[ https://issues.apache.org/jira/browse/FLINK-6233?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16191911#comment-16191911 ]
ASF GitHub Bot commented on FLINK-6233: --------------------------------------- Github user fhueske commented on a diff in the pull request: https://github.com/apache/flink/pull/4625#discussion_r142705185 --- Diff: flink-libraries/flink-table/src/main/scala/org/apache/flink/table/runtime/join/TimeBoundedStreamInnerJoin.scala --- @@ -0,0 +1,410 @@ +/* + * 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.util.{ArrayList, List => JList} + +import org.apache.flink.api.common.functions.FlatJoinFunction +import org.apache.flink.api.common.state._ +import org.apache.flink.api.common.typeinfo.TypeInformation +import org.apache.flink.api.java.typeutils.ListTypeInfo +import org.apache.flink.configuration.Configuration +import org.apache.flink.streaming.api.functions.co.CoProcessFunction +import org.apache.flink.table.api.Types +import org.apache.flink.table.codegen.Compiler +import org.apache.flink.table.runtime.CRowWrappingCollector +import org.apache.flink.table.runtime.types.CRow +import org.apache.flink.table.util.Logging +import org.apache.flink.types.Row +import org.apache.flink.util.Collector + +/** + * A CoProcessFunction to execute time-bounded stream inner-join. + * Two kinds of time criteria: + * "L.time between R.time + X and R.time + Y" or "R.time between L.time - Y and L.time - X". + * + * @param leftLowerBound the lower bound for the left stream (X in the criteria) + * @param leftUpperBound the upper bound for the left stream (Y in the criteria) + * @param allowedLateness the lateness allowed for the two streams + * @param leftType the input type of left stream + * @param rightType the input type of right stream + * @param genJoinFuncName the function code of other non-equi conditions + * @param genJoinFuncCode the function name of other non-equi conditions + * + */ +abstract class TimeBoundedStreamInnerJoin( + private val leftLowerBound: Long, + private val leftUpperBound: Long, + private val allowedLateness: Long, + private val leftType: TypeInformation[Row], + private val rightType: TypeInformation[Row], + private val genJoinFuncName: String, + private val genJoinFuncCode: String, + private val leftTimeIdx: Int, + private val rightTimeIdx: Int) + extends CoProcessFunction[CRow, CRow, CRow] + with Compiler[FlatJoinFunction[Row, Row, Row]] + with Logging { + + private var cRowWrapper: CRowWrappingCollector = _ + + // the join function for other conditions + private var joinFunction: FlatJoinFunction[Row, Row, Row] = _ + + // cache to store rows from the left stream + private var leftCache: MapState[Long, JList[Row]] = _ + // cache to store rows from the right stream + private var rightCache: MapState[Long, JList[Row]] = _ + + // state to record the timer on the left stream. 0 means no timer set + private var leftTimerState: ValueState[Long] = _ + // state to record the timer on the right stream. 0 means no timer set + private var rightTimerState: ValueState[Long] = _ + + private val leftRelativeSize: Long = -leftLowerBound + private val rightRelativeSize: Long = leftUpperBound + + private var leftExpirationTime: Long = 0L; + private var rightExpirationTime: Long = 0L; + + protected var leftOperatorTime: Long = 0L + protected var rightOperatorTime: Long = 0L + + + // for delayed cleanup + private val cleanupDelay = (leftRelativeSize + rightRelativeSize) / 2 + + if (allowedLateness < 0) { + throw new IllegalArgumentException("The allowed lateness must be non-negative.") + } + + /** + * Get the maximum interval between receiving a row and emitting it (as part of a joined result). + * Only reasonable for row time join. + * + * @return the maximum delay for the outputs + */ + def getMaxOutputDelay: Long = Math.max(leftRelativeSize, rightRelativeSize) + allowedLateness + + override def open(config: Configuration) { + LOG.debug(s"Compiling JoinFunction: $genJoinFuncName \n\n " + + s"Code:\n$genJoinFuncCode") + val clazz = compile( + getRuntimeContext.getUserCodeClassLoader, + genJoinFuncName, + genJoinFuncCode) + LOG.debug("Instantiating JoinFunction.") + joinFunction = clazz.newInstance() + + cRowWrapper = new CRowWrappingCollector() + cRowWrapper.setChange(true) + + // Initialize the data caches. + val leftListTypeInfo: TypeInformation[JList[Row]] = new ListTypeInfo[Row](leftType) + val leftStateDescriptor: MapStateDescriptor[Long, JList[Row]] = + new MapStateDescriptor[Long, JList[Row]]( + "InnerJoinLeftCache", + Types.LONG.asInstanceOf[TypeInformation[Long]], + leftListTypeInfo) + leftCache = getRuntimeContext.getMapState(leftStateDescriptor) + + val rightListTypeInfo: TypeInformation[JList[Row]] = new ListTypeInfo[Row](rightType) + val rightStateDescriptor: MapStateDescriptor[Long, JList[Row]] = + new MapStateDescriptor[Long, JList[Row]]( + "InnerJoinRightCache", + Types.LONG.asInstanceOf[TypeInformation[Long]], + rightListTypeInfo) + rightCache = getRuntimeContext.getMapState(rightStateDescriptor) + + // Initialize the timer states. + val leftTimerStateDesc: ValueStateDescriptor[Long] = + new ValueStateDescriptor[Long]("InnerJoinLeftTimerState", classOf[Long]) + leftTimerState = getRuntimeContext.getState(leftTimerStateDesc) + + val rightTimerStateDesc: ValueStateDescriptor[Long] = + new ValueStateDescriptor[Long]("InnerJoinRightTimerState", classOf[Long]) + rightTimerState = getRuntimeContext.getState(rightTimerStateDesc) + } + + /** + * Process rows from the left stream. + */ + override def processElement1( + cRowValue: CRow, + ctx: CoProcessFunction[CRow, CRow, CRow]#Context, + out: Collector[CRow]): Unit = { + updateOperatorTime(ctx) + val leftRow = cRowValue.row + val timeForLeftRow: Long = getTimeForLeftStream(ctx, leftRow) + val rightQualifiedLowerBound: Long = timeForLeftRow - rightRelativeSize + val rightQualifiedUpperBound: Long = timeForLeftRow + leftRelativeSize + cRowWrapper.out = out + --- End diff -- Add a comment: "Check if we need to cache the current row." > Support rowtime inner equi-join between two streams in the SQL API > ------------------------------------------------------------------ > > Key: FLINK-6233 > URL: https://issues.apache.org/jira/browse/FLINK-6233 > Project: Flink > Issue Type: Sub-task > Components: Table API & SQL > Reporter: hongyuhong > Assignee: Xingcan Cui > > The goal of this issue is to add support for inner equi-join on proc time > streams to the SQL interface. > Queries similar to the following should be supported: > {code} > SELECT o.rowtime , o.productId, o.orderId, s.rowtime AS shipTime > FROM Orders AS o > JOIN Shipments AS s > ON o.orderId = s.orderId > AND o.rowtime BETWEEN s.rowtime AND s.rowtime + INTERVAL '1' HOUR; > {code} > The following restrictions should initially apply: > * The join hint only support inner join > * The ON clause should include equi-join condition > * The time-condition {{o.rowtime BETWEEN s.rowtime AND s.rowtime + INTERVAL > '1' HOUR}} only can use rowtime that is a system attribute, the time > condition only support bounded time range like {{o.rowtime BETWEEN s.rowtime > - INTERVAL '1' HOUR AND s.rowtime + INTERVAL '1' HOUR}}, not support > unbounded like {{o.rowtime < s.rowtime}} , and should include both two > stream's rowtime attribute, {{o.rowtime between rowtime () and rowtime () + > 1}} should also not be supported. > An row-time streams join will not be able to handle late data, because this > would mean in insert a row into a sorted order shift all other computations. > This would be too expensive to maintain. Therefore, we will throw an error if > a user tries to use an row-time stream join with late data handling. > This issue includes: > * Design of the DataStream operator to deal with stream join > * Translation from Calcite's RelNode representation (LogicalJoin). -- This message was sent by Atlassian JIRA (v6.4.14#64029)