json 2.16.0 → 2.17.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: d8bd00bbc63367659eb90ee7e3f4f1394fecdf713fc724f7908c6a2c7b79bbf7
-  data.tar.gz: efa1d81f687277247649de777f25be01e56ad8dfac1bdd9ce61111e10231185e
+  metadata.gz: a35454680e80a622fb539155325b2f693e90b352aad3f24b7f9eafc5ca797963
+  data.tar.gz: 9eee4f7ce5348a1eb44100766d5a7b326a4e76691fbfd3db8a2410020fb75ec1
 SHA512:
-  metadata.gz: cb8b505c08b88115a87aa0b6ab8b7f34d6e0611af7d37277cef243131990db6cb77ca2508045104588607d54bd58b0c2e67ce0728bd72c4c32d2d77a56b9762e
-  data.tar.gz: 8dec0d5ba99f33e68049328f060a600da2fe0545caf47c44c2e812a8316cbe64a99152f365be8654cb4af45493de3bed97baa61a0b2a71ae71a7466d0ee11e8c
+  metadata.gz: 0771fad39331fe9d47dbd855a02f662053a79ff08679c8bc76855dcfd3edbcdb32746cc0db435e6ea14815eacdbe97bf1152b704eacb7ddf1423917eb2545923
+  data.tar.gz: 742dc87f4cc0306e613cfb65824aaee600aa53f0715135190ffaa4ca1f25aa6820a10b1860befcf90f3dcf6676bd5bb304f0e2858ec1e5933f66b089fca82e78

data/CHANGES.md

@@ -2,6 +2,18 @@
 
 ### Unreleased
 
+### 2025-12-03 (2.17.0)
+
+* Improve `JSON.load` and `JSON.unsafe_load` to allow passing options as second argument.
+* Fix the parser to no longer ignore invalid escapes in strings.
+  Only `\"`, `\\`, `\b`, `\f`, `\n`, `\r`, `\t` and `\u` are valid JSON escapes.
+* Fixed `JSON::Coder` to use the depth it was initialized with.
+* On TruffleRuby, fix the generator to not call `to_json` on the return value of `as_json` for `Float::NAN`.
+* Fixed handling of `state.depth`: when `to_json` changes `state.depth` but does not restore it, it is reset
+  automatically to its initial value.
+  In particular, when a `NestingError` is raised, `depth` is no longer equal to `max_nesting` after the call to
+  generate, and is reset to its initial value. Similarly when `to_json` raises an exception.
+
 ### 2025-11-07 (2.16.0)
 
 * Deprecate `JSON::State#[]` and `JSON::State#[]=`. Consider using `JSON::Coder` instead.

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

@@ -14,7 +14,7 @@ typedef struct FBufferStruct {
     unsigned long initial_length;
     unsigned long len;
     unsigned long capa;
-#ifdef JSON_DEBUG
+#if JSON_DEBUG
     unsigned long requested;
 #endif
     char *ptr;
@@ -45,14 +45,14 @@ static void fbuffer_stack_init(FBuffer *fb, unsigned long initial_length, char *
         fb->ptr = stack_buffer;
         fb->capa = stack_buffer_size;
     }
-#ifdef JSON_DEBUG
+#if JSON_DEBUG
     fb->requested = 0;
 #endif
 }
 
 static inline void fbuffer_consumed(FBuffer *fb, unsigned long consumed)
 {
-#ifdef JSON_DEBUG
+#if JSON_DEBUG
     if (consumed > fb->requested) {
         rb_bug("fbuffer: Out of bound write");
     }
@@ -122,7 +122,7 @@ 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
+#if JSON_DEBUG
     fb->requested = requested;
 #endif
 
@@ -148,7 +148,7 @@ static inline void fbuffer_append(FBuffer *fb, const char *newstr, unsigned long
 /* 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 JSON_DEBUG
     if (fb->requested < 1) {
         rb_bug("fbuffer: unreserved write");
     }
@@ -174,7 +174,7 @@ static void fbuffer_append_str_repeat(FBuffer *fb, VALUE str, size_t repeat)
 
     fbuffer_inc_capa(fb, repeat * len);
     while (repeat) {
-#ifdef JSON_DEBUG
+#if JSON_DEBUG
         fb->requested = len;
 #endif
         fbuffer_append_reserved(fb, newstr, len);

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

@@ -6,7 +6,7 @@ if RUBY_ENGINE == 'truffleruby'
 else
   append_cflags("-std=c99")
   $defs << "-DJSON_GENERATOR"
-  $defs << "-DJSON_DEBUG" if ENV["JSON_DEBUG"]
+  $defs << "-DJSON_DEBUG" if ENV.fetch("JSON_DEBUG", "0") != "0"
 
   if enable_config('generator-use-simd', default=!ENV["JSON_DISABLE_SIMD"])
     load __dir__ + "/../simd/conf.rb"

data/ext/json/ext/generator/generator.c

@@ -38,7 +38,7 @@ typedef struct JSON_Generator_StateStruct {
 
 static VALUE mJSON, cState, cFragment, eGeneratorError, eNestingError, Encoding_UTF_8;
 
-static ID i_to_s, i_to_json, i_new, i_pack, i_unpack, i_create_id, i_extend, i_encode;
+static ID i_to_s, i_to_json, i_new, i_encode;
 static VALUE sym_indent, sym_space, sym_space_before, sym_object_nl, sym_array_nl, sym_max_nesting, sym_allow_nan, sym_allow_duplicate_key,
              sym_ascii_only, sym_depth, sym_buffer_initial_length, sym_script_safe, sym_escape_slash, sym_strict, sym_as_json;
 
@@ -60,6 +60,7 @@ struct generate_json_data {
     JSON_Generator_State *state;
     VALUE obj;
     generator_func func;
+    long depth;
 };
 
 static VALUE cState_from_state_s(VALUE self, VALUE opts);
@@ -127,7 +128,7 @@ typedef struct _search_state {
 #endif /* HAVE_SIMD */
 } search_state;
 
-static ALWAYS_INLINE() void search_flush(search_state *search)
+ALWAYS_INLINE(static) void search_flush(search_state *search)
 {
     // Do not remove this conditional without profiling, specifically escape-heavy text.
     // escape_UTF8_char_basic will advance search->ptr and search->cursor (effectively a search_flush).
@@ -170,7 +171,7 @@ static inline unsigned char search_escape_basic(search_state *search)
     return 0;
 }
 
-static ALWAYS_INLINE() void escape_UTF8_char_basic(search_state *search)
+ALWAYS_INLINE(static) void escape_UTF8_char_basic(search_state *search)
 {
     const unsigned char ch = (unsigned char)*search->ptr;
     switch (ch) {
@@ -257,7 +258,7 @@ static inline void escape_UTF8_char(search_state *search, unsigned char ch_len)
 
 #ifdef HAVE_SIMD
 
-static ALWAYS_INLINE() char *copy_remaining_bytes(search_state *search, unsigned long vec_len, unsigned long len)
+ALWAYS_INLINE(static) char *copy_remaining_bytes(search_state *search, unsigned long vec_len, unsigned long len)
 {
     // Flush the buffer so everything up until the last 'len' characters are unflushed.
     search_flush(search);
@@ -280,7 +281,7 @@ static ALWAYS_INLINE() char *copy_remaining_bytes(search_state *search, unsigned
 
 #ifdef HAVE_SIMD_NEON
 
-static ALWAYS_INLINE() unsigned char neon_next_match(search_state *search)
+ALWAYS_INLINE(static) unsigned char neon_next_match(search_state *search)
 {
     uint64_t mask = search->matches_mask;
     uint32_t index = trailing_zeros64(mask) >> 2;
@@ -394,7 +395,7 @@ static inline unsigned char search_escape_basic_neon(search_state *search)
 
 #ifdef HAVE_SIMD_SSE2
 
-static ALWAYS_INLINE() unsigned char sse2_next_match(search_state *search)
+ALWAYS_INLINE(static) unsigned char sse2_next_match(search_state *search)
 {
     int mask = search->matches_mask;
     int index = trailing_zeros(mask);
@@ -418,7 +419,7 @@ static ALWAYS_INLINE() unsigned char sse2_next_match(search_state *search)
 #define TARGET_SSE2
 #endif
 
-static TARGET_SSE2 ALWAYS_INLINE() unsigned char search_escape_basic_sse2(search_state *search)
+ALWAYS_INLINE(static) TARGET_SSE2 unsigned char search_escape_basic_sse2(search_state *search)
 {
     if (RB_UNLIKELY(search->has_matches)) {
         // There are more matches if search->matches_mask > 0.
@@ -926,11 +927,6 @@ static size_t State_memsize(const void *ptr)
     return sizeof(JSON_Generator_State);
 }
 
-#ifndef HAVE_RB_EXT_RACTOR_SAFE
-#   undef RUBY_TYPED_FROZEN_SHAREABLE
-#   define RUBY_TYPED_FROZEN_SHAREABLE 0
-#endif
-
 static const rb_data_type_t JSON_Generator_State_type = {
     "JSON/Generator/State",
     {
@@ -972,12 +968,16 @@ static void vstate_spill(struct generate_json_data *data)
     RB_OBJ_WRITTEN(vstate, Qundef, state->as_json);
 }
 
-static inline VALUE vstate_get(struct generate_json_data *data)
+static inline VALUE json_call_to_json(struct generate_json_data *data, VALUE obj)
 {
     if (RB_UNLIKELY(!data->vstate)) {
         vstate_spill(data);
     }
-    return data->vstate;
+    GET_STATE(data->vstate);
+    state->depth = data->depth;
+    VALUE tmp = rb_funcall(obj, i_to_json, 1, data->vstate);
+    // no need to restore state->depth, vstate is just a temporary State
+    return tmp;
 }
 
 static VALUE
@@ -1125,8 +1125,7 @@ struct hash_foreach_arg {
     bool mixed_keys_encountered;
 };
 
-NOINLINE()
-static void
+NOINLINE(static) void
 json_inspect_hash_with_mixed_keys(struct hash_foreach_arg *arg)
 {
     if (arg->mixed_keys_encountered) {
@@ -1150,7 +1149,7 @@ json_object_i(VALUE key, VALUE val, VALUE _arg)
     FBuffer *buffer = data->buffer;
     JSON_Generator_State *state = data->state;
 
-    long depth = state->depth;
+    long depth = data->depth;
     int key_type = rb_type(key);
 
     if (arg->first) {
@@ -1224,9 +1223,9 @@ json_object_i(VALUE key, VALUE val, VALUE _arg)
 static inline long increase_depth(struct generate_json_data *data)
 {
     JSON_Generator_State *state = data->state;
-    long depth = ++state->depth;
+    long depth = ++data->depth;
     if (RB_UNLIKELY(depth > state->max_nesting && state->max_nesting)) {
-        rb_raise(eNestingError, "nesting of %ld is too deep. Did you try to serialize objects with circular references?", --state->depth);
+        rb_raise(eNestingError, "nesting of %ld is too deep. Did you try to serialize objects with circular references?", --data->depth);
     }
     return depth;
 }
@@ -1237,7 +1236,7 @@ static void generate_json_object(FBuffer *buffer, struct generate_json_data *dat
 
     if (RHASH_SIZE(obj) == 0) {
         fbuffer_append(buffer, "{}", 2);
-        --data->state->depth;
+        --data->depth;
         return;
     }
 
@@ -1250,7 +1249,7 @@ static void generate_json_object(FBuffer *buffer, struct generate_json_data *dat
     };
     rb_hash_foreach(obj, json_object_i, (VALUE)&arg);
 
-    depth = --data->state->depth;
+    depth = --data->depth;
     if (RB_UNLIKELY(data->state->object_nl)) {
         fbuffer_append_str(buffer, data->state->object_nl);
         if (RB_UNLIKELY(data->state->indent)) {
@@ -1266,7 +1265,7 @@ static void generate_json_array(FBuffer *buffer, struct generate_json_data *data
 
     if (RARRAY_LEN(obj) == 0) {
         fbuffer_append(buffer, "[]", 2);
-        --data->state->depth;
+        --data->depth;
         return;
     }
 
@@ -1282,7 +1281,7 @@ static void generate_json_array(FBuffer *buffer, struct generate_json_data *data
         }
         generate_json(buffer, data, RARRAY_AREF(obj, i));
     }
-    data->state->depth = --depth;
+    data->depth = --depth;
     if (RB_UNLIKELY(data->state->array_nl)) {
         fbuffer_append_str(buffer, data->state->array_nl);
         if (RB_UNLIKELY(data->state->indent)) {
@@ -1296,7 +1295,7 @@ static void generate_json_fallback(FBuffer *buffer, struct generate_json_data *d
 {
     VALUE tmp;
     if (rb_respond_to(obj, i_to_json)) {
-        tmp = rb_funcall(obj, i_to_json, 1, vstate_get(data));
+        tmp = json_call_to_json(data, obj);
         Check_Type(tmp, T_STRING);
         fbuffer_append_str(buffer, tmp);
     } else {
@@ -1363,7 +1362,7 @@ static void generate_json_float(FBuffer *buffer, struct generate_json_data *data
                 if (casted_obj != obj) {
                     increase_depth(data);
                     generate_json(buffer, data, casted_obj);
-                    data->state->depth--;
+                    data->depth--;
                     return;
                 }
             }
@@ -1498,8 +1497,9 @@ static VALUE cState_partial_generate(VALUE self, VALUE obj, generator_func func,
 
     struct generate_json_data data = {
         .buffer = &buffer,
-        .vstate = self,
+        .vstate = Qfalse, // don't use self as it may be frozen and its depth is mutated when calling to_json
         .state = state,
+        .depth = state->depth,
         .obj = obj,
         .func = func
     };
@@ -1522,36 +1522,6 @@ static VALUE cState_generate(int argc, VALUE *argv, VALUE self)
     return cState_partial_generate(self, obj, generate_json, io);
 }
 
-static VALUE cState_generate_new(int argc, VALUE *argv, VALUE self)
-{
-    rb_check_arity(argc, 1, 2);
-    VALUE obj = argv[0];
-    VALUE io = argc > 1 ? argv[1] : Qnil;
-
-    GET_STATE(self);
-
-    JSON_Generator_State new_state;
-    MEMCPY(&new_state, state, JSON_Generator_State, 1);
-
-    // FIXME: depth shouldn't be part of JSON_Generator_State, as that prevents it from being used concurrently.
-    new_state.depth = 0;
-
-    char stack_buffer[FBUFFER_STACK_SIZE];
-    FBuffer buffer = {
-        .io = RTEST(io) ? io : Qfalse,
-    };
-    fbuffer_stack_init(&buffer, state->buffer_initial_length, stack_buffer, FBUFFER_STACK_SIZE);
-
-    struct generate_json_data data = {
-        .buffer = &buffer,
-        .vstate = Qfalse,
-        .state = &new_state,
-        .obj = obj,
-        .func = generate_json
-    };
-    return rb_ensure(generate_json_try, (VALUE)&data, generate_json_ensure, (VALUE)&data);
-}
-
 static VALUE cState_initialize(int argc, VALUE *argv, VALUE self)
 {
     rb_warn("The json gem extension was loaded with the stdlib ruby code. You should upgrade rubygems with `gem update --system`");
@@ -1630,6 +1600,7 @@ static VALUE string_config(VALUE config)
  */
 static VALUE cState_indent_set(VALUE self, VALUE indent)
 {
+    rb_check_frozen(self);
     GET_STATE(self);
     RB_OBJ_WRITE(self, &state->indent, string_config(indent));
     return Qnil;
@@ -1655,6 +1626,7 @@ static VALUE cState_space(VALUE self)
  */
 static VALUE cState_space_set(VALUE self, VALUE space)
 {
+    rb_check_frozen(self);
     GET_STATE(self);
     RB_OBJ_WRITE(self, &state->space, string_config(space));
     return Qnil;
@@ -1678,6 +1650,7 @@ static VALUE cState_space_before(VALUE self)
  */
 static VALUE cState_space_before_set(VALUE self, VALUE space_before)
 {
+    rb_check_frozen(self);
     GET_STATE(self);
     RB_OBJ_WRITE(self, &state->space_before, string_config(space_before));
     return Qnil;
@@ -1703,6 +1676,7 @@ static VALUE cState_object_nl(VALUE self)
  */
 static VALUE cState_object_nl_set(VALUE self, VALUE object_nl)
 {
+    rb_check_frozen(self);
     GET_STATE(self);
     RB_OBJ_WRITE(self, &state->object_nl, string_config(object_nl));
     return Qnil;
@@ -1726,6 +1700,7 @@ static VALUE cState_array_nl(VALUE self)
  */
 static VALUE cState_array_nl_set(VALUE self, VALUE array_nl)
 {
+    rb_check_frozen(self);
     GET_STATE(self);
     RB_OBJ_WRITE(self, &state->array_nl, string_config(array_nl));
     return Qnil;
@@ -1749,6 +1724,7 @@ static VALUE cState_as_json(VALUE self)
  */
 static VALUE cState_as_json_set(VALUE self, VALUE as_json)
 {
+    rb_check_frozen(self);
     GET_STATE(self);
     RB_OBJ_WRITE(self, &state->as_json, rb_convert_type(as_json, T_DATA, "Proc", "to_proc"));
     return Qnil;
@@ -1791,6 +1767,7 @@ static long long_config(VALUE num)
  */
 static VALUE cState_max_nesting_set(VALUE self, VALUE depth)
 {
+    rb_check_frozen(self);
     GET_STATE(self);
     state->max_nesting = long_config(depth);
     return Qnil;
@@ -1816,6 +1793,7 @@ static VALUE cState_script_safe(VALUE self)
  */
 static VALUE cState_script_safe_set(VALUE self, VALUE enable)
 {
+    rb_check_frozen(self);
     GET_STATE(self);
     state->script_safe = RTEST(enable);
     return Qnil;
@@ -1847,6 +1825,7 @@ static VALUE cState_strict(VALUE self)
  */
 static VALUE cState_strict_set(VALUE self, VALUE enable)
 {
+    rb_check_frozen(self);
     GET_STATE(self);
     state->strict = RTEST(enable);
     return Qnil;
@@ -1871,6 +1850,7 @@ static VALUE cState_allow_nan_p(VALUE self)
  */
 static VALUE cState_allow_nan_set(VALUE self, VALUE enable)
 {
+    rb_check_frozen(self);
     GET_STATE(self);
     state->allow_nan = RTEST(enable);
     return Qnil;
@@ -1895,6 +1875,7 @@ static VALUE cState_ascii_only_p(VALUE self)
  */
 static VALUE cState_ascii_only_set(VALUE self, VALUE enable)
 {
+    rb_check_frozen(self);
     GET_STATE(self);
     state->ascii_only = RTEST(enable);
     return Qnil;
@@ -1932,6 +1913,7 @@ static VALUE cState_depth(VALUE self)
  */
 static VALUE cState_depth_set(VALUE self, VALUE depth)
 {
+    rb_check_frozen(self);
     GET_STATE(self);
     state->depth = long_config(depth);
     return Qnil;
@@ -1965,6 +1947,7 @@ static void buffer_initial_length_set(JSON_Generator_State *state, VALUE buffer_
  */
 static VALUE cState_buffer_initial_length_set(VALUE self, VALUE buffer_initial_length)
 {
+    rb_check_frozen(self);
     GET_STATE(self);
     buffer_initial_length_set(state, buffer_initial_length);
     return Qnil;
@@ -2031,6 +2014,7 @@ static void configure_state(JSON_Generator_State *state, VALUE vstate, VALUE con
 
 static VALUE cState_configure(VALUE self, VALUE opts)
 {
+    rb_check_frozen(self);
     GET_STATE(self);
     configure_state(state, self, opts);
     return self;
@@ -2052,6 +2036,7 @@ static VALUE cState_m_generate(VALUE klass, VALUE obj, VALUE opts, VALUE io)
         .buffer = &buffer,
         .vstate = Qfalse,
         .state = &state,
+        .depth = state.depth,
         .obj = obj,
         .func = generate_json,
     };
@@ -2125,7 +2110,6 @@ void Init_generator(void)
     rb_define_method(cState, "buffer_initial_length", cState_buffer_initial_length, 0);
     rb_define_method(cState, "buffer_initial_length=", cState_buffer_initial_length_set, 1);
     rb_define_method(cState, "generate", cState_generate, -1);
-    rb_define_method(cState, "generate_new", cState_generate_new, -1); // :nodoc:
 
     rb_define_private_method(cState, "allow_duplicate_key?", cState_allow_duplicate_key_p, 0);
 
@@ -2173,10 +2157,6 @@ void Init_generator(void)
     i_to_s = rb_intern("to_s");
     i_to_json = rb_intern("to_json");
     i_new = rb_intern("new");
-    i_pack = rb_intern("pack");
-    i_unpack = rb_intern("unpack");
-    i_create_id = rb_intern("create_id");
-    i_extend = rb_intern("extend");
     i_encode = rb_intern("encode");
 
     sym_indent = ID2SYM(rb_intern("indent"));

data/ext/json/ext/json.h

@@ -45,6 +45,11 @@ typedef unsigned char _Bool;
 #endif
 #endif
 
+#ifndef HAVE_RB_EXT_RACTOR_SAFE
+#   undef RUBY_TYPED_FROZEN_SHAREABLE
+#   define RUBY_TYPED_FROZEN_SHAREABLE 0
+#endif
+
 #ifndef NORETURN
 #define NORETURN(x) x
 #endif

data/ext/json/ext/parser/extconf.rb

@@ -1,12 +1,11 @@
 # frozen_string_literal: true
 require 'mkmf'
 
-$defs << "-DJSON_DEBUG" if ENV["JSON_DEBUG"]
+$defs << "-DJSON_DEBUG" if ENV.fetch("JSON_DEBUG", "0") != "0"
 have_func("rb_enc_interned_str", "ruby/encoding.h") # RUBY_VERSION >= 3.0
 have_func("rb_str_to_interned_str", "ruby.h") # RUBY_VERSION >= 3.0
 have_func("rb_hash_new_capa", "ruby.h") # RUBY_VERSION >= 3.2
 have_func("rb_hash_bulk_insert", "ruby.h") # Missing on TruffleRuby
-have_func("strnlen", "string.h") # Missing on Solaris 10
 
 append_cflags("-std=c99")
 

data/ext/json/ext/parser/parser.c

@@ -5,8 +5,7 @@
 static VALUE mJSON, eNestingError, Encoding_UTF_8;
 static VALUE CNaN, CInfinity, CMinusInfinity;
 
-static ID i_chr, i_aset, i_aref,
-          i_leftshift, i_new, i_try_convert, i_uminus, i_encode;
+static ID i_new, i_try_convert, i_uminus, i_encode;
 
 static VALUE sym_max_nesting, sym_allow_nan, sym_allow_trailing_comma, sym_symbolize_names, sym_freeze,
              sym_decimal_class, sym_on_load, sym_allow_duplicate_key;
@@ -89,7 +88,7 @@ static void rvalue_cache_insert_at(rvalue_cache *cache, int index, VALUE rstring
 #if JSON_CPU_LITTLE_ENDIAN_64BITS
 #if __has_builtin(__builtin_bswap64)
 #undef rstring_cache_memcmp
-static ALWAYS_INLINE() int rstring_cache_memcmp(const char *str, const char *rptr, const long length)
+ALWAYS_INLINE(static) int rstring_cache_memcmp(const char *str, const char *rptr, const long length)
 {
     // The libc memcmp has numerous complex optimizations, but in this particular case,
     // we know the string is small (JSON_RVALUE_CACHE_MAX_ENTRY_LENGTH), so being able to
@@ -118,7 +117,7 @@ static ALWAYS_INLINE() int rstring_cache_memcmp(const char *str, const char *rpt
 #endif
 #endif
 
-static ALWAYS_INLINE() int rstring_cache_cmp(const char *str, const long length, VALUE rstring)
+ALWAYS_INLINE(static) int rstring_cache_cmp(const char *str, const long length, VALUE rstring)
 {
     const char *rstring_ptr;
     long rstring_length;
@@ -132,7 +131,7 @@ static ALWAYS_INLINE() int rstring_cache_cmp(const char *str, const long length,
     }
 }
 
-static ALWAYS_INLINE() VALUE rstring_cache_fetch(rvalue_cache *cache, const char *str, const long length)
+ALWAYS_INLINE(static) VALUE rstring_cache_fetch(rvalue_cache *cache, const char *str, const long length)
 {
     int low = 0;
     int high = cache->length - 1;
@@ -296,15 +295,6 @@ static void rvalue_stack_eagerly_release(VALUE handle)
     }
 }
 
-
-#ifndef HAVE_STRNLEN
-static size_t strnlen(const char *s, size_t maxlen)
-{
-    char *p;
-    return ((p = memchr(s, '\0', maxlen)) ? p - s : maxlen);
-}
-#endif
-
 static int convert_UTF32_to_UTF8(char *buf, uint32_t ch)
 {
     int len = 1;
@@ -345,7 +335,6 @@ typedef struct JSON_ParserStruct {
     int max_nesting;
     bool allow_nan;
     bool allow_trailing_comma;
-    bool parsing_name;
     bool symbolize_names;
     bool freeze;
 } JSON_ParserConfig;
@@ -551,7 +540,7 @@ json_eat_comments(JSON_ParserState *state)
     }
 }
 
-static ALWAYS_INLINE() void
+ALWAYS_INLINE(static) void
 json_eat_whitespace(JSON_ParserState *state)
 {
     while (true) {
@@ -627,8 +616,10 @@ static inline bool json_string_cacheable_p(const char *string, size_t length)
     return length <= JSON_RVALUE_CACHE_MAX_ENTRY_LENGTH && rb_isalpha(string[0]);
 }
 
-static inline VALUE json_string_fastpath(JSON_ParserState *state, const char *string, const char *stringEnd, bool is_name, bool intern, bool symbolize)
+static inline VALUE json_string_fastpath(JSON_ParserState *state, JSON_ParserConfig *config, const char *string, const char *stringEnd, bool is_name)
 {
+    bool intern = is_name || config->freeze;
+    bool symbolize = is_name && config->symbolize_names;
     size_t bufferSize = stringEnd - string;
 
     if (is_name && state->in_array && RB_LIKELY(json_string_cacheable_p(string, bufferSize))) {
@@ -647,47 +638,71 @@ static inline VALUE json_string_fastpath(JSON_ParserState *state, const char *st
     return build_string(string, stringEnd, intern, symbolize);
 }
 
-static VALUE json_string_unescape(JSON_ParserState *state, const char *string, const char *stringEnd, bool is_name, bool intern, bool symbolize)
+#define JSON_MAX_UNESCAPE_POSITIONS 16
+typedef struct _json_unescape_positions {
+    long size;
+    const char **positions;
+    bool has_more;
+} JSON_UnescapePositions;
+
+static inline const char *json_next_backslash(const char *pe, const char *stringEnd, JSON_UnescapePositions *positions)
+{
+    while (positions->size) {
+        positions->size--;
+        const char *next_position = positions->positions[0];
+        positions->positions++;
+        return next_position;
+    }
+
+    if (positions->has_more) {
+        return memchr(pe, '\\', stringEnd - pe);
+    }
+
+    return NULL;
+}
+
+NOINLINE(static) VALUE json_string_unescape(JSON_ParserState *state, JSON_ParserConfig *config, const char *string, const char *stringEnd, bool is_name, JSON_UnescapePositions *positions)
 {
+    bool intern = is_name || config->freeze;
+    bool symbolize = is_name && config->symbolize_names;
     size_t bufferSize = stringEnd - string;
-    const char *p = string, *pe = string, *unescape, *bufferStart;
+    const char *p = string, *pe = string, *bufferStart;
     char *buffer;
-    int unescape_len;
-    char buf[4];
 
     VALUE result = rb_str_buf_new(bufferSize);
     rb_enc_associate_index(result, utf8_encindex);
     buffer = RSTRING_PTR(result);
     bufferStart = buffer;
 
-    while (pe < stringEnd && (pe = memchr(pe, '\\', stringEnd - pe))) {
-        unescape = (char *) "?";
-        unescape_len = 1;
+#define APPEND_CHAR(chr) *buffer++ = chr; p = ++pe;
+
+    while (pe < stringEnd && (pe = json_next_backslash(pe, stringEnd, positions))) {
         if (pe > p) {
           MEMCPY(buffer, p, char, pe - p);
           buffer += pe - p;
         }
         switch (*++pe) {
+            case '"':
+            case '/':
+                p = pe; // nothing to unescape just need to skip the backslash
+                break;
+            case '\\':
+                APPEND_CHAR('\\');
+                break;
             case 'n':
-                unescape = (char *) "\n";
+                APPEND_CHAR('\n');
                 break;
             case 'r':
-                unescape = (char *) "\r";
+                APPEND_CHAR('\r');
                 break;
             case 't':
-                unescape = (char *) "\t";
-                break;
-            case '"':
-                unescape = (char *) "\"";
-                break;
-            case '\\':
-                unescape = (char *) "\\";
+                APPEND_CHAR('\t');
                 break;
             case 'b':
-                unescape = (char *) "\b";
+                APPEND_CHAR('\b');
                 break;
             case 'f':
-                unescape = (char *) "\f";
+                APPEND_CHAR('\f');
                 break;
             case 'u':
                 if (pe > stringEnd - 5) {
@@ -725,18 +740,23 @@ static VALUE json_string_unescape(JSON_ParserState *state, const char *string, c
                             break;
                         }
                     }
-                    unescape_len = convert_UTF32_to_UTF8(buf, ch);
-                    unescape = buf;
+
+                    char buf[4];
+                    int unescape_len = convert_UTF32_to_UTF8(buf, ch);
+                    MEMCPY(buffer, buf, char, unescape_len);
+                    buffer += unescape_len;
+                    p = ++pe;
                 }
                 break;
             default:
-                p = pe;
-                continue;
+                if ((unsigned char)*pe < 0x20) {
+                    raise_parse_error_at("invalid ASCII control character in string: %s", state, pe - 1);
+                }
+                raise_parse_error_at("invalid escape character in string: %s", state, pe - 1);
+                break;
         }
-        MEMCPY(buffer, unescape, char, unescape_len);
-        buffer += unescape_len;
-        p = ++pe;
     }
+#undef APPEND_CHAR
 
     if (stringEnd > p) {
       MEMCPY(buffer, p, char, stringEnd - p);
@@ -900,20 +920,6 @@ static inline VALUE json_decode_object(JSON_ParserState *state, JSON_ParserConfi
     return object;
 }
 
-static inline VALUE json_decode_string(JSON_ParserState *state, JSON_ParserConfig *config, const char *start, const char *end, bool escaped, bool is_name)
-{
-    VALUE string;
-    bool intern = is_name || config->freeze;
-    bool symbolize = is_name && config->symbolize_names;
-    if (escaped) {
-        string = json_string_unescape(state, start, end, is_name, intern, symbolize);
-    } else {
-        string = json_string_fastpath(state, start, end, is_name, intern, symbolize);
-    }
-
-    return string;
-}
-
 static inline VALUE json_push_value(JSON_ParserState *state, JSON_ParserConfig *config, VALUE value)
 {
     if (RB_UNLIKELY(config->on_load_proc)) {
@@ -940,7 +946,7 @@ static const bool string_scan_table[256] = {
 static SIMD_Implementation simd_impl = SIMD_NONE;
 #endif /* HAVE_SIMD */
 
-static ALWAYS_INLINE() bool string_scan(JSON_ParserState *state)
+ALWAYS_INLINE(static) bool string_scan(JSON_ParserState *state)
 {
 #ifdef HAVE_SIMD
 #if defined(HAVE_SIMD_NEON)
@@ -948,7 +954,7 @@ static ALWAYS_INLINE() bool string_scan(JSON_ParserState *state)
     uint64_t mask = 0;
     if (string_scan_simd_neon(&state->cursor, state->end, &mask)) {
         state->cursor += trailing_zeros64(mask) >> 2;
-        return 1;
+        return true;
     }
 
 #elif defined(HAVE_SIMD_SSE2)
@@ -956,7 +962,7 @@ static ALWAYS_INLINE() bool string_scan(JSON_ParserState *state)
         int mask = 0;
         if (string_scan_simd_sse2(&state->cursor, state->end, &mask)) {
             state->cursor += trailing_zeros(mask);
-            return 1;
+            return true;
         }
     }
 #endif /* HAVE_SIMD_NEON or HAVE_SIMD_SSE2 */
@@ -964,32 +970,37 @@ static ALWAYS_INLINE() bool string_scan(JSON_ParserState *state)
 
     while (!eos(state)) {
         if (RB_UNLIKELY(string_scan_table[(unsigned char)*state->cursor])) {
-            return 1;
+            return true;
         }
         state->cursor++;
     }
-    return 0;
+    return false;
 }
 
-static inline VALUE json_parse_string(JSON_ParserState *state, JSON_ParserConfig *config, bool is_name)
+static VALUE json_parse_escaped_string(JSON_ParserState *state, JSON_ParserConfig *config, bool is_name, const char *start)
 {
-    state->cursor++;
-    const char *start = state->cursor;
-    bool escaped = false;
+    const char *backslashes[JSON_MAX_UNESCAPE_POSITIONS];
+    JSON_UnescapePositions positions = {
+        .size = 0,
+        .positions = backslashes,
+        .has_more = false,
+    };
 
-    while (RB_UNLIKELY(string_scan(state))) {
+    do {
         switch (*state->cursor) {
             case '"': {
-                VALUE string = json_decode_string(state, config, start, state->cursor, escaped, is_name);
+                VALUE string = json_string_unescape(state, config, start, state->cursor, is_name, &positions);
                 state->cursor++;
                 return json_push_value(state, config, string);
             }
             case '\\': {
-                state->cursor++;
-                escaped = true;
-                if ((unsigned char)*state->cursor < 0x20) {
-                    raise_parse_error("invalid ASCII control character in string: %s", state);
+                if (RB_LIKELY(positions.size < JSON_MAX_UNESCAPE_POSITIONS)) {
+                    backslashes[positions.size] = state->cursor;
+                    positions.size++;
+                } else {
+                    positions.has_more = true;
                 }
+                state->cursor++;
                 break;
             }
             default:
@@ -998,12 +1009,29 @@ static inline VALUE json_parse_string(JSON_ParserState *state, JSON_ParserConfig
         }
 
         state->cursor++;
-    }
+    } while (string_scan(state));
 
     raise_parse_error("unexpected end of input, expected closing \"", state);
     return Qfalse;
 }
 
+ALWAYS_INLINE(static) VALUE json_parse_string(JSON_ParserState *state, JSON_ParserConfig *config, bool is_name)
+{
+    state->cursor++;
+    const char *start = state->cursor;
+
+    if (RB_UNLIKELY(!string_scan(state))) {
+        raise_parse_error("unexpected end of input, expected closing \"", state);
+    }
+
+    if (RB_LIKELY(*state->cursor == '"')) {
+        VALUE string = json_string_fastpath(state, config, start, state->cursor, is_name);
+        state->cursor++;
+        return json_push_value(state, config, string);
+    }
+    return json_parse_escaped_string(state, config, is_name, start);
+}
+
 #if JSON_CPU_LITTLE_ENDIAN_64BITS
 // From: https://lemire.me/blog/2022/01/21/swar-explained-parsing-eight-digits/
 // Additional References:
@@ -1477,6 +1505,7 @@ static void parser_config_init(JSON_ParserConfig *config, VALUE opts)
  */
 static VALUE cParserConfig_initialize(VALUE self, VALUE opts)
 {
+    rb_check_frozen(self);
     GET_PARSER_CONFIG;
 
     parser_config_init(config, opts);
@@ -1572,7 +1601,7 @@ static const rb_data_type_t JSON_ParserConfig_type = {
         JSON_ParserConfig_memsize,
     },
     0, 0,
-    RUBY_TYPED_FREE_IMMEDIATELY | RUBY_TYPED_WB_PROTECTED,
+    RUBY_TYPED_FREE_IMMEDIATELY | RUBY_TYPED_WB_PROTECTED | RUBY_TYPED_FROZEN_SHAREABLE,
 };
 
 static VALUE cJSON_parser_s_allocate(VALUE klass)
@@ -1622,10 +1651,6 @@ void Init_parser(void)
     sym_decimal_class = ID2SYM(rb_intern("decimal_class"));
     sym_allow_duplicate_key = ID2SYM(rb_intern("allow_duplicate_key"));
 
-    i_chr = rb_intern("chr");
-    i_aset = rb_intern("[]=");
-    i_aref = rb_intern("[]");
-    i_leftshift = rb_intern("<<");
     i_new = rb_intern("new");
     i_try_convert = rb_intern("try_convert");
     i_uminus = rb_intern("-@");

data/ext/json/ext/simd/simd.h

@@ -73,14 +73,14 @@ static inline SIMD_Implementation find_simd_implementation(void)
 #define HAVE_SIMD_NEON 1
 
 // See: https://community.arm.com/arm-community-blogs/b/servers-and-cloud-computing-blog/posts/porting-x86-vector-bitmask-optimizations-to-arm-neon
-static ALWAYS_INLINE() uint64_t neon_match_mask(uint8x16_t matches)
+ALWAYS_INLINE(static) uint64_t neon_match_mask(uint8x16_t matches)
 {
     const uint8x8_t res = vshrn_n_u16(vreinterpretq_u16_u8(matches), 4);
     const uint64_t mask = vget_lane_u64(vreinterpret_u64_u8(res), 0);
     return mask & 0x8888888888888888ull;
 }
 
-static ALWAYS_INLINE() uint64_t compute_chunk_mask_neon(const char *ptr)
+ALWAYS_INLINE(static) uint64_t compute_chunk_mask_neon(const char *ptr)
 {
     uint8x16_t chunk = vld1q_u8((const unsigned char *)ptr);
 
@@ -93,7 +93,7 @@ static ALWAYS_INLINE() uint64_t compute_chunk_mask_neon(const char *ptr)
     return neon_match_mask(needs_escape);
 }
 
-static ALWAYS_INLINE() int string_scan_simd_neon(const char **ptr, const char *end, uint64_t *mask)
+ALWAYS_INLINE(static) int string_scan_simd_neon(const char **ptr, const char *end, uint64_t *mask)
 {
     while (*ptr + sizeof(uint8x16_t) <= end) {
         uint64_t chunk_mask = compute_chunk_mask_neon(*ptr);
@@ -140,7 +140,7 @@ static inline uint8x16x4_t load_uint8x16_4(const unsigned char *table)
 #define _mm_cmpgt_epu8(a, b) _mm_xor_si128(_mm_cmple_epu8(a, b), _mm_set1_epi8(-1))
 #define _mm_cmplt_epu8(a, b) _mm_cmpgt_epu8(b, a)
 
-static TARGET_SSE2 ALWAYS_INLINE() int compute_chunk_mask_sse2(const char *ptr)
+ALWAYS_INLINE(static) TARGET_SSE2 int compute_chunk_mask_sse2(const char *ptr)
 {
     __m128i chunk         = _mm_loadu_si128((__m128i const*)ptr);
     // Trick: c < 32 || c == 34 can be factored as c ^ 2 < 33
@@ -151,7 +151,7 @@ static TARGET_SSE2 ALWAYS_INLINE() int compute_chunk_mask_sse2(const char *ptr)
     return _mm_movemask_epi8(needs_escape);
 }
 
-static TARGET_SSE2 ALWAYS_INLINE() int string_scan_simd_sse2(const char **ptr, const char *end, int *mask)
+ALWAYS_INLINE(static) TARGET_SSE2 int string_scan_simd_sse2(const char **ptr, const char *end, int *mask)
 {
     while (*ptr + sizeof(__m128i) <= end) {
         int chunk_mask = compute_chunk_mask_sse2(*ptr);

data/ext/json/ext/vendor/fpconv.c

@@ -29,7 +29,7 @@
 #include <string.h>
 #include <stdint.h>
 
-#ifdef JSON_DEBUG
+#if JSON_DEBUG
 #include <assert.h>
 #endif
 
@@ -472,7 +472,7 @@ static int fpconv_dtoa(double d, char dest[28])
     int ndigits = grisu2(d, digits, &K);
 
     str_len += emit_digits(digits, ndigits, dest + str_len, K, neg);
-#ifdef JSON_DEBUG
+#if JSON_DEBUG
     assert(str_len <= 32);
 #endif
 

data/lib/json/common.rb

@@ -550,6 +550,7 @@ module JSON
     :create_additions => nil,
   }
   # :call-seq:
+  #   JSON.unsafe_load(source, options = {}) -> object
   #   JSON.unsafe_load(source, proc = nil, options = {}) -> object
   #
   # Returns the Ruby objects created by parsing the given +source+.
@@ -681,7 +682,12 @@ module JSON
   #
   def unsafe_load(source, proc = nil, options = nil)
     opts = if options.nil?
-      _unsafe_load_default_options
+      if proc && proc.is_a?(Hash)
+        options, proc = proc, nil
+        options
+      else
+        _unsafe_load_default_options
+      end
     else
       _unsafe_load_default_options.merge(options)
     end
@@ -709,6 +715,7 @@ module JSON
   end
 
   # :call-seq:
+  #   JSON.load(source, options = {}) -> object
   #   JSON.load(source, proc = nil, options = {}) -> object
   #
   # Returns the Ruby objects created by parsing the given +source+.
@@ -845,8 +852,18 @@ module JSON
   #      @attributes={"type"=>"Admin", "password"=>"0wn3d"}>}
   #
   def load(source, proc = nil, options = nil)
+    if proc && options.nil? && proc.is_a?(Hash)
+      options = proc
+      proc = nil
+    end
+
     opts = if options.nil?
-      _load_default_options
+      if proc && proc.is_a?(Hash)
+        options, proc = proc, nil
+        options
+      else
+        _load_default_options
+      end
     else
       _load_default_options.merge(options)
     end
@@ -1048,7 +1065,7 @@ module JSON
       options[:as_json] = as_json if as_json
 
       @state = State.new(options).freeze
-      @parser_config = Ext::Parser::Config.new(ParserOptions.prepare(options))
+      @parser_config = Ext::Parser::Config.new(ParserOptions.prepare(options)).freeze
     end
 
     # call-seq:
@@ -1057,7 +1074,7 @@ module JSON
     #
     # Serialize the given object into a \JSON document.
     def dump(object, io = nil)
-      @state.generate_new(object, io)
+      @state.generate(object, io)
     end
     alias_method :generate, :dump
 

data/lib/json/truffle_ruby/generator.rb

@@ -312,8 +312,8 @@ module JSON
         def to_h
           result = {}
           instance_variables.each do |iv|
-            iv = iv.to_s[1..-1]
-            result[iv.to_sym] = self[iv]
+            key = iv.to_s[1..-1]
+            result[key.to_sym] = instance_variable_get(iv)
           end
 
           if result[:allow_duplicate_key].nil?
@@ -330,6 +330,9 @@ module JSON
         # created this method raises a
         # GeneratorError exception.
         def generate(obj, anIO = nil)
+          return dup.generate(obj, anIO) if frozen?
+
+          depth = @depth
           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)
             result = generate_json(obj, ''.dup)
@@ -346,14 +349,8 @@ module JSON
           else
             result
           end
-        end
-
-        def generate_new(obj, anIO = nil) # :nodoc:
-          dup.generate(obj, anIO)
-        end
-
-        private def initialize_copy(_orig)
-          @depth = 0
+        ensure
+          @depth = depth unless frozen?
         end
 
         # Handles @allow_nan, @buffer_initial_length, other ivars must be the default value (see above)
@@ -494,8 +491,11 @@ module JSON
           # _depth_ is used to find out nesting depth, to indent accordingly.
           def to_json(state = nil, *)
             state = State.from_state(state)
+            depth = state.depth
             state.check_max_nesting
             json_transform(state)
+          ensure
+            state.depth = depth
           end
 
           private
@@ -559,17 +559,19 @@ module JSON
                     raise GeneratorError.new("#{value.class} returned by #{state.as_json} not allowed in JSON", value)
                   end
                   result << value.to_json(state)
+                  state.depth = depth
                 else
                   raise GeneratorError.new("#{value.class} not allowed in JSON", value)
                 end
               elsif value.respond_to?(:to_json)
                 result << value.to_json(state)
+                state.depth = depth
               else
                 result << %{"#{String(value)}"}
               end
               first = false
             }
-            depth = state.depth -= 1
+            depth -= 1
             unless first
               result << state.object_nl
               result << state.indent * depth if indent
@@ -586,8 +588,11 @@ module JSON
           # produced JSON string output further.
           def to_json(state = nil, *)
             state = State.from_state(state)
+            depth = state.depth
             state.check_max_nesting
             json_transform(state)
+          ensure
+            state.depth = depth
           end
 
           private
@@ -625,12 +630,13 @@ module JSON
                 end
               elsif value.respond_to?(:to_json)
                 result << value.to_json(state)
+                state.depth = depth
               else
                 result << %{"#{String(value)}"}
               end
               first = false
             }
-            depth = state.depth -= 1
+            depth -= 1
             result << state.array_nl
             result << state.indent * depth if indent
             result << ']'
@@ -655,6 +661,9 @@ module JSON
                 if casted_value.equal?(self)
                   raise GeneratorError.new("#{self} not allowed in JSON", self)
                 end
+                unless Generator.native_type?(casted_value)
+                  raise GeneratorError.new("#{casted_value.class} returned by #{state.as_json} not allowed in JSON", casted_value)
+                end
 
                 state.check_max_nesting
                 state.depth += 1

data/lib/json/version.rb

@@ -1,5 +1,5 @@
 # frozen_string_literal: true
 
 module JSON
-  VERSION = '2.16.0'
+  VERSION = '2.17.0'
 end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: json
 version: !ruby/object:Gem::Version
-  version: 2.16.0
+  version: 2.17.0
 platform: ruby
 authors:
 - Florian Frank