json-repair 0.24.0__tar.gz → 0.25.0__tar.gz

{json_repair-0.24.0/src/json_repair.egg-info → json_repair-0.25.0}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: json_repair
-Version: 0.24.0
+Version: 0.25.0
 Summary: A package to repair broken json strings
 Author-email: Stefano Baccianella <4247706+mangiucugna@users.noreply.github.com>
 License: MIT License

{json_repair-0.24.0 → json_repair-0.25.0}/pyproject.toml

@@ -3,7 +3,7 @@ requires = ["setuptools>=61.0"]
 build-backend = "setuptools.build_meta"
 [project]
 name = "json_repair"
-version = "0.24.0"
+version = "0.25.0"
 license = {file = "LICENSE"}
 authors = [
   { name="Stefano Baccianella", email="4247706+mangiucugna@users.noreply.github.com" },

{json_repair-0.24.0 → json_repair-0.25.0}/src/json_repair/json_repair.py

@@ -370,6 +370,10 @@ class JSONParser:
                 if self.get_context() == "object_key" and (
                     char == ":" or char.isspace()
                 ):
+                    self.log(
+                        "While parsing a string missing the left delimiter in object key context, we found a :, stopping here",
+                        "info",
+                    )
                     break
                 elif self.get_context() == "object_value" and char in [",", "}"]:
                     rstring_delimiter_missing = True
@@ -389,6 +393,10 @@ class JSONParser:
                         if next_c and next_c in [",", "}"]:
                             rstring_delimiter_missing = False
                     if rstring_delimiter_missing:
+                        self.log(
+                            "While parsing a string missing the left delimiter in object value context, we found a , or } and we couldn't determine that a right delimiter was present. Stopping here",
+                            "info",
+                        )
                         break
             string_acc += char
             self.index += 1
@@ -410,6 +418,33 @@ class JSONParser:
                         "While parsing a string, we found a doubled quote, ignoring it",
                         "info",
                     )
+                elif missing_quotes and self.get_context() == "object_value":
+                    # In case of missing starting quote I need to check if the delimeter is the end or the beginning of a key
+                    i = 1
+                    next_c = self.get_char_at(i)
+                    while next_c and next_c not in [
+                        rstring_delimiter,
+                        lstring_delimiter,
+                    ]:
+                        i += 1
+                        next_c = self.get_char_at(i)
+                    if next_c:
+                        # We found a quote, now let's make sure there's a ":" following
+                        i += 1
+                        next_c = self.get_char_at(i)
+                        # found a delimiter, now we need to check that is followed strictly by a comma or brace
+                        while next_c and next_c.isspace():
+                            i += 1
+                            next_c = self.get_char_at(i)
+                        if next_c and next_c == ":":
+                            # Reset the cursor
+                            self.index -= 1
+                            char = self.get_char_at()
+                            self.log(
+                                "In a string with missing quotes and object value context, I found a delimeter but it turns out it was the beginning on the next key. Stopping here.",
+                                "info",
+                            )
+                            break
                 else:
                     # Check if eventually there is a rstring delimiter, otherwise we bail
                     i = 1

{json_repair-0.24.0 → json_repair-0.25.0/src/json_repair.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: json_repair
-Version: 0.24.0
+Version: 0.25.0
 Summary: A package to repair broken json strings
 Author-email: Stefano Baccianella <4247706+mangiucugna@users.noreply.github.com>
 License: MIT License

{json_repair-0.24.0 → json_repair-0.25.0}/tests/test_json_repair.py

@@ -142,6 +142,7 @@ def test_object_edge_cases():
     assert repair_json("""{"b": "xxxxx" true}""") == '{"b": "xxxxx"}'
     assert repair_json('{"key": "Lorem "ipsum" s,"}') == '{"key": "Lorem \\"ipsum\\" s,"}'
     assert repair_json('{"lorem": ipsum, sic, datum.",}') == '{"lorem": "ipsum, sic, datum."}'
+    assert repair_json('{"lorem": sic tamet. "ipsum": sic tamet, quick brown fox. "sic": ipsum}') == '{"lorem": "sic tamet.", "ipsum": "sic tamet", "sic": "ipsum"}'
 
 def test_number_edge_cases():
     assert repair_json(' - { "test_key": ["test_value", "test_value2"] }') == '{"test_key": ["test_value", "test_value2"]}'
@@ -243,7 +244,7 @@ def test_repair_json_from_file():
     # Write content to the temporary file
         with os.fdopen(temp_fd, 'w') as tmp:
             tmp.write("{key:value}")
-        assert(from_file(temp_path, logging=True)) == ({'key': 'value'}, [{'text': 'While parsing a string, we found a literal instead of a quote', 'context': '{key:value}'}, {'text': 'While parsing a string, we found no starting quote. Will add the quote back', 'context': '{key:value}'}, {'text': 'While parsing a string, we missed the closing quote, ignoring', 'context': '{key:value}'}, {'text': 'While parsing a string, we found a literal instead of a quote', 'context': '{key:value}'}, {'text': 'While parsing a string, we found no starting quote. Will add the quote back', 'context': '{key:value}'}, {'text': 'While parsing a string, we missed the closing quote, ignoring', 'context': '{key:value}'}])
+        assert(from_file(temp_path, logging=True)) == ({'key': 'value'}, [{'text': 'While parsing a string, we found a literal instead of a quote', 'context': '{key:value}'}, {'text': 'While parsing a string, we found no starting quote. Will add the quote back', 'context': '{key:value}'}, {'context': '{key:value}', 'text': 'While parsing a string missing the left delimiter in object key context, we found a :, stopping here',}, {'text': 'While parsing a string, we missed the closing quote, ignoring', 'context': '{key:value}'}, {'text': 'While parsing a string, we found a literal instead of a quote', 'context': '{key:value}'}, {'text': 'While parsing a string, we found no starting quote. Will add the quote back', 'context': '{key:value}'}, {'context': '{key:value}', 'text': 'While parsing a string missing the left delimiter in object value context, we found a , or } and we couldn\'t determine that a right delimiter was present. Stopping here'}, {'text': 'While parsing a string, we missed the closing quote, ignoring', 'context': '{key:value}'}])
     finally:
         # Clean up - delete the temporary file
         os.remove(temp_path)