[
https://issues.apache.org/jira/browse/FLINK-6968?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16405945#comment-16405945
]
ASF GitHub Bot commented on FLINK-6968:
---------------------------------------
Github user liurenjie1024 commented on a diff in the pull request:
https://github.com/apache/flink/pull/5688#discussion_r175680499
--- Diff:
flink-libraries/flink-table/src/main/scala/org/apache/flink/table/sinks/QueryableTableSink.scala
---
@@ -0,0 +1,162 @@
+/*
+ * 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.sinks
+
+import java.lang.{Boolean => JBool}
+
+import org.apache.flink.api.common.state.{ValueState, ValueStateDescriptor}
+import org.apache.flink.api.common.typeinfo.TypeInformation
+import org.apache.flink.api.java.functions.KeySelector
+import org.apache.flink.api.java.tuple.{Tuple2 => JTuple2}
+import org.apache.flink.api.java.typeutils.{ResultTypeQueryable,
RowTypeInfo}
+import org.apache.flink.configuration.Configuration
+import org.apache.flink.streaming.api.datastream.DataStream
+import org.apache.flink.streaming.api.functions.ProcessFunction
+import org.apache.flink.table.api.StreamQueryConfig
+import
org.apache.flink.table.runtime.aggregate.ProcessFunctionWithCleanupState
+import org.apache.flink.types.Row
+import org.apache.flink.util.Collector
+import org.slf4j.LoggerFactory
+
+class QueryableTableSink(
+ private val namePrefix: String,
+ private val queryConfig: StreamQueryConfig)
+ extends UpsertStreamTableSink[Row]
+ with TableSinkBase[JTuple2[JBool, Row]] {
+ private var keys: Array[String] = _
+
+ override def setKeyFields(keys: Array[String]): Unit = {
+ if (keys == null) {
+ throw new IllegalArgumentException("keys can't be null!")
+ }
+ this.keys = keys
+ }
+
+ override def setIsAppendOnly(isAppendOnly: JBool): Unit = {
+ if (isAppendOnly) {
+ throw new IllegalArgumentException("A QueryableTableSink can not be
used with append-only " +
+ "tables as the table would grow infinitely")
+ }
+ }
+
+ override def getRecordType: TypeInformation[Row] = new
RowTypeInfo(getFieldTypes, getFieldNames)
+
+ override def emitDataStream(dataStream: DataStream[JTuple2[JBool,
Row]]): Unit = {
+ val keyIndices = keys.map(getFieldNames.indexOf(_))
+ val keyTypes = keyIndices.map(getFieldTypes(_))
+
+ val keySelectorType = new RowTypeInfo(keyTypes, keys)
+
+ val processFunction = new QueryableStateProcessFunction(
+ namePrefix,
+ queryConfig,
+ keys,
+ getFieldNames,
+ getFieldTypes)
+
+ dataStream.keyBy(new RowKeySelector(keyIndices, keySelectorType))
+ .process(processFunction)
+ }
+
+ override protected def copy: TableSinkBase[JTuple2[JBool, Row]] = {
+ new QueryableTableSink(this.namePrefix, this.queryConfig)
+ }
+}
+
+class RowKeySelector(
+ private val keyIndices: Array[Int],
+ @transient private val returnType: TypeInformation[Row])
+ extends KeySelector[JTuple2[JBool, Row], Row]
+ with ResultTypeQueryable[Row] {
+
+ override def getKey(value: JTuple2[JBool, Row]): Row = {
+ val keys = keyIndices
+
+ val srcRow = value.f1
+
+ val destRow = new Row(keys.length)
+ var i = 0
+ while (i < keys.length) {
+ destRow.setField(i, srcRow.getField(keys(i)))
+ i += 1
+ }
+
+ destRow
+ }
+
+ override def getProducedType: TypeInformation[Row] = returnType
+}
+
+class QueryableStateProcessFunction(
+ private val namePrefix: String,
+ private val queryConfig: StreamQueryConfig,
+ private val keyNames: Array[String],
+ private val fieldNames: Array[String],
+ private val fieldTypes: Array[TypeInformation[_]])
+ extends ProcessFunctionWithCleanupState[JTuple2[JBool, Row],
Void](queryConfig) {
+
+ @transient private var states = Array[ValueState[AnyRef]]()
+ @transient private var nonKeyIndices = Array[Int]()
+
+ override def open(parameters: Configuration): Unit = {
+ super.open(parameters)
+
+ nonKeyIndices = fieldNames.indices
+ .filter(idx => !keyNames.contains(fieldNames(idx)))
+ .toArray
+
+ val statesBuilder = Array.newBuilder[ValueState[AnyRef]]
+
+ for (i <- nonKeyIndices) {
+ val stateDesc = new ValueStateDescriptor(fieldNames(i),
fieldTypes(i))
+
stateDesc.initializeSerializerUnlessSet(getRuntimeContext.getExecutionConfig)
+ stateDesc.setQueryable(fieldNames(i))
+ statesBuilder +=
getRuntimeContext.getState(stateDesc).asInstanceOf[ValueState[AnyRef]]
+ }
+
+ states = statesBuilder.result()
+
+ initCleanupTimeState("QueryableStateCleanupTime")
+ }
+
+ override def processElement(
+ value: JTuple2[JBool, Row],
+ ctx: ProcessFunction[JTuple2[JBool, Row], Void]#Context,
+ out: Collector[Void]): Unit = {
+ if (value.f0) {
+ for (i <- nonKeyIndices.indices) {
+ states(i).update(value.f1.getField(nonKeyIndices(i)))
+ }
+
+ val currentTime = ctx.timerService().currentProcessingTime()
--- End diff --
I think it's right to be able to process both time domain, and I've added.
> Store streaming, updating tables with unique key in queryable state
> -------------------------------------------------------------------
>
> Key: FLINK-6968
> URL: https://issues.apache.org/jira/browse/FLINK-6968
> Project: Flink
> Issue Type: New Feature
> Components: Table API & SQL
> Reporter: Fabian Hueske
> Assignee: Renjie Liu
> Priority: Major
>
> Streaming tables with unique key are continuously updated. For example
> queries with a non-windowed aggregation generate such tables. Commonly, such
> updating tables are emitted via an upsert table sink to an external datastore
> (k-v store, database) to make it accessible to applications.
> This issue is about adding a feature to store and maintain such a table as
> queryable state in Flink. By storing the table in Flnk's queryable state, we
> do not need an external data store to access the results of the query but can
> query the results directly from Flink.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
