json_mend 0.3.4 → 0.3.6

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: cf7c2d2fd3815321d7d80912e4ee200cb94a0399220c17219a02f55305f6eccb
-  data.tar.gz: 896e1c154c05ce80032072e33d75ddf86335212b0659b9dac93c9c31959da634
+  metadata.gz: 0db6ff9aaecf22de21a04b136e235fa04d12d43f3c1db614f96eec2c881ce7b9
+  data.tar.gz: 00aa675a9392fdcd9d528f788144ad7a3064c689e9a6a698132c15ff0a9b4130
 SHA512:
-  metadata.gz: 54571b75e4f4448f8c3d87ff2ac2a9d569f5a75e602845ccbc3dda7ff00a935dd8cc2263c18e557e2ffc59ded871467867b0b1fd0371198d05fbaa6f7df46453
-  data.tar.gz: 0f0ef4172ffcbeeb0c28220d6af20603c4fdc7917e833dd2bd111343ec257040767d9a9d21d28d416e8a5096cdfda9baf39a63bf7612ff6cf3b7fb332e8036e2
+  metadata.gz: 22e7132bcfe8f2109be0684afc22bcd21ac8d358dfb0f3b04ad41b14bba5853d82a9d6f1a54d996f8c4ecd9ab0fded3675c147b98fa218513336d7f1856c2a6b
+  data.tar.gz: f347eb625959326c68c2ffb283ba0d7a170a981927540d16c1b9aa526cde70fb1755cd40e8587af308ed17c83adccddca4a0ea49f986c8536419cfaf99af88bb

data/.rubocop.yml

@@ -1,9 +1,18 @@
+inherit_mode:
+  merge:
+    - Exclude
+
 plugins:
   - rubocop-performance
   - rubocop-rspec
+  - rubocop-rake
 
 AllCops:
   NewCops: enable
+  ParserEngine: parser_prism
+  CacheRootDirectory: .rubocop_cache
+  MaxFilesInCache: 8000
+  UseCache: false
   SuggestExtensions: false
 
 Metrics/AbcSize:

data/.tool-versions

@@ -1 +1 @@
-ruby 4.0.2
+ruby 4.0.6

data/lib/json_mend/parser.rb

@@ -42,12 +42,19 @@ module JsonMend
     INVALID_NUMBER_TRAILERS = ['-', 'e', 'E', ','].freeze
 
     # Pre-compile regexes for performance
+    STRING_START_REGEX = /[\p{L}$_]/
     NUMBER_REGEX = /[#{Regexp.escape(NUMBER_CHARS.to_a.join)}]+/
     NUMBER_NO_COMMA_REGEX = /[#{Regexp.escape(NUMBER_CHARS.dup.tap { |s| s.delete(',') }.to_a.join)}]+/
+    INVALID_NUMBER_TRAILERS_REGEX = /[#{Regexp.union(*INVALID_NUMBER_TRAILERS)}]+\z/
+    HEX_ESCAPE_REGEXES = {
+      'u' => /[0-9a-fA-F]{4}/,
+      'x' => /[0-9a-fA-F]{2}/
+    }.freeze
 
     def initialize(json_string)
       @scanner = StringScanner.new(json_string)
       @context = []
+      @current_context = nil
       @depth = 0
     end
 
@@ -72,7 +79,7 @@ module JsonMend
             # Ignore strings that look like closing braces garbage (e.g. "}", " ] ")
             next if new_json.is_a?(String) && new_json.match?(/\A\s*[}\]]+\s*\z/)
 
-            if both_hash?(json.last, new_json)
+            if json.last.is_a?(Hash) && new_json.is_a?(Hash)
               json[-1] = deep_merge_hashes(json.last, new_json)
             else
               json << new_json
@@ -173,7 +180,7 @@ module JsonMend
       with_depth_check do
         object = {}
 
-        @context.push(:object)
+        push_context(:object)
 
         loop do
           skip_whitespaces
@@ -217,7 +224,7 @@ module JsonMend
           object[key] = value
         end
 
-        @context.pop
+        pop_context
 
         object
       end
@@ -294,7 +301,7 @@ module JsonMend
       end
 
       # If no merge happened, proceed with standard key parsing.
-      @context.push(:object_key)
+      push_context(:object_key)
       is_bracketed = false
 
       if char == '['
@@ -305,7 +312,7 @@ module JsonMend
       else
         key = parse_string.to_s
       end
-      @context.pop
+      pop_context
 
       # If the key is empty, consume any stray characters to prevent infinite loops.
       @scanner.getch if key.empty? && !@scanner.check(/[:{\[}\]]/) && !@scanner.eos?
@@ -315,18 +322,18 @@ module JsonMend
 
     # Parses the value part of a key-value pair.
     def parse_object_value(colon_found: true)
-      @context.push(:object_value)
+      push_context(:object_value)
       skip_whitespaces
 
       # Handle cases where the value is missing (e.g. "key": } or "key": ,)
       if @scanner.eos? || @scanner.check(/[,}]/)
-        @context.pop
+        pop_context
         return colon_found ? '' : :inferred_true
       end
 
       # Delegate to the main JSON value parser.
       value = parse_json
-      @context.pop
+      pop_context
 
       # If parse_json returned JSON_STOP_TOKEN (nothing found due to garbage->terminator),
       # treat it as empty string for object values to be safe.
@@ -337,7 +344,7 @@ module JsonMend
     def try_to_merge_dangling_array(object)
       return false unless peek_char == '['
 
-      prev_key = object.keys.last
+      prev_key = object.keys[-1]
       return false unless prev_key && object[prev_key].is_a?(Array)
 
       @scanner.getch # Consume '['
@@ -360,7 +367,7 @@ module JsonMend
     def parse_array
       with_depth_check do
         arr = []
-        @context.push(:array)
+        push_context(:array)
         char = peek_char
         # Stop when you find the closing bracket or an invalid character like '}'
         while !@scanner.eos? && !TERMINATORS_ARRAY.include?(char)
@@ -396,18 +403,15 @@ module JsonMend
             arr << value
           end
 
+          @scanner.skip(/[\s,]+/)
           char = peek_char
-          while char && char != ']' && (char.match?(/\s/) || char == ',')
-            @scanner.getch
-            char = peek_char
-          end
         end
 
         # Handle a potentially missing closing bracket, a common LLM error.
         unless @scanner.scan(']')
           @scanner.scan('}') # Consume } if it was the closer
         end
-        @context.pop
+        pop_context
 
         arr
       end
@@ -445,7 +449,7 @@ module JsonMend
 
       doubled_quotes = rest.first
 
-      string_parts = []
+      string_parts = +''
 
       # Here things get a bit hairy because a string missing the final quote can also be a key or a value in an object
       # In that case we need to use the ":|,|}" characters as terminators of the string
@@ -574,12 +578,12 @@ module JsonMend
         end
 
         break if context_termination_reached?(
-          char:,
-          missing_quotes:
+          char,
+          missing_quotes
         )
 
         if current_context?(:object_value) && TERMINATORS_OBJECT_VALUE.include?(char) &&
-           (string_parts.empty? || string_parts.last != rstring_delimiter)
+           (string_parts.empty? || string_parts[-1] != rstring_delimiter)
 
           is_break = check_rstring_delimiter_missing(
             string_parts:,
@@ -590,7 +594,7 @@ module JsonMend
           break if is_break
         end
 
-        if char == ']' && context_contain?(:array) && string_parts.last != rstring_delimiter
+        if char == ']' && context_contain?(:array) && string_parts[-1] != rstring_delimiter
           i = skip_to_character(rstring_delimiter)
           # No delimiter found
           break unless peek_char(i)
@@ -608,7 +612,7 @@ module JsonMend
         @scanner.getch # Consume the character
         char = peek_char
 
-        if !@scanner.eos? && string_parts.last == '\\'
+        if !@scanner.eos? && string_parts[-1] == '\\'
           # This is a special case, if people use real strings this might happen
           is_next, string_parts, char = parse_escape_sequence(
             string_parts:,
@@ -627,7 +631,7 @@ module JsonMend
           break if is_break
         end
 
-        if char == rstring_delimiter && string_parts.last != '\\'
+        if char == rstring_delimiter && string_parts[-1] != '\\'
           if check_doubled_quotes(doubled_quotes, rstring_delimiter)
             # Consumed in helper
           elsif check_missing_quotes_in_object_value(missing_quotes, lstring_delimiter, rstring_delimiter)
@@ -919,13 +923,7 @@ module JsonMend
           rstring_delimiter_missing = false
         elsif peek_char(j)
           # Check for an unmatched opening brace in string_parts
-          string_parts.reverse_each do |c|
-            next unless c == '{'
-
-            # Ok then this is part of the string
-            rstring_delimiter_missing = false
-            break
-          end
+          rstring_delimiter_missing = false if string_parts.include?('{')
         end
 
       end
@@ -938,18 +936,18 @@ module JsonMend
       char:,
       rstring_delimiter:
     )
-      if !@scanner.eos? && string_parts.last == '\\'
+      if !@scanner.eos? && string_parts[-1] == '\\'
         # This is a special case, if people use real strings this might happen
         if char == rstring_delimiter || ESCAPE_START_CHARS.include?(char)
-          string_parts.pop
+          string_parts.chop!
           string_parts << ESCAPE_MAPPING.fetch(char, char)
 
           @scanner.getch # Consume the character
           char = peek_char
-          while !@scanner.eos? && string_parts.last == '\\' && (char == rstring_delimiter || char == '\\')
+          while !@scanner.eos? && string_parts[-1] == '\\' && (char == rstring_delimiter || char == '\\')
             # this is a bit of a special case, if I don't do this it will close the loop or create a train of \\
             # I don't love it though
-            string_parts.pop
+            string_parts.chop!
             string_parts << char
             @scanner.getch # Consume the character
             char = peek_char
@@ -959,21 +957,12 @@ module JsonMend
           entry_pos = @scanner.pos
           @scanner.getch # consume 'u' or 'x'
 
-          num_chars = (char == 'u' ? 4 : 2)
-          hex_parts = []
-
-          # Use getch in loop to correctly extract chars (handling multibyte)
-          num_chars.times do
-            c = @scanner.getch
-            break unless c
-
-            hex_parts << c
-          end
+          hex_regex = HEX_ESCAPE_REGEXES.fetch(char)
 
           # Validate valid hex digits
-          if hex_parts.length == num_chars && hex_parts.all? { |c| c.match?(/[0-9a-fA-F]/) }
-            string_parts.pop
-            hex_val = hex_parts.join.to_i(16)
+          if (hex_str = @scanner.scan(hex_regex))
+            string_parts.chop!
+            hex_val = hex_str.to_i(16)
 
             if char == 'u' && hex_val.between?(0xD800, 0xDBFF)
               # Handle high surrogate pair
@@ -1013,7 +1002,7 @@ module JsonMend
             @scanner.pos = entry_pos
           end
         elsif STRING_DELIMITERS.include?(char) && char != rstring_delimiter
-          string_parts.pop
+          string_parts.chop!
           string_parts << char
           @scanner.getch # Consume the character
           char = peek_char
@@ -1058,19 +1047,19 @@ module JsonMend
       # we need to update the index only if we had a closing quote
       if char == rstring_delimiter
         @scanner.getch
-      elsif missing_quotes && current_context?(:object_key) && string_parts.last == ','
-        string_parts.pop
+      elsif missing_quotes && current_context?(:object_key) && string_parts[-1] == ','
+        string_parts.chop!
       end
 
-      final_str = string_parts.join
+      final_str = string_parts
       final_str = final_str.rstrip if missing_quotes || final_str.end_with?("\n")
 
       final_str
     end
 
     def context_termination_reached?(
-      char:,
-      missing_quotes:
+      char,
+      missing_quotes
     )
       return false unless missing_quotes
 
@@ -1118,9 +1107,9 @@ module JsonMend
       end
 
       # Handle cases where the number ends with one or more invalid characters.
-      if !scanned_str.empty? && INVALID_NUMBER_TRAILERS.include?(scanned_str[-1])
+      if !scanned_str.empty? && scanned_str.match?(INVALID_NUMBER_TRAILERS_REGEX)
         # Do not rewind scanner, simply discard the invalid trailing chars (garbage)
-        scanned_str = scanned_str[0...-1] while !scanned_str.empty? && INVALID_NUMBER_TRAILERS.include?(scanned_str[-1])
+        scanned_str.sub!(INVALID_NUMBER_TRAILERS_REGEX, '')
       end
 
       # Reject non-numbers (e.g., stray periods "." or dashes "-" from LLM conversational text)
@@ -1306,10 +1295,6 @@ module JsonMend
       res
     end
 
-    def both_hash?(obj1, obj2)
-      obj1.is_a?(Hash) && obj2.is_a?(Hash)
-    end
-
     # Skips whitespaces
     def skip_whitespaces
       @scanner.skip(/\s+/)
@@ -1319,12 +1304,13 @@ module JsonMend
     def peek_char(offset = 0)
       # Handle the common 0-offset case
       if offset.zero?
-        # peek(1) returns the next BYTE, not character
-        byte_str = @scanner.peek(1)
-        return nil if byte_str.empty?
+        pos = @scanner.pos
+        str = @scanner.string
+        return nil if pos >= str.bytesize
 
+        byte = str.getbyte(pos)
         # Fast path: If it's a standard ASCII char (0-127), return it directly.
-        return byte_str if byte_str.getbyte(0) < 128
+        return byte.chr if byte < 128
 
         # Slow path: If it's a multibyte char (e.g. “), use regex to match the full character.
         return @scanner.check(/./m)
@@ -1341,8 +1327,18 @@ module JsonMend
       res
     end
 
+    def push_context(value)
+      @context.push(value)
+      @current_context = value
+    end
+
+    def pop_context
+      @context.pop
+      @current_context = @context.last
+    end
+
     def current_context?(value)
-      @context&.last == value
+      @current_context == value
     end
 
     def context_contain?(value)
@@ -1351,7 +1347,7 @@ module JsonMend
 
     # Checks if the character signifies the start of a string or literal
     def string_start?(char)
-      STRING_DELIMITERS.include?(char) || char&.match?(/[\p{L}$_]/)
+      STRING_DELIMITERS.include?(char) || (char && STRING_START_REGEX.match?(char))
     end
 
     # Checks if the character signifies the start of a number

data/lib/json_mend/version.rb

@@ -1,5 +1,5 @@
 # frozen_string_literal: true
 
 module JsonMend
-  VERSION = '0.3.4'
+  VERSION = '0.3.6'
 end

data/lib/json_mend.rb

@@ -15,21 +15,22 @@ module JsonMend
     # @return [Object, String] The repaired JSON object or string.
     def repair(json_string, return_objects: false)
       # First, attempt to parse the string with the standard library.
-      repaired_json = begin
+      begin
         parsed = JSON.parse(
           json_string,
           allow_trailing_comma: true,
           allow_control_characters: true
         )
 
-        # Verify the native parser didn't produce invalid UTF-8 (like unpaired surrogates)
-        # by ensuring it can safely dump its own output.
-        JSON.generate(parsed)
+        # If the user wants objects, return them immediately
+        return parsed if return_objects
 
-        parsed
+        # Otherwise, generate the string once. This acts as both the
+        # UTF-8 verification step AND the final string output.
+        return JSON.generate(parsed)
       rescue JSON::ParserError, JSON::GeneratorError
         parser = Parser.new(json_string)
-        parser.parse
+        repaired_json = parser.parse
       end
 
       # Avoids returning `null` for empty results, returns the object directly

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: json_mend
 version: !ruby/object:Gem::Version
-  version: 0.3.4
+  version: 0.3.6
 platform: ruby
 authors:
 - Oleksii Vasyliev
@@ -77,14 +77,14 @@ required_ruby_version: !ruby/object:Gem::Requirement
   requirements:
   - - ">="
     - !ruby/object:Gem::Version
-      version: 3.2.0
+      version: 3.3.0
 required_rubygems_version: !ruby/object:Gem::Requirement
   requirements:
   - - ">="
     - !ruby/object:Gem::Version
       version: '0'
 requirements: []
-rubygems_version: 4.0.6
+rubygems_version: 4.0.16
 specification_version: 4
 summary: Repair broken JSON
 test_files: []