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 1.58 by root, Fri Nov 25 13:27:29 2016 UTC vs.
Revision 1.75 by root, Thu Sep 7 23:52:24 2023 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
32
33// perl's is_utf8_string interprets len=0 as "calculate len", but we want it to mean 0
34#define cbor_is_utf8_string(str,len) (!(len) || is_utf8_string ((str), (len)))
28 35
29// known major and minor types 36// known major and minor types
30enum cbor_type 37enum cbor_type
31{ 38{
32 MAJOR_SHIFT = 5, 39 MAJOR_SHIFT = 5,
… …
95 CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8 102 CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8
96 103
97 CBOR_TAG_MAGIC = 55799, // self-describe cbor 104 CBOR_TAG_MAGIC = 55799, // self-describe cbor
98}; 105};
99 106
107// known forced types, also hardcoded in CBOR.pm
108enum
109{
110 AS_CBOR = 0,
111 AS_INT = 1,
112 AS_BYTES = 2,
113 AS_TEXT = 3,
114 AS_FLOAT16 = 4,
115 AS_FLOAT32 = 5,
116 AS_FLOAT64 = 6,
117 AS_MAP = 7,
118 // possibly future enhancements: (generic) float, (generic) string
119};
120
100#define F_SHRINK 0x00000001UL 121#define F_SHRINK 0x00000001UL
101#define F_ALLOW_UNKNOWN 0x00000002UL 122#define F_ALLOW_UNKNOWN 0x00000002UL
102#define F_ALLOW_SHARING 0x00000004UL 123#define F_ALLOW_SHARING 0x00000004UL
103#define F_ALLOW_CYCLES 0x00000008UL 124#define F_ALLOW_CYCLES 0x00000008UL
125#define F_FORBID_OBJECTS 0x00000010UL
104#define F_PACK_STRINGS 0x00000010UL 126#define F_PACK_STRINGS 0x00000020UL
105#define F_TEXT_KEYS 0x00000020UL 127#define F_TEXT_KEYS 0x00000040UL
106#define F_TEXT_STRINGS 0x00000040UL 128#define F_TEXT_STRINGS 0x00000080UL
107#define F_VALIDATE_UTF8 0x00000080UL 129#define F_VALIDATE_UTF8 0x00000100UL
108 130
109#define INIT_SIZE 32 // initial scalar size to be allocated 131#define INIT_SIZE 32 // initial scalar size to be allocated
110 132
111#define SB do { 133#define SB do {
112#define SE } while (0) 134#define SE } while (0)
… …
182#endif 204#endif
183 } 205 }
184} 206}
185 207
186// minimum length of a string to be registered for stringref 208// minimum length of a string to be registered for stringref
187ecb_inline int 209ecb_inline STRLEN
188minimum_string_length (UV idx) 210minimum_string_length (UV idx)
189{ 211{
190 return idx > 23 212 return idx <= 23 ? 3
191 ? idx > 0xffU 213 : idx <= 0xffU ? 4
192 ? idx > 0xffffU 214 : idx <= 0xffffU ? 5
193 ? idx > 0xffffffffU 215 : idx <= 0xffffffffU ? 7
194 ? 11 216 : 11;
195 : 7
196 : 5
197 : 4
198 : 3;
199} 217}
200 218
201///////////////////////////////////////////////////////////////////////////// 219/////////////////////////////////////////////////////////////////////////////
202// encoder 220// encoder
203 221
… …
216} enc_t; 234} enc_t;
217 235
218ecb_inline void 236ecb_inline void
219need (enc_t *enc, STRLEN len) 237need (enc_t *enc, STRLEN len)
220{ 238{
221 if (ecb_expect_false (enc->cur + len >= enc->end)) 239 if (ecb_expect_false ((uintptr_t)(enc->end - enc->cur) < len))
222 { 240 {
223 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv); 241 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
224 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1); 242 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
225 enc->cur = SvPVX (enc->sv) + cur; 243 enc->cur = SvPVX (enc->sv) + cur;
226 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; 244 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
227 } 245 }
228} 246}
229 247
248static void encode_sv (enc_t *enc, SV *sv);
249
230ecb_inline void 250ecb_inline void
231encode_ch (enc_t *enc, char ch) 251encode_ch (enc_t *enc, char ch)
232{ 252{
233 need (enc, 1); 253 need (enc, 1);
234 *enc->cur++ = ch; 254 *enc->cur++ = ch;
235} 255}
236 256
257// used for tags, intregers, element counts and so on
237static void 258static void
238encode_uint (enc_t *enc, int major, UV len) 259encode_uint (enc_t *enc, int major, UV len)
239{ 260{
240 need (enc, 9); 261 need (enc, 9);
241 262
… …
272 *enc->cur++ = len >> 8; 293 *enc->cur++ = len >> 8;
273 *enc->cur++ = len; 294 *enc->cur++ = len;
274 } 295 }
275} 296}
276 297
298// encodes a perl value into a CBOR integer
299ecb_inline void
300encode_int (enc_t *enc, SV *sv)
301{
302 if (SvIsUV (sv))
303 encode_uint (enc, MAJOR_POS_INT, SvUVX (sv));
304 else if (SvIVX (sv) >= 0)
305 encode_uint (enc, MAJOR_POS_INT, SvIVX (sv));
306 else
307 encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1));
308}
309
277ecb_inline void 310ecb_inline void
278encode_tag (enc_t *enc, UV tag) 311encode_tag (enc_t *enc, UV tag)
279{ 312{
280 encode_uint (enc, MAJOR_TAG, tag); 313 encode_uint (enc, MAJOR_TAG, tag);
281} 314}
… …
342 } 375 }
343 376
344 encode_str (enc, upgrade_utf8, utf8, str, len); 377 encode_str (enc, upgrade_utf8, utf8, str, len);
345} 378}
346 379
347static void encode_sv (enc_t *enc, SV *sv); 380ecb_inline void
381encode_float16 (enc_t *enc, NV nv)
382{
383 need (enc, 1+2);
384
385 *enc->cur++ = MAJOR_MISC | MISC_FLOAT16;
386
387 uint16_t fp = ecb_float_to_binary16 (nv);
388
389 if (!ecb_big_endian ())
390 fp = ecb_bswap16 (fp);
391
392 memcpy (enc->cur, &fp, 2);
393 enc->cur += 2;
394}
395
396ecb_inline void
397encode_float32 (enc_t *enc, NV nv)
398{
399 need (enc, 1+4);
400
401 *enc->cur++ = MAJOR_MISC | MISC_FLOAT32;
402
403 uint32_t fp = ecb_float_to_binary32 (nv);
404
405 if (!ecb_big_endian ())
406 fp = ecb_bswap32 (fp);
407
408 memcpy (enc->cur, &fp, 4);
409 enc->cur += 4;
410}
411
412ecb_inline void
413encode_float64 (enc_t *enc, NV nv)
414{
415 need (enc, 1+8);
416
417 *enc->cur++ = MAJOR_MISC | MISC_FLOAT64;
418
419 uint64_t fp = ecb_double_to_binary64 (nv);
420
421 if (!ecb_big_endian ())
422 fp = ecb_bswap64 (fp);
423
424 memcpy (enc->cur, &fp, 8);
425 enc->cur += 8;
426}
427
428ecb_inline void
429encode_bool (enc_t *enc, int istrue)
430{
431 encode_ch (enc, istrue ? MAJOR_MISC | SIMPLE_TRUE : MAJOR_MISC | SIMPLE_FALSE);
432}
433
434// encodes an arrayref containing key-value pairs as CBOR map
435ecb_inline void
436encode_array_as_map (enc_t *enc, SV *sv)
437{
438 if (enc->depth >= enc->cbor.max_depth)
439 croak (ERR_NESTING_EXCEEDED);
440
441 ++enc->depth;
442
443 // as_map does error checking for us, but we re-check in case
444 // things have changed.
445
446 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
447 croak ("CBOR::XS::as_map requires an array reference (did you change the array after calling as_map?)");
448
449 AV *av = (AV *)SvRV (sv);
450 int i, len = av_len (av);
451
452 if (!(len & 1))
453 croak ("CBOR::XS::as_map requires an even number of elements (did you change the array after calling as_map?)");
454
455 encode_uint (enc, MAJOR_MAP, (len + 1) >> 1);
456
457 for (i = 0; i <= len; ++i)
458 {
459 SV **svp = av_fetch (av, i, 0);
460 encode_sv (enc, svp ? *svp : &PL_sv_undef);
461 }
462
463 --enc->depth;
464}
465
466ecb_inline void
467encode_forced (enc_t *enc, UV type, SV *sv)
468{
469 switch (type)
470 {
471 case AS_CBOR:
472 {
473 STRLEN len;
474 char *str = SvPVbyte (sv, len);
475
476 need (enc, len);
477 memcpy (enc->cur, str, len);
478 enc->cur += len;
479 }
480 break;
481
482 case AS_BYTES:
483 {
484 STRLEN len;
485 char *str = SvPVbyte (sv, len);
486 encode_strref (enc, 0, 0, str, len);
487 }
488 break;
489
490 case AS_TEXT:
491 {
492 STRLEN len;
493 char *str = SvPVutf8 (sv, len);
494 encode_strref (enc, 1, 1, str, len);
495 }
496 break;
497
498 case AS_INT: encode_int (enc, sv); break;
499
500 case AS_FLOAT16: encode_float16 (enc, SvNV (sv)); break;
501 case AS_FLOAT32: encode_float32 (enc, SvNV (sv)); break;
502 case AS_FLOAT64: encode_float64 (enc, SvNV (sv)); break;
503
504 case AS_MAP: encode_array_as_map (enc, sv); break;
505
506 default:
507 croak ("encountered malformed CBOR::XS::Tagged object");
508 }
509}
348 510
349static void 511static void
350encode_av (enc_t *enc, AV *av) 512encode_av (enc_t *enc, AV *av)
351{ 513{
352 int i, len = av_len (av); 514 int i, len = av_len (av);
… …
356 518
357 ++enc->depth; 519 ++enc->depth;
358 520
359 encode_uint (enc, MAJOR_ARRAY, len + 1); 521 encode_uint (enc, MAJOR_ARRAY, len + 1);
360 522
361 if (SvMAGICAL (av)) 523 if (ecb_expect_false (SvMAGICAL (av)))
362 for (i = 0; i <= len; ++i) 524 for (i = 0; i <= len; ++i)
363 { 525 {
364 SV **svp = av_fetch (av, i, 0); 526 SV **svp = av_fetch (av, i, 0);
365 encode_sv (enc, svp ? *svp : &PL_sv_undef); 527 encode_sv (enc, svp ? *svp : &PL_sv_undef);
366 } 528 }
… …
385 ++enc->depth; 547 ++enc->depth;
386 548
387 int pairs = hv_iterinit (hv); 549 int pairs = hv_iterinit (hv);
388 int mg = SvMAGICAL (hv); 550 int mg = SvMAGICAL (hv);
389 551
390 if (mg) 552 if (ecb_expect_false (mg))
391 encode_ch (enc, MAJOR_MAP | MINOR_INDEF); 553 encode_ch (enc, MAJOR_MAP | MINOR_INDEF);
392 else 554 else
393 encode_uint (enc, MAJOR_MAP, pairs); 555 encode_uint (enc, MAJOR_MAP, pairs);
394 556
395 while ((he = hv_iternext (hv))) 557 while ((he = hv_iternext (hv)))
… …
400 encode_strref (enc, enc->cbor.flags & (F_TEXT_KEYS | F_TEXT_STRINGS), HeKUTF8 (he), HeKEY (he), HeKLEN (he)); 562 encode_strref (enc, enc->cbor.flags & (F_TEXT_KEYS | F_TEXT_STRINGS), HeKUTF8 (he), HeKEY (he), HeKLEN (he));
401 563
402 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he)); 564 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he));
403 } 565 }
404 566
405 if (mg) 567 if (ecb_expect_false (mg))
406 encode_ch (enc, MAJOR_MISC | MINOR_INDEF); 568 encode_ch (enc, MAJOR_MISC | MINOR_INDEF);
407 569
408 --enc->depth; 570 --enc->depth;
409} 571}
410 572
… …
430 592
431 HV *stash = SvSTASH (sv); 593 HV *stash = SvSTASH (sv);
432 594
433 if (stash == boolean_stash) 595 if (stash == boolean_stash)
434 { 596 {
435 encode_ch (enc, SvIV (sv) ? MAJOR_MISC | SIMPLE_TRUE : MAJOR_MISC | SIMPLE_FALSE); 597 encode_bool (enc, SvIV (sv));
436 return; 598 return;
437 } 599 }
438 else if (stash == error_stash) 600 else if (stash == error_stash)
439 { 601 {
440 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF); 602 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
… …
443 else if (stash == tagged_stash) 605 else if (stash == tagged_stash)
444 { 606 {
445 if (svt != SVt_PVAV) 607 if (svt != SVt_PVAV)
446 croak ("encountered CBOR::XS::Tagged object that isn't an array"); 608 croak ("encountered CBOR::XS::Tagged object that isn't an array");
447 609
610 switch (av_len ((AV *)sv))
611 {
612 case 2-1:
613 // actually a tagged value
448 encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1))); 614 encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1)));
449 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1)); 615 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1));
616 break;
617
618 case 3-1:
619 // a forced type [value, type, undef]
620 encode_forced (enc, SvUV (*av_fetch ((AV *)sv, 1, 1)), *av_fetch ((AV *)sv, 0, 1));
621 break;
622
623 default:
624 croak ("encountered malformed CBOR::XS::Tagged object");
625 }
450 626
451 return; 627 return;
452 } 628 }
453 } 629 }
454 630
455 if (ecb_expect_false (SvREFCNT (sv) > 1) 631 if (ecb_expect_false (SvREFCNT (sv) > 1)
456 && ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING)) 632 && ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING))
457 { 633 {
458 if (!enc->shareable) 634 if (ecb_expect_false (!enc->shareable))
459 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ()); 635 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
460 636
461 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1); 637 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
462 638
463 if (SvOK (*svp)) 639 if (SvOK (*svp))
… …
477 if (ecb_expect_false (SvOBJECT (sv))) 653 if (ecb_expect_false (SvOBJECT (sv)))
478 { 654 {
479 HV *stash = SvSTASH (sv); 655 HV *stash = SvSTASH (sv);
480 GV *method; 656 GV *method;
481 657
658 if (enc->cbor.flags & F_FORBID_OBJECTS)
659 croak ("encountered object '%s', but forbid_objects is enabled",
660 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
482 if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0))) 661 else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
483 { 662 {
484 dSP; 663 dSP;
485 664
486 ENTER; SAVETMPS; 665 ENTER; SAVETMPS;
487 PUSHMARK (SP); 666 PUSHMARK (SP);
… …
506 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0) 685 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)
507 { 686 {
508 dSP; 687 dSP;
509 688
510 ENTER; SAVETMPS; 689 ENTER; SAVETMPS;
511 SAVESTACK_POS ();
512 PUSHMARK (SP); 690 PUSHMARK (SP);
513 EXTEND (SP, 2); 691 EXTEND (SP, 2);
514 // we re-bless the reference to get overload and other niceties right 692 // we re-bless the reference to get overload and other niceties right
515 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); 693 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
516 PUSHs (sv_cbor); 694 PUSHs (sv_cbor);
… …
525 703
526 encode_tag (enc, CBOR_TAG_PERL_OBJECT); 704 encode_tag (enc, CBOR_TAG_PERL_OBJECT);
527 encode_uint (enc, MAJOR_ARRAY, count + 1); 705 encode_uint (enc, MAJOR_ARRAY, count + 1);
528 encode_strref (enc, 0, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash)); 706 encode_strref (enc, 0, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
529 707
530 while (count) 708 {
709 int i;
710
711 for (i = 0; i < count; ++i)
531 encode_sv (enc, SP[1 - count--]); 712 encode_sv (enc, SP[i + 1 - count]);
713
714 SP -= count;
715 }
532 716
533 PUTBACK; 717 PUTBACK;
534 718
535 FREETMPS; LEAVE; 719 FREETMPS; LEAVE;
536 } 720 }
… …
558 742
559 if (ecb_expect_false (nv == (NV)(U32)nv)) 743 if (ecb_expect_false (nv == (NV)(U32)nv))
560 encode_uint (enc, MAJOR_POS_INT, (U32)nv); 744 encode_uint (enc, MAJOR_POS_INT, (U32)nv);
561 //TODO: maybe I32? 745 //TODO: maybe I32?
562 else if (ecb_expect_false (nv == (float)nv)) 746 else if (ecb_expect_false (nv == (float)nv))
563 { 747 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 748 else
575 { 749 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} 750}
587 751
588static void 752static void
589encode_sv (enc_t *enc, SV *sv) 753encode_sv (enc_t *enc, SV *sv)
590{ 754{
… …
597 encode_strref (enc, enc->cbor.flags & F_TEXT_STRINGS, SvUTF8 (sv), str, len); 761 encode_strref (enc, enc->cbor.flags & F_TEXT_STRINGS, SvUTF8 (sv), str, len);
598 } 762 }
599 else if (SvNOKp (sv)) 763 else if (SvNOKp (sv))
600 encode_nv (enc, sv); 764 encode_nv (enc, sv);
601 else if (SvIOKp (sv)) 765 else if (SvIOKp (sv))
602 { 766 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)) 767 else if (SvROK (sv))
611 encode_rv (enc, SvRV (sv)); 768 encode_rv (enc, SvRV (sv));
612 else if (!SvOK (sv)) 769 else if (!SvOK (sv))
613 encode_ch (enc, MAJOR_MISC | SIMPLE_NULL); 770 encode_ch (enc, MAJOR_MISC | SIMPLE_NULL);
614 else if (enc->cbor.flags & F_ALLOW_UNKNOWN) 771 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
… …
621static SV * 778static SV *
622encode_cbor (SV *scalar, CBOR *cbor) 779encode_cbor (SV *scalar, CBOR *cbor)
623{ 780{
624 enc_t enc = { 0 }; 781 enc_t enc = { 0 };
625 782
626 enc.cbor = *cbor; 783 enc.cbor = *cbor;
627 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 784 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
628 enc.cur = SvPVX (enc.sv); 785 enc.cur = SvPVX (enc.sv);
629 enc.end = SvEND (enc.sv); 786 enc.end = SvEND (enc.sv);
630 787
631 SvPOK_only (enc.sv); 788 SvPOK_only (enc.sv);
632 789
633 if (cbor->flags & F_PACK_STRINGS) 790 if (cbor->flags & F_PACK_STRINGS)
634 { 791 {
… …
661 U32 depth; // recursion depth 818 U32 depth; // recursion depth
662 U32 maxdepth; // recursion depth limit 819 U32 maxdepth; // recursion depth limit
663 AV *shareable; 820 AV *shareable;
664 AV *stringref; 821 AV *stringref;
665 SV *decode_tagged; 822 SV *decode_tagged;
823 SV *err_sv; // optional sv for error, needs to be freed
666} dec_t; 824} dec_t;
667 825
668#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE 826// set dec->err to ERRSV
827ecb_cold static void
828err_errsv (dec_t *dec)
829{
830 if (!dec->err)
831 {
832 dec->err_sv = newSVsv (ERRSV);
669 833
834 // chop off the trailing \n
835 SvCUR_set (dec->err_sv, SvCUR (dec->err_sv) - 1);
836 *SvEND (dec->err_sv) = 0;
837
838 dec->err = SvPVutf8_nolen (dec->err_sv);
839 }
840}
841
842// the following functions are used to reduce code size and help the compiler to optimise
843ecb_cold static void
844err_set (dec_t *dec, const char *reason)
845{
846 if (!dec->err)
847 dec->err = reason;
848}
849
850ecb_cold static void
851err_unexpected_end (dec_t *dec)
852{
853 err_set (dec, "unexpected end of CBOR data");
854}
855
856#define ERR_DO(do) SB do; goto fail; SE
857#define ERR(reason) ERR_DO (err_set (dec, reason))
858#define ERR_ERRSV ERR_DO (err_errsv (dec))
859
670#define WANT(len) if (ecb_expect_false ((UV)(dec->end - dec->cur) < (UV)len)) ERR ("unexpected end of CBOR data") 860#define WANT(len) if (ecb_expect_false ((uintptr_t)(dec->end - dec->cur) < (STRLEN)len)) ERR_DO (err_unexpected_end (dec))
671 861
672#define DEC_INC_DEPTH if (ecb_expect_false (++dec->depth > dec->cbor.max_depth)) ERR (ERR_NESTING_EXCEEDED) 862#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 863#define DEC_DEC_DEPTH --dec->depth
674 864
675static UV 865static UV
… …
743 933
744 for (;;) 934 for (;;)
745 { 935 {
746 WANT (1); 936 WANT (1);
747 937
748 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF)) 938 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF) || dec->err)
749 { 939 {
750 ++dec->cur; 940 ++dec->cur;
751 break; 941 break;
752 } 942 }
753 943
… …
767 957
768 DEC_DEC_DEPTH; 958 DEC_DEC_DEPTH;
769 return newRV_noinc ((SV *)av); 959 return newRV_noinc ((SV *)av);
770 960
771fail: 961fail:
772 SvREFCNT_dec (av); 962 SvREFCNT_dec_NN (av);
773 DEC_DEC_DEPTH; 963 DEC_DEC_DEPTH;
774 return &PL_sv_undef; 964 return &PL_sv_undef;
775} 965}
776 966
777static void 967static void
… …
800 990
801 WANT (len); 991 WANT (len);
802 dec->cur += len; 992 dec->cur += len;
803 993
804 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8)) 994 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
805 if (!is_utf8_string (key, len)) 995 if (!cbor_is_utf8_string ((U8 *)key, len))
806 ERR ("corrupted CBOR data (invalid UTF-8 in map key)"); 996 ERR ("corrupted CBOR data (invalid UTF-8 in map key)");
807 997
808 hv_store (hv, key, -len, decode_sv (dec), 0); 998 hv_store (hv, key, -len, decode_sv (dec), 0);
809 999
810 return; 1000 return;
811 } 1001 }
812 1002
813 SV *k = decode_sv (dec); 1003 SV *k = decode_sv (dec);
814 SV *v = decode_sv (dec); 1004 SV *v = decode_sv (dec);
815 1005
1006 // we leak memory if uncaught exceptions are thrown by random magical
1007 // methods, and this is hopefully the only place where it can happen,
1008 // so if there is a chance of an exception, take the very slow path.
1009 // since catching exceptions is "undocumented/internal/forbidden" by
1010 // the new p5p powers, we need to call out to a perl function :/
1011 if (ecb_expect_false (SvAMAGIC (k)))
1012 {
1013 dSP;
1014
1015 ENTER; SAVETMPS;
1016 PUSHMARK (SP);
1017 EXTEND (SP, 3);
1018 PUSHs (sv_2mortal (newRV_inc ((SV *)hv)));
1019 PUSHs (sv_2mortal (k));
1020 PUSHs (sv_2mortal (v));
1021
1022 PUTBACK;
1023 call_pv ("CBOR::XS::_hv_store", G_VOID | G_DISCARD | G_EVAL);
1024 SPAGAIN;
1025
1026 FREETMPS; LEAVE;
1027
1028 if (SvTRUE (ERRSV))
1029 ERR_ERRSV;
1030
1031 return;
1032 }
1033
816 hv_store_ent (hv, k, v, 0); 1034 hv_store_ent (hv, k, v, 0);
817 SvREFCNT_dec (k); 1035 SvREFCNT_dec_NN (k);
818 1036
819fail: 1037fail:
820 ; 1038 ;
821} 1039}
822 1040
… …
833 1051
834 for (;;) 1052 for (;;)
835 { 1053 {
836 WANT (1); 1054 WANT (1);
837 1055
838 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF)) 1056 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF) || dec->err)
839 { 1057 {
840 ++dec->cur; 1058 ++dec->cur;
841 break; 1059 break;
842 } 1060 }
843 1061
… …
856 1074
857 DEC_DEC_DEPTH; 1075 DEC_DEC_DEPTH;
858 return newRV_noinc ((SV *)hv); 1076 return newRV_noinc ((SV *)hv);
859 1077
860fail: 1078fail:
861 SvREFCNT_dec (hv); 1079 SvREFCNT_dec_NN (hv);
862 DEC_DEC_DEPTH; 1080 DEC_DEC_DEPTH;
863 return &PL_sv_undef; 1081 return &PL_sv_undef;
864} 1082}
865 1083
866static SV * 1084static SV *
867decode_str (dec_t *dec, int utf8) 1085decode_str (dec_t *dec, int utf8)
868{ 1086{
869 SV *sv = 0; 1087 SV *sv = 0;
870 1088
871 if ((*dec->cur & MINOR_MASK) == MINOR_INDEF) 1089 if (ecb_expect_false ((*dec->cur & MINOR_MASK) == MINOR_INDEF))
872 { 1090 {
873 // indefinite length strings 1091 // indefinite length strings
874 ++dec->cur; 1092 ++dec->cur;
875 1093
876 U8 major = *dec->cur & MAJOR_MISC; 1094 U8 major = *dec->cur & MAJOR_MISC;
… …
911 } 1129 }
912 1130
913 if (utf8) 1131 if (utf8)
914 { 1132 {
915 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8)) 1133 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
916 if (!is_utf8_string (SvPVX (sv), SvCUR (sv))) 1134 if (!cbor_is_utf8_string (SvPVX (sv), SvCUR (sv)))
917 ERR ("corrupted CBOR data (invalid UTF-8 in text string)"); 1135 ERR ("corrupted CBOR data (invalid UTF-8 in text string)");
918 1136
919 SvUTF8_on (sv); 1137 SvUTF8_on (sv);
920 } 1138 }
921 1139
… …
944 sv = newRV_noinc (decode_sv (dec)); 1162 sv = newRV_noinc (decode_sv (dec));
945 break; 1163 break;
946 1164
947 case CBOR_TAG_STRINGREF_NAMESPACE: 1165 case CBOR_TAG_STRINGREF_NAMESPACE:
948 { 1166 {
949 // do nmot use SAVETMPS/FREETMPS, as these will 1167 // do not use SAVETMPS/FREETMPS, as these will
950 // erase mortalised caches, e.g. "shareable" 1168 // erase mortalised caches, e.g. "shareable"
951 ENTER; 1169 ENTER;
952 1170
953 SAVESPTR (dec->stringref); 1171 SAVESPTR (dec->stringref);
954 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ()); 1172 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ());
… …
964 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT)) 1182 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
965 ERR ("corrupted CBOR data (stringref index not an unsigned integer)"); 1183 ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
966 1184
967 UV idx = decode_uint (dec); 1185 UV idx = decode_uint (dec);
968 1186
969 if (!dec->stringref || (int)idx > AvFILLp (dec->stringref)) 1187 if (!dec->stringref || idx >= (UV)(1 + AvFILLp (dec->stringref)))
970 ERR ("corrupted CBOR data (stringref index out of bounds or outside namespace)"); 1188 ERR ("corrupted CBOR data (stringref index out of bounds or outside namespace)");
971 1189
972 sv = newSVsv (AvARRAY (dec->stringref)[idx]); 1190 sv = newSVsv (AvARRAY (dec->stringref)[idx]);
973 } 1191 }
974 break; 1192 break;
… …
1002 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT)) 1220 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
1003 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)"); 1221 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
1004 1222
1005 UV idx = decode_uint (dec); 1223 UV idx = decode_uint (dec);
1006 1224
1007 if (!dec->shareable || (int)idx > AvFILLp (dec->shareable)) 1225 if (!dec->shareable || idx >= (UV)(1 + AvFILLp (dec->shareable)))
1008 ERR ("corrupted CBOR data (sharedref index out of bounds)"); 1226 ERR ("corrupted CBOR data (sharedref index out of bounds)");
1009 1227
1010 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]); 1228 sv = newRV_inc (AvARRAY (dec->shareable)[idx]);
1011 1229
1012 if (sv == &PL_sv_undef) 1230 if (sv == &PL_sv_undef)
1013 ERR ("cyclic CBOR data structure found, but allow_cycles is not enabled"); 1231 ERR ("cyclic CBOR data structure found, but allow_cycles is not enabled");
1014 } 1232 }
1015 break; 1233 break;
1016 1234
1017 case CBOR_TAG_PERL_OBJECT: 1235 case CBOR_TAG_PERL_OBJECT:
1018 { 1236 {
1237 if (dec->cbor.flags & F_FORBID_OBJECTS)
1238 goto filter;
1239
1019 sv = decode_sv (dec); 1240 sv = decode_sv (dec);
1020 1241
1021 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV) 1242 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
1022 ERR ("corrupted CBOR data (non-array perl object)"); 1243 ERR ("corrupted CBOR data (non-array perl object)");
1023 1244
… …
1052 SPAGAIN; 1273 SPAGAIN;
1053 1274
1054 if (SvTRUE (ERRSV)) 1275 if (SvTRUE (ERRSV))
1055 { 1276 {
1056 FREETMPS; LEAVE; 1277 FREETMPS; LEAVE;
1057 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV)))); 1278 ERR_ERRSV;
1058 } 1279 }
1059 1280
1060 SvREFCNT_dec (sv); 1281 SvREFCNT_dec_NN (sv);
1061 sv = SvREFCNT_inc (POPs); 1282 sv = SvREFCNT_inc (POPs);
1062 1283
1063 PUTBACK; 1284 PUTBACK;
1064 1285
1065 FREETMPS; LEAVE; 1286 FREETMPS; LEAVE;
1066 } 1287 }
1067 break; 1288 break;
1068 1289
1069 default: 1290 default:
1291 filter:
1070 { 1292 {
1071 SV *tag_sv = newSVuv (tag); 1293 SV *tag_sv = newSVuv (tag);
1072 1294
1073 sv = decode_sv (dec); 1295 sv = decode_sv (dec);
1074 1296
1075 dSP; 1297 dSP;
1076 ENTER; SAVETMPS; 1298 ENTER; SAVETMPS;
1077 SAVESTACK_POS ();
1078 PUSHMARK (SP); 1299 PUSHMARK (SP);
1079 EXTEND (SP, 2); 1300 EXTEND (SP, 2);
1080 PUSHs (tag_sv); 1301 PUSHs (tag_sv);
1081 PUSHs (sv); 1302 PUSHs (sv);
1082 1303
… …
1084 int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL); 1305 int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL);
1085 SPAGAIN; 1306 SPAGAIN;
1086 1307
1087 if (SvTRUE (ERRSV)) 1308 if (SvTRUE (ERRSV))
1088 { 1309 {
1089 SvREFCNT_dec (tag_sv); 1310 SvREFCNT_dec_NN (tag_sv);
1090 FREETMPS; LEAVE; 1311 FREETMPS; LEAVE;
1091 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV)))); 1312 ERR_ERRSV;
1092 } 1313 }
1093 1314
1094 if (count) 1315 if (count)
1095 { 1316 {
1096 SvREFCNT_dec (tag_sv); 1317 SvREFCNT_dec_NN (tag_sv);
1097 SvREFCNT_dec (sv); 1318 SvREFCNT_dec_NN (sv);
1098 sv = SvREFCNT_inc (POPs); 1319 sv = SvREFCNT_inc_NN (TOPs);
1320 SP -= count;
1099 } 1321 }
1100 else 1322 else
1101 { 1323 {
1102 AV *av = newAV (); 1324 AV *av = newAV ();
1103 av_push (av, tag_sv); 1325 av_push (av, tag_sv);
… …
1246 for (i = av_len (dec.shareable) + 1; i--; ) 1468 for (i = av_len (dec.shareable) + 1; i--; )
1247 if ((svp = av_fetch (dec.shareable, i, 0))) 1469 if ((svp = av_fetch (dec.shareable, i, 0)))
1248 sv_setsv (*svp, &PL_sv_undef); 1470 sv_setsv (*svp, &PL_sv_undef);
1249 } 1471 }
1250 1472
1251 SvREFCNT_dec (sv); 1473 SvREFCNT_dec_NN (sv);
1474
1475 if (dec.err_sv)
1476 sv_2mortal (dec.err_sv);
1477
1252 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur); 1478 croak ("%s, at offset %ld (octet 0x%02x)", dec.err, (long)(dec.cur - (U8 *)data), (int)(uint8_t)*dec.cur);
1253 } 1479 }
1254 1480
1255 sv = sv_2mortal (sv); 1481 sv = sv_2mortal (sv);
1256 1482
1257 return sv; 1483 return sv;
… …
1346 1572
1347 break; 1573 break;
1348 1574
1349 case MAJOR_MAP >> MAJOR_SHIFT: 1575 case MAJOR_MAP >> MAJOR_SHIFT:
1350 len <<= 1; 1576 len <<= 1;
1577 /* FALLTHROUGH */
1351 case MAJOR_ARRAY >> MAJOR_SHIFT: 1578 case MAJOR_ARRAY >> MAJOR_SHIFT:
1352 if (len) 1579 if (len)
1353 { 1580 {
1354 av_push (self->incr_count, newSViv (len + 1)); //TODO: nest 1581 av_push (self->incr_count, newSViv (len + 1)); //TODO: nest
1355 count = len + 1; 1582 count = len + 1;
… …
1431 ALIAS: 1658 ALIAS:
1432 shrink = F_SHRINK 1659 shrink = F_SHRINK
1433 allow_unknown = F_ALLOW_UNKNOWN 1660 allow_unknown = F_ALLOW_UNKNOWN
1434 allow_sharing = F_ALLOW_SHARING 1661 allow_sharing = F_ALLOW_SHARING
1435 allow_cycles = F_ALLOW_CYCLES 1662 allow_cycles = F_ALLOW_CYCLES
1663 forbid_objects = F_FORBID_OBJECTS
1436 pack_strings = F_PACK_STRINGS 1664 pack_strings = F_PACK_STRINGS
1437 text_keys = F_TEXT_KEYS 1665 text_keys = F_TEXT_KEYS
1438 text_strings = F_TEXT_STRINGS 1666 text_strings = F_TEXT_STRINGS
1439 validate_utf8 = F_VALIDATE_UTF8 1667 validate_utf8 = F_VALIDATE_UTF8
1440 PPCODE: 1668 PPCODE:
… …
1451 ALIAS: 1679 ALIAS:
1452 get_shrink = F_SHRINK 1680 get_shrink = F_SHRINK
1453 get_allow_unknown = F_ALLOW_UNKNOWN 1681 get_allow_unknown = F_ALLOW_UNKNOWN
1454 get_allow_sharing = F_ALLOW_SHARING 1682 get_allow_sharing = F_ALLOW_SHARING
1455 get_allow_cycles = F_ALLOW_CYCLES 1683 get_allow_cycles = F_ALLOW_CYCLES
1684 get_forbid_objects = F_FORBID_OBJECTS
1456 get_pack_strings = F_PACK_STRINGS 1685 get_pack_strings = F_PACK_STRINGS
1457 get_text_keys = F_TEXT_KEYS 1686 get_text_keys = F_TEXT_KEYS
1458 get_text_strings = F_TEXT_STRINGS 1687 get_text_strings = F_TEXT_STRINGS
1459 get_validate_utf8 = F_VALIDATE_UTF8 1688 get_validate_utf8 = F_VALIDATE_UTF8
1460 PPCODE: 1689 PPCODE:
… …
1593 cbor_init (&cbor); 1822 cbor_init (&cbor);
1594 PUTBACK; cborstr = decode_cbor (cborstr, &cbor, 0); SPAGAIN; 1823 PUTBACK; cborstr = decode_cbor (cborstr, &cbor, 0); SPAGAIN;
1595 XPUSHs (cborstr); 1824 XPUSHs (cborstr);
1596} 1825}
1597 1826
1827#ifdef __AFL_COMPILER
1828
1829void
1830afl_init ()
1831 CODE:
1832 __AFL_INIT ();
1833
1834int
1835afl_loop (unsigned int count = 10000)
1836 CODE:
1837 RETVAL = __AFL_LOOP (count);
1838 OUTPUT:
1839 RETVAL
1840
1841#endif
1842

Diff Legend

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