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 1.36 by root, Sat Nov 30 17:37:45 2013 UTC vs.
Revision 1.60 by root, Sat Nov 26 00:47:02 2016 UTC

… …
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 11
12#define ECB_NO_THREADS 1
12#include "ecb.h" 13#include "ecb.h"
13 14
14// compatibility with perl <5.18 15// compatibility with perl <5.18
15#ifndef HvNAMELEN_get 16#ifndef HvNAMELEN_get
16# define HvNAMELEN_get(hv) strlen (HvNAME (hv)) 17# define HvNAMELEN_get(hv) strlen (HvNAME (hv))
… …
97}; 98};
98 99
99#define F_SHRINK 0x00000001UL 100#define F_SHRINK 0x00000001UL
100#define F_ALLOW_UNKNOWN 0x00000002UL 101#define F_ALLOW_UNKNOWN 0x00000002UL
101#define F_ALLOW_SHARING 0x00000004UL 102#define F_ALLOW_SHARING 0x00000004UL
103#define F_ALLOW_CYCLES 0x00000008UL
104#define F_FORBID_OBJECTS 0x00000010UL
102#define F_PACK_STRINGS 0x00000008UL 105#define F_PACK_STRINGS 0x00000020UL
106#define F_TEXT_KEYS 0x00000040UL
107#define F_TEXT_STRINGS 0x00000080UL
108#define F_VALIDATE_UTF8 0x00000100UL
103 109
104#define INIT_SIZE 32 // initial scalar size to be allocated 110#define INIT_SIZE 32 // initial scalar size to be allocated
105 111
106#define SB do { 112#define SB do {
107#define SE } while (0) 113#define SE } while (0)
… …
126typedef struct { 132typedef struct {
127 U32 flags; 133 U32 flags;
128 U32 max_depth; 134 U32 max_depth;
129 STRLEN max_size; 135 STRLEN max_size;
130 SV *filter; 136 SV *filter;
137
138 // for the incremental parser
139 STRLEN incr_pos; // the current offset into the text
140 STRLEN incr_need; // minimum bytes needed to decode
141 AV *incr_count; // for every nesting level, the number of outstanding values, or -1 for indef.
131} CBOR; 142} CBOR;
132 143
133ecb_inline void 144ecb_inline void
134cbor_init (CBOR *cbor) 145cbor_init (CBOR *cbor)
135{ 146{
… …
139 150
140ecb_inline void 151ecb_inline void
141cbor_free (CBOR *cbor) 152cbor_free (CBOR *cbor)
142{ 153{
143 SvREFCNT_dec (cbor->filter); 154 SvREFCNT_dec (cbor->filter);
155 SvREFCNT_dec (cbor->incr_count);
144} 156}
145 157
146///////////////////////////////////////////////////////////////////////////// 158/////////////////////////////////////////////////////////////////////////////
147// utility functions 159// utility functions
148 160
… …
267encode_tag (enc_t *enc, UV tag) 279encode_tag (enc_t *enc, UV tag)
268{ 280{
269 encode_uint (enc, MAJOR_TAG, tag); 281 encode_uint (enc, MAJOR_TAG, tag);
270} 282}
271 283
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
272ecb_inline void 307ecb_inline void
273encode_str (enc_t *enc, int utf8, char *str, STRLEN len) 308encode_str (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len)
274{ 309{
310 if (ecb_expect_false (upgrade_utf8))
311 if (!utf8)
312 {
313 encode_str_utf8 (enc, utf8, str, len);
314 return;
315 }
316
275 encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len); 317 encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len);
276 need (enc, len); 318 need (enc, len);
277 memcpy (enc->cur, str, len); 319 memcpy (enc->cur, str, len);
278 enc->cur += len; 320 enc->cur += len;
279} 321}
280 322
281static void 323ecb_inline void
282encode_strref (enc_t *enc, int utf8, char *str, STRLEN len) 324encode_strref (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len)
283{ 325{
284 if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS)) 326 if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS))
285 { 327 {
286 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1); 328 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1);
287 329
… …
298 sv_setuv (*svp, enc->stringref_idx); 340 sv_setuv (*svp, enc->stringref_idx);
299 ++enc->stringref_idx; 341 ++enc->stringref_idx;
300 } 342 }
301 } 343 }
302 344
303 encode_str (enc, utf8, str, len); 345 encode_str (enc, upgrade_utf8, utf8, str, len);
304} 346}
305 347
306static void encode_sv (enc_t *enc, SV *sv); 348static void encode_sv (enc_t *enc, SV *sv);
307 349
308static void 350static void
… …
315 357
316 ++enc->depth; 358 ++enc->depth;
317 359
318 encode_uint (enc, MAJOR_ARRAY, len + 1); 360 encode_uint (enc, MAJOR_ARRAY, len + 1);
319 361
362 if (SvMAGICAL (av))
320 for (i = 0; i <= len; ++i) 363 for (i = 0; i <= len; ++i)
321 { 364 {
322 SV **svp = av_fetch (av, i, 0); 365 SV **svp = av_fetch (av, i, 0);
323 encode_sv (enc, svp ? *svp : &PL_sv_undef); 366 encode_sv (enc, svp ? *svp : &PL_sv_undef);
324 } 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 }
325 374
326 --enc->depth; 375 --enc->depth;
327} 376}
328 377
329static void 378static void
… …
347 while ((he = hv_iternext (hv))) 396 while ((he = hv_iternext (hv)))
348 { 397 {
349 if (HeKLEN (he) == HEf_SVKEY) 398 if (HeKLEN (he) == HEf_SVKEY)
350 encode_sv (enc, HeSVKEY (he)); 399 encode_sv (enc, HeSVKEY (he));
351 else 400 else
352 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));
353 402
354 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));
355 } 404 }
356 405
357 if (mg) 406 if (mg)
… …
429 if (ecb_expect_false (SvOBJECT (sv))) 478 if (ecb_expect_false (SvOBJECT (sv)))
430 { 479 {
431 HV *stash = SvSTASH (sv); 480 HV *stash = SvSTASH (sv);
432 GV *method; 481 GV *method;
433 482
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))));
434 if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0))) 486 else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
435 { 487 {
436 dSP; 488 dSP;
437 489
438 ENTER; SAVETMPS; PUSHMARK (SP); 490 ENTER; SAVETMPS;
491 PUSHMARK (SP);
439 // 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
440 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); 493 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
441 494
442 PUTBACK; 495 PUTBACK;
443 // G_SCALAR ensures that return value is 1 496 // G_SCALAR ensures that return value is 1
… …
456 } 509 }
457 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0) 510 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)
458 { 511 {
459 dSP; 512 dSP;
460 513
461 ENTER; SAVETMPS; PUSHMARK (SP); 514 ENTER; SAVETMPS;
515 SAVESTACK_POS ();
516 PUSHMARK (SP);
462 EXTEND (SP, 2); 517 EXTEND (SP, 2);
463 // 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
464 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); 519 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
465 PUSHs (sv_cbor); 520 PUSHs (sv_cbor);
466 521
… …
472 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv) 527 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)
473 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));
474 529
475 encode_tag (enc, CBOR_TAG_PERL_OBJECT); 530 encode_tag (enc, CBOR_TAG_PERL_OBJECT);
476 encode_uint (enc, MAJOR_ARRAY, count + 1); 531 encode_uint (enc, MAJOR_ARRAY, count + 1);
477 encode_strref (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash)); 532 encode_strref (enc, 0, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
478 533
479 while (count) 534 while (count)
480 encode_sv (enc, SP[1 - count--]); 535 encode_sv (enc, SP[1 - count--]);
481 536
482 PUTBACK; 537 PUTBACK;
… …
541 596
542 if (SvPOKp (sv)) 597 if (SvPOKp (sv))
543 { 598 {
544 STRLEN len; 599 STRLEN len;
545 char *str = SvPV (sv, len); 600 char *str = SvPV (sv, len);
546 encode_strref (enc, SvUTF8 (sv), str, len); 601 encode_strref (enc, enc->cbor.flags & F_TEXT_STRINGS, SvUTF8 (sv), str, len);
547 } 602 }
548 else if (SvNOKp (sv)) 603 else if (SvNOKp (sv))
549 encode_nv (enc, sv); 604 encode_nv (enc, sv);
550 else if (SvIOKp (sv)) 605 else if (SvIOKp (sv))
551 { 606 {
… …
568} 623}
569 624
570static SV * 625static SV *
571encode_cbor (SV *scalar, CBOR *cbor) 626encode_cbor (SV *scalar, CBOR *cbor)
572{ 627{
573 enc_t enc = { }; 628 enc_t enc = { 0 };
574 629
575 enc.cbor = *cbor; 630 enc.cbor = *cbor;
576 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 631 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
577 enc.cur = SvPVX (enc.sv); 632 enc.cur = SvPVX (enc.sv);
578 enc.end = SvEND (enc.sv); 633 enc.end = SvEND (enc.sv);
… …
610 U32 depth; // recursion depth 665 U32 depth; // recursion depth
611 U32 maxdepth; // recursion depth limit 666 U32 maxdepth; // recursion depth limit
612 AV *shareable; 667 AV *shareable;
613 AV *stringref; 668 AV *stringref;
614 SV *decode_tagged; 669 SV *decode_tagged;
670 SV *err_sv; // optional sv for error, needs to be freed
615} dec_t; 671} dec_t;
616 672
617#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);
618 680
619#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;
620 684
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
703ecb_cold static void
704err_nesting_exceeded (dec_t *dec)
705{
706 err_set (dec, ERR_NESTING_EXCEEDED);
707}
708
709#define ERR_DO(do) SB do; goto fail; SE
710#define ERR(reason) ERR_DO (err_set (dec, reason))
711#define ERR_ERRSV ERR_DO (err_errsv (dec))
712
713#define WANT(len) if (ecb_expect_false ((UV)(dec->end - dec->cur) < (UV)len)) ERR_DO (err_unexpected_end (dec))
714
621#define DEC_INC_DEPTH if (++dec->depth > dec->cbor.max_depth) ERR (ERR_NESTING_EXCEEDED) 715#define DEC_INC_DEPTH if (ecb_expect_false (++dec->depth > dec->cbor.max_depth)) ERR (ERR_NESTING_EXCEEDED)
622#define DEC_DEC_DEPTH --dec->depth 716#define DEC_DEC_DEPTH --dec->depth
623 717
624static UV 718static UV
625decode_uint (dec_t *dec) 719decode_uint (dec_t *dec)
626{ 720{
… …
703 av_push (av, decode_sv (dec)); 797 av_push (av, decode_sv (dec));
704 } 798 }
705 } 799 }
706 else 800 else
707 { 801 {
708 int i, len = decode_uint (dec); 802 UV i, len = decode_uint (dec);
709 803
710 WANT (len); // complexity check for av_fill - need at least one byte per value, do not allow supersize arrays 804 WANT (len); // complexity check for av_fill - need at least one byte per value, do not allow supersize arrays
711 av_fill (av, len - 1); 805 av_fill (av, len - 1);
712 806
713 for (i = 0; i < len; ++i) 807 for (i = 0; i < len; ++i)
… …
728{ 822{
729 // for speed reasons, we specialcase single-string 823 // for speed reasons, we specialcase single-string
730 // byte or utf-8 strings as keys, but only when !stringref 824 // byte or utf-8 strings as keys, but only when !stringref
731 825
732 if (ecb_expect_true (!dec->stringref)) 826 if (ecb_expect_true (!dec->stringref))
733 if (ecb_expect_true ((*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8)) 827 if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8))
734 { 828 {
735 I32 len = decode_uint (dec); 829 STRLEN len = decode_uint (dec);
736 char *key = (char *)dec->cur; 830 char *key = (char *)dec->cur;
737 831
832 WANT (len);
738 dec->cur += len; 833 dec->cur += len;
739 834
740 if (ecb_expect_false (dec->stringref))
741 av_push (dec->stringref, newSVpvn (key, len));
742
743 hv_store (hv, key, len, decode_sv (dec), 0); 835 hv_store (hv, key, len, decode_sv (dec), 0);
744 836
745 return; 837 return;
746 } 838 }
747 else if (ecb_expect_true ((*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8)) 839 else if (ecb_expect_true ((U8)(*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8))
748 { 840 {
749 I32 len = decode_uint (dec); 841 STRLEN len = decode_uint (dec);
750 char *key = (char *)dec->cur; 842 char *key = (char *)dec->cur;
751 843
844 WANT (len);
752 dec->cur += len; 845 dec->cur += len;
753 846
754 if (ecb_expect_false (dec->stringref)) 847 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
755 av_push (dec->stringref, newSVpvn_utf8 (key, len, 1)); 848 if (!is_utf8_string (key, len))
849 ERR ("corrupted CBOR data (invalid UTF-8 in map key)");
756 850
757 hv_store (hv, key, -len, decode_sv (dec), 0); 851 hv_store (hv, key, -len, decode_sv (dec), 0);
758 852
759 return; 853 return;
760 } 854 }
761 855
762 SV *k = decode_sv (dec); 856 SV *k = decode_sv (dec);
763 SV *v = decode_sv (dec); 857 SV *v = decode_sv (dec);
764 858
859 // we leak memory if uncaught exceptions are thrown by random magical
860 // methods, and this is hopefully the only place where it can happen,
861 // so if there is a chance of an exception, take the very slow path.
862 // since catching exceptions is "undocumented/internal/forbidden" by
863 // the new p5p powers, we need to call out to a perl function :/
864 if (ecb_expect_false (SvAMAGIC (k)))
865 {
866 dSP;
867
868 ENTER; SAVETMPS;
869 PUSHMARK (SP);
870 EXTEND (SP, 3);
871 PUSHs (sv_2mortal (newRV_inc ((SV *)hv)));
872 PUSHs (sv_2mortal (k));
873 PUSHs (sv_2mortal (v));
874
875 PUTBACK;
876 call_pv ("CBOR::XS::_hv_store", G_VOID | G_DISCARD | G_EVAL);
877 SPAGAIN;
878
879 FREETMPS; LEAVE;
880
881 if (SvTRUE (ERRSV))
882 ERR_ERRSV;
883
884 return;
885 }
886
765 hv_store_ent (hv, k, v, 0); 887 hv_store_ent (hv, k, v, 0);
766 SvREFCNT_dec (k); 888 SvREFCNT_dec (k);
889
890fail:
891 ;
767} 892}
768 893
769static SV * 894static SV *
770decode_hv (dec_t *dec) 895decode_hv (dec_t *dec)
771{ 896{
… …
790 decode_he (dec, hv); 915 decode_he (dec, hv);
791 } 916 }
792 } 917 }
793 else 918 else
794 { 919 {
795 int pairs = decode_uint (dec); 920 UV pairs = decode_uint (dec);
921
922 WANT (pairs); // complexity check - need at least one byte per value, do not allow supersize hashes
796 923
797 while (pairs--) 924 while (pairs--)
798 decode_he (dec, hv); 925 decode_he (dec, hv);
799 } 926 }
800 927
… …
853 && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1)) 980 && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1))
854 av_push (dec->stringref, SvREFCNT_inc_NN (sv)); 981 av_push (dec->stringref, SvREFCNT_inc_NN (sv));
855 } 982 }
856 983
857 if (utf8) 984 if (utf8)
985 {
986 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
987 if (!is_utf8_string (SvPVX (sv), SvCUR (sv)))
988 ERR ("corrupted CBOR data (invalid UTF-8 in text string)");
989
858 SvUTF8_on (sv); 990 SvUTF8_on (sv);
991 }
859 992
860 return sv; 993 return sv;
861 994
862fail: 995fail:
863 SvREFCNT_dec (sv); 996 SvREFCNT_dec (sv);
… …
882 sv = newRV_noinc (decode_sv (dec)); 1015 sv = newRV_noinc (decode_sv (dec));
883 break; 1016 break;
884 1017
885 case CBOR_TAG_STRINGREF_NAMESPACE: 1018 case CBOR_TAG_STRINGREF_NAMESPACE:
886 { 1019 {
1020 // do nmot use SAVETMPS/FREETMPS, as these will
1021 // erase mortalised caches, e.g. "shareable"
887 ENTER; SAVETMPS; 1022 ENTER;
888 1023
889 SAVESPTR (dec->stringref); 1024 SAVESPTR (dec->stringref);
890 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ()); 1025 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ());
891 1026
892 sv = decode_sv (dec); 1027 sv = decode_sv (dec);
893 1028
894 FREETMPS; LEAVE; 1029 LEAVE;
895 } 1030 }
896 break; 1031 break;
897 1032
898 case CBOR_TAG_STRINGREF: 1033 case CBOR_TAG_STRINGREF:
899 { 1034 {
… …
912 case CBOR_TAG_VALUE_SHAREABLE: 1047 case CBOR_TAG_VALUE_SHAREABLE:
913 { 1048 {
914 if (ecb_expect_false (!dec->shareable)) 1049 if (ecb_expect_false (!dec->shareable))
915 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ()); 1050 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ());
916 1051
1052 if (dec->cbor.flags & F_ALLOW_CYCLES)
1053 {
917 sv = newSV (0); 1054 sv = newSV (0);
918 av_push (dec->shareable, SvREFCNT_inc_NN (sv)); 1055 av_push (dec->shareable, SvREFCNT_inc_NN (sv));
919 1056
920 SV *osv = decode_sv (dec); 1057 SV *osv = decode_sv (dec);
921 sv_setsv (sv, osv); 1058 sv_setsv (sv, osv);
922 SvREFCNT_dec_NN (osv); 1059 SvREFCNT_dec_NN (osv);
1060 }
1061 else
1062 {
1063 av_push (dec->shareable, &PL_sv_undef);
1064 int idx = AvFILLp (dec->shareable);
1065 sv = decode_sv (dec);
1066 av_store (dec->shareable, idx, SvREFCNT_inc_NN (sv));
1067 }
923 } 1068 }
924 break; 1069 break;
925 1070
926 case CBOR_TAG_VALUE_SHAREDREF: 1071 case CBOR_TAG_VALUE_SHAREDREF:
927 { 1072 {
… …
932 1077
933 if (!dec->shareable || (int)idx > AvFILLp (dec->shareable)) 1078 if (!dec->shareable || (int)idx > AvFILLp (dec->shareable))
934 ERR ("corrupted CBOR data (sharedref index out of bounds)"); 1079 ERR ("corrupted CBOR data (sharedref index out of bounds)");
935 1080
936 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]); 1081 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]);
1082
1083 if (sv == &PL_sv_undef)
1084 ERR ("cyclic CBOR data structure found, but allow_cycles is not enabled");
937 } 1085 }
938 break; 1086 break;
939 1087
940 case CBOR_TAG_PERL_OBJECT: 1088 case CBOR_TAG_PERL_OBJECT:
941 { 1089 {
1090 if (dec->cbor.flags & F_FORBID_OBJECTS)
1091 goto filter;
1092
942 sv = decode_sv (dec); 1093 sv = decode_sv (dec);
943 1094
944 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV) 1095 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
945 ERR ("corrupted CBOR data (non-array perl object)"); 1096 ERR ("corrupted CBOR data (non-array perl object)");
946 1097
… …
956 if (!method) 1107 if (!method)
957 ERR ("cannot decode perl-object (package does not have a THAW method)"); 1108 ERR ("cannot decode perl-object (package does not have a THAW method)");
958 1109
959 dSP; 1110 dSP;
960 1111
961 ENTER; SAVETMPS; PUSHMARK (SP); 1112 ENTER; SAVETMPS;
1113 PUSHMARK (SP);
962 EXTEND (SP, len + 1); 1114 EXTEND (SP, len + 1);
963 // we re-bless the reference to get overload and other niceties right 1115 // we re-bless the reference to get overload and other niceties right
964 PUSHs (*av_fetch (av, 0, 1)); 1116 PUSHs (*av_fetch (av, 0, 1));
965 PUSHs (sv_cbor); 1117 PUSHs (sv_cbor);
966 1118
… …
974 SPAGAIN; 1126 SPAGAIN;
975 1127
976 if (SvTRUE (ERRSV)) 1128 if (SvTRUE (ERRSV))
977 { 1129 {
978 FREETMPS; LEAVE; 1130 FREETMPS; LEAVE;
979 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV)))); 1131 ERR_ERRSV;
980 } 1132 }
981 1133
982 SvREFCNT_dec (sv); 1134 SvREFCNT_dec (sv);
983 sv = SvREFCNT_inc (POPs); 1135 sv = SvREFCNT_inc (POPs);
984 1136
… …
987 FREETMPS; LEAVE; 1139 FREETMPS; LEAVE;
988 } 1140 }
989 break; 1141 break;
990 1142
991 default: 1143 default:
1144 filter:
992 { 1145 {
1146 SV *tag_sv = newSVuv (tag);
1147
993 sv = decode_sv (dec); 1148 sv = decode_sv (dec);
994 1149
995 dSP; 1150 dSP;
996 ENTER; SAVETMPS; PUSHMARK (SP); 1151 ENTER; SAVETMPS;
1152 SAVESTACK_POS ();
1153 PUSHMARK (SP);
997 EXTEND (SP, 2); 1154 EXTEND (SP, 2);
998 PUSHs (newSVuv (tag)); 1155 PUSHs (tag_sv);
999 PUSHs (sv); 1156 PUSHs (sv);
1000 1157
1001 PUTBACK; 1158 PUTBACK;
1002 int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL); 1159 int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL);
1003 SPAGAIN; 1160 SPAGAIN;
1004 1161
1005 if (SvTRUE (ERRSV)) 1162 if (SvTRUE (ERRSV))
1006 { 1163 {
1164 SvREFCNT_dec (tag_sv);
1007 FREETMPS; LEAVE; 1165 FREETMPS; LEAVE;
1008 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV)))); 1166 ERR_ERRSV;
1009 } 1167 }
1010 1168
1011 if (count) 1169 if (count)
1012 { 1170 {
1171 SvREFCNT_dec (tag_sv);
1013 SvREFCNT_dec (sv); 1172 SvREFCNT_dec (sv);
1014 sv = SvREFCNT_inc (POPs); 1173 sv = SvREFCNT_inc (POPs);
1015 } 1174 }
1016 else 1175 else
1017 { 1176 {
1018 AV *av = newAV (); 1177 AV *av = newAV ();
1019 av_push (av, newSVuv (tag)); 1178 av_push (av, tag_sv);
1020 av_push (av, sv); 1179 av_push (av, sv);
1021 1180
1022 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 1181 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
1023 ? cbor_tagged_stash 1182 ? cbor_tagged_stash
1024 : gv_stashpv ("CBOR::XS::Tagged" , 1); 1183 : gv_stashpv ("CBOR::XS::Tagged" , 1);
… …
1110 1269
1111 return newSVnv (ecb_binary64_to_double (fp)); 1270 return newSVnv (ecb_binary64_to_double (fp));
1112 } 1271 }
1113 1272
1114 // 0..19 unassigned simple 1273 // 0..19 unassigned simple
1115 // 24 reserved + unassigned (reserved values are not encodable) 1274 // 24 reserved + unassigned simple (reserved values are not encodable)
1275 // 28-30 unassigned misc
1276 // 31 break code
1116 default: 1277 default:
1117 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)"); 1278 ERR ("corrupted CBOR data (reserved/unassigned/unexpected major 7 value)");
1118 } 1279 }
1119 1280
1120 break; 1281 break;
1121 } 1282 }
1122 1283
… …
1125} 1286}
1126 1287
1127static SV * 1288static SV *
1128decode_cbor (SV *string, CBOR *cbor, char **offset_return) 1289decode_cbor (SV *string, CBOR *cbor, char **offset_return)
1129{ 1290{
1130 dec_t dec = { }; 1291 dec_t dec = { 0 };
1131 SV *sv; 1292 SV *sv;
1132 STRLEN len; 1293 STRLEN len;
1133 char *data = SvPVbyte (string, len); 1294 char *data = SvPVbyte (string, len);
1134 1295
1135 if (len > cbor->max_size && cbor->max_size) 1296 if (len > cbor->max_size && cbor->max_size)
… …
1149 if (dec.cur != dec.end && !dec.err) 1310 if (dec.cur != dec.end && !dec.err)
1150 dec.err = "garbage after CBOR object"; 1311 dec.err = "garbage after CBOR object";
1151 1312
1152 if (dec.err) 1313 if (dec.err)
1153 { 1314 {
1315 if (dec.shareable)
1316 {
1317 // need to break cyclic links, which would all be in shareable
1318 int i;
1319 SV **svp;
1320
1321 for (i = av_len (dec.shareable) + 1; i--; )
1322 if ((svp = av_fetch (dec.shareable, i, 0)))
1323 sv_setsv (*svp, &PL_sv_undef);
1324 }
1325
1154 SvREFCNT_dec (sv); 1326 SvREFCNT_dec (sv);
1327
1328 if (dec.err_sv)
1329 sv_2mortal (dec.err_sv);
1330
1155 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur); 1331 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur);
1156 } 1332 }
1157 1333
1158 sv = sv_2mortal (sv); 1334 sv = sv_2mortal (sv);
1159 1335
1160 return sv; 1336 return sv;
1161} 1337}
1162 1338
1339/////////////////////////////////////////////////////////////////////////////
1340// incremental parser
1341
1342#define INCR_DONE(cbor) (AvFILLp (cbor->incr_count) < 0)
1343
1344// returns 0 for notyet, 1 for success or error
1345static int
1346incr_parse (CBOR *self, SV *cborstr)
1347{
1348 STRLEN cur;
1349 SvPV (cborstr, cur);
1350
1351 while (ecb_expect_true (self->incr_need <= cur))
1352 {
1353 // table of integer count bytes
1354 static I8 incr_len[MINOR_MASK + 1] = {
1355 0, 0, 0, 0, 0, 0, 0, 0,
1356 0, 0, 0, 0, 0, 0, 0, 0,
1357 0, 0, 0, 0, 0, 0, 0, 0,
1358 1, 2, 4, 8,-1,-1,-1,-2
1359 };
1360
1361 const U8 *p = SvPVX (cborstr) + self->incr_pos;
1362 U8 m = *p & MINOR_MASK;
1363 IV count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1364 I8 ilen = incr_len[m];
1365
1366 self->incr_need = self->incr_pos + 1;
1367
1368 if (ecb_expect_false (ilen < 0))
1369 {
1370 if (m != MINOR_INDEF)
1371 return 1; // error
1372
1373 if (*p == (MAJOR_MISC | MINOR_INDEF))
1374 {
1375 if (count >= 0)
1376 return 1; // error
1377
1378 count = 1;
1379 }
1380 else
1381 {
1382 av_push (self->incr_count, newSViv (-1)); //TODO: nest
1383 count = -1;
1384 }
1385 }
1386 else
1387 {
1388 self->incr_need += ilen;
1389 if (ecb_expect_false (self->incr_need > cur))
1390 return 0;
1391
1392 int major = *p >> MAJOR_SHIFT;
1393
1394 switch (major)
1395 {
1396 case MAJOR_TAG >> MAJOR_SHIFT:
1397 ++count; // tags merely prefix another value
1398 break;
1399
1400 case MAJOR_BYTES >> MAJOR_SHIFT:
1401 case MAJOR_TEXT >> MAJOR_SHIFT:
1402 case MAJOR_ARRAY >> MAJOR_SHIFT:
1403 case MAJOR_MAP >> MAJOR_SHIFT:
1404 {
1405 UV len;
1406
1407 if (ecb_expect_false (ilen))
1408 {
1409 len = 0;
1410
1411 do {
1412 len = (len << 8) | *++p;
1413 } while (--ilen);
1414 }
1415 else
1416 len = m;
1417
1418 switch (major)
1419 {
1420 case MAJOR_BYTES >> MAJOR_SHIFT:
1421 case MAJOR_TEXT >> MAJOR_SHIFT:
1422 self->incr_need += len;
1423 if (ecb_expect_false (self->incr_need > cur))
1424 return 0;
1425
1426 break;
1427
1428 case MAJOR_MAP >> MAJOR_SHIFT:
1429 len <<= 1;
1430 case MAJOR_ARRAY >> MAJOR_SHIFT:
1431 if (len)
1432 {
1433 av_push (self->incr_count, newSViv (len + 1)); //TODO: nest
1434 count = len + 1;
1435 }
1436 break;
1437 }
1438 }
1439 }
1440 }
1441
1442 self->incr_pos = self->incr_need;
1443
1444 if (count > 0)
1445 {
1446 while (!--count)
1447 {
1448 if (!AvFILLp (self->incr_count))
1449 return 1; // done
1450
1451 SvREFCNT_dec_NN (av_pop (self->incr_count));
1452 count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1453 }
1454
1455 SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]) = count;
1456 }
1457 }
1458
1459 return 0;
1460}
1461
1462
1163///////////////////////////////////////////////////////////////////////////// 1463/////////////////////////////////////////////////////////////////////////////
1164// XS interface functions 1464// XS interface functions
1165 1465
1166MODULE = CBOR::XS PACKAGE = CBOR::XS 1466MODULE = CBOR::XS PACKAGE = CBOR::XS
1167 1467
… …
1179 1479
1180 default_filter = newSVpv ("CBOR::XS::default_filter", 0); 1480 default_filter = newSVpv ("CBOR::XS::default_filter", 0);
1181 1481
1182 sv_cbor = newSVpv ("CBOR", 0); 1482 sv_cbor = newSVpv ("CBOR", 0);
1183 SvREADONLY_on (sv_cbor); 1483 SvREADONLY_on (sv_cbor);
1484
1485 assert (("STRLEN must be an unsigned type", 0 <= (STRLEN)-1));
1184} 1486}
1185 1487
1186PROTOTYPES: DISABLE 1488PROTOTYPES: DISABLE
1187 1489
1188void CLONE (...) 1490void CLONE (...)
… …
1207void shrink (CBOR *self, int enable = 1) 1509void shrink (CBOR *self, int enable = 1)
1208 ALIAS: 1510 ALIAS:
1209 shrink = F_SHRINK 1511 shrink = F_SHRINK
1210 allow_unknown = F_ALLOW_UNKNOWN 1512 allow_unknown = F_ALLOW_UNKNOWN
1211 allow_sharing = F_ALLOW_SHARING 1513 allow_sharing = F_ALLOW_SHARING
1514 allow_cycles = F_ALLOW_CYCLES
1515 forbid_objects = F_FORBID_OBJECTS
1212 pack_strings = F_PACK_STRINGS 1516 pack_strings = F_PACK_STRINGS
1517 text_keys = F_TEXT_KEYS
1518 text_strings = F_TEXT_STRINGS
1519 validate_utf8 = F_VALIDATE_UTF8
1213 PPCODE: 1520 PPCODE:
1214{ 1521{
1215 if (enable) 1522 if (enable)
1216 self->flags |= ix; 1523 self->flags |= ix;
1217 else 1524 else
… …
1223void get_shrink (CBOR *self) 1530void get_shrink (CBOR *self)
1224 ALIAS: 1531 ALIAS:
1225 get_shrink = F_SHRINK 1532 get_shrink = F_SHRINK
1226 get_allow_unknown = F_ALLOW_UNKNOWN 1533 get_allow_unknown = F_ALLOW_UNKNOWN
1227 get_allow_sharing = F_ALLOW_SHARING 1534 get_allow_sharing = F_ALLOW_SHARING
1535 get_allow_cycles = F_ALLOW_CYCLES
1536 get_forbid_objects = F_FORBID_OBJECTS
1228 get_pack_strings = F_PACK_STRINGS 1537 get_pack_strings = F_PACK_STRINGS
1538 get_text_keys = F_TEXT_KEYS
1539 get_text_strings = F_TEXT_STRINGS
1540 get_validate_utf8 = F_VALIDATE_UTF8
1229 PPCODE: 1541 PPCODE:
1230 XPUSHs (boolSV (self->flags & ix)); 1542 XPUSHs (boolSV (self->flags & ix));
1231 1543
1232void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) 1544void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
1233 PPCODE: 1545 PPCODE:
… …
1282 EXTEND (SP, 2); 1594 EXTEND (SP, 2);
1283 PUSHs (sv); 1595 PUSHs (sv);
1284 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); 1596 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
1285} 1597}
1286 1598
1599void incr_parse (CBOR *self, SV *cborstr)
1600 ALIAS:
1601 incr_parse_multiple = 1
1602 PPCODE:
1603{
1604 if (SvUTF8 (cborstr))
1605 sv_utf8_downgrade (cborstr, 0);
1606
1607 if (!self->incr_count)
1608 {
1609 self->incr_count = newAV ();
1610 self->incr_pos = 0;
1611 self->incr_need = 1;
1612
1613 av_push (self->incr_count, newSViv (1));
1614 }
1615
1616 do
1617 {
1618 if (!incr_parse (self, cborstr))
1619 {
1620 if (self->incr_need > self->max_size && self->max_size)
1621 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
1622 (unsigned long)self->incr_need, (unsigned long)self->max_size);
1623
1624 break;
1625 }
1626
1627 SV *sv;
1628 char *offset;
1629
1630 PUTBACK; sv = decode_cbor (cborstr, self, &offset); SPAGAIN;
1631 XPUSHs (sv);
1632
1633 sv_chop (cborstr, offset);
1634
1635 av_clear (self->incr_count);
1636 av_push (self->incr_count, newSViv (1));
1637
1638 self->incr_pos = 0;
1639 self->incr_need = self->incr_pos + 1;
1640 }
1641 while (ix);
1642}
1643
1644void incr_reset (CBOR *self)
1645 CODE:
1646{
1647 SvREFCNT_dec (self->incr_count);
1648 self->incr_count = 0;
1649}
1650
1287void DESTROY (CBOR *self) 1651void DESTROY (CBOR *self)
1288 PPCODE: 1652 PPCODE:
1289 cbor_free (self); 1653 cbor_free (self);
1290 1654
1291PROTOTYPES: ENABLE 1655PROTOTYPES: ENABLE

Diff Legend

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