json 2.10.2 → 2.15.1

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: b909ed90787afa31835c9926529808ca3c6b161755e1f6126b7fe203231fa0c2
-  data.tar.gz: efd6a873e98f9e4b2b8c078210f141dfde5f3773884bd5851f228c184ff94ff7
+  metadata.gz: 981ed5b9d1d88ec48021ed8ab7757ab595dbb2bb5a16e9b3595dc4aae5c1f193
+  data.tar.gz: 1fc86464fa02dead5539e47dc37c3dcb3d068707ee8d1e32c574893e1bb702de
 SHA512:
-  metadata.gz: bff39363d5d7fa2f209d5bf97a5728738058bc83600b732d95487e263a5a3901b1e18d5c42d7d2be019ac55ef069a50cedc1d9a8d22b11e6061f2eef747dc94d
-  data.tar.gz: 693e480f8f2489e26c43eef56ad8977ab3f00fada03318a48b88ac55e2b82b15ffae92fd05dd47d86686fae39169b079f59a68fffc6efaf44ee3503a45060e30
+  metadata.gz: c06d2e27a811c7b2368ed364f1fd72c0423922094ed29ec0de07a75db7aa53d3f2662aa80220f6d094fe39400603d2919b97308010fcc5c0efab86307089848b
+  data.tar.gz: c032759c67a721b378518dd3c555e4e1c3f1ca34ff00f1bb21a2f136b626399efdd9c5a701aa018c3c9d5da050d0c110ee98e29853d4949db125f524fad1fdae

data/CHANGES.md

@@ -1,10 +1,112 @@
 # Changes
 
+### Unreleased
+
+### 2025-10-07 (2.15.1)
+
+* Fix incorrect escaping in the JRuby extension when encoding shared strings.
+
+### 2025-09-22 (2.15.0)
+
+* `JSON::Coder` callback now receive a second argument to convey whether the object is a hash key.
+* Tuned the floating point number generator to not use scientific notation as aggressively.
+
+### 2025-09-18 (2.14.1)
+
+* Fix `IndexOutOfBoundsException` in the JRuby extension when encoding shared strings.
+
+### 2025-09-18 (2.14.0)
+
+* Add new `allow_duplicate_key` generator options. By default a warning is now emitted when a duplicated key is encountered.
+  In `json 3.0` an error will be raised.
+  ```ruby
+  >> Warning[:deprecated] = true
+  >> puts JSON.generate({ foo: 1, "foo" => 2 })
+  (irb):2: warning: detected duplicate key "foo" in {foo: 1, "foo" => 2}.
+  This will raise an error in json 3.0 unless enabled via `allow_duplicate_key: true`
+  {"foo":1,"foo":2}
+  >> JSON.generate({ foo: 1, "foo" => 2 }, allow_duplicate_key: false)
+  detected duplicate key "foo" in {foo: 1, "foo" => 2} (JSON::GeneratorError)
+  ```
+* Fix `JSON.generate` `strict: true` mode to also restrict hash keys.
+* Fix `JSON::Coder` to also invoke block for hash keys that aren't strings nor symbols.
+* Fix `JSON.unsafe_load` usage with proc
+* Fix the parser to more consistently reject invalid UTF-16 surogate pairs. 
+* Stop defining `String.json_create`, `String#to_json_raw`, `String#to_json_raw_object` when `json/add` isn't loaded.
+
+### 2025-07-28 (2.13.2)
+
+* Improve duplicate key warning and errors to include the key name and point to the right caller.
+
+### 2025-07-24 (2.13.1)
+
+* Fix support for older compilers without `__builtin_cpu_supports`.
+
+### 2025-07-17 (2.13.0)
+
+* Add new `allow_duplicate_key` parsing options. By default a warning is now emitted when a duplicated key is encountered.
+  In `json 3.0` an error will be raised.
+* Optimize parsing further using SIMD to scan strings.
+
+### 2025-05-23 (2.12.2)
+
+* Fix compiler optimization level.
+
+### 2025-05-23 (2.12.1)
+
+* Fix a potential crash in large negative floating point number generation.
+* Fix for JSON.pretty_generate to use passed state object's generate instead of state class as the required parameters aren't available.
+
+### 2025-05-12 (2.12.0)
+
+* Improve floating point generation to not use scientific notation as much.
+* Include line and column in parser errors. Both in the message and as exception attributes.
+* Handle non-string hash keys with broken `to_s` implementations.
+* `JSON.generate` now uses SSE2 (x86) or NEON (arm64) instructions when available to escape strings.
+
+### 2025-04-25 (2.11.3)
+
+* Fix a regression in `JSON.pretty_generate` that could cause indentation to be off once some `#to_json` has been called.
+
+### 2025-04-24 (2.11.2)
+
+* Add back `JSON::PRETTY_STATE_PROTOTYPE`. This constant was private API but is used by popular gems like `multi_json`.
+  It now emits a deprecation warning.
+
+### 2025-04-24 (2.11.1)
+
+* Add back `JSON.restore`, `JSON.unparse`, `JSON.fast_unparse` and `JSON.pretty_unparse`.
+  These were deprecated 16 years ago, but never emitted warnings, only undocumented, so are
+  still used by a few gems.
+
+### 2025-04-24 (2.11.0)
+
+* Optimize Integer generation to be ~1.8x faster.
+* Optimize Float generation to be ~10x faster.
+* Fix `JSON.load` proc argument to substitute the parsed object with the return value.
+  This better match `Marshal.load` behavior.
+* Deprecate `JSON.fast_generate` (it's not any faster, so pointless).
+* Deprecate `JSON.load_default_options`.
+* Deprecate `JSON.unsafe_load_default_options`.
+* Deprecate `JSON.dump_default_options`.
+* Deprecate `Kernel#j`
+* Deprecate `Kernel#jj`
+* Remove outdated `JSON.iconv`.
+* Remove `Class#json_creatable?` monkey patch.
+* Remove deprecated `JSON.restore` method.
+* Remove deprecated `JSON.unparse` method.
+* Remove deprecated `JSON.fast_unparse` method.
+* Remove deprecated `JSON.pretty_unparse` method.
+* Remove deprecated `JSON::UnparserError` constant.
+* Remove outdated `JSON::MissingUnicodeSupport` constant.
+
 ### 2025-03-12 (2.10.2)
 
 * Fix a potential crash in the C extension parser.
-* Raise a ParserError on all incomplete unicode escape sequence. This was the behavior until `2.10.0` unadvertently changed it.
+* Raise a ParserError on all incomplete unicode escape sequence. This was the behavior until `2.10.0` inadvertently changed it.
 * Ensure document snippets that are included in parser errors don't include truncated multibyte characters.
+* Ensure parser error snippets are valid UTF-8.
+* Fix `JSON::GeneratorError#detailed_message` on Ruby < 3.2
 
 ### 2025-02-10 (2.10.1)
 
@@ -32,7 +134,7 @@
 
 ### 2024-11-14 (2.8.2)
 
-* `JSON.load_file` explictly read the file as UTF-8.
+* `JSON.load_file` explicitly read the file as UTF-8.
 
 ### 2024-11-06 (2.8.1)
 
@@ -40,7 +142,7 @@
 
 ### 2024-11-06 (2.8.0)
 
-* Emit a deprecation warning when `JSON.load` create custom types without the `create_additions` option being explictly enabled.
+* Emit a deprecation warning when `JSON.load` create custom types without the `create_additions` option being explicitly enabled.
   * Prefer to use `JSON.unsafe_load(string)` or `JSON.load(string, create_additions: true)`.
 * Emit a deprecation warning when serializing valid UTF-8 strings encoded in `ASCII_8BIT` aka `BINARY`.
 * Bump required Ruby version to 2.7.
@@ -48,7 +150,7 @@
   pre-existing support for comments, make it suitable to parse `jsonc` documents.
 * Many performance improvements to `JSON.parse` and `JSON.load`, up to `1.7x` faster on real world documents.
 * Some minor performance improvements to `JSON.dump` and `JSON.generate`.
-* `JSON.pretty_generate` no longer include newline inside empty object and arrays. 
+* `JSON.pretty_generate` no longer includes newlines inside empty object and arrays.
 
 ### 2024-11-04 (2.7.6)
 
@@ -65,13 +167,13 @@
 * Workaround a bug in 3.4.8 and older https://github.com/rubygems/rubygems/pull/6490.
   This bug would cause some gems with native extension to fail during compilation.
 * Workaround different versions of `json` and `json_pure` being loaded (not officially supported).
-* Make `json_pure` Ractor compatible. 
+* Make `json_pure` Ractor compatible.
 
 ### 2024-10-24 (2.7.3)
 
 * Numerous performance optimizations in `JSON.generate` and `JSON.dump` (up to 2 times faster).
-* Limit the size of ParserError exception messages, only include up to 32 bytes of the unparseable source.
-* Fix json-pure's `Object#to_json` to accept non state arguments 
+* Limit the size of ParserError exception messages, only include up to 32 bytes of the unparsable source.
+* Fix json-pure's `Object#to_json` to accept non-state arguments.
 * Fix multiline comment support in `json-pure`.
 * Fix `JSON.parse` to no longer mutate the argument encoding when passed an ASCII-8BIT string.
 * Fix `String#to_json` to raise on invalid encoding in `json-pure`.
@@ -216,6 +318,7 @@
 ## 2015-09-11 (2.0.0)
   * Now complies to newest JSON RFC 7159.
   * Implements compatibility to ruby 2.4 integer unification.
+  * Removed support for `quirks_mode` option.
   * Drops support for old rubies whose life has ended, that is rubies < 2.0.
     Also see https://www.ruby-lang.org/en/news/2014/07/01/eol-for-1-8-7-and-1-9-2/
   * There were still some mentions of dual GPL licensing in the source, but JSON

data/README.md

@@ -97,7 +97,7 @@ Instead it is recommended to use the newer `JSON::Coder` API:
 
 ```ruby
 module MyApp
-  API_JSON_CODER = JSON::Coder.new do |object|
+  API_JSON_CODER = JSON::Coder.new do |object, is_object_key|
     case object
     when Time
       object.iso8601(3)
@@ -113,6 +113,8 @@ puts MyApp::API_JSON_CODER.dump(Time.now.utc) # => "2025-01-21T08:41:44.286Z"
 The provided block is called for all objects that don't have a native JSON equivalent, and
 must return a Ruby object that has a native JSON equivalent.
 
+It is also called for objects that do have a JSON equivalent, but are used as Hash keys, for instance `{ 1 => 2}`.
+
 ## Combining JSON fragments
 
 To combine JSON fragments into a bigger JSON document, you can use `JSON::Fragment`:
@@ -233,6 +235,19 @@ the `pp` library's `pp` methods.
 
 ## Development
 
+### Prerequisites
+
+1. Clone the repository
+2. Install dependencies with `bundle install`
+
+### Testing
+
+The full test suite can be run with:
+
+```bash
+bundle exec rake test
+```
+
 ### Release
 
 Update the `lib/json/version.rb` file.

data/ext/json/ext/fbuffer/fbuffer.h

@@ -3,6 +3,7 @@
 
 #include "ruby.h"
 #include "ruby/encoding.h"
+#include "../vendor/jeaiii-ltoa.h"
 
 /* shims */
 /* This is the fallback definition from Ruby 3.4 */
@@ -23,6 +24,14 @@ typedef unsigned char _Bool;
 #endif
 #endif
 
+#ifndef NOINLINE
+#if defined(__has_attribute) && __has_attribute(noinline)
+#define NOINLINE() __attribute__((noinline))
+#else
+#define NOINLINE()
+#endif
+#endif
+
 #ifndef RB_UNLIKELY
 #define RB_UNLIKELY(expr) expr
 #endif
@@ -35,6 +44,12 @@ typedef unsigned char _Bool;
 # define MAYBE_UNUSED(x) x
 #endif
 
+#ifdef RUBY_DEBUG
+#ifndef JSON_DEBUG
+#define JSON_DEBUG RUBY_DEBUG
+#endif
+#endif
+
 enum fbuffer_type {
     FBUFFER_HEAP_ALLOCATED = 0,
     FBUFFER_STACK_ALLOCATED = 1,
@@ -45,6 +60,9 @@ typedef struct FBufferStruct {
     unsigned long initial_length;
     unsigned long len;
     unsigned long capa;
+#ifdef JSON_DEBUG
+    unsigned long requested;
+#endif
     char *ptr;
     VALUE io;
 } FBuffer;
@@ -73,6 +91,20 @@ static void fbuffer_stack_init(FBuffer *fb, unsigned long initial_length, char *
         fb->ptr = stack_buffer;
         fb->capa = stack_buffer_size;
     }
+#ifdef JSON_DEBUG
+    fb->requested = 0;
+#endif
+}
+
+static inline void fbuffer_consumed(FBuffer *fb, unsigned long consumed)
+{
+#ifdef JSON_DEBUG
+    if (consumed > fb->requested) {
+        rb_bug("fbuffer: Out of bound write");
+    }
+    fb->requested = 0;
+#endif
+    fb->len += consumed;
 }
 
 static void fbuffer_free(FBuffer *fb)
@@ -136,56 +168,115 @@ static void fbuffer_do_inc_capa(FBuffer *fb, unsigned long requested)
 
 static inline void fbuffer_inc_capa(FBuffer *fb, unsigned long requested)
 {
+#ifdef JSON_DEBUG
+    fb->requested = requested;
+#endif
+
     if (RB_UNLIKELY(requested > fb->capa - fb->len)) {
         fbuffer_do_inc_capa(fb, requested);
     }
 }
 
-static void fbuffer_append(FBuffer *fb, const char *newstr, unsigned long len)
+static inline void fbuffer_append_reserved(FBuffer *fb, const char *newstr, unsigned long len)
+{
+    MEMCPY(fb->ptr + fb->len, newstr, char, len);
+    fbuffer_consumed(fb, len);
+}
+
+static inline void fbuffer_append(FBuffer *fb, const char *newstr, unsigned long len)
 {
     if (len > 0) {
         fbuffer_inc_capa(fb, len);
-        MEMCPY(fb->ptr + fb->len, newstr, char, len);
-        fb->len += len;
+        fbuffer_append_reserved(fb, newstr, len);
     }
 }
 
+/* Appends a character into a buffer. The buffer needs to have sufficient capacity, via fbuffer_inc_capa(...). */
+static inline void fbuffer_append_reserved_char(FBuffer *fb, char chr)
+{
+#ifdef JSON_DEBUG
+    if (fb->requested < 1) {
+        rb_bug("fbuffer: unreserved write");
+    }
+    fb->requested--;
+#endif
+
+    fb->ptr[fb->len] = chr;
+    fb->len++;
+}
+
 static void fbuffer_append_str(FBuffer *fb, VALUE str)
 {
     const char *newstr = StringValuePtr(str);
     unsigned long len = RSTRING_LEN(str);
 
-    RB_GC_GUARD(str);
-
     fbuffer_append(fb, newstr, len);
 }
 
+static void fbuffer_append_str_repeat(FBuffer *fb, VALUE str, size_t repeat)
+{
+    const char *newstr = StringValuePtr(str);
+    unsigned long len = RSTRING_LEN(str);
+
+    fbuffer_inc_capa(fb, repeat * len);
+    while (repeat) {
+#ifdef JSON_DEBUG
+        fb->requested = len;
+#endif
+        fbuffer_append_reserved(fb, newstr, len);
+        repeat--;
+    }
+}
+
 static inline void fbuffer_append_char(FBuffer *fb, char newchr)
 {
     fbuffer_inc_capa(fb, 1);
     *(fb->ptr + fb->len) = newchr;
-    fb->len++;
+    fbuffer_consumed(fb, 1);
 }
 
-static long fltoa(long number, char *buf)
+static inline char *fbuffer_cursor(FBuffer *fb)
 {
-    static const char digits[] = "0123456789";
-    long sign = number;
-    char* tmp = buf;
+    return fb->ptr + fb->len;
+}
 
-    if (sign < 0) number = -number;
-    do *tmp-- = digits[number % 10]; while (number /= 10);
-    if (sign < 0) *tmp-- = '-';
-    return buf - tmp;
+static inline void fbuffer_advance_to(FBuffer *fb, char *end)
+{
+    fbuffer_consumed(fb, (end - fb->ptr) - fb->len);
 }
 
-#define LONG_BUFFER_SIZE 20
+/*
+ * Appends the decimal string representation of \a number into the buffer.
+ */
 static void fbuffer_append_long(FBuffer *fb, long number)
 {
-    char buf[LONG_BUFFER_SIZE];
-    char *buffer_end = buf + LONG_BUFFER_SIZE;
-    long len = fltoa(number, buffer_end - 1);
-    fbuffer_append(fb, buffer_end - len, len);
+    /*
+     * The jeaiii_ultoa() function produces digits left-to-right,
+     * allowing us to write directly into the buffer, but we don't know
+     * the number of resulting characters.
+     *
+     * We do know, however, that the `number` argument is always in the
+     * range 0xc000000000000000 to 0x3fffffffffffffff, or, in decimal,
+     * -4611686018427387904 to 4611686018427387903. The max number of chars
+     * generated is therefore 20 (including a potential sign character).
+     */
+
+    static const int MAX_CHARS_FOR_LONG = 20;
+
+    fbuffer_inc_capa(fb, MAX_CHARS_FOR_LONG);
+
+    if (number < 0) {
+        fbuffer_append_reserved_char(fb, '-');
+
+        /*
+         * Since number is always > LONG_MIN, `-number` will not overflow
+         * and is always the positive abs() value.
+         */
+        number = -number;
+    }
+
+    char *end = jeaiii_ultoa(fbuffer_cursor(fb), number);
+    fbuffer_advance_to(fb, end);
 }
 
 static VALUE fbuffer_finalize(FBuffer *fb)

data/ext/json/ext/generator/extconf.rb

@@ -6,5 +6,11 @@ if RUBY_ENGINE == 'truffleruby'
 else
   append_cflags("-std=c99")
   $defs << "-DJSON_GENERATOR"
+  $defs << "-DJSON_DEBUG" if ENV["JSON_DEBUG"]
+
+  if enable_config('generator-use-simd', default=!ENV["JSON_DISABLE_SIMD"])
+    load __dir__ + "/../simd/conf.rb"
+  end
+
   create_makefile 'json/ext/generator'
 end