdk.io.FileIO.MatchConfiguration;
+import org.apache.beam.sdk.io.ReadAllViaFileBasedSource;
+import org.apache.beam.sdk.io.fs.EmptyMatchTreatment;
+import org.apache.beam.sdk.options.ValueProvider;
+import org.apache.beam.sdk.options.ValueProvider.StaticValueProvider;
+import org.apache.beam.sdk.schemas.NoSuchSchemaException;
+import org.apache.beam.sdk.schemas.SchemaCoder;
+import org.apache.beam.sdk.transforms.Count;
+import org.apache.beam.sdk.transforms.Create;
+import org.apache.beam.sdk.transforms.DoFn;
+import org.apache.beam.sdk.transforms.PTransform;
+import org.apache.beam.sdk.transforms.ParDo;
+import org.apache.beam.sdk.transforms.SerializableFunction;
+import org.apache.beam.sdk.transforms.View;
+import org.apache.beam.sdk.transforms.Watch.Growth.TerminationCondition;
+import org.apache.beam.sdk.transforms.display.DisplayData;
+import org.apache.beam.sdk.values.KV;
+import org.apache.beam.sdk.values.PBegin;
+import org.apache.beam.sdk.values.PCollection;
+import org.apache.beam.sdk.values.PCollectionView;
+import org.apache.beam.vendor.guava.v26_0_jre.com.google.common.annotations.VisibleForTesting;
+import org.checkerframework.checker.nullness.qual.Nullable;
+import org.joda.time.Duration;
+
+/**
+ * {@link PTransform}s that read text files and collect contextual information of the elements in
+ * the input.
+ *
+ * <h2>Reading from text files</h2>
+ *
+ * <p>To read a {@link PCollection} from one or more text files, use {@code
+ * ContextualTextIO.read()}. To instantiate a transform use {@link
+ * ContextualTextIO.Read#from(String)} and specify the path of the file(s) to be read.
+ * Alternatively, if the filenames to be read are themselves in a {@link PCollection} you can use
+ * {@link FileIO} to match them and {@link ContextualTextIO#readFiles()} to read them.
+ *
+ * <p>{@link #read} returns a {@link PCollection} of {@link RecordWithMetadata RecordWithMetadata},
+ * each corresponding to one line of an inout UTF-8 text file (split into lines delimited by '\n',
+ * '\r', '\r\n', or specified delimiter see {@link ContextualTextIO.Read#withDelimiter})
+ *
+ * <h3>Filepattern expansion and watching</h3>
+ *
+ * <p>By default, the filepatterns are expanded only once. The combination of {@link
+ * FileIO.Match#continuously(Duration, TerminationCondition)} and {@link #readFiles()} allow
+ * streaming of new files matching the filepattern(s).
+ *
+ * <p>By default, {@link #read} prohibits filepatterns that match no files, and {@link #readFiles()}
+ * allows them in case the filepattern contains a glob wildcard character. Use {@link
+ * ContextualTextIO.Read#withEmptyMatchTreatment} or {@link
+ * FileIO.Match#withEmptyMatchTreatment(EmptyMatchTreatment)} plus {@link #readFiles()} to configure
+ * this behavior.
+ *
+ * <p>Example 1: reading a file or filepattern.
+ *
+ * <pre>{@code
+ * Pipeline p = ...;
+ *
+ * // A simple Read of a file:
+ * PCollection<RecordWithMetadata> lines = p.apply(ContextualTextIO.read().from("/local/path/to/file.txt"));
+ * }</pre>
+ *
+ * <p>Example 2: reading a PCollection of filenames.
+ *
+ * <pre>{@code
+ * Pipeline p = ...;
+ *
+ * // E.g. the filenames might be computed from other data in the pipeline, or
+ * // read from a data source.
+ * PCollection<String> filenames = ...;
+ *
+ * // Read all files in the collection.
+ * PCollection<RecordWithMetadata> lines =
+ * filenames
+ * .apply(FileIO.matchAll())
+ * .apply(FileIO.readMatches())
+ * .apply(ContextualTextIO.readFiles());
+ * }</pre>
+ *
+ * <p>Example 3: streaming new files matching a filepattern.
+ *
+ * <pre>{@code
+ * Pipeline p = ...;
+ *
+ * PCollection<RecordWithMetadata> lines = p.apply(ContextualTextIO.read()
+ * .from("/local/path/to/files/*")
+ * .watchForNewFiles(
+ * // Check for new files every minute
+ * Duration.standardMinutes(1),
+ * // Stop watching the filepattern if no new files appear within an hour
+ * afterTimeSinceNewOutput(Duration.standardHours(1))));
+ * }</pre>
+ *
+ * <p>Example 4: reading a file or file pattern of RFC4180-compliant CSV files with fields that may
+ * contain line breaks.
+ *
+ * <pre>{@code
+ * Pipeline p = ...;
+ *
+ * PCollection<RecordWithMetadata> lines = p.apply(ContextualTextIO.read()
+ * .from("/local/path/to/files/*.csv")
+ * .withHasMultilineCSVRecords(true));
+ * }</pre>
+ *
+ * <p>Example 5: reading while watching for new files
+ *
+ * <pre>{@code
+ * Pipeline p = ...;
+ *
+ * PCollection<RecordWithMetadata> lines = p.apply(FileIO.match()
+ * .filepattern("filepattern")
+ * .continuously(
+ * Duration.millis(100),
+ * Watch.Growth.afterTimeSinceNewOutput(Duration.standardSeconds(3))))
+ * .apply(FileIO.readMatches())
+ * .apply(ContextualTextIO.readFiles());
+ * }</pre>
+ *
+ * NOTE: Using {@link ContextualTextIO.Read#withHasMultilineCSVRecords(Boolean)} introduces a
+ * performance penalty: when this option is enabled, the input cannot be split and read in parallel.
+ *
+ * <h3>Reading a very large number of files</h3>
+ *
+ * <p>If it is known that the filepattern will match a very large number of files (e.g. tens of
+ * thousands or more), use {@link ContextualTextIO.Read#withHintMatchesManyFiles} for better
+ * performance and scalability. Note that it may decrease performance if the filepattern matches
+ * only a small number of files.
+ */
+public class ContextualTextIO {
+ private static final long DEFAULT_BUNDLE_SIZE_BYTES = 64 * 1024 * 1024L;
+
+ /**
+ * A {@link PTransform} that reads from one or more text files and returns a bounded {@link
+ * PCollection} containing one {@link RecordWithMetadata}element for each line of the input files.
+ */
+ public static Read read() {
+ return new AutoValue_ContextualTextIO_Read.Builder()
+ .setCompression(Compression.AUTO)
+ .setHintMatchesManyFiles(false)
+ .setWithoutLineNumMetadata(false)
+ .setMatchConfiguration(MatchConfiguration.create(EmptyMatchTreatment.DISALLOW))
+ .setHasMultilineCSVRecords(false)
+ .build();
+ }
+
+ /**
+ * Like {@link #read}, but reads each file in a {@link PCollection} of {@link
+ * FileIO.ReadableFile}, returned by {@link FileIO#readMatches}.
+ */
+ public static ReadFiles readFiles() {
+ return new AutoValue_ContextualTextIO_ReadFiles.Builder()
+ // 64MB is a reasonable value that allows to amortize the cost of opening files,
+ // but is not so large as to exhaust a typical runner's maximum amount of output per
+ // ProcessElement call.
+ .setDesiredBundleSizeBytes(DEFAULT_BUNDLE_SIZE_BYTES)
+ .setHasMultilineCSVRecords(false)
+ .build();
+ }
+
+ /** Implementation of {@link #read}. */
+ @AutoValue
+ public abstract static class Read extends PTransform<PBegin, PCollection<RecordWithMetadata>> {
+ abstract @Nullable ValueProvider<String> getFilepattern();
+
+ abstract MatchConfiguration getMatchConfiguration();
+
+ abstract boolean getHintMatchesManyFiles();
+
+ abstract boolean getWithoutLineNumMetadata();
+
+ abstract Compression getCompression();
+
+ abstract @Nullable Boolean getHasMultilineCSVRecords();
+
+ @SuppressWarnings("mutable") // this returns an array that can be mutated by the caller
+ abstract byte @Nullable [] getDelimiter();
+
+ abstract Builder toBuilder();
+
+ @AutoValue.Builder
+ abstract static class Builder {
+ abstract Builder setFilepattern(ValueProvider<String> filepattern);
+
+ abstract Builder setMatchConfiguration(MatchConfiguration matchConfiguration);
+
+ abstract Builder setHintMatchesManyFiles(boolean hintManyFiles);
+
+ abstract Builder setWithoutLineNumMetadata(boolean withoutLineNumMetadata);
+
+ abstract Builder setCompression(Compression compression);
+
+ abstract Builder setDelimiter(byte @Nullable [] delimiter);
+
+ abstract Builder setHasMultilineCSVRecords(Boolean hasMultilineCSVRecords);
+
+ abstract Read build();
+ }
+
+ /**
+ * Reads text from the file(s) with the given filename or filename pattern.
+ *
+ * <p>This can be a local path (if running locally), or a Google Cloud Storage filename or
+ * filename pattern of the form {@code "gs://<bucket>/<filepath>"} (if running locally or using
+ * remote execution service).
+ *
+ * <p>Standard <a href="http://docs.oracle.com/javase/tutorial/essential/io/find.html" >Java
+ * Filesystem glob patterns</a> ("*", "?", "[..]") are supported.
+ *
+ * <p>If it is known that the filepattern will match a very large number of files (at least tens
+ * of thousands), use {@link #withHintMatchesManyFiles} for better performance and scalability.
+ */
+ public Read from(String filepattern) {
+ checkArgument(filepattern != null, "filepattern can not be null");
+ return from(StaticValueProvider.of(filepattern));
+ }
+
+ /** Same as {@code from(filepattern)}, but accepting a {@link ValueProvider}. */
+ public Read from(ValueProvider<String> filepattern) {
+ checkArgument(filepattern != null, "filepattern can not be null");
+ return toBuilder().setFilepattern(filepattern).build();
+ }
+
+ /** Sets the {@link MatchConfiguration}. */
+ public Read withMatchConfiguration(MatchConfiguration matchConfiguration) {
+ return toBuilder().setMatchConfiguration(matchConfiguration).build();
+ }
+
+ /**
+ * When reading RFC4180 CSV files that have values that span multiple lines, set this to true.
+ * Note: this reduces the read performance (see: {@link ContextualTextIO}).
+ */
+ public Read withHasMultilineCSVRecords(Boolean hasMultilineCSVRecords) {
+ return toBuilder().setHasMultilineCSVRecords(hasMultilineCSVRecords).build();
+ }
+
+ /**
+ * Reads from input sources using the specified compression type.
+ *
+ * <p>If no compression type is specified, the default is {@link Compression#AUTO}.
+ */
+ public Read withCompression(Compression compression) {
+ return toBuilder().setCompression(compression).build();
+ }
+
+ /**
+ * Hints that the filepattern specified in {@link #from(String)} matches a very large number of
+ * files.
+ *
+ * <p>This hint may cause a runner to execute the transform differently, in a way that improves
+ * performance for this case, but it may worsen performance if the filepattern matches only a
+ * small number of files (e.g., in a runner that supports dynamic work rebalancing, it will
+ * happen less efficiently within individual files).
+ */
+ public Read withHintMatchesManyFiles() {
+ return toBuilder().setHintMatchesManyFiles(true).build();
+ }
+
+ /**
+ * Allows the user the opt out of getting recordNums associated with each record.
+ *
+ * <p>ContextualTextIO uses a shuffle step to assemble the recordNums for each record which may
Review comment:
I think it would be better to add something like, this will introduce a shuffle step, which will increase the resources used by the pipeline.
##########
File path: sdks/java/io/contextual-text-io/src/main/java/org/apache/beam/sdk/io/contextualtextio/ContextualTextIO.java
##########
@@ -0,0 +1,594 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one
+ * or more contributor license agreements. See the NOTICE file
+ * distributed with this work for additional information
+ * regarding copyright ownership. The ASF licenses this file
+ * to you under the Apache License, Version 2.0 (the
+ * "License"); you may not use this file except in compliance
+ * with the License. You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+package org.apache.beam.sdk.io.contextualtextio;
+
+import static org.apache.beam.sdk.io.FileIO.ReadMatches.DirectoryTreatment;
+import static org.apache.beam.vendor.guava.v26_0_jre.com.google.common.base.Preconditions.checkArgument;
+import static org.apache.beam.vendor.guava.v26_0_jre.com.google.common.base.Preconditions.checkNotNull;
+
+import com.google.auto.value.AutoValue;
+import java.nio.ByteBuffer;
+import java.util.Arrays;
+import java.util.Comparator;
+import java.util.HashMap;
+import java.util.Map;
+import java.util.SortedMap;
+import java.util.TreeMap;
+import org.apache.beam.sdk.coders.StringUtf8Coder;
+import org.apache.beam.sdk.io.CompressedSource;
+import org.apache.beam.sdk.io.Compression;
+import org.apache.beam.sdk.io.FileBasedSource;
+import org.apache.beam.sdk.io.FileIO;
+import org.apache.beam.sdk.io.FileIO.MatchConfiguration;
+import org.apache.beam.sdk.io.ReadAllViaFileBasedSource;
+import org.apache.beam.sdk.io.fs.EmptyMatchTreatment;
+import org.apache.beam.sdk.options.ValueProvider;
+import org.apache.beam.sdk.options.ValueProvider.StaticValueProvider;
+import org.apache.beam.sdk.schemas.NoSuchSchemaException;
+import org.apache.beam.sdk.schemas.SchemaCoder;
+import org.apache.beam.sdk.transforms.Count;
+import org.apache.beam.sdk.transforms.Create;
+import org.apache.beam.sdk.transforms.DoFn;
+import org.apache.beam.sdk.transforms.PTransform;
+import org.apache.beam.sdk.transforms.ParDo;
+import org.apache.beam.sdk.transforms.SerializableFunction;
+import org.apache.beam.sdk.transforms.View;
+import org.apache.beam.sdk.transforms.Watch.Growth.TerminationCondition;
+import org.apache.beam.sdk.transforms.display.DisplayData;
+import org.apache.beam.sdk.values.KV;
+import org.apache.beam.sdk.values.PBegin;
+import org.apache.beam.sdk.values.PCollection;
+import org.apache.beam.sdk.values.PCollectionView;
+import org.apache.beam.vendor.guava.v26_0_jre.com.google.common.annotations.VisibleForTesting;
+import org.checkerframework.checker.nullness.qual.Nullable;
+import org.joda.time.Duration;
+
+/**
+ * {@link PTransform}s that read text files and collect contextual information of the elements in
+ * the input.
+ *
+ * <h2>Reading from text files</h2>
+ *
+ * <p>To read a {@link PCollection} from one or more text files, use {@code
+ * ContextualTextIO.read()}. To instantiate a transform use {@link
+ * ContextualTextIO.Read#from(String)} and specify the path of the file(s) to be read.
+ * Alternatively, if the filenames to be read are themselves in a {@link PCollection} you can use
+ * {@link FileIO} to match them and {@link ContextualTextIO#readFiles()} to read them.
+ *
+ * <p>{@link #read} returns a {@link PCollection} of {@link RecordWithMetadata RecordWithMetadata},
+ * each corresponding to one line of an inout UTF-8 text file (split into lines delimited by '\n',
+ * '\r', '\r\n', or specified delimiter see {@link ContextualTextIO.Read#withDelimiter})
+ *
+ * <h3>Filepattern expansion and watching</h3>
+ *
+ * <p>By default, the filepatterns are expanded only once. The combination of {@link
+ * FileIO.Match#continuously(Duration, TerminationCondition)} and {@link #readFiles()} allow
+ * streaming of new files matching the filepattern(s).
+ *
+ * <p>By default, {@link #read} prohibits filepatterns that match no files, and {@link #readFiles()}
+ * allows them in case the filepattern contains a glob wildcard character. Use {@link
+ * ContextualTextIO.Read#withEmptyMatchTreatment} or {@link
+ * FileIO.Match#withEmptyMatchTreatment(EmptyMatchTreatment)} plus {@link #readFiles()} to configure
+ * this behavior.
+ *
+ * <p>Example 1: reading a file or filepattern.
+ *
+ * <pre>{@code
+ * Pipeline p = ...;
+ *
+ * // A simple Read of a file:
+ * PCollection<RecordWithMetadata> lines = p.apply(ContextualTextIO.read().from("/local/path/to/file.txt"));
+ * }</pre>
+ *
+ * <p>Example 2: reading a PCollection of filenames.
+ *
+ * <pre>{@code
+ * Pipeline p = ...;
+ *
+ * // E.g. the filenames might be computed from other data in the pipeline, or
+ * // read from a data source.
+ * PCollection<String> filenames = ...;
+ *
+ * // Read all files in the collection.
+ * PCollection<RecordWithMetadata> lines =
+ * filenames
+ * .apply(FileIO.matchAll())
+ * .apply(FileIO.readMatches())
+ * .apply(ContextualTextIO.readFiles());
+ * }</pre>
+ *
+ * <p>Example 3: streaming new files matching a filepattern.
+ *
+ * <pre>{@code
+ * Pipeline p = ...;
+ *
+ * PCollection<RecordWithMetadata> lines = p.apply(ContextualTextIO.read()
+ * .from("/local/path/to/files/*")
+ * .watchForNewFiles(
+ * // Check for new files every minute
+ * Duration.standardMinutes(1),
+ * // Stop watching the filepattern if no new files appear within an hour
+ * afterTimeSinceNewOutput(Duration.standardHours(1))));
+ * }</pre>
+ *
+ * <p>Example 4: reading a file or file pattern of RFC4180-compliant CSV files with fields that may
Review comment:
Show example of what the field may look like, I think the example from the RFC is fine.
##########
File path: sdks/java/io/contextual-text-io/src/main/java/org/apache/beam/sdk/io/contextualtextio/ContextualTextIOSource.java
##########
@@ -0,0 +1,363 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one
+ * or more contributor license agreements. See the NOTICE file
+ * distributed with this work for additional information
+ * regarding copyright ownership. The ASF licenses this file
+ * to you under the Apache License, Version 2.0 (the
+ * "License"); you may not use this file except in compliance
+ * with the License. You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+package org.apache.beam.sdk.io.contextualtextio;
+
+import java.io.IOException;
+import java.nio.ByteBuffer;
+import java.nio.channels.ReadableByteChannel;
+import java.nio.channels.SeekableByteChannel;
+import java.util.NoSuchElementException;
+import org.apache.beam.sdk.coders.Coder;
+import org.apache.beam.sdk.io.FileBasedSource;
+import org.apache.beam.sdk.io.fs.EmptyMatchTreatment;
+import org.apache.beam.sdk.io.fs.MatchResult;
+import org.apache.beam.sdk.options.PipelineOptions;
+import org.apache.beam.sdk.options.ValueProvider;
+import org.apache.beam.sdk.schemas.NoSuchSchemaException;
+import org.apache.beam.sdk.schemas.SchemaCoder;
+import org.apache.beam.sdk.schemas.SchemaRegistry;
+import org.apache.beam.vendor.grpc.v1p26p0.com.google.protobuf.ByteString;
+import org.apache.beam.vendor.guava.v26_0_jre.com.google.common.annotations.VisibleForTesting;
+import org.apache.beam.vendor.guava.v26_0_jre.com.google.common.base.Preconditions;
+import org.checkerframework.checker.nullness.qual.Nullable;
+
+/**
+ * Implementation detail of {@link ContextualTextIO.Read}.
+ *
+ * <p>A {@link FileBasedSource} which can decode records delimited by newline characters.
+ *
+ * <p>This source splits the data into records using {@code UTF-8} {@code \n}, {@code \r}, or {@code
+ * \r\n} as the delimiter. This source is not strict and supports decoding the last record even if
+ * it is not delimited. Finally, no records are decoded if the stream is empty.
+ *
+ * <p>This source supports reading from any arbitrary byte position within the stream. If the
+ * starting position is not {@code 0}, then bytes are skipped until the first delimiter is found
+ * representing the beginning of the first record to be decoded.
+ */
+@VisibleForTesting
+class ContextualTextIOSource extends FileBasedSource<RecordWithMetadata> {
+ byte[] delimiter;
+
+ // Used to Override isSplittable
+ private boolean hasMultilineCSVRecords;
+
+ @Override
+ protected boolean isSplittable() throws Exception {
+ if (hasMultilineCSVRecords) {
Review comment:
Create comment explaing why we return false
##########
File path: sdks/java/io/contextual-text-io/src/main/java/org/apache/beam/sdk/io/contextualtextio/ContextualTextIO.java
##########
@@ -0,0 +1,594 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one
+ * or more contributor license agreements. See the NOTICE file
+ * distributed with this work for additional information
+ * regarding copyright ownership. The ASF licenses this file
+ * to you under the Apache License, Version 2.0 (the
+ * "License"); you may not use this file except in compliance
+ * with the License. You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+package org.apache.beam.sdk.io.contextualtextio;
+
+import static org.apache.beam.sdk.io.FileIO.ReadMatches.DirectoryTreatment;
+import static org.apache.beam.vendor.guava.v26_0_jre.com.google.common.base.Preconditions.checkArgument;
+import static org.apache.beam.vendor.guava.v26_0_jre.com.google.common.base.Preconditions.checkNotNull;
+
+import com.google.auto.value.AutoValue;
+import java.nio.ByteBuffer;
+import java.util.Arrays;
+import java.util.Comparator;
+import java.util.HashMap;
+import java.util.Map;
+import java.util.SortedMap;
+import java.util.TreeMap;
+import org.apache.beam.sdk.coders.StringUtf8Coder;
+import org.apache.beam.sdk.io.CompressedSource;
+import org.apache.beam.sdk.io.Compression;
+import org.apache.beam.sdk.io.FileBasedSource;
+import org.apache.beam.sdk.io.FileIO;
+import org.apache.beam.sdk.io.FileIO.MatchConfiguration;
+import org.apache.beam.sdk.io.ReadAllViaFileBasedSource;
+import org.apache.beam.sdk.io.fs.EmptyMatchTreatment;
+import org.apache.beam.sdk.options.ValueProvider;
+import org.apache.beam.sdk.options.ValueProvider.StaticValueProvider;
+import org.apache.beam.sdk.schemas.NoSuchSchemaException;
+import org.apache.beam.sdk.schemas.SchemaCoder;
+import org.apache.beam.sdk.transforms.Count;
+import org.apache.beam.sdk.transforms.Create;
+import org.apache.beam.sdk.transforms.DoFn;
+import org.apache.beam.sdk.transforms.PTransform;
+import org.apache.beam.sdk.transforms.ParDo;
+import org.apache.beam.sdk.transforms.SerializableFunction;
+import org.apache.beam.sdk.transforms.View;
+import org.apache.beam.sdk.transforms.Watch.Growth.TerminationCondition;
+import org.apache.beam.sdk.transforms.display.DisplayData;
+import org.apache.beam.sdk.values.KV;
+import org.apache.beam.sdk.values.PBegin;
+import org.apache.beam.sdk.values.PCollection;
+import org.apache.beam.sdk.values.PCollectionView;
+import org.apache.beam.vendor.guava.v26_0_jre.com.google.common.annotations.VisibleForTesting;
+import org.checkerframework.checker.nullness.qual.Nullable;
+import org.joda.time.Duration;
+
+/**
+ * {@link PTransform}s that read text files and collect contextual information of the elements in
+ * the input.
+ *
+ * <h2>Reading from text files</h2>
+ *
+ * <p>To read a {@link PCollection} from one or more text files, use {@code
+ * ContextualTextIO.read()}. To instantiate a transform use {@link
+ * ContextualTextIO.Read#from(String)} and specify the path of the file(s) to be read.
+ * Alternatively, if the filenames to be read are themselves in a {@link PCollection} you can use
+ * {@link FileIO} to match them and {@link ContextualTextIO#readFiles()} to read them.
+ *
+ * <p>{@link #read} returns a {@link PCollection} of {@link RecordWithMetadata RecordWithMetadata},
+ * each corresponding to one line of an inout UTF-8 text file (split into lines delimited by '\n',
Review comment:
sorry not sure if inout is a typo?
##########
File path: sdks/java/io/contextual-text-io/src/main/java/org/apache/beam/sdk/io/contextualtextio/ContextualTextIO.java
##########
@@ -0,0 +1,594 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one
+ * or more contributor license agreements. See the NOTICE file
+ * distributed with this work for additional information
+ * regarding copyright ownership. The ASF licenses this file
+ * to you under the Apache License, Version 2.0 (the
+ * "License"); you may not use this file except in compliance
+ * with the License. You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+package org.apache.beam.sdk.io.contextualtextio;
+
+import static org.apache.beam.sdk.io.FileIO.ReadMatches.DirectoryTreatment;
+import static org.apache.beam.vendor.guava.v26_0_jre.com.google.common.base.Preconditions.checkArgument;
+import static org.apache.beam.vendor.guava.v26_0_jre.com.google.common.base.Preconditions.checkNotNull;
+
+import com.google.auto.value.AutoValue;
+import java.nio.ByteBuffer;
+import java.util.Arrays;
+import java.util.Comparator;
+import java.util.HashMap;
+import java.util.Map;
+import java.util.SortedMap;
+import java.util.TreeMap;
+import org.apache.beam.sdk.coders.StringUtf8Coder;
+import org.apache.beam.sdk.io.CompressedSource;
+import org.apache.beam.sdk.io.Compression;
+import org.apache.beam.sdk.io.FileBasedSource;
+import org.apache.beam.sdk.io.FileIO;
+import org.apache.beam.sdk.io.FileIO.MatchConfiguration;
+import org.apache.beam.sdk.io.ReadAllViaFileBasedSource;
+import org.apache.beam.sdk.io.fs.EmptyMatchTreatment;
+import org.apache.beam.sdk.options.ValueProvider;
+import org.apache.beam.sdk.options.ValueProvider.StaticValueProvider;
+import org.apache.beam.sdk.schemas.NoSuchSchemaException;
+import org.apache.beam.sdk.schemas.SchemaCoder;
+import org.apache.beam.sdk.transforms.Count;
+import org.apache.beam.sdk.transforms.Create;
+import org.apache.beam.sdk.transforms.DoFn;
+import org.apache.beam.sdk.transforms.PTransform;
+import org.apache.beam.sdk.transforms.ParDo;
+import org.apache.beam.sdk.transforms.SerializableFunction;
+import org.apache.beam.sdk.transforms.View;
+import org.apache.beam.sdk.transforms.Watch.Growth.TerminationCondition;
+import org.apache.beam.sdk.transforms.display.DisplayData;
+import org.apache.beam.sdk.values.KV;
+import org.apache.beam.sdk.values.PBegin;
+import org.apache.beam.sdk.values.PCollection;
+import org.apache.beam.sdk.values.PCollectionView;
+import org.apache.beam.vendor.guava.v26_0_jre.com.google.common.annotations.VisibleForTesting;
+import org.checkerframework.checker.nullness.qual.Nullable;
+import org.joda.time.Duration;
+
+/**
+ * {@link PTransform}s that read text files and collect contextual information of the elements in
Review comment:
Add note that ' if contextual information or multiline file reading is not needed, then make use of TextIO'
##########
File path: sdks/java/io/contextual-text-io/src/main/java/org/apache/beam/sdk/io/contextualtextio/ContextualTextIO.java
##########
@@ -0,0 +1,594 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one
+ * or more contributor license agreements. See the NOTICE file
+ * distributed with this work for additional information
+ * regarding copyright ownership. The ASF licenses this file
+ * to you under the Apache License, Version 2.0 (the
+ * "License"); you may not use this file except in compliance
+ * with the License. You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+package org.apache.beam.sdk.io.contextualtextio;
+
+import static org.apache.beam.sdk.io.FileIO.ReadMatches.DirectoryTreatment;
+import static org.apache.beam.vendor.guava.v26_0_jre.com.google.common.base.Preconditions.checkArgument;
+import static org.apache.beam.vendor.guava.v26_0_jre.com.google.common.base.Preconditions.checkNotNull;
+
+import com.google.auto.value.AutoValue;
+import java.nio.ByteBuffer;
+import java.util.Arrays;
+import java.util.Comparator;
+import java.util.HashMap;
+import java.util.Map;
+import java.util.SortedMap;
+import java.util.TreeMap;
+import org.apache.beam.sdk.coders.StringUtf8Coder;
+import org.apache.beam.sdk.io.CompressedSource;
+import org.apache.beam.sdk.io.Compression;
+import org.apache.beam.sdk.io.FileBasedSource;
+import org.apache.beam.sdk.io.FileIO;
+import org.apache.beam.sdk.io.FileIO.MatchConfiguration;
+import org.apache.beam.sdk.io.ReadAllViaFileBasedSource;
+import org.apache.beam.sdk.io.fs.EmptyMatchTreatment;
+import org.apache.beam.sdk.options.ValueProvider;
+import org.apache.beam.sdk.options.ValueProvider.StaticValueProvider;
+import org.apache.beam.sdk.schemas.NoSuchSchemaException;
+import org.apache.beam.sdk.schemas.SchemaCoder;
+import org.apache.beam.sdk.transforms.Count;
+import org.apache.beam.sdk.transforms.Create;
+import org.apache.beam.sdk.transforms.DoFn;
+import org.apache.beam.sdk.transforms.PTransform;
+import org.apache.beam.sdk.transforms.ParDo;
+import org.apache.beam.sdk.transforms.SerializableFunction;
+import org.apache.beam.sdk.transforms.View;
+import org.apache.beam.sdk.transforms.Watch.Growth.TerminationCondition;
+import org.apache.beam.sdk.transforms.display.DisplayData;
+import org.apache.beam.sdk.values.KV;
+import org.apache.beam.sdk.values.PBegin;
+import org.apache.beam.sdk.values.PCollection;
+import org.apache.beam.sdk.values.PCollectionView;
+import org.apache.beam.vendor.guava.v26_0_jre.com.google.common.annotations.VisibleForTesting;
+import org.checkerframework.checker.nullness.qual.Nullable;
+import org.joda.time.Duration;
+
+/**
+ * {@link PTransform}s that read text files and collect contextual information of the elements in
+ * the input.
+ *
+ * <h2>Reading from text files</h2>
+ *
+ * <p>To read a {@link PCollection} from one or more text files, use {@code
+ * ContextualTextIO.read()}. To instantiate a transform use {@link
+ * ContextualTextIO.Read#from(String)} and specify the path of the file(s) to be read.
+ * Alternatively, if the filenames to be read are themselves in a {@link PCollection} you can use
+ * {@link FileIO} to match them and {@link ContextualTextIO#readFiles()} to read them.
+ *
+ * <p>{@link #read} returns a {@link PCollection} of {@link RecordWithMetadata RecordWithMetadata},
+ * each corresponding to one line of an inout UTF-8 text file (split into lines delimited by '\n',
+ * '\r', '\r\n', or specified delimiter see {@link ContextualTextIO.Read#withDelimiter})
+ *
+ * <h3>Filepattern expansion and watching</h3>
+ *
+ * <p>By default, the filepatterns are expanded only once. The combination of {@link
+ * FileIO.Match#continuously(Duration, TerminationCondition)} and {@link #readFiles()} allow
+ * streaming of new files matching the filepattern(s).
+ *
+ * <p>By default, {@link #read} prohibits filepatterns that match no files, and {@link #readFiles()}
+ * allows them in case the filepattern contains a glob wildcard character. Use {@link
+ * ContextualTextIO.Read#withEmptyMatchTreatment} or {@link
+ * FileIO.Match#withEmptyMatchTreatment(EmptyMatchTreatment)} plus {@link #readFiles()} to configure
+ * this behavior.
+ *
+ * <p>Example 1: reading a file or filepattern.
+ *
+ * <pre>{@code
+ * Pipeline p = ...;
+ *
+ * // A simple Read of a file:
+ * PCollection<RecordWithMetadata> lines = p.apply(ContextualTextIO.read().from("/local/path/to/file.txt"));
+ * }</pre>
+ *
+ * <p>Example 2: reading a PCollection of filenames.
+ *
+ * <pre>{@code
+ * Pipeline p = ...;
+ *
+ * // E.g. the filenames might be computed from other data in the pipeline, or
+ * // read from a data source.
+ * PCollection<String> filenames = ...;
+ *
+ * // Read all files in the collection.
+ * PCollection<RecordWithMetadata> lines =
+ * filenames
+ * .apply(FileIO.matchAll())
+ * .apply(FileIO.readMatches())
+ * .apply(ContextualTextIO.readFiles());
+ * }</pre>
+ *
+ * <p>Example 3: streaming new files matching a filepattern.
+ *
+ * <pre>{@code
+ * Pipeline p = ...;
+ *
+ * PCollection<RecordWithMetadata> lines = p.apply(ContextualTextIO.read()
+ * .from("/local/path/to/files/*")
+ * .watchForNewFiles(
+ * // Check for new files every minute
+ * Duration.standardMinutes(1),
+ * // Stop watching the filepattern if no new files appear within an hour
+ * afterTimeSinceNewOutput(Duration.standardHours(1))));
+ * }</pre>
+ *
+ * <p>Example 4: reading a file or file pattern of RFC4180-compliant CSV files with fields that may
+ * contain line breaks.
+ *
+ * <pre>{@code
+ * Pipeline p = ...;
+ *
+ * PCollection<RecordWithMetadata> lines = p.apply(ContextualTextIO.read()
+ * .from("/local/path/to/files/*.csv")
+ * .withHasMultilineCSVRecords(true));
+ * }</pre>
+ *
+ * <p>Example 5: reading while watching for new files
+ *
+ * <pre>{@code
+ * Pipeline p = ...;
+ *
+ * PCollection<RecordWithMetadata> lines = p.apply(FileIO.match()
+ * .filepattern("filepattern")
+ * .continuously(
+ * Duration.millis(100),
+ * Watch.Growth.afterTimeSinceNewOutput(Duration.standardSeconds(3))))
+ * .apply(FileIO.readMatches())
+ * .apply(ContextualTextIO.readFiles());
+ * }</pre>
+ *
+ * NOTE: Using {@link ContextualTextIO.Read#withHasMultilineCSVRecords(Boolean)} introduces a
Review comment:
maybe change to :
when this option is used a single reader will be used to process the file, rather than multiple readers which can read from different offsets. For a large file this can result in lower performance.
##########
File path: sdks/java/io/contextual-text-io/src/main/java/org/apache/beam/sdk/io/contextualtextio/ContextualTextIO.java
##########
@@ -0,0 +1,594 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one
+ * or more contributor license agreements. See the NOTICE file
+ * distributed with this work for additional information
+ * regarding copyright ownership. The ASF licenses this file
+ * to you under the Apache License, Version 2.0 (the
+ * "License"); you may not use this file except in compliance
+ * with the License. You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+package org.apache.beam.sdk.io.contextualtextio;
+
+import static org.apache.beam.sdk.io.FileIO.ReadMatches.DirectoryTreatment;
+import static org.apache.beam.vendor.guava.v26_0_jre.com.google.common.base.Preconditions.checkArgument;
+import static org.apache.beam.vendor.guava.v26_0_jre.com.google.common.base.Preconditions.checkNotNull;
+
+import com.google.auto.value.AutoValue;
+import java.nio.ByteBuffer;
+import java.util.Arrays;
+import java.util.Comparator;
+import java.util.HashMap;
+import java.util.Map;
+import java.util.SortedMap;
+import java.util.TreeMap;
+import org.apache.beam.sdk.coders.StringUtf8Coder;
+import org.apache.beam.sdk.io.CompressedSource;
+import org.apache.beam.sdk.io.Compression;
+import org.apache.beam.sdk.io.FileBasedSource;
+import org.apache.beam.sdk.io.FileIO;
+import org.apache.beam.sdk.io.FileIO.MatchConfiguration;
+import org.apache.beam.sdk.io.ReadAllViaFileBasedSource;
+import org.apache.beam.sdk.io.fs.EmptyMatchTreatment;
+import org.apache.beam.sdk.options.ValueProvider;
+import org.apache.beam.sdk.options.ValueProvider.StaticValueProvider;
+import org.apache.beam.sdk.schemas.NoSuchSchemaException;
+import org.apache.beam.sdk.schemas.SchemaCoder;
+import org.apache.beam.sdk.transforms.Count;
+import org.apache.beam.sdk.transforms.Create;
+import org.apache.beam.sdk.transforms.DoFn;
+import org.apache.beam.sdk.transforms.PTransform;
+import org.apache.beam.sdk.transforms.ParDo;
+import org.apache.beam.sdk.transforms.SerializableFunction;
+import org.apache.beam.sdk.transforms.View;
+import org.apache.beam.sdk.transforms.Watch.Growth.TerminationCondition;
+import org.apache.beam.sdk.transforms.display.DisplayData;
+import org.apache.beam.sdk.values.KV;
+import org.apache.beam.sdk.values.PBegin;
+import org.apache.beam.sdk.values.PCollection;
+import org.apache.beam.sdk.values.PCollectionView;
+import org.apache.beam.vendor.guava.v26_0_jre.com.google.common.annotations.VisibleForTesting;
+import org.checkerframework.checker.nullness.qual.Nullable;
+import org.joda.time.Duration;
+
+/**
+ * {@link PTransform}s that read text files and collect contextual information of the elements in
+ * the input.
+ *
+ * <h2>Reading from text files</h2>
+ *
+ * <p>To read a {@link PCollection} from one or more text files, use {@code
+ * ContextualTextIO.read()}. To instantiate a transform use {@link
+ * ContextualTextIO.Read#from(String)} and specify the path of the file(s) to be read.
+ * Alternatively, if the filenames to be read are themselves in a {@link PCollection} you can use
+ * {@link FileIO} to match them and {@link ContextualTextIO#readFiles()} to read them.
+ *
+ * <p>{@link #read} returns a {@link PCollection} of {@link RecordWithMetadata RecordWithMetadata},
+ * each corresponding to one line of an inout UTF-8 text file (split into lines delimited by '\n',
+ * '\r', '\r\n', or specified delimiter see {@link ContextualTextIO.Read#withDelimiter})
+ *
+ * <h3>Filepattern expansion and watching</h3>
+ *
+ * <p>By default, the filepatterns are expanded only once. The combination of {@link
+ * FileIO.Match#continuously(Duration, TerminationCondition)} and {@link #readFiles()} allow
+ * streaming of new files matching the filepattern(s).
+ *
+ * <p>By default, {@link #read} prohibits filepatterns that match no files, and {@link #readFiles()}
+ * allows them in case the filepattern contains a glob wildcard character. Use {@link
+ * ContextualTextIO.Read#withEmptyMatchTreatment} or {@link
+ * FileIO.Match#withEmptyMatchTreatment(EmptyMatchTreatment)} plus {@link #readFiles()} to configure
+ * this behavior.
+ *
+ * <p>Example 1: reading a file or filepattern.
+ *
+ * <pre>{@code
+ * Pipeline p = ...;
+ *
+ * // A simple Read of a file:
+ * PCollection<RecordWithMetadata> lines = p.apply(ContextualTextIO.read().from("/local/path/to/file.txt"));
+ * }</pre>
+ *
+ * <p>Example 2: reading a PCollection of filenames.
+ *
+ * <pre>{@code
+ * Pipeline p = ...;
+ *
+ * // E.g. the filenames might be computed from other data in the pipeline, or
+ * // read from a data source.
+ * PCollection<String> filenames = ...;
+ *
+ * // Read all files in the collection.
+ * PCollection<RecordWithMetadata> lines =
+ * filenames
+ * .apply(FileIO.matchAll())
+ * .apply(FileIO.readMatches())
+ * .apply(ContextualTextIO.readFiles());
+ * }</pre>
+ *
+ * <p>Example 3: streaming new files matching a filepattern.
+ *
+ * <pre>{@code
+ * Pipeline p = ...;
+ *
+ * PCollection<RecordWithMetadata> lines = p.apply(ContextualTextIO.read()
+ * .from("/local/path/to/files/*")
+ * .watchForNewFiles(
+ * // Check for new files every minute
+ * Duration.standardMinutes(1),
+ * // Stop watching the filepattern if no new files appear within an hour
+ * afterTimeSinceNewOutput(Duration.standardHours(1))));
+ * }</pre>
+ *
+ * <p>Example 4: reading a file or file pattern of RFC4180-compliant CSV files with fields that may
+ * contain line breaks.
+ *
+ * <pre>{@code
+ * Pipeline p = ...;
+ *
+ * PCollection<RecordWithMetadata> lines = p.apply(ContextualTextIO.read()
+ * .from("/local/path/to/files/*.csv")
+ * .withHasMultilineCSVRecords(true));
+ * }</pre>
+ *
+ * <p>Example 5: reading while watching for new files
+ *
+ * <pre>{@code
+ * Pipeline p = ...;
+ *
+ * PCollection<RecordWithMetadata> lines = p.apply(FileIO.match()
+ * .filepattern("filepattern")
+ * .continuously(
+ * Duration.millis(100),
+ * Watch.Growth.afterTimeSinceNewOutput(Duration.standardSeconds(3))))
+ * .apply(FileIO.readMatches())
+ * .apply(ContextualTextIO.readFiles());
+ * }</pre>
+ *
+ * NOTE: Using {@link ContextualTextIO.Read#withHasMultilineCSVRecords(Boolean)} introduces a
+ * performance penalty: when this option is enabled, the input cannot be split and read in parallel.
+ *
+ * <h3>Reading a very large number of files</h3>
+ *
+ * <p>If it is known that the filepattern will match a very large number of files (e.g. tens of
+ * thousands or more), use {@link ContextualTextIO.Read#withHintMatchesManyFiles} for better
+ * performance and scalability. Note that it may decrease performance if the filepattern matches
+ * only a small number of files.
+ */
+public class ContextualTextIO {
+ private static final long DEFAULT_BUNDLE_SIZE_BYTES = 64 * 1024 * 1024L;
+
+ /**
+ * A {@link PTransform} that reads from one or more text files and returns a bounded {@link
Review comment:
explain the other config options used in the example, WithoutLineNumMetadata etc... Also as we have changed from lineContext to Record Contexr should this now be WithoutRecordMetadata?
@tvalentyn fyi
##########
File path: sdks/java/io/contextual-text-io/src/main/java/org/apache/beam/sdk/io/contextualtextio/ContextualTextIO.java
##########
@@ -0,0 +1,594 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one
+ * or more contributor license agreements. See the NOTICE file
+ * distributed with this work for additional information
+ * regarding copyright ownership. The ASF licenses this file
+ * to you under the Apache License, Version 2.0 (the
+ * "License"); you may not use this file except in compliance
+ * with the License. You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+package org.apache.beam.sdk.io.contextualtextio;
+
+import static org.apache.beam.sdk.io.FileIO.ReadMatches.DirectoryTreatment;
+import static org.apache.beam.vendor.guava.v26_0_jre.com.google.common.base.Preconditions.checkArgument;
+import static org.apache.beam.vendor.guava.v26_0_jre.com.google.common.base.Preconditions.checkNotNull;
+
+import com.google.auto.value.AutoValue;
+import java.nio.ByteBuffer;
+import java.util.Arrays;
+import java.util.Comparator;
+import java.util.HashMap;
+import java.util.Map;
+import java.util.SortedMap;
+import java.util.TreeMap;
+import org.apache.beam.sdk.coders.StringUtf8Coder;
+import org.apache.beam.sdk.io.CompressedSource;
+import org.apache.beam.sdk.io.Compression;
+import org.apache.beam.sdk.io.FileBasedSource;
+import org.apache.beam.sdk.io.FileIO;
+import org.apache.beam.sdk.io.FileIO.MatchConfiguration;
+import org.apache.beam.sdk.io.ReadAllViaFileBasedSource;
+import org.apache.beam.sdk.io.fs.EmptyMatchTreatment;
+import org.apache.beam.sdk.options.ValueProvider;
+import org.apache.beam.sdk.options.ValueProvider.StaticValueProvider;
+import org.apache.beam.sdk.schemas.NoSuchSchemaException;
+import org.apache.beam.sdk.schemas.SchemaCoder;
+import org.apache.beam.sdk.transforms.Count;
+import org.apache.beam.sdk.transforms.Create;
+import org.apache.beam.sdk.transforms.DoFn;
+import org.apache.beam.sdk.transforms.PTransform;
+import org.apache.beam.sdk.transforms.ParDo;
+import org.apache.beam.sdk.transforms.SerializableFunction;
+import org.apache.beam.sdk.transforms.View;
+import org.apache.beam.sdk.transforms.Watch.Growth.TerminationCondition;
+import org.apache.beam.sdk.transforms.display.DisplayData;
+import org.apache.beam.sdk.values.KV;
+import org.apache.beam.sdk.values.PBegin;
+import org.apache.beam.sdk.values.PCollection;
+import org.apache.beam.sdk.values.PCollectionView;
+import org.apache.beam.vendor.guava.v26_0_jre.com.google.common.annotations.VisibleForTesting;
+import org.checkerframework.checker.nullness.qual.Nullable;
+import org.joda.time.Duration;
+
+/**
+ * {@link PTransform}s that read text files and collect contextual information of the elements in
+ * the input.
+ *
+ * <h2>Reading from text files</h2>
+ *
+ * <p>To read a {@link PCollection} from one or more text files, use {@code
+ * ContextualTextIO.read()}. To instantiate a transform use {@link
+ * ContextualTextIO.Read#from(String)} and specify the path of the file(s) to be read.
+ * Alternatively, if the filenames to be read are themselves in a {@link PCollection} you can use
+ * {@link FileIO} to match them and {@link ContextualTextIO#readFiles()} to read them.
+ *
+ * <p>{@link #read} returns a {@link PCollection} of {@link RecordWithMetadata RecordWithMetadata},
+ * each corresponding to one line of an inout UTF-8 text file (split into lines delimited by '\n',
+ * '\r', '\r\n', or specified delimiter see {@link ContextualTextIO.Read#withDelimiter})
+ *
+ * <h3>Filepattern expansion and watching</h3>
+ *
+ * <p>By default, the filepatterns are expanded only once. The combination of {@link
+ * FileIO.Match#continuously(Duration, TerminationCondition)} and {@link #readFiles()} allow
+ * streaming of new files matching the filepattern(s).
+ *
+ * <p>By default, {@link #read} prohibits filepatterns that match no files, and {@link #readFiles()}
+ * allows them in case the filepattern contains a glob wildcard character. Use {@link
+ * ContextualTextIO.Read#withEmptyMatchTreatment} or {@link
+ * FileIO.Match#withEmptyMatchTreatment(EmptyMatchTreatment)} plus {@link #readFiles()} to configure
+ * this behavior.
+ *
+ * <p>Example 1: reading a file or filepattern.
+ *
+ * <pre>{@code
+ * Pipeline p = ...;
+ *
+ * // A simple Read of a file:
+ * PCollection<RecordWithMetadata> lines = p.apply(ContextualTextIO.read().from("/local/path/to/file.txt"));
+ * }</pre>
+ *
+ * <p>Example 2: reading a PCollection of filenames.
+ *
+ * <pre>{@code
+ * Pipeline p = ...;
+ *
+ * // E.g. the filenames might be computed from other data in the pipeline, or
+ * // read from a data source.
+ * PCollection<String> filenames = ...;
+ *
+ * // Read all files in the collection.
+ * PCollection<RecordWithMetadata> lines =
+ * filenames
+ * .apply(FileIO.matchAll())
+ * .apply(FileIO.readMatches())
+ * .apply(ContextualTextIO.readFiles());
+ * }</pre>
+ *
+ * <p>Example 3: streaming new files matching a filepattern.
+ *
+ * <pre>{@code
+ * Pipeline p = ...;
+ *
+ * PCollection<RecordWithMetadata> lines = p.apply(ContextualTextIO.read()
+ * .from("/local/path/to/files/*")
+ * .watchForNewFiles(
+ * // Check for new files every minute
+ * Duration.standardMinutes(1),
+ * // Stop watching the filepattern if no new files appear within an hour
+ * afterTimeSinceNewOutput(Duration.standardHours(1))));
+ * }</pre>
+ *
+ * <p>Example 4: reading a file or file pattern of RFC4180-compliant CSV files with fields that may
+ * contain line breaks.
+ *
+ * <pre>{@code
+ * Pipeline p = ...;
+ *
+ * PCollection<RecordWithMetadata> lines = p.apply(ContextualTextIO.read()
+ * .from("/local/path/to/files/*.csv")
+ * .withHasMultilineCSVRecords(true));
+ * }</pre>
+ *
+ * <p>Example 5: reading while watching for new files
+ *
+ * <pre>{@code
+ * Pipeline p = ...;
+ *
+ * PCollection<RecordWithMetadata> lines = p.apply(FileIO.match()
+ * .filepattern("filepattern")
+ * .continuously(
+ * Duration.millis(100),
+ * Watch.Growth.afterTimeSinceNewOutput(Duration.standardSeconds(3))))
+ * .apply(FileIO.readMatches())
+ * .apply(ContextualTextIO.readFiles());
+ * }</pre>
+ *
+ * NOTE: Using {@link ContextualTextIO.Read#withHasMultilineCSVRecords(Boolean)} introduces a
+ * performance penalty: when this option is enabled, the input cannot be split and read in parallel.
+ *
+ * <h3>Reading a very large number of files</h3>
+ *
+ * <p>If it is known that the filepattern will match a very large number of files (e.g. tens of
+ * thousands or more), use {@link ContextualTextIO.Read#withHintMatchesManyFiles} for better
+ * performance and scalability. Note that it may decrease performance if the filepattern matches
+ * only a small number of files.
+ */
+public class ContextualTextIO {
+ private static final long DEFAULT_BUNDLE_SIZE_BYTES = 64 * 1024 * 1024L;
+
+ /**
+ * A {@link PTransform} that reads from one or more text files and returns a bounded {@link
+ * PCollection} containing one {@link RecordWithMetadata}element for each line of the input files.
+ */
+ public static Read read() {
+ return new AutoValue_ContextualTextIO_Read.Builder()
+ .setCompression(Compression.AUTO)
+ .setHintMatchesManyFiles(false)
+ .setWithoutLineNumMetadata(false)
+ .setMatchConfiguration(MatchConfiguration.create(EmptyMatchTreatment.DISALLOW))
+ .setHasMultilineCSVRecords(false)
+ .build();
+ }
+
+ /**
+ * Like {@link #read}, but reads each file in a {@link PCollection} of {@link
+ * FileIO.ReadableFile}, returned by {@link FileIO#readMatches}.
+ */
+ public static ReadFiles readFiles() {
Review comment:
If the default is false, do we need to set it for every call?
##########
File path: sdks/java/io/contextual-text-io/src/main/java/org/apache/beam/sdk/io/contextualtextio/ContextualTextIOSource.java
##########
@@ -0,0 +1,363 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one
+ * or more contributor license agreements. See the NOTICE file
+ * distributed with this work for additional information
+ * regarding copyright ownership. The ASF licenses this file
+ * to you under the Apache License, Version 2.0 (the
+ * "License"); you may not use this file except in compliance
+ * with the License. You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+package org.apache.beam.sdk.io.contextualtextio;
+
+import java.io.IOException;
+import java.nio.ByteBuffer;
+import java.nio.channels.ReadableByteChannel;
+import java.nio.channels.SeekableByteChannel;
+import java.util.NoSuchElementException;
+import org.apache.beam.sdk.coders.Coder;
+import org.apache.beam.sdk.io.FileBasedSource;
+import org.apache.beam.sdk.io.fs.EmptyMatchTreatment;
+import org.apache.beam.sdk.io.fs.MatchResult;
+import org.apache.beam.sdk.options.PipelineOptions;
+import org.apache.beam.sdk.options.ValueProvider;
+import org.apache.beam.sdk.schemas.NoSuchSchemaException;
+import org.apache.beam.sdk.schemas.SchemaCoder;
+import org.apache.beam.sdk.schemas.SchemaRegistry;
+import org.apache.beam.vendor.grpc.v1p26p0.com.google.protobuf.ByteString;
+import org.apache.beam.vendor.guava.v26_0_jre.com.google.common.annotations.VisibleForTesting;
+import org.apache.beam.vendor.guava.v26_0_jre.com.google.common.base.Preconditions;
+import org.checkerframework.checker.nullness.qual.Nullable;
+
+/**
+ * Implementation detail of {@link ContextualTextIO.Read}.
+ *
+ * <p>A {@link FileBasedSource} which can decode records delimited by newline characters.
+ *
+ * <p>This source splits the data into records using {@code UTF-8} {@code \n}, {@code \r}, or {@code
+ * \r\n} as the delimiter. This source is not strict and supports decoding the last record even if
+ * it is not delimited. Finally, no records are decoded if the stream is empty.
+ *
+ * <p>This source supports reading from any arbitrary byte position within the stream. If the
+ * starting position is not {@code 0}, then bytes are skipped until the first delimiter is found
+ * representing the beginning of the first record to be decoded.
+ */
+@VisibleForTesting
+class ContextualTextIOSource extends FileBasedSource<RecordWithMetadata> {
+ byte[] delimiter;
+
+ // Used to Override isSplittable
+ private boolean hasMultilineCSVRecords;
+
+ @Override
+ protected boolean isSplittable() throws Exception {
+ if (hasMultilineCSVRecords) {
+ return false;
+ }
+ return super.isSplittable();
+ }
+
+ ContextualTextIOSource(
+ ValueProvider<String> fileSpec,
+ EmptyMatchTreatment emptyMatchTreatment,
+ byte[] delimiter,
+ boolean hasMultilineCSVRecords) {
+ super(fileSpec, emptyMatchTreatment, 1L);
+ this.delimiter = delimiter;
+ this.hasMultilineCSVRecords = hasMultilineCSVRecords;
+ }
+
+ private ContextualTextIOSource(
+ MatchResult.Metadata metadata,
+ long start,
+ long end,
+ byte[] delimiter,
+ boolean hasMultilineCSVRecords) {
+ super(metadata, 1L, start, end);
+ this.delimiter = delimiter;
+ this.hasMultilineCSVRecords = hasMultilineCSVRecords;
+ }
+
+ @Override
+ protected FileBasedSource<RecordWithMetadata> createForSubrangeOfFile(
+ MatchResult.Metadata metadata, long start, long end) {
+ return new ContextualTextIOSource(metadata, start, end, delimiter, hasMultilineCSVRecords);
+ }
+
+ @Override
+ protected FileBasedReader<RecordWithMetadata> createSingleFileReader(PipelineOptions options) {
+ return new MultiLineTextBasedReader(this, delimiter, hasMultilineCSVRecords);
+ }
+
+ @Override
+ public Coder<RecordWithMetadata> getOutputCoder() {
+ SchemaCoder<RecordWithMetadata> coder = null;
+ try {
+ coder = SchemaRegistry.createDefault().getSchemaCoder(RecordWithMetadata.class);
+ } catch (NoSuchSchemaException e) {
+ System.out.println("No Coder!");
+ }
+ return coder;
+ }
+
+ /**
+ * A {@link FileBasedReader FileBasedReader} which can decode records delimited by delimiter
+ * characters.
+ *
+ * <p>See {@link ContextualTextIOSource } for further details.
+ */
+ @VisibleForTesting
+ static class MultiLineTextBasedReader extends FileBasedReader<RecordWithMetadata> {
+ public static final int READ_BUFFER_SIZE = 8192;
+ private static final ByteString UTF8_BOM =
+ ByteString.copyFrom(new byte[] {(byte) 0xEF, (byte) 0xBB, (byte) 0xBF});
+ private final ByteBuffer readBuffer = ByteBuffer.allocate(READ_BUFFER_SIZE);
+ private ByteString buffer;
+ private int startOfDelimiterInBuffer;
+ private int endOfDelimiterInBuffer;
+ private long startOfRecord;
+ private volatile long startOfNextRecord;
+ private volatile boolean eof;
+ private volatile boolean elementIsPresent;
+ private @Nullable RecordWithMetadata currentValue;
+ private @Nullable ReadableByteChannel inChannel;
+ private byte @Nullable [] delimiter;
+
+ // Add to override the isSplittable
+ private boolean hasRFC4180MultiLineColumn;
+
+ private long startingOffset;
+ private long readerlineNum;
+
+ private MultiLineTextBasedReader(
+ ContextualTextIOSource source, byte[] delimiter, boolean hasRFC4180MultiLineColumn) {
+ super(source);
+ buffer = ByteString.EMPTY;
+ this.delimiter = delimiter;
+ this.hasRFC4180MultiLineColumn = hasRFC4180MultiLineColumn;
+ startingOffset = getCurrentSource().getStartOffset(); // Start offset;
+ }
+
+ @Override
+ protected long getCurrentOffset() throws NoSuchElementException {
+ if (!elementIsPresent) {
+ throw new NoSuchElementException();
+ }
+ return startOfRecord;
+ }
+
+ @Override
+ public long getSplitPointsRemaining() {
+ if (isStarted() && startOfNextRecord >= getCurrentSource().getEndOffset()) {
+ return isDone() ? 0 : 1;
+ }
+ return super.getSplitPointsRemaining();
+ }
+
+ @Override
+ public RecordWithMetadata getCurrent() throws NoSuchElementException {
+ if (!elementIsPresent) {
+ throw new NoSuchElementException();
+ }
+ return currentValue;
+ }
+
+ @Override
+ protected void startReading(ReadableByteChannel channel) throws IOException {
+ this.inChannel = channel;
+ // If the first offset is greater than zero, we need to skip bytes until we see our
+ // first delimiter.
+ long startOffset = getCurrentSource().getStartOffset();
+ if (startOffset > 0) {
+ Preconditions.checkState(
+ channel instanceof SeekableByteChannel,
+ "%s only supports reading from a SeekableByteChannel when given a start offset"
+ + " greater than 0.",
+ ContextualTextIOSource.class.getSimpleName());
+ long requiredPosition = startOffset - 1;
+ if (delimiter != null && startOffset >= delimiter.length) {
+ // we need to move back the offset of at worse delimiter.size to be sure to see
+ // all the bytes of the delimiter in the call to findDelimiterBounds() below
+ requiredPosition = startOffset - delimiter.length;
+ }
+ ((SeekableByteChannel) channel).position(requiredPosition);
+ findDelimiterBoundsWithMultiLineCheck();
+ buffer = buffer.substring(endOfDelimiterInBuffer);
+ startOfNextRecord = requiredPosition + endOfDelimiterInBuffer;
+ endOfDelimiterInBuffer = 0;
+ startOfDelimiterInBuffer = 0;
+ }
+ }
+
+ private void findDelimiterBoundsWithMultiLineCheck() throws IOException {
+ findDelimiterBounds();
+ }
+
+ /**
+ * Locates the start position and end position of the next delimiter. Will consume the channel
+ * till either EOF or the delimiter bounds are found.
+ *
+ * <p>If {@link ContextualTextIOSource#hasMultilineCSVRecords} is set then the behaviour will
+ * change from the standard read seen in {@link org.apache.beam.sdk.io.TextIO}. The assumption
+ * when {@link ContextualTextIOSource#hasMultilineCSVRecords} is set is that the file is being
+ * read with a single thread.
+ *
+ * <p>This fills the buffer and updates the positions as follows:
+ *
+ * <pre>{@code
+ * ------------------------------------------------------
+ * | element bytes | delimiter bytes | unconsumed bytes |
+ * ------------------------------------------------------
+ * 0 start of end of buffer
+ * delimiter delimiter size
+ * in buffer in buffer
+ * }</pre>
+ */
+ private void findDelimiterBounds() throws IOException {
+ int bytePositionInBuffer = 0;
+ boolean doubleQuoteClosed = true;
+
+ while (true) {
+ if (!tryToEnsureNumberOfBytesInBuffer(bytePositionInBuffer + 1)) {
+ startOfDelimiterInBuffer = endOfDelimiterInBuffer = bytePositionInBuffer;
+ break;
+ }
+
+ byte currentByte = buffer.byteAt(bytePositionInBuffer);
+ if (hasRFC4180MultiLineColumn) {
+ // Check if we are inside an open Quote
+ if (currentByte == '"') {
+ doubleQuoteClosed = !doubleQuoteClosed;
+ }
+ } else {
+ doubleQuoteClosed = true;
+ }
+
+ if (delimiter == null) {
+ // default delimiter
+ if (currentByte == '\n') {
+ startOfDelimiterInBuffer = bytePositionInBuffer;
+ endOfDelimiterInBuffer = startOfDelimiterInBuffer + 1;
+ if (doubleQuoteClosed) {
+ break;
+ }
+ } else if (currentByte == '\r') {
+ startOfDelimiterInBuffer = bytePositionInBuffer;
+ endOfDelimiterInBuffer = startOfDelimiterInBuffer + 1;
+ if (tryToEnsureNumberOfBytesInBuffer(bytePositionInBuffer + 2)) {
+ currentByte = buffer.byteAt(bytePositionInBuffer + 1);
+ if (currentByte == '\n') {
+ endOfDelimiterInBuffer += 1;
+ }
+ }
+ if (doubleQuoteClosed) {
+ break;
+ }
+ }
+ } else {
+ // when the user defines a delimiter
+ int i = 0;
+ startOfDelimiterInBuffer = endOfDelimiterInBuffer = bytePositionInBuffer;
+ while ((i < delimiter.length) && (currentByte == delimiter[i])) {
+ // read next byte;
+ i++;
+ if (tryToEnsureNumberOfBytesInBuffer(bytePositionInBuffer + i + 1)) {
+ currentByte = buffer.byteAt(bytePositionInBuffer + i);
+ } else {
+ // corner case: delimiter truncate at the end of file
+ startOfDelimiterInBuffer = endOfDelimiterInBuffer = bytePositionInBuffer;
+ break;
+ }
+ }
+ if (i == delimiter.length) {
+ endOfDelimiterInBuffer = bytePositionInBuffer + i;
+ if (doubleQuoteClosed) {
+ break;
+ }
+ }
+ }
+ bytePositionInBuffer += 1;
+ }
+ }
+
+ @Override
+ protected boolean readNextRecord() throws IOException {
+ startOfRecord = startOfNextRecord;
+
+ findDelimiterBoundsWithMultiLineCheck();
+
+ // If we have reached EOF file and consumed all of the buffer then we know
+ // that there are no more records.
+ if (eof && buffer.isEmpty()) {
+ elementIsPresent = false;
+ return false;
+ }
+
+ decodeCurrentElement();
+ startOfNextRecord = startOfRecord + endOfDelimiterInBuffer;
+ return true;
+ }
+
+ /**
+ * Decodes the current element updating the buffer to only contain the unconsumed bytes.
+ *
+ * <p>This invalidates the currently stored {@code startOfDelimiterInBuffer} and {@code
+ * endOfDelimiterInBuffer}.
+ */
+ private void decodeCurrentElement() throws IOException {
+ ByteString dataToDecode = buffer.substring(0, startOfDelimiterInBuffer);
+ // If present, the UTF8 Byte Order Mark (BOM) will be removed.
+ if (startOfRecord == 0 && dataToDecode.startsWith(UTF8_BOM)) {
+ dataToDecode = dataToDecode.substring(UTF8_BOM.size());
+ }
+
+ /////////////////////////////////////////////
+
+ // Data of the Current Line
+ // dataToDecode.toStringUtf8();
+
+ // The line num is:
+ Long lineUniqueLineNum = readerlineNum++;
+ // The Complete FileName (with uri if this is a web url eg: temp/abc.txt) is:
+ String fileName = getCurrentSource().getSingleFileMetadata().resourceId().toString();
Review comment:
@tvalentyn @abhiy13 what do you think of having a getUri field and a getFileName field, where the filename is just the filename,
My concern is that if we have a very long URI with a file that contains a couple of cols of ints, we are massively expanding the size of the output. In the future having a id + sideinput with the filenames would solve this issue better. But this would be a nice way to reduce this problem?
##########
File path: sdks/java/io/contextual-text-io/src/main/java/org/apache/beam/sdk/io/contextualtextio/ContextualTextIO.java
##########
@@ -0,0 +1,594 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one
+ * or more contributor license agreements. See the NOTICE file
+ * distributed with this work for additional information
+ * regarding copyright ownership. The ASF licenses this file
+ * to you under the Apache License, Version 2.0 (the
+ * "License"); you may not use this file except in compliance
+ * with the License. You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+package org.apache.beam.sdk.io.contextualtextio;
+
+import static org.apache.beam.sdk.io.FileIO.ReadMatches.DirectoryTreatment;
+import static org.apache.beam.vendor.guava.v26_0_jre.com.google.common.base.Preconditions.checkArgument;
+import static org.apache.beam.vendor.guava.v26_0_jre.com.google.common.base.Preconditions.checkNotNull;
+
+import com.google.auto.value.AutoValue;
+import java.nio.ByteBuffer;
+import java.util.Arrays;
+import java.util.Comparator;
+import java.util.HashMap;
+import java.util.Map;
+import java.util.SortedMap;
+import java.util.TreeMap;
+import org.apache.beam.sdk.coders.StringUtf8Coder;
+import org.apache.beam.sdk.io.CompressedSource;
+import org.apache.beam.sdk.io.Compression;
+import org.apache.beam.sdk.io.FileBasedSource;
+import org.apache.beam.sdk.io.FileIO;
+import org.apache.beam.sdk.io.FileIO.MatchConfiguration;
+import org.apache.beam.sdk.io.ReadAllViaFileBasedSource;
+import org.apache.beam.sdk.io.fs.EmptyMatchTreatment;
+import org.apache.beam.sdk.options.ValueProvider;
+import org.apache.beam.sdk.options.ValueProvider.StaticValueProvider;
+import org.apache.beam.sdk.schemas.NoSuchSchemaException;
+import org.apache.beam.sdk.schemas.SchemaCoder;
+import org.apache.beam.sdk.transforms.Count;
+import org.apache.beam.sdk.transforms.Create;
+import org.apache.beam.sdk.transforms.DoFn;
+import org.apache.beam.sdk.transforms.PTransform;
+import org.apache.beam.sdk.transforms.ParDo;
+import org.apache.beam.sdk.transforms.SerializableFunction;
+import org.apache.beam.sdk.transforms.View;
+import org.apache.beam.sdk.transforms.Watch.Growth.TerminationCondition;
+import org.apache.beam.sdk.transforms.display.DisplayData;
+import org.apache.beam.sdk.values.KV;
+import org.apache.beam.sdk.values.PBegin;
+import org.apache.beam.sdk.values.PCollection;
+import org.apache.beam.sdk.values.PCollectionView;
+import org.apache.beam.vendor.guava.v26_0_jre.com.google.common.annotations.VisibleForTesting;
+import org.checkerframework.checker.nullness.qual.Nullable;
+import org.joda.time.Duration;
+
+/**
+ * {@link PTransform}s that read text files and collect contextual information of the elements in
+ * the input.
+ *
+ * <h2>Reading from text files</h2>
+ *
+ * <p>To read a {@link PCollection} from one or more text files, use {@code
+ * ContextualTextIO.read()}. To instantiate a transform use {@link
+ * ContextualTextIO.Read#from(String)} and specify the path of the file(s) to be read.
+ * Alternatively, if the filenames to be read are themselves in a {@link PCollection} you can use
+ * {@link FileIO} to match them and {@link ContextualTextIO#readFiles()} to read them.
+ *
+ * <p>{@link #read} returns a {@link PCollection} of {@link RecordWithMetadata RecordWithMetadata},
+ * each corresponding to one line of an inout UTF-8 text file (split into lines delimited by '\n',
+ * '\r', '\r\n', or specified delimiter see {@link ContextualTextIO.Read#withDelimiter})
+ *
+ * <h3>Filepattern expansion and watching</h3>
+ *
+ * <p>By default, the filepatterns are expanded only once. The combination of {@link
+ * FileIO.Match#continuously(Duration, TerminationCondition)} and {@link #readFiles()} allow
+ * streaming of new files matching the filepattern(s).
+ *
+ * <p>By default, {@link #read} prohibits filepatterns that match no files, and {@link #readFiles()}
+ * allows them in case the filepattern contains a glob wildcard character. Use {@link
+ * ContextualTextIO.Read#withEmptyMatchTreatment} or {@link
+ * FileIO.Match#withEmptyMatchTreatment(EmptyMatchTreatment)} plus {@link #readFiles()} to configure
+ * this behavior.
+ *
+ * <p>Example 1: reading a file or filepattern.
+ *
+ * <pre>{@code
+ * Pipeline p = ...;
+ *
+ * // A simple Read of a file:
+ * PCollection<RecordWithMetadata> lines = p.apply(ContextualTextIO.read().from("/local/path/to/file.txt"));
+ * }</pre>
+ *
+ * <p>Example 2: reading a PCollection of filenames.
+ *
+ * <pre>{@code
+ * Pipeline p = ...;
+ *
+ * // E.g. the filenames might be computed from other data in the pipeline, or
+ * // read from a data source.
+ * PCollection<String> filenames = ...;
+ *
+ * // Read all files in the collection.
+ * PCollection<RecordWithMetadata> lines =
+ * filenames
+ * .apply(FileIO.matchAll())
+ * .apply(FileIO.readMatches())
+ * .apply(ContextualTextIO.readFiles());
+ * }</pre>
+ *
+ * <p>Example 3: streaming new files matching a filepattern.
+ *
+ * <pre>{@code
+ * Pipeline p = ...;
+ *
+ * PCollection<RecordWithMetadata> lines = p.apply(ContextualTextIO.read()
+ * .from("/local/path/to/files/*")
+ * .watchForNewFiles(
+ * // Check for new files every minute
+ * Duration.standardMinutes(1),
+ * // Stop watching the filepattern if no new files appear within an hour
+ * afterTimeSinceNewOutput(Duration.standardHours(1))));
+ * }</pre>
+ *
+ * <p>Example 4: reading a file or file pattern of RFC4180-compliant CSV files with fields that may
+ * contain line breaks.
+ *
+ * <pre>{@code
+ * Pipeline p = ...;
+ *
+ * PCollection<RecordWithMetadata> lines = p.apply(ContextualTextIO.read()
+ * .from("/local/path/to/files/*.csv")
+ * .withHasMultilineCSVRecords(true));
+ * }</pre>
+ *
+ * <p>Example 5: reading while watching for new files
+ *
+ * <pre>{@code
+ * Pipeline p = ...;
+ *
+ * PCollection<RecordWithMetadata> lines = p.apply(FileIO.match()
+ * .filepattern("filepattern")
+ * .continuously(
+ * Duration.millis(100),
+ * Watch.Growth.afterTimeSinceNewOutput(Duration.standardSeconds(3))))
+ * .apply(FileIO.readMatches())
+ * .apply(ContextualTextIO.readFiles());
+ * }</pre>
+ *
+ * NOTE: Using {@link ContextualTextIO.Read#withHasMultilineCSVRecords(Boolean)} introduces a
+ * performance penalty: when this option is enabled, the input cannot be split and read in parallel.
+ *
+ * <h3>Reading a very large number of files</h3>
+ *
+ * <p>If it is known that the filepattern will match a very large number of files (e.g. tens of
+ * thousands or more), use {@link ContextualTextIO.Read#withHintMatchesManyFiles} for better
+ * performance and scalability. Note that it may decrease performance if the filepattern matches
+ * only a small number of files.
+ */
+public class ContextualTextIO {
+ private static final long DEFAULT_BUNDLE_SIZE_BYTES = 64 * 1024 * 1024L;
+
+ /**
+ * A {@link PTransform} that reads from one or more text files and returns a bounded {@link
+ * PCollection} containing one {@link RecordWithMetadata}element for each line of the input files.
+ */
+ public static Read read() {
+ return new AutoValue_ContextualTextIO_Read.Builder()
+ .setCompression(Compression.AUTO)
+ .setHintMatchesManyFiles(false)
+ .setWithoutLineNumMetadata(false)
+ .setMatchConfiguration(MatchConfiguration.create(EmptyMatchTreatment.DISALLOW))
+ .setHasMultilineCSVRecords(false)
+ .build();
+ }
+
+ /**
+ * Like {@link #read}, but reads each file in a {@link PCollection} of {@link
+ * FileIO.ReadableFile}, returned by {@link FileIO#readMatches}.
+ */
+ public static ReadFiles readFiles() {
+ return new AutoValue_ContextualTextIO_ReadFiles.Builder()
+ // 64MB is a reasonable value that allows to amortize the cost of opening files,
+ // but is not so large as to exhaust a typical runner's maximum amount of output per
+ // ProcessElement call.
+ .setDesiredBundleSizeBytes(DEFAULT_BUNDLE_SIZE_BYTES)
+ .setHasMultilineCSVRecords(false)
+ .build();
+ }
+
+ /** Implementation of {@link #read}. */
+ @AutoValue
+ public abstract static class Read extends PTransform<PBegin, PCollection<RecordWithMetadata>> {
+ abstract @Nullable ValueProvider<String> getFilepattern();
+
+ abstract MatchConfiguration getMatchConfiguration();
+
+ abstract boolean getHintMatchesManyFiles();
+
+ abstract boolean getWithoutLineNumMetadata();
+
+ abstract Compression getCompression();
+
+ abstract @Nullable Boolean getHasMultilineCSVRecords();
+
+ @SuppressWarnings("mutable") // this returns an array that can be mutated by the caller
+ abstract byte @Nullable [] getDelimiter();
+
+ abstract Builder toBuilder();
+
+ @AutoValue.Builder
+ abstract static class Builder {
+ abstract Builder setFilepattern(ValueProvider<String> filepattern);
+
+ abstract Builder setMatchConfiguration(MatchConfiguration matchConfiguration);
+
+ abstract Builder setHintMatchesManyFiles(boolean hintManyFiles);
+
+ abstract Builder setWithoutLineNumMetadata(boolean withoutLineNumMetadata);
+
+ abstract Builder setCompression(Compression compression);
+
+ abstract Builder setDelimiter(byte @Nullable [] delimiter);
+
+ abstract Builder setHasMultilineCSVRecords(Boolean hasMultilineCSVRecords);
+
+ abstract Read build();
+ }
+
+ /**
+ * Reads text from the file(s) with the given filename or filename pattern.
+ *
+ * <p>This can be a local path (if running locally), or a Google Cloud Storage filename or
+ * filename pattern of the form {@code "gs://<bucket>/<filepath>"} (if running locally or using
+ * remote execution service).
+ *
+ * <p>Standard <a href="http://docs.oracle.com/javase/tutorial/essential/io/find.html" >Java
+ * Filesystem glob patterns</a> ("*", "?", "[..]") are supported.
+ *
+ * <p>If it is known that the filepattern will match a very large number of files (at least tens
+ * of thousands), use {@link #withHintMatchesManyFiles} for better performance and scalability.
+ */
+ public Read from(String filepattern) {
+ checkArgument(filepattern != null, "filepattern can not be null");
+ return from(StaticValueProvider.of(filepattern));
+ }
+
+ /** Same as {@code from(filepattern)}, but accepting a {@link ValueProvider}. */
+ public Read from(ValueProvider<String> filepattern) {
+ checkArgument(filepattern != null, "filepattern can not be null");
+ return toBuilder().setFilepattern(filepattern).build();
+ }
+
+ /** Sets the {@link MatchConfiguration}. */
+ public Read withMatchConfiguration(MatchConfiguration matchConfiguration) {
+ return toBuilder().setMatchConfiguration(matchConfiguration).build();
+ }
+
+ /**
+ * When reading RFC4180 CSV files that have values that span multiple lines, set this to true.
+ * Note: this reduces the read performance (see: {@link ContextualTextIO}).
+ */
+ public Read withHasMultilineCSVRecords(Boolean hasMultilineCSVRecords) {
+ return toBuilder().setHasMultilineCSVRecords(hasMultilineCSVRecords).build();
+ }
+
+ /**
+ * Reads from input sources using the specified compression type.
+ *
+ * <p>If no compression type is specified, the default is {@link Compression#AUTO}.
+ */
+ public Read withCompression(Compression compression) {
+ return toBuilder().setCompression(compression).build();
+ }
+
+ /**
+ * Hints that the filepattern specified in {@link #from(String)} matches a very large number of
+ * files.
+ *
+ * <p>This hint may cause a runner to execute the transform differently, in a way that improves
+ * performance for this case, but it may worsen performance if the filepattern matches only a
+ * small number of files (e.g., in a runner that supports dynamic work rebalancing, it will
+ * happen less efficiently within individual files).
+ */
+ public Read withHintMatchesManyFiles() {
+ return toBuilder().setHintMatchesManyFiles(true).build();
+ }
+
+ /**
+ * Allows the user the opt out of getting recordNums associated with each record.
+ *
+ * <p>ContextualTextIO uses a shuffle step to assemble the recordNums for each record which may
+ * result in some performance loss.
+ *
+ * <p>Use this when metadata like fileNames are required and their position/order can be
+ * ignored.
+ */
+ public Read withoutLineNumMetadata() {
+ return toBuilder().setWithoutLineNumMetadata(true).build();
+ }
+
+ /** See {@link MatchConfiguration#withEmptyMatchTreatment}. */
+ public Read withEmptyMatchTreatment(EmptyMatchTreatment treatment) {
+ return withMatchConfiguration(getMatchConfiguration().withEmptyMatchTreatment(treatment));
+ }
+
+ /** Set the custom delimiter to be used in place of the default ones ('\r', '\n' or '\r\n'). */
+ public Read withDelimiter(byte[] delimiter) {
+ checkArgument(delimiter != null, "delimiter can not be null");
+ checkArgument(!isSelfOverlapping(delimiter), "delimiter must not self-overlap");
+ return toBuilder().setDelimiter(delimiter).build();
+ }
+
+ static boolean isSelfOverlapping(byte[] s) {
+ // s self-overlaps if v exists such as s = vu = wv with u and w non empty
+ for (int i = 1; i < s.length - 1; ++i) {
+ if (ByteBuffer.wrap(s, 0, i).equals(ByteBuffer.wrap(s, s.length - i, i))) {
+ return true;
+ }
+ }
+ return false;
+ }
+
+ @Override
+ public PCollection<RecordWithMetadata> expand(PBegin input) {
+ checkNotNull(
+ getFilepattern(), "need to set the filepattern of a ContextualTextIO.Read transform");
+ PCollection<RecordWithMetadata> lines = null;
+ if (getMatchConfiguration().getWatchInterval() == null && !getHintMatchesManyFiles()) {
+ lines = input.apply("Read", org.apache.beam.sdk.io.Read.from(getSource()));
+ } else {
+ // All other cases go through FileIO + ReadFiles
+ lines =
+ input
+ .apply(
+ "Create filepattern", Create.ofProvider(getFilepattern(), StringUtf8Coder.of()))
+ .apply("Match All", FileIO.matchAll().withConfiguration(getMatchConfiguration()))
+ .apply(
+ "Read Matches",
+ FileIO.readMatches()
+ .withCompression(getCompression())
+ .withDirectoryTreatment(DirectoryTreatment.PROHIBIT))
+ .apply("Via ReadFiles", readFiles().withDelimiter(getDelimiter()));
+ }
+
+ // Check if the user decided to opt out of recordNums associated with records
+ if (getWithoutLineNumMetadata()) {
+ return lines;
+ }
+
+ // At this point the line number in RecordWithMetadata contains the relative line offset from
+ // the
+ // beginning of the read range.
+
+ // To compute the absolute position from the beginning of the input,
+ // we group the lines within the same ranges, and evaluate the size of each range.
+
+ PCollection<KV<KV<String, Long>, RecordWithMetadata>> linesGroupedByFileAndRange =
+ lines.apply("AddFileNameAndRange", ParDo.of(new AddFileNameAndRange()));
+
+ PCollectionView<Map<KV<String, Long>, Long>> rangeSizes =
+ linesGroupedByFileAndRange
+ .apply("CountLinesForEachFileRange", Count.perKey())
+ .apply("SizesAsView", View.asMap());
+
+ // Get Pipeline to create a dummy PCollection with one element to help compute the lines
+ // before each Range
+ PCollection<Integer> singletonPcoll =
+ input.getPipeline().apply("CreateSingletonPcoll", Create.of(Arrays.asList(1)));
+
+ // For each (File, Offset) pair, calculate the number of lines occurring before the Range for
+ // each File
+
+ // After computing the number of lines before each range, we can find the line number in
+ // original file as numLiesBeforeOffset + lineNumInCurrentOffset
+ PCollectionView<Map<KV<String, Long>, Long>> numLinesBeforeEachRange =
+ singletonPcoll
+ .apply(
+ "ComputeLinesBeforeRange",
+ ParDo.of(new ComputeLinesBeforeEachRange(rangeSizes)).withSideInputs(rangeSizes))
+ .apply("NumLinesBeforeEachRangeAsView", View.asMap());
+
+ return linesGroupedByFileAndRange.apply(
+ "AssignLineNums",
+ ParDo.of(new AssignLineNums(numLinesBeforeEachRange))
+ .withSideInputs(numLinesBeforeEachRange));
+ }
+
+ @VisibleForTesting
+ static class AddFileNameAndRange
+ extends DoFn<RecordWithMetadata, KV<KV<String, Long>, RecordWithMetadata>> {
+ @ProcessElement
+ public void processElement(
+ @Element RecordWithMetadata line,
+ OutputReceiver<KV<KV<String, Long>, RecordWithMetadata>> out) {
+ out.output(KV.of(KV.of(line.getFileName(), line.getRange().getRangeNum()), line));
+ }
+ }
+
+ /**
+ * Helper class for computing number of lines in the File preceding the beginning of the Range
+ * in this file.
+ */
+ @VisibleForTesting
+ static class ComputeLinesBeforeEachRange extends DoFn<Integer, KV<KV<String, Long>, Long>> {
+ private final PCollectionView<Map<KV<String, Long>, Long>> rangeSizes;
+
+ public ComputeLinesBeforeEachRange(PCollectionView<Map<KV<String, Long>, Long>> rangeSizes) {
+ this.rangeSizes = rangeSizes;
+ }
+
+ // Add custom comparator as KV<K, V> is not comparable by default
+ private static class FileRangeComparator<K extends Comparable<K>, V extends Comparable<V>>
+ implements Comparator<KV<K, V>> {
+ @Override
+ public int compare(KV<K, V> a, KV<K, V> b) {
+ if (a.getKey().compareTo(b.getKey()) == 0) {
+ return a.getValue().compareTo(b.getValue());
+ }
+ return a.getKey().compareTo(b.getKey());
+ }
+ }
+
+ @ProcessElement
+ public void processElement(ProcessContext p) {
+ // Get the Map Containing the size from side-input
+ Map<KV<String, Long>, Long> rangeSizesMap = p.sideInput(rangeSizes);
+
+ // The FileRange Pair must be sorted
+ SortedMap<KV<String, Long>, Long> sorted = new TreeMap<>(new FileRangeComparator<>());
+
+ // Initialize sorted map with values
+ for (Map.Entry<KV<String, Long>, Long> entry : rangeSizesMap.entrySet()) {
+ sorted.put(entry.getKey(), entry.getValue());
+ }
+
+ // HashMap that tracks lines passed for each file
+ Map<String, Long> pastLines = new HashMap<>();
+
+ // For each (File, Range) Pair, compute the number of lines before it
+ for (Map.Entry entry : sorted.entrySet()) {
+ Long lines = (long) entry.getValue();
+ KV<String, Long> fileRange = (KV<String, Long>) entry.getKey();
+ String file = fileRange.getKey();
+ Long linesBefore = 0L;
+ if (pastLines.containsKey(file)) {
+ linesBefore = pastLines.get(file);
+ }
+ p.output(KV.of(fileRange, linesBefore));
+ pastLines.put(file, linesBefore + lines);
+ }
+ }
+ }
+
+ static class AssignLineNums
+ extends DoFn<KV<KV<String, Long>, RecordWithMetadata>, RecordWithMetadata> {
+ PCollectionView<Map<KV<String, Long>, Long>> numLinesBeforeEachRange;
+
+ public AssignLineNums(PCollectionView<Map<KV<String, Long>, Long>> numLinesBeforeEachRange) {
+ this.numLinesBeforeEachRange = numLinesBeforeEachRange;
+ }
+
+ @ProcessElement
+ public void processElement(ProcessContext p) {
+ Long range = p.element().getKey().getValue();
+ String file = p.element().getKey().getKey();
+ RecordWithMetadata line = p.element().getValue();
+ Long linesLessThanThisRange = p.sideInput(numLinesBeforeEachRange).get(KV.of(file, range));
+ RecordWithMetadata newLine =
+ RecordWithMetadata.newBuilder()
+ .setRecordValue(line.getRecordValue())
+ .setRecordNum(line.getRange().getRangeLineNum() + linesLessThanThisRange)
+ .setFileName(line.getFileName())
+ .setRange(line.getRange())
+ .build();
+ p.output(newLine);
+ }
+ }
+
+ // Helper to create a source specific to the requested compression type.
+ protected FileBasedSource<RecordWithMetadata> getSource() {
+ return CompressedSource.from(
+ new ContextualTextIOSource(
+ getFilepattern(),
+ getMatchConfiguration().getEmptyMatchTreatment(),
+ getDelimiter(),
+ getHasMultilineCSVRecords()))
+ .withCompression(getCompression());
+ }
+
+ @Override
+ public void populateDisplayData(DisplayData.Builder builder) {
+ super.populateDisplayData(builder);
+ builder
+ .add(
+ DisplayData.item("compressionType", getCompression().toString())
+ .withLabel("Compression Type"))
+ .addIfNotNull(DisplayData.item("filePattern", getFilepattern()).withLabel("File Pattern"))
+ .include("matchConfiguration", getMatchConfiguration())
+ .addIfNotNull(
+ DisplayData.item("delimiter", Arrays.toString(getDelimiter()))
+ .withLabel("Custom delimiter to split records"))
+ .addIfNotNull(
+ DisplayData.item("hasMultilineCSVRecords", getHasMultilineCSVRecords())
+ .withLabel("Has RFC4180 MultiLineCSV Records"));
+ }
+ }
+
+ /** Implementation of {@link #readFiles}. */
+ @AutoValue
+ public abstract static class ReadFiles
+ extends PTransform<PCollection<FileIO.ReadableFile>, PCollection<RecordWithMetadata>> {
+ abstract long getDesiredBundleSizeBytes();
+
+ @SuppressWarnings("mutable") // this returns an array that can be mutated by the caller
+ abstract byte @Nullable [] getDelimiter();
+
+ abstract boolean getHasMultilineCSVRecords();
+
+ abstract Builder toBuilder();
+
+ @AutoValue.Builder
+ abstract static class Builder {
+ abstract Builder setDesiredBundleSizeBytes(long desiredBundleSizeBytes);
+
+ abstract Builder setHasMultilineCSVRecords(boolean hasMultilineCSVRecords);
+
+ abstract Builder setDelimiter(byte @Nullable [] delimiter);
+
+ abstract ReadFiles build();
+ }
+
+ @VisibleForTesting
+ ReadFiles withDesiredBundleSizeBytes(long desiredBundleSizeBytes) {
+ return toBuilder().setDesiredBundleSizeBytes(desiredBundleSizeBytes).build();
+ }
+
+ /** Like {@link Read#withDelimiter}. */
+ public ReadFiles withDelimiter(byte[] delimiter) {
+ return toBuilder().setDelimiter(delimiter).build();
+ }
+
+ @Override
+ public PCollection<RecordWithMetadata> expand(PCollection<FileIO.ReadableFile> input) {
+ SchemaCoder<RecordWithMetadata> coder = null;
+ try {
+ coder = input.getPipeline().getSchemaRegistry().getSchemaCoder(RecordWithMetadata.class);
+ } catch (NoSuchSchemaException e) {
+ System.out.println("No Coder!");
Review comment:
Need to use LOGGER
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
[GitHub] [beam] tvalentyn commented on a change in pull request #12645: [BEAM-10124] Add ContextualTextIO
Posted by GitBox <gi...@apache.org>.
tvalentyn commented on a change in pull request #12645:
URL: https://github.com/apache/beam/pull/12645#discussion_r477476918
##########
File path: sdks/java/io/contextual-text-io/src/main/java/org/apache/beam/sdk/io/contextualtextio/RecordWithMetadata.java
##########
@@ -0,0 +1,84 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one
+ * or more contributor license agreements. See the NOTICE file
+ * distributed with this work for additional information
+ * regarding copyright ownership. The ASF licenses this file
+ * to you under the Apache License, Version 2.0 (the
+ * "License"); you may not use this file except in compliance
+ * with the License. You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+package org.apache.beam.sdk.io.contextualtextio;
+
+import com.google.auto.value.AutoValue;
+import org.apache.beam.sdk.annotations.Experimental;
+import org.apache.beam.sdk.schemas.AutoValueSchema;
+import org.apache.beam.sdk.schemas.annotations.DefaultSchema;
+
+/**
+ * Helper Class based on {@link AutoValueSchema}, it provides Metadata associated with each Record
+ * when reading from file(s) using {@link ContextualTextIO}.
+ *
+ * <h3>Fields:</h3>
+ *
+ * <ul>
+ * <li>recordOffset: The offset of a record (the byte at which the record begins) in a file. This
+ * information can be useful if you wish to reconstruct the file. {@link
+ * RecordWithMetadata#getRecordOffset()}
+ * <li>recordNum: The ordinal number of the record in its file. {@link
+ * RecordWithMetadata#getRecordNum()}
+ * <li>recordValue: The value / contents of the record {@link RecordWithMetadata#getRecordValue()}
+ * <li>rangeOffset: The starting offset of the range (split), which contained the record, when the
+ * record was read. {@link RecordWithMetadata#getRangeOffset()}
+ * <li>recordNumInOffset: The record number relative to the Range. (line number within the range)
+ * {@link RecordWithMetadata#getRecordNumInOffset()}
+ * <li>fileName: Name of the file to which the record belongs (this is the full filename,
+ * eg:path/to/file.txt) {@link RecordWithMetadata#getFileName()}
+ * </ul>
+ */
+@Experimental(Experimental.Kind.SCHEMAS)
+@DefaultSchema(AutoValueSchema.class)
+@AutoValue
+public abstract class RecordWithMetadata {
Review comment:
Not sure I follow this comment - do you suggest to consider a different name than `RecordWithMetadata`?
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
[GitHub] [beam] tvalentyn commented on a change in pull request #12645: [BEAM-10124] Add ContextualTextIO
Posted by GitBox <gi...@apache.org>.
tvalentyn commented on a change in pull request #12645:
URL: https://github.com/apache/beam/pull/12645#discussion_r474411845
##########
File path: sdks/java/io/contextual-text-io/src/main/java/org/apache/beam/sdk/io/contextualtextio/ContextualTextIO.java
##########
@@ -151,8 +157,27 @@
* .apply(ContextualTextIO.readFiles());
* }</pre>
*
- * NOTE: Using {@link ContextualTextIO.Read#withHasMultilineCSVRecords(Boolean)} introduces a
- * performance penalty: when this option is enabled, the input cannot be split and read in parallel.
+ * <p>Example 6: reading without recordNum metadata, or only fileName associated Metadata. (the
+ * Objects would still contain recordNums, but these recordNums would correspond to their positions
+ * in their respective offsets rather than their positions within the entire file).
+ *
+ * <pre>{@code
+ * Pipeline p = ...;
+ *
+ * PCollection<RecordWithMetadata> records = p.apply(ContextualTextIO.read()
+ * .from("/local/path/to/files/*.csv")
+ * .setWithoutRecordNumMetadata(true));
+ * }</pre>
+ *
+ * <p>NOTE: When using {@link ContextualTextIO.Read#withHasMultilineCSVRecords(Boolean)} this
+ * option, a single reader will be used to process the file, rather than multiple readers which can
+ * read from different offsets. For a large file this can result in lower performance.
+ *
+ * <p>NOTE: Use {@link Read#withoutRecordNumMetadata()} when recordNum metadata is not required or
+ * when only metadata associated with filenames is required, Not using this option introduces a
+ * shuffle step which increases the resources used by the pipeline. <b>This option is set to false
+ * by default. Meaning that the shuffle step will be performed, set it to false to avoid the shuffle
Review comment:
s/false/true
how about:
```
Use {@link Read#withoutRecordNumMetadata()} when recordNum metadata is not required, for example, when when only filename metadata is required. Computing record positions currently introduces a shuffle step, which increases the resources used by the pipeline. <b> By default withoutRecordNumMetadata is set to false, so the shuffle step is performed.</b>
```
##########
File path: sdks/java/io/contextual-text-io/src/main/java/org/apache/beam/sdk/io/contextualtextio/RecordWithMetadata.java
##########
@@ -19,21 +19,45 @@
import com.google.auto.value.AutoValue;
import org.apache.beam.sdk.annotations.Experimental;
-import org.apache.beam.sdk.annotations.Internal;
import org.apache.beam.sdk.schemas.AutoValueSchema;
import org.apache.beam.sdk.schemas.annotations.DefaultSchema;
-@Internal
+/**
+ * Helper Class based on {@link AutoValueSchema}, it provides Metadata associated with each Record
+ * when reading from file(s) using {@link ContextualTextIO}.
+ *
+ * <h3>Fields:</h3>
+ *
+ * <ul>
+ * <li>recordOffset: The offset of a record (the byte at which the record begins) in a file. This
+ * information can be useful if you wish to reconstruct the file. {@link
+ * RecordWithMetadata#getRecordOffset()}
+ * <li>recordNum: The record number of the record from its ordinal file. {@link
Review comment:
The ordinal number of the record in its file.
##########
File path: sdks/java/io/contextual-text-io/src/main/java/org/apache/beam/sdk/io/contextualtextio/RecordWithMetadata.java
##########
@@ -19,21 +19,45 @@
import com.google.auto.value.AutoValue;
import org.apache.beam.sdk.annotations.Experimental;
-import org.apache.beam.sdk.annotations.Internal;
import org.apache.beam.sdk.schemas.AutoValueSchema;
import org.apache.beam.sdk.schemas.annotations.DefaultSchema;
-@Internal
+/**
+ * Helper Class based on {@link AutoValueSchema}, it provides Metadata associated with each Record
+ * when reading from file(s) using {@link ContextualTextIO}.
+ *
+ * <h3>Fields:</h3>
+ *
+ * <ul>
+ * <li>recordOffset: The offset of a record (the byte at which the record begins) in a file. This
+ * information can be useful if you wish to reconstruct the file. {@link
+ * RecordWithMetadata#getRecordOffset()}
+ * <li>recordNum: The record number of the record from its ordinal file. {@link
+ * RecordWithMetadata#getRecordNum()}
+ * <li>recordValue: The value / contents of the records {@link
Review comment:
of the record
##########
File path: sdks/java/io/contextual-text-io/src/main/java/org/apache/beam/sdk/io/contextualtextio/RecordWithMetadata.java
##########
@@ -19,21 +19,45 @@
import com.google.auto.value.AutoValue;
import org.apache.beam.sdk.annotations.Experimental;
-import org.apache.beam.sdk.annotations.Internal;
import org.apache.beam.sdk.schemas.AutoValueSchema;
import org.apache.beam.sdk.schemas.annotations.DefaultSchema;
-@Internal
+/**
+ * Helper Class based on {@link AutoValueSchema}, it provides Metadata associated with each Record
+ * when reading from file(s) using {@link ContextualTextIO}.
+ *
+ * <h3>Fields:</h3>
+ *
+ * <ul>
+ * <li>recordOffset: The offset of a record (the byte at which the record begins) in a file. This
+ * information can be useful if you wish to reconstruct the file. {@link
+ * RecordWithMetadata#getRecordOffset()}
+ * <li>recordNum: The record number of the record from its ordinal file. {@link
+ * RecordWithMetadata#getRecordNum()}
+ * <li>recordValue: The value / contents of the records {@link
+ * RecordWithMetadata#getRecordValue()}
+ * <li>startingOffset: The starting offset of the range to which the record belongs. {@link
Review comment:
this name may be confusing with recordOffset.
Consider:
`rangeOffset`
`recordNumInRange`
`The starting offset of the range (split), which contained the record, when the record was read.`
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
[GitHub] [beam] rezarokni commented on a change in pull request #12645: [BEAM-10124] Add ContextualTextIO
Posted by GitBox <gi...@apache.org>.
rezarokni commented on a change in pull request #12645:
URL: https://github.com/apache/beam/pull/12645#discussion_r486156062
##########
File path: sdks/java/io/contextual-text-io/src/main/java/org/apache/beam/sdk/io/contextualtextio/ContextualTextIOSource.java
##########
@@ -0,0 +1,364 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one
+ * or more contributor license agreements. See the NOTICE file
+ * distributed with this work for additional information
+ * regarding copyright ownership. The ASF licenses this file
+ * to you under the Apache License, Version 2.0 (the
+ * "License"); you may not use this file except in compliance
+ * with the License. You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+package org.apache.beam.sdk.io.contextualtextio;
+
+import java.io.IOException;
+import java.nio.ByteBuffer;
+import java.nio.channels.ReadableByteChannel;
+import java.nio.channels.SeekableByteChannel;
+import java.util.NoSuchElementException;
+import org.apache.beam.sdk.coders.Coder;
+import org.apache.beam.sdk.io.FileBasedSource;
+import org.apache.beam.sdk.io.fs.EmptyMatchTreatment;
+import org.apache.beam.sdk.io.fs.MatchResult;
+import org.apache.beam.sdk.options.PipelineOptions;
+import org.apache.beam.sdk.options.ValueProvider;
+import org.apache.beam.sdk.schemas.NoSuchSchemaException;
+import org.apache.beam.sdk.schemas.SchemaCoder;
+import org.apache.beam.sdk.schemas.SchemaRegistry;
+import org.apache.beam.vendor.grpc.v1p26p0.com.google.protobuf.ByteString;
+import org.apache.beam.vendor.guava.v26_0_jre.com.google.common.annotations.VisibleForTesting;
+import org.apache.beam.vendor.guava.v26_0_jre.com.google.common.base.Preconditions;
+import org.checkerframework.checker.nullness.qual.Nullable;
+import org.slf4j.Logger;
+import org.slf4j.LoggerFactory;
+
+/**
+ * Implementation detail of {@link ContextualTextIO.Read}.
+ *
+ * <p>A {@link FileBasedSource} which can decode records delimited by newline characters.
+ *
+ * <p>This source splits the data into records using {@code UTF-8} {@code \n}, {@code \r}, or {@code
+ * \r\n} as the delimiter. This source is not strict and supports decoding the last record even if
+ * it is not delimited. Finally, no records are decoded if the stream is empty.
+ *
+ * <p>This source supports reading from any arbitrary byte position within the stream. If the
+ * starting position is not {@code 0}, then bytes are skipped until the first delimiter is found
+ * representing the beginning of the first record to be decoded.
+ */
+@VisibleForTesting
+class ContextualTextIOSource extends FileBasedSource<RecordWithMetadata> {
+ byte[] delimiter;
+
+ private static final Logger LOG = LoggerFactory.getLogger(ContextualTextIOSource.class);
+
+ // Used to Override isSplittable
+ private boolean hasMultilineCSVRecords;
+
+ @Override
+ protected boolean isSplittable() throws Exception {
+ if (hasMultilineCSVRecords) {
+ // When Having Multiline CSV Records,
+ // Splitting the file may cause a split to be within a record,
+ // Disabling split prevents this from happening
+ return false;
+ }
+ return super.isSplittable();
+ }
+
+ ContextualTextIOSource(
+ ValueProvider<String> fileSpec,
+ EmptyMatchTreatment emptyMatchTreatment,
+ byte[] delimiter,
+ boolean hasMultilineCSVRecords) {
+ super(fileSpec, emptyMatchTreatment, 1L);
+ this.delimiter = delimiter;
+ this.hasMultilineCSVRecords = hasMultilineCSVRecords;
+ }
+
+ private ContextualTextIOSource(
+ MatchResult.Metadata metadata,
+ long start,
+ long end,
+ byte[] delimiter,
+ boolean hasMultilineCSVRecords) {
+ super(metadata, 1L, start, end);
+ this.delimiter = delimiter;
+ this.hasMultilineCSVRecords = hasMultilineCSVRecords;
+ }
+
+ @Override
+ protected FileBasedSource<RecordWithMetadata> createForSubrangeOfFile(
+ MatchResult.Metadata metadata, long start, long end) {
+ return new ContextualTextIOSource(metadata, start, end, delimiter, hasMultilineCSVRecords);
+ }
+
+ @Override
+ protected FileBasedReader<RecordWithMetadata> createSingleFileReader(PipelineOptions options) {
+ return new MultiLineTextBasedReader(this, delimiter, hasMultilineCSVRecords);
+ }
+
+ @Override
+ public Coder<RecordWithMetadata> getOutputCoder() {
Review comment:
@abhiy13 short javadoc please.
##########
File path: sdks/java/io/contextual-text-io/src/main/java/org/apache/beam/sdk/io/contextualtextio/RecordWithMetadata.java
##########
@@ -0,0 +1,85 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one
+ * or more contributor license agreements. See the NOTICE file
+ * distributed with this work for additional information
+ * regarding copyright ownership. The ASF licenses this file
+ * to you under the Apache License, Version 2.0 (the
+ * "License"); you may not use this file except in compliance
+ * with the License. You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+package org.apache.beam.sdk.io.contextualtextio;
+
+import com.google.auto.value.AutoValue;
+import org.apache.beam.sdk.annotations.Experimental;
+import org.apache.beam.sdk.io.fs.ResourceId;
+import org.apache.beam.sdk.schemas.AutoValueSchema;
+import org.apache.beam.sdk.schemas.annotations.DefaultSchema;
+
+/**
+ * Helper Class based on {@link AutoValueSchema}, it provides Metadata associated with each Record
+ * when reading from file(s) using {@link ContextualTextIO}.
+ *
+ * <h3>Fields:</h3>
+ *
+ * <ul>
+ * <li>recordOffset: The offset of a record (the byte at which the record begins) in a file. This
+ * information can be useful if you wish to reconstruct the file. {@link
+ * RecordWithMetadata#getRecordOffset()}
+ * <li>recordNum: The ordinal number of the record in its file. {@link
+ * RecordWithMetadata#getRecordNum()}
+ * <li>recordValue: The value / contents of the record {@link RecordWithMetadata#getValue()}
+ * <li>rangeOffset: The starting offset of the range (split), which contained the record, when the
+ * record was read. {@link RecordWithMetadata#getRangeOffset()}
+ * <li>recordNumInOffset: The record number relative to the Range. (line number within the range)
+ * {@link RecordWithMetadata#getRecordNumInOffset()}
+ * <li>fileName: Name of the file to which the record belongs (this is the full filename,
+ * eg:path/to/file.txt) {@link RecordWithMetadata#getFileName()}
+ * </ul>
+ */
+@Experimental(Experimental.Kind.SCHEMAS)
+@DefaultSchema(AutoValueSchema.class)
+@AutoValue
+public abstract class RecordWithMetadata {
+ public abstract long getRecordOffset();
Review comment:
@abhiy13 javadoc for public items. thanx
##########
File path: sdks/java/io/contextual-text-io/src/main/java/org/apache/beam/sdk/io/contextualtextio/ContextualTextIOSource.java
##########
@@ -152,18 +188,27 @@ protected void startReading(ReadableByteChannel channel) throws IOException {
requiredPosition = startOffset - delimiter.length;
}
((SeekableByteChannel) channel).position(requiredPosition);
- findDelimiterBounds();
+ findDelimiterBoundsWithMultiLineCheck();
buffer = buffer.substring(endOfDelimiterInBuffer);
startOfNextRecord = requiredPosition + endOfDelimiterInBuffer;
endOfDelimiterInBuffer = 0;
startOfDelimiterInBuffer = 0;
}
}
+ private void findDelimiterBoundsWithMultiLineCheck() throws IOException {
Review comment:
@abhiy13 I think this no longer needs to be used as the checks have now been moved?
##########
File path: sdks/java/io/contextual-text-io/src/main/java/org/apache/beam/sdk/io/contextualtextio/RecordWithMetadata.java
##########
@@ -0,0 +1,84 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one
+ * or more contributor license agreements. See the NOTICE file
+ * distributed with this work for additional information
+ * regarding copyright ownership. The ASF licenses this file
+ * to you under the Apache License, Version 2.0 (the
+ * "License"); you may not use this file except in compliance
+ * with the License. You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+package org.apache.beam.sdk.io.contextualtextio;
+
+import com.google.auto.value.AutoValue;
+import org.apache.beam.sdk.annotations.Experimental;
+import org.apache.beam.sdk.schemas.AutoValueSchema;
+import org.apache.beam.sdk.schemas.annotations.DefaultSchema;
+
+/**
+ * Helper Class based on {@link AutoValueSchema}, it provides Metadata associated with each Record
+ * when reading from file(s) using {@link ContextualTextIO}.
+ *
+ * <h3>Fields:</h3>
+ *
+ * <ul>
+ * <li>recordOffset: The offset of a record (the byte at which the record begins) in a file. This
+ * information can be useful if you wish to reconstruct the file. {@link
+ * RecordWithMetadata#getRecordOffset()}
+ * <li>recordNum: The ordinal number of the record in its file. {@link
+ * RecordWithMetadata#getRecordNum()}
+ * <li>recordValue: The value / contents of the record {@link RecordWithMetadata#getRecordValue()}
+ * <li>rangeOffset: The starting offset of the range (split), which contained the record, when the
+ * record was read. {@link RecordWithMetadata#getRangeOffset()}
+ * <li>recordNumInOffset: The record number relative to the Range. (line number within the range)
+ * {@link RecordWithMetadata#getRecordNumInOffset()}
+ * <li>fileName: Name of the file to which the record belongs (this is the full filename,
+ * eg:path/to/file.txt) {@link RecordWithMetadata#getFileName()}
+ * </ul>
+ */
+@Experimental(Experimental.Kind.SCHEMAS)
+@DefaultSchema(AutoValueSchema.class)
+@AutoValue
+public abstract class RecordWithMetadata {
+ public abstract Long getRecordOffset();
+
+ public abstract Long getRecordNum();
+
+ public abstract String getRecordValue();
+
+ public abstract Long getRangeOffset();
+
+ public abstract Long getRecordNumInOffset();
+
+ public abstract Builder toBuilder();
+
+ public abstract String getFileName();
Review comment:
@abhiy13 similar to above, lets pull resource id through.
##########
File path: sdks/java/io/contextual-text-io/src/main/java/org/apache/beam/sdk/io/contextualtextio/ContextualTextIOSource.java
##########
@@ -0,0 +1,364 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one
+ * or more contributor license agreements. See the NOTICE file
+ * distributed with this work for additional information
+ * regarding copyright ownership. The ASF licenses this file
+ * to you under the Apache License, Version 2.0 (the
+ * "License"); you may not use this file except in compliance
+ * with the License. You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+package org.apache.beam.sdk.io.contextualtextio;
+
+import java.io.IOException;
+import java.nio.ByteBuffer;
+import java.nio.channels.ReadableByteChannel;
+import java.nio.channels.SeekableByteChannel;
+import java.util.NoSuchElementException;
+import org.apache.beam.sdk.coders.Coder;
+import org.apache.beam.sdk.io.FileBasedSource;
+import org.apache.beam.sdk.io.fs.EmptyMatchTreatment;
+import org.apache.beam.sdk.io.fs.MatchResult;
+import org.apache.beam.sdk.options.PipelineOptions;
+import org.apache.beam.sdk.options.ValueProvider;
+import org.apache.beam.sdk.schemas.NoSuchSchemaException;
+import org.apache.beam.sdk.schemas.SchemaCoder;
+import org.apache.beam.sdk.schemas.SchemaRegistry;
+import org.apache.beam.vendor.grpc.v1p26p0.com.google.protobuf.ByteString;
+import org.apache.beam.vendor.guava.v26_0_jre.com.google.common.annotations.VisibleForTesting;
+import org.apache.beam.vendor.guava.v26_0_jre.com.google.common.base.Preconditions;
+import org.checkerframework.checker.nullness.qual.Nullable;
+import org.slf4j.Logger;
+import org.slf4j.LoggerFactory;
+
+/**
+ * Implementation detail of {@link ContextualTextIO.Read}.
+ *
+ * <p>A {@link FileBasedSource} which can decode records delimited by newline characters.
+ *
+ * <p>This source splits the data into records using {@code UTF-8} {@code \n}, {@code \r}, or {@code
+ * \r\n} as the delimiter. This source is not strict and supports decoding the last record even if
+ * it is not delimited. Finally, no records are decoded if the stream is empty.
+ *
+ * <p>This source supports reading from any arbitrary byte position within the stream. If the
+ * starting position is not {@code 0}, then bytes are skipped until the first delimiter is found
+ * representing the beginning of the first record to be decoded.
+ */
+@VisibleForTesting
+class ContextualTextIOSource extends FileBasedSource<RecordWithMetadata> {
+ byte[] delimiter;
+
+ private static final Logger LOG = LoggerFactory.getLogger(ContextualTextIOSource.class);
+
+ // Used to Override isSplittable
+ private boolean hasMultilineCSVRecords;
+
+ @Override
+ protected boolean isSplittable() throws Exception {
+ if (hasMultilineCSVRecords) {
+ // When Having Multiline CSV Records,
+ // Splitting the file may cause a split to be within a record,
+ // Disabling split prevents this from happening
+ return false;
+ }
+ return super.isSplittable();
+ }
+
+ ContextualTextIOSource(
+ ValueProvider<String> fileSpec,
+ EmptyMatchTreatment emptyMatchTreatment,
+ byte[] delimiter,
+ boolean hasMultilineCSVRecords) {
+ super(fileSpec, emptyMatchTreatment, 1L);
+ this.delimiter = delimiter;
+ this.hasMultilineCSVRecords = hasMultilineCSVRecords;
+ }
+
+ private ContextualTextIOSource(
+ MatchResult.Metadata metadata,
+ long start,
+ long end,
+ byte[] delimiter,
+ boolean hasMultilineCSVRecords) {
+ super(metadata, 1L, start, end);
+ this.delimiter = delimiter;
+ this.hasMultilineCSVRecords = hasMultilineCSVRecords;
+ }
+
+ @Override
+ protected FileBasedSource<RecordWithMetadata> createForSubrangeOfFile(
+ MatchResult.Metadata metadata, long start, long end) {
+ return new ContextualTextIOSource(metadata, start, end, delimiter, hasMultilineCSVRecords);
+ }
+
+ @Override
+ protected FileBasedReader<RecordWithMetadata> createSingleFileReader(PipelineOptions options) {
+ return new MultiLineTextBasedReader(this, delimiter, hasMultilineCSVRecords);
+ }
+
+ @Override
+ public Coder<RecordWithMetadata> getOutputCoder() {
+ SchemaCoder<RecordWithMetadata> coder = null;
+ try {
+ coder = SchemaRegistry.createDefault().getSchemaCoder(RecordWithMetadata.class);
+ } catch (NoSuchSchemaException e) {
+ LOG.error("No Coder Found for RecordWithMetadata");
+ }
+ return coder;
+ }
+
+ /**
+ * A {@link FileBasedReader FileBasedReader} which can decode records delimited by delimiter
+ * characters.
+ *
+ * <p>See {@link ContextualTextIOSource } for further details.
+ */
+ @VisibleForTesting
+ static class MultiLineTextBasedReader extends FileBasedReader<RecordWithMetadata> {
+ public static final int READ_BUFFER_SIZE = 8192;
+ private static final ByteString UTF8_BOM =
+ ByteString.copyFrom(new byte[] {(byte) 0xEF, (byte) 0xBB, (byte) 0xBF});
+ private final ByteBuffer readBuffer = ByteBuffer.allocate(READ_BUFFER_SIZE);
+ private ByteString buffer;
+ private int startOfDelimiterInBuffer;
+ private int endOfDelimiterInBuffer;
+ private long startOfRecord;
+ private volatile long startOfNextRecord;
+ private volatile boolean eof;
+ private volatile boolean elementIsPresent;
+ private @Nullable RecordWithMetadata currentValue;
+ private @Nullable ReadableByteChannel inChannel;
+ private byte @Nullable [] delimiter;
+
+ // Add to override the isSplittable
+ private boolean hasMultilineCSVRecords;
+
+ private long startingOffset;
+ private long totalRecordCount;
+
+ private MultiLineTextBasedReader(
+ ContextualTextIOSource source, byte[] delimiter, boolean hasMultilineCSVRecords) {
+ super(source);
+ buffer = ByteString.EMPTY;
+ this.delimiter = delimiter;
+ this.hasMultilineCSVRecords = hasMultilineCSVRecords;
+ startingOffset = getCurrentSource().getStartOffset(); // Start offset;
+ }
+
+ @Override
+ protected long getCurrentOffset() throws NoSuchElementException {
+ if (!elementIsPresent) {
+ throw new NoSuchElementException();
+ }
+ return startOfRecord;
+ }
+
+ @Override
+ public long getSplitPointsRemaining() {
+ if (isStarted() && startOfNextRecord >= getCurrentSource().getEndOffset()) {
+ return isDone() ? 0 : 1;
+ }
+ return super.getSplitPointsRemaining();
+ }
+
+ @Override
+ public RecordWithMetadata getCurrent() throws NoSuchElementException {
+ if (!elementIsPresent) {
+ throw new NoSuchElementException();
+ }
+ return currentValue;
+ }
+
+ @Override
+ protected void startReading(ReadableByteChannel channel) throws IOException {
+ this.inChannel = channel;
+ // If the first offset is greater than zero, we need to skip bytes until we see our
+ // first delimiter.
+ long startOffset = getCurrentSource().getStartOffset();
+ if (startOffset > 0) {
+ Preconditions.checkState(
+ channel instanceof SeekableByteChannel,
+ "%s only supports reading from a SeekableByteChannel when given a start offset"
+ + " greater than 0.",
+ ContextualTextIOSource.class.getSimpleName());
+ long requiredPosition = startOffset - 1;
+ if (delimiter != null && startOffset >= delimiter.length) {
+ // we need to move back the offset of at worse delimiter.size to be sure to see
+ // all the bytes of the delimiter in the call to findDelimiterBounds() below
+ requiredPosition = startOffset - delimiter.length;
+ }
+ ((SeekableByteChannel) channel).position(requiredPosition);
+ findDelimiterBoundsWithMultiLineCheck();
+ buffer = buffer.substring(endOfDelimiterInBuffer);
+ startOfNextRecord = requiredPosition + endOfDelimiterInBuffer;
+ endOfDelimiterInBuffer = 0;
+ startOfDelimiterInBuffer = 0;
+ }
+ }
+
+ private void findDelimiterBoundsWithMultiLineCheck() throws IOException {
+ findDelimiterBounds();
+ }
+
+ /**
+ * Locates the start position and end position of the next delimiter. Will consume the channel
+ * till either EOF or the delimiter bounds are found.
+ *
+ * <p>If {@link ContextualTextIOSource#hasMultilineCSVRecords} is set then the behaviour will
+ * change from the standard read seen in {@link org.apache.beam.sdk.io.TextIO}. The assumption
+ * when {@link ContextualTextIOSource#hasMultilineCSVRecords} is set is that the file is being
+ * read with a single thread.
+ *
+ * <p>This fills the buffer and updates the positions as follows:
+ *
+ * <pre>{@code
+ * ------------------------------------------------------
+ * | element bytes | delimiter bytes | unconsumed bytes |
+ * ------------------------------------------------------
+ * 0 start of end of buffer
+ * delimiter delimiter size
+ * in buffer in buffer
+ * }</pre>
+ */
+ private void findDelimiterBounds() throws IOException {
+ int bytePositionInBuffer = 0;
+ boolean doubleQuoteClosed = true;
+
+ while (true) {
+ if (!tryToEnsureNumberOfBytesInBuffer(bytePositionInBuffer + 1)) {
+ startOfDelimiterInBuffer = endOfDelimiterInBuffer = bytePositionInBuffer;
+ break;
+ }
+
+ byte currentByte = buffer.byteAt(bytePositionInBuffer);
+ if (hasMultilineCSVRecords) {
+ // Check if we are inside an open Quote
+ if (currentByte == '"') {
+ doubleQuoteClosed = !doubleQuoteClosed;
+ }
+ } else {
+ doubleQuoteClosed = true;
+ }
+
+ if (delimiter == null) {
+ // default delimiter
+ if (currentByte == '\n') {
+ startOfDelimiterInBuffer = bytePositionInBuffer;
+ endOfDelimiterInBuffer = startOfDelimiterInBuffer + 1;
+ if (doubleQuoteClosed) {
+ break;
+ }
+ } else if (currentByte == '\r') {
+ startOfDelimiterInBuffer = bytePositionInBuffer;
+ endOfDelimiterInBuffer = startOfDelimiterInBuffer + 1;
+ if (tryToEnsureNumberOfBytesInBuffer(bytePositionInBuffer + 2)) {
+ currentByte = buffer.byteAt(bytePositionInBuffer + 1);
+ if (currentByte == '\n') {
+ endOfDelimiterInBuffer += 1;
+ }
+ }
+ if (doubleQuoteClosed) {
+ break;
+ }
+ }
+ } else {
+ // when the user defines a delimiter
+ int i = 0;
+ startOfDelimiterInBuffer = endOfDelimiterInBuffer = bytePositionInBuffer;
+ while ((i < delimiter.length) && (currentByte == delimiter[i])) {
+ // read next byte;
+ i++;
+ if (tryToEnsureNumberOfBytesInBuffer(bytePositionInBuffer + i + 1)) {
+ currentByte = buffer.byteAt(bytePositionInBuffer + i);
+ } else {
+ // corner case: delimiter truncate at the end of file
+ startOfDelimiterInBuffer = endOfDelimiterInBuffer = bytePositionInBuffer;
+ break;
+ }
+ }
+ if (i == delimiter.length) {
+ endOfDelimiterInBuffer = bytePositionInBuffer + i;
+ if (doubleQuoteClosed) {
+ break;
+ }
+ }
+ }
+ bytePositionInBuffer += 1;
+ }
+ }
+
+ @Override
+ protected boolean readNextRecord() throws IOException {
+ startOfRecord = startOfNextRecord;
+
+ findDelimiterBoundsWithMultiLineCheck();
+
+ // If we have reached EOF file and consumed all of the buffer then we know
+ // that there are no more records.
+ if (eof && buffer.isEmpty()) {
+ elementIsPresent = false;
+ return false;
+ }
+
+ decodeCurrentElement();
+ startOfNextRecord = startOfRecord + endOfDelimiterInBuffer;
+ return true;
+ }
+
+ /**
+ * Decodes the current element updating the buffer to only contain the unconsumed bytes.
+ *
+ * <p>This invalidates the currently stored {@code startOfDelimiterInBuffer} and {@code
+ * endOfDelimiterInBuffer}.
+ */
+ private void decodeCurrentElement() throws IOException {
+ ByteString dataToDecode = buffer.substring(0, startOfDelimiterInBuffer);
+ // If present, the UTF8 Byte Order Mark (BOM) will be removed.
+ if (startOfRecord == 0 && dataToDecode.startsWith(UTF8_BOM)) {
+ dataToDecode = dataToDecode.substring(UTF8_BOM.size());
+ }
+
+ // The line num is:
+ Long recordUniqueNum = totalRecordCount++;
+ // The Complete FileName (with uri if this is a web url eg: temp/abc.txt) is:
+ String fileName = getCurrentSource().getSingleFileMetadata().resourceId().toString();
Review comment:
@abhiy13 Please use the resource ID
##########
File path: sdks/java/io/contextual-text-io/src/main/java/org/apache/beam/sdk/io/contextualtextio/RecordWithMetadata.java
##########
@@ -0,0 +1,84 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one
+ * or more contributor license agreements. See the NOTICE file
+ * distributed with this work for additional information
+ * regarding copyright ownership. The ASF licenses this file
+ * to you under the Apache License, Version 2.0 (the
+ * "License"); you may not use this file except in compliance
+ * with the License. You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+package org.apache.beam.sdk.io.contextualtextio;
+
+import com.google.auto.value.AutoValue;
+import org.apache.beam.sdk.annotations.Experimental;
+import org.apache.beam.sdk.schemas.AutoValueSchema;
+import org.apache.beam.sdk.schemas.annotations.DefaultSchema;
+
+/**
+ * Helper Class based on {@link AutoValueSchema}, it provides Metadata associated with each Record
+ * when reading from file(s) using {@link ContextualTextIO}.
+ *
+ * <h3>Fields:</h3>
+ *
+ * <ul>
+ * <li>recordOffset: The offset of a record (the byte at which the record begins) in a file. This
+ * information can be useful if you wish to reconstruct the file. {@link
+ * RecordWithMetadata#getRecordOffset()}
+ * <li>recordNum: The ordinal number of the record in its file. {@link
+ * RecordWithMetadata#getRecordNum()}
+ * <li>recordValue: The value / contents of the record {@link RecordWithMetadata#getRecordValue()}
+ * <li>rangeOffset: The starting offset of the range (split), which contained the record, when the
+ * record was read. {@link RecordWithMetadata#getRangeOffset()}
+ * <li>recordNumInOffset: The record number relative to the Range. (line number within the range)
+ * {@link RecordWithMetadata#getRecordNumInOffset()}
+ * <li>fileName: Name of the file to which the record belongs (this is the full filename,
+ * eg:path/to/file.txt) {@link RecordWithMetadata#getFileName()}
+ * </ul>
+ */
+@Experimental(Experimental.Kind.SCHEMAS)
+@DefaultSchema(AutoValueSchema.class)
+@AutoValue
+public abstract class RecordWithMetadata {
Review comment:
@abhiy13 please make this generic as per Luke request.
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
[GitHub] [beam] abhiy13 commented on a change in pull request #12645: [BEAM-10124] Add ContextualTextIO
Posted by GitBox <gi...@apache.org>.
abhiy13 commented on a change in pull request #12645:
URL: https://github.com/apache/beam/pull/12645#discussion_r492474814
##########
File path: sdks/java/io/contextual-text-io/src/main/java/org/apache/beam/sdk/io/contextualtextio/ContextualTextIO.java
##########
@@ -0,0 +1,631 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one
+ * or more contributor license agreements. See the NOTICE file
+ * distributed with this work for additional information
+ * regarding copyright ownership. The ASF licenses this file
+ * to you under the Apache License, Version 2.0 (the
+ * "License"); you may not use this file except in compliance
+ * with the License. You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+package org.apache.beam.sdk.io.contextualtextio;
+
+import static org.apache.beam.sdk.io.FileIO.ReadMatches.DirectoryTreatment;
+import static org.apache.beam.vendor.guava.v26_0_jre.com.google.common.base.Preconditions.checkArgument;
+import static org.apache.beam.vendor.guava.v26_0_jre.com.google.common.base.Preconditions.checkNotNull;
+
+import com.google.auto.value.AutoValue;
+import java.nio.ByteBuffer;
+import java.util.Arrays;
+import java.util.Comparator;
+import java.util.HashMap;
+import java.util.Map;
+import java.util.SortedMap;
+import java.util.TreeMap;
+import org.apache.beam.sdk.coders.StringUtf8Coder;
+import org.apache.beam.sdk.io.CompressedSource;
+import org.apache.beam.sdk.io.Compression;
+import org.apache.beam.sdk.io.FileBasedSource;
+import org.apache.beam.sdk.io.FileIO;
+import org.apache.beam.sdk.io.FileIO.MatchConfiguration;
+import org.apache.beam.sdk.io.ReadAllViaFileBasedSource;
+import org.apache.beam.sdk.io.TextIO;
+import org.apache.beam.sdk.io.fs.EmptyMatchTreatment;
+import org.apache.beam.sdk.options.ValueProvider;
+import org.apache.beam.sdk.options.ValueProvider.StaticValueProvider;
+import org.apache.beam.sdk.schemas.NoSuchSchemaException;
+import org.apache.beam.sdk.schemas.SchemaCoder;
+import org.apache.beam.sdk.transforms.Count;
+import org.apache.beam.sdk.transforms.Create;
+import org.apache.beam.sdk.transforms.DoFn;
+import org.apache.beam.sdk.transforms.PTransform;
+import org.apache.beam.sdk.transforms.ParDo;
+import org.apache.beam.sdk.transforms.SerializableFunction;
+import org.apache.beam.sdk.transforms.View;
+import org.apache.beam.sdk.transforms.Watch.Growth.TerminationCondition;
+import org.apache.beam.sdk.transforms.display.DisplayData;
+import org.apache.beam.sdk.values.KV;
+import org.apache.beam.sdk.values.PBegin;
+import org.apache.beam.sdk.values.PCollection;
+import org.apache.beam.sdk.values.PCollectionView;
+import org.apache.beam.vendor.guava.v26_0_jre.com.google.common.annotations.VisibleForTesting;
+import org.checkerframework.checker.nullness.qual.Nullable;
+import org.joda.time.Duration;
+import org.slf4j.Logger;
+import org.slf4j.LoggerFactory;
+
+/**
+ * {@link PTransform}s that read text files and collect contextual information of the elements in
+ * the input.
+ *
+ * <p>Use {@link TextIO} when not reading file with Multiline Records or additional metadata is not
+ * required.
+ *
+ * <h2>Reading from text files</h2>
+ *
+ * <p>To read a {@link PCollection} from one or more text files, use {@code
+ * ContextualTextIO.read()}. To instantiate a transform use {@link
+ * ContextualTextIO.Read#from(String)} and specify the path of the file(s) to be read.
+ * Alternatively, if the filenames to be read are themselves in a {@link PCollection} you can use
+ * {@link FileIO} to match them and {@link ContextualTextIO#readFiles()} to read them.
+ *
+ * <p>{@link #read} returns a {@link PCollection} of {@link RecordWithMetadata RecordWithMetadata},
+ * each corresponding to one line of an input UTF-8 text file (split into lines delimited by '\n',
+ * '\r', '\r\n', or specified delimiter see {@link ContextualTextIO.Read#withDelimiter})
+ *
+ * <h3>Filepattern expansion and watching</h3>
+ *
+ * <p>By default, the filepatterns are expanded only once. The combination of {@link
+ * FileIO.Match#continuously(Duration, TerminationCondition)} and {@link #readFiles()} allow
+ * streaming of new files matching the filepattern(s).
+ *
+ * <p>By default, {@link #read} prohibits filepatterns that match no files, and {@link #readFiles()}
+ * allows them in case the filepattern contains a glob wildcard character. Use {@link
+ * ContextualTextIO.Read#withEmptyMatchTreatment} or {@link
+ * FileIO.Match#withEmptyMatchTreatment(EmptyMatchTreatment)} plus {@link #readFiles()} to configure
+ * this behavior.
+ *
+ * <p>Example 1: reading a file or filepattern.
+ *
+ * <pre>{@code
+ * Pipeline p = ...;
+ *
+ * // A simple Read of a file:
+ * PCollection<RecordWithMetadata> records = p.apply(ContextualTextIO.read().from("/local/path/to/file.txt"));
+ * }</pre>
+ *
+ * <p>Example 2: reading a PCollection of filenames.
+ *
+ * <pre>{@code
+ * Pipeline p = ...;
+ *
+ * // E.g. the filenames might be computed from other data in the pipeline, or
+ * // read from a data source.
+ * PCollection<String> filenames = ...;
+ *
+ * // Read all files in the collection.
+ * PCollection<RecordWithMetadata> records =
+ * filenames
+ * .apply(FileIO.matchAll())
+ * .apply(FileIO.readMatches())
+ * .apply(ContextualTextIO.readFiles());
+ * }</pre>
+ *
+ * <p>Example 3: streaming new files matching a filepattern.
+ *
+ * <pre>{@code
+ * Pipeline p = ...;
+ *
+ * PCollection<RecordWithMetadata> records = p.apply(ContextualTextIO.read()
+ * .from("/local/path/to/files/*")
+ * .watchForNewFiles(
+ * // Check for new files every minute
+ * Duration.standardMinutes(1),
+ * // Stop watching the filepattern if no new files appear within an hour
+ * afterTimeSinceNewOutput(Duration.standardHours(1))));
+ * }</pre>
+ *
+ * <p>Example 4: reading a file or file pattern of RFC4180-compliant CSV files with fields that may
+ * contain line breaks.
+ *
+ * Example of such a file could be:
+ *
+ * "aaa","b CRLF
+ * bb","ccc" CRLF
+ * zzz,yyy,xxx
+ *
+ * <pre>{@code
+ * Pipeline p = ...;
+ *
+ * PCollection<RecordWithMetadata> records = p.apply(ContextualTextIO.read()
+ * .from("/local/path/to/files/*.csv")
+ * .withHasMultilineCSVRecords(true));
+ * }</pre>
+ *
+ * <p>Example 5: reading while watching for new files
+ *
+ * <pre>{@code
+ * Pipeline p = ...;
+ *
+ * PCollection<RecordWithMetadata> records = p.apply(FileIO.match()
+ * .filepattern("filepattern")
+ * .continuously(
+ * Duration.millis(100),
+ * Watch.Growth.afterTimeSinceNewOutput(Duration.standardSeconds(3))))
+ * .apply(FileIO.readMatches())
+ * .apply(ContextualTextIO.readFiles());
+ * }</pre>
+ *
+ * <p>Example 6: reading without recordNum metadata, or only fileName associated Metadata. (the
+ * Objects would still contain recordNums, but these recordNums would correspond to their positions
+ * in their respective offsets rather than their positions within the entire file).
+ *
+ * <pre>{@code
+ * Pipeline p = ...;
+ *
+ * PCollection<RecordWithMetadata> records = p.apply(ContextualTextIO.read()
+ * .from("/local/path/to/files/*.csv")
+ * .setWithoutRecordNumMetadata(true));
+ * }</pre>
+ *
+ * <p>NOTE: When using {@link ContextualTextIO.Read#withHasMultilineCSVRecords(Boolean)} this
+ * option, a single reader will be used to process the file, rather than multiple readers which can
+ * read from different offsets. For a large file this can result in lower performance.
+ *
+ * <p>NOTE: Use {@link Read#withoutRecordNumMetadata()} when recordNum metadata is not required, for
+ * example, when when only filename metadata is required. Computing record positions currently
+ * introduces a shuffle step, which increases the resources used by the pipeline. <b> By default
+ * withoutRecordNumMetadata is set to false, so the shuffle step is performed.</b>
+ *
+ * <h3>Reading a very large number of files</h3>
+ *
+ * <p>If it is known that the filepattern will match a very large number of files (e.g. tens of
+ * thousands or more), use {@link ContextualTextIO.Read#withHintMatchesManyFiles} for better
+ * performance and scalability. Note that it may decrease performance if the filepattern matches
+ * only a small number of files.
+ */
+public class ContextualTextIO {
+ private static final long DEFAULT_BUNDLE_SIZE_BYTES = 64 * 1024 * 1024L;
+ private static final Logger LOG = LoggerFactory.getLogger(ContextualTextIO.class);
+
+ /**
+ * A {@link PTransform} that reads from one or more text files and returns a bounded {@link
+ * PCollection} containing one {@link RecordWithMetadata}element for each line of the input files.
+ */
+ public static Read read() {
+ return new AutoValue_ContextualTextIO_Read.Builder()
+ .setCompression(Compression.AUTO)
+ .setHintMatchesManyFiles(false)
+ .setWithoutRecordNumMetadata(false)
+ .setMatchConfiguration(MatchConfiguration.create(EmptyMatchTreatment.DISALLOW))
+ .setHasMultilineCSVRecords(false)
+ .build();
+ }
+
+ /**
+ * Like {@link #read}, but reads each file in a {@link PCollection} of {@link
+ * FileIO.ReadableFile}, returned by {@link FileIO#readMatches}.
+ */
+ public static ReadFiles readFiles() {
+ return new AutoValue_ContextualTextIO_ReadFiles.Builder()
+ // 64MB is a reasonable value that allows to amortize the cost of opening files,
+ // but is not so large as to exhaust a typical runner's maximum amount of output per
+ // ProcessElement call.
+ .setDesiredBundleSizeBytes(DEFAULT_BUNDLE_SIZE_BYTES)
+ .setHasMultilineCSVRecords(false)
+ .build();
+ }
+
+ /** Implementation of {@link #read}. */
+ @AutoValue
+ public abstract static class Read extends PTransform<PBegin, PCollection<RecordWithMetadata>> {
+ abstract @Nullable ValueProvider<String> getFilepattern();
+
+ abstract MatchConfiguration getMatchConfiguration();
+
+ abstract boolean getHintMatchesManyFiles();
+
+ abstract boolean getWithoutRecordNumMetadata();
+
+ abstract Compression getCompression();
+
+ abstract @Nullable Boolean getHasMultilineCSVRecords();
+
+ @SuppressWarnings("mutable") // this returns an array that can be mutated by the caller
+ abstract byte @Nullable [] getDelimiter();
+
+ abstract Builder toBuilder();
+
+ @AutoValue.Builder
+ abstract static class Builder {
+ abstract Builder setFilepattern(ValueProvider<String> filepattern);
+
+ abstract Builder setMatchConfiguration(MatchConfiguration matchConfiguration);
+
+ abstract Builder setHintMatchesManyFiles(boolean hintManyFiles);
+
+ abstract Builder setWithoutRecordNumMetadata(boolean withoutLineNumMetadata);
+
+ abstract Builder setCompression(Compression compression);
+
+ abstract Builder setDelimiter(byte @Nullable [] delimiter);
+
+ abstract Builder setHasMultilineCSVRecords(Boolean hasMultilineCSVRecords);
+
+ abstract Read build();
+ }
+
+ /**
+ * Reads text from the file(s) with the given filename or filename pattern.
+ *
+ * <p>This can be a local path (if running locally), or a Google Cloud Storage filename or
+ * filename pattern of the form {@code "gs://<bucket>/<filepath>"} (if running locally or using
+ * remote execution service).
+ *
+ * <p>Standard <a href="http://docs.oracle.com/javase/tutorial/essential/io/find.html" >Java
+ * Filesystem glob patterns</a> ("*", "?", "[..]") are supported.
+ *
+ * <p>If it is known that the filepattern will match a very large number of files (at least tens
+ * of thousands), use {@link #withHintMatchesManyFiles} for better performance and scalability.
+ */
+ public Read from(String filepattern) {
+ checkArgument(filepattern != null, "filepattern can not be null");
+ return from(StaticValueProvider.of(filepattern));
+ }
+
+ /** Same as {@code from(filepattern)}, but accepting a {@link ValueProvider}. */
+ public Read from(ValueProvider<String> filepattern) {
+ checkArgument(filepattern != null, "filepattern can not be null");
+ return toBuilder().setFilepattern(filepattern).build();
+ }
+
+ /** Sets the {@link MatchConfiguration}. */
+ public Read withMatchConfiguration(MatchConfiguration matchConfiguration) {
+ return toBuilder().setMatchConfiguration(matchConfiguration).build();
+ }
+
+ /**
+ * When reading RFC4180 CSV files that have values that span multiple lines, set this to true.
+ * Note: this reduces the read performance (see: {@link ContextualTextIO}).
+ */
+ public Read withHasMultilineCSVRecords(Boolean hasMultilineCSVRecords) {
+ return toBuilder().setHasMultilineCSVRecords(hasMultilineCSVRecords).build();
+ }
+
+ /**
+ * Reads from input sources using the specified compression type.
+ *
+ * <p>If no compression type is specified, the default is {@link Compression#AUTO}.
+ */
+ public Read withCompression(Compression compression) {
+ return toBuilder().setCompression(compression).build();
+ }
+
+ /**
+ * Hints that the filepattern specified in {@link #from(String)} matches a very large number of
+ * files.
+ *
+ * <p>This hint may cause a runner to execute the transform differently, in a way that improves
+ * performance for this case, but it may worsen performance if the filepattern matches only a
+ * small number of files (e.g., in a runner that supports dynamic work rebalancing, it will
+ * happen less efficiently within individual files).
+ */
+ public Read withHintMatchesManyFiles() {
+ return toBuilder().setHintMatchesManyFiles(true).build();
+ }
+
+ /**
+ * Allows the user to opt out of getting recordNums associated with each record.
+ *
+ * <p>When set to true, it will introduce a shuffle step to assemble the recordNums for each
+ * record, which will increase the resources used by the pipeline.
+ *
+ * <p>Use this when metadata like fileNames are required and their position/order can be
+ * ignored.
+ */
+ public Read withoutRecordNumMetadata() {
+ return toBuilder().setWithoutRecordNumMetadata(true).build();
+ }
+
+ /** See {@link MatchConfiguration#withEmptyMatchTreatment}. */
+ public Read withEmptyMatchTreatment(EmptyMatchTreatment treatment) {
+ return withMatchConfiguration(getMatchConfiguration().withEmptyMatchTreatment(treatment));
+ }
+
+ /** Set the custom delimiter to be used in place of the default ones ('\r', '\n' or '\r\n'). */
+ public Read withDelimiter(byte[] delimiter) {
+ checkArgument(delimiter != null, "delimiter can not be null");
+ checkArgument(!isSelfOverlapping(delimiter), "delimiter must not self-overlap");
+ return toBuilder().setDelimiter(delimiter).build();
+ }
+
+ static boolean isSelfOverlapping(byte[] s) {
+ // s self-overlaps if v exists such as s = vu = wv with u and w non empty
+ for (int i = 1; i < s.length - 1; ++i) {
+ if (ByteBuffer.wrap(s, 0, i).equals(ByteBuffer.wrap(s, s.length - i, i))) {
+ return true;
+ }
+ }
+ return false;
+ }
+
+ @Override
+ public PCollection<RecordWithMetadata> expand(PBegin input) {
+ checkNotNull(
+ getFilepattern(), "need to set the filepattern of a ContextualTextIO.Read transform");
+ PCollection<RecordWithMetadata> records = null;
+ if (getMatchConfiguration().getWatchInterval() == null && !getHintMatchesManyFiles()) {
+ records = input.apply("Read", org.apache.beam.sdk.io.Read.from(getSource()));
+ } else {
+ // All other cases go through FileIO + ReadFiles
+ records =
+ input
+ .apply(
+ "Create filepattern", Create.ofProvider(getFilepattern(), StringUtf8Coder.of()))
+ .apply("Match All", FileIO.matchAll().withConfiguration(getMatchConfiguration()))
+ .apply(
+ "Read Matches",
+ FileIO.readMatches()
+ .withCompression(getCompression())
+ .withDirectoryTreatment(DirectoryTreatment.PROHIBIT))
+ .apply("Via ReadFiles", readFiles().withDelimiter(getDelimiter()));
+ }
+
+ // Check if the user decided to opt out of recordNums associated with records
+ if (getWithoutRecordNumMetadata()) {
+ return records;
+ }
+
+ /*
+ * At this point the line number in RecordWithMetadata contains the relative line offset from the beginning of the read range.
+ *
+ * To compute the absolute position from the beginning of the input we group the lines within the same ranges, and evaluate the size of each range.
+ */
+
+ PCollection<KV<KV<String, Long>, RecordWithMetadata>> recordsGroupedByFileAndRange =
+ records.apply("AddFileNameAndRange", ParDo.of(new AddFileNameAndRange()));
+
+ PCollectionView<Map<KV<String, Long>, Long>> rangeSizes =
+ recordsGroupedByFileAndRange
+ .apply("CountRecordsForEachFileRange", Count.perKey())
+ .apply("SizesAsView", View.asMap());
+
+ // Get Pipeline to create a dummy PCollection with one element to help compute the lines
+ // before each Range
+ PCollection<Integer> singletonPcoll =
+ input.getPipeline().apply("CreateSingletonPcoll", Create.of(Arrays.asList(1)));
+
+ /*
+ * For each (File, Offset) pair, calculate the number of lines occurring before the Range for each file
+ *
+ * After computing the number of lines before each range, we can find the line number in original file as numLiesBeforeOffset + lineNumInCurrentOffset
+ */
+
+ PCollectionView<Map<KV<String, Long>, Long>> numRecordsBeforeEachRange =
+ singletonPcoll
+ .apply(
+ "ComputeRecordsBeforeRange",
+ ParDo.of(new ComputeRecordsBeforeEachRange(rangeSizes))
+ .withSideInputs(rangeSizes))
+ .apply("NumRecordsBeforeEachRangeAsView", View.asMap());
+
+ return recordsGroupedByFileAndRange.apply(
+ "AssignLineNums",
+ ParDo.of(new AssignRecordNums(numRecordsBeforeEachRange))
+ .withSideInputs(numRecordsBeforeEachRange));
+ }
+
+ @VisibleForTesting
+ static class AddFileNameAndRange
+ extends DoFn<RecordWithMetadata, KV<KV<String, Long>, RecordWithMetadata>> {
+ @ProcessElement
+ public void processElement(
+ @Element RecordWithMetadata record,
+ OutputReceiver<KV<KV<String, Long>, RecordWithMetadata>> out) {
+ out.output(KV.of(KV.of(record.getFileName().toString(), record.getRangeOffset()), record));
+ }
+ }
+
+ /**
+ * Helper class for computing number of record in the File preceding the beginning of the Range
+ * in this file.
+ */
+ @VisibleForTesting
+ static class ComputeRecordsBeforeEachRange extends DoFn<Integer, KV<KV<String, Long>, Long>> {
+ private final PCollectionView<Map<KV<String, Long>, Long>> rangeSizes;
+
+ public ComputeRecordsBeforeEachRange(
+ PCollectionView<Map<KV<String, Long>, Long>> rangeSizes) {
+ this.rangeSizes = rangeSizes;
+ }
+
+ // Add custom comparator as KV<K, V> is not comparable by default
+ private static class FileRangeComparator<K extends Comparable<K>, V extends Comparable<V>>
+ implements Comparator<KV<K, V>> {
+ @Override
+ public int compare(KV<K, V> a, KV<K, V> b) {
+ if (a.getKey().compareTo(b.getKey()) == 0) {
+ return a.getValue().compareTo(b.getValue());
+ }
+ return a.getKey().compareTo(b.getKey());
+ }
+ }
+
+ @ProcessElement
+ public void processElement(ProcessContext p) {
+ // Get the Map Containing the size from side-input
+ Map<KV<String, Long>, Long> rangeSizesMap = p.sideInput(rangeSizes);
+
+ // The FileRange Pair must be sorted
+ SortedMap<KV<String, Long>, Long> sorted = new TreeMap<>(new FileRangeComparator<>());
+
+ // Initialize sorted map with values
+ for (Map.Entry<KV<String, Long>, Long> entry : rangeSizesMap.entrySet()) {
+ sorted.put(entry.getKey(), entry.getValue());
+ }
+
+ // HashMap that tracks number of records passed for each file
+ Map<String, Long> pastRecords = new HashMap<>();
+
+ // For each (File, Range) Pair, compute the number of records before it
+ for (Map.Entry entry : sorted.entrySet()) {
+ Long numRecords = (long) entry.getValue();
+ KV<String, Long> fileRange = (KV<String, Long>) entry.getKey();
+ String file = fileRange.getKey();
+ Long numRecordsBefore = 0L;
+ if (pastRecords.containsKey(file)) {
+ numRecordsBefore = pastRecords.get(file);
+ }
+ p.output(KV.of(fileRange, numRecordsBefore));
+ pastRecords.put(file, numRecordsBefore + numRecords);
+ }
+ }
+ }
+
+ static class AssignRecordNums
+ extends DoFn<KV<KV<String, Long>, RecordWithMetadata>, RecordWithMetadata> {
+ PCollectionView<Map<KV<String, Long>, Long>> numRecordsBeforeEachRange;
+
+ public AssignRecordNums(
+ PCollectionView<Map<KV<String, Long>, Long>> numRecordsBeforeEachRange) {
+ this.numRecordsBeforeEachRange = numRecordsBeforeEachRange;
+ }
+
+ @ProcessElement
+ public void processElement(ProcessContext p) {
+ Long range = p.element().getKey().getValue();
+ String file = p.element().getKey().getKey();
+ RecordWithMetadata record = p.element().getValue();
+ Long numRecordsLessThanThisRange =
+ p.sideInput(numRecordsBeforeEachRange).get(KV.of(file, range));
+ RecordWithMetadata newLine =
+ RecordWithMetadata.newBuilder()
Review comment:
Thanks !
Done.
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
[GitHub] [beam] tvalentyn commented on pull request #12645: [BEAM-10124] Add ContextualTextIO
Posted by GitBox <gi...@apache.org>.
tvalentyn commented on pull request #12645:
URL: https://github.com/apache/beam/pull/12645#issuecomment-678547893
Run Java PreCommit
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
[GitHub] [beam] lukecwik commented on a change in pull request #12645: [BEAM-10124] Add ContextualTextIO
Posted by GitBox <gi...@apache.org>.
lukecwik commented on a change in pull request #12645:
URL: https://github.com/apache/beam/pull/12645#discussion_r477459276
##########
File path: sdks/java/core/src/main/java/org/apache/beam/sdk/io/FileIO.java
##########
@@ -472,9 +472,9 @@ public static MatchConfiguration create(EmptyMatchTreatment emptyMatchTreatment)
.build();
}
- abstract EmptyMatchTreatment getEmptyMatchTreatment();
+ public abstract EmptyMatchTreatment getEmptyMatchTreatment();
- abstract @Nullable Duration getWatchInterval();
+ public abstract @Nullable Duration getWatchInterval();
abstract @Nullable TerminationCondition<String, ?> getWatchTerminationCondition();
Review comment:
might as well and make this public too
##########
File path: sdks/java/io/contextual-text-io/src/main/java/org/apache/beam/sdk/io/contextualtextio/RecordWithMetadata.java
##########
@@ -0,0 +1,84 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one
+ * or more contributor license agreements. See the NOTICE file
+ * distributed with this work for additional information
+ * regarding copyright ownership. The ASF licenses this file
+ * to you under the Apache License, Version 2.0 (the
+ * "License"); you may not use this file except in compliance
+ * with the License. You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+package org.apache.beam.sdk.io.contextualtextio;
+
+import com.google.auto.value.AutoValue;
+import org.apache.beam.sdk.annotations.Experimental;
+import org.apache.beam.sdk.schemas.AutoValueSchema;
+import org.apache.beam.sdk.schemas.annotations.DefaultSchema;
+
+/**
+ * Helper Class based on {@link AutoValueSchema}, it provides Metadata associated with each Record
+ * when reading from file(s) using {@link ContextualTextIO}.
+ *
+ * <h3>Fields:</h3>
+ *
+ * <ul>
+ * <li>recordOffset: The offset of a record (the byte at which the record begins) in a file. This
+ * information can be useful if you wish to reconstruct the file. {@link
+ * RecordWithMetadata#getRecordOffset()}
+ * <li>recordNum: The ordinal number of the record in its file. {@link
+ * RecordWithMetadata#getRecordNum()}
+ * <li>recordValue: The value / contents of the record {@link RecordWithMetadata#getRecordValue()}
+ * <li>rangeOffset: The starting offset of the range (split), which contained the record, when the
+ * record was read. {@link RecordWithMetadata#getRangeOffset()}
+ * <li>recordNumInOffset: The record number relative to the Range. (line number within the range)
+ * {@link RecordWithMetadata#getRecordNumInOffset()}
+ * <li>fileName: Name of the file to which the record belongs (this is the full filename,
+ * eg:path/to/file.txt) {@link RecordWithMetadata#getFileName()}
+ * </ul>
+ */
+@Experimental(Experimental.Kind.SCHEMAS)
+@DefaultSchema(AutoValueSchema.class)
+@AutoValue
+public abstract class RecordWithMetadata {
Review comment:
Should this be RecordWithMetadata<T>?
##########
File path: sdks/java/io/contextual-text-io/src/main/java/org/apache/beam/sdk/io/contextualtextio/RecordWithMetadata.java
##########
@@ -0,0 +1,84 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one
+ * or more contributor license agreements. See the NOTICE file
+ * distributed with this work for additional information
+ * regarding copyright ownership. The ASF licenses this file
+ * to you under the Apache License, Version 2.0 (the
+ * "License"); you may not use this file except in compliance
+ * with the License. You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+package org.apache.beam.sdk.io.contextualtextio;
+
+import com.google.auto.value.AutoValue;
+import org.apache.beam.sdk.annotations.Experimental;
+import org.apache.beam.sdk.schemas.AutoValueSchema;
+import org.apache.beam.sdk.schemas.annotations.DefaultSchema;
+
+/**
+ * Helper Class based on {@link AutoValueSchema}, it provides Metadata associated with each Record
+ * when reading from file(s) using {@link ContextualTextIO}.
+ *
+ * <h3>Fields:</h3>
+ *
+ * <ul>
+ * <li>recordOffset: The offset of a record (the byte at which the record begins) in a file. This
+ * information can be useful if you wish to reconstruct the file. {@link
+ * RecordWithMetadata#getRecordOffset()}
+ * <li>recordNum: The ordinal number of the record in its file. {@link
+ * RecordWithMetadata#getRecordNum()}
+ * <li>recordValue: The value / contents of the record {@link RecordWithMetadata#getRecordValue()}
+ * <li>rangeOffset: The starting offset of the range (split), which contained the record, when the
+ * record was read. {@link RecordWithMetadata#getRangeOffset()}
+ * <li>recordNumInOffset: The record number relative to the Range. (line number within the range)
+ * {@link RecordWithMetadata#getRecordNumInOffset()}
+ * <li>fileName: Name of the file to which the record belongs (this is the full filename,
+ * eg:path/to/file.txt) {@link RecordWithMetadata#getFileName()}
+ * </ul>
+ */
+@Experimental(Experimental.Kind.SCHEMAS)
+@DefaultSchema(AutoValueSchema.class)
+@AutoValue
+public abstract class RecordWithMetadata {
+ public abstract Long getRecordOffset();
Review comment:
please use `long` instead of `Long` since we don't expect any of these `Long` fields to be nullable
##########
File path: sdks/java/io/contextual-text-io/src/main/java/org/apache/beam/sdk/io/contextualtextio/RecordWithMetadata.java
##########
@@ -0,0 +1,84 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one
+ * or more contributor license agreements. See the NOTICE file
+ * distributed with this work for additional information
+ * regarding copyright ownership. The ASF licenses this file
+ * to you under the Apache License, Version 2.0 (the
+ * "License"); you may not use this file except in compliance
+ * with the License. You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+package org.apache.beam.sdk.io.contextualtextio;
+
+import com.google.auto.value.AutoValue;
+import org.apache.beam.sdk.annotations.Experimental;
+import org.apache.beam.sdk.schemas.AutoValueSchema;
+import org.apache.beam.sdk.schemas.annotations.DefaultSchema;
+
+/**
+ * Helper Class based on {@link AutoValueSchema}, it provides Metadata associated with each Record
+ * when reading from file(s) using {@link ContextualTextIO}.
+ *
+ * <h3>Fields:</h3>
+ *
+ * <ul>
+ * <li>recordOffset: The offset of a record (the byte at which the record begins) in a file. This
+ * information can be useful if you wish to reconstruct the file. {@link
+ * RecordWithMetadata#getRecordOffset()}
+ * <li>recordNum: The ordinal number of the record in its file. {@link
+ * RecordWithMetadata#getRecordNum()}
+ * <li>recordValue: The value / contents of the record {@link RecordWithMetadata#getRecordValue()}
+ * <li>rangeOffset: The starting offset of the range (split), which contained the record, when the
+ * record was read. {@link RecordWithMetadata#getRangeOffset()}
+ * <li>recordNumInOffset: The record number relative to the Range. (line number within the range)
+ * {@link RecordWithMetadata#getRecordNumInOffset()}
+ * <li>fileName: Name of the file to which the record belongs (this is the full filename,
+ * eg:path/to/file.txt) {@link RecordWithMetadata#getFileName()}
+ * </ul>
+ */
+@Experimental(Experimental.Kind.SCHEMAS)
+@DefaultSchema(AutoValueSchema.class)
+@AutoValue
+public abstract class RecordWithMetadata {
+ public abstract Long getRecordOffset();
+
+ public abstract Long getRecordNum();
+
+ public abstract String getRecordValue();
+
+ public abstract Long getRangeOffset();
+
+ public abstract Long getRecordNumInOffset();
+
+ public abstract Builder toBuilder();
+
+ public abstract String getFileName();
Review comment:
We should be using ResourceId here.
##########
File path: sdks/java/io/contextual-text-io/build.gradle
##########
@@ -0,0 +1,41 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one
+ * or more contributor license agreements. See the NOTICE file
+ * distributed with this work for additional information
+ * regarding copyright ownership. The ASF licenses this file
+ * to you under the Apache License, Version 2.0 (the
+ * License); you may not use this file except in compliance
+ * with the License. You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an AS IS BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+plugins { id 'org.apache.beam.module' }
+applyJavaNature(
+ automaticModuleName: 'org.apache.beam.sdk.io.contextual-text-io',
+ enableChecker: false,
+ ignoreRawtypeErrors: true)
+
+description = "Apache Beam :: SDKs :: Java :: Contextual-Text-IO"
+ext.summary = "Context-aware Text IO."
+
+dependencies {
+
Review comment:
```suggestion
```
##########
File path: sdks/java/io/contextual-text-io/build.gradle
##########
@@ -0,0 +1,41 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one
+ * or more contributor license agreements. See the NOTICE file
+ * distributed with this work for additional information
+ * regarding copyright ownership. The ASF licenses this file
+ * to you under the Apache License, Version 2.0 (the
+ * License); you may not use this file except in compliance
+ * with the License. You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an AS IS BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+plugins { id 'org.apache.beam.module' }
+applyJavaNature(
+ automaticModuleName: 'org.apache.beam.sdk.io.contextual-text-io',
+ enableChecker: false,
+ ignoreRawtypeErrors: true)
+
+description = "Apache Beam :: SDKs :: Java :: Contextual-Text-IO"
+ext.summary = "Context-aware Text IO."
+
+dependencies {
+
+ compile library.java.vendored_guava_26_0_jre
+ compile library.java.protobuf_java
+ compile project(path: ":sdks:java:core", configuration: "shadow")
+ testCompile project(path: ":sdks:java:core", configuration: "shadowTest")
+
+ testCompile library.java.guava_testlib
+ testCompile library.java.junit
+ testCompile library.java.hamcrest_core
+ testRuntimeOnly library.java.slf4j_jdk14
+ testCompile project(path: ":runners:direct-java", configuration: "shadow")
+
+}
Review comment:
```suggestion
}
```
##########
File path: sdks/java/io/contextual-text-io/src/main/java/org/apache/beam/sdk/io/contextualtextio/ContextualTextIOSource.java
##########
@@ -0,0 +1,364 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one
+ * or more contributor license agreements. See the NOTICE file
+ * distributed with this work for additional information
+ * regarding copyright ownership. The ASF licenses this file
+ * to you under the Apache License, Version 2.0 (the
+ * "License"); you may not use this file except in compliance
+ * with the License. You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+package org.apache.beam.sdk.io.contextualtextio;
+
+import java.io.IOException;
+import java.nio.ByteBuffer;
+import java.nio.channels.ReadableByteChannel;
+import java.nio.channels.SeekableByteChannel;
+import java.util.NoSuchElementException;
+import org.apache.beam.sdk.coders.Coder;
+import org.apache.beam.sdk.io.FileBasedSource;
+import org.apache.beam.sdk.io.fs.EmptyMatchTreatment;
+import org.apache.beam.sdk.io.fs.MatchResult;
+import org.apache.beam.sdk.options.PipelineOptions;
+import org.apache.beam.sdk.options.ValueProvider;
+import org.apache.beam.sdk.schemas.NoSuchSchemaException;
+import org.apache.beam.sdk.schemas.SchemaCoder;
+import org.apache.beam.sdk.schemas.SchemaRegistry;
+import org.apache.beam.vendor.grpc.v1p26p0.com.google.protobuf.ByteString;
+import org.apache.beam.vendor.guava.v26_0_jre.com.google.common.annotations.VisibleForTesting;
+import org.apache.beam.vendor.guava.v26_0_jre.com.google.common.base.Preconditions;
+import org.checkerframework.checker.nullness.qual.Nullable;
+import org.slf4j.Logger;
+import org.slf4j.LoggerFactory;
+
+/**
+ * Implementation detail of {@link ContextualTextIO.Read}.
+ *
+ * <p>A {@link FileBasedSource} which can decode records delimited by newline characters.
+ *
+ * <p>This source splits the data into records using {@code UTF-8} {@code \n}, {@code \r}, or {@code
+ * \r\n} as the delimiter. This source is not strict and supports decoding the last record even if
+ * it is not delimited. Finally, no records are decoded if the stream is empty.
+ *
+ * <p>This source supports reading from any arbitrary byte position within the stream. If the
+ * starting position is not {@code 0}, then bytes are skipped until the first delimiter is found
+ * representing the beginning of the first record to be decoded.
+ */
+@VisibleForTesting
+class ContextualTextIOSource extends FileBasedSource<RecordWithMetadata> {
Review comment:
We should have made this a splittable DoFn.
##########
File path: sdks/java/io/contextual-text-io/src/main/java/org/apache/beam/sdk/io/contextualtextio/ContextualTextIOSource.java
##########
@@ -0,0 +1,364 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one
+ * or more contributor license agreements. See the NOTICE file
+ * distributed with this work for additional information
+ * regarding copyright ownership. The ASF licenses this file
+ * to you under the Apache License, Version 2.0 (the
+ * "License"); you may not use this file except in compliance
+ * with the License. You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+package org.apache.beam.sdk.io.contextualtextio;
+
+import java.io.IOException;
+import java.nio.ByteBuffer;
+import java.nio.channels.ReadableByteChannel;
+import java.nio.channels.SeekableByteChannel;
+import java.util.NoSuchElementException;
+import org.apache.beam.sdk.coders.Coder;
+import org.apache.beam.sdk.io.FileBasedSource;
+import org.apache.beam.sdk.io.fs.EmptyMatchTreatment;
+import org.apache.beam.sdk.io.fs.MatchResult;
+import org.apache.beam.sdk.options.PipelineOptions;
+import org.apache.beam.sdk.options.ValueProvider;
+import org.apache.beam.sdk.schemas.NoSuchSchemaException;
+import org.apache.beam.sdk.schemas.SchemaCoder;
+import org.apache.beam.sdk.schemas.SchemaRegistry;
+import org.apache.beam.vendor.grpc.v1p26p0.com.google.protobuf.ByteString;
+import org.apache.beam.vendor.guava.v26_0_jre.com.google.common.annotations.VisibleForTesting;
+import org.apache.beam.vendor.guava.v26_0_jre.com.google.common.base.Preconditions;
+import org.checkerframework.checker.nullness.qual.Nullable;
+import org.slf4j.Logger;
+import org.slf4j.LoggerFactory;
+
+/**
+ * Implementation detail of {@link ContextualTextIO.Read}.
+ *
+ * <p>A {@link FileBasedSource} which can decode records delimited by newline characters.
+ *
+ * <p>This source splits the data into records using {@code UTF-8} {@code \n}, {@code \r}, or {@code
+ * \r\n} as the delimiter. This source is not strict and supports decoding the last record even if
+ * it is not delimited. Finally, no records are decoded if the stream is empty.
+ *
+ * <p>This source supports reading from any arbitrary byte position within the stream. If the
+ * starting position is not {@code 0}, then bytes are skipped until the first delimiter is found
+ * representing the beginning of the first record to be decoded.
+ */
+@VisibleForTesting
+class ContextualTextIOSource extends FileBasedSource<RecordWithMetadata> {
+ byte[] delimiter;
+
+ private static final Logger LOG = LoggerFactory.getLogger(ContextualTextIOSource.class);
+
+ // Used to Override isSplittable
+ private boolean hasMultilineCSVRecords;
+
+ @Override
+ protected boolean isSplittable() throws Exception {
+ if (hasMultilineCSVRecords) {
+ // When Having Multiline CSV Records,
+ // Splitting the file may cause a split to be within a record,
+ // Disabling split prevents this from happening
+ return false;
+ }
+ return super.isSplittable();
+ }
+
+ ContextualTextIOSource(
+ ValueProvider<String> fileSpec,
+ EmptyMatchTreatment emptyMatchTreatment,
+ byte[] delimiter,
+ boolean hasMultilineCSVRecords) {
+ super(fileSpec, emptyMatchTreatment, 1L);
+ this.delimiter = delimiter;
+ this.hasMultilineCSVRecords = hasMultilineCSVRecords;
+ }
+
+ private ContextualTextIOSource(
+ MatchResult.Metadata metadata,
+ long start,
+ long end,
+ byte[] delimiter,
+ boolean hasMultilineCSVRecords) {
+ super(metadata, 1L, start, end);
+ this.delimiter = delimiter;
+ this.hasMultilineCSVRecords = hasMultilineCSVRecords;
+ }
+
+ @Override
+ protected FileBasedSource<RecordWithMetadata> createForSubrangeOfFile(
+ MatchResult.Metadata metadata, long start, long end) {
+ return new ContextualTextIOSource(metadata, start, end, delimiter, hasMultilineCSVRecords);
+ }
+
+ @Override
+ protected FileBasedReader<RecordWithMetadata> createSingleFileReader(PipelineOptions options) {
+ return new MultiLineTextBasedReader(this, delimiter, hasMultilineCSVRecords);
+ }
+
+ @Override
+ public Coder<RecordWithMetadata> getOutputCoder() {
+ SchemaCoder<RecordWithMetadata> coder = null;
+ try {
+ coder = SchemaRegistry.createDefault().getSchemaCoder(RecordWithMetadata.class);
+ } catch (NoSuchSchemaException e) {
+ LOG.error("No Coder Found for RecordWithMetadata");
+ }
+ return coder;
+ }
+
+ /**
+ * A {@link FileBasedReader FileBasedReader} which can decode records delimited by delimiter
+ * characters.
+ *
+ * <p>See {@link ContextualTextIOSource } for further details.
+ */
+ @VisibleForTesting
+ static class MultiLineTextBasedReader extends FileBasedReader<RecordWithMetadata> {
+ public static final int READ_BUFFER_SIZE = 8192;
+ private static final ByteString UTF8_BOM =
+ ByteString.copyFrom(new byte[] {(byte) 0xEF, (byte) 0xBB, (byte) 0xBF});
+ private final ByteBuffer readBuffer = ByteBuffer.allocate(READ_BUFFER_SIZE);
+ private ByteString buffer;
+ private int startOfDelimiterInBuffer;
+ private int endOfDelimiterInBuffer;
+ private long startOfRecord;
+ private volatile long startOfNextRecord;
+ private volatile boolean eof;
+ private volatile boolean elementIsPresent;
+ private @Nullable RecordWithMetadata currentValue;
+ private @Nullable ReadableByteChannel inChannel;
+ private byte @Nullable [] delimiter;
+
+ // Add to override the isSplittable
+ private boolean hasMultilineCSVRecords;
+
+ private long startingOffset;
+ private long totalRecordCount;
+
+ private MultiLineTextBasedReader(
+ ContextualTextIOSource source, byte[] delimiter, boolean hasMultilineCSVRecords) {
+ super(source);
+ buffer = ByteString.EMPTY;
+ this.delimiter = delimiter;
+ this.hasMultilineCSVRecords = hasMultilineCSVRecords;
+ startingOffset = getCurrentSource().getStartOffset(); // Start offset;
+ }
+
+ @Override
+ protected long getCurrentOffset() throws NoSuchElementException {
+ if (!elementIsPresent) {
+ throw new NoSuchElementException();
+ }
+ return startOfRecord;
+ }
+
+ @Override
+ public long getSplitPointsRemaining() {
+ if (isStarted() && startOfNextRecord >= getCurrentSource().getEndOffset()) {
+ return isDone() ? 0 : 1;
+ }
+ return super.getSplitPointsRemaining();
+ }
+
+ @Override
+ public RecordWithMetadata getCurrent() throws NoSuchElementException {
+ if (!elementIsPresent) {
+ throw new NoSuchElementException();
+ }
+ return currentValue;
+ }
+
+ @Override
+ protected void startReading(ReadableByteChannel channel) throws IOException {
+ this.inChannel = channel;
+ // If the first offset is greater than zero, we need to skip bytes until we see our
+ // first delimiter.
+ long startOffset = getCurrentSource().getStartOffset();
+ if (startOffset > 0) {
+ Preconditions.checkState(
+ channel instanceof SeekableByteChannel,
+ "%s only supports reading from a SeekableByteChannel when given a start offset"
+ + " greater than 0.",
+ ContextualTextIOSource.class.getSimpleName());
+ long requiredPosition = startOffset - 1;
+ if (delimiter != null && startOffset >= delimiter.length) {
+ // we need to move back the offset of at worse delimiter.size to be sure to see
+ // all the bytes of the delimiter in the call to findDelimiterBounds() below
+ requiredPosition = startOffset - delimiter.length;
+ }
+ ((SeekableByteChannel) channel).position(requiredPosition);
+ findDelimiterBoundsWithMultiLineCheck();
+ buffer = buffer.substring(endOfDelimiterInBuffer);
+ startOfNextRecord = requiredPosition + endOfDelimiterInBuffer;
+ endOfDelimiterInBuffer = 0;
+ startOfDelimiterInBuffer = 0;
+ }
+ }
+
+ private void findDelimiterBoundsWithMultiLineCheck() throws IOException {
+ findDelimiterBounds();
+ }
+
+ /**
+ * Locates the start position and end position of the next delimiter. Will consume the channel
+ * till either EOF or the delimiter bounds are found.
+ *
+ * <p>If {@link ContextualTextIOSource#hasMultilineCSVRecords} is set then the behaviour will
+ * change from the standard read seen in {@link org.apache.beam.sdk.io.TextIO}. The assumption
+ * when {@link ContextualTextIOSource#hasMultilineCSVRecords} is set is that the file is being
+ * read with a single thread.
+ *
+ * <p>This fills the buffer and updates the positions as follows:
+ *
+ * <pre>{@code
+ * ------------------------------------------------------
+ * | element bytes | delimiter bytes | unconsumed bytes |
+ * ------------------------------------------------------
+ * 0 start of end of buffer
+ * delimiter delimiter size
+ * in buffer in buffer
+ * }</pre>
+ */
+ private void findDelimiterBounds() throws IOException {
+ int bytePositionInBuffer = 0;
+ boolean doubleQuoteClosed = true;
+
+ while (true) {
+ if (!tryToEnsureNumberOfBytesInBuffer(bytePositionInBuffer + 1)) {
+ startOfDelimiterInBuffer = endOfDelimiterInBuffer = bytePositionInBuffer;
+ break;
+ }
+
+ byte currentByte = buffer.byteAt(bytePositionInBuffer);
+ if (hasMultilineCSVRecords) {
+ // Check if we are inside an open Quote
+ if (currentByte == '"') {
+ doubleQuoteClosed = !doubleQuoteClosed;
+ }
+ } else {
+ doubleQuoteClosed = true;
+ }
+
+ if (delimiter == null) {
+ // default delimiter
+ if (currentByte == '\n') {
+ startOfDelimiterInBuffer = bytePositionInBuffer;
+ endOfDelimiterInBuffer = startOfDelimiterInBuffer + 1;
+ if (doubleQuoteClosed) {
+ break;
+ }
+ } else if (currentByte == '\r') {
+ startOfDelimiterInBuffer = bytePositionInBuffer;
+ endOfDelimiterInBuffer = startOfDelimiterInBuffer + 1;
+ if (tryToEnsureNumberOfBytesInBuffer(bytePositionInBuffer + 2)) {
+ currentByte = buffer.byteAt(bytePositionInBuffer + 1);
+ if (currentByte == '\n') {
+ endOfDelimiterInBuffer += 1;
+ }
+ }
+ if (doubleQuoteClosed) {
+ break;
+ }
+ }
+ } else {
+ // when the user defines a delimiter
+ int i = 0;
+ startOfDelimiterInBuffer = endOfDelimiterInBuffer = bytePositionInBuffer;
+ while ((i < delimiter.length) && (currentByte == delimiter[i])) {
+ // read next byte;
+ i++;
+ if (tryToEnsureNumberOfBytesInBuffer(bytePositionInBuffer + i + 1)) {
+ currentByte = buffer.byteAt(bytePositionInBuffer + i);
+ } else {
+ // corner case: delimiter truncate at the end of file
+ startOfDelimiterInBuffer = endOfDelimiterInBuffer = bytePositionInBuffer;
+ break;
+ }
+ }
+ if (i == delimiter.length) {
+ endOfDelimiterInBuffer = bytePositionInBuffer + i;
+ if (doubleQuoteClosed) {
+ break;
+ }
+ }
+ }
+ bytePositionInBuffer += 1;
+ }
+ }
+
+ @Override
+ protected boolean readNextRecord() throws IOException {
+ startOfRecord = startOfNextRecord;
+
+ findDelimiterBoundsWithMultiLineCheck();
+
+ // If we have reached EOF file and consumed all of the buffer then we know
+ // that there are no more records.
+ if (eof && buffer.isEmpty()) {
+ elementIsPresent = false;
+ return false;
+ }
+
+ decodeCurrentElement();
+ startOfNextRecord = startOfRecord + endOfDelimiterInBuffer;
+ return true;
+ }
+
+ /**
+ * Decodes the current element updating the buffer to only contain the unconsumed bytes.
+ *
+ * <p>This invalidates the currently stored {@code startOfDelimiterInBuffer} and {@code
+ * endOfDelimiterInBuffer}.
+ */
+ private void decodeCurrentElement() throws IOException {
+ ByteString dataToDecode = buffer.substring(0, startOfDelimiterInBuffer);
+ // If present, the UTF8 Byte Order Mark (BOM) will be removed.
+ if (startOfRecord == 0 && dataToDecode.startsWith(UTF8_BOM)) {
+ dataToDecode = dataToDecode.substring(UTF8_BOM.size());
+ }
+
+ // The line num is:
+ Long recordUniqueNum = totalRecordCount++;
+ // The Complete FileName (with uri if this is a web url eg: temp/abc.txt) is:
+ String fileName = getCurrentSource().getSingleFileMetadata().resourceId().toString();
Review comment:
We should be passing the resourceId through and not pulling out the filename by itself.
##########
File path: sdks/java/io/contextual-text-io/src/main/java/org/apache/beam/sdk/io/contextualtextio/RecordWithMetadata.java
##########
@@ -0,0 +1,84 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one
+ * or more contributor license agreements. See the NOTICE file
+ * distributed with this work for additional information
+ * regarding copyright ownership. The ASF licenses this file
+ * to you under the Apache License, Version 2.0 (the
+ * "License"); you may not use this file except in compliance
+ * with the License. You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+package org.apache.beam.sdk.io.contextualtextio;
+
+import com.google.auto.value.AutoValue;
+import org.apache.beam.sdk.annotations.Experimental;
+import org.apache.beam.sdk.schemas.AutoValueSchema;
+import org.apache.beam.sdk.schemas.annotations.DefaultSchema;
+
+/**
+ * Helper Class based on {@link AutoValueSchema}, it provides Metadata associated with each Record
+ * when reading from file(s) using {@link ContextualTextIO}.
+ *
+ * <h3>Fields:</h3>
+ *
+ * <ul>
+ * <li>recordOffset: The offset of a record (the byte at which the record begins) in a file. This
+ * information can be useful if you wish to reconstruct the file. {@link
+ * RecordWithMetadata#getRecordOffset()}
+ * <li>recordNum: The ordinal number of the record in its file. {@link
+ * RecordWithMetadata#getRecordNum()}
+ * <li>recordValue: The value / contents of the record {@link RecordWithMetadata#getRecordValue()}
+ * <li>rangeOffset: The starting offset of the range (split), which contained the record, when the
+ * record was read. {@link RecordWithMetadata#getRangeOffset()}
+ * <li>recordNumInOffset: The record number relative to the Range. (line number within the range)
+ * {@link RecordWithMetadata#getRecordNumInOffset()}
+ * <li>fileName: Name of the file to which the record belongs (this is the full filename,
+ * eg:path/to/file.txt) {@link RecordWithMetadata#getFileName()}
+ * </ul>
+ */
+@Experimental(Experimental.Kind.SCHEMAS)
+@DefaultSchema(AutoValueSchema.class)
+@AutoValue
+public abstract class RecordWithMetadata {
+ public abstract Long getRecordOffset();
+
+ public abstract Long getRecordNum();
+
+ public abstract String getRecordValue();
Review comment:
nit: getRecordValue -> getValue
##########
File path: sdks/java/io/contextual-text-io/src/main/java/org/apache/beam/sdk/io/contextualtextio/ContextualTextIOSource.java
##########
@@ -0,0 +1,364 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one
+ * or more contributor license agreements. See the NOTICE file
+ * distributed with this work for additional information
+ * regarding copyright ownership. The ASF licenses this file
+ * to you under the Apache License, Version 2.0 (the
+ * "License"); you may not use this file except in compliance
+ * with the License. You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+package org.apache.beam.sdk.io.contextualtextio;
+
+import java.io.IOException;
+import java.nio.ByteBuffer;
+import java.nio.channels.ReadableByteChannel;
+import java.nio.channels.SeekableByteChannel;
+import java.util.NoSuchElementException;
+import org.apache.beam.sdk.coders.Coder;
+import org.apache.beam.sdk.io.FileBasedSource;
+import org.apache.beam.sdk.io.fs.EmptyMatchTreatment;
+import org.apache.beam.sdk.io.fs.MatchResult;
+import org.apache.beam.sdk.options.PipelineOptions;
+import org.apache.beam.sdk.options.ValueProvider;
+import org.apache.beam.sdk.schemas.NoSuchSchemaException;
+import org.apache.beam.sdk.schemas.SchemaCoder;
+import org.apache.beam.sdk.schemas.SchemaRegistry;
+import org.apache.beam.vendor.grpc.v1p26p0.com.google.protobuf.ByteString;
+import org.apache.beam.vendor.guava.v26_0_jre.com.google.common.annotations.VisibleForTesting;
+import org.apache.beam.vendor.guava.v26_0_jre.com.google.common.base.Preconditions;
+import org.checkerframework.checker.nullness.qual.Nullable;
+import org.slf4j.Logger;
+import org.slf4j.LoggerFactory;
+
+/**
+ * Implementation detail of {@link ContextualTextIO.Read}.
+ *
+ * <p>A {@link FileBasedSource} which can decode records delimited by newline characters.
+ *
+ * <p>This source splits the data into records using {@code UTF-8} {@code \n}, {@code \r}, or {@code
+ * \r\n} as the delimiter. This source is not strict and supports decoding the last record even if
+ * it is not delimited. Finally, no records are decoded if the stream is empty.
+ *
+ * <p>This source supports reading from any arbitrary byte position within the stream. If the
+ * starting position is not {@code 0}, then bytes are skipped until the first delimiter is found
+ * representing the beginning of the first record to be decoded.
+ */
+@VisibleForTesting
+class ContextualTextIOSource extends FileBasedSource<RecordWithMetadata> {
+ byte[] delimiter;
+
+ private static final Logger LOG = LoggerFactory.getLogger(ContextualTextIOSource.class);
+
+ // Used to Override isSplittable
+ private boolean hasMultilineCSVRecords;
+
+ @Override
+ protected boolean isSplittable() throws Exception {
+ if (hasMultilineCSVRecords) {
+ // When Having Multiline CSV Records,
+ // Splitting the file may cause a split to be within a record,
+ // Disabling split prevents this from happening
+ return false;
+ }
+ return super.isSplittable();
+ }
+
+ ContextualTextIOSource(
+ ValueProvider<String> fileSpec,
+ EmptyMatchTreatment emptyMatchTreatment,
+ byte[] delimiter,
+ boolean hasMultilineCSVRecords) {
+ super(fileSpec, emptyMatchTreatment, 1L);
+ this.delimiter = delimiter;
+ this.hasMultilineCSVRecords = hasMultilineCSVRecords;
+ }
+
+ private ContextualTextIOSource(
+ MatchResult.Metadata metadata,
+ long start,
+ long end,
+ byte[] delimiter,
+ boolean hasMultilineCSVRecords) {
+ super(metadata, 1L, start, end);
+ this.delimiter = delimiter;
+ this.hasMultilineCSVRecords = hasMultilineCSVRecords;
+ }
+
+ @Override
+ protected FileBasedSource<RecordWithMetadata> createForSubrangeOfFile(
+ MatchResult.Metadata metadata, long start, long end) {
+ return new ContextualTextIOSource(metadata, start, end, delimiter, hasMultilineCSVRecords);
+ }
+
+ @Override
+ protected FileBasedReader<RecordWithMetadata> createSingleFileReader(PipelineOptions options) {
+ return new MultiLineTextBasedReader(this, delimiter, hasMultilineCSVRecords);
+ }
+
+ @Override
+ public Coder<RecordWithMetadata> getOutputCoder() {
+ SchemaCoder<RecordWithMetadata> coder = null;
+ try {
+ coder = SchemaRegistry.createDefault().getSchemaCoder(RecordWithMetadata.class);
+ } catch (NoSuchSchemaException e) {
+ LOG.error("No Coder Found for RecordWithMetadata");
+ }
+ return coder;
+ }
+
+ /**
+ * A {@link FileBasedReader FileBasedReader} which can decode records delimited by delimiter
+ * characters.
+ *
+ * <p>See {@link ContextualTextIOSource } for further details.
Review comment:
```suggestion
* <p>See {@link ContextualTextIOSource} for further details.
```
##########
File path: sdks/java/io/contextual-text-io/src/main/java/org/apache/beam/sdk/io/contextualtextio/ContextualTextIO.java
##########
@@ -319,31 +326,169 @@ static boolean isSelfOverlapping(byte[] s) {
}
@Override
- public PCollection<String> expand(PBegin input) {
- checkNotNull(getFilepattern(), "need to set the filepattern of a TextIO.Read transform");
+ public PCollection<RecordWithMetadata> expand(PBegin input) {
+ checkNotNull(
+ getFilepattern(), "need to set the filepattern of a ContextualTextIO.Read transform");
+ PCollection<RecordWithMetadata> lines = null;
if (getMatchConfiguration().getWatchInterval() == null && !getHintMatchesManyFiles()) {
- return input.apply("Read", org.apache.beam.sdk.io.Read.from(getSource()));
+ lines = input.apply("Read", org.apache.beam.sdk.io.Read.from(getSource()));
+ } else {
+ // All other cases go through FileIO + ReadFiles
+ lines =
+ input
+ .apply(
+ "Create filepattern", Create.ofProvider(getFilepattern(), StringUtf8Coder.of()))
+ .apply("Match All", FileIO.matchAll().withConfiguration(getMatchConfiguration()))
+ .apply(
+ "Read Matches",
+ FileIO.readMatches()
+ .withCompression(getCompression())
+ .withDirectoryTreatment(DirectoryTreatment.PROHIBIT))
+ .apply("Via ReadFiles", readFiles().withDelimiter(getDelimiter()));
}
- // All other cases go through FileIO + ReadFiles
- return input
- .apply("Create filepattern", Create.ofProvider(getFilepattern(), StringUtf8Coder.of()))
- .apply("Match All", FileIO.matchAll().withConfiguration(getMatchConfiguration()))
- .apply(
- "Read Matches",
- FileIO.readMatches()
- .withCompression(getCompression())
- .withDirectoryTreatment(DirectoryTreatment.PROHIBIT))
- .apply("Via ReadFiles", readFiles().withDelimiter(getDelimiter()));
+ // Check if the user decided to opt out of recordNums associated with records
+ if (getWithoutLineNumMetadata()) {
+ return lines;
+ }
+
+ // At this point the line number in RecordWithMetadata contains the relative line offset from
Review comment:
I think using:
```
/*
*
*
*/
```
will allow formatting tools to reformat multi-line comments automatically.
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
[GitHub] [beam] abhiy13 commented on a change in pull request #12645: [BEAM-10124] Add ContextualTextIO
Posted by GitBox <gi...@apache.org>.
abhiy13 commented on a change in pull request #12645:
URL: https://github.com/apache/beam/pull/12645#discussion_r474444099
##########
File path: sdks/java/io/contextual-text-io/src/main/java/org/apache/beam/sdk/io/contextualtextio/ContextualTextIO.java
##########
@@ -0,0 +1,594 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one
+ * or more contributor license agreements. See the NOTICE file
+ * distributed with this work for additional information
+ * regarding copyright ownership. The ASF licenses this file
+ * to you under the Apache License, Version 2.0 (the
+ * "License"); you may not use this file except in compliance
+ * with the License. You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+package org.apache.beam.sdk.io.contextualtextio;
+
+import static org.apache.beam.sdk.io.FileIO.ReadMatches.DirectoryTreatment;
+import static org.apache.beam.vendor.guava.v26_0_jre.com.google.common.base.Preconditions.checkArgument;
+import static org.apache.beam.vendor.guava.v26_0_jre.com.google.common.base.Preconditions.checkNotNull;
+
+import com.google.auto.value.AutoValue;
+import java.nio.ByteBuffer;
+import java.util.Arrays;
+import java.util.Comparator;
+import java.util.HashMap;
+import java.util.Map;
+import java.util.SortedMap;
+import java.util.TreeMap;
+import org.apache.beam.sdk.coders.StringUtf8Coder;
+import org.apache.beam.sdk.io.CompressedSource;
+import org.apache.beam.sdk.io.Compression;
+import org.apache.beam.sdk.io.FileBasedSource;
+import org.apache.beam.sdk.io.FileIO;
+import org.apache.beam.sdk.io.FileIO.MatchConfiguration;
+import org.apache.beam.sdk.io.ReadAllViaFileBasedSource;
+import org.apache.beam.sdk.io.fs.EmptyMatchTreatment;
+import org.apache.beam.sdk.options.ValueProvider;
+import org.apache.beam.sdk.options.ValueProvider.StaticValueProvider;
+import org.apache.beam.sdk.schemas.NoSuchSchemaException;
+import org.apache.beam.sdk.schemas.SchemaCoder;
+import org.apache.beam.sdk.transforms.Count;
+import org.apache.beam.sdk.transforms.Create;
+import org.apache.beam.sdk.transforms.DoFn;
+import org.apache.beam.sdk.transforms.PTransform;
+import org.apache.beam.sdk.transforms.ParDo;
+import org.apache.beam.sdk.transforms.SerializableFunction;
+import org.apache.beam.sdk.transforms.View;
+import org.apache.beam.sdk.transforms.Watch.Growth.TerminationCondition;
+import org.apache.beam.sdk.transforms.display.DisplayData;
+import org.apache.beam.sdk.values.KV;
+import org.apache.beam.sdk.values.PBegin;
+import org.apache.beam.sdk.values.PCollection;
+import org.apache.beam.sdk.values.PCollectionView;
+import org.apache.beam.vendor.guava.v26_0_jre.com.google.common.annotations.VisibleForTesting;
+import org.checkerframework.checker.nullness.qual.Nullable;
+import org.joda.time.Duration;
+
+/**
+ * {@link PTransform}s that read text files and collect contextual information of the elements in
+ * the input.
+ *
+ * <h2>Reading from text files</h2>
+ *
+ * <p>To read a {@link PCollection} from one or more text files, use {@code
+ * ContextualTextIO.read()}. To instantiate a transform use {@link
+ * ContextualTextIO.Read#from(String)} and specify the path of the file(s) to be read.
+ * Alternatively, if the filenames to be read are themselves in a {@link PCollection} you can use
+ * {@link FileIO} to match them and {@link ContextualTextIO#readFiles()} to read them.
+ *
+ * <p>{@link #read} returns a {@link PCollection} of {@link RecordWithMetadata RecordWithMetadata},
+ * each corresponding to one line of an inout UTF-8 text file (split into lines delimited by '\n',
+ * '\r', '\r\n', or specified delimiter see {@link ContextualTextIO.Read#withDelimiter})
+ *
+ * <h3>Filepattern expansion and watching</h3>
+ *
+ * <p>By default, the filepatterns are expanded only once. The combination of {@link
+ * FileIO.Match#continuously(Duration, TerminationCondition)} and {@link #readFiles()} allow
+ * streaming of new files matching the filepattern(s).
+ *
+ * <p>By default, {@link #read} prohibits filepatterns that match no files, and {@link #readFiles()}
+ * allows them in case the filepattern contains a glob wildcard character. Use {@link
+ * ContextualTextIO.Read#withEmptyMatchTreatment} or {@link
+ * FileIO.Match#withEmptyMatchTreatment(EmptyMatchTreatment)} plus {@link #readFiles()} to configure
+ * this behavior.
+ *
+ * <p>Example 1: reading a file or filepattern.
+ *
+ * <pre>{@code
+ * Pipeline p = ...;
+ *
+ * // A simple Read of a file:
+ * PCollection<RecordWithMetadata> lines = p.apply(ContextualTextIO.read().from("/local/path/to/file.txt"));
+ * }</pre>
+ *
+ * <p>Example 2: reading a PCollection of filenames.
+ *
+ * <pre>{@code
+ * Pipeline p = ...;
+ *
+ * // E.g. the filenames might be computed from other data in the pipeline, or
+ * // read from a data source.
+ * PCollection<String> filenames = ...;
+ *
+ * // Read all files in the collection.
+ * PCollection<RecordWithMetadata> lines =
+ * filenames
+ * .apply(FileIO.matchAll())
+ * .apply(FileIO.readMatches())
+ * .apply(ContextualTextIO.readFiles());
+ * }</pre>
+ *
+ * <p>Example 3: streaming new files matching a filepattern.
+ *
+ * <pre>{@code
+ * Pipeline p = ...;
+ *
+ * PCollection<RecordWithMetadata> lines = p.apply(ContextualTextIO.read()
+ * .from("/local/path/to/files/*")
+ * .watchForNewFiles(
+ * // Check for new files every minute
+ * Duration.standardMinutes(1),
+ * // Stop watching the filepattern if no new files appear within an hour
+ * afterTimeSinceNewOutput(Duration.standardHours(1))));
+ * }</pre>
+ *
+ * <p>Example 4: reading a file or file pattern of RFC4180-compliant CSV files with fields that may
+ * contain line breaks.
+ *
+ * <pre>{@code
+ * Pipeline p = ...;
+ *
+ * PCollection<RecordWithMetadata> lines = p.apply(ContextualTextIO.read()
+ * .from("/local/path/to/files/*.csv")
+ * .withHasMultilineCSVRecords(true));
+ * }</pre>
+ *
+ * <p>Example 5: reading while watching for new files
+ *
+ * <pre>{@code
+ * Pipeline p = ...;
+ *
+ * PCollection<RecordWithMetadata> lines = p.apply(FileIO.match()
+ * .filepattern("filepattern")
+ * .continuously(
+ * Duration.millis(100),
+ * Watch.Growth.afterTimeSinceNewOutput(Duration.standardSeconds(3))))
+ * .apply(FileIO.readMatches())
+ * .apply(ContextualTextIO.readFiles());
+ * }</pre>
+ *
+ * NOTE: Using {@link ContextualTextIO.Read#withHasMultilineCSVRecords(Boolean)} introduces a
+ * performance penalty: when this option is enabled, the input cannot be split and read in parallel.
+ *
+ * <h3>Reading a very large number of files</h3>
+ *
+ * <p>If it is known that the filepattern will match a very large number of files (e.g. tens of
+ * thousands or more), use {@link ContextualTextIO.Read#withHintMatchesManyFiles} for better
+ * performance and scalability. Note that it may decrease performance if the filepattern matches
+ * only a small number of files.
+ */
+public class ContextualTextIO {
+ private static final long DEFAULT_BUNDLE_SIZE_BYTES = 64 * 1024 * 1024L;
+
+ /**
+ * A {@link PTransform} that reads from one or more text files and returns a bounded {@link
Review comment:
Added the example.
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
[GitHub] [beam] rezarokni commented on a change in pull request #12645: [BEAM-10124] Add ContextualTextIO
Posted by GitBox <gi...@apache.org>.
rezarokni commented on a change in pull request #12645:
URL: https://github.com/apache/beam/pull/12645#discussion_r492513409
##########
File path: sdks/java/io/contextual-text-io/src/main/java/org/apache/beam/sdk/io/contextualtextio/ContextualTextIOSource.java
##########
@@ -0,0 +1,364 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one
+ * or more contributor license agreements. See the NOTICE file
+ * distributed with this work for additional information
+ * regarding copyright ownership. The ASF licenses this file
+ * to you under the Apache License, Version 2.0 (the
+ * "License"); you may not use this file except in compliance
+ * with the License. You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+package org.apache.beam.sdk.io.contextualtextio;
+
+import java.io.IOException;
+import java.nio.ByteBuffer;
+import java.nio.channels.ReadableByteChannel;
+import java.nio.channels.SeekableByteChannel;
+import java.util.NoSuchElementException;
+import org.apache.beam.sdk.coders.Coder;
+import org.apache.beam.sdk.io.FileBasedSource;
+import org.apache.beam.sdk.io.fs.EmptyMatchTreatment;
+import org.apache.beam.sdk.io.fs.MatchResult;
+import org.apache.beam.sdk.options.PipelineOptions;
+import org.apache.beam.sdk.options.ValueProvider;
+import org.apache.beam.sdk.schemas.NoSuchSchemaException;
+import org.apache.beam.sdk.schemas.SchemaCoder;
+import org.apache.beam.sdk.schemas.SchemaRegistry;
+import org.apache.beam.vendor.grpc.v1p26p0.com.google.protobuf.ByteString;
+import org.apache.beam.vendor.guava.v26_0_jre.com.google.common.annotations.VisibleForTesting;
+import org.apache.beam.vendor.guava.v26_0_jre.com.google.common.base.Preconditions;
+import org.checkerframework.checker.nullness.qual.Nullable;
+import org.slf4j.Logger;
+import org.slf4j.LoggerFactory;
+
+/**
+ * Implementation detail of {@link ContextualTextIO.Read}.
+ *
+ * <p>A {@link FileBasedSource} which can decode records delimited by newline characters.
+ *
+ * <p>This source splits the data into records using {@code UTF-8} {@code \n}, {@code \r}, or {@code
+ * \r\n} as the delimiter. This source is not strict and supports decoding the last record even if
+ * it is not delimited. Finally, no records are decoded if the stream is empty.
+ *
+ * <p>This source supports reading from any arbitrary byte position within the stream. If the
+ * starting position is not {@code 0}, then bytes are skipped until the first delimiter is found
+ * representing the beginning of the first record to be decoded.
+ */
+@VisibleForTesting
+class ContextualTextIOSource extends FileBasedSource<RecordWithMetadata> {
+ byte[] delimiter;
+
+ private static final Logger LOG = LoggerFactory.getLogger(ContextualTextIOSource.class);
+
+ // Used to Override isSplittable
+ private boolean hasMultilineCSVRecords;
+
+ @Override
+ protected boolean isSplittable() throws Exception {
+ if (hasMultilineCSVRecords) {
+ // When Having Multiline CSV Records,
+ // Splitting the file may cause a split to be within a record,
+ // Disabling split prevents this from happening
+ return false;
+ }
+ return super.isSplittable();
+ }
+
+ ContextualTextIOSource(
+ ValueProvider<String> fileSpec,
+ EmptyMatchTreatment emptyMatchTreatment,
+ byte[] delimiter,
+ boolean hasMultilineCSVRecords) {
+ super(fileSpec, emptyMatchTreatment, 1L);
+ this.delimiter = delimiter;
+ this.hasMultilineCSVRecords = hasMultilineCSVRecords;
+ }
+
+ private ContextualTextIOSource(
+ MatchResult.Metadata metadata,
+ long start,
+ long end,
+ byte[] delimiter,
+ boolean hasMultilineCSVRecords) {
+ super(metadata, 1L, start, end);
+ this.delimiter = delimiter;
+ this.hasMultilineCSVRecords = hasMultilineCSVRecords;
+ }
+
+ @Override
+ protected FileBasedSource<RecordWithMetadata> createForSubrangeOfFile(
+ MatchResult.Metadata metadata, long start, long end) {
+ return new ContextualTextIOSource(metadata, start, end, delimiter, hasMultilineCSVRecords);
+ }
+
+ @Override
+ protected FileBasedReader<RecordWithMetadata> createSingleFileReader(PipelineOptions options) {
+ return new MultiLineTextBasedReader(this, delimiter, hasMultilineCSVRecords);
+ }
+
+ @Override
+ public Coder<RecordWithMetadata> getOutputCoder() {
+ SchemaCoder<RecordWithMetadata> coder = null;
+ try {
+ coder = SchemaRegistry.createDefault().getSchemaCoder(RecordWithMetadata.class);
+ } catch (NoSuchSchemaException e) {
+ LOG.error("No Coder Found for RecordWithMetadata");
+ }
+ return coder;
+ }
+
+ /**
+ * A {@link FileBasedReader FileBasedReader} which can decode records delimited by delimiter
+ * characters.
+ *
+ * <p>See {@link ContextualTextIOSource } for further details.
+ */
+ @VisibleForTesting
+ static class MultiLineTextBasedReader extends FileBasedReader<RecordWithMetadata> {
+ public static final int READ_BUFFER_SIZE = 8192;
+ private static final ByteString UTF8_BOM =
+ ByteString.copyFrom(new byte[] {(byte) 0xEF, (byte) 0xBB, (byte) 0xBF});
+ private final ByteBuffer readBuffer = ByteBuffer.allocate(READ_BUFFER_SIZE);
+ private ByteString buffer;
+ private int startOfDelimiterInBuffer;
+ private int endOfDelimiterInBuffer;
+ private long startOfRecord;
+ private volatile long startOfNextRecord;
+ private volatile boolean eof;
+ private volatile boolean elementIsPresent;
+ private @Nullable RecordWithMetadata currentValue;
+ private @Nullable ReadableByteChannel inChannel;
+ private byte @Nullable [] delimiter;
+
+ // Add to override the isSplittable
+ private boolean hasMultilineCSVRecords;
+
+ private long startingOffset;
+ private long totalRecordCount;
+
+ private MultiLineTextBasedReader(
+ ContextualTextIOSource source, byte[] delimiter, boolean hasMultilineCSVRecords) {
+ super(source);
+ buffer = ByteString.EMPTY;
+ this.delimiter = delimiter;
+ this.hasMultilineCSVRecords = hasMultilineCSVRecords;
+ startingOffset = getCurrentSource().getStartOffset(); // Start offset;
+ }
+
+ @Override
+ protected long getCurrentOffset() throws NoSuchElementException {
+ if (!elementIsPresent) {
+ throw new NoSuchElementException();
+ }
+ return startOfRecord;
+ }
+
+ @Override
+ public long getSplitPointsRemaining() {
+ if (isStarted() && startOfNextRecord >= getCurrentSource().getEndOffset()) {
+ return isDone() ? 0 : 1;
+ }
+ return super.getSplitPointsRemaining();
+ }
+
+ @Override
+ public RecordWithMetadata getCurrent() throws NoSuchElementException {
+ if (!elementIsPresent) {
+ throw new NoSuchElementException();
+ }
+ return currentValue;
+ }
+
+ @Override
+ protected void startReading(ReadableByteChannel channel) throws IOException {
+ this.inChannel = channel;
+ // If the first offset is greater than zero, we need to skip bytes until we see our
+ // first delimiter.
+ long startOffset = getCurrentSource().getStartOffset();
+ if (startOffset > 0) {
+ Preconditions.checkState(
+ channel instanceof SeekableByteChannel,
+ "%s only supports reading from a SeekableByteChannel when given a start offset"
+ + " greater than 0.",
+ ContextualTextIOSource.class.getSimpleName());
+ long requiredPosition = startOffset - 1;
+ if (delimiter != null && startOffset >= delimiter.length) {
+ // we need to move back the offset of at worse delimiter.size to be sure to see
+ // all the bytes of the delimiter in the call to findDelimiterBounds() below
+ requiredPosition = startOffset - delimiter.length;
+ }
+ ((SeekableByteChannel) channel).position(requiredPosition);
+ findDelimiterBoundsWithMultiLineCheck();
+ buffer = buffer.substring(endOfDelimiterInBuffer);
+ startOfNextRecord = requiredPosition + endOfDelimiterInBuffer;
+ endOfDelimiterInBuffer = 0;
+ startOfDelimiterInBuffer = 0;
+ }
+ }
+
+ private void findDelimiterBoundsWithMultiLineCheck() throws IOException {
+ findDelimiterBounds();
+ }
+
+ /**
+ * Locates the start position and end position of the next delimiter. Will consume the channel
+ * till either EOF or the delimiter bounds are found.
+ *
+ * <p>If {@link ContextualTextIOSource#hasMultilineCSVRecords} is set then the behaviour will
+ * change from the standard read seen in {@link org.apache.beam.sdk.io.TextIO}. The assumption
+ * when {@link ContextualTextIOSource#hasMultilineCSVRecords} is set is that the file is being
+ * read with a single thread.
+ *
+ * <p>This fills the buffer and updates the positions as follows:
+ *
+ * <pre>{@code
+ * ------------------------------------------------------
+ * | element bytes | delimiter bytes | unconsumed bytes |
+ * ------------------------------------------------------
+ * 0 start of end of buffer
+ * delimiter delimiter size
+ * in buffer in buffer
+ * }</pre>
+ */
+ private void findDelimiterBounds() throws IOException {
+ int bytePositionInBuffer = 0;
+ boolean doubleQuoteClosed = true;
+
+ while (true) {
+ if (!tryToEnsureNumberOfBytesInBuffer(bytePositionInBuffer + 1)) {
+ startOfDelimiterInBuffer = endOfDelimiterInBuffer = bytePositionInBuffer;
+ break;
+ }
+
+ byte currentByte = buffer.byteAt(bytePositionInBuffer);
+ if (hasMultilineCSVRecords) {
+ // Check if we are inside an open Quote
+ if (currentByte == '"') {
+ doubleQuoteClosed = !doubleQuoteClosed;
+ }
+ } else {
+ doubleQuoteClosed = true;
+ }
+
+ if (delimiter == null) {
+ // default delimiter
+ if (currentByte == '\n') {
+ startOfDelimiterInBuffer = bytePositionInBuffer;
+ endOfDelimiterInBuffer = startOfDelimiterInBuffer + 1;
+ if (doubleQuoteClosed) {
+ break;
+ }
+ } else if (currentByte == '\r') {
+ startOfDelimiterInBuffer = bytePositionInBuffer;
+ endOfDelimiterInBuffer = startOfDelimiterInBuffer + 1;
+ if (tryToEnsureNumberOfBytesInBuffer(bytePositionInBuffer + 2)) {
+ currentByte = buffer.byteAt(bytePositionInBuffer + 1);
+ if (currentByte == '\n') {
+ endOfDelimiterInBuffer += 1;
+ }
+ }
+ if (doubleQuoteClosed) {
+ break;
+ }
+ }
+ } else {
+ // when the user defines a delimiter
+ int i = 0;
+ startOfDelimiterInBuffer = endOfDelimiterInBuffer = bytePositionInBuffer;
+ while ((i < delimiter.length) && (currentByte == delimiter[i])) {
+ // read next byte;
+ i++;
+ if (tryToEnsureNumberOfBytesInBuffer(bytePositionInBuffer + i + 1)) {
+ currentByte = buffer.byteAt(bytePositionInBuffer + i);
+ } else {
+ // corner case: delimiter truncate at the end of file
+ startOfDelimiterInBuffer = endOfDelimiterInBuffer = bytePositionInBuffer;
+ break;
+ }
+ }
+ if (i == delimiter.length) {
+ endOfDelimiterInBuffer = bytePositionInBuffer + i;
+ if (doubleQuoteClosed) {
+ break;
+ }
+ }
+ }
+ bytePositionInBuffer += 1;
+ }
+ }
+
+ @Override
+ protected boolean readNextRecord() throws IOException {
+ startOfRecord = startOfNextRecord;
+
+ findDelimiterBoundsWithMultiLineCheck();
+
+ // If we have reached EOF file and consumed all of the buffer then we know
+ // that there are no more records.
+ if (eof && buffer.isEmpty()) {
+ elementIsPresent = false;
+ return false;
+ }
+
+ decodeCurrentElement();
+ startOfNextRecord = startOfRecord + endOfDelimiterInBuffer;
+ return true;
+ }
+
+ /**
+ * Decodes the current element updating the buffer to only contain the unconsumed bytes.
+ *
+ * <p>This invalidates the currently stored {@code startOfDelimiterInBuffer} and {@code
+ * endOfDelimiterInBuffer}.
+ */
+ private void decodeCurrentElement() throws IOException {
+ ByteString dataToDecode = buffer.substring(0, startOfDelimiterInBuffer);
+ // If present, the UTF8 Byte Order Mark (BOM) will be removed.
+ if (startOfRecord == 0 && dataToDecode.startsWith(UTF8_BOM)) {
+ dataToDecode = dataToDecode.substring(UTF8_BOM.size());
+ }
+
+ // The line num is:
+ Long recordUniqueNum = totalRecordCount++;
+ // The Complete FileName (with uri if this is a web url eg: temp/abc.txt) is:
+ String fileName = getCurrentSource().getSingleFileMetadata().resourceId().toString();
Review comment:
@abhiy13 I raised a PR in your repo with some fixes that includes changing from <RecordWithMetadata> to <Row>, do you have time to merge that PR into yours? The Java doc still needs to be updated.
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
[GitHub] [beam] rezarokni commented on a change in pull request #12645: [BEAM-10124] Add ContextualTextIO
Posted by GitBox <gi...@apache.org>.
rezarokni commented on a change in pull request #12645:
URL: https://github.com/apache/beam/pull/12645#discussion_r487933083
##########
File path: sdks/java/io/contextual-text-io/src/main/java/org/apache/beam/sdk/io/contextualtextio/ContextualTextIO.java
##########
@@ -0,0 +1,631 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one
+ * or more contributor license agreements. See the NOTICE file
+ * distributed with this work for additional information
+ * regarding copyright ownership. The ASF licenses this file
+ * to you under the Apache License, Version 2.0 (the
+ * "License"); you may not use this file except in compliance
+ * with the License. You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+package org.apache.beam.sdk.io.contextualtextio;
+
+import static org.apache.beam.sdk.io.FileIO.ReadMatches.DirectoryTreatment;
+import static org.apache.beam.vendor.guava.v26_0_jre.com.google.common.base.Preconditions.checkArgument;
+import static org.apache.beam.vendor.guava.v26_0_jre.com.google.common.base.Preconditions.checkNotNull;
+
+import com.google.auto.value.AutoValue;
+import java.nio.ByteBuffer;
+import java.util.Arrays;
+import java.util.Comparator;
+import java.util.HashMap;
+import java.util.Map;
+import java.util.SortedMap;
+import java.util.TreeMap;
+import org.apache.beam.sdk.coders.StringUtf8Coder;
+import org.apache.beam.sdk.io.CompressedSource;
+import org.apache.beam.sdk.io.Compression;
+import org.apache.beam.sdk.io.FileBasedSource;
+import org.apache.beam.sdk.io.FileIO;
+import org.apache.beam.sdk.io.FileIO.MatchConfiguration;
+import org.apache.beam.sdk.io.ReadAllViaFileBasedSource;
+import org.apache.beam.sdk.io.TextIO;
+import org.apache.beam.sdk.io.fs.EmptyMatchTreatment;
+import org.apache.beam.sdk.options.ValueProvider;
+import org.apache.beam.sdk.options.ValueProvider.StaticValueProvider;
+import org.apache.beam.sdk.schemas.NoSuchSchemaException;
+import org.apache.beam.sdk.schemas.SchemaCoder;
+import org.apache.beam.sdk.transforms.Count;
+import org.apache.beam.sdk.transforms.Create;
+import org.apache.beam.sdk.transforms.DoFn;
+import org.apache.beam.sdk.transforms.PTransform;
+import org.apache.beam.sdk.transforms.ParDo;
+import org.apache.beam.sdk.transforms.SerializableFunction;
+import org.apache.beam.sdk.transforms.View;
+import org.apache.beam.sdk.transforms.Watch.Growth.TerminationCondition;
+import org.apache.beam.sdk.transforms.display.DisplayData;
+import org.apache.beam.sdk.values.KV;
+import org.apache.beam.sdk.values.PBegin;
+import org.apache.beam.sdk.values.PCollection;
+import org.apache.beam.sdk.values.PCollectionView;
+import org.apache.beam.vendor.guava.v26_0_jre.com.google.common.annotations.VisibleForTesting;
+import org.checkerframework.checker.nullness.qual.Nullable;
+import org.joda.time.Duration;
+import org.slf4j.Logger;
+import org.slf4j.LoggerFactory;
+
+/**
+ * {@link PTransform}s that read text files and collect contextual information of the elements in
+ * the input.
+ *
+ * <p>Use {@link TextIO} when not reading file with Multiline Records or additional metadata is not
+ * required.
+ *
+ * <h2>Reading from text files</h2>
+ *
+ * <p>To read a {@link PCollection} from one or more text files, use {@code
+ * ContextualTextIO.read()}. To instantiate a transform use {@link
+ * ContextualTextIO.Read#from(String)} and specify the path of the file(s) to be read.
+ * Alternatively, if the filenames to be read are themselves in a {@link PCollection} you can use
+ * {@link FileIO} to match them and {@link ContextualTextIO#readFiles()} to read them.
+ *
+ * <p>{@link #read} returns a {@link PCollection} of {@link RecordWithMetadata RecordWithMetadata},
+ * each corresponding to one line of an input UTF-8 text file (split into lines delimited by '\n',
+ * '\r', '\r\n', or specified delimiter see {@link ContextualTextIO.Read#withDelimiter})
+ *
+ * <h3>Filepattern expansion and watching</h3>
+ *
+ * <p>By default, the filepatterns are expanded only once. The combination of {@link
+ * FileIO.Match#continuously(Duration, TerminationCondition)} and {@link #readFiles()} allow
+ * streaming of new files matching the filepattern(s).
+ *
+ * <p>By default, {@link #read} prohibits filepatterns that match no files, and {@link #readFiles()}
+ * allows them in case the filepattern contains a glob wildcard character. Use {@link
+ * ContextualTextIO.Read#withEmptyMatchTreatment} or {@link
+ * FileIO.Match#withEmptyMatchTreatment(EmptyMatchTreatment)} plus {@link #readFiles()} to configure
+ * this behavior.
+ *
+ * <p>Example 1: reading a file or filepattern.
+ *
+ * <pre>{@code
+ * Pipeline p = ...;
+ *
+ * // A simple Read of a file:
+ * PCollection<RecordWithMetadata> records = p.apply(ContextualTextIO.read().from("/local/path/to/file.txt"));
+ * }</pre>
+ *
+ * <p>Example 2: reading a PCollection of filenames.
+ *
+ * <pre>{@code
+ * Pipeline p = ...;
+ *
+ * // E.g. the filenames might be computed from other data in the pipeline, or
+ * // read from a data source.
+ * PCollection<String> filenames = ...;
+ *
+ * // Read all files in the collection.
+ * PCollection<RecordWithMetadata> records =
+ * filenames
+ * .apply(FileIO.matchAll())
+ * .apply(FileIO.readMatches())
+ * .apply(ContextualTextIO.readFiles());
+ * }</pre>
+ *
+ * <p>Example 3: streaming new files matching a filepattern.
+ *
+ * <pre>{@code
+ * Pipeline p = ...;
+ *
+ * PCollection<RecordWithMetadata> records = p.apply(ContextualTextIO.read()
+ * .from("/local/path/to/files/*")
+ * .watchForNewFiles(
+ * // Check for new files every minute
+ * Duration.standardMinutes(1),
+ * // Stop watching the filepattern if no new files appear within an hour
+ * afterTimeSinceNewOutput(Duration.standardHours(1))));
+ * }</pre>
+ *
+ * <p>Example 4: reading a file or file pattern of RFC4180-compliant CSV files with fields that may
+ * contain line breaks.
+ *
+ * Example of such a file could be:
+ *
+ * "aaa","b CRLF
+ * bb","ccc" CRLF
+ * zzz,yyy,xxx
+ *
+ * <pre>{@code
+ * Pipeline p = ...;
+ *
+ * PCollection<RecordWithMetadata> records = p.apply(ContextualTextIO.read()
+ * .from("/local/path/to/files/*.csv")
+ * .withHasMultilineCSVRecords(true));
+ * }</pre>
+ *
+ * <p>Example 5: reading while watching for new files
+ *
+ * <pre>{@code
+ * Pipeline p = ...;
+ *
+ * PCollection<RecordWithMetadata> records = p.apply(FileIO.match()
+ * .filepattern("filepattern")
+ * .continuously(
+ * Duration.millis(100),
+ * Watch.Growth.afterTimeSinceNewOutput(Duration.standardSeconds(3))))
+ * .apply(FileIO.readMatches())
+ * .apply(ContextualTextIO.readFiles());
+ * }</pre>
+ *
+ * <p>Example 6: reading without recordNum metadata, or only fileName associated Metadata. (the
+ * Objects would still contain recordNums, but these recordNums would correspond to their positions
+ * in their respective offsets rather than their positions within the entire file).
+ *
+ * <pre>{@code
+ * Pipeline p = ...;
+ *
+ * PCollection<RecordWithMetadata> records = p.apply(ContextualTextIO.read()
+ * .from("/local/path/to/files/*.csv")
+ * .setWithoutRecordNumMetadata(true));
+ * }</pre>
+ *
+ * <p>NOTE: When using {@link ContextualTextIO.Read#withHasMultilineCSVRecords(Boolean)} this
+ * option, a single reader will be used to process the file, rather than multiple readers which can
+ * read from different offsets. For a large file this can result in lower performance.
+ *
+ * <p>NOTE: Use {@link Read#withoutRecordNumMetadata()} when recordNum metadata is not required, for
+ * example, when when only filename metadata is required. Computing record positions currently
+ * introduces a shuffle step, which increases the resources used by the pipeline. <b> By default
+ * withoutRecordNumMetadata is set to false, so the shuffle step is performed.</b>
+ *
+ * <h3>Reading a very large number of files</h3>
+ *
+ * <p>If it is known that the filepattern will match a very large number of files (e.g. tens of
+ * thousands or more), use {@link ContextualTextIO.Read#withHintMatchesManyFiles} for better
+ * performance and scalability. Note that it may decrease performance if the filepattern matches
+ * only a small number of files.
+ */
+public class ContextualTextIO {
+ private static final long DEFAULT_BUNDLE_SIZE_BYTES = 64 * 1024 * 1024L;
+ private static final Logger LOG = LoggerFactory.getLogger(ContextualTextIO.class);
+
+ /**
+ * A {@link PTransform} that reads from one or more text files and returns a bounded {@link
+ * PCollection} containing one {@link RecordWithMetadata}element for each line of the input files.
+ */
+ public static Read read() {
+ return new AutoValue_ContextualTextIO_Read.Builder()
+ .setCompression(Compression.AUTO)
+ .setHintMatchesManyFiles(false)
+ .setWithoutRecordNumMetadata(false)
+ .setMatchConfiguration(MatchConfiguration.create(EmptyMatchTreatment.DISALLOW))
+ .setHasMultilineCSVRecords(false)
+ .build();
+ }
+
+ /**
+ * Like {@link #read}, but reads each file in a {@link PCollection} of {@link
+ * FileIO.ReadableFile}, returned by {@link FileIO#readMatches}.
+ */
+ public static ReadFiles readFiles() {
+ return new AutoValue_ContextualTextIO_ReadFiles.Builder()
+ // 64MB is a reasonable value that allows to amortize the cost of opening files,
+ // but is not so large as to exhaust a typical runner's maximum amount of output per
+ // ProcessElement call.
+ .setDesiredBundleSizeBytes(DEFAULT_BUNDLE_SIZE_BYTES)
+ .setHasMultilineCSVRecords(false)
+ .build();
+ }
+
+ /** Implementation of {@link #read}. */
+ @AutoValue
+ public abstract static class Read extends PTransform<PBegin, PCollection<RecordWithMetadata>> {
+ abstract @Nullable ValueProvider<String> getFilepattern();
+
+ abstract MatchConfiguration getMatchConfiguration();
+
+ abstract boolean getHintMatchesManyFiles();
+
+ abstract boolean getWithoutRecordNumMetadata();
+
+ abstract Compression getCompression();
+
+ abstract @Nullable Boolean getHasMultilineCSVRecords();
+
+ @SuppressWarnings("mutable") // this returns an array that can be mutated by the caller
+ abstract byte @Nullable [] getDelimiter();
+
+ abstract Builder toBuilder();
+
+ @AutoValue.Builder
+ abstract static class Builder {
+ abstract Builder setFilepattern(ValueProvider<String> filepattern);
+
+ abstract Builder setMatchConfiguration(MatchConfiguration matchConfiguration);
+
+ abstract Builder setHintMatchesManyFiles(boolean hintManyFiles);
+
+ abstract Builder setWithoutRecordNumMetadata(boolean withoutLineNumMetadata);
+
+ abstract Builder setCompression(Compression compression);
+
+ abstract Builder setDelimiter(byte @Nullable [] delimiter);
+
+ abstract Builder setHasMultilineCSVRecords(Boolean hasMultilineCSVRecords);
+
+ abstract Read build();
+ }
+
+ /**
+ * Reads text from the file(s) with the given filename or filename pattern.
+ *
+ * <p>This can be a local path (if running locally), or a Google Cloud Storage filename or
+ * filename pattern of the form {@code "gs://<bucket>/<filepath>"} (if running locally or using
+ * remote execution service).
+ *
+ * <p>Standard <a href="http://docs.oracle.com/javase/tutorial/essential/io/find.html" >Java
+ * Filesystem glob patterns</a> ("*", "?", "[..]") are supported.
+ *
+ * <p>If it is known that the filepattern will match a very large number of files (at least tens
+ * of thousands), use {@link #withHintMatchesManyFiles} for better performance and scalability.
+ */
+ public Read from(String filepattern) {
+ checkArgument(filepattern != null, "filepattern can not be null");
+ return from(StaticValueProvider.of(filepattern));
+ }
+
+ /** Same as {@code from(filepattern)}, but accepting a {@link ValueProvider}. */
+ public Read from(ValueProvider<String> filepattern) {
+ checkArgument(filepattern != null, "filepattern can not be null");
+ return toBuilder().setFilepattern(filepattern).build();
+ }
+
+ /** Sets the {@link MatchConfiguration}. */
+ public Read withMatchConfiguration(MatchConfiguration matchConfiguration) {
+ return toBuilder().setMatchConfiguration(matchConfiguration).build();
+ }
+
+ /**
+ * When reading RFC4180 CSV files that have values that span multiple lines, set this to true.
+ * Note: this reduces the read performance (see: {@link ContextualTextIO}).
+ */
+ public Read withHasMultilineCSVRecords(Boolean hasMultilineCSVRecords) {
+ return toBuilder().setHasMultilineCSVRecords(hasMultilineCSVRecords).build();
+ }
+
+ /**
+ * Reads from input sources using the specified compression type.
+ *
+ * <p>If no compression type is specified, the default is {@link Compression#AUTO}.
+ */
+ public Read withCompression(Compression compression) {
+ return toBuilder().setCompression(compression).build();
+ }
+
+ /**
+ * Hints that the filepattern specified in {@link #from(String)} matches a very large number of
+ * files.
+ *
+ * <p>This hint may cause a runner to execute the transform differently, in a way that improves
+ * performance for this case, but it may worsen performance if the filepattern matches only a
+ * small number of files (e.g., in a runner that supports dynamic work rebalancing, it will
+ * happen less efficiently within individual files).
+ */
+ public Read withHintMatchesManyFiles() {
+ return toBuilder().setHintMatchesManyFiles(true).build();
+ }
+
+ /**
+ * Allows the user to opt out of getting recordNums associated with each record.
+ *
+ * <p>When set to true, it will introduce a shuffle step to assemble the recordNums for each
+ * record, which will increase the resources used by the pipeline.
+ *
+ * <p>Use this when metadata like fileNames are required and their position/order can be
+ * ignored.
+ */
+ public Read withoutRecordNumMetadata() {
+ return toBuilder().setWithoutRecordNumMetadata(true).build();
+ }
+
+ /** See {@link MatchConfiguration#withEmptyMatchTreatment}. */
+ public Read withEmptyMatchTreatment(EmptyMatchTreatment treatment) {
+ return withMatchConfiguration(getMatchConfiguration().withEmptyMatchTreatment(treatment));
+ }
+
+ /** Set the custom delimiter to be used in place of the default ones ('\r', '\n' or '\r\n'). */
+ public Read withDelimiter(byte[] delimiter) {
+ checkArgument(delimiter != null, "delimiter can not be null");
+ checkArgument(!isSelfOverlapping(delimiter), "delimiter must not self-overlap");
+ return toBuilder().setDelimiter(delimiter).build();
+ }
+
+ static boolean isSelfOverlapping(byte[] s) {
+ // s self-overlaps if v exists such as s = vu = wv with u and w non empty
+ for (int i = 1; i < s.length - 1; ++i) {
+ if (ByteBuffer.wrap(s, 0, i).equals(ByteBuffer.wrap(s, s.length - i, i))) {
+ return true;
+ }
+ }
+ return false;
+ }
+
+ @Override
+ public PCollection<RecordWithMetadata> expand(PBegin input) {
+ checkNotNull(
+ getFilepattern(), "need to set the filepattern of a ContextualTextIO.Read transform");
+ PCollection<RecordWithMetadata> records = null;
+ if (getMatchConfiguration().getWatchInterval() == null && !getHintMatchesManyFiles()) {
+ records = input.apply("Read", org.apache.beam.sdk.io.Read.from(getSource()));
+ } else {
+ // All other cases go through FileIO + ReadFiles
+ records =
+ input
+ .apply(
+ "Create filepattern", Create.ofProvider(getFilepattern(), StringUtf8Coder.of()))
+ .apply("Match All", FileIO.matchAll().withConfiguration(getMatchConfiguration()))
+ .apply(
+ "Read Matches",
+ FileIO.readMatches()
+ .withCompression(getCompression())
+ .withDirectoryTreatment(DirectoryTreatment.PROHIBIT))
+ .apply("Via ReadFiles", readFiles().withDelimiter(getDelimiter()));
+ }
+
+ // Check if the user decided to opt out of recordNums associated with records
+ if (getWithoutRecordNumMetadata()) {
+ return records;
+ }
+
+ /*
+ * At this point the line number in RecordWithMetadata contains the relative line offset from the beginning of the read range.
+ *
+ * To compute the absolute position from the beginning of the input we group the lines within the same ranges, and evaluate the size of each range.
+ */
+
+ PCollection<KV<KV<String, Long>, RecordWithMetadata>> recordsGroupedByFileAndRange =
+ records.apply("AddFileNameAndRange", ParDo.of(new AddFileNameAndRange()));
+
+ PCollectionView<Map<KV<String, Long>, Long>> rangeSizes =
+ recordsGroupedByFileAndRange
+ .apply("CountRecordsForEachFileRange", Count.perKey())
+ .apply("SizesAsView", View.asMap());
+
+ // Get Pipeline to create a dummy PCollection with one element to help compute the lines
+ // before each Range
+ PCollection<Integer> singletonPcoll =
+ input.getPipeline().apply("CreateSingletonPcoll", Create.of(Arrays.asList(1)));
+
+ /*
+ * For each (File, Offset) pair, calculate the number of lines occurring before the Range for each file
+ *
+ * After computing the number of lines before each range, we can find the line number in original file as numLiesBeforeOffset + lineNumInCurrentOffset
+ */
+
+ PCollectionView<Map<KV<String, Long>, Long>> numRecordsBeforeEachRange =
+ singletonPcoll
+ .apply(
+ "ComputeRecordsBeforeRange",
+ ParDo.of(new ComputeRecordsBeforeEachRange(rangeSizes))
+ .withSideInputs(rangeSizes))
+ .apply("NumRecordsBeforeEachRangeAsView", View.asMap());
+
+ return recordsGroupedByFileAndRange.apply(
+ "AssignLineNums",
+ ParDo.of(new AssignRecordNums(numRecordsBeforeEachRange))
+ .withSideInputs(numRecordsBeforeEachRange));
+ }
+
+ @VisibleForTesting
+ static class AddFileNameAndRange
+ extends DoFn<RecordWithMetadata, KV<KV<String, Long>, RecordWithMetadata>> {
+ @ProcessElement
+ public void processElement(
+ @Element RecordWithMetadata record,
+ OutputReceiver<KV<KV<String, Long>, RecordWithMetadata>> out) {
+ out.output(KV.of(KV.of(record.getFileName().toString(), record.getRangeOffset()), record));
+ }
+ }
+
+ /**
+ * Helper class for computing number of record in the File preceding the beginning of the Range
+ * in this file.
+ */
+ @VisibleForTesting
+ static class ComputeRecordsBeforeEachRange extends DoFn<Integer, KV<KV<String, Long>, Long>> {
+ private final PCollectionView<Map<KV<String, Long>, Long>> rangeSizes;
+
+ public ComputeRecordsBeforeEachRange(
+ PCollectionView<Map<KV<String, Long>, Long>> rangeSizes) {
+ this.rangeSizes = rangeSizes;
+ }
+
+ // Add custom comparator as KV<K, V> is not comparable by default
+ private static class FileRangeComparator<K extends Comparable<K>, V extends Comparable<V>>
+ implements Comparator<KV<K, V>> {
+ @Override
+ public int compare(KV<K, V> a, KV<K, V> b) {
+ if (a.getKey().compareTo(b.getKey()) == 0) {
+ return a.getValue().compareTo(b.getValue());
+ }
+ return a.getKey().compareTo(b.getKey());
+ }
+ }
+
+ @ProcessElement
+ public void processElement(ProcessContext p) {
+ // Get the Map Containing the size from side-input
+ Map<KV<String, Long>, Long> rangeSizesMap = p.sideInput(rangeSizes);
+
+ // The FileRange Pair must be sorted
+ SortedMap<KV<String, Long>, Long> sorted = new TreeMap<>(new FileRangeComparator<>());
+
+ // Initialize sorted map with values
+ for (Map.Entry<KV<String, Long>, Long> entry : rangeSizesMap.entrySet()) {
+ sorted.put(entry.getKey(), entry.getValue());
+ }
+
+ // HashMap that tracks number of records passed for each file
+ Map<String, Long> pastRecords = new HashMap<>();
+
+ // For each (File, Range) Pair, compute the number of records before it
+ for (Map.Entry entry : sorted.entrySet()) {
+ Long numRecords = (long) entry.getValue();
+ KV<String, Long> fileRange = (KV<String, Long>) entry.getKey();
+ String file = fileRange.getKey();
+ Long numRecordsBefore = 0L;
+ if (pastRecords.containsKey(file)) {
+ numRecordsBefore = pastRecords.get(file);
+ }
+ p.output(KV.of(fileRange, numRecordsBefore));
+ pastRecords.put(file, numRecordsBefore + numRecords);
+ }
+ }
+ }
+
+ static class AssignRecordNums
+ extends DoFn<KV<KV<String, Long>, RecordWithMetadata>, RecordWithMetadata> {
+ PCollectionView<Map<KV<String, Long>, Long>> numRecordsBeforeEachRange;
+
+ public AssignRecordNums(
+ PCollectionView<Map<KV<String, Long>, Long>> numRecordsBeforeEachRange) {
+ this.numRecordsBeforeEachRange = numRecordsBeforeEachRange;
+ }
+
+ @ProcessElement
+ public void processElement(ProcessContext p) {
+ Long range = p.element().getKey().getValue();
+ String file = p.element().getKey().getKey();
+ RecordWithMetadata record = p.element().getValue();
+ Long numRecordsLessThanThisRange =
+ p.sideInput(numRecordsBeforeEachRange).get(KV.of(file, range));
+ RecordWithMetadata newLine =
+ RecordWithMetadata.newBuilder()
Review comment:
@abhiy13 You can make a builder from the record object itself, saves having to copy everything over. You can just mutate the bit you need.
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
[GitHub] [beam] tvalentyn commented on pull request #12645: [BEAM-10124] Add ContextualTextIO
Posted by GitBox <gi...@apache.org>.
tvalentyn commented on pull request #12645:
URL: https://github.com/apache/beam/pull/12645#issuecomment-698059215
Let's continue the discussion on #12924 which includes commits from https://github.com/apache/beam/pull/12645 and https://github.com/abhiy13/beam/pull/2.
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
[GitHub] [beam] codecov[bot] commented on pull request #12645: [BEAM-10124] Add ContextualTextIO
Posted by GitBox <gi...@apache.org>.
codecov[bot] commented on pull request #12645:
URL: https://github.com/apache/beam/pull/12645#issuecomment-688630083
# [Codecov](https://codecov.io/gh/apache/beam/pull/12645?src=pr&el=h1) Report
> :exclamation: No coverage uploaded for pull request base (`master@2b2b8e7`). [Click here to learn what that means](https://docs.codecov.io/docs/error-reference#section-missing-base-commit).
> The diff coverage is `n/a`.
[](https://codecov.io/gh/apache/beam/pull/12645?src=pr&el=tree)
```diff
@@ Coverage Diff @@
## master #12645 +/- ##
=========================================
Coverage ? 34.47%
=========================================
Files ? 684
Lines ? 81483
Branches ? 9180
=========================================
Hits ? 28090
Misses ? 52972
Partials ? 421
```
| [Impacted Files](https://codecov.io/gh/apache/beam/pull/12645?src=pr&el=tree) | Coverage Δ | |
|---|---|---|
| [transforms/external.py](https://codecov.io/gh/apache/beam/pull/12645/diff?src=pr&el=tree#diff-dHJhbnNmb3Jtcy9leHRlcm5hbC5weQ==) | `32.03% <0.00%> (ø)` | |
| [io/textio.py](https://codecov.io/gh/apache/beam/pull/12645/diff?src=pr&el=tree#diff-aW8vdGV4dGlvLnB5) | `30.69% <0.00%> (ø)` | |
| [internal/module\_test.py](https://codecov.io/gh/apache/beam/pull/12645/diff?src=pr&el=tree#diff-aW50ZXJuYWwvbW9kdWxlX3Rlc3QucHk=) | `67.64% <0.00%> (ø)` | |
| [testing/benchmarks/chicago\_taxi/trainer/task.py](https://codecov.io/gh/apache/beam/pull/12645/diff?src=pr&el=tree#diff-dGVzdGluZy9iZW5jaG1hcmtzL2NoaWNhZ29fdGF4aS90cmFpbmVyL3Rhc2sucHk=) | `0.00% <0.00%> (ø)` | |
| [examples/snippets/transforms/aggregation/latest.py](https://codecov.io/gh/apache/beam/pull/12645/diff?src=pr&el=tree#diff-ZXhhbXBsZXMvc25pcHBldHMvdHJhbnNmb3Jtcy9hZ2dyZWdhdGlvbi9sYXRlc3QucHk=) | `20.00% <0.00%> (ø)` | |
| [transforms/deduplicate\_test.py](https://codecov.io/gh/apache/beam/pull/12645/diff?src=pr&el=tree#diff-dHJhbnNmb3Jtcy9kZWR1cGxpY2F0ZV90ZXN0LnB5) | `49.05% <0.00%> (ø)` | |
| [testing/benchmarks/nexmark/nexmark\_launcher.py](https://codecov.io/gh/apache/beam/pull/12645/diff?src=pr&el=tree#diff-dGVzdGluZy9iZW5jaG1hcmtzL25leG1hcmsvbmV4bWFya19sYXVuY2hlci5weQ==) | `0.00% <0.00%> (ø)` | |
| [portability/utils.py](https://codecov.io/gh/apache/beam/pull/12645/diff?src=pr&el=tree#diff-cG9ydGFiaWxpdHkvdXRpbHMucHk=) | `71.42% <0.00%> (ø)` | |
| [transforms/\_\_init\_\_.py](https://codecov.io/gh/apache/beam/pull/12645/diff?src=pr&el=tree#diff-dHJhbnNmb3Jtcy9fX2luaXRfXy5weQ==) | `100.00% <0.00%> (ø)` | |
| [runners/direct/sdf\_direct\_runner.py](https://codecov.io/gh/apache/beam/pull/12645/diff?src=pr&el=tree#diff-cnVubmVycy9kaXJlY3Qvc2RmX2RpcmVjdF9ydW5uZXIucHk=) | `33.19% <0.00%> (ø)` | |
| ... and [674 more](https://codecov.io/gh/apache/beam/pull/12645/diff?src=pr&el=tree-more) | |
------
[Continue to review full report at Codecov](https://codecov.io/gh/apache/beam/pull/12645?src=pr&el=continue).
> **Legend** - [Click here to learn more](https://docs.codecov.io/docs/codecov-delta)
> `Δ = absolute <relative> (impact)`, `ø = not affected`, `? = missing data`
> Powered by [Codecov](https://codecov.io/gh/apache/beam/pull/12645?src=pr&el=footer). Last update [2b2b8e7...7a096dd](https://codecov.io/gh/apache/beam/pull/12645?src=pr&el=lastupdated). Read the [comment docs](https://docs.codecov.io/docs/pull-request-comments).
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
[GitHub] [beam] abhiy13 commented on a change in pull request #12645: [BEAM-10124] Add ContextualTextIO
Posted by GitBox <gi...@apache.org>.
abhiy13 commented on a change in pull request #12645:
URL: https://github.com/apache/beam/pull/12645#discussion_r492480914
##########
File path: sdks/java/io/contextual-text-io/src/main/java/org/apache/beam/sdk/io/contextualtextio/ContextualTextIOSource.java
##########
@@ -0,0 +1,364 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one
+ * or more contributor license agreements. See the NOTICE file
+ * distributed with this work for additional information
+ * regarding copyright ownership. The ASF licenses this file
+ * to you under the Apache License, Version 2.0 (the
+ * "License"); you may not use this file except in compliance
+ * with the License. You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+package org.apache.beam.sdk.io.contextualtextio;
+
+import java.io.IOException;
+import java.nio.ByteBuffer;
+import java.nio.channels.ReadableByteChannel;
+import java.nio.channels.SeekableByteChannel;
+import java.util.NoSuchElementException;
+import org.apache.beam.sdk.coders.Coder;
+import org.apache.beam.sdk.io.FileBasedSource;
+import org.apache.beam.sdk.io.fs.EmptyMatchTreatment;
+import org.apache.beam.sdk.io.fs.MatchResult;
+import org.apache.beam.sdk.options.PipelineOptions;
+import org.apache.beam.sdk.options.ValueProvider;
+import org.apache.beam.sdk.schemas.NoSuchSchemaException;
+import org.apache.beam.sdk.schemas.SchemaCoder;
+import org.apache.beam.sdk.schemas.SchemaRegistry;
+import org.apache.beam.vendor.grpc.v1p26p0.com.google.protobuf.ByteString;
+import org.apache.beam.vendor.guava.v26_0_jre.com.google.common.annotations.VisibleForTesting;
+import org.apache.beam.vendor.guava.v26_0_jre.com.google.common.base.Preconditions;
+import org.checkerframework.checker.nullness.qual.Nullable;
+import org.slf4j.Logger;
+import org.slf4j.LoggerFactory;
+
+/**
+ * Implementation detail of {@link ContextualTextIO.Read}.
+ *
+ * <p>A {@link FileBasedSource} which can decode records delimited by newline characters.
+ *
+ * <p>This source splits the data into records using {@code UTF-8} {@code \n}, {@code \r}, or {@code
+ * \r\n} as the delimiter. This source is not strict and supports decoding the last record even if
+ * it is not delimited. Finally, no records are decoded if the stream is empty.
+ *
+ * <p>This source supports reading from any arbitrary byte position within the stream. If the
+ * starting position is not {@code 0}, then bytes are skipped until the first delimiter is found
+ * representing the beginning of the first record to be decoded.
+ */
+@VisibleForTesting
+class ContextualTextIOSource extends FileBasedSource<RecordWithMetadata> {
+ byte[] delimiter;
+
+ private static final Logger LOG = LoggerFactory.getLogger(ContextualTextIOSource.class);
+
+ // Used to Override isSplittable
+ private boolean hasMultilineCSVRecords;
+
+ @Override
+ protected boolean isSplittable() throws Exception {
+ if (hasMultilineCSVRecords) {
+ // When Having Multiline CSV Records,
+ // Splitting the file may cause a split to be within a record,
+ // Disabling split prevents this from happening
+ return false;
+ }
+ return super.isSplittable();
+ }
+
+ ContextualTextIOSource(
+ ValueProvider<String> fileSpec,
+ EmptyMatchTreatment emptyMatchTreatment,
+ byte[] delimiter,
+ boolean hasMultilineCSVRecords) {
+ super(fileSpec, emptyMatchTreatment, 1L);
+ this.delimiter = delimiter;
+ this.hasMultilineCSVRecords = hasMultilineCSVRecords;
+ }
+
+ private ContextualTextIOSource(
+ MatchResult.Metadata metadata,
+ long start,
+ long end,
+ byte[] delimiter,
+ boolean hasMultilineCSVRecords) {
+ super(metadata, 1L, start, end);
+ this.delimiter = delimiter;
+ this.hasMultilineCSVRecords = hasMultilineCSVRecords;
+ }
+
+ @Override
+ protected FileBasedSource<RecordWithMetadata> createForSubrangeOfFile(
+ MatchResult.Metadata metadata, long start, long end) {
+ return new ContextualTextIOSource(metadata, start, end, delimiter, hasMultilineCSVRecords);
+ }
+
+ @Override
+ protected FileBasedReader<RecordWithMetadata> createSingleFileReader(PipelineOptions options) {
+ return new MultiLineTextBasedReader(this, delimiter, hasMultilineCSVRecords);
+ }
+
+ @Override
+ public Coder<RecordWithMetadata> getOutputCoder() {
Review comment:
Done
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
[GitHub] [beam] lukecwik commented on a change in pull request #12645: [BEAM-10124] Add ContextualTextIO
Posted by GitBox <gi...@apache.org>.
lukecwik commented on a change in pull request #12645:
URL: https://github.com/apache/beam/pull/12645#discussion_r477514368
##########
File path: sdks/java/io/contextual-text-io/src/main/java/org/apache/beam/sdk/io/contextualtextio/ContextualTextIOSource.java
##########
@@ -0,0 +1,364 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one
+ * or more contributor license agreements. See the NOTICE file
+ * distributed with this work for additional information
+ * regarding copyright ownership. The ASF licenses this file
+ * to you under the Apache License, Version 2.0 (the
+ * "License"); you may not use this file except in compliance
+ * with the License. You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+package org.apache.beam.sdk.io.contextualtextio;
+
+import java.io.IOException;
+import java.nio.ByteBuffer;
+import java.nio.channels.ReadableByteChannel;
+import java.nio.channels.SeekableByteChannel;
+import java.util.NoSuchElementException;
+import org.apache.beam.sdk.coders.Coder;
+import org.apache.beam.sdk.io.FileBasedSource;
+import org.apache.beam.sdk.io.fs.EmptyMatchTreatment;
+import org.apache.beam.sdk.io.fs.MatchResult;
+import org.apache.beam.sdk.options.PipelineOptions;
+import org.apache.beam.sdk.options.ValueProvider;
+import org.apache.beam.sdk.schemas.NoSuchSchemaException;
+import org.apache.beam.sdk.schemas.SchemaCoder;
+import org.apache.beam.sdk.schemas.SchemaRegistry;
+import org.apache.beam.vendor.grpc.v1p26p0.com.google.protobuf.ByteString;
+import org.apache.beam.vendor.guava.v26_0_jre.com.google.common.annotations.VisibleForTesting;
+import org.apache.beam.vendor.guava.v26_0_jre.com.google.common.base.Preconditions;
+import org.checkerframework.checker.nullness.qual.Nullable;
+import org.slf4j.Logger;
+import org.slf4j.LoggerFactory;
+
+/**
+ * Implementation detail of {@link ContextualTextIO.Read}.
+ *
+ * <p>A {@link FileBasedSource} which can decode records delimited by newline characters.
+ *
+ * <p>This source splits the data into records using {@code UTF-8} {@code \n}, {@code \r}, or {@code
+ * \r\n} as the delimiter. This source is not strict and supports decoding the last record even if
+ * it is not delimited. Finally, no records are decoded if the stream is empty.
+ *
+ * <p>This source supports reading from any arbitrary byte position within the stream. If the
+ * starting position is not {@code 0}, then bytes are skipped until the first delimiter is found
+ * representing the beginning of the first record to be decoded.
+ */
+@VisibleForTesting
+class ContextualTextIOSource extends FileBasedSource<RecordWithMetadata> {
Review comment:
Typically yes. The swap to using a splittable DoFn shouldn't be onerous from what I see from the implementation.
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
[GitHub] [beam] abhiy13 commented on a change in pull request #12645: [BEAM-10124] Add ContextualTextIO
Posted by GitBox <gi...@apache.org>.
abhiy13 commented on a change in pull request #12645:
URL: https://github.com/apache/beam/pull/12645#discussion_r475174600
##########
File path: sdks/java/io/contextual-text-io/src/main/java/org/apache/beam/sdk/io/contextualtextio/Range.java
##########
@@ -0,0 +1,42 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one
+ * or more contributor license agreements. See the NOTICE file
+ * distributed with this work for additional information
+ * regarding copyright ownership. The ASF licenses this file
+ * to you under the Apache License, Version 2.0 (the
+ * "License"); you may not use this file except in compliance
+ * with the License. You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+package org.apache.beam.sdk.io.contextualtextio;
+
+import com.google.auto.value.AutoValue;
+
+@AutoValue
+public abstract class Range {
+ public abstract Long getRangeNum();
+
+ public abstract Long getRangeLineNum();
Review comment:
Added docstring and more meaningful naming.
##########
File path: sdks/java/io/contextual-text-io/src/main/java/org/apache/beam/sdk/io/contextualtextio/Range.java
##########
@@ -0,0 +1,42 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one
+ * or more contributor license agreements. See the NOTICE file
+ * distributed with this work for additional information
+ * regarding copyright ownership. The ASF licenses this file
+ * to you under the Apache License, Version 2.0 (the
+ * "License"); you may not use this file except in compliance
+ * with the License. You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+package org.apache.beam.sdk.io.contextualtextio;
+
+import com.google.auto.value.AutoValue;
+
+@AutoValue
+public abstract class Range {
+ public abstract Long getRangeNum();
Review comment:
Added docstring and more meaningful naming.
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
[GitHub] [beam] codecov[bot] edited a comment on pull request #12645: [BEAM-10124] Add ContextualTextIO
Posted by GitBox <gi...@apache.org>.
codecov[bot] edited a comment on pull request #12645:
URL: https://github.com/apache/beam/pull/12645#issuecomment-688630083
# [Codecov](https://codecov.io/gh/apache/beam/pull/12645?src=pr&el=h1) Report
> :exclamation: No coverage uploaded for pull request base (`master@2b2b8e7`). [Click here to learn what that means](https://docs.codecov.io/docs/error-reference#section-missing-base-commit).
> The diff coverage is `n/a`.
[](https://codecov.io/gh/apache/beam/pull/12645?src=pr&el=tree)
```diff
@@ Coverage Diff @@
## master #12645 +/- ##
=========================================
Coverage ? 82.33%
=========================================
Files ? 453
Lines ? 54054
Branches ? 0
=========================================
Hits ? 44506
Misses ? 9548
Partials ? 0
```
| [Impacted Files](https://codecov.io/gh/apache/beam/pull/12645?src=pr&el=tree) | Coverage Δ | |
|---|---|---|
| [sdks/python/apache\_beam/io/jdbc.py](https://codecov.io/gh/apache/beam/pull/12645/diff?src=pr&el=tree#diff-c2Rrcy9weXRob24vYXBhY2hlX2JlYW0vaW8vamRiYy5weQ==) | `86.36% <0.00%> (ø)` | |
| [...ache\_beam/portability/api/beam\_artifact\_api\_pb2.py](https://codecov.io/gh/apache/beam/pull/12645/diff?src=pr&el=tree#diff-c2Rrcy9weXRob24vYXBhY2hlX2JlYW0vcG9ydGFiaWxpdHkvYXBpL2JlYW1fYXJ0aWZhY3RfYXBpX3BiMi5weQ==) | `100.00% <0.00%> (ø)` | |
| [...ython/apache\_beam/io/gcp/datastore/v1new/helper.py](https://codecov.io/gh/apache/beam/pull/12645/diff?src=pr&el=tree#diff-c2Rrcy9weXRob24vYXBhY2hlX2JlYW0vaW8vZ2NwL2RhdGFzdG9yZS92MW5ldy9oZWxwZXIucHk=) | `86.20% <0.00%> (ø)` | |
| [.../python/apache\_beam/testing/benchmarks/\_\_init\_\_.py](https://codecov.io/gh/apache/beam/pull/12645/diff?src=pr&el=tree#diff-c2Rrcy9weXRob24vYXBhY2hlX2JlYW0vdGVzdGluZy9iZW5jaG1hcmtzL19faW5pdF9fLnB5) | `0.00% <0.00%> (ø)` | |
| [sdks/python/apache\_beam/examples/avro\_bitcoin.py](https://codecov.io/gh/apache/beam/pull/12645/diff?src=pr&el=tree#diff-c2Rrcy9weXRob24vYXBhY2hlX2JlYW0vZXhhbXBsZXMvYXZyb19iaXRjb2luLnB5) | `0.00% <0.00%> (ø)` | |
| [...beam/testing/benchmarks/nexmark/models/\_\_init\_\_.py](https://codecov.io/gh/apache/beam/pull/12645/diff?src=pr&el=tree#diff-c2Rrcy9weXRob24vYXBhY2hlX2JlYW0vdGVzdGluZy9iZW5jaG1hcmtzL25leG1hcmsvbW9kZWxzL19faW5pdF9fLnB5) | `0.00% <0.00%> (ø)` | |
| [...hon/apache\_beam/examples/wordcount\_with\_metrics.py](https://codecov.io/gh/apache/beam/pull/12645/diff?src=pr&el=tree#diff-c2Rrcy9weXRob24vYXBhY2hlX2JlYW0vZXhhbXBsZXMvd29yZGNvdW50X3dpdGhfbWV0cmljcy5weQ==) | `28.12% <0.00%> (ø)` | |
| [...ks/python/apache\_beam/io/gcp/pubsub\_it\_pipeline.py](https://codecov.io/gh/apache/beam/pull/12645/diff?src=pr&el=tree#diff-c2Rrcy9weXRob24vYXBhY2hlX2JlYW0vaW8vZ2NwL3B1YnN1Yl9pdF9waXBlbGluZS5weQ==) | `20.58% <0.00%> (ø)` | |
| [sdks/python/apache\_beam/io/gcp/dicomio.py](https://codecov.io/gh/apache/beam/pull/12645/diff?src=pr&el=tree#diff-c2Rrcy9weXRob24vYXBhY2hlX2JlYW0vaW8vZ2NwL2RpY29taW8ucHk=) | `94.92% <0.00%> (ø)` | |
| [...s/python/apache\_beam/portability/api/schema\_pb2.py](https://codecov.io/gh/apache/beam/pull/12645/diff?src=pr&el=tree#diff-c2Rrcy9weXRob24vYXBhY2hlX2JlYW0vcG9ydGFiaWxpdHkvYXBpL3NjaGVtYV9wYjIucHk=) | `100.00% <0.00%> (ø)` | |
| ... and [443 more](https://codecov.io/gh/apache/beam/pull/12645/diff?src=pr&el=tree-more) | |
------
[Continue to review full report at Codecov](https://codecov.io/gh/apache/beam/pull/12645?src=pr&el=continue).
> **Legend** - [Click here to learn more](https://docs.codecov.io/docs/codecov-delta)
> `Δ = absolute <relative> (impact)`, `ø = not affected`, `? = missing data`
> Powered by [Codecov](https://codecov.io/gh/apache/beam/pull/12645?src=pr&el=footer). Last update [2b2b8e7...85c579f](https://codecov.io/gh/apache/beam/pull/12645?src=pr&el=lastupdated). Read the [comment docs](https://docs.codecov.io/docs/pull-request-comments).
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
[GitHub] [beam] rezarokni commented on a change in pull request #12645: [BEAM-10124] Add ContextualTextIO
Posted by GitBox <gi...@apache.org>.
rezarokni commented on a change in pull request #12645:
URL: https://github.com/apache/beam/pull/12645#discussion_r487897086
##########
File path: sdks/java/io/contextual-text-io/src/main/java/org/apache/beam/sdk/io/contextualtextio/ContextualTextIOSource.java
##########
@@ -0,0 +1,364 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one
+ * or more contributor license agreements. See the NOTICE file
+ * distributed with this work for additional information
+ * regarding copyright ownership. The ASF licenses this file
+ * to you under the Apache License, Version 2.0 (the
+ * "License"); you may not use this file except in compliance
+ * with the License. You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+package org.apache.beam.sdk.io.contextualtextio;
+
+import java.io.IOException;
+import java.nio.ByteBuffer;
+import java.nio.channels.ReadableByteChannel;
+import java.nio.channels.SeekableByteChannel;
+import java.util.NoSuchElementException;
+import org.apache.beam.sdk.coders.Coder;
+import org.apache.beam.sdk.io.FileBasedSource;
+import org.apache.beam.sdk.io.fs.EmptyMatchTreatment;
+import org.apache.beam.sdk.io.fs.MatchResult;
+import org.apache.beam.sdk.options.PipelineOptions;
+import org.apache.beam.sdk.options.ValueProvider;
+import org.apache.beam.sdk.schemas.NoSuchSchemaException;
+import org.apache.beam.sdk.schemas.SchemaCoder;
+import org.apache.beam.sdk.schemas.SchemaRegistry;
+import org.apache.beam.vendor.grpc.v1p26p0.com.google.protobuf.ByteString;
+import org.apache.beam.vendor.guava.v26_0_jre.com.google.common.annotations.VisibleForTesting;
+import org.apache.beam.vendor.guava.v26_0_jre.com.google.common.base.Preconditions;
+import org.checkerframework.checker.nullness.qual.Nullable;
+import org.slf4j.Logger;
+import org.slf4j.LoggerFactory;
+
+/**
+ * Implementation detail of {@link ContextualTextIO.Read}.
+ *
+ * <p>A {@link FileBasedSource} which can decode records delimited by newline characters.
+ *
+ * <p>This source splits the data into records using {@code UTF-8} {@code \n}, {@code \r}, or {@code
+ * \r\n} as the delimiter. This source is not strict and supports decoding the last record even if
+ * it is not delimited. Finally, no records are decoded if the stream is empty.
+ *
+ * <p>This source supports reading from any arbitrary byte position within the stream. If the
+ * starting position is not {@code 0}, then bytes are skipped until the first delimiter is found
+ * representing the beginning of the first record to be decoded.
+ */
+@VisibleForTesting
+class ContextualTextIOSource extends FileBasedSource<RecordWithMetadata> {
+ byte[] delimiter;
+
+ private static final Logger LOG = LoggerFactory.getLogger(ContextualTextIOSource.class);
+
+ // Used to Override isSplittable
+ private boolean hasMultilineCSVRecords;
+
+ @Override
+ protected boolean isSplittable() throws Exception {
+ if (hasMultilineCSVRecords) {
+ // When Having Multiline CSV Records,
+ // Splitting the file may cause a split to be within a record,
+ // Disabling split prevents this from happening
+ return false;
+ }
+ return super.isSplittable();
+ }
+
+ ContextualTextIOSource(
+ ValueProvider<String> fileSpec,
+ EmptyMatchTreatment emptyMatchTreatment,
+ byte[] delimiter,
+ boolean hasMultilineCSVRecords) {
+ super(fileSpec, emptyMatchTreatment, 1L);
+ this.delimiter = delimiter;
+ this.hasMultilineCSVRecords = hasMultilineCSVRecords;
+ }
+
+ private ContextualTextIOSource(
+ MatchResult.Metadata metadata,
+ long start,
+ long end,
+ byte[] delimiter,
+ boolean hasMultilineCSVRecords) {
+ super(metadata, 1L, start, end);
+ this.delimiter = delimiter;
+ this.hasMultilineCSVRecords = hasMultilineCSVRecords;
+ }
+
+ @Override
+ protected FileBasedSource<RecordWithMetadata> createForSubrangeOfFile(
+ MatchResult.Metadata metadata, long start, long end) {
+ return new ContextualTextIOSource(metadata, start, end, delimiter, hasMultilineCSVRecords);
+ }
+
+ @Override
+ protected FileBasedReader<RecordWithMetadata> createSingleFileReader(PipelineOptions options) {
+ return new MultiLineTextBasedReader(this, delimiter, hasMultilineCSVRecords);
+ }
+
+ @Override
+ public Coder<RecordWithMetadata> getOutputCoder() {
+ SchemaCoder<RecordWithMetadata> coder = null;
+ try {
+ coder = SchemaRegistry.createDefault().getSchemaCoder(RecordWithMetadata.class);
+ } catch (NoSuchSchemaException e) {
+ LOG.error("No Coder Found for RecordWithMetadata");
+ }
+ return coder;
+ }
+
+ /**
+ * A {@link FileBasedReader FileBasedReader} which can decode records delimited by delimiter
+ * characters.
+ *
+ * <p>See {@link ContextualTextIOSource } for further details.
+ */
+ @VisibleForTesting
+ static class MultiLineTextBasedReader extends FileBasedReader<RecordWithMetadata> {
+ public static final int READ_BUFFER_SIZE = 8192;
+ private static final ByteString UTF8_BOM =
+ ByteString.copyFrom(new byte[] {(byte) 0xEF, (byte) 0xBB, (byte) 0xBF});
+ private final ByteBuffer readBuffer = ByteBuffer.allocate(READ_BUFFER_SIZE);
+ private ByteString buffer;
+ private int startOfDelimiterInBuffer;
+ private int endOfDelimiterInBuffer;
+ private long startOfRecord;
+ private volatile long startOfNextRecord;
+ private volatile boolean eof;
+ private volatile boolean elementIsPresent;
+ private @Nullable RecordWithMetadata currentValue;
+ private @Nullable ReadableByteChannel inChannel;
+ private byte @Nullable [] delimiter;
+
+ // Add to override the isSplittable
+ private boolean hasMultilineCSVRecords;
+
+ private long startingOffset;
+ private long totalRecordCount;
+
+ private MultiLineTextBasedReader(
+ ContextualTextIOSource source, byte[] delimiter, boolean hasMultilineCSVRecords) {
+ super(source);
+ buffer = ByteString.EMPTY;
+ this.delimiter = delimiter;
+ this.hasMultilineCSVRecords = hasMultilineCSVRecords;
+ startingOffset = getCurrentSource().getStartOffset(); // Start offset;
+ }
+
+ @Override
+ protected long getCurrentOffset() throws NoSuchElementException {
+ if (!elementIsPresent) {
+ throw new NoSuchElementException();
+ }
+ return startOfRecord;
+ }
+
+ @Override
+ public long getSplitPointsRemaining() {
+ if (isStarted() && startOfNextRecord >= getCurrentSource().getEndOffset()) {
+ return isDone() ? 0 : 1;
+ }
+ return super.getSplitPointsRemaining();
+ }
+
+ @Override
+ public RecordWithMetadata getCurrent() throws NoSuchElementException {
+ if (!elementIsPresent) {
+ throw new NoSuchElementException();
+ }
+ return currentValue;
+ }
+
+ @Override
+ protected void startReading(ReadableByteChannel channel) throws IOException {
+ this.inChannel = channel;
+ // If the first offset is greater than zero, we need to skip bytes until we see our
+ // first delimiter.
+ long startOffset = getCurrentSource().getStartOffset();
+ if (startOffset > 0) {
+ Preconditions.checkState(
+ channel instanceof SeekableByteChannel,
+ "%s only supports reading from a SeekableByteChannel when given a start offset"
+ + " greater than 0.",
+ ContextualTextIOSource.class.getSimpleName());
+ long requiredPosition = startOffset - 1;
+ if (delimiter != null && startOffset >= delimiter.length) {
+ // we need to move back the offset of at worse delimiter.size to be sure to see
+ // all the bytes of the delimiter in the call to findDelimiterBounds() below
+ requiredPosition = startOffset - delimiter.length;
+ }
+ ((SeekableByteChannel) channel).position(requiredPosition);
+ findDelimiterBoundsWithMultiLineCheck();
+ buffer = buffer.substring(endOfDelimiterInBuffer);
+ startOfNextRecord = requiredPosition + endOfDelimiterInBuffer;
+ endOfDelimiterInBuffer = 0;
+ startOfDelimiterInBuffer = 0;
+ }
+ }
+
+ private void findDelimiterBoundsWithMultiLineCheck() throws IOException {
+ findDelimiterBounds();
+ }
+
+ /**
+ * Locates the start position and end position of the next delimiter. Will consume the channel
+ * till either EOF or the delimiter bounds are found.
+ *
+ * <p>If {@link ContextualTextIOSource#hasMultilineCSVRecords} is set then the behaviour will
+ * change from the standard read seen in {@link org.apache.beam.sdk.io.TextIO}. The assumption
+ * when {@link ContextualTextIOSource#hasMultilineCSVRecords} is set is that the file is being
+ * read with a single thread.
+ *
+ * <p>This fills the buffer and updates the positions as follows:
+ *
+ * <pre>{@code
+ * ------------------------------------------------------
+ * | element bytes | delimiter bytes | unconsumed bytes |
+ * ------------------------------------------------------
+ * 0 start of end of buffer
+ * delimiter delimiter size
+ * in buffer in buffer
+ * }</pre>
+ */
+ private void findDelimiterBounds() throws IOException {
+ int bytePositionInBuffer = 0;
+ boolean doubleQuoteClosed = true;
+
+ while (true) {
+ if (!tryToEnsureNumberOfBytesInBuffer(bytePositionInBuffer + 1)) {
+ startOfDelimiterInBuffer = endOfDelimiterInBuffer = bytePositionInBuffer;
+ break;
+ }
+
+ byte currentByte = buffer.byteAt(bytePositionInBuffer);
+ if (hasMultilineCSVRecords) {
+ // Check if we are inside an open Quote
+ if (currentByte == '"') {
+ doubleQuoteClosed = !doubleQuoteClosed;
+ }
+ } else {
+ doubleQuoteClosed = true;
+ }
+
+ if (delimiter == null) {
+ // default delimiter
+ if (currentByte == '\n') {
+ startOfDelimiterInBuffer = bytePositionInBuffer;
+ endOfDelimiterInBuffer = startOfDelimiterInBuffer + 1;
+ if (doubleQuoteClosed) {
+ break;
+ }
+ } else if (currentByte == '\r') {
+ startOfDelimiterInBuffer = bytePositionInBuffer;
+ endOfDelimiterInBuffer = startOfDelimiterInBuffer + 1;
+ if (tryToEnsureNumberOfBytesInBuffer(bytePositionInBuffer + 2)) {
+ currentByte = buffer.byteAt(bytePositionInBuffer + 1);
+ if (currentByte == '\n') {
+ endOfDelimiterInBuffer += 1;
+ }
+ }
+ if (doubleQuoteClosed) {
+ break;
+ }
+ }
+ } else {
+ // when the user defines a delimiter
+ int i = 0;
+ startOfDelimiterInBuffer = endOfDelimiterInBuffer = bytePositionInBuffer;
+ while ((i < delimiter.length) && (currentByte == delimiter[i])) {
+ // read next byte;
+ i++;
+ if (tryToEnsureNumberOfBytesInBuffer(bytePositionInBuffer + i + 1)) {
+ currentByte = buffer.byteAt(bytePositionInBuffer + i);
+ } else {
+ // corner case: delimiter truncate at the end of file
+ startOfDelimiterInBuffer = endOfDelimiterInBuffer = bytePositionInBuffer;
+ break;
+ }
+ }
+ if (i == delimiter.length) {
+ endOfDelimiterInBuffer = bytePositionInBuffer + i;
+ if (doubleQuoteClosed) {
+ break;
+ }
+ }
+ }
+ bytePositionInBuffer += 1;
+ }
+ }
+
+ @Override
+ protected boolean readNextRecord() throws IOException {
+ startOfRecord = startOfNextRecord;
+
+ findDelimiterBoundsWithMultiLineCheck();
+
+ // If we have reached EOF file and consumed all of the buffer then we know
+ // that there are no more records.
+ if (eof && buffer.isEmpty()) {
+ elementIsPresent = false;
+ return false;
+ }
+
+ decodeCurrentElement();
+ startOfNextRecord = startOfRecord + endOfDelimiterInBuffer;
+ return true;
+ }
+
+ /**
+ * Decodes the current element updating the buffer to only contain the unconsumed bytes.
+ *
+ * <p>This invalidates the currently stored {@code startOfDelimiterInBuffer} and {@code
+ * endOfDelimiterInBuffer}.
+ */
+ private void decodeCurrentElement() throws IOException {
+ ByteString dataToDecode = buffer.substring(0, startOfDelimiterInBuffer);
+ // If present, the UTF8 Byte Order Mark (BOM) will be removed.
+ if (startOfRecord == 0 && dataToDecode.startsWith(UTF8_BOM)) {
+ dataToDecode = dataToDecode.substring(UTF8_BOM.size());
+ }
+
+ // The line num is:
+ Long recordUniqueNum = totalRecordCount++;
+ // The Complete FileName (with uri if this is a web url eg: temp/abc.txt) is:
+ String fileName = getCurrentSource().getSingleFileMetadata().resourceId().toString();
Review comment:
@lukecwik AutoValueSchema has issues with schemas that reference 'self' ResourceID references itself so can not be used... FileName looks like best option for now, well if we want to make use of Row, which I think we do.
@tvalentyn FYI
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
[GitHub] [beam] lukecwik commented on a change in pull request #12645: [BEAM-10124] Add ContextualTextIO
Posted by GitBox <gi...@apache.org>.
lukecwik commented on a change in pull request #12645:
URL: https://github.com/apache/beam/pull/12645#discussion_r477495350
##########
File path: sdks/java/io/contextual-text-io/src/main/java/org/apache/beam/sdk/io/contextualtextio/ContextualTextIOSource.java
##########
@@ -0,0 +1,364 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one
+ * or more contributor license agreements. See the NOTICE file
+ * distributed with this work for additional information
+ * regarding copyright ownership. The ASF licenses this file
+ * to you under the Apache License, Version 2.0 (the
+ * "License"); you may not use this file except in compliance
+ * with the License. You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+package org.apache.beam.sdk.io.contextualtextio;
+
+import java.io.IOException;
+import java.nio.ByteBuffer;
+import java.nio.channels.ReadableByteChannel;
+import java.nio.channels.SeekableByteChannel;
+import java.util.NoSuchElementException;
+import org.apache.beam.sdk.coders.Coder;
+import org.apache.beam.sdk.io.FileBasedSource;
+import org.apache.beam.sdk.io.fs.EmptyMatchTreatment;
+import org.apache.beam.sdk.io.fs.MatchResult;
+import org.apache.beam.sdk.options.PipelineOptions;
+import org.apache.beam.sdk.options.ValueProvider;
+import org.apache.beam.sdk.schemas.NoSuchSchemaException;
+import org.apache.beam.sdk.schemas.SchemaCoder;
+import org.apache.beam.sdk.schemas.SchemaRegistry;
+import org.apache.beam.vendor.grpc.v1p26p0.com.google.protobuf.ByteString;
+import org.apache.beam.vendor.guava.v26_0_jre.com.google.common.annotations.VisibleForTesting;
+import org.apache.beam.vendor.guava.v26_0_jre.com.google.common.base.Preconditions;
+import org.checkerframework.checker.nullness.qual.Nullable;
+import org.slf4j.Logger;
+import org.slf4j.LoggerFactory;
+
+/**
+ * Implementation detail of {@link ContextualTextIO.Read}.
+ *
+ * <p>A {@link FileBasedSource} which can decode records delimited by newline characters.
+ *
+ * <p>This source splits the data into records using {@code UTF-8} {@code \n}, {@code \r}, or {@code
+ * \r\n} as the delimiter. This source is not strict and supports decoding the last record even if
+ * it is not delimited. Finally, no records are decoded if the stream is empty.
+ *
+ * <p>This source supports reading from any arbitrary byte position within the stream. If the
+ * starting position is not {@code 0}, then bytes are skipped until the first delimiter is found
+ * representing the beginning of the first record to be decoded.
+ */
+@VisibleForTesting
+class ContextualTextIOSource extends FileBasedSource<RecordWithMetadata> {
Review comment:
Swapping will lead be update incompatible for runners supporting update so we will have to live with both versions .
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
[GitHub] [beam] tvalentyn commented on a change in pull request #12645: [BEAM-10124] Add ContextualTextIO
Posted by GitBox <gi...@apache.org>.
tvalentyn commented on a change in pull request #12645:
URL: https://github.com/apache/beam/pull/12645#discussion_r477509213
##########
File path: sdks/java/io/contextual-text-io/src/main/java/org/apache/beam/sdk/io/contextualtextio/ContextualTextIOSource.java
##########
@@ -0,0 +1,364 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one
+ * or more contributor license agreements. See the NOTICE file
+ * distributed with this work for additional information
+ * regarding copyright ownership. The ASF licenses this file
+ * to you under the Apache License, Version 2.0 (the
+ * "License"); you may not use this file except in compliance
+ * with the License. You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+package org.apache.beam.sdk.io.contextualtextio;
+
+import java.io.IOException;
+import java.nio.ByteBuffer;
+import java.nio.channels.ReadableByteChannel;
+import java.nio.channels.SeekableByteChannel;
+import java.util.NoSuchElementException;
+import org.apache.beam.sdk.coders.Coder;
+import org.apache.beam.sdk.io.FileBasedSource;
+import org.apache.beam.sdk.io.fs.EmptyMatchTreatment;
+import org.apache.beam.sdk.io.fs.MatchResult;
+import org.apache.beam.sdk.options.PipelineOptions;
+import org.apache.beam.sdk.options.ValueProvider;
+import org.apache.beam.sdk.schemas.NoSuchSchemaException;
+import org.apache.beam.sdk.schemas.SchemaCoder;
+import org.apache.beam.sdk.schemas.SchemaRegistry;
+import org.apache.beam.vendor.grpc.v1p26p0.com.google.protobuf.ByteString;
+import org.apache.beam.vendor.guava.v26_0_jre.com.google.common.annotations.VisibleForTesting;
+import org.apache.beam.vendor.guava.v26_0_jre.com.google.common.base.Preconditions;
+import org.checkerframework.checker.nullness.qual.Nullable;
+import org.slf4j.Logger;
+import org.slf4j.LoggerFactory;
+
+/**
+ * Implementation detail of {@link ContextualTextIO.Read}.
+ *
+ * <p>A {@link FileBasedSource} which can decode records delimited by newline characters.
+ *
+ * <p>This source splits the data into records using {@code UTF-8} {@code \n}, {@code \r}, or {@code
+ * \r\n} as the delimiter. This source is not strict and supports decoding the last record even if
+ * it is not delimited. Finally, no records are decoded if the stream is empty.
+ *
+ * <p>This source supports reading from any arbitrary byte position within the stream. If the
+ * starting position is not {@code 0}, then bytes are skipped until the first delimiter is found
+ * representing the beginning of the first record to be decoded.
+ */
+@VisibleForTesting
+class ContextualTextIOSource extends FileBasedSource<RecordWithMetadata> {
Review comment:
Thanks, to clarify you mean streaming job updates will fail?
Does experimental annotation in contextualtextio/package-info.java allow for making future breaking changes of this kind?
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
[GitHub] [beam] rezarokni commented on a change in pull request #12645: [BEAM-10124] Add ContextualTextIO
Posted by GitBox <gi...@apache.org>.
rezarokni commented on a change in pull request #12645:
URL: https://github.com/apache/beam/pull/12645#discussion_r487935345
##########
File path: sdks/java/io/contextual-text-io/src/main/java/org/apache/beam/sdk/io/contextualtextio/RecordWithMetadata.java
##########
@@ -0,0 +1,85 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one
+ * or more contributor license agreements. See the NOTICE file
+ * distributed with this work for additional information
+ * regarding copyright ownership. The ASF licenses this file
+ * to you under the Apache License, Version 2.0 (the
+ * "License"); you may not use this file except in compliance
+ * with the License. You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+package org.apache.beam.sdk.io.contextualtextio;
+
+import com.google.auto.value.AutoValue;
+import org.apache.beam.sdk.annotations.Experimental;
+import org.apache.beam.sdk.io.fs.ResourceId;
+import org.apache.beam.sdk.schemas.AutoValueSchema;
+import org.apache.beam.sdk.schemas.annotations.DefaultSchema;
+
+/**
+ * Helper Class based on {@link AutoValueSchema}, it provides Metadata associated with each Record
+ * when reading from file(s) using {@link ContextualTextIO}.
+ *
+ * <h3>Fields:</h3>
+ *
+ * <ul>
+ * <li>recordOffset: The offset of a record (the byte at which the record begins) in a file. This
+ * information can be useful if you wish to reconstruct the file. {@link
+ * RecordWithMetadata#getRecordOffset()}
+ * <li>recordNum: The ordinal number of the record in its file. {@link
+ * RecordWithMetadata#getRecordNum()}
+ * <li>recordValue: The value / contents of the record {@link RecordWithMetadata#getValue()}
+ * <li>rangeOffset: The starting offset of the range (split), which contained the record, when the
+ * record was read. {@link RecordWithMetadata#getRangeOffset()}
+ * <li>recordNumInOffset: The record number relative to the Range. (line number within the range)
+ * {@link RecordWithMetadata#getRecordNumInOffset()}
+ * <li>fileName: Name of the file to which the record belongs (this is the full filename,
+ * eg:path/to/file.txt) {@link RecordWithMetadata#getFileName()}
+ * </ul>
+ */
+@Experimental(Experimental.Kind.SCHEMAS)
+@DefaultSchema(AutoValueSchema.class)
+@AutoValue
+public abstract class RecordWithMetadata {
+ public abstract long getRecordOffset();
+
+ public abstract long getRecordNum();
+
+ public abstract String getValue();
+
+ public abstract long getRangeOffset();
+
+ public abstract long getRecordNumInOffset();
+
+ public abstract Builder toBuilder();
+
+ public abstract String getFileName();
+
+ public static Builder newBuilder() {
+ return new AutoValue_RecordWithMetadata.Builder();
+ }
+
+ @AutoValue.Builder
+ public abstract static class Builder {
+ public abstract Builder setRecordNum(long lineNum);
+
+ public abstract Builder setRecordOffset(long recordOffset);
+
+ public abstract Builder setValue(String Value);
Review comment:
String value instead of String Value.
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
users@infra.apache.org