json-repair 0.36.1__py3-none-any.whl → 0.38.0__py3-none-any.whl

json_repair/json_parser.py

@@ -101,6 +101,8 @@ class JSONParser:
                 char.isdigit() or char == "-" or char == "."
             ):
                 return self.parse_number()
+            elif char in ["#", "/"]:
+                return self.parse_comment()
             # If everything else fails, we just ignore and move on
             else:
                 self.index += 1
@@ -138,8 +140,9 @@ class JSONParser:
                 # The rollback index needs to be updated here in case the key is empty
                 rollback_index = self.index
                 key = str(self.parse_string())
-
-                if key != "" or (key == "" and self.get_char_at() == ":"):
+                if key == "":
+                    self.skip_whitespaces_at()
+                if key != "" or (key == "" and self.get_char_at() in [":", "}"]):
                     # If the string is empty but there is a object divider, we are done here
                     break
             if ContextValues.ARRAY in self.context.context and key in obj:
@@ -199,11 +202,10 @@ class JSONParser:
             self.skip_whitespaces_at()
             value = self.parse_json()
 
-            # It is possible that parse_json() returns nothing valid, so we stop
+            # It is possible that parse_json() returns nothing valid, so we increase by 1
             if value == "":
-                break
-
-            if value == "..." and self.get_char_at(-1) == ".":
+                self.index += 1
+            elif value == "..." and self.get_char_at(-1) == ".":
                 self.log(
                     "While parsing an array, found a stray '...'; ignoring it",
                 )
@@ -243,6 +245,8 @@ class JSONParser:
         lstring_delimiter = rstring_delimiter = '"'
 
         char = self.get_char_at()
+        if char in ["#", "/"]:
+            return self.parse_comment()
         # A valid string can only start with a valid quote or, in our case, with a literal
         while char and char not in self.STRING_DELIMITERS and not char.isalnum():
             self.index += 1
@@ -438,7 +442,7 @@ class JSONParser:
             string_acc += char
             self.index += 1
             char = self.get_char_at()
-            if char and len(string_acc) > 0 and string_acc[-1] == "\\":
+            if char and string_acc[-1] == "\\":
                 # This is a special case, if people use real strings this might happen
                 self.log("Found a stray escape sequence, normalizing it")
                 if char in [rstring_delimiter, "t", "n", "r", "b", "\\"]:
@@ -646,10 +650,15 @@ class JSONParser:
             self.log(
                 "While parsing a string, we missed the closing quote, ignoring",
             )
+            string_acc = string_acc.rstrip()
         else:
             self.index += 1
 
-        return string_acc.rstrip()
+        if missing_quotes or (string_acc and string_acc[-1] == "\n"):
+            # Clean the whitespaces for some corner cases
+            string_acc = string_acc.rstrip()
+
+        return string_acc
 
     def parse_number(self) -> Union[float, int, str, JSONReturnType]:
         # <number> is a valid real number expressed in one of a number of given formats
@@ -661,7 +670,7 @@ class JSONParser:
             number_str += char
             self.index += 1
             char = self.get_char_at()
-        if len(number_str) > 1 and number_str[-1] in "-eE/,":
+        if number_str and number_str[-1] in "-eE/,":
             # The number ends with a non valid character for a number/currency, rolling back one
             number_str = number_str[:-1]
             self.index -= 1
@@ -748,6 +757,76 @@ class JSONParser:
             return self.skip_to_character(character=character, idx=idx + 1)
         return idx
 
+    def parse_comment(self) -> str:
+        """
+        Parse code-like comments:
+
+        - "# comment": A line comment that continues until a newline.
+        - "// comment": A line comment that continues until a newline.
+        - "/* comment */": A block comment that continues until the closing delimiter "*/".
+
+        The comment is skipped over and an empty string is returned so that comments do not interfere
+        with the actual JSON elements.
+        """
+        char = self.get_char_at()
+        termination_characters = ["\n", "\r"]
+        if ContextValues.ARRAY in self.context.context:
+            termination_characters.append("]")
+        if ContextValues.OBJECT_VALUE in self.context.context:
+            termination_characters.append("}")
+        if ContextValues.OBJECT_KEY in self.context.context:
+            termination_characters.append(":")
+        # Line comment starting with #
+        if char == "#":
+            comment = ""
+            while char and char not in termination_characters:
+                comment += char
+                self.index += 1
+                char = self.get_char_at()
+            self.log(f"Found line comment: {comment}")
+            return ""
+
+        # Comments starting with '/'
+        elif char == "/":
+            next_char = self.get_char_at(1)
+            # Handle line comment starting with //
+            if next_char == "/":
+                comment = "//"
+                self.index += 2  # Skip both slashes.
+                char = self.get_char_at()
+                while char and char not in termination_characters:
+                    comment += char
+                    self.index += 1
+                    char = self.get_char_at()
+                self.log(f"Found line comment: {comment}")
+                return ""
+            # Handle block comment starting with /*
+            elif next_char == "*":
+                comment = "/*"
+                self.index += 2  # Skip '/*'
+                while True:
+                    char = self.get_char_at()
+                    if not char:
+                        self.log(
+                            "Reached end-of-string while parsing block comment; unclosed block comment."
+                        )
+                        break
+                    comment += char
+                    self.index += 1
+                    if comment.endswith("*/"):
+                        break
+                self.log(f"Found block comment: {comment}")
+                return ""
+            else:
+                # Not a recognized comment pattern, skip the slash.
+                self.index += 1
+                return ""
+
+        else:
+            # Should not be reached: if for some reason the current character does not start a comment, skip it.
+            self.index += 1
+            return ""
+
     def _log(self, text: str) -> None:
         window: int = 10
         start: int = max(self.index - window, 0)

{json_repair-0.36.1.dist-info → json_repair-0.38.0.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.2
 Name: json_repair
-Version: 0.36.1
+Version: 0.38.0
 Summary: A package to repair broken json strings
 Author-email: Stefano Baccianella <4247706+mangiucugna@users.noreply.github.com>
 License: MIT License

{json_repair-0.36.1.dist-info → json_repair-0.38.0.dist-info}/RECORD

@@ -1,13 +1,13 @@
 json_repair/__init__.py,sha256=c4L2kZrHvWEKfj_ODU2naliNuvU6FlFVxtF0hbLe6s8,178
 json_repair/__main__.py,sha256=EsJb-y89uZEvGQQg1GdIDWzfDwfOMvVekKEtdguQXCM,67
 json_repair/json_context.py,sha256=mm6dOyrPJ1sDskTORZSXCW7W9-5veMlUKqXQ3Hw3EG4,971
-json_repair/json_parser.py,sha256=VPpOVHHQwFLzI5iX3lLWHrNos2MRJ7cSdjW-lRaI1MA,35771
+json_repair/json_parser.py,sha256=BQsH8CRy59C2176bMwVerfqbHDXfLoEC1v5frmCiv7M,39020
 json_repair/json_repair.py,sha256=LINLSJBs3cJMfs1YRDaIpfWR5PJLs87Oe06G5yQjY18,9729
 json_repair/py.typed,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 json_repair/string_file_wrapper.py,sha256=koZmdq2-Z5K7XF1bDqX6dEbNaVMJYcMTjq-aGe6NQvA,4526
-json_repair-0.36.1.dist-info/LICENSE,sha256=wrjQo8MhNrNCicXtMe3MHmS-fx8AmQk1ue8AQwiiFV8,1076
-json_repair-0.36.1.dist-info/METADATA,sha256=VI-dvPj7yc3-Vq397F1nVCFxEaG6LzVu1t9zkQGloiE,11794
-json_repair-0.36.1.dist-info/WHEEL,sha256=In9FTNxeP60KnTkGw7wk6mJPYd_dQSjEZmXdBdMCI-8,91
-json_repair-0.36.1.dist-info/entry_points.txt,sha256=SNfge3zPSP-ASqriYU9r3NAPaXdseYr7ciPMKdV2uSw,57
-json_repair-0.36.1.dist-info/top_level.txt,sha256=7-VZwZN2CgB_n0NlSLk-rEUFh8ug21lESbsblOYuZqw,12
-json_repair-0.36.1.dist-info/RECORD,,
+json_repair-0.38.0.dist-info/LICENSE,sha256=wrjQo8MhNrNCicXtMe3MHmS-fx8AmQk1ue8AQwiiFV8,1076
+json_repair-0.38.0.dist-info/METADATA,sha256=GBsf-E48jHl2gfFs7kks6AluILNnN43en2QpQc_hmwE,11794
+json_repair-0.38.0.dist-info/WHEEL,sha256=In9FTNxeP60KnTkGw7wk6mJPYd_dQSjEZmXdBdMCI-8,91
+json_repair-0.38.0.dist-info/entry_points.txt,sha256=SNfge3zPSP-ASqriYU9r3NAPaXdseYr7ciPMKdV2uSw,57
+json_repair-0.38.0.dist-info/top_level.txt,sha256=7-VZwZN2CgB_n0NlSLk-rEUFh8ug21lESbsblOYuZqw,12
+json_repair-0.38.0.dist-info/RECORD,,