feat: Make windows-compatible

[Danipulok]

Summary

Hey here!
Thank you very much for the library, it is really useful and helpful

This PR fixes how pytest-examples works on Windows with sane (utf-8) default

Motivation

I'm using pytest-examples library in a small library I will release, and noticed that on Windows I get errors when try to run tests in docs with emojis
Windows has a thing where if you don't specify encoding, the default regional encoding is used, which is never utf-8

Traceback in my project, where I use pytest-examples with current code:

FAILED tests/test_docs.py::test_docs_examples[docs\\lax.md:9-22] - UnicodeEncodeError: 'charmap' codec can't encode character '\u2713' in position 237: character maps to <undefined>

Result of a test I have added in this PR with current code:

self = <encodings.cp1252.IncrementalEncoder object at 0x0000023B17AF9BB0>
input = "print('🚀 ✓')\r\n#> 🚀 ✓\r\n", final = False

    def encode(self, input, final=False):
>       return codecs.charmap_encode(input,self.errors,encoding_table)[0]
E       UnicodeEncodeError: 'charmap' codec can't encode character '\U0001f680' in position 7: character maps to <undefined>

..\..\..\..\..\Roaming\uv\python\cpython-3.14.0-windows-x86_64-none\Lib\encodings\cp1252.py:19: UnicodeEncodeError

[Danipulok]

Good day, @samuelcolvin!
I know you are really busy and have a lot to do, but is there any way this PR would get merged?
I tested in on Windows with all possible ways and added a test that failed on Windows before my PR
I really like the library and IMHO this would be helpful to a lot of developers on Windows

[hramezani]

Thanks for this, and sorry for the slow response! 🙏

I reproduced the bug on Windows (cp1252) and confirmed the fix works with no regressions.

One suggestion before we merge: rather than adding a configurable encoding option, I think we should just hardcode encoding='utf-8' at the write/subprocess sites. Reasoning: the read path already hardcodes read_text('utf-8') in find_examples.py, so the library is effectively utf-8-only already — a configurable write encoding could produce files the reader can't read back. Python source is utf-8 by definition anyway, so utf-8 is really the only correct value here.

Concretely that means: drop the encoding field/param from ExamplesConfig, set_config, and _modify_files, and just pass encoding='utf-8' in _write_file, the ruff_check Popen, and _modify_files. Smaller diff, symmetric with the read path, nothing to misconfigure. The regression test is great as-is.

Happy to push this tweak myself if that's easier. Thanks again!

[Danipulok]

@hramezani, hey!

Thank you very much for the response!
I will do it myself and change it today in a few hours

I think this is a nice idea to simply hardcode it. Because since 3.15 we will have "utf-8" by default anyway

[Danipulok]

@hramezani, should be good, please check