urllib.request.url2pathname() mishandles colons in URLs on Windows

[barneygale]

Bug report

Bug description:

On Windows, urllib.request.url2pathname() has overly simplistic handling of DOS drives in URLs.

Per @eryksun:

>>> from urllib.request import url2pathname
>>> url2pathname('//host/share/spam.txt:eggs')
'T:\\eggs'  # expected: r'\\host\share\spam.txt:eggs'

Also:

>>> url2pathname('///c:/spam.txt:eggs')
OSError: Bad URL: ///c|/spam.txt|eggs  # expected: r'c:\spam.txt:eggs'

CPython versions tested on:

CPython main branch

Operating systems tested on:

Windows

[samtstephens]

Hi, could this be assigned to me? I'm an undergrad working to make an open source contribution to a software library. Thanks!

[barneygale]

Hey, yes that's absolutely fine, thank you for offering! A couple of pointers:

• You might want to wait for GH-126212: Fix removal of slashes in file URIs on Windows #126214 to land first
• You probably want to use urllib.parse.urlsplit(), or some other urllib function, to split the URL authority and path.

[barneygale]

I should also note the Python Dev Guide's Lifecycle of a Pull Request guide :)

[samtstephens]

Thank you! Will check it out.

[barneygale]

Just FYI, the PR I mentioned previously has now landed, so feel free to have a crack at solving this issue! Let me know if you need a hand.

[barneygale]

Hey @samtstephens, if you need a hint, you could look at this code in pathlib:

cpython/Lib/pathlib/_local.py

Lines 909 to 920 in 307c633

	if path[:3] == '///':
	# Remove empty authority
	path = path[2:]
	elif path[:12] == '//localhost/':
	# Remove 'localhost' authority
	path = path[11:]
	if path[:3] == '///' or (path[:1] == '/' and path[2:3] in ':|'):
	# Remove slash before DOS device/UNC path
	path = path[1:]
	if path[1:2] == '|':
	# Replace bar with colon in DOS drive
	path = path[:1] + ':' + path[2:]

I think it does roughly what we want to happen in nturl2path.url2pathname() :)

[samtstephens]

Thanks! Me and my partner are trying this out!