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 1.23 by root, Wed Nov 20 16:19:47 2013 UTC

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 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 = 24, // http://cbor.schmorp.de/perl-object
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
36
37 // rfc7049
38 CBOR_TAG_DATETIME = 0, // rfc4287, utf-8
39 CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any
40 CBOR_TAG_POS_BIGNUM = 2, // byte string
41 CBOR_TAG_NEG_BIGNUM = 3, // byte string
42 CBOR_TAG_DECIMAL = 4, // decimal fraction, array
43 CBOR_TAG_BIGFLOAT = 5, // array
44
45 CBOR_TAG_CONV_B64U = 21, // base64url, any
46 CBOR_TAG_CONV_B64 = 22, // base64, any
47 CBOR_TAG_CONV_HEX = 23, // base16, any
48 CBOR_TAG_CBOR = 24, // embedded cbor, byte string
49
50 CBOR_TAG_URI = 32, // URI rfc3986, utf-8
51 CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8
52 CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8
53 CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8
54 CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8
55
56 CBOR_TAG_MAGIC = 55799 // self-describe cbor
57};
58
14#define F_SHRINK 0x00000200UL 59#define F_SHRINK 0x00000001UL
15#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
16 63
17#define INIT_SIZE 32 // initial scalar size to be allocated 64#define INIT_SIZE 32 // initial scalar size to be allocated
18 65
19#define SB do { 66#define SB do {
20#define SE } while (0) 67#define SE } while (0)
31#else 78#else
32# define CBOR_SLOW 0 79# define CBOR_SLOW 0
33# define CBOR_STASH cbor_stash 80# define CBOR_STASH cbor_stash
34#endif 81#endif
35 82
36static HV *cbor_stash, *cbor_boolean_stash, *cbor_tagged_stash; // CBOR::XS:: 83static HV *cbor_stash, *types_boolean_stash, *types_error_stash, *cbor_tagged_stash; // CBOR::XS::
37static SV *cbor_true, *cbor_false; 84static SV *types_true, *types_false, *types_error, *sv_cbor;
38 85
39typedef struct { 86typedef struct {
40 U32 flags; 87 U32 flags;
41 U32 max_depth; 88 U32 max_depth;
42 STRLEN max_size; 89 STRLEN max_size;
76 SvPV_renew (sv, SvCUR (sv) + 1); 123 SvPV_renew (sv, SvCUR (sv) + 1);
77#endif 124#endif
78 } 125 }
79} 126}
80 127
81///////////////////////////////////////////////////////////////////////////// 128// minimum length of a string to be registered for stringref
82// fp hell 129ecb_inline int
83 130minimum_string_length (UV idx)
84//TODO 131{
132 return idx > 23
133 ? idx > 0xffU
134 ? idx > 0xffffU
135 ? idx > 0xffffffffU
136 ? 7
137 : 6
138 : 5
139 : 4
140 : 3;
141}
85 142
86///////////////////////////////////////////////////////////////////////////// 143/////////////////////////////////////////////////////////////////////////////
87// encoder 144// encoder
88 145
89// structure used for encoding CBOR 146// structure used for encoding CBOR
92 char *cur; // SvPVX (sv) + current output position 149 char *cur; // SvPVX (sv) + current output position
93 char *end; // SvEND (sv) 150 char *end; // SvEND (sv)
94 SV *sv; // result scalar 151 SV *sv; // result scalar
95 CBOR cbor; 152 CBOR cbor;
96 U32 depth; // recursion level 153 U32 depth; // recursion level
154 HV *stringref[2]; // string => index, or 0 ([0] = bytes, [1] = utf-8)
155 UV stringref_idx;
156 HV *shareable; // ptr => index, or 0
157 UV shareable_idx;
97} enc_t; 158} enc_t;
98 159
99ecb_inline void 160ecb_inline void
100need (enc_t *enc, STRLEN len) 161need (enc_t *enc, STRLEN len)
101{ 162{
153 *enc->cur++ = len >> 8; 214 *enc->cur++ = len >> 8;
154 *enc->cur++ = len; 215 *enc->cur++ = len;
155 } 216 }
156} 217}
157 218
219ecb_inline void
220encode_tag (enc_t *enc, UV tag)
221{
222 encode_uint (enc, 0xc0, tag);
223}
224
158static void 225static void
159encode_str (enc_t *enc, int utf8, char *str, STRLEN len) 226encode_str (enc_t *enc, int utf8, char *str, STRLEN len)
160{ 227{
228 if (ecb_expect_false (enc->cbor.flags & F_DEDUP_STRINGS))
229 {
230 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1);
231
232 if (SvOK (*svp))
233 {
234 // already registered, use stringref
235 encode_tag (enc, CBOR_TAG_STRINGREF);
236 encode_uint (enc, 0x00, SvUV (*svp));
237 return;
238 }
239 else if (len >= minimum_string_length (enc->stringref_idx))
240 {
241 // register only
242 sv_setuv (*svp, enc->stringref_idx);
243 ++enc->stringref_idx;
244 }
245 }
246
161 encode_uint (enc, utf8 ? 0x60 : 0x40, len); 247 encode_uint (enc, utf8 ? 0x60 : 0x40, len);
162 need (enc, len); 248 need (enc, len);
163 memcpy (enc->cur, str, len); 249 memcpy (enc->cur, str, len);
164 enc->cur += len; 250 enc->cur += len;
165} 251}
185 } 271 }
186 272
187 --enc->depth; 273 --enc->depth;
188} 274}
189 275
276ecb_inline void
277encode_he (enc_t *enc, HE *he)
278{
279}
280
190static void 281static void
191encode_hv (enc_t *enc, HV *hv) 282encode_hv (enc_t *enc, HV *hv)
192{ 283{
193 HE *he; 284 HE *he;
194 285
223 314
224// encode objects, arrays and special \0=false and \1=true values. 315// encode objects, arrays and special \0=false and \1=true values.
225static void 316static void
226encode_rv (enc_t *enc, SV *sv) 317encode_rv (enc_t *enc, SV *sv)
227{ 318{
228 svtype svt;
229
230 SvGETMAGIC (sv); 319 SvGETMAGIC (sv);
320
321 if (ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING)
322 && ecb_expect_false (SvREFCNT (sv) > 1))
323 {
324 if (!enc->shareable)
325 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
326
327 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
328
329 if (SvOK (*svp))
330 {
331 encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF);
332 encode_uint (enc, 0x00, SvUV (*svp));
333 return;
334 }
335 else
336 {
337 sv_setuv (*svp, enc->shareable_idx);
338 ++enc->shareable_idx;
339 encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE);
340 }
341 }
342
231 svt = SvTYPE (sv); 343 svtype svt = SvTYPE (sv);
232 344
233 if (ecb_expect_false (SvOBJECT (sv))) 345 if (ecb_expect_false (SvOBJECT (sv)))
234 { 346 {
235 HV *boolean_stash = !CBOR_SLOW || cbor_boolean_stash 347 HV *boolean_stash = !CBOR_SLOW || types_boolean_stash
236 ? cbor_boolean_stash 348 ? types_boolean_stash
237 : gv_stashpv ("CBOR::XS::Boolean", 1); 349 : gv_stashpv ("Types::Serialiser::Boolean", 1);
350 HV *error_stash = !CBOR_SLOW || types_error_stash
351 ? types_error_stash
352 : gv_stashpv ("Types::Serialiser::Error", 1);
238 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 353 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
239 ? cbor_tagged_stash 354 ? cbor_tagged_stash
240 : gv_stashpv ("CBOR::XS::Tagged" , 1); 355 : gv_stashpv ("CBOR::XS::Tagged" , 1);
241 356
357 HV *stash = SvSTASH (sv);
358 GV *method;
359
242 if (SvSTASH (sv) == boolean_stash) 360 if (stash == boolean_stash)
243 encode_ch (enc, SvIV (sv) ? 0xe0 | 21 : 0xe0 | 20); 361 encode_ch (enc, SvIV (sv) ? 0xe0 | 21 : 0xe0 | 20);
362 else if (stash == error_stash)
363 encode_ch (enc, 0xe0 | 23);
244 else if (SvSTASH (sv) == tagged_stash) 364 else if (stash == tagged_stash)
245 { 365 {
246 if (svt != SVt_PVAV) 366 if (svt != SVt_PVAV)
247 croak ("encountered CBOR::XS::Tagged object that isn't an array"); 367 croak ("encountered CBOR::XS::Tagged object that isn't an array");
248 368
249 encode_uint (enc, 0xc0, SvUV (*av_fetch ((AV *)sv, 0, 1))); 369 encode_uint (enc, 0xc0, SvUV (*av_fetch ((AV *)sv, 0, 1)));
250 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1)); 370 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1));
251 } 371 }
372 else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
373 {
374 dSP;
375
376 ENTER; SAVETMPS; PUSHMARK (SP);
377 // we re-bless the reference to get overload and other niceties right
378 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
379
380 PUTBACK;
381 // G_SCALAR ensures that return value is 1
382 call_sv ((SV *)GvCV (method), G_SCALAR);
383 SPAGAIN;
384
385 // catch this surprisingly common error
386 if (SvROK (TOPs) && SvRV (TOPs) == sv)
387 croak ("%s::TO_CBOR method returned same object as was passed instead of a new one", HvNAME (stash));
388
389 encode_sv (enc, POPs);
390
391 PUTBACK;
392
393 FREETMPS; LEAVE;
394 }
395 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)
396 {
397 dSP;
398
399 ENTER; SAVETMPS; PUSHMARK (SP);
400 EXTEND (SP, 2);
401 // we re-bless the reference to get overload and other niceties right
402 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
403 PUSHs (sv_cbor);
404
405 PUTBACK;
406 int count = call_sv ((SV *)GvCV (method), G_ARRAY);
407 SPAGAIN;
408
409 // catch this surprisingly common error
410 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)
411 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));
412
413 encode_tag (enc, CBOR_TAG_PERL_OBJECT);
414 encode_uint (enc, 0x80, count + 1);
415 encode_str (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
416
417 while (count)
418 encode_sv (enc, SP[1 - count--]);
419
420 PUTBACK;
421
422 FREETMPS; LEAVE;
423 }
252 else 424 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", 425 croak ("encountered object '%s', but no TO_CBOR or FREEZE methods available on it",
282 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 426 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
283 }
284 } 427 }
285 else if (svt == SVt_PVHV) 428 else if (svt == SVt_PVHV)
286 encode_hv (enc, (HV *)sv); 429 encode_hv (enc, (HV *)sv);
287 else if (svt == SVt_PVAV) 430 else if (svt == SVt_PVAV)
288 encode_av (enc, (AV *)sv); 431 encode_av (enc, (AV *)sv);
289 else if (svt < SVt_PVAV)
290 {
291 STRLEN len = 0;
292 char *pv = svt ? SvPV (sv, len) : 0;
293
294 if (len == 1 && *pv == '1')
295 encode_ch (enc, 0xe0 | 21);
296 else if (len == 1 && *pv == '0')
297 encode_ch (enc, 0xe0 | 20);
298 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
299 encode_ch (enc, 0xe0 | 23);
300 else
301 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
302 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
303 }
304 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
305 encode_ch (enc, 0xe0 | 23);
306 else 432 else
307 croak ("encountered %s, but CBOR can only represent references to arrays or hashes", 433 {
308 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 434 encode_tag (enc, CBOR_TAG_INDIRECTION);
435 encode_sv (enc, sv);
436 }
309} 437}
310 438
311static void 439static void
312encode_nv (enc_t *enc, SV *sv) 440encode_nv (enc_t *enc, SV *sv)
313{ 441{
378} 506}
379 507
380static SV * 508static SV *
381encode_cbor (SV *scalar, CBOR *cbor) 509encode_cbor (SV *scalar, CBOR *cbor)
382{ 510{
383 enc_t enc; 511 enc_t enc = { };
384 512
385 enc.cbor = *cbor; 513 enc.cbor = *cbor;
386 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 514 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
387 enc.cur = SvPVX (enc.sv); 515 enc.cur = SvPVX (enc.sv);
388 enc.end = SvEND (enc.sv); 516 enc.end = SvEND (enc.sv);
389 enc.depth = 0;
390 517
391 SvPOK_only (enc.sv); 518 SvPOK_only (enc.sv);
519
520 if (cbor->flags & F_DEDUP_STRINGS)
521 {
522 encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE);
523 enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ());
524 enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ());
525 }
526
392 encode_sv (&enc, scalar); 527 encode_sv (&enc, scalar);
393 528
394 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 529 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
395 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings 530 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
396 531
410 U8 *end; // end of input string 545 U8 *end; // end of input string
411 const char *err; // parse error, if != 0 546 const char *err; // parse error, if != 0
412 CBOR cbor; 547 CBOR cbor;
413 U32 depth; // recursion depth 548 U32 depth; // recursion depth
414 U32 maxdepth; // recursion depth limit 549 U32 maxdepth; // recursion depth limit
550 AV *shareable;
551 AV *stringref;
415} dec_t; 552} dec_t;
416 553
417#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE 554#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE
418 555
419#define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data") 556#define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data")
513 SvREFCNT_dec (av); 650 SvREFCNT_dec (av);
514 DEC_DEC_DEPTH; 651 DEC_DEC_DEPTH;
515 return &PL_sv_undef; 652 return &PL_sv_undef;
516} 653}
517 654
655static void
656decode_he (dec_t *dec, HV *hv)
657{
658 // for speed reasons, we specialcase single-string
659 // byte or utf-8 strings as keys, but only when !stringref
660
661 if (ecb_expect_true (!dec->stringref))
662 if (*dec->cur >= 0x40 && *dec->cur <= 0x40 + 27)
663 {
664 I32 len = decode_uint (dec);
665 char *key = (char *)dec->cur;
666
667 dec->cur += len;
668
669 if (ecb_expect_false (dec->stringref))
670 av_push (dec->stringref, newSVpvn (key, len));
671
672 hv_store (hv, key, len, decode_sv (dec), 0);
673
674 return;
675 }
676 else if (*dec->cur >= 0x60 && *dec->cur <= 0x60 + 27)
677 {
678 I32 len = decode_uint (dec);
679 char *key = (char *)dec->cur;
680
681 dec->cur += len;
682
683 if (ecb_expect_false (dec->stringref))
684 av_push (dec->stringref, newSVpvn_utf8 (key, len, 1));
685
686 hv_store (hv, key, -len, decode_sv (dec), 0);
687
688 return;
689 }
690
691 SV *k = decode_sv (dec);
692 SV *v = decode_sv (dec);
693
694 hv_store_ent (hv, k, v, 0);
695 SvREFCNT_dec (k);
696}
697
518static SV * 698static SV *
519decode_hv (dec_t *dec) 699decode_hv (dec_t *dec)
520{ 700{
521 HV *hv = newHV (); 701 HV *hv = newHV ();
522 702
534 { 714 {
535 ++dec->cur; 715 ++dec->cur;
536 break; 716 break;
537 } 717 }
538 718
539 SV *k = decode_sv (dec); 719 decode_he (dec, hv);
540 SV *v = decode_sv (dec);
541
542 hv_store_ent (hv, k, v, 0);
543 } 720 }
544 } 721 }
545 else 722 else
546 { 723 {
547 int len = decode_uint (dec); 724 int pairs = decode_uint (dec);
548 725
549 while (len--) 726 while (pairs--)
550 { 727 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 } 728 }
557 729
558 DEC_DEC_DEPTH; 730 DEC_DEC_DEPTH;
559 return newRV_noinc ((SV *)hv); 731 return newRV_noinc ((SV *)hv);
560 732
599 } 771 }
600 772
601 if (utf8) 773 if (utf8)
602 SvUTF8_on (sv); 774 SvUTF8_on (sv);
603 775
776 if (ecb_expect_false (dec->stringref)
777 && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1))
778 av_push (dec->stringref, SvREFCNT_inc_NN (sv));
779
604 return sv; 780 return sv;
605 781
606fail: 782fail:
607 SvREFCNT_dec (sv); 783 SvREFCNT_dec (sv);
608 return &PL_sv_undef; 784 return &PL_sv_undef;
609} 785}
610 786
611static SV * 787static SV *
612decode_tagged (dec_t *dec) 788decode_tagged (dec_t *dec)
613{ 789{
790 SV *sv = 0;
614 UV tag = decode_uint (dec); 791 UV tag = decode_uint (dec);
792
793 WANT (1);
794
795 switch (tag)
796 {
797 case CBOR_TAG_MAGIC:
798 sv = decode_sv (dec);
799 break;
800
801 case CBOR_TAG_INDIRECTION:
802 sv = newRV_noinc (decode_sv (dec));
803 break;
804
805 case CBOR_TAG_STRINGREF_NAMESPACE:
806 {
807 ENTER; SAVETMPS;
808
809 SAVESPTR (dec->stringref);
810 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ());
811
812 sv = decode_sv (dec);
813
814 FREETMPS; LEAVE;
815 }
816 break;
817
818 case CBOR_TAG_STRINGREF:
819 {
820 if ((*dec->cur >> 5) != 0)
821 ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
822
823 UV idx = decode_uint (dec);
824
825 if (!dec->stringref || (int)idx > AvFILLp (dec->stringref))
826 ERR ("corrupted CBOR data (stringref index out of bounds or outside namespace)");
827
828 sv = newSVsv (AvARRAY (dec->stringref)[idx]);
829 }
830 break;
831
832 case CBOR_TAG_VALUE_SHAREABLE:
833 {
834 if (ecb_expect_false (!dec->shareable))
835 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ());
836
837 sv = newSV (0);
838 av_push (dec->shareable, SvREFCNT_inc_NN (sv));
839
615 SV *sv = decode_sv (dec); 840 SV *osv = decode_sv (dec);
841 sv_setsv (sv, osv);
842 SvREFCNT_dec_NN (osv);
843 }
844 break;
616 845
617 if (tag == 55799) // 2.4.5 Self-Describe CBOR 846 case CBOR_TAG_VALUE_SHAREDREF:
847 {
848 if ((*dec->cur >> 5) != 0)
849 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
850
851 UV idx = decode_uint (dec);
852
853 if (!dec->shareable || (int)idx > AvFILLp (dec->shareable))
854 ERR ("corrupted CBOR data (sharedref index out of bounds)");
855
856 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]);
857 }
858 break;
859
860 case CBOR_TAG_PERL_OBJECT:
861 {
862 sv = decode_sv (dec);
863
864 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
865 ERR ("corrupted CBOR data (non-array perl object)");
866
867 AV *av = (AV *)SvRV (sv);
868 int len = av_len (av) + 1;
869 HV *stash = gv_stashsv (*av_fetch (av, 0, 1), 0);
870
871 if (!stash)
872 ERR ("cannot decode perl-object (package does not exist)");
873
874 GV *method = gv_fetchmethod_autoload (stash, "THAW", 0);
875
876 if (!method)
877 ERR ("cannot decode perl-object (package does not have a THAW method)");
878
879 dSP;
880
881 ENTER; SAVETMPS; PUSHMARK (SP);
882 EXTEND (SP, len + 1);
883 // we re-bless the reference to get overload and other niceties right
884 PUSHs (*av_fetch (av, 0, 1));
885 PUSHs (sv_cbor);
886
887 int i;
888
889 for (i = 1; i < len; ++i)
890 PUSHs (*av_fetch (av, i, 1));
891
892 PUTBACK;
893 call_sv ((SV *)GvCV (method), G_SCALAR | G_EVAL);
894 SPAGAIN;
895
896 if (SvTRUE (ERRSV))
897 {
898 FREETMPS; LEAVE;
899 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV))));
900 }
901
902 SvREFCNT_dec (sv);
903 sv = SvREFCNT_inc (POPs);
904
905 PUTBACK;
906
907 FREETMPS; LEAVE;
908 }
909 break;
910
911 default:
912 {
913 sv = decode_sv (dec);
914
915 AV *av = newAV ();
916 av_push (av, newSVuv (tag));
917 av_push (av, sv);
918
919 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
920 ? cbor_tagged_stash
921 : gv_stashpv ("CBOR::XS::Tagged" , 1);
922
923 sv = sv_bless (newRV_noinc ((SV *)av), tagged_stash);
924 }
925 break;
926 }
927
618 return sv; 928 return sv;
619 929
620 AV *av = newAV (); 930fail:
621 av_push (av, newSVuv (tag)); 931 SvREFCNT_dec (sv);
622 av_push (av, sv); 932 return &PL_sv_undef;
623 return newRV_noinc ((SV *)av);
624} 933}
625 934
626static SV * 935static SV *
627decode_sv (dec_t *dec) 936decode_sv (dec_t *dec)
628{ 937{
647 case 7: // misc 956 case 7: // misc
648 switch (*dec->cur++ & 31) 957 switch (*dec->cur++ & 31)
649 { 958 {
650 case 20: 959 case 20:
651#if CBOR_SLOW 960#if CBOR_SLOW
652 cbor_false = get_bool ("CBOR::XS::false"); 961 types_false = get_bool ("Types::Serialiser::false");
653#endif 962#endif
654 return newSVsv (cbor_false); 963 return newSVsv (types_false);
655 case 21: 964 case 21:
656#if CBOR_SLOW 965#if CBOR_SLOW
657 cbor_true = get_bool ("CBOR::XS::true"); 966 types_true = get_bool ("Types::Serialiser::true");
658#endif 967#endif
659 return newSVsv (cbor_true); 968 return newSVsv (types_true);
660 case 22: 969 case 22:
661 return newSVsv (&PL_sv_undef); 970 return newSVsv (&PL_sv_undef);
971 case 23:
972#if CBOR_SLOW
973 types_error = get_bool ("Types::Serialiser::error");
974#endif
975 return newSVsv (types_error);
662 976
663 case 25: 977 case 25:
664 { 978 {
665 WANT (2); 979 WANT (2);
666 980
710} 1024}
711 1025
712static SV * 1026static SV *
713decode_cbor (SV *string, CBOR *cbor, char **offset_return) 1027decode_cbor (SV *string, CBOR *cbor, char **offset_return)
714{ 1028{
715 dec_t dec; 1029 dec_t dec = { };
716 SV *sv; 1030 SV *sv;
1031 STRLEN len;
1032 char *data = SvPVbyte (string, len);
717 1033
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) 1034 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", 1035 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); 1036 (unsigned long)len, (unsigned long)cbor->max_size);
749 }
750
751 sv_utf8_downgrade (string, 0);
752 1037
753 dec.cbor = *cbor; 1038 dec.cbor = *cbor;
754 dec.cur = (U8 *)SvPVX (string); 1039 dec.cur = (U8 *)data;
755 dec.end = (U8 *)SvEND (string); 1040 dec.end = (U8 *)data + len;
756 dec.err = 0;
757 dec.depth = 0;
758 1041
759 sv = decode_sv (&dec); 1042 sv = decode_sv (&dec);
760 1043
761 if (offset_return) 1044 if (offset_return)
762 *offset_return = dec.cur; 1045 *offset_return = dec.cur;
766 dec.err = "garbage after CBOR object"; 1049 dec.err = "garbage after CBOR object";
767 1050
768 if (dec.err) 1051 if (dec.err)
769 { 1052 {
770 SvREFCNT_dec (sv); 1053 SvREFCNT_dec (sv);
771 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)SvPVX (string), (int)(uint8_t)*dec.cur); 1054 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur);
772 } 1055 }
773 1056
774 sv = sv_2mortal (sv); 1057 sv = sv_2mortal (sv);
775 1058
776 return sv; 1059 return sv;
782MODULE = CBOR::XS PACKAGE = CBOR::XS 1065MODULE = CBOR::XS PACKAGE = CBOR::XS
783 1066
784BOOT: 1067BOOT:
785{ 1068{
786 cbor_stash = gv_stashpv ("CBOR::XS" , 1); 1069 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); 1070 cbor_tagged_stash = gv_stashpv ("CBOR::XS::Tagged" , 1);
789 1071
790 cbor_true = get_bool ("CBOR::XS::true"); 1072 types_boolean_stash = gv_stashpv ("Types::Serialiser::Boolean", 1);
791 cbor_false = get_bool ("CBOR::XS::false"); 1073 types_error_stash = gv_stashpv ("Types::Serialiser::Error" , 1);
1074
1075 types_true = get_bool ("Types::Serialiser::true" );
1076 types_false = get_bool ("Types::Serialiser::false");
1077 types_error = get_bool ("Types::Serialiser::error");
1078
1079 sv_cbor = newSVpv ("CBOR", 0);
1080 SvREADONLY_on (sv_cbor);
792} 1081}
793 1082
794PROTOTYPES: DISABLE 1083PROTOTYPES: DISABLE
795 1084
796void CLONE (...) 1085void CLONE (...)
797 CODE: 1086 CODE:
798 cbor_stash = 0; 1087 cbor_stash = 0;
799 cbor_boolean_stash = 0;
800 cbor_tagged_stash = 0; 1088 cbor_tagged_stash = 0;
1089 types_error_stash = 0;
1090 types_boolean_stash = 0;
801 1091
802void new (char *klass) 1092void new (char *klass)
803 PPCODE: 1093 PPCODE:
804{ 1094{
805 SV *pv = NEWSV (0, sizeof (CBOR)); 1095 SV *pv = NEWSV (0, sizeof (CBOR));
813 1103
814void shrink (CBOR *self, int enable = 1) 1104void shrink (CBOR *self, int enable = 1)
815 ALIAS: 1105 ALIAS:
816 shrink = F_SHRINK 1106 shrink = F_SHRINK
817 allow_unknown = F_ALLOW_UNKNOWN 1107 allow_unknown = F_ALLOW_UNKNOWN
1108 allow_sharing = F_ALLOW_SHARING
1109 dedup_strings = F_DEDUP_STRINGS
818 PPCODE: 1110 PPCODE:
819{ 1111{
820 if (enable) 1112 if (enable)
821 self->flags |= ix; 1113 self->flags |= ix;
822 else 1114 else
827 1119
828void get_shrink (CBOR *self) 1120void get_shrink (CBOR *self)
829 ALIAS: 1121 ALIAS:
830 get_shrink = F_SHRINK 1122 get_shrink = F_SHRINK
831 get_allow_unknown = F_ALLOW_UNKNOWN 1123 get_allow_unknown = F_ALLOW_UNKNOWN
1124 get_allow_sharing = F_ALLOW_SHARING
1125 get_dedup_strings = F_DEDUP_STRINGS
832 PPCODE: 1126 PPCODE:
833 XPUSHs (boolSV (self->flags & ix)); 1127 XPUSHs (boolSV (self->flags & ix));
834 1128
835void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) 1129void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
836 PPCODE: 1130 PPCODE:
851int get_max_size (CBOR *self) 1145int get_max_size (CBOR *self)
852 CODE: 1146 CODE:
853 RETVAL = self->max_size; 1147 RETVAL = self->max_size;
854 OUTPUT: 1148 OUTPUT:
855 RETVAL 1149 RETVAL
856
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 1150
892void encode (CBOR *self, SV *scalar) 1151void encode (CBOR *self, SV *scalar)
893 PPCODE: 1152 PPCODE:
894 PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN; 1153 PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN;
895 XPUSHs (scalar); 1154 XPUSHs (scalar);
908 EXTEND (SP, 2); 1167 EXTEND (SP, 2);
909 PUSHs (sv); 1168 PUSHs (sv);
910 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); 1169 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
911} 1170}
912 1171
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 1172PROTOTYPES: ENABLE
923 1173
924void encode_cbor (SV *scalar) 1174void encode_cbor (SV *scalar)
925 PPCODE: 1175 PPCODE:
926{ 1176{

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines