json-repair 0.15.5__tar.gz → 0.16.0__tar.gz

{json_repair-0.15.5/src/json_repair.egg-info → json_repair-0.16.0}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: json_repair
-Version: 0.15.5
+Version: 0.16.0
 Summary: A package to repair broken json strings
 Author-email: Stefano Baccianella <4247706+mangiucugna@users.noreply.github.com>
 License: MIT License

{json_repair-0.15.5 → json_repair-0.16.0}/pyproject.toml

@@ -3,7 +3,7 @@ requires = ["setuptools>=61.0"]
 build-backend = "setuptools.build_meta"
 [project]
 name = "json_repair"
-version = "0.15.5"
+version = "0.16.0"
 license = {file = "LICENSE"}
 authors = [
   { name="Stefano Baccianella", email="4247706+mangiucugna@users.noreply.github.com" },

{json_repair-0.15.5 → json_repair-0.16.0}/src/json_repair/json_repair.py

@@ -64,7 +64,7 @@ class JSONParser:
             return self.parse_array()
         # there can be an edge case in which a key is empty and at the end of an object
         # like "key": }. We return an empty string here to close the object properly
-        elif char == "}" and self.get_context() == "object_value":
+        elif char == "}":
             self.log(
                 "At the end of an object we found a key with missing value, skipping",
                 "info",
@@ -78,13 +78,13 @@ class JSONParser:
         elif char == "“":
             return self.parse_string(string_quotes=["“", "”"])
         # <number> starts with [0-9] or minus
-        elif char.isdigit() or char == "-" or char == ".":
+        elif self.get_context() != "" and char.isdigit() or char == "-" or char == ".":
             return self.parse_number()
         # <boolean> could be (T)rue or (F)alse or (N)ull
-        elif char.lower() in ["t", "f", "n"]:
+        elif self.get_context() != "" and char.lower() in ["t", "f", "n"]:
             return self.parse_boolean_or_null()
         # This might be a <string> that is missing the starting '"'
-        elif char.isalpha():
+        elif self.get_context() != "" and char.isalpha():
             return self.parse_string()
         # If everything else fails, we just ignore and move on
         else:
@@ -131,6 +131,8 @@ class JSONParser:
                         "info",
                     )
                     break
+                elif key == "":
+                    self.index += 1
 
             # We reached the end here
             if (self.get_char_at() or "}") == "}":
@@ -172,8 +174,10 @@ class JSONParser:
     def parse_array(self) -> List[Any]:
         # <array> ::= '[' [ <json> *(', ' <json>) ] ']' ; A sequence of JSON values separated by commas
         arr = []
+        self.set_context("array")
         # Stop when you either find the closing parentheses or you have iterated over the entire string
         while (self.get_char_at() or "]") != "]":
+            self.skip_whitespaces_at()
             value = self.parse_json()
 
             # It is possible that parse_json() returns nothing valid, so we stop
@@ -209,6 +213,7 @@ class JSONParser:
             self.index -= 1
 
         self.index += 1
+        self.reset_context()
         return arr
 
     def parse_string(self, string_quotes=False) -> str:
@@ -218,6 +223,7 @@ class JSONParser:
 
         # Flag to manage corner cases related to missing starting quote
         fixed_quotes = False
+        doubled_quotes = False
         lstring_delimiter = rstring_delimiter = '"'
         if isinstance(string_quotes, list):
             lstring_delimiter = string_quotes[0]
@@ -239,6 +245,7 @@ class JSONParser:
                     "While parsing a string, we found a valid starting doubled quote, ignoring it",
                     "info",
                 )
+                doubled_quotes = True
                 self.index += 1
         char = self.get_char_at()
         if char != lstring_delimiter:
@@ -279,13 +286,9 @@ class JSONParser:
                     self.remove_char_at(-1)
                     self.index -= 1
             # ChatGPT sometimes forget to quote stuff in html tags or markdown, so we do this whole thing here
-            if (
-                char == rstring_delimiter
-                # Next character is not a delimiter
-                and self.get_char_at(1) not in [",", ":", "]", "}"]
-            ):
+            if char == rstring_delimiter:
                 # Special case here, in case of double quotes one after another
-                if self.get_char_at(1) == rstring_delimiter:
+                if doubled_quotes and self.get_char_at(1) == rstring_delimiter:
                     self.log(
                         "While parsing a string, we found a doubled quote, ignoring it",
                         "info",
@@ -294,13 +297,19 @@ class JSONParser:
                     self.remove_char_at()
                 else:
                     # Check if eventually there is a rstring delimiter, otherwise we bail
-                    i = 2
+                    i = 1
                     next_c = self.get_char_at(i)
                     while next_c and next_c != rstring_delimiter:
+                        # If we are in an object context, let's check for the right delimiters
+                        if (
+                            ("object_key" in self.context and next_c == ":")
+                            or ("object_value" in self.context and next_c in ["}", ","])
+                            or ("array" in self.context and next_c in ["]", ","])
+                        ):
+                            break
                         i += 1
                         next_c = self.get_char_at(i)
-                    # In that case we ignore this rstring delimiter
-                    if next_c:
+                    if next_c == rstring_delimiter:
                         self.log(
                             "While parsing a string, we a misplaced quote that would have closed the string but has a different meaning here, ignoring it",
                             "info",
@@ -416,7 +425,7 @@ class JSONParser:
 
     def get_context(self) -> str:
         try:
-            return self.context[0]
+            return self.context[-1]
         except Exception:
             return ""
 

{json_repair-0.15.5 → json_repair-0.16.0/src/json_repair.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: json_repair
-Version: 0.15.5
+Version: 0.16.0
 Summary: A package to repair broken json strings
 Author-email: Stefano Baccianella <4247706+mangiucugna@users.noreply.github.com>
 License: MIT License

{json_repair-0.15.5 → json_repair-0.16.0}/tests/test_json_repair.py

@@ -50,6 +50,8 @@ def test_repair_json():
     assert repair_json("[[1\n\n]") == "[[1]]"
     assert repair_json("{") == "{}"
     assert repair_json("}") == '""'
+    assert repair_json("string") == '""'
+    assert repair_json("stringbeforeobject {}") == '{}'
     assert repair_json('{"') == '{}'
     assert repair_json('["') == '[]'
     assert repair_json("'\"'") == '"\\\""'
@@ -112,6 +114,7 @@ def test_repair_json():
     assert repair_json('{"key": 1/3}') == '{"key": "1/3"}'
     assert repair_json('{"key": .25}') == '{"key": 0.25}'
     assert repair_json("""{    "a": "",    "b": [ { "c": 1} ] \n}```""") == '{"a": "", "b": [{"c": 1}]}'
+    assert repair_json("Based on the information extracted, here is the filled JSON output: ```json { 'a': 'b' } ```") == '{"a": "b"}'
 
 
 def test_repair_json_with_objects():
@@ -190,6 +193,18 @@ def test_repair_json_with_objects():
 }
 ''', True) == {"resourceType": "Bundle", "id": "1", "type": "collection", "entry": [{"resource": {"resourceType": "Patient", "id": "1", "name": [{"use": "official", "family": "Corwin", "given": ["Keisha", "Sunny"], "prefix": ["Mrs."]}, {"use": "maiden", "family": "Goodwin", "given": ["Keisha", "Sunny"], "prefix": ["Mrs."]}]}}]}
     assert repair_json('{\n"html": "<h3 id="aaa">Waarom meer dan 200 Technical Experts - "Passie voor techniek"?</h3>"}', True) == {'html': '<h3 id="aaa">Waarom meer dan 200 Technical Experts - "Passie voor techniek"?</h3>'}
+    assert repair_json("""
+        [
+            {
+                "foo": "Foo bar baz",
+                "tag": "#foo-bar-baz"
+            },
+            {
+                "foo": "foo bar "foobar" foo bar baz.",
+                "tag": "#foo-bar-foobar"
+            }
+        ]
+        """, True) == [{"foo": "Foo bar baz", "tag": "#foo-bar-baz"},{"foo": "foo bar \"foobar\" foo bar baz.", "tag": "#foo-bar-foobar" }]
 
 
 def test_repair_json_corner_cases_generate_by_gpt():