Support DECIMAL logical type in python SDK #23014

Abacn · 2022-09-02T18:13:27Z

The second part of #19817 (keep the issue open for as go support is still pending).

This change makes decimal type work for xlang transforms (such as jdbc) in python sdk. It also contains prerequisite changes for the minimum support of a portable logical type with argument in both Java and Python sdk.

Python SDK still cannot get the argument value from proto because SchemaTranslation.value_to_runner_api method does not exist yet (it is SchemaTranslation.fieldValueToProto in Java SDK). Nevertheless because the argument is just a wrapper, and the Decimal value can be recovered from the encoded data alone. This does not affect the decimal type.

Because the change is already large, value_to_runner_api and the complete support of argument values in python is let as followup.

migrate FixedPrecisionNumeric jdbclogicaltype to portable
logical type
Support DECIMAL type in python sdk xlang jdbc transform
Support standard logical type with argument in java sdk
proto support logical type's argument type in java sdk.
Support of logical type's argument value is still pending
Implement BigIntegerCoder and DecimalCoder in python sdk

Please add a meaningful description for your change here

Thank you for your contribution! Follow this checklist to help us incorporate your contribution quickly and easily:

Choose reviewer(s) and mention them in a comment (R: @username).
Mention the appropriate issue in your description (for example: addresses #123), if applicable. This will automatically add a link to the pull request in the issue. If you would like the issue to automatically close on merging the pull request, comment fixes #<ISSUE NUMBER> instead.
Update CHANGES.md with noteworthy changes.
If this contribution is large, please file an Apache Individual Contributor License Agreement.

See the Contributor Guide for more tips on how to make review process smoother.

To check the build health, please visit https://github.com/apache/beam/blob/master/.test-infra/BUILD_STATUS.md

GitHub Actions Tests Status (on master branch)

See CI.md for more information about GitHub Actions CI.

codecov · 2022-09-02T20:16:15Z

Codecov Report

Merging #23014 (b75ba4b) into master (ebacef9) will decrease coverage by 0.10%.
The diff coverage is 82.40%.

@@            Coverage Diff             @@
##           master   #23014      +/-   ##
==========================================
- Coverage   73.58%   73.48%   -0.11%     
==========================================
  Files         716      718       +2     
  Lines       95311    95692     +381     
==========================================
+ Hits        70138    70316     +178     
- Misses      23877    24080     +203     
  Partials     1296     1296

Flag	Coverage Δ
python	`83.20% <82.40%> (-0.21%)`	⬇️

Flags with carried forward coverage won't be shown. Click here to find out more.

Impacted Files	Coverage Δ
sdks/python/apache_beam/io/gcp/bigquery.py	`74.24% <ø> (ø)`
sdks/python/apache_beam/coders/coders.py	`87.36% <65.21%> (-0.65%)`	⬇️
sdks/python/apache_beam/typehints/schemas.py	`91.89% <76.08%> (-1.96%)`	⬇️
sdks/python/apache_beam/coders/coder_impl.py	`94.05% <100.00%> (+0.14%)`	⬆️
...ks/python/apache_beam/coders/coders_test_common.py	`98.58% <100.00%> (+0.03%)`	⬆️
sdks/python/apache_beam/coders/row_coder.py	`94.95% <100.00%> (+0.13%)`	⬆️
...s/python/apache_beam/io/gcp/bigquery_file_loads.py	`87.24% <100.00%> (-0.47%)`	⬇️
sdks/python/apache_beam/portability/common_urns.py	`100.00% <100.00%> (ø)`
...s/interactive/dataproc/dataproc_cluster_manager.py	`71.72% <0.00%> (-5.70%)`	⬇️
sdks/python/apache_beam/internal/gcp/auth.py	`73.33% <0.00%> (-5.34%)`	⬇️
... and 27 more

📣 We’re building smart automated test selection to slash your CI/CD build times. Learn more

Abacn · 2022-09-02T21:26:33Z

Run Java PreCommit

Abacn · 2022-09-02T21:26:43Z

Run Python_PVR_Flink PreCommit

Abacn · 2022-09-02T21:27:24Z

Run Python 3.8 PostCommit

Abacn · 2022-09-03T02:12:51Z

xlang integration test passed: https://ci-beam.apache.org/job/beam_PostCommit_Python38_PR/578/testReport/apache_beam.io.external.xlang_jdbcio_it_test/CrossLanguageJdbcIOTest/

Abacn · 2022-09-03T02:13:35Z

Run Java PreCommit

github-actions · 2022-09-03T02:56:27Z

Assigning reviewers. If you would like to opt out of this review, comment assign to next reviewer:

R: @y1chi for label python.
R: @apilloud for label java.
R: @chamikaramj for label io.

Available commands:

stop reviewer notifications - opt out of the automated review tooling
remind me after tests pass - tag the comment author after tests pass
waiting on author - shift the attention set back to the author (any comment or push by the author will return the attention set to the reviewers)

The PR bot will only process comments in the main thread (not review comments).

Abacn · 2022-09-08T14:17:18Z

R: @TheNeuralBit

github-actions · 2022-09-08T15:11:40Z

Stopping reviewer notifications for this pull request: review requested by someone other than the bot, ceding control

* migrate FixedPrecisionNumeric jdbclogicaltype to portable logical type * Support DECIMAL type in python sdk xlang jdbc transform * Support standard logical type with argument in java sdk * proto support logical type's argument type in java sdk. Support of logical type's argument value is still pending * Implement BigIntegerCoder and DecimalCoder in python sdk

Abacn · 2022-09-13T17:19:47Z

Run Python 3.8 PostCommit

Abacn · 2022-09-13T21:10:32Z

xlang jdbcio test passed: https://ci-beam.apache.org/job/beam_PostCommit_Python38_PR/598/testReport/apache_beam.io.external.xlang_jdbcio_it_test/CrossLanguageJdbcIOTest/

Abacn · 2022-09-13T21:10:59Z

Run Java PreCommit

Abacn · 2022-09-13T23:05:45Z

Java PreCommit build stuck twice at Task :sdks:java:core:buildNeeded. Also occasionally happens on master (e.g. https://ci-beam.apache.org/job/beam_PreCommit_Java_Cron/5900/, https://ci-beam.apache.org/job/beam_PreCommit_Java_Cron/5896/console) should be irrelevant. Read to review.

Abacn · 2022-09-14T00:58:39Z

Run Java PreCommit

TheNeuralBit

Thanks @Abacn, and sorry about the delay. I have a few suggestions

TheNeuralBit · 2022-09-26T18:14:17Z

sdks/java/core/src/main/java/org/apache/beam/sdk/schemas/SchemaTranslation.java

@@ -65,7 +67,7 @@
 })
 public class SchemaTranslation {

-  private static final String URN_BEAM_LOGICAL_DECIMAL = "beam:logical_type:decimal:v1";
+  private static final String URN_BEAM_LOGICAL_DECIMAL = FixedPrecisionNumeric.BASE_IDENTIFIER;


Could you add this in schema.proto and reference that here? Then we can document the format in schema.proto as well

added "beam:logical_type:decimal:v1" into schema.proto

TheNeuralBit · 2022-09-26T18:18:08Z

.../java/core/src/main/java/org/apache/beam/sdk/schemas/logicaltypes/FixedPrecisionNumeric.java

+        (base == null) ? null : base.precision(),
+        scale,
+        (base == null) ? null : base.scale());
+    return base;


The logic in this class is a little bit complicated, what do you think about adding a unit test just for this file?

This file is just moving org.apache.beam.sdk.io.jdbc.LogicalTypes.FixedPrecisionNumeric class to a separate file under schemas/logicaltypes. Yes will do add unit tests.

Ah sorry about that, thank you

.../java/core/src/main/java/org/apache/beam/sdk/schemas/logicaltypes/FixedPrecisionNumeric.java

TheNeuralBit · 2022-09-26T18:19:51Z

.../java/core/src/main/java/org/apache/beam/sdk/schemas/logicaltypes/FixedPrecisionNumeric.java

+  public static final String IDENTIFIER = "beam:logical_type:fixed_decimal:v1";
+
+  // TODO(https://github.com/apache/beam/issues/19817) implement beam:logical_type:decimal:v1 as
+  // CoderLogicalType (once CoderLogicalType is implemented).


What is CoderLogicalType? I don't a reference to it in that issue

ah the link in TODO is not accurate. This is a suggestion I got from @reuvenlax. The reference was here: #7865 (comment). Should we create an issue for it?

entered #23374

TheNeuralBit · 2022-09-26T18:24:49Z

sdks/python/apache_beam/typehints/schemas.py

+                    logical_type.representation_type())))
+      else:
+        # TODO(bhulette,yathu): Complete support for logical types that require
+        # arguments.


Please file an issue for this and link it here. Also note that the option_to_runner_api and option_from_runner_api has logic for working with FieldValues, we should re-use that when we take this on.

entered #23373

TheNeuralBit · 2022-09-26T18:25:57Z

sdks/python/apache_beam/io/external/xlang_jdbcio_it_test.py

-        "CREATE TABLE {}(f_int INTEGER, f_timestamp TIMESTAMP)".format(
-            table_name))
+        "CREATE TABLE {}(f_int INTEGER, f_timestamp TIMESTAMP, f_decimal DECIMAL(10,2))"  # pylint: disable=line-too-long
+        .format(table_name))


Can you also add a test(s) for this in standard_coders.yaml?

added in standard_coders.yaml

Abacn · 2022-09-29T02:11:56Z

Run Python 3.8 PostCommit

Abacn · 2022-09-29T02:37:10Z

Run Python 3.8 PostCommit

…from serialized

Abacn · 2022-09-29T03:25:39Z

Run Python 3.8 PostCommit

Abacn · 2022-09-29T13:39:24Z

PostCommit test passed. Ready for review

TheNeuralBit · 2022-09-29T22:04:37Z

Run Java PreCommit

TheNeuralBit · 2022-09-29T22:06:52Z

.../java/core/src/main/java/org/apache/beam/sdk/schemas/logicaltypes/FixedPrecisionNumeric.java

+        (base == null) ? null : base.precision(),
+        scale,
+        (base == null) ? null : base.scale());
+    return base;


Ah sorry about that, thank you

model/pipeline/src/main/proto/org/apache/beam/model/pipeline/v1/schema.proto

TheNeuralBit · 2022-09-29T22:09:02Z

.../java/core/src/main/java/org/apache/beam/sdk/schemas/logicaltypes/FixedPrecisionNumeric.java

+  public static final String IDENTIFIER = "beam:logical_type:fixed_decimal:v1";
+
+  // TODO(https://github.com/apache/beam/issues/19817) implement beam:logical_type:decimal:v1 as
+  // CoderLogicalType (once CoderLogicalType is implemented).


TheNeuralBit · 2022-09-29T22:10:44Z

sdks/java/core/src/test/java/org/apache/beam/sdk/schemas/logicaltypes/LogicalTypesTest.java

+    row = Row.withSchema(schema).addValues(decimal).build();
+    assertEquals(decimal, row.getLogicalTypeValue(0, BigDecimal.class));
+
+    decimal = BigDecimal.valueOf(random.nextLong() + 100_000_000_000L, scale);


nit: I'm a little nervous that the random component can make this flaky. Can we just select some constants istead?

ah I made a typo. These two cases were intended be (random.nextInt() + 100_000_000_000L) which guarantees to generate a positive number with at least 11 digits and beyond the precision of DECIMAL(precision=10). Yes we can just use 100_000_000_001L here

TheNeuralBit · 2022-09-29T22:10:54Z

sdks/java/core/src/test/java/org/apache/beam/sdk/schemas/logicaltypes/LogicalTypesTest.java

+
+    // check argument invalid case
+    decimal = BigDecimal.valueOf(random.nextLong() + 100_000_000_000L, scale);
+    assertThrows(IllegalArgumentException.class, Row.withSchema(schema).addValues(decimal)::build);


Nice, I like how concise this is :)

TheNeuralBit · 2022-09-29T22:13:06Z

sdks/java/core/src/test/java/org/apache/beam/sdk/schemas/logicaltypes/LogicalTypesTest.java

+        IllegalArgumentException.class,
+        () ->
+            FixedPrecisionNumeric.of(
+                Row.withSchema(invalidArgumentSchema).addValues(precision, scale).build()));


nit: I think unit testing best practice would be to put each of these test cases in a different test method. On the other hand this pattern is consistent with the rest of the file, so I'm fine if you want to leave it as-is.

Yes I agree in general this is best practice, I observed the file LogicalTypesTest.java has each test testing one logical type and keep the new test following this. If tests become more loaded we can split test source files.

TheNeuralBit · 2022-09-29T22:18:38Z

sdks/python/apache_beam/typehints/schemas.py

+
+  @classmethod
+  def _from_typing(cls, typ):
+    print(typ)


This looks like a debug statement that was left in?

oops thanks for the catch, will double check

TheNeuralBit · 2022-09-29T22:24:51Z

.../java/core/src/main/java/org/apache/beam/sdk/schemas/logicaltypes/FixedPrecisionNumeric.java

+
+/** Fixed precision numeric types used to represent jdbc NUMERIC and DECIMAL types. */
+public class FixedPrecisionNumeric extends PassThroughLogicalType<BigDecimal> {
+  public static final String IDENTIFIER = "beam:logical_type:fixed_decimal:v1";


Since this is being used cross-language can we define the URN in schema.proto and document it there as well?

I considered to put it in schema.proto as beam:logical_type:decimal:v1 but the problem is that I do not know how to generate a test case of it in standard_coders.yaml yet because the logical types with argument is currently not fully supported. I would like to leave it now and make it into schema.proto until we are confident that this implementation is standardized(i.e. stable). Open to opinions for sure.

I see, that's fine with me.

…1/schema.proto fix incorrect documentation of Decimal logical type Co-authored-by: Brian Hulette <[email protected]>

Abacn · 2022-09-30T01:04:57Z

go precommit fails due to

 FAIL	github.com/apache/beam/sdks/v2/go/pkg/beam/util/pubsubx	600.108s

first time see go precommit flakes

java precommit failed for another different test

org.apache.beam.runners.samza.runtime.AsyncDoFnRunnerTest.testSimplePipeline

should be irrelevant either

Abacn · 2022-09-30T01:05:29Z

Run Go PreCommit

Abacn · 2022-09-30T01:05:49Z

Run Java PreCommit

Abacn · 2022-09-30T16:25:09Z

New flakes seen on Java PreCommit and entered #23449

Abacn · 2022-09-30T16:26:06Z

Run Java PreCommit

Abacn · 2022-09-30T16:26:17Z

Run Python 3.8 PostCommit

TheNeuralBit · 2022-09-30T16:51:10Z

Run Python 3.8 PostCommit

TheNeuralBit · 2022-09-30T16:51:17Z

Run Java PreCommit

TheNeuralBit · 2022-09-30T20:54:51Z

Run Python 3.8 PostCommit

Abacn · 2022-09-30T21:12:08Z

Thanks for taking care of the tests @TheNeuralBit! Test reliability has been terrible. Last two Java PreCommit fails in different flakes (no test fails for both)

#5592:
org.apache.beam.sdk.transforms.ParDoLifecycleTest.testTeardownCalledAfterExceptionInProcessElementStateful
org.apache.beam.sdk.io.pulsar.PulsarIOTest.testReadFromSimpleTopic
#5593:
org.apache.beam.runners.samza.runtime.AsyncDoFnRunnerTest.testSimplePipeline
org.apache.beam.sdk.fn.data.BeamFnDataInboundObserver2Test.testCloseVisibleToAwaitCompletionCallerAndProducer

Python 3.8 PostCommit:

#621: bigquery integration tests with BigQuery API call HttpError.

there was a success run (rare) of postcommit on the third commit: https://ci-beam.apache.org/job/beam_PostCommit_Python38_PR/618/ the last two commits of this PR are trivial and should not introduce surprise. We could wait for the running postcommit and give it one more chance :).

Update: looks like too much post commit run hit the BigQuery quota.

Abacn · 2022-10-03T00:26:26Z

Run Python 3.8 PostCommit

TheNeuralBit · 2022-10-03T23:47:25Z

Ok, given the previous successful run I'm going to go ahead and merge. Thanks @Abacn!

github-actions bot added io java jdbc python labels Sep 2, 2022

Abacn force-pushed the portabledecimal branch 7 times, most recently from 3ce2ce8 to 801a3cd Compare September 2, 2022 19:46

Abacn marked this pull request as ready for review September 3, 2022 02:36

github-actions bot added the Next Action: Reviewers label Sep 3, 2022

Abacn force-pushed the portabledecimal branch from b897925 to 5158b36 Compare September 13, 2022 17:19

TheNeuralBit reviewed Sep 26, 2022

View reviewed changes

Abacn force-pushed the portabledecimal branch from 501bf54 to 5f182d1 Compare September 29, 2022 02:36

Fix static assignment causing NoClassDefFoundError when initializing …

e849f1f

…from serialized

Abacn force-pushed the portabledecimal branch from 5f182d1 to e849f1f Compare September 29, 2022 03:25

TheNeuralBit reviewed Sep 29, 2022

View reviewed changes

Abacn and others added 2 commits September 29, 2022 19:43

Add comments regarding to fixed_decimal URN, fix typo and leftover

e9df734

Update model/pipeline/src/main/proto/org/apache/beam/model/pipeline/v…

b75ba4b

…1/schema.proto fix incorrect documentation of Decimal logical type Co-authored-by: Brian Hulette <[email protected]>

TheNeuralBit approved these changes Sep 30, 2022

View reviewed changes

TheNeuralBit merged commit 2955954 into apache:master Oct 3, 2022

Abacn deleted the portabledecimal branch October 5, 2022 13:48

This was referenced Oct 6, 2022

[Task]: Migrate remaining io.jdbc.JdbcLogicalType to Portable Logical Types in core.schemas.logicaltypes #23526

Closed

[Task]: Run some BigQuery integration tests only on one Dataflow Python3.X PostCommit #23573

Closed

This was referenced Jan 5, 2023

[Bug]: Encoding issue when reading BQ query result set in DataflowRunner #24870

Closed

Attempt deserialize all non-standard portable logical types from proto #24910

Merged

Support DECIMAL logical type in python SDK #23014

Support DECIMAL logical type in python SDK #23014

Conversation

Abacn commented Sep 2, 2022 • edited Loading

GitHub Actions Tests Status (on master branch)

codecov bot commented Sep 2, 2022 • edited Loading

Codecov Report

Abacn commented Sep 2, 2022

Abacn commented Sep 2, 2022

Abacn commented Sep 2, 2022

Abacn commented Sep 3, 2022

Abacn commented Sep 3, 2022

github-actions bot commented Sep 3, 2022

Abacn commented Sep 8, 2022

github-actions bot commented Sep 8, 2022

Abacn commented Sep 13, 2022

Abacn commented Sep 13, 2022

Abacn commented Sep 13, 2022

Abacn commented Sep 13, 2022

Abacn commented Sep 14, 2022

TheNeuralBit left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Abacn commented Sep 29, 2022

Abacn commented Sep 29, 2022

Abacn commented Sep 29, 2022

Abacn commented Sep 29, 2022

TheNeuralBit commented Sep 29, 2022

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Abacn commented Sep 30, 2022

Abacn commented Sep 30, 2022

Abacn commented Sep 30, 2022

Abacn commented Sep 30, 2022

Abacn commented Sep 30, 2022

Abacn commented Sep 30, 2022

TheNeuralBit commented Sep 30, 2022

TheNeuralBit commented Sep 30, 2022

TheNeuralBit commented Sep 30, 2022

Abacn commented Sep 30, 2022 • edited Loading

Abacn commented Oct 3, 2022

TheNeuralBit commented Oct 3, 2022

Abacn commented Sep 2, 2022 •

edited

Loading

codecov bot commented Sep 2, 2022 •

edited

Loading

Abacn commented Sep 30, 2022 •

edited

Loading