Skip to content

[C++][Parquet] Error when printing file metadata #51673

Description

@pitrou

Describe the bug, including details regarding any error messages, version, and platform.

If you download prices.parquet from https://www.data.gouv.fr/datasets/open-prices and then give it to parquet-reader --only-metadata, you get the following error:

[...]
Column 29
  Values: 1024, Null Values: 1024, Distinct Values: 0
  Max (exact: unknown): Parquet error: Failed to parse decimal value: Length of byte array passed to Decimal256::FromBigEndian was 0, but must be between 1 and 32

Ironically, the file was generated using Arrow C++:

File Name: ./prices.parquet
Version: 2.6
Created By: parquet-cpp-arrow version 18.0.0-SNAPSHOT

Component(s)

C++, Parquet

Activity

  1. aryan9948 commented on Oct 1, 2026

    @aryan9948

    Hi @pitrou, I’d like to work on this issue and investigate the decimal metadata handling that leads to the Decimal256::FromBigEndian error.

    If it’s okay, could you assign this issue to me? I’d be happy to work on a fix and add the necessary tests. Thanks!

  2. pitrou commented on Oct 1, 2026

    @pitrou
    MemberAuthor

    Yes, feel free to open a PR.

  3. Sunny-Island commented on Oct 4, 2026

    @Sunny-Island
    Contributor

    Hi @pitrou , I’m interested in contributing a fix for this bug. I reproduced the error using the Open Prices prices.parquet file and traced it to the text metadata printer attempting to format missing min/max statistics for an all-null DECIMAL column. I’ve opened PR #51705 with a fix and a regression test. I’d be happy to address any feedback.

  4. aryan9948 commented on Oct 5, 2026

    @aryan9948

    Hi @pitrou, I see that PR #51705 has been opened for #51673 and addresses the issue with missing min/max statistics for all-null DECIMAL columns.

    Since I’m assigned to this issue, should I review/test the proposed fix, or would you prefer that I leave the implementation to the existing PR?

  5. Sunny-Island commented on Oct 5, 2026

    @Sunny-Island
    Contributor

    Hi @pitrou, I see that PR #51705 has been opened for #51673 and addresses the issue with missing min/max statistics for all-null DECIMAL columns.

    Since I’m assigned to this issue, should I review/test the proposed fix, or would you prefer that I leave the implementation to the existing PR?

    Hi @aryan9948 , I’d be glad to collaborate with you on this. I’d welcome your review or testing of PR #51705 and any feedback you have. I’m happy to make further changes to ensure the fix fully addresses the issue.

  6. added 4 commits that reference this issue on Oct 6, 2026
  7. added a commit that references this issue on Oct 7, 2026
  8. added this to the 27.0.0 milestone on Oct 7, 2026
  9. pitrou commented on Oct 7, 2026

    @pitrou
    MemberAuthor

    Issue resolved by pull request 51705
    #51705

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Assignees

Type

No type

Projects

No projects

    Milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions