json-repair 0.5.0 → 0.7.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 83ede681caac8ebbc294902937c8af11ebc90fe066beed41e15013782fbd0b96
-  data.tar.gz: da309d72a1c564f05dd405c02115985572c38c8b2c28aad85f778b94cc2a68e9
+  metadata.gz: fdf2958528b936eba0faf6c724a20d10a3a3e0b95329bc1866740c99432f6fe3
+  data.tar.gz: 33fa97d2c7689ea594723e5d0d412646cf57a2ebdbf79f3b62947d4928963f32
 SHA512:
-  metadata.gz: da63df705106d20701700dd55712ed06676cdb4e6615f36d915cae12f3bdf7ac085ea919a553c77e8e0876663e5ad7d0067b46b00fddd3a7cca837eb38e54fc8
-  data.tar.gz: d4471c07b5a3a39505ec890b8afc359ca12fa3042c953143675af2d88bd8a4f56fe6225f00e0d4665eaf66465862f2b273929a2bb7fd54a81f79f1abf39ecfa1
+  metadata.gz: 587323c57caac3e53af1da24cfeee47df18ff738516d8fd2862d37a547c6e864cf5653c8b43427ec58aa47be3f2e958c991208b36de098c5ffc9503f2c64a0a2
+  data.tar.gz: f5380cfdca6a4f60833ab57fc8465715b546415fa4f6ed5781b6bb24653ce324d2b6d928f9a5ff87b4deead145391303c1e805bf0fbff238b9ef1c7024a1f4aa

data/.rubocop.yml

@@ -19,6 +19,7 @@ Metrics/AbcSize:
 Metrics/MethodLength:
   Exclude:
     - lib/json/repairer.rb
+    - benchmark/**/*
 
 Metrics/CyclomaticComplexity:
   Exclude:

data/CHANGELOG.md

@@ -1,5 +1,41 @@
 # Changes
 
+### 2026-05-12 (0.7.0)
+
+* `JSON.repair` now always returns canonical JSON via
+  `JSON.generate`. When the input is already valid JSON, stdlib
+  `JSON.parse` handles it directly; when it isn't, the repairer
+  produces an intermediate string that's then re-parsed and serialized
+  the same way. Both paths converge on the same output for any given
+  input, so `JSON.repair(json)` and
+  `JSON.repair(json, skip_json_loads: true)` agree on result and only
+  differ in how they got there.
+* **Breaking:** outputs are now canonical instead of preserving the
+  input's exact formatting. Whitespace is collapsed
+  (`'{"a": 1}'` → `'{"a":1}'`), numbers are normalized
+  (`2300e3` → `2300000.0`, `-0` → `0`), `\uXXXX` escapes are decoded
+  to their literal characters, `\/` is unescaped to `/`, and objects
+  with duplicate keys are collapsed to the last-write-wins form
+  (`{"a":1,"a":2}` → `{"a":2}`). Callers that need a parsed Ruby
+  value can opt out of the final `JSON.generate` step with
+  `return_objects: true`.
+* `skip_json_loads:` keyword argument added (default `false`,
+  mirroring Python's
+  [`json_repair`](https://github.com/mangiucugna/json_repair)
+  option). Passing `true` skips the stdlib `JSON.parse` fast attempt
+  and routes the input through the repairer first; the final output
+  is identical, so the option is purely a performance knob for
+  callers who know their input will need repair.
+
+### 2026-05-12 (0.6.0)
+
+* `JSON.repair` accepts a `return_objects:` keyword argument. Pass
+  `return_objects: true` to receive the parsed Ruby value (Hash, Array,
+  or scalar) instead of a serialized JSON string. Default is `false`,
+  preserving the existing return-a-string contract. Mirrors Python's
+  `return_objects` option on
+  [`json_repair`](https://github.com/mangiucugna/json_repair).
+
 ### 2026-05-12 (0.5.0)
 
 * `JSON::JSONRepairError#position` returns the input index at which the

data/CLAUDE.md

@@ -5,13 +5,13 @@ This file provides guidance to Claude Code (claude.ai/code) when working with co
 ## Commands
 
 - `bin/setup` — install dependencies via Bundler.
-- `bundle exec rake` — default task; runs both RSpec and RuboCop.
+- `bundle exec rake` — default task; runs RSpec, RuboCop, RBS validate, and Steep.
 - `bundle exec rspec` — run the test suite.
 - `bundle exec rspec spec/json_spec.rb:42` — run a single example by line number; nearly all behavioral specs live in `spec/json_spec.rb`.
 - `bundle exec rubocop` — lint. Project-specific exclusions in `.rubocop.yml` deliberately disable several `Metrics/*` cops for `lib/json/repairer.rb` and `lib/json/repair/string_utils.rb` because the parser is long by design — don't try to "fix" it by chopping methods up.
 - `bin/console` — IRB with the gem preloaded.
 - `bundle exec rake install` / `bundle exec rake release` — local install / publish to rubygems.org.
-- Type checking: `Steepfile` checks `lib/` against `sig/`. Run `bundle exec steep check` if/when steep is installed (not in `Gemfile` by default).
+- Type checking: `Steepfile` checks `lib/` against `sig/`. `bundle exec steep check` (typecheck) and `bundle exec rbs validate` (sig syntax) both run in CI and as part of the default rake task. `steep` and `rbs` are dev dependencies in the `Gemfile`.
 
 Ruby `>= 3.0.0` is required (per gemspec). CI runs against Ruby 3.3.1.
 

data/README.md

@@ -26,11 +26,33 @@ require 'json/repair'
 # Example of repairing a JSON string
 broken_json = '{name: Alice, "age": 25,}'
 repaired_json = JSON.repair(broken_json)
-puts repaired_json  # Outputs: {"name": "Alice", "age": 25}
+puts repaired_json  # Outputs: {"name":"Alice","age":25}
 ```
 
 The `repair` method takes a string containing JSON data and returns a corrected version of this string, ensuring it is valid JSON.
 
+Pass `return_objects: true` to get the parsed Ruby value (Hash, Array, or scalar) instead of a string:
+
+```ruby
+JSON.repair('{a: 1, b: [2, 3,]}', return_objects: true)
+# => {"a" => 1, "b" => [2, 3]}
+```
+
+### Canonical output
+
+`JSON.repair` returns canonical JSON via `JSON.generate`. When the input is already valid, stdlib `JSON.parse` handles it; otherwise the repairer fixes it up and the result is re-serialized the same way. Either way, the output is the canonical form of the parsed value — whitespace is collapsed, numbers are normalized, `\uXXXX` escapes are decoded to literal characters, and objects with duplicate keys collapse to last-write-wins.
+
+```ruby
+JSON.repair('{"a": 1}')          # => '{"a":1}'
+JSON.repair('{a:1}')             # => '{"a":1}'
+JSON.repair('2300e3')            # => '2300000.0'
+JSON.repair('{"a":1,"a":2}')     # => '{"a":2}'
+```
+
+If you need the parsed Ruby value instead of a string, pass `return_objects: true` (covered above).
+
+`skip_json_loads: true` skips the stdlib `JSON.parse` attempt and routes the input straight through the repairer. The output is the same; the option is purely a performance knob for callers who know their input will need repair.
+
 ## Command line
 
 The gem ships a `json-repair` executable. It reads from stdin or a file and writes to stdout, `--output FILE`, or back over the input file with `--overwrite`.
@@ -50,6 +72,8 @@ Run `json-repair --help` for the full list of options.
 
 After checking out the repo, run `bin/setup` to install dependencies. Then, run `rake spec` to run the tests. You can also run `bin/console` for an interactive prompt that will allow you to experiment.
 
+Run `bundle exec rake bench` for a `benchmark-ips` regression baseline across four canned scenarios (valid mixed JSON, broken LLM-style output, a large array, deeply nested objects). The harness lives under `benchmark/` and is not shipped in the published gem.
+
 To install this gem onto your local machine, run `bundle exec rake install`. To release a new version, update the version number in `version.rb`, and then run `bundle exec rake release`, which will create a git tag for the version, push git commits and the created tag, and push the `.gem` file to [rubygems.org](https://rubygems.org).
 
 ## Contributing

data/Rakefile

@@ -9,4 +9,19 @@ require 'rubocop/rake_task'
 
 RuboCop::RakeTask.new
 
-task default: %i[spec rubocop]
+desc 'Validate RBS signatures in sig/'
+task :rbs do
+  sh 'bundle exec rbs validate'
+end
+
+desc 'Run Steep type check against sig/'
+task :steep do
+  sh 'bundle exec steep check'
+end
+
+desc 'Run benchmark/run.rb (regression baseline for JSON.repair)'
+task :bench do
+  ruby '-Ilib', 'benchmark/run.rb'
+end
+
+task default: %i[spec rubocop rbs steep]

data/Steepfile

@@ -3,4 +3,9 @@
 target :lib do
   signature 'sig'
   check 'lib'
+
+  library 'optparse'
+  library 'tempfile'
+  library 'fileutils'
+  library 'json'
 end

data/lib/json/repair/cli.rb

@@ -34,7 +34,8 @@ module JSON
 
       def run(argv)
         positional = catch(:halt) { parser.parse(argv) }
-        return @halt if @halt
+        halt = @halt
+        return halt if halt
 
         input_path = positional.first
         return 1 unless validate(positional, input_path)
@@ -61,7 +62,7 @@ module JSON
       end
 
       def read_input(input_path)
-        raw = input_path ? File.read(input_path) : @stdin.read
+        raw = (input_path ? File.read(input_path) : @stdin.read).to_s
         raw.force_encoding(Encoding::UTF_8)
         raise JSON::JSONRepairError, 'input is not valid UTF-8' unless raw.valid_encoding?
 
@@ -69,13 +70,12 @@ module JSON
       end
 
       def write_output(repaired, input_path)
-        if @overwrite
+        if @overwrite && input_path
           replace_in_place(input_path, repaired)
-        elsif @output_path
-          File.write(@output_path, repaired)
+        elsif (output_path = @output_path)
+          File.write(output_path, repaired)
         else
-          @stdout.write(repaired)
-          @stdout.write("\n") unless repaired.end_with?("\n")
+          @stdout.write(repaired, "\n")
         end
       end
 

data/lib/json/repair/version.rb

@@ -2,6 +2,6 @@
 
 module JSON
   module Repair
-    VERSION = '0.5.0'
+    VERSION = '0.7.0'
   end
 end

data/lib/json/repair.rb

@@ -1,5 +1,7 @@
 # frozen_string_literal: true
 
+require 'json'
+
 require_relative 'repair/version'
 require_relative 'repairer'
 
@@ -13,7 +15,20 @@ module JSON
     end
   end
 
-  def self.repair(json)
-    Repairer.new(json).repair
+  def self.repair(json, return_objects: false, skip_json_loads: false)
+    parsed = skip_json_loads ? repaired_parse(json) : tolerant_parse(json)
+    return_objects ? parsed : JSON.generate(parsed)
+  end
+
+  def self.tolerant_parse(json)
+    JSON.parse(json)
+  rescue JSON::ParserError
+    repaired_parse(json)
+  end
+  private_class_method :tolerant_parse
+
+  def self.repaired_parse(json)
+    JSON.parse(Repairer.new(json).repair)
   end
+  private_class_method :repaired_parse
 end

data/lib/json/repairer.rb

@@ -227,9 +227,19 @@ module JSON
           if processed_colon || truncated_text
             # repair missing object value
             @output << 'null'
+          # :nocov:
           else
+            # Unreachable through JSON.repair: if we got here, the colon-repair
+            # branch above ran, which required start_of_value? to be true. Every
+            # char that satisfies start_of_value? (see REGEX_START_OF_VALUE plus
+            # quote chars) is consumable by some parse_* method, so parse_value
+            # cannot return false in this state. Preserved for parity with the
+            # upstream JS parser; if a future change to REGEX_START_OF_VALUE or
+            # parse_unquoted_string invalidates that invariant, this branch
+            # becomes live and the :nocov: will hide it.
             throw_colon_expected
           end
+          # :nocov:
         end
       end
 
@@ -725,10 +735,10 @@ module JSON
         processed_value = parse_value
       end
 
-      unless processed_value
-        # repair: remove trailing comma
-        @output = strip_last_occurrence(@output, ',')
-      end
+      # repair: remove trailing comma
+      # (the `while processed_value` loop above only exits when processed_value
+      # is falsy, so the upstream JS `if (!processedValue)` guard is redundant)
+      @output = strip_last_occurrence(@output, ',')
 
       # repair: wrap the output inside array brackets
       @output = "[\n#{@output}\n]"

data/sig/json/repair/cli.rbs

@@ -1,16 +1,47 @@
 module JSON
   module Repair
     class CLI
-      # `::IO | ::StringIO` because `::StringIO` is not an `::IO` subclass; the
-      # specs inject `::StringIO` instances and any other duck-typed stream
-      # that implements `#read` / `#write` / `#puts` would work too.
+      # `::IO | ::StringIO` because `::StringIO` is not an `::IO` subclass
+      # and the specs inject `::StringIO` instances.
       type stream = ::IO | ::StringIO
 
+      # Marked `private_constant` in `lib/json/repair/cli.rb`. RBS has no
+      # `private_constant` syntax, so the declaration is unavoidably public
+      # in the signature; do not rely on it from outside this class.
+      OVERWRITE_DESC: ::String
+
+      @stdin: stream
+      @stdout: stream
+      @stderr: stream
+      @output_path: ::String?
+      @halt: ::Integer?
+      @overwrite: bool
+
       def self.call: (::Array[::String] argv, ?stdin: stream, ?stdout: stream, ?stderr: stream) -> ::Integer
 
       def initialize: (?stdin: stream, ?stdout: stream, ?stderr: stream) -> void
 
       def call: (::Array[::String] argv) -> ::Integer
+
+      private
+
+      def run: (::Array[::String] argv) -> ::Integer
+
+      def validate: (::Array[::String] positional, ::String? input_path) -> bool
+
+      def validation_error: (::Array[::String] positional, ::String? input_path) -> ::String?
+
+      def read_input: (::String? input_path) -> ::String
+
+      def write_output: (::String repaired, ::String? input_path) -> void
+
+      def replace_in_place: (::String input_path, ::String repaired) -> void
+
+      def parser: () -> ::OptionParser
+
+      def define_options: (::OptionParser opts) -> void
+
+      def halt_with: (::String message) -> void
     end
   end
 end

data/sig/json/repair.rbs

@@ -9,5 +9,13 @@ module JSON
     VERSION: ::String
   end
 
-  def self.repair: (::String json) -> ::String
+  def self.repair: (::String json, return_objects: false, ?skip_json_loads: bool) -> ::String
+                 | (::String json, return_objects: true, ?skip_json_loads: bool) -> untyped
+                 | (::String json, ?skip_json_loads: bool) -> ::String
+
+  private
+
+  def self.tolerant_parse: (::String json) -> untyped
+
+  def self.repaired_parse: (::String json) -> untyped
 end

data/sig/json/repairer.rbs

@@ -1,6 +1,15 @@
 module JSON
   class Repairer
-    @json: ::String
+    # `untyped` (not `::String`) because the parser idiomatically uses
+    # `@json[@index]` past EOF, relying on Ruby returning `nil` as a
+    # sentinel. Declaring `@json: ::String` makes steep infer `String?`
+    # from `String#[]` and rejects ~15 downstream call sites that pass
+    # the result to methods expecting `String`. Tightening would either
+    # require pervasive nil-extraction at every indexing site or a
+    # private accessor — both of which break parity with the upstream
+    # JS source (see CLAUDE.md and `.rubocop.yml` exceptions for
+    # `lib/json/repairer.rb`).
+    @json: untyped
 
     @index: Integer
 

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: json-repair
 version: !ruby/object:Gem::Version
-  version: 0.5.0
+  version: 0.7.0
 platform: ruby
 authors:
 - Aleksandr Zykov