feat: resolve issue #13 by adding support for 'in' operator and regex matching (=~)

Author: sean2077

jsonpath/jsonpath.py

@@ -58,7 +58,7 @@ class JSONPath:
     # operators
     REP_SLICE_CONTENT = re.compile(r"^(-?\d*)?:(-?\d*)?(:-?\d*)?$")
     REP_SELECT_CONTENT = re.compile(r"^([\w.']+)(, ?[\w.']+)+$")
-    REP_FILTER_CONTENT = re.compile(r"@([.\[].*?)(?=<=|>=|==|!=|>|<| in| not| is)|len\(@([.\[].*?)\)")
+    REP_FILTER_CONTENT = re.compile(r"@([.\[].*?)(?=<=|>=|==|!=|>|<| in| not| is|\s|\)|$)|len\(@([.\[].*?)\)")
 
     # annotations
     f: list
@@ -216,7 +216,7 @@ def key_func(t, k):
     def _filter(self, obj, i: int, path: str, step: str):
         r = False
         try:
-            r = self.eval_func(step, None, {"__obj": obj})
+            r = self.eval_func(step, None, {"__obj": obj, "RegexPattern": RegexPattern})
         except Exception:
             pass
         if r:
@@ -293,6 +293,10 @@ def _trace(self, obj, i: int, path):
         if step.startswith("?(") and step.endswith(")"):
             step = step[2:-1]
             step = JSONPath.REP_FILTER_CONTENT.sub(self._gen_obj, step)
+
+            if "=~" in step:
+                step = re.sub(r"=~\s*/(.*?)/", r"@ RegexPattern(r'\1')", step)
+
             if isinstance(obj, dict):
                 self._filter(obj, i + 1, path, step)
             self._traverse(self._filter, obj, i + 1, path, step)
@@ -332,6 +336,16 @@ def _trace(self, obj, i: int, path):
             return
 
 
+class RegexPattern:
+    def __init__(self, pattern):
+        self.pattern = pattern
+
+    def __rmatmul__(self, other):
+        if isinstance(other, str):
+            return bool(re.search(self.pattern, other))
+        return False
+
+
 def compile(expr):
     return JSONPath(expr)
 

tests/test_issues.py

@@ -143,3 +143,32 @@ def test_sorting_with_missing_keys():
     data2 = [{"other": 1}, {"other": 2}]
     with pytest.raises(JSONPathTypeError):
         JSONPath("$./(val)").parse(data2)
+
+
+def test_issue_13_contains_and_regex():
+    data = {
+        "products": [
+            {"name": "Apple", "tags": ["fruit", "red"]},
+            {"name": "Banana", "tags": ["fruit", "yellow"]},
+            {"name": "Carrot", "tags": ["vegetable", "orange"]},
+        ]
+    }
+
+    # Case 1: 'in' operator (list contains)
+    # Check if 'fruit' is in tags list
+    expr_in_list = "$.products[?('fruit' in @.tags)].name"
+    assert JSONPath(expr_in_list).parse(data) == ["Apple", "Banana"]
+
+    # Case 2: 'in' operator (string contains)
+    # Check if 'App' is in name
+    expr_in_str = "$.products[?('App' in @.name)].name"
+    assert JSONPath(expr_in_str).parse(data) == ["Apple"]
+
+    # Case 3: Regex operator =~
+    # Check if name starts with 'B'
+    expr_regex = "$.products[?(@.name =~ /B.*/)].name"
+    assert JSONPath(expr_regex).parse(data) == ["Banana"]
+
+    # Case 4: Case insensitive regex
+    expr_regex_case = "$.products[?(@.name =~ /(?i)apple/)].name"
+    assert JSONPath(expr_regex_case).parse(data) == ["Apple"]