[GitHub] [flink] dianfu commented on a change in pull request #9748: [FLINK-14016][python][flink-table-planner] Introduce DataStreamPythonCalc for Python function execution

[GitBox <gi...@apache.org>]

dianfu commented on a change in pull request #9748: [FLINK-14016][python][flink-table-planner] Introduce DataStreamPythonCalc for Python function execution
URL: https://github.com/apache/flink/pull/9748#discussion_r328420024
 
 

 ##########
 File path: flink-table/flink-table-planner/src/main/scala/org/apache/flink/table/plan/rules/logical/PythonScalarFunctionSplitRule.scala
 ##########
 @@ -0,0 +1,211 @@
+/*
+ * 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.plan.rules.logical
+
+import org.apache.calcite.plan.RelOptRule.{any, operand}
+import org.apache.calcite.plan.{RelOptRule, RelOptRuleCall}
+import org.apache.calcite.rex.{RexCall, RexInputRef, RexNode, RexProgram}
+import org.apache.calcite.sql.validate.SqlValidatorUtil
+import org.apache.flink.table.functions.FunctionLanguage
+import org.apache.flink.table.functions.ScalarFunction
+import org.apache.flink.table.functions.utils.ScalarSqlFunction
+import org.apache.flink.table.plan.nodes.logical.FlinkLogicalCalc
+import org.apache.flink.table.plan.util.PythonUtil.containsFunctionOf
+import org.apache.flink.table.plan.util.{InputRefVisitor, RexDefaultVisitor}
+import org.apache.flink.table.plan.rules.logical.PythonScalarFunctionSplitRule.extractRefInputFields
+
+import scala.collection.JavaConverters._
+import scala.collection.JavaConversions._
+import scala.collection.mutable
+
+/**
+  * Rule that splits [[FlinkLogicalCalc]] into multiple [[FlinkLogicalCalc]]s. After this rule
+  * is applied, each [[FlinkLogicalCalc]] will only contain Python [[ScalarFunction]]s or Java
+  * [[ScalarFunction]]s. This is to ensure that the Python [[ScalarFunction]]s which could be
+  * executed in a batch are grouped into the same [[FlinkLogicalCalc]] node.
+  */
+class PythonScalarFunctionSplitRule extends RelOptRule(
+  operand(classOf[FlinkLogicalCalc], any),
+  "PythonScalarFunctionSplitRule") {
+
+  override def matches(call: RelOptRuleCall): Boolean = {
+    val calc: FlinkLogicalCalc = call.rel(0).asInstanceOf[FlinkLogicalCalc]
+    val program = calc.getProgram
+    program.getExprList.exists(containsFunctionOf(_, FunctionLanguage.PYTHON)) &&
+    program.getExprList.exists(containsFunctionOf(_, FunctionLanguage.JVM))
+  }
+
+  override def onMatch(call: RelOptRuleCall): Unit = {
+    val calc: FlinkLogicalCalc = call.rel(0).asInstanceOf[FlinkLogicalCalc]
+    val input = calc.getInput
+    val rexBuilder = call.builder().getRexBuilder
+    val program = calc.getProgram
+    val extractedRexCalls = new mutable.ArrayBuffer[RexCall]()
+
+    val outerCallContainsJavaFunction =
+      program.getProjectList
+        .map(program.expandLocalRef)
+        .exists(containsFunctionOf(_, FunctionLanguage.JVM, recursive = false)) ||
+      Option(program.getCondition)
+        .map(program.expandLocalRef)
+        .exists(containsFunctionOf(_, FunctionLanguage.JVM, recursive = false))
+
+    val extractedFunctionOffset = input.getRowType.getFieldCount
+    val splitter = new ScalarFunctionSplitter(
+      extractedFunctionOffset,
+      extractedRexCalls,
+      outerCallContainsJavaFunction)
+
+    val newProjects = program.getProjectList
+      .map(program.expandLocalRef)
+      .map(_.accept(splitter))
+
+    val newCondition = Option(program.getCondition)
+      .map(program.expandLocalRef)
+      .map(_.accept(splitter))
+
+    val accessedFields = extractRefInputFields(
+      newProjects, newCondition, extractedFunctionOffset)
+
+    val bottomCalcProjects =
+      accessedFields.map(RexInputRef.of(_, input.getRowType)) ++ extractedRexCalls
+    val bottomCalcFieldNames = SqlValidatorUtil.uniquify(
+      accessedFields.map(i => input.getRowType.getFieldNames.get(i)).toSeq ++
+        extractedRexCalls.indices.map("f" + _),
+      rexBuilder.getTypeFactory.getTypeSystem.isSchemaCaseSensitive)
+
+    val bottomCalc = new FlinkLogicalCalc(
 
 Review comment:
   Good catch. Fixed.

----------------------------------------------------------------
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