fix(catalog): refine PYTHON_FORMAT regex to avoid prose false positives

Allow space flag only when followed by explicit width or precision. Add test cases in test_catalog.py and test_checkers.py.

Author: goodluck-ry

babel/messages/catalog.py

@@ -74,10 +74,11 @@ def get_close_matches(word, possibilities, n=3, cutoff=0.6):
     \%
         (?:\(([\w]*)\))?
         (
-            [-#0+]?(?:\*|[\d]+)?
-            (?:\.(?:\*|[\d]+))?
-            [hlL]?
+            (?:[-#0+]?(?:\*|[\d]+)?(?:\.(?:\*|[\d]+))?)
+            |
+            (?:\ +(?:\.[\d]+|[\d]+(?:\.[\d]+)?))
         )
+        [hlL]?
         ([diouxXeEfFgGcrs%])
 ''',
     re.VERBOSE,

tests/messages/test_catalog.py

@@ -38,6 +38,18 @@ def test_message_python_format():
     assert catalog.PYTHON_FORMAT.search('foo %()s')
 
 
+def test_message_python_format_prose_percent():
+    # literal percent signs in prose must not be mistaken for placeholders
+    assert not catalog.PYTHON_FORMAT.search('100 % done')
+    assert not catalog.PYTHON_FORMAT.search('50 % off')
+    assert not catalog.PYTHON_FORMAT.search('10% of')
+    assert not catalog.PYTHON_FORMAT.search('10% der')
+    # space-flag placeholders with explicit width/precision are still valid
+    assert catalog.PYTHON_FORMAT.search('% 5d')
+    assert catalog.PYTHON_FORMAT.search('% .2f')
+    assert catalog.PYTHON_FORMAT.search('% 5.2f')
+
+
 def test_message_python_brace_format():
     assert not catalog._has_python_brace_format('')
     assert not catalog._has_python_brace_format('foo')

tests/messages/test_checkers.py

@@ -390,6 +390,7 @@ def test_python_format_valid(msgid, msgstr):
             '%(foo)d',
             "incompatible format for placeholder 'foo': 'd' and 's' are not compatible",
         ),
+        ('% 5d', '% 5s', "incompatible format for placeholder 1: 'd' and 's' are not compatible"),
     ],
 )
 def test__validate_format_invalid(msgid, msgstr, error):
@@ -412,6 +413,11 @@ def test__validate_format_invalid(msgid, msgstr, error):
         ('%(foo)s', 'foo'),
         ('%(foo)s', '%(foo)s %(foo)s'),
         ('%(bar)s foo %(n)d', '%(n)d foo %(bar)s'),
+        # literal percent signs in prose are not placeholders
+        ('100 % done', '100 % erledigt'),
+        ('50 % off', '50 % Rabatt'),
+        # space-flag placeholders with explicit width/precision are still checked
+        ('% 5d', '% 5d'),
     ],
 )
 def test__validate_format_valid(msgid, msgstr):