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 1.8 by root, Sun Oct 27 10:12:01 2013 UTC vs.
Revision 1.24 by root, Wed Nov 20 16:29:02 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// known tags, rfc7049 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
15enum cbor_tag 26enum cbor_tag
16{ 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
17 CBOR_TAG_DATETIME = 0, // rfc4287, utf-8 38 CBOR_TAG_DATETIME = 0, // rfc4287, utf-8
18 CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any 39 CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any
19 CBOR_TAG_POS_BIGNUM = 2, // byte string 40 CBOR_TAG_POS_BIGNUM = 2, // byte string
20 CBOR_TAG_NEG_BIGNUM = 3, // byte string 41 CBOR_TAG_NEG_BIGNUM = 3, // byte string
21 CBOR_TAG_DECIMAL = 4, // decimal fraction, array 42 CBOR_TAG_DECIMAL = 4, // decimal fraction, array
22 CBOR_TAG_BIGFLOAT = 5, // array 43 CBOR_TAG_BIGFLOAT = 5, // array
23 44
24 CBOR_TAG_CONV_B64U = 21, // base64url, any 45 CBOR_TAG_CONV_B64U = 21, // base64url, any
25 CBOR_TAG_CONV_B64 = 22, // base64, any 46 CBOR_TAG_CONV_B64 = 22, // base64, any
26 CBOR_TAG_CONV_HEX = 23, // base16, any 47 CBOR_TAG_CONV_HEX = 23, // base16, any
27 CBOR_TAG_CBOR = 24, // embedded cbor, byte string 48 CBOR_TAG_CBOR = 24, // embedded cbor, byte string
28 49
29 CBOR_TAG_URI = 32, // URI rfc3986, utf-8 50 CBOR_TAG_URI = 32, // URI rfc3986, utf-8
30 CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8 51 CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8
31 CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8 52 CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8
32 CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8 53 CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8
33 CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8 54 CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8
34 55
35 CBOR_TAG_MAGIC = 55799 56 CBOR_TAG_MAGIC = 55799 // self-describe cbor
36}; 57};
37 58
38#define F_SHRINK 0x00000200UL 59#define F_SHRINK 0x00000001UL
39#define F_ALLOW_UNKNOWN 0x00002000UL 60#define F_ALLOW_UNKNOWN 0x00000002UL
61#define F_ALLOW_SHARING 0x00000004UL //TODO
62#define F_ALLOW_STRINGREF 0x00000008UL //TODO
40 63
41#define INIT_SIZE 32 // initial scalar size to be allocated 64#define INIT_SIZE 32 // initial scalar size to be allocated
42 65
43#define SB do { 66#define SB do {
44#define SE } while (0) 67#define SE } while (0)
55#else 78#else
56# define CBOR_SLOW 0 79# define CBOR_SLOW 0
57# define CBOR_STASH cbor_stash 80# define CBOR_STASH cbor_stash
58#endif 81#endif
59 82
60static 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::
61static SV *cbor_true, *cbor_false; 84static SV *types_true, *types_false, *types_error, *sv_cbor;
62 85
63typedef struct { 86typedef struct {
64 U32 flags; 87 U32 flags;
65 U32 max_depth; 88 U32 max_depth;
66 STRLEN max_size; 89 STRLEN max_size;
100 SvPV_renew (sv, SvCUR (sv) + 1); 123 SvPV_renew (sv, SvCUR (sv) + 1);
101#endif 124#endif
102 } 125 }
103} 126}
104 127
105///////////////////////////////////////////////////////////////////////////// 128// minimum length of a string to be registered for stringref
106// fp hell 129ecb_inline int
107 130minimum_string_length (UV idx)
108//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}
109 142
110///////////////////////////////////////////////////////////////////////////// 143/////////////////////////////////////////////////////////////////////////////
111// encoder 144// encoder
112 145
113// structure used for encoding CBOR 146// structure used for encoding CBOR
116 char *cur; // SvPVX (sv) + current output position 149 char *cur; // SvPVX (sv) + current output position
117 char *end; // SvEND (sv) 150 char *end; // SvEND (sv)
118 SV *sv; // result scalar 151 SV *sv; // result scalar
119 CBOR cbor; 152 CBOR cbor;
120 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;
121} enc_t; 158} enc_t;
122 159
123ecb_inline void 160ecb_inline void
124need (enc_t *enc, STRLEN len) 161need (enc_t *enc, STRLEN len)
125{ 162{
177 *enc->cur++ = len >> 8; 214 *enc->cur++ = len >> 8;
178 *enc->cur++ = len; 215 *enc->cur++ = len;
179 } 216 }
180} 217}
181 218
219ecb_inline void
220encode_tag (enc_t *enc, UV tag)
221{
222 encode_uint (enc, 0xc0, tag);
223}
224
182static void 225static void
183encode_str (enc_t *enc, int utf8, char *str, STRLEN len) 226encode_str (enc_t *enc, int utf8, char *str, STRLEN len)
184{ 227{
228 if (ecb_expect_false (enc->cbor.flags & F_ALLOW_STRINGREF))
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
185 encode_uint (enc, utf8 ? 0x60 : 0x40, len); 247 encode_uint (enc, utf8 ? 0x60 : 0x40, len);
186 need (enc, len); 248 need (enc, len);
187 memcpy (enc->cur, str, len); 249 memcpy (enc->cur, str, len);
188 enc->cur += len; 250 enc->cur += len;
189} 251}
209 } 271 }
210 272
211 --enc->depth; 273 --enc->depth;
212} 274}
213 275
276ecb_inline void
277encode_he (enc_t *enc, HE *he)
278{
279}
280
214static void 281static void
215encode_hv (enc_t *enc, HV *hv) 282encode_hv (enc_t *enc, HV *hv)
216{ 283{
217 HE *he; 284 HE *he;
218 285
247 314
248// encode objects, arrays and special \0=false and \1=true values. 315// encode objects, arrays and special \0=false and \1=true values.
249static void 316static void
250encode_rv (enc_t *enc, SV *sv) 317encode_rv (enc_t *enc, SV *sv)
251{ 318{
252 svtype svt;
253
254 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
255 svt = SvTYPE (sv); 343 svtype svt = SvTYPE (sv);
256 344
257 if (ecb_expect_false (SvOBJECT (sv))) 345 if (ecb_expect_false (SvOBJECT (sv)))
258 { 346 {
259 HV *boolean_stash = !CBOR_SLOW || cbor_boolean_stash 347 HV *boolean_stash = !CBOR_SLOW || types_boolean_stash
260 ? cbor_boolean_stash 348 ? types_boolean_stash
261 : 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);
262 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 353 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
263 ? cbor_tagged_stash 354 ? cbor_tagged_stash
264 : gv_stashpv ("CBOR::XS::Tagged" , 1); 355 : gv_stashpv ("CBOR::XS::Tagged" , 1);
265 356
357 HV *stash = SvSTASH (sv);
358 GV *method;
359
266 if (SvSTASH (sv) == boolean_stash) 360 if (stash == boolean_stash)
267 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);
268 else if (SvSTASH (sv) == tagged_stash) 364 else if (stash == tagged_stash)
269 { 365 {
270 if (svt != SVt_PVAV) 366 if (svt != SVt_PVAV)
271 croak ("encountered CBOR::XS::Tagged object that isn't an array"); 367 croak ("encountered CBOR::XS::Tagged object that isn't an array");
272 368
273 encode_uint (enc, 0xc0, SvUV (*av_fetch ((AV *)sv, 0, 1))); 369 encode_uint (enc, 0xc0, SvUV (*av_fetch ((AV *)sv, 0, 1)));
274 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1)); 370 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1));
275 } 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 }
276 else 424 else
277 {
278 // we re-bless the reference to get overload and other niceties right
279 GV *to_cbor = gv_fetchmethod_autoload (SvSTASH (sv), "TO_CBOR", 0);
280
281 if (to_cbor)
282 {
283 dSP;
284
285 ENTER; SAVETMPS; PUSHMARK (SP);
286 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv)));
287
288 // calling with G_SCALAR ensures that we always get a 1 return value
289 PUTBACK;
290 call_sv ((SV *)GvCV (to_cbor), G_SCALAR);
291 SPAGAIN;
292
293 // catch this surprisingly common error
294 if (SvROK (TOPs) && SvRV (TOPs) == sv)
295 croak ("%s::TO_CBOR method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv)));
296
297 sv = POPs;
298 PUTBACK;
299
300 encode_sv (enc, sv);
301
302 FREETMPS; LEAVE;
303 }
304 else
305 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",
306 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 426 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
307 }
308 } 427 }
309 else if (svt == SVt_PVHV) 428 else if (svt == SVt_PVHV)
310 encode_hv (enc, (HV *)sv); 429 encode_hv (enc, (HV *)sv);
311 else if (svt == SVt_PVAV) 430 else if (svt == SVt_PVAV)
312 encode_av (enc, (AV *)sv); 431 encode_av (enc, (AV *)sv);
313 else if (svt < SVt_PVAV)
314 {
315 STRLEN len = 0;
316 char *pv = svt ? SvPV (sv, len) : 0;
317
318 if (len == 1 && *pv == '1')
319 encode_ch (enc, 0xe0 | 21);
320 else if (len == 1 && *pv == '0')
321 encode_ch (enc, 0xe0 | 20);
322 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
323 encode_ch (enc, 0xe0 | 23);
324 else
325 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
326 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
327 }
328 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
329 encode_ch (enc, 0xe0 | 23);
330 else 432 else
331 croak ("encountered %s, but CBOR can only represent references to arrays or hashes", 433 {
332 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 434 encode_tag (enc, CBOR_TAG_INDIRECTION);
435 encode_sv (enc, sv);
436 }
333} 437}
334 438
335static void 439static void
336encode_nv (enc_t *enc, SV *sv) 440encode_nv (enc_t *enc, SV *sv)
337{ 441{
402} 506}
403 507
404static SV * 508static SV *
405encode_cbor (SV *scalar, CBOR *cbor) 509encode_cbor (SV *scalar, CBOR *cbor)
406{ 510{
407 enc_t enc; 511 enc_t enc = { };
408 512
409 enc.cbor = *cbor; 513 enc.cbor = *cbor;
410 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 514 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
411 enc.cur = SvPVX (enc.sv); 515 enc.cur = SvPVX (enc.sv);
412 enc.end = SvEND (enc.sv); 516 enc.end = SvEND (enc.sv);
413 enc.depth = 0;
414 517
415 SvPOK_only (enc.sv); 518 SvPOK_only (enc.sv);
519
520 if (cbor->flags & F_ALLOW_STRINGREF)
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
416 encode_sv (&enc, scalar); 527 encode_sv (&enc, scalar);
417 528
418 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 529 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
419 *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
420 531
434 U8 *end; // end of input string 545 U8 *end; // end of input string
435 const char *err; // parse error, if != 0 546 const char *err; // parse error, if != 0
436 CBOR cbor; 547 CBOR cbor;
437 U32 depth; // recursion depth 548 U32 depth; // recursion depth
438 U32 maxdepth; // recursion depth limit 549 U32 maxdepth; // recursion depth limit
550 AV *shareable;
551 AV *stringref;
439} dec_t; 552} dec_t;
440 553
441#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
442 555
443#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")
537 SvREFCNT_dec (av); 650 SvREFCNT_dec (av);
538 DEC_DEC_DEPTH; 651 DEC_DEC_DEPTH;
539 return &PL_sv_undef; 652 return &PL_sv_undef;
540} 653}
541 654
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
542static SV * 698static SV *
543decode_hv (dec_t *dec) 699decode_hv (dec_t *dec)
544{ 700{
545 HV *hv = newHV (); 701 HV *hv = newHV ();
546 702
558 { 714 {
559 ++dec->cur; 715 ++dec->cur;
560 break; 716 break;
561 } 717 }
562 718
563 SV *k = decode_sv (dec); 719 decode_he (dec, hv);
564 SV *v = decode_sv (dec);
565
566 hv_store_ent (hv, k, v, 0);
567 } 720 }
568 } 721 }
569 else 722 else
570 { 723 {
571 int len = decode_uint (dec); 724 int pairs = decode_uint (dec);
572 725
573 while (len--) 726 while (pairs--)
574 { 727 decode_he (dec, hv);
575 SV *k = decode_sv (dec);
576 SV *v = decode_sv (dec);
577
578 hv_store_ent (hv, k, v, 0);
579 }
580 } 728 }
581 729
582 DEC_DEC_DEPTH; 730 DEC_DEC_DEPTH;
583 return newRV_noinc ((SV *)hv); 731 return newRV_noinc ((SV *)hv);
584 732
623 } 771 }
624 772
625 if (utf8) 773 if (utf8)
626 SvUTF8_on (sv); 774 SvUTF8_on (sv);
627 775
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
628 return sv; 780 return sv;
629 781
630fail: 782fail:
631 SvREFCNT_dec (sv); 783 SvREFCNT_dec (sv);
632 return &PL_sv_undef; 784 return &PL_sv_undef;
633} 785}
634 786
635static SV * 787static SV *
636decode_tagged (dec_t *dec) 788decode_tagged (dec_t *dec)
637{ 789{
790 SV *sv = 0;
638 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
639 SV *sv = decode_sv (dec); 840 SV *osv = decode_sv (dec);
841 sv_setsv (sv, osv);
842 SvREFCNT_dec_NN (osv);
843 }
844 break;
640 845
641 if (tag == CBOR_TAG_MAGIC) // 2.4.5 Self-Describe CBOR 846 case CBOR_TAG_VALUE_SHAREDREF:
642 return sv; 847 {
848 if ((*dec->cur >> 5) != 0)
849 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
643 850
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
644 AV *av = newAV (); 915 AV *av = newAV ();
645 av_push (av, newSVuv (tag)); 916 av_push (av, newSVuv (tag));
646 av_push (av, sv); 917 av_push (av, sv);
647 918
648 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 919 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
649 ? cbor_tagged_stash 920 ? cbor_tagged_stash
650 : gv_stashpv ("CBOR::XS::Tagged" , 1); 921 : gv_stashpv ("CBOR::XS::Tagged" , 1);
651 922
652 return sv_bless (newRV_noinc ((SV *)av), tagged_stash); 923 sv = sv_bless (newRV_noinc ((SV *)av), tagged_stash);
924 }
925 break;
926 }
927
928 return sv;
929
930fail:
931 SvREFCNT_dec (sv);
932 return &PL_sv_undef;
653} 933}
654 934
655static SV * 935static SV *
656decode_sv (dec_t *dec) 936decode_sv (dec_t *dec)
657{ 937{
676 case 7: // misc 956 case 7: // misc
677 switch (*dec->cur++ & 31) 957 switch (*dec->cur++ & 31)
678 { 958 {
679 case 20: 959 case 20:
680#if CBOR_SLOW 960#if CBOR_SLOW
681 cbor_false = get_bool ("CBOR::XS::false"); 961 types_false = get_bool ("Types::Serialiser::false");
682#endif 962#endif
683 return newSVsv (cbor_false); 963 return newSVsv (types_false);
684 case 21: 964 case 21:
685#if CBOR_SLOW 965#if CBOR_SLOW
686 cbor_true = get_bool ("CBOR::XS::true"); 966 types_true = get_bool ("Types::Serialiser::true");
687#endif 967#endif
688 return newSVsv (cbor_true); 968 return newSVsv (types_true);
689 case 22: 969 case 22:
690 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);
691 976
692 case 25: 977 case 25:
693 { 978 {
694 WANT (2); 979 WANT (2);
695 980
739} 1024}
740 1025
741static SV * 1026static SV *
742decode_cbor (SV *string, CBOR *cbor, char **offset_return) 1027decode_cbor (SV *string, CBOR *cbor, char **offset_return)
743{ 1028{
744 dec_t dec; 1029 dec_t dec = { };
745 SV *sv; 1030 SV *sv;
1031 STRLEN len;
1032 char *data = SvPVbyte (string, len);
746 1033
747 /* work around bugs in 5.10 where manipulating magic values
748 * makes perl ignore the magic in subsequent accesses.
749 * also make a copy of non-PV values, to get them into a clean
750 * state (SvPV should do that, but it's buggy, see below).
751 */
752 /*SvGETMAGIC (string);*/
753 if (SvMAGICAL (string) || !SvPOK (string))
754 string = sv_2mortal (newSVsv (string));
755
756 SvUPGRADE (string, SVt_PV);
757
758 /* work around a bug in perl 5.10, which causes SvCUR to fail an
759 * assertion with -DDEBUGGING, although SvCUR is documented to
760 * return the xpv_cur field which certainly exists after upgrading.
761 * according to nicholas clark, calling SvPOK fixes this.
762 * But it doesn't fix it, so try another workaround, call SvPV_nolen
763 * and hope for the best.
764 * Damnit, SvPV_nolen still trips over yet another assertion. This
765 * assertion business is seriously broken, try yet another workaround
766 * for the broken -DDEBUGGING.
767 */
768 {
769#ifdef DEBUGGING
770 STRLEN offset = SvOK (string) ? sv_len (string) : 0;
771#else
772 STRLEN offset = SvCUR (string);
773#endif
774
775 if (offset > cbor->max_size && cbor->max_size) 1034 if (len > cbor->max_size && cbor->max_size)
776 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",
777 (unsigned long)SvCUR (string), (unsigned long)cbor->max_size); 1036 (unsigned long)len, (unsigned long)cbor->max_size);
778 }
779
780 sv_utf8_downgrade (string, 0);
781 1037
782 dec.cbor = *cbor; 1038 dec.cbor = *cbor;
783 dec.cur = (U8 *)SvPVX (string); 1039 dec.cur = (U8 *)data;
784 dec.end = (U8 *)SvEND (string); 1040 dec.end = (U8 *)data + len;
785 dec.err = 0;
786 dec.depth = 0;
787 1041
788 sv = decode_sv (&dec); 1042 sv = decode_sv (&dec);
789 1043
790 if (offset_return) 1044 if (offset_return)
791 *offset_return = dec.cur; 1045 *offset_return = dec.cur;
795 dec.err = "garbage after CBOR object"; 1049 dec.err = "garbage after CBOR object";
796 1050
797 if (dec.err) 1051 if (dec.err)
798 { 1052 {
799 SvREFCNT_dec (sv); 1053 SvREFCNT_dec (sv);
800 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);
801 } 1055 }
802 1056
803 sv = sv_2mortal (sv); 1057 sv = sv_2mortal (sv);
804 1058
805 return sv; 1059 return sv;
811MODULE = CBOR::XS PACKAGE = CBOR::XS 1065MODULE = CBOR::XS PACKAGE = CBOR::XS
812 1066
813BOOT: 1067BOOT:
814{ 1068{
815 cbor_stash = gv_stashpv ("CBOR::XS" , 1); 1069 cbor_stash = gv_stashpv ("CBOR::XS" , 1);
816 cbor_boolean_stash = gv_stashpv ("CBOR::XS::Boolean", 1);
817 cbor_tagged_stash = gv_stashpv ("CBOR::XS::Tagged" , 1); 1070 cbor_tagged_stash = gv_stashpv ("CBOR::XS::Tagged" , 1);
818 1071
819 cbor_true = get_bool ("CBOR::XS::true"); 1072 types_boolean_stash = gv_stashpv ("Types::Serialiser::Boolean", 1);
820 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);
821} 1081}
822 1082
823PROTOTYPES: DISABLE 1083PROTOTYPES: DISABLE
824 1084
825void CLONE (...) 1085void CLONE (...)
826 CODE: 1086 CODE:
827 cbor_stash = 0; 1087 cbor_stash = 0;
828 cbor_boolean_stash = 0;
829 cbor_tagged_stash = 0; 1088 cbor_tagged_stash = 0;
1089 types_error_stash = 0;
1090 types_boolean_stash = 0;
830 1091
831void new (char *klass) 1092void new (char *klass)
832 PPCODE: 1093 PPCODE:
833{ 1094{
834 SV *pv = NEWSV (0, sizeof (CBOR)); 1095 SV *pv = NEWSV (0, sizeof (CBOR));
842 1103
843void shrink (CBOR *self, int enable = 1) 1104void shrink (CBOR *self, int enable = 1)
844 ALIAS: 1105 ALIAS:
845 shrink = F_SHRINK 1106 shrink = F_SHRINK
846 allow_unknown = F_ALLOW_UNKNOWN 1107 allow_unknown = F_ALLOW_UNKNOWN
1108 allow_sharing = F_ALLOW_SHARING
1109 allow_stringref = F_ALLOW_STRINGREF
847 PPCODE: 1110 PPCODE:
848{ 1111{
849 if (enable) 1112 if (enable)
850 self->flags |= ix; 1113 self->flags |= ix;
851 else 1114 else
856 1119
857void get_shrink (CBOR *self) 1120void get_shrink (CBOR *self)
858 ALIAS: 1121 ALIAS:
859 get_shrink = F_SHRINK 1122 get_shrink = F_SHRINK
860 get_allow_unknown = F_ALLOW_UNKNOWN 1123 get_allow_unknown = F_ALLOW_UNKNOWN
1124 get_allow_sharing = F_ALLOW_SHARING
1125 get_allow_stringref = F_ALLOW_STRINGREF
861 PPCODE: 1126 PPCODE:
862 XPUSHs (boolSV (self->flags & ix)); 1127 XPUSHs (boolSV (self->flags & ix));
863 1128
864void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) 1129void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
865 PPCODE: 1130 PPCODE:
880int get_max_size (CBOR *self) 1145int get_max_size (CBOR *self)
881 CODE: 1146 CODE:
882 RETVAL = self->max_size; 1147 RETVAL = self->max_size;
883 OUTPUT: 1148 OUTPUT:
884 RETVAL 1149 RETVAL
885
886#if 0 //TODO
887
888void filter_cbor_object (CBOR *self, SV *cb = &PL_sv_undef)
889 PPCODE:
890{
891 SvREFCNT_dec (self->cb_object);
892 self->cb_object = SvOK (cb) ? newSVsv (cb) : 0;
893
894 XPUSHs (ST (0));
895}
896
897void filter_cbor_single_key_object (CBOR *self, SV *key, SV *cb = &PL_sv_undef)
898 PPCODE:
899{
900 if (!self->cb_sk_object)
901 self->cb_sk_object = newHV ();
902
903 if (SvOK (cb))
904 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);
905 else
906 {
907 hv_delete_ent (self->cb_sk_object, key, G_DISCARD, 0);
908
909 if (!HvKEYS (self->cb_sk_object))
910 {
911 SvREFCNT_dec (self->cb_sk_object);
912 self->cb_sk_object = 0;
913 }
914 }
915
916 XPUSHs (ST (0));
917}
918
919#endif
920 1150
921void encode (CBOR *self, SV *scalar) 1151void encode (CBOR *self, SV *scalar)
922 PPCODE: 1152 PPCODE:
923 PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN; 1153 PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN;
924 XPUSHs (scalar); 1154 XPUSHs (scalar);
937 EXTEND (SP, 2); 1167 EXTEND (SP, 2);
938 PUSHs (sv); 1168 PUSHs (sv);
939 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); 1169 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
940} 1170}
941 1171
942#if 0
943
944void DESTROY (CBOR *self)
945 CODE:
946 SvREFCNT_dec (self->cb_sk_object);
947 SvREFCNT_dec (self->cb_object);
948
949#endif
950
951PROTOTYPES: ENABLE 1172PROTOTYPES: ENABLE
952 1173
953void encode_cbor (SV *scalar) 1174void encode_cbor (SV *scalar)
954 PPCODE: 1175 PPCODE:
955{ 1176{

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines