Capture the directory marker in the */** regex shortcut so !*/ can re-include directories

pathspec/patterns/gitignore/spec.py

@@ -39,6 +39,14 @@
 This regular expression matches the optional directory marker and sub-path.
 """
 
+_MATCH_ALL = f'^(?:.+/)?[^/]+{_DIR_MARK_OPT}'
+"""
+This regular expression matches every path. It is the expansion of the patterns
+"*" and "**" (i.e., "**/{any name}"), and it has to capture the directory marker
+like any other pattern so that :class:`.GitIgnoreSpec` can tell a directory
+match from a file match.
+"""
+
 
 class GitIgnoreSpecPattern(_GitIgnoreBasePattern):
 	"""
@@ -121,7 +129,7 @@ def __normalize_segments(
 				return (None, _DIR_MARK_CG)
 			else:
 				# The pattern "**" will match every path. Special case this pattern.
-				return (None, '.')
+				return (None, _MATCH_ALL)
 
 		elif (
 			seg_count == 2
@@ -130,7 +138,7 @@ def __normalize_segments(
 		):
 			# The pattern "*" will be normalized to "**/*" and will match every
 			# path. Special case this pattern for efficiency.
-			return (None, '.')
+			return (None, _MATCH_ALL)
 
 		elif (
 			seg_count == 3

tests/test_04_gitignore_spec.py

@@ -14,8 +14,10 @@
 	_BYTES_ENCODING)
 from pathspec.patterns.gitignore.spec import (
 	GitIgnoreSpecPattern,
+	_DIR_MARK,
 	_DIR_MARK_CG,
-	_DIR_MARK_OPT)
+	_DIR_MARK_OPT,
+	_MATCH_ALL)
 from pathspec.patterns.gitwildmatch import (
 	GitWildMatchPattern)
 from pathspec.util import (
@@ -275,7 +277,7 @@ def test_03_only_double_asterisk(self):
 		"""
 		regex, include = GitIgnoreSpecPattern.pattern_to_regex('**')
 		self.assertTrue(include)
-		self.assertEqual(regex, '.')
+		self.assertEqual(regex, _MATCH_ALL)
 
 		pattern = GitIgnoreSpecPattern(re.compile(regex), include)
 		results = set(filter(pattern.match_file, [
@@ -332,7 +334,7 @@ def test_03_duplicate_leading_double_asterisk_edge_case(self):
 		"""
 		regex, include = GitIgnoreSpecPattern.pattern_to_regex('**')
 		self.assertTrue(include)
-		self.assertEqual(regex, '.')
+		self.assertEqual(regex, _MATCH_ALL)
 
 		equiv_regex, include = GitIgnoreSpecPattern.pattern_to_regex('**/**')
 		self.assertTrue(include)
@@ -753,7 +755,23 @@ def test_12_asterisk_1_regex(self):
 		"""
 		regex, include = GitIgnoreSpecPattern.pattern_to_regex('*')
 		self.assertTrue(include)
-		self.assertEqual(regex, '.')
+		self.assertEqual(regex, _MATCH_ALL)
+
+	def test_12_asterisk_1b_regex_marks_directories(self):
+		"""
+		Test that the relative asterisk path pattern captures the directory marker.
+
+		Without the marker, "*" outranks a later directory-only pattern (e.g.
+		"!*/") and :class:`.GitIgnoreSpec` reports a directory as ignored where Git
+		does not.
+		"""
+		regex, include = GitIgnoreSpecPattern.pattern_to_regex('*')
+		self.assertTrue(include)
+
+		compiled = re.compile(regex)
+		self.assertIsNotNone(compiled.search('dirA/').group(_DIR_MARK))
+		self.assertIsNone(compiled.search('fileA').group(_DIR_MARK))
+		self.assertIsNone(compiled.search('dirA/fileB').group(_DIR_MARK))
 
 	def test_12_asterisk_2_regex_equivalent(self):
 		"""

tests/test_06_gitignore.py

@@ -205,6 +205,47 @@ def test_02_dir_exclusions(self):
 					'test2/b.bin',
 				}, debug)
 
+	def test_02_dir_reinclusion_whitelist(self):
+		"""
+		Test that a directory re-included by a directory-only pattern is not
+		reported as ignored.
+
+		The whitelist idiom ("*" ignores everything, "!*/" keeps descending into
+		directories, "!*.py" keeps the files of interest) only works if asking
+		about the directory answers what Git answers. A consumer asks about the
+		directory precisely to decide whether to descend, so reporting "sub/" as
+		ignored silently drops every file below it.
+		"""
+		for sub_test in self.parameterize_from_lines([
+			'*',
+			'!*/',
+			'!*.py',
+		]):
+			with sub_test() as spec:
+				# Confirmed results with git check-ignore (v2.55.0).
+				dirs = {
+					'sub/',
+					'sub/d/',
+				}
+				self.assertEqual({_dir for _dir in dirs if spec.match_file(_dir)}, set())
+
+				files = {
+					'a.py',
+					'a.txt',
+					'sub/b.py',
+					'sub/b.txt',
+					'sub/d/c.py',
+				}
+
+				results = list(spec.check_files(files))
+				ignores = get_includes(results)
+				debug = debug_results(spec, results)
+
+				self.assertEqual(ignores, {
+					'a.txt',
+					'sub/b.txt',
+				}, debug)
+
 	def test_02_file_exclusions(self):
 		"""
 		Test file exclusions.