Skip to content

Drop _tmp_metadata_row_index column from the output of Delta Scan on GPU #13991

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
razajafri merged 7 commits into from
Dec 12, 2025
Merged

Drop _tmp_metadata_row_index column from the output of Delta Scan on GPU #13991

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
7 commits
Select commit Hold shift + click to select a range
0cf8bb5
drop _metadata_row_index column
razajafri Dec 11, 2025
31743d2
Signing off
razajafri Dec 11, 2025
75bb8f5
added tests
razajafri Dec 11, 2025
baea0c2
Remove unnecessary commented line
razajafri Dec 11, 2025
1671116
Updated the test to be more robust
razajafri Dec 12, 2025
2af9c5d
Addressed review comments
razajafri Dec 12, 2025
e67c990
assert test has deletion vector scan
razajafri Dec 12, 2025
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
2 changes: 2 additions & 0 deletions ...src/main/delta-33x-40x/scala/com/nvidia/spark/rapids/delta/common/DeltaProviderBase.scala
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -162,6 +162,8 @@ abstract class DeltaProviderBase extends DeltaIOProvider {
dvFilterInput.copy(projectList = inputList.filterNot(_.name == "_metadata"))
.withNewChildren(Seq(
fsse.copy(
originalOutput =
fsse.originalOutput.filterNot(_.name == "_tmp_metadata_row_index"),
requiredSchema = StructType(
fsse.requiredSchema.filterNot(_.name == "_tmp_metadata_row_index")
))(fsse.rapidsConf)))))))
Expand Down
23 changes: 23 additions & 0 deletions integration_tests/src/main/python/delta_lake_test.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -160,3 +160,26 @@ def convert_and_setup_name_mapping(spark):
with_cpu_session(setup_parquet_table, {"spark.sql.parquet.fieldId.write.enabled": str(enable_deletion_vectors).lower()})
with_cpu_session(convert_and_setup_name_mapping, conf={"spark.databricks.delta.properties.defaults.enableDeletionVectors": "false"})
assert_gpu_and_cpu_are_equal_collect(lambda spark: spark.read.format("delta").load(data_path))

@allow_non_gpu(*delta_meta_allow)
@delta_lake
@ignore_order(local=True)
@pytest.mark.skipif(not is_spark_340_or_later(), reason="Deletion Vectors only supported on Spark 3.4.0+")
def test_delta_filter_out_metadata_col(spark_tmp_path):
data_path = spark_tmp_path + "/DELTA_DATA"

def create_delta(spark):
two_col_df(spark, int_gen, int_gen).coalesce(1).write.format("delta") \

@gerashegalov gerashegalov Dec 12, 2025

Copy link
Copy Markdown
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Is int_gen guaranteed to have a 0 that you are later deleting?

Is coalesce(1) the same as passing num_slices=1 to two_col_df ?

@razajafri razajafri Dec 12, 2025

Copy link
Copy Markdown
Collaborator Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

0 is guaranteed to be present as it's a special_case in the IntegerGen

.option("delta.enableDeletionVectors", "true") \
.partitionBy("a").save(data_path)

count = spark.sql(f"DELETE FROM delta.`{data_path}` WHERE b = 0").collect()[0][0]
assert(count > 0)

def read_table(spark):
df = spark.sql(f"SELECT * FROM delta.`{data_path}`")
assert "__delta_internal_is_row_deleted" in df._sc._jvm.PythonSQLUtils.explainString(df._jdf.queryExecution(), "extended")
return df

with_cpu_session(create_delta)
assert_gpu_and_cpu_are_equal_collect(read_table)