Fix empty list and empty dict handling

json_fingerprint/_jfpv1.py

@@ -25,7 +25,7 @@ def _create_json_hash(data: Any, hash_function: str) -> str:
     return m.hexdigest()
 
 
-def _create_sorted_hash_list(data: Dict, hash_function: str) -> List[Dict]:
+def _create_sorted_hash_list(data: List, hash_function: str) -> List[str]:
     """Create a sorted sha256 hash list."""
     out = []
     for obj in data:
@@ -42,7 +42,7 @@ def _build_path(key: str, base_path: str):
     return key
 
 
-def _build_element(path: str, siblings: str, value: Any):
+def _build_element(path: str, siblings: List[str], value: Any):
     """Build an element dictionary based on presence of sibling data."""
     if siblings:
         return {
@@ -57,20 +57,29 @@ def _build_element(path: str, siblings: str, value: Any):
     }
 
 
-def _flatten_json(data: Dict, hash_function: str, path: str = "", siblings: List = [], debug: bool = False) -> List:
+def _flatten_json(data: Any, hash_function: str, path: str = "", siblings: List = [], debug: bool = False) -> List:
     """Flatten json data structures into a sibling-aware data element list."""
     out = []
-    if type(data) is dict:
+
+    # Process non-empty dicts
+    if type(data) is dict and data:
         for key in data.keys():
             p = _build_path(key=f"{{{key}}}", base_path=path)
-            output = _flatten_json(data=data[key], hash_function=hash_function, path=p, siblings=siblings, debug=debug)
+            output = _flatten_json(
+                data=data[key],
+                hash_function=hash_function,
+                path=p,
+                siblings=siblings,
+                debug=debug,
+            )
             out.extend(output)
         return out
 
-    if type(data) is list:
+    # Process non-empty lists
+    if type(data) is list and data:
         p = _build_path(key=f"[{len(data)}]", base_path=path)
 
-        # Iterate and collect sibling structures, which'll be then attached to each sibling element
+        # Iterate and collect sibling structures, which will be then attached to each sibling element
         siblings = []
         for item in data:
             output = _flatten_json(data=item, hash_function=hash_function, path=p, debug=debug)
@@ -83,7 +92,13 @@ def _flatten_json(data: Dict, hash_function: str, path: str = "", siblings: List
 
         # Recurse with each value in list to typecheck it and eventually get the element value
         for item in data:
-            output = _flatten_json(data=item, hash_function=hash_function, path=p, siblings=siblings, debug=debug)
+            output = _flatten_json(
+                data=item,
+                hash_function=hash_function,
+                path=p,
+                siblings=siblings,
+                debug=debug,
+            )
             out.extend(output)
         return out
 

json_fingerprint/_validators.py

@@ -21,7 +21,7 @@
 
 
 def _validate_hash_function(hash_function: str, version: int):
-    if hash_function not in JFPV1_HASH_FUNCTIONS:
+    if version == 1 and hash_function not in JFPV1_HASH_FUNCTIONS:
         err = f"Expected one of supported hash functions '{JFPV1_HASH_FUNCTIONS}', " f"instead got '{hash_function}'"
         raise FingerprintHashFunctionError(err)
 

json_fingerprint/tests/test_create.py

@@ -64,7 +64,7 @@ def test_jfpv1_sha256_structural_distinction_1(self):
         """Test jfpv1 json flattener's structural value distinction.
 
         Verify that:
-        - Identical value content in identical depths, but in different structures,
+        - Identical values at identical depths, but held in different data structures,
           don't produce identical outputs"""
         obj_in_1 = [
             1,
@@ -85,7 +85,7 @@ def test_jfpv1_sha256_structural_distinction_2(self):
         """Test jfpv1 json flattener's structural value distinction.
 
         Verify that:
-        - Values in identical paths/structures but different sibling values don't get matched"""
+        - Values in identical data structure paths, but different sibling values, don't get matched"""
         obj_in_1 = [
             [1, ["x", "x"]],
             [2, ["y", "y"]],
@@ -100,6 +100,40 @@ def test_jfpv1_sha256_structural_distinction_2(self):
 
         self.assertNotEqual(fp_1, fp_2)
 
+    def test_jfpv1_empty_list_as_value(self):
+        """Test jfpv1 json flattener's ability to handle empty lists as values.
+
+        Versions up to 0.12.2 did not acknowledge empty lists as values.
+        Related issue: https://github.com/cobaltine/json-fingerprint/issues/33
+
+        Verify that:
+        - Empty lists (and, as such, underlying data structure paths) are not ignored by the json flattener"""
+
+        obj_in_1 = {"field1": "yes"}
+        fp_1 = create(input=json.dumps(obj_in_1), hash_function="sha256", version=1)
+
+        obj_in_2 = {"field1": "yes", "field2": []}
+        fp_2 = create(input=json.dumps(obj_in_2), hash_function="sha256", version=1)
+
+        self.assertNotEqual(fp_1, fp_2)
+
+    def test_jfpv1_empty_dict_as_value(self):
+        """Test jfpv1 json flattener's ability to handle empty dicts as values.
+
+        Versions up to 0.12.2 did not acknowledge empty dicts as values.
+        Related issue: https://github.com/cobaltine/json-fingerprint/issues/33
+
+        Verify that:
+        - Empty dicts (and, as such, underlying data structure paths) are not ignored by the json flattener"""
+
+        obj_in_1 = {"field1": "yes"}
+        fp_1 = create(input=json.dumps(obj_in_1), hash_function="sha256", version=1)
+
+        obj_in_2 = {"field1": "yes", "field2": {}}
+        fp_2 = create(input=json.dumps(obj_in_2), hash_function="sha256", version=1)
+
+        self.assertNotEqual(fp_1, fp_2)
+
 
 if __name__ == "__main__":
     unittest.main()

json_fingerprint/tests/test_jfpv1.py

@@ -192,6 +192,34 @@ def test_jfpv1_json_flattener_primitive_none_handling(self):
         expected_bool_out_raw = [{"path": "", "value": none_val}]
         self.assertEqual(none_out_raw, expected_bool_out_raw)
 
+    def test_jfpv1_json_flattener_empty_list_handling(self):
+        """Test jfpv1 json flattener's ability to handle empty lists as values.
+
+        Versions up to v0.12.2 did not acknowledge empty lists as values.
+        Related issue: https://github.com/cobaltine/json-fingerprint/issues/33
+
+        Verify that:
+        - Empty lists are considered to be values"""
+
+        empty_list_val = []
+        empty_list_out_raw = _jfpv1._flatten_json(data=empty_list_val, hash_function="sha256")
+        expected_emtpy_list_out_raw = [{"path": "", "value": empty_list_val}]
+        self.assertEqual(empty_list_out_raw, expected_emtpy_list_out_raw)
+
+    def test_jfpv1_json_flattener_empty_dict_handling(self):
+        """Test jfpv1 json flattener's ability to handle empty dicts as values.
+
+        Versions up to v0.12.2 did not acknowledge empty dicts as values.
+        Related issue: https://github.com/cobaltine/json-fingerprint/issues/33
+
+        Verify that:
+        - Empty dicts are considered to be values"""
+
+        empty_dict_val = {}
+        empty_dict_out_raw = _jfpv1._flatten_json(data=empty_dict_val, hash_function="sha256")
+        expected_empty_dict_out_raw = [{"path": "", "value": empty_dict_val}]
+        self.assertEqual(empty_dict_out_raw, expected_empty_dict_out_raw)
+
 
 if __name__ == "__main__":
     unittest.main()