urllib should fsdecode percent-encoded parts of file URIs on Unix

[manueljacob]

BPO	40996
Nosy	@ezio-melotti, @manueljacob

^{Note: these values reflect the state of the issue at the time it was migrated and might not reflect the current state.}

Show more details

GitHub fields:

assignee = None
closed_at = None
created_at = <Date 2020-06-17.00:19:10.953>
labels = ['type-bug', 'library', 'expert-unicode']
title = 'urllib should fsdecode percent-encoded parts of file URIs on Unix'
updated_at = <Date 2020-06-17.10:14:42.633>
user = 'https://github.com/manueljacob'

bugs.python.org fields:

activity = <Date 2020-06-17.10:14:42.633>
actor = 'vstinner'
assignee = 'none'
closed = False
closed_date = None
closer = None
components = ['Library (Lib)', 'Unicode']
creation = <Date 2020-06-17.00:19:10.953>
creator = 'mjacob'
dependencies = []
files = []
hgrepos = []
issue_num = 40996
keywords = []
message_count = 1.0
messages = ['371702']
nosy_count = 2.0
nosy_names = ['ezio.melotti', 'mjacob']
pr_nums = []
priority = 'normal'
resolution = None
stage = None
status = 'open'
superseder = None
type = 'behavior'
url = 'https://bugs.python.org/issue40996'
versions = []

Linked PRs

• GH-85168: Use filesystem encoding when converting to/from file URIs #126852
• [3.13] GH-85168: Use filesystem encoding when converting to/from file URIs (GH-126852) #127039
• [3.12] GH-85168: Use filesystem encoding when converting to/from file URIs (GH-126852) #127040

[manueljacob]

On Unix, file names are bytes. Python mostly prefers to use unicode for file names. On the Python <-> system boundary, os.fsencode() / os.fsdecode() are used.

In URIs, bytes can be percent-encoded. On Unix, most applications pass the percent-decoded bytes in file URIs to the file system unchanged. The remainder of this issue description is about Unix, except for the last paragraph.

Pathlib fsencodes the path when making a file URI, roundtripping the bytes e.g. passed as an argument:
% python3 -c 'import pathlib, sys; print(pathlib.Path(sys.argv[1]).as_uri())' /tmp/a$(echo -e '\xE4')
file:///tmp/a%E4

Example with curl using this URL:
% echo 'Hello, World!' > /tmp/a$(echo -e '\xE4')
% curl file:///tmp/a%E4
Hello, World!

Python 2’s urllib works the same:
% python2 -c 'from urllib import urlopen; print(repr(urlopen("file:///tmp/a%E4").read()))'
'Hello, World!\n'

However, Python 3’s urllib fails:
% python3 -c 'from urllib.request import urlopen; print(repr(urlopen("file:///tmp/a%E4").read()))' 
Traceback (most recent call last):
  File "/usr/lib/python3.8/urllib/request.py", line 1507, in open_local_file
    stats = os.stat(localfile)
FileNotFoundError: [Errno 2] No such file or directory: '/tmp/a�'

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
  File "<string>", line 1, in <module>
  File "/usr/lib/python3.8/urllib/request.py", line 222, in urlopen
    return opener.open(url, data, timeout)
  File "/usr/lib/python3.8/urllib/request.py", line 525, in open
    response = self._open(req, data)
  File "/usr/lib/python3.8/urllib/request.py", line 542, in _open
    result = self._call_chain(self.handle_open, protocol, protocol +
  File "/usr/lib/python3.8/urllib/request.py", line 502, in _call_chain
    result = func(*args)
  File "/usr/lib/python3.8/urllib/request.py", line 1485, in file_open
    return self.open_local_file(req)
  File "/usr/lib/python3.8/urllib/request.py", line 1524, in open_local_file
    raise URLError(exp)
urllib.error.URLError: <urlopen error [Errno 2] No such file or directory: '/tmp/a�'>

urllib.request.url2pathname() is the function converting the path of the file URI to a file name. On Unix, it uses urllib.parse.unquote() with the default settings (UTF-8 encoding and the "replace" error handler).

I think that on Unix, the settings from os.fsdecode() should be used, so that it roundtrips with pathlib.Path.as_uri() and so that the percent-decoded bytes are passed to the file system as-is.

On Windows, I couldn’t do experiments, but using UTF-8 seems like the right thing (according to https://en.wikipedia.org/wiki/File_URI_scheme#Windows_2). I’m not sure that the "replace" error handler is a good idea. I prefer "errors should never pass silently" from the Zen of Python, but I don’t a have a strong opinion on this.