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 1.44 by root, Mon Jun 25 04:08:17 2007 UTC vs.
Revision 1.64 by root, Mon Aug 27 02:03:23 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
25#define F_SPACE_BEFORE 0x00000020UL 26#define F_SPACE_BEFORE 0x00000020UL
26#define F_SPACE_AFTER 0x00000040UL 27#define F_SPACE_AFTER 0x00000040UL
27#define F_ALLOW_NONREF 0x00000100UL 28#define F_ALLOW_NONREF 0x00000100UL
28#define F_SHRINK 0x00000200UL 29#define F_SHRINK 0x00000200UL
29#define F_ALLOW_BLESSED 0x00000400UL 30#define F_ALLOW_BLESSED 0x00000400UL
30#define F_CONV_BLESSED 0x00000800UL // NYI 31#define F_CONV_BLESSED 0x00000800UL
32#define F_RELAXED 0x00001000UL
33
31#define F_MAXDEPTH 0xf8000000UL 34#define F_MAXDEPTH 0xf8000000UL
32#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
33 39
34#define DEC_DEPTH(flags) (1UL << ((flags & F_MAXDEPTH) >> S_MAXDEPTH)) 40#define DEC_DEPTH(flags) (1UL << ((flags & F_MAXDEPTH) >> S_MAXDEPTH))
41#define DEC_SIZE(flags) (1UL << ((flags & F_MAXSIZE ) >> S_MAXSIZE ))
35 42
36#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER 43#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER
37#define F_DEFAULT (9UL << S_MAXDEPTH) 44#define F_DEFAULT (9UL << S_MAXDEPTH)
38 45
39#define INIT_SIZE 32 // initial scalar size to be allocated 46#define INIT_SIZE 32 // initial scalar size to be allocated
53#endif 60#endif
54 61
55#define expect_false(expr) expect ((expr) != 0, 0) 62#define expect_false(expr) expect ((expr) != 0, 0)
56#define expect_true(expr) expect ((expr) != 0, 1) 63#define expect_true(expr) expect ((expr) != 0, 1)
57 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
58static HV *json_stash, *json_boolean_stash; // JSON::XS:: 73static HV *json_stash, *json_boolean_stash; // JSON::XS::
59static SV *json_true, *json_false; 74static SV *json_true, *json_false;
60 75
76typedef struct {
77 U32 flags;
78 SV *cb_object;
79 HV *cb_sk_object;
80} JSON;
81
61///////////////////////////////////////////////////////////////////////////// 82/////////////////////////////////////////////////////////////////////////////
62// utility functions 83// utility functions
63 84
64static UV * 85inline void
65SvJSON (SV *sv)
66{
67 if (!(SvROK (sv) && SvOBJECT (SvRV (sv)) && SvSTASH (SvRV (sv)) == json_stash))
68 croak ("object is not of type JSON::XS");
69
70 return &SvUVX (SvRV (sv));
71}
72
73static void
74shrink (SV *sv) 86shrink (SV *sv)
75{ 87{
76 sv_utf8_downgrade (sv, 1); 88 sv_utf8_downgrade (sv, 1);
77 if (SvLEN (sv) > SvCUR (sv) + 1) 89 if (SvLEN (sv) > SvCUR (sv) + 1)
78 { 90 {
113typedef struct 125typedef struct
114{ 126{
115 char *cur; // SvPVX (sv) + current output position 127 char *cur; // SvPVX (sv) + current output position
116 char *end; // SvEND (sv) 128 char *end; // SvEND (sv)
117 SV *sv; // result scalar 129 SV *sv; // result scalar
118 U32 flags; // F_* 130 JSON json;
119 U32 indent; // indentation level 131 U32 indent; // indentation level
120 U32 maxdepth; // max. indentation/recursion level 132 U32 maxdepth; // max. indentation/recursion level
121} enc_t; 133} enc_t;
122 134
123inline void 135inline void
197 } 209 }
198 210
199 if (uch > 0x10FFFFUL) 211 if (uch > 0x10FFFFUL)
200 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);
201 213
202 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))
203 { 215 {
204 if (uch > 0xFFFFUL) 216 if (uch > 0xFFFFUL)
205 { 217 {
206 need (enc, len += 11); 218 need (enc, len += 11);
207 sprintf (enc->cur, "\\u%04x\\u%04x", 219 sprintf (enc->cur, "\\u%04x\\u%04x",
221 *enc->cur++ = hexdigit [(uch >> 0) & 15]; 233 *enc->cur++ = hexdigit [(uch >> 0) & 15];
222 } 234 }
223 235
224 str += clen; 236 str += clen;
225 } 237 }
226 else if (enc->flags & F_LATIN1) 238 else if (enc->json.flags & F_LATIN1)
227 { 239 {
228 *enc->cur++ = uch; 240 *enc->cur++ = uch;
229 str += clen; 241 str += clen;
230 } 242 }
231 else if (is_utf8) 243 else if (is_utf8)
252} 264}
253 265
254inline void 266inline void
255encode_indent (enc_t *enc) 267encode_indent (enc_t *enc)
256{ 268{
257 if (enc->flags & F_INDENT) 269 if (enc->json.flags & F_INDENT)
258 { 270 {
259 int spaces = enc->indent * INDENT_STEP; 271 int spaces = enc->indent * INDENT_STEP;
260 272
261 need (enc, spaces); 273 need (enc, spaces);
262 memset (enc->cur, ' ', spaces); 274 memset (enc->cur, ' ', spaces);
272} 284}
273 285
274inline void 286inline void
275encode_nl (enc_t *enc) 287encode_nl (enc_t *enc)
276{ 288{
277 if (enc->flags & F_INDENT) 289 if (enc->json.flags & F_INDENT)
278 { 290 {
279 need (enc, 1); 291 need (enc, 1);
280 encode_ch (enc, '\n'); 292 encode_ch (enc, '\n');
281 } 293 }
282} 294}
284inline void 296inline void
285encode_comma (enc_t *enc) 297encode_comma (enc_t *enc)
286{ 298{
287 encode_ch (enc, ','); 299 encode_ch (enc, ',');
288 300
289 if (enc->flags & F_INDENT) 301 if (enc->json.flags & F_INDENT)
290 encode_nl (enc); 302 encode_nl (enc);
291 else if (enc->flags & F_SPACE_AFTER) 303 else if (enc->json.flags & F_SPACE_AFTER)
292 encode_space (enc); 304 encode_space (enc);
293} 305}
294 306
295static void encode_sv (enc_t *enc, SV *sv); 307static void encode_sv (enc_t *enc, SV *sv);
296 308
305 encode_ch (enc, '['); encode_nl (enc); 317 encode_ch (enc, '['); encode_nl (enc);
306 ++enc->indent; 318 ++enc->indent;
307 319
308 for (i = 0; i <= len; ++i) 320 for (i = 0; i <= len; ++i)
309 { 321 {
322 SV **svp = av_fetch (av, i, 0);
323
310 encode_indent (enc); 324 encode_indent (enc);
311 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);
312 330
313 if (i < len) 331 if (i < len)
314 encode_comma (enc); 332 encode_comma (enc);
315 } 333 }
316 334
319 --enc->indent; 337 --enc->indent;
320 encode_indent (enc); encode_ch (enc, ']'); 338 encode_indent (enc); encode_ch (enc, ']');
321} 339}
322 340
323static void 341static void
324encode_he (enc_t *enc, HE *he) 342encode_hk (enc_t *enc, HE *he)
325{ 343{
326 encode_ch (enc, '"'); 344 encode_ch (enc, '"');
327 345
328 if (HeKLEN (he) == HEf_SVKEY) 346 if (HeKLEN (he) == HEf_SVKEY)
329 { 347 {
339 else 357 else
340 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he)); 358 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he));
341 359
342 encode_ch (enc, '"'); 360 encode_ch (enc, '"');
343 361
344 if (enc->flags & F_SPACE_BEFORE) encode_space (enc); 362 if (enc->json.flags & F_SPACE_BEFORE) encode_space (enc);
345 encode_ch (enc, ':'); 363 encode_ch (enc, ':');
346 if (enc->flags & F_SPACE_AFTER ) encode_space (enc); 364 if (enc->json.flags & F_SPACE_AFTER ) encode_space (enc);
347 encode_sv (enc, HeVAL (he));
348} 365}
349 366
350// compare hash entries, used when all keys are bytestrings 367// compare hash entries, used when all keys are bytestrings
351static int 368static int
352he_cmp_fast (const void *a_, const void *b_) 369he_cmp_fast (const void *a_, const void *b_)
357 HE *b = *(HE **)b_; 374 HE *b = *(HE **)b_;
358 375
359 STRLEN la = HeKLEN (a); 376 STRLEN la = HeKLEN (a);
360 STRLEN lb = HeKLEN (b); 377 STRLEN lb = HeKLEN (b);
361 378
362 if (!(cmp = memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb))) 379 if (!(cmp = memcmp (HeKEY (b), HeKEY (a), lb < la ? lb : la)))
363 cmp = la - lb; 380 cmp = lb - la;
364 381
365 return cmp; 382 return cmp;
366} 383}
367 384
368// 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
369static int 386static int
370he_cmp_slow (const void *a, const void *b) 387he_cmp_slow (const void *a, const void *b)
371{ 388{
372 return sv_cmp (HeSVKEY_force (*(HE **)a), HeSVKEY_force (*(HE **)b)); 389 return sv_cmp (HeSVKEY_force (*(HE **)b), HeSVKEY_force (*(HE **)a));
373} 390}
374 391
375static void 392static void
376encode_hv (enc_t *enc, HV *hv) 393encode_hv (enc_t *enc, HV *hv)
377{ 394{
395 HE *he;
378 int count, i; 396 int count;
379 397
380 if (enc->indent >= enc->maxdepth) 398 if (enc->indent >= enc->maxdepth)
381 croak ("data structure too deep (hit recursion limit)"); 399 croak ("data structure too deep (hit recursion limit)");
382 400
383 encode_ch (enc, '{'); encode_nl (enc); ++enc->indent; 401 encode_ch (enc, '{'); encode_nl (enc); ++enc->indent;
384 402
385 if ((count = hv_iterinit (hv)))
386 {
387 // for canonical output we have to sort by keys first 403 // for canonical output we have to sort by keys first
388 // actually, this is mostly due to the stupid so-called 404 // actually, this is mostly due to the stupid so-called
389 // security workaround added somewhere in 5.8.x. 405 // security workaround added somewhere in 5.8.x.
390 // that randomises hash orderings 406 // that randomises hash orderings
391 if (enc->flags & F_CANONICAL) 407 if (enc->json.flags & F_CANONICAL)
408 {
409 int count = hv_iterinit (hv);
410
411 if (SvMAGICAL (hv))
392 { 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 {
393 int fast = 1; 427 int i, fast = 1;
394 HE *he;
395#if defined(__BORLANDC__) || defined(_MSC_VER) 428#if defined(__BORLANDC__) || defined(_MSC_VER)
396 HE **hes = _alloca (count * sizeof (HE)); 429 HE **hes = _alloca (count * sizeof (HE));
397#else 430#else
398 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
399#endif 432#endif
426 459
427 FREETMPS; 460 FREETMPS;
428 LEAVE; 461 LEAVE;
429 } 462 }
430 463
431 for (i = 0; i < count; ++i) 464 while (count--)
432 { 465 {
433 encode_indent (enc); 466 encode_indent (enc);
467 he = hes [count];
434 encode_he (enc, hes [i]); 468 encode_hk (enc, he);
469 encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he));
435 470
436 if (i < count - 1) 471 if (count)
437 encode_comma (enc); 472 encode_comma (enc);
438 } 473 }
439
440 encode_nl (enc);
441 } 474 }
475 }
442 else 476 else
443 { 477 {
478 if (hv_iterinit (hv) || SvMAGICAL (hv))
444 HE *he = hv_iternext (hv); 479 if ((he = hv_iternext (hv)))
445
446 for (;;) 480 for (;;)
447 { 481 {
448 encode_indent (enc); 482 encode_indent (enc);
449 encode_he (enc, he); 483 encode_hk (enc, he);
484 encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he));
450 485
451 if (!(he = hv_iternext (hv))) 486 if (!(he = hv_iternext (hv)))
452 break; 487 break;
453 488
454 encode_comma (enc); 489 encode_comma (enc);
455 } 490 }
491 }
456 492
457 encode_nl (enc); 493 encode_nl (enc);
458 }
459 }
460 494
461 --enc->indent; encode_indent (enc); encode_ch (enc, '}'); 495 --enc->indent; encode_indent (enc); encode_ch (enc, '}');
462} 496}
463 497
464// encode objects, arrays and special \0=false and \1=true values. 498// encode objects, arrays and special \0=false and \1=true values.
470 SvGETMAGIC (sv); 504 SvGETMAGIC (sv);
471 svt = SvTYPE (sv); 505 svt = SvTYPE (sv);
472 506
473 if (expect_false (SvOBJECT (sv))) 507 if (expect_false (SvOBJECT (sv)))
474 { 508 {
509 HV *stash = !JSON_SLOW || json_boolean_stash
510 ? json_boolean_stash
511 : gv_stashpv ("JSON::XS::Boolean", 1);
512
475 if (SvSTASH (sv) == json_boolean_stash) 513 if (SvSTASH (sv) == stash)
476 { 514 {
477 if (SvIV (sv) == 0) 515 if (SvIV (sv))
516 encode_str (enc, "true", 4, 0);
517 else
478 encode_str (enc, "false", 5, 0); 518 encode_str (enc, "false", 5, 0);
479 else
480 encode_str (enc, "true", 4, 0);
481 } 519 }
482 else 520 else
483 { 521 {
484#if 0 522#if 0
485 if (0 && sv_derived_from (rv, "JSON::Literal")) 523 if (0 && sv_derived_from (rv, "JSON::Literal"))
486 { 524 {
487 // not yet 525 // not yet
488 } 526 }
489#endif 527#endif
490 if (enc->flags & F_CONV_BLESSED) 528 if (enc->json.flags & F_CONV_BLESSED)
491 { 529 {
492 // we re-bless the reference to get overload and other niceties right 530 // we re-bless the reference to get overload and other niceties right
493 GV *to_json = gv_fetchmethod_autoload (SvSTASH (sv), "TO_JSON", 1); 531 GV *to_json = gv_fetchmethod_autoload (SvSTASH (sv), "TO_JSON", 0);
494 532
495 if (to_json) 533 if (to_json)
496 { 534 {
497 dSP; 535 dSP;
498 ENTER; 536
499 SAVETMPS;
500 PUSHMARK (SP); 537 ENTER; SAVETMPS; PUSHMARK (SP);
501 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv))); 538 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv)));
502 539
503 // calling with G_SCALAR ensures that we always get a 1 reutrn value 540 // calling with G_SCALAR ensures that we always get a 1 return value
504 // check anyways.
505 PUTBACK; 541 PUTBACK;
506 assert (1 == call_sv ((SV *)GvCV (to_json), G_SCALAR)); 542 call_sv ((SV *)GvCV (to_json), G_SCALAR);
507 SPAGAIN; 543 SPAGAIN;
508 544
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
509 encode_sv (enc, POPs); 552 encode_sv (enc, sv);
510 553
511 FREETMPS; 554 FREETMPS; LEAVE;
512 LEAVE;
513 } 555 }
514 else if (enc->flags & F_ALLOW_BLESSED) 556 else if (enc->json.flags & F_ALLOW_BLESSED)
515 encode_str (enc, "null", 4, 0); 557 encode_str (enc, "null", 4, 0);
516 else 558 else
517 croak ("encountered object '%s', but neither allow_blessed enabled nor TO_JSON method available on it", 559 croak ("encountered object '%s', but neither allow_blessed enabled nor TO_JSON method available on it",
518 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 560 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
519 } 561 }
520 else if (enc->flags & F_ALLOW_BLESSED) 562 else if (enc->json.flags & F_ALLOW_BLESSED)
521 encode_str (enc, "null", 4, 0); 563 encode_str (enc, "null", 4, 0);
522 else 564 else
523 croak ("encountered object '%s', but neither allow_blessed nor convert_blessed settings are enabled", 565 croak ("encountered object '%s', but neither allow_blessed nor convert_blessed settings are enabled",
524 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 566 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
525 } 567 }
528 encode_hv (enc, (HV *)sv); 570 encode_hv (enc, (HV *)sv);
529 else if (svt == SVt_PVAV) 571 else if (svt == SVt_PVAV)
530 encode_av (enc, (AV *)sv); 572 encode_av (enc, (AV *)sv);
531 else if (svt < SVt_PVAV) 573 else if (svt < SVt_PVAV)
532 { 574 {
533 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')
534 encode_str (enc, "false", 5, 0); 581 encode_str (enc, "false", 5, 0);
535 else if (SvNIOK (sv) && SvIV (sv) == 1)
536 encode_str (enc, "true", 4, 0);
537 else 582 else
538 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",
539 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 584 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
540 } 585 }
541 else 586 else
611 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",
612 SvPV_nolen (sv), SvFLAGS (sv)); 657 SvPV_nolen (sv), SvFLAGS (sv));
613} 658}
614 659
615static SV * 660static SV *
616encode_json (SV *scalar, U32 flags) 661encode_json (SV *scalar, JSON *json)
617{ 662{
618 enc_t enc; 663 enc_t enc;
619 664
620 if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar)) 665 if (!(json->flags & F_ALLOW_NONREF) && !SvROK (scalar))
621 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)");
622 667
623 enc.flags = flags; 668 enc.json = *json;
624 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 669 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
625 enc.cur = SvPVX (enc.sv); 670 enc.cur = SvPVX (enc.sv);
626 enc.end = SvEND (enc.sv); 671 enc.end = SvEND (enc.sv);
627 enc.indent = 0; 672 enc.indent = 0;
628 enc.maxdepth = DEC_DEPTH (flags); 673 enc.maxdepth = DEC_DEPTH (enc.json.flags);
629 674
630 SvPOK_only (enc.sv); 675 SvPOK_only (enc.sv);
631 encode_sv (&enc, scalar); 676 encode_sv (&enc, scalar);
632 677
633 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 678 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
634 *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
635 680
636 if (!(flags & (F_ASCII | F_LATIN1 | F_UTF8))) 681 if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8)))
637 SvUTF8_on (enc.sv); 682 SvUTF8_on (enc.sv);
638 683
639 if (enc.flags & F_SHRINK) 684 if (enc.json.flags & F_SHRINK)
640 shrink (enc.sv); 685 shrink (enc.sv);
641 686
642 return enc.sv; 687 return enc.sv;
643} 688}
644 689
649typedef struct 694typedef struct
650{ 695{
651 char *cur; // current parser pointer 696 char *cur; // current parser pointer
652 char *end; // end of input string 697 char *end; // end of input string
653 const char *err; // parse error, if != 0 698 const char *err; // parse error, if != 0
654 U32 flags; // F_* 699 JSON json;
655 U32 depth; // recursion depth 700 U32 depth; // recursion depth
656 U32 maxdepth; // recursion depth limit 701 U32 maxdepth; // recursion depth limit
657} dec_t; 702} dec_t;
658 703
659inline void 704inline void
705decode_comment (dec_t *dec)
706{
707 // only '#'-style comments allowed a.t.m.
708
709 while (*dec->cur && *dec->cur != 0x0a && *dec->cur != 0x0d)
710 ++dec->cur;
711}
712
713inline void
660decode_ws (dec_t *dec) 714decode_ws (dec_t *dec)
661{ 715{
662 for (;;) 716 for (;;)
663 { 717 {
664 char ch = *dec->cur; 718 char ch = *dec->cur;
665 719
666 if (ch > 0x20 720 if (ch > 0x20)
721 {
722 if (expect_false (ch == '#'))
723 {
724 if (dec->json.flags & F_RELAXED)
725 decode_comment (dec);
726 else
727 break;
728 }
729 else
730 break;
731 }
667 || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)) 732 else if (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)
668 break; 733 break; // parse error, but let higher level handle it, gives better error messages
669 734
670 ++dec->cur; 735 ++dec->cur;
671 } 736 }
672} 737}
673 738
922 is_nv = 1; 987 is_nv = 1;
923 } 988 }
924 989
925 if (!is_nv) 990 if (!is_nv)
926 { 991 {
992 int len = dec->cur - start;
993
927 // special case the rather common 1..4-digit-int case, assumes 32 bit ints or so 994 // special case the rather common 1..4-digit-int case, assumes 32 bit ints or so
928 if (*start == '-') 995 if (*start == '-')
929 switch (dec->cur - start) 996 switch (len)
930 { 997 {
931 case 2: return newSViv (-( start [1] - '0' * 1)); 998 case 2: return newSViv (-( start [1] - '0' * 1));
932 case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11)); 999 case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11));
933 case 4: return newSViv (-( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111)); 1000 case 4: return newSViv (-( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111));
934 case 5: return newSViv (-(start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111)); 1001 case 5: return newSViv (-(start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111));
935 } 1002 }
936 else 1003 else
937 switch (dec->cur - start) 1004 switch (len)
938 { 1005 {
939 case 1: return newSViv ( start [0] - '0' * 1); 1006 case 1: return newSViv ( start [0] - '0' * 1);
940 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11); 1007 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11);
941 case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111); 1008 case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111);
942 case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111); 1009 case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111);
943 } 1010 }
944 1011
945 { 1012 {
946 UV uv; 1013 UV uv;
947 int numtype = grok_number (start, dec->cur - start, &uv); 1014 int numtype = grok_number (start, len, &uv);
948 if (numtype & IS_NUMBER_IN_UV) 1015 if (numtype & IS_NUMBER_IN_UV)
949 if (numtype & IS_NUMBER_NEG) 1016 if (numtype & IS_NUMBER_NEG)
950 { 1017 {
951 if (uv < (UV)IV_MIN) 1018 if (uv < (UV)IV_MIN)
952 return newSViv (-(IV)uv); 1019 return newSViv (-(IV)uv);
953 } 1020 }
954 else 1021 else
955 return newSVuv (uv); 1022 return newSVuv (uv);
956
957 // here would likely be the place for bigint support
958 } 1023 }
959 }
960 1024
961 // if we ever support bigint or bigfloat, this is the place for bigfloat 1025 len -= *start == '-' ? 1 : 0;
1026
1027 // does not fit into IV or UV, try NV
1028 if ((sizeof (NV) == sizeof (double) && DBL_DIG >= len)
1029 #if defined (LDBL_DIG)
1030 || (sizeof (NV) == sizeof (long double) && LDBL_DIG >= len)
1031 #endif
1032 )
1033 // fits into NV without loss of precision
1034 return newSVnv (Atof (start));
1035
1036 // everything else fails, convert it to a string
1037 return newSVpvn (start, dec->cur - start);
1038 }
1039
1040 // loss of precision here
962 return newSVnv (Atof (start)); 1041 return newSVnv (Atof (start));
963 1042
964fail: 1043fail:
965 return 0; 1044 return 0;
966} 1045}
996 1075
997 if (*dec->cur != ',') 1076 if (*dec->cur != ',')
998 ERR (", or ] expected while parsing array"); 1077 ERR (", or ] expected while parsing array");
999 1078
1000 ++dec->cur; 1079 ++dec->cur;
1080
1081 decode_ws (dec);
1082
1083 if (*dec->cur == ']' && dec->json.flags & F_RELAXED)
1084 {
1085 ++dec->cur;
1086 break;
1087 }
1001 } 1088 }
1002 1089
1003 DEC_DEC_DEPTH; 1090 DEC_DEC_DEPTH;
1004 return newRV_noinc ((SV *)av); 1091 return newRV_noinc ((SV *)av);
1005 1092
1010} 1097}
1011 1098
1012static SV * 1099static SV *
1013decode_hv (dec_t *dec) 1100decode_hv (dec_t *dec)
1014{ 1101{
1102 SV *sv;
1015 HV *hv = newHV (); 1103 HV *hv = newHV ();
1016 1104
1017 DEC_INC_DEPTH; 1105 DEC_INC_DEPTH;
1018 decode_ws (dec); 1106 decode_ws (dec);
1019 1107
1020 if (*dec->cur == '}') 1108 if (*dec->cur == '}')
1021 ++dec->cur; 1109 ++dec->cur;
1022 else 1110 else
1023 for (;;) 1111 for (;;)
1024 { 1112 {
1113 EXPECT_CH ('"');
1114
1115 // heuristic: assume that
1116 // a) decode_str + hv_store_ent are abysmally slow.
1117 // b) most hash keys are short, simple ascii text.
1118 // => try to "fast-match" such strings to avoid
1119 // the overhead of decode_str + hv_store_ent.
1120 {
1025 SV *key, *value; 1121 SV *value;
1122 char *p = dec->cur;
1123 char *e = p + 24; // only try up to 24 bytes
1026 1124
1027 decode_ws (dec); EXPECT_CH ('"'); 1125 for (;;)
1028
1029 key = decode_str (dec);
1030 if (!key)
1031 goto fail;
1032
1033 decode_ws (dec); EXPECT_CH (':');
1034
1035 value = decode_sv (dec);
1036 if (!value)
1037 { 1126 {
1127 // the >= 0x80 is true on most architectures
1128 if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\')
1129 {
1130 // slow path, back up and use decode_str
1131 SV *key = decode_str (dec);
1132 if (!key)
1133 goto fail;
1134
1135 decode_ws (dec); EXPECT_CH (':');
1136
1137 decode_ws (dec);
1138 value = decode_sv (dec);
1139 if (!value)
1140 {
1141 SvREFCNT_dec (key);
1142 goto fail;
1143 }
1144
1145 hv_store_ent (hv, key, value, 0);
1038 SvREFCNT_dec (key); 1146 SvREFCNT_dec (key);
1147
1148 break;
1149 }
1150 else if (*p == '"')
1151 {
1152 // fast path, got a simple key
1153 char *key = dec->cur;
1154 int len = p - key;
1155 dec->cur = p + 1;
1156
1157 decode_ws (dec); EXPECT_CH (':');
1158
1159 decode_ws (dec);
1160 value = decode_sv (dec);
1161 if (!value)
1039 goto fail; 1162 goto fail;
1163
1164 hv_store (hv, key, len, value, 0);
1165
1166 break;
1167 }
1168
1169 ++p;
1040 } 1170 }
1041 1171 }
1042 hv_store_ent (hv, key, value, 0);
1043 SvREFCNT_dec (key);
1044 1172
1045 decode_ws (dec); 1173 decode_ws (dec);
1046 1174
1047 if (*dec->cur == '}') 1175 if (*dec->cur == '}')
1048 { 1176 {
1052 1180
1053 if (*dec->cur != ',') 1181 if (*dec->cur != ',')
1054 ERR (", or } expected while parsing object/hash"); 1182 ERR (", or } expected while parsing object/hash");
1055 1183
1056 ++dec->cur; 1184 ++dec->cur;
1185
1186 decode_ws (dec);
1187
1188 if (*dec->cur == '}' && dec->json.flags & F_RELAXED)
1189 {
1190 ++dec->cur;
1191 break;
1192 }
1057 } 1193 }
1058 1194
1059 DEC_DEC_DEPTH; 1195 DEC_DEC_DEPTH;
1060 return newRV_noinc ((SV *)hv); 1196 sv = newRV_noinc ((SV *)hv);
1197
1198 // check filter callbacks
1199 if (dec->json.flags & F_HOOK)
1200 {
1201 if (dec->json.cb_sk_object && HvKEYS (hv) == 1)
1202 {
1203 HE *cb, *he;
1204
1205 hv_iterinit (hv);
1206 he = hv_iternext (hv);
1207 hv_iterinit (hv);
1208
1209 // the next line creates a mortal sv each time its called.
1210 // might want to optimise this for common cases.
1211 cb = hv_fetch_ent (dec->json.cb_sk_object, hv_iterkeysv (he), 0, 0);
1212
1213 if (cb)
1214 {
1215 dSP;
1216 int count;
1217
1218 ENTER; SAVETMPS; PUSHMARK (SP);
1219 XPUSHs (HeVAL (he));
1220
1221 PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN;
1222
1223 if (count == 1)
1224 {
1225 sv = newSVsv (POPs);
1226 FREETMPS; LEAVE;
1227 return sv;
1228 }
1229
1230 FREETMPS; LEAVE;
1231 }
1232 }
1233
1234 if (dec->json.cb_object)
1235 {
1236 dSP;
1237 int count;
1238
1239 ENTER; SAVETMPS; PUSHMARK (SP);
1240 XPUSHs (sv_2mortal (sv));
1241
1242 PUTBACK; count = call_sv (dec->json.cb_object, G_ARRAY); SPAGAIN;
1243
1244 if (count == 1)
1245 {
1246 sv = newSVsv (POPs);
1247 FREETMPS; LEAVE;
1248 return sv;
1249 }
1250
1251 SvREFCNT_inc (sv);
1252 FREETMPS; LEAVE;
1253 }
1254 }
1255
1256 return sv;
1061 1257
1062fail: 1258fail:
1063 SvREFCNT_dec (hv); 1259 SvREFCNT_dec (hv);
1064 DEC_DEC_DEPTH; 1260 DEC_DEC_DEPTH;
1065 return 0; 1261 return 0;
1066} 1262}
1067 1263
1068static SV * 1264static SV *
1069decode_sv (dec_t *dec) 1265decode_sv (dec_t *dec)
1070{ 1266{
1071 decode_ws (dec);
1072
1073 // the beauty of JSON: you need exactly one character lookahead 1267 // the beauty of JSON: you need exactly one character lookahead
1074 // to parse anything. 1268 // to parse anything.
1075 switch (*dec->cur) 1269 switch (*dec->cur)
1076 { 1270 {
1077 case '"': ++dec->cur; return decode_str (dec); 1271 case '"': ++dec->cur; return decode_str (dec);
1085 1279
1086 case 't': 1280 case 't':
1087 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) 1281 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
1088 { 1282 {
1089 dec->cur += 4; 1283 dec->cur += 4;
1284#if JSON_SLOW
1285 json_true = get_sv ("JSON::XS::true", 1); SvREADONLY_on (json_true);
1286#endif
1090 return SvREFCNT_inc (json_true); 1287 return SvREFCNT_inc (json_true);
1091 } 1288 }
1092 else 1289 else
1093 ERR ("'true' expected"); 1290 ERR ("'true' expected");
1094 1291
1096 1293
1097 case 'f': 1294 case 'f':
1098 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) 1295 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
1099 { 1296 {
1100 dec->cur += 5; 1297 dec->cur += 5;
1298#if JSON_SLOW
1299 json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false);
1300#endif
1101 return SvREFCNT_inc (json_false); 1301 return SvREFCNT_inc (json_false);
1102 } 1302 }
1103 else 1303 else
1104 ERR ("'false' expected"); 1304 ERR ("'false' expected");
1105 1305
1124fail: 1324fail:
1125 return 0; 1325 return 0;
1126} 1326}
1127 1327
1128static SV * 1328static SV *
1129decode_json (SV *string, U32 flags, UV *offset_return) 1329decode_json (SV *string, JSON *json, UV *offset_return)
1130{ 1330{
1131 dec_t dec; 1331 dec_t dec;
1132 UV offset; 1332 UV offset;
1133 SV *sv; 1333 SV *sv;
1134 1334
1135 SvGETMAGIC (string); 1335 SvGETMAGIC (string);
1136 SvUPGRADE (string, SVt_PV); 1336 SvUPGRADE (string, SVt_PV);
1137 1337
1338 if (json->flags & F_MAXSIZE && SvCUR (string) > DEC_SIZE (json->flags))
1339 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
1340 (unsigned long)SvCUR (string), (unsigned long)DEC_SIZE (json->flags));
1341
1138 if (flags & F_UTF8) 1342 if (json->flags & F_UTF8)
1139 sv_utf8_downgrade (string, 0); 1343 sv_utf8_downgrade (string, 0);
1140 else 1344 else
1141 sv_utf8_upgrade (string); 1345 sv_utf8_upgrade (string);
1142 1346
1143 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP 1347 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
1144 1348
1145 dec.flags = flags; 1349 dec.json = *json;
1146 dec.cur = SvPVX (string); 1350 dec.cur = SvPVX (string);
1147 dec.end = SvEND (string); 1351 dec.end = SvEND (string);
1148 dec.err = 0; 1352 dec.err = 0;
1149 dec.depth = 0; 1353 dec.depth = 0;
1150 dec.maxdepth = DEC_DEPTH (dec.flags); 1354 dec.maxdepth = DEC_DEPTH (dec.json.flags);
1355
1356 if (dec.json.cb_object || dec.json.cb_sk_object)
1357 dec.json.flags |= F_HOOK;
1151 1358
1152 *dec.end = 0; // this should basically be a nop, too, but make sure it's there 1359 *dec.end = 0; // this should basically be a nop, too, but make sure it's there
1360
1361 decode_ws (&dec);
1153 sv = decode_sv (&dec); 1362 sv = decode_sv (&dec);
1154 1363
1155 if (!(offset_return || !sv)) 1364 if (!(offset_return || !sv))
1156 { 1365 {
1157 // check for trailing garbage 1366 // check for trailing garbage
1165 } 1374 }
1166 } 1375 }
1167 1376
1168 if (offset_return || !sv) 1377 if (offset_return || !sv)
1169 { 1378 {
1170 offset = dec.flags & F_UTF8 1379 offset = dec.json.flags & F_UTF8
1171 ? dec.cur - SvPVX (string) 1380 ? dec.cur - SvPVX (string)
1172 : utf8_distance (dec.cur, SvPVX (string)); 1381 : utf8_distance (dec.cur, SvPVX (string));
1173 1382
1174 if (offset_return) 1383 if (offset_return)
1175 *offset_return = offset; 1384 *offset_return = offset;
1194 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); 1403 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
1195 } 1404 }
1196 1405
1197 sv = sv_2mortal (sv); 1406 sv = sv_2mortal (sv);
1198 1407
1199 if (!(dec.flags & F_ALLOW_NONREF) && !SvROK (sv)) 1408 if (!(dec.json.flags & F_ALLOW_NONREF) && !SvROK (sv))
1200 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)"); 1409 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)");
1201 1410
1202 return sv; 1411 return sv;
1203} 1412}
1204 1413
1225 json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false); 1434 json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false);
1226} 1435}
1227 1436
1228PROTOTYPES: DISABLE 1437PROTOTYPES: DISABLE
1229 1438
1230SV *new (char *dummy) 1439void CLONE (...)
1231 CODE: 1440 CODE:
1232 RETVAL = sv_bless (newRV_noinc (newSVuv (F_DEFAULT)), json_stash); 1441 json_stash = 0;
1233 OUTPUT: 1442 json_boolean_stash = 0;
1234 RETVAL
1235 1443
1444void new (char *klass)
1445 PPCODE:
1446{
1447 SV *pv = NEWSV (0, sizeof (JSON));
1448 SvPOK_only (pv);
1449 Zero (SvPVX (pv), 1, JSON);
1450 ((JSON *)SvPVX (pv))->flags = F_DEFAULT;
1451 XPUSHs (sv_2mortal (sv_bless (newRV_noinc (pv), JSON_STASH)));
1452}
1453
1236SV *ascii (SV *self, int enable = 1) 1454void ascii (JSON *self, int enable = 1)
1237 ALIAS: 1455 ALIAS:
1238 ascii = F_ASCII 1456 ascii = F_ASCII
1239 latin1 = F_LATIN1 1457 latin1 = F_LATIN1
1240 utf8 = F_UTF8 1458 utf8 = F_UTF8
1241 indent = F_INDENT 1459 indent = F_INDENT
1245 pretty = F_PRETTY 1463 pretty = F_PRETTY
1246 allow_nonref = F_ALLOW_NONREF 1464 allow_nonref = F_ALLOW_NONREF
1247 shrink = F_SHRINK 1465 shrink = F_SHRINK
1248 allow_blessed = F_ALLOW_BLESSED 1466 allow_blessed = F_ALLOW_BLESSED
1249 convert_blessed = F_CONV_BLESSED 1467 convert_blessed = F_CONV_BLESSED
1468 relaxed = F_RELAXED
1250 CODE: 1469 PPCODE:
1251{ 1470{
1252 UV *uv = SvJSON (self);
1253 if (enable) 1471 if (enable)
1254 *uv |= ix; 1472 self->flags |= ix;
1255 else 1473 else
1256 *uv &= ~ix; 1474 self->flags &= ~ix;
1257 1475
1258 RETVAL = newSVsv (self); 1476 XPUSHs (ST (0));
1259} 1477}
1260 OUTPUT:
1261 RETVAL
1262 1478
1263SV *max_depth (SV *self, UV max_depth = 0x80000000UL) 1479void max_depth (JSON *self, UV max_depth = 0x80000000UL)
1264 CODE: 1480 PPCODE:
1265{ 1481{
1266 UV *uv = SvJSON (self);
1267 UV log2 = 0; 1482 UV log2 = 0;
1268 1483
1269 if (max_depth > 0x80000000UL) max_depth = 0x80000000UL; 1484 if (max_depth > 0x80000000UL) max_depth = 0x80000000UL;
1270 1485
1271 while ((1UL << log2) < max_depth) 1486 while ((1UL << log2) < max_depth)
1272 ++log2; 1487 ++log2;
1273 1488
1274 *uv = *uv & ~F_MAXDEPTH | (log2 << S_MAXDEPTH); 1489 self->flags = self->flags & ~F_MAXDEPTH | (log2 << S_MAXDEPTH);
1275 1490
1276 RETVAL = newSVsv (self); 1491 XPUSHs (ST (0));
1277} 1492}
1278 OUTPUT:
1279 RETVAL
1280 1493
1281void encode (SV *self, SV *scalar) 1494void max_size (JSON *self, UV max_size = 0)
1282 PPCODE: 1495 PPCODE:
1283 XPUSHs (encode_json (scalar, *SvJSON (self))); 1496{
1497 UV log2 = 0;
1284 1498
1285void decode (SV *self, SV *jsonstr) 1499 if (max_size > 0x80000000UL) max_size = 0x80000000UL;
1500 if (max_size == 1) max_size = 2;
1501
1502 while ((1UL << log2) < max_size)
1503 ++log2;
1504
1505 self->flags = self->flags & ~F_MAXSIZE | (log2 << S_MAXSIZE);
1506
1507 XPUSHs (ST (0));
1508}
1509
1510void filter_json_object (JSON *self, SV *cb = &PL_sv_undef)
1286 PPCODE: 1511 PPCODE:
1512{
1513 SvREFCNT_dec (self->cb_object);
1514 self->cb_object = SvOK (cb) ? newSVsv (cb) : 0;
1515
1516 XPUSHs (ST (0));
1517}
1518
1519void filter_json_single_key_object (JSON *self, SV *key, SV *cb = &PL_sv_undef)
1520 PPCODE:
1521{
1522 if (!self->cb_sk_object)
1523 self->cb_sk_object = newHV ();
1524
1525 if (SvOK (cb))
1526 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);
1527 else
1528 {
1529 hv_delete_ent (self->cb_sk_object, key, G_DISCARD, 0);
1530
1531 if (!HvKEYS (self->cb_sk_object))
1532 {
1533 SvREFCNT_dec (self->cb_sk_object);
1534 self->cb_sk_object = 0;
1535 }
1536 }
1537
1538 XPUSHs (ST (0));
1539}
1540
1541void encode (JSON *self, SV *scalar)
1542 PPCODE:
1543 XPUSHs (encode_json (scalar, self));
1544
1545void decode (JSON *self, SV *jsonstr)
1546 PPCODE:
1287 XPUSHs (decode_json (jsonstr, *SvJSON (self), 0)); 1547 XPUSHs (decode_json (jsonstr, self, 0));
1288 1548
1289void decode_prefix (SV *self, SV *jsonstr) 1549void decode_prefix (JSON *self, SV *jsonstr)
1290 PPCODE: 1550 PPCODE:
1291{ 1551{
1292 UV offset; 1552 UV offset;
1293 EXTEND (SP, 2); 1553 EXTEND (SP, 2);
1294 PUSHs (decode_json (jsonstr, *SvJSON (self), &offset)); 1554 PUSHs (decode_json (jsonstr, self, &offset));
1295 PUSHs (sv_2mortal (newSVuv (offset))); 1555 PUSHs (sv_2mortal (newSVuv (offset)));
1296} 1556}
1297 1557
1558void DESTROY (JSON *self)
1559 CODE:
1560 SvREFCNT_dec (self->cb_sk_object);
1561 SvREFCNT_dec (self->cb_object);
1562
1298PROTOTYPES: ENABLE 1563PROTOTYPES: ENABLE
1299 1564
1300void to_json (SV *scalar) 1565void to_json (SV *scalar)
1301 ALIAS:
1302 objToJson = 0
1303 PPCODE: 1566 PPCODE:
1567{
1568 JSON json = { F_DEFAULT | F_UTF8 };
1304 XPUSHs (encode_json (scalar, F_DEFAULT | F_UTF8)); 1569 XPUSHs (encode_json (scalar, &json));
1570}
1305 1571
1306void from_json (SV *jsonstr) 1572void from_json (SV *jsonstr)
1307 ALIAS:
1308 jsonToObj = 0
1309 PPCODE: 1573 PPCODE:
1574{
1575 JSON json = { F_DEFAULT | F_UTF8 };
1310 XPUSHs (decode_json (jsonstr, F_DEFAULT | F_UTF8, 0)); 1576 XPUSHs (decode_json (jsonstr, &json, 0));
1577}
1311 1578

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines