flink-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From GitBox <...@apache.org>
Subject [GitHub] [flink] hequn8128 commented on a change in pull request #9890: [FLINK-14272][python][table-planner-blink] Support Blink planner for Python UDF
Date Thu, 17 Oct 2019 09:26:48 GMT
hequn8128 commented on a change in pull request #9890: [FLINK-14272][python][table-planner-blink]
Support Blink planner for Python UDF
URL: https://github.com/apache/flink/pull/9890#discussion_r335893209
 
 

 ##########
 File path: flink-table/flink-table-planner-blink/src/main/scala/org/apache/flink/table/planner/plan/nodes/common/CommonPythonCalc.scala
 ##########
 @@ -0,0 +1,184 @@
+/*
+ * 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.planner.plan.nodes.common
+
+import org.apache.calcite.plan.RelOptCluster
+import org.apache.calcite.rex.{RexCall, RexInputRef, RexLiteral, RexNode, RexProgram}
+import org.apache.calcite.sql.`type`.SqlTypeName
+import org.apache.flink.api.dag.Transformation
+import org.apache.flink.streaming.api.operators.OneInputStreamOperator
+import org.apache.flink.streaming.api.transformations.OneInputTransformation
+import org.apache.flink.table.dataformat.BaseRow
+import org.apache.flink.table.functions.FunctionLanguage
+import org.apache.flink.table.functions.python.{PythonFunction, PythonFunctionInfo, SimplePythonFunction}
+import org.apache.flink.table.planner.calcite.{FlinkTypeFactory, FlinkTypeSystem}
+import org.apache.flink.table.planner.functions.utils.ScalarSqlFunction
+import org.apache.flink.table.planner.plan.nodes.common.CommonPythonCalc.PYTHON_SCALAR_FUNCTION_OPERATOR_NAME
+import org.apache.flink.table.runtime.typeutils.BaseRowTypeInfo
+import org.apache.flink.table.types.logical.RowType
+
+import scala.collection.JavaConversions._
+import scala.collection.mutable
+
+trait CommonPythonCalc {
+
+  private lazy val convertLiteralToPython = {
+    val clazz = Class.forName("org.apache.flink.api.common.python.PythonBridgeUtils")
+    clazz.getMethod("convertLiteralToPython", classOf[RexLiteral], classOf[SqlTypeName])
+  }
+
+  private[flink] def extractPythonScalarFunctionInfos(
+      rexCalls: Array[RexCall]): (Array[Int], Array[PythonFunctionInfo]) = {
+    // using LinkedHashMap to keep the insert order
+    val inputNodes = new mutable.LinkedHashMap[RexNode, Integer]()
+    val pythonFunctionInfos = rexCalls.map(createPythonScalarFunctionInfo(_, inputNodes))
+
+    val udfInputOffsets = inputNodes.toArray
+      .map(_._1)
+      .filter(_.isInstanceOf[RexInputRef])
+      .map(_.asInstanceOf[RexInputRef].getIndex)
+    (udfInputOffsets, pythonFunctionInfos)
+  }
+
+  private[flink] def createPythonScalarFunctionInfo(
+      rexCall: RexCall,
+      inputNodes: mutable.Map[RexNode, Integer]): PythonFunctionInfo = rexCall.getOperator
match {
+    case sfc: ScalarSqlFunction if sfc.scalarFunction.getLanguage == FunctionLanguage.PYTHON
=>
+      val inputs = new mutable.ArrayBuffer[AnyRef]()
+      rexCall.getOperands.foreach {
+        case pythonRexCall: RexCall if pythonRexCall.getOperator.asInstanceOf[ScalarSqlFunction]
+          .scalarFunction.getLanguage == FunctionLanguage.PYTHON =>
+          // Continuous Python UDFs can be chained together
+          val argPythonInfo = createPythonScalarFunctionInfo(pythonRexCall, inputNodes)
+          inputs.append(argPythonInfo)
+
+        case literal: RexLiteral =>
+          inputs.append(
+            convertLiteralToPython.invoke(null, literal, literal.getType.getSqlTypeName))
+
+        case argNode: RexNode =>
+          // For input arguments of RexInputRef, it's replaced with an offset into the input
row
+          inputNodes.get(argNode) match {
+            case Some(existing) => inputs.append(existing)
+            case None =>
+              val inputOffset = Integer.valueOf(inputNodes.size)
+              inputs.append(inputOffset)
+              inputNodes.put(argNode, inputOffset)
+          }
+      }
+
+      // Extracts the necessary information for Python function execution, such as
+      // the serialized Python function, the Python env, etc
+      val pythonFunction = new SimplePythonFunction(
+        sfc.scalarFunction.asInstanceOf[PythonFunction].getSerializedPythonFunction,
+        sfc.scalarFunction.asInstanceOf[PythonFunction].getPythonEnv)
+      new PythonFunctionInfo(pythonFunction, inputs.toArray)
+  }
+
+  private[flink] def getPythonScalarFunctionOperator(
+      inputRowTypeInfo: BaseRowTypeInfo,
+      outputRowTypeInfo: BaseRowTypeInfo,
+      udfInputOffsets: Array[Int],
+      pythonFunctionInfos: Array[PythonFunctionInfo],
+      forwardedFields: Array[Int])= {
+    val clazz = Class.forName(PYTHON_SCALAR_FUNCTION_OPERATOR_NAME)
+    val ctor = clazz.getConstructor(
+      classOf[Array[PythonFunctionInfo]],
+      classOf[RowType],
+      classOf[RowType],
+      classOf[Array[Int]],
+      classOf[Array[Int]])
+    ctor.newInstance(
+      pythonFunctionInfos,
+      inputRowTypeInfo.toRowType,
+      outputRowTypeInfo.toRowType,
+      udfInputOffsets,
+      forwardedFields)
+      .asInstanceOf[OneInputStreamOperator[BaseRow, BaseRow]]
+  }
+
+  private [flink] def generatePythonOneInputStream(
 
 Review comment:
   Maybe change to protected? Same for `createProjectionRexProgram`

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

Mime
View raw message