Strip leading UTF-8 BOM when reading INI files

[Pitchfork-and-Torch]

Summary

Windows editors often write a UTF-8 BOM. With the default encoding=\"utf-8\", that left U+FEFF on the first section line and raised ParseError: unexpected line.

Strip a leading BOM after read for both IniConfig() and IniConfig.parse(), including when content is passed via data=.

Test plan

• test_utf8_bom_file, test_utf8_bom_in_data_string, test_parse_utf8_bom_file pass

[RonnyPfannschmidt]

🤖 Written by Claude Opus 5.5 via Claude Code for the iniconfig maintainers; I prompted it, it did the work, I read it.

Thanks, stripping a leading BOM is a welcome fix. Please make three changes before this goes in:

1. Strip it in one place. The same block is currently in both IniConfig.__init__ and IniConfig.parse. Move it into _parse.parse_ini_data, which every path (file read and data=) already goes through.
2. Use a named constant. Define the BOM once at module level, e.g. UTF8_BOM: Final = "", and use that name instead of an inline literal.
3. Write it as an escape. The current diff contains a raw U+FEFF character inside "...", which is invisible in editors and in review. Spell it "" so it is readable. The startswith check is then unnecessary: data = data.removeprefix(UTF8_BOM) does the same thing.

Separately, pre-commit.ci is red: mypy wants annotations on the three new tests (-> None and tmp_path: Path).

---

Generated by Claude Code

[RonnyPfannschmidt]

🤖 Written by Claude Opus 5.5 via Claude Code for the iniconfig maintainers; I prompted it, it did the work, I read it.

Correction to my review: both snippets in it were mangled into the raw invisible BOM. The intended code is:

UTF8_BOM: Final = "\N{BYTE ORDER MARK}"

data = data.removeprefix(UTF8_BOM)

---

Generated by Claude Code

[Pitchfork-and-Torch]

Moved the strip into parse_ini_data. The constant is UTF8_BOM: Final = "\N{BYTE ORDER MARK}", and the three new tests now have return annotations plus a Path type on tmp_path.

[RonnyPfannschmidt]

Well done thanks

do you want to squash or should I squash merge

[Pitchfork-and-Torch]

Please squash merge — happy either way as long as it lands as one commit. Thanks!