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

Comparing CBOR-XS/XS.xs (file contents):
Revision 1.40 by root, Tue Dec 10 13:54:40 2013 UTC vs.
Revision 1.63 by root, Sat Nov 26 04:50:58 2016 UTC

… …
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 <limits.h> 9#include <limits.h>
10#include <float.h> 10#include <float.h>
11#include <inttypes.h>
11 12
13#define ECB_NO_THREADS 1
12#include "ecb.h" 14#include "ecb.h"
13 15
14// compatibility with perl <5.18 16// compatibility with perl <5.18
15#ifndef HvNAMELEN_get 17#ifndef HvNAMELEN_get
16# define HvNAMELEN_get(hv) strlen (HvNAME (hv)) 18# define HvNAMELEN_get(hv) strlen (HvNAME (hv))
… …
19# define HvNAMELEN(hv) HvNAMELEN_get (hv) 21# define HvNAMELEN(hv) HvNAMELEN_get (hv)
20#endif 22#endif
21#ifndef HvNAMEUTF8 23#ifndef HvNAMEUTF8
22# define HvNAMEUTF8(hv) 0 24# define HvNAMEUTF8(hv) 0
23#endif 25#endif
26#ifndef SvREFCNT_inc_NN
27# define SvREFCNT_inc_NN(sv) SvREFCNT_inc (sv)
28#endif
24#ifndef SvREFCNT_dec_NN 29#ifndef SvREFCNT_dec_NN
25# define SvREFCNT_dec_NN(sv) SvREFCNT_dec (sv) 30# define SvREFCNT_dec_NN(sv) SvREFCNT_dec (sv)
26#endif 31#endif
27 32
28// known major and minor types 33// known major and minor types
… …
98 103
99#define F_SHRINK 0x00000001UL 104#define F_SHRINK 0x00000001UL
100#define F_ALLOW_UNKNOWN 0x00000002UL 105#define F_ALLOW_UNKNOWN 0x00000002UL
101#define F_ALLOW_SHARING 0x00000004UL 106#define F_ALLOW_SHARING 0x00000004UL
102#define F_ALLOW_CYCLES 0x00000008UL 107#define F_ALLOW_CYCLES 0x00000008UL
108#define F_FORBID_OBJECTS 0x00000010UL
103#define F_PACK_STRINGS 0x00000010UL 109#define F_PACK_STRINGS 0x00000020UL
110#define F_TEXT_KEYS 0x00000040UL
111#define F_TEXT_STRINGS 0x00000080UL
104#define F_VALIDATE_UTF8 0x00000020UL 112#define F_VALIDATE_UTF8 0x00000100UL
105 113
106#define INIT_SIZE 32 // initial scalar size to be allocated 114#define INIT_SIZE 32 // initial scalar size to be allocated
107 115
108#define SB do { 116#define SB do {
109#define SE } while (0) 117#define SE } while (0)
… …
182 190
183// minimum length of a string to be registered for stringref 191// minimum length of a string to be registered for stringref
184ecb_inline int 192ecb_inline int
185minimum_string_length (UV idx) 193minimum_string_length (UV idx)
186{ 194{
187 return idx > 23 195 return idx <= 23 ? 3
188 ? idx > 0xffU 196 : idx <= 0xffU ? 4
189 ? idx > 0xffffU 197 : idx <= 0xffffU ? 5
190 ? idx > 0xffffffffU 198 : idx <= 0xffffffffU ? 7
191 ? 11 199 : 11;
192 : 7
193 : 5
194 : 4
195 : 3;
196} 200}
197 201
198///////////////////////////////////////////////////////////////////////////// 202/////////////////////////////////////////////////////////////////////////////
199// encoder 203// encoder
200 204
… …
213} enc_t; 217} enc_t;
214 218
215ecb_inline void 219ecb_inline void
216need (enc_t *enc, STRLEN len) 220need (enc_t *enc, STRLEN len)
217{ 221{
218 if (ecb_expect_false (enc->cur + len >= enc->end)) 222 if (ecb_expect_false ((uintptr_t)(enc->end - enc->cur) < len))
219 { 223 {
220 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv); 224 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
221 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1); 225 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
222 enc->cur = SvPVX (enc->sv) + cur; 226 enc->cur = SvPVX (enc->sv) + cur;
223 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; 227 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
… …
275encode_tag (enc_t *enc, UV tag) 279encode_tag (enc_t *enc, UV tag)
276{ 280{
277 encode_uint (enc, MAJOR_TAG, tag); 281 encode_uint (enc, MAJOR_TAG, tag);
278} 282}
279 283
284// exceptional (hopefully) slow path for byte strings that need to be utf8-encoded
285ecb_noinline static void
286encode_str_utf8 (enc_t *enc, int utf8, char *str, STRLEN len)
287{
288 STRLEN ulen = len;
289 U8 *p, *pend = (U8 *)str + len;
290
291 for (p = (U8 *)str; p < pend; ++p)
292 ulen += *p >> 7; // count set high bits
293
294 encode_uint (enc, MAJOR_TEXT, ulen);
295
296 need (enc, ulen);
297 for (p = (U8 *)str; p < pend; ++p)
298 if (*p < 0x80)
299 *enc->cur++ = *p;
300 else
301 {
302 *enc->cur++ = 0xc0 + (*p >> 6);
303 *enc->cur++ = 0x80 + (*p & 63);
304 }
305}
306
280ecb_inline void 307ecb_inline void
281encode_str (enc_t *enc, int utf8, char *str, STRLEN len) 308encode_str (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len)
282{ 309{
310 if (ecb_expect_false (upgrade_utf8))
311 if (!utf8)
312 {
313 encode_str_utf8 (enc, utf8, str, len);
314 return;
315 }
316
283 encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len); 317 encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len);
284 need (enc, len); 318 need (enc, len);
285 memcpy (enc->cur, str, len); 319 memcpy (enc->cur, str, len);
286 enc->cur += len; 320 enc->cur += len;
287} 321}
288 322
289static void 323ecb_inline void
290encode_strref (enc_t *enc, int utf8, char *str, STRLEN len) 324encode_strref (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len)
291{ 325{
292 if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS)) 326 if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS))
293 { 327 {
294 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1); 328 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1);
295 329
… …
306 sv_setuv (*svp, enc->stringref_idx); 340 sv_setuv (*svp, enc->stringref_idx);
307 ++enc->stringref_idx; 341 ++enc->stringref_idx;
308 } 342 }
309 } 343 }
310 344
311 encode_str (enc, utf8, str, len); 345 encode_str (enc, upgrade_utf8, utf8, str, len);
312} 346}
313 347
314static void encode_sv (enc_t *enc, SV *sv); 348static void encode_sv (enc_t *enc, SV *sv);
315 349
316static void 350static void
… …
323 357
324 ++enc->depth; 358 ++enc->depth;
325 359
326 encode_uint (enc, MAJOR_ARRAY, len + 1); 360 encode_uint (enc, MAJOR_ARRAY, len + 1);
327 361
362 if (ecb_expect_false (SvMAGICAL (av)))
328 for (i = 0; i <= len; ++i) 363 for (i = 0; i <= len; ++i)
329 { 364 {
330 SV **svp = av_fetch (av, i, 0); 365 SV **svp = av_fetch (av, i, 0);
331 encode_sv (enc, svp ? *svp : &PL_sv_undef); 366 encode_sv (enc, svp ? *svp : &PL_sv_undef);
332 } 367 }
368 else
369 for (i = 0; i <= len; ++i)
370 {
371 SV *sv = AvARRAY (av)[i];
372 encode_sv (enc, sv ? sv : &PL_sv_undef);
373 }
333 374
334 --enc->depth; 375 --enc->depth;
335} 376}
336 377
337static void 378static void
… …
345 ++enc->depth; 386 ++enc->depth;
346 387
347 int pairs = hv_iterinit (hv); 388 int pairs = hv_iterinit (hv);
348 int mg = SvMAGICAL (hv); 389 int mg = SvMAGICAL (hv);
349 390
350 if (mg) 391 if (ecb_expect_false (mg))
351 encode_ch (enc, MAJOR_MAP | MINOR_INDEF); 392 encode_ch (enc, MAJOR_MAP | MINOR_INDEF);
352 else 393 else
353 encode_uint (enc, MAJOR_MAP, pairs); 394 encode_uint (enc, MAJOR_MAP, pairs);
354 395
355 while ((he = hv_iternext (hv))) 396 while ((he = hv_iternext (hv)))
356 { 397 {
357 if (HeKLEN (he) == HEf_SVKEY) 398 if (HeKLEN (he) == HEf_SVKEY)
358 encode_sv (enc, HeSVKEY (he)); 399 encode_sv (enc, HeSVKEY (he));
359 else 400 else
360 encode_strref (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he)); 401 encode_strref (enc, enc->cbor.flags & (F_TEXT_KEYS | F_TEXT_STRINGS), HeKUTF8 (he), HeKEY (he), HeKLEN (he));
361 402
362 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he)); 403 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he));
363 } 404 }
364 405
365 if (mg) 406 if (ecb_expect_false (mg))
366 encode_ch (enc, MAJOR_MISC | MINOR_INDEF); 407 encode_ch (enc, MAJOR_MISC | MINOR_INDEF);
367 408
368 --enc->depth; 409 --enc->depth;
369} 410}
370 411
… …
413 } 454 }
414 455
415 if (ecb_expect_false (SvREFCNT (sv) > 1) 456 if (ecb_expect_false (SvREFCNT (sv) > 1)
416 && ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING)) 457 && ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING))
417 { 458 {
418 if (!enc->shareable) 459 if (ecb_expect_false (!enc->shareable))
419 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ()); 460 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
420 461
421 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1); 462 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
422 463
423 if (SvOK (*svp)) 464 if (SvOK (*svp))
… …
437 if (ecb_expect_false (SvOBJECT (sv))) 478 if (ecb_expect_false (SvOBJECT (sv)))
438 { 479 {
439 HV *stash = SvSTASH (sv); 480 HV *stash = SvSTASH (sv);
440 GV *method; 481 GV *method;
441 482
483 if (enc->cbor.flags & F_FORBID_OBJECTS)
484 croak ("encountered object '%s', but forbid_objects is enabled",
485 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
442 if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0))) 486 else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
443 { 487 {
444 dSP; 488 dSP;
445 489
446 ENTER; SAVETMPS; PUSHMARK (SP); 490 ENTER; SAVETMPS;
491 PUSHMARK (SP);
447 // we re-bless the reference to get overload and other niceties right 492 // we re-bless the reference to get overload and other niceties right
448 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); 493 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
449 494
450 PUTBACK; 495 PUTBACK;
451 // G_SCALAR ensures that return value is 1 496 // G_SCALAR ensures that return value is 1
… …
464 } 509 }
465 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0) 510 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)
466 { 511 {
467 dSP; 512 dSP;
468 513
469 ENTER; SAVETMPS; PUSHMARK (SP); 514 ENTER; SAVETMPS;
515 SAVESTACK_POS ();
516 PUSHMARK (SP);
470 EXTEND (SP, 2); 517 EXTEND (SP, 2);
471 // we re-bless the reference to get overload and other niceties right 518 // we re-bless the reference to get overload and other niceties right
472 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); 519 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
473 PUSHs (sv_cbor); 520 PUSHs (sv_cbor);
474 521
… …
480 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv) 527 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)
481 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash)); 528 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));
482 529
483 encode_tag (enc, CBOR_TAG_PERL_OBJECT); 530 encode_tag (enc, CBOR_TAG_PERL_OBJECT);
484 encode_uint (enc, MAJOR_ARRAY, count + 1); 531 encode_uint (enc, MAJOR_ARRAY, count + 1);
485 encode_strref (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash)); 532 encode_strref (enc, 0, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
486 533
487 while (count) 534 while (count)
488 encode_sv (enc, SP[1 - count--]); 535 encode_sv (enc, SP[1 - count--]);
489 536
490 PUTBACK; 537 PUTBACK;
… …
516 if (ecb_expect_false (nv == (NV)(U32)nv)) 563 if (ecb_expect_false (nv == (NV)(U32)nv))
517 encode_uint (enc, MAJOR_POS_INT, (U32)nv); 564 encode_uint (enc, MAJOR_POS_INT, (U32)nv);
518 //TODO: maybe I32? 565 //TODO: maybe I32?
519 else if (ecb_expect_false (nv == (float)nv)) 566 else if (ecb_expect_false (nv == (float)nv))
520 { 567 {
568 *enc->cur++ = MAJOR_MISC | MISC_FLOAT32;
569
521 uint32_t fp = ecb_float_to_binary32 (nv); 570 uint32_t fp = ecb_float_to_binary32 (nv);
522
523 *enc->cur++ = MAJOR_MISC | MISC_FLOAT32;
524 571
525 if (!ecb_big_endian ()) 572 if (!ecb_big_endian ())
526 fp = ecb_bswap32 (fp); 573 fp = ecb_bswap32 (fp);
527 574
528 memcpy (enc->cur, &fp, 4); 575 memcpy (enc->cur, &fp, 4);
529 enc->cur += 4; 576 enc->cur += 4;
530 } 577 }
531 else 578 else
532 { 579 {
580 *enc->cur++ = MAJOR_MISC | MISC_FLOAT64;
581
533 uint64_t fp = ecb_double_to_binary64 (nv); 582 uint64_t fp = ecb_double_to_binary64 (nv);
534
535 *enc->cur++ = MAJOR_MISC | MISC_FLOAT64;
536 583
537 if (!ecb_big_endian ()) 584 if (!ecb_big_endian ())
538 fp = ecb_bswap64 (fp); 585 fp = ecb_bswap64 (fp);
539 586
540 memcpy (enc->cur, &fp, 8); 587 memcpy (enc->cur, &fp, 8);
… …
549 596
550 if (SvPOKp (sv)) 597 if (SvPOKp (sv))
551 { 598 {
552 STRLEN len; 599 STRLEN len;
553 char *str = SvPV (sv, len); 600 char *str = SvPV (sv, len);
554 encode_strref (enc, SvUTF8 (sv), str, len); 601 encode_strref (enc, enc->cbor.flags & F_TEXT_STRINGS, SvUTF8 (sv), str, len);
555 } 602 }
556 else if (SvNOKp (sv)) 603 else if (SvNOKp (sv))
557 encode_nv (enc, sv); 604 encode_nv (enc, sv);
558 else if (SvIOKp (sv)) 605 else if (SvIOKp (sv))
559 { 606 {
… …
576} 623}
577 624
578static SV * 625static SV *
579encode_cbor (SV *scalar, CBOR *cbor) 626encode_cbor (SV *scalar, CBOR *cbor)
580{ 627{
581 enc_t enc = { }; 628 enc_t enc = { 0 };
582 629
583 enc.cbor = *cbor; 630 enc.cbor = *cbor;
584 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 631 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
585 enc.cur = SvPVX (enc.sv); 632 enc.cur = SvPVX (enc.sv);
586 enc.end = SvEND (enc.sv); 633 enc.end = SvEND (enc.sv);
587 634
588 SvPOK_only (enc.sv); 635 SvPOK_only (enc.sv);
589 636
590 if (cbor->flags & F_PACK_STRINGS) 637 if (cbor->flags & F_PACK_STRINGS)
591 { 638 {
… …
618 U32 depth; // recursion depth 665 U32 depth; // recursion depth
619 U32 maxdepth; // recursion depth limit 666 U32 maxdepth; // recursion depth limit
620 AV *shareable; 667 AV *shareable;
621 AV *stringref; 668 AV *stringref;
622 SV *decode_tagged; 669 SV *decode_tagged;
670 SV *err_sv; // optional sv for error, needs to be freed
623} dec_t; 671} dec_t;
624 672
625#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE 673// set dec->err to ERRSV
674ecb_cold static void
675err_errsv (dec_t *dec)
676{
677 if (!dec->err)
678 {
679 dec->err_sv = newSVsv (ERRSV);
626 680
627#define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data") 681 // chop off the trailing \n
682 SvCUR_set (dec->err_sv, SvCUR (dec->err_sv) - 1);
683 *SvEND (dec->err_sv) = 0;
628 684
685 dec->err = SvPVutf8_nolen (dec->err_sv);
686 }
687}
688
689// the following functions are used to reduce code size and help the compiler to optimise
690ecb_cold static void
691err_set (dec_t *dec, const char *reason)
692{
693 if (!dec->err)
694 dec->err = reason;
695}
696
697ecb_cold static void
698err_unexpected_end (dec_t *dec)
699{
700 err_set (dec, "unexpected end of CBOR data");
701}
702
703#define ERR_DO(do) SB do; goto fail; SE
704#define ERR(reason) ERR_DO (err_set (dec, reason))
705#define ERR_ERRSV ERR_DO (err_errsv (dec))
706
707#define WANT(len) if (ecb_expect_false ((uintptr_t)(dec->end - dec->cur) < (STRLEN)len)) ERR_DO (err_unexpected_end (dec))
708
629#define DEC_INC_DEPTH if (++dec->depth > dec->cbor.max_depth) ERR (ERR_NESTING_EXCEEDED) 709#define DEC_INC_DEPTH if (ecb_expect_false (++dec->depth > dec->cbor.max_depth)) ERR (ERR_NESTING_EXCEEDED)
630#define DEC_DEC_DEPTH --dec->depth 710#define DEC_DEC_DEPTH --dec->depth
631 711
632static UV 712static UV
633decode_uint (dec_t *dec) 713decode_uint (dec_t *dec)
634{ 714{
… …
711 av_push (av, decode_sv (dec)); 791 av_push (av, decode_sv (dec));
712 } 792 }
713 } 793 }
714 else 794 else
715 { 795 {
716 int i, len = decode_uint (dec); 796 UV i, len = decode_uint (dec);
717 797
718 WANT (len); // complexity check for av_fill - need at least one byte per value, do not allow supersize arrays 798 WANT (len); // complexity check for av_fill - need at least one byte per value, do not allow supersize arrays
719 av_fill (av, len - 1); 799 av_fill (av, len - 1);
720 800
721 for (i = 0; i < len; ++i) 801 for (i = 0; i < len; ++i)
… …
724 804
725 DEC_DEC_DEPTH; 805 DEC_DEC_DEPTH;
726 return newRV_noinc ((SV *)av); 806 return newRV_noinc ((SV *)av);
727 807
728fail: 808fail:
729 SvREFCNT_dec (av); 809 SvREFCNT_dec_NN (av);
730 DEC_DEC_DEPTH; 810 DEC_DEC_DEPTH;
731 return &PL_sv_undef; 811 return &PL_sv_undef;
732} 812}
733 813
734static void 814static void
… …
736{ 816{
737 // for speed reasons, we specialcase single-string 817 // for speed reasons, we specialcase single-string
738 // byte or utf-8 strings as keys, but only when !stringref 818 // byte or utf-8 strings as keys, but only when !stringref
739 819
740 if (ecb_expect_true (!dec->stringref)) 820 if (ecb_expect_true (!dec->stringref))
741 if (ecb_expect_true ((*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8)) 821 if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8))
742 { 822 {
743 I32 len = decode_uint (dec); 823 STRLEN len = decode_uint (dec);
744 char *key = (char *)dec->cur; 824 char *key = (char *)dec->cur;
745 825
826 WANT (len);
746 dec->cur += len; 827 dec->cur += len;
747 828
748 hv_store (hv, key, len, decode_sv (dec), 0); 829 hv_store (hv, key, len, decode_sv (dec), 0);
749 830
750 return; 831 return;
751 } 832 }
752 else if (ecb_expect_true ((*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8)) 833 else if (ecb_expect_true ((U8)(*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8))
753 { 834 {
754 I32 len = decode_uint (dec); 835 STRLEN len = decode_uint (dec);
755 char *key = (char *)dec->cur; 836 char *key = (char *)dec->cur;
756 837
838 WANT (len);
757 dec->cur += len; 839 dec->cur += len;
758 840
759 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8)) 841 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
760 if (!is_utf8_string (key, len)) 842 if (!is_utf8_string (key, len))
761 ERR ("corrupted CBOR data (invalid UTF-8 in map key)"); 843 ERR ("corrupted CBOR data (invalid UTF-8 in map key)");
… …
766 } 848 }
767 849
768 SV *k = decode_sv (dec); 850 SV *k = decode_sv (dec);
769 SV *v = decode_sv (dec); 851 SV *v = decode_sv (dec);
770 852
853 // we leak memory if uncaught exceptions are thrown by random magical
854 // methods, and this is hopefully the only place where it can happen,
855 // so if there is a chance of an exception, take the very slow path.
856 // since catching exceptions is "undocumented/internal/forbidden" by
857 // the new p5p powers, we need to call out to a perl function :/
858 if (ecb_expect_false (SvAMAGIC (k)))
859 {
860 dSP;
861
862 ENTER; SAVETMPS;
863 PUSHMARK (SP);
864 EXTEND (SP, 3);
865 PUSHs (sv_2mortal (newRV_inc ((SV *)hv)));
866 PUSHs (sv_2mortal (k));
867 PUSHs (sv_2mortal (v));
868
869 PUTBACK;
870 call_pv ("CBOR::XS::_hv_store", G_VOID | G_DISCARD | G_EVAL);
871 SPAGAIN;
872
873 FREETMPS; LEAVE;
874
875 if (SvTRUE (ERRSV))
876 ERR_ERRSV;
877
878 return;
879 }
880
771 hv_store_ent (hv, k, v, 0); 881 hv_store_ent (hv, k, v, 0);
772 SvREFCNT_dec (k); 882 SvREFCNT_dec_NN (k);
773 883
774fail: 884fail:
775 ; 885 ;
776} 886}
777 887
… …
799 decode_he (dec, hv); 909 decode_he (dec, hv);
800 } 910 }
801 } 911 }
802 else 912 else
803 { 913 {
804 int pairs = decode_uint (dec); 914 UV pairs = decode_uint (dec);
915
916 WANT (pairs); // complexity check - need at least one byte per value, do not allow supersize hashes
805 917
806 while (pairs--) 918 while (pairs--)
807 decode_he (dec, hv); 919 decode_he (dec, hv);
808 } 920 }
809 921
810 DEC_DEC_DEPTH; 922 DEC_DEC_DEPTH;
811 return newRV_noinc ((SV *)hv); 923 return newRV_noinc ((SV *)hv);
812 924
813fail: 925fail:
814 SvREFCNT_dec (hv); 926 SvREFCNT_dec_NN (hv);
815 DEC_DEC_DEPTH; 927 DEC_DEC_DEPTH;
816 return &PL_sv_undef; 928 return &PL_sv_undef;
817} 929}
818 930
819static SV * 931static SV *
820decode_str (dec_t *dec, int utf8) 932decode_str (dec_t *dec, int utf8)
821{ 933{
822 SV *sv = 0; 934 SV *sv = 0;
823 935
824 if ((*dec->cur & MINOR_MASK) == MINOR_INDEF) 936 if (ecb_expect_false ((*dec->cur & MINOR_MASK) == MINOR_INDEF))
825 { 937 {
826 // indefinite length strings 938 // indefinite length strings
827 ++dec->cur; 939 ++dec->cur;
828 940
829 U8 major = *dec->cur & MAJOR_MISC; 941 U8 major = *dec->cur & MAJOR_MISC;
… …
897 sv = newRV_noinc (decode_sv (dec)); 1009 sv = newRV_noinc (decode_sv (dec));
898 break; 1010 break;
899 1011
900 case CBOR_TAG_STRINGREF_NAMESPACE: 1012 case CBOR_TAG_STRINGREF_NAMESPACE:
901 { 1013 {
1014 // do not use SAVETMPS/FREETMPS, as these will
1015 // erase mortalised caches, e.g. "shareable"
902 ENTER; SAVETMPS; 1016 ENTER;
903 1017
904 SAVESPTR (dec->stringref); 1018 SAVESPTR (dec->stringref);
905 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ()); 1019 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ());
906 1020
907 sv = decode_sv (dec); 1021 sv = decode_sv (dec);
908 1022
909 FREETMPS; LEAVE; 1023 LEAVE;
910 } 1024 }
911 break; 1025 break;
912 1026
913 case CBOR_TAG_STRINGREF: 1027 case CBOR_TAG_STRINGREF:
914 { 1028 {
… …
965 } 1079 }
966 break; 1080 break;
967 1081
968 case CBOR_TAG_PERL_OBJECT: 1082 case CBOR_TAG_PERL_OBJECT:
969 { 1083 {
1084 if (dec->cbor.flags & F_FORBID_OBJECTS)
1085 goto filter;
1086
970 sv = decode_sv (dec); 1087 sv = decode_sv (dec);
971 1088
972 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV) 1089 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
973 ERR ("corrupted CBOR data (non-array perl object)"); 1090 ERR ("corrupted CBOR data (non-array perl object)");
974 1091
… …
984 if (!method) 1101 if (!method)
985 ERR ("cannot decode perl-object (package does not have a THAW method)"); 1102 ERR ("cannot decode perl-object (package does not have a THAW method)");
986 1103
987 dSP; 1104 dSP;
988 1105
989 ENTER; SAVETMPS; PUSHMARK (SP); 1106 ENTER; SAVETMPS;
1107 PUSHMARK (SP);
990 EXTEND (SP, len + 1); 1108 EXTEND (SP, len + 1);
991 // we re-bless the reference to get overload and other niceties right 1109 // we re-bless the reference to get overload and other niceties right
992 PUSHs (*av_fetch (av, 0, 1)); 1110 PUSHs (*av_fetch (av, 0, 1));
993 PUSHs (sv_cbor); 1111 PUSHs (sv_cbor);
994 1112
… …
1002 SPAGAIN; 1120 SPAGAIN;
1003 1121
1004 if (SvTRUE (ERRSV)) 1122 if (SvTRUE (ERRSV))
1005 { 1123 {
1006 FREETMPS; LEAVE; 1124 FREETMPS; LEAVE;
1007 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV)))); 1125 ERR_ERRSV;
1008 } 1126 }
1009 1127
1010 SvREFCNT_dec (sv); 1128 SvREFCNT_dec_NN (sv);
1011 sv = SvREFCNT_inc (POPs); 1129 sv = SvREFCNT_inc (POPs);
1012 1130
1013 PUTBACK; 1131 PUTBACK;
1014 1132
1015 FREETMPS; LEAVE; 1133 FREETMPS; LEAVE;
1016 } 1134 }
1017 break; 1135 break;
1018 1136
1019 default: 1137 default:
1138 filter:
1020 { 1139 {
1140 SV *tag_sv = newSVuv (tag);
1141
1021 sv = decode_sv (dec); 1142 sv = decode_sv (dec);
1022 1143
1023 dSP; 1144 dSP;
1024 ENTER; SAVETMPS; PUSHMARK (SP); 1145 ENTER; SAVETMPS;
1146 SAVESTACK_POS ();
1147 PUSHMARK (SP);
1025 EXTEND (SP, 2); 1148 EXTEND (SP, 2);
1026 PUSHs (newSVuv (tag)); 1149 PUSHs (tag_sv);
1027 PUSHs (sv); 1150 PUSHs (sv);
1028 1151
1029 PUTBACK; 1152 PUTBACK;
1030 int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL); 1153 int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL);
1031 SPAGAIN; 1154 SPAGAIN;
1032 1155
1033 if (SvTRUE (ERRSV)) 1156 if (SvTRUE (ERRSV))
1034 { 1157 {
1158 SvREFCNT_dec_NN (tag_sv);
1035 FREETMPS; LEAVE; 1159 FREETMPS; LEAVE;
1036 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV)))); 1160 ERR_ERRSV;
1037 } 1161 }
1038 1162
1039 if (count) 1163 if (count)
1040 { 1164 {
1165 SvREFCNT_dec_NN (tag_sv);
1041 SvREFCNT_dec (sv); 1166 SvREFCNT_dec_NN (sv);
1042 sv = SvREFCNT_inc (POPs); 1167 sv = SvREFCNT_inc_NN (POPs);
1043 } 1168 }
1044 else 1169 else
1045 { 1170 {
1046 AV *av = newAV (); 1171 AV *av = newAV ();
1047 av_push (av, newSVuv (tag)); 1172 av_push (av, tag_sv);
1048 av_push (av, sv); 1173 av_push (av, sv);
1049 1174
1050 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 1175 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
1051 ? cbor_tagged_stash 1176 ? cbor_tagged_stash
1052 : gv_stashpv ("CBOR::XS::Tagged" , 1); 1177 : gv_stashpv ("CBOR::XS::Tagged" , 1);
… …
1155} 1280}
1156 1281
1157static SV * 1282static SV *
1158decode_cbor (SV *string, CBOR *cbor, char **offset_return) 1283decode_cbor (SV *string, CBOR *cbor, char **offset_return)
1159{ 1284{
1160 dec_t dec = { }; 1285 dec_t dec = { 0 };
1161 SV *sv; 1286 SV *sv;
1162 STRLEN len; 1287 STRLEN len;
1163 char *data = SvPVbyte (string, len); 1288 char *data = SvPVbyte (string, len);
1164 1289
1165 if (len > cbor->max_size && cbor->max_size) 1290 if (len > cbor->max_size && cbor->max_size)
… …
1181 1306
1182 if (dec.err) 1307 if (dec.err)
1183 { 1308 {
1184 if (dec.shareable) 1309 if (dec.shareable)
1185 { 1310 {
1186 // need to break cyclic links, which whould all be in shareable 1311 // need to break cyclic links, which would all be in shareable
1187 int i; 1312 int i;
1188 SV **svp; 1313 SV **svp;
1189 1314
1190 for (i = av_len (dec.shareable) + 1; i--; ) 1315 for (i = av_len (dec.shareable) + 1; i--; )
1191 if ((svp = av_fetch (dec.shareable, i, 0))) 1316 if ((svp = av_fetch (dec.shareable, i, 0)))
1192 sv_setsv (*svp, &PL_sv_undef); 1317 sv_setsv (*svp, &PL_sv_undef);
1193 } 1318 }
1194 1319
1195 SvREFCNT_dec (sv); 1320 SvREFCNT_dec_NN (sv);
1321
1322 if (dec.err_sv)
1323 sv_2mortal (dec.err_sv);
1324
1196 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur); 1325 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur);
1197 } 1326 }
1198 1327
1199 sv = sv_2mortal (sv); 1328 sv = sv_2mortal (sv);
1200 1329
… …
1256 1385
1257 int major = *p >> MAJOR_SHIFT; 1386 int major = *p >> MAJOR_SHIFT;
1258 1387
1259 switch (major) 1388 switch (major)
1260 { 1389 {
1390 case MAJOR_TAG >> MAJOR_SHIFT:
1391 ++count; // tags merely prefix another value
1392 break;
1393
1261 case MAJOR_BYTES >> MAJOR_SHIFT: 1394 case MAJOR_BYTES >> MAJOR_SHIFT:
1262 case MAJOR_TEXT >> MAJOR_SHIFT: 1395 case MAJOR_TEXT >> MAJOR_SHIFT:
1263 case MAJOR_ARRAY >> MAJOR_SHIFT: 1396 case MAJOR_ARRAY >> MAJOR_SHIFT:
1264 case MAJOR_MAP >> MAJOR_SHIFT: 1397 case MAJOR_MAP >> MAJOR_SHIFT:
1265 { 1398 {
… …
1340 1473
1341 default_filter = newSVpv ("CBOR::XS::default_filter", 0); 1474 default_filter = newSVpv ("CBOR::XS::default_filter", 0);
1342 1475
1343 sv_cbor = newSVpv ("CBOR", 0); 1476 sv_cbor = newSVpv ("CBOR", 0);
1344 SvREADONLY_on (sv_cbor); 1477 SvREADONLY_on (sv_cbor);
1478
1479 assert (("STRLEN must be an unsigned type", 0 <= (STRLEN)-1));
1345} 1480}
1346 1481
1347PROTOTYPES: DISABLE 1482PROTOTYPES: DISABLE
1348 1483
1349void CLONE (...) 1484void CLONE (...)
… …
1369 ALIAS: 1504 ALIAS:
1370 shrink = F_SHRINK 1505 shrink = F_SHRINK
1371 allow_unknown = F_ALLOW_UNKNOWN 1506 allow_unknown = F_ALLOW_UNKNOWN
1372 allow_sharing = F_ALLOW_SHARING 1507 allow_sharing = F_ALLOW_SHARING
1373 allow_cycles = F_ALLOW_CYCLES 1508 allow_cycles = F_ALLOW_CYCLES
1509 forbid_objects = F_FORBID_OBJECTS
1374 pack_strings = F_PACK_STRINGS 1510 pack_strings = F_PACK_STRINGS
1511 text_keys = F_TEXT_KEYS
1512 text_strings = F_TEXT_STRINGS
1375 validate_utf8 = F_VALIDATE_UTF8 1513 validate_utf8 = F_VALIDATE_UTF8
1376 PPCODE: 1514 PPCODE:
1377{ 1515{
1378 if (enable) 1516 if (enable)
1379 self->flags |= ix; 1517 self->flags |= ix;
… …
1387 ALIAS: 1525 ALIAS:
1388 get_shrink = F_SHRINK 1526 get_shrink = F_SHRINK
1389 get_allow_unknown = F_ALLOW_UNKNOWN 1527 get_allow_unknown = F_ALLOW_UNKNOWN
1390 get_allow_sharing = F_ALLOW_SHARING 1528 get_allow_sharing = F_ALLOW_SHARING
1391 get_allow_cycles = F_ALLOW_CYCLES 1529 get_allow_cycles = F_ALLOW_CYCLES
1530 get_forbid_objects = F_FORBID_OBJECTS
1392 get_pack_strings = F_PACK_STRINGS 1531 get_pack_strings = F_PACK_STRINGS
1532 get_text_keys = F_TEXT_KEYS
1533 get_text_strings = F_TEXT_STRINGS
1393 get_validate_utf8 = F_VALIDATE_UTF8 1534 get_validate_utf8 = F_VALIDATE_UTF8
1394 PPCODE: 1535 PPCODE:
1395 XPUSHs (boolSV (self->flags & ix)); 1536 XPUSHs (boolSV (self->flags & ix));
1396 1537
1397void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) 1538void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
… …
1447 EXTEND (SP, 2); 1588 EXTEND (SP, 2);
1448 PUSHs (sv); 1589 PUSHs (sv);
1449 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); 1590 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
1450} 1591}
1451 1592
1452void incr_parse (CBOR *self, SV *cborstr, int chop = 0) 1593void incr_parse (CBOR *self, SV *cborstr)
1594 ALIAS:
1595 incr_parse_multiple = 1
1453 PPCODE: 1596 PPCODE:
1454{ 1597{
1455 if (SvUTF8 (cborstr)) 1598 if (SvUTF8 (cborstr))
1456 sv_utf8_downgrade (cborstr, 0); 1599 sv_utf8_downgrade (cborstr, 0);
1457 1600
… …
1462 self->incr_need = 1; 1605 self->incr_need = 1;
1463 1606
1464 av_push (self->incr_count, newSViv (1)); 1607 av_push (self->incr_count, newSViv (1));
1465 } 1608 }
1466 1609
1467 for (;;) 1610 do
1468 { 1611 {
1469 if (!incr_parse (self, cborstr)) 1612 if (!incr_parse (self, cborstr))
1470 { 1613 {
1471 if (self->incr_need > self->max_size && self->max_size) 1614 if (self->incr_need > self->max_size && self->max_size)
1472 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu", 1615 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
1473 (unsigned long)self->incr_need, (unsigned long)self->max_size); 1616 (unsigned long)self->incr_need, (unsigned long)self->max_size);
1474 1617
1475 break; 1618 break;
1476 } 1619 }
1477 1620
1478 SV *sv; 1621 SV *sv;
1479 char *offset; 1622 char *offset;
1480 1623
1481 PUTBACK; sv = decode_cbor (cborstr, self, &offset); SPAGAIN; 1624 PUTBACK; sv = decode_cbor (cborstr, self, &offset); SPAGAIN;
1482 XPUSHs (sv); 1625 XPUSHs (sv);
1483 1626
1627 sv_chop (cborstr, offset);
1628
1484 av_clear (self->incr_count); 1629 av_clear (self->incr_count);
1485 av_push (self->incr_count, newSViv (1)); 1630 av_push (self->incr_count, newSViv (1));
1486 1631
1487 if (chop)
1488 {
1489 self->incr_pos = 0; 1632 self->incr_pos = 0;
1490 sv_chop (cborstr, offset);
1491 }
1492 else
1493 self->incr_pos = offset - SvPVX (cborstr);
1494
1495 self->incr_need = self->incr_pos + 1; 1633 self->incr_need = self->incr_pos + 1;
1496 } 1634 }
1635 while (ix);
1497} 1636}
1498 1637
1499void incr_reset (CBOR *self) 1638void incr_reset (CBOR *self)
1500 CODE: 1639 CODE:
1501{ 1640{

Diff Legend

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