ViewVC Help
View File | Revision Log | Show Annotations | Download File
/cvs/JSON-XS/XS.xs
(Generate patch)

Comparing JSON-XS/XS.xs (file contents):
Revision 1.56 by root, Thu Jul 26 11:33:35 2007 UTC vs.
Revision 1.72 by root, Wed Mar 19 04:08:22 2008 UTC

… …
27#define F_SPACE_AFTER 0x00000040UL 27#define F_SPACE_AFTER 0x00000040UL
28#define F_ALLOW_NONREF 0x00000100UL 28#define F_ALLOW_NONREF 0x00000100UL
29#define F_SHRINK 0x00000200UL 29#define F_SHRINK 0x00000200UL
30#define F_ALLOW_BLESSED 0x00000400UL 30#define F_ALLOW_BLESSED 0x00000400UL
31#define F_CONV_BLESSED 0x00000800UL 31#define F_CONV_BLESSED 0x00000800UL
32#define F_RELAXED 0x00001000UL
33
32#define F_MAXDEPTH 0xf8000000UL 34#define F_MAXDEPTH 0xf8000000UL
33#define S_MAXDEPTH 27 35#define S_MAXDEPTH 27
34#define F_MAXSIZE 0x01f00000UL 36#define F_MAXSIZE 0x01f00000UL
35#define S_MAXSIZE 20 37#define S_MAXSIZE 20
36#define F_HOOK 0x00080000UL // some hooks exist, so slow-path processing 38#define F_HOOK 0x00080000UL // some hooks exist, so slow-path processing
… …
48 50
49#define SB do { 51#define SB do {
50#define SE } while (0) 52#define SE } while (0)
51 53
52#if __GNUC__ >= 3 54#if __GNUC__ >= 3
53# define expect(expr,value) __builtin_expect ((expr),(value)) 55# define expect(expr,value) __builtin_expect ((expr), (value))
54# define inline inline 56# define INLINE static inline
55#else 57#else
56# define expect(expr,value) (expr) 58# define expect(expr,value) (expr)
57# define inline static 59# define INLINE static
58#endif 60#endif
59 61
60#define expect_false(expr) expect ((expr) != 0, 0) 62#define expect_false(expr) expect ((expr) != 0, 0)
61#define expect_true(expr) expect ((expr) != 0, 1) 63#define expect_true(expr) expect ((expr) != 0, 1)
64
65#define IN_RANGE_INC(type,val,beg,end) \
66 ((unsigned type)((unsigned type)(val) - (unsigned type)(beg)) \
67 <= (unsigned type)((unsigned type)(end) - (unsigned type)(beg)))
68
69#ifdef USE_ITHREADS
70# define JSON_SLOW 1
71# define JSON_STASH (json_stash ? json_stash : gv_stashpv ("JSON::XS", 1))
72#else
73# define JSON_SLOW 0
74# define JSON_STASH json_stash
75#endif
62 76
63static HV *json_stash, *json_boolean_stash; // JSON::XS:: 77static HV *json_stash, *json_boolean_stash; // JSON::XS::
64static SV *json_true, *json_false; 78static SV *json_true, *json_false;
65 79
66typedef struct { 80typedef struct {
… …
70} JSON; 84} JSON;
71 85
72///////////////////////////////////////////////////////////////////////////// 86/////////////////////////////////////////////////////////////////////////////
73// utility functions 87// utility functions
74 88
75inline void 89INLINE void
76shrink (SV *sv) 90shrink (SV *sv)
77{ 91{
78 sv_utf8_downgrade (sv, 1); 92 sv_utf8_downgrade (sv, 1);
79 if (SvLEN (sv) > SvCUR (sv) + 1) 93 if (SvLEN (sv) > SvCUR (sv) + 1)
80 { 94 {
… …
89// decode an utf-8 character and return it, or (UV)-1 in 103// decode an utf-8 character and return it, or (UV)-1 in
90// case of an error. 104// case of an error.
91// we special-case "safe" characters from U+80 .. U+7FF, 105// we special-case "safe" characters from U+80 .. U+7FF,
92// but use the very good perl function to parse anything else. 106// but use the very good perl function to parse anything else.
93// note that we never call this function for a ascii codepoints 107// note that we never call this function for a ascii codepoints
94inline UV 108INLINE UV
95decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen) 109decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen)
96{ 110{
97 if (expect_false (s[0] > 0xdf || s[0] < 0xc2)) 111 if (expect_true (len >= 2
98 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY); 112 && IN_RANGE_INC (char, s[0], 0xc2, 0xdf)
99 else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf) 113 && IN_RANGE_INC (char, s[1], 0x80, 0xbf)))
100 { 114 {
101 *clen = 2; 115 *clen = 2;
102 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f); 116 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f);
103 } 117 }
104 else 118 else
105 { 119 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY);
106 *clen = (STRLEN)-1; 120}
107 return (UV)-1; 121
122// likewise for encoding, also never called for ascii codepoints
123// this function takes advantage of this fact, although current gccs
124// seem to optimise the check for >= 0x80 away anyways
125INLINE unsigned char *
126encode_utf8 (unsigned char *s, UV ch)
127{
128 if (ch <= 0x7FF)
108 } 129 {
130 *s++ = (ch >> 6) | 0xc0;
131 *s++ = (ch & 0x3f) | 0x80;
132 }
133 else
134 s = uvuni_to_utf8_flags (s, ch, 0);
135
136 return s;
109} 137}
110 138
111///////////////////////////////////////////////////////////////////////////// 139/////////////////////////////////////////////////////////////////////////////
112// encoder 140// encoder
113 141
… …
118 char *end; // SvEND (sv) 146 char *end; // SvEND (sv)
119 SV *sv; // result scalar 147 SV *sv; // result scalar
120 JSON json; 148 JSON json;
121 U32 indent; // indentation level 149 U32 indent; // indentation level
122 U32 maxdepth; // max. indentation/recursion level 150 U32 maxdepth; // max. indentation/recursion level
151 UV limit; // escape character values >= this value when encoding
123} enc_t; 152} enc_t;
124 153
125inline void 154INLINE void
126need (enc_t *enc, STRLEN len) 155need (enc_t *enc, STRLEN len)
127{ 156{
128 if (expect_false (enc->cur + len >= enc->end)) 157 if (expect_false (enc->cur + len >= enc->end))
129 { 158 {
130 STRLEN cur = enc->cur - SvPVX (enc->sv); 159 STRLEN cur = enc->cur - SvPVX (enc->sv);
… …
132 enc->cur = SvPVX (enc->sv) + cur; 161 enc->cur = SvPVX (enc->sv) + cur;
133 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; 162 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
134 } 163 }
135} 164}
136 165
137inline void 166INLINE void
138encode_ch (enc_t *enc, char ch) 167encode_ch (enc_t *enc, char ch)
139{ 168{
140 need (enc, 1); 169 need (enc, 1);
141 *enc->cur++ = ch; 170 *enc->cur++ = ch;
142} 171}
… …
196 { 225 {
197 uch = ch; 226 uch = ch;
198 clen = 1; 227 clen = 1;
199 } 228 }
200 229
201 if (uch > 0x10FFFFUL) 230 if (uch < 0x80/*0x20*/ || uch >= enc->limit)
202 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
203
204 if (uch < 0x80 || enc->json.flags & F_ASCII || (enc->json.flags & F_LATIN1 && uch > 0xFF))
205 { 231 {
206 if (uch > 0xFFFFUL) 232 if (uch > 0xFFFFUL)
207 { 233 {
234 if (uch > 0x10FFFFUL)
235 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
236
208 need (enc, len += 11); 237 need (enc, len += 11);
209 sprintf (enc->cur, "\\u%04x\\u%04x", 238 sprintf (enc->cur, "\\u%04x\\u%04x",
210 (int)((uch - 0x10000) / 0x400 + 0xD800), 239 (int)((uch - 0x10000) / 0x400 + 0xD800),
211 (int)((uch - 0x10000) % 0x400 + 0xDC00)); 240 (int)((uch - 0x10000) % 0x400 + 0xDC00));
212 enc->cur += 12; 241 enc->cur += 12;
… …
240 while (--clen); 269 while (--clen);
241 } 270 }
242 else 271 else
243 { 272 {
244 need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed 273 need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed
245 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0); 274 enc->cur = encode_utf8 (enc->cur, uch);
246 ++str; 275 ++str;
247 } 276 }
248 } 277 }
249 } 278 }
250 } 279 }
251 280
252 --len; 281 --len;
253 } 282 }
254} 283}
255 284
256inline void 285INLINE void
257encode_indent (enc_t *enc) 286encode_indent (enc_t *enc)
258{ 287{
259 if (enc->json.flags & F_INDENT) 288 if (enc->json.flags & F_INDENT)
260 { 289 {
261 int spaces = enc->indent * INDENT_STEP; 290 int spaces = enc->indent * INDENT_STEP;
… …
264 memset (enc->cur, ' ', spaces); 293 memset (enc->cur, ' ', spaces);
265 enc->cur += spaces; 294 enc->cur += spaces;
266 } 295 }
267} 296}
268 297
269inline void 298INLINE void
270encode_space (enc_t *enc) 299encode_space (enc_t *enc)
271{ 300{
272 need (enc, 1); 301 need (enc, 1);
273 encode_ch (enc, ' '); 302 encode_ch (enc, ' ');
274} 303}
275 304
276inline void 305INLINE void
277encode_nl (enc_t *enc) 306encode_nl (enc_t *enc)
278{ 307{
279 if (enc->json.flags & F_INDENT) 308 if (enc->json.flags & F_INDENT)
280 { 309 {
281 need (enc, 1); 310 need (enc, 1);
282 encode_ch (enc, '\n'); 311 encode_ch (enc, '\n');
283 } 312 }
284} 313}
285 314
286inline void 315INLINE void
287encode_comma (enc_t *enc) 316encode_comma (enc_t *enc)
288{ 317{
289 encode_ch (enc, ','); 318 encode_ch (enc, ',');
290 319
291 if (enc->json.flags & F_INDENT) 320 if (enc->json.flags & F_INDENT)
… …
302 int i, len = av_len (av); 331 int i, len = av_len (av);
303 332
304 if (enc->indent >= enc->maxdepth) 333 if (enc->indent >= enc->maxdepth)
305 croak ("data structure too deep (hit recursion limit)"); 334 croak ("data structure too deep (hit recursion limit)");
306 335
307 encode_ch (enc, '['); encode_nl (enc); 336 encode_ch (enc, '[');
308 ++enc->indent; 337
338 if (len >= 0)
339 {
340 encode_nl (enc); ++enc->indent;
309 341
310 for (i = 0; i <= len; ++i) 342 for (i = 0; i <= len; ++i)
311 { 343 {
312 SV **svp = av_fetch (av, i, 0); 344 SV **svp = av_fetch (av, i, 0);
313 345
314 encode_indent (enc); 346 encode_indent (enc);
315 347
316 if (svp) 348 if (svp)
317 encode_sv (enc, *svp); 349 encode_sv (enc, *svp);
318 else 350 else
319 encode_str (enc, "null", 4, 0); 351 encode_str (enc, "null", 4, 0);
320 352
321 if (i < len) 353 if (i < len)
322 encode_comma (enc); 354 encode_comma (enc);
323 } 355 }
324 356
357 encode_nl (enc); --enc->indent; encode_indent (enc);
358 }
359
325 encode_nl (enc); 360 encode_ch (enc, ']');
326
327 --enc->indent;
328 encode_indent (enc); encode_ch (enc, ']');
329} 361}
330 362
331static void 363static void
332encode_he (enc_t *enc, HE *he) 364encode_hk (enc_t *enc, HE *he)
333{ 365{
334 encode_ch (enc, '"'); 366 encode_ch (enc, '"');
335 367
336 if (HeKLEN (he) == HEf_SVKEY) 368 if (HeKLEN (he) == HEf_SVKEY)
337 { 369 {
… …
350 encode_ch (enc, '"'); 382 encode_ch (enc, '"');
351 383
352 if (enc->json.flags & F_SPACE_BEFORE) encode_space (enc); 384 if (enc->json.flags & F_SPACE_BEFORE) encode_space (enc);
353 encode_ch (enc, ':'); 385 encode_ch (enc, ':');
354 if (enc->json.flags & F_SPACE_AFTER ) encode_space (enc); 386 if (enc->json.flags & F_SPACE_AFTER ) encode_space (enc);
355 encode_sv (enc, HeVAL (he));
356} 387}
357 388
358// compare hash entries, used when all keys are bytestrings 389// compare hash entries, used when all keys are bytestrings
359static int 390static int
360he_cmp_fast (const void *a_, const void *b_) 391he_cmp_fast (const void *a_, const void *b_)
… …
365 HE *b = *(HE **)b_; 396 HE *b = *(HE **)b_;
366 397
367 STRLEN la = HeKLEN (a); 398 STRLEN la = HeKLEN (a);
368 STRLEN lb = HeKLEN (b); 399 STRLEN lb = HeKLEN (b);
369 400
370 if (!(cmp = memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb))) 401 if (!(cmp = memcmp (HeKEY (b), HeKEY (a), lb < la ? lb : la)))
371 cmp = la - lb; 402 cmp = lb - la;
372 403
373 return cmp; 404 return cmp;
374} 405}
375 406
376// compare hash entries, used when some keys are sv's or utf-x 407// compare hash entries, used when some keys are sv's or utf-x
377static int 408static int
378he_cmp_slow (const void *a, const void *b) 409he_cmp_slow (const void *a, const void *b)
379{ 410{
380 return sv_cmp (HeSVKEY_force (*(HE **)a), HeSVKEY_force (*(HE **)b)); 411 return sv_cmp (HeSVKEY_force (*(HE **)b), HeSVKEY_force (*(HE **)a));
381} 412}
382 413
383static void 414static void
384encode_hv (enc_t *enc, HV *hv) 415encode_hv (enc_t *enc, HV *hv)
385{ 416{
417 HE *he;
386 int count, i; 418 int count;
387 419
388 if (enc->indent >= enc->maxdepth) 420 if (enc->indent >= enc->maxdepth)
389 croak ("data structure too deep (hit recursion limit)"); 421 croak ("data structure too deep (hit recursion limit)");
390 422
391 encode_ch (enc, '{'); encode_nl (enc); ++enc->indent; 423 encode_ch (enc, '{');
392 424
393 if ((count = hv_iterinit (hv)))
394 {
395 // for canonical output we have to sort by keys first 425 // for canonical output we have to sort by keys first
396 // actually, this is mostly due to the stupid so-called 426 // actually, this is mostly due to the stupid so-called
397 // security workaround added somewhere in 5.8.x. 427 // security workaround added somewhere in 5.8.x.
398 // that randomises hash orderings 428 // that randomises hash orderings
399 if (enc->json.flags & F_CANONICAL) 429 if (enc->json.flags & F_CANONICAL)
430 {
431 int count = hv_iterinit (hv);
432
433 if (SvMAGICAL (hv))
400 { 434 {
435 // need to count by iterating. could improve by dynamically building the vector below
436 // but I don't care for the speed of this special case.
437 // note also that we will run into undefined behaviour when the two iterations
438 // do not result in the same count, something I might care for in some later release.
439
440 count = 0;
441 while (hv_iternext (hv))
442 ++count;
443
444 hv_iterinit (hv);
445 }
446
447 if (count)
448 {
401 int fast = 1; 449 int i, fast = 1;
402 HE *he;
403#if defined(__BORLANDC__) || defined(_MSC_VER) 450#if defined(__BORLANDC__) || defined(_MSC_VER)
404 HE **hes = _alloca (count * sizeof (HE)); 451 HE **hes = _alloca (count * sizeof (HE));
405#else 452#else
406 HE *hes [count]; // if your compiler dies here, you need to enable C99 mode 453 HE *hes [count]; // if your compiler dies here, you need to enable C99 mode
407#endif 454#endif
… …
434 481
435 FREETMPS; 482 FREETMPS;
436 LEAVE; 483 LEAVE;
437 } 484 }
438 485
439 for (i = 0; i < count; ++i) 486 encode_nl (enc); ++enc->indent;
487
488 while (count--)
440 { 489 {
441 encode_indent (enc); 490 encode_indent (enc);
491 he = hes [count];
442 encode_he (enc, hes [i]); 492 encode_hk (enc, he);
493 encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he));
443 494
444 if (i < count - 1) 495 if (count)
445 encode_comma (enc); 496 encode_comma (enc);
446 } 497 }
447 498
448 encode_nl (enc); 499 encode_nl (enc); --enc->indent; encode_indent (enc);
449 } 500 }
501 }
450 else 502 else
503 {
504 if (hv_iterinit (hv) || SvMAGICAL (hv))
505 if ((he = hv_iternext (hv)))
451 { 506 {
452 HE *he = hv_iternext (hv); 507 encode_nl (enc); ++enc->indent;
453 508
454 for (;;) 509 for (;;)
455 { 510 {
456 encode_indent (enc); 511 encode_indent (enc);
457 encode_he (enc, he); 512 encode_hk (enc, he);
513 encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he));
458 514
459 if (!(he = hv_iternext (hv))) 515 if (!(he = hv_iternext (hv)))
460 break; 516 break;
461 517
462 encode_comma (enc); 518 encode_comma (enc);
463 } 519 }
464 520
465 encode_nl (enc); 521 encode_nl (enc); --enc->indent; encode_indent (enc);
466 } 522 }
467 } 523 }
468 524
469 --enc->indent; encode_indent (enc); encode_ch (enc, '}'); 525 encode_ch (enc, '}');
470} 526}
471 527
472// encode objects, arrays and special \0=false and \1=true values. 528// encode objects, arrays and special \0=false and \1=true values.
473static void 529static void
474encode_rv (enc_t *enc, SV *sv) 530encode_rv (enc_t *enc, SV *sv)
… …
478 SvGETMAGIC (sv); 534 SvGETMAGIC (sv);
479 svt = SvTYPE (sv); 535 svt = SvTYPE (sv);
480 536
481 if (expect_false (SvOBJECT (sv))) 537 if (expect_false (SvOBJECT (sv)))
482 { 538 {
539 HV *stash = !JSON_SLOW || json_boolean_stash
540 ? json_boolean_stash
541 : gv_stashpv ("JSON::XS::Boolean", 1);
542
483 if (SvSTASH (sv) == json_boolean_stash) 543 if (SvSTASH (sv) == stash)
484 { 544 {
485 if (SvIV (sv)) 545 if (SvIV (sv))
486 encode_str (enc, "true", 4, 0); 546 encode_str (enc, "true", 4, 0);
487 else 547 else
488 encode_str (enc, "false", 5, 0); 548 encode_str (enc, "false", 5, 0);
… …
500 // we re-bless the reference to get overload and other niceties right 560 // we re-bless the reference to get overload and other niceties right
501 GV *to_json = gv_fetchmethod_autoload (SvSTASH (sv), "TO_JSON", 0); 561 GV *to_json = gv_fetchmethod_autoload (SvSTASH (sv), "TO_JSON", 0);
502 562
503 if (to_json) 563 if (to_json)
504 { 564 {
505 int count;
506 dSP; 565 dSP;
507 566
508 ENTER; SAVETMPS; PUSHMARK (SP); 567 ENTER; SAVETMPS; PUSHMARK (SP);
509 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv))); 568 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv)));
510 569
… …
640 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 699 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
641 enc.cur = SvPVX (enc.sv); 700 enc.cur = SvPVX (enc.sv);
642 enc.end = SvEND (enc.sv); 701 enc.end = SvEND (enc.sv);
643 enc.indent = 0; 702 enc.indent = 0;
644 enc.maxdepth = DEC_DEPTH (enc.json.flags); 703 enc.maxdepth = DEC_DEPTH (enc.json.flags);
704 enc.limit = enc.json.flags & F_ASCII ? 0x000080UL
705 : enc.json.flags & F_LATIN1 ? 0x000100UL
706 : 0x10FFFFUL;
645 707
646 SvPOK_only (enc.sv); 708 SvPOK_only (enc.sv);
647 encode_sv (&enc, scalar); 709 encode_sv (&enc, scalar);
648 710
649 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 711 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
… …
670 JSON json; 732 JSON json;
671 U32 depth; // recursion depth 733 U32 depth; // recursion depth
672 U32 maxdepth; // recursion depth limit 734 U32 maxdepth; // recursion depth limit
673} dec_t; 735} dec_t;
674 736
675inline void 737INLINE void
738decode_comment (dec_t *dec)
739{
740 // only '#'-style comments allowed a.t.m.
741
742 while (*dec->cur && *dec->cur != 0x0a && *dec->cur != 0x0d)
743 ++dec->cur;
744}
745
746INLINE void
676decode_ws (dec_t *dec) 747decode_ws (dec_t *dec)
677{ 748{
678 for (;;) 749 for (;;)
679 { 750 {
680 char ch = *dec->cur; 751 char ch = *dec->cur;
681 752
682 if (ch > 0x20 753 if (ch > 0x20)
754 {
755 if (expect_false (ch == '#'))
756 {
757 if (dec->json.flags & F_RELAXED)
758 decode_comment (dec);
759 else
760 break;
761 }
762 else
763 break;
764 }
683 || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)) 765 else if (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)
684 break; 766 break; // parse error, but let higher level handle it, gives better error messages
685 767
686 ++dec->cur; 768 ++dec->cur;
687 } 769 }
688} 770}
689 771
… …
795 877
796 if (hi >= 0x80) 878 if (hi >= 0x80)
797 { 879 {
798 utf8 = 1; 880 utf8 = 1;
799 881
800 cur = (char *)uvuni_to_utf8_flags (cur, hi, 0); 882 cur = encode_utf8 (cur, hi);
801 } 883 }
802 else 884 else
803 *cur++ = hi; 885 *cur++ = hi;
804 } 886 }
805 break; 887 break;
… …
940 1022
941 if (!is_nv) 1023 if (!is_nv)
942 { 1024 {
943 int len = dec->cur - start; 1025 int len = dec->cur - start;
944 1026
945 // special case the rather common 1..4-digit-int case, assumes 32 bit ints or so 1027 // special case the rather common 1..5-digit-int case
946 if (*start == '-') 1028 if (*start == '-')
947 switch (len) 1029 switch (len)
948 { 1030 {
949 case 2: return newSViv (-( start [1] - '0' * 1)); 1031 case 2: return newSViv (-( start [1] - '0' * 1));
950 case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11)); 1032 case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11));
951 case 4: return newSViv (-( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111)); 1033 case 4: return newSViv (-( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111));
952 case 5: return newSViv (-(start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111)); 1034 case 5: return newSViv (-( start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111));
1035 case 6: return newSViv (-(start [1] * 10000 + start [2] * 1000 + start [3] * 100 + start [4] * 10 + start [5] - '0' * 11111));
953 } 1036 }
954 else 1037 else
955 switch (len) 1038 switch (len)
956 { 1039 {
957 case 1: return newSViv ( start [0] - '0' * 1); 1040 case 1: return newSViv ( start [0] - '0' * 1);
958 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11); 1041 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11);
959 case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111); 1042 case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111);
960 case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111); 1043 case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111);
1044 case 5: return newSViv ( start [0] * 10000 + start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 11111);
961 } 1045 }
962 1046
963 { 1047 {
964 UV uv; 1048 UV uv;
965 int numtype = grok_number (start, len, &uv); 1049 int numtype = grok_number (start, len, &uv);
… …
1026 1110
1027 if (*dec->cur != ',') 1111 if (*dec->cur != ',')
1028 ERR (", or ] expected while parsing array"); 1112 ERR (", or ] expected while parsing array");
1029 1113
1030 ++dec->cur; 1114 ++dec->cur;
1115
1116 decode_ws (dec);
1117
1118 if (*dec->cur == ']' && dec->json.flags & F_RELAXED)
1119 {
1120 ++dec->cur;
1121 break;
1122 }
1031 } 1123 }
1032 1124
1033 DEC_DEC_DEPTH; 1125 DEC_DEC_DEPTH;
1034 return newRV_noinc ((SV *)av); 1126 return newRV_noinc ((SV *)av);
1035 1127
… …
1051 if (*dec->cur == '}') 1143 if (*dec->cur == '}')
1052 ++dec->cur; 1144 ++dec->cur;
1053 else 1145 else
1054 for (;;) 1146 for (;;)
1055 { 1147 {
1056 decode_ws (dec); EXPECT_CH ('"'); 1148 EXPECT_CH ('"');
1057 1149
1058 // heuristic: assume that 1150 // heuristic: assume that
1059 // a) decode_str + hv_store_ent are abysmally slow. 1151 // a) decode_str + hv_store_ent are abysmally slow.
1060 // b) most hash keys are short, simple ascii text. 1152 // b) most hash keys are short, simple ascii text.
1061 // => try to "fast-match" such strings to avoid 1153 // => try to "fast-match" such strings to avoid
… …
1075 if (!key) 1167 if (!key)
1076 goto fail; 1168 goto fail;
1077 1169
1078 decode_ws (dec); EXPECT_CH (':'); 1170 decode_ws (dec); EXPECT_CH (':');
1079 1171
1172 decode_ws (dec);
1080 value = decode_sv (dec); 1173 value = decode_sv (dec);
1081 if (!value) 1174 if (!value)
1082 { 1175 {
1083 SvREFCNT_dec (key); 1176 SvREFCNT_dec (key);
1084 goto fail; 1177 goto fail;
… …
1096 int len = p - key; 1189 int len = p - key;
1097 dec->cur = p + 1; 1190 dec->cur = p + 1;
1098 1191
1099 decode_ws (dec); EXPECT_CH (':'); 1192 decode_ws (dec); EXPECT_CH (':');
1100 1193
1194 decode_ws (dec);
1101 value = decode_sv (dec); 1195 value = decode_sv (dec);
1102 if (!value) 1196 if (!value)
1103 goto fail; 1197 goto fail;
1104 1198
1105 hv_store (hv, key, len, value, 0); 1199 hv_store (hv, key, len, value, 0);
… …
1121 1215
1122 if (*dec->cur != ',') 1216 if (*dec->cur != ',')
1123 ERR (", or } expected while parsing object/hash"); 1217 ERR (", or } expected while parsing object/hash");
1124 1218
1125 ++dec->cur; 1219 ++dec->cur;
1220
1221 decode_ws (dec);
1222
1223 if (*dec->cur == '}' && dec->json.flags & F_RELAXED)
1224 {
1225 ++dec->cur;
1226 break;
1227 }
1126 } 1228 }
1127 1229
1128 DEC_DEC_DEPTH; 1230 DEC_DEC_DEPTH;
1129 sv = newRV_noinc ((SV *)hv); 1231 sv = newRV_noinc ((SV *)hv);
1130 1232
… …
1195} 1297}
1196 1298
1197static SV * 1299static SV *
1198decode_sv (dec_t *dec) 1300decode_sv (dec_t *dec)
1199{ 1301{
1200 decode_ws (dec);
1201
1202 // the beauty of JSON: you need exactly one character lookahead 1302 // the beauty of JSON: you need exactly one character lookahead
1203 // to parse anything. 1303 // to parse anything.
1204 switch (*dec->cur) 1304 switch (*dec->cur)
1205 { 1305 {
1206 case '"': ++dec->cur; return decode_str (dec); 1306 case '"': ++dec->cur; return decode_str (dec);
… …
1214 1314
1215 case 't': 1315 case 't':
1216 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) 1316 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
1217 { 1317 {
1218 dec->cur += 4; 1318 dec->cur += 4;
1319#if JSON_SLOW
1320 json_true = get_sv ("JSON::XS::true", 1); SvREADONLY_on (json_true);
1321#endif
1219 return SvREFCNT_inc (json_true); 1322 return SvREFCNT_inc (json_true);
1220 } 1323 }
1221 else 1324 else
1222 ERR ("'true' expected"); 1325 ERR ("'true' expected");
1223 1326
… …
1225 1328
1226 case 'f': 1329 case 'f':
1227 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) 1330 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
1228 { 1331 {
1229 dec->cur += 5; 1332 dec->cur += 5;
1333#if JSON_SLOW
1334 json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false);
1335#endif
1230 return SvREFCNT_inc (json_false); 1336 return SvREFCNT_inc (json_false);
1231 } 1337 }
1232 else 1338 else
1233 ERR ("'false' expected"); 1339 ERR ("'false' expected");
1234 1340
… …
1284 1390
1285 if (dec.json.cb_object || dec.json.cb_sk_object) 1391 if (dec.json.cb_object || dec.json.cb_sk_object)
1286 dec.json.flags |= F_HOOK; 1392 dec.json.flags |= F_HOOK;
1287 1393
1288 *dec.end = 0; // this should basically be a nop, too, but make sure it's there 1394 *dec.end = 0; // this should basically be a nop, too, but make sure it's there
1395
1396 decode_ws (&dec);
1289 sv = decode_sv (&dec); 1397 sv = decode_sv (&dec);
1290 1398
1291 if (!(offset_return || !sv)) 1399 if (!(offset_return || !sv))
1292 { 1400 {
1293 // check for trailing garbage 1401 // check for trailing garbage
… …
1361 json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false); 1469 json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false);
1362} 1470}
1363 1471
1364PROTOTYPES: DISABLE 1472PROTOTYPES: DISABLE
1365 1473
1474void CLONE (...)
1475 CODE:
1476 json_stash = 0;
1477 json_boolean_stash = 0;
1478
1366void new (char *klass) 1479void new (char *klass)
1367 PPCODE: 1480 PPCODE:
1368{ 1481{
1369 SV *pv = NEWSV (0, sizeof (JSON)); 1482 SV *pv = NEWSV (0, sizeof (JSON));
1370 SvPOK_only (pv); 1483 SvPOK_only (pv);
1371 Zero (SvPVX (pv), 1, JSON); 1484 Zero (SvPVX (pv), 1, JSON);
1372 ((JSON *)SvPVX (pv))->flags = F_DEFAULT; 1485 ((JSON *)SvPVX (pv))->flags = F_DEFAULT;
1373 XPUSHs (sv_2mortal (sv_bless (newRV_noinc (pv), json_stash))); 1486 XPUSHs (sv_2mortal (sv_bless (
1487 newRV_noinc (pv),
1488 strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1)
1489 )));
1374} 1490}
1375 1491
1376void ascii (JSON *self, int enable = 1) 1492void ascii (JSON *self, int enable = 1)
1377 ALIAS: 1493 ALIAS:
1378 ascii = F_ASCII 1494 ascii = F_ASCII
… …
1385 pretty = F_PRETTY 1501 pretty = F_PRETTY
1386 allow_nonref = F_ALLOW_NONREF 1502 allow_nonref = F_ALLOW_NONREF
1387 shrink = F_SHRINK 1503 shrink = F_SHRINK
1388 allow_blessed = F_ALLOW_BLESSED 1504 allow_blessed = F_ALLOW_BLESSED
1389 convert_blessed = F_CONV_BLESSED 1505 convert_blessed = F_CONV_BLESSED
1506 relaxed = F_RELAXED
1390 PPCODE: 1507 PPCODE:
1391{ 1508{
1392 if (enable) 1509 if (enable)
1393 self->flags |= ix; 1510 self->flags |= ix;
1394 else 1511 else
1395 self->flags &= ~ix; 1512 self->flags &= ~ix;
1396 1513
1397 XPUSHs (ST (0)); 1514 XPUSHs (ST (0));
1398} 1515}
1399 1516
1517void get_ascii (JSON *self)
1518 ALIAS:
1519 get_ascii = F_ASCII
1520 get_latin1 = F_LATIN1
1521 get_utf8 = F_UTF8
1522 get_indent = F_INDENT
1523 get_canonical = F_CANONICAL
1524 get_space_before = F_SPACE_BEFORE
1525 get_space_after = F_SPACE_AFTER
1526 get_allow_nonref = F_ALLOW_NONREF
1527 get_shrink = F_SHRINK
1528 get_allow_blessed = F_ALLOW_BLESSED
1529 get_convert_blessed = F_CONV_BLESSED
1530 get_relaxed = F_RELAXED
1531 PPCODE:
1532 XPUSHs (boolSV (self->flags & ix));
1533
1400void max_depth (JSON *self, UV max_depth = 0x80000000UL) 1534void max_depth (JSON *self, UV max_depth = 0x80000000UL)
1401 PPCODE: 1535 PPCODE:
1402{ 1536{
1403 UV log2 = 0; 1537 UV log2 = 0;
1404 1538
… …
1410 self->flags = self->flags & ~F_MAXDEPTH | (log2 << S_MAXDEPTH); 1544 self->flags = self->flags & ~F_MAXDEPTH | (log2 << S_MAXDEPTH);
1411 1545
1412 XPUSHs (ST (0)); 1546 XPUSHs (ST (0));
1413} 1547}
1414 1548
1549U32 get_max_depth (JSON *self)
1550 CODE:
1551 RETVAL = DEC_DEPTH (self->flags);
1552 OUTPUT:
1553 RETVAL
1554
1415void max_size (JSON *self, UV max_size = 0) 1555void max_size (JSON *self, UV max_size = 0)
1416 PPCODE: 1556 PPCODE:
1417{ 1557{
1418 UV log2 = 0; 1558 UV log2 = 0;
1419 1559
… …
1425 1565
1426 self->flags = self->flags & ~F_MAXSIZE | (log2 << S_MAXSIZE); 1566 self->flags = self->flags & ~F_MAXSIZE | (log2 << S_MAXSIZE);
1427 1567
1428 XPUSHs (ST (0)); 1568 XPUSHs (ST (0));
1429} 1569}
1570
1571int get_max_size (JSON *self)
1572 CODE:
1573 RETVAL = DEC_SIZE (self->flags);
1574 OUTPUT:
1575 RETVAL
1430 1576
1431void filter_json_object (JSON *self, SV *cb = &PL_sv_undef) 1577void filter_json_object (JSON *self, SV *cb = &PL_sv_undef)
1432 PPCODE: 1578 PPCODE:
1433{ 1579{
1434 SvREFCNT_dec (self->cb_object); 1580 SvREFCNT_dec (self->cb_object);
… …
1481 SvREFCNT_dec (self->cb_sk_object); 1627 SvREFCNT_dec (self->cb_sk_object);
1482 SvREFCNT_dec (self->cb_object); 1628 SvREFCNT_dec (self->cb_object);
1483 1629
1484PROTOTYPES: ENABLE 1630PROTOTYPES: ENABLE
1485 1631
1486void to_json (SV *scalar) 1632void encode_json (SV *scalar)
1487 PPCODE: 1633 PPCODE:
1488{ 1634{
1489 JSON json = { F_DEFAULT | F_UTF8 }; 1635 JSON json = { F_DEFAULT | F_UTF8 };
1490 XPUSHs (encode_json (scalar, &json)); 1636 XPUSHs (encode_json (scalar, &json));
1491} 1637}
1492 1638
1493void from_json (SV *jsonstr) 1639void decode_json (SV *jsonstr)
1494 PPCODE: 1640 PPCODE:
1495{ 1641{
1496 JSON json = { F_DEFAULT | F_UTF8 }; 1642 JSON json = { F_DEFAULT | F_UTF8 };
1497 XPUSHs (decode_json (jsonstr, &json, 0)); 1643 XPUSHs (decode_json (jsonstr, &json, 0));
1498} 1644}

Diff Legend

– Removed lines
+ Added lines
< Changed lines
> Changed lines