json 2.16.0 → 2.19.3

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

@@ -5,11 +5,11 @@
 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;
+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;
@@ -89,7 +89,7 @@ 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 +118,7 @@ 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 +132,7 @@ 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 +296,6 @@ 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 +336,8 @@ typedef struct JSON_ParserStruct {
     int max_nesting;
     bool allow_nan;
     bool allow_trailing_comma;
-    bool parsing_name;
+    bool allow_control_characters;
+    bool allow_invalid_escape;
     bool symbolize_names;
     bool freeze;
 } JSON_ParserConfig;
@@ -410,14 +402,9 @@ 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)
+static VALUE build_parse_error_message(const char *format, JSON_ParserState *state, long line, long column)
 {
     unsigned char buffer[PARSE_ERROR_FRAGMENT_LEN + 3];
-    long line, column;
-    cursor_position(state, &line, &column);
 
     const char *ptr = "EOF";
     if (state->cursor && state->cursor < state->end) {
@@ -452,17 +439,26 @@ static void raise_parse_error(const char *format, JSON_ParserState *state)
     VALUE msg = rb_sprintf(format, ptr);
     VALUE message = rb_enc_sprintf(enc_utf8, "%s at line %ld column %ld", RSTRING_PTR(msg), line, column);
     RB_GC_GUARD(msg);
+    return message;
+}
 
+static VALUE parse_error_new(VALUE message, long line, long column)
+{
     VALUE exc = rb_exc_new_str(rb_path2class("JSON::ParserError"), message);
     rb_ivar_set(exc, rb_intern("@line"), LONG2NUM(line));
     rb_ivar_set(exc, rb_intern("@column"), LONG2NUM(column));
-    rb_exc_raise(exc);
+    return 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(const char *format, JSON_ParserState *state)
+{
+    long line, column;
+    cursor_position(state, &line, &column);
+    VALUE message = build_parse_error_message(format, state, line, column);
+    rb_exc_raise(parse_error_new(message, line, column));
+}
+
+NORETURN(static) void raise_parse_error_at(const char *format, JSON_ParserState *state, const char *at)
 {
     state->cursor = at;
     raise_parse_error(format, state);
@@ -487,23 +483,24 @@ 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                          \
@@ -551,7 +548,7 @@ json_eat_comments(JSON_ParserState *state)
     }
 }
 
-static ALWAYS_INLINE() void
+ALWAYS_INLINE(static) void
 json_eat_whitespace(JSON_ParserState *state)
 {
     while (true) {
@@ -627,8 +624,10 @@ 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))) {
@@ -647,96 +646,129 @@ 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;
+    unsigned long additional_backslashes;
+} 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++;
+        if (next_position >= pe) {
+            return next_position;
+        }
+    }
+
+    if (positions->additional_backslashes) {
+        positions->additional_backslashes--;
+        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) {
-                    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);
                         }
+
+                        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;
                     }
-                    unescape_len = convert_UTF32_to_UTF8(buf, ch);
-                    unescape = buf;
                 }
+
+                int unescape_len = convert_UTF32_to_UTF8(buffer, ch);
+                buffer += unescape_len;
+                p = ++pe;
                 break;
+            }
             default:
-                p = pe;
-                continue;
+                if ((unsigned char)*pe < 0x20) {
+                    if (!config->allow_control_characters) {
+                        if (*pe == '\n') {
+                            raise_parse_error_at("Invalid unescaped newline character (\\n) in string: %s", state, pe - 1);
+                        }
+                        raise_parse_error_at("invalid ASCII control character in string: %s", state, pe - 1);
+                    }
+                }
+
+                if (config->allow_invalid_escape) {
+                    APPEND_CHAR(*pe);
+                } else {
+                    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);
@@ -754,20 +786,39 @@ static VALUE json_string_unescape(JSON_ParserState *state, const char *string, c
 }
 
 #define MAX_FAST_INTEGER_SIZE 18
+#define MAX_NUMBER_STACK_BUFFER 128
+
+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)
+{
+    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 VALUE json_decode_large_integer(const char *start, long len)
+static VALUE json_decode_inum(const char *buffer)
 {
-    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;
+    return rb_cstr2inum(buffer, 10);
 }
 
-static inline VALUE
-json_decode_integer(uint64_t mantissa, int mantissa_digits, bool negative, const char *start, const char *end)
+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) {
@@ -779,22 +830,14 @@ 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
@@ -846,7 +889,7 @@ 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`",
@@ -857,16 +900,18 @@ 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,
         rb_inspect(duplicate_key)
     );
 
+    long line, column;
+    cursor_position(state, &line, &column);
+    rb_str_concat(message, build_parse_error_message("", state, line, column)) ;
+    rb_exc_raise(parse_error_new(message, line, column));
+
     raise_parse_error(RSTRING_PTR(message), state);
     RB_GC_GUARD(message);
 }
@@ -900,20 +945,6 @@ 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 +971,7 @@ 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 +979,7 @@ 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 +987,7 @@ 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 */
@@ -964,46 +995,70 @@ 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,
+        .additional_backslashes = 0,
+    };
 
-    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.additional_backslashes++;
                 }
+                state->cursor++;
                 break;
             }
             default:
-                raise_parse_error("invalid ASCII control character in string: %s", state);
+                if (!config->allow_control_characters) {
+                    raise_parse_error("invalid ASCII control character in string: %s", state);
+                }
                 break;
         }
 
         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:
@@ -1397,14 +1452,16 @@ static int parser_config_init_i(VALUE key, VALUE val, VALUE data)
 {
     JSON_ParserConfig *config = (JSON_ParserConfig *)data;
 
-         if (key == sym_max_nesting)          { config->max_nesting = RTEST(val) ? FIX2INT(val) : 0; }
-    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_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; }
-    else if (key == sym_allow_duplicate_key)  { config->on_duplicate_key = RTEST(val) ? JSON_IGNORE : JSON_RAISE; }
-    else if (key == sym_decimal_class)        {
+         if (key == sym_max_nesting)                { config->max_nesting = RTEST(val) ? FIX2INT(val) : 0; }
+    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; }
+    else if (key == sym_allow_duplicate_key)        { config->on_duplicate_key = RTEST(val) ? JSON_IGNORE : JSON_RAISE; }
+    else if (key == sym_decimal_class)              {
         if (RTEST(val)) {
             if (rb_respond_to(val, i_try_convert)) {
                 config->decimal_class = val;
@@ -1477,6 +1534,7 @@ 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 +1630,7 @@ 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)
@@ -1616,16 +1674,14 @@ void Init_parser(void)
     sym_max_nesting = ID2SYM(rb_intern("max_nesting"));
     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"));
     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("-@");