fix: anchor string patterns in match() to consume full keypath. (#579)

String patterns like '*a' were matched with regex.match() which only
checks a prefix, so '(.)*a' matched 'ab' (consuming just 'a', leaving
'b' unconsumed). Fix by appending '$' to string-derived patterns so the
entire keypath must be consumed. Raw compiled-regex patterns are
unaffected and continue to use match() as before.

Author: gaoflow

benedict/core/match.py

@@ -21,6 +21,8 @@ def match(
         pattern = re.sub(r"([\*]{1})", "(.)*", pattern)
         # escape square brackets
         pattern = re.sub(r"(\[([^\[\]]*)\])", "\\[\\g<2>\\]", pattern)
+        # anchor to end so the full keypath must be consumed
+        pattern = pattern + "$"
         regex = re.compile(pattern, flags=re.DOTALL)
     else:
         raise ValueError(f"Expected regex or string, found: {type(pattern)}")

tests/core/test_match.py

@@ -57,6 +57,17 @@ def test_match_with_regex_pattern(self) -> None:
         ]
         self.assertEqual(values, expected_values)
 
+    def test_match_with_suffix_wildcard(self) -> None:
+        # Suffix wildcard like "*.jpg" must not match "IMG_0001.jpg.bak" –
+        # i.e. the full keypath must be consumed, not just a prefix.
+        d = {
+            "IMG_0001.jpg": "IMG_0001.jpg",
+            "IMG_0001.jpg.bak": "IMG_0001.jpg.bak",
+            "DOC_0001.pdf": "DOC_0001.pdf",
+        }
+        values = _match(d, "*.jpg")
+        self.assertEqual(values, ["IMG_0001.jpg"])
+
     def test_match_with_invalid_pattern(self) -> None:
         d = self._get_dict()
         with self.assertRaises(ValueError):