Mailing-List: contact issues-help@flink.apache.org; run by ezmlm
Precedence: bulk
Reply-To: dev@flink.apache.org
From: fhueske <git@git.apache.org>
To: issues@flink.incubator.apache.org
Reply-To: issues@flink.incubator.apache.org
References: <git-pr-3641-flink@git.apache.org>
In-Reply-To: <git-pr-3641-flink@git.apache.org>
Subject: [GitHub] flink pull request #3641: [FLINK-5654] - Add processing time OVER RANGE BETW...
Content-Type: text/plain
Message-Id: <20170329165956.24492E9652@git1-us-west.apache.org>
Date: Wed, 29 Mar 2017 16:59:56 +0000 (UTC)
archived-at: Wed, 29 Mar 2017 17:00:07 -0000

Github user fhueske commented on a diff in the pull request:

    https://github.com/apache/flink/pull/3641#discussion_r108727511
  
    --- Diff: flink-libraries/flink-table/src/main/scala/org/apache/flink/table/runtime/aggregate/ProcTimeBoundedProcessingOverProcessFunction.scala ---
    @@ -0,0 +1,166 @@
    +/*
    + * 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 org.apache.flink.api.common.state.{ ListState, ListStateDescriptor }
    +import org.apache.flink.api.java.typeutils.RowTypeInfo
    +import org.apache.flink.configuration.Configuration
    +import org.apache.flink.runtime.state.{ FunctionInitializationContext, FunctionSnapshotContext }
    +import org.apache.flink.streaming.api.functions.ProcessFunction
    +import org.apache.flink.table.functions.{ Accumulator, AggregateFunction }
    +import org.apache.flink.types.Row
    +import org.apache.flink.util.{ Collector, Preconditions }
    +import org.apache.flink.api.common.state.ValueState
    +import org.apache.flink.api.common.state.ValueStateDescriptor
    +import scala.util.control.Breaks._
    +import org.apache.flink.api.java.tuple.{ Tuple2 => JTuple2 }
    +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.{ ArrayList, LinkedList, List => JList }
    +import org.apache.flink.api.common.typeinfo.BasicTypeInfo
    +
    +/**
    + * Process Function used for the aggregate in bounded proc-time OVER window
    + * [[org.apache.flink.streaming.api.datastream.DataStream]]
    + *
    + * @param aggregates the list of all [[org.apache.flink.table.functions.AggregateFunction]]
    + *                   used for this aggregation
    + * @param aggFields  the position (in the input Row) of the input value for each aggregate
    + * @param forwardedFieldCount Is used to indicate fields in the current element to forward
    + * @param rowTypeInfo Is used to indicate the field schema
    + * @param timeBoundary Is used to indicate the processing time boundaries
    + * @param inputType It is used to mark the Row type of the input
    + */
    +class ProcTimeBoundedProcessingOverProcessFunction(
    +  private val aggregates: Array[AggregateFunction[_]],
    +  private val aggFields: Array[Int],
    +  private val forwardedFieldCount: Int,
    +  private val rowTypeInfo: RowTypeInfo,
    +  private val timeBoundary: Long,
    +  private val inputType: TypeInformation[Row])
    +    extends ProcessFunction[Row, Row] {
    +
    +  Preconditions.checkNotNull(aggregates)
    +  Preconditions.checkNotNull(aggFields)
    +  Preconditions.checkArgument(aggregates.length == aggFields.length)
    +
    +  private var output: Row = _
    +  private var accumulatorState: ValueState[Row] = _
    +  private var rowMapState: MapState[Long, JList[Row]] = _
    +
    +  override def open(config: Configuration) {
    +    output = new Row(forwardedFieldCount + aggregates.length)
    +
    +    // We keep the elements received in a list state 
    +    // together with the ingestion time in the operator
    +    val rowListTypeInfo: TypeInformation[JList[Row]] =
    +      new ListTypeInfo[Row](inputType).asInstanceOf[TypeInformation[JList[Row]]]
    +    val mapStateDescriptor: MapStateDescriptor[Long, JList[Row]] =
    +      new MapStateDescriptor[Long, JList[Row]]("rowmapstate",
    +        BasicTypeInfo.LONG_TYPE_INFO.asInstanceOf[TypeInformation[Long]], rowListTypeInfo)
    +    rowMapState = getRuntimeContext.getMapState(mapStateDescriptor)
    +
    +    val stateDescriptor: ValueStateDescriptor[Row] =
    +      new ValueStateDescriptor[Row]("overState", rowTypeInfo)
    +    accumulatorState = getRuntimeContext.getState(stateDescriptor)
    +  }
    +
    +  override def processElement(
    +    input: Row,
    +    ctx: ProcessFunction[Row, Row]#Context,
    +    out: Collector[Row]): Unit = {
    +
    +    val currentTime = ctx.timerService().currentProcessingTime()
    +    //buffer the event incoming event
    +
    +    var i = 0
    +
    +    //initialize the accumulators 
    +    var accumulators = accumulatorState.value()
    +
    +    if (null == accumulators) {
    +      accumulators = new Row(aggregates.length)
    +      i = 0
    +      while (i < aggregates.length) {
    +        accumulators.setField(i, aggregates(i).createAccumulator())
    +        i += 1
    +      }
    +    }
    +
    +    //set the fields of the last event to carry on with the aggregates
    +    i = 0
    +    while (i < forwardedFieldCount) {
    +      output.setField(i, input.getField(i))
    +      i += 1
    +    }
    +
    +    //update the elements to be removed and retract them from aggregators
    +    val limit = currentTime - timeBoundary
    +    
    +    // we iterate through all elements in the window buffer based on timestampt keys
    +    // when we find timestamps that are out of interest, we need to get the corresponding elements
    +    // and eliminate them. Multiple elements can be received at the same timestamp
    +    val iter = rowMapState.keys.iterator
    --- End diff --
    
    We could add another `ValueState` to remember the highest key that had been removed. If we have this information, we only need to iterate once per millisecond over the map keys. Would help if multiple rows arrive in the same millisecond.


---
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.
---