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 1.62 by root, Sun Aug 26 22:27:32 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
31#define F_MAXDEPTH 0xf8000000UL 32#define F_MAXDEPTH 0xf8000000UL
32#define S_MAXDEPTH 27 33#define S_MAXDEPTH 27
34#define F_MAXSIZE 0x01f00000UL
35#define S_MAXSIZE 20
36#define F_HOOK 0x00080000UL // some hooks exist, so slow-path processing
33 37
34#define DEC_DEPTH(flags) (1UL << ((flags & F_MAXDEPTH) >> S_MAXDEPTH)) 38#define DEC_DEPTH(flags) (1UL << ((flags & F_MAXDEPTH) >> S_MAXDEPTH))
39#define DEC_SIZE(flags) (1UL << ((flags & F_MAXSIZE ) >> S_MAXSIZE ))
35 40
36#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER 41#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER
37#define F_DEFAULT (9UL << S_MAXDEPTH) 42#define F_DEFAULT (9UL << S_MAXDEPTH)
38 43
39#define INIT_SIZE 32 // initial scalar size to be allocated 44#define INIT_SIZE 32 // initial scalar size to be allocated
53#endif 58#endif
54 59
55#define expect_false(expr) expect ((expr) != 0, 0) 60#define expect_false(expr) expect ((expr) != 0, 0)
56#define expect_true(expr) expect ((expr) != 0, 1) 61#define expect_true(expr) expect ((expr) != 0, 1)
57 62
63#ifdef USE_ITHREADS
64# define JSON_SLOW 1
65# define JSON_STASH (json_stash ? json_stash : gv_stashpv ("JSON::XS", 1))
66#else
67# define JSON_SLOW 0
68# define JSON_STASH json_stash
69#endif
70
58static HV *json_stash, *json_boolean_stash; // JSON::XS:: 71static HV *json_stash, *json_boolean_stash; // JSON::XS::
59static SV *json_true, *json_false; 72static SV *json_true, *json_false;
60 73
74typedef struct {
75 U32 flags;
76 SV *cb_object;
77 HV *cb_sk_object;
78} JSON;
79
61///////////////////////////////////////////////////////////////////////////// 80/////////////////////////////////////////////////////////////////////////////
62// utility functions 81// utility functions
63 82
64static UV * 83inline 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) 84shrink (SV *sv)
75{ 85{
76 sv_utf8_downgrade (sv, 1); 86 sv_utf8_downgrade (sv, 1);
77 if (SvLEN (sv) > SvCUR (sv) + 1) 87 if (SvLEN (sv) > SvCUR (sv) + 1)
78 { 88 {
113typedef struct 123typedef struct
114{ 124{
115 char *cur; // SvPVX (sv) + current output position 125 char *cur; // SvPVX (sv) + current output position
116 char *end; // SvEND (sv) 126 char *end; // SvEND (sv)
117 SV *sv; // result scalar 127 SV *sv; // result scalar
118 U32 flags; // F_* 128 JSON json;
119 U32 indent; // indentation level 129 U32 indent; // indentation level
120 U32 maxdepth; // max. indentation/recursion level 130 U32 maxdepth; // max. indentation/recursion level
121} enc_t; 131} enc_t;
122 132
123inline void 133inline void
197 } 207 }
198 208
199 if (uch > 0x10FFFFUL) 209 if (uch > 0x10FFFFUL)
200 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch); 210 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
201 211
202 if (uch < 0x80 || enc->flags & F_ASCII || (enc->flags & F_LATIN1 && uch > 0xFF)) 212 if (uch < 0x80 || enc->json.flags & F_ASCII || (enc->json.flags & F_LATIN1 && uch > 0xFF))
203 { 213 {
204 if (uch > 0xFFFFUL) 214 if (uch > 0xFFFFUL)
205 { 215 {
206 need (enc, len += 11); 216 need (enc, len += 11);
207 sprintf (enc->cur, "\\u%04x\\u%04x", 217 sprintf (enc->cur, "\\u%04x\\u%04x",
221 *enc->cur++ = hexdigit [(uch >> 0) & 15]; 231 *enc->cur++ = hexdigit [(uch >> 0) & 15];
222 } 232 }
223 233
224 str += clen; 234 str += clen;
225 } 235 }
226 else if (enc->flags & F_LATIN1) 236 else if (enc->json.flags & F_LATIN1)
227 { 237 {
228 *enc->cur++ = uch; 238 *enc->cur++ = uch;
229 str += clen; 239 str += clen;
230 } 240 }
231 else if (is_utf8) 241 else if (is_utf8)
252} 262}
253 263
254inline void 264inline void
255encode_indent (enc_t *enc) 265encode_indent (enc_t *enc)
256{ 266{
257 if (enc->flags & F_INDENT) 267 if (enc->json.flags & F_INDENT)
258 { 268 {
259 int spaces = enc->indent * INDENT_STEP; 269 int spaces = enc->indent * INDENT_STEP;
260 270
261 need (enc, spaces); 271 need (enc, spaces);
262 memset (enc->cur, ' ', spaces); 272 memset (enc->cur, ' ', spaces);
272} 282}
273 283
274inline void 284inline void
275encode_nl (enc_t *enc) 285encode_nl (enc_t *enc)
276{ 286{
277 if (enc->flags & F_INDENT) 287 if (enc->json.flags & F_INDENT)
278 { 288 {
279 need (enc, 1); 289 need (enc, 1);
280 encode_ch (enc, '\n'); 290 encode_ch (enc, '\n');
281 } 291 }
282} 292}
284inline void 294inline void
285encode_comma (enc_t *enc) 295encode_comma (enc_t *enc)
286{ 296{
287 encode_ch (enc, ','); 297 encode_ch (enc, ',');
288 298
289 if (enc->flags & F_INDENT) 299 if (enc->json.flags & F_INDENT)
290 encode_nl (enc); 300 encode_nl (enc);
291 else if (enc->flags & F_SPACE_AFTER) 301 else if (enc->json.flags & F_SPACE_AFTER)
292 encode_space (enc); 302 encode_space (enc);
293} 303}
294 304
295static void encode_sv (enc_t *enc, SV *sv); 305static void encode_sv (enc_t *enc, SV *sv);
296 306
305 encode_ch (enc, '['); encode_nl (enc); 315 encode_ch (enc, '['); encode_nl (enc);
306 ++enc->indent; 316 ++enc->indent;
307 317
308 for (i = 0; i <= len; ++i) 318 for (i = 0; i <= len; ++i)
309 { 319 {
320 SV **svp = av_fetch (av, i, 0);
321
310 encode_indent (enc); 322 encode_indent (enc);
311 encode_sv (enc, *av_fetch (av, i, 0)); 323
324 if (svp)
325 encode_sv (enc, *svp);
326 else
327 encode_str (enc, "null", 4, 0);
312 328
313 if (i < len) 329 if (i < len)
314 encode_comma (enc); 330 encode_comma (enc);
315 } 331 }
316 332
319 --enc->indent; 335 --enc->indent;
320 encode_indent (enc); encode_ch (enc, ']'); 336 encode_indent (enc); encode_ch (enc, ']');
321} 337}
322 338
323static void 339static void
324encode_he (enc_t *enc, HE *he) 340encode_hk (enc_t *enc, HE *he)
325{ 341{
326 encode_ch (enc, '"'); 342 encode_ch (enc, '"');
327 343
328 if (HeKLEN (he) == HEf_SVKEY) 344 if (HeKLEN (he) == HEf_SVKEY)
329 { 345 {
339 else 355 else
340 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he)); 356 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he));
341 357
342 encode_ch (enc, '"'); 358 encode_ch (enc, '"');
343 359
344 if (enc->flags & F_SPACE_BEFORE) encode_space (enc); 360 if (enc->json.flags & F_SPACE_BEFORE) encode_space (enc);
345 encode_ch (enc, ':'); 361 encode_ch (enc, ':');
346 if (enc->flags & F_SPACE_AFTER ) encode_space (enc); 362 if (enc->json.flags & F_SPACE_AFTER ) encode_space (enc);
347 encode_sv (enc, HeVAL (he));
348} 363}
349 364
350// compare hash entries, used when all keys are bytestrings 365// compare hash entries, used when all keys are bytestrings
351static int 366static int
352he_cmp_fast (const void *a_, const void *b_) 367he_cmp_fast (const void *a_, const void *b_)
357 HE *b = *(HE **)b_; 372 HE *b = *(HE **)b_;
358 373
359 STRLEN la = HeKLEN (a); 374 STRLEN la = HeKLEN (a);
360 STRLEN lb = HeKLEN (b); 375 STRLEN lb = HeKLEN (b);
361 376
362 if (!(cmp = memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb))) 377 if (!(cmp = memcmp (HeKEY (b), HeKEY (a), lb < la ? lb : la)))
363 cmp = la - lb; 378 cmp = lb - la;
364 379
365 return cmp; 380 return cmp;
366} 381}
367 382
368// compare hash entries, used when some keys are sv's or utf-x 383// compare hash entries, used when some keys are sv's or utf-x
369static int 384static int
370he_cmp_slow (const void *a, const void *b) 385he_cmp_slow (const void *a, const void *b)
371{ 386{
372 return sv_cmp (HeSVKEY_force (*(HE **)a), HeSVKEY_force (*(HE **)b)); 387 return sv_cmp (HeSVKEY_force (*(HE **)b), HeSVKEY_force (*(HE **)a));
373} 388}
374 389
375static void 390static void
376encode_hv (enc_t *enc, HV *hv) 391encode_hv (enc_t *enc, HV *hv)
377{ 392{
393 HE *he;
378 int count, i; 394 int count;
379 395
380 if (enc->indent >= enc->maxdepth) 396 if (enc->indent >= enc->maxdepth)
381 croak ("data structure too deep (hit recursion limit)"); 397 croak ("data structure too deep (hit recursion limit)");
382 398
383 encode_ch (enc, '{'); encode_nl (enc); ++enc->indent; 399 encode_ch (enc, '{'); encode_nl (enc); ++enc->indent;
384 400
385 if ((count = hv_iterinit (hv)))
386 {
387 // for canonical output we have to sort by keys first 401 // for canonical output we have to sort by keys first
388 // actually, this is mostly due to the stupid so-called 402 // actually, this is mostly due to the stupid so-called
389 // security workaround added somewhere in 5.8.x. 403 // security workaround added somewhere in 5.8.x.
390 // that randomises hash orderings 404 // that randomises hash orderings
391 if (enc->flags & F_CANONICAL) 405 if (enc->json.flags & F_CANONICAL)
406 {
407 int count = hv_iterinit (hv);
408
409 if (SvMAGICAL (hv))
392 { 410 {
411 // need to count by iterating. could improve by dynamically building the vector below
412 // but I don't care for the speed of this special case.
413 // note also that we will run into undefined behaviour when the two iterations
414 // do not result in the same count, something I might care for in some later release.
415
416 count = 0;
417 while (hv_iternext (hv))
418 ++count;
419
420 hv_iterinit (hv);
421 }
422
423 if (count)
424 {
393 int fast = 1; 425 int i, fast = 1;
394 HE *he;
395#if defined(__BORLANDC__) || defined(_MSC_VER) 426#if defined(__BORLANDC__) || defined(_MSC_VER)
396 HE **hes = _alloca (count * sizeof (HE)); 427 HE **hes = _alloca (count * sizeof (HE));
397#else 428#else
398 HE *hes [count]; // if your compiler dies here, you need to enable C99 mode 429 HE *hes [count]; // if your compiler dies here, you need to enable C99 mode
399#endif 430#endif
426 457
427 FREETMPS; 458 FREETMPS;
428 LEAVE; 459 LEAVE;
429 } 460 }
430 461
431 for (i = 0; i < count; ++i) 462 while (count--)
432 { 463 {
433 encode_indent (enc); 464 encode_indent (enc);
465 he = hes [count];
434 encode_he (enc, hes [i]); 466 encode_hk (enc, he);
467 encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he));
435 468
436 if (i < count - 1) 469 if (count)
437 encode_comma (enc); 470 encode_comma (enc);
438 } 471 }
439
440 encode_nl (enc);
441 } 472 }
473 }
442 else 474 else
443 { 475 {
476 if (hv_iterinit (hv) || SvMAGICAL (hv))
444 HE *he = hv_iternext (hv); 477 if ((he = hv_iternext (hv)))
445
446 for (;;) 478 for (;;)
447 { 479 {
448 encode_indent (enc); 480 encode_indent (enc);
449 encode_he (enc, he); 481 encode_hk (enc, he);
482 encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he));
450 483
451 if (!(he = hv_iternext (hv))) 484 if (!(he = hv_iternext (hv)))
452 break; 485 break;
453 486
454 encode_comma (enc); 487 encode_comma (enc);
455 } 488 }
489 }
456 490
457 encode_nl (enc); 491 encode_nl (enc);
458 }
459 }
460 492
461 --enc->indent; encode_indent (enc); encode_ch (enc, '}'); 493 --enc->indent; encode_indent (enc); encode_ch (enc, '}');
462} 494}
463 495
464// encode objects, arrays and special \0=false and \1=true values. 496// encode objects, arrays and special \0=false and \1=true values.
470 SvGETMAGIC (sv); 502 SvGETMAGIC (sv);
471 svt = SvTYPE (sv); 503 svt = SvTYPE (sv);
472 504
473 if (expect_false (SvOBJECT (sv))) 505 if (expect_false (SvOBJECT (sv)))
474 { 506 {
507 HV *stash = !JSON_SLOW || json_boolean_stash
508 ? json_boolean_stash
509 : gv_stashpv ("JSON::XS::Boolean", 1);
510
475 if (SvSTASH (sv) == json_boolean_stash) 511 if (SvSTASH (sv) == stash)
476 { 512 {
477 if (SvIV (sv) == 0) 513 if (SvIV (sv))
514 encode_str (enc, "true", 4, 0);
515 else
478 encode_str (enc, "false", 5, 0); 516 encode_str (enc, "false", 5, 0);
479 else
480 encode_str (enc, "true", 4, 0);
481 } 517 }
482 else 518 else
483 { 519 {
484#if 0 520#if 0
485 if (0 && sv_derived_from (rv, "JSON::Literal")) 521 if (0 && sv_derived_from (rv, "JSON::Literal"))
486 { 522 {
487 // not yet 523 // not yet
488 } 524 }
489#endif 525#endif
490 if (enc->flags & F_CONV_BLESSED) 526 if (enc->json.flags & F_CONV_BLESSED)
491 { 527 {
492 // we re-bless the reference to get overload and other niceties right 528 // we re-bless the reference to get overload and other niceties right
493 GV *to_json = gv_fetchmethod_autoload (SvSTASH (sv), "TO_JSON", 1); 529 GV *to_json = gv_fetchmethod_autoload (SvSTASH (sv), "TO_JSON", 0);
494 530
495 if (to_json) 531 if (to_json)
496 { 532 {
497 dSP; 533 dSP;
498 ENTER; 534
499 SAVETMPS;
500 PUSHMARK (SP); 535 ENTER; SAVETMPS; PUSHMARK (SP);
501 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv))); 536 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv)));
502 537
503 // calling with G_SCALAR ensures that we always get a 1 reutrn value 538 // calling with G_SCALAR ensures that we always get a 1 return value
504 // check anyways.
505 PUTBACK; 539 PUTBACK;
506 assert (1 == call_sv ((SV *)GvCV (to_json), G_SCALAR)); 540 call_sv ((SV *)GvCV (to_json), G_SCALAR);
507 SPAGAIN; 541 SPAGAIN;
508 542
543 // catch this surprisingly common error
544 if (SvROK (TOPs) && SvRV (TOPs) == sv)
545 croak ("%s::TO_JSON method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv)));
546
547 sv = POPs;
548 PUTBACK;
549
509 encode_sv (enc, POPs); 550 encode_sv (enc, sv);
510 551
511 FREETMPS; 552 FREETMPS; LEAVE;
512 LEAVE;
513 } 553 }
514 else if (enc->flags & F_ALLOW_BLESSED) 554 else if (enc->json.flags & F_ALLOW_BLESSED)
515 encode_str (enc, "null", 4, 0); 555 encode_str (enc, "null", 4, 0);
516 else 556 else
517 croak ("encountered object '%s', but neither allow_blessed enabled nor TO_JSON method available on it", 557 croak ("encountered object '%s', but neither allow_blessed enabled nor TO_JSON method available on it",
518 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 558 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
519 } 559 }
520 else if (enc->flags & F_ALLOW_BLESSED) 560 else if (enc->json.flags & F_ALLOW_BLESSED)
521 encode_str (enc, "null", 4, 0); 561 encode_str (enc, "null", 4, 0);
522 else 562 else
523 croak ("encountered object '%s', but neither allow_blessed nor convert_blessed settings are enabled", 563 croak ("encountered object '%s', but neither allow_blessed nor convert_blessed settings are enabled",
524 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 564 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
525 } 565 }
528 encode_hv (enc, (HV *)sv); 568 encode_hv (enc, (HV *)sv);
529 else if (svt == SVt_PVAV) 569 else if (svt == SVt_PVAV)
530 encode_av (enc, (AV *)sv); 570 encode_av (enc, (AV *)sv);
531 else if (svt < SVt_PVAV) 571 else if (svt < SVt_PVAV)
532 { 572 {
533 if (SvNIOK (sv) && SvIV (sv) == 0) 573 STRLEN len = 0;
574 char *pv = svt ? SvPV (sv, len) : 0;
575
576 if (len == 1 && *pv == '1')
577 encode_str (enc, "true", 4, 0);
578 else if (len == 1 && *pv == '0')
534 encode_str (enc, "false", 5, 0); 579 encode_str (enc, "false", 5, 0);
535 else if (SvNIOK (sv) && SvIV (sv) == 1)
536 encode_str (enc, "true", 4, 0);
537 else 580 else
538 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1", 581 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
539 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 582 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
540 } 583 }
541 else 584 else
611 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this", 654 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this",
612 SvPV_nolen (sv), SvFLAGS (sv)); 655 SvPV_nolen (sv), SvFLAGS (sv));
613} 656}
614 657
615static SV * 658static SV *
616encode_json (SV *scalar, U32 flags) 659encode_json (SV *scalar, JSON *json)
617{ 660{
618 enc_t enc; 661 enc_t enc;
619 662
620 if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar)) 663 if (!(json->flags & F_ALLOW_NONREF) && !SvROK (scalar))
621 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)"); 664 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)");
622 665
623 enc.flags = flags; 666 enc.json = *json;
624 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 667 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
625 enc.cur = SvPVX (enc.sv); 668 enc.cur = SvPVX (enc.sv);
626 enc.end = SvEND (enc.sv); 669 enc.end = SvEND (enc.sv);
627 enc.indent = 0; 670 enc.indent = 0;
628 enc.maxdepth = DEC_DEPTH (flags); 671 enc.maxdepth = DEC_DEPTH (enc.json.flags);
629 672
630 SvPOK_only (enc.sv); 673 SvPOK_only (enc.sv);
631 encode_sv (&enc, scalar); 674 encode_sv (&enc, scalar);
632 675
633 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 676 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
634 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings 677 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
635 678
636 if (!(flags & (F_ASCII | F_LATIN1 | F_UTF8))) 679 if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8)))
637 SvUTF8_on (enc.sv); 680 SvUTF8_on (enc.sv);
638 681
639 if (enc.flags & F_SHRINK) 682 if (enc.json.flags & F_SHRINK)
640 shrink (enc.sv); 683 shrink (enc.sv);
641 684
642 return enc.sv; 685 return enc.sv;
643} 686}
644 687
649typedef struct 692typedef struct
650{ 693{
651 char *cur; // current parser pointer 694 char *cur; // current parser pointer
652 char *end; // end of input string 695 char *end; // end of input string
653 const char *err; // parse error, if != 0 696 const char *err; // parse error, if != 0
654 U32 flags; // F_* 697 JSON json;
655 U32 depth; // recursion depth 698 U32 depth; // recursion depth
656 U32 maxdepth; // recursion depth limit 699 U32 maxdepth; // recursion depth limit
657} dec_t; 700} dec_t;
658 701
659inline void 702inline void
922 is_nv = 1; 965 is_nv = 1;
923 } 966 }
924 967
925 if (!is_nv) 968 if (!is_nv)
926 { 969 {
970 int len = dec->cur - start;
971
927 // special case the rather common 1..4-digit-int case, assumes 32 bit ints or so 972 // special case the rather common 1..4-digit-int case, assumes 32 bit ints or so
928 if (*start == '-') 973 if (*start == '-')
929 switch (dec->cur - start) 974 switch (len)
930 { 975 {
931 case 2: return newSViv (-( start [1] - '0' * 1)); 976 case 2: return newSViv (-( start [1] - '0' * 1));
932 case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11)); 977 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)); 978 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)); 979 case 5: return newSViv (-(start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111));
935 } 980 }
936 else 981 else
937 switch (dec->cur - start) 982 switch (len)
938 { 983 {
939 case 1: return newSViv ( start [0] - '0' * 1); 984 case 1: return newSViv ( start [0] - '0' * 1);
940 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11); 985 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); 986 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); 987 case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111);
943 } 988 }
944 989
945 { 990 {
946 UV uv; 991 UV uv;
947 int numtype = grok_number (start, dec->cur - start, &uv); 992 int numtype = grok_number (start, len, &uv);
948 if (numtype & IS_NUMBER_IN_UV) 993 if (numtype & IS_NUMBER_IN_UV)
949 if (numtype & IS_NUMBER_NEG) 994 if (numtype & IS_NUMBER_NEG)
950 { 995 {
951 if (uv < (UV)IV_MIN) 996 if (uv < (UV)IV_MIN)
952 return newSViv (-(IV)uv); 997 return newSViv (-(IV)uv);
953 } 998 }
954 else 999 else
955 return newSVuv (uv); 1000 return newSVuv (uv);
956
957 // here would likely be the place for bigint support
958 } 1001 }
959 }
960 1002
961 // if we ever support bigint or bigfloat, this is the place for bigfloat 1003 len -= *start == '-' ? 1 : 0;
1004
1005 // does not fit into IV or UV, try NV
1006 if ((sizeof (NV) == sizeof (double) && DBL_DIG >= len)
1007 #if defined (LDBL_DIG)
1008 || (sizeof (NV) == sizeof (long double) && LDBL_DIG >= len)
1009 #endif
1010 )
1011 // fits into NV without loss of precision
1012 return newSVnv (Atof (start));
1013
1014 // everything else fails, convert it to a string
1015 return newSVpvn (start, dec->cur - start);
1016 }
1017
1018 // loss of precision here
962 return newSVnv (Atof (start)); 1019 return newSVnv (Atof (start));
963 1020
964fail: 1021fail:
965 return 0; 1022 return 0;
966} 1023}
1010} 1067}
1011 1068
1012static SV * 1069static SV *
1013decode_hv (dec_t *dec) 1070decode_hv (dec_t *dec)
1014{ 1071{
1072 SV *sv;
1015 HV *hv = newHV (); 1073 HV *hv = newHV ();
1016 1074
1017 DEC_INC_DEPTH; 1075 DEC_INC_DEPTH;
1018 decode_ws (dec); 1076 decode_ws (dec);
1019 1077
1020 if (*dec->cur == '}') 1078 if (*dec->cur == '}')
1021 ++dec->cur; 1079 ++dec->cur;
1022 else 1080 else
1023 for (;;) 1081 for (;;)
1024 { 1082 {
1025 SV *key, *value;
1026
1027 decode_ws (dec); EXPECT_CH ('"'); 1083 decode_ws (dec); EXPECT_CH ('"');
1028 1084
1029 key = decode_str (dec); 1085 // heuristic: assume that
1030 if (!key) 1086 // a) decode_str + hv_store_ent are abysmally slow.
1031 goto fail; 1087 // b) most hash keys are short, simple ascii text.
1088 // => try to "fast-match" such strings to avoid
1089 // the overhead of decode_str + hv_store_ent.
1090 {
1091 SV *value;
1092 char *p = dec->cur;
1093 char *e = p + 24; // only try up to 24 bytes
1032 1094
1033 decode_ws (dec); EXPECT_CH (':'); 1095 for (;;)
1034
1035 value = decode_sv (dec);
1036 if (!value)
1037 { 1096 {
1097 // the >= 0x80 is true on most architectures
1098 if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\')
1099 {
1100 // slow path, back up and use decode_str
1101 SV *key = decode_str (dec);
1102 if (!key)
1103 goto fail;
1104
1105 decode_ws (dec); EXPECT_CH (':');
1106
1107 value = decode_sv (dec);
1108 if (!value)
1109 {
1110 SvREFCNT_dec (key);
1111 goto fail;
1112 }
1113
1114 hv_store_ent (hv, key, value, 0);
1038 SvREFCNT_dec (key); 1115 SvREFCNT_dec (key);
1116
1117 break;
1118 }
1119 else if (*p == '"')
1120 {
1121 // fast path, got a simple key
1122 char *key = dec->cur;
1123 int len = p - key;
1124 dec->cur = p + 1;
1125
1126 decode_ws (dec); EXPECT_CH (':');
1127
1128 value = decode_sv (dec);
1129 if (!value)
1039 goto fail; 1130 goto fail;
1131
1132 hv_store (hv, key, len, value, 0);
1133
1134 break;
1135 }
1136
1137 ++p;
1040 } 1138 }
1041 1139 }
1042 hv_store_ent (hv, key, value, 0);
1043 SvREFCNT_dec (key);
1044 1140
1045 decode_ws (dec); 1141 decode_ws (dec);
1046 1142
1047 if (*dec->cur == '}') 1143 if (*dec->cur == '}')
1048 { 1144 {
1055 1151
1056 ++dec->cur; 1152 ++dec->cur;
1057 } 1153 }
1058 1154
1059 DEC_DEC_DEPTH; 1155 DEC_DEC_DEPTH;
1060 return newRV_noinc ((SV *)hv); 1156 sv = newRV_noinc ((SV *)hv);
1157
1158 // check filter callbacks
1159 if (dec->json.flags & F_HOOK)
1160 {
1161 if (dec->json.cb_sk_object && HvKEYS (hv) == 1)
1162 {
1163 HE *cb, *he;
1164
1165 hv_iterinit (hv);
1166 he = hv_iternext (hv);
1167 hv_iterinit (hv);
1168
1169 // the next line creates a mortal sv each time its called.
1170 // might want to optimise this for common cases.
1171 cb = hv_fetch_ent (dec->json.cb_sk_object, hv_iterkeysv (he), 0, 0);
1172
1173 if (cb)
1174 {
1175 dSP;
1176 int count;
1177
1178 ENTER; SAVETMPS; PUSHMARK (SP);
1179 XPUSHs (HeVAL (he));
1180
1181 PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN;
1182
1183 if (count == 1)
1184 {
1185 sv = newSVsv (POPs);
1186 FREETMPS; LEAVE;
1187 return sv;
1188 }
1189
1190 FREETMPS; LEAVE;
1191 }
1192 }
1193
1194 if (dec->json.cb_object)
1195 {
1196 dSP;
1197 int count;
1198
1199 ENTER; SAVETMPS; PUSHMARK (SP);
1200 XPUSHs (sv_2mortal (sv));
1201
1202 PUTBACK; count = call_sv (dec->json.cb_object, G_ARRAY); SPAGAIN;
1203
1204 if (count == 1)
1205 {
1206 sv = newSVsv (POPs);
1207 FREETMPS; LEAVE;
1208 return sv;
1209 }
1210
1211 SvREFCNT_inc (sv);
1212 FREETMPS; LEAVE;
1213 }
1214 }
1215
1216 return sv;
1061 1217
1062fail: 1218fail:
1063 SvREFCNT_dec (hv); 1219 SvREFCNT_dec (hv);
1064 DEC_DEC_DEPTH; 1220 DEC_DEC_DEPTH;
1065 return 0; 1221 return 0;
1085 1241
1086 case 't': 1242 case 't':
1087 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) 1243 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
1088 { 1244 {
1089 dec->cur += 4; 1245 dec->cur += 4;
1246#if JSON_SLOW
1247 json_true = get_sv ("JSON::XS::true", 1); SvREADONLY_on (json_true);
1248#endif
1090 return SvREFCNT_inc (json_true); 1249 return SvREFCNT_inc (json_true);
1091 } 1250 }
1092 else 1251 else
1093 ERR ("'true' expected"); 1252 ERR ("'true' expected");
1094 1253
1096 1255
1097 case 'f': 1256 case 'f':
1098 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) 1257 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
1099 { 1258 {
1100 dec->cur += 5; 1259 dec->cur += 5;
1260#if JSON_SLOW
1261 json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false);
1262#endif
1101 return SvREFCNT_inc (json_false); 1263 return SvREFCNT_inc (json_false);
1102 } 1264 }
1103 else 1265 else
1104 ERR ("'false' expected"); 1266 ERR ("'false' expected");
1105 1267
1124fail: 1286fail:
1125 return 0; 1287 return 0;
1126} 1288}
1127 1289
1128static SV * 1290static SV *
1129decode_json (SV *string, U32 flags, UV *offset_return) 1291decode_json (SV *string, JSON *json, UV *offset_return)
1130{ 1292{
1131 dec_t dec; 1293 dec_t dec;
1132 UV offset; 1294 UV offset;
1133 SV *sv; 1295 SV *sv;
1134 1296
1135 SvGETMAGIC (string); 1297 SvGETMAGIC (string);
1136 SvUPGRADE (string, SVt_PV); 1298 SvUPGRADE (string, SVt_PV);
1137 1299
1300 if (json->flags & F_MAXSIZE && SvCUR (string) > DEC_SIZE (json->flags))
1301 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
1302 (unsigned long)SvCUR (string), (unsigned long)DEC_SIZE (json->flags));
1303
1138 if (flags & F_UTF8) 1304 if (json->flags & F_UTF8)
1139 sv_utf8_downgrade (string, 0); 1305 sv_utf8_downgrade (string, 0);
1140 else 1306 else
1141 sv_utf8_upgrade (string); 1307 sv_utf8_upgrade (string);
1142 1308
1143 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP 1309 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
1144 1310
1145 dec.flags = flags; 1311 dec.json = *json;
1146 dec.cur = SvPVX (string); 1312 dec.cur = SvPVX (string);
1147 dec.end = SvEND (string); 1313 dec.end = SvEND (string);
1148 dec.err = 0; 1314 dec.err = 0;
1149 dec.depth = 0; 1315 dec.depth = 0;
1150 dec.maxdepth = DEC_DEPTH (dec.flags); 1316 dec.maxdepth = DEC_DEPTH (dec.json.flags);
1317
1318 if (dec.json.cb_object || dec.json.cb_sk_object)
1319 dec.json.flags |= F_HOOK;
1151 1320
1152 *dec.end = 0; // this should basically be a nop, too, but make sure it's there 1321 *dec.end = 0; // this should basically be a nop, too, but make sure it's there
1153 sv = decode_sv (&dec); 1322 sv = decode_sv (&dec);
1154 1323
1155 if (!(offset_return || !sv)) 1324 if (!(offset_return || !sv))
1165 } 1334 }
1166 } 1335 }
1167 1336
1168 if (offset_return || !sv) 1337 if (offset_return || !sv)
1169 { 1338 {
1170 offset = dec.flags & F_UTF8 1339 offset = dec.json.flags & F_UTF8
1171 ? dec.cur - SvPVX (string) 1340 ? dec.cur - SvPVX (string)
1172 : utf8_distance (dec.cur, SvPVX (string)); 1341 : utf8_distance (dec.cur, SvPVX (string));
1173 1342
1174 if (offset_return) 1343 if (offset_return)
1175 *offset_return = offset; 1344 *offset_return = offset;
1194 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); 1363 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
1195 } 1364 }
1196 1365
1197 sv = sv_2mortal (sv); 1366 sv = sv_2mortal (sv);
1198 1367
1199 if (!(dec.flags & F_ALLOW_NONREF) && !SvROK (sv)) 1368 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)"); 1369 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)");
1201 1370
1202 return sv; 1371 return sv;
1203} 1372}
1204 1373
1225 json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false); 1394 json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false);
1226} 1395}
1227 1396
1228PROTOTYPES: DISABLE 1397PROTOTYPES: DISABLE
1229 1398
1230SV *new (char *dummy) 1399void CLONE (...)
1231 CODE: 1400 CODE:
1232 RETVAL = sv_bless (newRV_noinc (newSVuv (F_DEFAULT)), json_stash); 1401 json_stash = 0;
1233 OUTPUT: 1402 json_boolean_stash = 0;
1234 RETVAL
1235 1403
1404void new (char *klass)
1405 PPCODE:
1406{
1407 SV *pv = NEWSV (0, sizeof (JSON));
1408 SvPOK_only (pv);
1409 Zero (SvPVX (pv), 1, JSON);
1410 ((JSON *)SvPVX (pv))->flags = F_DEFAULT;
1411 XPUSHs (sv_2mortal (sv_bless (newRV_noinc (pv), JSON_STASH)));
1412}
1413
1236SV *ascii (SV *self, int enable = 1) 1414void ascii (JSON *self, int enable = 1)
1237 ALIAS: 1415 ALIAS:
1238 ascii = F_ASCII 1416 ascii = F_ASCII
1239 latin1 = F_LATIN1 1417 latin1 = F_LATIN1
1240 utf8 = F_UTF8 1418 utf8 = F_UTF8
1241 indent = F_INDENT 1419 indent = F_INDENT
1245 pretty = F_PRETTY 1423 pretty = F_PRETTY
1246 allow_nonref = F_ALLOW_NONREF 1424 allow_nonref = F_ALLOW_NONREF
1247 shrink = F_SHRINK 1425 shrink = F_SHRINK
1248 allow_blessed = F_ALLOW_BLESSED 1426 allow_blessed = F_ALLOW_BLESSED
1249 convert_blessed = F_CONV_BLESSED 1427 convert_blessed = F_CONV_BLESSED
1250 CODE: 1428 PPCODE:
1251{ 1429{
1252 UV *uv = SvJSON (self);
1253 if (enable) 1430 if (enable)
1254 *uv |= ix; 1431 self->flags |= ix;
1255 else 1432 else
1256 *uv &= ~ix; 1433 self->flags &= ~ix;
1257 1434
1258 RETVAL = newSVsv (self); 1435 XPUSHs (ST (0));
1259} 1436}
1260 OUTPUT:
1261 RETVAL
1262 1437
1263SV *max_depth (SV *self, UV max_depth = 0x80000000UL) 1438void max_depth (JSON *self, UV max_depth = 0x80000000UL)
1264 CODE: 1439 PPCODE:
1265{ 1440{
1266 UV *uv = SvJSON (self);
1267 UV log2 = 0; 1441 UV log2 = 0;
1268 1442
1269 if (max_depth > 0x80000000UL) max_depth = 0x80000000UL; 1443 if (max_depth > 0x80000000UL) max_depth = 0x80000000UL;
1270 1444
1271 while ((1UL << log2) < max_depth) 1445 while ((1UL << log2) < max_depth)
1272 ++log2; 1446 ++log2;
1273 1447
1274 *uv = *uv & ~F_MAXDEPTH | (log2 << S_MAXDEPTH); 1448 self->flags = self->flags & ~F_MAXDEPTH | (log2 << S_MAXDEPTH);
1275 1449
1276 RETVAL = newSVsv (self); 1450 XPUSHs (ST (0));
1277} 1451}
1278 OUTPUT:
1279 RETVAL
1280 1452
1281void encode (SV *self, SV *scalar) 1453void max_size (JSON *self, UV max_size = 0)
1282 PPCODE: 1454 PPCODE:
1283 XPUSHs (encode_json (scalar, *SvJSON (self))); 1455{
1456 UV log2 = 0;
1284 1457
1285void decode (SV *self, SV *jsonstr) 1458 if (max_size > 0x80000000UL) max_size = 0x80000000UL;
1459 if (max_size == 1) max_size = 2;
1460
1461 while ((1UL << log2) < max_size)
1462 ++log2;
1463
1464 self->flags = self->flags & ~F_MAXSIZE | (log2 << S_MAXSIZE);
1465
1466 XPUSHs (ST (0));
1467}
1468
1469void filter_json_object (JSON *self, SV *cb = &PL_sv_undef)
1286 PPCODE: 1470 PPCODE:
1471{
1472 SvREFCNT_dec (self->cb_object);
1473 self->cb_object = SvOK (cb) ? newSVsv (cb) : 0;
1474
1475 XPUSHs (ST (0));
1476}
1477
1478void filter_json_single_key_object (JSON *self, SV *key, SV *cb = &PL_sv_undef)
1479 PPCODE:
1480{
1481 if (!self->cb_sk_object)
1482 self->cb_sk_object = newHV ();
1483
1484 if (SvOK (cb))
1485 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);
1486 else
1487 {
1488 hv_delete_ent (self->cb_sk_object, key, G_DISCARD, 0);
1489
1490 if (!HvKEYS (self->cb_sk_object))
1491 {
1492 SvREFCNT_dec (self->cb_sk_object);
1493 self->cb_sk_object = 0;
1494 }
1495 }
1496
1497 XPUSHs (ST (0));
1498}
1499
1500void encode (JSON *self, SV *scalar)
1501 PPCODE:
1502 XPUSHs (encode_json (scalar, self));
1503
1504void decode (JSON *self, SV *jsonstr)
1505 PPCODE:
1287 XPUSHs (decode_json (jsonstr, *SvJSON (self), 0)); 1506 XPUSHs (decode_json (jsonstr, self, 0));
1288 1507
1289void decode_prefix (SV *self, SV *jsonstr) 1508void decode_prefix (JSON *self, SV *jsonstr)
1290 PPCODE: 1509 PPCODE:
1291{ 1510{
1292 UV offset; 1511 UV offset;
1293 EXTEND (SP, 2); 1512 EXTEND (SP, 2);
1294 PUSHs (decode_json (jsonstr, *SvJSON (self), &offset)); 1513 PUSHs (decode_json (jsonstr, self, &offset));
1295 PUSHs (sv_2mortal (newSVuv (offset))); 1514 PUSHs (sv_2mortal (newSVuv (offset)));
1296} 1515}
1297 1516
1517void DESTROY (JSON *self)
1518 CODE:
1519 SvREFCNT_dec (self->cb_sk_object);
1520 SvREFCNT_dec (self->cb_object);
1521
1298PROTOTYPES: ENABLE 1522PROTOTYPES: ENABLE
1299 1523
1300void to_json (SV *scalar) 1524void to_json (SV *scalar)
1301 ALIAS:
1302 objToJson = 0
1303 PPCODE: 1525 PPCODE:
1526{
1527 JSON json = { F_DEFAULT | F_UTF8 };
1304 XPUSHs (encode_json (scalar, F_DEFAULT | F_UTF8)); 1528 XPUSHs (encode_json (scalar, &json));
1529}
1305 1530
1306void from_json (SV *jsonstr) 1531void from_json (SV *jsonstr)
1307 ALIAS:
1308 jsonToObj = 0
1309 PPCODE: 1532 PPCODE:
1533{
1534 JSON json = { F_DEFAULT | F_UTF8 };
1310 XPUSHs (decode_json (jsonstr, F_DEFAULT | F_UTF8, 0)); 1535 XPUSHs (decode_json (jsonstr, &json, 0));
1536}
1311 1537

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines