json 2.18.0 → 2.19.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 51eab66896e862b679d424133f11e1367d5d8e71add943e67cf0673d0d562fcd
-  data.tar.gz: 7b69d4a42137897fe9a45bd60a21b759d133c112cb4ba16020099f27074ac2fd
+  metadata.gz: 2e0f63481c8ba1c4f76f44a86ac9f1814e82fb396125e41e95efddc8e259fe64
+  data.tar.gz: a070ae0776f2db0519ec672d3a32be7dee2be2ee10d0bc35c9dde6581ec8c4a8
 SHA512:
-  metadata.gz: ea3b026c8ccd6cb477858bf06f07f8b5adc5bcf7b52a175487c19dc2835ef63db2e4f87074a00ec2fe2c70e588c205f679116536da40f15e767f35351a52fc5c
-  data.tar.gz: f58144a5329ad95128e00bbc5670280f6a699e04cf05060bdfc7acaf112e62eb44c14c36328a9683d86e138c6eb2f3599f5ec35347867ac99ab9f0e16813c4df
+  metadata.gz: fc49905c26e2173018856dba7c4ae10ef74f015233b607f977b42ceea5f1d9e511f9ad4a7c22d4aed66e4ef8c60ad23a506d8bb0e053c0155ea2f7eb40a82d06
+  data.tar.gz: 61c1367d7b91621a34c0fc9fc29802534f14d27c1b9c50ecfa09d68d100fcda840847393063ec4c477437261398c559b93f292d2594bdf1e597c4b5e33959322

data/CHANGES.md

@@ -2,6 +2,16 @@
 
 ### Unreleased
 
+### 2026-03-06 (2.19.0)
+
+* Fix `allow_blank` parsing option to no longer allow invalid types (e.g. `load([], allow_blank: true)` now raise a type error).
+* Add `allow_invalid_escape` parsing option to ignore backslashes that aren't followed by one of the valid escape characters.
+
+### 2026-02-03 (2.18.1)
+
+* Fix a potential crash in very specific circumstance if GC triggers during a call to `to_json`
+  without first invoking a user defined `#to_json` method.
+
 ### 2025-12-11 (2.18.0)
 
 * Add `:allow_control_characters` parser options, to allow JSON strings containing unescaped ASCII control characters (e.g. newlines).
@@ -66,7 +76,7 @@
 * 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. 
+* 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)

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

@@ -11,11 +11,11 @@ enum fbuffer_type {
 
 typedef struct FBufferStruct {
     enum fbuffer_type type;
-    unsigned long initial_length;
-    unsigned long len;
-    unsigned long capa;
+    size_t initial_length;
+    size_t len;
+    size_t capa;
 #if JSON_DEBUG
-    unsigned long requested;
+    size_t requested;
 #endif
     char *ptr;
     VALUE io;
@@ -32,12 +32,12 @@ typedef struct FBufferStruct {
 
 static void fbuffer_free(FBuffer *fb);
 static void fbuffer_clear(FBuffer *fb);
-static void fbuffer_append(FBuffer *fb, const char *newstr, unsigned long len);
+static void fbuffer_append(FBuffer *fb, const char *newstr, size_t len);
 static void fbuffer_append_long(FBuffer *fb, long number);
 static inline void fbuffer_append_char(FBuffer *fb, char newchr);
 static VALUE fbuffer_finalize(FBuffer *fb);
 
-static void fbuffer_stack_init(FBuffer *fb, unsigned long initial_length, char *stack_buffer, long stack_buffer_size)
+static void fbuffer_stack_init(FBuffer *fb, size_t initial_length, char *stack_buffer, size_t stack_buffer_size)
 {
     fb->initial_length = (initial_length > 0) ? initial_length : FBUFFER_INITIAL_LENGTH_DEFAULT;
     if (stack_buffer) {
@@ -50,7 +50,7 @@ static void fbuffer_stack_init(FBuffer *fb, unsigned long initial_length, char *
 #endif
 }
 
-static inline void fbuffer_consumed(FBuffer *fb, unsigned long consumed)
+static inline void fbuffer_consumed(FBuffer *fb, size_t consumed)
 {
 #if JSON_DEBUG
     if (consumed > fb->requested) {
@@ -79,7 +79,7 @@ static void fbuffer_flush(FBuffer *fb)
     fbuffer_clear(fb);
 }
 
-static void fbuffer_realloc(FBuffer *fb, unsigned long required)
+static void fbuffer_realloc(FBuffer *fb, size_t required)
 {
     if (required > fb->capa) {
         if (fb->type == FBUFFER_STACK_ALLOCATED) {
@@ -94,7 +94,7 @@ static void fbuffer_realloc(FBuffer *fb, unsigned long required)
     }
 }
 
-static void fbuffer_do_inc_capa(FBuffer *fb, unsigned long requested)
+static void fbuffer_do_inc_capa(FBuffer *fb, size_t requested)
 {
     if (RB_UNLIKELY(fb->io)) {
         if (fb->capa < FBUFFER_IO_BUFFER_SIZE) {
@@ -108,7 +108,7 @@ static void fbuffer_do_inc_capa(FBuffer *fb, unsigned long requested)
         }
     }
 
-    unsigned long required;
+    size_t required;
 
     if (RB_UNLIKELY(!fb->ptr)) {
         fb->ptr = ALLOC_N(char, fb->initial_length);
@@ -120,7 +120,7 @@ static void fbuffer_do_inc_capa(FBuffer *fb, unsigned long requested)
     fbuffer_realloc(fb, required);
 }
 
-static inline void fbuffer_inc_capa(FBuffer *fb, unsigned long requested)
+static inline void fbuffer_inc_capa(FBuffer *fb, size_t requested)
 {
 #if JSON_DEBUG
     fb->requested = requested;
@@ -131,13 +131,13 @@ static inline void fbuffer_inc_capa(FBuffer *fb, unsigned long requested)
     }
 }
 
-static inline void fbuffer_append_reserved(FBuffer *fb, const char *newstr, unsigned long len)
+static inline void fbuffer_append_reserved(FBuffer *fb, const char *newstr, size_t 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)
+static inline void fbuffer_append(FBuffer *fb, const char *newstr, size_t len)
 {
     if (len > 0) {
         fbuffer_inc_capa(fb, len);
@@ -161,23 +161,25 @@ static inline void fbuffer_append_reserved_char(FBuffer *fb, char chr)
 
 static void fbuffer_append_str(FBuffer *fb, VALUE str)
 {
-    const char *newstr = StringValuePtr(str);
-    unsigned long len = RSTRING_LEN(str);
+    const char *ptr;
+    size_t len;
+    RSTRING_GETMEM(str, ptr, len);
 
-    fbuffer_append(fb, newstr, len);
+    fbuffer_append(fb, ptr, 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);
+    const char *ptr;
+    size_t len;
+    RSTRING_GETMEM(str, ptr, len);
 
     fbuffer_inc_capa(fb, repeat * len);
     while (repeat) {
 #if JSON_DEBUG
         fb->requested = len;
 #endif
-        fbuffer_append_reserved(fb, newstr, len);
+        fbuffer_append_reserved(fb, ptr, len);
         repeat--;
     }
 }

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

@@ -63,6 +63,8 @@ struct generate_json_data {
     long depth;
 };
 
+static SIMD_Implementation simd_impl;
+
 static VALUE cState_from_state_s(VALUE self, VALUE opts);
 static VALUE cState_partial_generate(VALUE self, VALUE obj, generator_func, VALUE io);
 static void generate_json(FBuffer *buffer, struct generate_json_data *data, VALUE obj);
@@ -72,9 +74,6 @@ static void generate_json_string(FBuffer *buffer, struct generate_json_data *dat
 static void generate_json_null(FBuffer *buffer, struct generate_json_data *data, VALUE obj);
 static void generate_json_false(FBuffer *buffer, struct generate_json_data *data, VALUE obj);
 static void generate_json_true(FBuffer *buffer, struct generate_json_data *data, VALUE obj);
-#ifdef RUBY_INTEGER_UNIFICATION
-static void generate_json_integer(FBuffer *buffer, struct generate_json_data *data, VALUE obj);
-#endif
 static void generate_json_fixnum(FBuffer *buffer, struct generate_json_data *data, VALUE obj);
 static void generate_json_bignum(FBuffer *buffer, struct generate_json_data *data, VALUE obj);
 static void generate_json_float(FBuffer *buffer, struct generate_json_data *data, VALUE obj);
@@ -155,8 +154,6 @@ static const unsigned char escape_table_basic[256] = {
      0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
 };
 
-static unsigned char (*search_escape_basic_impl)(search_state *);
-
 static inline unsigned char search_escape_basic(search_state *search)
 {
     while (search->ptr < search->end) {
@@ -212,11 +209,39 @@ ALWAYS_INLINE(static) void escape_UTF8_char_basic(search_state *search)
  * Everything else (should be UTF-8) is just passed through and
  * appended to the result.
  */
+
+
+#if defined(HAVE_SIMD_NEON)
+static inline unsigned char search_escape_basic_neon(search_state *search);
+#elif defined(HAVE_SIMD_SSE2)
+static inline unsigned char search_escape_basic_sse2(search_state *search);
+#endif
+
+static inline unsigned char search_escape_basic(search_state *search);
+
 static inline void convert_UTF8_to_JSON(search_state *search)
 {
-    while (search_escape_basic_impl(search)) {
+#ifdef HAVE_SIMD
+#if defined(HAVE_SIMD_NEON)
+    while (search_escape_basic_neon(search)) {
         escape_UTF8_char_basic(search);
     }
+#elif defined(HAVE_SIMD_SSE2)
+    if (simd_impl == SIMD_SSE2) {
+        while (search_escape_basic_sse2(search)) {
+            escape_UTF8_char_basic(search);
+        }
+        return;
+    }
+    while (search_escape_basic(search)) {
+        escape_UTF8_char_basic(search);
+    }
+#endif
+#else
+    while (search_escape_basic(search)) {
+        escape_UTF8_char_basic(search);
+    }
+#endif /* HAVE_SIMD */
 }
 
 static inline void escape_UTF8_char(search_state *search, unsigned char ch_len)
@@ -260,6 +285,8 @@ static inline void escape_UTF8_char(search_state *search, unsigned char ch_len)
 
 ALWAYS_INLINE(static) char *copy_remaining_bytes(search_state *search, unsigned long vec_len, unsigned long len)
 {
+    RBIMPL_ASSERT_OR_ASSUME(len < vec_len);
+
     // Flush the buffer so everything up until the last 'len' characters are unflushed.
     search_flush(search);
 
@@ -269,12 +296,18 @@ ALWAYS_INLINE(static) char *copy_remaining_bytes(search_state *search, unsigned
     char *s = (buf->ptr + buf->len);
 
     // Pad the buffer with dummy characters that won't need escaping.
-    // This seem wateful at first sight, but memset of vector length is very fast.
-    memset(s, 'X', vec_len);
+    // This seem wasteful at first sight, but memset of vector length is very fast.
+    // This is a space as it can be directly represented as an immediate on AArch64.
+    memset(s, ' ', vec_len);
 
     // Optimistically copy the remaining 'len' characters to the output FBuffer. If there are no characters
     // to escape, then everything ends up in the correct spot. Otherwise it was convenient temporary storage.
-    MEMCPY(s, search->ptr, char, len);
+    if (vec_len == 16) {
+        RBIMPL_ASSERT_OR_ASSUME(len >= SIMD_MINIMUM_THRESHOLD);
+        json_fast_memcpy16(s, search->ptr, len);
+    } else {
+        MEMCPY(s, search->ptr, char, len);
+    }
 
     return s;
 }
@@ -667,233 +700,6 @@ static void convert_UTF8_to_ASCII_only_JSON(search_state *search, const unsigned
     }
 }
 
-/*
- * Document-module: JSON::Ext::Generator
- *
- * This is the JSON generator implemented as a C extension. It can be
- * configured to be used by setting
- *
- *  JSON.generator = JSON::Ext::Generator
- *
- * with the method generator= in JSON.
- *
- */
-
-/* Explanation of the following: that's the only way to not pollute
- * standard library's docs with GeneratorMethods::<ClassName> which
- * are uninformative and take a large place in a list of classes
- */
-
-/*
- * Document-module: JSON::Ext::Generator::GeneratorMethods
- * :nodoc:
- */
-
-/*
- * Document-module: JSON::Ext::Generator::GeneratorMethods::Array
- * :nodoc:
- */
-
-/*
- * Document-module: JSON::Ext::Generator::GeneratorMethods::Bignum
- * :nodoc:
- */
-
-/*
- * Document-module: JSON::Ext::Generator::GeneratorMethods::FalseClass
- * :nodoc:
- */
-
-/*
- * Document-module: JSON::Ext::Generator::GeneratorMethods::Fixnum
- * :nodoc:
- */
-
-/*
- * Document-module: JSON::Ext::Generator::GeneratorMethods::Float
- * :nodoc:
- */
-
-/*
- * Document-module: JSON::Ext::Generator::GeneratorMethods::Hash
- * :nodoc:
- */
-
-/*
- * Document-module: JSON::Ext::Generator::GeneratorMethods::Integer
- * :nodoc:
- */
-
-/*
- * Document-module: JSON::Ext::Generator::GeneratorMethods::NilClass
- * :nodoc:
- */
-
-/*
- * Document-module: JSON::Ext::Generator::GeneratorMethods::Object
- * :nodoc:
- */
-
-/*
- * Document-module: JSON::Ext::Generator::GeneratorMethods::String
- * :nodoc:
- */
-
-/*
- * Document-module: JSON::Ext::Generator::GeneratorMethods::String::Extend
- * :nodoc:
- */
-
-/*
- * Document-module: JSON::Ext::Generator::GeneratorMethods::TrueClass
- * :nodoc:
- */
-
-/*
- * call-seq: to_json(state = nil)
- *
- * Returns a JSON string containing a JSON object, that is generated from
- * this Hash instance.
- * _state_ is a JSON::State object, that can also be used to configure the
- * produced JSON string output further.
- */
-static VALUE mHash_to_json(int argc, VALUE *argv, VALUE self)
-{
-    rb_check_arity(argc, 0, 1);
-    VALUE Vstate = cState_from_state_s(cState, argc == 1 ? argv[0] : Qnil);
-    return cState_partial_generate(Vstate, self, generate_json_object, Qfalse);
-}
-
-/*
- * call-seq: to_json(state = nil)
- *
- * Returns a JSON string containing a JSON array, that is generated from
- * this Array instance.
- * _state_ is a JSON::State object, that can also be used to configure the
- * produced JSON string output further.
- */
-static VALUE mArray_to_json(int argc, VALUE *argv, VALUE self)
-{
-    rb_check_arity(argc, 0, 1);
-    VALUE Vstate = cState_from_state_s(cState, argc == 1 ? argv[0] : Qnil);
-    return cState_partial_generate(Vstate, self, generate_json_array, Qfalse);
-}
-
-#ifdef RUBY_INTEGER_UNIFICATION
-/*
- * call-seq: to_json(*)
- *
- * Returns a JSON string representation for this Integer number.
- */
-static VALUE mInteger_to_json(int argc, VALUE *argv, VALUE self)
-{
-    rb_check_arity(argc, 0, 1);
-    VALUE Vstate = cState_from_state_s(cState, argc == 1 ? argv[0] : Qnil);
-    return cState_partial_generate(Vstate, self, generate_json_integer, Qfalse);
-}
-
-#else
-/*
- * call-seq: to_json(*)
- *
- * Returns a JSON string representation for this Integer number.
- */
-static VALUE mFixnum_to_json(int argc, VALUE *argv, VALUE self)
-{
-    rb_check_arity(argc, 0, 1);
-    VALUE Vstate = cState_from_state_s(cState, argc == 1 ? argv[0] : Qnil);
-    return cState_partial_generate(Vstate, self, generate_json_fixnum, Qfalse);
-}
-
-/*
- * call-seq: to_json(*)
- *
- * Returns a JSON string representation for this Integer number.
- */
-static VALUE mBignum_to_json(int argc, VALUE *argv, VALUE self)
-{
-    rb_check_arity(argc, 0, 1);
-    VALUE Vstate = cState_from_state_s(cState, argc == 1 ? argv[0] : Qnil);
-    return cState_partial_generate(Vstate, self, generate_json_bignum, Qfalse);
-}
-#endif
-
-/*
- * call-seq: to_json(*)
- *
- * Returns a JSON string representation for this Float number.
- */
-static VALUE mFloat_to_json(int argc, VALUE *argv, VALUE self)
-{
-    rb_check_arity(argc, 0, 1);
-    VALUE Vstate = cState_from_state_s(cState, argc == 1 ? argv[0] : Qnil);
-    return cState_partial_generate(Vstate, self, generate_json_float, Qfalse);
-}
-
-/*
- * call-seq: to_json(*)
- *
- * This string should be encoded with UTF-8 A call to this method
- * returns a JSON string encoded with UTF16 big endian characters as
- * \u????.
- */
-static VALUE mString_to_json(int argc, VALUE *argv, VALUE self)
-{
-    rb_check_arity(argc, 0, 1);
-    VALUE Vstate = cState_from_state_s(cState, argc == 1 ? argv[0] : Qnil);
-    return cState_partial_generate(Vstate, self, generate_json_string, Qfalse);
-}
-
-/*
- * call-seq: to_json(*)
- *
- * Returns a JSON string for true: 'true'.
- */
-static VALUE mTrueClass_to_json(int argc, VALUE *argv, VALUE self)
-{
-    rb_check_arity(argc, 0, 1);
-    return rb_utf8_str_new("true", 4);
-}
-
-/*
- * call-seq: to_json(*)
- *
- * Returns a JSON string for false: 'false'.
- */
-static VALUE mFalseClass_to_json(int argc, VALUE *argv, VALUE self)
-{
-    rb_check_arity(argc, 0, 1);
-    return rb_utf8_str_new("false", 5);
-}
-
-/*
- * call-seq: to_json(*)
- *
- * Returns a JSON string for nil: 'null'.
- */
-static VALUE mNilClass_to_json(int argc, VALUE *argv, VALUE self)
-{
-    rb_check_arity(argc, 0, 1);
-    return rb_utf8_str_new("null", 4);
-}
-
-/*
- * call-seq: to_json(*)
- *
- * Converts this object to a string (calling #to_s), converts
- * it to a JSON string, and returns the result. This is a fallback, if no
- * special method #to_json was defined for some object.
- */
-static VALUE mObject_to_json(int argc, VALUE *argv, VALUE self)
-{
-    VALUE state;
-    VALUE string = rb_funcall(self, i_to_s, 0);
-    rb_scan_args(argc, argv, "01", &state);
-    Check_Type(string, T_STRING);
-    state = cState_from_state_s(cState, state);
-    return cState_partial_generate(state, string, generate_json_string, Qfalse);
-}
-
 static void State_mark(void *ptr)
 {
     JSON_Generator_State *state = ptr;
@@ -1091,6 +897,7 @@ static void raw_generate_json_string(FBuffer *buffer, struct generate_json_data
     search.matches_mask = 0;
     search.has_matches = false;
     search.chunk_base = NULL;
+    search.chunk_end = NULL;
 #endif /* HAVE_SIMD */
 
     switch (rb_enc_str_coderange(obj)) {
@@ -1337,18 +1144,8 @@ static void generate_json_fixnum(FBuffer *buffer, struct generate_json_data *dat
 static void generate_json_bignum(FBuffer *buffer, struct generate_json_data *data, VALUE obj)
 {
     VALUE tmp = rb_funcall(obj, i_to_s, 0);
-    fbuffer_append_str(buffer, tmp);
-}
-
-#ifdef RUBY_INTEGER_UNIFICATION
-static void generate_json_integer(FBuffer *buffer, struct generate_json_data *data, VALUE obj)
-{
-    if (FIXNUM_P(obj))
-        generate_json_fixnum(buffer, data, obj);
-    else
-        generate_json_bignum(buffer, data, obj);
+    fbuffer_append_str(buffer, StringValue(tmp));
 }
-#endif
 
 static void generate_json_float(FBuffer *buffer, struct generate_json_data *data, VALUE obj)
 {
@@ -1393,7 +1190,7 @@ static void generate_json_fragment(FBuffer *buffer, struct generate_json_data *d
     fbuffer_append_str(buffer, fragment);
 }
 
-static void generate_json(FBuffer *buffer, struct generate_json_data *data, VALUE obj)
+static inline void generate_json_general(FBuffer *buffer, struct generate_json_data *data, VALUE obj, bool fallback)
 {
     bool as_json_called = false;
 start:
@@ -1420,15 +1217,15 @@ start:
                 generate_json_bignum(buffer, data, obj);
                 break;
             case T_HASH:
-                if (klass != rb_cHash) goto general;
+                if (fallback && klass != rb_cHash) goto general;
                 generate_json_object(buffer, data, obj);
                 break;
             case T_ARRAY:
-                if (klass != rb_cArray) goto general;
+                if (fallback && klass != rb_cArray) goto general;
                 generate_json_array(buffer, data, obj);
                 break;
             case T_STRING:
-                if (klass != rb_cString) goto general;
+                if (fallback && klass != rb_cString) goto general;
 
                 if (RB_LIKELY(valid_json_string_p(obj))) {
                     raw_generate_json_string(buffer, data, obj);
@@ -1444,7 +1241,7 @@ start:
                 generate_json_symbol(buffer, data, obj);
                 break;
             case T_FLOAT:
-                if (klass != rb_cFloat) goto general;
+                if (fallback && klass != rb_cFloat) goto general;
                 generate_json_float(buffer, data, obj);
                 break;
             case T_STRUCT:
@@ -1468,6 +1265,16 @@ start:
     }
 }
 
+static void generate_json(FBuffer *buffer, struct generate_json_data *data, VALUE obj)
+{
+    generate_json_general(buffer, data, obj, true);
+}
+
+static void generate_json_no_fallback(FBuffer *buffer, struct generate_json_data *data, VALUE obj)
+{
+    generate_json_general(buffer, data, obj, false);
+}
+
 static VALUE generate_json_try(VALUE d)
 {
     struct generate_json_data *data = (struct generate_json_data *)d;
@@ -1485,7 +1292,7 @@ static VALUE generate_json_ensure(VALUE d)
     return Qundef;
 }
 
-static VALUE cState_partial_generate(VALUE self, VALUE obj, generator_func func, VALUE io)
+static inline VALUE cState_partial_generate(VALUE self, VALUE obj, generator_func func, VALUE io)
 {
     GET_STATE(self);
 
@@ -1522,6 +1329,15 @@ static VALUE cState_generate(int argc, VALUE *argv, VALUE self)
     return cState_partial_generate(self, obj, generate_json, io);
 }
 
+/* :nodoc: */
+static VALUE cState_generate_no_fallback(int argc, VALUE *argv, VALUE self)
+{
+    rb_check_arity(argc, 1, 2);
+    VALUE obj = argv[0];
+    VALUE io = argc > 1 ? argv[1] : Qnil;
+    return cState_partial_generate(self, obj, generate_json_no_fallback, io);
+}
+
 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`");
@@ -1759,6 +1575,17 @@ static long long_config(VALUE num)
     return RTEST(num) ? FIX2LONG(num) : 0;
 }
 
+// depth must never be negative; reject early with a clear error.
+static long depth_config(VALUE num)
+{
+    if (!RTEST(num)) return 0;
+    long d = NUM2LONG(num);
+    if (RB_UNLIKELY(d < 0)) {
+        rb_raise(rb_eArgError, "depth must be >= 0 (got %ld)", d);
+    }
+    return d;
+}
+
 /*
  * call-seq: max_nesting=(depth)
  *
@@ -1915,7 +1742,7 @@ static VALUE cState_depth_set(VALUE self, VALUE depth)
 {
     rb_check_frozen(self);
     GET_STATE(self);
-    state->depth = long_config(depth);
+    state->depth = depth_config(depth);
     return Qnil;
 }
 
@@ -1980,7 +1807,7 @@ static int configure_state_i(VALUE key, VALUE val, VALUE _arg)
     else if (key == sym_max_nesting)           { state->max_nesting = long_config(val); }
     else if (key == sym_allow_nan)             { state->allow_nan = RTEST(val); }
     else if (key == sym_ascii_only)            { state->ascii_only = RTEST(val); }
-    else if (key == sym_depth)                 { state->depth = long_config(val); }
+    else if (key == sym_depth)                 { state->depth = depth_config(val); }
     else if (key == sym_buffer_initial_length) { buffer_initial_length_set(state, val); }
     else if (key == sym_script_safe)           { state->script_safe = RTEST(val); }
     else if (key == sym_escape_slash)          { state->script_safe = RTEST(val); }
@@ -2020,7 +1847,7 @@ static VALUE cState_configure(VALUE self, VALUE opts)
     return self;
 }
 
-static VALUE cState_m_generate(VALUE klass, VALUE obj, VALUE opts, VALUE io)
+static VALUE cState_m_do_generate(VALUE klass, VALUE obj, VALUE opts, VALUE io, generator_func func)
 {
     JSON_Generator_State state = {0};
     state_init(&state);
@@ -2038,14 +1865,21 @@ static VALUE cState_m_generate(VALUE klass, VALUE obj, VALUE opts, VALUE io)
         .state = &state,
         .depth = state.depth,
         .obj = obj,
-        .func = generate_json,
+        .func = func,
     };
     return rb_ensure(generate_json_try, (VALUE)&data, generate_json_ensure, (VALUE)&data);
 }
 
-/*
- *
- */
+static VALUE cState_m_generate(VALUE klass, VALUE obj, VALUE opts, VALUE io)
+{
+    return cState_m_do_generate(klass, obj, opts, io, generate_json);
+}
+
+static VALUE cState_m_generate_no_fallback(VALUE klass, VALUE obj, VALUE opts, VALUE io)
+{
+    return cState_m_do_generate(klass, obj, opts, io, generate_json_no_fallback);
+}
+
 void Init_generator(void)
 {
 #ifdef HAVE_RB_EXT_RACTOR_SAFE
@@ -2110,46 +1944,12 @@ 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_no_fallback", cState_generate_no_fallback, -1);
 
     rb_define_private_method(cState, "allow_duplicate_key?", cState_allow_duplicate_key_p, 0);
 
     rb_define_singleton_method(cState, "generate", cState_m_generate, 3);
-
-    VALUE mGeneratorMethods = rb_define_module_under(mGenerator, "GeneratorMethods");
-
-    VALUE mObject = rb_define_module_under(mGeneratorMethods, "Object");
-    rb_define_method(mObject, "to_json", mObject_to_json, -1);
-
-    VALUE mHash = rb_define_module_under(mGeneratorMethods, "Hash");
-    rb_define_method(mHash, "to_json", mHash_to_json, -1);
-
-    VALUE mArray = rb_define_module_under(mGeneratorMethods, "Array");
-    rb_define_method(mArray, "to_json", mArray_to_json, -1);
-
-#ifdef RUBY_INTEGER_UNIFICATION
-    VALUE mInteger = rb_define_module_under(mGeneratorMethods, "Integer");
-    rb_define_method(mInteger, "to_json", mInteger_to_json, -1);
-#else
-    VALUE mFixnum = rb_define_module_under(mGeneratorMethods, "Fixnum");
-    rb_define_method(mFixnum, "to_json", mFixnum_to_json, -1);
-
-    VALUE mBignum = rb_define_module_under(mGeneratorMethods, "Bignum");
-    rb_define_method(mBignum, "to_json", mBignum_to_json, -1);
-#endif
-    VALUE mFloat = rb_define_module_under(mGeneratorMethods, "Float");
-    rb_define_method(mFloat, "to_json", mFloat_to_json, -1);
-
-    VALUE mString = rb_define_module_under(mGeneratorMethods, "String");
-    rb_define_method(mString, "to_json", mString_to_json, -1);
-
-    VALUE mTrueClass = rb_define_module_under(mGeneratorMethods, "TrueClass");
-    rb_define_method(mTrueClass, "to_json", mTrueClass_to_json, -1);
-
-    VALUE mFalseClass = rb_define_module_under(mGeneratorMethods, "FalseClass");
-    rb_define_method(mFalseClass, "to_json", mFalseClass_to_json, -1);
-
-    VALUE mNilClass = rb_define_module_under(mGeneratorMethods, "NilClass");
-    rb_define_method(mNilClass, "to_json", mNilClass_to_json, -1);
+    rb_define_singleton_method(cState, "_generate_no_fallback", cState_m_generate_no_fallback, 3);
 
     rb_global_variable(&Encoding_UTF_8);
     Encoding_UTF_8 = rb_const_get(rb_path2class("Encoding"), rb_intern("UTF_8"));
@@ -2181,22 +1981,5 @@ void Init_generator(void)
 
     rb_require("json/ext/generator/state");
 
-
-    switch (find_simd_implementation()) {
-#ifdef HAVE_SIMD
-#ifdef HAVE_SIMD_NEON
-        case SIMD_NEON:
-            search_escape_basic_impl = search_escape_basic_neon;
-            break;
-#endif /* HAVE_SIMD_NEON */
-#ifdef HAVE_SIMD_SSE2
-        case SIMD_SSE2:
-            search_escape_basic_impl = search_escape_basic_sse2;
-            break;
-#endif /* HAVE_SIMD_SSE2 */
-#endif /* HAVE_SIMD */
-        default:
-            search_escape_basic_impl = search_escape_basic;
-            break;
-    }
+    simd_impl = find_simd_implementation();
 }

data/ext/json/ext/json.h

@@ -5,6 +5,10 @@
 #include "ruby/encoding.h"
 #include <stdint.h>
 
+#ifndef RBIMPL_ASSERT_OR_ASSUME
+# define RBIMPL_ASSERT_OR_ASSUME(x)
+#endif
+
 #if defined(RUBY_DEBUG) && RUBY_DEBUG
 # define JSON_ASSERT RUBY_ASSERT
 #else
@@ -51,8 +55,12 @@ typedef unsigned char _Bool;
 #endif
 
 #ifndef NORETURN
+#if defined(__has_attribute) && __has_attribute(noreturn)
+#define NORETURN(x) __attribute__((noreturn)) x
+#else
 #define NORETURN(x) x
 #endif
+#endif
 
 #ifndef NOINLINE
 #if defined(__has_attribute) && __has_attribute(noinline)

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

@@ -7,8 +7,9 @@ static VALUE CNaN, CInfinity, CMinusInfinity;
 
 static ID i_new, i_try_convert, i_uminus, i_encode;
 
-static VALUE sym_max_nesting, sym_allow_nan, sym_allow_trailing_comma, sym_allow_control_characters, sym_symbolize_names, sym_freeze,
-             sym_decimal_class, sym_on_load, sym_allow_duplicate_key;
+static VALUE sym_max_nesting, sym_allow_nan, sym_allow_trailing_comma, sym_allow_control_characters,
+             sym_allow_invalid_escape, sym_symbolize_names, sym_freeze, sym_decimal_class, sym_on_load,
+             sym_allow_duplicate_key;
 
 static int binary_encindex;
 static int utf8_encindex;
@@ -336,6 +337,7 @@ typedef struct JSON_ParserStruct {
     bool allow_nan;
     bool allow_trailing_comma;
     bool allow_control_characters;
+    bool allow_invalid_escape;
     bool symbolize_names;
     bool freeze;
 } JSON_ParserConfig;
@@ -400,10 +402,7 @@ static void emit_parse_warning(const char *message, JSON_ParserState *state)
 
 #define PARSE_ERROR_FRAGMENT_LEN 32
 
-#ifdef RBIMPL_ATTR_NORETURN
-RBIMPL_ATTR_NORETURN()
-#endif
-static void raise_parse_error(const char *format, JSON_ParserState *state)
+NORETURN(static) void raise_parse_error(const char *format, JSON_ParserState *state)
 {
     unsigned char buffer[PARSE_ERROR_FRAGMENT_LEN + 3];
     long line, column;
@@ -449,10 +448,7 @@ static void raise_parse_error(const char *format, JSON_ParserState *state)
     rb_exc_raise(exc);
 }
 
-#ifdef RBIMPL_ATTR_NORETURN
-RBIMPL_ATTR_NORETURN()
-#endif
-static void raise_parse_error_at(const char *format, JSON_ParserState *state, const char *at)
+NORETURN(static) void raise_parse_error_at(const char *format, JSON_ParserState *state, const char *at)
 {
     state->cursor = at;
     raise_parse_error(format, state);
@@ -477,23 +473,24 @@ static const signed char digit_values[256] = {
     -1, -1, -1, -1, -1, -1, -1
 };
 
-static uint32_t unescape_unicode(JSON_ParserState *state, const unsigned char *p)
-{
-    signed char b;
-    uint32_t result = 0;
-    b = digit_values[p[0]];
-    if (b < 0) raise_parse_error_at("incomplete unicode character escape sequence at %s", state, (char *)p - 2);
-    result = (result << 4) | (unsigned char)b;
-    b = digit_values[p[1]];
-    if (b < 0) raise_parse_error_at("incomplete unicode character escape sequence at %s", state, (char *)p - 2);
-    result = (result << 4) | (unsigned char)b;
-    b = digit_values[p[2]];
-    if (b < 0) raise_parse_error_at("incomplete unicode character escape sequence at %s", state, (char *)p - 2);
-    result = (result << 4) | (unsigned char)b;
-    b = digit_values[p[3]];
-    if (b < 0) raise_parse_error_at("incomplete unicode character escape sequence at %s", state, (char *)p - 2);
-    result = (result << 4) | (unsigned char)b;
-    return result;
+static uint32_t unescape_unicode(JSON_ParserState *state, const char *sp, const char *spe)
+{
+    if (RB_UNLIKELY(sp > spe - 4)) {
+        raise_parse_error_at("incomplete unicode character escape sequence at %s", state, sp - 2);
+    }
+
+    const unsigned char *p = (const unsigned char *)sp;
+
+    const signed char b0 = digit_values[p[0]];
+    const signed char b1 = digit_values[p[1]];
+    const signed char b2 = digit_values[p[2]];
+    const signed char b3 = digit_values[p[3]];
+
+    if (RB_UNLIKELY((signed char)(b0 | b1 | b2 | b3) < 0)) {
+        raise_parse_error_at("incomplete unicode character escape sequence at %s", state, sp - 2);
+    }
+
+    return ((uint32_t)b0 << 12) | ((uint32_t)b1 << 8) | ((uint32_t)b2 << 4) | (uint32_t)b3;
 }
 
 #define GET_PARSER_CONFIG                          \
@@ -643,7 +640,7 @@ static inline VALUE json_string_fastpath(JSON_ParserState *state, JSON_ParserCon
 typedef struct _json_unescape_positions {
     long size;
     const char **positions;
-    bool has_more;
+    unsigned long additional_backslashes;
 } JSON_UnescapePositions;
 
 static inline const char *json_next_backslash(const char *pe, const char *stringEnd, JSON_UnescapePositions *positions)
@@ -657,7 +654,8 @@ static inline const char *json_next_backslash(const char *pe, const char *string
         }
     }
 
-    if (positions->has_more) {
+    if (positions->additional_backslashes) {
+        positions->additional_backslashes--;
         return memchr(pe, '\\', stringEnd - pe);
     }
 
@@ -707,50 +705,41 @@ NOINLINE(static) VALUE json_string_unescape(JSON_ParserState *state, JSON_Parser
             case 'f':
                 APPEND_CHAR('\f');
                 break;
-            case 'u':
-                if (pe > stringEnd - 5) {
-                    raise_parse_error_at("incomplete unicode character escape sequence at %s", state, p);
-                } else {
-                    uint32_t ch = unescape_unicode(state, (unsigned char *) ++pe);
-                    pe += 3;
-                    /* To handle values above U+FFFF, we take a sequence of
-                     * \uXXXX escapes in the U+D800..U+DBFF then
-                     * U+DC00..U+DFFF ranges, take the low 10 bits from each
-                     * to make a 20-bit number, then add 0x10000 to get the
-                     * final codepoint.
-                     *
-                     * See Unicode 15: 3.8 "Surrogates", 5.3 "Handling
-                     * Surrogate Pairs in UTF-16", and 23.6 "Surrogates
-                     * Area".
-                     */
-                    if ((ch & 0xFC00) == 0xD800) {
-                        pe++;
-                        if (pe > stringEnd - 6) {
-                            raise_parse_error_at("incomplete surrogate pair at %s", state, p);
-                        }
-                        if (pe[0] == '\\' && pe[1] == 'u') {
-                            uint32_t sur = unescape_unicode(state, (unsigned char *) pe + 2);
-
-                            if ((sur & 0xFC00) != 0xDC00) {
-                                raise_parse_error_at("invalid surrogate pair at %s", state, p);
-                            }
-
-                            ch = (((ch & 0x3F) << 10) | ((((ch >> 6) & 0xF) + 1) << 16)
-                                    | (sur & 0x3FF));
-                            pe += 5;
-                        } else {
-                            raise_parse_error_at("incomplete surrogate pair at %s", state, p);
-                            break;
+            case 'u': {
+                uint32_t ch = unescape_unicode(state, ++pe, stringEnd);
+                pe += 3;
+                /* To handle values above U+FFFF, we take a sequence of
+                 * \uXXXX escapes in the U+D800..U+DBFF then
+                 * U+DC00..U+DFFF ranges, take the low 10 bits from each
+                 * to make a 20-bit number, then add 0x10000 to get the
+                 * final codepoint.
+                 *
+                 * See Unicode 15: 3.8 "Surrogates", 5.3 "Handling
+                 * Surrogate Pairs in UTF-16", and 23.6 "Surrogates
+                 * Area".
+                 */
+                if ((ch & 0xFC00) == 0xD800) {
+                    pe++;
+                    if (RB_LIKELY((pe <= stringEnd - 6) && memcmp(pe, "\\u", 2) == 0)) {
+                        uint32_t sur = unescape_unicode(state, pe + 2, stringEnd);
+
+                        if (RB_UNLIKELY((sur & 0xFC00) != 0xDC00)) {
+                            raise_parse_error_at("invalid surrogate pair at %s", state, p);
                         }
-                    }
 
-                    char buf[4];
-                    int unescape_len = convert_UTF32_to_UTF8(buf, ch);
-                    MEMCPY(buffer, buf, char, unescape_len);
-                    buffer += unescape_len;
-                    p = ++pe;
+                        ch = (((ch & 0x3F) << 10) | ((((ch >> 6) & 0xF) + 1) << 16) | (sur & 0x3FF));
+                        pe += 5;
+                    } else {
+                        raise_parse_error_at("incomplete surrogate pair at %s", state, p);
+                        break;
+                    }
                 }
+
+                int unescape_len = convert_UTF32_to_UTF8(buffer, ch);
+                buffer += unescape_len;
+                p = ++pe;
                 break;
+            }
             default:
                 if ((unsigned char)*pe < 0x20) {
                     if (!config->allow_control_characters) {
@@ -759,6 +748,8 @@ NOINLINE(static) VALUE json_string_unescape(JSON_ParserState *state, JSON_Parser
                         }
                         raise_parse_error_at("invalid ASCII control character in string: %s", state, pe - 1);
                     }
+                } else if (config->allow_invalid_escape) {
+                    APPEND_CHAR(*pe);
                 } else {
                     raise_parse_error_at("invalid escape character in string: %s", state, pe - 1);
                 }
@@ -783,20 +774,39 @@ NOINLINE(static) VALUE json_string_unescape(JSON_ParserState *state, JSON_Parser
 }
 
 #define MAX_FAST_INTEGER_SIZE 18
+#define MAX_NUMBER_STACK_BUFFER 128
 
-static VALUE json_decode_large_integer(const char *start, long len)
+typedef VALUE (*json_number_decode_func_t)(const char *ptr);
+
+static inline VALUE json_decode_large_number(const char *start, long len, json_number_decode_func_t func)
 {
-    VALUE buffer_v;
-    char *buffer = RB_ALLOCV_N(char, buffer_v, len + 1);
-    MEMCPY(buffer, start, char, len);
-    buffer[len] = '\0';
-    VALUE number = rb_cstr2inum(buffer, 10);
-    RB_ALLOCV_END(buffer_v);
-    return number;
+    if (RB_LIKELY(len < MAX_NUMBER_STACK_BUFFER)) {
+        char buffer[MAX_NUMBER_STACK_BUFFER];
+        MEMCPY(buffer, start, char, len);
+        buffer[len] = '\0';
+        return func(buffer);
+    } else {
+        VALUE buffer_v = rb_str_tmp_new(len);
+        char *buffer = RSTRING_PTR(buffer_v);
+        MEMCPY(buffer, start, char, len);
+        buffer[len] = '\0';
+        VALUE number = func(buffer);
+        RB_GC_GUARD(buffer_v);
+        return number;
+    }
 }
 
-static inline VALUE
-json_decode_integer(uint64_t mantissa, int mantissa_digits, bool negative, const char *start, const char *end)
+static VALUE json_decode_inum(const char *buffer)
+{
+    return rb_cstr2inum(buffer, 10);
+}
+
+NOINLINE(static) VALUE json_decode_large_integer(const char *start, long len)
+{
+    return json_decode_large_number(start, len, json_decode_inum);
+}
+
+static inline VALUE json_decode_integer(uint64_t mantissa, int mantissa_digits, bool negative, const char *start, const char *end)
 {
     if (RB_LIKELY(mantissa_digits < MAX_FAST_INTEGER_SIZE)) {
         if (negative) {
@@ -808,22 +818,14 @@ json_decode_integer(uint64_t mantissa, int mantissa_digits, bool negative, const
     return json_decode_large_integer(start, end - start);
 }
 
-static VALUE json_decode_large_float(const char *start, long len)
+static VALUE json_decode_dnum(const char *buffer)
 {
-    if (RB_LIKELY(len < 64)) {
-        char buffer[64];
-        MEMCPY(buffer, start, char, len);
-        buffer[len] = '\0';
-        return DBL2NUM(rb_cstr_to_dbl(buffer, 1));
-    }
+    return DBL2NUM(rb_cstr_to_dbl(buffer, 1));
+}
 
-    VALUE buffer_v;
-    char *buffer = RB_ALLOCV_N(char, buffer_v, len + 1);
-    MEMCPY(buffer, start, char, len);
-    buffer[len] = '\0';
-    VALUE number = DBL2NUM(rb_cstr_to_dbl(buffer, 1));
-    RB_ALLOCV_END(buffer_v);
-    return number;
+NOINLINE(static) VALUE json_decode_large_float(const char *start, long len)
+{
+    return json_decode_large_number(start, len, json_decode_dnum);
 }
 
 /* Ruby JSON optimized float decoder using vendored Ryu algorithm
@@ -875,7 +877,7 @@ static VALUE json_find_duplicated_key(size_t count, const VALUE *pairs)
     return Qfalse;
 }
 
-static void emit_duplicate_key_warning(JSON_ParserState *state, VALUE duplicate_key)
+NOINLINE(static) void emit_duplicate_key_warning(JSON_ParserState *state, VALUE duplicate_key)
 {
     VALUE message = rb_sprintf(
         "detected duplicate key %"PRIsVALUE" in JSON object. This will raise an error in json 3.0 unless enabled via `allow_duplicate_key: true`",
@@ -886,10 +888,7 @@ static void emit_duplicate_key_warning(JSON_ParserState *state, VALUE duplicate_
     RB_GC_GUARD(message);
 }
 
-#ifdef RBIMPL_ATTR_NORETURN
-RBIMPL_ATTR_NORETURN()
-#endif
-static void raise_duplicate_key_error(JSON_ParserState *state, VALUE duplicate_key)
+NORETURN(static) void raise_duplicate_key_error(JSON_ParserState *state, VALUE duplicate_key)
 {
     VALUE message = rb_sprintf(
         "duplicate key %"PRIsVALUE,
@@ -992,7 +991,7 @@ static VALUE json_parse_escaped_string(JSON_ParserState *state, JSON_ParserConfi
     JSON_UnescapePositions positions = {
         .size = 0,
         .positions = backslashes,
-        .has_more = false,
+        .additional_backslashes = 0,
     };
 
     do {
@@ -1007,7 +1006,7 @@ static VALUE json_parse_escaped_string(JSON_ParserState *state, JSON_ParserConfi
                     backslashes[positions.size] = state->cursor;
                     positions.size++;
                 } else {
-                    positions.has_more = true;
+                    positions.additional_backslashes++;
                 }
                 state->cursor++;
                 break;
@@ -1440,6 +1439,7 @@ static int parser_config_init_i(VALUE key, VALUE val, VALUE data)
     else if (key == sym_allow_nan)                  { config->allow_nan = RTEST(val); }
     else if (key == sym_allow_trailing_comma)       { config->allow_trailing_comma = RTEST(val); }
     else if (key == sym_allow_control_characters)   { config->allow_control_characters = RTEST(val); }
+    else if (key == sym_allow_invalid_escape)       { config->allow_invalid_escape = RTEST(val); }
     else if (key == sym_symbolize_names)            { config->symbolize_names = RTEST(val); }
     else if (key == sym_freeze)                     { config->freeze = RTEST(val); }
     else if (key == sym_on_load)                    { config->on_load_proc = RTEST(val) ? val : Qfalse; }
@@ -1658,6 +1658,7 @@ void Init_parser(void)
     sym_allow_nan = ID2SYM(rb_intern("allow_nan"));
     sym_allow_trailing_comma = ID2SYM(rb_intern("allow_trailing_comma"));
     sym_allow_control_characters = ID2SYM(rb_intern("allow_control_characters"));
+    sym_allow_invalid_escape = ID2SYM(rb_intern("allow_invalid_escape"));
     sym_symbolize_names = ID2SYM(rb_intern("symbolize_names"));
     sym_freeze = ID2SYM(rb_intern("freeze"));
     sym_on_load = ID2SYM(rb_intern("on_load"));

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

@@ -58,7 +58,34 @@ static inline int trailing_zeros(int input)
 
 #ifdef JSON_ENABLE_SIMD
 
-#define SIMD_MINIMUM_THRESHOLD 6
+#define SIMD_MINIMUM_THRESHOLD 4
+
+ALWAYS_INLINE(static) void json_fast_memcpy16(char *dst, const char *src, size_t len)
+{
+    RBIMPL_ASSERT_OR_ASSUME(len < 16);
+    RBIMPL_ASSERT_OR_ASSUME(len >= SIMD_MINIMUM_THRESHOLD); // 4
+#if defined(__has_builtin) && __has_builtin(__builtin_memcpy)
+    // If __builtin_memcpy is available, use it to copy between SIMD_MINIMUM_THRESHOLD (4) and vec_len-1 (15) bytes.
+    // These copies overlap. The first copy will copy the first 8 (or 4) bytes. The second copy will copy
+    // the last 8 (or 4) bytes but overlap with the first copy. The overlapping bytes will be in the correct
+    // position in both copies.
+
+    // Please do not attempt to replace __builtin_memcpy with memcpy without profiling and/or looking at the
+    // generated assembly. On clang-specifically (tested on Apple clang version 17.0.0 (clang-1700.0.13.3)),
+    // when using memcpy, the compiler will notice the only difference is a 4 or 8 and generate a conditional
+    // select instruction instead of direct loads and stores with a branch. This ends up slower than the branch
+    // plus two loads and stores generated when using __builtin_memcpy.
+    if (len >= 8) {
+        __builtin_memcpy(dst, src, 8);
+        __builtin_memcpy(dst + len - 8, src + len - 8, 8);
+    } else {
+        __builtin_memcpy(dst, src, 4);
+        __builtin_memcpy(dst + len - 4, src + len - 4, 4);
+    }
+#else
+    MEMCPY(dst, src, char, len);
+#endif
+}
 
 #if defined(__ARM_NEON) || defined(__ARM_NEON__) || defined(__aarch64__) || defined(_M_ARM64)
 #include <arm_neon.h>
@@ -106,16 +133,6 @@ ALWAYS_INLINE(static) int string_scan_simd_neon(const char **ptr, const char *en
     return 0;
 }
 
-static inline uint8x16x4_t load_uint8x16_4(const unsigned char *table)
-{
-    uint8x16x4_t tab;
-    tab.val[0] = vld1q_u8(table);
-    tab.val[1] = vld1q_u8(table+16);
-    tab.val[2] = vld1q_u8(table+32);
-    tab.val[3] = vld1q_u8(table+48);
-    return tab;
-}
-
 #endif /* ARM Neon Support.*/
 
 #if defined(__amd64__) || defined(__amd64) || defined(__x86_64__) || defined(__x86_64) || defined(_M_X64) || defined(_M_AMD64)

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

@@ -449,7 +449,7 @@ static int filter_special(double fp, char* dest)
  * }
  *
  */
-static int fpconv_dtoa(double d, char dest[28])
+static int fpconv_dtoa(double d, char dest[32])
 {
     char digits[18];
 

data/lib/json/common.rb

@@ -156,15 +156,17 @@ module JSON
     def generator=(generator) # :nodoc:
       old, $VERBOSE = $VERBOSE, nil
       @generator = generator
-      generator_methods = generator::GeneratorMethods
-      for const in generator_methods.constants
-        klass = const_get(const)
-        modul = generator_methods.const_get(const)
-        klass.class_eval do
-          instance_methods(false).each do |m|
-            m.to_s == 'to_json' and remove_method m
+      if generator.const_defined?(:GeneratorMethods)
+        generator_methods = generator::GeneratorMethods
+        for const in generator_methods.constants
+          klass = const_get(const)
+          modul = generator_methods.const_get(const)
+          klass.class_eval do
+            instance_methods(false).each do |m|
+              m.to_s == 'to_json' and remove_method m
+            end
+            include modul
           end
-          include modul
         end
       end
       self.state = generator::State
@@ -878,7 +880,7 @@ module JSON
       end
     end
 
-    if opts[:allow_blank] && (source.nil? || source.empty?)
+    if opts[:allow_blank] && (source.nil? || (String === source && source.empty?))
       source = 'null'
     end
 
@@ -1036,7 +1038,8 @@ module JSON
     #   JSON.new(options = nil, &block)
     #
     # Argument +options+, if given, contains a \Hash of options for both parsing and generating.
-    # See {Parsing Options}[#module-JSON-label-Parsing+Options], and {Generating Options}[#module-JSON-label-Generating+Options].
+    # See {Parsing Options}[rdoc-ref:JSON@Parsing+Options],
+    # and {Generating Options}[rdoc-ref:JSON@Generating+Options].
     #
     # For generation, the <tt>strict: true</tt> option is always set. When a Ruby object with no native \JSON counterpart is
     # encountered, the block provided to the initialize method is invoked, and must return a Ruby object that has a native
@@ -1095,6 +1098,30 @@ module JSON
       load(File.read(path, encoding: Encoding::UTF_8))
     end
   end
+
+  module GeneratorMethods
+    # call-seq: to_json(*)
+    #
+    # Converts this object into a JSON string.
+    # If this object doesn't directly maps to a JSON native type,
+    # first convert it to a string (calling #to_s), then converts
+    # it to a JSON string, and returns the result.
+    # This is a fallback, if no special method #to_json was defined for some object.
+    def to_json(state = nil, *)
+      obj = case self
+      when nil, false, true, Integer, Float, Array, Hash
+        self
+      else
+        "#{self}"
+      end
+
+      if state.nil?
+        JSON::State._generate_no_fallback(obj, nil, nil)
+      else
+        JSON::State.from_state(state)._generate_no_fallback(obj)
+      end
+    end
+  end
 end
 
 module ::Kernel
@@ -1140,3 +1167,7 @@ module ::Kernel
     JSON[object, opts]
   end
 end
+
+class Object
+  include JSON::GeneratorMethods
+end

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

@@ -9,7 +9,7 @@ module JSON
         # Instantiates a new State object, configured by _opts_.
         #
         # Argument +opts+, if given, contains a \Hash of options for the generation.
-        # See {Generating Options}[#module-JSON-label-Generating+Options].
+        # See {Generating Options}[rdoc-ref:JSON@Generating+Options].
         def initialize(opts = nil)
           if opts && !opts.empty?
             configure(opts)

data/lib/json/truffle_ruby/generator.rb

@@ -211,7 +211,14 @@ module JSON
 
         # This integer returns the current depth data structure nesting in the
         # generated JSON.
-        attr_accessor :depth
+        attr_reader :depth
+
+        def depth=(depth)
+          if depth.negative?
+            raise ArgumentError, "depth must be >= 0 (got #{depth})"
+          end
+          @depth = depth
+        end
 
         def check_max_nesting # :nodoc:
           return if @max_nesting.zero?
@@ -260,6 +267,11 @@ module JSON
           else
             raise TypeError, "can't convert #{opts.class} into Hash"
           end
+
+          if opts[:depth]&.negative?
+            raise ArgumentError, "depth must be >= 0 (got #{opts[:depth]})"
+          end
+
           opts.each do |key, value|
             instance_variable_set "@#{key}", value
           end

data/lib/json/version.rb

@@ -1,5 +1,5 @@
 # frozen_string_literal: true
 
 module JSON
-  VERSION = '2.18.0'
+  VERSION = '2.19.0'
 end

data/lib/json.rb

@@ -6,6 +6,15 @@ require 'json/common'
 #
 # \JSON is a lightweight data-interchange format.
 #
+# \JSON is easy for us humans to read and write,
+# and equally simple for machines to read (parse) and write (generate).
+#
+# \JSON is language-independent, making it an ideal interchange format
+# for applications in differing programming languages
+# and on differing operating systems.
+#
+# == \JSON Values
+#
 # A \JSON value is one of the following:
 # - Double-quoted text:  <tt>"foo"</tt>.
 # - Number:  +1+, +1.0+, +2.0e2+.
@@ -173,6 +182,30 @@ require 'json/common'
 # When enabled:
 #   JSON.parse('[1,]', allow_trailing_comma: true) # => [1]
 #
+# ---
+#
+# Option +allow_control_characters+ (boolean) specifies whether to allow
+# unescaped ASCII control characters, such as newlines, in strings;
+# defaults to +false+.
+#
+# With the default, +false+:
+#   JSON.parse(%{"Hello\nWorld"}) # invalid ASCII control character in string (JSON::ParserError)
+#
+# When enabled:
+#   JSON.parse(%{"Hello\nWorld"}, allow_control_characters: true) # => "Hello\nWorld"
+#
+# ---
+#
+# Option +allow_invalid_escape+ (boolean) specifies whether to ignore backslahes that are followed
+# by an invalid escape character in strings;
+# defaults to +false+.
+#
+# With the default, +false+:
+#   JSON.parse('"Hell\o"') # invalid escape character in string (JSON::ParserError)
+#
+# When enabled:
+#   JSON.parse('"Hell\o"', allow_invalid_escape: true) # => "Hello"
+#
 # ====== Output Options
 #
 # Option +freeze+ (boolean) specifies whether the returned objects will be frozen;

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: json
 version: !ruby/object:Gem::Version
-  version: 2.18.0
+  version: 2.19.0
 platform: ruby
 authors:
 - Florian Frank
@@ -84,7 +84,7 @@ required_rubygems_version: !ruby/object:Gem::Requirement
     - !ruby/object:Gem::Version
       version: '0'
 requirements: []
-rubygems_version: 3.6.9
+rubygems_version: 4.0.3
 specification_version: 4
 summary: JSON Implementation for Ruby
 test_files: []