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 1.50 by root, Thu Feb 25 02:29:22 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
106#define F_TEXT_KEYS 0x00000040UL
107#define F_TEXT_STRINGS 0x00000080UL
105#define F_VALIDATE_UTF8 0x00000020UL 108#define F_VALIDATE_UTF8 0x00000100UL
106 109
107#define INIT_SIZE 32 // initial scalar size to be allocated 110#define INIT_SIZE 32 // initial scalar size to be allocated
108 111
109#define SB do { 112#define SB do {
110#define SE } while (0) 113#define SE } while (0)
… …
276encode_tag (enc_t *enc, UV tag) 279encode_tag (enc_t *enc, UV tag)
277{ 280{
278 encode_uint (enc, MAJOR_TAG, tag); 281 encode_uint (enc, MAJOR_TAG, tag);
279} 282}
280 283
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
281ecb_inline void 307ecb_inline void
282encode_str (enc_t *enc, int utf8, char *str, STRLEN len) 308encode_str (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len)
283{ 309{
310 if (ecb_expect_false (upgrade_utf8))
311 if (!utf8)
312 {
313 encode_str_utf8 (enc, utf8, str, len);
314 return;
315 }
316
284 encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len); 317 encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len);
285 need (enc, len); 318 need (enc, len);
286 memcpy (enc->cur, str, len); 319 memcpy (enc->cur, str, len);
287 enc->cur += len; 320 enc->cur += len;
288} 321}
289 322
290static void 323ecb_inline void
291encode_strref (enc_t *enc, int utf8, char *str, STRLEN len) 324encode_strref (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len)
292{ 325{
293 if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS)) 326 if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS))
294 { 327 {
295 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1); 328 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1);
296 329
… …
307 sv_setuv (*svp, enc->stringref_idx); 340 sv_setuv (*svp, enc->stringref_idx);
308 ++enc->stringref_idx; 341 ++enc->stringref_idx;
309 } 342 }
310 } 343 }
311 344
312 encode_str (enc, utf8, str, len); 345 encode_str (enc, upgrade_utf8, utf8, str, len);
313} 346}
314 347
315static void encode_sv (enc_t *enc, SV *sv); 348static void encode_sv (enc_t *enc, SV *sv);
316 349
317static void 350static void
… …
363 while ((he = hv_iternext (hv))) 396 while ((he = hv_iternext (hv)))
364 { 397 {
365 if (HeKLEN (he) == HEf_SVKEY) 398 if (HeKLEN (he) == HEf_SVKEY)
366 encode_sv (enc, HeSVKEY (he)); 399 encode_sv (enc, HeSVKEY (he));
367 else 400 else
368 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));
369 402
370 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));
371 } 404 }
372 405
373 if (mg) 406 if (mg)
… …
445 if (ecb_expect_false (SvOBJECT (sv))) 478 if (ecb_expect_false (SvOBJECT (sv)))
446 { 479 {
447 HV *stash = SvSTASH (sv); 480 HV *stash = SvSTASH (sv);
448 GV *method; 481 GV *method;
449 482
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))));
450 if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0))) 486 else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
451 { 487 {
452 dSP; 488 dSP;
453 489
454 ENTER; SAVETMPS; 490 ENTER; SAVETMPS;
455 PUSHMARK (SP); 491 PUSHMARK (SP);
… …
491 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv) 527 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)
492 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));
493 529
494 encode_tag (enc, CBOR_TAG_PERL_OBJECT); 530 encode_tag (enc, CBOR_TAG_PERL_OBJECT);
495 encode_uint (enc, MAJOR_ARRAY, count + 1); 531 encode_uint (enc, MAJOR_ARRAY, count + 1);
496 encode_strref (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash)); 532 encode_strref (enc, 0, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
497 533
498 while (count) 534 while (count)
499 encode_sv (enc, SP[1 - count--]); 535 encode_sv (enc, SP[1 - count--]);
500 536
501 PUTBACK; 537 PUTBACK;
… …
560 596
561 if (SvPOKp (sv)) 597 if (SvPOKp (sv))
562 { 598 {
563 STRLEN len; 599 STRLEN len;
564 char *str = SvPV (sv, len); 600 char *str = SvPV (sv, len);
565 encode_strref (enc, SvUTF8 (sv), str, len); 601 encode_strref (enc, enc->cbor.flags & F_TEXT_STRINGS, SvUTF8 (sv), str, len);
566 } 602 }
567 else if (SvNOKp (sv)) 603 else if (SvNOKp (sv))
568 encode_nv (enc, sv); 604 encode_nv (enc, sv);
569 else if (SvIOKp (sv)) 605 else if (SvIOKp (sv))
570 { 606 {
… …
629 U32 depth; // recursion depth 665 U32 depth; // recursion depth
630 U32 maxdepth; // recursion depth limit 666 U32 maxdepth; // recursion depth limit
631 AV *shareable; 667 AV *shareable;
632 AV *stringref; 668 AV *stringref;
633 SV *decode_tagged; 669 SV *decode_tagged;
670 SV *err_sv; // optional sv for error, needs to be freed
634} dec_t; 671} dec_t;
635 672
636#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);
637 680
638#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;
639 684
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
640#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)
641#define DEC_DEC_DEPTH --dec->depth 716#define DEC_DEC_DEPTH --dec->depth
642 717
643static UV 718static UV
644decode_uint (dec_t *dec) 719decode_uint (dec_t *dec)
645{ 720{
… …
722 av_push (av, decode_sv (dec)); 797 av_push (av, decode_sv (dec));
723 } 798 }
724 } 799 }
725 else 800 else
726 { 801 {
727 int i, len = decode_uint (dec); 802 UV i, len = decode_uint (dec);
728 803
729 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
730 av_fill (av, len - 1); 805 av_fill (av, len - 1);
731 806
732 for (i = 0; i < len; ++i) 807 for (i = 0; i < len; ++i)
… …
749 // byte or utf-8 strings as keys, but only when !stringref 824 // byte or utf-8 strings as keys, but only when !stringref
750 825
751 if (ecb_expect_true (!dec->stringref)) 826 if (ecb_expect_true (!dec->stringref))
752 if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8)) 827 if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8))
753 { 828 {
754 I32 len = decode_uint (dec); 829 STRLEN len = decode_uint (dec);
755 char *key = (char *)dec->cur; 830 char *key = (char *)dec->cur;
756 831
757 WANT (len); 832 WANT (len);
758 dec->cur += len; 833 dec->cur += len;
759 834
… …
761 836
762 return; 837 return;
763 } 838 }
764 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))
765 { 840 {
766 I32 len = decode_uint (dec); 841 STRLEN len = decode_uint (dec);
767 char *key = (char *)dec->cur; 842 char *key = (char *)dec->cur;
768 843
769 WANT (len); 844 WANT (len);
770 dec->cur += len; 845 dec->cur += len;
771 846
… …
778 return; 853 return;
779 } 854 }
780 855
781 SV *k = decode_sv (dec); 856 SV *k = decode_sv (dec);
782 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 }
783 886
784 hv_store_ent (hv, k, v, 0); 887 hv_store_ent (hv, k, v, 0);
785 SvREFCNT_dec (k); 888 SvREFCNT_dec (k);
786 889
787fail: 890fail:
… …
812 decode_he (dec, hv); 915 decode_he (dec, hv);
813 } 916 }
814 } 917 }
815 else 918 else
816 { 919 {
817 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
818 923
819 while (pairs--) 924 while (pairs--)
820 decode_he (dec, hv); 925 decode_he (dec, hv);
821 } 926 }
822 927
… …
910 sv = newRV_noinc (decode_sv (dec)); 1015 sv = newRV_noinc (decode_sv (dec));
911 break; 1016 break;
912 1017
913 case CBOR_TAG_STRINGREF_NAMESPACE: 1018 case CBOR_TAG_STRINGREF_NAMESPACE:
914 { 1019 {
1020 // do nmot use SAVETMPS/FREETMPS, as these will
1021 // erase mortalised caches, e.g. "shareable"
915 ENTER; SAVETMPS; 1022 ENTER;
916 1023
917 SAVESPTR (dec->stringref); 1024 SAVESPTR (dec->stringref);
918 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ()); 1025 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ());
919 1026
920 sv = decode_sv (dec); 1027 sv = decode_sv (dec);
921 1028
922 FREETMPS; LEAVE; 1029 LEAVE;
923 } 1030 }
924 break; 1031 break;
925 1032
926 case CBOR_TAG_STRINGREF: 1033 case CBOR_TAG_STRINGREF:
927 { 1034 {
… …
978 } 1085 }
979 break; 1086 break;
980 1087
981 case CBOR_TAG_PERL_OBJECT: 1088 case CBOR_TAG_PERL_OBJECT:
982 { 1089 {
1090 if (dec->cbor.flags & F_FORBID_OBJECTS)
1091 goto filter;
1092
983 sv = decode_sv (dec); 1093 sv = decode_sv (dec);
984 1094
985 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV) 1095 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
986 ERR ("corrupted CBOR data (non-array perl object)"); 1096 ERR ("corrupted CBOR data (non-array perl object)");
987 1097
… …
1016 SPAGAIN; 1126 SPAGAIN;
1017 1127
1018 if (SvTRUE (ERRSV)) 1128 if (SvTRUE (ERRSV))
1019 { 1129 {
1020 FREETMPS; LEAVE; 1130 FREETMPS; LEAVE;
1021 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV)))); 1131 ERR_ERRSV;
1022 } 1132 }
1023 1133
1024 SvREFCNT_dec (sv); 1134 SvREFCNT_dec (sv);
1025 sv = SvREFCNT_inc (POPs); 1135 sv = SvREFCNT_inc (POPs);
1026 1136
… …
1029 FREETMPS; LEAVE; 1139 FREETMPS; LEAVE;
1030 } 1140 }
1031 break; 1141 break;
1032 1142
1033 default: 1143 default:
1144 filter:
1034 { 1145 {
1146 SV *tag_sv = newSVuv (tag);
1147
1035 sv = decode_sv (dec); 1148 sv = decode_sv (dec);
1036 1149
1037 dSP; 1150 dSP;
1038 ENTER; SAVETMPS; 1151 ENTER; SAVETMPS;
1039 SAVESTACK_POS (); 1152 SAVESTACK_POS ();
1040 PUSHMARK (SP); 1153 PUSHMARK (SP);
1041 EXTEND (SP, 2); 1154 EXTEND (SP, 2);
1042 PUSHs (newSVuv (tag)); 1155 PUSHs (tag_sv);
1043 PUSHs (sv); 1156 PUSHs (sv);
1044 1157
1045 PUTBACK; 1158 PUTBACK;
1046 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);
1047 SPAGAIN; 1160 SPAGAIN;
1048 1161
1049 if (SvTRUE (ERRSV)) 1162 if (SvTRUE (ERRSV))
1050 { 1163 {
1164 SvREFCNT_dec (tag_sv);
1051 FREETMPS; LEAVE; 1165 FREETMPS; LEAVE;
1052 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV)))); 1166 ERR_ERRSV;
1053 } 1167 }
1054 1168
1055 if (count) 1169 if (count)
1056 { 1170 {
1171 SvREFCNT_dec (tag_sv);
1057 SvREFCNT_dec (sv); 1172 SvREFCNT_dec (sv);
1058 sv = SvREFCNT_inc (POPs); 1173 sv = SvREFCNT_inc (POPs);
1059 } 1174 }
1060 else 1175 else
1061 { 1176 {
1062 AV *av = newAV (); 1177 AV *av = newAV ();
1063 av_push (av, newSVuv (tag)); 1178 av_push (av, tag_sv);
1064 av_push (av, sv); 1179 av_push (av, sv);
1065 1180
1066 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 1181 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
1067 ? cbor_tagged_stash 1182 ? cbor_tagged_stash
1068 : gv_stashpv ("CBOR::XS::Tagged" , 1); 1183 : gv_stashpv ("CBOR::XS::Tagged" , 1);
… …
1197 1312
1198 if (dec.err) 1313 if (dec.err)
1199 { 1314 {
1200 if (dec.shareable) 1315 if (dec.shareable)
1201 { 1316 {
1202 // need to break cyclic links, which whould all be in shareable 1317 // need to break cyclic links, which would all be in shareable
1203 int i; 1318 int i;
1204 SV **svp; 1319 SV **svp;
1205 1320
1206 for (i = av_len (dec.shareable) + 1; i--; ) 1321 for (i = av_len (dec.shareable) + 1; i--; )
1207 if ((svp = av_fetch (dec.shareable, i, 0))) 1322 if ((svp = av_fetch (dec.shareable, i, 0)))
1208 sv_setsv (*svp, &PL_sv_undef); 1323 sv_setsv (*svp, &PL_sv_undef);
1209 } 1324 }
1210 1325
1211 SvREFCNT_dec (sv); 1326 SvREFCNT_dec (sv);
1327
1328 if (dec.err_sv)
1329 sv_2mortal (dec.err_sv);
1330
1212 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);
1213 } 1332 }
1214 1333
1215 sv = sv_2mortal (sv); 1334 sv = sv_2mortal (sv);
1216 1335
… …
1360 1479
1361 default_filter = newSVpv ("CBOR::XS::default_filter", 0); 1480 default_filter = newSVpv ("CBOR::XS::default_filter", 0);
1362 1481
1363 sv_cbor = newSVpv ("CBOR", 0); 1482 sv_cbor = newSVpv ("CBOR", 0);
1364 SvREADONLY_on (sv_cbor); 1483 SvREADONLY_on (sv_cbor);
1484
1485 assert (("STRLEN must be an unsigned type", 0 <= (STRLEN)-1));
1365} 1486}
1366 1487
1367PROTOTYPES: DISABLE 1488PROTOTYPES: DISABLE
1368 1489
1369void CLONE (...) 1490void CLONE (...)
… …
1389 ALIAS: 1510 ALIAS:
1390 shrink = F_SHRINK 1511 shrink = F_SHRINK
1391 allow_unknown = F_ALLOW_UNKNOWN 1512 allow_unknown = F_ALLOW_UNKNOWN
1392 allow_sharing = F_ALLOW_SHARING 1513 allow_sharing = F_ALLOW_SHARING
1393 allow_cycles = F_ALLOW_CYCLES 1514 allow_cycles = F_ALLOW_CYCLES
1515 forbid_objects = F_FORBID_OBJECTS
1394 pack_strings = F_PACK_STRINGS 1516 pack_strings = F_PACK_STRINGS
1517 text_keys = F_TEXT_KEYS
1518 text_strings = F_TEXT_STRINGS
1395 validate_utf8 = F_VALIDATE_UTF8 1519 validate_utf8 = F_VALIDATE_UTF8
1396 PPCODE: 1520 PPCODE:
1397{ 1521{
1398 if (enable) 1522 if (enable)
1399 self->flags |= ix; 1523 self->flags |= ix;
… …
1407 ALIAS: 1531 ALIAS:
1408 get_shrink = F_SHRINK 1532 get_shrink = F_SHRINK
1409 get_allow_unknown = F_ALLOW_UNKNOWN 1533 get_allow_unknown = F_ALLOW_UNKNOWN
1410 get_allow_sharing = F_ALLOW_SHARING 1534 get_allow_sharing = F_ALLOW_SHARING
1411 get_allow_cycles = F_ALLOW_CYCLES 1535 get_allow_cycles = F_ALLOW_CYCLES
1536 get_forbid_objects = F_FORBID_OBJECTS
1412 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
1413 get_validate_utf8 = F_VALIDATE_UTF8 1540 get_validate_utf8 = F_VALIDATE_UTF8
1414 PPCODE: 1541 PPCODE:
1415 XPUSHs (boolSV (self->flags & ix)); 1542 XPUSHs (boolSV (self->flags & ix));
1416 1543
1417void 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