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 1.28 by root, Sat Nov 23 18:30:59 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#ifndef SvREFCNT_dec_NN
25# define SvREFCNT_dec_NN(sv) SvREFCNT_dec (sv)
26#endif
27
28// known tags
15enum cbor_tag 29enum cbor_tag
16{ 30{
31 // inofficial extensions (pending iana registration)
32 CBOR_TAG_PERL_OBJECT = 24, // http://cbor.schmorp.de/perl-object
33 CBOR_TAG_GENERIC_OBJECT = 25, // http://cbor.schmorp.de/generic-object
34 CBOR_TAG_VALUE_SHAREABLE = 26, // http://cbor.schmorp.de/value-sharing
35 CBOR_TAG_VALUE_SHAREDREF = 27, // http://cbor.schmorp.de/value-sharing
36 CBOR_TAG_STRINGREF_NAMESPACE = 65537, // http://cbor.schmorp.de/stringref
37 CBOR_TAG_STRINGREF = 28, // http://cbor.schmorp.de/stringref
38 CBOR_TAG_INDIRECTION = 22098, // http://cbor.schmorp.de/indirection
39
40 // rfc7049
17 CBOR_TAG_DATETIME = 0, // rfc4287, utf-8 41 CBOR_TAG_DATETIME = 0, // rfc4287, utf-8
18 CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any 42 CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any
19 CBOR_TAG_POS_BIGNUM = 2, // byte string 43 CBOR_TAG_POS_BIGNUM = 2, // byte string
20 CBOR_TAG_NEG_BIGNUM = 3, // byte string 44 CBOR_TAG_NEG_BIGNUM = 3, // byte string
21 CBOR_TAG_DECIMAL = 4, // decimal fraction, array 45 CBOR_TAG_DECIMAL = 4, // decimal fraction, array
22 CBOR_TAG_BIGFLOAT = 5, // array 46 CBOR_TAG_BIGFLOAT = 5, // array
23 47
24 CBOR_TAG_CONV_B64U = 21, // base64url, any 48 CBOR_TAG_CONV_B64U = 21, // base64url, any
25 CBOR_TAG_CONV_B64 = 22, // base64, any 49 CBOR_TAG_CONV_B64 = 22, // base64, any
26 CBOR_TAG_CONV_HEX = 23, // base16, any 50 CBOR_TAG_CONV_HEX = 23, // base16, any
27 CBOR_TAG_CBOR = 24, // embedded cbor, byte string 51 CBOR_TAG_CBOR = 24, // embedded cbor, byte string
28 52
29 CBOR_TAG_URI = 32, // URI rfc3986, utf-8 53 CBOR_TAG_URI = 32, // URI rfc3986, utf-8
30 CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8 54 CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8
31 CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8 55 CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8
32 CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8 56 CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8
33 CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8 57 CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8
34 58
35 CBOR_TAG_MAGIC = 55799 59 CBOR_TAG_MAGIC = 55799 // self-describe cbor
36}; 60};
37 61
38#define F_SHRINK 0x00000200UL 62#define F_SHRINK 0x00000001UL
39#define F_ALLOW_UNKNOWN 0x00002000UL 63#define F_ALLOW_UNKNOWN 0x00000002UL
64#define F_ALLOW_SHARING 0x00000004UL //TODO
65#define F_ALLOW_STRINGREF 0x00000008UL //TODO
40 66
41#define INIT_SIZE 32 // initial scalar size to be allocated 67#define INIT_SIZE 32 // initial scalar size to be allocated
42 68
43#define SB do { 69#define SB do {
44#define SE } while (0) 70#define SE } while (0)
55#else 81#else
56# define CBOR_SLOW 0 82# define CBOR_SLOW 0
57# define CBOR_STASH cbor_stash 83# define CBOR_STASH cbor_stash
58#endif 84#endif
59 85
60static HV *cbor_stash, *cbor_boolean_stash, *cbor_tagged_stash; // CBOR::XS:: 86static HV *cbor_stash, *types_boolean_stash, *types_error_stash, *cbor_tagged_stash; // CBOR::XS::
61static SV *cbor_true, *cbor_false; 87static SV *types_true, *types_false, *types_error, *sv_cbor, *default_filter;
62 88
63typedef struct { 89typedef struct {
64 U32 flags; 90 U32 flags;
65 U32 max_depth; 91 U32 max_depth;
66 STRLEN max_size; 92 STRLEN max_size;
93 SV *filter;
67} CBOR; 94} CBOR;
68 95
69ecb_inline void 96ecb_inline void
70cbor_init (CBOR *cbor) 97cbor_init (CBOR *cbor)
71{ 98{
72 Zero (cbor, 1, CBOR); 99 Zero (cbor, 1, CBOR);
73 cbor->max_depth = 512; 100 cbor->max_depth = 512;
101}
102
103ecb_inline void
104cbor_free (CBOR *cbor)
105{
106 SvREFCNT_dec (cbor->filter);
74} 107}
75 108
76///////////////////////////////////////////////////////////////////////////// 109/////////////////////////////////////////////////////////////////////////////
77// utility functions 110// utility functions
78 111
100 SvPV_renew (sv, SvCUR (sv) + 1); 133 SvPV_renew (sv, SvCUR (sv) + 1);
101#endif 134#endif
102 } 135 }
103} 136}
104 137
105///////////////////////////////////////////////////////////////////////////// 138// minimum length of a string to be registered for stringref
106// fp hell 139ecb_inline int
107 140minimum_string_length (UV idx)
108//TODO 141{
142 return idx > 23
143 ? idx > 0xffU
144 ? idx > 0xffffU
145 ? idx > 0xffffffffU
146 ? 7
147 : 6
148 : 5
149 : 4
150 : 3;
151}
109 152
110///////////////////////////////////////////////////////////////////////////// 153/////////////////////////////////////////////////////////////////////////////
111// encoder 154// encoder
112 155
113// structure used for encoding CBOR 156// structure used for encoding CBOR
116 char *cur; // SvPVX (sv) + current output position 159 char *cur; // SvPVX (sv) + current output position
117 char *end; // SvEND (sv) 160 char *end; // SvEND (sv)
118 SV *sv; // result scalar 161 SV *sv; // result scalar
119 CBOR cbor; 162 CBOR cbor;
120 U32 depth; // recursion level 163 U32 depth; // recursion level
164 HV *stringref[2]; // string => index, or 0 ([0] = bytes, [1] = utf-8)
165 UV stringref_idx;
166 HV *shareable; // ptr => index, or 0
167 UV shareable_idx;
121} enc_t; 168} enc_t;
122 169
123ecb_inline void 170ecb_inline void
124need (enc_t *enc, STRLEN len) 171need (enc_t *enc, STRLEN len)
125{ 172{
177 *enc->cur++ = len >> 8; 224 *enc->cur++ = len >> 8;
178 *enc->cur++ = len; 225 *enc->cur++ = len;
179 } 226 }
180} 227}
181 228
229ecb_inline void
230encode_tag (enc_t *enc, UV tag)
231{
232 encode_uint (enc, 0xc0, tag);
233}
234
182static void 235static void
183encode_str (enc_t *enc, int utf8, char *str, STRLEN len) 236encode_str (enc_t *enc, int utf8, char *str, STRLEN len)
184{ 237{
238 if (ecb_expect_false (enc->cbor.flags & F_ALLOW_STRINGREF))
239 {
240 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1);
241
242 if (SvOK (*svp))
243 {
244 // already registered, use stringref
245 encode_tag (enc, CBOR_TAG_STRINGREF);
246 encode_uint (enc, 0x00, SvUV (*svp));
247 return;
248 }
249 else if (len >= minimum_string_length (enc->stringref_idx))
250 {
251 // register only
252 sv_setuv (*svp, enc->stringref_idx);
253 ++enc->stringref_idx;
254 }
255 }
256
185 encode_uint (enc, utf8 ? 0x60 : 0x40, len); 257 encode_uint (enc, utf8 ? 0x60 : 0x40, len);
186 need (enc, len); 258 need (enc, len);
187 memcpy (enc->cur, str, len); 259 memcpy (enc->cur, str, len);
188 enc->cur += len; 260 enc->cur += len;
189} 261}
247 319
248// encode objects, arrays and special \0=false and \1=true values. 320// encode objects, arrays and special \0=false and \1=true values.
249static void 321static void
250encode_rv (enc_t *enc, SV *sv) 322encode_rv (enc_t *enc, SV *sv)
251{ 323{
252 svtype svt;
253
254 SvGETMAGIC (sv); 324 SvGETMAGIC (sv);
325
326 if (ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING)
327 && ecb_expect_false (SvREFCNT (sv) > 1))
328 {
329 if (!enc->shareable)
330 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
331
332 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
333
334 if (SvOK (*svp))
335 {
336 encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF);
337 encode_uint (enc, 0x00, SvUV (*svp));
338 return;
339 }
340 else
341 {
342 sv_setuv (*svp, enc->shareable_idx);
343 ++enc->shareable_idx;
344 encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE);
345 }
346 }
347
255 svt = SvTYPE (sv); 348 svtype svt = SvTYPE (sv);
256 349
257 if (ecb_expect_false (SvOBJECT (sv))) 350 if (ecb_expect_false (SvOBJECT (sv)))
258 { 351 {
259 HV *boolean_stash = !CBOR_SLOW || cbor_boolean_stash 352 HV *boolean_stash = !CBOR_SLOW || types_boolean_stash
260 ? cbor_boolean_stash 353 ? types_boolean_stash
261 : gv_stashpv ("CBOR::XS::Boolean", 1); 354 : gv_stashpv ("Types::Serialiser::Boolean", 1);
355 HV *error_stash = !CBOR_SLOW || types_error_stash
356 ? types_error_stash
357 : gv_stashpv ("Types::Serialiser::Error", 1);
262 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 358 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
263 ? cbor_tagged_stash 359 ? cbor_tagged_stash
264 : gv_stashpv ("CBOR::XS::Tagged" , 1); 360 : gv_stashpv ("CBOR::XS::Tagged" , 1);
265 361
362 HV *stash = SvSTASH (sv);
363 GV *method;
364
266 if (SvSTASH (sv) == boolean_stash) 365 if (stash == boolean_stash)
267 encode_ch (enc, SvIV (sv) ? 0xe0 | 21 : 0xe0 | 20); 366 encode_ch (enc, SvIV (sv) ? 0xe0 | 21 : 0xe0 | 20);
367 else if (stash == error_stash)
368 encode_ch (enc, 0xe0 | 23);
268 else if (SvSTASH (sv) == tagged_stash) 369 else if (stash == tagged_stash)
269 { 370 {
270 if (svt != SVt_PVAV) 371 if (svt != SVt_PVAV)
271 croak ("encountered CBOR::XS::Tagged object that isn't an array"); 372 croak ("encountered CBOR::XS::Tagged object that isn't an array");
272 373
273 encode_uint (enc, 0xc0, SvUV (*av_fetch ((AV *)sv, 0, 1))); 374 encode_uint (enc, 0xc0, SvUV (*av_fetch ((AV *)sv, 0, 1)));
274 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1)); 375 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1));
275 } 376 }
377 else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
378 {
379 dSP;
380
381 ENTER; SAVETMPS; PUSHMARK (SP);
382 // we re-bless the reference to get overload and other niceties right
383 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
384
385 PUTBACK;
386 // G_SCALAR ensures that return value is 1
387 call_sv ((SV *)GvCV (method), G_SCALAR);
388 SPAGAIN;
389
390 // catch this surprisingly common error
391 if (SvROK (TOPs) && SvRV (TOPs) == sv)
392 croak ("%s::TO_CBOR method returned same object as was passed instead of a new one", HvNAME (stash));
393
394 encode_sv (enc, POPs);
395
396 PUTBACK;
397
398 FREETMPS; LEAVE;
399 }
400 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)
401 {
402 dSP;
403
404 ENTER; SAVETMPS; PUSHMARK (SP);
405 EXTEND (SP, 2);
406 // we re-bless the reference to get overload and other niceties right
407 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
408 PUSHs (sv_cbor);
409
410 PUTBACK;
411 int count = call_sv ((SV *)GvCV (method), G_ARRAY);
412 SPAGAIN;
413
414 // catch this surprisingly common error
415 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)
416 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));
417
418 encode_tag (enc, CBOR_TAG_PERL_OBJECT);
419 encode_uint (enc, 0x80, count + 1);
420 encode_str (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
421
422 while (count)
423 encode_sv (enc, SP[1 - count--]);
424
425 PUTBACK;
426
427 FREETMPS; LEAVE;
428 }
276 else 429 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", 430 croak ("encountered object '%s', but no TO_CBOR or FREEZE methods available on it",
306 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 431 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
307 }
308 } 432 }
309 else if (svt == SVt_PVHV) 433 else if (svt == SVt_PVHV)
310 encode_hv (enc, (HV *)sv); 434 encode_hv (enc, (HV *)sv);
311 else if (svt == SVt_PVAV) 435 else if (svt == SVt_PVAV)
312 encode_av (enc, (AV *)sv); 436 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 437 else
331 croak ("encountered %s, but CBOR can only represent references to arrays or hashes", 438 {
332 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 439 encode_tag (enc, CBOR_TAG_INDIRECTION);
440 encode_sv (enc, sv);
441 }
333} 442}
334 443
335static void 444static void
336encode_nv (enc_t *enc, SV *sv) 445encode_nv (enc_t *enc, SV *sv)
337{ 446{
402} 511}
403 512
404static SV * 513static SV *
405encode_cbor (SV *scalar, CBOR *cbor) 514encode_cbor (SV *scalar, CBOR *cbor)
406{ 515{
407 enc_t enc; 516 enc_t enc = { };
408 517
409 enc.cbor = *cbor; 518 enc.cbor = *cbor;
410 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 519 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
411 enc.cur = SvPVX (enc.sv); 520 enc.cur = SvPVX (enc.sv);
412 enc.end = SvEND (enc.sv); 521 enc.end = SvEND (enc.sv);
413 enc.depth = 0;
414 522
415 SvPOK_only (enc.sv); 523 SvPOK_only (enc.sv);
524
525 if (cbor->flags & F_ALLOW_STRINGREF)
526 {
527 encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE);
528 enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ());
529 enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ());
530 }
531
416 encode_sv (&enc, scalar); 532 encode_sv (&enc, scalar);
417 533
418 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 534 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
419 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings 535 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
420 536
434 U8 *end; // end of input string 550 U8 *end; // end of input string
435 const char *err; // parse error, if != 0 551 const char *err; // parse error, if != 0
436 CBOR cbor; 552 CBOR cbor;
437 U32 depth; // recursion depth 553 U32 depth; // recursion depth
438 U32 maxdepth; // recursion depth limit 554 U32 maxdepth; // recursion depth limit
555 AV *shareable;
556 AV *stringref;
557 SV *decode_tagged;
439} dec_t; 558} dec_t;
440 559
441#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE 560#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE
442 561
443#define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data") 562#define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data")
537 SvREFCNT_dec (av); 656 SvREFCNT_dec (av);
538 DEC_DEC_DEPTH; 657 DEC_DEC_DEPTH;
539 return &PL_sv_undef; 658 return &PL_sv_undef;
540} 659}
541 660
661static void
662decode_he (dec_t *dec, HV *hv)
663{
664 // for speed reasons, we specialcase single-string
665 // byte or utf-8 strings as keys, but only when !stringref
666
667 if (ecb_expect_true (!dec->stringref))
668 if (*dec->cur >= 0x40 && *dec->cur <= 0x40 + 27)
669 {
670 I32 len = decode_uint (dec);
671 char *key = (char *)dec->cur;
672
673 dec->cur += len;
674
675 if (ecb_expect_false (dec->stringref))
676 av_push (dec->stringref, newSVpvn (key, len));
677
678 hv_store (hv, key, len, decode_sv (dec), 0);
679
680 return;
681 }
682 else if (*dec->cur >= 0x60 && *dec->cur <= 0x60 + 27)
683 {
684 I32 len = decode_uint (dec);
685 char *key = (char *)dec->cur;
686
687 dec->cur += len;
688
689 if (ecb_expect_false (dec->stringref))
690 av_push (dec->stringref, newSVpvn_utf8 (key, len, 1));
691
692 hv_store (hv, key, -len, decode_sv (dec), 0);
693
694 return;
695 }
696
697 SV *k = decode_sv (dec);
698 SV *v = decode_sv (dec);
699
700 hv_store_ent (hv, k, v, 0);
701 SvREFCNT_dec (k);
702}
703
542static SV * 704static SV *
543decode_hv (dec_t *dec) 705decode_hv (dec_t *dec)
544{ 706{
545 HV *hv = newHV (); 707 HV *hv = newHV ();
546 708
558 { 720 {
559 ++dec->cur; 721 ++dec->cur;
560 break; 722 break;
561 } 723 }
562 724
563 SV *k = decode_sv (dec); 725 decode_he (dec, hv);
564 SV *v = decode_sv (dec);
565
566 hv_store_ent (hv, k, v, 0);
567 } 726 }
568 } 727 }
569 else 728 else
570 { 729 {
571 int len = decode_uint (dec); 730 int pairs = decode_uint (dec);
572 731
573 while (len--) 732 while (pairs--)
574 { 733 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 } 734 }
581 735
582 DEC_DEC_DEPTH; 736 DEC_DEC_DEPTH;
583 return newRV_noinc ((SV *)hv); 737 return newRV_noinc ((SV *)hv);
584 738
618 STRLEN len = decode_uint (dec); 772 STRLEN len = decode_uint (dec);
619 773
620 WANT (len); 774 WANT (len);
621 sv = newSVpvn (dec->cur, len); 775 sv = newSVpvn (dec->cur, len);
622 dec->cur += len; 776 dec->cur += len;
777
778 if (ecb_expect_false (dec->stringref)
779 && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1))
780 av_push (dec->stringref, SvREFCNT_inc_NN (sv));
623 } 781 }
624 782
625 if (utf8) 783 if (utf8)
626 SvUTF8_on (sv); 784 SvUTF8_on (sv);
627 785
633} 791}
634 792
635static SV * 793static SV *
636decode_tagged (dec_t *dec) 794decode_tagged (dec_t *dec)
637{ 795{
796 SV *sv = 0;
638 UV tag = decode_uint (dec); 797 UV tag = decode_uint (dec);
798
799 WANT (1);
800
801 switch (tag)
802 {
803 case CBOR_TAG_MAGIC:
804 sv = decode_sv (dec);
805 break;
806
807 case CBOR_TAG_INDIRECTION:
808 sv = newRV_noinc (decode_sv (dec));
809 break;
810
811 case CBOR_TAG_STRINGREF_NAMESPACE:
812 {
813 ENTER; SAVETMPS;
814
815 SAVESPTR (dec->stringref);
816 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ());
817
818 sv = decode_sv (dec);
819
820 FREETMPS; LEAVE;
821 }
822 break;
823
824 case CBOR_TAG_STRINGREF:
825 {
826 if ((*dec->cur >> 5) != 0)
827 ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
828
829 UV idx = decode_uint (dec);
830
831 if (!dec->stringref || (int)idx > AvFILLp (dec->stringref))
832 ERR ("corrupted CBOR data (stringref index out of bounds or outside namespace)");
833
834 sv = newSVsv (AvARRAY (dec->stringref)[idx]);
835 }
836 break;
837
838 case CBOR_TAG_VALUE_SHAREABLE:
839 {
840 if (ecb_expect_false (!dec->shareable))
841 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ());
842
843 sv = newSV (0);
844 av_push (dec->shareable, SvREFCNT_inc_NN (sv));
845
639 SV *sv = decode_sv (dec); 846 SV *osv = decode_sv (dec);
847 sv_setsv (sv, osv);
848 SvREFCNT_dec_NN (osv);
849 }
850 break;
640 851
641 if (tag == CBOR_TAG_MAGIC) // 2.4.5 Self-Describe CBOR 852 case CBOR_TAG_VALUE_SHAREDREF:
853 {
854 if ((*dec->cur >> 5) != 0)
855 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
856
857 UV idx = decode_uint (dec);
858
859 if (!dec->shareable || (int)idx > AvFILLp (dec->shareable))
860 ERR ("corrupted CBOR data (sharedref index out of bounds)");
861
862 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]);
863 }
864 break;
865
866 case CBOR_TAG_PERL_OBJECT:
867 {
868 sv = decode_sv (dec);
869
870 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
871 ERR ("corrupted CBOR data (non-array perl object)");
872
873 AV *av = (AV *)SvRV (sv);
874 int len = av_len (av) + 1;
875 HV *stash = gv_stashsv (*av_fetch (av, 0, 1), 0);
876
877 if (!stash)
878 ERR ("cannot decode perl-object (package does not exist)");
879
880 GV *method = gv_fetchmethod_autoload (stash, "THAW", 0);
881
882 if (!method)
883 ERR ("cannot decode perl-object (package does not have a THAW method)");
884
885 dSP;
886
887 ENTER; SAVETMPS; PUSHMARK (SP);
888 EXTEND (SP, len + 1);
889 // we re-bless the reference to get overload and other niceties right
890 PUSHs (*av_fetch (av, 0, 1));
891 PUSHs (sv_cbor);
892
893 int i;
894
895 for (i = 1; i < len; ++i)
896 PUSHs (*av_fetch (av, i, 1));
897
898 PUTBACK;
899 call_sv ((SV *)GvCV (method), G_SCALAR | G_EVAL);
900 SPAGAIN;
901
902 if (SvTRUE (ERRSV))
903 {
904 FREETMPS; LEAVE;
905 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV))));
906 }
907
908 SvREFCNT_dec (sv);
909 sv = SvREFCNT_inc (POPs);
910
911 PUTBACK;
912
913 FREETMPS; LEAVE;
914 }
915 break;
916
917 default:
918 {
919 sv = decode_sv (dec);
920
921 dSP;
922 ENTER; SAVETMPS; PUSHMARK (SP);
923 EXTEND (SP, 2);
924 PUSHs (newSVuv (tag));
925 PUSHs (sv);
926
927 PUTBACK;
928 int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL);
929 SPAGAIN;
930
931 if (SvTRUE (ERRSV))
932 {
933 FREETMPS; LEAVE;
934 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV))));
935 }
936
937 if (count)
938 {
939 SvREFCNT_dec (sv);
940 sv = SvREFCNT_inc (POPs);
941 }
942 else
943 {
944 AV *av = newAV ();
945 av_push (av, newSVuv (tag));
946 av_push (av, sv);
947
948 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
949 ? cbor_tagged_stash
950 : gv_stashpv ("CBOR::XS::Tagged" , 1);
951 sv = sv_bless (newRV_noinc ((SV *)av), tagged_stash);
952 }
953
954 PUTBACK;
955
956 FREETMPS; LEAVE;
957 }
958 break;
959 }
960
642 return sv; 961 return sv;
643 962
644 AV *av = newAV (); 963fail:
645 av_push (av, newSVuv (tag)); 964 SvREFCNT_dec (sv);
646 av_push (av, sv); 965 return &PL_sv_undef;
647
648 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
649 ? cbor_tagged_stash
650 : gv_stashpv ("CBOR::XS::Tagged" , 1);
651
652 return sv_bless (newRV_noinc ((SV *)av), tagged_stash);
653} 966}
654 967
655static SV * 968static SV *
656decode_sv (dec_t *dec) 969decode_sv (dec_t *dec)
657{ 970{
676 case 7: // misc 989 case 7: // misc
677 switch (*dec->cur++ & 31) 990 switch (*dec->cur++ & 31)
678 { 991 {
679 case 20: 992 case 20:
680#if CBOR_SLOW 993#if CBOR_SLOW
681 cbor_false = get_bool ("CBOR::XS::false"); 994 types_false = get_bool ("Types::Serialiser::false");
682#endif 995#endif
683 return newSVsv (cbor_false); 996 return newSVsv (types_false);
684 case 21: 997 case 21:
685#if CBOR_SLOW 998#if CBOR_SLOW
686 cbor_true = get_bool ("CBOR::XS::true"); 999 types_true = get_bool ("Types::Serialiser::true");
687#endif 1000#endif
688 return newSVsv (cbor_true); 1001 return newSVsv (types_true);
689 case 22: 1002 case 22:
690 return newSVsv (&PL_sv_undef); 1003 return newSVsv (&PL_sv_undef);
1004 case 23:
1005#if CBOR_SLOW
1006 types_error = get_bool ("Types::Serialiser::error");
1007#endif
1008 return newSVsv (types_error);
691 1009
692 case 25: 1010 case 25:
693 { 1011 {
694 WANT (2); 1012 WANT (2);
695 1013
739} 1057}
740 1058
741static SV * 1059static SV *
742decode_cbor (SV *string, CBOR *cbor, char **offset_return) 1060decode_cbor (SV *string, CBOR *cbor, char **offset_return)
743{ 1061{
744 dec_t dec; 1062 dec_t dec = { };
745 SV *sv; 1063 SV *sv;
1064 STRLEN len;
1065 char *data = SvPVbyte (string, len);
746 1066
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) 1067 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", 1068 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); 1069 (unsigned long)len, (unsigned long)cbor->max_size);
778 }
779
780 sv_utf8_downgrade (string, 0);
781 1070
782 dec.cbor = *cbor; 1071 dec.cbor = *cbor;
783 dec.cur = (U8 *)SvPVX (string); 1072 dec.cur = (U8 *)data;
784 dec.end = (U8 *)SvEND (string); 1073 dec.end = (U8 *)data + len;
785 dec.err = 0;
786 dec.depth = 0;
787 1074
788 sv = decode_sv (&dec); 1075 sv = decode_sv (&dec);
789 1076
790 if (offset_return) 1077 if (offset_return)
791 *offset_return = dec.cur; 1078 *offset_return = dec.cur;
795 dec.err = "garbage after CBOR object"; 1082 dec.err = "garbage after CBOR object";
796 1083
797 if (dec.err) 1084 if (dec.err)
798 { 1085 {
799 SvREFCNT_dec (sv); 1086 SvREFCNT_dec (sv);
800 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)SvPVX (string), (int)(uint8_t)*dec.cur); 1087 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur);
801 } 1088 }
802 1089
803 sv = sv_2mortal (sv); 1090 sv = sv_2mortal (sv);
804 1091
805 return sv; 1092 return sv;
811MODULE = CBOR::XS PACKAGE = CBOR::XS 1098MODULE = CBOR::XS PACKAGE = CBOR::XS
812 1099
813BOOT: 1100BOOT:
814{ 1101{
815 cbor_stash = gv_stashpv ("CBOR::XS" , 1); 1102 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); 1103 cbor_tagged_stash = gv_stashpv ("CBOR::XS::Tagged" , 1);
818 1104
819 cbor_true = get_bool ("CBOR::XS::true"); 1105 types_boolean_stash = gv_stashpv ("Types::Serialiser::Boolean", 1);
820 cbor_false = get_bool ("CBOR::XS::false"); 1106 types_error_stash = gv_stashpv ("Types::Serialiser::Error" , 1);
1107
1108 types_true = get_bool ("Types::Serialiser::true" );
1109 types_false = get_bool ("Types::Serialiser::false");
1110 types_error = get_bool ("Types::Serialiser::error");
1111
1112 default_filter = newSVpv ("CBOR::XS::default_filter", 0);
1113
1114 sv_cbor = newSVpv ("CBOR", 0);
1115 SvREADONLY_on (sv_cbor);
821} 1116}
822 1117
823PROTOTYPES: DISABLE 1118PROTOTYPES: DISABLE
824 1119
825void CLONE (...) 1120void CLONE (...)
826 CODE: 1121 CODE:
827 cbor_stash = 0; 1122 cbor_stash = 0;
828 cbor_boolean_stash = 0;
829 cbor_tagged_stash = 0; 1123 cbor_tagged_stash = 0;
1124 types_error_stash = 0;
1125 types_boolean_stash = 0;
830 1126
831void new (char *klass) 1127void new (char *klass)
832 PPCODE: 1128 PPCODE:
833{ 1129{
834 SV *pv = NEWSV (0, sizeof (CBOR)); 1130 SV *pv = NEWSV (0, sizeof (CBOR));
842 1138
843void shrink (CBOR *self, int enable = 1) 1139void shrink (CBOR *self, int enable = 1)
844 ALIAS: 1140 ALIAS:
845 shrink = F_SHRINK 1141 shrink = F_SHRINK
846 allow_unknown = F_ALLOW_UNKNOWN 1142 allow_unknown = F_ALLOW_UNKNOWN
1143 allow_sharing = F_ALLOW_SHARING
1144 allow_stringref = F_ALLOW_STRINGREF
847 PPCODE: 1145 PPCODE:
848{ 1146{
849 if (enable) 1147 if (enable)
850 self->flags |= ix; 1148 self->flags |= ix;
851 else 1149 else
856 1154
857void get_shrink (CBOR *self) 1155void get_shrink (CBOR *self)
858 ALIAS: 1156 ALIAS:
859 get_shrink = F_SHRINK 1157 get_shrink = F_SHRINK
860 get_allow_unknown = F_ALLOW_UNKNOWN 1158 get_allow_unknown = F_ALLOW_UNKNOWN
1159 get_allow_sharing = F_ALLOW_SHARING
1160 get_allow_stringref = F_ALLOW_STRINGREF
861 PPCODE: 1161 PPCODE:
862 XPUSHs (boolSV (self->flags & ix)); 1162 XPUSHs (boolSV (self->flags & ix));
863 1163
864void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) 1164void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
865 PPCODE: 1165 PPCODE:
881 CODE: 1181 CODE:
882 RETVAL = self->max_size; 1182 RETVAL = self->max_size;
883 OUTPUT: 1183 OUTPUT:
884 RETVAL 1184 RETVAL
885 1185
886#if 0 //TODO 1186void filter (CBOR *self, SV *filter = 0)
887
888void filter_cbor_object (CBOR *self, SV *cb = &PL_sv_undef)
889 PPCODE: 1187 PPCODE:
890{
891 SvREFCNT_dec (self->cb_object); 1188 SvREFCNT_dec (self->filter);
892 self->cb_object = SvOK (cb) ? newSVsv (cb) : 0; 1189 self->filter = filter ? newSVsv (filter) : filter;
893
894 XPUSHs (ST (0)); 1190 XPUSHs (ST (0));
895}
896 1191
897void filter_cbor_single_key_object (CBOR *self, SV *key, SV *cb = &PL_sv_undef) 1192SV *get_filter (CBOR *self)
898 PPCODE: 1193 CODE:
899{ 1194 RETVAL = self->filter ? self->filter : NEWSV (0, 0);
900 if (!self->cb_sk_object) 1195 OUTPUT:
901 self->cb_sk_object = newHV (); 1196 RETVAL
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 1197
921void encode (CBOR *self, SV *scalar) 1198void encode (CBOR *self, SV *scalar)
922 PPCODE: 1199 PPCODE:
923 PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN; 1200 PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN;
924 XPUSHs (scalar); 1201 XPUSHs (scalar);
937 EXTEND (SP, 2); 1214 EXTEND (SP, 2);
938 PUSHs (sv); 1215 PUSHs (sv);
939 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); 1216 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
940} 1217}
941 1218
942#if 0
943
944void DESTROY (CBOR *self) 1219void DESTROY (CBOR *self)
945 CODE: 1220 PPCODE:
946 SvREFCNT_dec (self->cb_sk_object); 1221 cbor_free (self);
947 SvREFCNT_dec (self->cb_object);
948
949#endif
950 1222
951PROTOTYPES: ENABLE 1223PROTOTYPES: ENABLE
952 1224
953void encode_cbor (SV *scalar) 1225void encode_cbor (SV *scalar)
954 PPCODE: 1226 PPCODE:

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines