HeartSaVioR commented on code in PR #46247:
URL: https://github.com/apache/spark/pull/46247#discussion_r1599101022
##########
sql/core/src/test/scala/org/apache/spark/sql/streaming/StreamingDeduplicationSuite.scala:
##########
@@ -484,6 +486,52 @@ class StreamingDeduplicationSuite extends
StateStoreMetricsTest {
CheckLastBatch(("c", 9, "c"))
)
}
+
+ test("collation aware deduplication") {
+ val inputData = MemoryStream[(String, Int)]
+ val result = inputData.toDF()
+ .select(col("_1")
+ .try_cast(StringType("UNICODE")).as("str"),
Review Comment:
While we are here, I see UNICODE is binary equality but non-binary ordering.
Does this still ensure that we can put this into RocksDB which key is binary
sorted and find the key group based on prefix of key including this column?
E.g. Say we have two columns, dept (String with UNICODE collation), session
start (timestamp) as grouping key, and want to scan all grouping keys which are
having dept as 'dept1'. This is required for several operations like session
window aggregation.
My gut feeling is yes, but I would like to double confirm.
##########
sql/core/src/test/scala/org/apache/spark/sql/streaming/StreamingDeduplicationSuite.scala:
##########
@@ -484,6 +486,52 @@ class StreamingDeduplicationSuite extends
StateStoreMetricsTest {
CheckLastBatch(("c", 9, "c"))
)
}
+
+ test("collation aware deduplication") {
+ val inputData = MemoryStream[(String, Int)]
+ val result = inputData.toDF()
+ .select(col("_1")
+ .try_cast(StringType("UNICODE")).as("str"),
+ col("_2").as("int"))
+ .dropDuplicates("str")
+
+ testStream(result, Append)(
+ AddData(inputData, "a" -> 1),
+ CheckLastBatch("a" -> 1),
+ assertNumStateRows(total = 1, updated = 1, droppedByWatermark = 0),
+ AddData(inputData, "a" -> 2), // Dropped
+ CheckLastBatch(),
+ assertNumStateRows(total = 1, updated = 0, droppedByWatermark = 0),
+ // scalastyle:off
+ AddData(inputData, "ä" -> 1),
+ CheckLastBatch("ä" -> 1),
+ // scalastyle:on
+ assertNumStateRows(total = 2, updated = 1, droppedByWatermark = 0)
+ )
+ }
+
+ test("non-binary collation aware deduplication not supported") {
+ val inputData = MemoryStream[(String)]
+ val result = inputData.toDF()
+ .select(col("value")
+ .try_cast(StringType("UTF8_BINARY_LCASE")).as("str"))
Review Comment:
AFAIK, LCASE means comparing as lowercase, so yes it's bound to non-binary
equality.
##########
sql/core/src/test/scala/org/apache/spark/sql/streaming/StreamingQuerySuite.scala:
##########
@@ -1364,6 +1364,35 @@ class StreamingQuerySuite extends StreamTest with
BeforeAndAfter with Logging wi
)
}
+ test("Collation aware streaming") {
+ withTable("parquet_streaming_tbl") {
+ spark.sql(
+ """
+ |CREATE TABLE parquet_streaming_tbl
+ |(
+ | key STRING COLLATE UTF8_BINARY_LCASE,
+ | value_stream INTEGER
+ |) USING parquet""".stripMargin)
+
+ val streamDf = spark.readStream.table("parquet_streaming_tbl")
+ val filteredDf = streamDf.filter("key = 'aaa'")
+
+ val clock = new StreamManualClock()
+ testStream(filteredDf)(
+ StartStream(triggerClock = clock, trigger =
Trigger.ProcessingTime(100)),
+ Execute { _ =>
+ spark.createDataFrame(Seq("aaa" -> 1, "AAA" -> 2, "bbb" -> 3, "aa"
-> 4))
Review Comment:
Not sure I understand. streamDf has non-default collation,
UTF8_BINARY_LCASE. What do you mean by "incoming" stream?
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
