voonhous commented on code in PR #19648:
URL: https://github.com/apache/hudi/pull/19648#discussion_r3819038112
##########
hudi-spark-datasource/hudi-spark/src/test/scala/org/apache/spark/sql/hudi/common/TestSlashSeparatedPartitionValue.scala:
##########
@@ -90,6 +90,113 @@ class TestSlashSeparatedPartitionValue extends
HoodieSparkSqlTestBase {
}
}
+ test("Test slash separated date partitions written through the row writer") {
+ withSQLConf("hoodie.spark.sql.insert.into.operation" -> "bulk_insert") {
+ withTempDir { tmp =>
+ val targetTable = generateTableName
+ val tablePath = s"${tmp.getCanonicalPath}/$targetTable"
+
+ spark.sql(
+ s"""
+ |create table $targetTable (
+ | `id` string,
+ | `name` string,
+ | `ts` bigint,
+ | `datestr` STRING
+ |) using hudi
+ | tblproperties (
+ | 'primaryKey' = 'id',
+ | 'type' = 'COW',
+ | 'preCombineField'='ts',
+ |
'hoodie.datasource.write.slash.separated.date.partitioning'='true'
+ | )
+ | partitioned by (`datestr`)
+ | location '$tablePath'
+ """.stripMargin)
+
+ // NOTE: The row writer derives the partition path off of an
[[InternalRow]], which used to
+ // blow up with a [[ClassCastException]]; a null partition value
used to NPE
+ spark.sql(
+ s"""
+ | insert into $targetTable values
+ | (1, 'a1', 1000, "2026-01-05"),
+ | (2, 'a2', 2000, "2026-01-06"),
+ | (3, 'a3', 3000, null)
+ """.stripMargin)
+
+ checkAnswer(s"select id, name, ts, _hoodie_partition_path, datestr
from $targetTable order by id")(
+ Seq("1", "a1", 1000, "2026/01/05", "2026-01-05"),
+ Seq("2", "a2", 2000, "2026/01/06", "2026-01-06"),
+ Seq("3", "a3", 3000, "__HIVE_DEFAULT_PARTITION__", null)
+ )
+
+ val metaClient = HoodieTableMetaClient.builder()
+
.setConf(HadoopFSUtils.getStorageConfWithCopy(spark.sparkContext.hadoopConfiguration))
+ .setBasePath(tablePath)
+ .build()
+ assertTrue(metaClient.getStorage.exists(new StoragePath(tablePath,
"2026/01/05")),
Review Comment:
The partition DDL commands are slash-unaware on the table this test creates:
`HoodieSqlCommonUtils#makePartitionPath` (`HoodieSqlCommonUtils.scala:420-427`)
reads hive-style and urlencode from table config but never
`getSlashSeparatedDatePartitioning`, so `alter table ... drop partition
(datestr='2026-01-05')` targets `2026-01-05` while the directory is
`2026/01/05`: the replacecommit replaces zero file groups and the drop is a
silent no-op, and `add partition` creates a bogus dashed directory. Distinct
from #19666/#19668/#19669.
Optional, nothing needed in this PR: this is a pre-existing bug in a file
this PR does not touch, so feel free to leave it. Flagging here so it is on
record next to the tests that create such tables.
--
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]
