flink-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From rtudoran <...@git.apache.org>
Subject [GitHub] flink pull request #3889: [FLINK-6075] - Support Limit/Top(Sort) for Stream ...
Date Wed, 28 Jun 2017 14:52:49 GMT
Github user rtudoran commented on a diff in the pull request:

    https://github.com/apache/flink/pull/3889#discussion_r124563186
  
    --- Diff: flink-libraries/flink-table/src/main/scala/org/apache/flink/table/runtime/aggregate/SortUtil.scala
---
    @@ -0,0 +1,247 @@
    +/*
    + * 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.table.calcite.FlinkTypeFactory
    +import org.apache.flink.types.Row
    +import org.apache.calcite.rel.`type`._
    +import org.apache.calcite.rel.RelCollation
    +import org.apache.flink.streaming.api.functions.ProcessFunction
    +import org.apache.flink.table.functions.AggregateFunction
    +import org.apache.calcite.sql.`type`.SqlTypeName
    +import org.apache.flink.table.api.TableException
    +import org.apache.calcite.sql.`type`.SqlTypeName._
    +import java.util.{ List => JList, ArrayList }
    +import org.apache.flink.api.common.typeinfo.{SqlTimeTypeInfo, TypeInformation}
    +import org.apache.flink.api.java.typeutils.RowTypeInfo
    +import java.sql.Timestamp
    +import org.apache.calcite.rel.RelFieldCollation
    +import org.apache.calcite.rel.RelFieldCollation.Direction
    +import java.util.Comparator
    +import org.apache.flink.api.common.typeutils.TypeComparator
    +import org.apache.flink.api.common.typeinfo.BasicTypeInfo._
    +import java.lang.{Byte=>JByte,Integer=>JInt,Long=>JLong,Double=>JDouble,Short=>JShort,String=>JString,Float=>JFloat}
    +import java.math.{BigDecimal=>JBigDecimal}
    +import org.apache.flink.api.common.functions.MapFunction
    +import org.apache.flink.api.common.operators.Order
    +import org.apache.calcite.rex.{RexLiteral, RexNode}
    +import org.apache.flink.api.common.ExecutionConfig
    +import org.apache.flink.api.common.typeinfo.AtomicType
    +import org.apache.flink.api.java.typeutils.runtime.RowComparator
    +import org.apache.flink.api.common.typeutils.TypeSerializer
    +import org.apache.flink.table.runtime.types.{CRow, CRowTypeInfo}
    +
    +/**
    + * Class represents a collection of helper methods to build the sort logic.
    + * It encapsulates as well the implementation for ordering and generic interfaces
    + */
    +object SortUtil {
    +
    +  /**
    +   * Function creates [org.apache.flink.streaming.api.functions.ProcessFunction] for
sorting 
    +   * elements based on rowtime and potentially other fields
    +   * @param collationSort The Sort collation list
    +   * @param inputType input row type
    +   * @param execCfg table environment execution configuration
    +   * @return org.apache.flink.streaming.api.functions.ProcessFunction
    +   */
    +  private[flink] def createRowTimeSortFunction(
    +    collationSort: RelCollation,
    +    inputType: RelDataType,
    +    inputTypeInfo: TypeInformation[Row],
    +    execCfg: ExecutionConfig): ProcessFunction[CRow, CRow] = {
    +
    +    val keySortFields = getSortFieldIndexList(collationSort)
    +    //drop time from comparison as we sort on time in the states and result emission
    +    val keyIndexesNoTime = keySortFields.slice(1, keySortFields.size)
    +    val booleanOrderings = getSortFieldDirectionBooleanList(collationSort)
    +    val booleanDirectionsNoTime = booleanOrderings.slice(1, booleanOrderings.size)
    +    
    +    val fieldComps = createFieldComparators(
    +      inputType, 
    +      keyIndexesNoTime, 
    +      booleanDirectionsNoTime,
    +      execCfg)
    +    val fieldCompsRefs = fieldComps.asInstanceOf[Array[TypeComparator[AnyRef]]]
    +    
    +    val rowComp = new RowComparator(
    +       inputType.getFieldCount,
    +       keyIndexesNoTime,
    +       fieldCompsRefs,
    +       new Array[TypeSerializer[AnyRef]](0), //used only for object comparisons
    +       booleanDirectionsNoTime)
    +      
    +    val collectionRowComparator = new CollectionRowComparator(rowComp)
    +    
    +    val inputCRowType = CRowTypeInfo(inputTypeInfo)
    + 
    +    new RowTimeSortProcessFunction(
    +      inputCRowType,
    +      collectionRowComparator)
    +
    +  }
    +  
    +  
    +  /**
    +   * Function creates [org.apache.flink.streaming.api.functions.ProcessFunction] for
sorting 
    +   * elements based on proctime and potentially other fields
    +   * @param collationSort The Sort collation list
    +   * @param inputType input row type
    +   * @param execCfg table environment execution configuration
    +   * @return org.apache.flink.streaming.api.functions.ProcessFunction
    +   */
    +  private[flink] def createProcTimeSortFunction(
    +    collationSort: RelCollation,
    +    inputType: RelDataType,
    +    inputTypeInfo: TypeInformation[Row],
    +    execCfg: ExecutionConfig): ProcessFunction[CRow, CRow] = {
    +
    +    val keySortFields = getSortFieldIndexList(collationSort)
    --- End diff --
    
    I am not sure I understand what you mean. But I think the current setup to create the
processfunction based on rowtime/ proctime are needed and is useful also for adding later
the cases when we have offset/fetch. I would not like to modify this as it does not bring
value (but it might be that I misunderstood what you mean)


---
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