Github user lw-lin commented on a diff in the pull request:
https://github.com/apache/spark/pull/16987#discussion_r103603935
--- Diff:
sql/core/src/test/scala/org/apache/spark/sql/streaming/FileStreamSourceSuite.scala
---
@@ -662,6 +665,154 @@ class FileStreamSourceSuite extends
FileStreamSourceTest {
}
}
+ test("read data from outputs of another streaming query") {
+ withSQLConf(SQLConf.FILE_SINK_LOG_COMPACT_INTERVAL.key -> "3") {
+ withTempDirs { case (dir, tmp) =>
+ // q1 is a streaming query that reads from memory and writes to
text files
+ val q1_source = MemoryStream[String]
+ val q1_checkpointDir = new File(dir,
"q1_checkpointDir").getCanonicalPath
+ val q1_outputDir = new File(dir, "q1_outputDir").getCanonicalPath
+ val q1 =
+ q1_source
+ .toDF()
+ .writeStream
+ .option("checkpointLocation", q1_checkpointDir)
+ .format("text")
+ .start(q1_outputDir)
+
+ // q2 is a streaming query that reads q1's text outputs
+ val q2 = createFileStream("text", q1_outputDir).filter($"value"
contains "keep")
+
+ def q1AddData(data: String*): StreamAction =
+ Execute { _ =>
+ q1_source.addData(data)
+ q1.processAllAvailable()
+ }
+ def q2ProcessAllAvailable(): StreamAction = Execute { q2 =>
q2.processAllAvailable() }
+
+ testStream(q2)(
+ // batch 0
+ q1AddData("drop1", "keep2"),
+ q2ProcessAllAvailable(),
+ CheckAnswer("keep2"),
+
+ // batch 1
+ Assert {
+ // create a text file that won't be on q1's sink log
+ // thus even if its contents contains "keep", it should NOT
appear in q2's answer
+ val shouldNotKeep = new File(q1_outputDir,
"should_not_keep.txt")
+ stringToFile(shouldNotKeep, "should_not_keep!!!")
+ shouldNotKeep.exists()
+ },
+ q1AddData("keep3"),
+ q2ProcessAllAvailable(),
+ CheckAnswer("keep2", "keep3"),
+
+ // batch 2: check that things work well when the sink log gets
compacted
+ q1AddData("keep4"),
+ Assert {
+ // compact interval is 3, so file "2.compact" should exist
+ new File(q1_outputDir,
s"${FileStreamSink.metadataDir}/2.compact").exists()
+ },
+ q2ProcessAllAvailable(),
+ CheckAnswer("keep2", "keep3", "keep4"),
+
+ // stop q1 manually
+ Execute { _ => q1.stop() }
+ )
+ }
+ }
+ }
+
+ test("read partitioned data from outputs of another streaming query") {
+ withTempDirs { case (dir, tmp) =>
+ // q1 is a streaming query that reads from memory and writes to
partitioned json files
+ val q1_source = MemoryStream[(String, String)]
+ val q1_checkpointDir = new File(dir,
"q1_checkpointDir").getCanonicalPath
+ val q1_outputDir = new File(dir, "q1_outputDir").getCanonicalPath
+ val q1 =
+ q1_source
+ .toDF()
+ .select($"_1" as "partition", $"_2" as "value")
+ .writeStream
+ .option("checkpointLocation", q1_checkpointDir)
+ .partitionBy("partition")
+ .format("json")
+ .start(q1_outputDir)
+
+ // q2 is a streaming query that reads q1's partitioned json outputs
+ val schema = new StructType().add("value",
StringType).add("partition", StringType)
+ val q2 = createFileStream("json", q1_outputDir,
Some(schema)).filter($"value" contains "keep")
+
+ def q1AddData(data: (String, String)*): StreamAction =
+ Execute { _ =>
+ q1_source.addData(data)
+ q1.processAllAvailable()
+ }
+ def q2ProcessAllAvailable(): StreamAction = Execute { q2 =>
q2.processAllAvailable() }
+
+ testStream(q2)(
+ // batch 0: append to a new partition=foo, should read value and
partition
+ q1AddData(("foo", "drop1"), ("foo", "keep2")),
+ q2ProcessAllAvailable(),
+ CheckAnswer(("keep2", "foo")),
+
+ // batch 1: append to same partition=foo, should read value and
partition
+ q1AddData(("foo", "keep3")),
+ q2ProcessAllAvailable(),
+ CheckAnswer(("keep2", "foo"), ("keep3", "foo")),
+
+ // batch 2: append to a different partition=bar, should read value
and partition
+ q1AddData(("bar", "keep4")),
+ q2ProcessAllAvailable(),
+ CheckAnswer(("keep2", "foo"), ("keep3", "foo"), ("keep4", "bar")),
+
+ // stop q1 manually
+ Execute { _ => q1.stop() }
+ )
+ }
+ }
+
+ test("start before another streaming query, and read its output") {
+ withTempDirs { case (dir, tmp) =>
+ // q1 is a streaming query that reads from memory and writes to text
files
+ val q1_source = MemoryStream[String]
+ val q1_checkpointDir = new File(dir,
"q1_checkpointDir").getCanonicalPath
+ val q1_outputDir = new File(dir, "q1_outputDir")
+ assert(q1_outputDir.mkdir()) // prepare the
output dir for q2 to read
+ val q1_write = q1_source
+ .toDF()
+ .writeStream
+ .option("checkpointLocation", q1_checkpointDir)
+ .format("text") // define q1, but
don't start it for now
+ var q1: StreamingQuery = null
+
+ val q2 =
+ createFileStream("text",
q1_outputDir.getCanonicalPath).filter($"value" contains "keep")
+
+ testStream(q2)(
+ AssertOnQuery { q2 =>
+ val fileSource = getSourcesFromStreamingQuery(q2).head
+ fileSource.sourceHasMetadata === None // q1 has not
started yet, verify that q2
--- End diff --
fixed
---
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 infrastruct...@apache.org or file a JIRA ticket
with INFRA.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org
For additional commands, e-mail: reviews-h...@spark.apache.org
