diff options
-rwxr-xr-x | benchmarks/generator2_benchmark.rb | 2 | ||||
-rwxr-xr-x | benchmarks/generator_benchmark.rb | 2 | ||||
-rwxr-xr-x | benchmarks/parser2_benchmark.rb | 2 | ||||
-rwxr-xr-x | benchmarks/parser_benchmark.rb | 2 | ||||
-rw-r--r-- | ext/json/ext/fbuffer.h | 156 | ||||
-rw-r--r-- | ext/json/ext/generator/generator.c | 187 | ||||
-rw-r--r-- | ext/json/ext/generator/generator.h | 41 | ||||
-rw-r--r-- | ext/json/ext/parser/parser.c | 177 | ||||
-rw-r--r-- | ext/json/ext/parser/parser.h | 7 | ||||
-rw-r--r-- | ext/json/ext/parser/parser.rl | 13 | ||||
-rw-r--r-- | java/src/json/ext/GeneratorState.java | 21 | ||||
-rw-r--r-- | lib/json/pure/generator.rb | 29 | ||||
-rwxr-xr-x | tests/test_json_generate.rb | 74 |
13 files changed, 408 insertions, 305 deletions
diff --git a/benchmarks/generator2_benchmark.rb b/benchmarks/generator2_benchmark.rb index 9885143..90fc711 100755 --- a/benchmarks/generator2_benchmark.rb +++ b/benchmarks/generator2_benchmark.rb @@ -199,7 +199,7 @@ if $0 == __FILE__ system "#{RAKE_PATH} clean" system "#{RUBY_PATH} #$0 rails" system "#{RUBY_PATH} #$0 pure" - system "#{RAKE_PATH} compile_ext" + system "#{RAKE_PATH} compile" system "#{RUBY_PATH} #$0 ext" system "#{RUBY_PATH} #$0 yajl" Bullshit.compare do diff --git a/benchmarks/generator_benchmark.rb b/benchmarks/generator_benchmark.rb index 83fa577..0554bca 100755 --- a/benchmarks/generator_benchmark.rb +++ b/benchmarks/generator_benchmark.rb @@ -201,7 +201,7 @@ if $0 == __FILE__ system "#{RAKE_PATH} clean" system "#{RUBY_PATH} #$0 rails" system "#{RUBY_PATH} #$0 pure" - system "#{RAKE_PATH} compile_ext" + system "#{RAKE_PATH} compile" system "#{RUBY_PATH} #$0 ext" system "#{RUBY_PATH} #$0 yajl" Bullshit.compare do diff --git a/benchmarks/parser2_benchmark.rb b/benchmarks/parser2_benchmark.rb index 95a510d..8798660 100755 --- a/benchmarks/parser2_benchmark.rb +++ b/benchmarks/parser2_benchmark.rb @@ -233,7 +233,7 @@ if $0 == __FILE__ system "#{RUBY_PATH} #$0 yaml" system "#{RUBY_PATH} #$0 rails" system "#{RUBY_PATH} #$0 pure" - system "#{RAKE_PATH} compile_ext" + system "#{RAKE_PATH} compile" system "#{RUBY_PATH} #$0 ext" system "#{RUBY_PATH} #$0 yajl" Bullshit.compare do diff --git a/benchmarks/parser_benchmark.rb b/benchmarks/parser_benchmark.rb index 9ce7e25..ed2331a 100755 --- a/benchmarks/parser_benchmark.rb +++ b/benchmarks/parser_benchmark.rb @@ -241,7 +241,7 @@ if $0 == __FILE__ system "#{RUBY_PATH} #$0 yaml" system "#{RUBY_PATH} #$0 rails" system "#{RUBY_PATH} #$0 pure" - system "#{RAKE_PATH} compile_ext" + system "#{RAKE_PATH} compile" system "#{RUBY_PATH} #$0 ext" system "#{RUBY_PATH} #$0 yajl" Bullshit.compare do diff --git a/ext/json/ext/fbuffer.h b/ext/json/ext/fbuffer.h new file mode 100644 index 0000000..7d1dad7 --- /dev/null +++ b/ext/json/ext/fbuffer.h @@ -0,0 +1,156 @@ + +#ifndef _FBUFFER_H_ +#define _FBUFFER_H_ + +#include <assert.h> +#include "ruby.h" + +#ifdef HAVE_RUBY_ENCODING_H +#include "ruby/encoding.h" +#define FORCE_UTF8(obj) rb_enc_associate((obj), rb_utf8_encoding()) +#else +#define FORCE_UTF8(obj) +#endif + +/* We don't need to guard objects for rbx, so let's do nothing at all. */ +#ifndef RB_GC_GUARD +#define RB_GC_GUARD(object) +#endif + +typedef struct FBufferStruct { + unsigned long initial_length; + char *ptr; + unsigned long len; + unsigned long capa; +} FBuffer; + +#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 FBuffer *fbuffer_alloc(unsigned long initial_length); +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 void fbuffer_append_char(FBuffer *fb, char newchr); +static FBuffer *fbuffer_dup(FBuffer *fb); +static VALUE fbuffer_to_s(FBuffer *fb); + +static FBuffer *fbuffer_alloc(unsigned long initial_length) +{ + FBuffer *fb; + if (initial_length <= 0) initial_length = FBUFFER_INITIAL_LENGTH_DEFAULT; + fb = ALLOC(FBuffer); + memset((void *) fb, 0, sizeof(FBuffer)); + fb->initial_length = initial_length; + return fb; +} + +static void fbuffer_free(FBuffer *fb) +{ + if (fb->ptr) ruby_xfree(fb->ptr); + ruby_xfree(fb); +} + +static void fbuffer_clear(FBuffer *fb) +{ + fb->len = 0; +} + +static void fbuffer_inc_capa(FBuffer *fb, unsigned long requested) +{ + unsigned long required; + + if (!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); + + if (required > fb->capa) { + REALLOC_N(fb->ptr, char, required); + fb->capa = required; + } +} + +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); + fb->len += len; + } +} + +static void fbuffer_append_str(FBuffer *fb, VALUE str) +{ + const char *newstr = StringValuePtr(str); + unsigned long len = RSTRING_LEN(str); + + RB_GC_GUARD(str); + + fbuffer_append(fb, newstr, len); +} + +static void fbuffer_append_char(FBuffer *fb, char newchr) +{ + fbuffer_inc_capa(fb, 1); + *(fb->ptr + fb->len) = newchr; + fb->len++; +} + +static void freverse(char *start, char *end) +{ + char c; + + while (end > start) { + c = *end, *end-- = *start, *start++ = c; + } +} + +static long fltoa(long number, char *buf) +{ + static char digits[] = "0123456789"; + long sign = number; + char* tmp = buf; + + if (sign < 0) number = -number; + do *tmp++ = digits[number % 10]; while (number /= 10); + if (sign < 0) *tmp++ = '-'; + freverse(buf, tmp - 1); + return tmp - buf; +} + +static void fbuffer_append_long(FBuffer *fb, long number) +{ + char buf[20]; + unsigned long len = fltoa(number, buf); + fbuffer_append(fb, buf, len); +} + +static FBuffer *fbuffer_dup(FBuffer *fb) +{ + unsigned long len = fb->len; + FBuffer *result; + + assert(len > 0); + if (len > 0) { + result = fbuffer_alloc(len); + fbuffer_append(result, FBUFFER_PAIR(fb)); + } + return result; +} + +static VALUE fbuffer_to_s(FBuffer *fb) +{ + VALUE result = rb_str_new(FBUFFER_PAIR(fb)); + fbuffer_free(fb); + FORCE_UTF8(result); + return result; +} +#endif diff --git a/ext/json/ext/generator/generator.c b/ext/json/ext/generator/generator.c index 781e9e6..2f768ac 100644 --- a/ext/json/ext/generator/generator.c +++ b/ext/json/ext/generator/generator.c @@ -1,3 +1,4 @@ +#include "../fbuffer.h" #include "generator.h" #ifdef HAVE_RUBY_ENCODING_H @@ -14,7 +15,8 @@ static VALUE mJSON, mExt, mGenerator, cState, mGeneratorMethods, mObject, static ID i_to_s, i_to_json, i_new, i_indent, i_space, i_space_before, i_object_nl, i_array_nl, i_max_nesting, i_allow_nan, i_ascii_only, i_quirks_mode, i_pack, i_unpack, i_create_id, i_extend, i_key_p, - i_aref, i_send, i_respond_to_p, i_match, i_keys, i_depth, i_dup; + i_aref, i_send, i_respond_to_p, i_match, i_keys, i_depth, + i_buffer_initial_length, i_dup; /* * Copyright 2001-2004 Unicode, Inc. @@ -292,123 +294,6 @@ static char *fstrndup(const char *ptr, unsigned long len) { return result; } -/* fbuffer implementation */ - -static FBuffer *fbuffer_alloc() -{ - FBuffer *fb = ALLOC(FBuffer); - memset((void *) fb, 0, sizeof(FBuffer)); - fb->initial_length = FBUFFER_INITIAL_LENGTH; - return fb; -} - -static FBuffer *fbuffer_alloc_with_length(unsigned long initial_length) -{ - FBuffer *fb; - assert(initial_length > 0); - fb = ALLOC(FBuffer); - memset((void *) fb, 0, sizeof(FBuffer)); - fb->initial_length = initial_length; - return fb; -} - -static void fbuffer_free(FBuffer *fb) -{ - if (fb->ptr) ruby_xfree(fb->ptr); - ruby_xfree(fb); -} - -static void fbuffer_clear(FBuffer *fb) -{ - fb->len = 0; -} - -static void fbuffer_inc_capa(FBuffer *fb, unsigned long requested) -{ - unsigned long required; - - if (!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); - - if (required > fb->capa) { - REALLOC_N(fb->ptr, char, required); - fb->capa = required; - } -} - -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); - fb->len += len; - } -} - -static void fbuffer_append_str(FBuffer *fb, VALUE str) -{ - const char *newstr = StringValuePtr(str); - unsigned long len = RSTRING_LEN(str); - - RB_GC_GUARD(str); - - fbuffer_append(fb, newstr, len); -} - -static void fbuffer_append_char(FBuffer *fb, char newchr) -{ - fbuffer_inc_capa(fb, 1); - *(fb->ptr + fb->len) = newchr; - fb->len++; -} - -static void freverse(char *start, char *end) -{ - char c; - - while (end > start) { - c = *end, *end-- = *start, *start++ = c; - } -} - -static long fltoa(long number, char *buf) -{ - static char digits[] = "0123456789"; - long sign = number; - char* tmp = buf; - - if (sign < 0) number = -number; - do *tmp++ = digits[number % 10]; while (number /= 10); - if (sign < 0) *tmp++ = '-'; - freverse(buf, tmp - 1); - return tmp - buf; -} - -static void fbuffer_append_long(FBuffer *fb, long number) -{ - char buf[20]; - unsigned long len = fltoa(number, buf); - fbuffer_append(fb, buf, len); -} - -static FBuffer *fbuffer_dup(FBuffer *fb) -{ - unsigned long len = fb->len; - FBuffer *result; - - if (len > 0) { - result = fbuffer_alloc_with_length(len); - fbuffer_append(result, FBUFFER_PAIR(fb)); - } else { - result = fbuffer_alloc(); - } - return result; -} - /* * Document-module: JSON::Ext::Generator * @@ -694,6 +579,16 @@ static VALUE cState_configure(VALUE self, VALUE opts) state->depth = 0; } } + tmp = ID2SYM(i_buffer_initial_length); + if (option_given_p(opts, tmp)) { + VALUE buffer_initial_length = rb_hash_aref(opts, tmp); + if (RTEST(buffer_initial_length)) { + long initial_length; + Check_Type(buffer_initial_length, T_FIXNUM); + initial_length = FIX2LONG(buffer_initial_length); + if (initial_length > 0) state->buffer_initial_length = initial_length; + } + } tmp = rb_hash_aref(opts, ID2SYM(i_allow_nan)); state->allow_nan = RTEST(tmp); tmp = rb_hash_aref(opts, ID2SYM(i_ascii_only)); @@ -723,6 +618,7 @@ static VALUE cState_to_h(VALUE self) rb_hash_aset(result, ID2SYM(i_quirks_mode), state->quirks_mode ? Qtrue : Qfalse); rb_hash_aset(result, ID2SYM(i_max_nesting), LONG2FIX(state->max_nesting)); rb_hash_aset(result, ID2SYM(i_depth), LONG2FIX(state->depth)); + rb_hash_aset(result, ID2SYM(i_buffer_initial_length), LONG2FIX(state->buffer_initial_length)); return result; } @@ -920,19 +816,20 @@ static void generate_json(FBuffer *buffer, VALUE Vstate, JSON_Generator_State *s static FBuffer *cState_prepare_buffer(VALUE self) { - FBuffer *buffer = fbuffer_alloc(); + FBuffer *buffer; GET_STATE(self); + buffer = fbuffer_alloc(state->buffer_initial_length); if (state->object_delim) { fbuffer_clear(state->object_delim); } else { - state->object_delim = fbuffer_alloc_with_length(16); + state->object_delim = fbuffer_alloc(16); } fbuffer_append_char(state->object_delim, ','); if (state->object_delim2) { fbuffer_clear(state->object_delim2); } else { - state->object_delim2 = fbuffer_alloc_with_length(16); + state->object_delim2 = fbuffer_alloc(16); } fbuffer_append_char(state->object_delim2, ':'); if (state->space) fbuffer_append(state->object_delim2, state->space, state->space_len); @@ -940,21 +837,13 @@ static FBuffer *cState_prepare_buffer(VALUE self) if (state->array_delim) { fbuffer_clear(state->array_delim); } else { - state->array_delim = fbuffer_alloc_with_length(16); + state->array_delim = fbuffer_alloc(16); } fbuffer_append_char(state->array_delim, ','); if (state->array_nl) fbuffer_append(state->array_delim, state->array_nl, state->array_nl_len); return buffer; } -static VALUE fbuffer_to_s(FBuffer *fb) -{ - VALUE result = rb_str_new(FBUFFER_PAIR(fb)); - fbuffer_free(fb); - FORCE_UTF8(result); - return result; -} - static VALUE cState_partial_generate(VALUE self, VALUE obj) { FBuffer *buffer = cState_prepare_buffer(self); @@ -1003,12 +892,15 @@ static VALUE cState_generate(VALUE self, VALUE obj) * encountered. This options defaults to false. * * *quirks_mode*: Enables quirks_mode for parser, that is for example * generating single JSON values instead of documents is possible. + * * *buffer_initial_length*: sets the initial length of the generator's + * internal buffer. */ static VALUE cState_initialize(int argc, VALUE *argv, VALUE self) { VALUE opts; GET_STATE(self); state->max_nesting = 19; + state->buffer_initial_length = FBUFFER_INITIAL_LENGTH_DEFAULT; rb_scan_args(argc, argv, "01", &opts); if (!NIL_P(opts)) cState_configure(self, opts); return self; @@ -1349,7 +1241,37 @@ static VALUE cState_depth_set(VALUE self, VALUE depth) { GET_STATE(self); Check_Type(depth, T_FIXNUM); - return state->depth = FIX2LONG(depth); + state->depth = FIX2LONG(depth); + return Qnil; +} + +/* + * call-seq: buffer_initial_length + * + * This integer returns the current inital length of the buffer. + */ +static VALUE cState_buffer_initial_length(VALUE self) +{ + GET_STATE(self); + return LONG2FIX(state->buffer_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) +{ + long initial_length; + GET_STATE(self); + Check_Type(buffer_initial_length, T_FIXNUM); + initial_length = FIX2LONG(buffer_initial_length); + if (initial_length > 0) { + state->buffer_initial_length = initial_length; + } + return Qnil; } /* @@ -1391,6 +1313,8 @@ void Init_generator() rb_define_method(cState, "quirks_mode=", cState_quirks_mode_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, "configure", cState_configure, 1); rb_define_alias(cState, "merge", "configure"); rb_define_method(cState, "to_h", cState_to_h, 0); @@ -1438,6 +1362,7 @@ void Init_generator() i_ascii_only = rb_intern("ascii_only"); i_quirks_mode = rb_intern("quirks_mode"); i_depth = rb_intern("depth"); + i_buffer_initial_length = rb_intern("buffer_initial_length"); i_pack = rb_intern("pack"); i_unpack = rb_intern("unpack"); i_create_id = rb_intern("create_id"); diff --git a/ext/json/ext/generator/generator.h b/ext/json/ext/generator/generator.h index 1fdd351..3220178 100644 --- a/ext/json/ext/generator/generator.h +++ b/ext/json/ext/generator/generator.h @@ -13,13 +13,6 @@ #include "re.h" #endif -#ifdef HAVE_RUBY_ENCODING_H -#include "ruby/encoding.h" -#define FORCE_UTF8(obj) rb_enc_associate((obj), rb_utf8_encoding()) -#else -#define FORCE_UTF8(obj) -#endif - #define option_given_p(opts, key) RTEST(rb_funcall(opts, i_key_p, 1, key)) #ifndef RHASH_SIZE @@ -43,38 +36,6 @@ #define RSTRING_LEN(string) RSTRING(string)->len #endif -/* We don't need to guard objects for rbx, so let's do nothing at all. */ -#ifndef RB_GC_GUARD -#define RB_GC_GUARD(object) -#endif - -/* fbuffer implementation */ - -typedef struct FBufferStruct { - unsigned long initial_length; - char *ptr; - unsigned long len; - unsigned long capa; -} FBuffer; - -#define FBUFFER_INITIAL_LENGTH 4096 - -#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 char *fstrndup(const char *ptr, unsigned long len); -static FBuffer *fbuffer_alloc(); -static FBuffer *fbuffer_alloc_with_length(unsigned long initial_length); -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 void fbuffer_append_char(FBuffer *fb, char newchr); -static FBuffer *fbuffer_dup(FBuffer *fb); -static VALUE fbuffer_to_s(FBuffer *fb); - /* unicode defintions */ #define UNI_STRICT_CONVERSION 1 @@ -104,6 +65,7 @@ static void unicode_escape(char *buf, UTF16 character); static void unicode_escape_to_buffer(FBuffer *buffer, char buf[6], UTF16 character); static void convert_UTF8_to_JSON_ASCII(FBuffer *buffer, VALUE string); static void convert_UTF8_to_JSON(FBuffer *buffer, VALUE string); +static char *fstrndup(const char *ptr, unsigned long len); /* ruby api and some helpers */ @@ -126,6 +88,7 @@ typedef struct JSON_Generator_StateStruct { char ascii_only; char quirks_mode; long depth; + long buffer_initial_length; } JSON_Generator_State; #define GET_STATE(self) \ diff --git a/ext/json/ext/parser/parser.c b/ext/json/ext/parser/parser.c index 4bbc0c7..067aef4 100644 --- a/ext/json/ext/parser/parser.c +++ b/ext/json/ext/parser/parser.c @@ -1,5 +1,6 @@ #line 1 "parser.rl" +#include "../fbuffer.h" #include "parser.h" /* unicode */ @@ -83,11 +84,11 @@ static ID i_json_creatable_p, i_json_create, i_create_id, i_create_additions, i_match_string, i_aset, i_leftshift; -#line 109 "parser.rl" +#line 110 "parser.rl" -#line 91 "parser.c" +#line 92 "parser.c" static const int JSON_object_start = 1; static const int JSON_object_first_final = 27; static const int JSON_object_error = 0; @@ -95,7 +96,7 @@ static const int JSON_object_error = 0; static const int JSON_object_en_main = 1; -#line 150 "parser.rl" +#line 151 "parser.rl" static char *JSON_parse_object(JSON_Parser *json, char *p, char *pe, VALUE *result) @@ -111,14 +112,14 @@ static char *JSON_parse_object(JSON_Parser *json, char *p, char *pe, VALUE *resu *result = NIL_P(object_class) ? rb_hash_new() : rb_class_new_instance(0, 0, object_class); -#line 115 "parser.c" +#line 116 "parser.c" { cs = JSON_object_start; } -#line 165 "parser.rl" +#line 166 "parser.rl" -#line 122 "parser.c" +#line 123 "parser.c" { if ( p == pe ) goto _test_eof; @@ -146,7 +147,7 @@ case 2: goto st2; goto st0; tr2: -#line 132 "parser.rl" +#line 133 "parser.rl" { char *np; json->parsing_name = 1; @@ -159,7 +160,7 @@ st3: if ( ++p == pe ) goto _test_eof3; case 3: -#line 163 "parser.c" +#line 164 "parser.c" switch( (*p) ) { case 13: goto st3; case 32: goto st3; @@ -226,7 +227,7 @@ case 8: goto st8; goto st0; tr11: -#line 117 "parser.rl" +#line 118 "parser.rl" { VALUE v = Qnil; char *np = JSON_parse_value(json, p, pe, &v); @@ -246,7 +247,7 @@ st9: if ( ++p == pe ) goto _test_eof9; case 9: -#line 250 "parser.c" +#line 251 "parser.c" switch( (*p) ) { case 13: goto st9; case 32: goto st9; @@ -335,14 +336,14 @@ case 18: goto st9; goto st18; tr4: -#line 140 "parser.rl" +#line 141 "parser.rl" { p--; {p++; cs = 27; goto _out;} } goto st27; st27: if ( ++p == pe ) goto _test_eof27; case 27: -#line 346 "parser.c" +#line 347 "parser.c" goto st0; st19: if ( ++p == pe ) @@ -440,7 +441,7 @@ case 26: _out: {} } -#line 166 "parser.rl" +#line 167 "parser.rl" if (cs >= JSON_object_first_final) { if (json->create_additions) { @@ -460,7 +461,7 @@ case 26: -#line 464 "parser.c" +#line 465 "parser.c" static const int JSON_value_start = 1; static const int JSON_value_first_final = 21; static const int JSON_value_error = 0; @@ -468,7 +469,7 @@ static const int JSON_value_error = 0; static const int JSON_value_en_main = 1; -#line 265 "parser.rl" +#line 266 "parser.rl" static char *JSON_parse_value(JSON_Parser *json, char *p, char *pe, VALUE *result) @@ -476,14 +477,14 @@ static char *JSON_parse_value(JSON_Parser *json, char *p, char *pe, VALUE *resul int cs = EVIL; -#line 480 "parser.c" +#line 481 "parser.c" { cs = JSON_value_start; } -#line 272 "parser.rl" +#line 273 "parser.rl" -#line 487 "parser.c" +#line 488 "parser.c" { if ( p == pe ) goto _test_eof; @@ -508,14 +509,14 @@ st0: cs = 0; goto _out; tr0: -#line 213 "parser.rl" +#line 214 "parser.rl" { char *np = JSON_parse_string(json, p, pe, result); if (np == NULL) { p--; {p++; cs = 21; goto _out;} } else {p = (( np))-1;} } goto st21; tr2: -#line 218 "parser.rl" +#line 219 "parser.rl" { char *np; if(pe > p + 9 - json->quirks_mode && !strncmp(MinusInfinity, p, 9)) { @@ -535,7 +536,7 @@ tr2: } goto st21; tr5: -#line 236 "parser.rl" +#line 237 "parser.rl" { char *np; json->current_nesting++; @@ -545,7 +546,7 @@ tr5: } goto st21; tr9: -#line 244 "parser.rl" +#line 245 "parser.rl" { char *np; json->current_nesting++; @@ -555,7 +556,7 @@ tr9: } goto st21; tr16: -#line 206 "parser.rl" +#line 207 "parser.rl" { if (json->allow_nan) { *result = CInfinity; @@ -565,7 +566,7 @@ tr16: } goto st21; tr18: -#line 199 "parser.rl" +#line 200 "parser.rl" { if (json->allow_nan) { *result = CNaN; @@ -575,19 +576,19 @@ tr18: } goto st21; tr22: -#line 193 "parser.rl" +#line 194 "parser.rl" { *result = Qfalse; } goto st21; tr25: -#line 190 "parser.rl" +#line 191 "parser.rl" { *result = Qnil; } goto st21; tr28: -#line 196 "parser.rl" +#line 197 "parser.rl" { *result = Qtrue; } @@ -596,9 +597,9 @@ st21: if ( ++p == pe ) goto _test_eof21; case 21: -#line 252 "parser.rl" +#line 253 "parser.rl" { p--; {p++; cs = 21; goto _out;} } -#line 602 "parser.c" +#line 603 "parser.c" goto st0; st2: if ( ++p == pe ) @@ -759,7 +760,7 @@ case 20: _out: {} } -#line 273 "parser.rl" +#line 274 "parser.rl" if (cs >= JSON_value_first_final) { return p; @@ -769,7 +770,7 @@ case 20: } -#line 773 "parser.c" +#line 774 "parser.c" static const int JSON_integer_start = 1; static const int JSON_integer_first_final = 3; static const int JSON_integer_error = 0; @@ -777,7 +778,7 @@ static const int JSON_integer_error = 0; static const int JSON_integer_en_main = 1; -#line 289 "parser.rl" +#line 290 "parser.rl" static char *JSON_parse_integer(JSON_Parser *json, char *p, char *pe, VALUE *result) @@ -785,15 +786,15 @@ static char *JSON_parse_integer(JSON_Parser *json, char *p, char *pe, VALUE *res int cs = EVIL; -#line 789 "parser.c" +#line 790 "parser.c" { cs = JSON_integer_start; } -#line 296 "parser.rl" +#line 297 "parser.rl" json->memo = p; -#line 797 "parser.c" +#line 798 "parser.c" { if ( p == pe ) goto _test_eof; @@ -827,14 +828,14 @@ case 3: goto st0; goto tr4; tr4: -#line 286 "parser.rl" +#line 287 "parser.rl" { p--; {p++; cs = 4; goto _out;} } goto st4; st4: if ( ++p == pe ) goto _test_eof4; case 4: -#line 838 "parser.c" +#line 839 "parser.c" goto st0; st5: if ( ++p == pe ) @@ -853,11 +854,14 @@ case 5: _out: {} } -#line 298 "parser.rl" +#line 299 "parser.rl" if (cs >= JSON_integer_first_final) { long len = p - json->memo; - *result = rb_Integer(rb_str_new(json->memo, len)); + fbuffer_clear(json->fbuffer); + fbuffer_append(json->fbuffer, json->memo, len); + fbuffer_append_char(json->fbuffer, '\0'); + *result = rb_cstr2inum(FBUFFER_PTR(json->fbuffer), 10); return p + 1; } else { return NULL; @@ -865,7 +869,7 @@ case 5: } -#line 869 "parser.c" +#line 873 "parser.c" static const int JSON_float_start = 1; static const int JSON_float_first_final = 8; static const int JSON_float_error = 0; @@ -873,7 +877,7 @@ static const int JSON_float_error = 0; static const int JSON_float_en_main = 1; -#line 320 "parser.rl" +#line 324 "parser.rl" static char *JSON_parse_float(JSON_Parser *json, char *p, char *pe, VALUE *result) @@ -881,15 +885,15 @@ static char *JSON_parse_float(JSON_Parser *json, char *p, char *pe, VALUE *resul int cs = EVIL; -#line 885 "parser.c" +#line 889 "parser.c" { cs = JSON_float_start; } -#line 327 "parser.rl" +#line 331 "parser.rl" json->memo = p; -#line 893 "parser.c" +#line 897 "parser.c" { if ( p == pe ) goto _test_eof; @@ -947,14 +951,14 @@ case 8: goto st0; goto tr9; tr9: -#line 314 "parser.rl" +#line 318 "parser.rl" { p--; {p++; cs = 9; goto _out;} } goto st9; st9: if ( ++p == pe ) goto _test_eof9; case 9: -#line 958 "parser.c" +#line 962 "parser.c" goto st0; st5: if ( ++p == pe ) @@ -1015,11 +1019,14 @@ case 7: _out: {} } -#line 329 "parser.rl" +#line 333 "parser.rl" if (cs >= JSON_float_first_final) { long len = p - json->memo; - *result = rb_Float(rb_str_new(json->memo, len)); + fbuffer_clear(json->fbuffer); + fbuffer_append(json->fbuffer, json->memo, len); + fbuffer_append_char(json->fbuffer, '\0'); + *result = rb_float_new(rb_cstr_to_dbl(FBUFFER_PTR(json->fbuffer), 1)); return p + 1; } else { return NULL; @@ -1028,7 +1035,7 @@ case 7: -#line 1032 "parser.c" +#line 1039 "parser.c" static const int JSON_array_start = 1; static const int JSON_array_first_final = 17; static const int JSON_array_error = 0; @@ -1036,7 +1043,7 @@ static const int JSON_array_error = 0; static const int JSON_array_en_main = 1; -#line 369 "parser.rl" +#line 376 "parser.rl" static char *JSON_parse_array(JSON_Parser *json, char *p, char *pe, VALUE *result) @@ -1050,14 +1057,14 @@ static char *JSON_parse_array(JSON_Parser *json, char *p, char *pe, VALUE *resul *result = NIL_P(array_class) ? rb_ary_new() : rb_class_new_instance(0, 0, array_class); -#line 1054 "parser.c" +#line 1061 "parser.c" { cs = JSON_array_start; } -#line 382 "parser.rl" +#line 389 "parser.rl" -#line 1061 "parser.c" +#line 1068 "parser.c" { if ( p == pe ) goto _test_eof; @@ -1096,7 +1103,7 @@ case 2: goto st2; goto st0; tr2: -#line 346 "parser.rl" +#line 353 "parser.rl" { VALUE v = Qnil; char *np = JSON_parse_value(json, p, pe, &v); @@ -1116,7 +1123,7 @@ st3: if ( ++p == pe ) goto _test_eof3; case 3: -#line 1120 "parser.c" +#line 1127 "parser.c" switch( (*p) ) { case 13: goto st3; case 32: goto st3; @@ -1216,14 +1223,14 @@ case 12: goto st3; goto st12; tr4: -#line 361 "parser.rl" +#line 368 "parser.rl" { p--; {p++; cs = 17; goto _out;} } goto st17; st17: if ( ++p == pe ) goto _test_eof17; case 17: -#line 1227 "parser.c" +#line 1234 "parser.c" goto st0; st13: if ( ++p == pe ) @@ -1279,7 +1286,7 @@ case 16: _out: {} } -#line 383 "parser.rl" +#line 390 "parser.rl" if(cs >= JSON_array_first_final) { return p + 1; @@ -1360,7 +1367,7 @@ static VALUE json_string_unescape(VALUE result, char *string, char *stringEnd) } -#line 1364 "parser.c" +#line 1371 "parser.c" static const int JSON_string_start = 1; static const int JSON_string_first_final = 8; static const int JSON_string_error = 0; @@ -1368,7 +1375,7 @@ static const int JSON_string_error = 0; static const int JSON_string_en_main = 1; -#line 482 "parser.rl" +#line 489 "parser.rl" static int @@ -1390,15 +1397,15 @@ static char *JSON_parse_string(JSON_Parser *json, char *p, char *pe, VALUE *resu *result = rb_str_buf_new(0); -#line 1394 "parser.c" +#line 1401 "parser.c" { cs = JSON_string_start; } -#line 503 "parser.rl" +#line 510 "parser.rl" json->memo = p; -#line 1402 "parser.c" +#line 1409 "parser.c" { if ( p == pe ) goto _test_eof; @@ -1423,7 +1430,7 @@ case 2: goto st0; goto st2; tr2: -#line 468 "parser.rl" +#line 475 "parser.rl" { *result = json_string_unescape(*result, json->memo + 1, p); if (NIL_P(*result)) { @@ -1434,14 +1441,14 @@ tr2: {p = (( p + 1))-1;} } } -#line 479 "parser.rl" +#line 486 "parser.rl" { p--; {p++; cs = 8; goto _out;} } goto st8; st8: if ( ++p == pe ) goto _test_eof8; case 8: -#line 1445 "parser.c" +#line 1452 "parser.c" goto st0; st3: if ( ++p == pe ) @@ -1517,7 +1524,7 @@ case 7: _out: {} } -#line 505 "parser.rl" +#line 512 "parser.rl" if (json->create_additions && RTEST(match_string = json->match_string)) { VALUE klass; @@ -1716,7 +1723,7 @@ static VALUE cParser_initialize(int argc, VALUE *argv, VALUE self) } -#line 1720 "parser.c" +#line 1727 "parser.c" static const int JSON_start = 1; static const int JSON_first_final = 10; static const int JSON_error = 0; @@ -1724,7 +1731,7 @@ static const int JSON_error = 0; static const int JSON_en_main = 1; -#line 727 "parser.rl" +#line 734 "parser.rl" static VALUE cParser_parse_strict(VALUE self) @@ -1735,16 +1742,16 @@ static VALUE cParser_parse_strict(VALUE self) GET_PARSER; -#line 1739 "parser.c" +#line 1746 "parser.c" { cs = JSON_start; } -#line 737 "parser.rl" +#line 744 "parser.rl" p = json->source; pe = p + json->len; -#line 1748 "parser.c" +#line 1755 "parser.c" { if ( p == pe ) goto _test_eof; @@ -1800,7 +1807,7 @@ case 5: goto st1; goto st5; tr3: -#line 716 "parser.rl" +#line 723 "parser.rl" { char *np; json->current_nesting = 1; @@ -1809,7 +1816,7 @@ tr3: } goto st10; tr4: -#line 709 "parser.rl" +#line 716 "parser.rl" { char *np; json->current_nesting = 1; @@ -1821,7 +1828,7 @@ st10: if ( ++p == pe ) goto _test_eof10; case 10: -#line 1825 "parser.c" +#line 1832 "parser.c" switch( (*p) ) { case 13: goto st10; case 32: goto st10; @@ -1878,7 +1885,7 @@ case 9: _out: {} } -#line 740 "parser.rl" +#line 747 "parser.rl" if (cs >= JSON_first_final && p == pe) { return result; @@ -1890,7 +1897,7 @@ case 9: -#line 1894 "parser.c" +#line 1901 "parser.c" static const int JSON_quirks_mode_start = 1; static const int JSON_quirks_mode_first_final = 10; static const int JSON_quirks_mode_error = 0; @@ -1898,7 +1905,7 @@ static const int JSON_quirks_mode_error = 0; static const int JSON_quirks_mode_en_main = 1; -#line 765 "parser.rl" +#line 772 "parser.rl" static VALUE cParser_parse_quirks_mode(VALUE self) @@ -1909,16 +1916,16 @@ static VALUE cParser_parse_quirks_mode(VALUE self) GET_PARSER; -#line 1913 "parser.c" +#line 1920 "parser.c" { cs = JSON_quirks_mode_start; } -#line 775 "parser.rl" +#line 782 "parser.rl" p = json->source; pe = p + json->len; -#line 1922 "parser.c" +#line 1929 "parser.c" { if ( p == pe ) goto _test_eof; @@ -1952,7 +1959,7 @@ st0: cs = 0; goto _out; tr2: -#line 757 "parser.rl" +#line 764 "parser.rl" { char *np = JSON_parse_value(json, p, pe, &result); if (np == NULL) { p--; {p++; cs = 10; goto _out;} } else {p = (( np))-1;} @@ -1962,7 +1969,7 @@ st10: if ( ++p == pe ) goto _test_eof10; case 10: -#line 1966 "parser.c" +#line 1973 "parser.c" switch( (*p) ) { case 13: goto st10; case 32: goto st10; @@ -2051,7 +2058,7 @@ case 9: _out: {} } -#line 778 "parser.rl" +#line 785 "parser.rl" if (cs >= JSON_quirks_mode_first_final && p == pe) { return result; @@ -2083,6 +2090,7 @@ static JSON_Parser *JSON_allocate() { JSON_Parser *json = ALLOC(JSON_Parser); MEMZERO(json, JSON_Parser, 1); + json->fbuffer = fbuffer_alloc(0); return json; } @@ -2097,6 +2105,7 @@ static void JSON_mark(JSON_Parser *json) static void JSON_free(JSON_Parser *json) { + fbuffer_free(json->fbuffer); ruby_xfree(json); } diff --git a/ext/json/ext/parser/parser.h b/ext/json/ext/parser/parser.h index da6fc5c..b192064 100644 --- a/ext/json/ext/parser/parser.h +++ b/ext/json/ext/parser/parser.h @@ -7,12 +7,6 @@ #include "re.h" #endif -#ifdef HAVE_RUBY_ENCODING_H -#include "ruby/encoding.h" -#define FORCE_UTF8(obj) ((obj) = rb_enc_associate(rb_str_dup(obj), rb_utf8_encoding())) -#else -#define FORCE_UTF8(obj) -#endif #ifdef HAVE_RUBY_ST_H #include "ruby/st.h" #else @@ -49,6 +43,7 @@ typedef struct JSON_ParserStruct { VALUE array_class; int create_additions; VALUE match_string; + FBuffer *fbuffer; } JSON_Parser; #define GET_PARSER \ diff --git a/ext/json/ext/parser/parser.rl b/ext/json/ext/parser/parser.rl index 8c4681d..d9ae84c 100644 --- a/ext/json/ext/parser/parser.rl +++ b/ext/json/ext/parser/parser.rl @@ -1,3 +1,4 @@ +#include "../fbuffer.h" #include "parser.h" /* unicode */ @@ -298,7 +299,10 @@ static char *JSON_parse_integer(JSON_Parser *json, char *p, char *pe, VALUE *res if (cs >= JSON_integer_first_final) { long len = p - json->memo; - *result = rb_Integer(rb_str_new(json->memo, len)); + fbuffer_clear(json->fbuffer); + fbuffer_append(json->fbuffer, json->memo, len); + fbuffer_append_char(json->fbuffer, '\0'); + *result = rb_cstr2inum(FBUFFER_PTR(json->fbuffer), 10); return p + 1; } else { return NULL; @@ -329,7 +333,10 @@ static char *JSON_parse_float(JSON_Parser *json, char *p, char *pe, VALUE *resul if (cs >= JSON_float_first_final) { long len = p - json->memo; - *result = rb_Float(rb_str_new(json->memo, len)); + fbuffer_clear(json->fbuffer); + fbuffer_append(json->fbuffer, json->memo, len); + fbuffer_append_char(json->fbuffer, '\0'); + *result = rb_float_new(rb_cstr_to_dbl(FBUFFER_PTR(json->fbuffer), 1)); return p + 1; } else { return NULL; @@ -806,6 +813,7 @@ static JSON_Parser *JSON_allocate() { JSON_Parser *json = ALLOC(JSON_Parser); MEMZERO(json, JSON_Parser, 1); + json->fbuffer = fbuffer_alloc(0); return json; } @@ -820,6 +828,7 @@ static void JSON_mark(JSON_Parser *json) static void JSON_free(JSON_Parser *json) { + fbuffer_free(json->fbuffer); ruby_xfree(json); } diff --git a/java/src/json/ext/GeneratorState.java b/java/src/json/ext/GeneratorState.java index 78524a1..65ee984 100644 --- a/java/src/json/ext/GeneratorState.java +++ b/java/src/json/ext/GeneratorState.java @@ -83,6 +83,12 @@ public class GeneratorState extends RubyObject { */ private boolean quirksMode = DEFAULT_QUIRKS_MODE; static final boolean DEFAULT_QUIRKS_MODE = false; + /** + * The initial buffer length of this state. (This isn't really used on all + * non-C implementations.) + */ + private int bufferInitialLength = DEFAULT_BUFFER_INITIAL_LENGTH; + static final int DEFAULT_BUFFER_INITIAL_LENGTH = 1024; /** * The current depth (inside a #to_json call) @@ -189,6 +195,7 @@ public class GeneratorState extends RubyObject { this.allowNaN = orig.allowNaN; this.asciiOnly = orig.asciiOnly; this.quirksMode = orig.quirksMode; + this.bufferInitialLength = orig.bufferInitialLength; this.depth = orig.depth; return this; } @@ -385,6 +392,18 @@ public class GeneratorState extends RubyObject { return quirks_mode.getRuntime().newBoolean(quirksMode); } + @JRubyMethod(name="buffer_initial_length") + public RubyInteger buffer_initial_length_get(ThreadContext context) { + return context.getRuntime().newFixnum(bufferInitialLength); + } + + @JRubyMethod(name="buffer_initial_length=") + public IRubyObject buffer_initial_length_set(IRubyObject buffer_initial_length) { + int newLength = RubyNumeric.fix2int(buffer_initial_length); + if (newLength > 0) bufferInitialLength = newLength; + return buffer_initial_length; + } + @JRubyMethod(name="quirks_mode?") public RubyBoolean quirks_mode_p(ThreadContext context) { return context.getRuntime().newBoolean(quirksMode); @@ -445,6 +464,7 @@ public class GeneratorState extends RubyObject { allowNaN = opts.getBool("allow_nan", DEFAULT_ALLOW_NAN); asciiOnly = opts.getBool("ascii_only", DEFAULT_ASCII_ONLY); quirksMode = opts.getBool("quirks_mode", DEFAULT_QUIRKS_MODE); + bufferInitialLength = opts.getInt("buffer_initial_length", DEFAULT_BUFFER_INITIAL_LENGTH); depth = opts.getInt("depth", 0); @@ -473,6 +493,7 @@ public class GeneratorState extends RubyObject { result.op_aset(context, runtime.newSymbol("quirks_mode"), quirks_mode_p(context)); result.op_aset(context, runtime.newSymbol("max_nesting"), max_nesting_get(context)); result.op_aset(context, runtime.newSymbol("depth"), depth_get(context)); + result.op_aset(context, runtime.newSymbol("buffer_initial_length"), buffer_initial_length_get(context)); return result; } diff --git a/lib/json/pure/generator.rb b/lib/json/pure/generator.rb index 7c9b2ad..9141ae5 100644 --- a/lib/json/pure/generator.rb +++ b/lib/json/pure/generator.rb @@ -136,14 +136,15 @@ module JSON # * *quirks_mode*: Enables quirks_mode for parser, that is for example # generating single JSON values instead of documents is possible. def initialize(opts = {}) - @indent = '' - @space = '' - @space_before = '' - @object_nl = '' - @array_nl = '' - @allow_nan = false - @ascii_only = false - @quirks_mode = false + @indent = '' + @space = '' + @space_before = '' + @object_nl = '' + @array_nl = '' + @allow_nan = false + @ascii_only = false + @quirks_mode = false + @buffer_initial_length = 1024 configure opts end @@ -172,6 +173,16 @@ module JSON # it's disabled. attr_accessor :quirks_mode + # :stopdoc: + attr_reader :buffer_initial_length + + def buffer_initial_length=(length) + if length > 0 + @buffer_initial_length = length + end + end + # :startdoc: + # This integer returns the current depth data structure nesting in the # generated JSON. attr_accessor :depth @@ -233,7 +244,7 @@ module JSON # passed to the configure method. def to_h result = {} - for iv in %w[indent space space_before object_nl array_nl allow_nan max_nesting ascii_only quirks_mode depth] + for iv in %w[indent space space_before object_nl array_nl allow_nan max_nesting ascii_only quirks_mode buffer_initial_length depth] result[iv.intern] = instance_variable_get("@#{iv}") end result diff --git a/tests/test_json_generate.rb b/tests/test_json_generate.rb index a8741a5..b363de1 100755 --- a/tests/test_json_generate.rb +++ b/tests/test_json_generate.rb @@ -123,48 +123,51 @@ EOT def test_pretty_state state = PRETTY_STATE_PROTOTYPE.dup assert_equal({ - :allow_nan => false, - :array_nl => "\n", - :ascii_only => false, - :quirks_mode => false, - :depth => 0, - :indent => " ", - :max_nesting => 19, - :object_nl => "\n", - :space => " ", - :space_before => "", + :allow_nan => false, + :array_nl => "\n", + :ascii_only => false, + :buffer_initial_length => 1024, + :quirks_mode => false, + :depth => 0, + :indent => " ", + :max_nesting => 19, + :object_nl => "\n", + :space => " ", + :space_before => "", }.sort_by { |n,| n.to_s }, state.to_h.sort_by { |n,| n.to_s }) end def test_safe_state state = SAFE_STATE_PROTOTYPE.dup assert_equal({ - :allow_nan => false, - :array_nl => "", - :ascii_only => false, - :quirks_mode => false, - :depth => 0, - :indent => "", - :max_nesting => 19, - :object_nl => "", - :space => "", - :space_before => "", + :allow_nan => false, + :array_nl => "", + :ascii_only => false, + :buffer_initial_length => 1024, + :quirks_mode => false, + :depth => 0, + :indent => "", + :max_nesting => 19, + :object_nl => "", + :space => "", + :space_before => "", }.sort_by { |n,| n.to_s }, state.to_h.sort_by { |n,| n.to_s }) end def test_fast_state state = FAST_STATE_PROTOTYPE.dup assert_equal({ - :allow_nan => false, - :array_nl => "", - :ascii_only => false, - :quirks_mode => false, - :depth => 0, - :indent => "", - :max_nesting => 0, - :object_nl => "", - :space => "", - :space_before => "", + :allow_nan => false, + :array_nl => "", + :ascii_only => false, + :buffer_initial_length => 1024, + :quirks_mode => false, + :depth => 0, + :indent => "", + :max_nesting => 0, + :object_nl => "", + :space => "", + :space_before => "", }.sort_by { |n,| n.to_s }, state.to_h.sort_by { |n,| n.to_s }) end @@ -200,6 +203,17 @@ EOT assert_equal 19, s.depth end + def test_buffer_initial_length + s = JSON.state.new + assert_equal 1024, s.buffer_initial_length + s.buffer_initial_length = 0 + assert_equal 1024, s.buffer_initial_length + s.buffer_initial_length = -1 + assert_equal 1024, s.buffer_initial_length + s.buffer_initial_length = 128 + assert_equal 128, s.buffer_initial_length + end + def test_gc bignum_too_long_to_embed_as_string = 1234567890123456789012345 expect = bignum_too_long_to_embed_as_string.to_s |