json-repair 0.15.6__py3-none-any.whl → 0.16.1__py3-none-any.whl

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 "}") == "}":
@@ -172,6 +174,7 @@ 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()
@@ -210,6 +213,7 @@ class JSONParser:
             self.index -= 1
 
         self.index += 1
+        self.reset_context()
         return arr
 
     def parse_string(self, string_quotes=False) -> str:
@@ -294,25 +298,41 @@ class JSONParser:
                 else:
                     # Check if eventually there is a rstring delimiter, otherwise we bail
                     i = 1
-                    context = self.get_context()
                     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 (
-                            (context == "object_key" and next_c == ":")
-                            or (context == "object_value" and next_c in ["}", ","])
-                            or (context == "" and next_c in ["]", ","])
+                            next_c == lstring_delimiter
+                            or ("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)
                     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",
-                        )
-                        self.index += 1
-                        char = self.get_char_at()
+                        # But this might not be it! This could be just a missing comma
+                        # We need to check if we find a rstring_delimiter and a colon after
+                        i += 1
+                        next_c = self.get_char_at(i)
+                        while next_c and next_c != rstring_delimiter:
+                            i += 1
+                            next_c = self.get_char_at(i)
+                        i += 1
+                        next_c = self.get_char_at(i)
+                        while next_c and next_c != ":":
+                            if next_c in [lstring_delimiter, rstring_delimiter, ","]:
+                                break
+                            i += 1
+                            next_c = self.get_char_at(i)
+                        # Only if we fail to find a ':' then we know this is misplaced quote
+                        if next_c != ":":
+                            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",
+                            )
+                            self.index += 1
+                            char = self.get_char_at()
 
         if (
             char

{json_repair-0.15.6.dist-info → json_repair-0.16.1.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: json_repair
-Version: 0.15.6
+Version: 0.16.1
 Summary: A package to repair broken json strings
 Author-email: Stefano Baccianella <4247706+mangiucugna@users.noreply.github.com>
 License: MIT License

json_repair-0.16.1.dist-info/RECORD

@@ -0,0 +1,7 @@
+json_repair/__init__.py,sha256=AlNie5y6BZBioGi5fzTAUvum_y0U5aL5aNsuQ_68LQc,175
+json_repair/json_repair.py,sha256=P3jzLaa70VyyE-kjRhijJ4gEPMXba3FmsocJF7bPB9I,21417
+json_repair-0.16.1.dist-info/LICENSE,sha256=wrjQo8MhNrNCicXtMe3MHmS-fx8AmQk1ue8AQwiiFV8,1076
+json_repair-0.16.1.dist-info/METADATA,sha256=UCkuATUR-x0kFLbunLN_htTmUy5T5ZAzXOFhEVeISNc,7355
+json_repair-0.16.1.dist-info/WHEEL,sha256=GJ7t_kWBFywbagK5eo9IoUwLW6oyOeTKmQ-9iHFVNxQ,92
+json_repair-0.16.1.dist-info/top_level.txt,sha256=7-VZwZN2CgB_n0NlSLk-rEUFh8ug21lESbsblOYuZqw,12
+json_repair-0.16.1.dist-info/RECORD,,

json_repair-0.15.6.dist-info/RECORD

@@ -1,7 +0,0 @@
-json_repair/__init__.py,sha256=AlNie5y6BZBioGi5fzTAUvum_y0U5aL5aNsuQ_68LQc,175
-json_repair/json_repair.py,sha256=qQosSMpX3r1adMubgHNZ9gT0lSlZBn3Id_98yS1ITwk,20270
-json_repair-0.15.6.dist-info/LICENSE,sha256=wrjQo8MhNrNCicXtMe3MHmS-fx8AmQk1ue8AQwiiFV8,1076
-json_repair-0.15.6.dist-info/METADATA,sha256=BNtrdGCSn8NC9gBrWdmX2bRQv6VJgEcwgCM5_SWLXUY,7355
-json_repair-0.15.6.dist-info/WHEEL,sha256=GJ7t_kWBFywbagK5eo9IoUwLW6oyOeTKmQ-9iHFVNxQ,92
-json_repair-0.15.6.dist-info/top_level.txt,sha256=7-VZwZN2CgB_n0NlSLk-rEUFh8ug21lESbsblOYuZqw,12
-json_repair-0.15.6.dist-info/RECORD,,