You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@flink.apache.org by GitBox <gi...@apache.org> on 2020/01/21 06:17:58 UTC
[GitHub] [flink] hequn8128 opened a new pull request #10913:
[FLINK-15636][python] Supports Python UDF in old planner under batch mode
hequn8128 opened a new pull request #10913: [FLINK-15636][python] Supports Python UDF in old planner under batch mode
URL: https://github.com/apache/flink/pull/10913
## What is the purpose of the change
Currently, Python UDF has been supported under old planner(only stream) and blink planner(stream&batch). This pull request dedicates to add Python UDF support in old planner under batch mode.
## Brief change log
- Add rules to convert RelNode to DataSetPythonCalc Node.
- Add PythonScalarFunctionFlatMap function to invoke Python ScalarFunctions for the legacy planner.
- Add IT Tests.
## Verifying this change
This change added tests and can be verified as follows:
- Added integration tests for Python UDF(test_udf.test_chaining_scalar_function) and dependency management(test_dependency.test_add_python_file).
## Does this pull request potentially affect one of the following parts:
- Dependencies (does it add or upgrade a dependency): (no)
- The public API, i.e., is any changed class annotated with `@Public(Evolving)`: ( no)
- The serializers: (no)
- The runtime per-record code paths (performance sensitive): (no)
- Anything that affects deployment or recovery: JobManager (and its components), Checkpointing, Yarn/Mesos, ZooKeeper: (no)
- The S3 file system connector: (no)
## Documentation
- Does this pull request introduce a new feature? (yes)
- If yes, how is the feature documented? (docs)
----------------------------------------------------------------
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
With regards,
Apache Git Services
[GitHub] [flink] hequn8128 commented on issue #10913: [FLINK-15636][python]
Supports Python UDF in old planner under batch mode
Posted by GitBox <gi...@apache.org>.
hequn8128 commented on issue #10913: [FLINK-15636][python] Supports Python UDF in old planner under batch mode
URL: https://github.com/apache/flink/pull/10913#issuecomment-577079384
@dianfu Thanks a lot for the review and suggestions. I have addressed your comments and updated the 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
With regards,
Apache Git Services
[GitHub] [flink] dianfu commented on a change in pull request #10913:
[FLINK-15636][python] Supports Python UDF in old planner under batch mode
Posted by GitBox <gi...@apache.org>.
dianfu commented on a change in pull request #10913: [FLINK-15636][python] Supports Python UDF in old planner under batch mode
URL: https://github.com/apache/flink/pull/10913#discussion_r369412012
##########
File path: flink-python/src/main/java/org/apache/flink/table/runtime/functions/python/PythonScalarFunctionFlatMap.java
##########
@@ -0,0 +1,319 @@
+/*
+ * 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.flink.table.runtime.functions.python;
+
+import org.apache.flink.annotation.Internal;
+import org.apache.flink.api.common.functions.RichFlatMapFunction;
+import org.apache.flink.api.common.typeinfo.TypeInformation;
+import org.apache.flink.api.common.typeutils.TypeSerializer;
+import org.apache.flink.api.java.typeutils.ResultTypeQueryable;
+import org.apache.flink.api.java.typeutils.RowTypeInfo;
+import org.apache.flink.configuration.Configuration;
+import org.apache.flink.configuration.ConfigurationUtils;
+import org.apache.flink.python.PythonConfig;
+import org.apache.flink.python.PythonFunctionRunner;
+import org.apache.flink.python.PythonOptions;
+import org.apache.flink.python.env.ProcessPythonEnvironmentManager;
+import org.apache.flink.python.env.PythonDependencyInfo;
+import org.apache.flink.python.env.PythonEnvironmentManager;
+import org.apache.flink.table.functions.ScalarFunction;
+import org.apache.flink.table.functions.python.PythonEnv;
+import org.apache.flink.table.functions.python.PythonFunctionInfo;
+import org.apache.flink.table.runtime.runners.python.PythonScalarFunctionRunner;
+import org.apache.flink.table.types.logical.RowType;
+import org.apache.flink.table.types.utils.LegacyTypeInfoDataTypeConverter;
+import org.apache.flink.table.types.utils.LogicalTypeDataTypeConverter;
+import org.apache.flink.table.types.utils.TypeConversions;
+import org.apache.flink.types.Row;
+import org.apache.flink.util.Collector;
+import org.apache.flink.util.Preconditions;
+
+import org.apache.beam.sdk.fn.data.FnDataReceiver;
+import org.slf4j.Logger;
+import org.slf4j.LoggerFactory;
+
+import java.io.IOException;
+import java.util.Arrays;
+import java.util.concurrent.LinkedBlockingQueue;
+import java.util.concurrent.atomic.AtomicBoolean;
+import java.util.stream.Collectors;
+
+/**
+ * The {@link RichFlatMapFunction} used to invoke Python {@link ScalarFunction} functions for the
+ * old planner.
+ */
+@Internal
+public final class PythonScalarFunctionFlatMap
+ extends RichFlatMapFunction<Row, Row> implements ResultTypeQueryable<Row> {
+
+ private static final long serialVersionUID = 1L;
+
+ protected static final Logger LOG = LoggerFactory.getLogger(PythonScalarFunctionFlatMap.class);
+
+ /**
+ * The type serializer for the forwarded fields.
+ */
+ private transient TypeSerializer<Row> forwardedInputSerializer;
+
+ /**
+ * The Python {@link ScalarFunction}s to be executed.
+ */
+ protected final PythonFunctionInfo[] scalarFunctions;
+
+ /**
+ * The input logical type.
+ */
+ protected final RowType inputType;
+
+ /**
+ * The output logical type.
+ */
+ protected final RowType outputType;
+
+ /**
+ * The offsets of udf inputs.
+ */
+ protected final int[] udfInputOffsets;
+
+ /**
+ * The offset of the fields which should be forwarded.
+ */
+ protected final int[] forwardedFields;
+
+ /**
+ * The udf input logical type.
+ */
+ protected transient RowType udfInputType;
+
+ /**
+ * The udf output logical type.
+ */
+ protected transient RowType udfOutputType;
+
+ /**
+ * The queue holding the input elements for which the execution results have not been received.
+ */
+ protected transient LinkedBlockingQueue<Row> forwardedInputQueue;
+
+ /**
+ * The queue holding the user-defined function execution results. The execution results are in
+ * the same order as the input elements.
+ */
+ protected transient LinkedBlockingQueue<Row> udfResultQueue;
+
+ /**
+ * The python config.
+ */
+ private final PythonConfig config;
+
+ /**
+ * Use an AtomicBoolean because we start/stop bundles by a timer thread.
+ */
+ private transient AtomicBoolean bundleStarted;
+
+ /**
+ * Max number of elements to include in a bundle.
+ */
+ private transient int maxBundleSize;
+
+ /**
+ * The collector used to collect records.
+ */
+ private transient Collector<Row> resultCollector;
+
+ /**
+ * Number of processed elements in the current bundle.
+ */
+ private transient int elementCount;
+
+ /**
+ * The {@link PythonFunctionRunner} which is responsible for Python user-defined function execution.
+ */
+ private transient PythonFunctionRunner<Row> pythonFunctionRunner;
+
+ public PythonScalarFunctionFlatMap(
+ Configuration config,
+ PythonFunctionInfo[] scalarFunctions,
+ RowType inputType,
+ RowType outputType,
+ int[] udfInputOffsets,
+ int[] forwardedFields) {
+ this.scalarFunctions = Preconditions.checkNotNull(scalarFunctions);
+ this.inputType = Preconditions.checkNotNull(inputType);
+ this.outputType = Preconditions.checkNotNull(outputType);
+ this.udfInputOffsets = Preconditions.checkNotNull(udfInputOffsets);
+ this.forwardedFields = Preconditions.checkNotNull(forwardedFields);
+ this.config = new PythonConfig(Preconditions.checkNotNull(config));
+ }
+
+ @Override
+ public void open(Configuration parameters) throws Exception {
+ super.open(parameters);
+
+ this.elementCount = 0;
+ this.bundleStarted = new AtomicBoolean(false);
+ this.maxBundleSize = config.getMaxBundleSize();
+ if (this.maxBundleSize <= 0) {
+ this.maxBundleSize = PythonOptions.MAX_BUNDLE_SIZE.defaultValue();
+ LOG.error("Invalid value for the maximum bundle size. Using default value of " +
+ this.maxBundleSize + '.');
+ } else {
+ LOG.info("The maximum bundle size is configured to {}.", this.maxBundleSize);
+ }
+
+ if (config.getMaxBundleTimeMills() != PythonOptions.MAX_BUNDLE_TIME_MILLS.defaultValue()) {
+ LOG.info("Maximum bundle time takes no effect in old planner under batch mode. " +
+ "Config maximum bundle size instead! " +
+ "Under batch mode, bundle size should be enough to control both throughput and latency.");
+ }
+
+ forwardedInputQueue = new LinkedBlockingQueue<>();
+ udfResultQueue = new LinkedBlockingQueue<>();
+ udfInputType = new RowType(
+ Arrays.stream(udfInputOffsets)
+ .mapToObj(i -> inputType.getFields().get(i))
+ .collect(Collectors.toList()));
+ udfOutputType = new RowType(outputType.getFields().subList(forwardedFields.length, outputType.getFieldCount()));
+
+ RowTypeInfo forwardedInputTypeInfo = new RowTypeInfo(
+ Arrays.stream(forwardedFields)
+ .mapToObj(i -> inputType.getFields().get(i))
+ .map(RowType.RowField::getType)
+ .map(TypeConversions::fromLogicalToDataType)
+ .map(TypeConversions::fromDataTypeToLegacyInfo)
+ .toArray(TypeInformation[]::new));
+ forwardedInputSerializer = forwardedInputTypeInfo.createSerializer(getRuntimeContext().getExecutionConfig());
+
+ this.pythonFunctionRunner = createPythonFunctionRunner();
+ this.pythonFunctionRunner.open();
+ this.resultCollector = null;
Review comment:
It seems that this line is not necessary. What about removing it?
----------------------------------------------------------------
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
With regards,
Apache Git Services
[GitHub] [flink] flinkbot edited a comment on issue #10913:
[FLINK-15636][python] Supports Python UDF in old planner under batch mode
Posted by GitBox <gi...@apache.org>.
flinkbot edited a comment on issue #10913: [FLINK-15636][python] Supports Python UDF in old planner under batch mode
URL: https://github.com/apache/flink/pull/10913#issuecomment-576547242
<!--
Meta data
Hash:bdaf386d48c4ef0493a0c1c61a7e6aea12f148d1 Status:PENDING URL:https://dev.azure.com/rmetzger/5bd3ef0a-4359-41af-abca-811b04098d2e/_build/results?buildId=4527 TriggerType:PUSH TriggerID:bdaf386d48c4ef0493a0c1c61a7e6aea12f148d1
Hash:bdaf386d48c4ef0493a0c1c61a7e6aea12f148d1 Status:PENDING URL:https://travis-ci.com/flink-ci/flink/builds/145316229 TriggerType:PUSH TriggerID:bdaf386d48c4ef0493a0c1c61a7e6aea12f148d1
-->
## CI report:
* bdaf386d48c4ef0493a0c1c61a7e6aea12f148d1 Travis: [PENDING](https://travis-ci.com/flink-ci/flink/builds/145316229) Azure: [PENDING](https://dev.azure.com/rmetzger/5bd3ef0a-4359-41af-abca-811b04098d2e/_build/results?buildId=4527)
<details>
<summary>Bot commands</summary>
The @flinkbot bot supports the following commands:
- `@flinkbot run travis` re-run the last Travis build
- `@flinkbot run azure` re-run the last Azure build
</details>
----------------------------------------------------------------
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
With regards,
Apache Git Services
[GitHub] [flink] dianfu commented on a change in pull request #10913:
[FLINK-15636][python] Supports Python UDF in old planner under batch mode
Posted by GitBox <gi...@apache.org>.
dianfu commented on a change in pull request #10913: [FLINK-15636][python] Supports Python UDF in old planner under batch mode
URL: https://github.com/apache/flink/pull/10913#discussion_r369424851
##########
File path: flink-table/flink-table-planner/src/test/scala/org/apache/flink/table/plan/PythonCalcSplitRuleTest.scala
##########
@@ -266,4 +266,26 @@ class PythonCalcSplitRuleTest extends TableTestBase {
util.verifyTable(resultTable, expected)
}
+
+ @Test
+ def testSplitRuleForBatch(): Unit = {
Review comment:
I guess we can remove this test case as the split rule is applied for both the streaming and batch plans. The original tests for streaming plans are enough. What do you think?
----------------------------------------------------------------
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
With regards,
Apache Git Services
[GitHub] [flink] flinkbot commented on issue #10913: [FLINK-15636][python]
Supports Python UDF in old planner under batch mode
Posted by GitBox <gi...@apache.org>.
flinkbot commented on issue #10913: [FLINK-15636][python] Supports Python UDF in old planner under batch mode
URL: https://github.com/apache/flink/pull/10913#issuecomment-576547242
<!--
Meta data
Hash:bdaf386d48c4ef0493a0c1c61a7e6aea12f148d1 Status:UNKNOWN URL:TBD TriggerType:PUSH TriggerID:bdaf386d48c4ef0493a0c1c61a7e6aea12f148d1
-->
## CI report:
* bdaf386d48c4ef0493a0c1c61a7e6aea12f148d1 UNKNOWN
<details>
<summary>Bot commands</summary>
The @flinkbot bot supports the following commands:
- `@flinkbot run travis` re-run the last Travis build
- `@flinkbot run azure` re-run the last Azure build
</details>
----------------------------------------------------------------
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
With regards,
Apache Git Services
[GitHub] [flink] dianfu commented on a change in pull request #10913:
[FLINK-15636][python] Supports Python UDF in old planner under batch mode
Posted by GitBox <gi...@apache.org>.
dianfu commented on a change in pull request #10913: [FLINK-15636][python] Supports Python UDF in old planner under batch mode
URL: https://github.com/apache/flink/pull/10913#discussion_r369407291
##########
File path: flink-python/src/main/java/org/apache/flink/table/runtime/functions/python/PythonScalarFunctionFlatMap.java
##########
@@ -0,0 +1,319 @@
+/*
+ * 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.flink.table.runtime.functions.python;
+
+import org.apache.flink.annotation.Internal;
+import org.apache.flink.api.common.functions.RichFlatMapFunction;
+import org.apache.flink.api.common.typeinfo.TypeInformation;
+import org.apache.flink.api.common.typeutils.TypeSerializer;
+import org.apache.flink.api.java.typeutils.ResultTypeQueryable;
+import org.apache.flink.api.java.typeutils.RowTypeInfo;
+import org.apache.flink.configuration.Configuration;
+import org.apache.flink.configuration.ConfigurationUtils;
+import org.apache.flink.python.PythonConfig;
+import org.apache.flink.python.PythonFunctionRunner;
+import org.apache.flink.python.PythonOptions;
+import org.apache.flink.python.env.ProcessPythonEnvironmentManager;
+import org.apache.flink.python.env.PythonDependencyInfo;
+import org.apache.flink.python.env.PythonEnvironmentManager;
+import org.apache.flink.table.functions.ScalarFunction;
+import org.apache.flink.table.functions.python.PythonEnv;
+import org.apache.flink.table.functions.python.PythonFunctionInfo;
+import org.apache.flink.table.runtime.runners.python.PythonScalarFunctionRunner;
+import org.apache.flink.table.types.logical.RowType;
+import org.apache.flink.table.types.utils.LegacyTypeInfoDataTypeConverter;
+import org.apache.flink.table.types.utils.LogicalTypeDataTypeConverter;
+import org.apache.flink.table.types.utils.TypeConversions;
+import org.apache.flink.types.Row;
+import org.apache.flink.util.Collector;
+import org.apache.flink.util.Preconditions;
+
+import org.apache.beam.sdk.fn.data.FnDataReceiver;
+import org.slf4j.Logger;
+import org.slf4j.LoggerFactory;
+
+import java.io.IOException;
+import java.util.Arrays;
+import java.util.concurrent.LinkedBlockingQueue;
+import java.util.concurrent.atomic.AtomicBoolean;
+import java.util.stream.Collectors;
+
+/**
+ * The {@link RichFlatMapFunction} used to invoke Python {@link ScalarFunction} functions for the
+ * old planner.
+ */
+@Internal
+public final class PythonScalarFunctionFlatMap
+ extends RichFlatMapFunction<Row, Row> implements ResultTypeQueryable<Row> {
+
+ private static final long serialVersionUID = 1L;
+
+ protected static final Logger LOG = LoggerFactory.getLogger(PythonScalarFunctionFlatMap.class);
+
+ /**
+ * The type serializer for the forwarded fields.
+ */
+ private transient TypeSerializer<Row> forwardedInputSerializer;
+
+ /**
+ * The Python {@link ScalarFunction}s to be executed.
+ */
+ protected final PythonFunctionInfo[] scalarFunctions;
Review comment:
ditto
----------------------------------------------------------------
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
With regards,
Apache Git Services
[GitHub] [flink] dianfu closed pull request #10913: [FLINK-15636][python]
Supports Python UDF in old planner under batch mode
Posted by GitBox <gi...@apache.org>.
dianfu closed pull request #10913: [FLINK-15636][python] Supports Python UDF in old planner under batch mode
URL: https://github.com/apache/flink/pull/10913
----------------------------------------------------------------
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
With regards,
Apache Git Services
[GitHub] [flink] flinkbot commented on issue #10913: [FLINK-15636][python]
Supports Python UDF in old planner under batch mode
Posted by GitBox <gi...@apache.org>.
flinkbot commented on issue #10913: [FLINK-15636][python] Supports Python UDF in old planner under batch mode
URL: https://github.com/apache/flink/pull/10913#issuecomment-576535977
Thanks a lot for your contribution to the Apache Flink project. I'm the @flinkbot. I help the community
to review your pull request. We will use this comment to track the progress of the review.
## Automated Checks
Last check on commit bdaf386d48c4ef0493a0c1c61a7e6aea12f148d1 (Tue Jan 21 06:22:00 UTC 2020)
✅no warnings
<sub>Mention the bot in a comment to re-run the automated checks.</sub>
## Review Progress
* ❓ 1. The [description] looks good.
* ❓ 2. There is [consensus] that the contribution should go into to Flink.
* ❓ 3. Needs [attention] from.
* ❓ 4. The change fits into the overall [architecture].
* ❓ 5. Overall code [quality] is good.
Please see the [Pull Request Review Guide](https://flink.apache.org/contributing/reviewing-prs.html) for a full explanation of the review process.<details>
The Bot is tracking the review progress through labels. Labels are applied according to the order of the review items. For consensus, approval by a Flink committer of PMC member is required <summary>Bot commands</summary>
The @flinkbot bot supports the following commands:
- `@flinkbot approve description` to approve one or more aspects (aspects: `description`, `consensus`, `architecture` and `quality`)
- `@flinkbot approve all` to approve all aspects
- `@flinkbot approve-until architecture` to approve everything until `architecture`
- `@flinkbot attention @username1 [@username2 ..]` to require somebody's attention
- `@flinkbot disapprove architecture` to remove an approval you gave earlier
</details>
----------------------------------------------------------------
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
With regards,
Apache Git Services
[GitHub] [flink] dianfu commented on issue #10913: [FLINK-15636][python]
Supports Python UDF in old planner under batch mode
Posted by GitBox <gi...@apache.org>.
dianfu commented on issue #10913: [FLINK-15636][python] Supports Python UDF in old planner under batch mode
URL: https://github.com/apache/flink/pull/10913#issuecomment-577081127
@hequn8128 Thanks for the update. LGTM. Merging...
----------------------------------------------------------------
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
With regards,
Apache Git Services
[GitHub] [flink] dianfu commented on a change in pull request #10913:
[FLINK-15636][python] Supports Python UDF in old planner under batch mode
Posted by GitBox <gi...@apache.org>.
dianfu commented on a change in pull request #10913: [FLINK-15636][python] Supports Python UDF in old planner under batch mode
URL: https://github.com/apache/flink/pull/10913#discussion_r369403872
##########
File path: flink-python/src/main/java/org/apache/flink/table/runtime/functions/python/PythonScalarFunctionFlatMap.java
##########
@@ -0,0 +1,319 @@
+/*
+ * 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.flink.table.runtime.functions.python;
+
+import org.apache.flink.annotation.Internal;
+import org.apache.flink.api.common.functions.RichFlatMapFunction;
+import org.apache.flink.api.common.typeinfo.TypeInformation;
+import org.apache.flink.api.common.typeutils.TypeSerializer;
+import org.apache.flink.api.java.typeutils.ResultTypeQueryable;
+import org.apache.flink.api.java.typeutils.RowTypeInfo;
+import org.apache.flink.configuration.Configuration;
+import org.apache.flink.configuration.ConfigurationUtils;
+import org.apache.flink.python.PythonConfig;
+import org.apache.flink.python.PythonFunctionRunner;
+import org.apache.flink.python.PythonOptions;
+import org.apache.flink.python.env.ProcessPythonEnvironmentManager;
+import org.apache.flink.python.env.PythonDependencyInfo;
+import org.apache.flink.python.env.PythonEnvironmentManager;
+import org.apache.flink.table.functions.ScalarFunction;
+import org.apache.flink.table.functions.python.PythonEnv;
+import org.apache.flink.table.functions.python.PythonFunctionInfo;
+import org.apache.flink.table.runtime.runners.python.PythonScalarFunctionRunner;
+import org.apache.flink.table.types.logical.RowType;
+import org.apache.flink.table.types.utils.LegacyTypeInfoDataTypeConverter;
+import org.apache.flink.table.types.utils.LogicalTypeDataTypeConverter;
+import org.apache.flink.table.types.utils.TypeConversions;
+import org.apache.flink.types.Row;
+import org.apache.flink.util.Collector;
+import org.apache.flink.util.Preconditions;
+
+import org.apache.beam.sdk.fn.data.FnDataReceiver;
+import org.slf4j.Logger;
+import org.slf4j.LoggerFactory;
+
+import java.io.IOException;
+import java.util.Arrays;
+import java.util.concurrent.LinkedBlockingQueue;
+import java.util.concurrent.atomic.AtomicBoolean;
+import java.util.stream.Collectors;
+
+/**
+ * The {@link RichFlatMapFunction} used to invoke Python {@link ScalarFunction} functions for the
+ * old planner.
+ */
+@Internal
+public final class PythonScalarFunctionFlatMap
+ extends RichFlatMapFunction<Row, Row> implements ResultTypeQueryable<Row> {
+
+ private static final long serialVersionUID = 1L;
+
+ protected static final Logger LOG = LoggerFactory.getLogger(PythonScalarFunctionFlatMap.class);
Review comment:
protected -> private
----------------------------------------------------------------
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
With regards,
Apache Git Services
[GitHub] [flink] flinkbot edited a comment on issue #10913:
[FLINK-15636][python] Supports Python UDF in old planner under batch mode
Posted by GitBox <gi...@apache.org>.
flinkbot edited a comment on issue #10913: [FLINK-15636][python] Supports Python UDF in old planner under batch mode
URL: https://github.com/apache/flink/pull/10913#issuecomment-576547242
<!--
Meta data
Hash:bdaf386d48c4ef0493a0c1c61a7e6aea12f148d1 Status:PENDING URL:https://dev.azure.com/rmetzger/5bd3ef0a-4359-41af-abca-811b04098d2e/_build/results?buildId=4527 TriggerType:PUSH TriggerID:bdaf386d48c4ef0493a0c1c61a7e6aea12f148d1
Hash:bdaf386d48c4ef0493a0c1c61a7e6aea12f148d1 Status:SUCCESS URL:https://travis-ci.com/flink-ci/flink/builds/145316229 TriggerType:PUSH TriggerID:bdaf386d48c4ef0493a0c1c61a7e6aea12f148d1
-->
## CI report:
* bdaf386d48c4ef0493a0c1c61a7e6aea12f148d1 Travis: [SUCCESS](https://travis-ci.com/flink-ci/flink/builds/145316229) Azure: [PENDING](https://dev.azure.com/rmetzger/5bd3ef0a-4359-41af-abca-811b04098d2e/_build/results?buildId=4527)
<details>
<summary>Bot commands</summary>
The @flinkbot bot supports the following commands:
- `@flinkbot run travis` re-run the last Travis build
- `@flinkbot run azure` re-run the last Azure build
</details>
----------------------------------------------------------------
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
With regards,
Apache Git Services
[GitHub] [flink] dianfu commented on a change in pull request #10913:
[FLINK-15636][python] Supports Python UDF in old planner under batch mode
Posted by GitBox <gi...@apache.org>.
dianfu commented on a change in pull request #10913: [FLINK-15636][python] Supports Python UDF in old planner under batch mode
URL: https://github.com/apache/flink/pull/10913#discussion_r369420517
##########
File path: flink-table/flink-table-planner/src/main/scala/org/apache/flink/table/plan/nodes/CommonPythonCalc.scala
##########
@@ -89,4 +89,18 @@ trait CommonPythonCalc {
new PythonFunctionInfo(pythonFunction, inputs.toArray)
}
}
+
+ private[flink] def getPythonRexCalls(calcProgram: RexProgram): Array[RexCall] = {
+ calcProgram.getProjectList
+ .map(calcProgram.expandLocalRef)
+ .collect { case call: RexCall => call }
+ .toArray
+ }
+
+ private [flink] def getForwardFields(calcProgram: RexProgram): Array[Int] = {
Review comment:
getForwardFields -> getForwardedFields
----------------------------------------------------------------
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
With regards,
Apache Git Services
[GitHub] [flink] flinkbot edited a comment on issue #10913:
[FLINK-15636][python] Supports Python UDF in old planner under batch mode
Posted by GitBox <gi...@apache.org>.
flinkbot edited a comment on issue #10913: [FLINK-15636][python] Supports Python UDF in old planner under batch mode
URL: https://github.com/apache/flink/pull/10913#issuecomment-576547242
<!--
Meta data
Hash:bdaf386d48c4ef0493a0c1c61a7e6aea12f148d1 Status:SUCCESS URL:https://dev.azure.com/rmetzger/5bd3ef0a-4359-41af-abca-811b04098d2e/_build/results?buildId=4527 TriggerType:PUSH TriggerID:bdaf386d48c4ef0493a0c1c61a7e6aea12f148d1
Hash:bdaf386d48c4ef0493a0c1c61a7e6aea12f148d1 Status:SUCCESS URL:https://travis-ci.com/flink-ci/flink/builds/145316229 TriggerType:PUSH TriggerID:bdaf386d48c4ef0493a0c1c61a7e6aea12f148d1
-->
## CI report:
* bdaf386d48c4ef0493a0c1c61a7e6aea12f148d1 Travis: [SUCCESS](https://travis-ci.com/flink-ci/flink/builds/145316229) Azure: [SUCCESS](https://dev.azure.com/rmetzger/5bd3ef0a-4359-41af-abca-811b04098d2e/_build/results?buildId=4527)
<details>
<summary>Bot commands</summary>
The @flinkbot bot supports the following commands:
- `@flinkbot run travis` re-run the last Travis build
- `@flinkbot run azure` re-run the last Azure build
</details>
----------------------------------------------------------------
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
With regards,
Apache Git Services
[GitHub] [flink] dianfu commented on a change in pull request #10913:
[FLINK-15636][python] Supports Python UDF in old planner under batch mode
Posted by GitBox <gi...@apache.org>.
dianfu commented on a change in pull request #10913: [FLINK-15636][python] Supports Python UDF in old planner under batch mode
URL: https://github.com/apache/flink/pull/10913#discussion_r369413656
##########
File path: flink-python/src/main/java/org/apache/flink/table/runtime/functions/python/PythonScalarFunctionFlatMap.java
##########
@@ -0,0 +1,319 @@
+/*
+ * 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.flink.table.runtime.functions.python;
+
+import org.apache.flink.annotation.Internal;
+import org.apache.flink.api.common.functions.RichFlatMapFunction;
+import org.apache.flink.api.common.typeinfo.TypeInformation;
+import org.apache.flink.api.common.typeutils.TypeSerializer;
+import org.apache.flink.api.java.typeutils.ResultTypeQueryable;
+import org.apache.flink.api.java.typeutils.RowTypeInfo;
+import org.apache.flink.configuration.Configuration;
+import org.apache.flink.configuration.ConfigurationUtils;
+import org.apache.flink.python.PythonConfig;
+import org.apache.flink.python.PythonFunctionRunner;
+import org.apache.flink.python.PythonOptions;
+import org.apache.flink.python.env.ProcessPythonEnvironmentManager;
+import org.apache.flink.python.env.PythonDependencyInfo;
+import org.apache.flink.python.env.PythonEnvironmentManager;
+import org.apache.flink.table.functions.ScalarFunction;
+import org.apache.flink.table.functions.python.PythonEnv;
+import org.apache.flink.table.functions.python.PythonFunctionInfo;
+import org.apache.flink.table.runtime.runners.python.PythonScalarFunctionRunner;
+import org.apache.flink.table.types.logical.RowType;
+import org.apache.flink.table.types.utils.LegacyTypeInfoDataTypeConverter;
+import org.apache.flink.table.types.utils.LogicalTypeDataTypeConverter;
+import org.apache.flink.table.types.utils.TypeConversions;
+import org.apache.flink.types.Row;
+import org.apache.flink.util.Collector;
+import org.apache.flink.util.Preconditions;
+
+import org.apache.beam.sdk.fn.data.FnDataReceiver;
+import org.slf4j.Logger;
+import org.slf4j.LoggerFactory;
+
+import java.io.IOException;
+import java.util.Arrays;
+import java.util.concurrent.LinkedBlockingQueue;
+import java.util.concurrent.atomic.AtomicBoolean;
+import java.util.stream.Collectors;
+
+/**
+ * The {@link RichFlatMapFunction} used to invoke Python {@link ScalarFunction} functions for the
+ * old planner.
+ */
+@Internal
+public final class PythonScalarFunctionFlatMap
+ extends RichFlatMapFunction<Row, Row> implements ResultTypeQueryable<Row> {
+
+ private static final long serialVersionUID = 1L;
+
+ protected static final Logger LOG = LoggerFactory.getLogger(PythonScalarFunctionFlatMap.class);
+
+ /**
+ * The type serializer for the forwarded fields.
+ */
+ private transient TypeSerializer<Row> forwardedInputSerializer;
+
+ /**
+ * The Python {@link ScalarFunction}s to be executed.
+ */
+ protected final PythonFunctionInfo[] scalarFunctions;
+
+ /**
+ * The input logical type.
+ */
+ protected final RowType inputType;
+
+ /**
+ * The output logical type.
+ */
+ protected final RowType outputType;
+
+ /**
+ * The offsets of udf inputs.
+ */
+ protected final int[] udfInputOffsets;
+
+ /**
+ * The offset of the fields which should be forwarded.
+ */
+ protected final int[] forwardedFields;
+
+ /**
+ * The udf input logical type.
+ */
+ protected transient RowType udfInputType;
+
+ /**
+ * The udf output logical type.
+ */
+ protected transient RowType udfOutputType;
+
+ /**
+ * The queue holding the input elements for which the execution results have not been received.
+ */
+ protected transient LinkedBlockingQueue<Row> forwardedInputQueue;
+
+ /**
+ * The queue holding the user-defined function execution results. The execution results are in
+ * the same order as the input elements.
+ */
+ protected transient LinkedBlockingQueue<Row> udfResultQueue;
+
+ /**
+ * The python config.
+ */
+ private final PythonConfig config;
+
+ /**
+ * Use an AtomicBoolean because we start/stop bundles by a timer thread.
+ */
+ private transient AtomicBoolean bundleStarted;
+
+ /**
+ * Max number of elements to include in a bundle.
+ */
+ private transient int maxBundleSize;
+
+ /**
+ * The collector used to collect records.
+ */
+ private transient Collector<Row> resultCollector;
+
+ /**
+ * Number of processed elements in the current bundle.
+ */
+ private transient int elementCount;
+
+ /**
+ * The {@link PythonFunctionRunner} which is responsible for Python user-defined function execution.
+ */
+ private transient PythonFunctionRunner<Row> pythonFunctionRunner;
+
+ public PythonScalarFunctionFlatMap(
+ Configuration config,
+ PythonFunctionInfo[] scalarFunctions,
+ RowType inputType,
+ RowType outputType,
+ int[] udfInputOffsets,
+ int[] forwardedFields) {
+ this.scalarFunctions = Preconditions.checkNotNull(scalarFunctions);
+ this.inputType = Preconditions.checkNotNull(inputType);
+ this.outputType = Preconditions.checkNotNull(outputType);
+ this.udfInputOffsets = Preconditions.checkNotNull(udfInputOffsets);
+ this.forwardedFields = Preconditions.checkNotNull(forwardedFields);
+ this.config = new PythonConfig(Preconditions.checkNotNull(config));
+ }
+
+ @Override
+ public void open(Configuration parameters) throws Exception {
+ super.open(parameters);
+
+ this.elementCount = 0;
+ this.bundleStarted = new AtomicBoolean(false);
+ this.maxBundleSize = config.getMaxBundleSize();
+ if (this.maxBundleSize <= 0) {
+ this.maxBundleSize = PythonOptions.MAX_BUNDLE_SIZE.defaultValue();
+ LOG.error("Invalid value for the maximum bundle size. Using default value of " +
+ this.maxBundleSize + '.');
+ } else {
+ LOG.info("The maximum bundle size is configured to {}.", this.maxBundleSize);
+ }
+
+ if (config.getMaxBundleTimeMills() != PythonOptions.MAX_BUNDLE_TIME_MILLS.defaultValue()) {
+ LOG.info("Maximum bundle time takes no effect in old planner under batch mode. " +
+ "Config maximum bundle size instead! " +
+ "Under batch mode, bundle size should be enough to control both throughput and latency.");
+ }
+
+ forwardedInputQueue = new LinkedBlockingQueue<>();
+ udfResultQueue = new LinkedBlockingQueue<>();
+ udfInputType = new RowType(
+ Arrays.stream(udfInputOffsets)
+ .mapToObj(i -> inputType.getFields().get(i))
+ .collect(Collectors.toList()));
+ udfOutputType = new RowType(outputType.getFields().subList(forwardedFields.length, outputType.getFieldCount()));
+
+ RowTypeInfo forwardedInputTypeInfo = new RowTypeInfo(
+ Arrays.stream(forwardedFields)
+ .mapToObj(i -> inputType.getFields().get(i))
+ .map(RowType.RowField::getType)
+ .map(TypeConversions::fromLogicalToDataType)
+ .map(TypeConversions::fromDataTypeToLegacyInfo)
+ .toArray(TypeInformation[]::new));
+ forwardedInputSerializer = forwardedInputTypeInfo.createSerializer(getRuntimeContext().getExecutionConfig());
+
+ this.pythonFunctionRunner = createPythonFunctionRunner();
+ this.pythonFunctionRunner.open();
+ this.resultCollector = null;
+ }
+
+ @Override
+ public void flatMap(Row value, Collector<Row> out) throws Exception {
+ this.resultCollector = out;
+ bufferInput(value);
+
+ checkInvokeStartBundle();
+ pythonFunctionRunner.processElement(getUdfInput(value));
+ checkInvokeFinishBundleByCount();
+ emitResults();
+ }
+
+ /**
+ * Checks whether to invoke startBundle.
+ */
+ private void checkInvokeStartBundle() throws Exception {
+ if (bundleStarted.compareAndSet(false, true)) {
+ pythonFunctionRunner.startBundle();
+ }
+ }
+
+ /**
+ * Checks whether to invoke finishBundle by elements count. Called in processElement.
Review comment:
Correct the Java doc: Called in processElement -> Called in flatMap
----------------------------------------------------------------
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
With regards,
Apache Git Services