[PARQUET-895] Reading of nested columns is broken

• Type: Bug
• Status: Resolved
• Priority: Major
• Resolution: Fixed
• Component/s: parquet-cpp
• Assignee: Marc Vertes
• Reporter: Marc Vertes
• Created: 23/Feb/17 13:31
• Updated: 23/Jun/24 03:29
• Resolved: 23/Feb/17 17:37

Description

Problem occurs when reading a nested column with repeated values, specially when there is much more levels in that column than the number of global rows.

Citing @peshopetrov, who filed a github pull request identifying the problem and proposing a fix:

Nested repeated columns' count is incorrectly read from row group's metadata. That's correct in cases where there aren't any nested repeated fields but is generally not correct. Instead the num_values from the column's metadata should be used.

Issue Links

• blocks: PARQUET-713 parquet-cpp 1.0.0 release