[ice-pring]

Reading a UTF-8 BOM-prefixed .gitignore with encoding="utf-8" leaves the BOM in the first pattern. For example:

GitIgnoreSpec.from_lines(["\ufeff*.log\n"]).match_file("test.log")
# False before this change; Git ignores test.log.

Git skips one UTF-8 BOM at the start of an ignore file before parsing its lines (implementation in Git 2.54.0). Strip it at the GitIgnoreSpec.from_lines() boundary, where the first line can be distinguished from later patterns. Comments, negations and anchored directory rules then retain their meaning. A second BOM or a BOM on a later line remains literal. Both text and byte lines are supported, including single-pass iterators and the reversed argument form.

Added regression and control tests, a docstring note and a changelog entry.

Validation on Windows:

• Focused tests fail before the fix and pass afterward.
• Tox py312 and py314: the full base suite succeeds on CPython 3.12.14 and 3.14.5, with optional backend/platform skips.
• With both optional backends installed, python -m unittest discover -t . -s tests/ on CPython 3.14.5 runs all 300 tests successfully, without skips.
• mypy pathspec tests/check_usage.py passes for all 32 source files.
• 90 results across 10 representative ignore files agree with git check-ignore --no-index --verbose --non-matching -z using Git 2.54.0.windows.1.
• git diff --check passes.

[cpburnz]

The appropriate way to handle a UTF-8 BOM is to open the file with encoding="utf-8-sig". That will remove the BOM.

[ice-pring]

Thanks for clarifying the intended behavior. Using encoding="utf-8-sig" when opening the file makes sense, so I'll close this PR and leave BOM handling to the caller. I appreciate your review.