Check regex patterns against I-Regexp if available

Author: jg-rp

jsonpath/function_extensions/match.py

@@ -9,6 +9,13 @@
 
     REGEX_AVAILABLE = False
 
+try:
+    from iregexp_check import check
+
+    IREGEXP_AVAILABLE = True
+except ImportError:
+    IREGEXP_AVAILABLE = False
+
 from jsonpath.function_extensions import ExpressionType
 from jsonpath.function_extensions import FilterFunction
 
@@ -23,8 +30,14 @@ class Match(FilterFunction):
 
     def __call__(self, string: str, pattern: str) -> bool:
         """Return `True` if _string_ matches _pattern_, or `False` otherwise."""
-        # XXX: re.fullmatch caches compiled patterns internally, but `map_re` is not
-        # cached.
+        # TODO: re.match caches compiled patterns internally, but `map_re` and `check`
+        # are not cached.
+
+        # TODO: validate literal patterns ar compile time?
+
+        if IREGEXP_AVAILABLE and (not isinstance(pattern, str) or not check(pattern)):
+            return False
+
         if REGEX_AVAILABLE:
             try:
                 pattern = map_re(pattern)

jsonpath/function_extensions/search.py

@@ -9,6 +9,13 @@
 
     REGEX_AVAILABLE = False
 
+try:
+    from iregexp_check import check
+
+    IREGEXP_AVAILABLE = True
+except ImportError:
+    IREGEXP_AVAILABLE = False
+
 from jsonpath.function_extensions import ExpressionType
 from jsonpath.function_extensions import FilterFunction
 
@@ -23,8 +30,14 @@ class Search(FilterFunction):
 
     def __call__(self, string: str, pattern: str) -> bool:
         """Return `True` if _string_ contains _pattern_, or `False` otherwise."""
-        # XXX: re.search caches compiled patterns internally, but `map_re` is not
-        # cached.
+        # TODO: re.search caches compiled patterns internally, but `map_re` and `check`
+        # are not cached.
+
+        # TODO: validate literal patterns ar compile time?
+
+        if IREGEXP_AVAILABLE and (not isinstance(pattern, str) or not check(pattern)):
+            return False
+
         if REGEX_AVAILABLE:
             try:
                 pattern = map_re(pattern)

jsonpath/parse.py

@@ -860,7 +860,8 @@ def _decode_string_literal(self, token: Token) -> str:
                 assert isinstance(rv, str)
                 return rv
             except json.JSONDecodeError as err:
-                raise JSONPathSyntaxError(str(err).split(":")[1], token=token) from None
+                message = f"decode error: {str(err).split(':')[1]}"
+                raise JSONPathSyntaxError(message, token=token) from None
 
         return token.value
 

tests/test_iregexp.py

@@ -0,0 +1,34 @@
+import pytest
+
+try:
+    import iregexp_check  # noqa: F401
+
+    IREGEXP_AVAILABLE = True
+except ImportError:
+    IREGEXP_AVAILABLE = False
+
+import jsonpath
+
+
+@pytest.mark.skipif(IREGEXP_AVAILABLE is False, reason="requires iregexp_check")
+def test_iregexp_check() -> None:
+    # Character classes are OK.
+    query = "$[?match(@, '[0-9]+')]"
+    data = ["123", "abc", "abc123"]
+    assert jsonpath.findall(query, data) == ["123"]
+
+    # Multi character escapes are not.
+    query = "$[?match(@, '\\\\d+')]"
+    assert jsonpath.findall(query, data) == []
+
+
+@pytest.mark.skipif(IREGEXP_AVAILABLE, reason="iregexp_check is available")
+def test_no_iregexp_check() -> None:
+    # Character classes are OK.
+    query = "$[?match(@, '[0-9]+')]"
+    data = ["123", "abc", "abc123"]
+    assert jsonpath.findall(query, data) == ["123"]
+
+    # Multi character escapes are OK when iregexp_check is not installed.
+    query = "$[?match(@, '\\\\d+')]"
+    assert jsonpath.findall(query, data) == ["123"]