consistently set encoding=utf-8 for opening files

[h-vetinari]

I haven't had issues with running the linter on windows, but today I ran into encoding errors while linting a multi-output rattler recipe (no idea what triggered it exactly). Since python won't default to utf-8 until 3.15(!), it's better to be explicit about the encoding where we're interacting with files. I only needed the changes in lints.py, but while I was at it, I also looked for other uses of bare open() (with the exception of some obviously unix-only files, e.g. those using paths like ~/...)

[hmaarrfk]

I use this in my codebase as well. The only thing is that some LICENSE.txt from packages are going to be non-utf8 encoded. do these changes touch any of those codepaths?

[h-vetinari]

I don't see a reason why license files cannot be encoded in utf-8 themselves, it's not like the encoding is part of the license. Also, I don't think we do any parsing (or even open()-ing`) of license files, so the question shouldn't arise AFAICT.

[h-vetinari]

BTW, I'm now hitting this in rerendering too, which makes this a fair bit more urgent for me...

[hmaarrfk]

I don't see a reason why license files cannot be encoded in utf-8

I'm just saying I saw feedstocks' source code, from the package they point to, have a license with characters that are illegal for UTF-8

[wolfv]

We could maybe also use export PYTHONIOENCODING=utf8?

Or perhaps:

import sys
sys.setdefaultencoding('utf-8')

[h-vetinari]

I don't mind which way we do it. I used the explicit encoding because we already have that in a bunch of places anyway. 🤷

[hmaarrfk]

Looks good.

I don’t like setting the default encoding since it is easy to undo