GH-116377: Stop raising `ValueError` from `glob.translate()`. (#116378)

Stop raising `ValueError` from `glob.translate()` when a `**` sub-string
appears in a non-recursive pattern segment. This matches `glob.glob()`
behaviour.
This commit is contained in:
Barney Gale 2024-03-17 17:09:35 +00:00 committed by GitHub
parent 1cf0301086
commit 0634201f53
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
4 changed files with 18 additions and 26 deletions

View File

@ -136,8 +136,7 @@ The :mod:`glob` module defines the following functions:
separators, and ``*`` pattern segments match precisely one path segment. separators, and ``*`` pattern segments match precisely one path segment.
If *recursive* is true, the pattern segment "``**``" will match any number If *recursive* is true, the pattern segment "``**``" will match any number
of path segments. If "``**``" occurs in any position other than a full of path segments.
pattern segment, :exc:`ValueError` is raised.
If *include_hidden* is true, wildcards can match path segments that start If *include_hidden* is true, wildcards can match path segments that start
with a dot (``.``). with a dot (``.``).

View File

@ -256,8 +256,7 @@ def translate(pat, *, recursive=False, include_hidden=False, seps=None):
"""Translate a pathname with shell wildcards to a regular expression. """Translate a pathname with shell wildcards to a regular expression.
If `recursive` is true, the pattern segment '**' will match any number of If `recursive` is true, the pattern segment '**' will match any number of
path segments; if '**' appears outside its own segment, ValueError will be path segments.
raised.
If `include_hidden` is true, wildcards can match path segments beginning If `include_hidden` is true, wildcards can match path segments beginning
with a dot ('.'). with a dot ('.').
@ -291,22 +290,18 @@ def translate(pat, *, recursive=False, include_hidden=False, seps=None):
for idx, part in enumerate(parts): for idx, part in enumerate(parts):
if part == '*': if part == '*':
results.append(one_segment if idx < last_part_idx else one_last_segment) results.append(one_segment if idx < last_part_idx else one_last_segment)
continue elif recursive and part == '**':
if recursive: if idx < last_part_idx:
if part == '**': if parts[idx + 1] != '**':
if idx < last_part_idx: results.append(any_segments)
if parts[idx + 1] != '**': else:
results.append(any_segments) results.append(any_last_segments)
else: else:
results.append(any_last_segments) if part:
continue if not include_hidden and part[0] in '*?':
elif '**' in part: results.append(r'(?!\.)')
raise ValueError("Invalid pattern: '**' can only be an entire path component") results.extend(fnmatch._translate(part, f'{not_sep}*', not_sep))
if part: if idx < last_part_idx:
if not include_hidden and part[0] in '*?': results.append(any_sep)
results.append(r'(?!\.)')
results.extend(fnmatch._translate(part, f'{not_sep}*', not_sep))
if idx < last_part_idx:
results.append(any_sep)
res = ''.join(results) res = ''.join(results)
return fr'(?s:{res})\Z' return fr'(?s:{res})\Z'

View File

@ -452,9 +452,9 @@ class GlobTests(unittest.TestCase):
self.assertEqual(fn('?'), r'(?s:[^/])\Z') self.assertEqual(fn('?'), r'(?s:[^/])\Z')
self.assertEqual(fn('**'), r'(?s:.*)\Z') self.assertEqual(fn('**'), r'(?s:.*)\Z')
self.assertEqual(fn('**/**'), r'(?s:.*)\Z') self.assertEqual(fn('**/**'), r'(?s:.*)\Z')
self.assertRaises(ValueError, fn, '***') self.assertEqual(fn('***'), r'(?s:[^/]*)\Z')
self.assertRaises(ValueError, fn, 'a**') self.assertEqual(fn('a**'), r'(?s:a[^/]*)\Z')
self.assertRaises(ValueError, fn, '**b') self.assertEqual(fn('**b'), r'(?s:[^/]*b)\Z')
self.assertEqual(fn('/**/*/*.*/**'), r'(?s:/(?:.+/)?[^/]+/[^/]*\.[^/]*/.*)\Z') self.assertEqual(fn('/**/*/*.*/**'), r'(?s:/(?:.+/)?[^/]+/[^/]*\.[^/]*/.*)\Z')
def test_translate_seps(self): def test_translate_seps(self):

View File

@ -512,8 +512,6 @@ class DummyPurePathTest(unittest.TestCase):
self.assertFalse(P('a/b/c.py').full_match('**/a/b/c./**')) self.assertFalse(P('a/b/c.py').full_match('**/a/b/c./**'))
self.assertFalse(P('a/b/c.py').full_match('/a/b/c.py/**')) self.assertFalse(P('a/b/c.py').full_match('/a/b/c.py/**'))
self.assertFalse(P('a/b/c.py').full_match('/**/a/b/c.py')) self.assertFalse(P('a/b/c.py').full_match('/**/a/b/c.py'))
self.assertRaises(ValueError, P('a').full_match, '**a/b/c')
self.assertRaises(ValueError, P('a').full_match, 'a/b/c**')
# Case-sensitive flag # Case-sensitive flag
self.assertFalse(P('A.py').full_match('a.PY', case_sensitive=True)) self.assertFalse(P('A.py').full_match('a.PY', case_sensitive=True))
self.assertTrue(P('A.py').full_match('a.PY', case_sensitive=False)) self.assertTrue(P('A.py').full_match('a.PY', case_sensitive=False))