Fix #155, handle an edge case with code fences in object value context

Author: mangiucugna

pyproject.toml

@@ -3,7 +3,7 @@ requires = ["setuptools>=61.0"]
 build-backend = "setuptools.build_meta"
 [project]
 name = "json_repair"
-version = "0.52.1"
+version = "0.52.2"
 license = "MIT"
 license-files = ["LICENSE"]
 authors = [

src/json_repair/parse_string.py

@@ -206,6 +206,17 @@ def parse_string(self: "JSONParser") -> JSONReturnType:
             # Check if the object is really over, to avoid doubling the closing brace
             i = self.skip_whitespaces_at(idx=1, move_main_index=False)
             next_c = self.get_char_at(i)
+            if next_c and next_c == "`":
+                # This could be a special case in which the LLM added code fences after the object
+                # So we need to check if there are another two ` after this one`
+                next_c = self.get_char_at(i + 1)
+                if next_c and next_c == "`":
+                    next_c = self.get_char_at(i + 2)
+                    if next_c and next_c == "`":
+                        self.log(
+                            "While parsing a string in object value context, we found a } that closes the object before code fences, stopping here",
+                        )
+                        break
             if not next_c:
                 self.log(
                     "While parsing a string in object value context, we found a } that closes the object, stopping here",

tests/test_parse_object.py

@@ -63,6 +63,7 @@ def test_parse_object_edge_cases():
     assert repair_json("{text:words{words in brackets}}") == '{"text": "words{words in brackets}"}'
     assert repair_json("{text:words{words in brackets}m}") == '{"text": "words{words in brackets}m"}'
     assert repair_json('{"key": "value, value2"```') == '{"key": "value, value2"}'
+    assert repair_json('{"key": "value}```') == '{"key": "value"}'
     assert repair_json("{key:value,key2:value2}") == '{"key": "value", "key2": "value2"}'
     assert repair_json('{"key:"value"}') == '{"key": "value"}'
     assert repair_json('{"key:value}') == '{"key": "value"}'