The documentation for "ZipFile.writestr(zinfo_or_arcname, data[, compress_type])" says: "Write the string data to the archive; [...]" --> https://docs.python.org/3/library/zipfile.html I fails to mention that data could also be bytes. The source code does mention it however, that is how I found out: def writestr(self, zinfo_or_arcname, data, compress_type=None): """Write a file into the archive. The contents is 'data', which may be either a 'str' or a 'bytes' instance; if it is a 'str', it is encoded as UTF-8 first. 'zinfo_or_arcname' is either a ZipInfo instance or the name of the file in the archive.""" I believe this should be added to the documentation as it's an important information to those who try to add non-UTF8 StringIO data to in-memory ZipFiles using BytesIO.
The change in current PR is very minor and looks good enough that aligns the docs with the docstring and functionality of `ZipFile.writestr` as described. Can a core-contributor, please look at this? Also, I'm adding 3.8 for this change.
Since the docstring contains more information, it is better to just copy words from the docstring to the documentation. There are also other places where string is improperly used instead of bytes or the type is not specified. The note about file name encoding doesn't make sense in Python 3, since ZipFile.write() doesn't support bytes as a file name. PR 10592 fixed this.
