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 1.40 by root, Tue Jun 12 01:27:02 2007 UTC vs.
Revision 1.63 by root, Mon Aug 27 01:49:01 2007 UTC

1#include "EXTERN.h" 1#include "EXTERN.h"
2#include "perl.h" 2#include "perl.h"
3#include "XSUB.h" 3#include "XSUB.h"
4 4
5#include "assert.h" 5#include <assert.h>
6#include "string.h" 6#include <string.h>
7#include "stdlib.h" 7#include <stdlib.h>
8#include "stdio.h" 8#include <stdio.h>
9#include <float.h>
9 10
10#if defined(__BORLANDC__) || defined(_MSC_VER) 11#if defined(__BORLANDC__) || defined(_MSC_VER)
11# define snprintf _snprintf // C compilers have this in stdio.h 12# define snprintf _snprintf // C compilers have this in stdio.h
12#endif 13#endif
13 14
15// some old perls do not have this, try to make it work, no
16// guarentees, though. if it breaks, you get to keep the pieces.
17#ifndef UTF8_MAXBYTES
18# define UTF8_MAXBYTES 13
19#endif
20
14#define F_ASCII 0x00000001UL 21#define F_ASCII 0x00000001UL
15#define F_LATIN1 0x00000002UL 22#define F_LATIN1 0x00000002UL
16#define F_UTF8 0x00000004UL 23#define F_UTF8 0x00000004UL
17#define F_INDENT 0x00000008UL 24#define F_INDENT 0x00000008UL
18#define F_CANONICAL 0x00000010UL 25#define F_CANONICAL 0x00000010UL
19#define F_SPACE_BEFORE 0x00000020UL 26#define F_SPACE_BEFORE 0x00000020UL
20#define F_SPACE_AFTER 0x00000040UL 27#define F_SPACE_AFTER 0x00000040UL
21#define F_ALLOW_NONREF 0x00000100UL 28#define F_ALLOW_NONREF 0x00000100UL
22#define F_SHRINK 0x00000200UL 29#define F_SHRINK 0x00000200UL
30#define F_ALLOW_BLESSED 0x00000400UL
31#define F_CONV_BLESSED 0x00000800UL
32#define F_RELAXED 0x00001000UL
33
23#define F_MAXDEPTH 0xf8000000UL 34#define F_MAXDEPTH 0xf8000000UL
24#define S_MAXDEPTH 27 35#define S_MAXDEPTH 27
36#define F_MAXSIZE 0x01f00000UL
37#define S_MAXSIZE 20
38#define F_HOOK 0x00080000UL // some hooks exist, so slow-path processing
25 39
26#define DEC_DEPTH(flags) (1UL << ((flags & F_MAXDEPTH) >> S_MAXDEPTH)) 40#define DEC_DEPTH(flags) (1UL << ((flags & F_MAXDEPTH) >> S_MAXDEPTH))
27 41#define DEC_SIZE(flags) (1UL << ((flags & F_MAXSIZE ) >> S_MAXSIZE ))
28// F_SELFCONVERT? <=> to_json/toJson
29// F_BLESSED? <=> { $__class__$ => }
30 42
31#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER 43#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER
32#define F_DEFAULT (9UL << S_MAXDEPTH) 44#define F_DEFAULT (9UL << S_MAXDEPTH)
33 45
34#define INIT_SIZE 32 // initial scalar size to be allocated 46#define INIT_SIZE 32 // initial scalar size to be allocated
48#endif 60#endif
49 61
50#define expect_false(expr) expect ((expr) != 0, 0) 62#define expect_false(expr) expect ((expr) != 0, 0)
51#define expect_true(expr) expect ((expr) != 0, 1) 63#define expect_true(expr) expect ((expr) != 0, 1)
52 64
65#ifdef USE_ITHREADS
66# define JSON_SLOW 1
67# define JSON_STASH (json_stash ? json_stash : gv_stashpv ("JSON::XS", 1))
68#else
69# define JSON_SLOW 0
70# define JSON_STASH json_stash
71#endif
72
53static HV *json_stash; // JSON::XS:: 73static HV *json_stash, *json_boolean_stash; // JSON::XS::
74static SV *json_true, *json_false;
75
76typedef struct {
77 U32 flags;
78 SV *cb_object;
79 HV *cb_sk_object;
80} JSON;
54 81
55///////////////////////////////////////////////////////////////////////////// 82/////////////////////////////////////////////////////////////////////////////
56// utility functions 83// utility functions
57 84
58static UV * 85inline void
59SvJSON (SV *sv)
60{
61 if (!(SvROK (sv) && SvOBJECT (SvRV (sv)) && SvSTASH (SvRV (sv)) == json_stash))
62 croak ("object is not of type JSON::XS");
63
64 return &SvUVX (SvRV (sv));
65}
66
67static void
68shrink (SV *sv) 86shrink (SV *sv)
69{ 87{
70 sv_utf8_downgrade (sv, 1); 88 sv_utf8_downgrade (sv, 1);
71 if (SvLEN (sv) > SvCUR (sv) + 1) 89 if (SvLEN (sv) > SvCUR (sv) + 1)
72 { 90 {
107typedef struct 125typedef struct
108{ 126{
109 char *cur; // SvPVX (sv) + current output position 127 char *cur; // SvPVX (sv) + current output position
110 char *end; // SvEND (sv) 128 char *end; // SvEND (sv)
111 SV *sv; // result scalar 129 SV *sv; // result scalar
112 U32 flags; // F_* 130 JSON json;
113 U32 indent; // indentation level 131 U32 indent; // indentation level
114 U32 maxdepth; // max. indentation/recursion level 132 U32 maxdepth; // max. indentation/recursion level
115} enc_t; 133} enc_t;
116 134
117inline void 135inline void
191 } 209 }
192 210
193 if (uch > 0x10FFFFUL) 211 if (uch > 0x10FFFFUL)
194 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch); 212 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
195 213
196 if (uch < 0x80 || enc->flags & F_ASCII || (enc->flags & F_LATIN1 && uch > 0xFF)) 214 if (uch < 0x80 || enc->json.flags & F_ASCII || (enc->json.flags & F_LATIN1 && uch > 0xFF))
197 { 215 {
198 if (uch > 0xFFFFUL) 216 if (uch > 0xFFFFUL)
199 { 217 {
200 need (enc, len += 11); 218 need (enc, len += 11);
201 sprintf (enc->cur, "\\u%04x\\u%04x", 219 sprintf (enc->cur, "\\u%04x\\u%04x",
215 *enc->cur++ = hexdigit [(uch >> 0) & 15]; 233 *enc->cur++ = hexdigit [(uch >> 0) & 15];
216 } 234 }
217 235
218 str += clen; 236 str += clen;
219 } 237 }
220 else if (enc->flags & F_LATIN1) 238 else if (enc->json.flags & F_LATIN1)
221 { 239 {
222 *enc->cur++ = uch; 240 *enc->cur++ = uch;
223 str += clen; 241 str += clen;
224 } 242 }
225 else if (is_utf8) 243 else if (is_utf8)
246} 264}
247 265
248inline void 266inline void
249encode_indent (enc_t *enc) 267encode_indent (enc_t *enc)
250{ 268{
251 if (enc->flags & F_INDENT) 269 if (enc->json.flags & F_INDENT)
252 { 270 {
253 int spaces = enc->indent * INDENT_STEP; 271 int spaces = enc->indent * INDENT_STEP;
254 272
255 need (enc, spaces); 273 need (enc, spaces);
256 memset (enc->cur, ' ', spaces); 274 memset (enc->cur, ' ', spaces);
266} 284}
267 285
268inline void 286inline void
269encode_nl (enc_t *enc) 287encode_nl (enc_t *enc)
270{ 288{
271 if (enc->flags & F_INDENT) 289 if (enc->json.flags & F_INDENT)
272 { 290 {
273 need (enc, 1); 291 need (enc, 1);
274 encode_ch (enc, '\n'); 292 encode_ch (enc, '\n');
275 } 293 }
276} 294}
278inline void 296inline void
279encode_comma (enc_t *enc) 297encode_comma (enc_t *enc)
280{ 298{
281 encode_ch (enc, ','); 299 encode_ch (enc, ',');
282 300
283 if (enc->flags & F_INDENT) 301 if (enc->json.flags & F_INDENT)
284 encode_nl (enc); 302 encode_nl (enc);
285 else if (enc->flags & F_SPACE_AFTER) 303 else if (enc->json.flags & F_SPACE_AFTER)
286 encode_space (enc); 304 encode_space (enc);
287} 305}
288 306
289static void encode_sv (enc_t *enc, SV *sv); 307static void encode_sv (enc_t *enc, SV *sv);
290 308
299 encode_ch (enc, '['); encode_nl (enc); 317 encode_ch (enc, '['); encode_nl (enc);
300 ++enc->indent; 318 ++enc->indent;
301 319
302 for (i = 0; i <= len; ++i) 320 for (i = 0; i <= len; ++i)
303 { 321 {
322 SV **svp = av_fetch (av, i, 0);
323
304 encode_indent (enc); 324 encode_indent (enc);
305 encode_sv (enc, *av_fetch (av, i, 0)); 325
326 if (svp)
327 encode_sv (enc, *svp);
328 else
329 encode_str (enc, "null", 4, 0);
306 330
307 if (i < len) 331 if (i < len)
308 encode_comma (enc); 332 encode_comma (enc);
309 } 333 }
310 334
313 --enc->indent; 337 --enc->indent;
314 encode_indent (enc); encode_ch (enc, ']'); 338 encode_indent (enc); encode_ch (enc, ']');
315} 339}
316 340
317static void 341static void
318encode_he (enc_t *enc, HE *he) 342encode_hk (enc_t *enc, HE *he)
319{ 343{
320 encode_ch (enc, '"'); 344 encode_ch (enc, '"');
321 345
322 if (HeKLEN (he) == HEf_SVKEY) 346 if (HeKLEN (he) == HEf_SVKEY)
323 { 347 {
333 else 357 else
334 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he)); 358 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he));
335 359
336 encode_ch (enc, '"'); 360 encode_ch (enc, '"');
337 361
338 if (enc->flags & F_SPACE_BEFORE) encode_space (enc); 362 if (enc->json.flags & F_SPACE_BEFORE) encode_space (enc);
339 encode_ch (enc, ':'); 363 encode_ch (enc, ':');
340 if (enc->flags & F_SPACE_AFTER ) encode_space (enc); 364 if (enc->json.flags & F_SPACE_AFTER ) encode_space (enc);
341 encode_sv (enc, HeVAL (he));
342} 365}
343 366
344// compare hash entries, used when all keys are bytestrings 367// compare hash entries, used when all keys are bytestrings
345static int 368static int
346he_cmp_fast (const void *a_, const void *b_) 369he_cmp_fast (const void *a_, const void *b_)
351 HE *b = *(HE **)b_; 374 HE *b = *(HE **)b_;
352 375
353 STRLEN la = HeKLEN (a); 376 STRLEN la = HeKLEN (a);
354 STRLEN lb = HeKLEN (b); 377 STRLEN lb = HeKLEN (b);
355 378
356 if (!(cmp = memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb))) 379 if (!(cmp = memcmp (HeKEY (b), HeKEY (a), lb < la ? lb : la)))
357 cmp = la - lb; 380 cmp = lb - la;
358 381
359 return cmp; 382 return cmp;
360} 383}
361 384
362// compare hash entries, used when some keys are sv's or utf-x 385// compare hash entries, used when some keys are sv's or utf-x
363static int 386static int
364he_cmp_slow (const void *a, const void *b) 387he_cmp_slow (const void *a, const void *b)
365{ 388{
366 return sv_cmp (HeSVKEY_force (*(HE **)a), HeSVKEY_force (*(HE **)b)); 389 return sv_cmp (HeSVKEY_force (*(HE **)b), HeSVKEY_force (*(HE **)a));
367} 390}
368 391
369static void 392static void
370encode_hv (enc_t *enc, HV *hv) 393encode_hv (enc_t *enc, HV *hv)
371{ 394{
395 HE *he;
372 int count, i; 396 int count;
373 397
374 if (enc->indent >= enc->maxdepth) 398 if (enc->indent >= enc->maxdepth)
375 croak ("data structure too deep (hit recursion limit)"); 399 croak ("data structure too deep (hit recursion limit)");
376 400
377 encode_ch (enc, '{'); encode_nl (enc); ++enc->indent; 401 encode_ch (enc, '{'); encode_nl (enc); ++enc->indent;
378 402
379 if ((count = hv_iterinit (hv)))
380 {
381 // for canonical output we have to sort by keys first 403 // for canonical output we have to sort by keys first
382 // actually, this is mostly due to the stupid so-called 404 // actually, this is mostly due to the stupid so-called
383 // security workaround added somewhere in 5.8.x. 405 // security workaround added somewhere in 5.8.x.
384 // that randomises hash orderings 406 // that randomises hash orderings
385 if (enc->flags & F_CANONICAL) 407 if (enc->json.flags & F_CANONICAL)
408 {
409 int count = hv_iterinit (hv);
410
411 if (SvMAGICAL (hv))
386 { 412 {
413 // need to count by iterating. could improve by dynamically building the vector below
414 // but I don't care for the speed of this special case.
415 // note also that we will run into undefined behaviour when the two iterations
416 // do not result in the same count, something I might care for in some later release.
417
418 count = 0;
419 while (hv_iternext (hv))
420 ++count;
421
422 hv_iterinit (hv);
423 }
424
425 if (count)
426 {
387 int fast = 1; 427 int i, fast = 1;
388 HE *he;
389#if defined(__BORLANDC__) || defined(_MSC_VER) 428#if defined(__BORLANDC__) || defined(_MSC_VER)
390 HE **hes = _alloca (count * sizeof (HE)); 429 HE **hes = _alloca (count * sizeof (HE));
391#else 430#else
392 HE *hes [count]; // if your compiler dies here, you need to enable C99 mode 431 HE *hes [count]; // if your compiler dies here, you need to enable C99 mode
393#endif 432#endif
420 459
421 FREETMPS; 460 FREETMPS;
422 LEAVE; 461 LEAVE;
423 } 462 }
424 463
425 for (i = 0; i < count; ++i) 464 while (count--)
426 { 465 {
427 encode_indent (enc); 466 encode_indent (enc);
467 he = hes [count];
428 encode_he (enc, hes [i]); 468 encode_hk (enc, he);
469 encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he));
429 470
430 if (i < count - 1) 471 if (count)
431 encode_comma (enc); 472 encode_comma (enc);
432 } 473 }
433 474 }
475 }
476 else
477 {
478 if (hv_iterinit (hv) || SvMAGICAL (hv))
479 if ((he = hv_iternext (hv)))
480 for (;;)
481 {
434 encode_nl (enc); 482 encode_indent (enc);
483 encode_hk (enc, he);
484 encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he));
485
486 if (!(he = hv_iternext (hv)))
487 break;
488
489 encode_comma (enc);
490 }
491 }
492
493 encode_nl (enc);
494
495 --enc->indent; encode_indent (enc); encode_ch (enc, '}');
496}
497
498// encode objects, arrays and special \0=false and \1=true values.
499static void
500encode_rv (enc_t *enc, SV *sv)
501{
502 svtype svt;
503
504 SvGETMAGIC (sv);
505 svt = SvTYPE (sv);
506
507 if (expect_false (SvOBJECT (sv)))
508 {
509 HV *stash = !JSON_SLOW || json_boolean_stash
510 ? json_boolean_stash
511 : gv_stashpv ("JSON::XS::Boolean", 1);
512
513 if (SvSTASH (sv) == stash)
514 {
515 if (SvIV (sv))
516 encode_str (enc, "true", 4, 0);
517 else
518 encode_str (enc, "false", 5, 0);
435 } 519 }
436 else 520 else
437 { 521 {
438 HE *he = hv_iternext (hv); 522#if 0
439 523 if (0 && sv_derived_from (rv, "JSON::Literal"))
440 for (;;)
441 {
442 encode_indent (enc);
443 encode_he (enc, he);
444
445 if (!(he = hv_iternext (hv)))
446 break;
447
448 encode_comma (enc);
449 } 524 {
525 // not yet
526 }
527#endif
528 if (enc->json.flags & F_CONV_BLESSED)
529 {
530 // we re-bless the reference to get overload and other niceties right
531 GV *to_json = gv_fetchmethod_autoload (SvSTASH (sv), "TO_JSON", 0);
450 532
533 if (to_json)
534 {
535 dSP;
536
537 ENTER; SAVETMPS; PUSHMARK (SP);
538 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv)));
539
540 // calling with G_SCALAR ensures that we always get a 1 return value
541 PUTBACK;
542 call_sv ((SV *)GvCV (to_json), G_SCALAR);
543 SPAGAIN;
544
545 // catch this surprisingly common error
546 if (SvROK (TOPs) && SvRV (TOPs) == sv)
547 croak ("%s::TO_JSON method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv)));
548
549 sv = POPs;
550 PUTBACK;
551
451 encode_nl (enc); 552 encode_sv (enc, sv);
553
554 FREETMPS; LEAVE;
555 }
556 else if (enc->json.flags & F_ALLOW_BLESSED)
557 encode_str (enc, "null", 4, 0);
558 else
559 croak ("encountered object '%s', but neither allow_blessed enabled nor TO_JSON method available on it",
560 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
561 }
562 else if (enc->json.flags & F_ALLOW_BLESSED)
563 encode_str (enc, "null", 4, 0);
564 else
565 croak ("encountered object '%s', but neither allow_blessed nor convert_blessed settings are enabled",
566 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
452 } 567 }
453 } 568 }
454
455 --enc->indent; encode_indent (enc); encode_ch (enc, '}');
456}
457
458// encode objects, arrays and special \0=false and \1=true values.
459static void
460encode_rv (enc_t *enc, SV *sv)
461{
462 svtype svt;
463
464 SvGETMAGIC (sv);
465 svt = SvTYPE (sv);
466
467 if (svt == SVt_PVHV) 569 else if (svt == SVt_PVHV)
468 encode_hv (enc, (HV *)sv); 570 encode_hv (enc, (HV *)sv);
469 else if (svt == SVt_PVAV) 571 else if (svt == SVt_PVAV)
470 encode_av (enc, (AV *)sv); 572 encode_av (enc, (AV *)sv);
471 else if (svt < SVt_PVAV) 573 else if (svt < SVt_PVAV)
472 { 574 {
473 if (SvNIOK (sv) && SvIV (sv) == 0) 575 STRLEN len = 0;
576 char *pv = svt ? SvPV (sv, len) : 0;
577
578 if (len == 1 && *pv == '1')
579 encode_str (enc, "true", 4, 0);
580 else if (len == 1 && *pv == '0')
474 encode_str (enc, "false", 5, 0); 581 encode_str (enc, "false", 5, 0);
475 else if (SvNIOK (sv) && SvIV (sv) == 1)
476 encode_str (enc, "true", 4, 0);
477 else 582 else
478 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1", 583 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
479 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 584 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
480 } 585 }
481 else 586 else
551 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this", 656 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this",
552 SvPV_nolen (sv), SvFLAGS (sv)); 657 SvPV_nolen (sv), SvFLAGS (sv));
553} 658}
554 659
555static SV * 660static SV *
556encode_json (SV *scalar, U32 flags) 661encode_json (SV *scalar, JSON *json)
557{ 662{
558 enc_t enc; 663 enc_t enc;
559 664
560 if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar)) 665 if (!(json->flags & F_ALLOW_NONREF) && !SvROK (scalar))
561 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)"); 666 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)");
562 667
563 enc.flags = flags; 668 enc.json = *json;
564 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 669 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
565 enc.cur = SvPVX (enc.sv); 670 enc.cur = SvPVX (enc.sv);
566 enc.end = SvEND (enc.sv); 671 enc.end = SvEND (enc.sv);
567 enc.indent = 0; 672 enc.indent = 0;
568 enc.maxdepth = DEC_DEPTH (flags); 673 enc.maxdepth = DEC_DEPTH (enc.json.flags);
569 674
570 SvPOK_only (enc.sv); 675 SvPOK_only (enc.sv);
571 encode_sv (&enc, scalar); 676 encode_sv (&enc, scalar);
572 677
573 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 678 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
574 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings 679 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
575 680
576 if (!(flags & (F_ASCII | F_LATIN1 | F_UTF8))) 681 if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8)))
577 SvUTF8_on (enc.sv); 682 SvUTF8_on (enc.sv);
578 683
579 if (enc.flags & F_SHRINK) 684 if (enc.json.flags & F_SHRINK)
580 shrink (enc.sv); 685 shrink (enc.sv);
581 686
582 return enc.sv; 687 return enc.sv;
583} 688}
584 689
589typedef struct 694typedef struct
590{ 695{
591 char *cur; // current parser pointer 696 char *cur; // current parser pointer
592 char *end; // end of input string 697 char *end; // end of input string
593 const char *err; // parse error, if != 0 698 const char *err; // parse error, if != 0
594 U32 flags; // F_* 699 JSON json;
595 U32 depth; // recursion depth 700 U32 depth; // recursion depth
596 U32 maxdepth; // recursion depth limit 701 U32 maxdepth; // recursion depth limit
597} dec_t; 702} dec_t;
598 703
599inline void 704inline void
604 char ch = *dec->cur; 709 char ch = *dec->cur;
605 710
606 if (ch > 0x20 711 if (ch > 0x20
607 || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)) 712 || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09))
608 break; 713 break;
714
715 if (ch == '#' && dec->json.flags & F_RELAXED)
716 ++dec->cur;
609 717
610 ++dec->cur; 718 ++dec->cur;
611 } 719 }
612} 720}
613 721
862 is_nv = 1; 970 is_nv = 1;
863 } 971 }
864 972
865 if (!is_nv) 973 if (!is_nv)
866 { 974 {
975 int len = dec->cur - start;
976
867 // special case the rather common 1..4-digit-int case, assumes 32 bit ints or so 977 // special case the rather common 1..4-digit-int case, assumes 32 bit ints or so
868 if (*start == '-') 978 if (*start == '-')
869 switch (dec->cur - start) 979 switch (len)
870 { 980 {
871 case 2: return newSViv (-( start [1] - '0' * 1)); 981 case 2: return newSViv (-( start [1] - '0' * 1));
872 case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11)); 982 case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11));
873 case 4: return newSViv (-( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111)); 983 case 4: return newSViv (-( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111));
874 case 5: return newSViv (-(start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111)); 984 case 5: return newSViv (-(start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111));
875 } 985 }
876 else 986 else
877 switch (dec->cur - start) 987 switch (len)
878 { 988 {
879 case 1: return newSViv ( start [0] - '0' * 1); 989 case 1: return newSViv ( start [0] - '0' * 1);
880 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11); 990 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11);
881 case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111); 991 case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111);
882 case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111); 992 case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111);
883 } 993 }
884 994
885 { 995 {
886 UV uv; 996 UV uv;
887 int numtype = grok_number (start, dec->cur - start, &uv); 997 int numtype = grok_number (start, len, &uv);
888 if (numtype & IS_NUMBER_IN_UV) 998 if (numtype & IS_NUMBER_IN_UV)
889 if (numtype & IS_NUMBER_NEG) 999 if (numtype & IS_NUMBER_NEG)
890 { 1000 {
891 if (uv < (UV)IV_MIN) 1001 if (uv < (UV)IV_MIN)
892 return newSViv (-(IV)uv); 1002 return newSViv (-(IV)uv);
893 } 1003 }
894 else 1004 else
895 return newSVuv (uv); 1005 return newSVuv (uv);
896
897 // here would likely be the place for bigint support
898 } 1006 }
899 }
900 1007
901 // if we ever support bigint or bigfloat, this is the place for bigfloat 1008 len -= *start == '-' ? 1 : 0;
1009
1010 // does not fit into IV or UV, try NV
1011 if ((sizeof (NV) == sizeof (double) && DBL_DIG >= len)
1012 #if defined (LDBL_DIG)
1013 || (sizeof (NV) == sizeof (long double) && LDBL_DIG >= len)
1014 #endif
1015 )
1016 // fits into NV without loss of precision
1017 return newSVnv (Atof (start));
1018
1019 // everything else fails, convert it to a string
1020 return newSVpvn (start, dec->cur - start);
1021 }
1022
1023 // loss of precision here
902 return newSVnv (Atof (start)); 1024 return newSVnv (Atof (start));
903 1025
904fail: 1026fail:
905 return 0; 1027 return 0;
906} 1028}
936 1058
937 if (*dec->cur != ',') 1059 if (*dec->cur != ',')
938 ERR (", or ] expected while parsing array"); 1060 ERR (", or ] expected while parsing array");
939 1061
940 ++dec->cur; 1062 ++dec->cur;
1063
1064 decode_ws (dec);
1065
1066 if (*dec->cur == ']' && dec->json.flags & F_RELAXED)
1067 {
1068 ++dec->cur;
1069 break;
1070 }
941 } 1071 }
942 1072
943 DEC_DEC_DEPTH; 1073 DEC_DEC_DEPTH;
944 return newRV_noinc ((SV *)av); 1074 return newRV_noinc ((SV *)av);
945 1075
950} 1080}
951 1081
952static SV * 1082static SV *
953decode_hv (dec_t *dec) 1083decode_hv (dec_t *dec)
954{ 1084{
1085 SV *sv;
955 HV *hv = newHV (); 1086 HV *hv = newHV ();
956 1087
957 DEC_INC_DEPTH; 1088 DEC_INC_DEPTH;
958 decode_ws (dec); 1089 decode_ws (dec);
959 1090
960 if (*dec->cur == '}') 1091 if (*dec->cur == '}')
961 ++dec->cur; 1092 ++dec->cur;
962 else 1093 else
963 for (;;) 1094 for (;;)
964 { 1095 {
1096 EXPECT_CH ('"');
1097
1098 // heuristic: assume that
1099 // a) decode_str + hv_store_ent are abysmally slow.
1100 // b) most hash keys are short, simple ascii text.
1101 // => try to "fast-match" such strings to avoid
1102 // the overhead of decode_str + hv_store_ent.
1103 {
965 SV *key, *value; 1104 SV *value;
1105 char *p = dec->cur;
1106 char *e = p + 24; // only try up to 24 bytes
966 1107
967 decode_ws (dec); EXPECT_CH ('"'); 1108 for (;;)
968
969 key = decode_str (dec);
970 if (!key)
971 goto fail;
972
973 decode_ws (dec); EXPECT_CH (':');
974
975 value = decode_sv (dec);
976 if (!value)
977 { 1109 {
1110 // the >= 0x80 is true on most architectures
1111 if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\')
1112 {
1113 // slow path, back up and use decode_str
1114 SV *key = decode_str (dec);
1115 if (!key)
1116 goto fail;
1117
1118 decode_ws (dec); EXPECT_CH (':');
1119
1120 decode_ws (dec);
1121 value = decode_sv (dec);
1122 if (!value)
1123 {
1124 SvREFCNT_dec (key);
1125 goto fail;
1126 }
1127
1128 hv_store_ent (hv, key, value, 0);
978 SvREFCNT_dec (key); 1129 SvREFCNT_dec (key);
1130
1131 break;
1132 }
1133 else if (*p == '"')
1134 {
1135 // fast path, got a simple key
1136 char *key = dec->cur;
1137 int len = p - key;
1138 dec->cur = p + 1;
1139
1140 decode_ws (dec); EXPECT_CH (':');
1141
1142 decode_ws (dec);
1143 value = decode_sv (dec);
1144 if (!value)
979 goto fail; 1145 goto fail;
1146
1147 hv_store (hv, key, len, value, 0);
1148
1149 break;
1150 }
1151
1152 ++p;
980 } 1153 }
981 1154 }
982 hv_store_ent (hv, key, value, 0);
983 SvREFCNT_dec (key);
984 1155
985 decode_ws (dec); 1156 decode_ws (dec);
986 1157
987 if (*dec->cur == '}') 1158 if (*dec->cur == '}')
988 { 1159 {
992 1163
993 if (*dec->cur != ',') 1164 if (*dec->cur != ',')
994 ERR (", or } expected while parsing object/hash"); 1165 ERR (", or } expected while parsing object/hash");
995 1166
996 ++dec->cur; 1167 ++dec->cur;
1168
1169 decode_ws (dec);
1170
1171 if (*dec->cur == '}' && dec->json.flags & F_RELAXED)
1172 {
1173 ++dec->cur;
1174 break;
1175 }
997 } 1176 }
998 1177
999 DEC_DEC_DEPTH; 1178 DEC_DEC_DEPTH;
1000 return newRV_noinc ((SV *)hv); 1179 sv = newRV_noinc ((SV *)hv);
1180
1181 // check filter callbacks
1182 if (dec->json.flags & F_HOOK)
1183 {
1184 if (dec->json.cb_sk_object && HvKEYS (hv) == 1)
1185 {
1186 HE *cb, *he;
1187
1188 hv_iterinit (hv);
1189 he = hv_iternext (hv);
1190 hv_iterinit (hv);
1191
1192 // the next line creates a mortal sv each time its called.
1193 // might want to optimise this for common cases.
1194 cb = hv_fetch_ent (dec->json.cb_sk_object, hv_iterkeysv (he), 0, 0);
1195
1196 if (cb)
1197 {
1198 dSP;
1199 int count;
1200
1201 ENTER; SAVETMPS; PUSHMARK (SP);
1202 XPUSHs (HeVAL (he));
1203
1204 PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN;
1205
1206 if (count == 1)
1207 {
1208 sv = newSVsv (POPs);
1209 FREETMPS; LEAVE;
1210 return sv;
1211 }
1212
1213 FREETMPS; LEAVE;
1214 }
1215 }
1216
1217 if (dec->json.cb_object)
1218 {
1219 dSP;
1220 int count;
1221
1222 ENTER; SAVETMPS; PUSHMARK (SP);
1223 XPUSHs (sv_2mortal (sv));
1224
1225 PUTBACK; count = call_sv (dec->json.cb_object, G_ARRAY); SPAGAIN;
1226
1227 if (count == 1)
1228 {
1229 sv = newSVsv (POPs);
1230 FREETMPS; LEAVE;
1231 return sv;
1232 }
1233
1234 SvREFCNT_inc (sv);
1235 FREETMPS; LEAVE;
1236 }
1237 }
1238
1239 return sv;
1001 1240
1002fail: 1241fail:
1003 SvREFCNT_dec (hv); 1242 SvREFCNT_dec (hv);
1004 DEC_DEC_DEPTH; 1243 DEC_DEC_DEPTH;
1005 return 0; 1244 return 0;
1006} 1245}
1007 1246
1008static SV * 1247static SV *
1009decode_sv (dec_t *dec) 1248decode_sv (dec_t *dec)
1010{ 1249{
1011 decode_ws (dec);
1012
1013 // the beauty of JSON: you need exactly one character lookahead 1250 // the beauty of JSON: you need exactly one character lookahead
1014 // to parse anything. 1251 // to parse anything.
1015 switch (*dec->cur) 1252 switch (*dec->cur)
1016 { 1253 {
1017 case '"': ++dec->cur; return decode_str (dec); 1254 case '"': ++dec->cur; return decode_str (dec);
1025 1262
1026 case 't': 1263 case 't':
1027 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) 1264 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
1028 { 1265 {
1029 dec->cur += 4; 1266 dec->cur += 4;
1030 return newSViv (1); 1267#if JSON_SLOW
1268 json_true = get_sv ("JSON::XS::true", 1); SvREADONLY_on (json_true);
1269#endif
1270 return SvREFCNT_inc (json_true);
1031 } 1271 }
1032 else 1272 else
1033 ERR ("'true' expected"); 1273 ERR ("'true' expected");
1034 1274
1035 break; 1275 break;
1036 1276
1037 case 'f': 1277 case 'f':
1038 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) 1278 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
1039 { 1279 {
1040 dec->cur += 5; 1280 dec->cur += 5;
1041 return newSViv (0); 1281#if JSON_SLOW
1282 json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false);
1283#endif
1284 return SvREFCNT_inc (json_false);
1042 } 1285 }
1043 else 1286 else
1044 ERR ("'false' expected"); 1287 ERR ("'false' expected");
1045 1288
1046 break; 1289 break;
1064fail: 1307fail:
1065 return 0; 1308 return 0;
1066} 1309}
1067 1310
1068static SV * 1311static SV *
1069decode_json (SV *string, U32 flags, UV *offset_return) 1312decode_json (SV *string, JSON *json, UV *offset_return)
1070{ 1313{
1071 dec_t dec; 1314 dec_t dec;
1072 UV offset; 1315 UV offset;
1073 SV *sv; 1316 SV *sv;
1074 1317
1075 SvGETMAGIC (string); 1318 SvGETMAGIC (string);
1076 SvUPGRADE (string, SVt_PV); 1319 SvUPGRADE (string, SVt_PV);
1077 1320
1321 if (json->flags & F_MAXSIZE && SvCUR (string) > DEC_SIZE (json->flags))
1322 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
1323 (unsigned long)SvCUR (string), (unsigned long)DEC_SIZE (json->flags));
1324
1078 if (flags & F_UTF8) 1325 if (json->flags & F_UTF8)
1079 sv_utf8_downgrade (string, 0); 1326 sv_utf8_downgrade (string, 0);
1080 else 1327 else
1081 sv_utf8_upgrade (string); 1328 sv_utf8_upgrade (string);
1082 1329
1083 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP 1330 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
1084 1331
1085 dec.flags = flags; 1332 dec.json = *json;
1086 dec.cur = SvPVX (string); 1333 dec.cur = SvPVX (string);
1087 dec.end = SvEND (string); 1334 dec.end = SvEND (string);
1088 dec.err = 0; 1335 dec.err = 0;
1089 dec.depth = 0; 1336 dec.depth = 0;
1090 dec.maxdepth = DEC_DEPTH (dec.flags); 1337 dec.maxdepth = DEC_DEPTH (dec.json.flags);
1338
1339 if (dec.json.cb_object || dec.json.cb_sk_object)
1340 dec.json.flags |= F_HOOK;
1091 1341
1092 *dec.end = 0; // this should basically be a nop, too, but make sure it's there 1342 *dec.end = 0; // this should basically be a nop, too, but make sure it's there
1343
1344 decode_ws (&dec);
1093 sv = decode_sv (&dec); 1345 sv = decode_sv (&dec);
1094 1346
1095 if (!(offset_return || !sv)) 1347 if (!(offset_return || !sv))
1096 { 1348 {
1097 // check for trailing garbage 1349 // check for trailing garbage
1105 } 1357 }
1106 } 1358 }
1107 1359
1108 if (offset_return || !sv) 1360 if (offset_return || !sv)
1109 { 1361 {
1110 offset = dec.flags & F_UTF8 1362 offset = dec.json.flags & F_UTF8
1111 ? dec.cur - SvPVX (string) 1363 ? dec.cur - SvPVX (string)
1112 : utf8_distance (dec.cur, SvPVX (string)); 1364 : utf8_distance (dec.cur, SvPVX (string));
1113 1365
1114 if (offset_return) 1366 if (offset_return)
1115 *offset_return = offset; 1367 *offset_return = offset;
1134 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); 1386 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
1135 } 1387 }
1136 1388
1137 sv = sv_2mortal (sv); 1389 sv = sv_2mortal (sv);
1138 1390
1139 if (!(dec.flags & F_ALLOW_NONREF) && !SvROK (sv)) 1391 if (!(dec.json.flags & F_ALLOW_NONREF) && !SvROK (sv))
1140 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)"); 1392 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)");
1141 1393
1142 return sv; 1394 return sv;
1143} 1395}
1144 1396
1156 i >= '0' && i <= '9' ? i - '0' 1408 i >= '0' && i <= '9' ? i - '0'
1157 : i >= 'a' && i <= 'f' ? i - 'a' + 10 1409 : i >= 'a' && i <= 'f' ? i - 'a' + 10
1158 : i >= 'A' && i <= 'F' ? i - 'A' + 10 1410 : i >= 'A' && i <= 'F' ? i - 'A' + 10
1159 : -1; 1411 : -1;
1160 1412
1161 json_stash = gv_stashpv ("JSON::XS", 1); 1413 json_stash = gv_stashpv ("JSON::XS" , 1);
1414 json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1);
1415
1416 json_true = get_sv ("JSON::XS::true" , 1); SvREADONLY_on (json_true );
1417 json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false);
1162} 1418}
1163 1419
1164PROTOTYPES: DISABLE 1420PROTOTYPES: DISABLE
1165 1421
1166SV *new (char *dummy) 1422void CLONE (...)
1167 CODE: 1423 CODE:
1168 RETVAL = sv_bless (newRV_noinc (newSVuv (F_DEFAULT)), json_stash); 1424 json_stash = 0;
1169 OUTPUT: 1425 json_boolean_stash = 0;
1170 RETVAL
1171 1426
1427void new (char *klass)
1428 PPCODE:
1429{
1430 SV *pv = NEWSV (0, sizeof (JSON));
1431 SvPOK_only (pv);
1432 Zero (SvPVX (pv), 1, JSON);
1433 ((JSON *)SvPVX (pv))->flags = F_DEFAULT;
1434 XPUSHs (sv_2mortal (sv_bless (newRV_noinc (pv), JSON_STASH)));
1435}
1436
1172SV *ascii (SV *self, int enable = 1) 1437void ascii (JSON *self, int enable = 1)
1173 ALIAS: 1438 ALIAS:
1174 ascii = F_ASCII 1439 ascii = F_ASCII
1175 latin1 = F_LATIN1 1440 latin1 = F_LATIN1
1176 utf8 = F_UTF8 1441 utf8 = F_UTF8
1177 indent = F_INDENT 1442 indent = F_INDENT
1178 canonical = F_CANONICAL 1443 canonical = F_CANONICAL
1179 space_before = F_SPACE_BEFORE 1444 space_before = F_SPACE_BEFORE
1180 space_after = F_SPACE_AFTER 1445 space_after = F_SPACE_AFTER
1181 pretty = F_PRETTY 1446 pretty = F_PRETTY
1182 allow_nonref = F_ALLOW_NONREF 1447 allow_nonref = F_ALLOW_NONREF
1183 shrink = F_SHRINK 1448 shrink = F_SHRINK
1449 allow_blessed = F_ALLOW_BLESSED
1450 convert_blessed = F_CONV_BLESSED
1451 relaxed = F_RELAXED
1184 CODE: 1452 PPCODE:
1185{ 1453{
1186 UV *uv = SvJSON (self);
1187 if (enable) 1454 if (enable)
1188 *uv |= ix; 1455 self->flags |= ix;
1189 else 1456 else
1190 *uv &= ~ix; 1457 self->flags &= ~ix;
1191 1458
1192 RETVAL = newSVsv (self); 1459 XPUSHs (ST (0));
1193} 1460}
1194 OUTPUT:
1195 RETVAL
1196 1461
1197SV *max_depth (SV *self, UV max_depth = 0x80000000UL) 1462void max_depth (JSON *self, UV max_depth = 0x80000000UL)
1198 CODE: 1463 PPCODE:
1199{ 1464{
1200 UV *uv = SvJSON (self);
1201 UV log2 = 0; 1465 UV log2 = 0;
1202 1466
1203 if (max_depth > 0x80000000UL) max_depth = 0x80000000UL; 1467 if (max_depth > 0x80000000UL) max_depth = 0x80000000UL;
1204 1468
1205 while ((1UL << log2) < max_depth) 1469 while ((1UL << log2) < max_depth)
1206 ++log2; 1470 ++log2;
1207 1471
1208 *uv = *uv & ~F_MAXDEPTH | (log2 << S_MAXDEPTH); 1472 self->flags = self->flags & ~F_MAXDEPTH | (log2 << S_MAXDEPTH);
1209 1473
1210 RETVAL = newSVsv (self); 1474 XPUSHs (ST (0));
1211} 1475}
1212 OUTPUT:
1213 RETVAL
1214 1476
1215void encode (SV *self, SV *scalar) 1477void max_size (JSON *self, UV max_size = 0)
1216 PPCODE: 1478 PPCODE:
1217 XPUSHs (encode_json (scalar, *SvJSON (self))); 1479{
1480 UV log2 = 0;
1218 1481
1219void decode (SV *self, SV *jsonstr) 1482 if (max_size > 0x80000000UL) max_size = 0x80000000UL;
1483 if (max_size == 1) max_size = 2;
1484
1485 while ((1UL << log2) < max_size)
1486 ++log2;
1487
1488 self->flags = self->flags & ~F_MAXSIZE | (log2 << S_MAXSIZE);
1489
1490 XPUSHs (ST (0));
1491}
1492
1493void filter_json_object (JSON *self, SV *cb = &PL_sv_undef)
1220 PPCODE: 1494 PPCODE:
1495{
1496 SvREFCNT_dec (self->cb_object);
1497 self->cb_object = SvOK (cb) ? newSVsv (cb) : 0;
1498
1499 XPUSHs (ST (0));
1500}
1501
1502void filter_json_single_key_object (JSON *self, SV *key, SV *cb = &PL_sv_undef)
1503 PPCODE:
1504{
1505 if (!self->cb_sk_object)
1506 self->cb_sk_object = newHV ();
1507
1508 if (SvOK (cb))
1509 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);
1510 else
1511 {
1512 hv_delete_ent (self->cb_sk_object, key, G_DISCARD, 0);
1513
1514 if (!HvKEYS (self->cb_sk_object))
1515 {
1516 SvREFCNT_dec (self->cb_sk_object);
1517 self->cb_sk_object = 0;
1518 }
1519 }
1520
1521 XPUSHs (ST (0));
1522}
1523
1524void encode (JSON *self, SV *scalar)
1525 PPCODE:
1526 XPUSHs (encode_json (scalar, self));
1527
1528void decode (JSON *self, SV *jsonstr)
1529 PPCODE:
1221 XPUSHs (decode_json (jsonstr, *SvJSON (self), 0)); 1530 XPUSHs (decode_json (jsonstr, self, 0));
1222 1531
1223void decode_prefix (SV *self, SV *jsonstr) 1532void decode_prefix (JSON *self, SV *jsonstr)
1224 PPCODE: 1533 PPCODE:
1225{ 1534{
1226 UV offset; 1535 UV offset;
1227 EXTEND (SP, 2); 1536 EXTEND (SP, 2);
1228 PUSHs (decode_json (jsonstr, *SvJSON (self), &offset)); 1537 PUSHs (decode_json (jsonstr, self, &offset));
1229 PUSHs (sv_2mortal (newSVuv (offset))); 1538 PUSHs (sv_2mortal (newSVuv (offset)));
1230} 1539}
1231 1540
1541void DESTROY (JSON *self)
1542 CODE:
1543 SvREFCNT_dec (self->cb_sk_object);
1544 SvREFCNT_dec (self->cb_object);
1545
1232PROTOTYPES: ENABLE 1546PROTOTYPES: ENABLE
1233 1547
1234void to_json (SV *scalar) 1548void to_json (SV *scalar)
1235 ALIAS:
1236 objToJson = 0
1237 PPCODE: 1549 PPCODE:
1550{
1551 JSON json = { F_DEFAULT | F_UTF8 };
1238 XPUSHs (encode_json (scalar, F_DEFAULT | F_UTF8)); 1552 XPUSHs (encode_json (scalar, &json));
1553}
1239 1554
1240void from_json (SV *jsonstr) 1555void from_json (SV *jsonstr)
1241 ALIAS:
1242 jsonToObj = 0
1243 PPCODE: 1556 PPCODE:
1557{
1558 JSON json = { F_DEFAULT | F_UTF8 };
1244 XPUSHs (decode_json (jsonstr, F_DEFAULT | F_UTF8, 0)); 1559 XPUSHs (decode_json (jsonstr, &json, 0));
1560}
1245 1561

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines