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

Comparing cvsroot/CBOR-XS/XS.xs (file contents):
Revision 1.36 by root, Sat Nov 30 17:37:45 2013 UTC vs.
Revision 1.59 by root, Fri Nov 25 23:37:27 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
102#define F_PACK_STRINGS 0x00000008UL 104#define F_PACK_STRINGS 0x00000010UL
105#define F_TEXT_KEYS 0x00000020UL
106#define F_TEXT_STRINGS 0x00000040UL
107#define F_VALIDATE_UTF8 0x00000080UL
103 108
104#define INIT_SIZE 32 // initial scalar size to be allocated 109#define INIT_SIZE 32 // initial scalar size to be allocated
105 110
106#define SB do { 111#define SB do {
107#define SE } while (0) 112#define SE } while (0)
… …
126typedef struct { 131typedef struct {
127 U32 flags; 132 U32 flags;
128 U32 max_depth; 133 U32 max_depth;
129 STRLEN max_size; 134 STRLEN max_size;
130 SV *filter; 135 SV *filter;
136
137 // for the incremental parser
138 STRLEN incr_pos; // the current offset into the text
139 STRLEN incr_need; // minimum bytes needed to decode
140 AV *incr_count; // for every nesting level, the number of outstanding values, or -1 for indef.
131} CBOR; 141} CBOR;
132 142
133ecb_inline void 143ecb_inline void
134cbor_init (CBOR *cbor) 144cbor_init (CBOR *cbor)
135{ 145{
… …
139 149
140ecb_inline void 150ecb_inline void
141cbor_free (CBOR *cbor) 151cbor_free (CBOR *cbor)
142{ 152{
143 SvREFCNT_dec (cbor->filter); 153 SvREFCNT_dec (cbor->filter);
154 SvREFCNT_dec (cbor->incr_count);
144} 155}
145 156
146///////////////////////////////////////////////////////////////////////////// 157/////////////////////////////////////////////////////////////////////////////
147// utility functions 158// utility functions
148 159
… …
267encode_tag (enc_t *enc, UV tag) 278encode_tag (enc_t *enc, UV tag)
268{ 279{
269 encode_uint (enc, MAJOR_TAG, tag); 280 encode_uint (enc, MAJOR_TAG, tag);
270} 281}
271 282
283// exceptional (hopefully) slow path for byte strings that need to be utf8-encoded
284ecb_noinline static void
285encode_str_utf8 (enc_t *enc, int utf8, char *str, STRLEN len)
286{
287 STRLEN ulen = len;
288 U8 *p, *pend = (U8 *)str + len;
289
290 for (p = (U8 *)str; p < pend; ++p)
291 ulen += *p >> 7; // count set high bits
292
293 encode_uint (enc, MAJOR_TEXT, ulen);
294
295 need (enc, ulen);
296 for (p = (U8 *)str; p < pend; ++p)
297 if (*p < 0x80)
298 *enc->cur++ = *p;
299 else
300 {
301 *enc->cur++ = 0xc0 + (*p >> 6);
302 *enc->cur++ = 0x80 + (*p & 63);
303 }
304}
305
272ecb_inline void 306ecb_inline void
273encode_str (enc_t *enc, int utf8, char *str, STRLEN len) 307encode_str (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len)
274{ 308{
309 if (ecb_expect_false (upgrade_utf8))
310 if (!utf8)
311 {
312 encode_str_utf8 (enc, utf8, str, len);
313 return;
314 }
315
275 encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len); 316 encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len);
276 need (enc, len); 317 need (enc, len);
277 memcpy (enc->cur, str, len); 318 memcpy (enc->cur, str, len);
278 enc->cur += len; 319 enc->cur += len;
279} 320}
280 321
281static void 322ecb_inline void
282encode_strref (enc_t *enc, int utf8, char *str, STRLEN len) 323encode_strref (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len)
283{ 324{
284 if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS)) 325 if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS))
285 { 326 {
286 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1); 327 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1);
287 328
… …
298 sv_setuv (*svp, enc->stringref_idx); 339 sv_setuv (*svp, enc->stringref_idx);
299 ++enc->stringref_idx; 340 ++enc->stringref_idx;
300 } 341 }
301 } 342 }
302 343
303 encode_str (enc, utf8, str, len); 344 encode_str (enc, upgrade_utf8, utf8, str, len);
304} 345}
305 346
306static void encode_sv (enc_t *enc, SV *sv); 347static void encode_sv (enc_t *enc, SV *sv);
307 348
308static void 349static void
… …
315 356
316 ++enc->depth; 357 ++enc->depth;
317 358
318 encode_uint (enc, MAJOR_ARRAY, len + 1); 359 encode_uint (enc, MAJOR_ARRAY, len + 1);
319 360
361 if (SvMAGICAL (av))
320 for (i = 0; i <= len; ++i) 362 for (i = 0; i <= len; ++i)
321 { 363 {
322 SV **svp = av_fetch (av, i, 0); 364 SV **svp = av_fetch (av, i, 0);
323 encode_sv (enc, svp ? *svp : &PL_sv_undef); 365 encode_sv (enc, svp ? *svp : &PL_sv_undef);
324 } 366 }
367 else
368 for (i = 0; i <= len; ++i)
369 {
370 SV *sv = AvARRAY (av)[i];
371 encode_sv (enc, sv ? sv : &PL_sv_undef);
372 }
325 373
326 --enc->depth; 374 --enc->depth;
327} 375}
328 376
329static void 377static void
… …
347 while ((he = hv_iternext (hv))) 395 while ((he = hv_iternext (hv)))
348 { 396 {
349 if (HeKLEN (he) == HEf_SVKEY) 397 if (HeKLEN (he) == HEf_SVKEY)
350 encode_sv (enc, HeSVKEY (he)); 398 encode_sv (enc, HeSVKEY (he));
351 else 399 else
352 encode_strref (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he)); 400 encode_strref (enc, enc->cbor.flags & (F_TEXT_KEYS | F_TEXT_STRINGS), HeKUTF8 (he), HeKEY (he), HeKLEN (he));
353 401
354 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he)); 402 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he));
355 } 403 }
356 404
357 if (mg) 405 if (mg)
… …
433 481
434 if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0))) 482 if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
435 { 483 {
436 dSP; 484 dSP;
437 485
438 ENTER; SAVETMPS; PUSHMARK (SP); 486 ENTER; SAVETMPS;
487 PUSHMARK (SP);
439 // we re-bless the reference to get overload and other niceties right 488 // we re-bless the reference to get overload and other niceties right
440 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); 489 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
441 490
442 PUTBACK; 491 PUTBACK;
443 // G_SCALAR ensures that return value is 1 492 // G_SCALAR ensures that return value is 1
… …
456 } 505 }
457 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0) 506 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)
458 { 507 {
459 dSP; 508 dSP;
460 509
461 ENTER; SAVETMPS; PUSHMARK (SP); 510 ENTER; SAVETMPS;
511 SAVESTACK_POS ();
512 PUSHMARK (SP);
462 EXTEND (SP, 2); 513 EXTEND (SP, 2);
463 // we re-bless the reference to get overload and other niceties right 514 // we re-bless the reference to get overload and other niceties right
464 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); 515 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
465 PUSHs (sv_cbor); 516 PUSHs (sv_cbor);
466 517
… …
472 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv) 523 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)); 524 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));
474 525
475 encode_tag (enc, CBOR_TAG_PERL_OBJECT); 526 encode_tag (enc, CBOR_TAG_PERL_OBJECT);
476 encode_uint (enc, MAJOR_ARRAY, count + 1); 527 encode_uint (enc, MAJOR_ARRAY, count + 1);
477 encode_strref (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash)); 528 encode_strref (enc, 0, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
478 529
479 while (count) 530 while (count)
480 encode_sv (enc, SP[1 - count--]); 531 encode_sv (enc, SP[1 - count--]);
481 532
482 PUTBACK; 533 PUTBACK;
… …
541 592
542 if (SvPOKp (sv)) 593 if (SvPOKp (sv))
543 { 594 {
544 STRLEN len; 595 STRLEN len;
545 char *str = SvPV (sv, len); 596 char *str = SvPV (sv, len);
546 encode_strref (enc, SvUTF8 (sv), str, len); 597 encode_strref (enc, enc->cbor.flags & F_TEXT_STRINGS, SvUTF8 (sv), str, len);
547 } 598 }
548 else if (SvNOKp (sv)) 599 else if (SvNOKp (sv))
549 encode_nv (enc, sv); 600 encode_nv (enc, sv);
550 else if (SvIOKp (sv)) 601 else if (SvIOKp (sv))
551 { 602 {
… …
568} 619}
569 620
570static SV * 621static SV *
571encode_cbor (SV *scalar, CBOR *cbor) 622encode_cbor (SV *scalar, CBOR *cbor)
572{ 623{
573 enc_t enc = { }; 624 enc_t enc = { 0 };
574 625
575 enc.cbor = *cbor; 626 enc.cbor = *cbor;
576 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 627 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
577 enc.cur = SvPVX (enc.sv); 628 enc.cur = SvPVX (enc.sv);
578 enc.end = SvEND (enc.sv); 629 enc.end = SvEND (enc.sv);
… …
610 U32 depth; // recursion depth 661 U32 depth; // recursion depth
611 U32 maxdepth; // recursion depth limit 662 U32 maxdepth; // recursion depth limit
612 AV *shareable; 663 AV *shareable;
613 AV *stringref; 664 AV *stringref;
614 SV *decode_tagged; 665 SV *decode_tagged;
666 SV *err_sv; // optional sv for error, needs to be freed
615} dec_t; 667} dec_t;
616 668
617#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE 669// set dec->err to ERRSV
670ecb_cold static void
671err_errsv (dec_t *dec)
672{
673 if (!dec->err)
674 {
675 dec->err_sv = newSVsv (ERRSV);
618 676
619#define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data") 677 // chop off the trailing \n
678 SvCUR_set (dec->err_sv, SvCUR (dec->err_sv) - 1);
679 *SvEND (dec->err_sv) = 0;
620 680
681 dec->err = SvPVutf8_nolen (dec->err_sv);
682 }
683}
684
685// the following functions are used to reduce code size and help the compiler to optimise
686ecb_cold static void
687err_set (dec_t *dec, const char *reason)
688{
689 if (!dec->err)
690 dec->err = reason;
691}
692
693ecb_cold static void
694err_unexpected_end (dec_t *dec)
695{
696 err_set (dec, "unexpected end of CBOR data");
697}
698
699ecb_cold static void
700err_nesting_exceeded (dec_t *dec)
701{
702 err_set (dec, ERR_NESTING_EXCEEDED);
703}
704
705#define ERR_DO(do) SB do; goto fail; SE
706#define ERR(reason) ERR_DO (err_set (dec, reason))
707#define ERR_ERRSV ERR_DO (err_errsv (dec))
708
709#define WANT(len) if (ecb_expect_false ((UV)(dec->end - dec->cur) < (UV)len)) ERR_DO (err_unexpected_end (dec))
710
621#define DEC_INC_DEPTH if (++dec->depth > dec->cbor.max_depth) ERR (ERR_NESTING_EXCEEDED) 711#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 712#define DEC_DEC_DEPTH --dec->depth
623 713
624static UV 714static UV
625decode_uint (dec_t *dec) 715decode_uint (dec_t *dec)
626{ 716{
… …
703 av_push (av, decode_sv (dec)); 793 av_push (av, decode_sv (dec));
704 } 794 }
705 } 795 }
706 else 796 else
707 { 797 {
708 int i, len = decode_uint (dec); 798 UV i, len = decode_uint (dec);
709 799
710 WANT (len); // complexity check for av_fill - need at least one byte per value, do not allow supersize arrays 800 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); 801 av_fill (av, len - 1);
712 802
713 for (i = 0; i < len; ++i) 803 for (i = 0; i < len; ++i)
… …
728{ 818{
729 // for speed reasons, we specialcase single-string 819 // for speed reasons, we specialcase single-string
730 // byte or utf-8 strings as keys, but only when !stringref 820 // byte or utf-8 strings as keys, but only when !stringref
731 821
732 if (ecb_expect_true (!dec->stringref)) 822 if (ecb_expect_true (!dec->stringref))
733 if (ecb_expect_true ((*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8)) 823 if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8))
734 { 824 {
735 I32 len = decode_uint (dec); 825 STRLEN len = decode_uint (dec);
736 char *key = (char *)dec->cur; 826 char *key = (char *)dec->cur;
737 827
828 WANT (len);
738 dec->cur += len; 829 dec->cur += len;
739 830
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); 831 hv_store (hv, key, len, decode_sv (dec), 0);
744 832
745 return; 833 return;
746 } 834 }
747 else if (ecb_expect_true ((*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8)) 835 else if (ecb_expect_true ((U8)(*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8))
748 { 836 {
749 I32 len = decode_uint (dec); 837 STRLEN len = decode_uint (dec);
750 char *key = (char *)dec->cur; 838 char *key = (char *)dec->cur;
751 839
840 WANT (len);
752 dec->cur += len; 841 dec->cur += len;
753 842
754 if (ecb_expect_false (dec->stringref)) 843 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
755 av_push (dec->stringref, newSVpvn_utf8 (key, len, 1)); 844 if (!is_utf8_string (key, len))
845 ERR ("corrupted CBOR data (invalid UTF-8 in map key)");
756 846
757 hv_store (hv, key, -len, decode_sv (dec), 0); 847 hv_store (hv, key, -len, decode_sv (dec), 0);
758 848
759 return; 849 return;
760 } 850 }
761 851
762 SV *k = decode_sv (dec); 852 SV *k = decode_sv (dec);
763 SV *v = decode_sv (dec); 853 SV *v = decode_sv (dec);
764 854
855 // we leak memory if uncaught exceptions are thrown by random magical
856 // methods, and this is hopefully the only place where it can happen,
857 // so if there is a chance of an exception, take the very slow path.
858 // since catching exceptions is "undocumented/internal/forbidden" by
859 // the new p5p powers, we need to call out to a perl function :/
860 if (ecb_expect_false (SvAMAGIC (k)))
861 {
862 dSP;
863
864 ENTER; SAVETMPS;
865 PUSHMARK (SP);
866 EXTEND (SP, 3);
867 PUSHs (sv_2mortal (newRV_inc ((SV *)hv)));
868 PUSHs (sv_2mortal (k));
869 PUSHs (sv_2mortal (v));
870
871 PUTBACK;
872 call_pv ("CBOR::XS::_hv_store", G_VOID | G_DISCARD | G_EVAL);
873 SPAGAIN;
874
875 FREETMPS; LEAVE;
876
877 if (SvTRUE (ERRSV))
878 ERR_ERRSV;
879
880 return;
881 }
882
765 hv_store_ent (hv, k, v, 0); 883 hv_store_ent (hv, k, v, 0);
766 SvREFCNT_dec (k); 884 SvREFCNT_dec (k);
885
886fail:
887 ;
767} 888}
768 889
769static SV * 890static SV *
770decode_hv (dec_t *dec) 891decode_hv (dec_t *dec)
771{ 892{
… …
790 decode_he (dec, hv); 911 decode_he (dec, hv);
791 } 912 }
792 } 913 }
793 else 914 else
794 { 915 {
795 int pairs = decode_uint (dec); 916 UV pairs = decode_uint (dec);
917
918 WANT (pairs); // complexity check - need at least one byte per value, do not allow supersize hashes
796 919
797 while (pairs--) 920 while (pairs--)
798 decode_he (dec, hv); 921 decode_he (dec, hv);
799 } 922 }
800 923
… …
853 && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1)) 976 && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1))
854 av_push (dec->stringref, SvREFCNT_inc_NN (sv)); 977 av_push (dec->stringref, SvREFCNT_inc_NN (sv));
855 } 978 }
856 979
857 if (utf8) 980 if (utf8)
981 {
982 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
983 if (!is_utf8_string (SvPVX (sv), SvCUR (sv)))
984 ERR ("corrupted CBOR data (invalid UTF-8 in text string)");
985
858 SvUTF8_on (sv); 986 SvUTF8_on (sv);
987 }
859 988
860 return sv; 989 return sv;
861 990
862fail: 991fail:
863 SvREFCNT_dec (sv); 992 SvREFCNT_dec (sv);
… …
882 sv = newRV_noinc (decode_sv (dec)); 1011 sv = newRV_noinc (decode_sv (dec));
883 break; 1012 break;
884 1013
885 case CBOR_TAG_STRINGREF_NAMESPACE: 1014 case CBOR_TAG_STRINGREF_NAMESPACE:
886 { 1015 {
1016 // do nmot use SAVETMPS/FREETMPS, as these will
1017 // erase mortalised caches, e.g. "shareable"
887 ENTER; SAVETMPS; 1018 ENTER;
888 1019
889 SAVESPTR (dec->stringref); 1020 SAVESPTR (dec->stringref);
890 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ()); 1021 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ());
891 1022
892 sv = decode_sv (dec); 1023 sv = decode_sv (dec);
893 1024
894 FREETMPS; LEAVE; 1025 LEAVE;
895 } 1026 }
896 break; 1027 break;
897 1028
898 case CBOR_TAG_STRINGREF: 1029 case CBOR_TAG_STRINGREF:
899 { 1030 {
… …
912 case CBOR_TAG_VALUE_SHAREABLE: 1043 case CBOR_TAG_VALUE_SHAREABLE:
913 { 1044 {
914 if (ecb_expect_false (!dec->shareable)) 1045 if (ecb_expect_false (!dec->shareable))
915 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ()); 1046 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ());
916 1047
1048 if (dec->cbor.flags & F_ALLOW_CYCLES)
1049 {
917 sv = newSV (0); 1050 sv = newSV (0);
918 av_push (dec->shareable, SvREFCNT_inc_NN (sv)); 1051 av_push (dec->shareable, SvREFCNT_inc_NN (sv));
919 1052
920 SV *osv = decode_sv (dec); 1053 SV *osv = decode_sv (dec);
921 sv_setsv (sv, osv); 1054 sv_setsv (sv, osv);
922 SvREFCNT_dec_NN (osv); 1055 SvREFCNT_dec_NN (osv);
1056 }
1057 else
1058 {
1059 av_push (dec->shareable, &PL_sv_undef);
1060 int idx = AvFILLp (dec->shareable);
1061 sv = decode_sv (dec);
1062 av_store (dec->shareable, idx, SvREFCNT_inc_NN (sv));
1063 }
923 } 1064 }
924 break; 1065 break;
925 1066
926 case CBOR_TAG_VALUE_SHAREDREF: 1067 case CBOR_TAG_VALUE_SHAREDREF:
927 { 1068 {
… …
932 1073
933 if (!dec->shareable || (int)idx > AvFILLp (dec->shareable)) 1074 if (!dec->shareable || (int)idx > AvFILLp (dec->shareable))
934 ERR ("corrupted CBOR data (sharedref index out of bounds)"); 1075 ERR ("corrupted CBOR data (sharedref index out of bounds)");
935 1076
936 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]); 1077 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]);
1078
1079 if (sv == &PL_sv_undef)
1080 ERR ("cyclic CBOR data structure found, but allow_cycles is not enabled");
937 } 1081 }
938 break; 1082 break;
939 1083
940 case CBOR_TAG_PERL_OBJECT: 1084 case CBOR_TAG_PERL_OBJECT:
941 { 1085 {
… …
956 if (!method) 1100 if (!method)
957 ERR ("cannot decode perl-object (package does not have a THAW method)"); 1101 ERR ("cannot decode perl-object (package does not have a THAW method)");
958 1102
959 dSP; 1103 dSP;
960 1104
961 ENTER; SAVETMPS; PUSHMARK (SP); 1105 ENTER; SAVETMPS;
1106 PUSHMARK (SP);
962 EXTEND (SP, len + 1); 1107 EXTEND (SP, len + 1);
963 // we re-bless the reference to get overload and other niceties right 1108 // we re-bless the reference to get overload and other niceties right
964 PUSHs (*av_fetch (av, 0, 1)); 1109 PUSHs (*av_fetch (av, 0, 1));
965 PUSHs (sv_cbor); 1110 PUSHs (sv_cbor);
966 1111
… …
974 SPAGAIN; 1119 SPAGAIN;
975 1120
976 if (SvTRUE (ERRSV)) 1121 if (SvTRUE (ERRSV))
977 { 1122 {
978 FREETMPS; LEAVE; 1123 FREETMPS; LEAVE;
979 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV)))); 1124 ERR_ERRSV;
980 } 1125 }
981 1126
982 SvREFCNT_dec (sv); 1127 SvREFCNT_dec (sv);
983 sv = SvREFCNT_inc (POPs); 1128 sv = SvREFCNT_inc (POPs);
984 1129
… …
988 } 1133 }
989 break; 1134 break;
990 1135
991 default: 1136 default:
992 { 1137 {
1138 SV *tag_sv = newSVuv (tag);
1139
993 sv = decode_sv (dec); 1140 sv = decode_sv (dec);
994 1141
995 dSP; 1142 dSP;
996 ENTER; SAVETMPS; PUSHMARK (SP); 1143 ENTER; SAVETMPS;
1144 SAVESTACK_POS ();
1145 PUSHMARK (SP);
997 EXTEND (SP, 2); 1146 EXTEND (SP, 2);
998 PUSHs (newSVuv (tag)); 1147 PUSHs (tag_sv);
999 PUSHs (sv); 1148 PUSHs (sv);
1000 1149
1001 PUTBACK; 1150 PUTBACK;
1002 int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL); 1151 int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL);
1003 SPAGAIN; 1152 SPAGAIN;
1004 1153
1005 if (SvTRUE (ERRSV)) 1154 if (SvTRUE (ERRSV))
1006 { 1155 {
1156 SvREFCNT_dec (tag_sv);
1007 FREETMPS; LEAVE; 1157 FREETMPS; LEAVE;
1008 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV)))); 1158 ERR_ERRSV;
1009 } 1159 }
1010 1160
1011 if (count) 1161 if (count)
1012 { 1162 {
1163 SvREFCNT_dec (tag_sv);
1013 SvREFCNT_dec (sv); 1164 SvREFCNT_dec (sv);
1014 sv = SvREFCNT_inc (POPs); 1165 sv = SvREFCNT_inc (POPs);
1015 } 1166 }
1016 else 1167 else
1017 { 1168 {
1018 AV *av = newAV (); 1169 AV *av = newAV ();
1019 av_push (av, newSVuv (tag)); 1170 av_push (av, tag_sv);
1020 av_push (av, sv); 1171 av_push (av, sv);
1021 1172
1022 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 1173 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
1023 ? cbor_tagged_stash 1174 ? cbor_tagged_stash
1024 : gv_stashpv ("CBOR::XS::Tagged" , 1); 1175 : gv_stashpv ("CBOR::XS::Tagged" , 1);
… …
1110 1261
1111 return newSVnv (ecb_binary64_to_double (fp)); 1262 return newSVnv (ecb_binary64_to_double (fp));
1112 } 1263 }
1113 1264
1114 // 0..19 unassigned simple 1265 // 0..19 unassigned simple
1115 // 24 reserved + unassigned (reserved values are not encodable) 1266 // 24 reserved + unassigned simple (reserved values are not encodable)
1267 // 28-30 unassigned misc
1268 // 31 break code
1116 default: 1269 default:
1117 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)"); 1270 ERR ("corrupted CBOR data (reserved/unassigned/unexpected major 7 value)");
1118 } 1271 }
1119 1272
1120 break; 1273 break;
1121 } 1274 }
1122 1275
… …
1125} 1278}
1126 1279
1127static SV * 1280static SV *
1128decode_cbor (SV *string, CBOR *cbor, char **offset_return) 1281decode_cbor (SV *string, CBOR *cbor, char **offset_return)
1129{ 1282{
1130 dec_t dec = { }; 1283 dec_t dec = { 0 };
1131 SV *sv; 1284 SV *sv;
1132 STRLEN len; 1285 STRLEN len;
1133 char *data = SvPVbyte (string, len); 1286 char *data = SvPVbyte (string, len);
1134 1287
1135 if (len > cbor->max_size && cbor->max_size) 1288 if (len > cbor->max_size && cbor->max_size)
… …
1149 if (dec.cur != dec.end && !dec.err) 1302 if (dec.cur != dec.end && !dec.err)
1150 dec.err = "garbage after CBOR object"; 1303 dec.err = "garbage after CBOR object";
1151 1304
1152 if (dec.err) 1305 if (dec.err)
1153 { 1306 {
1307 if (dec.shareable)
1308 {
1309 // need to break cyclic links, which would all be in shareable
1310 int i;
1311 SV **svp;
1312
1313 for (i = av_len (dec.shareable) + 1; i--; )
1314 if ((svp = av_fetch (dec.shareable, i, 0)))
1315 sv_setsv (*svp, &PL_sv_undef);
1316 }
1317
1154 SvREFCNT_dec (sv); 1318 SvREFCNT_dec (sv);
1319
1320 if (dec.err_sv)
1321 sv_2mortal (dec.err_sv);
1322
1155 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur); 1323 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur);
1156 } 1324 }
1157 1325
1158 sv = sv_2mortal (sv); 1326 sv = sv_2mortal (sv);
1159 1327
1160 return sv; 1328 return sv;
1161} 1329}
1162 1330
1331/////////////////////////////////////////////////////////////////////////////
1332// incremental parser
1333
1334#define INCR_DONE(cbor) (AvFILLp (cbor->incr_count) < 0)
1335
1336// returns 0 for notyet, 1 for success or error
1337static int
1338incr_parse (CBOR *self, SV *cborstr)
1339{
1340 STRLEN cur;
1341 SvPV (cborstr, cur);
1342
1343 while (ecb_expect_true (self->incr_need <= cur))
1344 {
1345 // table of integer count bytes
1346 static I8 incr_len[MINOR_MASK + 1] = {
1347 0, 0, 0, 0, 0, 0, 0, 0,
1348 0, 0, 0, 0, 0, 0, 0, 0,
1349 0, 0, 0, 0, 0, 0, 0, 0,
1350 1, 2, 4, 8,-1,-1,-1,-2
1351 };
1352
1353 const U8 *p = SvPVX (cborstr) + self->incr_pos;
1354 U8 m = *p & MINOR_MASK;
1355 IV count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1356 I8 ilen = incr_len[m];
1357
1358 self->incr_need = self->incr_pos + 1;
1359
1360 if (ecb_expect_false (ilen < 0))
1361 {
1362 if (m != MINOR_INDEF)
1363 return 1; // error
1364
1365 if (*p == (MAJOR_MISC | MINOR_INDEF))
1366 {
1367 if (count >= 0)
1368 return 1; // error
1369
1370 count = 1;
1371 }
1372 else
1373 {
1374 av_push (self->incr_count, newSViv (-1)); //TODO: nest
1375 count = -1;
1376 }
1377 }
1378 else
1379 {
1380 self->incr_need += ilen;
1381 if (ecb_expect_false (self->incr_need > cur))
1382 return 0;
1383
1384 int major = *p >> MAJOR_SHIFT;
1385
1386 switch (major)
1387 {
1388 case MAJOR_TAG >> MAJOR_SHIFT:
1389 ++count; // tags merely prefix another value
1390 break;
1391
1392 case MAJOR_BYTES >> MAJOR_SHIFT:
1393 case MAJOR_TEXT >> MAJOR_SHIFT:
1394 case MAJOR_ARRAY >> MAJOR_SHIFT:
1395 case MAJOR_MAP >> MAJOR_SHIFT:
1396 {
1397 UV len;
1398
1399 if (ecb_expect_false (ilen))
1400 {
1401 len = 0;
1402
1403 do {
1404 len = (len << 8) | *++p;
1405 } while (--ilen);
1406 }
1407 else
1408 len = m;
1409
1410 switch (major)
1411 {
1412 case MAJOR_BYTES >> MAJOR_SHIFT:
1413 case MAJOR_TEXT >> MAJOR_SHIFT:
1414 self->incr_need += len;
1415 if (ecb_expect_false (self->incr_need > cur))
1416 return 0;
1417
1418 break;
1419
1420 case MAJOR_MAP >> MAJOR_SHIFT:
1421 len <<= 1;
1422 case MAJOR_ARRAY >> MAJOR_SHIFT:
1423 if (len)
1424 {
1425 av_push (self->incr_count, newSViv (len + 1)); //TODO: nest
1426 count = len + 1;
1427 }
1428 break;
1429 }
1430 }
1431 }
1432 }
1433
1434 self->incr_pos = self->incr_need;
1435
1436 if (count > 0)
1437 {
1438 while (!--count)
1439 {
1440 if (!AvFILLp (self->incr_count))
1441 return 1; // done
1442
1443 SvREFCNT_dec_NN (av_pop (self->incr_count));
1444 count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1445 }
1446
1447 SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]) = count;
1448 }
1449 }
1450
1451 return 0;
1452}
1453
1454
1163///////////////////////////////////////////////////////////////////////////// 1455/////////////////////////////////////////////////////////////////////////////
1164// XS interface functions 1456// XS interface functions
1165 1457
1166MODULE = CBOR::XS PACKAGE = CBOR::XS 1458MODULE = CBOR::XS PACKAGE = CBOR::XS
1167 1459
… …
1179 1471
1180 default_filter = newSVpv ("CBOR::XS::default_filter", 0); 1472 default_filter = newSVpv ("CBOR::XS::default_filter", 0);
1181 1473
1182 sv_cbor = newSVpv ("CBOR", 0); 1474 sv_cbor = newSVpv ("CBOR", 0);
1183 SvREADONLY_on (sv_cbor); 1475 SvREADONLY_on (sv_cbor);
1476
1477 assert (("STRLEN must be an unsigned type", 0 <= (STRLEN)-1));
1184} 1478}
1185 1479
1186PROTOTYPES: DISABLE 1480PROTOTYPES: DISABLE
1187 1481
1188void CLONE (...) 1482void CLONE (...)
… …
1207void shrink (CBOR *self, int enable = 1) 1501void shrink (CBOR *self, int enable = 1)
1208 ALIAS: 1502 ALIAS:
1209 shrink = F_SHRINK 1503 shrink = F_SHRINK
1210 allow_unknown = F_ALLOW_UNKNOWN 1504 allow_unknown = F_ALLOW_UNKNOWN
1211 allow_sharing = F_ALLOW_SHARING 1505 allow_sharing = F_ALLOW_SHARING
1506 allow_cycles = F_ALLOW_CYCLES
1212 pack_strings = F_PACK_STRINGS 1507 pack_strings = F_PACK_STRINGS
1508 text_keys = F_TEXT_KEYS
1509 text_strings = F_TEXT_STRINGS
1510 validate_utf8 = F_VALIDATE_UTF8
1213 PPCODE: 1511 PPCODE:
1214{ 1512{
1215 if (enable) 1513 if (enable)
1216 self->flags |= ix; 1514 self->flags |= ix;
1217 else 1515 else
… …
1223void get_shrink (CBOR *self) 1521void get_shrink (CBOR *self)
1224 ALIAS: 1522 ALIAS:
1225 get_shrink = F_SHRINK 1523 get_shrink = F_SHRINK
1226 get_allow_unknown = F_ALLOW_UNKNOWN 1524 get_allow_unknown = F_ALLOW_UNKNOWN
1227 get_allow_sharing = F_ALLOW_SHARING 1525 get_allow_sharing = F_ALLOW_SHARING
1526 get_allow_cycles = F_ALLOW_CYCLES
1228 get_pack_strings = F_PACK_STRINGS 1527 get_pack_strings = F_PACK_STRINGS
1528 get_text_keys = F_TEXT_KEYS
1529 get_text_strings = F_TEXT_STRINGS
1530 get_validate_utf8 = F_VALIDATE_UTF8
1229 PPCODE: 1531 PPCODE:
1230 XPUSHs (boolSV (self->flags & ix)); 1532 XPUSHs (boolSV (self->flags & ix));
1231 1533
1232void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) 1534void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
1233 PPCODE: 1535 PPCODE:
… …
1282 EXTEND (SP, 2); 1584 EXTEND (SP, 2);
1283 PUSHs (sv); 1585 PUSHs (sv);
1284 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); 1586 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
1285} 1587}
1286 1588
1589void incr_parse (CBOR *self, SV *cborstr)
1590 ALIAS:
1591 incr_parse_multiple = 1
1592 PPCODE:
1593{
1594 if (SvUTF8 (cborstr))
1595 sv_utf8_downgrade (cborstr, 0);
1596
1597 if (!self->incr_count)
1598 {
1599 self->incr_count = newAV ();
1600 self->incr_pos = 0;
1601 self->incr_need = 1;
1602
1603 av_push (self->incr_count, newSViv (1));
1604 }
1605
1606 do
1607 {
1608 if (!incr_parse (self, cborstr))
1609 {
1610 if (self->incr_need > self->max_size && self->max_size)
1611 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
1612 (unsigned long)self->incr_need, (unsigned long)self->max_size);
1613
1614 break;
1615 }
1616
1617 SV *sv;
1618 char *offset;
1619
1620 PUTBACK; sv = decode_cbor (cborstr, self, &offset); SPAGAIN;
1621 XPUSHs (sv);
1622
1623 sv_chop (cborstr, offset);
1624
1625 av_clear (self->incr_count);
1626 av_push (self->incr_count, newSViv (1));
1627
1628 self->incr_pos = 0;
1629 self->incr_need = self->incr_pos + 1;
1630 }
1631 while (ix);
1632}
1633
1634void incr_reset (CBOR *self)
1635 CODE:
1636{
1637 SvREFCNT_dec (self->incr_count);
1638 self->incr_count = 0;
1639}
1640
1287void DESTROY (CBOR *self) 1641void DESTROY (CBOR *self)
1288 PPCODE: 1642 PPCODE:
1289 cbor_free (self); 1643 cbor_free (self);
1290 1644
1291PROTOTYPES: ENABLE 1645PROTOTYPES: ENABLE

Diff Legend

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