[
https://issues.apache.org/jira/browse/FLINK-7062?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16656512#comment-16656512
]
ASF GitHub Bot commented on FLINK-7062:
---------------------------------------
twalthr commented on a change in pull request #6815: [FLINK-7062][cep][table]
Added basic support for MATCH_RECOGNIZE
URL: https://github.com/apache/flink/pull/6815#discussion_r226558043
##########
File path:
flink-libraries/flink-table/src/main/scala/org/apache/flink/table/plan/nodes/datastream/DataStreamMatch.scala
##########
@@ -214,24 +160,145 @@ class DataStreamMatch(
patternStream.flatSelect[CRow](patternSelectFunction, outTypeInfo)
}
}
+
+ private def translateOrder(
+ tableEnv: StreamTableEnvironment,
+ crowInput: DataStream[CRow],
+ orderKeys: RelCollation)
+ : (DataStream[CRow], Option[RowComparator]) = {
+
+ if (orderKeys.getFieldCollations.size() == 0) {
+ throw new ValidationException("You must specify either rowtime or
proctime for order by.")
+ }
+
+ // need to identify time between others order fields. Time needs to be
first sort element
+ val timeOrderField = SortUtil.getFirstSortField(orderKeys,
inputSchema.relDataType)
+
+ if (!FlinkTypeFactory.isTimeIndicatorType(timeOrderField.getType)) {
+ throw new ValidationException(
+ "You must specify either rowtime or proctime for order by as the first
one.")
+ }
+
+ // time ordering needs to be ascending
+ if (SortUtil.getFirstSortDirection(orderKeys) != Direction.ASCENDING) {
+ throw new ValidationException(
+ "Primary sort order of a streaming table must be ascending on time.")
+ }
+
+ val rowComparator = if (orderKeys.getFieldCollations.size() > 1) {
+ Some(SortUtil
+ .createRowComparator(inputSchema.relDataType,
+ orderKeys.getFieldCollations.asScala.tail,
+ tableEnv.execEnv.getConfig))
+ } else {
+ None
+ }
+
+ timeOrderField.getType match {
+ case _ if
FlinkTypeFactory.isRowtimeIndicatorType(timeOrderField.getType) =>
+ (crowInput.process(
+ new RowtimeProcessFunction(timeOrderField.getIndex,
CRowTypeInfo(inputSchema.typeInfo))
+ ).setParallelism(crowInput.getParallelism),
+ rowComparator)
+ case _ =>
+ (crowInput, rowComparator)
+ }
+ }
+
+ private def applyPartitioning(partitionKeys: util.List[RexNode], inputDs:
DataStream[Row])
+ : DataStream[Row] = {
+ if (partitionKeys.size() > 0) {
+ val keys = partitionKeys.asScala.map {
+ case ref: RexInputRef => ref.getIndex
+ }.toArray
+ val keySelector = new RowKeySelector(keys,
inputSchema.projectedTypeInfo(keys))
+ inputDs.keyBy(keySelector)
+ } else {
+ inputDs
+ }
+ }
+
+ @VisibleForTesting private[flink] def getLogicalMatch = logicalMatch
}
+@VisibleForTesting
private[flink] class PatternVisitor(
config: TableConfig,
inputTypeInfo: TypeInformation[Row],
- logicalMatch: FlinkLogicalMatch)
Review comment:
Move this class to a separate class? Because it actually `logicalMatch` must
not be passed in the current implementation as the class is not "static".
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org
> Support the basic functionality of MATCH_RECOGNIZE
> --------------------------------------------------
>
> Key: FLINK-7062
> URL: https://issues.apache.org/jira/browse/FLINK-7062
> Project: Flink
> Issue Type: Sub-task
> Components: CEP, Table API & SQL
> Reporter: Dian Fu
> Assignee: Dian Fu
> Priority: Major
> Labels: pull-request-available
>
> In this JIRA, we will support the basic functionality of {{MATCH_RECOGNIZE}}
> in Flink SQL API which includes the support of syntax {{MEASURES}},
> {{PATTERN}} and {{DEFINE}}. This would allow users write basic cep use cases
> with SQL like the following example:
> {code}
> SELECT T.aid, T.bid, T.cid
> FROM MyTable
> MATCH_RECOGNIZE (
> MEASURES
> A.id AS aid,
> B.id AS bid,
> C.id AS cid
> PATTERN (A B C)
> DEFINE
> A AS A.name = 'a',
> B AS B.name = 'b',
> C AS C.name = 'c'
> ) AS T
> {code}
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
