json 1.8.3 → 2.6.0

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
Files changed (108) hide show
  1. checksums.yaml +5 -5
  2. data/{CHANGES → CHANGES.md} +244 -93
  3. data/{COPYING-json-jruby → LICENSE} +5 -6
  4. data/{README.rdoc → README.md} +202 -135
  5. data/VERSION +1 -1
  6. data/ext/json/ext/fbuffer/fbuffer.h +0 -3
  7. data/ext/json/ext/generator/generator.c +264 -114
  8. data/ext/json/ext/generator/generator.h +12 -4
  9. data/ext/json/ext/parser/extconf.rb +29 -0
  10. data/ext/json/ext/parser/parser.c +3002 -1880
  11. data/ext/json/ext/parser/parser.h +10 -6
  12. data/ext/json/ext/parser/parser.rl +230 -192
  13. data/ext/json/extconf.rb +1 -1
  14. data/json.gemspec +0 -0
  15. data/lib/json/add/bigdecimal.rb +3 -2
  16. data/lib/json/add/complex.rb +4 -4
  17. data/lib/json/add/core.rb +1 -0
  18. data/lib/json/add/date.rb +1 -1
  19. data/lib/json/add/date_time.rb +1 -1
  20. data/lib/json/add/exception.rb +1 -1
  21. data/lib/json/add/ostruct.rb +3 -3
  22. data/lib/json/add/range.rb +1 -1
  23. data/lib/json/add/rational.rb +3 -3
  24. data/lib/json/add/regexp.rb +3 -3
  25. data/lib/json/add/set.rb +29 -0
  26. data/lib/json/add/struct.rb +1 -1
  27. data/lib/json/add/symbol.rb +1 -1
  28. data/lib/json/add/time.rb +1 -1
  29. data/lib/json/common.rb +381 -162
  30. data/lib/json/ext.rb +0 -6
  31. data/lib/json/generic_object.rb +5 -4
  32. data/lib/json/pure/generator.rb +83 -126
  33. data/lib/json/pure/parser.rb +63 -85
  34. data/lib/json/pure.rb +2 -8
  35. data/lib/json/version.rb +2 -1
  36. data/lib/json.rb +550 -29
  37. metadata +23 -132
  38. data/.gitignore +0 -16
  39. data/.travis.yml +0 -26
  40. data/COPYING +0 -58
  41. data/GPL +0 -340
  42. data/Gemfile +0 -10
  43. data/README-json-jruby.markdown +0 -33
  44. data/Rakefile +0 -412
  45. data/TODO +0 -1
  46. data/data/example.json +0 -1
  47. data/data/index.html +0 -38
  48. data/data/prototype.js +0 -4184
  49. data/diagrams/.keep +0 -0
  50. data/install.rb +0 -23
  51. data/java/src/json/ext/ByteListTranscoder.java +0 -167
  52. data/java/src/json/ext/Generator.java +0 -444
  53. data/java/src/json/ext/GeneratorMethods.java +0 -232
  54. data/java/src/json/ext/GeneratorService.java +0 -43
  55. data/java/src/json/ext/GeneratorState.java +0 -543
  56. data/java/src/json/ext/OptionsReader.java +0 -114
  57. data/java/src/json/ext/Parser.java +0 -2645
  58. data/java/src/json/ext/Parser.rl +0 -969
  59. data/java/src/json/ext/ParserService.java +0 -35
  60. data/java/src/json/ext/RuntimeInfo.java +0 -121
  61. data/java/src/json/ext/StringDecoder.java +0 -167
  62. data/java/src/json/ext/StringEncoder.java +0 -106
  63. data/java/src/json/ext/Utils.java +0 -89
  64. data/json-java.gemspec +0 -23
  65. data/json_pure.gemspec +0 -40
  66. data/lib/json/ext/.keep +0 -0
  67. data/tests/fixtures/fail1.json +0 -1
  68. data/tests/fixtures/fail10.json +0 -1
  69. data/tests/fixtures/fail11.json +0 -1
  70. data/tests/fixtures/fail12.json +0 -1
  71. data/tests/fixtures/fail13.json +0 -1
  72. data/tests/fixtures/fail14.json +0 -1
  73. data/tests/fixtures/fail18.json +0 -1
  74. data/tests/fixtures/fail19.json +0 -1
  75. data/tests/fixtures/fail2.json +0 -1
  76. data/tests/fixtures/fail20.json +0 -1
  77. data/tests/fixtures/fail21.json +0 -1
  78. data/tests/fixtures/fail22.json +0 -1
  79. data/tests/fixtures/fail23.json +0 -1
  80. data/tests/fixtures/fail24.json +0 -1
  81. data/tests/fixtures/fail25.json +0 -1
  82. data/tests/fixtures/fail27.json +0 -2
  83. data/tests/fixtures/fail28.json +0 -2
  84. data/tests/fixtures/fail3.json +0 -1
  85. data/tests/fixtures/fail4.json +0 -1
  86. data/tests/fixtures/fail5.json +0 -1
  87. data/tests/fixtures/fail6.json +0 -1
  88. data/tests/fixtures/fail7.json +0 -1
  89. data/tests/fixtures/fail8.json +0 -1
  90. data/tests/fixtures/fail9.json +0 -1
  91. data/tests/fixtures/pass1.json +0 -56
  92. data/tests/fixtures/pass15.json +0 -1
  93. data/tests/fixtures/pass16.json +0 -1
  94. data/tests/fixtures/pass17.json +0 -1
  95. data/tests/fixtures/pass2.json +0 -1
  96. data/tests/fixtures/pass26.json +0 -1
  97. data/tests/fixtures/pass3.json +0 -6
  98. data/tests/setup_variant.rb +0 -11
  99. data/tests/test_json.rb +0 -553
  100. data/tests/test_json_addition.rb +0 -196
  101. data/tests/test_json_encoding.rb +0 -65
  102. data/tests/test_json_fixtures.rb +0 -35
  103. data/tests/test_json_generate.rb +0 -337
  104. data/tests/test_json_generic_object.rb +0 -75
  105. data/tests/test_json_string_matching.rb +0 -39
  106. data/tests/test_json_unicode.rb +0 -72
  107. data/tools/fuzz.rb +0 -139
  108. data/tools/server.rb +0 -62
@@ -1,9 +1,31 @@
1
1
  #include "../fbuffer/fbuffer.h"
2
2
  #include "parser.h"
3
3
 
4
+ #if defined HAVE_RUBY_ENCODING_H
5
+ # define EXC_ENCODING rb_utf8_encoding(),
6
+ # ifndef HAVE_RB_ENC_RAISE
7
+ static void
8
+ enc_raise(rb_encoding *enc, VALUE exc, const char *fmt, ...)
9
+ {
10
+ va_list args;
11
+ VALUE mesg;
12
+
13
+ va_start(args, fmt);
14
+ mesg = rb_enc_vsprintf(enc, fmt, args);
15
+ va_end(args);
16
+
17
+ rb_exc_raise(rb_exc_new3(exc, mesg));
18
+ }
19
+ # define rb_enc_raise enc_raise
20
+ # endif
21
+ #else
22
+ # define EXC_ENCODING /* nothing */
23
+ # define rb_enc_raise rb_raise
24
+ #endif
25
+
4
26
  /* unicode */
5
27
 
6
- static const char digit_values[256] = {
28
+ static const signed char digit_values[256] = {
7
29
  -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1,
8
30
  -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1,
9
31
  -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, -1,
@@ -22,7 +44,7 @@ static const char digit_values[256] = {
22
44
 
23
45
  static UTF32 unescape_unicode(const unsigned char *p)
24
46
  {
25
- char b;
47
+ signed char b;
26
48
  UTF32 result = 0;
27
49
  b = digit_values[p[0]];
28
50
  if (b < 0) return UNI_REPLACEMENT_CHAR;
@@ -65,21 +87,14 @@ static int convert_UTF32_to_UTF8(char *buf, UTF32 ch)
65
87
  return len;
66
88
  }
67
89
 
68
- #ifdef HAVE_RUBY_ENCODING_H
69
- static VALUE CEncoding_ASCII_8BIT, CEncoding_UTF_8, CEncoding_UTF_16BE,
70
- CEncoding_UTF_16LE, CEncoding_UTF_32BE, CEncoding_UTF_32LE;
71
- static ID i_encoding, i_encode;
72
- #else
73
- static ID i_iconv;
74
- #endif
75
-
76
90
  static VALUE mJSON, mExt, cParser, eParserError, eNestingError;
77
91
  static VALUE CNaN, CInfinity, CMinusInfinity;
78
92
 
79
93
  static ID i_json_creatable_p, i_json_create, i_create_id, i_create_additions,
80
- i_chr, i_max_nesting, i_allow_nan, i_symbolize_names, i_quirks_mode,
81
- i_object_class, i_array_class, i_key_p, i_deep_const_get, i_match,
82
- i_match_string, i_aset, i_aref, i_leftshift;
94
+ i_chr, i_max_nesting, i_allow_nan, i_symbolize_names,
95
+ i_object_class, i_array_class, i_decimal_class, i_key_p,
96
+ i_deep_const_get, i_match, i_match_string, i_aset, i_aref,
97
+ i_leftshift, i_new, i_try_convert, i_freeze, i_uminus;
83
98
 
84
99
  %%{
85
100
  machine JSON_common;
@@ -117,11 +132,12 @@ static ID i_json_creatable_p, i_json_create, i_create_id, i_create_additions,
117
132
 
118
133
  action parse_value {
119
134
  VALUE v = Qnil;
120
- char *np = JSON_parse_value(json, fpc, pe, &v);
135
+ char *np = JSON_parse_value(json, fpc, pe, &v, current_nesting);
121
136
  if (np == NULL) {
122
137
  fhold; fbreak;
123
138
  } else {
124
139
  if (NIL_P(json->object_class)) {
140
+ OBJ_FREEZE(last_name);
125
141
  rb_hash_aset(*result, last_name, v);
126
142
  } else {
127
143
  rb_funcall(*result, i_aset, 2, last_name, v);
@@ -150,14 +166,14 @@ static ID i_json_creatable_p, i_json_create, i_create_id, i_create_additions,
150
166
  ) @exit;
151
167
  }%%
152
168
 
153
- static char *JSON_parse_object(JSON_Parser *json, char *p, char *pe, VALUE *result)
169
+ static char *JSON_parse_object(JSON_Parser *json, char *p, char *pe, VALUE *result, int current_nesting)
154
170
  {
155
171
  int cs = EVIL;
156
172
  VALUE last_name = Qnil;
157
173
  VALUE object_class = json->object_class;
158
174
 
159
- if (json->max_nesting && json->current_nesting > json->max_nesting) {
160
- rb_raise(eNestingError, "nesting of %d is too deep", json->current_nesting);
175
+ if (json->max_nesting && current_nesting > json->max_nesting) {
176
+ rb_raise(eNestingError, "nesting of %d is too deep", current_nesting);
161
177
  }
162
178
 
163
179
  *result = NIL_P(object_class) ? rb_hash_new() : rb_class_new_instance(0, 0, object_class);
@@ -206,14 +222,14 @@ static char *JSON_parse_object(JSON_Parser *json, char *p, char *pe, VALUE *resu
206
222
  if (json->allow_nan) {
207
223
  *result = CNaN;
208
224
  } else {
209
- rb_raise(eParserError, "%u: unexpected token at '%s'", __LINE__, p - 2);
225
+ rb_enc_raise(EXC_ENCODING eParserError, "%u: unexpected token at '%s'", __LINE__, p - 2);
210
226
  }
211
227
  }
212
228
  action parse_infinity {
213
229
  if (json->allow_nan) {
214
230
  *result = CInfinity;
215
231
  } else {
216
- rb_raise(eParserError, "%u: unexpected token at '%s'", __LINE__, p - 8);
232
+ rb_enc_raise(EXC_ENCODING eParserError, "%u: unexpected token at '%s'", __LINE__, p - 8);
217
233
  }
218
234
  }
219
235
  action parse_string {
@@ -223,13 +239,13 @@ static char *JSON_parse_object(JSON_Parser *json, char *p, char *pe, VALUE *resu
223
239
 
224
240
  action parse_number {
225
241
  char *np;
226
- if(pe > fpc + 9 - json->quirks_mode && !strncmp(MinusInfinity, fpc, 9)) {
242
+ if(pe > fpc + 8 && !strncmp(MinusInfinity, fpc, 9)) {
227
243
  if (json->allow_nan) {
228
244
  *result = CMinusInfinity;
229
245
  fexec p + 10;
230
246
  fhold; fbreak;
231
247
  } else {
232
- rb_raise(eParserError, "%u: unexpected token at '%s'", __LINE__, p);
248
+ rb_enc_raise(EXC_ENCODING eParserError, "%u: unexpected token at '%s'", __LINE__, p);
233
249
  }
234
250
  }
235
251
  np = JSON_parse_float(json, fpc, pe, result);
@@ -241,23 +257,19 @@ static char *JSON_parse_object(JSON_Parser *json, char *p, char *pe, VALUE *resu
241
257
 
242
258
  action parse_array {
243
259
  char *np;
244
- json->current_nesting++;
245
- np = JSON_parse_array(json, fpc, pe, result);
246
- json->current_nesting--;
260
+ np = JSON_parse_array(json, fpc, pe, result, current_nesting + 1);
247
261
  if (np == NULL) { fhold; fbreak; } else fexec np;
248
262
  }
249
263
 
250
264
  action parse_object {
251
265
  char *np;
252
- json->current_nesting++;
253
- np = JSON_parse_object(json, fpc, pe, result);
254
- json->current_nesting--;
266
+ np = JSON_parse_object(json, fpc, pe, result, current_nesting + 1);
255
267
  if (np == NULL) { fhold; fbreak; } else fexec np;
256
268
  }
257
269
 
258
270
  action exit { fhold; fbreak; }
259
271
 
260
- main := (
272
+ main := ignore* (
261
273
  Vnull @parse_null |
262
274
  Vfalse @parse_false |
263
275
  Vtrue @parse_true |
@@ -267,16 +279,20 @@ main := (
267
279
  begin_string >parse_string |
268
280
  begin_array >parse_array |
269
281
  begin_object >parse_object
270
- ) %*exit;
282
+ ) ignore* %*exit;
271
283
  }%%
272
284
 
273
- static char *JSON_parse_value(JSON_Parser *json, char *p, char *pe, VALUE *result)
285
+ static char *JSON_parse_value(JSON_Parser *json, char *p, char *pe, VALUE *result, int current_nesting)
274
286
  {
275
287
  int cs = EVIL;
276
288
 
277
289
  %% write init;
278
290
  %% write exec;
279
291
 
292
+ if (json->freeze) {
293
+ OBJ_FREEZE(*result);
294
+ }
295
+
280
296
  if (cs >= JSON_value_first_final) {
281
297
  return p;
282
298
  } else {
@@ -337,11 +353,46 @@ static char *JSON_parse_float(JSON_Parser *json, char *p, char *pe, VALUE *resul
337
353
  %% write exec;
338
354
 
339
355
  if (cs >= JSON_float_first_final) {
356
+ VALUE mod = Qnil;
357
+ ID method_id = 0;
358
+ if (rb_respond_to(json->decimal_class, i_try_convert)) {
359
+ mod = json->decimal_class;
360
+ method_id = i_try_convert;
361
+ } else if (rb_respond_to(json->decimal_class, i_new)) {
362
+ mod = json->decimal_class;
363
+ method_id = i_new;
364
+ } else if (RB_TYPE_P(json->decimal_class, T_CLASS)) {
365
+ VALUE name = rb_class_name(json->decimal_class);
366
+ const char *name_cstr = RSTRING_PTR(name);
367
+ const char *last_colon = strrchr(name_cstr, ':');
368
+ if (last_colon) {
369
+ const char *mod_path_end = last_colon - 1;
370
+ VALUE mod_path = rb_str_substr(name, 0, mod_path_end - name_cstr);
371
+ mod = rb_path_to_class(mod_path);
372
+
373
+ const char *method_name_beg = last_colon + 1;
374
+ long before_len = method_name_beg - name_cstr;
375
+ long len = RSTRING_LEN(name) - before_len;
376
+ VALUE method_name = rb_str_substr(name, before_len, len);
377
+ method_id = SYM2ID(rb_str_intern(method_name));
378
+ } else {
379
+ mod = rb_mKernel;
380
+ method_id = SYM2ID(rb_str_intern(name));
381
+ }
382
+ }
383
+
340
384
  long len = p - json->memo;
341
385
  fbuffer_clear(json->fbuffer);
342
386
  fbuffer_append(json->fbuffer, json->memo, len);
343
387
  fbuffer_append_char(json->fbuffer, '\0');
344
- *result = rb_float_new(rb_cstr_to_dbl(FBUFFER_PTR(json->fbuffer), 1));
388
+
389
+ if (method_id) {
390
+ VALUE text = rb_str_new2(FBUFFER_PTR(json->fbuffer));
391
+ *result = rb_funcallv(mod, method_id, 1, &text);
392
+ } else {
393
+ *result = DBL2NUM(rb_cstr_to_dbl(FBUFFER_PTR(json->fbuffer), 1));
394
+ }
395
+
345
396
  return p + 1;
346
397
  } else {
347
398
  return NULL;
@@ -357,7 +408,7 @@ static char *JSON_parse_float(JSON_Parser *json, char *p, char *pe, VALUE *resul
357
408
 
358
409
  action parse_value {
359
410
  VALUE v = Qnil;
360
- char *np = JSON_parse_value(json, fpc, pe, &v);
411
+ char *np = JSON_parse_value(json, fpc, pe, &v, current_nesting);
361
412
  if (np == NULL) {
362
413
  fhold; fbreak;
363
414
  } else {
@@ -380,13 +431,13 @@ static char *JSON_parse_float(JSON_Parser *json, char *p, char *pe, VALUE *resul
380
431
  end_array @exit;
381
432
  }%%
382
433
 
383
- static char *JSON_parse_array(JSON_Parser *json, char *p, char *pe, VALUE *result)
434
+ static char *JSON_parse_array(JSON_Parser *json, char *p, char *pe, VALUE *result, int current_nesting)
384
435
  {
385
436
  int cs = EVIL;
386
437
  VALUE array_class = json->array_class;
387
438
 
388
- if (json->max_nesting && json->current_nesting > json->max_nesting) {
389
- rb_raise(eNestingError, "nesting of %d is too deep", json->current_nesting);
439
+ if (json->max_nesting && current_nesting > json->max_nesting) {
440
+ rb_raise(eNestingError, "nesting of %d is too deep", current_nesting);
390
441
  }
391
442
  *result = NIL_P(array_class) ? rb_ary_new() : rb_class_new_instance(0, 0, array_class);
392
443
 
@@ -396,22 +447,34 @@ static char *JSON_parse_array(JSON_Parser *json, char *p, char *pe, VALUE *resul
396
447
  if(cs >= JSON_array_first_final) {
397
448
  return p + 1;
398
449
  } else {
399
- rb_raise(eParserError, "%u: unexpected token at '%s'", __LINE__, p);
450
+ rb_enc_raise(EXC_ENCODING eParserError, "%u: unexpected token at '%s'", __LINE__, p);
400
451
  return NULL;
401
452
  }
402
453
  }
403
454
 
404
- static VALUE json_string_unescape(VALUE result, char *string, char *stringEnd)
455
+ static const size_t MAX_STACK_BUFFER_SIZE = 128;
456
+ static VALUE json_string_unescape(char *string, char *stringEnd, int intern, int symbolize)
405
457
  {
406
- char *p = string, *pe = string, *unescape;
458
+ VALUE result = Qnil;
459
+ size_t bufferSize = stringEnd - string;
460
+ char *p = string, *pe = string, *unescape, *bufferStart, *buffer;
407
461
  int unescape_len;
408
462
  char buf[4];
409
463
 
464
+ if (bufferSize > MAX_STACK_BUFFER_SIZE) {
465
+ bufferStart = buffer = ALLOC_N(char, bufferSize);
466
+ } else {
467
+ bufferStart = buffer = ALLOCA_N(char, bufferSize);
468
+ }
469
+
410
470
  while (pe < stringEnd) {
411
471
  if (*pe == '\\') {
412
472
  unescape = (char *) "?";
413
473
  unescape_len = 1;
414
- if (pe > p) rb_str_buf_cat(result, p, pe - p);
474
+ if (pe > p) {
475
+ MEMCPY(buffer, p, char, pe - p);
476
+ buffer += pe - p;
477
+ }
415
478
  switch (*++pe) {
416
479
  case 'n':
417
480
  unescape = (char *) "\n";
@@ -436,13 +499,27 @@ static VALUE json_string_unescape(VALUE result, char *string, char *stringEnd)
436
499
  break;
437
500
  case 'u':
438
501
  if (pe > stringEnd - 4) {
439
- return Qnil;
502
+ if (bufferSize > MAX_STACK_BUFFER_SIZE) {
503
+ free(bufferStart);
504
+ }
505
+ rb_enc_raise(
506
+ EXC_ENCODING eParserError,
507
+ "%u: incomplete unicode character escape sequence at '%s'", __LINE__, p
508
+ );
440
509
  } else {
441
510
  UTF32 ch = unescape_unicode((unsigned char *) ++pe);
442
511
  pe += 3;
443
512
  if (UNI_SUR_HIGH_START == (ch & 0xFC00)) {
444
513
  pe++;
445
- if (pe > stringEnd - 6) return Qnil;
514
+ if (pe > stringEnd - 6) {
515
+ if (bufferSize > MAX_STACK_BUFFER_SIZE) {
516
+ free(bufferStart);
517
+ }
518
+ rb_enc_raise(
519
+ EXC_ENCODING eParserError,
520
+ "%u: incomplete surrogate pair at '%s'", __LINE__, p
521
+ );
522
+ }
446
523
  if (pe[0] == '\\' && pe[1] == 'u') {
447
524
  UTF32 sur = unescape_unicode((unsigned char *) pe + 2);
448
525
  ch = (((ch & 0x3F) << 10) | ((((ch >> 6) & 0xF) + 1) << 16)
@@ -461,13 +538,55 @@ static VALUE json_string_unescape(VALUE result, char *string, char *stringEnd)
461
538
  p = pe;
462
539
  continue;
463
540
  }
464
- rb_str_buf_cat(result, unescape, unescape_len);
541
+ MEMCPY(buffer, unescape, char, unescape_len);
542
+ buffer += unescape_len;
465
543
  p = ++pe;
466
544
  } else {
467
545
  pe++;
468
546
  }
469
547
  }
470
- rb_str_buf_cat(result, p, pe - p);
548
+
549
+ if (pe > p) {
550
+ MEMCPY(buffer, p, char, pe - p);
551
+ buffer += pe - p;
552
+ }
553
+
554
+ # ifdef HAVE_RB_ENC_INTERNED_STR
555
+ if (intern) {
556
+ result = rb_enc_interned_str(bufferStart, (long)(buffer - bufferStart), rb_utf8_encoding());
557
+ } else {
558
+ result = rb_utf8_str_new(bufferStart, (long)(buffer - bufferStart));
559
+ }
560
+ if (bufferSize > MAX_STACK_BUFFER_SIZE) {
561
+ free(bufferStart);
562
+ }
563
+ # else
564
+ result = rb_utf8_str_new(bufferStart, (long)(buffer - bufferStart));
565
+
566
+ if (bufferSize > MAX_STACK_BUFFER_SIZE) {
567
+ free(bufferStart);
568
+ }
569
+
570
+ if (intern) {
571
+ # if STR_UMINUS_DEDUPE_FROZEN
572
+ // Starting from MRI 2.8 it is preferable to freeze the string
573
+ // before deduplication so that it can be interned directly
574
+ // otherwise it would be duplicated first which is wasteful.
575
+ result = rb_funcall(rb_str_freeze(result), i_uminus, 0);
576
+ # elif STR_UMINUS_DEDUPE
577
+ // MRI 2.5 and older do not deduplicate strings that are already
578
+ // frozen.
579
+ result = rb_funcall(result, i_uminus, 0);
580
+ # else
581
+ result = rb_str_freeze(result);
582
+ # endif
583
+ }
584
+ # endif
585
+
586
+ if (symbolize) {
587
+ result = rb_str_intern(result);
588
+ }
589
+
471
590
  return result;
472
591
  }
473
592
 
@@ -478,12 +597,11 @@ static VALUE json_string_unescape(VALUE result, char *string, char *stringEnd)
478
597
  write data;
479
598
 
480
599
  action parse_string {
481
- *result = json_string_unescape(*result, json->memo + 1, p);
600
+ *result = json_string_unescape(json->memo + 1, p, json->parsing_name || json-> freeze, json->parsing_name && json->symbolize_names);
482
601
  if (NIL_P(*result)) {
483
602
  fhold;
484
603
  fbreak;
485
604
  } else {
486
- FORCE_UTF8(*result);
487
605
  fexec p + 1;
488
606
  }
489
607
  }
@@ -510,7 +628,6 @@ static char *JSON_parse_string(JSON_Parser *json, char *p, char *pe, VALUE *resu
510
628
  int cs = EVIL;
511
629
  VALUE match_string;
512
630
 
513
- *result = rb_str_buf_new(0);
514
631
  %% write init;
515
632
  json->memo = p;
516
633
  %% write exec;
@@ -526,9 +643,6 @@ static char *JSON_parse_string(JSON_Parser *json, char *p, char *pe, VALUE *resu
526
643
  }
527
644
  }
528
645
 
529
- if (json->symbolize_names && json->parsing_name) {
530
- *result = rb_str_intern(*result);
531
- }
532
646
  if (cs >= JSON_string_first_final) {
533
647
  return p + 1;
534
648
  } else {
@@ -550,41 +664,16 @@ static char *JSON_parse_string(JSON_Parser *json, char *p, char *pe, VALUE *resu
550
664
 
551
665
  static VALUE convert_encoding(VALUE source)
552
666
  {
553
- char *ptr = RSTRING_PTR(source);
554
- long len = RSTRING_LEN(source);
555
- if (len < 2) {
556
- rb_raise(eParserError, "A JSON text must at least contain two octets!");
557
- }
558
667
  #ifdef HAVE_RUBY_ENCODING_H
559
- {
560
- VALUE encoding = rb_funcall(source, i_encoding, 0);
561
- if (encoding == CEncoding_ASCII_8BIT) {
562
- if (len >= 4 && ptr[0] == 0 && ptr[1] == 0 && ptr[2] == 0) {
563
- source = rb_funcall(source, i_encode, 2, CEncoding_UTF_8, CEncoding_UTF_32BE);
564
- } else if (len >= 4 && ptr[0] == 0 && ptr[2] == 0) {
565
- source = rb_funcall(source, i_encode, 2, CEncoding_UTF_8, CEncoding_UTF_16BE);
566
- } else if (len >= 4 && ptr[1] == 0 && ptr[2] == 0 && ptr[3] == 0) {
567
- source = rb_funcall(source, i_encode, 2, CEncoding_UTF_8, CEncoding_UTF_32LE);
568
- } else if (len >= 4 && ptr[1] == 0 && ptr[3] == 0) {
569
- source = rb_funcall(source, i_encode, 2, CEncoding_UTF_8, CEncoding_UTF_16LE);
570
- } else {
571
- source = rb_str_dup(source);
572
- FORCE_UTF8(source);
573
- }
574
- } else {
575
- source = rb_funcall(source, i_encode, 1, CEncoding_UTF_8);
576
- }
577
- }
578
- #else
579
- if (len >= 4 && ptr[0] == 0 && ptr[1] == 0 && ptr[2] == 0) {
580
- source = rb_funcall(mJSON, i_iconv, 3, rb_str_new2("utf-8"), rb_str_new2("utf-32be"), source);
581
- } else if (len >= 4 && ptr[0] == 0 && ptr[2] == 0) {
582
- source = rb_funcall(mJSON, i_iconv, 3, rb_str_new2("utf-8"), rb_str_new2("utf-16be"), source);
583
- } else if (len >= 4 && ptr[1] == 0 && ptr[2] == 0 && ptr[3] == 0) {
584
- source = rb_funcall(mJSON, i_iconv, 3, rb_str_new2("utf-8"), rb_str_new2("utf-32le"), source);
585
- } else if (len >= 4 && ptr[1] == 0 && ptr[3] == 0) {
586
- source = rb_funcall(mJSON, i_iconv, 3, rb_str_new2("utf-8"), rb_str_new2("utf-16le"), source);
668
+ rb_encoding *enc = rb_enc_get(source);
669
+ if (enc == rb_ascii8bit_encoding()) {
670
+ if (OBJ_FROZEN(source)) {
671
+ source = rb_str_dup(source);
587
672
  }
673
+ FORCE_UTF8(source);
674
+ } else {
675
+ source = rb_str_conv_enc(source, rb_enc_get(source), rb_utf8_encoding());
676
+ }
588
677
  #endif
589
678
  return source;
590
679
  }
@@ -607,8 +696,9 @@ static VALUE convert_encoding(VALUE source)
607
696
  * defiance of RFC 4627 to be parsed by the Parser. This option defaults to
608
697
  * false.
609
698
  * * *symbolize_names*: If set to true, returns symbols for the names
610
- * (keys) in a JSON object. Otherwise strings are returned, which is also
611
- * the default.
699
+ * (keys) in a JSON object. Otherwise strings are returned, which is
700
+ * also the default. It's not possible to use this option in
701
+ * conjunction with the *create_additions* option.
612
702
  * * *create_additions*: If set to false, the Parser doesn't create
613
703
  * additions even if a matching class and create_id was found. This option
614
704
  * defaults to false.
@@ -623,12 +713,18 @@ static VALUE cParser_initialize(int argc, VALUE *argv, VALUE self)
623
713
  if (json->Vsource) {
624
714
  rb_raise(rb_eTypeError, "already initialized instance");
625
715
  }
716
+ #ifdef HAVE_RB_SCAN_ARGS_OPTIONAL_HASH
717
+ rb_scan_args(argc, argv, "1:", &source, &opts);
718
+ #else
626
719
  rb_scan_args(argc, argv, "11", &source, &opts);
720
+ #endif
627
721
  if (!NIL_P(opts)) {
722
+ #ifndef HAVE_RB_SCAN_ARGS_OPTIONAL_HASH
628
723
  opts = rb_convert_type(opts, T_HASH, "Hash", "to_hash");
629
724
  if (NIL_P(opts)) {
630
725
  rb_raise(rb_eArgError, "opts needs to be like a hash");
631
726
  } else {
727
+ #endif
632
728
  VALUE tmp = ID2SYM(i_max_nesting);
633
729
  if (option_given_p(opts, tmp)) {
634
730
  VALUE max_nesting = rb_hash_aref(opts, tmp);
@@ -653,12 +749,11 @@ static VALUE cParser_initialize(int argc, VALUE *argv, VALUE self)
653
749
  } else {
654
750
  json->symbolize_names = 0;
655
751
  }
656
- tmp = ID2SYM(i_quirks_mode);
752
+ tmp = ID2SYM(i_freeze);
657
753
  if (option_given_p(opts, tmp)) {
658
- VALUE quirks_mode = rb_hash_aref(opts, tmp);
659
- json->quirks_mode = RTEST(quirks_mode) ? 1 : 0;
754
+ json->freeze = RTEST(rb_hash_aref(opts, tmp)) ? 1 : 0;
660
755
  } else {
661
- json->quirks_mode = 0;
756
+ json->freeze = 0;
662
757
  }
663
758
  tmp = ID2SYM(i_create_additions);
664
759
  if (option_given_p(opts, tmp)) {
@@ -666,6 +761,11 @@ static VALUE cParser_initialize(int argc, VALUE *argv, VALUE self)
666
761
  } else {
667
762
  json->create_additions = 0;
668
763
  }
764
+ if (json->symbolize_names && json->create_additions) {
765
+ rb_raise(rb_eArgError,
766
+ "options :symbolize_names and :create_additions cannot be "
767
+ " used in conjunction");
768
+ }
669
769
  tmp = ID2SYM(i_create_id);
670
770
  if (option_given_p(opts, tmp)) {
671
771
  json->create_id = rb_hash_aref(opts, tmp);
@@ -684,6 +784,12 @@ static VALUE cParser_initialize(int argc, VALUE *argv, VALUE self)
684
784
  } else {
685
785
  json->array_class = Qnil;
686
786
  }
787
+ tmp = ID2SYM(i_decimal_class);
788
+ if (option_given_p(opts, tmp)) {
789
+ json->decimal_class = rb_hash_aref(opts, tmp);
790
+ } else {
791
+ json->decimal_class = Qnil;
792
+ }
687
793
  tmp = ID2SYM(i_match_string);
688
794
  if (option_given_p(opts, tmp)) {
689
795
  VALUE match_string = rb_hash_aref(opts, tmp);
@@ -691,20 +797,19 @@ static VALUE cParser_initialize(int argc, VALUE *argv, VALUE self)
691
797
  } else {
692
798
  json->match_string = Qnil;
693
799
  }
800
+ #ifndef HAVE_RB_SCAN_ARGS_OPTIONAL_HASH
694
801
  }
802
+ #endif
695
803
  } else {
696
804
  json->max_nesting = 100;
697
805
  json->allow_nan = 0;
698
- json->create_additions = 1;
806
+ json->create_additions = 0;
699
807
  json->create_id = rb_funcall(mJSON, i_create_id, 0);
700
808
  json->object_class = Qnil;
701
809
  json->array_class = Qnil;
810
+ json->decimal_class = Qnil;
702
811
  }
703
- source = rb_convert_type(source, T_STRING, "String", "to_str");
704
- if (!json->quirks_mode) {
705
- source = convert_encoding(StringValue(source));
706
- }
707
- json->current_nesting = 0;
812
+ source = convert_encoding(StringValue(source));
708
813
  StringValue(source);
709
814
  json->len = RSTRING_LEN(source);
710
815
  json->source = RSTRING_PTR(source);;
@@ -719,56 +824,8 @@ static VALUE cParser_initialize(int argc, VALUE *argv, VALUE self)
719
824
 
720
825
  include JSON_common;
721
826
 
722
- action parse_object {
723
- char *np;
724
- json->current_nesting = 1;
725
- np = JSON_parse_object(json, fpc, pe, &result);
726
- if (np == NULL) { fhold; fbreak; } else fexec np;
727
- }
728
-
729
- action parse_array {
730
- char *np;
731
- json->current_nesting = 1;
732
- np = JSON_parse_array(json, fpc, pe, &result);
733
- if (np == NULL) { fhold; fbreak; } else fexec np;
734
- }
735
-
736
- main := ignore* (
737
- begin_object >parse_object |
738
- begin_array >parse_array
739
- ) ignore*;
740
- }%%
741
-
742
- static VALUE cParser_parse_strict(VALUE self)
743
- {
744
- char *p, *pe;
745
- int cs = EVIL;
746
- VALUE result = Qnil;
747
- GET_PARSER;
748
-
749
- %% write init;
750
- p = json->source;
751
- pe = p + json->len;
752
- %% write exec;
753
-
754
- if (cs >= JSON_first_final && p == pe) {
755
- return result;
756
- } else {
757
- rb_raise(eParserError, "%u: unexpected token at '%s'", __LINE__, p);
758
- return Qnil;
759
- }
760
- }
761
-
762
-
763
- %%{
764
- machine JSON_quirks_mode;
765
-
766
- write data;
767
-
768
- include JSON_common;
769
-
770
827
  action parse_value {
771
- char *np = JSON_parse_value(json, fpc, pe, &result);
828
+ char *np = JSON_parse_value(json, fpc, pe, &result, 0);
772
829
  if (np == NULL) { fhold; fbreak; } else fexec np;
773
830
  }
774
831
 
@@ -777,26 +834,6 @@ static VALUE cParser_parse_strict(VALUE self)
777
834
  ) ignore*;
778
835
  }%%
779
836
 
780
- static VALUE cParser_parse_quirks_mode(VALUE self)
781
- {
782
- char *p, *pe;
783
- int cs = EVIL;
784
- VALUE result = Qnil;
785
- GET_PARSER;
786
-
787
- %% write init;
788
- p = json->source;
789
- pe = p + json->len;
790
- %% write exec;
791
-
792
- if (cs >= JSON_quirks_mode_first_final && p == pe) {
793
- return result;
794
- } else {
795
- rb_raise(eParserError, "%u: unexpected token at '%s'", __LINE__, p);
796
- return Qnil;
797
- }
798
- }
799
-
800
837
  /*
801
838
  * call-seq: parse()
802
839
  *
@@ -805,12 +842,21 @@ static VALUE cParser_parse_quirks_mode(VALUE self)
805
842
  */
806
843
  static VALUE cParser_parse(VALUE self)
807
844
  {
845
+ char *p, *pe;
846
+ int cs = EVIL;
847
+ VALUE result = Qnil;
808
848
  GET_PARSER;
809
849
 
810
- if (json->quirks_mode) {
811
- return cParser_parse_quirks_mode(self);
850
+ %% write init;
851
+ p = json->source;
852
+ pe = p + json->len;
853
+ %% write exec;
854
+
855
+ if (cs >= JSON_first_final && p == pe) {
856
+ return result;
812
857
  } else {
813
- return cParser_parse_strict(self);
858
+ rb_enc_raise(EXC_ENCODING eParserError, "%u: unexpected token at '%s'", __LINE__, p);
859
+ return Qnil;
814
860
  }
815
861
  }
816
862
 
@@ -821,6 +867,7 @@ static void JSON_mark(void *ptr)
821
867
  rb_gc_mark_maybe(json->create_id);
822
868
  rb_gc_mark_maybe(json->object_class);
823
869
  rb_gc_mark_maybe(json->array_class);
870
+ rb_gc_mark_maybe(json->decimal_class);
824
871
  rb_gc_mark_maybe(json->match_string);
825
872
  }
826
873
 
@@ -868,35 +915,34 @@ static VALUE cParser_source(VALUE self)
868
915
  return rb_str_dup(json->Vsource);
869
916
  }
870
917
 
871
- /*
872
- * call-seq: quirks_mode?()
873
- *
874
- * Returns a true, if this parser is in quirks_mode, false otherwise.
875
- */
876
- static VALUE cParser_quirks_mode_p(VALUE self)
877
- {
878
- GET_PARSER;
879
- return json->quirks_mode ? Qtrue : Qfalse;
880
- }
881
-
882
-
883
918
  void Init_parser(void)
884
919
  {
920
+ #ifdef HAVE_RB_EXT_RACTOR_SAFE
921
+ rb_ext_ractor_safe(true);
922
+ #endif
923
+
924
+ #undef rb_intern
885
925
  rb_require("json/common");
886
926
  mJSON = rb_define_module("JSON");
887
927
  mExt = rb_define_module_under(mJSON, "Ext");
888
928
  cParser = rb_define_class_under(mExt, "Parser", rb_cObject);
889
929
  eParserError = rb_path2class("JSON::ParserError");
890
930
  eNestingError = rb_path2class("JSON::NestingError");
931
+ rb_gc_register_mark_object(eParserError);
932
+ rb_gc_register_mark_object(eNestingError);
891
933
  rb_define_alloc_func(cParser, cJSON_parser_s_allocate);
892
934
  rb_define_method(cParser, "initialize", cParser_initialize, -1);
893
935
  rb_define_method(cParser, "parse", cParser_parse, 0);
894
936
  rb_define_method(cParser, "source", cParser_source, 0);
895
- rb_define_method(cParser, "quirks_mode?", cParser_quirks_mode_p, 0);
896
937
 
897
938
  CNaN = rb_const_get(mJSON, rb_intern("NaN"));
939
+ rb_gc_register_mark_object(CNaN);
940
+
898
941
  CInfinity = rb_const_get(mJSON, rb_intern("Infinity"));
942
+ rb_gc_register_mark_object(CInfinity);
943
+
899
944
  CMinusInfinity = rb_const_get(mJSON, rb_intern("MinusInfinity"));
945
+ rb_gc_register_mark_object(CMinusInfinity);
900
946
 
901
947
  i_json_creatable_p = rb_intern("json_creatable?");
902
948
  i_json_create = rb_intern("json_create");
@@ -906,9 +952,9 @@ void Init_parser(void)
906
952
  i_max_nesting = rb_intern("max_nesting");
907
953
  i_allow_nan = rb_intern("allow_nan");
908
954
  i_symbolize_names = rb_intern("symbolize_names");
909
- i_quirks_mode = rb_intern("quirks_mode");
910
955
  i_object_class = rb_intern("object_class");
911
956
  i_array_class = rb_intern("array_class");
957
+ i_decimal_class = rb_intern("decimal_class");
912
958
  i_match = rb_intern("match");
913
959
  i_match_string = rb_intern("match_string");
914
960
  i_key_p = rb_intern("key?");
@@ -916,18 +962,10 @@ void Init_parser(void)
916
962
  i_aset = rb_intern("[]=");
917
963
  i_aref = rb_intern("[]");
918
964
  i_leftshift = rb_intern("<<");
919
- #ifdef HAVE_RUBY_ENCODING_H
920
- CEncoding_UTF_8 = rb_funcall(rb_path2class("Encoding"), rb_intern("find"), 1, rb_str_new2("utf-8"));
921
- CEncoding_UTF_16BE = rb_funcall(rb_path2class("Encoding"), rb_intern("find"), 1, rb_str_new2("utf-16be"));
922
- CEncoding_UTF_16LE = rb_funcall(rb_path2class("Encoding"), rb_intern("find"), 1, rb_str_new2("utf-16le"));
923
- CEncoding_UTF_32BE = rb_funcall(rb_path2class("Encoding"), rb_intern("find"), 1, rb_str_new2("utf-32be"));
924
- CEncoding_UTF_32LE = rb_funcall(rb_path2class("Encoding"), rb_intern("find"), 1, rb_str_new2("utf-32le"));
925
- CEncoding_ASCII_8BIT = rb_funcall(rb_path2class("Encoding"), rb_intern("find"), 1, rb_str_new2("ascii-8bit"));
926
- i_encoding = rb_intern("encoding");
927
- i_encode = rb_intern("encode");
928
- #else
929
- i_iconv = rb_intern("iconv");
930
- #endif
965
+ i_new = rb_intern("new");
966
+ i_try_convert = rb_intern("try_convert");
967
+ i_freeze = rb_intern("freeze");
968
+ i_uminus = rb_intern("-@");
931
969
  }
932
970
 
933
971
  /*