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 1.61 by root, Sat Nov 26 02:10:19 2016 UTC

6#include <string.h> 6#include <string.h>
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#include <inttypes.h>
11 12
12#define ECB_NO_THREADS 1 13#define ECB_NO_THREADS 1
13#include "ecb.h" 14#include "ecb.h"
14 15
15// compatibility with perl <5.18 16// compatibility with perl <5.18
99 100
100#define F_SHRINK 0x00000001UL 101#define F_SHRINK 0x00000001UL
101#define F_ALLOW_UNKNOWN 0x00000002UL 102#define F_ALLOW_UNKNOWN 0x00000002UL
102#define F_ALLOW_SHARING 0x00000004UL 103#define F_ALLOW_SHARING 0x00000004UL
103#define F_ALLOW_CYCLES 0x00000008UL 104#define F_ALLOW_CYCLES 0x00000008UL
105#define F_FORBID_OBJECTS 0x00000010UL
104#define F_PACK_STRINGS 0x00000010UL 106#define F_PACK_STRINGS 0x00000020UL
105#define F_UTF8_STRINGS 0x00000020UL
106#define F_UTF8_KEYS 0x00000040UL 107#define F_TEXT_KEYS 0x00000040UL
108#define F_TEXT_STRINGS 0x00000080UL
107#define F_VALIDATE_UTF8 0x00000080UL 109#define F_VALIDATE_UTF8 0x00000100UL
108 110
109#define INIT_SIZE 32 // initial scalar size to be allocated 111#define INIT_SIZE 32 // initial scalar size to be allocated
110 112
111#define SB do { 113#define SB do {
112#define SE } while (0) 114#define SE } while (0)
185 187
186// minimum length of a string to be registered for stringref 188// minimum length of a string to be registered for stringref
187ecb_inline int 189ecb_inline int
188minimum_string_length (UV idx) 190minimum_string_length (UV idx)
189{ 191{
190 return idx > 23 192
191 ? idx > 0xffU 193 return idx <= 23 ? 3
192 ? idx > 0xffffU 194 : idx <= 0xffU ? 4
195 : idx <= 0xffffU ? 5
193 ? idx > 0xffffffffU 196 : idx <= 0xffffffffU ? 7
194 ? 11 197 : 11;
195 : 7
196 : 5
197 : 4
198 : 3;
199} 198}
200 199
201///////////////////////////////////////////////////////////////////////////// 200/////////////////////////////////////////////////////////////////////////////
202// encoder 201// encoder
203 202
216} enc_t; 215} enc_t;
217 216
218ecb_inline void 217ecb_inline void
219need (enc_t *enc, STRLEN len) 218need (enc_t *enc, STRLEN len)
220{ 219{
221 if (ecb_expect_false (enc->cur + len >= enc->end)) 220 if (ecb_expect_false ((uintptr_t)(enc->end - enc->cur) < len))
222 { 221 {
223 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv); 222 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
224 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1); 223 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
225 enc->cur = SvPVX (enc->sv) + cur; 224 enc->cur = SvPVX (enc->sv) + cur;
226 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; 225 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
317 need (enc, len); 316 need (enc, len);
318 memcpy (enc->cur, str, len); 317 memcpy (enc->cur, str, len);
319 enc->cur += len; 318 enc->cur += len;
320} 319}
321 320
322ecb_inline 321ecb_inline void
323encode_strref (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len) 322encode_strref (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len)
324{ 323{
325 if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS)) 324 if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS))
326 { 325 {
327 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1); 326 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1);
395 while ((he = hv_iternext (hv))) 394 while ((he = hv_iternext (hv)))
396 { 395 {
397 if (HeKLEN (he) == HEf_SVKEY) 396 if (HeKLEN (he) == HEf_SVKEY)
398 encode_sv (enc, HeSVKEY (he)); 397 encode_sv (enc, HeSVKEY (he));
399 else 398 else
400 encode_strref (enc, enc->cbor.flags & (F_UTF8_KEYS | F_UTF8_STRINGS), HeKUTF8 (he), HeKEY (he), HeKLEN (he)); 399 encode_strref (enc, enc->cbor.flags & (F_TEXT_KEYS | F_TEXT_STRINGS), HeKUTF8 (he), HeKEY (he), HeKLEN (he));
401 400
402 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he)); 401 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he));
403 } 402 }
404 403
405 if (mg) 404 if (mg)
477 if (ecb_expect_false (SvOBJECT (sv))) 476 if (ecb_expect_false (SvOBJECT (sv)))
478 { 477 {
479 HV *stash = SvSTASH (sv); 478 HV *stash = SvSTASH (sv);
480 GV *method; 479 GV *method;
481 480
481 if (enc->cbor.flags & F_FORBID_OBJECTS)
482 croak ("encountered object '%s', but forbid_objects is enabled",
483 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
482 if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0))) 484 else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
483 { 485 {
484 dSP; 486 dSP;
485 487
486 ENTER; SAVETMPS; 488 ENTER; SAVETMPS;
487 PUSHMARK (SP); 489 PUSHMARK (SP);
592 594
593 if (SvPOKp (sv)) 595 if (SvPOKp (sv))
594 { 596 {
595 STRLEN len; 597 STRLEN len;
596 char *str = SvPV (sv, len); 598 char *str = SvPV (sv, len);
597 encode_strref (enc, enc->cbor.flags & F_UTF8_STRINGS, SvUTF8 (sv), str, len); 599 encode_strref (enc, enc->cbor.flags & F_TEXT_STRINGS, SvUTF8 (sv), str, len);
598 } 600 }
599 else if (SvNOKp (sv)) 601 else if (SvNOKp (sv))
600 encode_nv (enc, sv); 602 encode_nv (enc, sv);
601 else if (SvIOKp (sv)) 603 else if (SvIOKp (sv))
602 { 604 {
661 U32 depth; // recursion depth 663 U32 depth; // recursion depth
662 U32 maxdepth; // recursion depth limit 664 U32 maxdepth; // recursion depth limit
663 AV *shareable; 665 AV *shareable;
664 AV *stringref; 666 AV *stringref;
665 SV *decode_tagged; 667 SV *decode_tagged;
668 SV *err_sv; // optional sv for error, needs to be freed
666} dec_t; 669} dec_t;
667 670
668#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE 671// set dec->err to ERRSV
672ecb_cold static void
673err_errsv (dec_t *dec)
674{
675 if (!dec->err)
676 {
677 dec->err_sv = newSVsv (ERRSV);
669 678
670#define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data") 679 // chop off the trailing \n
680 SvCUR_set (dec->err_sv, SvCUR (dec->err_sv) - 1);
681 *SvEND (dec->err_sv) = 0;
671 682
683 dec->err = SvPVutf8_nolen (dec->err_sv);
684 }
685}
686
687// the following functions are used to reduce code size and help the compiler to optimise
688ecb_cold static void
689err_set (dec_t *dec, const char *reason)
690{
691 if (!dec->err)
692 dec->err = reason;
693}
694
695ecb_cold static void
696err_unexpected_end (dec_t *dec)
697{
698 err_set (dec, "unexpected end of CBOR data");
699}
700
701ecb_cold static void
702err_nesting_exceeded (dec_t *dec)
703{
704 err_set (dec, ERR_NESTING_EXCEEDED);
705}
706
707#define ERR_DO(do) SB do; goto fail; SE
708#define ERR(reason) ERR_DO (err_set (dec, reason))
709#define ERR_ERRSV ERR_DO (err_errsv (dec))
710
711#define WANT(len) if (ecb_expect_false ((uintptr_t)(dec->end - dec->cur) < (STRLEN)len)) ERR_DO (err_unexpected_end (dec))
712
672#define DEC_INC_DEPTH if (++dec->depth > dec->cbor.max_depth) ERR (ERR_NESTING_EXCEEDED) 713#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 714#define DEC_DEC_DEPTH --dec->depth
674 715
675static UV 716static UV
676decode_uint (dec_t *dec) 717decode_uint (dec_t *dec)
677{ 718{
754 av_push (av, decode_sv (dec)); 795 av_push (av, decode_sv (dec));
755 } 796 }
756 } 797 }
757 else 798 else
758 { 799 {
759 int i, len = decode_uint (dec); 800 UV i, len = decode_uint (dec);
760 801
761 WANT (len); // complexity check for av_fill - need at least one byte per value, do not allow supersize arrays 802 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); 803 av_fill (av, len - 1);
763 804
764 for (i = 0; i < len; ++i) 805 for (i = 0; i < len; ++i)
781 // byte or utf-8 strings as keys, but only when !stringref 822 // byte or utf-8 strings as keys, but only when !stringref
782 823
783 if (ecb_expect_true (!dec->stringref)) 824 if (ecb_expect_true (!dec->stringref))
784 if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8)) 825 if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8))
785 { 826 {
786 I32 len = decode_uint (dec); 827 STRLEN len = decode_uint (dec);
787 char *key = (char *)dec->cur; 828 char *key = (char *)dec->cur;
788 829
789 WANT (len); 830 WANT (len);
790 dec->cur += len; 831 dec->cur += len;
791 832
793 834
794 return; 835 return;
795 } 836 }
796 else if (ecb_expect_true ((U8)(*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8)) 837 else if (ecb_expect_true ((U8)(*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8))
797 { 838 {
798 I32 len = decode_uint (dec); 839 STRLEN len = decode_uint (dec);
799 char *key = (char *)dec->cur; 840 char *key = (char *)dec->cur;
800 841
801 WANT (len); 842 WANT (len);
802 dec->cur += len; 843 dec->cur += len;
803 844
810 return; 851 return;
811 } 852 }
812 853
813 SV *k = decode_sv (dec); 854 SV *k = decode_sv (dec);
814 SV *v = decode_sv (dec); 855 SV *v = decode_sv (dec);
856
857 // we leak memory if uncaught exceptions are thrown by random magical
858 // methods, and this is hopefully the only place where it can happen,
859 // so if there is a chance of an exception, take the very slow path.
860 // since catching exceptions is "undocumented/internal/forbidden" by
861 // the new p5p powers, we need to call out to a perl function :/
862 if (ecb_expect_false (SvAMAGIC (k)))
863 {
864 dSP;
865
866 ENTER; SAVETMPS;
867 PUSHMARK (SP);
868 EXTEND (SP, 3);
869 PUSHs (sv_2mortal (newRV_inc ((SV *)hv)));
870 PUSHs (sv_2mortal (k));
871 PUSHs (sv_2mortal (v));
872
873 PUTBACK;
874 call_pv ("CBOR::XS::_hv_store", G_VOID | G_DISCARD | G_EVAL);
875 SPAGAIN;
876
877 FREETMPS; LEAVE;
878
879 if (SvTRUE (ERRSV))
880 ERR_ERRSV;
881
882 return;
883 }
815 884
816 hv_store_ent (hv, k, v, 0); 885 hv_store_ent (hv, k, v, 0);
817 SvREFCNT_dec (k); 886 SvREFCNT_dec (k);
818 887
819fail: 888fail:
844 decode_he (dec, hv); 913 decode_he (dec, hv);
845 } 914 }
846 } 915 }
847 else 916 else
848 { 917 {
849 int pairs = decode_uint (dec); 918 UV pairs = decode_uint (dec);
919
920 WANT (pairs); // complexity check - need at least one byte per value, do not allow supersize hashes
850 921
851 while (pairs--) 922 while (pairs--)
852 decode_he (dec, hv); 923 decode_he (dec, hv);
853 } 924 }
854 925
942 sv = newRV_noinc (decode_sv (dec)); 1013 sv = newRV_noinc (decode_sv (dec));
943 break; 1014 break;
944 1015
945 case CBOR_TAG_STRINGREF_NAMESPACE: 1016 case CBOR_TAG_STRINGREF_NAMESPACE:
946 { 1017 {
1018 // do nmot use SAVETMPS/FREETMPS, as these will
1019 // erase mortalised caches, e.g. "shareable"
947 ENTER; SAVETMPS; 1020 ENTER;
948 1021
949 SAVESPTR (dec->stringref); 1022 SAVESPTR (dec->stringref);
950 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ()); 1023 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ());
951 1024
952 sv = decode_sv (dec); 1025 sv = decode_sv (dec);
953 1026
954 FREETMPS; LEAVE; 1027 LEAVE;
955 } 1028 }
956 break; 1029 break;
957 1030
958 case CBOR_TAG_STRINGREF: 1031 case CBOR_TAG_STRINGREF:
959 { 1032 {
1010 } 1083 }
1011 break; 1084 break;
1012 1085
1013 case CBOR_TAG_PERL_OBJECT: 1086 case CBOR_TAG_PERL_OBJECT:
1014 { 1087 {
1088 if (dec->cbor.flags & F_FORBID_OBJECTS)
1089 goto filter;
1090
1015 sv = decode_sv (dec); 1091 sv = decode_sv (dec);
1016 1092
1017 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV) 1093 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
1018 ERR ("corrupted CBOR data (non-array perl object)"); 1094 ERR ("corrupted CBOR data (non-array perl object)");
1019 1095
1048 SPAGAIN; 1124 SPAGAIN;
1049 1125
1050 if (SvTRUE (ERRSV)) 1126 if (SvTRUE (ERRSV))
1051 { 1127 {
1052 FREETMPS; LEAVE; 1128 FREETMPS; LEAVE;
1053 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV)))); 1129 ERR_ERRSV;
1054 } 1130 }
1055 1131
1056 SvREFCNT_dec (sv); 1132 SvREFCNT_dec (sv);
1057 sv = SvREFCNT_inc (POPs); 1133 sv = SvREFCNT_inc (POPs);
1058 1134
1061 FREETMPS; LEAVE; 1137 FREETMPS; LEAVE;
1062 } 1138 }
1063 break; 1139 break;
1064 1140
1065 default: 1141 default:
1142 filter:
1066 { 1143 {
1144 SV *tag_sv = newSVuv (tag);
1145
1067 sv = decode_sv (dec); 1146 sv = decode_sv (dec);
1068 1147
1069 dSP; 1148 dSP;
1070 ENTER; SAVETMPS; 1149 ENTER; SAVETMPS;
1071 SAVESTACK_POS (); 1150 SAVESTACK_POS ();
1072 PUSHMARK (SP); 1151 PUSHMARK (SP);
1073 EXTEND (SP, 2); 1152 EXTEND (SP, 2);
1074 PUSHs (newSVuv (tag)); 1153 PUSHs (tag_sv);
1075 PUSHs (sv); 1154 PUSHs (sv);
1076 1155
1077 PUTBACK; 1156 PUTBACK;
1078 int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL); 1157 int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL);
1079 SPAGAIN; 1158 SPAGAIN;
1080 1159
1081 if (SvTRUE (ERRSV)) 1160 if (SvTRUE (ERRSV))
1082 { 1161 {
1162 SvREFCNT_dec (tag_sv);
1083 FREETMPS; LEAVE; 1163 FREETMPS; LEAVE;
1084 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV)))); 1164 ERR_ERRSV;
1085 } 1165 }
1086 1166
1087 if (count) 1167 if (count)
1088 { 1168 {
1169 SvREFCNT_dec (tag_sv);
1089 SvREFCNT_dec (sv); 1170 SvREFCNT_dec (sv);
1090 sv = SvREFCNT_inc (POPs); 1171 sv = SvREFCNT_inc (POPs);
1091 } 1172 }
1092 else 1173 else
1093 { 1174 {
1094 AV *av = newAV (); 1175 AV *av = newAV ();
1095 av_push (av, newSVuv (tag)); 1176 av_push (av, tag_sv);
1096 av_push (av, sv); 1177 av_push (av, sv);
1097 1178
1098 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 1179 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
1099 ? cbor_tagged_stash 1180 ? cbor_tagged_stash
1100 : gv_stashpv ("CBOR::XS::Tagged" , 1); 1181 : gv_stashpv ("CBOR::XS::Tagged" , 1);
1229 1310
1230 if (dec.err) 1311 if (dec.err)
1231 { 1312 {
1232 if (dec.shareable) 1313 if (dec.shareable)
1233 { 1314 {
1234 // need to break cyclic links, which whould all be in shareable 1315 // need to break cyclic links, which would all be in shareable
1235 int i; 1316 int i;
1236 SV **svp; 1317 SV **svp;
1237 1318
1238 for (i = av_len (dec.shareable) + 1; i--; ) 1319 for (i = av_len (dec.shareable) + 1; i--; )
1239 if ((svp = av_fetch (dec.shareable, i, 0))) 1320 if ((svp = av_fetch (dec.shareable, i, 0)))
1240 sv_setsv (*svp, &PL_sv_undef); 1321 sv_setsv (*svp, &PL_sv_undef);
1241 } 1322 }
1242 1323
1243 SvREFCNT_dec (sv); 1324 SvREFCNT_dec (sv);
1325
1326 if (dec.err_sv)
1327 sv_2mortal (dec.err_sv);
1328
1244 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur); 1329 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur);
1245 } 1330 }
1246 1331
1247 sv = sv_2mortal (sv); 1332 sv = sv_2mortal (sv);
1248 1333
1392 1477
1393 default_filter = newSVpv ("CBOR::XS::default_filter", 0); 1478 default_filter = newSVpv ("CBOR::XS::default_filter", 0);
1394 1479
1395 sv_cbor = newSVpv ("CBOR", 0); 1480 sv_cbor = newSVpv ("CBOR", 0);
1396 SvREADONLY_on (sv_cbor); 1481 SvREADONLY_on (sv_cbor);
1482
1483 assert (("STRLEN must be an unsigned type", 0 <= (STRLEN)-1));
1397} 1484}
1398 1485
1399PROTOTYPES: DISABLE 1486PROTOTYPES: DISABLE
1400 1487
1401void CLONE (...) 1488void CLONE (...)
1421 ALIAS: 1508 ALIAS:
1422 shrink = F_SHRINK 1509 shrink = F_SHRINK
1423 allow_unknown = F_ALLOW_UNKNOWN 1510 allow_unknown = F_ALLOW_UNKNOWN
1424 allow_sharing = F_ALLOW_SHARING 1511 allow_sharing = F_ALLOW_SHARING
1425 allow_cycles = F_ALLOW_CYCLES 1512 allow_cycles = F_ALLOW_CYCLES
1513 forbid_objects = F_FORBID_OBJECTS
1426 pack_strings = F_PACK_STRINGS 1514 pack_strings = F_PACK_STRINGS
1515 text_keys = F_TEXT_KEYS
1427 utf8_strings = F_UTF8_STRINGS 1516 text_strings = F_TEXT_STRINGS
1428 validate_utf8 = F_VALIDATE_UTF8 1517 validate_utf8 = F_VALIDATE_UTF8
1429 PPCODE: 1518 PPCODE:
1430{ 1519{
1431 if (enable) 1520 if (enable)
1432 self->flags |= ix; 1521 self->flags |= ix;
1440 ALIAS: 1529 ALIAS:
1441 get_shrink = F_SHRINK 1530 get_shrink = F_SHRINK
1442 get_allow_unknown = F_ALLOW_UNKNOWN 1531 get_allow_unknown = F_ALLOW_UNKNOWN
1443 get_allow_sharing = F_ALLOW_SHARING 1532 get_allow_sharing = F_ALLOW_SHARING
1444 get_allow_cycles = F_ALLOW_CYCLES 1533 get_allow_cycles = F_ALLOW_CYCLES
1534 get_forbid_objects = F_FORBID_OBJECTS
1445 get_pack_strings = F_PACK_STRINGS 1535 get_pack_strings = F_PACK_STRINGS
1536 get_text_keys = F_TEXT_KEYS
1537 get_text_strings = F_TEXT_STRINGS
1446 get_validate_utf8 = F_VALIDATE_UTF8 1538 get_validate_utf8 = F_VALIDATE_UTF8
1447 PPCODE: 1539 PPCODE:
1448 XPUSHs (boolSV (self->flags & ix)); 1540 XPUSHs (boolSV (self->flags & ix));
1449 1541
1450void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) 1542void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines