Allow path to be pattern in list_dir_names()

[hagenw]

Considering #76

This extends audeer.list_dir_names(path) by the possibility to provide a pattern as path.
It raises then the same errors as audeer.list_file_names() does in #93.

The only difference is that due to how we recognize what a pattern is it behaves slightly different.
E.g. assume we have a file called path/abc.
Then we would get

>>> audeer.list_file_names('path/abc', basenames=True)
['abc']
>>> audeer.list_file_names('path/a*', basenames=True)
['abc']

Now assume we have a folder called path/abc.
Then we would get

>>> audeer.list_dir_names('path/abc', basenames=True)
[]
>>> audeer.list_dir_names('path/a*', basenames=True)
['abc']

[codecov]

Codecov Report

Merging #92 (b4d8b4d) into main (118498b) will not change coverage.
The diff coverage is 100.0%.

Impacted Files	Coverage Δ
audeer/core/io.py	100.0% <100.0%> (ø)

[frankenjoe]

Wouldn't it make sense we first find a solution for #76? I guess with the changes you propose here we no longer raise an error if a directory does not exist, but return an empty list instead.

[hagenw]

In my opinion #76 is solved by #93.

I guess it was not closed as I wrote "Closing" instead of "Closes" in the pull request ;)

[frankenjoe]

As far as I see in #93 we only extended the docstring, but we didn't solve the actual issue, namely that the following returns an empty list:

>>> audeer.list_file_names('./does-not-exist/')
[]

and I think the changes proposed here introduce now the same issue for list_dir_names().

[frankenjoe]

Since we are struggeling to detect if path is a folder or a pattern in #94, I think it is not a good idea to introduce the same issue here. We should consider to introduce a separate pattern argument instead.