Bug: `dbt-autofix-fix` introduces mojibake (`�"`) and then fails with UnicodeDecodeError on Windows

[tonyhoang-cot]

Bug: dbt-autofix-fix introduces mojibake (�") and then fails with UnicodeDecodeError on Windows

Summary

On Windows, dbt-autofix-fix appears to corrupt punctuation/characters in YAML text (example: introducing �" in-place), and then fails with a Unicode decode error.

Environment

• OS: Windows 11 (win32 10.0.26100)
• Python: 3.12.12
• pre-commit: 4.5.1
• dbt-autofix hook:
  • repo: https://github.com/dbt-labs/dbt-autofix
  • rev: v0.20.2
  • hook id: dbt-autofix-fix

Hook config

yaml:
repos:

• repo: https://github.com/dbt-labs/dbt-autofix
  rev: v0.20.2
  hooks:
  • id: dbt-autofix-fix

---

Steps to reproduce

1. pre-commit install
2. pre-commit run --all-files

Actual result

• Hook reports file refactors (replace_fancy_quotes).
• Character corruption is observed in edited text (e.g. unexpected �" sequence appears where normal punctuation/characters should be; screenshot attached).
• Run then crashes with:
  UnicodeDecodeError: 'charmap' codec can't decode byte 0x9d ... ... File "...dbt_autofix\refactor.py", line 150, in _apply_changesets yml_str = yml_file.read_text() ... File "...encodings\cp1252.py", line 23, in decode

Expected result

• No mojibake/corruption in file content.
• Hook should consistently handle UTF-8 content and complete without decode errors.