gh-92550 - Fix regression in `pathlib.Path.rglob()` (GH-92583)

We could try to remedy this by taking a slice, but we then run into an issue where the empty string will match altsep on POSIX. That rabbit hole could keep getting deeper.

A proper fix for the original issue involves making pathlib's path normalisation more configurable - in this case we want to retain trailing slashes, but in other we might want to preserve `./` prefixes, or elide `../` segments when we're sure we won't encounter symlinks.

This reverts commit ea2f5bcda1.
(cherry picked from commit dcdf250d2d)

Co-authored-by: Barney Gale <barney.gale@gmail.com>
This commit is contained in:
Miss Islington (bot) 2022-05-09 17:39:43 -07:00 committed by GitHub
parent a86229e824
commit a51baec9ce
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
5 changed files with 2 additions and 38 deletions

View File

@ -815,9 +815,6 @@ call fails (for example because the path doesn't exist).
.. audit-event:: pathlib.Path.glob self,pattern pathlib.Path.glob
.. versionchanged:: 3.11
Return only directories if *pattern* ends with a pathname components
separator (:data:`~os.sep` or :data:`~os.altsep`).
.. method:: Path.group()
@ -1107,9 +1104,6 @@ call fails (for example because the path doesn't exist).
.. audit-event:: pathlib.Path.rglob self,pattern pathlib.Path.rglob
.. versionchanged:: 3.11
Return only directories if *pattern* ends with a pathname components
separator (:data:`~os.sep` or :data:`~os.altsep`).
.. method:: Path.rmdir()

View File

@ -557,15 +557,6 @@ os
instead of ``CryptGenRandom()`` which is deprecated.
(Contributed by Dong-hee Na in :issue:`44611`.)
pathlib
-------
* :meth:`~pathlib.Path.glob` and :meth:`~pathlib.Path.rglob` return only
directories if *pattern* ends with a pathname components separator:
:data:`~os.sep` or :data:`~os.altsep`.
(Contributed by Eisuke Kawasima in :issue:`22276` and :issue:`33392`.)
re
--

View File

@ -281,8 +281,6 @@ _posix_flavour = _PosixFlavour()
def _make_selector(pattern_parts, flavour):
pat = pattern_parts[0]
child_parts = pattern_parts[1:]
if not pat:
return _TerminatingSelector()
if pat == '**':
cls = _RecursiveWildcardSelector
elif '**' in pat:
@ -945,8 +943,6 @@ class Path(PurePath):
drv, root, pattern_parts = self._flavour.parse_parts((pattern,))
if drv or root:
raise NotImplementedError("Non-relative patterns are unsupported")
if pattern[-1] in (self._flavour.sep, self._flavour.altsep):
pattern_parts.append('')
selector = _make_selector(tuple(pattern_parts), self._flavour)
for p in selector.select_from(self):
yield p
@ -960,8 +956,6 @@ class Path(PurePath):
drv, root, pattern_parts = self._flavour.parse_parts((pattern,))
if drv or root:
raise NotImplementedError("Non-relative patterns are unsupported")
if pattern[-1] in (self._flavour.sep, self._flavour.altsep):
pattern_parts.append('')
selector = _make_selector(("**",) + tuple(pattern_parts), self._flavour)
for p in selector.select_from(self):
yield p

View File

@ -1662,11 +1662,6 @@ class _BasePathTest(object):
else:
_check(p.glob("*/fileB"), ['dirB/fileB', 'linkB/fileB'])
if not os_helper.can_symlink():
_check(p.glob("*/"), ["dirA", "dirB", "dirC", "dirE"])
else:
_check(p.glob("*/"), ["dirA", "dirB", "dirC", "dirE", "linkB"])
def test_rglob_common(self):
def _check(glob, expected):
self.assertEqual(set(glob), { P(BASE, q) for q in expected })
@ -1684,16 +1679,6 @@ class _BasePathTest(object):
"linkB/fileB", "dirA/linkC/fileB"])
_check(p.rglob("file*"), ["fileA", "dirB/fileB",
"dirC/fileC", "dirC/dirD/fileD"])
if not os_helper.can_symlink():
_check(p.rglob("*/"), [
"dirA", "dirB", "dirC", "dirC/dirD", "dirE",
])
else:
_check(p.rglob("*/"), [
"dirA", "dirA/linkC", "dirB", "dirB/linkD", "dirC",
"dirC/dirD", "dirE", "linkB",
])
p = P(BASE, "dirC")
_check(p.rglob("file*"), ["dirC/fileC", "dirC/dirD/fileD"])
_check(p.rglob("*/*"), ["dirC/dirD/fileD"])
@ -2719,7 +2704,6 @@ class WindowsPathTest(_BasePathTest, unittest.TestCase):
P = self.cls
p = P(BASE)
self.assertEqual(set(p.glob("FILEa")), { P(BASE, "fileA") })
self.assertEqual(set(p.glob("*a\\")), { P(BASE, "dirA") })
self.assertEqual(set(p.glob("F*a")), { P(BASE, "fileA") })
self.assertEqual(set(map(str, p.glob("FILEa"))), {f"{p}\\FILEa"})
self.assertEqual(set(map(str, p.glob("F*a"))), {f"{p}\\fileA"})
@ -2728,7 +2712,6 @@ class WindowsPathTest(_BasePathTest, unittest.TestCase):
P = self.cls
p = P(BASE, "dirC")
self.assertEqual(set(p.rglob("FILEd")), { P(BASE, "dirC/dirD/fileD") })
self.assertEqual(set(p.rglob("*\\")), { P(BASE, "dirC/dirD") })
self.assertEqual(set(map(str, p.rglob("FILEd"))), {f"{p}\\dirD\\FILEd"})
def test_expanduser(self):

View File

@ -0,0 +1,2 @@
:meth:`pathlib.Path.rglob` raised :exc:`IndexError` when called with an
empty string. This regression was introduced in 3.11b1.