From c78570749bc3887db669c3771eb6d63287fd98a1 Mon Sep 17 00:00:00 2001 From: You-Cheng Lin Date: Tue, 27 Oct 2026 18:58:57 +0800 Subject: [PATCH] use make_fragment instead of _subset to create sub fragment Signed-off-by: You-Cheng Lin --- .../formats/parquet/parquet_file_chunking_utils.py | 8 +++++++- 1 file changed, 7 insertions(+), 1 deletion(-) diff --git a/python/ray/data/_internal/datasource_v2/formats/parquet/parquet_file_chunking_utils.py b/python/ray/data/_internal/datasource_v2/formats/parquet/parquet_file_chunking_utils.py index f6ceeb0182ec..f6a89525025e 100644 --- a/python/ray/data/_internal/datasource_v2/formats/parquet/parquet_file_chunking_utils.py +++ b/python/ray/data/_internal/datasource_v2/formats/parquet/parquet_file_chunking_utils.py @@ -79,7 +79,13 @@ def _subset(rg_ids: List[int]) -> pds.ParquetFileFragment: path = fragment.path if not per_row_group_offsets: - return [ReadUnitFragment(_subset(ids), ReadUnit(id=path, source=path, count=1))] + lazy = fragment.format.make_fragment( + path, + filesystem=fragment.filesystem, + partition_expression=fragment.partition_expression, + row_groups=ids, + ) + return [ReadUnitFragment(lazy, ReadUnit(id=path, source=path, count=1))] metadata = _with_io_retry( lambda: fragment.metadata, f"read Parquet footer for {path}"