A. This works (all ascii):
B. This breaks (non-ascii in filename):
Python3 setup:
>>> import urllib.parse
>>> url = r"https://raw.githubusercontent.com/maverickwoo/1015-overleaf/main/è.tex"Note that in Python3, / is not encoded by default:
>>> help(urllib.parse.quote)
Help on function quote in module urllib.parse:
quote(string, safe='/', encoding=None, errors=None)
C. This breaks because the %'s in %C3%A8 are not encoded; this is just like the
original example above:
>>> urllib.parse.quote(url)
'https%3A//raw.githubusercontent.com/maverickwoo/1015-overleaf/main/%C3%A8.tex'D. This breaks because the : in url become %3A in the inner call and then the
% in this %3A gets encoded in the outer call:
>>> urllib.parse.quote(urllib.parse.quote(url))
'https%253A//raw.githubusercontent.com/maverickwoo/1015-overleaf/main/%25C3%25A8.tex'E. This works because the : is not quoted in the inner call:
>>> urllib.parse.quote(urllib.parse.quote(url, safe="/:"), safe="/")
'https%3A//raw.githubusercontent.com/maverickwoo/1015-overleaf/main/%25C3%25A8.tex'F. This also works and looks somewhat better for human-written HTML:
>>> urllib.parse.quote(urllib.parse.quote(url, safe="/:"), safe="/:")
'https://raw.githubusercontent.com/maverickwoo/1015-overleaf/main/%25C3%25A8.tex'G. But really this is what should be done (and this is consistent with advice from an Overleaf developer):
>>> urllib.parse.quote(urllib.parse.quote(url, safe="/:"), safe="")
'https%3A%2F%2Fraw.githubusercontent.com%2Fmaverickwoo%2F1015-overleaf%2Fmain%2F%25C3%25A8.tex'