json 3.0.2-java → 3.1.0-java

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 4be775fd0d4221af4703313be47b36df730b542f95063025472d82386bc6f949
-  data.tar.gz: aa1136bf72ebb55174ab30a3ddcaf908364a7e0cde743b952435a5ab59195d0a
+  metadata.gz: ee1372af8bb19939f969cebf3d301b38b9d03c40e48ed62d9d0eda90eb04692f
+  data.tar.gz: b380bcb62bfd92420834b2ce75f6fc1e298be512a15f4a5ac5839e967b51c23f
 SHA512:
-  metadata.gz: 3b1da37df8ae7d0a8a40e41caed988f2b9f33c8ef147e66220e811adbc26199482189055badf0023bc5c4b8f201f2954a30cfa355cd7629604485c0424869f06
-  data.tar.gz: 7e50467d636c6a30b701d4778650355e6856660add1de37eef2519865977107e715cf19d3bcceff288ee068b5993908535243b6b799f7bc37bd7f41b8c270d37
+  metadata.gz: e9788ac086e362d08433e0027ffc269ec4dd58075165b86b0a4fa4c2da7d9de8583d155bf79b0b4389eaab3b10ff92e3b162b808e3dd413b22b2e3473e67fdd4
+  data.tar.gz: 51800b6ce2d20e0916b1244f9d18ab10ba3727e0bb03634f723774280fab00f731b0860dd97952e4cdc81ab02fe613260a29afa6547ff2640434df116fe7f9cd

data/CHANGES.md

@@ -2,6 +2,15 @@
 
 ### Unreleased
 
+### 2026-10-13 (3.1.0)
+
+* Added `rfc8785` generator option to canonicalize the output. Replaces the `json-canonicalization` with numerous correctness fixes.
+* Use the `rb_len_t` type introduced in Ruby 4.1 to allow for proper 64bit support on LP64 platforms (Windows).
+* JRuby: parser errors now include the position (`line`, `column` and the message suffix) as well as `json_path`, matching the C extension.
+* Improved the accuracy of parser error's `line`, `column` and snipets in presence of multibyte characters.
+* Fix `JSON.load` to work correctly when provided a `Zlib::GzipReader` as IO source.
+* Reject non-ASCII BINARY strings during JSON generation. This was deprecated since a long time and scheduled to be removed for 3.0.
+
 ### 2026-09-09 (3.0.2)
 
 * Fix `JSON.load_file` and `JSON.load_file!` to load on Ruby 2.7.0 through 2.7.2, which cannot parse a leading parameter before `...`.

data/lib/json/common.rb

@@ -80,14 +80,6 @@ module JSON
     def generator=(generator) # :nodoc:
       old, $VERBOSE = $VERBOSE, nil
 
-      # The default proc used when the +sort_keys+ generation option is +true+.
-      # It returns a new hash with the entries sorted by their keys.
-      sort_keys_proc = ->(hash) { hash.sort.to_h }
-      if defined?(::Ractor) && Ractor.respond_to?(:shareable_lambda)
-        sort_keys_proc = Ractor.shareable_lambda(&sort_keys_proc)
-      end
-      generator::State.default_sort_keys_proc = sort_keys_proc
-
       @generator = generator
       if generator.const_defined?(:GeneratorMethods)
         generator_methods = generator::GeneratorMethods
@@ -146,7 +138,7 @@ module JSON
     # Is +nil+ when raised by JSON::ResumableParser.
     attr_reader :line
 
-    # Column number where the parser encountered an error.
+    # One-based column number where the parser encountered an error, counted in Unicode codepoints.
     # Is +nil+ when raised by JSON::ResumableParser.
     attr_reader :column
 
@@ -179,7 +171,10 @@ module JSON
         when Integer
           "[#{segment}]"
         when String, Symbol
-          if segment.match?(/\A[a-zA-Z\$\_][a-zA-Z\$\_0-9]*\z/)
+          if !segment.valid_encoding?
+            segment = segment.b.gsub(/["\\]/n, { '"' => '\\"', '\\' => '\\\\' }).force_encoding(Encoding::UTF_8)
+            %{["#{segment}"]}
+          elsif segment.match?(/\A[a-zA-Z\$\_][a-zA-Z\$\_0-9]*\z/)
             ".#{segment}"
           else
             segment = segment.to_s.gsub(/["\\]/, { '"' => '\\"', '\\' => '\\\\' })
@@ -232,7 +227,7 @@ module JSON
   Fragment = Struct.new(:json) do
     def initialize(json)
       unless string = String.try_convert(json)
-        raise TypeError, " no implicit conversion of #{json.class} into String"
+        raise TypeError, "no implicit conversion of #{json.class} into String"
       end
 
       super(string)
@@ -707,10 +702,10 @@ module JSON
     unless source.is_a?(String)
       if source.respond_to? :to_str
         source = source.to_str
+      elsif source.respond_to? :read
+        source = source.read
       elsif source.respond_to? :to_io
         source = source.to_io.read
-      elsif source.respond_to?(:read)
-        source = source.read
       end
     end
 
@@ -771,7 +766,7 @@ module JSON
     end
 
     opts = {
-      allow_nan: true,
+      allow_nan: !(kwargs && kwargs[:rfc8785]),
     }
     opts[:max_nesting] = _deprecated_limit if _deprecated_limit
     opts.merge!(kwargs) if kwargs
@@ -835,7 +830,12 @@ module JSON
     #
     #  puts MyApp::API_JSON_CODER.dump(Time.now.utc) # => "2025-01-21T08:41:44.286Z"
     #
-    def initialize(object_class: nil, array_class: nil, on_load: nil, **options, &as_json)
+    # The conversion callback can also be passed as +as_json:+. If both are given, the block takes precedence.
+    # Passing +false+ or +nil+ for +strict:+ raises ArgumentError.
+    #
+    def initialize(object_class: nil, array_class: nil, on_load: nil, strict: true, **options, &as_json)
+      raise ArgumentError, "JSON::Coder requires strict mode" unless strict
+
       if object_class || array_class
         on_load = ParserOptions.on_load(on_load, object_class, array_class)
       end
@@ -847,7 +847,7 @@ module JSON
       @state = State.new(
         **generator_options,
         strict: true,
-        as_json: as_json,
+        as_json: as_json || generator_options[:as_json],
       ).freeze
     end
 

data/lib/json/ext/generator/state.rb

@@ -54,7 +54,8 @@ module JSON
             strict: strict?,
             depth: depth,
             buffer_initial_length: buffer_initial_length,
-            sort_keys: sort_keys
+            sort_keys: sort_keys,
+            rfc8785: rfc8785?,
           }
 
           allow_duplicate_key = allow_duplicate_key?
@@ -62,11 +63,6 @@ module JSON
             result[:allow_duplicate_key] = allow_duplicate_key
           end
 
-          instance_variables.each do |iv|
-            iv = iv.to_s[1..-1]
-            result[iv.to_sym] = self[iv]
-          end
-
           result
         end
 

data/lib/json/ext.rb

@@ -28,17 +28,143 @@ module JSON
       end
     end
 
+    class << self
+      if defined?(::Ractor) && Ractor.respond_to?(:shareable_lambda)
+        def shareable_lambda(block) # :nodoc:
+          Ractor.shareable_lambda(&block)
+        end
+      else
+        def shareable_lambda(block) # :nodoc:
+          block
+        end
+      end
+
+      def valid_encoding?(str) # :nodoc:
+        str.ascii_only? || (str.encoding == Encoding::UTF_8 && str.valid_encoding?)
+      end
+
+      def coerce_key(key, as_json) # :nodoc:
+        as_json_called = false
+
+        unless String === key || Symbol === key
+          key = as_json.call(key, true)
+          as_json_called = true
+        end
+
+        if Symbol === key
+          key = key.to_s
+        end
+
+        unless String === key
+          raise GeneratorError.new("#{key.class} not allowed as object key in JSON", key)
+        end
+
+        unless valid_encoding?(key)
+          if as_json_called
+            raise GeneratorError.new("source sequence is illegal/malformed utf-8", key)
+          end
+
+          key = as_json.call(key, true)
+          unless String === key
+            raise GeneratorError.new("#{key.class} not allowed as object key in JSON", key)
+          end
+
+          unless Ext.valid_encoding?(key)
+            raise GeneratorError.new("source sequence is illegal/malformed utf-8", key)
+          end
+        end
+
+        key
+      end
+
+      def on_duplicate_key(hash, as_json)
+        set = {}
+        hash.each_key do |key|
+          key = coerce_key(key, as_json)
+          if set[key]
+            raise GeneratorError.new("detected duplicate key #{key.inspect}", key)
+          else
+            set[key] = true
+          end
+        end
+      end
+    end
+
     require 'json/ext/parser'
     Ext::Parser::Config = Ext::ParserConfig
     JSON.parser = Ext::Parser
 
-    if RUBY_ENGINE == 'truffleruby'
+    generator = if RUBY_ENGINE == 'truffleruby'
       require 'json/truffle_ruby/generator'
-      JSON.generator = JSON::TruffleRuby::Generator
+      JSON::TruffleRuby::Generator
     else
       require 'json/ext/generator'
-      JSON.generator = Generator
+      Generator
     end
+
+    # The default proc used when the +sort_keys+ generation option is +true+.
+    # It returns a new hash with the entries sorted by their keys.
+    generator::State.default_sort_keys_proc = shareable_lambda(->(hash) {
+      hash.sort.to_h
+    })
+
+    # Directly lifted from Gregg Kellogg's json-canonicalization
+    generator::State.rfc8785_number_formatter_proc = shareable_lambda(->(num) {
+      if Integer === num
+        if num > Float::MAX || num < -Float::MAX
+          raise GeneratorError.new("Integer out of range for RFC 8785", num)
+        end
+        num = num.to_f
+      end
+      if num.zero?
+        "0"
+      else
+        if num < 0
+          num, sign = -num, '-'
+        end
+        native_rep = num.to_s
+        decimal, exponential = native_rep.split('e')
+        exp_val = exponential.to_i
+        exponential = exp_val > 0 ? ('+' + exp_val.to_s) : exp_val.to_s
+
+        integral, fractional = decimal.split('.')
+        fractional = fractional.sub(/0+$/, '')  # Remove trailing zeros
+
+        if exp_val > 0 && exp_val < 21
+          while exp_val > 0
+            integral += fractional.to_s[0] || '0'
+            fractional = fractional.to_s[1..-1]
+            exp_val -= 1
+          end
+          exponential = nil
+        elsif exp_val == 0
+          exponential = nil
+        elsif exp_val < 0 && exp_val > -7
+          # Small numbers are shown as 0.etc with e-6 as lower limit
+          fractional, integral, exponential = integral + fractional.to_s, '0', nil
+          fractional = ("0" * (-exp_val - 1)) + fractional
+        end
+
+        fractional = nil if fractional.to_s.empty?
+        sign.to_s + integral + (fractional ? ".#{fractional}" : '') + (exponential ? "e#{exponential}" : '')
+      end
+    })
+
+    generator::State.rfc8785_sort_keys_proc = shareable_lambda(->(hash, as_json) {
+      converted = if as_json
+        hash.transform_keys { |k| Ext.coerce_key(k, as_json) }
+      else
+        hash.transform_keys(&:to_s)
+      end
+
+      if converted.size < hash.size
+        Ext.on_duplicate_key(hash, as_json)
+      end
+
+      converted.sort_by { |k,| k.encode(Encoding::UTF_16) }.to_h
+    })
+
+    JSON.generator = generator
   end
 
   if defined?(ResumableParser) # Not yet available on JRuby

data/lib/json/truffle_ruby/generator.rb

@@ -112,6 +112,8 @@ module JSON
       # while generating a JSON text from a Ruby data structure.
       class State
         singleton_class.attr_accessor :default_sort_keys_proc # :nodoc:
+        singleton_class.attr_accessor :rfc8785_number_formatter_proc # :nodoc:
+        singleton_class.attr_accessor :rfc8785_sort_keys_proc # :nodoc:
 
         def self.generate(obj, opts = nil, io = nil)
           new(opts).generate(obj, io)
@@ -125,7 +127,7 @@ module JSON
           if opts
             case
             when self === opts
-              return opts
+              return opts.validate_rfc8785
             when opts.respond_to?(:to_hash)
               return new(opts.to_hash)
             when opts.respond_to?(:to_h)
@@ -168,6 +170,7 @@ module JSON
           @max_nesting           = 100
           @sort_keys             = false
           @allow_duplicate_key   = false
+          @rfc8785               = false
           _configure(**opts) if opts
         end
 
@@ -189,7 +192,16 @@ module JSON
         attr_accessor :array_nl
 
         # This proc converts unsupported types into native JSON types.
-        attr_accessor :as_json
+        attr_reader :as_json
+
+        def as_json=(value)
+          raise FrozenError, "can't modify frozen #{self.class}: #{inspect}" if frozen?
+          value = value.to_proc if !(Proc === value) && value.respond_to?(:to_proc)
+          unless Proc === value
+            raise TypeError, "as_json must be a Proc"
+          end
+          @as_json = value
+        end
 
         # This integer returns the maximum level of data structure nesting in
         # the generated JSON, max_nesting = 0 if no maximum is checked.
@@ -209,6 +221,9 @@ module JSON
         # desired order.
         attr_reader :sort_keys
 
+        # Controls whether RFC8785 (canonicalization) mode is enabled. Otherwise returns false.
+        attr_writer :rfc8785
+
         def sort_keys=(value) # :nodoc:
           type_error = false
           @sort_keys = case value
@@ -288,6 +303,27 @@ module JSON
           @strict
         end
 
+        # Returns true, if RFC8785 (canonicalization) mode is enabled. Otherwise returns false.
+        def rfc8785?
+          @rfc8785
+        end
+
+        def validate_rfc8785 # :nodoc:
+          if @rfc8785
+            option = if !@indent.empty? then :indent
+                     elsif !@space.empty? then :space
+                     elsif !@space_before.empty? then :space_before
+                     elsif !@object_nl.empty? then :object_nl
+                     elsif !@array_nl.empty? then :array_nl
+                     elsif @ascii_only then :ascii_only
+                     elsif @script_safe then :script_safe
+                     elsif @allow_nan then :allow_nan
+                     end
+            raise ArgumentError, "#{option} cannot be used with rfc8785" if option
+          end
+          self
+        end
+
         # Configure this State instance with the Hash _opts_, and return
         # itself.
         def configure(options)
@@ -302,10 +338,13 @@ module JSON
         end
         alias merge configure
 
+        # Defaults are the current values so that #configure only writes what it was given.
         private def _configure(
-          indent: '', space: '', space_before: '', object_nl: '', array_nl: '', allow_nan: false,
-          as_json: false, ascii_only: false, sort_keys: false, depth: 0, buffer_initial_length: 1024,
-          allow_duplicate_key: false, script_safe: false, strict: false, max_nesting: 100
+          indent: @indent, space: @space, space_before: @space_before, object_nl: @object_nl,
+          array_nl: @array_nl, allow_nan: @allow_nan, as_json: @as_json, ascii_only: @ascii_only,
+          sort_keys: @sort_keys, depth: @depth, buffer_initial_length: @buffer_initial_length,
+          allow_duplicate_key: @allow_duplicate_key, script_safe: @script_safe, strict: @strict,
+          max_nesting: @max_nesting, rfc8785: @rfc8785
         )
           if depth.negative?
             raise ArgumentError, "depth must be >= 0 (got #{depth})"
@@ -321,7 +360,11 @@ module JSON
           @object_nl = object_nl || ''
           @array_nl = array_nl || ''
           @allow_nan = allow_nan || false
-          @as_json = as_json || false
+          if as_json
+            self.as_json = as_json
+          else
+            @as_json = false
+          end
           @ascii_only = ascii_only || false
           self.sort_keys = sort_keys
           @depth = depth
@@ -330,7 +373,9 @@ module JSON
           @script_safe = script_safe
           @strict = strict
           @max_nesting = max_nesting || 0
-          self
+          self.rfc8785 = rfc8785
+
+          validate_rfc8785
         end
 
         def allow_duplicate_key? # :nodoc:
@@ -340,13 +385,24 @@ module JSON
         # Returns the configuration instance variables as a hash, that can be
         # passed to the configure method.
         def to_h
-          result = {}
-          instance_variables.each do |iv|
-            key = iv.to_s[1..-1]
-            result[key.to_sym] = instance_variable_get(iv)
-          end
-
-          result
+          {
+            indent: @indent,
+            space: @space,
+            space_before: @space_before,
+            object_nl: @object_nl,
+            array_nl: @array_nl,
+            as_json: @as_json,
+            allow_nan: @allow_nan,
+            ascii_only: @ascii_only,
+            max_nesting: @max_nesting,
+            script_safe: @script_safe,
+            strict: @strict,
+            depth: @depth,
+            buffer_initial_length: @buffer_initial_length,
+            sort_keys: @sort_keys,
+            allow_duplicate_key: @allow_duplicate_key,
+            rfc8785: @rfc8785,
+          }
         end
 
         alias to_hash to_h
@@ -359,16 +415,14 @@ module JSON
           return dup.generate(obj, anIO) if frozen?
 
           depth = @depth
+          validate_rfc8785
           if @indent.empty? and @space.empty? and @space_before.empty? and @object_nl.empty? and @array_nl.empty? and
               !@ascii_only and !@script_safe and @max_nesting == 0 and (!@strict || Symbol === obj) and !@sort_keys
             result = generate_json(obj, ''.dup)
           else
-            if @sort_keys
-              obj = @sort_keys.call(obj)
-            end
-
             result = obj.to_json(self)
           end
+
           JSON::TruffleRuby::Generator.valid_utf8?(result) or raise GeneratorError.new(
             "source sequence #{result.inspect} is illegal/malformed utf-8",
             obj
@@ -390,6 +444,11 @@ module JSON
             buf << '{'
             first = true
             key_type = nil
+            if @rfc8785
+              obj = State.rfc8785_sort_keys_proc.call(obj, @strict && @as_json)
+            elsif @sort_keys
+              obj = @sort_keys.call(obj)
+            end
             obj.each_pair do |k,v|
               if first
                 key_type = k.class
@@ -431,7 +490,7 @@ module JSON
               buf << obj.to_json(self)
             end
           when Integer
-            buf << obj.to_s
+            buf << (@rfc8785 ? State.rfc8785_number_formatter_proc.call(obj) : obj.to_s)
           when Symbol
             if @strict
               fast_serialize_string(obj.name, buf)
@@ -500,10 +559,12 @@ module JSON
           def to_json(state = nil, *)
             state = State.from_state(state)
             depth = state.depth
-            state.check_max_nesting
-            json_transform(state)
-          ensure
-            state.depth = depth
+            begin
+              state.check_max_nesting
+              json_transform(state)
+            ensure
+              state.depth = depth
+            end
           end
 
           private
@@ -521,13 +582,21 @@ module JSON
             first = true
             key_type = nil
             indent = !state.object_nl.empty?
-            each { |key, value|
+            hash = self
+
+            if state.rfc8785?
+              hash = State.rfc8785_sort_keys_proc.call(hash, state.strict? && state.as_json)
+            elsif state.sort_keys
+              hash = state.sort_keys.call(hash)
+            end
+
+            hash.each { |key, value|
               if first
                 key_type = key.class
               else
                 if key_type && !state.allow_duplicate_key? && key_type != key.class
                   key_type = nil # stop checking
-                  JSON.send(:on_mixed_keys_hash, self)
+                  JSON.send(:on_mixed_keys_hash, hash)
                 end
                 result << delim
               end
@@ -592,10 +661,12 @@ module JSON
           def to_json(state = nil, *)
             state = State.from_state(state)
             depth = state.depth
-            state.check_max_nesting
-            json_transform(state)
-          ensure
-            state.depth = depth
+            begin
+              state.check_max_nesting
+              json_transform(state)
+            ensure
+              state.depth = depth
+            end
           end
 
           private
@@ -648,13 +719,20 @@ module JSON
 
         module Integer
           # Returns a JSON string representation for this Integer number.
-          def to_json(*) to_s end
+          def to_json(state = nil, *)
+            if state && State.from_state(state).rfc8785?
+              State.rfc8785_number_formatter_proc.call(self)
+            else
+              to_s
+            end
+          end
         end
 
         module Float
           # Returns a JSON string representation for this Float number.
           def to_json(state = nil, *args)
             state = State.from_state(state)
+
             if infinite? || nan?
               if state.allow_nan?
                 to_s
@@ -676,6 +754,8 @@ module JSON
               else
                 raise GeneratorError.new("#{self} not allowed in JSON", self)
               end
+            elsif state.rfc8785?
+              State.rfc8785_number_formatter_proc.call(self)
             else
               to_s
             end

data/lib/json/version.rb

@@ -1,5 +1,5 @@
 # frozen_string_literal: true
 
 module JSON
-  VERSION = '3.0.2'
+  VERSION = '3.1.0'
 end

data/lib/json.rb

@@ -414,6 +414,14 @@ require 'json/common'
 #   When +true+, keys are sorted lexicographically. When a \Proc, it receives
 #   the entire \Hash and must return a \Hash with its pairs in the desired
 #   order, allowing for arbitrary sort orders.
+# - Option +rfc8785+ (boolean) controls whether the generated JSON will be canonicalized
+#   as defined in RFC8785.
+#   Integers are converted to Float before formatting and may be rounded.
+#   Integers outside the finite Float range raise JSON::GeneratorError.
+#   Nonempty +indent+, +space+, +space_before+, +object_nl+, or +array_nl+,
+#   and enabled +ascii_only+, +script_safe+, or +allow_nan+ raise ArgumentError.
+#   JSON.dump defaults +allow_nan+ to +false+ when +rfc8785+ is enabled.
+#   JSON::Fragment contents are inserted as is; the caller must ensure they are canonical.
 #
 # In this example, +obj+ is used first to generate the shortest
 # \JSON data (no whitespace), then again with all formatting options

metadata

@@ -1,13 +1,13 @@
 --- !ruby/object:Gem::Specification
 name: json
 version: !ruby/object:Gem::Version
-  version: 3.0.2
+  version: 3.1.0
 platform: java
 authors:
 - Daniel Luz
 bindir: bin
 cert_chain: []
-date: 2026-09-09 00:00:00.000000000 Z
+date: 2026-10-13 00:00:00.000000000 Z
 dependencies: []
 description: A JSON implementation as a JRuby extension.
 email: dev+ruby@mernen.com