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 1.39 by root, Sun Dec 1 14:45:03 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
104#define F_VALIDATE_UTF8 0x00000020UL 107#define F_VALIDATE_UTF8 0x00000080UL
105 108
106#define INIT_SIZE 32 // initial scalar size to be allocated 109#define INIT_SIZE 32 // initial scalar size to be allocated
107 110
108#define SB do { 111#define SB do {
109#define SE } while (0) 112#define SE } while (0)
… …
128typedef struct { 131typedef struct {
129 U32 flags; 132 U32 flags;
130 U32 max_depth; 133 U32 max_depth;
131 STRLEN max_size; 134 STRLEN max_size;
132 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.
133} CBOR; 141} CBOR;
134 142
135ecb_inline void 143ecb_inline void
136cbor_init (CBOR *cbor) 144cbor_init (CBOR *cbor)
137{ 145{
… …
141 149
142ecb_inline void 150ecb_inline void
143cbor_free (CBOR *cbor) 151cbor_free (CBOR *cbor)
144{ 152{
145 SvREFCNT_dec (cbor->filter); 153 SvREFCNT_dec (cbor->filter);
154 SvREFCNT_dec (cbor->incr_count);
146} 155}
147 156
148///////////////////////////////////////////////////////////////////////////// 157/////////////////////////////////////////////////////////////////////////////
149// utility functions 158// utility functions
150 159
… …
269encode_tag (enc_t *enc, UV tag) 278encode_tag (enc_t *enc, UV tag)
270{ 279{
271 encode_uint (enc, MAJOR_TAG, tag); 280 encode_uint (enc, MAJOR_TAG, tag);
272} 281}
273 282
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
274ecb_inline void 306ecb_inline void
275encode_str (enc_t *enc, int utf8, char *str, STRLEN len) 307encode_str (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len)
276{ 308{
309 if (ecb_expect_false (upgrade_utf8))
310 if (!utf8)
311 {
312 encode_str_utf8 (enc, utf8, str, len);
313 return;
314 }
315
277 encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len); 316 encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len);
278 need (enc, len); 317 need (enc, len);
279 memcpy (enc->cur, str, len); 318 memcpy (enc->cur, str, len);
280 enc->cur += len; 319 enc->cur += len;
281} 320}
282 321
283static void 322ecb_inline void
284encode_strref (enc_t *enc, int utf8, char *str, STRLEN len) 323encode_strref (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len)
285{ 324{
286 if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS)) 325 if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS))
287 { 326 {
288 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1); 327 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1);
289 328
… …
300 sv_setuv (*svp, enc->stringref_idx); 339 sv_setuv (*svp, enc->stringref_idx);
301 ++enc->stringref_idx; 340 ++enc->stringref_idx;
302 } 341 }
303 } 342 }
304 343
305 encode_str (enc, utf8, str, len); 344 encode_str (enc, upgrade_utf8, utf8, str, len);
306} 345}
307 346
308static void encode_sv (enc_t *enc, SV *sv); 347static void encode_sv (enc_t *enc, SV *sv);
309 348
310static void 349static void
… …
317 356
318 ++enc->depth; 357 ++enc->depth;
319 358
320 encode_uint (enc, MAJOR_ARRAY, len + 1); 359 encode_uint (enc, MAJOR_ARRAY, len + 1);
321 360
361 if (SvMAGICAL (av))
322 for (i = 0; i <= len; ++i) 362 for (i = 0; i <= len; ++i)
323 { 363 {
324 SV **svp = av_fetch (av, i, 0); 364 SV **svp = av_fetch (av, i, 0);
325 encode_sv (enc, svp ? *svp : &PL_sv_undef); 365 encode_sv (enc, svp ? *svp : &PL_sv_undef);
326 } 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 }
327 373
328 --enc->depth; 374 --enc->depth;
329} 375}
330 376
331static void 377static void
… …
349 while ((he = hv_iternext (hv))) 395 while ((he = hv_iternext (hv)))
350 { 396 {
351 if (HeKLEN (he) == HEf_SVKEY) 397 if (HeKLEN (he) == HEf_SVKEY)
352 encode_sv (enc, HeSVKEY (he)); 398 encode_sv (enc, HeSVKEY (he));
353 else 399 else
354 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));
355 401
356 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));
357 } 403 }
358 404
359 if (mg) 405 if (mg)
… …
435 481
436 if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0))) 482 if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
437 { 483 {
438 dSP; 484 dSP;
439 485
440 ENTER; SAVETMPS; PUSHMARK (SP); 486 ENTER; SAVETMPS;
487 PUSHMARK (SP);
441 // 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
442 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); 489 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
443 490
444 PUTBACK; 491 PUTBACK;
445 // G_SCALAR ensures that return value is 1 492 // G_SCALAR ensures that return value is 1
… …
458 } 505 }
459 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0) 506 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)
460 { 507 {
461 dSP; 508 dSP;
462 509
463 ENTER; SAVETMPS; PUSHMARK (SP); 510 ENTER; SAVETMPS;
511 SAVESTACK_POS ();
512 PUSHMARK (SP);
464 EXTEND (SP, 2); 513 EXTEND (SP, 2);
465 // 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
466 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); 515 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
467 PUSHs (sv_cbor); 516 PUSHs (sv_cbor);
468 517
… …
474 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv) 523 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)
475 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));
476 525
477 encode_tag (enc, CBOR_TAG_PERL_OBJECT); 526 encode_tag (enc, CBOR_TAG_PERL_OBJECT);
478 encode_uint (enc, MAJOR_ARRAY, count + 1); 527 encode_uint (enc, MAJOR_ARRAY, count + 1);
479 encode_strref (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash)); 528 encode_strref (enc, 0, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
480 529
481 while (count) 530 while (count)
482 encode_sv (enc, SP[1 - count--]); 531 encode_sv (enc, SP[1 - count--]);
483 532
484 PUTBACK; 533 PUTBACK;
… …
543 592
544 if (SvPOKp (sv)) 593 if (SvPOKp (sv))
545 { 594 {
546 STRLEN len; 595 STRLEN len;
547 char *str = SvPV (sv, len); 596 char *str = SvPV (sv, len);
548 encode_strref (enc, SvUTF8 (sv), str, len); 597 encode_strref (enc, enc->cbor.flags & F_TEXT_STRINGS, SvUTF8 (sv), str, len);
549 } 598 }
550 else if (SvNOKp (sv)) 599 else if (SvNOKp (sv))
551 encode_nv (enc, sv); 600 encode_nv (enc, sv);
552 else if (SvIOKp (sv)) 601 else if (SvIOKp (sv))
553 { 602 {
… …
570} 619}
571 620
572static SV * 621static SV *
573encode_cbor (SV *scalar, CBOR *cbor) 622encode_cbor (SV *scalar, CBOR *cbor)
574{ 623{
575 enc_t enc = { }; 624 enc_t enc = { 0 };
576 625
577 enc.cbor = *cbor; 626 enc.cbor = *cbor;
578 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 627 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
579 enc.cur = SvPVX (enc.sv); 628 enc.cur = SvPVX (enc.sv);
580 enc.end = SvEND (enc.sv); 629 enc.end = SvEND (enc.sv);
… …
612 U32 depth; // recursion depth 661 U32 depth; // recursion depth
613 U32 maxdepth; // recursion depth limit 662 U32 maxdepth; // recursion depth limit
614 AV *shareable; 663 AV *shareable;
615 AV *stringref; 664 AV *stringref;
616 SV *decode_tagged; 665 SV *decode_tagged;
666 SV *err_sv; // optional sv for error, needs to be freed
617} dec_t; 667} dec_t;
618 668
619#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);
620 676
621#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;
622 680
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
623#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)
624#define DEC_DEC_DEPTH --dec->depth 712#define DEC_DEC_DEPTH --dec->depth
625 713
626static UV 714static UV
627decode_uint (dec_t *dec) 715decode_uint (dec_t *dec)
628{ 716{
… …
705 av_push (av, decode_sv (dec)); 793 av_push (av, decode_sv (dec));
706 } 794 }
707 } 795 }
708 else 796 else
709 { 797 {
710 int i, len = decode_uint (dec); 798 UV i, len = decode_uint (dec);
711 799
712 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
713 av_fill (av, len - 1); 801 av_fill (av, len - 1);
714 802
715 for (i = 0; i < len; ++i) 803 for (i = 0; i < len; ++i)
… …
730{ 818{
731 // for speed reasons, we specialcase single-string 819 // for speed reasons, we specialcase single-string
732 // byte or utf-8 strings as keys, but only when !stringref 820 // byte or utf-8 strings as keys, but only when !stringref
733 821
734 if (ecb_expect_true (!dec->stringref)) 822 if (ecb_expect_true (!dec->stringref))
735 if (ecb_expect_true ((*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8)) 823 if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8))
736 { 824 {
737 I32 len = decode_uint (dec); 825 STRLEN len = decode_uint (dec);
738 char *key = (char *)dec->cur; 826 char *key = (char *)dec->cur;
739 827
828 WANT (len);
740 dec->cur += len; 829 dec->cur += len;
741 830
742 hv_store (hv, key, len, decode_sv (dec), 0); 831 hv_store (hv, key, len, decode_sv (dec), 0);
743 832
744 return; 833 return;
745 } 834 }
746 else if (ecb_expect_true ((*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8)) 835 else if (ecb_expect_true ((U8)(*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8))
747 { 836 {
748 I32 len = decode_uint (dec); 837 STRLEN len = decode_uint (dec);
749 char *key = (char *)dec->cur; 838 char *key = (char *)dec->cur;
750 839
840 WANT (len);
751 dec->cur += len; 841 dec->cur += len;
752 842
753 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8)) 843 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
754 if (!is_utf8_string (key, len)) 844 if (!is_utf8_string (key, len))
755 ERR ("corrupted CBOR data (invalid UTF-8 in map key)"); 845 ERR ("corrupted CBOR data (invalid UTF-8 in map key)");
… …
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);
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 }
764 882
765 hv_store_ent (hv, k, v, 0); 883 hv_store_ent (hv, k, v, 0);
766 SvREFCNT_dec (k); 884 SvREFCNT_dec (k);
767 885
768fail: 886fail:
… …
793 decode_he (dec, hv); 911 decode_he (dec, hv);
794 } 912 }
795 } 913 }
796 else 914 else
797 { 915 {
798 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
799 919
800 while (pairs--) 920 while (pairs--)
801 decode_he (dec, hv); 921 decode_he (dec, hv);
802 } 922 }
803 923
… …
891 sv = newRV_noinc (decode_sv (dec)); 1011 sv = newRV_noinc (decode_sv (dec));
892 break; 1012 break;
893 1013
894 case CBOR_TAG_STRINGREF_NAMESPACE: 1014 case CBOR_TAG_STRINGREF_NAMESPACE:
895 { 1015 {
1016 // do nmot use SAVETMPS/FREETMPS, as these will
1017 // erase mortalised caches, e.g. "shareable"
896 ENTER; SAVETMPS; 1018 ENTER;
897 1019
898 SAVESPTR (dec->stringref); 1020 SAVESPTR (dec->stringref);
899 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ()); 1021 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ());
900 1022
901 sv = decode_sv (dec); 1023 sv = decode_sv (dec);
902 1024
903 FREETMPS; LEAVE; 1025 LEAVE;
904 } 1026 }
905 break; 1027 break;
906 1028
907 case CBOR_TAG_STRINGREF: 1029 case CBOR_TAG_STRINGREF:
908 { 1030 {
… …
978 if (!method) 1100 if (!method)
979 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)");
980 1102
981 dSP; 1103 dSP;
982 1104
983 ENTER; SAVETMPS; PUSHMARK (SP); 1105 ENTER; SAVETMPS;
1106 PUSHMARK (SP);
984 EXTEND (SP, len + 1); 1107 EXTEND (SP, len + 1);
985 // 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
986 PUSHs (*av_fetch (av, 0, 1)); 1109 PUSHs (*av_fetch (av, 0, 1));
987 PUSHs (sv_cbor); 1110 PUSHs (sv_cbor);
988 1111
… …
996 SPAGAIN; 1119 SPAGAIN;
997 1120
998 if (SvTRUE (ERRSV)) 1121 if (SvTRUE (ERRSV))
999 { 1122 {
1000 FREETMPS; LEAVE; 1123 FREETMPS; LEAVE;
1001 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV)))); 1124 ERR_ERRSV;
1002 } 1125 }
1003 1126
1004 SvREFCNT_dec (sv); 1127 SvREFCNT_dec (sv);
1005 sv = SvREFCNT_inc (POPs); 1128 sv = SvREFCNT_inc (POPs);
1006 1129
… …
1010 } 1133 }
1011 break; 1134 break;
1012 1135
1013 default: 1136 default:
1014 { 1137 {
1138 SV *tag_sv = newSVuv (tag);
1139
1015 sv = decode_sv (dec); 1140 sv = decode_sv (dec);
1016 1141
1017 dSP; 1142 dSP;
1018 ENTER; SAVETMPS; PUSHMARK (SP); 1143 ENTER; SAVETMPS;
1144 SAVESTACK_POS ();
1145 PUSHMARK (SP);
1019 EXTEND (SP, 2); 1146 EXTEND (SP, 2);
1020 PUSHs (newSVuv (tag)); 1147 PUSHs (tag_sv);
1021 PUSHs (sv); 1148 PUSHs (sv);
1022 1149
1023 PUTBACK; 1150 PUTBACK;
1024 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);
1025 SPAGAIN; 1152 SPAGAIN;
1026 1153
1027 if (SvTRUE (ERRSV)) 1154 if (SvTRUE (ERRSV))
1028 { 1155 {
1156 SvREFCNT_dec (tag_sv);
1029 FREETMPS; LEAVE; 1157 FREETMPS; LEAVE;
1030 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV)))); 1158 ERR_ERRSV;
1031 } 1159 }
1032 1160
1033 if (count) 1161 if (count)
1034 { 1162 {
1163 SvREFCNT_dec (tag_sv);
1035 SvREFCNT_dec (sv); 1164 SvREFCNT_dec (sv);
1036 sv = SvREFCNT_inc (POPs); 1165 sv = SvREFCNT_inc (POPs);
1037 } 1166 }
1038 else 1167 else
1039 { 1168 {
1040 AV *av = newAV (); 1169 AV *av = newAV ();
1041 av_push (av, newSVuv (tag)); 1170 av_push (av, tag_sv);
1042 av_push (av, sv); 1171 av_push (av, sv);
1043 1172
1044 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 1173 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
1045 ? cbor_tagged_stash 1174 ? cbor_tagged_stash
1046 : gv_stashpv ("CBOR::XS::Tagged" , 1); 1175 : gv_stashpv ("CBOR::XS::Tagged" , 1);
… …
1132 1261
1133 return newSVnv (ecb_binary64_to_double (fp)); 1262 return newSVnv (ecb_binary64_to_double (fp));
1134 } 1263 }
1135 1264
1136 // 0..19 unassigned simple 1265 // 0..19 unassigned simple
1137 // 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
1138 default: 1269 default:
1139 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)"); 1270 ERR ("corrupted CBOR data (reserved/unassigned/unexpected major 7 value)");
1140 } 1271 }
1141 1272
1142 break; 1273 break;
1143 } 1274 }
1144 1275
… …
1147} 1278}
1148 1279
1149static SV * 1280static SV *
1150decode_cbor (SV *string, CBOR *cbor, char **offset_return) 1281decode_cbor (SV *string, CBOR *cbor, char **offset_return)
1151{ 1282{
1152 dec_t dec = { }; 1283 dec_t dec = { 0 };
1153 SV *sv; 1284 SV *sv;
1154 STRLEN len; 1285 STRLEN len;
1155 char *data = SvPVbyte (string, len); 1286 char *data = SvPVbyte (string, len);
1156 1287
1157 if (len > cbor->max_size && cbor->max_size) 1288 if (len > cbor->max_size && cbor->max_size)
… …
1173 1304
1174 if (dec.err) 1305 if (dec.err)
1175 { 1306 {
1176 if (dec.shareable) 1307 if (dec.shareable)
1177 { 1308 {
1178 // need to break cyclic links, which whould all be in shareable 1309 // need to break cyclic links, which would all be in shareable
1179 int i; 1310 int i;
1180 SV **svp; 1311 SV **svp;
1181 1312
1182 for (i = av_len (dec.shareable) + 1; i--; ) 1313 for (i = av_len (dec.shareable) + 1; i--; )
1183 if ((svp = av_fetch (dec.shareable, i, 0))) 1314 if ((svp = av_fetch (dec.shareable, i, 0)))
1184 sv_setsv (*svp, &PL_sv_undef); 1315 sv_setsv (*svp, &PL_sv_undef);
1185 } 1316 }
1186 1317
1187 SvREFCNT_dec (sv); 1318 SvREFCNT_dec (sv);
1319
1320 if (dec.err_sv)
1321 sv_2mortal (dec.err_sv);
1322
1188 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);
1189 } 1324 }
1190 1325
1191 sv = sv_2mortal (sv); 1326 sv = sv_2mortal (sv);
1192 1327
1193 return sv; 1328 return sv;
1194} 1329}
1195 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
1196///////////////////////////////////////////////////////////////////////////// 1455/////////////////////////////////////////////////////////////////////////////
1197// XS interface functions 1456// XS interface functions
1198 1457
1199MODULE = CBOR::XS PACKAGE = CBOR::XS 1458MODULE = CBOR::XS PACKAGE = CBOR::XS
1200 1459
… …
1212 1471
1213 default_filter = newSVpv ("CBOR::XS::default_filter", 0); 1472 default_filter = newSVpv ("CBOR::XS::default_filter", 0);
1214 1473
1215 sv_cbor = newSVpv ("CBOR", 0); 1474 sv_cbor = newSVpv ("CBOR", 0);
1216 SvREADONLY_on (sv_cbor); 1475 SvREADONLY_on (sv_cbor);
1476
1477 assert (("STRLEN must be an unsigned type", 0 <= (STRLEN)-1));
1217} 1478}
1218 1479
1219PROTOTYPES: DISABLE 1480PROTOTYPES: DISABLE
1220 1481
1221void CLONE (...) 1482void CLONE (...)
… …
1242 shrink = F_SHRINK 1503 shrink = F_SHRINK
1243 allow_unknown = F_ALLOW_UNKNOWN 1504 allow_unknown = F_ALLOW_UNKNOWN
1244 allow_sharing = F_ALLOW_SHARING 1505 allow_sharing = F_ALLOW_SHARING
1245 allow_cycles = F_ALLOW_CYCLES 1506 allow_cycles = F_ALLOW_CYCLES
1246 pack_strings = F_PACK_STRINGS 1507 pack_strings = F_PACK_STRINGS
1508 text_keys = F_TEXT_KEYS
1509 text_strings = F_TEXT_STRINGS
1247 validate_utf8 = F_VALIDATE_UTF8 1510 validate_utf8 = F_VALIDATE_UTF8
1248 PPCODE: 1511 PPCODE:
1249{ 1512{
1250 if (enable) 1513 if (enable)
1251 self->flags |= ix; 1514 self->flags |= ix;
… …
1260 get_shrink = F_SHRINK 1523 get_shrink = F_SHRINK
1261 get_allow_unknown = F_ALLOW_UNKNOWN 1524 get_allow_unknown = F_ALLOW_UNKNOWN
1262 get_allow_sharing = F_ALLOW_SHARING 1525 get_allow_sharing = F_ALLOW_SHARING
1263 get_allow_cycles = F_ALLOW_CYCLES 1526 get_allow_cycles = F_ALLOW_CYCLES
1264 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
1265 get_validate_utf8 = F_VALIDATE_UTF8 1530 get_validate_utf8 = F_VALIDATE_UTF8
1266 PPCODE: 1531 PPCODE:
1267 XPUSHs (boolSV (self->flags & ix)); 1532 XPUSHs (boolSV (self->flags & ix));
1268 1533
1269void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) 1534void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
… …
1319 EXTEND (SP, 2); 1584 EXTEND (SP, 2);
1320 PUSHs (sv); 1585 PUSHs (sv);
1321 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); 1586 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
1322} 1587}
1323 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
1324void DESTROY (CBOR *self) 1641void DESTROY (CBOR *self)
1325 PPCODE: 1642 PPCODE:
1326 cbor_free (self); 1643 cbor_free (self);
1327 1644
1328PROTOTYPES: ENABLE 1645PROTOTYPES: ENABLE

Diff Legend

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