GH-104102: Optimize `pathlib.Path.glob()` handling of `../` pattern segments (GH-104103)

These segments do not require a `stat()` call, as the selector's
`_select_from()` method is called after we've established that the
parent is a directory.
This commit is contained in:
Barney Gale 2023-05-03 00:16:04 +01:00 committed by GitHub
parent 47770a1e91
commit 65a49c6553
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
3 changed files with 19 additions and 0 deletions

View File

@ -74,6 +74,8 @@ def _make_selector(pattern_parts, flavour):
return _TerminatingSelector() return _TerminatingSelector()
if pat == '**': if pat == '**':
cls = _RecursiveWildcardSelector cls = _RecursiveWildcardSelector
elif pat == '..':
cls = _ParentSelector
elif '**' in pat: elif '**' in pat:
raise ValueError("Invalid pattern: '**' can only be an entire path component") raise ValueError("Invalid pattern: '**' can only be an entire path component")
elif _is_wildcard_pattern(pat): elif _is_wildcard_pattern(pat):
@ -114,6 +116,16 @@ class _TerminatingSelector:
yield parent_path yield parent_path
class _ParentSelector(_Selector):
def __init__(self, name, child_parts, flavour):
_Selector.__init__(self, child_parts, flavour)
def _select_from(self, parent_path, is_dir, exists, scandir):
path = parent_path._make_child_relpath('..')
for p in self.successor._select_from(path, is_dir, exists, scandir):
yield p
class _PreciseSelector(_Selector): class _PreciseSelector(_Selector):
def __init__(self, name, child_parts, flavour): def __init__(self, name, child_parts, flavour):

View File

@ -1892,8 +1892,13 @@ class _BasePathTest(object):
P = self.cls P = self.cls
p = P(BASE) p = P(BASE)
self.assertEqual(set(p.glob("..")), { P(BASE, "..") }) self.assertEqual(set(p.glob("..")), { P(BASE, "..") })
self.assertEqual(set(p.glob("../..")), { P(BASE, "..", "..") })
self.assertEqual(set(p.glob("dirA/..")), { P(BASE, "dirA", "..") })
self.assertEqual(set(p.glob("dirA/../file*")), { P(BASE, "dirA/../fileA") }) self.assertEqual(set(p.glob("dirA/../file*")), { P(BASE, "dirA/../fileA") })
self.assertEqual(set(p.glob("dirA/../file*/..")), set())
self.assertEqual(set(p.glob("../xyzzy")), set()) self.assertEqual(set(p.glob("../xyzzy")), set())
self.assertEqual(set(p.glob("xyzzy/..")), set())
self.assertEqual(set(p.glob("/".join([".."] * 50))), { P(BASE, *[".."] * 50)})
@os_helper.skip_unless_symlink @os_helper.skip_unless_symlink
def test_glob_permissions(self): def test_glob_permissions(self):

View File

@ -0,0 +1,2 @@
Improve performance of :meth:`pathlib.Path.glob` when evaluating patterns
that contain ``'../'`` segments.