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 1.53 by root, Sun Apr 24 19:51:41 2016 UTC vs.
Revision 1.60 by root, Sat Nov 26 00:47:02 2016 UTC

99 99
100#define F_SHRINK 0x00000001UL 100#define F_SHRINK 0x00000001UL
101#define F_ALLOW_UNKNOWN 0x00000002UL 101#define F_ALLOW_UNKNOWN 0x00000002UL
102#define F_ALLOW_SHARING 0x00000004UL 102#define F_ALLOW_SHARING 0x00000004UL
103#define F_ALLOW_CYCLES 0x00000008UL 103#define F_ALLOW_CYCLES 0x00000008UL
104#define F_FORBID_OBJECTS 0x00000010UL
104#define F_PACK_STRINGS 0x00000010UL 105#define F_PACK_STRINGS 0x00000020UL
105#define F_UTF8_STRINGS 0x00000020UL
106#define F_UTF8_KEYS 0x00000040UL 106#define F_TEXT_KEYS 0x00000040UL
107#define F_TEXT_STRINGS 0x00000080UL
107#define F_VALIDATE_UTF8 0x00000080UL 108#define F_VALIDATE_UTF8 0x00000100UL
108 109
109#define INIT_SIZE 32 // initial scalar size to be allocated 110#define INIT_SIZE 32 // initial scalar size to be allocated
110 111
111#define SB do { 112#define SB do {
112#define SE } while (0) 113#define SE } while (0)
317 need (enc, len); 318 need (enc, len);
318 memcpy (enc->cur, str, len); 319 memcpy (enc->cur, str, len);
319 enc->cur += len; 320 enc->cur += len;
320} 321}
321 322
322ecb_inline 323ecb_inline void
323encode_strref (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len) 324encode_strref (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len)
324{ 325{
325 if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS)) 326 if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS))
326 { 327 {
327 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1); 328 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1);
395 while ((he = hv_iternext (hv))) 396 while ((he = hv_iternext (hv)))
396 { 397 {
397 if (HeKLEN (he) == HEf_SVKEY) 398 if (HeKLEN (he) == HEf_SVKEY)
398 encode_sv (enc, HeSVKEY (he)); 399 encode_sv (enc, HeSVKEY (he));
399 else 400 else
400 encode_strref (enc, enc->cbor.flags & (F_UTF8_KEYS | F_UTF8_STRINGS), 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));
401 402
402 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));
403 } 404 }
404 405
405 if (mg) 406 if (mg)
477 if (ecb_expect_false (SvOBJECT (sv))) 478 if (ecb_expect_false (SvOBJECT (sv)))
478 { 479 {
479 HV *stash = SvSTASH (sv); 480 HV *stash = SvSTASH (sv);
480 GV *method; 481 GV *method;
481 482
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))));
482 if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0))) 486 else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
483 { 487 {
484 dSP; 488 dSP;
485 489
486 ENTER; SAVETMPS; 490 ENTER; SAVETMPS;
487 PUSHMARK (SP); 491 PUSHMARK (SP);
592 596
593 if (SvPOKp (sv)) 597 if (SvPOKp (sv))
594 { 598 {
595 STRLEN len; 599 STRLEN len;
596 char *str = SvPV (sv, len); 600 char *str = SvPV (sv, len);
597 encode_strref (enc, enc->cbor.flags & F_UTF8_STRINGS, SvUTF8 (sv), str, len); 601 encode_strref (enc, enc->cbor.flags & F_TEXT_STRINGS, SvUTF8 (sv), str, len);
598 } 602 }
599 else if (SvNOKp (sv)) 603 else if (SvNOKp (sv))
600 encode_nv (enc, sv); 604 encode_nv (enc, sv);
601 else if (SvIOKp (sv)) 605 else if (SvIOKp (sv))
602 { 606 {
661 U32 depth; // recursion depth 665 U32 depth; // recursion depth
662 U32 maxdepth; // recursion depth limit 666 U32 maxdepth; // recursion depth limit
663 AV *shareable; 667 AV *shareable;
664 AV *stringref; 668 AV *stringref;
665 SV *decode_tagged; 669 SV *decode_tagged;
670 SV *err_sv; // optional sv for error, needs to be freed
666} dec_t; 671} dec_t;
667 672
668#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);
669 680
670#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;
671 684
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
672#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)
673#define DEC_DEC_DEPTH --dec->depth 716#define DEC_DEC_DEPTH --dec->depth
674 717
675static UV 718static UV
676decode_uint (dec_t *dec) 719decode_uint (dec_t *dec)
677{ 720{
754 av_push (av, decode_sv (dec)); 797 av_push (av, decode_sv (dec));
755 } 798 }
756 } 799 }
757 else 800 else
758 { 801 {
759 int i, len = decode_uint (dec); 802 UV i, len = decode_uint (dec);
760 803
761 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
762 av_fill (av, len - 1); 805 av_fill (av, len - 1);
763 806
764 for (i = 0; i < len; ++i) 807 for (i = 0; i < len; ++i)
781 // byte or utf-8 strings as keys, but only when !stringref 824 // byte or utf-8 strings as keys, but only when !stringref
782 825
783 if (ecb_expect_true (!dec->stringref)) 826 if (ecb_expect_true (!dec->stringref))
784 if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8)) 827 if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8))
785 { 828 {
786 I32 len = decode_uint (dec); 829 STRLEN len = decode_uint (dec);
787 char *key = (char *)dec->cur; 830 char *key = (char *)dec->cur;
788 831
789 WANT (len); 832 WANT (len);
790 dec->cur += len; 833 dec->cur += len;
791 834
793 836
794 return; 837 return;
795 } 838 }
796 else if (ecb_expect_true ((U8)(*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8)) 839 else if (ecb_expect_true ((U8)(*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8))
797 { 840 {
798 I32 len = decode_uint (dec); 841 STRLEN len = decode_uint (dec);
799 char *key = (char *)dec->cur; 842 char *key = (char *)dec->cur;
800 843
801 WANT (len); 844 WANT (len);
802 dec->cur += len; 845 dec->cur += len;
803 846
810 return; 853 return;
811 } 854 }
812 855
813 SV *k = decode_sv (dec); 856 SV *k = decode_sv (dec);
814 SV *v = decode_sv (dec); 857 SV *v = decode_sv (dec);
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 }
815 886
816 hv_store_ent (hv, k, v, 0); 887 hv_store_ent (hv, k, v, 0);
817 SvREFCNT_dec (k); 888 SvREFCNT_dec (k);
818 889
819fail: 890fail:
844 decode_he (dec, hv); 915 decode_he (dec, hv);
845 } 916 }
846 } 917 }
847 else 918 else
848 { 919 {
849 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
850 923
851 while (pairs--) 924 while (pairs--)
852 decode_he (dec, hv); 925 decode_he (dec, hv);
853 } 926 }
854 927
942 sv = newRV_noinc (decode_sv (dec)); 1015 sv = newRV_noinc (decode_sv (dec));
943 break; 1016 break;
944 1017
945 case CBOR_TAG_STRINGREF_NAMESPACE: 1018 case CBOR_TAG_STRINGREF_NAMESPACE:
946 { 1019 {
1020 // do nmot use SAVETMPS/FREETMPS, as these will
1021 // erase mortalised caches, e.g. "shareable"
947 ENTER; SAVETMPS; 1022 ENTER;
948 1023
949 SAVESPTR (dec->stringref); 1024 SAVESPTR (dec->stringref);
950 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ()); 1025 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ());
951 1026
952 sv = decode_sv (dec); 1027 sv = decode_sv (dec);
953 1028
954 FREETMPS; LEAVE; 1029 LEAVE;
955 } 1030 }
956 break; 1031 break;
957 1032
958 case CBOR_TAG_STRINGREF: 1033 case CBOR_TAG_STRINGREF:
959 { 1034 {
1010 } 1085 }
1011 break; 1086 break;
1012 1087
1013 case CBOR_TAG_PERL_OBJECT: 1088 case CBOR_TAG_PERL_OBJECT:
1014 { 1089 {
1090 if (dec->cbor.flags & F_FORBID_OBJECTS)
1091 goto filter;
1092
1015 sv = decode_sv (dec); 1093 sv = decode_sv (dec);
1016 1094
1017 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV) 1095 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
1018 ERR ("corrupted CBOR data (non-array perl object)"); 1096 ERR ("corrupted CBOR data (non-array perl object)");
1019 1097
1048 SPAGAIN; 1126 SPAGAIN;
1049 1127
1050 if (SvTRUE (ERRSV)) 1128 if (SvTRUE (ERRSV))
1051 { 1129 {
1052 FREETMPS; LEAVE; 1130 FREETMPS; LEAVE;
1053 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV)))); 1131 ERR_ERRSV;
1054 } 1132 }
1055 1133
1056 SvREFCNT_dec (sv); 1134 SvREFCNT_dec (sv);
1057 sv = SvREFCNT_inc (POPs); 1135 sv = SvREFCNT_inc (POPs);
1058 1136
1061 FREETMPS; LEAVE; 1139 FREETMPS; LEAVE;
1062 } 1140 }
1063 break; 1141 break;
1064 1142
1065 default: 1143 default:
1144 filter:
1066 { 1145 {
1146 SV *tag_sv = newSVuv (tag);
1147
1067 sv = decode_sv (dec); 1148 sv = decode_sv (dec);
1068 1149
1069 dSP; 1150 dSP;
1070 ENTER; SAVETMPS; 1151 ENTER; SAVETMPS;
1071 SAVESTACK_POS (); 1152 SAVESTACK_POS ();
1072 PUSHMARK (SP); 1153 PUSHMARK (SP);
1073 EXTEND (SP, 2); 1154 EXTEND (SP, 2);
1074 PUSHs (newSVuv (tag)); 1155 PUSHs (tag_sv);
1075 PUSHs (sv); 1156 PUSHs (sv);
1076 1157
1077 PUTBACK; 1158 PUTBACK;
1078 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);
1079 SPAGAIN; 1160 SPAGAIN;
1080 1161
1081 if (SvTRUE (ERRSV)) 1162 if (SvTRUE (ERRSV))
1082 { 1163 {
1164 SvREFCNT_dec (tag_sv);
1083 FREETMPS; LEAVE; 1165 FREETMPS; LEAVE;
1084 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV)))); 1166 ERR_ERRSV;
1085 } 1167 }
1086 1168
1087 if (count) 1169 if (count)
1088 { 1170 {
1171 SvREFCNT_dec (tag_sv);
1089 SvREFCNT_dec (sv); 1172 SvREFCNT_dec (sv);
1090 sv = SvREFCNT_inc (POPs); 1173 sv = SvREFCNT_inc (POPs);
1091 } 1174 }
1092 else 1175 else
1093 { 1176 {
1094 AV *av = newAV (); 1177 AV *av = newAV ();
1095 av_push (av, newSVuv (tag)); 1178 av_push (av, tag_sv);
1096 av_push (av, sv); 1179 av_push (av, sv);
1097 1180
1098 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 1181 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
1099 ? cbor_tagged_stash 1182 ? cbor_tagged_stash
1100 : gv_stashpv ("CBOR::XS::Tagged" , 1); 1183 : gv_stashpv ("CBOR::XS::Tagged" , 1);
1229 1312
1230 if (dec.err) 1313 if (dec.err)
1231 { 1314 {
1232 if (dec.shareable) 1315 if (dec.shareable)
1233 { 1316 {
1234 // need to break cyclic links, which whould all be in shareable 1317 // need to break cyclic links, which would all be in shareable
1235 int i; 1318 int i;
1236 SV **svp; 1319 SV **svp;
1237 1320
1238 for (i = av_len (dec.shareable) + 1; i--; ) 1321 for (i = av_len (dec.shareable) + 1; i--; )
1239 if ((svp = av_fetch (dec.shareable, i, 0))) 1322 if ((svp = av_fetch (dec.shareable, i, 0)))
1240 sv_setsv (*svp, &PL_sv_undef); 1323 sv_setsv (*svp, &PL_sv_undef);
1241 } 1324 }
1242 1325
1243 SvREFCNT_dec (sv); 1326 SvREFCNT_dec (sv);
1327
1328 if (dec.err_sv)
1329 sv_2mortal (dec.err_sv);
1330
1244 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);
1245 } 1332 }
1246 1333
1247 sv = sv_2mortal (sv); 1334 sv = sv_2mortal (sv);
1248 1335
1392 1479
1393 default_filter = newSVpv ("CBOR::XS::default_filter", 0); 1480 default_filter = newSVpv ("CBOR::XS::default_filter", 0);
1394 1481
1395 sv_cbor = newSVpv ("CBOR", 0); 1482 sv_cbor = newSVpv ("CBOR", 0);
1396 SvREADONLY_on (sv_cbor); 1483 SvREADONLY_on (sv_cbor);
1484
1485 assert (("STRLEN must be an unsigned type", 0 <= (STRLEN)-1));
1397} 1486}
1398 1487
1399PROTOTYPES: DISABLE 1488PROTOTYPES: DISABLE
1400 1489
1401void CLONE (...) 1490void CLONE (...)
1421 ALIAS: 1510 ALIAS:
1422 shrink = F_SHRINK 1511 shrink = F_SHRINK
1423 allow_unknown = F_ALLOW_UNKNOWN 1512 allow_unknown = F_ALLOW_UNKNOWN
1424 allow_sharing = F_ALLOW_SHARING 1513 allow_sharing = F_ALLOW_SHARING
1425 allow_cycles = F_ALLOW_CYCLES 1514 allow_cycles = F_ALLOW_CYCLES
1515 forbid_objects = F_FORBID_OBJECTS
1426 pack_strings = F_PACK_STRINGS 1516 pack_strings = F_PACK_STRINGS
1517 text_keys = F_TEXT_KEYS
1427 utf8_strings = F_UTF8_STRINGS 1518 text_strings = F_TEXT_STRINGS
1428 validate_utf8 = F_VALIDATE_UTF8 1519 validate_utf8 = F_VALIDATE_UTF8
1429 PPCODE: 1520 PPCODE:
1430{ 1521{
1431 if (enable) 1522 if (enable)
1432 self->flags |= ix; 1523 self->flags |= ix;
1440 ALIAS: 1531 ALIAS:
1441 get_shrink = F_SHRINK 1532 get_shrink = F_SHRINK
1442 get_allow_unknown = F_ALLOW_UNKNOWN 1533 get_allow_unknown = F_ALLOW_UNKNOWN
1443 get_allow_sharing = F_ALLOW_SHARING 1534 get_allow_sharing = F_ALLOW_SHARING
1444 get_allow_cycles = F_ALLOW_CYCLES 1535 get_allow_cycles = F_ALLOW_CYCLES
1536 get_forbid_objects = F_FORBID_OBJECTS
1445 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
1446 get_validate_utf8 = F_VALIDATE_UTF8 1540 get_validate_utf8 = F_VALIDATE_UTF8
1447 PPCODE: 1541 PPCODE:
1448 XPUSHs (boolSV (self->flags & ix)); 1542 XPUSHs (boolSV (self->flags & ix));
1449 1543
1450void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) 1544void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines