json-repair 0.52.4__py3-none-any.whl → 0.53.0__py3-none-any.whl

json_repair/json_parser.py

@@ -158,23 +158,32 @@ class JSONParser:
 
     def skip_to_character(self, character: str | list[str], idx: int = 0) -> int:
         """
-        This function quickly iterates to find a character, syntactic sugar to make the code more concise
+        Advance from (self.index + idx) until we hit an *unescaped* target character.
+        Returns the offset (idx) from self.index to that position, or the distance to the end if not found.
         """
-        try:
-            char = self.json_str[self.index + idx]
-        except IndexError:
-            return idx
-        character_list = character if isinstance(character, list) else [character]
-        while char not in character_list:
-            idx += 1
-            try:
-                char = self.json_str[self.index + idx]
-            except IndexError:
-                return idx
-        if self.json_str[self.index + idx - 1] == "\\":
-            # Ah shoot this was actually escaped, continue
-            return self.skip_to_character(character, idx + 1)
-        return idx
+        targets = set(character) if isinstance(character, list) else {character}
+        i = self.index + idx
+        n = len(self.json_str)
+        backslashes = 0  # count of consecutive '\' immediately before current char
+
+        while i < n:
+            ch = self.json_str[i]
+
+            if ch == "\\":
+                backslashes += 1
+                i += 1
+                continue
+
+            # ch is not a backslash; if it's a target and not escaped (even backslashes), we're done
+            if ch in targets and (backslashes % 2 == 0):
+                return i - self.index
+
+            # reset backslash run when we see a non-backslash
+            backslashes = 0
+            i += 1
+
+        # not found; return distance to end
+        return n - self.index
 
     def _log(self, text: str) -> None:
         window: int = 10

json_repair/parse_object.py

@@ -7,7 +7,7 @@ if TYPE_CHECKING:
     from .json_parser import JSONParser
 
 
-def parse_object(self: "JSONParser") -> dict[str, JSONReturnType]:
+def parse_object(self: "JSONParser") -> JSONReturnType:
     # <object> ::= '{' [ <member> *(', ' <member>) ] '}' ; A sequence of 'members'
     obj: dict[str, JSONReturnType] = {}
     start_index = self.index

json_repair/parse_string.py

@@ -2,6 +2,7 @@ from typing import TYPE_CHECKING
 
 from .constants import STRING_DELIMITERS, JSONReturnType
 from .json_context import ContextValues
+from .parse_string_helpers.parse_json_llm_block import parse_json_llm_block
 
 if TYPE_CHECKING:
     from .json_parser import JSONParser
@@ -49,7 +50,14 @@ def parse_string(self: "JSONParser") -> JSONReturnType:
 
     if not missing_quotes:
         self.index += 1
-
+    if self.get_char_at() == "`":
+        ret_val = parse_json_llm_block(self)
+        # If we found a valid JSON block, return it, otherwise continue parsing the string
+        if ret_val is not False:
+            return ret_val
+        self.log(
+            "While parsing a string, we found code fences but they did not enclose valid JSON, continuing parsing the string",
+        )
     # There is sometimes a weird case of doubled quotes, we manage this also later in the while loop
     if self.get_char_at() in STRING_DELIMITERS and self.get_char_at() == lstring_delimiter:
         # If it's an empty key, this was easy

json_repair/parse_string_helpers/parse_json_llm_block.py

@@ -0,0 +1,19 @@
+from typing import TYPE_CHECKING
+
+from ..constants import JSONReturnType  # noqa: TID252
+
+if TYPE_CHECKING:
+    from ..json_parser import JSONParser  # noqa: TID252
+
+
+def parse_json_llm_block(self: "JSONParser") -> JSONReturnType:
+    """
+    Extracts and normalizes JSON enclosed in ```json ... ``` blocks.
+    """
+    # Try to find a ```json ... ``` block
+    if self.json_str[self.index : self.index + 7] == "```json":
+        i = self.skip_to_character("`", idx=7)
+        if self.json_str[self.index + i : self.index + i + 3] == "```":
+            self.index += 7  # Move past ```json
+            return self.parse_json()
+    return False

{json_repair-0.52.4.dist-info → json_repair-0.53.0.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: json_repair
-Version: 0.52.4
+Version: 0.53.0
 Summary: A package to repair broken json strings
 Author-email: Stefano Baccianella <4247706+mangiucugna@users.noreply.github.com>
 License-Expression: MIT

{json_repair-0.52.4.dist-info → json_repair-0.53.0.dist-info}/RECORD

@@ -2,20 +2,21 @@ json_repair/__init__.py,sha256=JdJIZNCKV3MfIviryqK8NH8yGssCta2-192CekcwH-o,174
 json_repair/__main__.py,sha256=EsJb-y89uZEvGQQg1GdIDWzfDwfOMvVekKEtdguQXCM,67
 json_repair/constants.py,sha256=cv2gvyosuq0me0600WyTysM9avrtfXPuXYR26tawcuo,158
 json_repair/json_context.py,sha256=WsMOjqpGSr6aaDONcrk8UFtTurzWon2Qq9AoBBYseoI,934
-json_repair/json_parser.py,sha256=S4Ogczq42CiYbuF06EsDGkVJx05ePgTvrKIUQsZrBzM,7674
+json_repair/json_parser.py,sha256=vy5Z8aiJUVhVmvYEgy0dkYy5WgUmyOeS6PEFiR3cW44,7948
 json_repair/json_repair.py,sha256=sDhXzDZxu0QmaFzICPTtf_q7yOY1A1Lf_iQG6Potsco,11572
 json_repair/object_comparer.py,sha256=XKV3MRab8H7_v4sm-wpEa5le0XX9OeycWo5S-MFm-GI,1716
 json_repair/parse_array.py,sha256=-rh65JcfT-FtXiR6s8RYlMfI-6LzVr08ytlDh6Z2CFE,2181
 json_repair/parse_boolean_or_null.py,sha256=WMSkvvxsp4wvauBcDqtt9WnLMD5SMoxeRfZFXp3FEBc,890
 json_repair/parse_comment.py,sha256=JHtQ_QlxOvPNnMh7lhUaoTjFGelqjhTNq7qn9xUE7SU,2648
 json_repair/parse_number.py,sha256=33zAtkbuVzi9Lqjxu7cXn9WlVzd3WjRx9Ln_LFzVL4o,1259
-json_repair/parse_object.py,sha256=xMpO64sYW0JmUtW75BTD0EdPQjY7PYVgu4tHVXrWB6s,5582
-json_repair/parse_string.py,sha256=banF0nNbONvO4C_SDQMDuCZO99UIQvKzRVki9xWVpfU,25686
+json_repair/parse_object.py,sha256=rnuH5Oxo98OrXhktF0wrOC1vRb5Th_m819Li1EFJzm4,5571
+json_repair/parse_string.py,sha256=--coxoyH4nxl7osxgs1fIu31IEtB0HHwVbbOewypG4g,26146
 json_repair/py.typed,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 json_repair/string_file_wrapper.py,sha256=tGkWBEUPE-CZPf4uSM5NE9oSDTpskX0myJiXsl-gbds,4333
-json_repair-0.52.4.dist-info/licenses/LICENSE,sha256=wrjQo8MhNrNCicXtMe3MHmS-fx8AmQk1ue8AQwiiFV8,1076
-json_repair-0.52.4.dist-info/METADATA,sha256=x1rE0e8rADu8qoy4k4vE2iKEamEsfh-ykMu7xMngX7I,11027
-json_repair-0.52.4.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
-json_repair-0.52.4.dist-info/entry_points.txt,sha256=SNfge3zPSP-ASqriYU9r3NAPaXdseYr7ciPMKdV2uSw,57
-json_repair-0.52.4.dist-info/top_level.txt,sha256=7-VZwZN2CgB_n0NlSLk-rEUFh8ug21lESbsblOYuZqw,12
-json_repair-0.52.4.dist-info/RECORD,,
+json_repair/parse_string_helpers/parse_json_llm_block.py,sha256=taREF3pwb35kGBGJYbUHkTybATX3GI-SOwOz3yXaEQs,644
+json_repair-0.53.0.dist-info/licenses/LICENSE,sha256=wrjQo8MhNrNCicXtMe3MHmS-fx8AmQk1ue8AQwiiFV8,1076
+json_repair-0.53.0.dist-info/METADATA,sha256=JvMUVYGDDIzmym7MqbQ6k6PjbnuuskW_myvk0EWp7V8,11027
+json_repair-0.53.0.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
+json_repair-0.53.0.dist-info/entry_points.txt,sha256=SNfge3zPSP-ASqriYU9r3NAPaXdseYr7ciPMKdV2uSw,57
+json_repair-0.53.0.dist-info/top_level.txt,sha256=7-VZwZN2CgB_n0NlSLk-rEUFh8ug21lESbsblOYuZqw,12
+json_repair-0.53.0.dist-info/RECORD,,