ViewVC Help
View File | Revision Log | Show Annotations | Download File
/cvs/cvsroot/CBOR-XS/XS.xs
(Generate patch)

Comparing cvsroot/CBOR-XS/XS.xs (file contents):
Revision 1.17 by root, Tue Oct 29 22:04:52 2013 UTC vs.
Revision 1.19 by root, Wed Nov 20 02:03:09 2013 UTC

24 24
25// known tags 25// known tags
26enum cbor_tag 26enum cbor_tag
27{ 27{
28 // inofficial extensions (pending iana registration) 28 // inofficial extensions (pending iana registration)
29 CBOR_TAG_PERL_OBJECT = 256, 29 CBOR_TAG_PERL_OBJECT = 24, // http://cbor.schmorp.de/perl-object
30 CBOR_TAG_GENERIC_OBJECT = 257, 30 CBOR_TAG_GENERIC_OBJECT = 25, // http://cbor.schmorp.de/generic-object
31 CBOR_TAG_VALUE_SHAREABLE = 26, // http://cbor.schmorp.de/value-sharing
32 CBOR_TAG_VALUE_SHAREDREF = 27, // http://cbor.schmorp.de/value-sharing
33 CBOR_TAG_STRINGREF_NAMESPACE = 65537, // http://cbor.schmorp.de/stringref
34 CBOR_TAG_STRINGREF = 28, // http://cbor.schmorp.de/stringref
35 CBOR_TAG_INDIRECTION = 22098, // http://cbor.schmorp.de/indirection
31 36
32 // rfc7049 37 // rfc7049
33 CBOR_TAG_DATETIME = 0, // rfc4287, utf-8 38 CBOR_TAG_DATETIME = 0, // rfc4287, utf-8
34 CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any 39 CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any
35 CBOR_TAG_POS_BIGNUM = 2, // byte string 40 CBOR_TAG_POS_BIGNUM = 2, // byte string
49 CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8 54 CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8
50 55
51 CBOR_TAG_MAGIC = 55799 // self-describe cbor 56 CBOR_TAG_MAGIC = 55799 // self-describe cbor
52}; 57};
53 58
54#define F_SHRINK 0x00000200UL 59#define F_SHRINK 0x00000001UL
55#define F_ALLOW_UNKNOWN 0x00002000UL 60#define F_ALLOW_UNKNOWN 0x00000002UL
61#define F_ALLOW_SHARING 0x00000004UL //TODO
62#define F_DEDUP_STRINGS 0x00000008UL //TODO
63#define F_DEDUP_KEYS 0x00000010UL //TODO
56 64
57#define INIT_SIZE 32 // initial scalar size to be allocated 65#define INIT_SIZE 32 // initial scalar size to be allocated
58 66
59#define SB do { 67#define SB do {
60#define SE } while (0) 68#define SE } while (0)
117#endif 125#endif
118 } 126 }
119} 127}
120 128
121///////////////////////////////////////////////////////////////////////////// 129/////////////////////////////////////////////////////////////////////////////
122// fp hell
123
124//TODO
125
126/////////////////////////////////////////////////////////////////////////////
127// encoder 130// encoder
128 131
129// structure used for encoding CBOR 132// structure used for encoding CBOR
130typedef struct 133typedef struct
131{ 134{
132 char *cur; // SvPVX (sv) + current output position 135 char *cur; // SvPVX (sv) + current output position
133 char *end; // SvEND (sv) 136 char *end; // SvEND (sv)
134 SV *sv; // result scalar 137 SV *sv; // result scalar
135 CBOR cbor; 138 CBOR cbor;
136 U32 depth; // recursion level 139 U32 depth; // recursion level
140 HV *stringref; // string => index, or 0
141 HV *shareable; // ptr => index, or 0
142 UV shareable_idx;
137} enc_t; 143} enc_t;
138 144
139ecb_inline void 145ecb_inline void
140need (enc_t *enc, STRLEN len) 146need (enc_t *enc, STRLEN len)
141{ 147{
202 need (enc, len); 208 need (enc, len);
203 memcpy (enc->cur, str, len); 209 memcpy (enc->cur, str, len);
204 enc->cur += len; 210 enc->cur += len;
205} 211}
206 212
213ecb_inline void
214encode_tag (enc_t *enc, UV tag)
215{
216 encode_uint (enc, 0xc0, tag);
217}
218
207static void encode_sv (enc_t *enc, SV *sv); 219static void encode_sv (enc_t *enc, SV *sv);
208 220
209static void 221static void
210encode_av (enc_t *enc, AV *av) 222encode_av (enc_t *enc, AV *av)
211{ 223{
263 275
264// encode objects, arrays and special \0=false and \1=true values. 276// encode objects, arrays and special \0=false and \1=true values.
265static void 277static void
266encode_rv (enc_t *enc, SV *sv) 278encode_rv (enc_t *enc, SV *sv)
267{ 279{
268 svtype svt;
269
270 SvGETMAGIC (sv); 280 SvGETMAGIC (sv);
281
282 if (ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING)
283 && ecb_expect_false (SvREFCNT (sv) > 1))
284 {
285 if (!enc->shareable)
286 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
287
288 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
289
290 if (SvOK (*svp))
291 {
292 encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF);
293 encode_uint (enc, 0x00, SvUV (*svp));
294 return;
295 }
296 else
297 {
298 sv_setuv (*svp, enc->shareable_idx);
299 ++enc->shareable_idx;
300 encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE);
301 }
302 }
303
271 svt = SvTYPE (sv); 304 svtype svt = SvTYPE (sv);
272 305
273 if (ecb_expect_false (SvOBJECT (sv))) 306 if (ecb_expect_false (SvOBJECT (sv)))
274 { 307 {
275 HV *boolean_stash = !CBOR_SLOW || types_boolean_stash 308 HV *boolean_stash = !CBOR_SLOW || types_boolean_stash
276 ? types_boolean_stash 309 ? types_boolean_stash
336 369
337 // catch this surprisingly common error 370 // catch this surprisingly common error
338 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv) 371 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)); 372 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));
340 373
341 encode_uint (enc, 0xc0, CBOR_TAG_PERL_OBJECT); 374 encode_tag (enc, CBOR_TAG_PERL_OBJECT);
342 encode_uint (enc, 0x80, count + 1); 375 encode_uint (enc, 0x80, count + 1);
343 encode_str (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash)); 376 encode_str (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
344 377
345 while (count) 378 while (count)
346 encode_sv (enc, SP[1 - count--]); 379 encode_sv (enc, SP[1 - count--]);
355 } 388 }
356 else if (svt == SVt_PVHV) 389 else if (svt == SVt_PVHV)
357 encode_hv (enc, (HV *)sv); 390 encode_hv (enc, (HV *)sv);
358 else if (svt == SVt_PVAV) 391 else if (svt == SVt_PVAV)
359 encode_av (enc, (AV *)sv); 392 encode_av (enc, (AV *)sv);
360 else if (svt < SVt_PVAV)
361 {
362 STRLEN len = 0;
363 char *pv = svt ? SvPV (sv, len) : 0;
364
365 if (len == 1 && *pv == '1')
366 encode_ch (enc, 0xe0 | 21);
367 else if (len == 1 && *pv == '0')
368 encode_ch (enc, 0xe0 | 20);
369 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
370 encode_ch (enc, 0xe0 | 23);
371 else
372 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
373 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
374 }
375 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
376 encode_ch (enc, 0xe0 | 23);
377 else 393 else
378 croak ("encountered %s, but CBOR can only represent references to arrays or hashes", 394 {
379 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 395 encode_tag (enc, CBOR_TAG_INDIRECTION);
396 encode_sv (enc, sv);
397 }
380} 398}
381 399
382static void 400static void
383encode_nv (enc_t *enc, SV *sv) 401encode_nv (enc_t *enc, SV *sv)
384{ 402{
449} 467}
450 468
451static SV * 469static SV *
452encode_cbor (SV *scalar, CBOR *cbor) 470encode_cbor (SV *scalar, CBOR *cbor)
453{ 471{
454 enc_t enc; 472 enc_t enc = { };
455 473
456 enc.cbor = *cbor; 474 enc.cbor = *cbor;
457 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 475 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
458 enc.cur = SvPVX (enc.sv); 476 enc.cur = SvPVX (enc.sv);
459 enc.end = SvEND (enc.sv); 477 enc.end = SvEND (enc.sv);
460 enc.depth = 0;
461 478
462 SvPOK_only (enc.sv); 479 SvPOK_only (enc.sv);
463 encode_sv (&enc, scalar); 480 encode_sv (&enc, scalar);
464 481
465 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 482 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
481 U8 *end; // end of input string 498 U8 *end; // end of input string
482 const char *err; // parse error, if != 0 499 const char *err; // parse error, if != 0
483 CBOR cbor; 500 CBOR cbor;
484 U32 depth; // recursion depth 501 U32 depth; // recursion depth
485 U32 maxdepth; // recursion depth limit 502 U32 maxdepth; // recursion depth limit
503 AV *shareable;
486} dec_t; 504} dec_t;
487 505
488#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE 506#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE
489 507
490#define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data") 508#define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data")
706} 724}
707 725
708static SV * 726static SV *
709decode_tagged (dec_t *dec) 727decode_tagged (dec_t *dec)
710{ 728{
729 SV *sv = 0;
711 UV tag = decode_uint (dec); 730 UV tag = decode_uint (dec);
712 SV *sv = decode_sv (dec);
713 731
714 if (tag == CBOR_TAG_MAGIC) 732 WANT (1);
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)");
720 733
721 AV *av = (AV *)SvRV (sv); 734 switch (tag)
722 int len = av_len (av) + 1; 735 {
723 HV *stash = gv_stashsv (*av_fetch (av, 0, 1), 0); 736 case CBOR_TAG_MAGIC:
737 return decode_sv (dec);
724 738
725 if (!stash) 739 case CBOR_TAG_INDIRECTION:
726 ERR ("cannot decode perl-object (package does not exist)"); 740 return newRV_noinc (decode_sv (dec));
727 741
728 GV *method = gv_fetchmethod_autoload (stash, "THAW", 0); 742 case CBOR_TAG_VALUE_SHAREABLE:
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 | G_EVAL);
748 SPAGAIN;
749
750 if (SvTRUE (ERRSV))
751 { 743 {
744 if (ecb_expect_false (!dec->shareable))
745 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ());
746
747 sv = newSV (0);
748 av_push (dec->shareable, SvREFCNT_inc_NN (sv));
749
750 SV *osv = decode_sv (dec);
751 sv_setsv (sv, osv);
752 SvREFCNT_dec_NN (osv);
753 }
754
755 return sv;
756
757 case CBOR_TAG_VALUE_SHAREDREF:
758 {
759 if ((*dec->cur >> 5) != 0)
760 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
761
762 UV idx = decode_uint (dec);
763
764 if (!dec->shareable || idx > AvFILLp (dec->shareable))
765 ERR ("corrupted CBOR data (sharedref index out of bounds)");
766
767 return SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]);
768 }
769
770 case CBOR_TAG_PERL_OBJECT:
771 {
772 sv = decode_sv (dec);
773
774 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
775 ERR ("corrupted CBOR data (non-array perl object)");
776
777 AV *av = (AV *)SvRV (sv);
778 int len = av_len (av) + 1;
779 HV *stash = gv_stashsv (*av_fetch (av, 0, 1), 0);
780
781 if (!stash)
782 ERR ("cannot decode perl-object (package does not exist)");
783
784 GV *method = gv_fetchmethod_autoload (stash, "THAW", 0);
785
786 if (!method)
787 ERR ("cannot decode perl-object (package does not have a THAW method)");
788
789 dSP;
790
791 ENTER; SAVETMPS; PUSHMARK (SP);
792 EXTEND (SP, len + 1);
793 // we re-bless the reference to get overload and other niceties right
794 PUSHs (*av_fetch (av, 0, 1));
795 PUSHs (sv_cbor);
796
797 int i;
798
799 for (i = 1; i < len; ++i)
800 PUSHs (*av_fetch (av, i, 1));
801
802 PUTBACK;
803 call_sv ((SV *)GvCV (method), G_SCALAR | G_EVAL);
804 SPAGAIN;
805
806 if (SvTRUE (ERRSV))
807 {
808 FREETMPS; LEAVE;
809 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV))));
810 }
811
812 SvREFCNT_dec (sv);
813 sv = SvREFCNT_inc (POPs);
814
815 PUTBACK;
816
752 FREETMPS; LEAVE; 817 FREETMPS; LEAVE;
753 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV)))); 818
819 return sv;
754 } 820 }
755 821
756 SvREFCNT_dec (sv); 822 default:
757 sv = SvREFCNT_inc (POPs); 823 {
824 sv = decode_sv (dec);
758 825
759 PUTBACK;
760
761 FREETMPS; LEAVE;
762
763 return sv;
764 }
765 else
766 {
767 AV *av = newAV (); 826 AV *av = newAV ();
768 av_push (av, newSVuv (tag)); 827 av_push (av, newSVuv (tag));
769 av_push (av, sv); 828 av_push (av, sv);
770 829
771 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 830 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
772 ? cbor_tagged_stash 831 ? cbor_tagged_stash
773 : gv_stashpv ("CBOR::XS::Tagged" , 1); 832 : gv_stashpv ("CBOR::XS::Tagged" , 1);
774 833
775 return sv_bless (newRV_noinc ((SV *)av), tagged_stash); 834 return sv_bless (newRV_noinc ((SV *)av), tagged_stash);
835 }
776 } 836 }
777 837
778fail: 838fail:
779 SvREFCNT_dec (sv); 839 SvREFCNT_dec (sv);
780 return &PL_sv_undef; 840 return &PL_sv_undef;
872} 932}
873 933
874static SV * 934static SV *
875decode_cbor (SV *string, CBOR *cbor, char **offset_return) 935decode_cbor (SV *string, CBOR *cbor, char **offset_return)
876{ 936{
877 dec_t dec; 937 dec_t dec = { };
878 SV *sv; 938 SV *sv;
879 STRLEN len; 939 STRLEN len;
880 char *data = SvPVbyte (string, len); 940 char *data = SvPVbyte (string, len);
881 941
882 if (len > cbor->max_size && cbor->max_size) 942 if (len > cbor->max_size && cbor->max_size)
884 (unsigned long)len, (unsigned long)cbor->max_size); 944 (unsigned long)len, (unsigned long)cbor->max_size);
885 945
886 dec.cbor = *cbor; 946 dec.cbor = *cbor;
887 dec.cur = (U8 *)data; 947 dec.cur = (U8 *)data;
888 dec.end = (U8 *)data + len; 948 dec.end = (U8 *)data + len;
889 dec.err = 0;
890 dec.depth = 0;
891 949
892 sv = decode_sv (&dec); 950 sv = decode_sv (&dec);
893 951
894 if (offset_return) 952 if (offset_return)
895 *offset_return = dec.cur; 953 *offset_return = dec.cur;
953 1011
954void shrink (CBOR *self, int enable = 1) 1012void shrink (CBOR *self, int enable = 1)
955 ALIAS: 1013 ALIAS:
956 shrink = F_SHRINK 1014 shrink = F_SHRINK
957 allow_unknown = F_ALLOW_UNKNOWN 1015 allow_unknown = F_ALLOW_UNKNOWN
1016 allow_sharing = F_ALLOW_SHARING
1017 dedup_keys = F_DEDUP_KEYS
1018 dedup_strings = F_DEDUP_STRINGS
958 PPCODE: 1019 PPCODE:
959{ 1020{
960 if (enable) 1021 if (enable)
961 self->flags |= ix; 1022 self->flags |= ix;
962 else 1023 else
967 1028
968void get_shrink (CBOR *self) 1029void get_shrink (CBOR *self)
969 ALIAS: 1030 ALIAS:
970 get_shrink = F_SHRINK 1031 get_shrink = F_SHRINK
971 get_allow_unknown = F_ALLOW_UNKNOWN 1032 get_allow_unknown = F_ALLOW_UNKNOWN
1033 get_allow_sharing = F_ALLOW_SHARING
1034 get_dedup_keys = F_DEDUP_KEYS
1035 get_dedup_strings = F_DEDUP_STRINGS
972 PPCODE: 1036 PPCODE:
973 XPUSHs (boolSV (self->flags & ix)); 1037 XPUSHs (boolSV (self->flags & ix));
974 1038
975void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) 1039void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
976 PPCODE: 1040 PPCODE:

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines