flink-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From huawei-flink <...@git.apache.org>
Subject [GitHub] flink pull request #3732: [FLINK-6250] Distinct procTime with Rows boundarie...
Date Wed, 19 Apr 2017 14:03:03 GMT
Github user huawei-flink commented on a diff in the pull request:

    https://github.com/apache/flink/pull/3732#discussion_r112210389
  
    --- Diff: flink-libraries/flink-table/src/main/scala/org/apache/flink/table/runtime/aggregate/ProcTimeBoundedDistinctRowsOver.scala
---
    @@ -0,0 +1,238 @@
    +/*
    + * 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.aggregate
    +
    +import java.util
    +
    +import org.apache.flink.configuration.Configuration
    +import org.apache.flink.streaming.api.functions.ProcessFunction
    +import org.apache.flink.types.Row
    +import org.apache.flink.util.{Collector, Preconditions}
    +import org.apache.flink.api.common.state.ValueStateDescriptor
    +import org.apache.flink.api.java.typeutils.RowTypeInfo
    +import org.apache.flink.api.common.state.ValueState
    +import org.apache.flink.table.functions.{Accumulator, AggregateFunction}
    +import org.apache.flink.api.common.state.MapState
    +import org.apache.flink.api.common.state.MapStateDescriptor
    +import org.apache.flink.api.common.typeinfo.TypeInformation
    +import org.apache.flink.api.java.typeutils.ListTypeInfo
    +import java.util.{List => JList}
    +
    +import org.apache.flink.api.common.typeinfo.BasicTypeInfo
    +
    +class ProcTimeBoundedDistinctRowsOver(
    +  private val aggregates: Array[AggregateFunction[_]],
    +  private val aggFields: Array[Array[Int]],
    +  private val distinctAggsFlag: Array[Boolean],
    +  private val precedingOffset: Long,
    +  private val forwardedFieldCount: Int,
    +  private val aggregatesTypeInfo: RowTypeInfo,
    +  private val inputType: TypeInformation[Row])
    +    extends ProcessFunction[Row, Row] {
    +
    +  Preconditions.checkNotNull(aggregates)
    +  Preconditions.checkNotNull(aggFields)
    +  Preconditions.checkNotNull(distinctAggsFlag)
    +  Preconditions.checkNotNull(distinctAggsFlag.length == aggregates.length)
    +  Preconditions.checkArgument(aggregates.length == aggFields.length)
    +  Preconditions.checkArgument(precedingOffset > 0)
    +
    +  private var accumulatorState: ValueState[Row] = _
    +  private var rowMapState: MapState[Long, JList[Row]] = _
    +  private var output: Row = _
    +  private var counterState: ValueState[Long] = _
    +  private var smallestTsState: ValueState[Long] = _
    +  private var distinctValueState: MapState[Any, Row] = _
    +
    +  override def open(config: Configuration) {
    +
    +    output = new Row(forwardedFieldCount + aggregates.length)
    +    // We keep the elements received in a Map state keyed
    +    // by the ingestion time in the operator.
    +    // we also keep counter of processed elements
    +    // and timestamp of oldest element
    +    val rowListTypeInfo: TypeInformation[JList[Row]] =
    +      new ListTypeInfo[Row](inputType).asInstanceOf[TypeInformation[JList[Row]]]
    +
    +    val mapStateDescriptor: MapStateDescriptor[Long, JList[Row]] =
    +      new MapStateDescriptor[Long, JList[Row]]("windowBufferMapState",
    +        BasicTypeInfo.LONG_TYPE_INFO.asInstanceOf[TypeInformation[Long]], rowListTypeInfo)
    +    rowMapState = getRuntimeContext.getMapState(mapStateDescriptor)
    +
    +    val aggregationStateDescriptor: ValueStateDescriptor[Row] =
    +      new ValueStateDescriptor[Row]("aggregationState", aggregatesTypeInfo)
    +    accumulatorState = getRuntimeContext.getState(aggregationStateDescriptor)
    +
    +    val processedCountDescriptor : ValueStateDescriptor[Long] =
    +       new ValueStateDescriptor[Long]("processedCountState", classOf[Long])
    +    counterState = getRuntimeContext.getState(processedCountDescriptor)
    +
    +    val smallestTimestampDescriptor : ValueStateDescriptor[Long] =
    +       new ValueStateDescriptor[Long]("smallestTSState", classOf[Long])
    +    smallestTsState = getRuntimeContext.getState(smallestTimestampDescriptor)
    +    
    +    val distinctValDescriptor : MapStateDescriptor[Any, Row] =
    +      new MapStateDescriptor[Any, Row]("distinctValuesBufferMapState", classOf[Any],
classOf[Row])
    +    distinctValueState = getRuntimeContext.getMapState(distinctValDescriptor)
    +  }
    +
    +  override def processElement(
    +    input: Row,
    +    ctx: ProcessFunction[Row, Row]#Context,
    +    out: Collector[Row]): Unit = {
    +
    +    val currentTime = ctx.timerService.currentProcessingTime
    +    var i = 0
    +
    +    // initialize state for the processed element
    +    var accumulators = accumulatorState.value
    +    if (accumulators == null) {
    +      accumulators = new Row(aggregates.length)
    +      while (i < aggregates.length) {
    +        accumulators.setField(i, aggregates(i).createAccumulator())
    +        i += 1
    +      }
    +    }
    +
    +    // get smallest timestamp
    +    var smallestTs = smallestTsState.value
    +    if (smallestTs == 0L) {
    +      smallestTs = currentTime
    +      smallestTsState.update(smallestTs)
    +    }
    +    // get previous counter value
    +    var counter = counterState.value
    +
    +    if (counter == precedingOffset) {
    +      val retractList = rowMapState.get(smallestTs)
    +
    +      // get oldest element beyond buffer size
    +      // and if oldest element exist, retract value
    +      var removeCounter :Integer = 0
    +      var distinctCounter : Integer = 0
    +      var retractVal : Object = null
    +      i = 0
    +      while (i < aggregates.length) {
    +        val accumulator = accumulators.getField(i).asInstanceOf[Accumulator]
    +        retractVal = retractList.get(0).getField(aggFields(i)(0))
    +        if(distinctAggsFlag(i)){
    +          distinctCounter += 1
    +          val counterRow = distinctValueState.get(retractVal)
    --- End diff --
    
    the test you suggested indeed failed. great point! the new version passes the problem.


---
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 infrastructure@apache.org or file a JIRA ticket
with INFRA.
---

Mime
View raw message