ntpath not adding slash after drive letters

[matan1008]

Bug report

Bug description:

import ntpath
print(ntpath.join('c:', 'x', 'y'))

I would expect the following code to print 'c:\\x\\y', yet it prints 'c:x\\y', Would you expect a PR that changes the current behaviour?

CPython versions tested on:

3.10, 3.11, 3.12, 3.13, CPython main branch

Operating systems tested on:

Linux, Windows

[Wulian233]

In my previous PR
#122372
I found that ntpath and nturl2path are not documented. In the comment of nturl2path, it is mentioned that it is only used by urllib, but it is not in the urllib folder and is not a private module with an _ prefix (like _colorize), so we can import and use it freely.

https://github.com/python/cpython/blob/main/Lib/nturl2path.py

Searching for "import nturl2path" on GitHub also shows that a considerable number of people use it. I am curious why these two modules do not have documentation, is it a historical legacy issue?

Discussion #120423

They are not restricted by the system, but the functions they provide are Windows-specific file paths. Should we add their documentation? Or keep it that way

I want to answer this question, but I don't know if it's a feature or a bug, so we have to read the source code

https://github.com/python/cpython/blob/main/Lib/ntpath.py

[Wulian233]

oh, ntpath is a part of os.path
https://github.com/python/cpython/blob/main/Doc%2Flibrary%2Fos.path.rst#L7-L8

[matan1008]

I actually found the issue using pathlib, so I guess it is a serious API change. I have a PR ready to be pushed, if it is something you would like 😀

[terryjreedy]

In CommandPrompt, cd C:Users fails. If same is true of os.chdir('C:Users'), then this seems a bug.

EDIT: The above is not quite right as cd e:Languages worked in the sense that when I switched to the e drive, Languages had become the working directory there. But c:xxx seems 'Not found' even when 'xxx' exists.

[barneygale]

https://learn.microsoft.com/en-us/windows/win32/fileio/naming-a-file says:

If a file name begins with only a disk designator but not the backslash after the colon, it is interpreted as a relative path to the current directory on the drive with the specified letter. Note that the current directory may or may not be the root directory depending on what it was set to during the most recent "change directory" operation on that disk. Examples of this format are as follows:

• "C:tmp.txt" refers to a file named "tmp.txt" in the current directory on drive C.
• "C:tempdir\tmp.txt" refers to a file in a subdirectory to the current directory on drive C.

[zware]

This is not a bug, it's an oddity of Windows file paths. On POSIX-ish platforms, there are relative (some/path/rooted/in/cwd/example.py) and absolute (/some/path/rooted/at/the/root/example.py) paths. On Windows each drive has its own current working directory (cwd), so there are relative (some\path\rooted\in\cwd\example.py), absolute (C:\some\path\on\drive_c\rooted\in\drive_c_root\example.py), drive-relative¹ (C:some\path\on\drive_c\rooted\in\drive_c_cwd\example.py), and drive-agnostic-absolute¹ (\some\path\on\current\drive\at\its\root\example.py). ntpath can't guess what kind you actually want, so if you want a rooted path you need to provide the root in addition to the drive.

Footnotes

1. Not sure what the real names should be ↩ ↩²

[matan1008]

So in the above example (or in pathlib) what is the correct way to get PureWindowsPath('c:/x/y')? PureWindowsPath('c:') / '\\x' / 'y'?

[zware]

That would be fine, as would PureWindowsPath('c:\\') / 'x' / 'y' or PureWindowsPath('c:') / '/' / 'x' / 'y'