Attribute pattern fix (#289)

Author: facelessuser

docs/src/markdown/about/changelog.md

@@ -3,6 +3,10 @@ icon: lucide/scroll-text
 ---
 # Changelog
 
+## 2.8.3
+
+-   **FIX**: Fix inefficient attribute pattern.
+
 ## 2.8.2
 
 -   **FIX**: Ensure custom selectors or namespace dictionaries reject non-string keys (@mundanevision20).

soupsieve/__meta__.py

@@ -193,5 +193,5 @@ def parse_version(ver: str) -> Version:
     return Version(major, minor, micro, release, pre, post, dev)
 
 
-__version_info__ = Version(2, 8, 2, "final")
+__version_info__ = Version(2, 8, 3, "final")
 __version__ = __version_info__._get_canonical()

soupsieve/css_parser.py

@@ -118,9 +118,9 @@
 # `nth` content
 NTH = fr'(?:[-+])?(?:[0-9]+n?|n)(?:(?<=n){WSC}*(?:[-+]){WSC}*(?:[0-9]+))?'
 # Value: quoted string or identifier
-VALUE = fr'''(?:"(?:\\(?:.|{NEWLINE})|[^\\"\r\n\f]+)*?"|'(?:\\(?:.|{NEWLINE})|[^\\'\r\n\f]+)*?'|{IDENTIFIER}+)'''
+VALUE = fr'''(?:"(?:\\(?:.|{NEWLINE})|[^\\"\r\n\f]+)*?"|'(?:\\(?:.|{NEWLINE})|[^\\'\r\n\f]+)*?'|{IDENTIFIER})'''
 # Attribute value comparison. `!=` is handled special as it is non-standard.
-ATTR = fr'(?:{WSC}*(?P<cmp>[!~^|*$]?=){WSC}*(?P<value>{VALUE})(?:{WSC}*(?P<case>[is]))?)?{WSC}*\]'
+ATTR = fr'(?:{WSC}*(?P<cmp>[!~^|*$]?=){WSC}*(?P<value>{VALUE})(?:{WSC}*(?P<case>[is]))?)?{WSC}*'
 
 # Selector patterns
 # IDs (`#id`)
@@ -130,7 +130,7 @@
 # Prefix:Tag (`prefix|tag`)
 PAT_TAG = fr'(?P<tag_ns>(?:{IDENTIFIER}|\*)?\|)?(?P<tag_name>{IDENTIFIER}|\*)'
 # Attributes (`[attr]`, `[attr=value]`, etc.)
-PAT_ATTR = fr'\[{WSC}*(?P<attr_ns>(?:{IDENTIFIER}|\*)?\|)?(?P<attr_name>{IDENTIFIER}){ATTR}'
+PAT_ATTR = fr'\[{WSC}*(?P<attr_ns>(?:{IDENTIFIER}|\*)?\|)?(?P<attr_name>{IDENTIFIER}){ATTR}\]'
 # Pseudo class (`:pseudo-class`, `:pseudo-class(`)
 PAT_PSEUDO_CLASS = fr'(?P<name>:{IDENTIFIER})(?P<open>\({WSC}*)?'
 # Pseudo class special patterns. Matches `:pseudo-class(` for special case pseudo classes.

tests/test_extra/test_attribute.py

@@ -1,5 +1,6 @@
 """Test attribute selectors."""
 from .. import util
+import soupsieve as sv
 
 
 class TestAttribute(util.TestCase):
@@ -50,3 +51,14 @@ def test_attribute_not_equal_double_quotes(self):
             ["div", "0", "1", "2", "3", "pre", "4", "6"],
             flags=util.HTML5
         )
+
+    def test_bad_attribute(self):
+        """Test bad attribute fails."""
+
+        with self.assertRaises(sv.SelectorSyntaxError) as cm:
+            sv.compile(r"[\]!=D4XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX")
+
+        e = cm.exception
+        self.assertEqual(e.context, '[\\]!=D4XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX\n^')
+        self.assertEqual(e.line, 1)
+        self.assertEqual(e.col, 1)