[PARQUET-2241] ByteStreamSplitDecoder broken in presence of nulls

• Type: Bug
• Status: Resolved
• Priority: Major
• Resolution: Fixed
• Affects Version/s: format-2.8.0
• Fix Version/s: 1.13.0, format-2.10.0
• Component/s: parquet-format, parquet-mr
• Assignee: Gang Wu
• Reporter: Xuwei Fu
• Created: 09/Feb/23 03:49
• Updated: 15/Nov/23 15:31
• Resolved: 21/Feb/23 15:27

Description

 
This problem is shown in this issue: https://github.com/apache/arrow/issues/15173
Let me talk about it briefly:

• Encoder doesn't write "num_values" on Page payload for BYTE_STREAM_SPLIT, but using "num_values" as stride in BYTE_STREAM_SPLIT
• When decoding, for DATA_PAGE_V2, it can now the num_values and num_nulls in the page, however, in DATA_PAGE_V1, without statistics, we should read def-levels and rep-levels to get the real num-of-values. And without the num-of-values, we aren't able to decode BYTE_STREAM_SPLIT correctly
   
  The bug-reproducing code is in the issue.