diff options
Diffstat (limited to 'vendor/bundle/ruby/3.4.0/gems/json-2.12.2/ext')
10 files changed, 5329 insertions, 0 deletions
diff --git a/vendor/bundle/ruby/3.4.0/gems/json-2.12.2/ext/json/ext/fbuffer/fbuffer.h b/vendor/bundle/ruby/3.4.0/gems/json-2.12.2/ext/json/ext/fbuffer/fbuffer.h new file mode 100644 index 0000000..d323714 --- /dev/null +++ b/vendor/bundle/ruby/3.4.0/gems/json-2.12.2/ext/json/ext/fbuffer/fbuffer.h @@ -0,0 +1,270 @@ +#ifndef _FBUFFER_H_ +#define _FBUFFER_H_ + +#include "ruby.h" +#include "ruby/encoding.h" +#include "../vendor/jeaiii-ltoa.h" + +/* shims */ +/* This is the fallback definition from Ruby 3.4 */ + +#ifndef RBIMPL_STDBOOL_H +#if defined(__cplusplus) +# if defined(HAVE_STDBOOL_H) && (__cplusplus >= 201103L) +# include <cstdbool> +# endif +#elif defined(HAVE_STDBOOL_H) +# include <stdbool.h> +#elif !defined(HAVE__BOOL) +typedef unsigned char _Bool; +# define bool _Bool +# define true ((_Bool)+1) +# define false ((_Bool)+0) +# define __bool_true_false_are_defined +#endif +#endif + +#ifndef RB_UNLIKELY +#define RB_UNLIKELY(expr) expr +#endif + +#ifndef RB_LIKELY +#define RB_LIKELY(expr) expr +#endif + +#ifndef MAYBE_UNUSED +# define MAYBE_UNUSED(x) x +#endif + +#ifdef RUBY_DEBUG +#ifndef JSON_DEBUG +#define JSON_DEBUG RUBY_DEBUG +#endif +#endif + +enum fbuffer_type { + FBUFFER_HEAP_ALLOCATED = 0, + FBUFFER_STACK_ALLOCATED = 1, +}; + +typedef struct FBufferStruct { + enum fbuffer_type type; + unsigned long initial_length; + unsigned long len; + unsigned long capa; +#ifdef JSON_DEBUG + unsigned long requested; +#endif + char *ptr; + VALUE io; +} FBuffer; + +#define FBUFFER_STACK_SIZE 512 +#define FBUFFER_IO_BUFFER_SIZE (16384 - 1) +#define FBUFFER_INITIAL_LENGTH_DEFAULT 1024 + +#define FBUFFER_PTR(fb) ((fb)->ptr) +#define FBUFFER_LEN(fb) ((fb)->len) +#define FBUFFER_CAPA(fb) ((fb)->capa) +#define FBUFFER_PAIR(fb) FBUFFER_PTR(fb), FBUFFER_LEN(fb) + +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_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) +{ + fb->initial_length = (initial_length > 0) ? initial_length : FBUFFER_INITIAL_LENGTH_DEFAULT; + if (stack_buffer) { + fb->type = FBUFFER_STACK_ALLOCATED; + fb->ptr = stack_buffer; + fb->capa = stack_buffer_size; + } +#ifdef JSON_DEBUG + fb->requested = 0; +#endif +} + +static inline void fbuffer_consumed(FBuffer *fb, unsigned long consumed) +{ +#ifdef JSON_DEBUG + if (consumed > fb->requested) { + rb_bug("fbuffer: Out of bound write"); + } + fb->requested = 0; +#endif + fb->len += consumed; +} + +static void fbuffer_free(FBuffer *fb) +{ + if (fb->ptr && fb->type == FBUFFER_HEAP_ALLOCATED) { + ruby_xfree(fb->ptr); + } +} + +static void fbuffer_clear(FBuffer *fb) +{ + fb->len = 0; +} + +static void fbuffer_flush(FBuffer *fb) +{ + rb_io_write(fb->io, rb_utf8_str_new(fb->ptr, fb->len)); + fbuffer_clear(fb); +} + +static void fbuffer_realloc(FBuffer *fb, unsigned long required) +{ + if (required > fb->capa) { + if (fb->type == FBUFFER_STACK_ALLOCATED) { + const char *old_buffer = fb->ptr; + fb->ptr = ALLOC_N(char, required); + fb->type = FBUFFER_HEAP_ALLOCATED; + MEMCPY(fb->ptr, old_buffer, char, fb->len); + } else { + REALLOC_N(fb->ptr, char, required); + } + fb->capa = required; + } +} + +static void fbuffer_do_inc_capa(FBuffer *fb, unsigned long requested) +{ + if (RB_UNLIKELY(fb->io)) { + if (fb->capa < FBUFFER_IO_BUFFER_SIZE) { + fbuffer_realloc(fb, FBUFFER_IO_BUFFER_SIZE); + } else { + fbuffer_flush(fb); + } + + if (RB_LIKELY(requested < fb->capa)) { + return; + } + } + + unsigned long required; + + if (RB_UNLIKELY(!fb->ptr)) { + fb->ptr = ALLOC_N(char, fb->initial_length); + fb->capa = fb->initial_length; + } + + for (required = fb->capa; requested > required - fb->len; required <<= 1); + + fbuffer_realloc(fb, required); +} + +static inline void fbuffer_inc_capa(FBuffer *fb, unsigned long requested) +{ +#ifdef JSON_DEBUG + fb->requested = requested; +#endif + + if (RB_UNLIKELY(requested > fb->capa - fb->len)) { + fbuffer_do_inc_capa(fb, requested); + } +} + +static void fbuffer_append(FBuffer *fb, const char *newstr, unsigned long len) +{ + if (len > 0) { + fbuffer_inc_capa(fb, len); + MEMCPY(fb->ptr + fb->len, newstr, char, len); + fbuffer_consumed(fb, len); + } +} + +/* Appends a character into a buffer. The buffer needs to have sufficient capacity, via fbuffer_inc_capa(...). */ +static inline void fbuffer_append_reserved_char(FBuffer *fb, char chr) +{ +#ifdef JSON_DEBUG + if (fb->requested < 1) { + rb_bug("fbuffer: unreserved write"); + } + fb->requested--; +#endif + + fb->ptr[fb->len] = chr; + fb->len++; +} + +static void fbuffer_append_str(FBuffer *fb, VALUE str) +{ + const char *newstr = StringValuePtr(str); + unsigned long len = RSTRING_LEN(str); + + RB_GC_GUARD(str); + + fbuffer_append(fb, newstr, len); +} + +static inline void fbuffer_append_char(FBuffer *fb, char newchr) +{ + fbuffer_inc_capa(fb, 1); + *(fb->ptr + fb->len) = newchr; + fbuffer_consumed(fb, 1); +} + +static inline char *fbuffer_cursor(FBuffer *fb) +{ + return fb->ptr + fb->len; +} + +static inline void fbuffer_advance_to(FBuffer *fb, char *end) +{ + fbuffer_consumed(fb, (end - fb->ptr) - fb->len); +} + +/* + * Appends the decimal string representation of \a number into the buffer. + */ +static void fbuffer_append_long(FBuffer *fb, long number) +{ + /* + * The jeaiii_ultoa() function produces digits left-to-right, + * allowing us to write directly into the buffer, but we don't know + * the number of resulting characters. + * + * We do know, however, that the `number` argument is always in the + * range 0xc000000000000000 to 0x3fffffffffffffff, or, in decimal, + * -4611686018427387904 to 4611686018427387903. The max number of chars + * generated is therefore 20 (including a potential sign character). + */ + + static const int MAX_CHARS_FOR_LONG = 20; + + fbuffer_inc_capa(fb, MAX_CHARS_FOR_LONG); + + if (number < 0) { + fbuffer_append_reserved_char(fb, '-'); + + /* + * Since number is always > LONG_MIN, `-number` will not overflow + * and is always the positive abs() value. + */ + number = -number; + } + + char *end = jeaiii_ultoa(fbuffer_cursor(fb), number); + fbuffer_advance_to(fb, end); +} + +static VALUE fbuffer_finalize(FBuffer *fb) +{ + if (fb->io) { + fbuffer_flush(fb); + fbuffer_free(fb); + rb_io_flush(fb->io); + return fb->io; + } else { + VALUE result = rb_utf8_str_new(FBUFFER_PTR(fb), FBUFFER_LEN(fb)); + fbuffer_free(fb); + return result; + } +} + +#endif diff --git a/vendor/bundle/ruby/3.4.0/gems/json-2.12.2/ext/json/ext/generator/Makefile b/vendor/bundle/ruby/3.4.0/gems/json-2.12.2/ext/json/ext/generator/Makefile new file mode 100644 index 0000000..37f250a --- /dev/null +++ b/vendor/bundle/ruby/3.4.0/gems/json-2.12.2/ext/json/ext/generator/Makefile @@ -0,0 +1,273 @@ + +SHELL = /bin/sh + +# V=0 quiet, V=1 verbose. other values don't work. +V = 0 +V0 = $(V:0=) +Q1 = $(V:1=) +Q = $(Q1:0=@) +ECHO1 = $(V:1=@ :) +ECHO = $(ECHO1:0=@ echo) +NULLCMD = : + +#### Start of system configuration section. #### + +srcdir = . +topdir = /Users/ben/.rubies/ruby-3.4.1/include/ruby-3.4.0 +hdrdir = $(topdir) +arch_hdrdir = /Users/ben/.rubies/ruby-3.4.1/include/ruby-3.4.0/arm64-darwin24 +PATH_SEPARATOR = : +VPATH = $(srcdir):$(arch_hdrdir)/ruby:$(hdrdir)/ruby +prefix = $(DESTDIR)/Users/ben/.rubies/ruby-3.4.1 +rubysitearchprefix = $(rubylibprefix)/$(sitearch) +rubyarchprefix = $(rubylibprefix)/$(arch) +rubylibprefix = $(libdir)/$(RUBY_BASE_NAME) +exec_prefix = $(prefix) +vendorarchhdrdir = $(vendorhdrdir)/$(sitearch) +sitearchhdrdir = $(sitehdrdir)/$(sitearch) +rubyarchhdrdir = $(rubyhdrdir)/$(arch) +vendorhdrdir = $(rubyhdrdir)/vendor_ruby +sitehdrdir = $(rubyhdrdir)/site_ruby +rubyhdrdir = $(includedir)/$(RUBY_VERSION_NAME) +vendorarchdir = $(vendorlibdir)/$(sitearch) +vendorlibdir = $(vendordir)/$(ruby_version) +vendordir = $(rubylibprefix)/vendor_ruby +sitearchdir = $(sitelibdir)/$(sitearch) +sitelibdir = $(sitedir)/$(ruby_version) +sitedir = $(rubylibprefix)/site_ruby +rubyarchdir = $(rubylibdir)/$(arch) +rubylibdir = $(rubylibprefix)/$(ruby_version) +sitearchincludedir = $(includedir)/$(sitearch) +archincludedir = $(includedir)/$(arch) +sitearchlibdir = $(libdir)/$(sitearch) +archlibdir = $(libdir)/$(arch) +ridir = $(datarootdir)/$(RI_BASE_NAME) +modular_gc_dir = $(DESTDIR) +mandir = $(datarootdir)/man +localedir = $(datarootdir)/locale +libdir = $(exec_prefix)/lib +psdir = $(docdir) +pdfdir = $(docdir) +dvidir = $(docdir) +htmldir = $(docdir) +infodir = $(datarootdir)/info +docdir = $(datarootdir)/doc/$(PACKAGE) +oldincludedir = $(DESTDIR)/usr/include +includedir = $(SDKROOT)$(prefix)/include +runstatedir = $(localstatedir)/run +localstatedir = $(prefix)/var +sharedstatedir = $(prefix)/com +sysconfdir = $(prefix)/etc +datadir = $(datarootdir) +datarootdir = $(prefix)/share +libexecdir = $(exec_prefix)/libexec +sbindir = $(exec_prefix)/sbin +bindir = $(exec_prefix)/bin +archdir = $(rubyarchdir) + + +CC_WRAPPER = +CC = clang +CXX = clang++ -std=gnu++11 +LIBRUBY = $(LIBRUBY_A) +LIBRUBY_A = lib$(RUBY_SO_NAME)-static.a +LIBRUBYARG_SHARED = +LIBRUBYARG_STATIC = -l$(RUBY_SO_NAME)-static -framework CoreFoundation $(MAINLIBS) +empty = +OUTFLAG = -o $(empty) +COUTFLAG = -o $(empty) +CSRCFLAG = $(empty) + +RUBY_EXTCONF_H = +cflags = $(hardenflags) -fdeclspec $(optflags) $(debugflags) $(warnflags) +cxxflags = +optflags = -O3 -fno-fast-math +debugflags = -ggdb3 +warnflags = -Wall -Wextra -Wextra-tokens -Wdeprecated-declarations -Wdivision-by-zero -Wdiv-by-zero -Wimplicit-function-declaration -Wimplicit-int -Wpointer-arith -Wshorten-64-to-32 -Wwrite-strings -Wold-style-definition -Wmissing-noreturn -Wno-cast-function-type -Wno-constant-logical-operand -Wno-long-long -Wno-missing-field-initializers -Wno-overlength-strings -Wno-parentheses-equality -Wno-self-assign -Wno-tautological-compare -Wno-unused-parameter -Wno-unused-value -Wunused-variable -Wmisleading-indentation -Wundef +cppflags = +CCDLFLAGS = -fno-common +CFLAGS = $(CCDLFLAGS) $(cflags) -pipe -std=c99 $(ARCH_FLAG) +INCFLAGS = -I. -I$(arch_hdrdir) -I$(hdrdir)/ruby/backward -I$(hdrdir) -I$(srcdir) +DEFS = +CPPFLAGS = -DJSON_GENERATOR -DHAVE_ARM_NEON_H -DHAVE_TYPE_UINT8X16_T -DJSON_ENABLE_SIMD -I/opt/homebrew/opt/readline/include -I/opt/homebrew/opt/libyaml/include -I/opt/homebrew/opt/gdbm/include -D_XOPEN_SOURCE -D_DARWIN_C_SOURCE -D_DARWIN_UNLIMITED_SELECT -D_REENTRANT $(DEFS) $(cppflags) +CXXFLAGS = $(CCDLFLAGS) -fdeclspec $(ARCH_FLAG) +ldflags = -L. -fstack-protector-strong -L/opt/homebrew/Cellar/readline/8.2.13/lib -L/opt/homebrew/Cellar/libyaml/0.2.5/lib -L/opt/homebrew/Cellar/gdbm/1.24/lib +dldflags = -L/opt/homebrew/Cellar/readline/8.2.13/lib -L/opt/homebrew/Cellar/libyaml/0.2.5/lib -L/opt/homebrew/Cellar/gdbm/1.24/lib -Wl,-undefined,dynamic_lookup -bundle_loader '$(BUILTRUBY)' +ARCH_FLAG = -arch arm64 +DLDFLAGS = $(ldflags) $(dldflags) $(ARCH_FLAG) +LDSHARED = $(CC) -dynamic -bundle +LDSHAREDXX = $(CXX) -dynamic -bundle +POSTLINK = dsymutil $@ 2>/dev/null; { test -z '$(RUBY_CODESIGN)' || codesign -s '$(RUBY_CODESIGN)' $@; } +AR = ar +LD = ld +EXEEXT = + +RUBY_INSTALL_NAME = $(RUBY_BASE_NAME) +RUBY_SO_NAME = ruby.3.4 +RUBYW_INSTALL_NAME = +RUBY_VERSION_NAME = $(RUBY_BASE_NAME)-$(ruby_version) +RUBYW_BASE_NAME = rubyw +RUBY_BASE_NAME = ruby + +arch = arm64-darwin24 +sitearch = $(arch) +ruby_version = 3.4.0 +ruby = $(bindir)/$(RUBY_BASE_NAME) +RUBY = $(ruby) +BUILTRUBY = $(bindir)/$(RUBY_BASE_NAME) +ruby_headers = $(hdrdir)/ruby.h $(hdrdir)/ruby/backward.h $(hdrdir)/ruby/ruby.h $(hdrdir)/ruby/defines.h $(hdrdir)/ruby/missing.h $(hdrdir)/ruby/intern.h $(hdrdir)/ruby/st.h $(hdrdir)/ruby/subst.h $(arch_hdrdir)/ruby/config.h + +RM = rm -f +RM_RF = rm -fr +RMDIRS = rmdir -p +MAKEDIRS = mkdir -p +INSTALL = /usr/bin/install -c +INSTALL_PROG = $(INSTALL) -m 0755 +INSTALL_DATA = $(INSTALL) -m 644 +COPY = cp +TOUCH = exit > + +#### End of system configuration section. #### + +preload = +libpath = . $(libdir) /opt/homebrew/opt/readline/lib /opt/homebrew/opt/libyaml/lib /opt/homebrew/opt/gdbm/lib +LIBPATH = -L. -L$(libdir) -L/opt/homebrew/opt/readline/lib -L/opt/homebrew/opt/libyaml/lib -L/opt/homebrew/opt/gdbm/lib +DEFFILE = + +CLEANFILES = mkmf.log +DISTCLEANFILES = +DISTCLEANDIRS = + +extout = +extout_prefix = +target_prefix = /json/ext +LOCAL_LIBS = +LIBS = -lpthread +ORIG_SRCS = generator.c +SRCS = $(ORIG_SRCS) +OBJS = generator.o +HDRS = $(srcdir)/simd.h +LOCAL_HDRS = +TARGET = generator +TARGET_NAME = generator +TARGET_ENTRY = Init_$(TARGET_NAME) +DLLIB = $(TARGET).bundle +EXTSTATIC = +STATIC_LIB = + +TIMESTAMP_DIR = . +BINDIR = $(bindir) +RUBYCOMMONDIR = $(sitedir)$(target_prefix) +RUBYLIBDIR = $(sitelibdir)$(target_prefix) +RUBYARCHDIR = $(sitearchdir)$(target_prefix) +HDRDIR = $(sitehdrdir)$(target_prefix) +ARCHHDRDIR = $(sitearchhdrdir)$(target_prefix) +TARGET_SO_DIR = +TARGET_SO = $(TARGET_SO_DIR)$(DLLIB) +CLEANLIBS = $(TARGET_SO) $(TARGET_SO:=.dSYM) +CLEANOBJS = $(OBJS) *.bak +TARGET_SO_DIR_TIMESTAMP = $(TIMESTAMP_DIR)/.sitearchdir.-.json.-.ext.time + +all: $(DLLIB) +static: $(STATIC_LIB) +.PHONY: all install static install-so install-rb +.PHONY: clean clean-so clean-static clean-rb + +clean-static:: +clean-rb-default:: +clean-rb:: +clean-so:: +clean: clean-so clean-static clean-rb-default clean-rb + -$(Q)$(RM_RF) $(CLEANLIBS) $(CLEANOBJS) $(CLEANFILES) .*.time + +distclean-rb-default:: +distclean-rb:: +distclean-so:: +distclean-static:: +distclean: clean distclean-so distclean-static distclean-rb-default distclean-rb + -$(Q)$(RM) Makefile $(RUBY_EXTCONF_H) conftest.* mkmf.log + -$(Q)$(RM) core ruby$(EXEEXT) *~ $(DISTCLEANFILES) + -$(Q)$(RMDIRS) $(DISTCLEANDIRS) 2> /dev/null || true + +realclean: distclean +install: install-so install-rb + +install-so: $(DLLIB) $(TARGET_SO_DIR_TIMESTAMP) + $(INSTALL_PROG) $(DLLIB) $(RUBYARCHDIR) +clean-static:: + -$(Q)$(RM) $(STATIC_LIB) +install-rb: pre-install-rb do-install-rb install-rb-default +install-rb-default: pre-install-rb-default do-install-rb-default +pre-install-rb: Makefile +pre-install-rb-default: Makefile +do-install-rb: +do-install-rb-default: +pre-install-rb-default: + @$(NULLCMD) +$(TARGET_SO_DIR_TIMESTAMP): + $(Q) $(MAKEDIRS) $(@D) $(RUBYARCHDIR) + $(Q) $(TOUCH) $@ + +site-install: site-install-so site-install-rb +site-install-so: install-so +site-install-rb: install-rb + +.SUFFIXES: .c .m .cc .mm .cxx .cpp .o .S + +.cc.o: + $(ECHO) compiling $(<) + $(Q) $(CXX) $(INCFLAGS) $(CPPFLAGS) $(CXXFLAGS) $(COUTFLAG)$@ -c $(CSRCFLAG)$< + +.cc.S: + $(ECHO) translating $(<) + $(Q) $(CXX) $(INCFLAGS) $(CPPFLAGS) $(CXXFLAGS) $(COUTFLAG)$@ -S $(CSRCFLAG)$< + +.mm.o: + $(ECHO) compiling $(<) + $(Q) $(CXX) $(INCFLAGS) $(CPPFLAGS) $(CXXFLAGS) $(COUTFLAG)$@ -c $(CSRCFLAG)$< + +.mm.S: + $(ECHO) translating $(<) + $(Q) $(CXX) $(INCFLAGS) $(CPPFLAGS) $(CXXFLAGS) $(COUTFLAG)$@ -S $(CSRCFLAG)$< + +.cxx.o: + $(ECHO) compiling $(<) + $(Q) $(CXX) $(INCFLAGS) $(CPPFLAGS) $(CXXFLAGS) $(COUTFLAG)$@ -c $(CSRCFLAG)$< + +.cxx.S: + $(ECHO) translating $(<) + $(Q) $(CXX) $(INCFLAGS) $(CPPFLAGS) $(CXXFLAGS) $(COUTFLAG)$@ -S $(CSRCFLAG)$< + +.cpp.o: + $(ECHO) compiling $(<) + $(Q) $(CXX) $(INCFLAGS) $(CPPFLAGS) $(CXXFLAGS) $(COUTFLAG)$@ -c $(CSRCFLAG)$< + +.cpp.S: + $(ECHO) translating $(<) + $(Q) $(CXX) $(INCFLAGS) $(CPPFLAGS) $(CXXFLAGS) $(COUTFLAG)$@ -S $(CSRCFLAG)$< + +.c.o: + $(ECHO) compiling $(<) + $(Q) $(CC) $(INCFLAGS) $(CPPFLAGS) $(CFLAGS) $(COUTFLAG)$@ -c $(CSRCFLAG)$< + +.c.S: + $(ECHO) translating $(<) + $(Q) $(CC) $(INCFLAGS) $(CPPFLAGS) $(CFLAGS) $(COUTFLAG)$@ -S $(CSRCFLAG)$< + +.m.o: + $(ECHO) compiling $(<) + $(Q) $(CC) $(INCFLAGS) $(CPPFLAGS) $(CFLAGS) $(COUTFLAG)$@ -c $(CSRCFLAG)$< + +.m.S: + $(ECHO) translating $(<) + $(Q) $(CC) $(INCFLAGS) $(CPPFLAGS) $(CFLAGS) $(COUTFLAG)$@ -S $(CSRCFLAG)$< + +$(TARGET_SO): $(OBJS) Makefile + $(ECHO) linking shared-object json/ext/$(DLLIB) + -$(Q)$(RM) $(@) + $(Q) $(LDSHARED) -o $@ $(OBJS) $(LIBPATH) $(DLDFLAGS) $(LOCAL_LIBS) $(LIBS) + $(Q) $(POSTLINK) + + + +$(OBJS): $(HDRS) $(ruby_headers) diff --git a/vendor/bundle/ruby/3.4.0/gems/json-2.12.2/ext/json/ext/generator/extconf.rb b/vendor/bundle/ruby/3.4.0/gems/json-2.12.2/ext/json/ext/generator/extconf.rb new file mode 100644 index 0000000..f58574a --- /dev/null +++ b/vendor/bundle/ruby/3.4.0/gems/json-2.12.2/ext/json/ext/generator/extconf.rb @@ -0,0 +1,40 @@ +require 'mkmf' + +if RUBY_ENGINE == 'truffleruby' + # The pure-Ruby generator is faster on TruffleRuby, so skip compiling the generator extension + File.write('Makefile', dummy_makefile("").join) +else + append_cflags("-std=c99") + $defs << "-DJSON_GENERATOR" + $defs << "-DJSON_DEBUG" if ENV["JSON_DEBUG"] + + if enable_config('generator-use-simd', default=!ENV["JSON_DISABLE_SIMD"]) + if RbConfig::CONFIG['host_cpu'] =~ /^(arm.*|aarch64.*)/ + # Try to compile a small program using NEON instructions + if have_header('arm_neon.h') + have_type('uint8x16_t', headers=['arm_neon.h']) && try_compile(<<~'SRC') + #include <arm_neon.h> + int main() { + uint8x16_t test = vdupq_n_u8(32); + return 0; + } + SRC + $defs.push("-DJSON_ENABLE_SIMD") + end + end + + if have_header('x86intrin.h') && have_type('__m128i', headers=['x86intrin.h']) && try_compile(<<~'SRC') + #include <x86intrin.h> + int main() { + __m128i test = _mm_set1_epi8(32); + return 0; + } + SRC + $defs.push("-DJSON_ENABLE_SIMD") + end + + have_header('cpuid.h') + end + + create_makefile 'json/ext/generator' +end diff --git a/vendor/bundle/ruby/3.4.0/gems/json-2.12.2/ext/json/ext/generator/generator.c b/vendor/bundle/ruby/3.4.0/gems/json-2.12.2/ext/json/ext/generator/generator.c new file mode 100644 index 0000000..f7690a2 --- /dev/null +++ b/vendor/bundle/ruby/3.4.0/gems/json-2.12.2/ext/json/ext/generator/generator.c @@ -0,0 +1,2188 @@ +#include "ruby.h" +#include "../fbuffer/fbuffer.h" +#include "../vendor/fpconv.c" + +#include <math.h> +#include <ctype.h> + +#include "simd.h" + +/* ruby api and some helpers */ + +typedef struct JSON_Generator_StateStruct { + VALUE indent; + VALUE space; + VALUE space_before; + VALUE object_nl; + VALUE array_nl; + VALUE as_json; + + long max_nesting; + long depth; + long buffer_initial_length; + + bool allow_nan; + bool ascii_only; + bool script_safe; + bool strict; +} JSON_Generator_State; + +#ifndef RB_UNLIKELY +#define RB_UNLIKELY(cond) (cond) +#endif + +static VALUE mJSON, cState, cFragment, mString_Extend, 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 VALUE sym_indent, sym_space, sym_space_before, sym_object_nl, sym_array_nl, sym_max_nesting, sym_allow_nan, + sym_ascii_only, sym_depth, sym_buffer_initial_length, sym_script_safe, sym_escape_slash, sym_strict, sym_as_json; + + +#define GET_STATE_TO(self, state) \ + TypedData_Get_Struct(self, JSON_Generator_State, &JSON_Generator_State_type, state) + +#define GET_STATE(self) \ + JSON_Generator_State *state; \ + GET_STATE_TO(self, state) + +struct generate_json_data; + +typedef void (*generator_func)(FBuffer *buffer, struct generate_json_data *data, VALUE obj); + +struct generate_json_data { + FBuffer *buffer; + VALUE vstate; + JSON_Generator_State *state; + VALUE obj; + generator_func func; +}; + +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); +static void generate_json_object(FBuffer *buffer, struct generate_json_data *data, VALUE obj); +static void generate_json_array(FBuffer *buffer, struct generate_json_data *data, VALUE obj); +static void generate_json_string(FBuffer *buffer, struct generate_json_data *data, VALUE obj); +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); +static void generate_json_fragment(FBuffer *buffer, struct generate_json_data *data, VALUE obj); + +static int usascii_encindex, utf8_encindex, binary_encindex; + +#ifdef RBIMPL_ATTR_NORETURN +RBIMPL_ATTR_NORETURN() +#endif +static void raise_generator_error_str(VALUE invalid_object, VALUE str) +{ + VALUE exc = rb_exc_new_str(eGeneratorError, str); + rb_ivar_set(exc, rb_intern("@invalid_object"), invalid_object); + rb_exc_raise(exc); +} + +#ifdef RBIMPL_ATTR_NORETURN +RBIMPL_ATTR_NORETURN() +#endif +#ifdef RBIMPL_ATTR_FORMAT +RBIMPL_ATTR_FORMAT(RBIMPL_PRINTF_FORMAT, 2, 3) +#endif +static void raise_generator_error(VALUE invalid_object, const char *fmt, ...) +{ + va_list args; + va_start(args, fmt); + VALUE str = rb_vsprintf(fmt, args); + va_end(args); + raise_generator_error_str(invalid_object, str); +} + +// 0 - single byte char that don't need to be escaped. +// (x | 8) - char that needs to be escaped. +static const unsigned char CHAR_LENGTH_MASK = 7; +static const unsigned char ESCAPE_MASK = 8; + +typedef struct _search_state { + const char *ptr; + const char *end; + const char *cursor; + FBuffer *buffer; + +#ifdef HAVE_SIMD + const char *chunk_base; + const char *chunk_end; + bool has_matches; + +#if defined(HAVE_SIMD_NEON) + uint64_t matches_mask; +#elif defined(HAVE_SIMD_SSE2) + int matches_mask; +#else +#error "Unknown SIMD Implementation." +#endif /* HAVE_SIMD_NEON */ +#endif /* HAVE_SIMD */ +} search_state; + +#if (defined(__GNUC__ ) || defined(__clang__)) +#define FORCE_INLINE __attribute__((always_inline)) +#else +#define FORCE_INLINE +#endif + +static inline FORCE_INLINE 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). + // For back-to-back characters that need to be escaped, specifcally for the SIMD code paths, this method + // will be called just before calling escape_UTF8_char_basic. There will be no characers to append for the + // consecutive characters that need to be escaped. While the fbuffer_append is a no-op if + // nothing needs to be flushed, we can save a few memory references with this conditional. + if (search->ptr > search->cursor) { + fbuffer_append(search->buffer, search->cursor, search->ptr - search->cursor); + search->cursor = search->ptr; + } +} + +static const unsigned char escape_table_basic[256] = { + // ASCII Control Characters + 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, + 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, + // ASCII Characters + 0, 0, 9, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, // '"' + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 9, 0, 0, 0, // '\\' + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, + 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) { + if (RB_UNLIKELY(escape_table_basic[(const unsigned char)*search->ptr])) { + search_flush(search); + return 1; + } else { + search->ptr++; + } + } + search_flush(search); + return 0; +} + +static inline FORCE_INLINE void escape_UTF8_char_basic(search_state *search) +{ + const unsigned char ch = (unsigned char)*search->ptr; + switch (ch) { + case '"': fbuffer_append(search->buffer, "\\\"", 2); break; + case '\\': fbuffer_append(search->buffer, "\\\\", 2); break; + case '/': fbuffer_append(search->buffer, "\\/", 2); break; + case '\b': fbuffer_append(search->buffer, "\\b", 2); break; + case '\f': fbuffer_append(search->buffer, "\\f", 2); break; + case '\n': fbuffer_append(search->buffer, "\\n", 2); break; + case '\r': fbuffer_append(search->buffer, "\\r", 2); break; + case '\t': fbuffer_append(search->buffer, "\\t", 2); break; + default: { + const char *hexdig = "0123456789abcdef"; + char scratch[6] = { '\\', 'u', '0', '0', 0, 0 }; + scratch[4] = hexdig[(ch >> 4) & 0xf]; + scratch[5] = hexdig[ch & 0xf]; + fbuffer_append(search->buffer, scratch, 6); + break; + } + } + search->ptr++; + search->cursor = search->ptr; +} + +/* Converts in_string to a JSON string (without the wrapping '"' + * characters) in FBuffer out_buffer. + * + * Character are JSON-escaped according to: + * + * - Always: ASCII control characters (0x00-0x1F), dquote, and + * backslash. + * + * - If out_ascii_only: non-ASCII characters (>0x7F) + * + * - If script_safe: forwardslash (/), line separator (U+2028), and + * paragraph separator (U+2029) + * + * Everything else (should be UTF-8) is just passed through and + * appended to the result. + */ +static inline void convert_UTF8_to_JSON(search_state *search) +{ + while (search_escape_basic_impl(search)) { + escape_UTF8_char_basic(search); + } +} + +static inline void escape_UTF8_char(search_state *search, unsigned char ch_len) +{ + const unsigned char ch = (unsigned char)*search->ptr; + switch (ch_len) { + case 1: { + switch (ch) { + case '"': fbuffer_append(search->buffer, "\\\"", 2); break; + case '\\': fbuffer_append(search->buffer, "\\\\", 2); break; + case '/': fbuffer_append(search->buffer, "\\/", 2); break; + case '\b': fbuffer_append(search->buffer, "\\b", 2); break; + case '\f': fbuffer_append(search->buffer, "\\f", 2); break; + case '\n': fbuffer_append(search->buffer, "\\n", 2); break; + case '\r': fbuffer_append(search->buffer, "\\r", 2); break; + case '\t': fbuffer_append(search->buffer, "\\t", 2); break; + default: { + const char *hexdig = "0123456789abcdef"; + char scratch[6] = { '\\', 'u', '0', '0', 0, 0 }; + scratch[4] = hexdig[(ch >> 4) & 0xf]; + scratch[5] = hexdig[ch & 0xf]; + fbuffer_append(search->buffer, scratch, 6); + break; + } + } + break; + } + case 3: { + if (search->ptr[2] & 1) { + fbuffer_append(search->buffer, "\\u2029", 6); + } else { + fbuffer_append(search->buffer, "\\u2028", 6); + } + break; + } + } + search->cursor = (search->ptr += ch_len); +} + +#ifdef HAVE_SIMD + +static inline FORCE_INLINE 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); + + FBuffer *buf = search->buffer; + fbuffer_inc_capa(buf, vec_len); + + 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); + + // 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); + + return s; +} + +#ifdef HAVE_SIMD_NEON + +static inline FORCE_INLINE unsigned char neon_next_match(search_state *search) +{ + uint64_t mask = search->matches_mask; + uint32_t index = trailing_zeros64(mask) >> 2; + + // It is assumed escape_UTF8_char_basic will only ever increase search->ptr by at most one character. + // If we want to use a similar approach for full escaping we'll need to ensure: + // search->chunk_base + index >= search->ptr + // However, since we know escape_UTF8_char_basic only increases search->ptr by one, if the next match + // is one byte after the previous match then: + // search->chunk_base + index == search->ptr + search->ptr = search->chunk_base + index; + mask &= mask - 1; + search->matches_mask = mask; + search_flush(search); + return 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 inline FORCE_INLINE 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 inline FORCE_INLINE uint64_t neon_rules_update(const char *ptr) +{ + uint8x16_t chunk = vld1q_u8((const unsigned char *)ptr); + + // Trick: c < 32 || c == 34 can be factored as c ^ 2 < 33 + // https://lemire.me/blog/2025/04/13/detect-control-characters-quotes-and-backslashes-efficiently-using-swar/ + const uint8x16_t too_low_or_dbl_quote = vcltq_u8(veorq_u8(chunk, vdupq_n_u8(2)), vdupq_n_u8(33)); + + uint8x16_t has_backslash = vceqq_u8(chunk, vdupq_n_u8('\\')); + uint8x16_t needs_escape = vorrq_u8(too_low_or_dbl_quote, has_backslash); + + return neon_match_mask(needs_escape); +} + +static inline unsigned char search_escape_basic_neon(search_state *search) +{ + if (RB_UNLIKELY(search->has_matches)) { + // There are more matches if search->matches_mask > 0. + if (search->matches_mask > 0) { + return neon_next_match(search); + } else { + // neon_next_match will only advance search->ptr up to the last matching character. + // Skip over any characters in the last chunk that occur after the last match. + search->has_matches = false; + search->ptr = search->chunk_end; + } + } + + /* + * The code below implements an SIMD-based algorithm to determine if N bytes at a time + * need to be escaped. + * + * Assume the ptr = "Te\sting!" (the double quotes are included in the string) + * + * The explanation will be limited to the first 8 bytes of the string for simplicity. However + * the vector insructions may work on larger vectors. + * + * First, we load three constants 'lower_bound', 'backslash' and 'dblquote" in vector registers. + * + * lower_bound: [20 20 20 20 20 20 20 20] + * backslash: [5C 5C 5C 5C 5C 5C 5C 5C] + * dblquote: [22 22 22 22 22 22 22 22] + * + * Next we load the first chunk of the ptr: + * [22 54 65 5C 73 74 69 6E] (" T e \ s t i n) + * + * First we check if any byte in chunk is less than 32 (0x20). This returns the following vector + * as no bytes are less than 32 (0x20): + * [0 0 0 0 0 0 0 0] + * + * Next, we check if any byte in chunk is equal to a backslash: + * [0 0 0 FF 0 0 0 0] + * + * Finally we check if any byte in chunk is equal to a double quote: + * [FF 0 0 0 0 0 0 0] + * + * Now we have three vectors where each byte indicates if the corresponding byte in chunk + * needs to be escaped. We combine these vectors with a series of logical OR instructions. + * This is the needs_escape vector and it is equal to: + * [FF 0 0 FF 0 0 0 0] + * + * Next we compute the bitwise AND between each byte and 0x1 and compute the horizontal sum of + * the values in the vector. This computes how many bytes need to be escaped within this chunk. + * + * Finally we compute a mask that indicates which bytes need to be escaped. If the mask is 0 then, + * no bytes need to be escaped and we can continue to the next chunk. If the mask is not 0 then we + * have at least one byte that needs to be escaped. + */ + while (search->ptr + sizeof(uint8x16_t) <= search->end) { + uint64_t mask = neon_rules_update(search->ptr); + + if (!mask) { + search->ptr += sizeof(uint8x16_t); + continue; + } + search->matches_mask = mask; + search->has_matches = true; + search->chunk_base = search->ptr; + search->chunk_end = search->ptr + sizeof(uint8x16_t); + return neon_next_match(search); + } + + // There are fewer than 16 bytes left. + unsigned long remaining = (search->end - search->ptr); + if (remaining >= SIMD_MINIMUM_THRESHOLD) { + char *s = copy_remaining_bytes(search, sizeof(uint8x16_t), remaining); + + uint64_t mask = neon_rules_update(s); + + if (!mask) { + // Nothing to escape, ensure search_flush doesn't do anything by setting + // search->cursor to search->ptr. + fbuffer_consumed(search->buffer, remaining); + search->ptr = search->end; + search->cursor = search->end; + return 0; + } + + search->matches_mask = mask; + search->has_matches = true; + search->chunk_end = search->end; + search->chunk_base = search->ptr; + return neon_next_match(search); + } + + if (search->ptr < search->end) { + return search_escape_basic(search); + } + + search_flush(search); + return 0; +} +#endif /* HAVE_SIMD_NEON */ + +#ifdef HAVE_SIMD_SSE2 + +#define _mm_cmpge_epu8(a, b) _mm_cmpeq_epi8(_mm_max_epu8(a, b), a) +#define _mm_cmple_epu8(a, b) _mm_cmpge_epu8(b, a) +#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 inline FORCE_INLINE unsigned char sse2_next_match(search_state *search) +{ + int mask = search->matches_mask; + int index = trailing_zeros(mask); + + // It is assumed escape_UTF8_char_basic will only ever increase search->ptr by at most one character. + // If we want to use a similar approach for full escaping we'll need to ensure: + // search->chunk_base + index >= search->ptr + // However, since we know escape_UTF8_char_basic only increases search->ptr by one, if the next match + // is one byte after the previous match then: + // search->chunk_base + index == search->ptr + search->ptr = search->chunk_base + index; + mask &= mask - 1; + search->matches_mask = mask; + search_flush(search); + return 1; +} + +#if defined(__clang__) || defined(__GNUC__) +#define TARGET_SSE2 __attribute__((target("sse2"))) +#else +#define TARGET_SSE2 +#endif + +static inline TARGET_SSE2 FORCE_INLINE int sse2_update(const char *ptr) +{ + __m128i chunk = _mm_loadu_si128((__m128i const*)ptr); + + // Trick: c < 32 || c == 34 can be factored as c ^ 2 < 33 + // https://lemire.me/blog/2025/04/13/detect-control-characters-quotes-and-backslashes-efficiently-using-swar/ + __m128i too_low_or_dbl_quote = _mm_cmplt_epu8(_mm_xor_si128(chunk, _mm_set1_epi8(2)), _mm_set1_epi8(33)); + __m128i has_backslash = _mm_cmpeq_epi8(chunk, _mm_set1_epi8('\\')); + __m128i needs_escape = _mm_or_si128(too_low_or_dbl_quote, has_backslash); + return _mm_movemask_epi8(needs_escape); +} + +static inline TARGET_SSE2 FORCE_INLINE unsigned char search_escape_basic_sse2(search_state *search) +{ + if (RB_UNLIKELY(search->has_matches)) { + // There are more matches if search->matches_mask > 0. + if (search->matches_mask > 0) { + return sse2_next_match(search); + } else { + // sse2_next_match will only advance search->ptr up to the last matching character. + // Skip over any characters in the last chunk that occur after the last match. + search->has_matches = false; + if (RB_UNLIKELY(search->chunk_base + sizeof(__m128i) >= search->end)) { + search->ptr = search->end; + } else { + search->ptr = search->chunk_base + sizeof(__m128i); + } + } + } + + while (search->ptr + sizeof(__m128i) <= search->end) { + int needs_escape_mask = sse2_update(search->ptr); + + if (needs_escape_mask == 0) { + search->ptr += sizeof(__m128i); + continue; + } + + search->has_matches = true; + search->matches_mask = needs_escape_mask; + search->chunk_base = search->ptr; + return sse2_next_match(search); + } + + // There are fewer than 16 bytes left. + unsigned long remaining = (search->end - search->ptr); + if (remaining >= SIMD_MINIMUM_THRESHOLD) { + char *s = copy_remaining_bytes(search, sizeof(__m128i), remaining); + + int needs_escape_mask = sse2_update(s); + + if (needs_escape_mask == 0) { + // Nothing to escape, ensure search_flush doesn't do anything by setting + // search->cursor to search->ptr. + fbuffer_consumed(search->buffer, remaining); + search->ptr = search->end; + search->cursor = search->end; + return 0; + } + + search->has_matches = true; + search->matches_mask = needs_escape_mask; + search->chunk_base = search->ptr; + return sse2_next_match(search); + } + + if (search->ptr < search->end) { + return search_escape_basic(search); + } + + search_flush(search); + return 0; +} + +#endif /* HAVE_SIMD_SSE2 */ + +#endif /* HAVE_SIMD */ + +static const unsigned char script_safe_escape_table[256] = { + // ASCII Control Characters + 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, + 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, + // ASCII Characters + 0, 0, 9, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 9, // '"' and '/' + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 9, 0, 0, 0, // '\\' + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, + // Continuation byte + 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, + 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, + 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, + 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, + // First byte of a 2-byte code point + 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, + 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, + // First byte of a 3-byte code point + 3, 3,11, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, // 0xE2 is the start of \u2028 and \u2029 + //First byte of a 4+ byte code point + 4, 4, 4, 4, 4, 4, 4, 4, 5, 5, 5, 5, 6, 6, 9, 9, +}; + +static inline unsigned char search_script_safe_escape(search_state *search) +{ + while (search->ptr < search->end) { + unsigned char ch = (unsigned char)*search->ptr; + unsigned char ch_len = script_safe_escape_table[ch]; + + if (RB_UNLIKELY(ch_len)) { + if (ch_len & ESCAPE_MASK) { + if (RB_UNLIKELY(ch_len == 11)) { + const unsigned char *uptr = (const unsigned char *)search->ptr; + if (!(uptr[1] == 0x80 && (uptr[2] >> 1) == 0x54)) { + search->ptr += 3; + continue; + } + } + search_flush(search); + return ch_len & CHAR_LENGTH_MASK; + } else { + search->ptr += ch_len; + } + } else { + search->ptr++; + } + } + search_flush(search); + return 0; +} + +static void convert_UTF8_to_script_safe_JSON(search_state *search) +{ + unsigned char ch_len; + while ((ch_len = search_script_safe_escape(search))) { + escape_UTF8_char(search, ch_len); + } +} + +static const unsigned char ascii_only_escape_table[256] = { + // ASCII Control Characters + 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, + 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, + // ASCII Characters + 0, 0, 9, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, // '"' + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 9, 0, 0, 0, // '\\' + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, + // Continuation byte + 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, + 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, + 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, + 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, + // First byte of a 2-byte code point + 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, + 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, + // First byte of a 3-byte code point + 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, + //First byte of a 4+ byte code point + 4, 4, 4, 4, 4, 4, 4, 4, 5, 5, 5, 5, 6, 6, 9, 9, +}; + +static inline unsigned char search_ascii_only_escape(search_state *search, const unsigned char escape_table[256]) +{ + while (search->ptr < search->end) { + unsigned char ch = (unsigned char)*search->ptr; + unsigned char ch_len = escape_table[ch]; + + if (RB_UNLIKELY(ch_len)) { + search_flush(search); + return ch_len & CHAR_LENGTH_MASK; + } else { + search->ptr++; + } + } + search_flush(search); + return 0; +} + +static inline void full_escape_UTF8_char(search_state *search, unsigned char ch_len) { + const unsigned char ch = (unsigned char)*search->ptr; + switch (ch_len) { + case 1: { + switch (ch) { + case '"': fbuffer_append(search->buffer, "\\\"", 2); break; + case '\\': fbuffer_append(search->buffer, "\\\\", 2); break; + case '/': fbuffer_append(search->buffer, "\\/", 2); break; + case '\b': fbuffer_append(search->buffer, "\\b", 2); break; + case '\f': fbuffer_append(search->buffer, "\\f", 2); break; + case '\n': fbuffer_append(search->buffer, "\\n", 2); break; + case '\r': fbuffer_append(search->buffer, "\\r", 2); break; + case '\t': fbuffer_append(search->buffer, "\\t", 2); break; + default: { + const char *hexdig = "0123456789abcdef"; + char scratch[6] = { '\\', 'u', '0', '0', 0, 0 }; + scratch[4] = hexdig[(ch >> 4) & 0xf]; + scratch[5] = hexdig[ch & 0xf]; + fbuffer_append(search->buffer, scratch, 6); + break; + } + } + break; + } + default: { + const char *hexdig = "0123456789abcdef"; + char scratch[12] = { '\\', 'u', 0, 0, 0, 0, '\\', 'u' }; + + uint32_t wchar = 0; + + switch(ch_len) { + case 2: + wchar = ch & 0x1F; + break; + case 3: + wchar = ch & 0x0F; + break; + case 4: + wchar = ch & 0x07; + break; + } + + for (short i = 1; i < ch_len; i++) { + wchar = (wchar << 6) | (search->ptr[i] & 0x3F); + } + + if (wchar <= 0xFFFF) { + scratch[2] = hexdig[wchar >> 12]; + scratch[3] = hexdig[(wchar >> 8) & 0xf]; + scratch[4] = hexdig[(wchar >> 4) & 0xf]; + scratch[5] = hexdig[wchar & 0xf]; + fbuffer_append(search->buffer, scratch, 6); + } else { + uint16_t hi, lo; + wchar -= 0x10000; + hi = 0xD800 + (uint16_t)(wchar >> 10); + lo = 0xDC00 + (uint16_t)(wchar & 0x3FF); + + scratch[2] = hexdig[hi >> 12]; + scratch[3] = hexdig[(hi >> 8) & 0xf]; + scratch[4] = hexdig[(hi >> 4) & 0xf]; + scratch[5] = hexdig[hi & 0xf]; + + scratch[8] = hexdig[lo >> 12]; + scratch[9] = hexdig[(lo >> 8) & 0xf]; + scratch[10] = hexdig[(lo >> 4) & 0xf]; + scratch[11] = hexdig[lo & 0xf]; + + fbuffer_append(search->buffer, scratch, 12); + } + + break; + } + } + search->cursor = (search->ptr += ch_len); +} + +static void convert_UTF8_to_ASCII_only_JSON(search_state *search, const unsigned char escape_table[256]) +{ + unsigned char ch_len; + while ((ch_len = search_ascii_only_escape(search, escape_table))) { + full_escape_UTF8_char(search, ch_len); + } +} + +/* + * 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: String.included(modul) + * + * Extends _modul_ with the String::Extend module. + */ +static VALUE mString_included_s(VALUE self, VALUE modul) { + VALUE result = rb_funcall(modul, i_extend, 1, mString_Extend); + rb_call_super(1, &modul); + return result; +} + +/* + * 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_raw_object() + * + * This method creates a raw object hash, that can be nested into + * other data structures and will be generated as a raw string. This + * method should be used, if you want to convert raw strings to JSON + * instead of UTF-8 strings, e. g. binary data. + */ +static VALUE mString_to_json_raw_object(VALUE self) +{ + VALUE ary; + VALUE result = rb_hash_new(); + rb_hash_aset(result, rb_funcall(mJSON, i_create_id, 0), rb_class_name(rb_obj_class(self))); + ary = rb_funcall(self, i_unpack, 1, rb_str_new2("C*")); + rb_hash_aset(result, rb_utf8_str_new_lit("raw"), ary); + return result; +} + +/* + * call-seq: to_json_raw(*args) + * + * This method creates a JSON text from the result of a call to + * to_json_raw_object of this String. + */ +static VALUE mString_to_json_raw(int argc, VALUE *argv, VALUE self) +{ + VALUE obj = mString_to_json_raw_object(self); + Check_Type(obj, T_HASH); + return mHash_to_json(argc, argv, obj); +} + +/* + * call-seq: json_create(o) + * + * Raw Strings are JSON Objects (the raw bytes are stored in an array for the + * key "raw"). The Ruby String can be created by this module method. + */ +static VALUE mString_Extend_json_create(VALUE self, VALUE o) +{ + VALUE ary; + Check_Type(o, T_HASH); + ary = rb_hash_aref(o, rb_str_new2("raw")); + return rb_funcall(ary, i_pack, 1, rb_str_new2("C*")); +} + +/* + * 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; + rb_gc_mark_movable(state->indent); + rb_gc_mark_movable(state->space); + rb_gc_mark_movable(state->space_before); + rb_gc_mark_movable(state->object_nl); + rb_gc_mark_movable(state->array_nl); + rb_gc_mark_movable(state->as_json); +} + +static void State_compact(void *ptr) +{ + JSON_Generator_State *state = ptr; + state->indent = rb_gc_location(state->indent); + state->space = rb_gc_location(state->space); + state->space_before = rb_gc_location(state->space_before); + state->object_nl = rb_gc_location(state->object_nl); + state->array_nl = rb_gc_location(state->array_nl); + state->as_json = rb_gc_location(state->as_json); +} + +static void State_free(void *ptr) +{ + JSON_Generator_State *state = ptr; + ruby_xfree(state); +} + +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", + { + .dmark = State_mark, + .dfree = State_free, + .dsize = State_memsize, + .dcompact = State_compact, + }, + 0, 0, + RUBY_TYPED_WB_PROTECTED | RUBY_TYPED_FREE_IMMEDIATELY | RUBY_TYPED_FROZEN_SHAREABLE, +}; + +static void state_init(JSON_Generator_State *state) +{ + state->max_nesting = 100; + state->buffer_initial_length = FBUFFER_INITIAL_LENGTH_DEFAULT; +} + +static VALUE cState_s_allocate(VALUE klass) +{ + JSON_Generator_State *state; + VALUE obj = TypedData_Make_Struct(klass, JSON_Generator_State, &JSON_Generator_State_type, state); + state_init(state); + return obj; +} + +static void vstate_spill(struct generate_json_data *data) +{ + VALUE vstate = cState_s_allocate(cState); + GET_STATE(vstate); + MEMCPY(state, data->state, JSON_Generator_State, 1); + data->state = state; + data->vstate = vstate; + RB_OBJ_WRITTEN(vstate, Qundef, state->indent); + RB_OBJ_WRITTEN(vstate, Qundef, state->space); + RB_OBJ_WRITTEN(vstate, Qundef, state->space_before); + RB_OBJ_WRITTEN(vstate, Qundef, state->object_nl); + RB_OBJ_WRITTEN(vstate, Qundef, state->array_nl); + RB_OBJ_WRITTEN(vstate, Qundef, state->as_json); +} + +static inline VALUE vstate_get(struct generate_json_data *data) +{ + if (RB_UNLIKELY(!data->vstate)) { + vstate_spill(data); + } + return data->vstate; +} + +struct hash_foreach_arg { + struct generate_json_data *data; + int iter; +}; + +static VALUE +convert_string_subclass(VALUE key) +{ + VALUE key_to_s = rb_funcall(key, i_to_s, 0); + + if (RB_UNLIKELY(!RB_TYPE_P(key_to_s, T_STRING))) { + VALUE cname = rb_obj_class(key); + rb_raise(rb_eTypeError, + "can't convert %"PRIsVALUE" to %s (%"PRIsVALUE"#%s gives %"PRIsVALUE")", + cname, "String", cname, "to_s", rb_obj_class(key_to_s)); + } + + return key_to_s; +} + +static int +json_object_i(VALUE key, VALUE val, VALUE _arg) +{ + struct hash_foreach_arg *arg = (struct hash_foreach_arg *)_arg; + struct generate_json_data *data = arg->data; + + FBuffer *buffer = data->buffer; + JSON_Generator_State *state = data->state; + + long depth = state->depth; + int j; + + if (arg->iter > 0) fbuffer_append_char(buffer, ','); + if (RB_UNLIKELY(data->state->object_nl)) { + fbuffer_append_str(buffer, data->state->object_nl); + } + if (RB_UNLIKELY(data->state->indent)) { + for (j = 0; j < depth; j++) { + fbuffer_append_str(buffer, data->state->indent); + } + } + + VALUE key_to_s; + switch(rb_type(key)) { + case T_STRING: + if (RB_LIKELY(RBASIC_CLASS(key) == rb_cString)) { + key_to_s = key; + } else { + key_to_s = convert_string_subclass(key); + } + break; + case T_SYMBOL: + key_to_s = rb_sym2str(key); + break; + default: + key_to_s = rb_convert_type(key, T_STRING, "String", "to_s"); + break; + } + + if (RB_LIKELY(RBASIC_CLASS(key_to_s) == rb_cString)) { + generate_json_string(buffer, data, key_to_s); + } else { + generate_json(buffer, data, key_to_s); + } + if (RB_UNLIKELY(state->space_before)) fbuffer_append_str(buffer, data->state->space_before); + fbuffer_append_char(buffer, ':'); + if (RB_UNLIKELY(state->space)) fbuffer_append_str(buffer, data->state->space); + generate_json(buffer, data, val); + + arg->iter++; + return ST_CONTINUE; +} + +static inline long increase_depth(struct generate_json_data *data) +{ + JSON_Generator_State *state = data->state; + long depth = ++state->depth; + if (RB_UNLIKELY(depth > state->max_nesting && state->max_nesting)) { + rb_raise(eNestingError, "nesting of %ld is too deep", --state->depth); + } + return depth; +} + +static void generate_json_object(FBuffer *buffer, struct generate_json_data *data, VALUE obj) +{ + int j; + long depth = increase_depth(data); + + if (RHASH_SIZE(obj) == 0) { + fbuffer_append(buffer, "{}", 2); + --data->state->depth; + return; + } + + fbuffer_append_char(buffer, '{'); + + struct hash_foreach_arg arg = { + .data = data, + .iter = 0, + }; + rb_hash_foreach(obj, json_object_i, (VALUE)&arg); + + depth = --data->state->depth; + if (RB_UNLIKELY(data->state->object_nl)) { + fbuffer_append_str(buffer, data->state->object_nl); + if (RB_UNLIKELY(data->state->indent)) { + for (j = 0; j < depth; j++) { + fbuffer_append_str(buffer, data->state->indent); + } + } + } + fbuffer_append_char(buffer, '}'); +} + +static void generate_json_array(FBuffer *buffer, struct generate_json_data *data, VALUE obj) +{ + int i, j; + long depth = increase_depth(data); + + if (RARRAY_LEN(obj) == 0) { + fbuffer_append(buffer, "[]", 2); + --data->state->depth; + return; + } + + fbuffer_append_char(buffer, '['); + if (RB_UNLIKELY(data->state->array_nl)) fbuffer_append_str(buffer, data->state->array_nl); + for(i = 0; i < RARRAY_LEN(obj); i++) { + if (i > 0) { + fbuffer_append_char(buffer, ','); + if (RB_UNLIKELY(data->state->array_nl)) fbuffer_append_str(buffer, data->state->array_nl); + } + if (RB_UNLIKELY(data->state->indent)) { + for (j = 0; j < depth; j++) { + fbuffer_append_str(buffer, data->state->indent); + } + } + generate_json(buffer, data, RARRAY_AREF(obj, i)); + } + data->state->depth = --depth; + if (RB_UNLIKELY(data->state->array_nl)) { + fbuffer_append_str(buffer, data->state->array_nl); + if (RB_UNLIKELY(data->state->indent)) { + for (j = 0; j < depth; j++) { + fbuffer_append_str(buffer, data->state->indent); + } + } + } + fbuffer_append_char(buffer, ']'); +} + +static inline int enc_utf8_compatible_p(int enc_idx) +{ + if (enc_idx == usascii_encindex) return 1; + if (enc_idx == utf8_encindex) return 1; + return 0; +} + +static VALUE encode_json_string_try(VALUE str) +{ + return rb_funcall(str, i_encode, 1, Encoding_UTF_8); +} + +static VALUE encode_json_string_rescue(VALUE str, VALUE exception) +{ + raise_generator_error_str(str, rb_funcall(exception, rb_intern("message"), 0)); + return Qundef; +} + +static inline VALUE ensure_valid_encoding(VALUE str) +{ + int encindex = RB_ENCODING_GET(str); + VALUE utf8_string; + if (RB_UNLIKELY(!enc_utf8_compatible_p(encindex))) { + if (encindex == binary_encindex) { + utf8_string = rb_enc_associate_index(rb_str_dup(str), utf8_encindex); + switch (rb_enc_str_coderange(utf8_string)) { + case ENC_CODERANGE_7BIT: + return utf8_string; + case ENC_CODERANGE_VALID: + // For historical reason, we silently reinterpret binary strings as UTF-8 if it would work. + // TODO: Raise in 3.0.0 + rb_warn("JSON.generate: UTF-8 string passed as BINARY, this will raise an encoding error in json 3.0"); + return utf8_string; + break; + } + } + + str = rb_rescue(encode_json_string_try, str, encode_json_string_rescue, str); + } + return str; +} + +static void generate_json_string(FBuffer *buffer, struct generate_json_data *data, VALUE obj) +{ + obj = ensure_valid_encoding(obj); + + fbuffer_append_char(buffer, '"'); + + long len; + search_state search; + search.buffer = buffer; + RSTRING_GETMEM(obj, search.ptr, len); + search.cursor = search.ptr; + search.end = search.ptr + len; + +#ifdef HAVE_SIMD + search.matches_mask = 0; + search.has_matches = false; + search.chunk_base = NULL; +#endif /* HAVE_SIMD */ + + switch(rb_enc_str_coderange(obj)) { + case ENC_CODERANGE_7BIT: + case ENC_CODERANGE_VALID: + if (RB_UNLIKELY(data->state->ascii_only)) { + convert_UTF8_to_ASCII_only_JSON(&search, data->state->script_safe ? script_safe_escape_table : ascii_only_escape_table); + } else if (RB_UNLIKELY(data->state->script_safe)) { + convert_UTF8_to_script_safe_JSON(&search); + } else { + convert_UTF8_to_JSON(&search); + } + break; + default: + raise_generator_error(obj, "source sequence is illegal/malformed utf-8"); + break; + } + fbuffer_append_char(buffer, '"'); +} + +static void generate_json_fallback(FBuffer *buffer, struct generate_json_data *data, VALUE obj) +{ + VALUE tmp; + if (rb_respond_to(obj, i_to_json)) { + tmp = rb_funcall(obj, i_to_json, 1, vstate_get(data)); + Check_Type(tmp, T_STRING); + fbuffer_append_str(buffer, tmp); + } else { + tmp = rb_funcall(obj, i_to_s, 0); + Check_Type(tmp, T_STRING); + generate_json_string(buffer, data, tmp); + } +} + +static inline void generate_json_symbol(FBuffer *buffer, struct generate_json_data *data, VALUE obj) +{ + if (data->state->strict) { + generate_json_string(buffer, data, rb_sym2str(obj)); + } else { + generate_json_fallback(buffer, data, obj); + } +} + +static void generate_json_null(FBuffer *buffer, struct generate_json_data *data, VALUE obj) +{ + fbuffer_append(buffer, "null", 4); +} + +static void generate_json_false(FBuffer *buffer, struct generate_json_data *data, VALUE obj) +{ + fbuffer_append(buffer, "false", 5); +} + +static void generate_json_true(FBuffer *buffer, struct generate_json_data *data, VALUE obj) +{ + fbuffer_append(buffer, "true", 4); +} + +static void generate_json_fixnum(FBuffer *buffer, struct generate_json_data *data, VALUE obj) +{ + fbuffer_append_long(buffer, FIX2LONG(obj)); +} + +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); +} +#endif + +static void generate_json_float(FBuffer *buffer, struct generate_json_data *data, VALUE obj) +{ + double value = RFLOAT_VALUE(obj); + char allow_nan = data->state->allow_nan; + if (isinf(value) || isnan(value)) { + /* for NaN and Infinity values we either raise an error or rely on Float#to_s. */ + if (!allow_nan) { + if (data->state->strict && data->state->as_json) { + VALUE casted_obj = rb_proc_call_with_block(data->state->as_json, 1, &obj, Qnil); + if (casted_obj != obj) { + increase_depth(data); + generate_json(buffer, data, casted_obj); + data->state->depth--; + return; + } + } + raise_generator_error(obj, "%"PRIsVALUE" not allowed in JSON", rb_funcall(obj, i_to_s, 0)); + } + + VALUE tmp = rb_funcall(obj, i_to_s, 0); + fbuffer_append_str(buffer, tmp); + return; + } + + /* This implementation writes directly into the buffer. We reserve + * the 28 characters that fpconv_dtoa states as its maximum. + */ + fbuffer_inc_capa(buffer, 28); + char* d = buffer->ptr + buffer->len; + int len = fpconv_dtoa(value, d); + + /* fpconv_dtoa converts a float to its shortest string representation, + * but it adds a ".0" if this is a plain integer. + */ + fbuffer_consumed(buffer, len); +} + +static void generate_json_fragment(FBuffer *buffer, struct generate_json_data *data, VALUE obj) +{ + VALUE fragment = RSTRUCT_GET(obj, 0); + Check_Type(fragment, T_STRING); + fbuffer_append_str(buffer, fragment); +} + +static void generate_json(FBuffer *buffer, struct generate_json_data *data, VALUE obj) +{ + bool as_json_called = false; +start: + if (obj == Qnil) { + generate_json_null(buffer, data, obj); + } else if (obj == Qfalse) { + generate_json_false(buffer, data, obj); + } else if (obj == Qtrue) { + generate_json_true(buffer, data, obj); + } else if (RB_SPECIAL_CONST_P(obj)) { + if (RB_FIXNUM_P(obj)) { + generate_json_fixnum(buffer, data, obj); + } else if (RB_FLONUM_P(obj)) { + generate_json_float(buffer, data, obj); + } else if (RB_STATIC_SYM_P(obj)) { + generate_json_symbol(buffer, data, obj); + } else { + goto general; + } + } else { + VALUE klass = RBASIC_CLASS(obj); + switch (RB_BUILTIN_TYPE(obj)) { + case T_BIGNUM: + generate_json_bignum(buffer, data, obj); + break; + case T_HASH: + if (klass != rb_cHash) goto general; + generate_json_object(buffer, data, obj); + break; + case T_ARRAY: + if (klass != rb_cArray) goto general; + generate_json_array(buffer, data, obj); + break; + case T_STRING: + if (klass != rb_cString) goto general; + generate_json_string(buffer, data, obj); + break; + case T_SYMBOL: + generate_json_symbol(buffer, data, obj); + break; + case T_FLOAT: + if (klass != rb_cFloat) goto general; + generate_json_float(buffer, data, obj); + break; + case T_STRUCT: + if (klass != cFragment) goto general; + generate_json_fragment(buffer, data, obj); + break; + default: + general: + if (data->state->strict) { + if (RTEST(data->state->as_json) && !as_json_called) { + obj = rb_proc_call_with_block(data->state->as_json, 1, &obj, Qnil); + as_json_called = true; + goto start; + } else { + raise_generator_error(obj, "%"PRIsVALUE" not allowed in JSON", CLASS_OF(obj)); + } + } else { + generate_json_fallback(buffer, data, obj); + } + } + } +} + +static VALUE generate_json_try(VALUE d) +{ + struct generate_json_data *data = (struct generate_json_data *)d; + + data->func(data->buffer, data, data->obj); + + return Qnil; +} + +static VALUE generate_json_rescue(VALUE d, VALUE exc) +{ + struct generate_json_data *data = (struct generate_json_data *)d; + fbuffer_free(data->buffer); + + rb_exc_raise(exc); + + return Qundef; +} + +static VALUE cState_partial_generate(VALUE self, VALUE obj, generator_func func, VALUE io) +{ + GET_STATE(self); + + 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 = self, + .state = state, + .obj = obj, + .func = func + }; + rb_rescue(generate_json_try, (VALUE)&data, generate_json_rescue, (VALUE)&data); + + return fbuffer_finalize(&buffer); +} + +/* call-seq: + * generate(obj) -> String + * generate(obj, anIO) -> anIO + * + * Generates a valid JSON document from object +obj+ and returns the + * result. If no valid JSON document can be created this method raises a + * GeneratorError exception. + */ +static VALUE cState_generate(int argc, VALUE *argv, VALUE self) +{ + rb_check_arity(argc, 1, 2); + VALUE obj = argv[0]; + VALUE io = argc > 1 ? argv[1] : Qnil; + VALUE result = cState_partial_generate(self, obj, generate_json, io); + GET_STATE(self); + (void)state; + return result; +} + +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`"); + return self; +} + +/* + * call-seq: initialize_copy(orig) + * + * Initializes this object from orig if it can be duplicated/cloned and returns + * it. +*/ +static VALUE cState_init_copy(VALUE obj, VALUE orig) +{ + JSON_Generator_State *objState, *origState; + + if (obj == orig) return obj; + GET_STATE_TO(obj, objState); + GET_STATE_TO(orig, origState); + if (!objState) rb_raise(rb_eArgError, "unallocated JSON::State"); + + MEMCPY(objState, origState, JSON_Generator_State, 1); + objState->indent = origState->indent; + objState->space = origState->space; + objState->space_before = origState->space_before; + objState->object_nl = origState->object_nl; + objState->array_nl = origState->array_nl; + objState->as_json = origState->as_json; + return obj; +} + +/* + * call-seq: from_state(opts) + * + * Creates a State object from _opts_, which ought to be Hash to create a + * new State instance configured by _opts_, something else to create an + * unconfigured instance. If _opts_ is a State object, it is just returned. + */ +static VALUE cState_from_state_s(VALUE self, VALUE opts) +{ + if (rb_obj_is_kind_of(opts, self)) { + return opts; + } else if (rb_obj_is_kind_of(opts, rb_cHash)) { + return rb_funcall(self, i_new, 1, opts); + } else { + return rb_class_new_instance(0, NULL, cState); + } +} + +/* + * call-seq: indent() + * + * Returns the string that is used to indent levels in the JSON text. + */ +static VALUE cState_indent(VALUE self) +{ + GET_STATE(self); + return state->indent ? state->indent : rb_str_freeze(rb_utf8_str_new("", 0)); +} + +static VALUE string_config(VALUE config) +{ + if (RTEST(config)) { + Check_Type(config, T_STRING); + if (RSTRING_LEN(config)) { + return rb_str_new_frozen(config); + } + } + return Qfalse; +} + +/* + * call-seq: indent=(indent) + * + * Sets the string that is used to indent levels in the JSON text. + */ +static VALUE cState_indent_set(VALUE self, VALUE indent) +{ + GET_STATE(self); + RB_OBJ_WRITE(self, &state->indent, string_config(indent)); + return Qnil; +} + +/* + * call-seq: space() + * + * Returns the string that is used to insert a space between the tokens in a JSON + * string. + */ +static VALUE cState_space(VALUE self) +{ + GET_STATE(self); + return state->space ? state->space : rb_str_freeze(rb_utf8_str_new("", 0)); +} + +/* + * call-seq: space=(space) + * + * Sets _space_ to the string that is used to insert a space between the tokens in a JSON + * string. + */ +static VALUE cState_space_set(VALUE self, VALUE space) +{ + GET_STATE(self); + RB_OBJ_WRITE(self, &state->space, string_config(space)); + return Qnil; +} + +/* + * call-seq: space_before() + * + * Returns the string that is used to insert a space before the ':' in JSON objects. + */ +static VALUE cState_space_before(VALUE self) +{ + GET_STATE(self); + return state->space_before ? state->space_before : rb_str_freeze(rb_utf8_str_new("", 0)); +} + +/* + * call-seq: space_before=(space_before) + * + * Sets the string that is used to insert a space before the ':' in JSON objects. + */ +static VALUE cState_space_before_set(VALUE self, VALUE space_before) +{ + GET_STATE(self); + RB_OBJ_WRITE(self, &state->space_before, string_config(space_before)); + return Qnil; +} + +/* + * call-seq: object_nl() + * + * This string is put at the end of a line that holds a JSON object (or + * Hash). + */ +static VALUE cState_object_nl(VALUE self) +{ + GET_STATE(self); + return state->object_nl ? state->object_nl : rb_str_freeze(rb_utf8_str_new("", 0)); +} + +/* + * call-seq: object_nl=(object_nl) + * + * This string is put at the end of a line that holds a JSON object (or + * Hash). + */ +static VALUE cState_object_nl_set(VALUE self, VALUE object_nl) +{ + GET_STATE(self); + RB_OBJ_WRITE(self, &state->object_nl, string_config(object_nl)); + return Qnil; +} + +/* + * call-seq: array_nl() + * + * This string is put at the end of a line that holds a JSON array. + */ +static VALUE cState_array_nl(VALUE self) +{ + GET_STATE(self); + return state->array_nl ? state->array_nl : rb_str_freeze(rb_utf8_str_new("", 0)); +} + +/* + * call-seq: array_nl=(array_nl) + * + * This string is put at the end of a line that holds a JSON array. + */ +static VALUE cState_array_nl_set(VALUE self, VALUE array_nl) +{ + GET_STATE(self); + RB_OBJ_WRITE(self, &state->array_nl, string_config(array_nl)); + return Qnil; +} + +/* + * call-seq: as_json() + * + * This string is put at the end of a line that holds a JSON array. + */ +static VALUE cState_as_json(VALUE self) +{ + GET_STATE(self); + return state->as_json; +} + +/* + * call-seq: as_json=(as_json) + * + * This string is put at the end of a line that holds a JSON array. + */ +static VALUE cState_as_json_set(VALUE self, VALUE as_json) +{ + GET_STATE(self); + RB_OBJ_WRITE(self, &state->as_json, rb_convert_type(as_json, T_DATA, "Proc", "to_proc")); + return Qnil; +} + +/* +* call-seq: check_circular? +* +* Returns true, if circular data structures should be checked, +* otherwise returns false. +*/ +static VALUE cState_check_circular_p(VALUE self) +{ + GET_STATE(self); + return state->max_nesting ? Qtrue : Qfalse; +} + +/* + * call-seq: max_nesting + * + * This integer returns the maximum level of data structure nesting in + * the generated JSON, max_nesting = 0 if no maximum is checked. + */ +static VALUE cState_max_nesting(VALUE self) +{ + GET_STATE(self); + return LONG2FIX(state->max_nesting); +} + +static long long_config(VALUE num) +{ + return RTEST(num) ? FIX2LONG(num) : 0; +} + +/* + * call-seq: max_nesting=(depth) + * + * This sets the maximum level of data structure nesting in the generated JSON + * to the integer depth, max_nesting = 0 if no maximum should be checked. + */ +static VALUE cState_max_nesting_set(VALUE self, VALUE depth) +{ + GET_STATE(self); + state->max_nesting = long_config(depth); + return Qnil; +} + +/* + * call-seq: script_safe + * + * If this boolean is true, the forward slashes will be escaped in + * the json output. + */ +static VALUE cState_script_safe(VALUE self) +{ + GET_STATE(self); + return state->script_safe ? Qtrue : Qfalse; +} + +/* + * call-seq: script_safe=(enable) + * + * This sets whether or not the forward slashes will be escaped in + * the json output. + */ +static VALUE cState_script_safe_set(VALUE self, VALUE enable) +{ + GET_STATE(self); + state->script_safe = RTEST(enable); + return Qnil; +} + +/* + * call-seq: strict + * + * If this boolean is false, types unsupported by the JSON format will + * be serialized as strings. + * If this boolean is true, types unsupported by the JSON format will + * raise a JSON::GeneratorError. + */ +static VALUE cState_strict(VALUE self) +{ + GET_STATE(self); + return state->strict ? Qtrue : Qfalse; +} + +/* + * call-seq: strict=(enable) + * + * This sets whether or not to serialize types unsupported by the + * JSON format as strings. + * If this boolean is false, types unsupported by the JSON format will + * be serialized as strings. + * If this boolean is true, types unsupported by the JSON format will + * raise a JSON::GeneratorError. + */ +static VALUE cState_strict_set(VALUE self, VALUE enable) +{ + GET_STATE(self); + state->strict = RTEST(enable); + return Qnil; +} + +/* + * call-seq: allow_nan? + * + * Returns true, if NaN, Infinity, and -Infinity should be generated, otherwise + * returns false. + */ +static VALUE cState_allow_nan_p(VALUE self) +{ + GET_STATE(self); + return state->allow_nan ? Qtrue : Qfalse; +} + +/* + * call-seq: allow_nan=(enable) + * + * This sets whether or not to serialize NaN, Infinity, and -Infinity + */ +static VALUE cState_allow_nan_set(VALUE self, VALUE enable) +{ + GET_STATE(self); + state->allow_nan = RTEST(enable); + return Qnil; +} + +/* + * call-seq: ascii_only? + * + * Returns true, if only ASCII characters should be generated. Otherwise + * returns false. + */ +static VALUE cState_ascii_only_p(VALUE self) +{ + GET_STATE(self); + return state->ascii_only ? Qtrue : Qfalse; +} + +/* + * call-seq: ascii_only=(enable) + * + * This sets whether only ASCII characters should be generated. + */ +static VALUE cState_ascii_only_set(VALUE self, VALUE enable) +{ + GET_STATE(self); + state->ascii_only = RTEST(enable); + return Qnil; +} + +/* + * call-seq: depth + * + * This integer returns the current depth of data structure nesting. + */ +static VALUE cState_depth(VALUE self) +{ + GET_STATE(self); + return LONG2FIX(state->depth); +} + +/* + * call-seq: depth=(depth) + * + * This sets the maximum level of data structure nesting in the generated JSON + * to the integer depth, max_nesting = 0 if no maximum should be checked. + */ +static VALUE cState_depth_set(VALUE self, VALUE depth) +{ + GET_STATE(self); + state->depth = long_config(depth); + return Qnil; +} + +/* + * call-seq: buffer_initial_length + * + * This integer returns the current initial length of the buffer. + */ +static VALUE cState_buffer_initial_length(VALUE self) +{ + GET_STATE(self); + return LONG2FIX(state->buffer_initial_length); +} + +static void buffer_initial_length_set(JSON_Generator_State *state, VALUE buffer_initial_length) +{ + Check_Type(buffer_initial_length, T_FIXNUM); + long initial_length = FIX2LONG(buffer_initial_length); + if (initial_length > 0) { + state->buffer_initial_length = initial_length; + } +} + +/* + * call-seq: buffer_initial_length=(length) + * + * This sets the initial length of the buffer to +length+, if +length+ > 0, + * otherwise its value isn't changed. + */ +static VALUE cState_buffer_initial_length_set(VALUE self, VALUE buffer_initial_length) +{ + GET_STATE(self); + buffer_initial_length_set(state, buffer_initial_length); + return Qnil; +} + +static int configure_state_i(VALUE key, VALUE val, VALUE _arg) +{ + JSON_Generator_State *state = (JSON_Generator_State *)_arg; + + if (key == sym_indent) { state->indent = string_config(val); } + else if (key == sym_space) { state->space = string_config(val); } + else if (key == sym_space_before) { state->space_before = string_config(val); } + else if (key == sym_object_nl) { state->object_nl = string_config(val); } + else if (key == sym_array_nl) { state->array_nl = string_config(val); } + 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_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); } + else if (key == sym_strict) { state->strict = RTEST(val); } + else if (key == sym_as_json) { state->as_json = RTEST(val) ? rb_convert_type(val, T_DATA, "Proc", "to_proc") : Qfalse; } + return ST_CONTINUE; +} + +static void configure_state(JSON_Generator_State *state, VALUE config) +{ + if (!RTEST(config)) return; + + Check_Type(config, T_HASH); + + if (!RHASH_SIZE(config)) return; + + // We assume in most cases few keys are set so it's faster to go over + // the provided keys than to check all possible keys. + rb_hash_foreach(config, configure_state_i, (VALUE)state); +} + +static VALUE cState_configure(VALUE self, VALUE opts) +{ + GET_STATE(self); + configure_state(state, opts); + return self; +} + +static VALUE cState_m_generate(VALUE klass, VALUE obj, VALUE opts, VALUE io) +{ + JSON_Generator_State state = {0}; + state_init(&state); + configure_state(&state, opts); + + 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 = &state, + .obj = obj, + .func = generate_json, + }; + rb_rescue(generate_json_try, (VALUE)&data, generate_json_rescue, (VALUE)&data); + + return fbuffer_finalize(&buffer); +} + +/* + * + */ +void Init_generator(void) +{ +#ifdef HAVE_RB_EXT_RACTOR_SAFE + rb_ext_ractor_safe(true); +#endif + +#undef rb_intern + rb_require("json/common"); + + mJSON = rb_define_module("JSON"); + + rb_global_variable(&cFragment); + cFragment = rb_const_get(mJSON, rb_intern("Fragment")); + + VALUE mExt = rb_define_module_under(mJSON, "Ext"); + VALUE mGenerator = rb_define_module_under(mExt, "Generator"); + + rb_global_variable(&eGeneratorError); + eGeneratorError = rb_path2class("JSON::GeneratorError"); + + rb_global_variable(&eNestingError); + eNestingError = rb_path2class("JSON::NestingError"); + + cState = rb_define_class_under(mGenerator, "State", rb_cObject); + rb_define_alloc_func(cState, cState_s_allocate); + rb_define_singleton_method(cState, "from_state", cState_from_state_s, 1); + rb_define_method(cState, "initialize", cState_initialize, -1); + rb_define_alias(cState, "initialize", "initialize"); // avoid method redefinition warnings + rb_define_private_method(cState, "_configure", cState_configure, 1); + + rb_define_method(cState, "initialize_copy", cState_init_copy, 1); + rb_define_method(cState, "indent", cState_indent, 0); + rb_define_method(cState, "indent=", cState_indent_set, 1); + rb_define_method(cState, "space", cState_space, 0); + rb_define_method(cState, "space=", cState_space_set, 1); + rb_define_method(cState, "space_before", cState_space_before, 0); + rb_define_method(cState, "space_before=", cState_space_before_set, 1); + rb_define_method(cState, "object_nl", cState_object_nl, 0); + rb_define_method(cState, "object_nl=", cState_object_nl_set, 1); + rb_define_method(cState, "array_nl", cState_array_nl, 0); + rb_define_method(cState, "array_nl=", cState_array_nl_set, 1); + rb_define_method(cState, "as_json", cState_as_json, 0); + rb_define_method(cState, "as_json=", cState_as_json_set, 1); + rb_define_method(cState, "max_nesting", cState_max_nesting, 0); + rb_define_method(cState, "max_nesting=", cState_max_nesting_set, 1); + rb_define_method(cState, "script_safe", cState_script_safe, 0); + rb_define_method(cState, "script_safe?", cState_script_safe, 0); + rb_define_method(cState, "script_safe=", cState_script_safe_set, 1); + rb_define_alias(cState, "escape_slash", "script_safe"); + rb_define_alias(cState, "escape_slash?", "script_safe?"); + rb_define_alias(cState, "escape_slash=", "script_safe="); + rb_define_method(cState, "strict", cState_strict, 0); + rb_define_method(cState, "strict?", cState_strict, 0); + rb_define_method(cState, "strict=", cState_strict_set, 1); + rb_define_method(cState, "check_circular?", cState_check_circular_p, 0); + rb_define_method(cState, "allow_nan?", cState_allow_nan_p, 0); + rb_define_method(cState, "allow_nan=", cState_allow_nan_set, 1); + rb_define_method(cState, "ascii_only?", cState_ascii_only_p, 0); + rb_define_method(cState, "ascii_only=", cState_ascii_only_set, 1); + rb_define_method(cState, "depth", cState_depth, 0); + rb_define_method(cState, "depth=", cState_depth_set, 1); + 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_alias(cState, "generate_new", "generate"); // :nodoc: + + 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_singleton_method(mString, "included", mString_included_s, 1); + rb_define_method(mString, "to_json", mString_to_json, -1); + rb_define_method(mString, "to_json_raw", mString_to_json_raw, -1); + rb_define_method(mString, "to_json_raw_object", mString_to_json_raw_object, 0); + + mString_Extend = rb_define_module_under(mString, "Extend"); + rb_define_method(mString_Extend, "json_create", mString_Extend_json_create, 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_global_variable(&Encoding_UTF_8); + Encoding_UTF_8 = rb_const_get(rb_path2class("Encoding"), rb_intern("UTF_8")); + + 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")); + sym_space = ID2SYM(rb_intern("space")); + sym_space_before = ID2SYM(rb_intern("space_before")); + sym_object_nl = ID2SYM(rb_intern("object_nl")); + sym_array_nl = ID2SYM(rb_intern("array_nl")); + sym_max_nesting = ID2SYM(rb_intern("max_nesting")); + sym_allow_nan = ID2SYM(rb_intern("allow_nan")); + sym_ascii_only = ID2SYM(rb_intern("ascii_only")); + sym_depth = ID2SYM(rb_intern("depth")); + sym_buffer_initial_length = ID2SYM(rb_intern("buffer_initial_length")); + sym_script_safe = ID2SYM(rb_intern("script_safe")); + sym_escape_slash = ID2SYM(rb_intern("escape_slash")); + sym_strict = ID2SYM(rb_intern("strict")); + sym_as_json = ID2SYM(rb_intern("as_json")); + + usascii_encindex = rb_usascii_encindex(); + utf8_encindex = rb_utf8_encindex(); + binary_encindex = rb_ascii8bit_encindex(); + + 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; + } +} diff --git a/vendor/bundle/ruby/3.4.0/gems/json-2.12.2/ext/json/ext/generator/simd.h b/vendor/bundle/ruby/3.4.0/gems/json-2.12.2/ext/json/ext/generator/simd.h new file mode 100644 index 0000000..b12890c --- /dev/null +++ b/vendor/bundle/ruby/3.4.0/gems/json-2.12.2/ext/json/ext/generator/simd.h @@ -0,0 +1,112 @@ +typedef enum { + SIMD_NONE, + SIMD_NEON, + SIMD_SSE2 +} SIMD_Implementation; + +#ifdef JSON_ENABLE_SIMD + +#ifdef __clang__ + #if __has_builtin(__builtin_ctzll) + #define HAVE_BUILTIN_CTZLL 1 + #else + #define HAVE_BUILTIN_CTZLL 0 + #endif +#elif defined(__GNUC__) && (__GNUC__ > 4 || (__GNUC__ == 4 && __GNUC_MINOR__ >= 3)) + #define HAVE_BUILTIN_CTZLL 1 +#else + #define HAVE_BUILTIN_CTZLL 0 +#endif + +static inline uint32_t trailing_zeros64(uint64_t input) { +#if HAVE_BUILTIN_CTZLL + return __builtin_ctzll(input); +#else + uint32_t trailing_zeros = 0; + uint64_t temp = input; + while ((temp & 1) == 0 && temp > 0) { + trailing_zeros++; + temp >>= 1; + } + return trailing_zeros; +#endif +} + +static inline int trailing_zeros(int input) { + #if HAVE_BUILTIN_CTZLL + return __builtin_ctz(input); + #else + int trailing_zeros = 0; + int temp = input; + while ((temp & 1) == 0 && temp > 0) { + trailing_zeros++; + temp >>= 1; + } + return trailing_zeros; + #endif +} + +#define SIMD_MINIMUM_THRESHOLD 6 + +#if defined(__ARM_NEON) || defined(__ARM_NEON__) || defined(__aarch64__) || defined(_M_ARM64) +#include <arm_neon.h> + +#define FIND_SIMD_IMPLEMENTATION_DEFINED 1 +static SIMD_Implementation find_simd_implementation(void) { + return SIMD_NEON; +} + +#define HAVE_SIMD 1 +#define HAVE_SIMD_NEON 1 + +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) + +#ifdef HAVE_X86INTRIN_H +#include <x86intrin.h> + +#define HAVE_SIMD 1 +#define HAVE_SIMD_SSE2 1 + +#ifdef HAVE_CPUID_H +#define FIND_SIMD_IMPLEMENTATION_DEFINED 1 + +#include <cpuid.h> +#endif /* HAVE_CPUID_H */ + +static SIMD_Implementation find_simd_implementation(void) { + +#if defined(__GNUC__ ) || defined(__clang__) +#ifdef __GNUC__ + __builtin_cpu_init(); +#endif /* __GNUC__ */ + + // TODO Revisit. I think the SSE version now only uses SSE2 instructions. + if (__builtin_cpu_supports("sse2")) { + return SIMD_SSE2; + } +#endif /* __GNUC__ || __clang__*/ + + return SIMD_NONE; +} + +#endif /* HAVE_X86INTRIN_H */ +#endif /* X86_64 Support */ + +#endif /* JSON_ENABLE_SIMD */ + +#ifndef FIND_SIMD_IMPLEMENTATION_DEFINED +static SIMD_Implementation find_simd_implementation(void) { + return SIMD_NONE; +} +#endif diff --git a/vendor/bundle/ruby/3.4.0/gems/json-2.12.2/ext/json/ext/parser/Makefile b/vendor/bundle/ruby/3.4.0/gems/json-2.12.2/ext/json/ext/parser/Makefile new file mode 100644 index 0000000..d6c0672 --- /dev/null +++ b/vendor/bundle/ruby/3.4.0/gems/json-2.12.2/ext/json/ext/parser/Makefile @@ -0,0 +1,273 @@ + +SHELL = /bin/sh + +# V=0 quiet, V=1 verbose. other values don't work. +V = 0 +V0 = $(V:0=) +Q1 = $(V:1=) +Q = $(Q1:0=@) +ECHO1 = $(V:1=@ :) +ECHO = $(ECHO1:0=@ echo) +NULLCMD = : + +#### Start of system configuration section. #### + +srcdir = . +topdir = /Users/ben/.rubies/ruby-3.4.1/include/ruby-3.4.0 +hdrdir = $(topdir) +arch_hdrdir = /Users/ben/.rubies/ruby-3.4.1/include/ruby-3.4.0/arm64-darwin24 +PATH_SEPARATOR = : +VPATH = $(srcdir):$(arch_hdrdir)/ruby:$(hdrdir)/ruby +prefix = $(DESTDIR)/Users/ben/.rubies/ruby-3.4.1 +rubysitearchprefix = $(rubylibprefix)/$(sitearch) +rubyarchprefix = $(rubylibprefix)/$(arch) +rubylibprefix = $(libdir)/$(RUBY_BASE_NAME) +exec_prefix = $(prefix) +vendorarchhdrdir = $(vendorhdrdir)/$(sitearch) +sitearchhdrdir = $(sitehdrdir)/$(sitearch) +rubyarchhdrdir = $(rubyhdrdir)/$(arch) +vendorhdrdir = $(rubyhdrdir)/vendor_ruby +sitehdrdir = $(rubyhdrdir)/site_ruby +rubyhdrdir = $(includedir)/$(RUBY_VERSION_NAME) +vendorarchdir = $(vendorlibdir)/$(sitearch) +vendorlibdir = $(vendordir)/$(ruby_version) +vendordir = $(rubylibprefix)/vendor_ruby +sitearchdir = $(sitelibdir)/$(sitearch) +sitelibdir = $(sitedir)/$(ruby_version) +sitedir = $(rubylibprefix)/site_ruby +rubyarchdir = $(rubylibdir)/$(arch) +rubylibdir = $(rubylibprefix)/$(ruby_version) +sitearchincludedir = $(includedir)/$(sitearch) +archincludedir = $(includedir)/$(arch) +sitearchlibdir = $(libdir)/$(sitearch) +archlibdir = $(libdir)/$(arch) +ridir = $(datarootdir)/$(RI_BASE_NAME) +modular_gc_dir = $(DESTDIR) +mandir = $(datarootdir)/man +localedir = $(datarootdir)/locale +libdir = $(exec_prefix)/lib +psdir = $(docdir) +pdfdir = $(docdir) +dvidir = $(docdir) +htmldir = $(docdir) +infodir = $(datarootdir)/info +docdir = $(datarootdir)/doc/$(PACKAGE) +oldincludedir = $(DESTDIR)/usr/include +includedir = $(SDKROOT)$(prefix)/include +runstatedir = $(localstatedir)/run +localstatedir = $(prefix)/var +sharedstatedir = $(prefix)/com +sysconfdir = $(prefix)/etc +datadir = $(datarootdir) +datarootdir = $(prefix)/share +libexecdir = $(exec_prefix)/libexec +sbindir = $(exec_prefix)/sbin +bindir = $(exec_prefix)/bin +archdir = $(rubyarchdir) + + +CC_WRAPPER = +CC = clang +CXX = clang++ -std=gnu++11 +LIBRUBY = $(LIBRUBY_A) +LIBRUBY_A = lib$(RUBY_SO_NAME)-static.a +LIBRUBYARG_SHARED = +LIBRUBYARG_STATIC = -l$(RUBY_SO_NAME)-static -framework CoreFoundation $(MAINLIBS) +empty = +OUTFLAG = -o $(empty) +COUTFLAG = -o $(empty) +CSRCFLAG = $(empty) + +RUBY_EXTCONF_H = +cflags = $(hardenflags) -fdeclspec $(optflags) $(debugflags) $(warnflags) +cxxflags = +optflags = -O3 -fno-fast-math +debugflags = -ggdb3 +warnflags = -Wall -Wextra -Wextra-tokens -Wdeprecated-declarations -Wdivision-by-zero -Wdiv-by-zero -Wimplicit-function-declaration -Wimplicit-int -Wpointer-arith -Wshorten-64-to-32 -Wwrite-strings -Wold-style-definition -Wmissing-noreturn -Wno-cast-function-type -Wno-constant-logical-operand -Wno-long-long -Wno-missing-field-initializers -Wno-overlength-strings -Wno-parentheses-equality -Wno-self-assign -Wno-tautological-compare -Wno-unused-parameter -Wno-unused-value -Wunused-variable -Wmisleading-indentation -Wundef +cppflags = +CCDLFLAGS = -fno-common +CFLAGS = $(CCDLFLAGS) $(cflags) -pipe -std=c99 $(ARCH_FLAG) +INCFLAGS = -I. -I$(arch_hdrdir) -I$(hdrdir)/ruby/backward -I$(hdrdir) -I$(srcdir) +DEFS = +CPPFLAGS = -DHAVE_RB_ENC_INTERNED_STR -DHAVE_RB_HASH_NEW_CAPA -DHAVE_RB_HASH_BULK_INSERT -DHAVE_STRNLEN -I/opt/homebrew/opt/readline/include -I/opt/homebrew/opt/libyaml/include -I/opt/homebrew/opt/gdbm/include -D_XOPEN_SOURCE -D_DARWIN_C_SOURCE -D_DARWIN_UNLIMITED_SELECT -D_REENTRANT $(DEFS) $(cppflags) +CXXFLAGS = $(CCDLFLAGS) -fdeclspec $(ARCH_FLAG) +ldflags = -L. -fstack-protector-strong -L/opt/homebrew/Cellar/readline/8.2.13/lib -L/opt/homebrew/Cellar/libyaml/0.2.5/lib -L/opt/homebrew/Cellar/gdbm/1.24/lib +dldflags = -L/opt/homebrew/Cellar/readline/8.2.13/lib -L/opt/homebrew/Cellar/libyaml/0.2.5/lib -L/opt/homebrew/Cellar/gdbm/1.24/lib -Wl,-undefined,dynamic_lookup -bundle_loader '$(BUILTRUBY)' +ARCH_FLAG = -arch arm64 +DLDFLAGS = $(ldflags) $(dldflags) $(ARCH_FLAG) +LDSHARED = $(CC) -dynamic -bundle +LDSHAREDXX = $(CXX) -dynamic -bundle +POSTLINK = dsymutil $@ 2>/dev/null; { test -z '$(RUBY_CODESIGN)' || codesign -s '$(RUBY_CODESIGN)' $@; } +AR = ar +LD = ld +EXEEXT = + +RUBY_INSTALL_NAME = $(RUBY_BASE_NAME) +RUBY_SO_NAME = ruby.3.4 +RUBYW_INSTALL_NAME = +RUBY_VERSION_NAME = $(RUBY_BASE_NAME)-$(ruby_version) +RUBYW_BASE_NAME = rubyw +RUBY_BASE_NAME = ruby + +arch = arm64-darwin24 +sitearch = $(arch) +ruby_version = 3.4.0 +ruby = $(bindir)/$(RUBY_BASE_NAME) +RUBY = $(ruby) +BUILTRUBY = $(bindir)/$(RUBY_BASE_NAME) +ruby_headers = $(hdrdir)/ruby.h $(hdrdir)/ruby/backward.h $(hdrdir)/ruby/ruby.h $(hdrdir)/ruby/defines.h $(hdrdir)/ruby/missing.h $(hdrdir)/ruby/intern.h $(hdrdir)/ruby/st.h $(hdrdir)/ruby/subst.h $(arch_hdrdir)/ruby/config.h + +RM = rm -f +RM_RF = rm -fr +RMDIRS = rmdir -p +MAKEDIRS = mkdir -p +INSTALL = /usr/bin/install -c +INSTALL_PROG = $(INSTALL) -m 0755 +INSTALL_DATA = $(INSTALL) -m 644 +COPY = cp +TOUCH = exit > + +#### End of system configuration section. #### + +preload = +libpath = . $(libdir) /opt/homebrew/opt/readline/lib /opt/homebrew/opt/libyaml/lib /opt/homebrew/opt/gdbm/lib +LIBPATH = -L. -L$(libdir) -L/opt/homebrew/opt/readline/lib -L/opt/homebrew/opt/libyaml/lib -L/opt/homebrew/opt/gdbm/lib +DEFFILE = + +CLEANFILES = mkmf.log +DISTCLEANFILES = +DISTCLEANDIRS = + +extout = +extout_prefix = +target_prefix = /json/ext +LOCAL_LIBS = +LIBS = -lpthread +ORIG_SRCS = parser.c +SRCS = $(ORIG_SRCS) +OBJS = parser.o +HDRS = +LOCAL_HDRS = +TARGET = parser +TARGET_NAME = parser +TARGET_ENTRY = Init_$(TARGET_NAME) +DLLIB = $(TARGET).bundle +EXTSTATIC = +STATIC_LIB = + +TIMESTAMP_DIR = . +BINDIR = $(bindir) +RUBYCOMMONDIR = $(sitedir)$(target_prefix) +RUBYLIBDIR = $(sitelibdir)$(target_prefix) +RUBYARCHDIR = $(sitearchdir)$(target_prefix) +HDRDIR = $(sitehdrdir)$(target_prefix) +ARCHHDRDIR = $(sitearchhdrdir)$(target_prefix) +TARGET_SO_DIR = +TARGET_SO = $(TARGET_SO_DIR)$(DLLIB) +CLEANLIBS = $(TARGET_SO) $(TARGET_SO:=.dSYM) +CLEANOBJS = $(OBJS) *.bak +TARGET_SO_DIR_TIMESTAMP = $(TIMESTAMP_DIR)/.sitearchdir.-.json.-.ext.time + +all: $(DLLIB) +static: $(STATIC_LIB) +.PHONY: all install static install-so install-rb +.PHONY: clean clean-so clean-static clean-rb + +clean-static:: +clean-rb-default:: +clean-rb:: +clean-so:: +clean: clean-so clean-static clean-rb-default clean-rb + -$(Q)$(RM_RF) $(CLEANLIBS) $(CLEANOBJS) $(CLEANFILES) .*.time + +distclean-rb-default:: +distclean-rb:: +distclean-so:: +distclean-static:: +distclean: clean distclean-so distclean-static distclean-rb-default distclean-rb + -$(Q)$(RM) Makefile $(RUBY_EXTCONF_H) conftest.* mkmf.log + -$(Q)$(RM) core ruby$(EXEEXT) *~ $(DISTCLEANFILES) + -$(Q)$(RMDIRS) $(DISTCLEANDIRS) 2> /dev/null || true + +realclean: distclean +install: install-so install-rb + +install-so: $(DLLIB) $(TARGET_SO_DIR_TIMESTAMP) + $(INSTALL_PROG) $(DLLIB) $(RUBYARCHDIR) +clean-static:: + -$(Q)$(RM) $(STATIC_LIB) +install-rb: pre-install-rb do-install-rb install-rb-default +install-rb-default: pre-install-rb-default do-install-rb-default +pre-install-rb: Makefile +pre-install-rb-default: Makefile +do-install-rb: +do-install-rb-default: +pre-install-rb-default: + @$(NULLCMD) +$(TARGET_SO_DIR_TIMESTAMP): + $(Q) $(MAKEDIRS) $(@D) $(RUBYARCHDIR) + $(Q) $(TOUCH) $@ + +site-install: site-install-so site-install-rb +site-install-so: install-so +site-install-rb: install-rb + +.SUFFIXES: .c .m .cc .mm .cxx .cpp .o .S + +.cc.o: + $(ECHO) compiling $(<) + $(Q) $(CXX) $(INCFLAGS) $(CPPFLAGS) $(CXXFLAGS) $(COUTFLAG)$@ -c $(CSRCFLAG)$< + +.cc.S: + $(ECHO) translating $(<) + $(Q) $(CXX) $(INCFLAGS) $(CPPFLAGS) $(CXXFLAGS) $(COUTFLAG)$@ -S $(CSRCFLAG)$< + +.mm.o: + $(ECHO) compiling $(<) + $(Q) $(CXX) $(INCFLAGS) $(CPPFLAGS) $(CXXFLAGS) $(COUTFLAG)$@ -c $(CSRCFLAG)$< + +.mm.S: + $(ECHO) translating $(<) + $(Q) $(CXX) $(INCFLAGS) $(CPPFLAGS) $(CXXFLAGS) $(COUTFLAG)$@ -S $(CSRCFLAG)$< + +.cxx.o: + $(ECHO) compiling $(<) + $(Q) $(CXX) $(INCFLAGS) $(CPPFLAGS) $(CXXFLAGS) $(COUTFLAG)$@ -c $(CSRCFLAG)$< + +.cxx.S: + $(ECHO) translating $(<) + $(Q) $(CXX) $(INCFLAGS) $(CPPFLAGS) $(CXXFLAGS) $(COUTFLAG)$@ -S $(CSRCFLAG)$< + +.cpp.o: + $(ECHO) compiling $(<) + $(Q) $(CXX) $(INCFLAGS) $(CPPFLAGS) $(CXXFLAGS) $(COUTFLAG)$@ -c $(CSRCFLAG)$< + +.cpp.S: + $(ECHO) translating $(<) + $(Q) $(CXX) $(INCFLAGS) $(CPPFLAGS) $(CXXFLAGS) $(COUTFLAG)$@ -S $(CSRCFLAG)$< + +.c.o: + $(ECHO) compiling $(<) + $(Q) $(CC) $(INCFLAGS) $(CPPFLAGS) $(CFLAGS) $(COUTFLAG)$@ -c $(CSRCFLAG)$< + +.c.S: + $(ECHO) translating $(<) + $(Q) $(CC) $(INCFLAGS) $(CPPFLAGS) $(CFLAGS) $(COUTFLAG)$@ -S $(CSRCFLAG)$< + +.m.o: + $(ECHO) compiling $(<) + $(Q) $(CC) $(INCFLAGS) $(CPPFLAGS) $(CFLAGS) $(COUTFLAG)$@ -c $(CSRCFLAG)$< + +.m.S: + $(ECHO) translating $(<) + $(Q) $(CC) $(INCFLAGS) $(CPPFLAGS) $(CFLAGS) $(COUTFLAG)$@ -S $(CSRCFLAG)$< + +$(TARGET_SO): $(OBJS) Makefile + $(ECHO) linking shared-object json/ext/$(DLLIB) + -$(Q)$(RM) $(@) + $(Q) $(LDSHARED) -o $@ $(OBJS) $(LIBPATH) $(DLDFLAGS) $(LOCAL_LIBS) $(LIBS) + $(Q) $(POSTLINK) + + + +$(OBJS): $(HDRS) $(ruby_headers) diff --git a/vendor/bundle/ruby/3.4.0/gems/json-2.12.2/ext/json/ext/parser/extconf.rb b/vendor/bundle/ruby/3.4.0/gems/json-2.12.2/ext/json/ext/parser/extconf.rb new file mode 100644 index 0000000..09c9637 --- /dev/null +++ b/vendor/bundle/ruby/3.4.0/gems/json-2.12.2/ext/json/ext/parser/extconf.rb @@ -0,0 +1,11 @@ +# frozen_string_literal: true +require 'mkmf' + +have_func("rb_enc_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") + +create_makefile 'json/ext/parser' diff --git a/vendor/bundle/ruby/3.4.0/gems/json-2.12.2/ext/json/ext/parser/parser.c b/vendor/bundle/ruby/3.4.0/gems/json-2.12.2/ext/json/ext/parser/parser.c new file mode 100644 index 0000000..c5f3001 --- /dev/null +++ b/vendor/bundle/ruby/3.4.0/gems/json-2.12.2/ext/json/ext/parser/parser.c @@ -0,0 +1,1416 @@ +#include "ruby.h" +#include "ruby/encoding.h" + +/* shims */ +/* This is the fallback definition from Ruby 3.4 */ + +#ifndef RBIMPL_STDBOOL_H +#if defined(__cplusplus) +# if defined(HAVE_STDBOOL_H) && (__cplusplus >= 201103L) +# include <cstdbool> +# endif +#elif defined(HAVE_STDBOOL_H) +# include <stdbool.h> +#elif !defined(HAVE__BOOL) +typedef unsigned char _Bool; +# define bool _Bool +# define true ((_Bool)+1) +# define false ((_Bool)+0) +# define __bool_true_false_are_defined +#endif +#endif + +#ifndef RB_UNLIKELY +#define RB_UNLIKELY(expr) expr +#endif + +#ifndef RB_LIKELY +#define RB_LIKELY(expr) expr +#endif + +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 VALUE sym_max_nesting, sym_allow_nan, sym_allow_trailing_comma, sym_symbolize_names, sym_freeze, + sym_decimal_class, sym_on_load; + +static int binary_encindex; +static int utf8_encindex; + +#ifndef HAVE_RB_HASH_BULK_INSERT +// For TruffleRuby +void +rb_hash_bulk_insert(long count, const VALUE *pairs, VALUE hash) +{ + long index = 0; + while (index < count) { + VALUE name = pairs[index++]; + VALUE value = pairs[index++]; + rb_hash_aset(hash, name, value); + } + RB_GC_GUARD(hash); +} +#endif + +#ifndef HAVE_RB_HASH_NEW_CAPA +#define rb_hash_new_capa(n) rb_hash_new() +#endif + + +/* name cache */ + +#include <string.h> +#include <ctype.h> + +// Object names are likely to be repeated, and are frozen. +// As such we can re-use them if we keep a cache of the ones we've seen so far, +// and save much more expensive lookups into the global fstring table. +// This cache implementation is deliberately simple, as we're optimizing for compactness, +// to be able to fit safely on the stack. +// As such, binary search into a sorted array gives a good tradeoff between compactness and +// performance. +#define JSON_RVALUE_CACHE_CAPA 63 +typedef struct rvalue_cache_struct { + int length; + VALUE entries[JSON_RVALUE_CACHE_CAPA]; +} rvalue_cache; + +static rb_encoding *enc_utf8; + +#define JSON_RVALUE_CACHE_MAX_ENTRY_LENGTH 55 + +static inline VALUE build_interned_string(const char *str, const long length) +{ +# ifdef HAVE_RB_ENC_INTERNED_STR + return rb_enc_interned_str(str, length, enc_utf8); +# else + VALUE rstring = rb_utf8_str_new(str, length); + return rb_funcall(rb_str_freeze(rstring), i_uminus, 0); +# endif +} + +static inline VALUE build_symbol(const char *str, const long length) +{ + return rb_str_intern(build_interned_string(str, length)); +} + +static void rvalue_cache_insert_at(rvalue_cache *cache, int index, VALUE rstring) +{ + MEMMOVE(&cache->entries[index + 1], &cache->entries[index], VALUE, cache->length - index); + cache->length++; + cache->entries[index] = rstring; +} + +static inline int rstring_cache_cmp(const char *str, const long length, VALUE rstring) +{ + long rstring_length = RSTRING_LEN(rstring); + if (length == rstring_length) { + return memcmp(str, RSTRING_PTR(rstring), length); + } else { + return (int)(length - rstring_length); + } +} + +static VALUE rstring_cache_fetch(rvalue_cache *cache, const char *str, const long length) +{ + if (RB_UNLIKELY(length > JSON_RVALUE_CACHE_MAX_ENTRY_LENGTH)) { + // Common names aren't likely to be very long. So we just don't + // cache names above an arbitrary threshold. + return Qfalse; + } + + if (RB_UNLIKELY(!isalpha((unsigned char)str[0]))) { + // Simple heuristic, if the first character isn't a letter, + // we're much less likely to see this string again. + // We mostly want to cache strings that are likely to be repeated. + return Qfalse; + } + + int low = 0; + int high = cache->length - 1; + int mid = 0; + int last_cmp = 0; + + while (low <= high) { + mid = (high + low) >> 1; + VALUE entry = cache->entries[mid]; + last_cmp = rstring_cache_cmp(str, length, entry); + + if (last_cmp == 0) { + return entry; + } else if (last_cmp > 0) { + low = mid + 1; + } else { + high = mid - 1; + } + } + + if (RB_UNLIKELY(memchr(str, '\\', length))) { + // We assume the overwhelming majority of names don't need to be escaped. + // But if they do, we have to fallback to the slow path. + return Qfalse; + } + + VALUE rstring = build_interned_string(str, length); + + if (cache->length < JSON_RVALUE_CACHE_CAPA) { + if (last_cmp > 0) { + mid += 1; + } + + rvalue_cache_insert_at(cache, mid, rstring); + } + return rstring; +} + +static VALUE rsymbol_cache_fetch(rvalue_cache *cache, const char *str, const long length) +{ + if (RB_UNLIKELY(length > JSON_RVALUE_CACHE_MAX_ENTRY_LENGTH)) { + // Common names aren't likely to be very long. So we just don't + // cache names above an arbitrary threshold. + return Qfalse; + } + + if (RB_UNLIKELY(!isalpha((unsigned char)str[0]))) { + // Simple heuristic, if the first character isn't a letter, + // we're much less likely to see this string again. + // We mostly want to cache strings that are likely to be repeated. + return Qfalse; + } + + int low = 0; + int high = cache->length - 1; + int mid = 0; + int last_cmp = 0; + + while (low <= high) { + mid = (high + low) >> 1; + VALUE entry = cache->entries[mid]; + last_cmp = rstring_cache_cmp(str, length, rb_sym2str(entry)); + + if (last_cmp == 0) { + return entry; + } else if (last_cmp > 0) { + low = mid + 1; + } else { + high = mid - 1; + } + } + + if (RB_UNLIKELY(memchr(str, '\\', length))) { + // We assume the overwhelming majority of names don't need to be escaped. + // But if they do, we have to fallback to the slow path. + return Qfalse; + } + + VALUE rsymbol = build_symbol(str, length); + + if (cache->length < JSON_RVALUE_CACHE_CAPA) { + if (last_cmp > 0) { + mid += 1; + } + + rvalue_cache_insert_at(cache, mid, rsymbol); + } + return rsymbol; +} + +/* rvalue stack */ + +#define RVALUE_STACK_INITIAL_CAPA 128 + +enum rvalue_stack_type { + RVALUE_STACK_HEAP_ALLOCATED = 0, + RVALUE_STACK_STACK_ALLOCATED = 1, +}; + +typedef struct rvalue_stack_struct { + enum rvalue_stack_type type; + long capa; + long head; + VALUE *ptr; +} rvalue_stack; + +static rvalue_stack *rvalue_stack_spill(rvalue_stack *old_stack, VALUE *handle, rvalue_stack **stack_ref); + +static rvalue_stack *rvalue_stack_grow(rvalue_stack *stack, VALUE *handle, rvalue_stack **stack_ref) +{ + long required = stack->capa * 2; + + if (stack->type == RVALUE_STACK_STACK_ALLOCATED) { + stack = rvalue_stack_spill(stack, handle, stack_ref); + } else { + REALLOC_N(stack->ptr, VALUE, required); + stack->capa = required; + } + return stack; +} + +static VALUE rvalue_stack_push(rvalue_stack *stack, VALUE value, VALUE *handle, rvalue_stack **stack_ref) +{ + if (RB_UNLIKELY(stack->head >= stack->capa)) { + stack = rvalue_stack_grow(stack, handle, stack_ref); + } + stack->ptr[stack->head] = value; + stack->head++; + return value; +} + +static inline VALUE *rvalue_stack_peek(rvalue_stack *stack, long count) +{ + return stack->ptr + (stack->head - count); +} + +static inline void rvalue_stack_pop(rvalue_stack *stack, long count) +{ + stack->head -= count; +} + +static void rvalue_stack_mark(void *ptr) +{ + rvalue_stack *stack = (rvalue_stack *)ptr; + long index; + for (index = 0; index < stack->head; index++) { + rb_gc_mark(stack->ptr[index]); + } +} + +static void rvalue_stack_free(void *ptr) +{ + rvalue_stack *stack = (rvalue_stack *)ptr; + if (stack) { + ruby_xfree(stack->ptr); + ruby_xfree(stack); + } +} + +static size_t rvalue_stack_memsize(const void *ptr) +{ + const rvalue_stack *stack = (const rvalue_stack *)ptr; + return sizeof(rvalue_stack) + sizeof(VALUE) * stack->capa; +} + +static const rb_data_type_t JSON_Parser_rvalue_stack_type = { + "JSON::Ext::Parser/rvalue_stack", + { + .dmark = rvalue_stack_mark, + .dfree = rvalue_stack_free, + .dsize = rvalue_stack_memsize, + }, + 0, 0, + RUBY_TYPED_FREE_IMMEDIATELY, +}; + +static rvalue_stack *rvalue_stack_spill(rvalue_stack *old_stack, VALUE *handle, rvalue_stack **stack_ref) +{ + rvalue_stack *stack; + *handle = TypedData_Make_Struct(0, rvalue_stack, &JSON_Parser_rvalue_stack_type, stack); + *stack_ref = stack; + MEMCPY(stack, old_stack, rvalue_stack, 1); + + stack->capa = old_stack->capa << 1; + stack->ptr = ALLOC_N(VALUE, stack->capa); + stack->type = RVALUE_STACK_HEAP_ALLOCATED; + MEMCPY(stack->ptr, old_stack->ptr, VALUE, old_stack->head); + return stack; +} + +static void rvalue_stack_eagerly_release(VALUE handle) +{ + if (handle) { + rvalue_stack *stack; + TypedData_Get_Struct(handle, rvalue_stack, &JSON_Parser_rvalue_stack_type, stack); + RTYPEDDATA_DATA(handle) = NULL; + rvalue_stack_free(stack); + } +} + + +#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; + if (ch <= 0x7F) { + buf[0] = (char) ch; + } else if (ch <= 0x07FF) { + buf[0] = (char) ((ch >> 6) | 0xC0); + buf[1] = (char) ((ch & 0x3F) | 0x80); + len++; + } else if (ch <= 0xFFFF) { + buf[0] = (char) ((ch >> 12) | 0xE0); + buf[1] = (char) (((ch >> 6) & 0x3F) | 0x80); + buf[2] = (char) ((ch & 0x3F) | 0x80); + len += 2; + } else if (ch <= 0x1fffff) { + buf[0] =(char) ((ch >> 18) | 0xF0); + buf[1] =(char) (((ch >> 12) & 0x3F) | 0x80); + buf[2] =(char) (((ch >> 6) & 0x3F) | 0x80); + buf[3] =(char) ((ch & 0x3F) | 0x80); + len += 3; + } else { + buf[0] = '?'; + } + return len; +} + +typedef struct JSON_ParserStruct { + VALUE on_load_proc; + VALUE decimal_class; + ID decimal_method_id; + int max_nesting; + bool allow_nan; + bool allow_trailing_comma; + bool parsing_name; + bool symbolize_names; + bool freeze; +} JSON_ParserConfig; + +typedef struct JSON_ParserStateStruct { + VALUE stack_handle; + const char *start; + const char *cursor; + const char *end; + rvalue_stack *stack; + rvalue_cache name_cache; + int in_array; + int current_nesting; +} JSON_ParserState; + + +#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) +{ + unsigned char buffer[PARSE_ERROR_FRAGMENT_LEN + 3]; + + const char *cursor = state->cursor; + long column = 0; + long line = 1; + + while (cursor >= state->start) { + if (*cursor-- == '\n') { + break; + } + column++; + } + + while (cursor >= state->start) { + if (*cursor-- == '\n') { + line++; + } + } + + const char *ptr = "EOF"; + if (state->cursor && state->cursor < state->end) { + ptr = state->cursor; + size_t len = 0; + while (len < PARSE_ERROR_FRAGMENT_LEN) { + char ch = ptr[len]; + if (!ch || ch == '\n' || ch == ' ' || ch == '\t' || ch == '\r') { + break; + } + len++; + } + + if (len) { + buffer[0] = '\''; + MEMCPY(buffer + 1, ptr, char, len); + + while (buffer[len] >= 0x80 && buffer[len] < 0xC0) { // Is continuation byte + len--; + } + + if (buffer[len] >= 0xC0) { // multibyte character start + len--; + } + + buffer[len + 1] = '\''; + buffer[len + 2] = '\0'; + ptr = (const char *)buffer; + } + } + + 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); + + 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); +} + +#ifdef RBIMPL_ATTR_NORETURN +RBIMPL_ATTR_NORETURN() +#endif +static void raise_parse_error_at(const char *format, JSON_ParserState *state, const char *at) +{ + state->cursor = at; + raise_parse_error(format, state); +} + +/* unicode */ + +static const signed char digit_values[256] = { + -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, + -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, + -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, -1, + -1, -1, -1, -1, -1, -1, 10, 11, 12, 13, 14, 15, -1, -1, -1, -1, -1, -1, -1, + -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, + 10, 11, 12, 13, 14, 15, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, + -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, + -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, + -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, + -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, + -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, + -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, + -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, + -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; +} + +#define GET_PARSER_CONFIG \ + JSON_ParserConfig *config; \ + TypedData_Get_Struct(self, JSON_ParserConfig, &JSON_ParserConfig_type, config) + +static const rb_data_type_t JSON_ParserConfig_type; + +static const bool whitespace[256] = { + [' '] = 1, + ['\t'] = 1, + ['\n'] = 1, + ['\r'] = 1, + ['/'] = 1, +}; + +static void +json_eat_comments(JSON_ParserState *state) +{ + if (state->cursor + 1 < state->end) { + switch(state->cursor[1]) { + case '/': { + state->cursor = memchr(state->cursor, '\n', state->end - state->cursor); + if (!state->cursor) { + state->cursor = state->end; + } else { + state->cursor++; + } + break; + } + case '*': { + state->cursor += 2; + while (true) { + state->cursor = memchr(state->cursor, '*', state->end - state->cursor); + if (!state->cursor) { + raise_parse_error_at("unexpected end of input, expected closing '*/'", state, state->end); + } else { + state->cursor++; + if (state->cursor < state->end && *state->cursor == '/') { + state->cursor++; + break; + } + } + } + break; + } + default: + raise_parse_error("unexpected token %s", state); + break; + } + } else { + raise_parse_error("unexpected token %s", state); + } +} + +static inline void +json_eat_whitespace(JSON_ParserState *state) +{ + while (state->cursor < state->end && RB_UNLIKELY(whitespace[(unsigned char)*state->cursor])) { + if (RB_LIKELY(*state->cursor != '/')) { + state->cursor++; + } else { + json_eat_comments(state); + } + } +} + +static inline VALUE build_string(const char *start, const char *end, bool intern, bool symbolize) +{ + if (symbolize) { + intern = true; + } + VALUE result; +# ifdef HAVE_RB_ENC_INTERNED_STR + if (intern) { + result = rb_enc_interned_str(start, (long)(end - start), enc_utf8); + } else { + result = rb_utf8_str_new(start, (long)(end - start)); + } +# else + result = rb_utf8_str_new(start, (long)(end - start)); + if (intern) { + result = rb_funcall(rb_str_freeze(result), i_uminus, 0); + } +# endif + + if (symbolize) { + result = rb_str_intern(result); + } + + return result; +} + +static inline VALUE json_string_fastpath(JSON_ParserState *state, const char *string, const char *stringEnd, bool is_name, bool intern, bool symbolize) +{ + size_t bufferSize = stringEnd - string; + + if (is_name && state->in_array) { + VALUE cached_key; + if (RB_UNLIKELY(symbolize)) { + cached_key = rsymbol_cache_fetch(&state->name_cache, string, bufferSize); + } else { + cached_key = rstring_cache_fetch(&state->name_cache, string, bufferSize); + } + + if (RB_LIKELY(cached_key)) { + return cached_key; + } + } + + 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) +{ + size_t bufferSize = stringEnd - string; + const char *p = string, *pe = string, *unescape, *bufferStart; + char *buffer; + int unescape_len; + char buf[4]; + + if (is_name && state->in_array) { + VALUE cached_key; + if (RB_UNLIKELY(symbolize)) { + cached_key = rsymbol_cache_fetch(&state->name_cache, string, bufferSize); + } else { + cached_key = rstring_cache_fetch(&state->name_cache, string, bufferSize); + } + + if (RB_LIKELY(cached_key)) { + return cached_key; + } + } + + 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; + if (pe > p) { + MEMCPY(buffer, p, char, pe - p); + buffer += pe - p; + } + switch (*++pe) { + case 'n': + unescape = (char *) "\n"; + break; + case 'r': + unescape = (char *) "\r"; + break; + case 't': + unescape = (char *) "\t"; + break; + case '"': + unescape = (char *) "\""; + break; + case '\\': + unescape = (char *) "\\"; + break; + case 'b': + unescape = (char *) "\b"; + break; + case 'f': + unescape = (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); + ch = (((ch & 0x3F) << 10) | ((((ch >> 6) & 0xF) + 1) << 16) + | (sur & 0x3FF)); + pe += 5; + } else { + unescape = (char *) "?"; + break; + } + } + unescape_len = convert_UTF32_to_UTF8(buf, ch); + unescape = buf; + } + break; + default: + p = pe; + continue; + } + MEMCPY(buffer, unescape, char, unescape_len); + buffer += unescape_len; + p = ++pe; + } + + if (stringEnd > p) { + MEMCPY(buffer, p, char, stringEnd - p); + buffer += stringEnd - p; + } + rb_str_set_len(result, buffer - bufferStart); + + if (symbolize) { + result = rb_str_intern(result); + } else if (intern) { + result = rb_funcall(rb_str_freeze(result), i_uminus, 0); + } + + return result; +} + +#define MAX_FAST_INTEGER_SIZE 18 +static inline VALUE fast_decode_integer(const char *p, const char *pe) +{ + bool negative = false; + if (*p == '-') { + negative = true; + p++; + } + + long long memo = 0; + while (p < pe) { + memo *= 10; + memo += *p - '0'; + p++; + } + + if (negative) { + memo = -memo; + } + return LL2NUM(memo); +} + +static VALUE json_decode_large_integer(const char *start, long len) +{ + 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; +} + +static inline VALUE +json_decode_integer(const char *start, const char *end) +{ + long len = end - start; + if (RB_LIKELY(len < MAX_FAST_INTEGER_SIZE)) { + return fast_decode_integer(start, end); + } + return json_decode_large_integer(start, len); +} + +static VALUE json_decode_large_float(const char *start, long len) +{ + 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; +} + +static VALUE json_decode_float(JSON_ParserConfig *config, const char *start, const char *end) +{ + long len = end - start; + + if (RB_UNLIKELY(config->decimal_class)) { + VALUE text = rb_str_new(start, len); + return rb_funcallv(config->decimal_class, config->decimal_method_id, 1, &text); + } else if (RB_LIKELY(len < 64)) { + char buffer[64]; + MEMCPY(buffer, start, char, len); + buffer[len] = '\0'; + return DBL2NUM(rb_cstr_to_dbl(buffer, 1)); + } else { + return json_decode_large_float(start, len); + } +} + +static inline VALUE json_decode_array(JSON_ParserState *state, JSON_ParserConfig *config, long count) +{ + VALUE array = rb_ary_new_from_values(count, rvalue_stack_peek(state->stack, count)); + rvalue_stack_pop(state->stack, count); + + if (config->freeze) { + RB_OBJ_FREEZE(array); + } + + return array; +} + +static inline VALUE json_decode_object(JSON_ParserState *state, JSON_ParserConfig *config, long count) +{ + VALUE object = rb_hash_new_capa(count); + rb_hash_bulk_insert(count, rvalue_stack_peek(state->stack, count), object); + + rvalue_stack_pop(state->stack, count); + + if (config->freeze) { + RB_OBJ_FREEZE(object); + } + + 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)) { + value = rb_proc_call_with_block(config->on_load_proc, 1, &value, Qnil); + } + rvalue_stack_push(state->stack, value, &state->stack_handle, &state->stack); + return value; +} + +static const bool string_scan[256] = { + // ASCII Control Characters + 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, + 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, + // ASCII Characters + 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, // '"' + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, // '\\' + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, +}; + +static inline VALUE json_parse_string(JSON_ParserState *state, JSON_ParserConfig *config, bool is_name) +{ + state->cursor++; + const char *start = state->cursor; + bool escaped = false; + + while (state->cursor < state->end) { + if (RB_UNLIKELY(string_scan[(unsigned char)*state->cursor])) { + switch (*state->cursor) { + case '"': { + VALUE string = json_decode_string(state, config, start, state->cursor, escaped, is_name); + 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); + } + break; + } + default: + raise_parse_error("invalid ASCII control character in string: %s", state); + break; + } + } + + state->cursor++; + } + + raise_parse_error("unexpected end of input, expected closing \"", state); + return Qfalse; +} + +static VALUE json_parse_any(JSON_ParserState *state, JSON_ParserConfig *config) +{ + json_eat_whitespace(state); + if (state->cursor >= state->end) { + raise_parse_error("unexpected end of input", state); + } + + switch (*state->cursor) { + case 'n': + if ((state->end - state->cursor >= 4) && (memcmp(state->cursor, "null", 4) == 0)) { + state->cursor += 4; + return json_push_value(state, config, Qnil); + } + + raise_parse_error("unexpected token %s", state); + break; + case 't': + if ((state->end - state->cursor >= 4) && (memcmp(state->cursor, "true", 4) == 0)) { + state->cursor += 4; + return json_push_value(state, config, Qtrue); + } + + raise_parse_error("unexpected token %s", state); + break; + case 'f': + // Note: memcmp with a small power of two compile to an integer comparison + if ((state->end - state->cursor >= 5) && (memcmp(state->cursor + 1, "alse", 4) == 0)) { + state->cursor += 5; + return json_push_value(state, config, Qfalse); + } + + raise_parse_error("unexpected token %s", state); + break; + case 'N': + // Note: memcmp with a small power of two compile to an integer comparison + if (config->allow_nan && (state->end - state->cursor >= 3) && (memcmp(state->cursor + 1, "aN", 2) == 0)) { + state->cursor += 3; + return json_push_value(state, config, CNaN); + } + + raise_parse_error("unexpected token %s", state); + break; + case 'I': + if (config->allow_nan && (state->end - state->cursor >= 8) && (memcmp(state->cursor, "Infinity", 8) == 0)) { + state->cursor += 8; + return json_push_value(state, config, CInfinity); + } + + raise_parse_error("unexpected token %s", state); + break; + case '-': + // Note: memcmp with a small power of two compile to an integer comparison + if ((state->end - state->cursor >= 9) && (memcmp(state->cursor + 1, "Infinity", 8) == 0)) { + if (config->allow_nan) { + state->cursor += 9; + return json_push_value(state, config, CMinusInfinity); + } else { + raise_parse_error("unexpected token %s", state); + } + } + // Fallthrough + case '0': case '1': case '2': case '3': case '4': case '5': case '6': case '7': case '8': case '9': { + bool integer = true; + + // /\A-?(0|[1-9]\d*)(\.\d+)?([Ee][-+]?\d+)?/ + const char *start = state->cursor; + state->cursor++; + + while ((state->cursor < state->end) && (*state->cursor >= '0') && (*state->cursor <= '9')) { + state->cursor++; + } + + long integer_length = state->cursor - start; + + if (RB_UNLIKELY(start[0] == '0' && integer_length > 1)) { + raise_parse_error_at("invalid number: %s", state, start); + } else if (RB_UNLIKELY(integer_length > 2 && start[0] == '-' && start[1] == '0')) { + raise_parse_error_at("invalid number: %s", state, start); + } else if (RB_UNLIKELY(integer_length == 1 && start[0] == '-')) { + raise_parse_error_at("invalid number: %s", state, start); + } + + if ((state->cursor < state->end) && (*state->cursor == '.')) { + integer = false; + state->cursor++; + + if (state->cursor == state->end || *state->cursor < '0' || *state->cursor > '9') { + raise_parse_error("invalid number: %s", state); + } + + while ((state->cursor < state->end) && (*state->cursor >= '0') && (*state->cursor <= '9')) { + state->cursor++; + } + } + + if ((state->cursor < state->end) && ((*state->cursor == 'e') || (*state->cursor == 'E'))) { + integer = false; + state->cursor++; + if ((state->cursor < state->end) && ((*state->cursor == '+') || (*state->cursor == '-'))) { + state->cursor++; + } + + if (state->cursor == state->end || *state->cursor < '0' || *state->cursor > '9') { + raise_parse_error("invalid number: %s", state); + } + + while ((state->cursor < state->end) && (*state->cursor >= '0') && (*state->cursor <= '9')) { + state->cursor++; + } + } + + if (integer) { + return json_push_value(state, config, json_decode_integer(start, state->cursor)); + } + return json_push_value(state, config, json_decode_float(config, start, state->cursor)); + } + case '"': { + // %r{\A"[^"\\\t\n\x00]*(?:\\[bfnrtu\\/"][^"\\]*)*"} + return json_parse_string(state, config, false); + break; + } + case '[': { + state->cursor++; + json_eat_whitespace(state); + long stack_head = state->stack->head; + + if ((state->cursor < state->end) && (*state->cursor == ']')) { + state->cursor++; + return json_push_value(state, config, json_decode_array(state, config, 0)); + } else { + state->current_nesting++; + if (RB_UNLIKELY(config->max_nesting && (config->max_nesting < state->current_nesting))) { + rb_raise(eNestingError, "nesting of %d is too deep", state->current_nesting); + } + state->in_array++; + json_parse_any(state, config); + } + + while (true) { + json_eat_whitespace(state); + + if (state->cursor < state->end) { + if (*state->cursor == ']') { + state->cursor++; + long count = state->stack->head - stack_head; + state->current_nesting--; + state->in_array--; + return json_push_value(state, config, json_decode_array(state, config, count)); + } + + if (*state->cursor == ',') { + state->cursor++; + if (config->allow_trailing_comma) { + json_eat_whitespace(state); + if ((state->cursor < state->end) && (*state->cursor == ']')) { + continue; + } + } + json_parse_any(state, config); + continue; + } + } + + raise_parse_error("expected ',' or ']' after array value", state); + } + break; + } + case '{': { + state->cursor++; + json_eat_whitespace(state); + long stack_head = state->stack->head; + + if ((state->cursor < state->end) && (*state->cursor == '}')) { + state->cursor++; + return json_push_value(state, config, json_decode_object(state, config, 0)); + } else { + state->current_nesting++; + if (RB_UNLIKELY(config->max_nesting && (config->max_nesting < state->current_nesting))) { + rb_raise(eNestingError, "nesting of %d is too deep", state->current_nesting); + } + + if (*state->cursor != '"') { + raise_parse_error("expected object key, got %s", state); + } + json_parse_string(state, config, true); + + json_eat_whitespace(state); + if ((state->cursor >= state->end) || (*state->cursor != ':')) { + raise_parse_error("expected ':' after object key", state); + } + state->cursor++; + + json_parse_any(state, config); + } + + while (true) { + json_eat_whitespace(state); + + if (state->cursor < state->end) { + if (*state->cursor == '}') { + state->cursor++; + state->current_nesting--; + long count = state->stack->head - stack_head; + return json_push_value(state, config, json_decode_object(state, config, count)); + } + + if (*state->cursor == ',') { + state->cursor++; + json_eat_whitespace(state); + + if (config->allow_trailing_comma) { + if ((state->cursor < state->end) && (*state->cursor == '}')) { + continue; + } + } + + if (*state->cursor != '"') { + raise_parse_error("expected object key, got: %s", state); + } + json_parse_string(state, config, true); + + json_eat_whitespace(state); + if ((state->cursor >= state->end) || (*state->cursor != ':')) { + raise_parse_error("expected ':' after object key, got: %s", state); + } + state->cursor++; + + json_parse_any(state, config); + + continue; + } + } + + raise_parse_error("expected ',' or '}' after object value, got: %s", state); + } + break; + } + + default: + raise_parse_error("unexpected character: %s", state); + break; + } + + raise_parse_error("unreacheable: %s", state); +} + +static void json_ensure_eof(JSON_ParserState *state) +{ + json_eat_whitespace(state); + if (state->cursor != state->end) { + raise_parse_error("unexpected token at end of stream %s", state); + } +} + +/* + * Document-class: JSON::Ext::Parser + * + * This is the JSON parser implemented as a C extension. It can be configured + * to be used by setting + * + * JSON.parser = JSON::Ext::Parser + * + * with the method parser= in JSON. + * + */ + +static VALUE convert_encoding(VALUE source) +{ + int encindex = RB_ENCODING_GET(source); + + if (RB_LIKELY(encindex == utf8_encindex)) { + return source; + } + + if (encindex == binary_encindex) { + // For historical reason, we silently reinterpret binary strings as UTF-8 + return rb_enc_associate_index(rb_str_dup(source), utf8_encindex); + } + + return rb_funcall(source, i_encode, 1, Encoding_UTF_8); +} + +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_decimal_class) { + if (RTEST(val)) { + if (rb_respond_to(val, i_try_convert)) { + config->decimal_class = val; + config->decimal_method_id = i_try_convert; + } else if (rb_respond_to(val, i_new)) { + config->decimal_class = val; + config->decimal_method_id = i_new; + } else if (RB_TYPE_P(val, T_CLASS)) { + VALUE name = rb_class_name(val); + const char *name_cstr = RSTRING_PTR(name); + const char *last_colon = strrchr(name_cstr, ':'); + if (last_colon) { + const char *mod_path_end = last_colon - 1; + VALUE mod_path = rb_str_substr(name, 0, mod_path_end - name_cstr); + config->decimal_class = rb_path_to_class(mod_path); + + const char *method_name_beg = last_colon + 1; + long before_len = method_name_beg - name_cstr; + long len = RSTRING_LEN(name) - before_len; + VALUE method_name = rb_str_substr(name, before_len, len); + config->decimal_method_id = SYM2ID(rb_str_intern(method_name)); + } else { + config->decimal_class = rb_mKernel; + config->decimal_method_id = SYM2ID(rb_str_intern(name)); + } + } + } + } + + return ST_CONTINUE; +} + +static void parser_config_init(JSON_ParserConfig *config, VALUE opts) +{ + config->max_nesting = 100; + + if (!NIL_P(opts)) { + Check_Type(opts, T_HASH); + if (RHASH_SIZE(opts) > 0) { + // We assume in most cases few keys are set so it's faster to go over + // the provided keys than to check all possible keys. + rb_hash_foreach(opts, parser_config_init_i, (VALUE)config); + } + + } +} + +/* + * call-seq: new(opts => {}) + * + * Creates a new JSON::Ext::ParserConfig instance. + * + * It will be configured by the _opts_ hash. _opts_ can have the following + * keys: + * + * _opts_ can have the following keys: + * * *max_nesting*: The maximum depth of nesting allowed in the parsed data + * structures. Disable depth checking with :max_nesting => false|nil|0, it + * defaults to 100. + * * *allow_nan*: If set to true, allow NaN, Infinity and -Infinity in + * defiance of RFC 4627 to be parsed by the Parser. This option defaults to + * false. + * * *symbolize_names*: If set to true, returns symbols for the names + * (keys) in a JSON object. Otherwise strings are returned, which is + * also the default. It's not possible to use this option in + * conjunction with the *create_additions* option. + * * *decimal_class*: Specifies which class to use instead of the default + * (Float) when parsing decimal numbers. This class must accept a single + * string argument in its constructor. + */ +static VALUE cParserConfig_initialize(VALUE self, VALUE opts) +{ + GET_PARSER_CONFIG; + + parser_config_init(config, opts); + + RB_OBJ_WRITTEN(self, Qundef, config->decimal_class); + + return self; +} + +static VALUE cParser_parse(JSON_ParserConfig *config, VALUE Vsource) +{ + Vsource = convert_encoding(StringValue(Vsource)); + StringValue(Vsource); + + VALUE rvalue_stack_buffer[RVALUE_STACK_INITIAL_CAPA]; + rvalue_stack stack = { + .type = RVALUE_STACK_STACK_ALLOCATED, + .ptr = rvalue_stack_buffer, + .capa = RVALUE_STACK_INITIAL_CAPA, + }; + + long len; + const char *start; + RSTRING_GETMEM(Vsource, start, len); + + JSON_ParserState _state = { + .start = start, + .cursor = start, + .end = start + len, + .stack = &stack, + }; + JSON_ParserState *state = &_state; + + VALUE result = json_parse_any(state, config); + + // This may be skipped in case of exception, but + // it won't cause a leak. + rvalue_stack_eagerly_release(state->stack_handle); + + json_ensure_eof(state); + + return result; +} + +/* + * call-seq: parse(source) + * + * Parses the current JSON text _source_ and returns the complete data + * structure as a result. + * It raises JSON::ParserError if fail to parse. + */ +static VALUE cParserConfig_parse(VALUE self, VALUE Vsource) +{ + GET_PARSER_CONFIG; + return cParser_parse(config, Vsource); +} + +static VALUE cParser_m_parse(VALUE klass, VALUE Vsource, VALUE opts) +{ + Vsource = convert_encoding(StringValue(Vsource)); + StringValue(Vsource); + + JSON_ParserConfig _config = {0}; + JSON_ParserConfig *config = &_config; + parser_config_init(config, opts); + + return cParser_parse(config, Vsource); +} + +static void JSON_ParserConfig_mark(void *ptr) +{ + JSON_ParserConfig *config = ptr; + rb_gc_mark(config->on_load_proc); + rb_gc_mark(config->decimal_class); +} + +static void JSON_ParserConfig_free(void *ptr) +{ + JSON_ParserConfig *config = ptr; + ruby_xfree(config); +} + +static size_t JSON_ParserConfig_memsize(const void *ptr) +{ + return sizeof(JSON_ParserConfig); +} + +static const rb_data_type_t JSON_ParserConfig_type = { + "JSON::Ext::Parser/ParserConfig", + { + JSON_ParserConfig_mark, + JSON_ParserConfig_free, + JSON_ParserConfig_memsize, + }, + 0, 0, + RUBY_TYPED_FREE_IMMEDIATELY | RUBY_TYPED_WB_PROTECTED, +}; + +static VALUE cJSON_parser_s_allocate(VALUE klass) +{ + JSON_ParserConfig *config; + return TypedData_Make_Struct(klass, JSON_ParserConfig, &JSON_ParserConfig_type, config); +} + +void Init_parser(void) +{ +#ifdef HAVE_RB_EXT_RACTOR_SAFE + rb_ext_ractor_safe(true); +#endif + +#undef rb_intern + rb_require("json/common"); + mJSON = rb_define_module("JSON"); + VALUE mExt = rb_define_module_under(mJSON, "Ext"); + VALUE cParserConfig = rb_define_class_under(mExt, "ParserConfig", rb_cObject); + eNestingError = rb_path2class("JSON::NestingError"); + rb_gc_register_mark_object(eNestingError); + rb_define_alloc_func(cParserConfig, cJSON_parser_s_allocate); + rb_define_method(cParserConfig, "initialize", cParserConfig_initialize, 1); + rb_define_method(cParserConfig, "parse", cParserConfig_parse, 1); + + VALUE cParser = rb_define_class_under(mExt, "Parser", rb_cObject); + rb_define_singleton_method(cParser, "parse", cParser_m_parse, 2); + + CNaN = rb_const_get(mJSON, rb_intern("NaN")); + rb_gc_register_mark_object(CNaN); + + CInfinity = rb_const_get(mJSON, rb_intern("Infinity")); + rb_gc_register_mark_object(CInfinity); + + CMinusInfinity = rb_const_get(mJSON, rb_intern("MinusInfinity")); + rb_gc_register_mark_object(CMinusInfinity); + + rb_global_variable(&Encoding_UTF_8); + Encoding_UTF_8 = rb_const_get(rb_path2class("Encoding"), rb_intern("UTF_8")); + + 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_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")); + + 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("-@"); + i_encode = rb_intern("encode"); + + binary_encindex = rb_ascii8bit_encindex(); + utf8_encindex = rb_utf8_encindex(); + enc_utf8 = rb_utf8_encoding(); +} diff --git a/vendor/bundle/ruby/3.4.0/gems/json-2.12.2/ext/json/ext/vendor/fpconv.c b/vendor/bundle/ruby/3.4.0/gems/json-2.12.2/ext/json/ext/vendor/fpconv.c new file mode 100644 index 0000000..75efd46 --- /dev/null +++ b/vendor/bundle/ruby/3.4.0/gems/json-2.12.2/ext/json/ext/vendor/fpconv.c @@ -0,0 +1,479 @@ +// Boost Software License - Version 1.0 - August 17th, 2003 +// +// Permission is hereby granted, free of charge, to any person or organization +// obtaining a copy of the software and accompanying documentation covered by +// this license (the "Software") to use, reproduce, display, distribute, +// execute, and transmit the Software, and to prepare derivative works of the +// Software, and to permit third-parties to whom the Software is furnished to +// do so, all subject to the following: +// +// The copyright notices in the Software and this entire statement, including +// the above license grant, this restriction and the following disclaimer, +// must be included in all copies of the Software, in whole or in part, and +// all derivative works of the Software, unless such copies or derivative +// works are solely in the form of machine-executable object code generated by +// a source language processor. +// +// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR +// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, +// FITNESS FOR A PARTICULAR PURPOSE, TITLE AND NON-INFRINGEMENT. IN NO EVENT +// SHALL THE COPYRIGHT HOLDERS OR ANYONE DISTRIBUTING THE SOFTWARE BE LIABLE +// FOR ANY DAMAGES OR OTHER LIABILITY, WHETHER IN CONTRACT, TORT OR OTHERWISE, +// ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER +// DEALINGS IN THE SOFTWARE. + +// The contents of this file is extracted from https://github.com/night-shift/fpconv +// It was slightly modified to append ".0" to plain floats, for use with the https://github.com/ruby/json package. + +#include <stdbool.h> +#include <string.h> +#include <stdint.h> + +#define npowers 87 +#define steppowers 8 +#define firstpower -348 /* 10 ^ -348 */ + +#define expmax -32 +#define expmin -60 + +typedef struct Fp { + uint64_t frac; + int exp; +} Fp; + +static const Fp powers_ten[] = { + { 18054884314459144840U, -1220 }, { 13451937075301367670U, -1193 }, + { 10022474136428063862U, -1166 }, { 14934650266808366570U, -1140 }, + { 11127181549972568877U, -1113 }, { 16580792590934885855U, -1087 }, + { 12353653155963782858U, -1060 }, { 18408377700990114895U, -1034 }, + { 13715310171984221708U, -1007 }, { 10218702384817765436U, -980 }, + { 15227053142812498563U, -954 }, { 11345038669416679861U, -927 }, + { 16905424996341287883U, -901 }, { 12595523146049147757U, -874 }, + { 9384396036005875287U, -847 }, { 13983839803942852151U, -821 }, + { 10418772551374772303U, -794 }, { 15525180923007089351U, -768 }, + { 11567161174868858868U, -741 }, { 17236413322193710309U, -715 }, + { 12842128665889583758U, -688 }, { 9568131466127621947U, -661 }, + { 14257626930069360058U, -635 }, { 10622759856335341974U, -608 }, + { 15829145694278690180U, -582 }, { 11793632577567316726U, -555 }, + { 17573882009934360870U, -529 }, { 13093562431584567480U, -502 }, + { 9755464219737475723U, -475 }, { 14536774485912137811U, -449 }, + { 10830740992659433045U, -422 }, { 16139061738043178685U, -396 }, + { 12024538023802026127U, -369 }, { 17917957937422433684U, -343 }, + { 13349918974505688015U, -316 }, { 9946464728195732843U, -289 }, + { 14821387422376473014U, -263 }, { 11042794154864902060U, -236 }, + { 16455045573212060422U, -210 }, { 12259964326927110867U, -183 }, + { 18268770466636286478U, -157 }, { 13611294676837538539U, -130 }, + { 10141204801825835212U, -103 }, { 15111572745182864684U, -77 }, + { 11258999068426240000U, -50 }, { 16777216000000000000U, -24 }, + { 12500000000000000000U, 3 }, { 9313225746154785156U, 30 }, + { 13877787807814456755U, 56 }, { 10339757656912845936U, 83 }, + { 15407439555097886824U, 109 }, { 11479437019748901445U, 136 }, + { 17105694144590052135U, 162 }, { 12744735289059618216U, 189 }, + { 9495567745759798747U, 216 }, { 14149498560666738074U, 242 }, + { 10542197943230523224U, 269 }, { 15709099088952724970U, 295 }, + { 11704190886730495818U, 322 }, { 17440603504673385349U, 348 }, + { 12994262207056124023U, 375 }, { 9681479787123295682U, 402 }, + { 14426529090290212157U, 428 }, { 10748601772107342003U, 455 }, + { 16016664761464807395U, 481 }, { 11933345169920330789U, 508 }, + { 17782069995880619868U, 534 }, { 13248674568444952270U, 561 }, + { 9871031767461413346U, 588 }, { 14708983551653345445U, 614 }, + { 10959046745042015199U, 641 }, { 16330252207878254650U, 667 }, + { 12166986024289022870U, 694 }, { 18130221999122236476U, 720 }, + { 13508068024458167312U, 747 }, { 10064294952495520794U, 774 }, + { 14996968138956309548U, 800 }, { 11173611982879273257U, 827 }, + { 16649979327439178909U, 853 }, { 12405201291620119593U, 880 }, + { 9242595204427927429U, 907 }, { 13772540099066387757U, 933 }, + { 10261342003245940623U, 960 }, { 15290591125556738113U, 986 }, + { 11392378155556871081U, 1013 }, { 16975966327722178521U, 1039 }, + { 12648080533535911531U, 1066 } +}; + +static Fp find_cachedpow10(int exp, int* k) +{ + const double one_log_ten = 0.30102999566398114; + + int approx = (int)(-(exp + npowers) * one_log_ten); + int idx = (approx - firstpower) / steppowers; + + while(1) { + int current = exp + powers_ten[idx].exp + 64; + + if(current < expmin) { + idx++; + continue; + } + + if(current > expmax) { + idx--; + continue; + } + + *k = (firstpower + idx * steppowers); + + return powers_ten[idx]; + } +} + +#define fracmask 0x000FFFFFFFFFFFFFU +#define expmask 0x7FF0000000000000U +#define hiddenbit 0x0010000000000000U +#define signmask 0x8000000000000000U +#define expbias (1023 + 52) + +#define absv(n) ((n) < 0 ? -(n) : (n)) +#define minv(a, b) ((a) < (b) ? (a) : (b)) + +static const uint64_t tens[] = { + 10000000000000000000U, 1000000000000000000U, 100000000000000000U, + 10000000000000000U, 1000000000000000U, 100000000000000U, + 10000000000000U, 1000000000000U, 100000000000U, + 10000000000U, 1000000000U, 100000000U, + 10000000U, 1000000U, 100000U, + 10000U, 1000U, 100U, + 10U, 1U +}; + +static inline uint64_t get_dbits(double d) +{ + union { + double dbl; + uint64_t i; + } dbl_bits = { d }; + + return dbl_bits.i; +} + +static Fp build_fp(double d) +{ + uint64_t bits = get_dbits(d); + + Fp fp; + fp.frac = bits & fracmask; + fp.exp = (bits & expmask) >> 52; + + if(fp.exp) { + fp.frac += hiddenbit; + fp.exp -= expbias; + + } else { + fp.exp = -expbias + 1; + } + + return fp; +} + +static void normalize(Fp* fp) +{ + while ((fp->frac & hiddenbit) == 0) { + fp->frac <<= 1; + fp->exp--; + } + + int shift = 64 - 52 - 1; + fp->frac <<= shift; + fp->exp -= shift; +} + +static void get_normalized_boundaries(Fp* fp, Fp* lower, Fp* upper) +{ + upper->frac = (fp->frac << 1) + 1; + upper->exp = fp->exp - 1; + + while ((upper->frac & (hiddenbit << 1)) == 0) { + upper->frac <<= 1; + upper->exp--; + } + + int u_shift = 64 - 52 - 2; + + upper->frac <<= u_shift; + upper->exp = upper->exp - u_shift; + + + int l_shift = fp->frac == hiddenbit ? 2 : 1; + + lower->frac = (fp->frac << l_shift) - 1; + lower->exp = fp->exp - l_shift; + + + lower->frac <<= lower->exp - upper->exp; + lower->exp = upper->exp; +} + +static Fp multiply(Fp* a, Fp* b) +{ + const uint64_t lomask = 0x00000000FFFFFFFF; + + uint64_t ah_bl = (a->frac >> 32) * (b->frac & lomask); + uint64_t al_bh = (a->frac & lomask) * (b->frac >> 32); + uint64_t al_bl = (a->frac & lomask) * (b->frac & lomask); + uint64_t ah_bh = (a->frac >> 32) * (b->frac >> 32); + + uint64_t tmp = (ah_bl & lomask) + (al_bh & lomask) + (al_bl >> 32); + /* round up */ + tmp += 1U << 31; + + Fp fp = { + ah_bh + (ah_bl >> 32) + (al_bh >> 32) + (tmp >> 32), + a->exp + b->exp + 64 + }; + + return fp; +} + +static void round_digit(char* digits, int ndigits, uint64_t delta, uint64_t rem, uint64_t kappa, uint64_t frac) +{ + while (rem < frac && delta - rem >= kappa && + (rem + kappa < frac || frac - rem > rem + kappa - frac)) { + + digits[ndigits - 1]--; + rem += kappa; + } +} + +static int generate_digits(Fp* fp, Fp* upper, Fp* lower, char* digits, int* K) +{ + uint64_t wfrac = upper->frac - fp->frac; + uint64_t delta = upper->frac - lower->frac; + + Fp one; + one.frac = 1ULL << -upper->exp; + one.exp = upper->exp; + + uint64_t part1 = upper->frac >> -one.exp; + uint64_t part2 = upper->frac & (one.frac - 1); + + int idx = 0, kappa = 10; + const uint64_t* divp; + /* 1000000000 */ + for(divp = tens + 10; kappa > 0; divp++) { + + uint64_t div = *divp; + unsigned digit = (unsigned) (part1 / div); + + if (digit || idx) { + digits[idx++] = digit + '0'; + } + + part1 -= digit * div; + kappa--; + + uint64_t tmp = (part1 <<-one.exp) + part2; + if (tmp <= delta) { + *K += kappa; + round_digit(digits, idx, delta, tmp, div << -one.exp, wfrac); + + return idx; + } + } + + /* 10 */ + const uint64_t* unit = tens + 18; + + while(true) { + part2 *= 10; + delta *= 10; + kappa--; + + unsigned digit = (unsigned) (part2 >> -one.exp); + if (digit || idx) { + digits[idx++] = digit + '0'; + } + + part2 &= one.frac - 1; + if (part2 < delta) { + *K += kappa; + round_digit(digits, idx, delta, part2, one.frac, wfrac * *unit); + + return idx; + } + + unit--; + } +} + +static int grisu2(double d, char* digits, int* K) +{ + Fp w = build_fp(d); + + Fp lower, upper; + get_normalized_boundaries(&w, &lower, &upper); + + normalize(&w); + + int k; + Fp cp = find_cachedpow10(upper.exp, &k); + + w = multiply(&w, &cp); + upper = multiply(&upper, &cp); + lower = multiply(&lower, &cp); + + lower.frac++; + upper.frac--; + + *K = -k; + + return generate_digits(&w, &upper, &lower, digits, K); +} + +static int emit_digits(char* digits, int ndigits, char* dest, int K, bool neg) +{ + int exp = absv(K + ndigits - 1); + + int max_trailing_zeros = 7; + + if(neg) { + max_trailing_zeros -= 1; + } + + /* write plain integer */ + if(K >= 0 && (exp < (ndigits + max_trailing_zeros))) { + + memcpy(dest, digits, ndigits); + memset(dest + ndigits, '0', K); + + /* add a .0 to mark this as a float. */ + dest[ndigits + K] = '.'; + dest[ndigits + K + 1] = '0'; + + return ndigits + K + 2; + } + + /* write decimal w/o scientific notation */ + if(K < 0 && (K > -7 || exp < 10)) { + int offset = ndigits - absv(K); + /* fp < 1.0 -> write leading zero */ + if(offset <= 0) { + offset = -offset; + dest[0] = '0'; + dest[1] = '.'; + memset(dest + 2, '0', offset); + memcpy(dest + offset + 2, digits, ndigits); + + return ndigits + 2 + offset; + + /* fp > 1.0 */ + } else { + memcpy(dest, digits, offset); + dest[offset] = '.'; + memcpy(dest + offset + 1, digits + offset, ndigits - offset); + + return ndigits + 1; + } + } + + /* write decimal w/ scientific notation */ + ndigits = minv(ndigits, 18 - neg); + + int idx = 0; + dest[idx++] = digits[0]; + + if(ndigits > 1) { + dest[idx++] = '.'; + memcpy(dest + idx, digits + 1, ndigits - 1); + idx += ndigits - 1; + } + + dest[idx++] = 'e'; + + char sign = K + ndigits - 1 < 0 ? '-' : '+'; + dest[idx++] = sign; + + int cent = 0; + + if(exp > 99) { + cent = exp / 100; + dest[idx++] = cent + '0'; + exp -= cent * 100; + } + if(exp > 9) { + int dec = exp / 10; + dest[idx++] = dec + '0'; + exp -= dec * 10; + + } else if(cent) { + dest[idx++] = '0'; + } + + dest[idx++] = exp % 10 + '0'; + + return idx; +} + +static int filter_special(double fp, char* dest) +{ + if(fp == 0.0) { + dest[0] = '0'; + dest[1] = '.'; + dest[2] = '0'; + return 3; + } + + uint64_t bits = get_dbits(fp); + + bool nan = (bits & expmask) == expmask; + + if(!nan) { + return 0; + } + + if(bits & fracmask) { + dest[0] = 'n'; dest[1] = 'a'; dest[2] = 'n'; + + } else { + dest[0] = 'i'; dest[1] = 'n'; dest[2] = 'f'; + } + + return 3; +} + +/* Fast and accurate double to string conversion based on Florian Loitsch's + * Grisu-algorithm[1]. + * + * Input: + * fp -> the double to convert, dest -> destination buffer. + * The generated string will never be longer than 28 characters. + * Make sure to pass a pointer to at least 28 bytes of memory. + * The emitted string will not be null terminated. + * + * Output: + * The number of written characters. + * + * Exemplary usage: + * + * void print(double d) + * { + * char buf[28 + 1] // plus null terminator + * int str_len = fpconv_dtoa(d, buf); + * + * buf[str_len] = '\0'; + * printf("%s", buf); + * } + * + */ +static int fpconv_dtoa(double d, char dest[28]) +{ + char digits[18]; + + int str_len = 0; + bool neg = false; + + if(get_dbits(d) & signmask) { + dest[0] = '-'; + str_len++; + neg = true; + } + + int spec = filter_special(d, dest + str_len); + + if(spec) { + return str_len + spec; + } + + int K = 0; + int ndigits = grisu2(d, digits, &K); + + str_len += emit_digits(digits, ndigits, dest + str_len, K, neg); + + return str_len; +} diff --git a/vendor/bundle/ruby/3.4.0/gems/json-2.12.2/ext/json/ext/vendor/jeaiii-ltoa.h b/vendor/bundle/ruby/3.4.0/gems/json-2.12.2/ext/json/ext/vendor/jeaiii-ltoa.h new file mode 100644 index 0000000..ba4f497 --- /dev/null +++ b/vendor/bundle/ruby/3.4.0/gems/json-2.12.2/ext/json/ext/vendor/jeaiii-ltoa.h @@ -0,0 +1,267 @@ +/* + +This file is released under the terms of the MIT License. It is based on the +work of James Edward Anhalt III, with the original license listed below. + +MIT License + +Copyright (c) 2024,2025 Enrico Thierbach - https://github.com/radiospiel +Copyright (c) 2022 James Edward Anhalt III - https://github.com/jeaiii/itoa + +Permission is hereby granted, free of charge, to any person obtaining a copy +of this software and associated documentation files (the "Software"), to deal +in the Software without restriction, including without limitation the rights +to use, copy, modify, merge, publish, distribute, sublicense, and/or sell +copies of the Software, and to permit persons to whom the Software is +furnished to do so, subject to the following conditions: + +The above copyright notice and this permission notice shall be included in all +copies or substantial portions of the Software. + +THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR +IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, +FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE +AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER +LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, +OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE +SOFTWARE. +*/ + +#ifndef JEAIII_TO_TEXT_H_ +#define JEAIII_TO_TEXT_H_ + +#include <stdint.h> + +typedef uint_fast32_t u32_t; +typedef uint_fast64_t u64_t; + +#define u32(x) ((u32_t)(x)) +#define u64(x) ((u64_t)(x)) + +struct digit_pair +{ + char dd[2]; +}; + +static const struct digit_pair *digits_dd = (struct digit_pair *)( + "00" "01" "02" "03" "04" "05" "06" "07" "08" "09" + "10" "11" "12" "13" "14" "15" "16" "17" "18" "19" + "20" "21" "22" "23" "24" "25" "26" "27" "28" "29" + "30" "31" "32" "33" "34" "35" "36" "37" "38" "39" + "40" "41" "42" "43" "44" "45" "46" "47" "48" "49" + "50" "51" "52" "53" "54" "55" "56" "57" "58" "59" + "60" "61" "62" "63" "64" "65" "66" "67" "68" "69" + "70" "71" "72" "73" "74" "75" "76" "77" "78" "79" + "80" "81" "82" "83" "84" "85" "86" "87" "88" "89" + "90" "91" "92" "93" "94" "95" "96" "97" "98" "99" +); + +static const struct digit_pair *digits_fd = (struct digit_pair *)( + "0_" "1_" "2_" "3_" "4_" "5_" "6_" "7_" "8_" "9_" + "10" "11" "12" "13" "14" "15" "16" "17" "18" "19" + "20" "21" "22" "23" "24" "25" "26" "27" "28" "29" + "30" "31" "32" "33" "34" "35" "36" "37" "38" "39" + "40" "41" "42" "43" "44" "45" "46" "47" "48" "49" + "50" "51" "52" "53" "54" "55" "56" "57" "58" "59" + "60" "61" "62" "63" "64" "65" "66" "67" "68" "69" + "70" "71" "72" "73" "74" "75" "76" "77" "78" "79" + "80" "81" "82" "83" "84" "85" "86" "87" "88" "89" + "90" "91" "92" "93" "94" "95" "96" "97" "98" "99" +); + +static const u64_t mask24 = (u64(1) << 24) - 1; +static const u64_t mask32 = (u64(1) << 32) - 1; +static const u64_t mask57 = (u64(1) << 57) - 1; + +#define COPY(buffer, digits) memcpy(buffer, &(digits), sizeof(struct digit_pair)) + +static char * +jeaiii_ultoa(char *b, u64_t n) +{ + if (n < u32(1e2)) { + COPY(b, digits_fd[n]); + return n < 10 ? b + 1 : b + 2; + } + + if (n < u32(1e6)) { + if (n < u32(1e4)) { + u32_t f0 = u32((10 * (1 << 24) / 1e3 + 1) * n); + COPY(b, digits_fd[f0 >> 24]); + + b -= n < u32(1e3); + u32_t f2 = (f0 & mask24) * 100; + COPY(b + 2, digits_dd[f2 >> 24]); + + return b + 4; + } + + u64_t f0 = u64(10 * (1ull << 32ull)/ 1e5 + 1) * n; + COPY(b, digits_fd[f0 >> 32]); + + b -= n < u32(1e5); + u64_t f2 = (f0 & mask32) * 100; + COPY(b + 2, digits_dd[f2 >> 32]); + + u64_t f4 = (f2 & mask32) * 100; + COPY(b + 4, digits_dd[f4 >> 32]); + return b + 6; + } + + if (n < u64(1ull << 32ull)) { + if (n < u32(1e8)) { + u64_t f0 = u64(10 * (1ull << 48ull) / 1e7 + 1) * n >> 16; + COPY(b, digits_fd[f0 >> 32]); + + b -= n < u32(1e7); + u64_t f2 = (f0 & mask32) * 100; + COPY(b + 2, digits_dd[f2 >> 32]); + + u64_t f4 = (f2 & mask32) * 100; + COPY(b + 4, digits_dd[f4 >> 32]); + + u64_t f6 = (f4 & mask32) * 100; + COPY(b + 6, digits_dd[f6 >> 32]); + + return b + 8; + } + + u64_t f0 = u64(10 * (1ull << 57ull) / 1e9 + 1) * n; + COPY(b, digits_fd[f0 >> 57]); + + b -= n < u32(1e9); + u64_t f2 = (f0 & mask57) * 100; + COPY(b + 2, digits_dd[f2 >> 57]); + + u64_t f4 = (f2 & mask57) * 100; + COPY(b + 4, digits_dd[f4 >> 57]); + + u64_t f6 = (f4 & mask57) * 100; + COPY(b + 6, digits_dd[f6 >> 57]); + + u64_t f8 = (f6 & mask57) * 100; + COPY(b + 8, digits_dd[f8 >> 57]); + + return b + 10; + } + + // if we get here U must be u64 but some compilers don't know that, so reassign n to a u64 to avoid warnings + u32_t z = n % u32(1e8); + u64_t u = n / u32(1e8); + + if (u < u32(1e2)) { + // u can't be 1 digit (if u < 10 it would have been handled above as a 9 digit 32bit number) + COPY(b, digits_dd[u]); + b += 2; + } + else if (u < u32(1e6)) { + if (u < u32(1e4)) { + u32_t f0 = u32((10 * (1 << 24) / 1e3 + 1) * u); + COPY(b, digits_fd[f0 >> 24]); + + b -= u < u32(1e3); + u32_t f2 = (f0 & mask24) * 100; + COPY(b + 2, digits_dd[f2 >> 24]); + b += 4; + } + else { + u64_t f0 = u64(10 * (1ull << 32ull) / 1e5 + 1) * u; + COPY(b, digits_fd[f0 >> 32]); + + b -= u < u32(1e5); + u64_t f2 = (f0 & mask32) * 100; + COPY(b + 2, digits_dd[f2 >> 32]); + + u64_t f4 = (f2 & mask32) * 100; + COPY(b + 4, digits_dd[f4 >> 32]); + b += 6; + } + } + else if (u < u32(1e8)) { + u64_t f0 = u64(10 * (1ull << 48ull) / 1e7 + 1) * u >> 16; + COPY(b, digits_fd[f0 >> 32]); + + b -= u < u32(1e7); + u64_t f2 = (f0 & mask32) * 100; + COPY(b + 2, digits_dd[f2 >> 32]); + + u64_t f4 = (f2 & mask32) * 100; + COPY(b + 4, digits_dd[f4 >> 32]); + + u64_t f6 = (f4 & mask32) * 100; + COPY(b + 6, digits_dd[f6 >> 32]); + + b += 8; + } + else if (u < u64(1ull << 32ull)) { + u64_t f0 = u64(10 * (1ull << 57ull) / 1e9 + 1) * u; + COPY(b, digits_fd[f0 >> 57]); + + b -= u < u32(1e9); + u64_t f2 = (f0 & mask57) * 100; + COPY(b + 2, digits_dd[f2 >> 57]); + + u64_t f4 = (f2 & mask57) * 100; + COPY(b + 4, digits_dd[f4 >> 57]); + + u64_t f6 = (f4 & mask57) * 100; + COPY(b + 6, digits_dd[f6 >> 57]); + + u64_t f8 = (f6 & mask57) * 100; + COPY(b + 8, digits_dd[f8 >> 57]); + b += 10; + } + else { + u32_t y = u % u32(1e8); + u /= u32(1e8); + + // u is 2, 3, or 4 digits (if u < 10 it would have been handled above) + if (u < u32(1e2)) { + COPY(b, digits_dd[u]); + b += 2; + } + else { + u32_t f0 = u32((10 * (1 << 24) / 1e3 + 1) * u); + COPY(b, digits_fd[f0 >> 24]); + + b -= u < u32(1e3); + u32_t f2 = (f0 & mask24) * 100; + COPY(b + 2, digits_dd[f2 >> 24]); + + b += 4; + } + // do 8 digits + u64_t f0 = (u64((1ull << 48ull) / 1e6 + 1) * y >> 16) + 1; + COPY(b, digits_dd[f0 >> 32]); + + u64_t f2 = (f0 & mask32) * 100; + COPY(b + 2, digits_dd[f2 >> 32]); + + u64_t f4 = (f2 & mask32) * 100; + COPY(b + 4, digits_dd[f4 >> 32]); + + u64_t f6 = (f4 & mask32) * 100; + COPY(b + 6, digits_dd[f6 >> 32]); + b += 8; + } + + // do 8 digits + u64_t f0 = (u64((1ull << 48ull) / 1e6 + 1) * z >> 16) + 1; + COPY(b, digits_dd[f0 >> 32]); + + u64_t f2 = (f0 & mask32) * 100; + COPY(b + 2, digits_dd[f2 >> 32]); + + u64_t f4 = (f2 & mask32) * 100; + COPY(b + 4, digits_dd[f4 >> 32]); + + u64_t f6 = (f4 & mask32) * 100; + COPY(b + 6, digits_dd[f6 >> 32]); + + return b + 8; +} + +#undef u32 +#undef u64 +#undef COPY + +#endif // JEAIII_TO_TEXT_H_ |