Martín Blech

@litios this sounds indeed like a valuable contribution for the library and the users. I would be more than happy to review and approve such PR.

I think the problem is that README is Markdown formatted but PyPI expects reStructuredText. I found some documentation here: https://packaging.python.org/guides/making-a-pypi-friendly-readme/ Feel free to take this on and send a Pull...

Thank you for reporting this. I will have to write a few test cases and review the way [CriticalCssMiddleware](https://github.com/martinblech/django-critical/blob/master/critical/middleware.py#L28) handles Unicode, it's a bit messy.

Thanks for debugging this! Happy to merge the MiddlewareMixin if you send a Pull Request.

About the "string pattern on a bytes-like object" this is probably a Python 3 issue. When I first wrote this experiment I was using Python 2 where strings are bytes...

Thank you!

Hey @programmdesign, thanks for your contribution! This PR addresses several issues at once, so I'll leave a few comments on the individual commits.

What would be the expected output if the node with no children appeared twice, though? E.g.: ```python xml = """ """ xmltodict.parse(xml, force_list=('server', )) ``` I agree that `[None, None]`...

@BlackMan82 You are right, it seems that `xmltodict` does not follow the spec for semi-structured XML nodes. We should detect such cases and have `xmltodict` behave as specified, it would...

@joohoi Not much, sorry! Since `xmltodict` uses SAX, not DOM, by the time it's able to detect semi-structured XML, it's already too late to process it differently. It would have...