Skip to content

Windows non-ASCII path support issue in indexing pipeline #357

Description

@bahadraksakal

Windows non-ASCII path support issue in indexing pipeline

Environment:

  • Windows
  • Repository path contains Turkish characters:
    D:\Masaüstü\Bahadır Aksakal\MetSims\Projeler\cageEpd

Symptom:

  • index_repository fails or downstream tools cannot resolve the project reliably.
  • Moving/aliasing the same repository to ASCII-only path works:
    D:\Projects\cageEpd

Expected:

  • Windows paths with non-ASCII UTF-8 characters should index correctly.

Likely areas:

  • filesystem/discovery layer should use UTF-8 -> UTF-16 conversion on Windows
  • Windows filesystem calls should use wide APIs:
    FindFirstFileW, FindNextFileW, GetFileAttributesW, CreateFileW, _wfopen/_wstat
  • all path normalization should preserve UTF-8 input consistently
  • errors should distinguish path encoding failure from “project not indexed”

Also related:

  • MCP tool schema should return a clear error if project_name is passed instead of project.

Metadata

Metadata

Assignees

No one assigned

    Labels

    bugSomething isn't workingwindowsWindows-specific issues

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions