Github user fhueske commented on a diff in the pull request:
https://github.com/apache/flink/pull/3715#discussion_r117006282
--- Diff:
flink-libraries/flink-table/src/main/scala/org/apache/flink/table/runtime/join/JoinUtil.scala
---
@@ -0,0 +1,488 @@
+/*
+ * 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.join
+
+import java.math.{BigDecimal => JBigDecimal}
+import java.util
+import java.util.EnumSet
+
+import org.apache.calcite.avatica.util.TimeUnit
+import org.apache.calcite.rel.`type`.RelDataType
+import org.apache.calcite.rel.core.JoinRelType
+import org.apache.calcite.rex._
+import org.apache.calcite.sql.fun.SqlStdOperatorTable
+import org.apache.calcite.sql.parser.SqlParserPos
+import org.apache.calcite.sql.{SqlIntervalQualifier, SqlKind}
+import org.apache.flink.api.common.functions.FilterFunction
+import org.apache.flink.api.common.typeinfo.{BasicTypeInfo,
TypeInformation}
+import org.apache.flink.table.api.{TableConfig, TableException}
+import org.apache.flink.table.calcite.FlinkTypeFactory
+import org.apache.flink.table.codegen.{CodeGenerator, ExpressionReducer}
+import org.apache.flink.table.plan.nodes.logical.FlinkLogicalJoin
+import org.apache.flink.table.runtime.FilterRunner
+import org.apache.flink.types.Row
+
+import scala.collection.JavaConversions._
+import scala.collection.mutable.ArrayBuffer
+
+
+object JoinUtil {
+
+ /**
+ * Analyze join condition to get equi-conditon and other condition
+ * @param joinNode logicaljoin node
+ * @param expression the function to generate condition string
+ */
+ private[flink] def analyzeJoinCondition(
+ joinNode: FlinkLogicalJoin,
+ expression: (RexNode, List[String], Option[List[RexNode]]) => String)
= {
+
+ val joinInfo = joinNode.analyzeCondition()
+ val keyPairs = joinInfo.pairs.toList
+ val otherCondition =
+ if(joinInfo.isEqui) null
+ else joinInfo.getRemaining(joinNode.getCluster.getRexBuilder)
+
+ val leftKeys = ArrayBuffer.empty[Int]
+ val rightKeys = ArrayBuffer.empty[Int]
+ if (!keyPairs.isEmpty) {
+ val leftFields = joinNode.getLeft.getRowType.getFieldList
+ val rightFields = joinNode.getRight.getRowType.getFieldList
+
+ keyPairs.foreach(pair => {
+ val leftKeyType =
leftFields.get(pair.source).getType.getSqlTypeName
+ val rightKeyType =
rightFields.get(pair.target).getType.getSqlTypeName
+
+ // check if keys are compatible
+ if (leftKeyType == rightKeyType) {
+ // add key pair
+ leftKeys.append(pair.source)
+ rightKeys.append(pair.target)
+ } else {
+ throw TableException(
+ "Equality join predicate on incompatible types.\n" +
+ s"\tLeft: ${joinNode.getLeft.toString},\n" +
+ s"\tRight: ${joinNode.getRight.toString},\n" +
+ s"\tCondition: (${expression(joinNode.getCondition,
+ joinNode.getRowType.getFieldNames.toList, None)})"
+ )
+ }
+ })
+ }
+ (leftKeys.toArray, rightKeys.toArray, otherCondition)
+ }
+
+ /**
+ * Analyze time-condtion to get time boundary for each stream and get
the time type
+ * and return condition without time-condition.
+ *
+ * @param condition other condtion include time-condition
+ * @param leftFieldCount left stream fields count
+ * @param inputType left and right connect stream type
+ * @param rexBuilder util to build rexNode
+ * @param config table environment config
+ */
+ private[flink] def analyzeTimeBoundary(
--- End diff --
I think the logic of this function is correct. However, I find it a bit
hard to follow because it starts with many conditions.
What do you think about the following approach:
1. convert condition to CNF
2. go over all conjunctive terms and split them into those referencing a
time field and those that don't reference a time field (those latter terms are
the remaining conditions)
3. Check that there are only two such terms, throw an exception otherwise
4. Check for each conjunctive term that is references time fields of both
inputs exactly once. If there is one that does not, throw an exception.
5. Extract the offset from each of the two terms (this is the complex step)
This way, we keep the logic simple until the end and don't need to
distinguish between greater / smaller, left input, right input, etc. from the
start.
What do you think?
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at infrastruct...@apache.org or file a JIRA ticket
with INFRA.
---
