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 1.37 by root, Sat Nov 30 18:13:53 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))
… …
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
102#define F_ALLOW_CYCLES 0x00000008UL 103#define F_ALLOW_CYCLES 0x00000008UL
103#define F_PACK_STRINGS 0x00000010UL 104#define F_PACK_STRINGS 0x00000010UL
105#define F_TEXT_KEYS 0x00000020UL
106#define F_TEXT_STRINGS 0x00000040UL
107#define F_VALIDATE_UTF8 0x00000080UL
104 108
105#define INIT_SIZE 32 // initial scalar size to be allocated 109#define INIT_SIZE 32 // initial scalar size to be allocated
106 110
107#define SB do { 111#define SB do {
108#define SE } while (0) 112#define SE } while (0)
… …
127typedef struct { 131typedef struct {
128 U32 flags; 132 U32 flags;
129 U32 max_depth; 133 U32 max_depth;
130 STRLEN max_size; 134 STRLEN max_size;
131 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.
132} CBOR; 141} CBOR;
133 142
134ecb_inline void 143ecb_inline void
135cbor_init (CBOR *cbor) 144cbor_init (CBOR *cbor)
136{ 145{
… …
140 149
141ecb_inline void 150ecb_inline void
142cbor_free (CBOR *cbor) 151cbor_free (CBOR *cbor)
143{ 152{
144 SvREFCNT_dec (cbor->filter); 153 SvREFCNT_dec (cbor->filter);
154 SvREFCNT_dec (cbor->incr_count);
145} 155}
146 156
147///////////////////////////////////////////////////////////////////////////// 157/////////////////////////////////////////////////////////////////////////////
148// utility functions 158// utility functions
149 159
… …
268encode_tag (enc_t *enc, UV tag) 278encode_tag (enc_t *enc, UV tag)
269{ 279{
270 encode_uint (enc, MAJOR_TAG, tag); 280 encode_uint (enc, MAJOR_TAG, tag);
271} 281}
272 282
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
273ecb_inline void 306ecb_inline void
274encode_str (enc_t *enc, int utf8, char *str, STRLEN len) 307encode_str (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len)
275{ 308{
309 if (ecb_expect_false (upgrade_utf8))
310 if (!utf8)
311 {
312 encode_str_utf8 (enc, utf8, str, len);
313 return;
314 }
315
276 encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len); 316 encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len);
277 need (enc, len); 317 need (enc, len);
278 memcpy (enc->cur, str, len); 318 memcpy (enc->cur, str, len);
279 enc->cur += len; 319 enc->cur += len;
280} 320}
281 321
282static void 322ecb_inline void
283encode_strref (enc_t *enc, int utf8, char *str, STRLEN len) 323encode_strref (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len)
284{ 324{
285 if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS)) 325 if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS))
286 { 326 {
287 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1); 327 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1);
288 328
… …
299 sv_setuv (*svp, enc->stringref_idx); 339 sv_setuv (*svp, enc->stringref_idx);
300 ++enc->stringref_idx; 340 ++enc->stringref_idx;
301 } 341 }
302 } 342 }
303 343
304 encode_str (enc, utf8, str, len); 344 encode_str (enc, upgrade_utf8, utf8, str, len);
305} 345}
306 346
307static void encode_sv (enc_t *enc, SV *sv); 347static void encode_sv (enc_t *enc, SV *sv);
308 348
309static void 349static void
… …
316 356
317 ++enc->depth; 357 ++enc->depth;
318 358
319 encode_uint (enc, MAJOR_ARRAY, len + 1); 359 encode_uint (enc, MAJOR_ARRAY, len + 1);
320 360
361 if (SvMAGICAL (av))
321 for (i = 0; i <= len; ++i) 362 for (i = 0; i <= len; ++i)
322 { 363 {
323 SV **svp = av_fetch (av, i, 0); 364 SV **svp = av_fetch (av, i, 0);
324 encode_sv (enc, svp ? *svp : &PL_sv_undef); 365 encode_sv (enc, svp ? *svp : &PL_sv_undef);
325 } 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 }
326 373
327 --enc->depth; 374 --enc->depth;
328} 375}
329 376
330static void 377static void
… …
348 while ((he = hv_iternext (hv))) 395 while ((he = hv_iternext (hv)))
349 { 396 {
350 if (HeKLEN (he) == HEf_SVKEY) 397 if (HeKLEN (he) == HEf_SVKEY)
351 encode_sv (enc, HeSVKEY (he)); 398 encode_sv (enc, HeSVKEY (he));
352 else 399 else
353 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));
354 401
355 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));
356 } 403 }
357 404
358 if (mg) 405 if (mg)
… …
434 481
435 if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0))) 482 if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
436 { 483 {
437 dSP; 484 dSP;
438 485
439 ENTER; SAVETMPS; PUSHMARK (SP); 486 ENTER; SAVETMPS;
487 PUSHMARK (SP);
440 // 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
441 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); 489 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
442 490
443 PUTBACK; 491 PUTBACK;
444 // G_SCALAR ensures that return value is 1 492 // G_SCALAR ensures that return value is 1
… …
457 } 505 }
458 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0) 506 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)
459 { 507 {
460 dSP; 508 dSP;
461 509
462 ENTER; SAVETMPS; PUSHMARK (SP); 510 ENTER; SAVETMPS;
511 SAVESTACK_POS ();
512 PUSHMARK (SP);
463 EXTEND (SP, 2); 513 EXTEND (SP, 2);
464 // 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
465 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); 515 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
466 PUSHs (sv_cbor); 516 PUSHs (sv_cbor);
467 517
… …
473 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv) 523 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)
474 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));
475 525
476 encode_tag (enc, CBOR_TAG_PERL_OBJECT); 526 encode_tag (enc, CBOR_TAG_PERL_OBJECT);
477 encode_uint (enc, MAJOR_ARRAY, count + 1); 527 encode_uint (enc, MAJOR_ARRAY, count + 1);
478 encode_strref (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash)); 528 encode_strref (enc, 0, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
479 529
480 while (count) 530 while (count)
481 encode_sv (enc, SP[1 - count--]); 531 encode_sv (enc, SP[1 - count--]);
482 532
483 PUTBACK; 533 PUTBACK;
… …
542 592
543 if (SvPOKp (sv)) 593 if (SvPOKp (sv))
544 { 594 {
545 STRLEN len; 595 STRLEN len;
546 char *str = SvPV (sv, len); 596 char *str = SvPV (sv, len);
547 encode_strref (enc, SvUTF8 (sv), str, len); 597 encode_strref (enc, enc->cbor.flags & F_TEXT_STRINGS, SvUTF8 (sv), str, len);
548 } 598 }
549 else if (SvNOKp (sv)) 599 else if (SvNOKp (sv))
550 encode_nv (enc, sv); 600 encode_nv (enc, sv);
551 else if (SvIOKp (sv)) 601 else if (SvIOKp (sv))
552 { 602 {
… …
569} 619}
570 620
571static SV * 621static SV *
572encode_cbor (SV *scalar, CBOR *cbor) 622encode_cbor (SV *scalar, CBOR *cbor)
573{ 623{
574 enc_t enc = { }; 624 enc_t enc = { 0 };
575 625
576 enc.cbor = *cbor; 626 enc.cbor = *cbor;
577 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 627 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
578 enc.cur = SvPVX (enc.sv); 628 enc.cur = SvPVX (enc.sv);
579 enc.end = SvEND (enc.sv); 629 enc.end = SvEND (enc.sv);
… …
611 U32 depth; // recursion depth 661 U32 depth; // recursion depth
612 U32 maxdepth; // recursion depth limit 662 U32 maxdepth; // recursion depth limit
613 AV *shareable; 663 AV *shareable;
614 AV *stringref; 664 AV *stringref;
615 SV *decode_tagged; 665 SV *decode_tagged;
666 SV *err_sv; // optional sv for error, needs to be freed
616} dec_t; 667} dec_t;
617 668
618#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);
619 676
620#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;
621 680
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
622#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)
623#define DEC_DEC_DEPTH --dec->depth 712#define DEC_DEC_DEPTH --dec->depth
624 713
625static UV 714static UV
626decode_uint (dec_t *dec) 715decode_uint (dec_t *dec)
627{ 716{
… …
704 av_push (av, decode_sv (dec)); 793 av_push (av, decode_sv (dec));
705 } 794 }
706 } 795 }
707 else 796 else
708 { 797 {
709 int i, len = decode_uint (dec); 798 UV i, len = decode_uint (dec);
710 799
711 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
712 av_fill (av, len - 1); 801 av_fill (av, len - 1);
713 802
714 for (i = 0; i < len; ++i) 803 for (i = 0; i < len; ++i)
… …
729{ 818{
730 // for speed reasons, we specialcase single-string 819 // for speed reasons, we specialcase single-string
731 // byte or utf-8 strings as keys, but only when !stringref 820 // byte or utf-8 strings as keys, but only when !stringref
732 821
733 if (ecb_expect_true (!dec->stringref)) 822 if (ecb_expect_true (!dec->stringref))
734 if (ecb_expect_true ((*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8)) 823 if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8))
735 { 824 {
736 I32 len = decode_uint (dec); 825 STRLEN len = decode_uint (dec);
737 char *key = (char *)dec->cur; 826 char *key = (char *)dec->cur;
738 827
828 WANT (len);
739 dec->cur += len; 829 dec->cur += len;
740 830
741 if (ecb_expect_false (dec->stringref))
742 av_push (dec->stringref, newSVpvn (key, len));
743
744 hv_store (hv, key, len, decode_sv (dec), 0); 831 hv_store (hv, key, len, decode_sv (dec), 0);
745 832
746 return; 833 return;
747 } 834 }
748 else if (ecb_expect_true ((*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8)) 835 else if (ecb_expect_true ((U8)(*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8))
749 { 836 {
750 I32 len = decode_uint (dec); 837 STRLEN len = decode_uint (dec);
751 char *key = (char *)dec->cur; 838 char *key = (char *)dec->cur;
752 839
840 WANT (len);
753 dec->cur += len; 841 dec->cur += len;
754 842
755 if (ecb_expect_false (dec->stringref)) 843 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
756 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)");
757 846
758 hv_store (hv, key, -len, decode_sv (dec), 0); 847 hv_store (hv, key, -len, decode_sv (dec), 0);
759 848
760 return; 849 return;
761 } 850 }
762 851
763 SV *k = decode_sv (dec); 852 SV *k = decode_sv (dec);
764 SV *v = decode_sv (dec); 853 SV *v = decode_sv (dec);
765 854
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
766 hv_store_ent (hv, k, v, 0); 883 hv_store_ent (hv, k, v, 0);
767 SvREFCNT_dec (k); 884 SvREFCNT_dec (k);
885
886fail:
887 ;
768} 888}
769 889
770static SV * 890static SV *
771decode_hv (dec_t *dec) 891decode_hv (dec_t *dec)
772{ 892{
… …
791 decode_he (dec, hv); 911 decode_he (dec, hv);
792 } 912 }
793 } 913 }
794 else 914 else
795 { 915 {
796 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
797 919
798 while (pairs--) 920 while (pairs--)
799 decode_he (dec, hv); 921 decode_he (dec, hv);
800 } 922 }
801 923
… …
854 && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1)) 976 && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1))
855 av_push (dec->stringref, SvREFCNT_inc_NN (sv)); 977 av_push (dec->stringref, SvREFCNT_inc_NN (sv));
856 } 978 }
857 979
858 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
859 SvUTF8_on (sv); 986 SvUTF8_on (sv);
987 }
860 988
861 return sv; 989 return sv;
862 990
863fail: 991fail:
864 SvREFCNT_dec (sv); 992 SvREFCNT_dec (sv);
… …
883 sv = newRV_noinc (decode_sv (dec)); 1011 sv = newRV_noinc (decode_sv (dec));
884 break; 1012 break;
885 1013
886 case CBOR_TAG_STRINGREF_NAMESPACE: 1014 case CBOR_TAG_STRINGREF_NAMESPACE:
887 { 1015 {
1016 // do nmot use SAVETMPS/FREETMPS, as these will
1017 // erase mortalised caches, e.g. "shareable"
888 ENTER; SAVETMPS; 1018 ENTER;
889 1019
890 SAVESPTR (dec->stringref); 1020 SAVESPTR (dec->stringref);
891 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ()); 1021 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ());
892 1022
893 sv = decode_sv (dec); 1023 sv = decode_sv (dec);
894 1024
895 FREETMPS; LEAVE; 1025 LEAVE;
896 } 1026 }
897 break; 1027 break;
898 1028
899 case CBOR_TAG_STRINGREF: 1029 case CBOR_TAG_STRINGREF:
900 { 1030 {
… …
970 if (!method) 1100 if (!method)
971 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)");
972 1102
973 dSP; 1103 dSP;
974 1104
975 ENTER; SAVETMPS; PUSHMARK (SP); 1105 ENTER; SAVETMPS;
1106 PUSHMARK (SP);
976 EXTEND (SP, len + 1); 1107 EXTEND (SP, len + 1);
977 // 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
978 PUSHs (*av_fetch (av, 0, 1)); 1109 PUSHs (*av_fetch (av, 0, 1));
979 PUSHs (sv_cbor); 1110 PUSHs (sv_cbor);
980 1111
… …
988 SPAGAIN; 1119 SPAGAIN;
989 1120
990 if (SvTRUE (ERRSV)) 1121 if (SvTRUE (ERRSV))
991 { 1122 {
992 FREETMPS; LEAVE; 1123 FREETMPS; LEAVE;
993 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV)))); 1124 ERR_ERRSV;
994 } 1125 }
995 1126
996 SvREFCNT_dec (sv); 1127 SvREFCNT_dec (sv);
997 sv = SvREFCNT_inc (POPs); 1128 sv = SvREFCNT_inc (POPs);
998 1129
… …
1002 } 1133 }
1003 break; 1134 break;
1004 1135
1005 default: 1136 default:
1006 { 1137 {
1138 SV *tag_sv = newSVuv (tag);
1139
1007 sv = decode_sv (dec); 1140 sv = decode_sv (dec);
1008 1141
1009 dSP; 1142 dSP;
1010 ENTER; SAVETMPS; PUSHMARK (SP); 1143 ENTER; SAVETMPS;
1144 SAVESTACK_POS ();
1145 PUSHMARK (SP);
1011 EXTEND (SP, 2); 1146 EXTEND (SP, 2);
1012 PUSHs (newSVuv (tag)); 1147 PUSHs (tag_sv);
1013 PUSHs (sv); 1148 PUSHs (sv);
1014 1149
1015 PUTBACK; 1150 PUTBACK;
1016 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);
1017 SPAGAIN; 1152 SPAGAIN;
1018 1153
1019 if (SvTRUE (ERRSV)) 1154 if (SvTRUE (ERRSV))
1020 { 1155 {
1156 SvREFCNT_dec (tag_sv);
1021 FREETMPS; LEAVE; 1157 FREETMPS; LEAVE;
1022 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV)))); 1158 ERR_ERRSV;
1023 } 1159 }
1024 1160
1025 if (count) 1161 if (count)
1026 { 1162 {
1163 SvREFCNT_dec (tag_sv);
1027 SvREFCNT_dec (sv); 1164 SvREFCNT_dec (sv);
1028 sv = SvREFCNT_inc (POPs); 1165 sv = SvREFCNT_inc (POPs);
1029 } 1166 }
1030 else 1167 else
1031 { 1168 {
1032 AV *av = newAV (); 1169 AV *av = newAV ();
1033 av_push (av, newSVuv (tag)); 1170 av_push (av, tag_sv);
1034 av_push (av, sv); 1171 av_push (av, sv);
1035 1172
1036 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 1173 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
1037 ? cbor_tagged_stash 1174 ? cbor_tagged_stash
1038 : gv_stashpv ("CBOR::XS::Tagged" , 1); 1175 : gv_stashpv ("CBOR::XS::Tagged" , 1);
… …
1124 1261
1125 return newSVnv (ecb_binary64_to_double (fp)); 1262 return newSVnv (ecb_binary64_to_double (fp));
1126 } 1263 }
1127 1264
1128 // 0..19 unassigned simple 1265 // 0..19 unassigned simple
1129 // 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
1130 default: 1269 default:
1131 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)"); 1270 ERR ("corrupted CBOR data (reserved/unassigned/unexpected major 7 value)");
1132 } 1271 }
1133 1272
1134 break; 1273 break;
1135 } 1274 }
1136 1275
… …
1139} 1278}
1140 1279
1141static SV * 1280static SV *
1142decode_cbor (SV *string, CBOR *cbor, char **offset_return) 1281decode_cbor (SV *string, CBOR *cbor, char **offset_return)
1143{ 1282{
1144 dec_t dec = { }; 1283 dec_t dec = { 0 };
1145 SV *sv; 1284 SV *sv;
1146 STRLEN len; 1285 STRLEN len;
1147 char *data = SvPVbyte (string, len); 1286 char *data = SvPVbyte (string, len);
1148 1287
1149 if (len > cbor->max_size && cbor->max_size) 1288 if (len > cbor->max_size && cbor->max_size)
… …
1163 if (dec.cur != dec.end && !dec.err) 1302 if (dec.cur != dec.end && !dec.err)
1164 dec.err = "garbage after CBOR object"; 1303 dec.err = "garbage after CBOR object";
1165 1304
1166 if (dec.err) 1305 if (dec.err)
1167 { 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
1168 SvREFCNT_dec (sv); 1318 SvREFCNT_dec (sv);
1319
1320 if (dec.err_sv)
1321 sv_2mortal (dec.err_sv);
1322
1169 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);
1170 } 1324 }
1171 1325
1172 sv = sv_2mortal (sv); 1326 sv = sv_2mortal (sv);
1173 1327
1174 return sv; 1328 return sv;
1175} 1329}
1176 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
1177///////////////////////////////////////////////////////////////////////////// 1455/////////////////////////////////////////////////////////////////////////////
1178// XS interface functions 1456// XS interface functions
1179 1457
1180MODULE = CBOR::XS PACKAGE = CBOR::XS 1458MODULE = CBOR::XS PACKAGE = CBOR::XS
1181 1459
… …
1193 1471
1194 default_filter = newSVpv ("CBOR::XS::default_filter", 0); 1472 default_filter = newSVpv ("CBOR::XS::default_filter", 0);
1195 1473
1196 sv_cbor = newSVpv ("CBOR", 0); 1474 sv_cbor = newSVpv ("CBOR", 0);
1197 SvREADONLY_on (sv_cbor); 1475 SvREADONLY_on (sv_cbor);
1476
1477 assert (("STRLEN must be an unsigned type", 0 <= (STRLEN)-1));
1198} 1478}
1199 1479
1200PROTOTYPES: DISABLE 1480PROTOTYPES: DISABLE
1201 1481
1202void CLONE (...) 1482void CLONE (...)
… …
1223 shrink = F_SHRINK 1503 shrink = F_SHRINK
1224 allow_unknown = F_ALLOW_UNKNOWN 1504 allow_unknown = F_ALLOW_UNKNOWN
1225 allow_sharing = F_ALLOW_SHARING 1505 allow_sharing = F_ALLOW_SHARING
1226 allow_cycles = F_ALLOW_CYCLES 1506 allow_cycles = F_ALLOW_CYCLES
1227 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
1228 PPCODE: 1511 PPCODE:
1229{ 1512{
1230 if (enable) 1513 if (enable)
1231 self->flags |= ix; 1514 self->flags |= ix;
1232 else 1515 else
… …
1240 get_shrink = F_SHRINK 1523 get_shrink = F_SHRINK
1241 get_allow_unknown = F_ALLOW_UNKNOWN 1524 get_allow_unknown = F_ALLOW_UNKNOWN
1242 get_allow_sharing = F_ALLOW_SHARING 1525 get_allow_sharing = F_ALLOW_SHARING
1243 get_allow_cycles = F_ALLOW_CYCLES 1526 get_allow_cycles = F_ALLOW_CYCLES
1244 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
1245 PPCODE: 1531 PPCODE:
1246 XPUSHs (boolSV (self->flags & ix)); 1532 XPUSHs (boolSV (self->flags & ix));
1247 1533
1248void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) 1534void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
1249 PPCODE: 1535 PPCODE:
… …
1298 EXTEND (SP, 2); 1584 EXTEND (SP, 2);
1299 PUSHs (sv); 1585 PUSHs (sv);
1300 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); 1586 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
1301} 1587}
1302 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
1303void DESTROY (CBOR *self) 1641void DESTROY (CBOR *self)
1304 PPCODE: 1642 PPCODE:
1305 cbor_free (self); 1643 cbor_free (self);
1306 1644
1307PROTOTYPES: ENABLE 1645PROTOTYPES: ENABLE

Diff Legend

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