Explicitly specify file encoding for windows

[dalito]

It seems that this place was overlooked. The file encoding is specified on all other non-binary open(...) calls except for this one.

Closes #2006

[sisp]

Thanks for raising this problem and offering a PR to fix it, @dalito! 🙏

I have an inline suggestion. And could you try adding a test that fails without the fix?

[sisp]

Or how about opening the file with "rb" mode and turning the markers into byte strings?

Suggested change

	with Path(fname).open(encoding="utf-8") as conflicts_candidate:
	if any(
	line.rstrip()
	in {"<<<<<<< before updating", ">>>>>>> after updating"}
	with Path(fname).open("rb") as conflicts_candidate:
	if any(
	line.rstrip()
	in {
	b"<<<<<<< before updating",
	b">>>>>>> after updating",
	}

This way, we're independent of encodings.

[dalito]

@sisp Can you point me to the approximate place where to add a test?

Ah, I just remember this: It may not be possible to reproduce this on the Windows runner as the encoding is different than on local Windows10/11.

[sisp]

@sisp Can you point me to the approximate place where to add a test?

I'd add a test at the bottom of https://github.com/copier-org/copier/blob/master/tests/test_updatediff.py.

Ah, I just remember this: It may not be possible to reproduce this on the Windows runner as the encoding is different than on local Windows10/11.

We can mock the default encoding. See, e.g., https://github.com/copier-org/copier/pull/2003/files#diff-7e70d7a9d4f971464d4b2f4f2bb1c5154cd6de349132cdf4acafc3f28b05bf4d. Does this help?