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 1.55 by root, Fri Nov 25 06:13:16 2016 UTC vs.
Revision 1.70 by root, Fri Dec 4 02:57:14 2020 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
12#define ECB_NO_THREADS 1 13#define ECB_NO_THREADS 1
13#include "ecb.h" 14#include "ecb.h"
14 15
15// compatibility with perl <5.18 16// compatibility with perl <5.18
… …
20# define HvNAMELEN(hv) HvNAMELEN_get (hv) 21# define HvNAMELEN(hv) HvNAMELEN_get (hv)
21#endif 22#endif
22#ifndef HvNAMEUTF8 23#ifndef HvNAMEUTF8
23# define HvNAMEUTF8(hv) 0 24# define HvNAMEUTF8(hv) 0
24#endif 25#endif
26#ifndef SvREFCNT_inc_NN
27# define SvREFCNT_inc_NN(sv) SvREFCNT_inc (sv)
28#endif
25#ifndef SvREFCNT_dec_NN 29#ifndef SvREFCNT_dec_NN
26# define SvREFCNT_dec_NN(sv) SvREFCNT_dec (sv) 30# define SvREFCNT_dec_NN(sv) SvREFCNT_dec (sv)
27#endif 31#endif
28 32
29// known major and minor types 33// known major and minor types
… …
95 CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8 99 CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8
96 100
97 CBOR_TAG_MAGIC = 55799, // self-describe cbor 101 CBOR_TAG_MAGIC = 55799, // self-describe cbor
98}; 102};
99 103
104// known forced types, also hardcoded in CBOR.pm
105enum
106{
107 AS_CBOR = 0,
108 AS_INT = 1,
109 AS_BYTES = 2,
110 AS_TEXT = 3,
111 AS_FLOAT16 = 4,
112 AS_FLOAT32 = 5,
113 AS_FLOAT64 = 6,
114 AS_MAP = 7,
115 // possibly future enhancements: (generic) float, (generic) string
116};
117
100#define F_SHRINK 0x00000001UL 118#define F_SHRINK 0x00000001UL
101#define F_ALLOW_UNKNOWN 0x00000002UL 119#define F_ALLOW_UNKNOWN 0x00000002UL
102#define F_ALLOW_SHARING 0x00000004UL 120#define F_ALLOW_SHARING 0x00000004UL
103#define F_ALLOW_CYCLES 0x00000008UL 121#define F_ALLOW_CYCLES 0x00000008UL
122#define F_FORBID_OBJECTS 0x00000010UL
104#define F_PACK_STRINGS 0x00000010UL 123#define F_PACK_STRINGS 0x00000020UL
105#define F_TEXT_KEYS 0x00000020UL 124#define F_TEXT_KEYS 0x00000040UL
106#define F_TEXT_STRINGS 0x00000040UL 125#define F_TEXT_STRINGS 0x00000080UL
107#define F_VALIDATE_UTF8 0x00000080UL 126#define F_VALIDATE_UTF8 0x00000100UL
108 127
109#define INIT_SIZE 32 // initial scalar size to be allocated 128#define INIT_SIZE 32 // initial scalar size to be allocated
110 129
111#define SB do { 130#define SB do {
112#define SE } while (0) 131#define SE } while (0)
… …
185 204
186// minimum length of a string to be registered for stringref 205// minimum length of a string to be registered for stringref
187ecb_inline int 206ecb_inline int
188minimum_string_length (UV idx) 207minimum_string_length (UV idx)
189{ 208{
190 return idx > 23 209 return idx <= 23 ? 3
191 ? idx > 0xffU 210 : idx <= 0xffU ? 4
192 ? idx > 0xffffU 211 : idx <= 0xffffU ? 5
193 ? idx > 0xffffffffU 212 : idx <= 0xffffffffU ? 7
194 ? 11 213 : 11;
195 : 7
196 : 5
197 : 4
198 : 3;
199} 214}
200 215
201///////////////////////////////////////////////////////////////////////////// 216/////////////////////////////////////////////////////////////////////////////
202// encoder 217// encoder
203 218
… …
216} enc_t; 231} enc_t;
217 232
218ecb_inline void 233ecb_inline void
219need (enc_t *enc, STRLEN len) 234need (enc_t *enc, STRLEN len)
220{ 235{
221 if (ecb_expect_false (enc->cur + len >= enc->end)) 236 if (ecb_expect_false ((uintptr_t)(enc->end - enc->cur) < len))
222 { 237 {
223 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv); 238 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
224 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1); 239 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
225 enc->cur = SvPVX (enc->sv) + cur; 240 enc->cur = SvPVX (enc->sv) + cur;
226 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; 241 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
227 } 242 }
228} 243}
229 244
245static void encode_sv (enc_t *enc, SV *sv);
246
230ecb_inline void 247ecb_inline void
231encode_ch (enc_t *enc, char ch) 248encode_ch (enc_t *enc, char ch)
232{ 249{
233 need (enc, 1); 250 need (enc, 1);
234 *enc->cur++ = ch; 251 *enc->cur++ = ch;
235} 252}
236 253
254// used for tags, intregers, element counts and so on
237static void 255static void
238encode_uint (enc_t *enc, int major, UV len) 256encode_uint (enc_t *enc, int major, UV len)
239{ 257{
240 need (enc, 9); 258 need (enc, 9);
241 259
… …
272 *enc->cur++ = len >> 8; 290 *enc->cur++ = len >> 8;
273 *enc->cur++ = len; 291 *enc->cur++ = len;
274 } 292 }
275} 293}
276 294
295// encodes a perl value into a CBOR integer
296ecb_inline void
297encode_int (enc_t *enc, SV *sv)
298{
299 if (SvIsUV (sv))
300 encode_uint (enc, MAJOR_POS_INT, SvUVX (sv));
301 else if (SvIVX (sv) >= 0)
302 encode_uint (enc, MAJOR_POS_INT, SvIVX (sv));
303 else
304 encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1));
305}
306
277ecb_inline void 307ecb_inline void
278encode_tag (enc_t *enc, UV tag) 308encode_tag (enc_t *enc, UV tag)
279{ 309{
280 encode_uint (enc, MAJOR_TAG, tag); 310 encode_uint (enc, MAJOR_TAG, tag);
281} 311}
… …
342 } 372 }
343 373
344 encode_str (enc, upgrade_utf8, utf8, str, len); 374 encode_str (enc, upgrade_utf8, utf8, str, len);
345} 375}
346 376
347static void encode_sv (enc_t *enc, SV *sv); 377ecb_inline void
378encode_float16 (enc_t *enc, NV nv)
379{
380 need (enc, 1+2);
381
382 *enc->cur++ = MAJOR_MISC | MISC_FLOAT16;
383
384 uint16_t fp = ecb_float_to_binary16 (nv);
385
386 if (!ecb_big_endian ())
387 fp = ecb_bswap16 (fp);
388
389 memcpy (enc->cur, &fp, 2);
390 enc->cur += 2;
391}
392
393ecb_inline void
394encode_float32 (enc_t *enc, NV nv)
395{
396 need (enc, 1+4);
397
398 *enc->cur++ = MAJOR_MISC | MISC_FLOAT32;
399
400 uint32_t fp = ecb_float_to_binary32 (nv);
401
402 if (!ecb_big_endian ())
403 fp = ecb_bswap32 (fp);
404
405 memcpy (enc->cur, &fp, 4);
406 enc->cur += 4;
407}
408
409ecb_inline void
410encode_float64 (enc_t *enc, NV nv)
411{
412 need (enc, 1+8);
413
414 *enc->cur++ = MAJOR_MISC | MISC_FLOAT64;
415
416 uint64_t fp = ecb_double_to_binary64 (nv);
417
418 if (!ecb_big_endian ())
419 fp = ecb_bswap64 (fp);
420
421 memcpy (enc->cur, &fp, 8);
422 enc->cur += 8;
423}
424
425ecb_inline void
426encode_bool (enc_t *enc, int istrue)
427{
428 encode_ch (enc, istrue ? MAJOR_MISC | SIMPLE_TRUE : MAJOR_MISC | SIMPLE_FALSE);
429}
430
431// encodes an arrayref containing key-value pairs as CBOR map
432ecb_inline void
433encode_array_as_map (enc_t *enc, SV *sv)
434{
435 if (enc->depth >= enc->cbor.max_depth)
436 croak (ERR_NESTING_EXCEEDED);
437
438 ++enc->depth;
439
440 // as_map does error checking for us, but we re-check in case
441 // things have changed.
442
443 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
444 croak ("CBOR::XS::as_map requires an array reference (did you change the array after calling as_map?)");
445
446 AV *av = (AV *)SvRV (sv);
447 int i, len = av_len (av);
448
449 if (!(len & 1))
450 croak ("CBOR::XS::as_map requires an even number of elements (did you change the array after calling as_map?)");
451
452 encode_uint (enc, MAJOR_MAP, (len + 1) >> 1);
453
454 for (i = 0; i <= len; ++i)
455 {
456 SV **svp = av_fetch (av, i, 0);
457 encode_sv (enc, svp ? *svp : &PL_sv_undef);
458 }
459
460 --enc->depth;
461}
462
463ecb_inline void
464encode_forced (enc_t *enc, UV type, SV *sv)
465{
466 switch (type)
467 {
468 case AS_CBOR:
469 {
470 STRLEN len;
471 char *str = SvPVbyte (sv, len);
472
473 need (enc, len);
474 memcpy (enc->cur, str, len);
475 enc->cur += len;
476 }
477 break;
478
479 case AS_BYTES:
480 {
481 STRLEN len;
482 char *str = SvPVbyte (sv, len);
483 encode_strref (enc, 0, 0, str, len);
484 }
485 break;
486
487 case AS_TEXT:
488 {
489 STRLEN len;
490 char *str = SvPVutf8 (sv, len);
491 encode_strref (enc, 1, 1, str, len);
492 }
493 break;
494
495 case AS_INT: encode_int (enc, sv); break;
496
497 case AS_FLOAT16: encode_float16 (enc, SvNV (sv)); break;
498 case AS_FLOAT32: encode_float32 (enc, SvNV (sv)); break;
499 case AS_FLOAT64: encode_float64 (enc, SvNV (sv)); break;
500
501 case AS_MAP: encode_array_as_map (enc, sv); break;
502
503 default:
504 croak ("encountered malformed CBOR::XS::Tagged object");
505 }
506}
348 507
349static void 508static void
350encode_av (enc_t *enc, AV *av) 509encode_av (enc_t *enc, AV *av)
351{ 510{
352 int i, len = av_len (av); 511 int i, len = av_len (av);
… …
356 515
357 ++enc->depth; 516 ++enc->depth;
358 517
359 encode_uint (enc, MAJOR_ARRAY, len + 1); 518 encode_uint (enc, MAJOR_ARRAY, len + 1);
360 519
361 if (SvMAGICAL (av)) 520 if (ecb_expect_false (SvMAGICAL (av)))
362 for (i = 0; i <= len; ++i) 521 for (i = 0; i <= len; ++i)
363 { 522 {
364 SV **svp = av_fetch (av, i, 0); 523 SV **svp = av_fetch (av, i, 0);
365 encode_sv (enc, svp ? *svp : &PL_sv_undef); 524 encode_sv (enc, svp ? *svp : &PL_sv_undef);
366 } 525 }
… …
385 ++enc->depth; 544 ++enc->depth;
386 545
387 int pairs = hv_iterinit (hv); 546 int pairs = hv_iterinit (hv);
388 int mg = SvMAGICAL (hv); 547 int mg = SvMAGICAL (hv);
389 548
390 if (mg) 549 if (ecb_expect_false (mg))
391 encode_ch (enc, MAJOR_MAP | MINOR_INDEF); 550 encode_ch (enc, MAJOR_MAP | MINOR_INDEF);
392 else 551 else
393 encode_uint (enc, MAJOR_MAP, pairs); 552 encode_uint (enc, MAJOR_MAP, pairs);
394 553
395 while ((he = hv_iternext (hv))) 554 while ((he = hv_iternext (hv)))
… …
400 encode_strref (enc, enc->cbor.flags & (F_TEXT_KEYS | F_TEXT_STRINGS), HeKUTF8 (he), HeKEY (he), HeKLEN (he)); 559 encode_strref (enc, enc->cbor.flags & (F_TEXT_KEYS | F_TEXT_STRINGS), HeKUTF8 (he), HeKEY (he), HeKLEN (he));
401 560
402 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he)); 561 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he));
403 } 562 }
404 563
405 if (mg) 564 if (ecb_expect_false (mg))
406 encode_ch (enc, MAJOR_MISC | MINOR_INDEF); 565 encode_ch (enc, MAJOR_MISC | MINOR_INDEF);
407 566
408 --enc->depth; 567 --enc->depth;
409} 568}
410 569
… …
430 589
431 HV *stash = SvSTASH (sv); 590 HV *stash = SvSTASH (sv);
432 591
433 if (stash == boolean_stash) 592 if (stash == boolean_stash)
434 { 593 {
435 encode_ch (enc, SvIV (sv) ? MAJOR_MISC | SIMPLE_TRUE : MAJOR_MISC | SIMPLE_FALSE); 594 encode_bool (enc, SvIV (sv));
436 return; 595 return;
437 } 596 }
438 else if (stash == error_stash) 597 else if (stash == error_stash)
439 { 598 {
440 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF); 599 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
… …
443 else if (stash == tagged_stash) 602 else if (stash == tagged_stash)
444 { 603 {
445 if (svt != SVt_PVAV) 604 if (svt != SVt_PVAV)
446 croak ("encountered CBOR::XS::Tagged object that isn't an array"); 605 croak ("encountered CBOR::XS::Tagged object that isn't an array");
447 606
607 switch (av_len ((AV *)sv))
608 {
609 case 2-1:
610 // actually a tagged value
448 encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1))); 611 encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1)));
449 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1)); 612 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1));
613 break;
614
615 case 3-1:
616 // a forced type [value, type, undef]
617 encode_forced (enc, SvUV (*av_fetch ((AV *)sv, 1, 1)), *av_fetch ((AV *)sv, 0, 1));
618 break;
619
620 default:
621 croak ("encountered malformed CBOR::XS::Tagged object");
622 }
450 623
451 return; 624 return;
452 } 625 }
453 } 626 }
454 627
455 if (ecb_expect_false (SvREFCNT (sv) > 1) 628 if (ecb_expect_false (SvREFCNT (sv) > 1)
456 && ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING)) 629 && ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING))
457 { 630 {
458 if (!enc->shareable) 631 if (ecb_expect_false (!enc->shareable))
459 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ()); 632 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
460 633
461 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1); 634 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
462 635
463 if (SvOK (*svp)) 636 if (SvOK (*svp))
… …
477 if (ecb_expect_false (SvOBJECT (sv))) 650 if (ecb_expect_false (SvOBJECT (sv)))
478 { 651 {
479 HV *stash = SvSTASH (sv); 652 HV *stash = SvSTASH (sv);
480 GV *method; 653 GV *method;
481 654
655 if (enc->cbor.flags & F_FORBID_OBJECTS)
656 croak ("encountered object '%s', but forbid_objects is enabled",
657 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
482 if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0))) 658 else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
483 { 659 {
484 dSP; 660 dSP;
485 661
486 ENTER; SAVETMPS; 662 ENTER; SAVETMPS;
487 PUSHMARK (SP); 663 PUSHMARK (SP);
… …
506 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0) 682 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)
507 { 683 {
508 dSP; 684 dSP;
509 685
510 ENTER; SAVETMPS; 686 ENTER; SAVETMPS;
511 SAVESTACK_POS ();
512 PUSHMARK (SP); 687 PUSHMARK (SP);
513 EXTEND (SP, 2); 688 EXTEND (SP, 2);
514 // we re-bless the reference to get overload and other niceties right 689 // we re-bless the reference to get overload and other niceties right
515 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); 690 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
516 PUSHs (sv_cbor); 691 PUSHs (sv_cbor);
… …
525 700
526 encode_tag (enc, CBOR_TAG_PERL_OBJECT); 701 encode_tag (enc, CBOR_TAG_PERL_OBJECT);
527 encode_uint (enc, MAJOR_ARRAY, count + 1); 702 encode_uint (enc, MAJOR_ARRAY, count + 1);
528 encode_strref (enc, 0, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash)); 703 encode_strref (enc, 0, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
529 704
530 while (count) 705 {
706 int i;
707
708 for (i = 0; i < count; ++i)
531 encode_sv (enc, SP[1 - count--]); 709 encode_sv (enc, SP[i + 1 - count]);
710
711 SP -= count;
712 }
532 713
533 PUTBACK; 714 PUTBACK;
534 715
535 FREETMPS; LEAVE; 716 FREETMPS; LEAVE;
536 } 717 }
… …
558 739
559 if (ecb_expect_false (nv == (NV)(U32)nv)) 740 if (ecb_expect_false (nv == (NV)(U32)nv))
560 encode_uint (enc, MAJOR_POS_INT, (U32)nv); 741 encode_uint (enc, MAJOR_POS_INT, (U32)nv);
561 //TODO: maybe I32? 742 //TODO: maybe I32?
562 else if (ecb_expect_false (nv == (float)nv)) 743 else if (ecb_expect_false (nv == (float)nv))
563 { 744 encode_float32 (enc, nv);
564 uint32_t fp = ecb_float_to_binary32 (nv);
565
566 *enc->cur++ = MAJOR_MISC | MISC_FLOAT32;
567
568 if (!ecb_big_endian ())
569 fp = ecb_bswap32 (fp);
570
571 memcpy (enc->cur, &fp, 4);
572 enc->cur += 4;
573 }
574 else 745 else
575 { 746 encode_float64 (enc, nv);
576 uint64_t fp = ecb_double_to_binary64 (nv);
577
578 *enc->cur++ = MAJOR_MISC | MISC_FLOAT64;
579
580 if (!ecb_big_endian ())
581 fp = ecb_bswap64 (fp);
582
583 memcpy (enc->cur, &fp, 8);
584 enc->cur += 8;
585 }
586} 747}
587 748
588static void 749static void
589encode_sv (enc_t *enc, SV *sv) 750encode_sv (enc_t *enc, SV *sv)
590{ 751{
… …
597 encode_strref (enc, enc->cbor.flags & F_TEXT_STRINGS, SvUTF8 (sv), str, len); 758 encode_strref (enc, enc->cbor.flags & F_TEXT_STRINGS, SvUTF8 (sv), str, len);
598 } 759 }
599 else if (SvNOKp (sv)) 760 else if (SvNOKp (sv))
600 encode_nv (enc, sv); 761 encode_nv (enc, sv);
601 else if (SvIOKp (sv)) 762 else if (SvIOKp (sv))
602 { 763 encode_int (enc, sv);
603 if (SvIsUV (sv))
604 encode_uint (enc, MAJOR_POS_INT, SvUVX (sv));
605 else if (SvIVX (sv) >= 0)
606 encode_uint (enc, MAJOR_POS_INT, SvIVX (sv));
607 else
608 encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1));
609 }
610 else if (SvROK (sv)) 764 else if (SvROK (sv))
611 encode_rv (enc, SvRV (sv)); 765 encode_rv (enc, SvRV (sv));
612 else if (!SvOK (sv)) 766 else if (!SvOK (sv))
613 encode_ch (enc, MAJOR_MISC | SIMPLE_NULL); 767 encode_ch (enc, MAJOR_MISC | SIMPLE_NULL);
614 else if (enc->cbor.flags & F_ALLOW_UNKNOWN) 768 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
… …
621static SV * 775static SV *
622encode_cbor (SV *scalar, CBOR *cbor) 776encode_cbor (SV *scalar, CBOR *cbor)
623{ 777{
624 enc_t enc = { 0 }; 778 enc_t enc = { 0 };
625 779
626 enc.cbor = *cbor; 780 enc.cbor = *cbor;
627 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 781 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
628 enc.cur = SvPVX (enc.sv); 782 enc.cur = SvPVX (enc.sv);
629 enc.end = SvEND (enc.sv); 783 enc.end = SvEND (enc.sv);
630 784
631 SvPOK_only (enc.sv); 785 SvPOK_only (enc.sv);
632 786
633 if (cbor->flags & F_PACK_STRINGS) 787 if (cbor->flags & F_PACK_STRINGS)
634 { 788 {
… …
661 U32 depth; // recursion depth 815 U32 depth; // recursion depth
662 U32 maxdepth; // recursion depth limit 816 U32 maxdepth; // recursion depth limit
663 AV *shareable; 817 AV *shareable;
664 AV *stringref; 818 AV *stringref;
665 SV *decode_tagged; 819 SV *decode_tagged;
820 SV *err_sv; // optional sv for error, needs to be freed
666} dec_t; 821} dec_t;
667 822
668#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE 823// set dec->err to ERRSV
824ecb_cold static void
825err_errsv (dec_t *dec)
826{
827 if (!dec->err)
828 {
829 dec->err_sv = newSVsv (ERRSV);
669 830
670#define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data") 831 // chop off the trailing \n
832 SvCUR_set (dec->err_sv, SvCUR (dec->err_sv) - 1);
833 *SvEND (dec->err_sv) = 0;
671 834
835 dec->err = SvPVutf8_nolen (dec->err_sv);
836 }
837}
838
839// the following functions are used to reduce code size and help the compiler to optimise
840ecb_cold static void
841err_set (dec_t *dec, const char *reason)
842{
843 if (!dec->err)
844 dec->err = reason;
845}
846
847ecb_cold static void
848err_unexpected_end (dec_t *dec)
849{
850 err_set (dec, "unexpected end of CBOR data");
851}
852
853#define ERR_DO(do) SB do; goto fail; SE
854#define ERR(reason) ERR_DO (err_set (dec, reason))
855#define ERR_ERRSV ERR_DO (err_errsv (dec))
856
857#define WANT(len) if (ecb_expect_false ((uintptr_t)(dec->end - dec->cur) < (STRLEN)len)) ERR_DO (err_unexpected_end (dec))
858
672#define DEC_INC_DEPTH if (++dec->depth > dec->cbor.max_depth) ERR (ERR_NESTING_EXCEEDED) 859#define DEC_INC_DEPTH if (ecb_expect_false (++dec->depth > dec->cbor.max_depth)) ERR (ERR_NESTING_EXCEEDED)
673#define DEC_DEC_DEPTH --dec->depth 860#define DEC_DEC_DEPTH --dec->depth
674 861
675static UV 862static UV
676decode_uint (dec_t *dec) 863decode_uint (dec_t *dec)
677{ 864{
… …
743 930
744 for (;;) 931 for (;;)
745 { 932 {
746 WANT (1); 933 WANT (1);
747 934
748 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF)) 935 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF) || dec->err)
749 { 936 {
750 ++dec->cur; 937 ++dec->cur;
751 break; 938 break;
752 } 939 }
753 940
754 av_push (av, decode_sv (dec)); 941 av_push (av, decode_sv (dec));
755 } 942 }
756 } 943 }
757 else 944 else
758 { 945 {
759 int i, len = decode_uint (dec); 946 UV i, len = decode_uint (dec);
760 947
761 WANT (len); // complexity check for av_fill - need at least one byte per value, do not allow supersize arrays 948 WANT (len); // complexity check for av_fill - need at least one byte per value, do not allow supersize arrays
762 av_fill (av, len - 1); 949 av_fill (av, len - 1);
763 950
764 for (i = 0; i < len; ++i) 951 for (i = 0; i < len; ++i)
… …
767 954
768 DEC_DEC_DEPTH; 955 DEC_DEC_DEPTH;
769 return newRV_noinc ((SV *)av); 956 return newRV_noinc ((SV *)av);
770 957
771fail: 958fail:
772 SvREFCNT_dec (av); 959 SvREFCNT_dec_NN (av);
773 DEC_DEC_DEPTH; 960 DEC_DEC_DEPTH;
774 return &PL_sv_undef; 961 return &PL_sv_undef;
775} 962}
776 963
777static void 964static void
… …
781 // byte or utf-8 strings as keys, but only when !stringref 968 // byte or utf-8 strings as keys, but only when !stringref
782 969
783 if (ecb_expect_true (!dec->stringref)) 970 if (ecb_expect_true (!dec->stringref))
784 if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8)) 971 if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8))
785 { 972 {
786 I32 len = decode_uint (dec); 973 STRLEN len = decode_uint (dec);
787 char *key = (char *)dec->cur; 974 char *key = (char *)dec->cur;
788 975
789 WANT (len); 976 WANT (len);
790 dec->cur += len; 977 dec->cur += len;
791 978
… …
793 980
794 return; 981 return;
795 } 982 }
796 else if (ecb_expect_true ((U8)(*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8)) 983 else if (ecb_expect_true ((U8)(*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8))
797 { 984 {
798 I32 len = decode_uint (dec); 985 STRLEN len = decode_uint (dec);
799 char *key = (char *)dec->cur; 986 char *key = (char *)dec->cur;
800 987
801 WANT (len); 988 WANT (len);
802 dec->cur += len; 989 dec->cur += len;
803 990
… …
811 } 998 }
812 999
813 SV *k = decode_sv (dec); 1000 SV *k = decode_sv (dec);
814 SV *v = decode_sv (dec); 1001 SV *v = decode_sv (dec);
815 1002
1003 // we leak memory if uncaught exceptions are thrown by random magical
1004 // methods, and this is hopefully the only place where it can happen,
1005 // so if there is a chance of an exception, take the very slow path.
1006 // since catching exceptions is "undocumented/internal/forbidden" by
1007 // the new p5p powers, we need to call out to a perl function :/
1008 if (ecb_expect_false (SvAMAGIC (k)))
1009 {
1010 dSP;
1011
1012 ENTER; SAVETMPS;
1013 PUSHMARK (SP);
1014 EXTEND (SP, 3);
1015 PUSHs (sv_2mortal (newRV_inc ((SV *)hv)));
1016 PUSHs (sv_2mortal (k));
1017 PUSHs (sv_2mortal (v));
1018
1019 PUTBACK;
1020 call_pv ("CBOR::XS::_hv_store", G_VOID | G_DISCARD | G_EVAL);
1021 SPAGAIN;
1022
1023 FREETMPS; LEAVE;
1024
1025 if (SvTRUE (ERRSV))
1026 ERR_ERRSV;
1027
1028 return;
1029 }
1030
816 hv_store_ent (hv, k, v, 0); 1031 hv_store_ent (hv, k, v, 0);
817 SvREFCNT_dec (k); 1032 SvREFCNT_dec_NN (k);
818 1033
819fail: 1034fail:
820 ; 1035 ;
821} 1036}
822 1037
… …
833 1048
834 for (;;) 1049 for (;;)
835 { 1050 {
836 WANT (1); 1051 WANT (1);
837 1052
838 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF)) 1053 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF) || dec->err)
839 { 1054 {
840 ++dec->cur; 1055 ++dec->cur;
841 break; 1056 break;
842 } 1057 }
843 1058
844 decode_he (dec, hv); 1059 decode_he (dec, hv);
845 } 1060 }
846 } 1061 }
847 else 1062 else
848 { 1063 {
849 int pairs = decode_uint (dec); 1064 UV pairs = decode_uint (dec);
1065
1066 WANT (pairs); // complexity check - need at least one byte per value, do not allow supersize hashes
850 1067
851 while (pairs--) 1068 while (pairs--)
852 decode_he (dec, hv); 1069 decode_he (dec, hv);
853 } 1070 }
854 1071
855 DEC_DEC_DEPTH; 1072 DEC_DEC_DEPTH;
856 return newRV_noinc ((SV *)hv); 1073 return newRV_noinc ((SV *)hv);
857 1074
858fail: 1075fail:
859 SvREFCNT_dec (hv); 1076 SvREFCNT_dec_NN (hv);
860 DEC_DEC_DEPTH; 1077 DEC_DEC_DEPTH;
861 return &PL_sv_undef; 1078 return &PL_sv_undef;
862} 1079}
863 1080
864static SV * 1081static SV *
865decode_str (dec_t *dec, int utf8) 1082decode_str (dec_t *dec, int utf8)
866{ 1083{
867 SV *sv = 0; 1084 SV *sv = 0;
868 1085
869 if ((*dec->cur & MINOR_MASK) == MINOR_INDEF) 1086 if (ecb_expect_false ((*dec->cur & MINOR_MASK) == MINOR_INDEF))
870 { 1087 {
871 // indefinite length strings 1088 // indefinite length strings
872 ++dec->cur; 1089 ++dec->cur;
873 1090
874 U8 major = *dec->cur & MAJOR_MISC; 1091 U8 major = *dec->cur & MAJOR_MISC;
… …
942 sv = newRV_noinc (decode_sv (dec)); 1159 sv = newRV_noinc (decode_sv (dec));
943 break; 1160 break;
944 1161
945 case CBOR_TAG_STRINGREF_NAMESPACE: 1162 case CBOR_TAG_STRINGREF_NAMESPACE:
946 { 1163 {
947 // do nmot use SAVETMPS/FREETMPS, as these will 1164 // do not use SAVETMPS/FREETMPS, as these will
948 // erase mortalised caches, e.g. "shareable" 1165 // erase mortalised caches, e.g. "shareable"
949 ENTER; 1166 ENTER;
950 1167
951 SAVESPTR (dec->stringref); 1168 SAVESPTR (dec->stringref);
952 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ()); 1169 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ());
… …
962 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT)) 1179 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
963 ERR ("corrupted CBOR data (stringref index not an unsigned integer)"); 1180 ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
964 1181
965 UV idx = decode_uint (dec); 1182 UV idx = decode_uint (dec);
966 1183
967 if (!dec->stringref || (int)idx > AvFILLp (dec->stringref)) 1184 if (!dec->stringref || idx >= (UV)(1 + AvFILLp (dec->stringref)))
968 ERR ("corrupted CBOR data (stringref index out of bounds or outside namespace)"); 1185 ERR ("corrupted CBOR data (stringref index out of bounds or outside namespace)");
969 1186
970 sv = newSVsv (AvARRAY (dec->stringref)[idx]); 1187 sv = newSVsv (AvARRAY (dec->stringref)[idx]);
971 } 1188 }
972 break; 1189 break;
… …
1000 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT)) 1217 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
1001 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)"); 1218 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
1002 1219
1003 UV idx = decode_uint (dec); 1220 UV idx = decode_uint (dec);
1004 1221
1005 if (!dec->shareable || (int)idx > AvFILLp (dec->shareable)) 1222 if (!dec->shareable || idx >= (UV)(1 + AvFILLp (dec->shareable)))
1006 ERR ("corrupted CBOR data (sharedref index out of bounds)"); 1223 ERR ("corrupted CBOR data (sharedref index out of bounds)");
1007 1224
1008 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]); 1225 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]);
1009 1226
1010 if (sv == &PL_sv_undef) 1227 if (sv == &PL_sv_undef)
… …
1012 } 1229 }
1013 break; 1230 break;
1014 1231
1015 case CBOR_TAG_PERL_OBJECT: 1232 case CBOR_TAG_PERL_OBJECT:
1016 { 1233 {
1234 if (dec->cbor.flags & F_FORBID_OBJECTS)
1235 goto filter;
1236
1017 sv = decode_sv (dec); 1237 sv = decode_sv (dec);
1018 1238
1019 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV) 1239 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
1020 ERR ("corrupted CBOR data (non-array perl object)"); 1240 ERR ("corrupted CBOR data (non-array perl object)");
1021 1241
… …
1050 SPAGAIN; 1270 SPAGAIN;
1051 1271
1052 if (SvTRUE (ERRSV)) 1272 if (SvTRUE (ERRSV))
1053 { 1273 {
1054 FREETMPS; LEAVE; 1274 FREETMPS; LEAVE;
1055 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV)))); 1275 ERR_ERRSV;
1056 } 1276 }
1057 1277
1058 SvREFCNT_dec (sv); 1278 SvREFCNT_dec_NN (sv);
1059 sv = SvREFCNT_inc (POPs); 1279 sv = SvREFCNT_inc (POPs);
1060 1280
1061 PUTBACK; 1281 PUTBACK;
1062 1282
1063 FREETMPS; LEAVE; 1283 FREETMPS; LEAVE;
1064 } 1284 }
1065 break; 1285 break;
1066 1286
1067 default: 1287 default:
1288 filter:
1068 { 1289 {
1290 SV *tag_sv = newSVuv (tag);
1291
1069 sv = decode_sv (dec); 1292 sv = decode_sv (dec);
1070 1293
1071 dSP; 1294 dSP;
1072 ENTER; SAVETMPS; 1295 ENTER; SAVETMPS;
1073 SAVESTACK_POS ();
1074 PUSHMARK (SP); 1296 PUSHMARK (SP);
1075 EXTEND (SP, 2); 1297 EXTEND (SP, 2);
1076 PUSHs (newSVuv (tag)); 1298 PUSHs (tag_sv);
1077 PUSHs (sv); 1299 PUSHs (sv);
1078 1300
1079 PUTBACK; 1301 PUTBACK;
1080 int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL); 1302 int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL);
1081 SPAGAIN; 1303 SPAGAIN;
1082 1304
1083 if (SvTRUE (ERRSV)) 1305 if (SvTRUE (ERRSV))
1084 { 1306 {
1307 SvREFCNT_dec_NN (tag_sv);
1085 FREETMPS; LEAVE; 1308 FREETMPS; LEAVE;
1086 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV)))); 1309 ERR_ERRSV;
1087 } 1310 }
1088 1311
1089 if (count) 1312 if (count)
1090 { 1313 {
1314 SvREFCNT_dec_NN (tag_sv);
1091 SvREFCNT_dec (sv); 1315 SvREFCNT_dec_NN (sv);
1092 sv = SvREFCNT_inc (POPs); 1316 sv = SvREFCNT_inc_NN (TOPs);
1317 SP -= count;
1093 } 1318 }
1094 else 1319 else
1095 { 1320 {
1096 AV *av = newAV (); 1321 AV *av = newAV ();
1097 av_push (av, newSVuv (tag)); 1322 av_push (av, tag_sv);
1098 av_push (av, sv); 1323 av_push (av, sv);
1099 1324
1100 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 1325 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
1101 ? cbor_tagged_stash 1326 ? cbor_tagged_stash
1102 : gv_stashpv ("CBOR::XS::Tagged" , 1); 1327 : gv_stashpv ("CBOR::XS::Tagged" , 1);
… …
1231 1456
1232 if (dec.err) 1457 if (dec.err)
1233 { 1458 {
1234 if (dec.shareable) 1459 if (dec.shareable)
1235 { 1460 {
1236 // need to break cyclic links, which whould all be in shareable 1461 // need to break cyclic links, which would all be in shareable
1237 int i; 1462 int i;
1238 SV **svp; 1463 SV **svp;
1239 1464
1240 for (i = av_len (dec.shareable) + 1; i--; ) 1465 for (i = av_len (dec.shareable) + 1; i--; )
1241 if ((svp = av_fetch (dec.shareable, i, 0))) 1466 if ((svp = av_fetch (dec.shareable, i, 0)))
1242 sv_setsv (*svp, &PL_sv_undef); 1467 sv_setsv (*svp, &PL_sv_undef);
1243 } 1468 }
1244 1469
1245 SvREFCNT_dec (sv); 1470 SvREFCNT_dec_NN (sv);
1471
1472 if (dec.err_sv)
1473 sv_2mortal (dec.err_sv);
1474
1246 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur); 1475 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur);
1247 } 1476 }
1248 1477
1249 sv = sv_2mortal (sv); 1478 sv = sv_2mortal (sv);
1250 1479
… …
1394 1623
1395 default_filter = newSVpv ("CBOR::XS::default_filter", 0); 1624 default_filter = newSVpv ("CBOR::XS::default_filter", 0);
1396 1625
1397 sv_cbor = newSVpv ("CBOR", 0); 1626 sv_cbor = newSVpv ("CBOR", 0);
1398 SvREADONLY_on (sv_cbor); 1627 SvREADONLY_on (sv_cbor);
1628
1629 assert (("STRLEN must be an unsigned type", 0 <= (STRLEN)-1));
1399} 1630}
1400 1631
1401PROTOTYPES: DISABLE 1632PROTOTYPES: DISABLE
1402 1633
1403void CLONE (...) 1634void CLONE (...)
… …
1423 ALIAS: 1654 ALIAS:
1424 shrink = F_SHRINK 1655 shrink = F_SHRINK
1425 allow_unknown = F_ALLOW_UNKNOWN 1656 allow_unknown = F_ALLOW_UNKNOWN
1426 allow_sharing = F_ALLOW_SHARING 1657 allow_sharing = F_ALLOW_SHARING
1427 allow_cycles = F_ALLOW_CYCLES 1658 allow_cycles = F_ALLOW_CYCLES
1659 forbid_objects = F_FORBID_OBJECTS
1428 pack_strings = F_PACK_STRINGS 1660 pack_strings = F_PACK_STRINGS
1429 text_keys = F_TEXT_KEYS 1661 text_keys = F_TEXT_KEYS
1430 text_strings = F_TEXT_STRINGS 1662 text_strings = F_TEXT_STRINGS
1431 validate_utf8 = F_VALIDATE_UTF8 1663 validate_utf8 = F_VALIDATE_UTF8
1432 PPCODE: 1664 PPCODE:
… …
1443 ALIAS: 1675 ALIAS:
1444 get_shrink = F_SHRINK 1676 get_shrink = F_SHRINK
1445 get_allow_unknown = F_ALLOW_UNKNOWN 1677 get_allow_unknown = F_ALLOW_UNKNOWN
1446 get_allow_sharing = F_ALLOW_SHARING 1678 get_allow_sharing = F_ALLOW_SHARING
1447 get_allow_cycles = F_ALLOW_CYCLES 1679 get_allow_cycles = F_ALLOW_CYCLES
1680 get_forbid_objects = F_FORBID_OBJECTS
1448 get_pack_strings = F_PACK_STRINGS 1681 get_pack_strings = F_PACK_STRINGS
1449 get_text_keys = F_TEXT_KEYS 1682 get_text_keys = F_TEXT_KEYS
1450 get_text_strings = F_TEXT_STRINGS 1683 get_text_strings = F_TEXT_STRINGS
1451 get_validate_utf8 = F_VALIDATE_UTF8 1684 get_validate_utf8 = F_VALIDATE_UTF8
1452 PPCODE: 1685 PPCODE:
… …
1585 cbor_init (&cbor); 1818 cbor_init (&cbor);
1586 PUTBACK; cborstr = decode_cbor (cborstr, &cbor, 0); SPAGAIN; 1819 PUTBACK; cborstr = decode_cbor (cborstr, &cbor, 0); SPAGAIN;
1587 XPUSHs (cborstr); 1820 XPUSHs (cborstr);
1588} 1821}
1589 1822
1823#ifdef __AFL_COMPILER
1824
1825void
1826afl_init ()
1827 CODE:
1828 __AFL_INIT ();
1829
1830int
1831afl_loop (unsigned int count = 10000)
1832 CODE:
1833 RETVAL = __AFL_LOOP (count);
1834 OUTPUT:
1835 RETVAL
1836
1837#endif
1838

Diff Legend

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