`/a/**` is understood by `Path.full_match` as `/a` followed by multiple segments, at least one

[macmarrum]

Bug report

Bug description:

/a/** is understood by Path.full_match as "/a followed by multiple segments, at least one", instead of "/a followed by multiple segments, including zero".

From https://docs.python.org/3/library/pathlib.html#pathlib-pattern-language

** (entire segment)
Matches any number of file or directory segments, including zero.

# one segment per **
Path('/a/b').full_match('/a/**') # True
# zero segments per **
Path('/a').full_match('/a/**') # False - expected True
# interestingly
Path('/').full_match('/**') # True

CPython versions tested on:

3.13

Operating systems tested on:

Linux

[picnixz]

Path('/a').full_match('/a/**') # False - expected True

I was wrong before, and /a/ is also not matched. I don't know if it's a bug in the code or in the docs.

cc @barneygale

[barneygale]

Path('/a').full_match('/a/**') # False - expected True

I think this is expected, because the path does not begin with /a/. The docs could be clearer on this point.

I was wrong before, and /a/ is also not matched. I don't know if it's a bug in the code or in the docs.

pathlib.Path removes trailing slashes so Path('/a/') becomes Path('/a') and fails to match.

[picnixz]

Oh, so I wasn't wrong before! I don't know how the docs could be improved though.

[macmarrum]

The ** pattern has different interpretations in glob and full_match:

1. In Path('/a').glob('**'), it means "all files and directories under '/a', including '/a' itself"
2. However, Path('/a').full_match('/a/**') interprets it differently, which I find unexpected

from pathlib import Path

from tempfile import gettempdir, mktempi

temp = Path(gettempdir()) / mktemp()
temp_a = temp / 'a'
temp_a.mkdir(parents=True)
temp_a_1 = temp_a / '1.txt'
temp_a_1.touch()
expected = {temp, temp_a, temp_a_1}
actual = {*temp.glob('**')}
print('Actual:', [f"{p}" for p in sorted(actual)])
assert actual == expected, f"\nExpected: {[f"{p}" for p in sorted(expected)]}"

# Actual: ['/tmp/tmpmcdga9my', '/tmp/tmpmcdga9my/a', '/tmp/tmpmcdga9my/a/1.txt']

[barneygale]

In Path('/a').glob('**'), it means "all files and directories under '/a', including '/a' itself"

If pathlib preserved trailing slashes, then the glob would yield '/a/', and that would match '/a/**'. The inconsistency comes from pathlib stripping trailing slashes unfortunately...