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 1.37 by root, Sat Nov 30 18:13:53 2013 UTC vs.
Revision 1.60 by root, Sat Nov 26 00:47:02 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))
… …
98 99
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
104#define F_FORBID_OBJECTS 0x00000010UL
103#define F_PACK_STRINGS 0x00000010UL 105#define F_PACK_STRINGS 0x00000020UL
106#define F_TEXT_KEYS 0x00000040UL
107#define F_TEXT_STRINGS 0x00000080UL
108#define F_VALIDATE_UTF8 0x00000100UL
104 109
105#define INIT_SIZE 32 // initial scalar size to be allocated 110#define INIT_SIZE 32 // initial scalar size to be allocated
106 111
107#define SB do { 112#define SB do {
108#define SE } while (0) 113#define SE } while (0)
… …
127typedef struct { 132typedef struct {
128 U32 flags; 133 U32 flags;
129 U32 max_depth; 134 U32 max_depth;
130 STRLEN max_size; 135 STRLEN max_size;
131 SV *filter; 136 SV *filter;
137
138 // for the incremental parser
139 STRLEN incr_pos; // the current offset into the text
140 STRLEN incr_need; // minimum bytes needed to decode
141 AV *incr_count; // for every nesting level, the number of outstanding values, or -1 for indef.
132} CBOR; 142} CBOR;
133 143
134ecb_inline void 144ecb_inline void
135cbor_init (CBOR *cbor) 145cbor_init (CBOR *cbor)
136{ 146{
… …
140 150
141ecb_inline void 151ecb_inline void
142cbor_free (CBOR *cbor) 152cbor_free (CBOR *cbor)
143{ 153{
144 SvREFCNT_dec (cbor->filter); 154 SvREFCNT_dec (cbor->filter);
155 SvREFCNT_dec (cbor->incr_count);
145} 156}
146 157
147///////////////////////////////////////////////////////////////////////////// 158/////////////////////////////////////////////////////////////////////////////
148// utility functions 159// utility functions
149 160
… …
268encode_tag (enc_t *enc, UV tag) 279encode_tag (enc_t *enc, UV tag)
269{ 280{
270 encode_uint (enc, MAJOR_TAG, tag); 281 encode_uint (enc, MAJOR_TAG, tag);
271} 282}
272 283
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
273ecb_inline void 307ecb_inline void
274encode_str (enc_t *enc, int utf8, char *str, STRLEN len) 308encode_str (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len)
275{ 309{
310 if (ecb_expect_false (upgrade_utf8))
311 if (!utf8)
312 {
313 encode_str_utf8 (enc, utf8, str, len);
314 return;
315 }
316
276 encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len); 317 encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len);
277 need (enc, len); 318 need (enc, len);
278 memcpy (enc->cur, str, len); 319 memcpy (enc->cur, str, len);
279 enc->cur += len; 320 enc->cur += len;
280} 321}
281 322
282static void 323ecb_inline void
283encode_strref (enc_t *enc, int utf8, char *str, STRLEN len) 324encode_strref (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len)
284{ 325{
285 if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS)) 326 if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS))
286 { 327 {
287 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1); 328 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1);
288 329
… …
299 sv_setuv (*svp, enc->stringref_idx); 340 sv_setuv (*svp, enc->stringref_idx);
300 ++enc->stringref_idx; 341 ++enc->stringref_idx;
301 } 342 }
302 } 343 }
303 344
304 encode_str (enc, utf8, str, len); 345 encode_str (enc, upgrade_utf8, utf8, str, len);
305} 346}
306 347
307static void encode_sv (enc_t *enc, SV *sv); 348static void encode_sv (enc_t *enc, SV *sv);
308 349
309static void 350static void
… …
316 357
317 ++enc->depth; 358 ++enc->depth;
318 359
319 encode_uint (enc, MAJOR_ARRAY, len + 1); 360 encode_uint (enc, MAJOR_ARRAY, len + 1);
320 361
362 if (SvMAGICAL (av))
321 for (i = 0; i <= len; ++i) 363 for (i = 0; i <= len; ++i)
322 { 364 {
323 SV **svp = av_fetch (av, i, 0); 365 SV **svp = av_fetch (av, i, 0);
324 encode_sv (enc, svp ? *svp : &PL_sv_undef); 366 encode_sv (enc, svp ? *svp : &PL_sv_undef);
325 } 367 }
368 else
369 for (i = 0; i <= len; ++i)
370 {
371 SV *sv = AvARRAY (av)[i];
372 encode_sv (enc, sv ? sv : &PL_sv_undef);
373 }
326 374
327 --enc->depth; 375 --enc->depth;
328} 376}
329 377
330static void 378static void
… …
348 while ((he = hv_iternext (hv))) 396 while ((he = hv_iternext (hv)))
349 { 397 {
350 if (HeKLEN (he) == HEf_SVKEY) 398 if (HeKLEN (he) == HEf_SVKEY)
351 encode_sv (enc, HeSVKEY (he)); 399 encode_sv (enc, HeSVKEY (he));
352 else 400 else
353 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));
354 402
355 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));
356 } 404 }
357 405
358 if (mg) 406 if (mg)
… …
430 if (ecb_expect_false (SvOBJECT (sv))) 478 if (ecb_expect_false (SvOBJECT (sv)))
431 { 479 {
432 HV *stash = SvSTASH (sv); 480 HV *stash = SvSTASH (sv);
433 GV *method; 481 GV *method;
434 482
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))));
435 if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0))) 486 else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
436 { 487 {
437 dSP; 488 dSP;
438 489
439 ENTER; SAVETMPS; PUSHMARK (SP); 490 ENTER; SAVETMPS;
491 PUSHMARK (SP);
440 // we re-bless the reference to get overload and other niceties right 492 // we re-bless the reference to get overload and other niceties right
441 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); 493 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
442 494
443 PUTBACK; 495 PUTBACK;
444 // G_SCALAR ensures that return value is 1 496 // G_SCALAR ensures that return value is 1
… …
457 } 509 }
458 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0) 510 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)
459 { 511 {
460 dSP; 512 dSP;
461 513
462 ENTER; SAVETMPS; PUSHMARK (SP); 514 ENTER; SAVETMPS;
515 SAVESTACK_POS ();
516 PUSHMARK (SP);
463 EXTEND (SP, 2); 517 EXTEND (SP, 2);
464 // we re-bless the reference to get overload and other niceties right 518 // we re-bless the reference to get overload and other niceties right
465 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); 519 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
466 PUSHs (sv_cbor); 520 PUSHs (sv_cbor);
467 521
… …
473 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv) 527 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)); 528 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));
475 529
476 encode_tag (enc, CBOR_TAG_PERL_OBJECT); 530 encode_tag (enc, CBOR_TAG_PERL_OBJECT);
477 encode_uint (enc, MAJOR_ARRAY, count + 1); 531 encode_uint (enc, MAJOR_ARRAY, count + 1);
478 encode_strref (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash)); 532 encode_strref (enc, 0, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
479 533
480 while (count) 534 while (count)
481 encode_sv (enc, SP[1 - count--]); 535 encode_sv (enc, SP[1 - count--]);
482 536
483 PUTBACK; 537 PUTBACK;
… …
542 596
543 if (SvPOKp (sv)) 597 if (SvPOKp (sv))
544 { 598 {
545 STRLEN len; 599 STRLEN len;
546 char *str = SvPV (sv, len); 600 char *str = SvPV (sv, len);
547 encode_strref (enc, SvUTF8 (sv), str, len); 601 encode_strref (enc, enc->cbor.flags & F_TEXT_STRINGS, SvUTF8 (sv), str, len);
548 } 602 }
549 else if (SvNOKp (sv)) 603 else if (SvNOKp (sv))
550 encode_nv (enc, sv); 604 encode_nv (enc, sv);
551 else if (SvIOKp (sv)) 605 else if (SvIOKp (sv))
552 { 606 {
… …
569} 623}
570 624
571static SV * 625static SV *
572encode_cbor (SV *scalar, CBOR *cbor) 626encode_cbor (SV *scalar, CBOR *cbor)
573{ 627{
574 enc_t enc = { }; 628 enc_t enc = { 0 };
575 629
576 enc.cbor = *cbor; 630 enc.cbor = *cbor;
577 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 631 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
578 enc.cur = SvPVX (enc.sv); 632 enc.cur = SvPVX (enc.sv);
579 enc.end = SvEND (enc.sv); 633 enc.end = SvEND (enc.sv);
… …
611 U32 depth; // recursion depth 665 U32 depth; // recursion depth
612 U32 maxdepth; // recursion depth limit 666 U32 maxdepth; // recursion depth limit
613 AV *shareable; 667 AV *shareable;
614 AV *stringref; 668 AV *stringref;
615 SV *decode_tagged; 669 SV *decode_tagged;
670 SV *err_sv; // optional sv for error, needs to be freed
616} dec_t; 671} dec_t;
617 672
618#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);
619 680
620#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;
621 684
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
622#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)
623#define DEC_DEC_DEPTH --dec->depth 716#define DEC_DEC_DEPTH --dec->depth
624 717
625static UV 718static UV
626decode_uint (dec_t *dec) 719decode_uint (dec_t *dec)
627{ 720{
… …
704 av_push (av, decode_sv (dec)); 797 av_push (av, decode_sv (dec));
705 } 798 }
706 } 799 }
707 else 800 else
708 { 801 {
709 int i, len = decode_uint (dec); 802 UV i, len = decode_uint (dec);
710 803
711 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
712 av_fill (av, len - 1); 805 av_fill (av, len - 1);
713 806
714 for (i = 0; i < len; ++i) 807 for (i = 0; i < len; ++i)
… …
729{ 822{
730 // for speed reasons, we specialcase single-string 823 // for speed reasons, we specialcase single-string
731 // byte or utf-8 strings as keys, but only when !stringref 824 // byte or utf-8 strings as keys, but only when !stringref
732 825
733 if (ecb_expect_true (!dec->stringref)) 826 if (ecb_expect_true (!dec->stringref))
734 if (ecb_expect_true ((*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8)) 827 if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8))
735 { 828 {
736 I32 len = decode_uint (dec); 829 STRLEN len = decode_uint (dec);
737 char *key = (char *)dec->cur; 830 char *key = (char *)dec->cur;
738 831
832 WANT (len);
739 dec->cur += len; 833 dec->cur += len;
740 834
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); 835 hv_store (hv, key, len, decode_sv (dec), 0);
745 836
746 return; 837 return;
747 } 838 }
748 else if (ecb_expect_true ((*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8)) 839 else if (ecb_expect_true ((U8)(*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8))
749 { 840 {
750 I32 len = decode_uint (dec); 841 STRLEN len = decode_uint (dec);
751 char *key = (char *)dec->cur; 842 char *key = (char *)dec->cur;
752 843
844 WANT (len);
753 dec->cur += len; 845 dec->cur += len;
754 846
755 if (ecb_expect_false (dec->stringref)) 847 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
756 av_push (dec->stringref, newSVpvn_utf8 (key, len, 1)); 848 if (!is_utf8_string (key, len))
849 ERR ("corrupted CBOR data (invalid UTF-8 in map key)");
757 850
758 hv_store (hv, key, -len, decode_sv (dec), 0); 851 hv_store (hv, key, -len, decode_sv (dec), 0);
759 852
760 return; 853 return;
761 } 854 }
762 855
763 SV *k = decode_sv (dec); 856 SV *k = decode_sv (dec);
764 SV *v = decode_sv (dec); 857 SV *v = decode_sv (dec);
765 858
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 }
886
766 hv_store_ent (hv, k, v, 0); 887 hv_store_ent (hv, k, v, 0);
767 SvREFCNT_dec (k); 888 SvREFCNT_dec (k);
889
890fail:
891 ;
768} 892}
769 893
770static SV * 894static SV *
771decode_hv (dec_t *dec) 895decode_hv (dec_t *dec)
772{ 896{
… …
791 decode_he (dec, hv); 915 decode_he (dec, hv);
792 } 916 }
793 } 917 }
794 else 918 else
795 { 919 {
796 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
797 923
798 while (pairs--) 924 while (pairs--)
799 decode_he (dec, hv); 925 decode_he (dec, hv);
800 } 926 }
801 927
… …
854 && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1)) 980 && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1))
855 av_push (dec->stringref, SvREFCNT_inc_NN (sv)); 981 av_push (dec->stringref, SvREFCNT_inc_NN (sv));
856 } 982 }
857 983
858 if (utf8) 984 if (utf8)
985 {
986 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
987 if (!is_utf8_string (SvPVX (sv), SvCUR (sv)))
988 ERR ("corrupted CBOR data (invalid UTF-8 in text string)");
989
859 SvUTF8_on (sv); 990 SvUTF8_on (sv);
991 }
860 992
861 return sv; 993 return sv;
862 994
863fail: 995fail:
864 SvREFCNT_dec (sv); 996 SvREFCNT_dec (sv);
… …
883 sv = newRV_noinc (decode_sv (dec)); 1015 sv = newRV_noinc (decode_sv (dec));
884 break; 1016 break;
885 1017
886 case CBOR_TAG_STRINGREF_NAMESPACE: 1018 case CBOR_TAG_STRINGREF_NAMESPACE:
887 { 1019 {
1020 // do nmot use SAVETMPS/FREETMPS, as these will
1021 // erase mortalised caches, e.g. "shareable"
888 ENTER; SAVETMPS; 1022 ENTER;
889 1023
890 SAVESPTR (dec->stringref); 1024 SAVESPTR (dec->stringref);
891 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ()); 1025 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ());
892 1026
893 sv = decode_sv (dec); 1027 sv = decode_sv (dec);
894 1028
895 FREETMPS; LEAVE; 1029 LEAVE;
896 } 1030 }
897 break; 1031 break;
898 1032
899 case CBOR_TAG_STRINGREF: 1033 case CBOR_TAG_STRINGREF:
900 { 1034 {
… …
951 } 1085 }
952 break; 1086 break;
953 1087
954 case CBOR_TAG_PERL_OBJECT: 1088 case CBOR_TAG_PERL_OBJECT:
955 { 1089 {
1090 if (dec->cbor.flags & F_FORBID_OBJECTS)
1091 goto filter;
1092
956 sv = decode_sv (dec); 1093 sv = decode_sv (dec);
957 1094
958 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV) 1095 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
959 ERR ("corrupted CBOR data (non-array perl object)"); 1096 ERR ("corrupted CBOR data (non-array perl object)");
960 1097
… …
970 if (!method) 1107 if (!method)
971 ERR ("cannot decode perl-object (package does not have a THAW method)"); 1108 ERR ("cannot decode perl-object (package does not have a THAW method)");
972 1109
973 dSP; 1110 dSP;
974 1111
975 ENTER; SAVETMPS; PUSHMARK (SP); 1112 ENTER; SAVETMPS;
1113 PUSHMARK (SP);
976 EXTEND (SP, len + 1); 1114 EXTEND (SP, len + 1);
977 // we re-bless the reference to get overload and other niceties right 1115 // we re-bless the reference to get overload and other niceties right
978 PUSHs (*av_fetch (av, 0, 1)); 1116 PUSHs (*av_fetch (av, 0, 1));
979 PUSHs (sv_cbor); 1117 PUSHs (sv_cbor);
980 1118
… …
988 SPAGAIN; 1126 SPAGAIN;
989 1127
990 if (SvTRUE (ERRSV)) 1128 if (SvTRUE (ERRSV))
991 { 1129 {
992 FREETMPS; LEAVE; 1130 FREETMPS; LEAVE;
993 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV)))); 1131 ERR_ERRSV;
994 } 1132 }
995 1133
996 SvREFCNT_dec (sv); 1134 SvREFCNT_dec (sv);
997 sv = SvREFCNT_inc (POPs); 1135 sv = SvREFCNT_inc (POPs);
998 1136
… …
1001 FREETMPS; LEAVE; 1139 FREETMPS; LEAVE;
1002 } 1140 }
1003 break; 1141 break;
1004 1142
1005 default: 1143 default:
1144 filter:
1006 { 1145 {
1146 SV *tag_sv = newSVuv (tag);
1147
1007 sv = decode_sv (dec); 1148 sv = decode_sv (dec);
1008 1149
1009 dSP; 1150 dSP;
1010 ENTER; SAVETMPS; PUSHMARK (SP); 1151 ENTER; SAVETMPS;
1152 SAVESTACK_POS ();
1153 PUSHMARK (SP);
1011 EXTEND (SP, 2); 1154 EXTEND (SP, 2);
1012 PUSHs (newSVuv (tag)); 1155 PUSHs (tag_sv);
1013 PUSHs (sv); 1156 PUSHs (sv);
1014 1157
1015 PUTBACK; 1158 PUTBACK;
1016 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);
1017 SPAGAIN; 1160 SPAGAIN;
1018 1161
1019 if (SvTRUE (ERRSV)) 1162 if (SvTRUE (ERRSV))
1020 { 1163 {
1164 SvREFCNT_dec (tag_sv);
1021 FREETMPS; LEAVE; 1165 FREETMPS; LEAVE;
1022 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV)))); 1166 ERR_ERRSV;
1023 } 1167 }
1024 1168
1025 if (count) 1169 if (count)
1026 { 1170 {
1171 SvREFCNT_dec (tag_sv);
1027 SvREFCNT_dec (sv); 1172 SvREFCNT_dec (sv);
1028 sv = SvREFCNT_inc (POPs); 1173 sv = SvREFCNT_inc (POPs);
1029 } 1174 }
1030 else 1175 else
1031 { 1176 {
1032 AV *av = newAV (); 1177 AV *av = newAV ();
1033 av_push (av, newSVuv (tag)); 1178 av_push (av, tag_sv);
1034 av_push (av, sv); 1179 av_push (av, sv);
1035 1180
1036 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 1181 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
1037 ? cbor_tagged_stash 1182 ? cbor_tagged_stash
1038 : gv_stashpv ("CBOR::XS::Tagged" , 1); 1183 : gv_stashpv ("CBOR::XS::Tagged" , 1);
… …
1124 1269
1125 return newSVnv (ecb_binary64_to_double (fp)); 1270 return newSVnv (ecb_binary64_to_double (fp));
1126 } 1271 }
1127 1272
1128 // 0..19 unassigned simple 1273 // 0..19 unassigned simple
1129 // 24 reserved + unassigned (reserved values are not encodable) 1274 // 24 reserved + unassigned simple (reserved values are not encodable)
1275 // 28-30 unassigned misc
1276 // 31 break code
1130 default: 1277 default:
1131 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)"); 1278 ERR ("corrupted CBOR data (reserved/unassigned/unexpected major 7 value)");
1132 } 1279 }
1133 1280
1134 break; 1281 break;
1135 } 1282 }
1136 1283
… …
1139} 1286}
1140 1287
1141static SV * 1288static SV *
1142decode_cbor (SV *string, CBOR *cbor, char **offset_return) 1289decode_cbor (SV *string, CBOR *cbor, char **offset_return)
1143{ 1290{
1144 dec_t dec = { }; 1291 dec_t dec = { 0 };
1145 SV *sv; 1292 SV *sv;
1146 STRLEN len; 1293 STRLEN len;
1147 char *data = SvPVbyte (string, len); 1294 char *data = SvPVbyte (string, len);
1148 1295
1149 if (len > cbor->max_size && cbor->max_size) 1296 if (len > cbor->max_size && cbor->max_size)
… …
1163 if (dec.cur != dec.end && !dec.err) 1310 if (dec.cur != dec.end && !dec.err)
1164 dec.err = "garbage after CBOR object"; 1311 dec.err = "garbage after CBOR object";
1165 1312
1166 if (dec.err) 1313 if (dec.err)
1167 { 1314 {
1315 if (dec.shareable)
1316 {
1317 // need to break cyclic links, which would all be in shareable
1318 int i;
1319 SV **svp;
1320
1321 for (i = av_len (dec.shareable) + 1; i--; )
1322 if ((svp = av_fetch (dec.shareable, i, 0)))
1323 sv_setsv (*svp, &PL_sv_undef);
1324 }
1325
1168 SvREFCNT_dec (sv); 1326 SvREFCNT_dec (sv);
1327
1328 if (dec.err_sv)
1329 sv_2mortal (dec.err_sv);
1330
1169 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);
1170 } 1332 }
1171 1333
1172 sv = sv_2mortal (sv); 1334 sv = sv_2mortal (sv);
1173 1335
1174 return sv; 1336 return sv;
1175} 1337}
1176 1338
1339/////////////////////////////////////////////////////////////////////////////
1340// incremental parser
1341
1342#define INCR_DONE(cbor) (AvFILLp (cbor->incr_count) < 0)
1343
1344// returns 0 for notyet, 1 for success or error
1345static int
1346incr_parse (CBOR *self, SV *cborstr)
1347{
1348 STRLEN cur;
1349 SvPV (cborstr, cur);
1350
1351 while (ecb_expect_true (self->incr_need <= cur))
1352 {
1353 // table of integer count bytes
1354 static I8 incr_len[MINOR_MASK + 1] = {
1355 0, 0, 0, 0, 0, 0, 0, 0,
1356 0, 0, 0, 0, 0, 0, 0, 0,
1357 0, 0, 0, 0, 0, 0, 0, 0,
1358 1, 2, 4, 8,-1,-1,-1,-2
1359 };
1360
1361 const U8 *p = SvPVX (cborstr) + self->incr_pos;
1362 U8 m = *p & MINOR_MASK;
1363 IV count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1364 I8 ilen = incr_len[m];
1365
1366 self->incr_need = self->incr_pos + 1;
1367
1368 if (ecb_expect_false (ilen < 0))
1369 {
1370 if (m != MINOR_INDEF)
1371 return 1; // error
1372
1373 if (*p == (MAJOR_MISC | MINOR_INDEF))
1374 {
1375 if (count >= 0)
1376 return 1; // error
1377
1378 count = 1;
1379 }
1380 else
1381 {
1382 av_push (self->incr_count, newSViv (-1)); //TODO: nest
1383 count = -1;
1384 }
1385 }
1386 else
1387 {
1388 self->incr_need += ilen;
1389 if (ecb_expect_false (self->incr_need > cur))
1390 return 0;
1391
1392 int major = *p >> MAJOR_SHIFT;
1393
1394 switch (major)
1395 {
1396 case MAJOR_TAG >> MAJOR_SHIFT:
1397 ++count; // tags merely prefix another value
1398 break;
1399
1400 case MAJOR_BYTES >> MAJOR_SHIFT:
1401 case MAJOR_TEXT >> MAJOR_SHIFT:
1402 case MAJOR_ARRAY >> MAJOR_SHIFT:
1403 case MAJOR_MAP >> MAJOR_SHIFT:
1404 {
1405 UV len;
1406
1407 if (ecb_expect_false (ilen))
1408 {
1409 len = 0;
1410
1411 do {
1412 len = (len << 8) | *++p;
1413 } while (--ilen);
1414 }
1415 else
1416 len = m;
1417
1418 switch (major)
1419 {
1420 case MAJOR_BYTES >> MAJOR_SHIFT:
1421 case MAJOR_TEXT >> MAJOR_SHIFT:
1422 self->incr_need += len;
1423 if (ecb_expect_false (self->incr_need > cur))
1424 return 0;
1425
1426 break;
1427
1428 case MAJOR_MAP >> MAJOR_SHIFT:
1429 len <<= 1;
1430 case MAJOR_ARRAY >> MAJOR_SHIFT:
1431 if (len)
1432 {
1433 av_push (self->incr_count, newSViv (len + 1)); //TODO: nest
1434 count = len + 1;
1435 }
1436 break;
1437 }
1438 }
1439 }
1440 }
1441
1442 self->incr_pos = self->incr_need;
1443
1444 if (count > 0)
1445 {
1446 while (!--count)
1447 {
1448 if (!AvFILLp (self->incr_count))
1449 return 1; // done
1450
1451 SvREFCNT_dec_NN (av_pop (self->incr_count));
1452 count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1453 }
1454
1455 SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]) = count;
1456 }
1457 }
1458
1459 return 0;
1460}
1461
1462
1177///////////////////////////////////////////////////////////////////////////// 1463/////////////////////////////////////////////////////////////////////////////
1178// XS interface functions 1464// XS interface functions
1179 1465
1180MODULE = CBOR::XS PACKAGE = CBOR::XS 1466MODULE = CBOR::XS PACKAGE = CBOR::XS
1181 1467
… …
1193 1479
1194 default_filter = newSVpv ("CBOR::XS::default_filter", 0); 1480 default_filter = newSVpv ("CBOR::XS::default_filter", 0);
1195 1481
1196 sv_cbor = newSVpv ("CBOR", 0); 1482 sv_cbor = newSVpv ("CBOR", 0);
1197 SvREADONLY_on (sv_cbor); 1483 SvREADONLY_on (sv_cbor);
1484
1485 assert (("STRLEN must be an unsigned type", 0 <= (STRLEN)-1));
1198} 1486}
1199 1487
1200PROTOTYPES: DISABLE 1488PROTOTYPES: DISABLE
1201 1489
1202void CLONE (...) 1490void CLONE (...)
… …
1222 ALIAS: 1510 ALIAS:
1223 shrink = F_SHRINK 1511 shrink = F_SHRINK
1224 allow_unknown = F_ALLOW_UNKNOWN 1512 allow_unknown = F_ALLOW_UNKNOWN
1225 allow_sharing = F_ALLOW_SHARING 1513 allow_sharing = F_ALLOW_SHARING
1226 allow_cycles = F_ALLOW_CYCLES 1514 allow_cycles = F_ALLOW_CYCLES
1515 forbid_objects = F_FORBID_OBJECTS
1227 pack_strings = F_PACK_STRINGS 1516 pack_strings = F_PACK_STRINGS
1517 text_keys = F_TEXT_KEYS
1518 text_strings = F_TEXT_STRINGS
1519 validate_utf8 = F_VALIDATE_UTF8
1228 PPCODE: 1520 PPCODE:
1229{ 1521{
1230 if (enable) 1522 if (enable)
1231 self->flags |= ix; 1523 self->flags |= ix;
1232 else 1524 else
… …
1239 ALIAS: 1531 ALIAS:
1240 get_shrink = F_SHRINK 1532 get_shrink = F_SHRINK
1241 get_allow_unknown = F_ALLOW_UNKNOWN 1533 get_allow_unknown = F_ALLOW_UNKNOWN
1242 get_allow_sharing = F_ALLOW_SHARING 1534 get_allow_sharing = F_ALLOW_SHARING
1243 get_allow_cycles = F_ALLOW_CYCLES 1535 get_allow_cycles = F_ALLOW_CYCLES
1536 get_forbid_objects = F_FORBID_OBJECTS
1244 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
1540 get_validate_utf8 = F_VALIDATE_UTF8
1245 PPCODE: 1541 PPCODE:
1246 XPUSHs (boolSV (self->flags & ix)); 1542 XPUSHs (boolSV (self->flags & ix));
1247 1543
1248void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) 1544void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
1249 PPCODE: 1545 PPCODE:
… …
1298 EXTEND (SP, 2); 1594 EXTEND (SP, 2);
1299 PUSHs (sv); 1595 PUSHs (sv);
1300 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); 1596 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
1301} 1597}
1302 1598
1599void incr_parse (CBOR *self, SV *cborstr)
1600 ALIAS:
1601 incr_parse_multiple = 1
1602 PPCODE:
1603{
1604 if (SvUTF8 (cborstr))
1605 sv_utf8_downgrade (cborstr, 0);
1606
1607 if (!self->incr_count)
1608 {
1609 self->incr_count = newAV ();
1610 self->incr_pos = 0;
1611 self->incr_need = 1;
1612
1613 av_push (self->incr_count, newSViv (1));
1614 }
1615
1616 do
1617 {
1618 if (!incr_parse (self, cborstr))
1619 {
1620 if (self->incr_need > self->max_size && self->max_size)
1621 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
1622 (unsigned long)self->incr_need, (unsigned long)self->max_size);
1623
1624 break;
1625 }
1626
1627 SV *sv;
1628 char *offset;
1629
1630 PUTBACK; sv = decode_cbor (cborstr, self, &offset); SPAGAIN;
1631 XPUSHs (sv);
1632
1633 sv_chop (cborstr, offset);
1634
1635 av_clear (self->incr_count);
1636 av_push (self->incr_count, newSViv (1));
1637
1638 self->incr_pos = 0;
1639 self->incr_need = self->incr_pos + 1;
1640 }
1641 while (ix);
1642}
1643
1644void incr_reset (CBOR *self)
1645 CODE:
1646{
1647 SvREFCNT_dec (self->incr_count);
1648 self->incr_count = 0;
1649}
1650
1303void DESTROY (CBOR *self) 1651void DESTROY (CBOR *self)
1304 PPCODE: 1652 PPCODE:
1305 cbor_free (self); 1653 cbor_free (self);
1306 1654
1307PROTOTYPES: ENABLE 1655PROTOTYPES: ENABLE

Diff Legend

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