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 1.6 by root, Sat Oct 26 22:25:47 2013 UTC vs.
Revision 1.16 by root, Tue Oct 29 20:59:16 2013 UTC

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#include "ecb.h" 12#include "ecb.h"
13
14// compatibility with perl <5.18
15#ifndef HvNAMELEN_get
16# define HvNAMELEN_get(hv) strlen (HvNAME (hv))
17#endif
18#ifndef HvNAMELEN
19# define HvNAMELEN(hv) HvNAMELEN_get (hv)
20#endif
21#ifndef HvNAMEUTF8
22# define HvNAMEUTF8(hv) 0
23#endif
24
25// known tags
26enum cbor_tag
27{
28 // inofficial extensions (pending iana registration)
29 CBOR_TAG_PERL_OBJECT = 256,
30 CBOR_TAG_GENERIC_OBJECT = 257,
31
32 // rfc7049
33 CBOR_TAG_DATETIME = 0, // rfc4287, utf-8
34 CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any
35 CBOR_TAG_POS_BIGNUM = 2, // byte string
36 CBOR_TAG_NEG_BIGNUM = 3, // byte string
37 CBOR_TAG_DECIMAL = 4, // decimal fraction, array
38 CBOR_TAG_BIGFLOAT = 5, // array
39
40 CBOR_TAG_CONV_B64U = 21, // base64url, any
41 CBOR_TAG_CONV_B64 = 22, // base64, any
42 CBOR_TAG_CONV_HEX = 23, // base16, any
43 CBOR_TAG_CBOR = 24, // embedded cbor, byte string
44
45 CBOR_TAG_URI = 32, // URI rfc3986, utf-8
46 CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8
47 CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8
48 CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8
49 CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8
50
51 CBOR_TAG_MAGIC = 55799 // self-describe cbor
52};
13 53
14#define F_SHRINK 0x00000200UL 54#define F_SHRINK 0x00000200UL
15#define F_ALLOW_UNKNOWN 0x00002000UL 55#define F_ALLOW_UNKNOWN 0x00002000UL
16 56
17#define INIT_SIZE 32 // initial scalar size to be allocated 57#define INIT_SIZE 32 // initial scalar size to be allocated
31#else 71#else
32# define CBOR_SLOW 0 72# define CBOR_SLOW 0
33# define CBOR_STASH cbor_stash 73# define CBOR_STASH cbor_stash
34#endif 74#endif
35 75
36static HV *cbor_stash, *cbor_boolean_stash, *cbor_tagged_stash; // CBOR::XS:: 76static HV *cbor_stash, *types_boolean_stash, *types_error_stash, *cbor_tagged_stash; // CBOR::XS::
37static SV *cbor_true, *cbor_false; 77static SV *types_true, *types_false, *types_error, *sv_cbor;
38 78
39typedef struct { 79typedef struct {
40 U32 flags; 80 U32 flags;
41 U32 max_depth; 81 U32 max_depth;
42 STRLEN max_size; 82 STRLEN max_size;
230 SvGETMAGIC (sv); 270 SvGETMAGIC (sv);
231 svt = SvTYPE (sv); 271 svt = SvTYPE (sv);
232 272
233 if (ecb_expect_false (SvOBJECT (sv))) 273 if (ecb_expect_false (SvOBJECT (sv)))
234 { 274 {
235 HV *boolean_stash = !CBOR_SLOW || cbor_boolean_stash 275 HV *boolean_stash = !CBOR_SLOW || types_boolean_stash
236 ? cbor_boolean_stash 276 ? types_boolean_stash
237 : gv_stashpv ("CBOR::XS::Boolean", 1); 277 : gv_stashpv ("Types::Serialiser::Boolean", 1);
278 HV *error_stash = !CBOR_SLOW || types_error_stash
279 ? types_error_stash
280 : gv_stashpv ("Types::Serialiser::Error", 1);
238 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 281 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
239 ? cbor_tagged_stash 282 ? cbor_tagged_stash
240 : gv_stashpv ("CBOR::XS::Tagged" , 1); 283 : gv_stashpv ("CBOR::XS::Tagged" , 1);
241 284
285 HV *stash = SvSTASH (sv);
286 GV *method;
287
242 if (SvSTASH (sv) == boolean_stash) 288 if (stash == boolean_stash)
243 encode_ch (enc, SvIV (sv) ? 0xe0 | 21 : 0xe0 | 20); 289 encode_ch (enc, SvIV (sv) ? 0xe0 | 21 : 0xe0 | 20);
290 else if (stash == error_stash)
291 encode_ch (enc, 0xe0 | 23);
244 else if (SvSTASH (sv) == tagged_stash) 292 else if (stash == tagged_stash)
245 { 293 {
246 if (svt != SVt_PVAV) 294 if (svt != SVt_PVAV)
247 croak ("encountered CBOR::XS::Tagged object that isn't an array"); 295 croak ("encountered CBOR::XS::Tagged object that isn't an array");
248 296
249 encode_uint (enc, 0xc0, SvUV (*av_fetch ((AV *)sv, 0, 1))); 297 encode_uint (enc, 0xc0, SvUV (*av_fetch ((AV *)sv, 0, 1)));
250 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1)); 298 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1));
251 } 299 }
300 else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
301 {
302 dSP;
303
304 ENTER; SAVETMPS; PUSHMARK (SP);
305 // we re-bless the reference to get overload and other niceties right
306 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
307
308 PUTBACK;
309 // G_SCALAR ensures that return value is 1
310 call_sv ((SV *)GvCV (method), G_SCALAR);
311 SPAGAIN;
312
313 // catch this surprisingly common error
314 if (SvROK (TOPs) && SvRV (TOPs) == sv)
315 croak ("%s::TO_CBOR method returned same object as was passed instead of a new one", HvNAME (stash));
316
317 encode_sv (enc, POPs);
318
319 PUTBACK;
320
321 FREETMPS; LEAVE;
322 }
323 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)
324 {
325 dSP;
326
327 ENTER; SAVETMPS; PUSHMARK (SP);
328 EXTEND (SP, 2);
329 // we re-bless the reference to get overload and other niceties right
330 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
331 PUSHs (sv_cbor);
332
333 PUTBACK;
334 int count = call_sv ((SV *)GvCV (method), G_ARRAY);
335 SPAGAIN;
336
337 // catch this surprisingly common error
338 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)
339 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));
340
341 encode_uint (enc, 0xc0, CBOR_TAG_PERL_OBJECT);
342 encode_uint (enc, 0x80, count + 1);
343 encode_str (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
344
345 while (count)
346 encode_sv (enc, SP[1 - count--]);
347
348 PUTBACK;
349
350 FREETMPS; LEAVE;
351 }
252 else 352 else
253 {
254 // we re-bless the reference to get overload and other niceties right
255 GV *to_cbor = gv_fetchmethod_autoload (SvSTASH (sv), "TO_CBOR", 0);
256
257 if (to_cbor)
258 {
259 dSP;
260
261 ENTER; SAVETMPS; PUSHMARK (SP);
262 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv)));
263
264 // calling with G_SCALAR ensures that we always get a 1 return value
265 PUTBACK;
266 call_sv ((SV *)GvCV (to_cbor), G_SCALAR);
267 SPAGAIN;
268
269 // catch this surprisingly common error
270 if (SvROK (TOPs) && SvRV (TOPs) == sv)
271 croak ("%s::TO_CBOR method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv)));
272
273 sv = POPs;
274 PUTBACK;
275
276 encode_sv (enc, sv);
277
278 FREETMPS; LEAVE;
279 }
280 else
281 croak ("encountered object '%s', but no TO_CBOR method available on it", 353 croak ("encountered object '%s', but no TO_CBOR or FREEZE methods available on it",
282 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 354 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
283 }
284 } 355 }
285 else if (svt == SVt_PVHV) 356 else if (svt == SVt_PVHV)
286 encode_hv (enc, (HV *)sv); 357 encode_hv (enc, (HV *)sv);
287 else if (svt == SVt_PVAV) 358 else if (svt == SVt_PVAV)
288 encode_av (enc, (AV *)sv); 359 encode_av (enc, (AV *)sv);
513 SvREFCNT_dec (av); 584 SvREFCNT_dec (av);
514 DEC_DEC_DEPTH; 585 DEC_DEC_DEPTH;
515 return &PL_sv_undef; 586 return &PL_sv_undef;
516} 587}
517 588
589static void
590decode_he (dec_t *dec, HV *hv)
591{
592 // for speed reasons, we specialcase single-string
593 // byte or utf-8 strings as keys.
594
595 if (*dec->cur >= 0x40 && *dec->cur <= 0x40 + 27)
596 {
597 I32 len = decode_uint (dec);
598 char *key = (char *)dec->cur;
599
600 dec->cur += len;
601
602 hv_store (hv, key, len, decode_sv (dec), 0);
603 }
604 else if (*dec->cur >= 0x60 && *dec->cur <= 0x60 + 27)
605 {
606 I32 len = decode_uint (dec);
607 char *key = (char *)dec->cur;
608
609 dec->cur += len;
610
611 hv_store (hv, key, -len, decode_sv (dec), 0);
612 }
613 else
614 {
615 SV *k = decode_sv (dec);
616 SV *v = decode_sv (dec);
617
618 hv_store_ent (hv, k, v, 0);
619 SvREFCNT_dec (k);
620 }
621}
622
518static SV * 623static SV *
519decode_hv (dec_t *dec) 624decode_hv (dec_t *dec)
520{ 625{
521 HV *hv = newHV (); 626 HV *hv = newHV ();
522 627
534 { 639 {
535 ++dec->cur; 640 ++dec->cur;
536 break; 641 break;
537 } 642 }
538 643
539 SV *k = decode_sv (dec); 644 decode_he (dec, hv);
540 SV *v = decode_sv (dec);
541
542 hv_store_ent (hv, k, v, 0);
543 } 645 }
544 } 646 }
545 else 647 else
546 { 648 {
547 int len = decode_uint (dec); 649 int pairs = decode_uint (dec);
548 650
549 while (len--) 651 while (pairs--)
550 { 652 decode_he (dec, hv);
551 SV *k = decode_sv (dec);
552 SV *v = decode_sv (dec);
553
554 hv_store_ent (hv, k, v, 0);
555 }
556 } 653 }
557 654
558 DEC_DEC_DEPTH; 655 DEC_DEC_DEPTH;
559 return newRV_noinc ((SV *)hv); 656 return newRV_noinc ((SV *)hv);
560 657
612decode_tagged (dec_t *dec) 709decode_tagged (dec_t *dec)
613{ 710{
614 UV tag = decode_uint (dec); 711 UV tag = decode_uint (dec);
615 SV *sv = decode_sv (dec); 712 SV *sv = decode_sv (dec);
616 713
617 if (tag == 55799) // 2.4.5 Self-Describe CBOR 714 if (tag == CBOR_TAG_MAGIC)
618 return sv; 715 return sv;
716 else if (tag == CBOR_TAG_PERL_OBJECT)
717 {
718 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
719 ERR ("corrupted CBOR data (non-array perl object)");
619 720
721 AV *av = (AV *)SvRV (sv);
722 int len = av_len (av) + 1;
723 HV *stash = gv_stashsv (*av_fetch (av, 0, 1), 0);
724
725 if (!stash)
726 ERR ("cannot decode perl-object (package does not exist)");
727
728 GV *method = gv_fetchmethod_autoload (stash, "THAW", 0);
729
730 if (!method)
731 ERR ("cannot decode perl-object (package does not have a THAW method)");
732
733 dSP;
734
735 ENTER; SAVETMPS; PUSHMARK (SP);
736 EXTEND (SP, len + 1);
737 // we re-bless the reference to get overload and other niceties right
738 PUSHs (*av_fetch (av, 0, 1));
739 PUSHs (sv_cbor);
740
741 int i;
742
743 for (i = 1; i < len; ++i)
744 PUSHs (*av_fetch (av, i, 1));
745
746 PUTBACK;
747 call_sv ((SV *)GvCV (method), G_SCALAR);
748 SPAGAIN;
749
750 SvREFCNT_dec (sv);
751 sv = SvREFCNT_inc (POPs);
752
753 PUTBACK;
754
755 FREETMPS; LEAVE;
756
757 return sv;
758 }
759 else
760 {
620 AV *av = newAV (); 761 AV *av = newAV ();
621 av_push (av, newSVuv (tag)); 762 av_push (av, newSVuv (tag));
622 av_push (av, sv); 763 av_push (av, sv);
623 return newRV_noinc ((SV *)av); 764
765 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
766 ? cbor_tagged_stash
767 : gv_stashpv ("CBOR::XS::Tagged" , 1);
768
769 return sv_bless (newRV_noinc ((SV *)av), tagged_stash);
770 }
771
772fail:
773 SvREFCNT_dec (sv);
774 return &PL_sv_undef;
624} 775}
625 776
626static SV * 777static SV *
627decode_sv (dec_t *dec) 778decode_sv (dec_t *dec)
628{ 779{
647 case 7: // misc 798 case 7: // misc
648 switch (*dec->cur++ & 31) 799 switch (*dec->cur++ & 31)
649 { 800 {
650 case 20: 801 case 20:
651#if CBOR_SLOW 802#if CBOR_SLOW
652 cbor_false = get_bool ("CBOR::XS::false"); 803 types_false = get_bool ("Types::Serialiser::false");
653#endif 804#endif
654 return newSVsv (cbor_false); 805 return newSVsv (types_false);
655 case 21: 806 case 21:
656#if CBOR_SLOW 807#if CBOR_SLOW
657 cbor_true = get_bool ("CBOR::XS::true"); 808 types_true = get_bool ("Types::Serialiser::true");
658#endif 809#endif
659 return newSVsv (cbor_true); 810 return newSVsv (types_true);
660 case 22: 811 case 22:
661 return newSVsv (&PL_sv_undef); 812 return newSVsv (&PL_sv_undef);
813 case 23:
814#if CBOR_SLOW
815 types_error = get_bool ("Types::Serialiser::error");
816#endif
817 return newSVsv (types_error);
662 818
663 case 25: 819 case 25:
664 { 820 {
665 WANT (2); 821 WANT (2);
666 822
712static SV * 868static SV *
713decode_cbor (SV *string, CBOR *cbor, char **offset_return) 869decode_cbor (SV *string, CBOR *cbor, char **offset_return)
714{ 870{
715 dec_t dec; 871 dec_t dec;
716 SV *sv; 872 SV *sv;
873 STRLEN len;
874 char *data = SvPVbyte (string, len);
717 875
718 /* work around bugs in 5.10 where manipulating magic values
719 * makes perl ignore the magic in subsequent accesses.
720 * also make a copy of non-PV values, to get them into a clean
721 * state (SvPV should do that, but it's buggy, see below).
722 */
723 /*SvGETMAGIC (string);*/
724 if (SvMAGICAL (string) || !SvPOK (string))
725 string = sv_2mortal (newSVsv (string));
726
727 SvUPGRADE (string, SVt_PV);
728
729 /* work around a bug in perl 5.10, which causes SvCUR to fail an
730 * assertion with -DDEBUGGING, although SvCUR is documented to
731 * return the xpv_cur field which certainly exists after upgrading.
732 * according to nicholas clark, calling SvPOK fixes this.
733 * But it doesn't fix it, so try another workaround, call SvPV_nolen
734 * and hope for the best.
735 * Damnit, SvPV_nolen still trips over yet another assertion. This
736 * assertion business is seriously broken, try yet another workaround
737 * for the broken -DDEBUGGING.
738 */
739 {
740#ifdef DEBUGGING
741 STRLEN offset = SvOK (string) ? sv_len (string) : 0;
742#else
743 STRLEN offset = SvCUR (string);
744#endif
745
746 if (offset > cbor->max_size && cbor->max_size) 876 if (len > cbor->max_size && cbor->max_size)
747 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu", 877 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
748 (unsigned long)SvCUR (string), (unsigned long)cbor->max_size); 878 (unsigned long)len, (unsigned long)cbor->max_size);
749 }
750
751 sv_utf8_downgrade (string, 0);
752 879
753 dec.cbor = *cbor; 880 dec.cbor = *cbor;
754 dec.cur = (U8 *)SvPVX (string); 881 dec.cur = (U8 *)data;
755 dec.end = (U8 *)SvEND (string); 882 dec.end = (U8 *)data + len;
756 dec.err = 0; 883 dec.err = 0;
757 dec.depth = 0; 884 dec.depth = 0;
758 885
759 sv = decode_sv (&dec); 886 sv = decode_sv (&dec);
760 887
766 dec.err = "garbage after CBOR object"; 893 dec.err = "garbage after CBOR object";
767 894
768 if (dec.err) 895 if (dec.err)
769 { 896 {
770 SvREFCNT_dec (sv); 897 SvREFCNT_dec (sv);
771 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)SvPVX (string), (int)(uint8_t)*dec.cur); 898 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur);
772 } 899 }
773 900
774 sv = sv_2mortal (sv); 901 sv = sv_2mortal (sv);
775 902
776 return sv; 903 return sv;
782MODULE = CBOR::XS PACKAGE = CBOR::XS 909MODULE = CBOR::XS PACKAGE = CBOR::XS
783 910
784BOOT: 911BOOT:
785{ 912{
786 cbor_stash = gv_stashpv ("CBOR::XS" , 1); 913 cbor_stash = gv_stashpv ("CBOR::XS" , 1);
787 cbor_boolean_stash = gv_stashpv ("CBOR::XS::Boolean", 1);
788 cbor_tagged_stash = gv_stashpv ("CBOR::XS::Tagged" , 1); 914 cbor_tagged_stash = gv_stashpv ("CBOR::XS::Tagged" , 1);
789 915
790 cbor_true = get_bool ("CBOR::XS::true"); 916 types_boolean_stash = gv_stashpv ("Types::Serialiser::Boolean", 1);
791 cbor_false = get_bool ("CBOR::XS::false"); 917 types_error_stash = gv_stashpv ("Types::Serialiser::Error" , 1);
918
919 types_true = get_bool ("Types::Serialiser::true" );
920 types_false = get_bool ("Types::Serialiser::false");
921 types_error = get_bool ("Types::Serialiser::error");
922
923 sv_cbor = newSVpv ("CBOR", 0);
924 SvREADONLY_on (sv_cbor);
792} 925}
793 926
794PROTOTYPES: DISABLE 927PROTOTYPES: DISABLE
795 928
796void CLONE (...) 929void CLONE (...)
797 CODE: 930 CODE:
798 cbor_stash = 0; 931 cbor_stash = 0;
799 cbor_boolean_stash = 0;
800 cbor_tagged_stash = 0; 932 cbor_tagged_stash = 0;
933 types_error_stash = 0;
934 types_boolean_stash = 0;
801 935
802void new (char *klass) 936void new (char *klass)
803 PPCODE: 937 PPCODE:
804{ 938{
805 SV *pv = NEWSV (0, sizeof (CBOR)); 939 SV *pv = NEWSV (0, sizeof (CBOR));
852 CODE: 986 CODE:
853 RETVAL = self->max_size; 987 RETVAL = self->max_size;
854 OUTPUT: 988 OUTPUT:
855 RETVAL 989 RETVAL
856 990
857#if 0 //TODO
858
859void filter_cbor_object (CBOR *self, SV *cb = &PL_sv_undef)
860 PPCODE:
861{
862 SvREFCNT_dec (self->cb_object);
863 self->cb_object = SvOK (cb) ? newSVsv (cb) : 0;
864
865 XPUSHs (ST (0));
866}
867
868void filter_cbor_single_key_object (CBOR *self, SV *key, SV *cb = &PL_sv_undef)
869 PPCODE:
870{
871 if (!self->cb_sk_object)
872 self->cb_sk_object = newHV ();
873
874 if (SvOK (cb))
875 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);
876 else
877 {
878 hv_delete_ent (self->cb_sk_object, key, G_DISCARD, 0);
879
880 if (!HvKEYS (self->cb_sk_object))
881 {
882 SvREFCNT_dec (self->cb_sk_object);
883 self->cb_sk_object = 0;
884 }
885 }
886
887 XPUSHs (ST (0));
888}
889
890#endif
891
892void encode (CBOR *self, SV *scalar) 991void encode (CBOR *self, SV *scalar)
893 PPCODE: 992 PPCODE:
894 PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN; 993 PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN;
895 XPUSHs (scalar); 994 XPUSHs (scalar);
896 995
908 EXTEND (SP, 2); 1007 EXTEND (SP, 2);
909 PUSHs (sv); 1008 PUSHs (sv);
910 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); 1009 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
911} 1010}
912 1011
913#if 0
914
915void DESTROY (CBOR *self)
916 CODE:
917 SvREFCNT_dec (self->cb_sk_object);
918 SvREFCNT_dec (self->cb_object);
919
920#endif
921
922PROTOTYPES: ENABLE 1012PROTOTYPES: ENABLE
923 1013
924void encode_cbor (SV *scalar) 1014void encode_cbor (SV *scalar)
925 PPCODE: 1015 PPCODE:
926{ 1016{

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines