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 1.20 by root, Wed Nov 20 11:06:42 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_DEDUP_STRINGS 0x00000008UL //TODO
63#define F_DEDUP_KEYS 0x00000010UL //TODO
40 64
41#define INIT_SIZE 32 // initial scalar size to be allocated 65#define INIT_SIZE 32 // initial scalar size to be allocated
42 66
43#define SB do { 67#define SB do {
44#define SE } while (0) 68#define SE } while (0)
55#else 79#else
56# define CBOR_SLOW 0 80# define CBOR_SLOW 0
57# define CBOR_STASH cbor_stash 81# define CBOR_STASH cbor_stash
58#endif 82#endif
59 83
60static HV *cbor_stash, *cbor_boolean_stash, *cbor_tagged_stash; // CBOR::XS:: 84static HV *cbor_stash, *types_boolean_stash, *types_error_stash, *cbor_tagged_stash; // CBOR::XS::
61static SV *cbor_true, *cbor_false; 85static SV *types_true, *types_false, *types_error, *sv_cbor;
62 86
63typedef struct { 87typedef struct {
64 U32 flags; 88 U32 flags;
65 U32 max_depth; 89 U32 max_depth;
66 STRLEN max_size; 90 STRLEN max_size;
101#endif 125#endif
102 } 126 }
103} 127}
104 128
105///////////////////////////////////////////////////////////////////////////// 129/////////////////////////////////////////////////////////////////////////////
106// fp hell
107
108//TODO
109
110/////////////////////////////////////////////////////////////////////////////
111// encoder 130// encoder
112 131
113// structure used for encoding CBOR 132// structure used for encoding CBOR
114typedef struct 133typedef struct
115{ 134{
116 char *cur; // SvPVX (sv) + current output position 135 char *cur; // SvPVX (sv) + current output position
117 char *end; // SvEND (sv) 136 char *end; // SvEND (sv)
118 SV *sv; // result scalar 137 SV *sv; // result scalar
119 CBOR cbor; 138 CBOR cbor;
120 U32 depth; // recursion level 139 U32 depth; // recursion level
140 HV *stringref[2]; // string => index, or 0 ([0] = bytes, [1] = utf-8)
141 UV stringref_idx;
142 HV *shareable; // ptr => index, or 0
143 UV shareable_idx;
121} enc_t; 144} enc_t;
122 145
123ecb_inline void 146ecb_inline void
124need (enc_t *enc, STRLEN len) 147need (enc_t *enc, STRLEN len)
125{ 148{
186 need (enc, len); 209 need (enc, len);
187 memcpy (enc->cur, str, len); 210 memcpy (enc->cur, str, len);
188 enc->cur += len; 211 enc->cur += len;
189} 212}
190 213
214ecb_inline void
215encode_tag (enc_t *enc, UV tag)
216{
217 encode_uint (enc, 0xc0, tag);
218}
219
191static void encode_sv (enc_t *enc, SV *sv); 220static void encode_sv (enc_t *enc, SV *sv);
192 221
193static void 222static void
194encode_av (enc_t *enc, AV *av) 223encode_av (enc_t *enc, AV *av)
195{ 224{
207 SV **svp = av_fetch (av, i, 0); 236 SV **svp = av_fetch (av, i, 0);
208 encode_sv (enc, svp ? *svp : &PL_sv_undef); 237 encode_sv (enc, svp ? *svp : &PL_sv_undef);
209 } 238 }
210 239
211 --enc->depth; 240 --enc->depth;
241}
242
243ecb_inline void
244encode_he (enc_t *enc, HE *he)
245{
246 if (HeKLEN (he) == HEf_SVKEY)
247 encode_sv (enc, HeSVKEY (he));
248 else
249 encode_str (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he));
212} 250}
213 251
214static void 252static void
215encode_hv (enc_t *enc, HV *hv) 253encode_hv (enc_t *enc, HV *hv)
216{ 254{
229 else 267 else
230 encode_uint (enc, 0xa0, pairs); 268 encode_uint (enc, 0xa0, pairs);
231 269
232 while ((he = hv_iternext (hv))) 270 while ((he = hv_iternext (hv)))
233 { 271 {
272 if (ecb_expect_false (enc->cbor.flags & (F_DEDUP_STRINGS | F_DEDUP_KEYS)))
273 {
274 SV **svp;
275
234 if (HeKLEN (he) == HEf_SVKEY) 276 if (HeKLEN (he) == HEf_SVKEY)
235 encode_sv (enc, HeSVKEY (he)); 277 svp = hv_fetch_ent (enc->stringref[!! SvUTF8 (HeSVKEY (he))], HeSVKEY (he) , 1, 0);//TODO return HE :/
278 else
279 svp = hv_fetch (enc->stringref[!! HeKUTF8 (he) ], HeKEY (he), HeKLEN (he), 1);
280
281 if (SvOK (*svp))
282 {
283 encode_tag (enc, CBOR_TAG_STRINGREF);
284 encode_uint (enc, 0x00, SvUV (*svp));
285 }
286 else
287 {
288 sv_setuv (*svp, enc->stringref_idx);
289 ++enc->stringref_idx;
290 encode_he (enc, he);
291 }
292 }
236 else 293 else
237 encode_str (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he)); 294 encode_he (enc, he);
238 295
239 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he)); 296 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he));
240 } 297 }
241 298
242 if (mg) 299 if (mg)
247 304
248// encode objects, arrays and special \0=false and \1=true values. 305// encode objects, arrays and special \0=false and \1=true values.
249static void 306static void
250encode_rv (enc_t *enc, SV *sv) 307encode_rv (enc_t *enc, SV *sv)
251{ 308{
252 svtype svt;
253
254 SvGETMAGIC (sv); 309 SvGETMAGIC (sv);
310
311 if (ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING)
312 && ecb_expect_false (SvREFCNT (sv) > 1))
313 {
314 if (!enc->shareable)
315 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
316
317 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
318
319 if (SvOK (*svp))
320 {
321 encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF);
322 encode_uint (enc, 0x00, SvUV (*svp));
323 return;
324 }
325 else
326 {
327 sv_setuv (*svp, enc->shareable_idx);
328 ++enc->shareable_idx;
329 encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE);
330 }
331 }
332
255 svt = SvTYPE (sv); 333 svtype svt = SvTYPE (sv);
256 334
257 if (ecb_expect_false (SvOBJECT (sv))) 335 if (ecb_expect_false (SvOBJECT (sv)))
258 { 336 {
259 HV *boolean_stash = !CBOR_SLOW || cbor_boolean_stash 337 HV *boolean_stash = !CBOR_SLOW || types_boolean_stash
260 ? cbor_boolean_stash 338 ? types_boolean_stash
261 : gv_stashpv ("CBOR::XS::Boolean", 1); 339 : gv_stashpv ("Types::Serialiser::Boolean", 1);
340 HV *error_stash = !CBOR_SLOW || types_error_stash
341 ? types_error_stash
342 : gv_stashpv ("Types::Serialiser::Error", 1);
262 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 343 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
263 ? cbor_tagged_stash 344 ? cbor_tagged_stash
264 : gv_stashpv ("CBOR::XS::Tagged" , 1); 345 : gv_stashpv ("CBOR::XS::Tagged" , 1);
265 346
347 HV *stash = SvSTASH (sv);
348 GV *method;
349
266 if (SvSTASH (sv) == boolean_stash) 350 if (stash == boolean_stash)
267 encode_ch (enc, SvIV (sv) ? 0xe0 | 21 : 0xe0 | 20); 351 encode_ch (enc, SvIV (sv) ? 0xe0 | 21 : 0xe0 | 20);
352 else if (stash == error_stash)
353 encode_ch (enc, 0xe0 | 23);
268 else if (SvSTASH (sv) == tagged_stash) 354 else if (stash == tagged_stash)
269 { 355 {
270 if (svt != SVt_PVAV) 356 if (svt != SVt_PVAV)
271 croak ("encountered CBOR::XS::Tagged object that isn't an array"); 357 croak ("encountered CBOR::XS::Tagged object that isn't an array");
272 358
273 encode_uint (enc, 0xc0, SvUV (*av_fetch ((AV *)sv, 0, 1))); 359 encode_uint (enc, 0xc0, SvUV (*av_fetch ((AV *)sv, 0, 1)));
274 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1)); 360 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1));
275 } 361 }
362 else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
363 {
364 dSP;
365
366 ENTER; SAVETMPS; PUSHMARK (SP);
367 // we re-bless the reference to get overload and other niceties right
368 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
369
370 PUTBACK;
371 // G_SCALAR ensures that return value is 1
372 call_sv ((SV *)GvCV (method), G_SCALAR);
373 SPAGAIN;
374
375 // catch this surprisingly common error
376 if (SvROK (TOPs) && SvRV (TOPs) == sv)
377 croak ("%s::TO_CBOR method returned same object as was passed instead of a new one", HvNAME (stash));
378
379 encode_sv (enc, POPs);
380
381 PUTBACK;
382
383 FREETMPS; LEAVE;
384 }
385 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)
386 {
387 dSP;
388
389 ENTER; SAVETMPS; PUSHMARK (SP);
390 EXTEND (SP, 2);
391 // we re-bless the reference to get overload and other niceties right
392 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
393 PUSHs (sv_cbor);
394
395 PUTBACK;
396 int count = call_sv ((SV *)GvCV (method), G_ARRAY);
397 SPAGAIN;
398
399 // catch this surprisingly common error
400 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)
401 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));
402
403 encode_tag (enc, CBOR_TAG_PERL_OBJECT);
404 encode_uint (enc, 0x80, count + 1);
405 encode_str (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
406
407 while (count)
408 encode_sv (enc, SP[1 - count--]);
409
410 PUTBACK;
411
412 FREETMPS; LEAVE;
413 }
276 else 414 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", 415 croak ("encountered object '%s', but no TO_CBOR or FREEZE methods available on it",
306 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 416 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
307 }
308 } 417 }
309 else if (svt == SVt_PVHV) 418 else if (svt == SVt_PVHV)
310 encode_hv (enc, (HV *)sv); 419 encode_hv (enc, (HV *)sv);
311 else if (svt == SVt_PVAV) 420 else if (svt == SVt_PVAV)
312 encode_av (enc, (AV *)sv); 421 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 422 else
331 croak ("encountered %s, but CBOR can only represent references to arrays or hashes", 423 {
332 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 424 encode_tag (enc, CBOR_TAG_INDIRECTION);
425 encode_sv (enc, sv);
426 }
333} 427}
334 428
335static void 429static void
336encode_nv (enc_t *enc, SV *sv) 430encode_nv (enc_t *enc, SV *sv)
337{ 431{
402} 496}
403 497
404static SV * 498static SV *
405encode_cbor (SV *scalar, CBOR *cbor) 499encode_cbor (SV *scalar, CBOR *cbor)
406{ 500{
407 enc_t enc; 501 enc_t enc = { };
408 502
409 enc.cbor = *cbor; 503 enc.cbor = *cbor;
410 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 504 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
411 enc.cur = SvPVX (enc.sv); 505 enc.cur = SvPVX (enc.sv);
412 enc.end = SvEND (enc.sv); 506 enc.end = SvEND (enc.sv);
413 enc.depth = 0;
414 507
415 SvPOK_only (enc.sv); 508 SvPOK_only (enc.sv);
509
510 if (cbor->flags & (F_DEDUP_STRINGS | F_DEDUP_KEYS))
511 {
512 encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE);
513 enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ());
514 enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ());
515 }
516
416 encode_sv (&enc, scalar); 517 encode_sv (&enc, scalar);
417 518
418 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 519 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
419 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings 520 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
420 521
434 U8 *end; // end of input string 535 U8 *end; // end of input string
435 const char *err; // parse error, if != 0 536 const char *err; // parse error, if != 0
436 CBOR cbor; 537 CBOR cbor;
437 U32 depth; // recursion depth 538 U32 depth; // recursion depth
438 U32 maxdepth; // recursion depth limit 539 U32 maxdepth; // recursion depth limit
540 AV *shareable;
541 AV *stringref;
439} dec_t; 542} dec_t;
440 543
441#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE 544#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE
442 545
443#define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data") 546#define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data")
537 SvREFCNT_dec (av); 640 SvREFCNT_dec (av);
538 DEC_DEC_DEPTH; 641 DEC_DEC_DEPTH;
539 return &PL_sv_undef; 642 return &PL_sv_undef;
540} 643}
541 644
645static void
646decode_he (dec_t *dec, HV *hv)
647{
648 // for speed reasons, we specialcase single-string
649 // byte or utf-8 strings as keys.
650
651 if (*dec->cur >= 0x40 && *dec->cur <= 0x40 + 27)
652 {
653 I32 len = decode_uint (dec);
654 char *key = (char *)dec->cur;
655
656 dec->cur += len;
657
658 if (ecb_expect_false (dec->stringref))
659 av_push (dec->stringref, newSVpvn (key, len));
660
661 hv_store (hv, key, len, decode_sv (dec), 0);
662 }
663 else if (*dec->cur >= 0x60 && *dec->cur <= 0x60 + 27)
664 {
665 I32 len = decode_uint (dec);
666 char *key = (char *)dec->cur;
667
668 dec->cur += len;
669
670 if (ecb_expect_false (dec->stringref))
671 av_push (dec->stringref, newSVpvn_utf8 (key, len, 1));
672
673 hv_store (hv, key, -len, decode_sv (dec), 0);
674 }
675 else
676 {
677 SV *k = decode_sv (dec);
678 SV *v = decode_sv (dec);
679
680 hv_store_ent (hv, k, v, 0);
681 SvREFCNT_dec (k);
682 }
683}
684
542static SV * 685static SV *
543decode_hv (dec_t *dec) 686decode_hv (dec_t *dec)
544{ 687{
545 HV *hv = newHV (); 688 HV *hv = newHV ();
546 689
558 { 701 {
559 ++dec->cur; 702 ++dec->cur;
560 break; 703 break;
561 } 704 }
562 705
563 SV *k = decode_sv (dec); 706 decode_he (dec, hv);
564 SV *v = decode_sv (dec);
565
566 hv_store_ent (hv, k, v, 0);
567 } 707 }
568 } 708 }
569 else 709 else
570 { 710 {
571 int len = decode_uint (dec); 711 int pairs = decode_uint (dec);
572 712
573 while (len--) 713 while (pairs--)
574 { 714 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 } 715 }
581 716
582 DEC_DEC_DEPTH; 717 DEC_DEC_DEPTH;
583 return newRV_noinc ((SV *)hv); 718 return newRV_noinc ((SV *)hv);
584 719
623 } 758 }
624 759
625 if (utf8) 760 if (utf8)
626 SvUTF8_on (sv); 761 SvUTF8_on (sv);
627 762
763 if (ecb_expect_false (dec->stringref))
764 av_push (dec->stringref, SvREFCNT_inc_NN (sv));
765
628 return sv; 766 return sv;
629 767
630fail: 768fail:
631 SvREFCNT_dec (sv); 769 SvREFCNT_dec (sv);
632 return &PL_sv_undef; 770 return &PL_sv_undef;
633} 771}
634 772
635static SV * 773static SV *
636decode_tagged (dec_t *dec) 774decode_tagged (dec_t *dec)
637{ 775{
776 SV *sv = 0;
638 UV tag = decode_uint (dec); 777 UV tag = decode_uint (dec);
778
779 WANT (1);
780
781 switch (tag)
782 {
783 case CBOR_TAG_MAGIC:
784 sv = decode_sv (dec);
785 break;
786
787 case CBOR_TAG_INDIRECTION:
788 sv = newRV_noinc (decode_sv (dec));
789 break;
790
791 case CBOR_TAG_STRINGREF_NAMESPACE:
792 {
793 ENTER; SAVETMPS;
794
795 SAVESPTR (dec->stringref);
796 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ());
797
798 sv = decode_sv (dec);
799
800 FREETMPS; LEAVE;
801 }
802 break;
803
804 case CBOR_TAG_STRINGREF:
805 {
806 if ((*dec->cur >> 5) != 0)
807 ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
808
809 UV idx = decode_uint (dec);
810
811 if (!dec->stringref || (int)idx > AvFILLp (dec->stringref))
812 ERR ("corrupted CBOR data (stringref index out of bounds or outside namespace)");
813
814 sv = newSVsv (AvARRAY (dec->stringref)[idx]);
815 }
816 break;
817
818 case CBOR_TAG_VALUE_SHAREABLE:
819 {
820 if (ecb_expect_false (!dec->shareable))
821 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ());
822
823 sv = newSV (0);
824 av_push (dec->shareable, SvREFCNT_inc_NN (sv));
825
639 SV *sv = decode_sv (dec); 826 SV *osv = decode_sv (dec);
827 sv_setsv (sv, osv);
828 SvREFCNT_dec_NN (osv);
829 }
830 break;
640 831
641 if (tag == CBOR_TAG_MAGIC) // 2.4.5 Self-Describe CBOR 832 case CBOR_TAG_VALUE_SHAREDREF:
642 return sv; 833 {
834 if ((*dec->cur >> 5) != 0)
835 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
643 836
837 UV idx = decode_uint (dec);
838
839 if (!dec->shareable || (int)idx > AvFILLp (dec->shareable))
840 ERR ("corrupted CBOR data (sharedref index out of bounds)");
841
842 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]);
843 }
844 break;
845
846 case CBOR_TAG_PERL_OBJECT:
847 {
848 sv = decode_sv (dec);
849
850 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
851 ERR ("corrupted CBOR data (non-array perl object)");
852
853 AV *av = (AV *)SvRV (sv);
854 int len = av_len (av) + 1;
855 HV *stash = gv_stashsv (*av_fetch (av, 0, 1), 0);
856
857 if (!stash)
858 ERR ("cannot decode perl-object (package does not exist)");
859
860 GV *method = gv_fetchmethod_autoload (stash, "THAW", 0);
861
862 if (!method)
863 ERR ("cannot decode perl-object (package does not have a THAW method)");
864
865 dSP;
866
867 ENTER; SAVETMPS; PUSHMARK (SP);
868 EXTEND (SP, len + 1);
869 // we re-bless the reference to get overload and other niceties right
870 PUSHs (*av_fetch (av, 0, 1));
871 PUSHs (sv_cbor);
872
873 int i;
874
875 for (i = 1; i < len; ++i)
876 PUSHs (*av_fetch (av, i, 1));
877
878 PUTBACK;
879 call_sv ((SV *)GvCV (method), G_SCALAR | G_EVAL);
880 SPAGAIN;
881
882 if (SvTRUE (ERRSV))
883 {
884 FREETMPS; LEAVE;
885 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV))));
886 }
887
888 SvREFCNT_dec (sv);
889 sv = SvREFCNT_inc (POPs);
890
891 PUTBACK;
892
893 FREETMPS; LEAVE;
894 }
895 break;
896
897 default:
898 {
899 sv = decode_sv (dec);
900
644 AV *av = newAV (); 901 AV *av = newAV ();
645 av_push (av, newSVuv (tag)); 902 av_push (av, newSVuv (tag));
646 av_push (av, sv); 903 av_push (av, sv);
647 904
648 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 905 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
649 ? cbor_tagged_stash 906 ? cbor_tagged_stash
650 : gv_stashpv ("CBOR::XS::Tagged" , 1); 907 : gv_stashpv ("CBOR::XS::Tagged" , 1);
651 908
652 return sv_bless (newRV_noinc ((SV *)av), tagged_stash); 909 sv = sv_bless (newRV_noinc ((SV *)av), tagged_stash);
910 }
911 break;
912 }
913
914 return sv;
915
916fail:
917 SvREFCNT_dec (sv);
918 return &PL_sv_undef;
653} 919}
654 920
655static SV * 921static SV *
656decode_sv (dec_t *dec) 922decode_sv (dec_t *dec)
657{ 923{
676 case 7: // misc 942 case 7: // misc
677 switch (*dec->cur++ & 31) 943 switch (*dec->cur++ & 31)
678 { 944 {
679 case 20: 945 case 20:
680#if CBOR_SLOW 946#if CBOR_SLOW
681 cbor_false = get_bool ("CBOR::XS::false"); 947 types_false = get_bool ("Types::Serialiser::false");
682#endif 948#endif
683 return newSVsv (cbor_false); 949 return newSVsv (types_false);
684 case 21: 950 case 21:
685#if CBOR_SLOW 951#if CBOR_SLOW
686 cbor_true = get_bool ("CBOR::XS::true"); 952 types_true = get_bool ("Types::Serialiser::true");
687#endif 953#endif
688 return newSVsv (cbor_true); 954 return newSVsv (types_true);
689 case 22: 955 case 22:
690 return newSVsv (&PL_sv_undef); 956 return newSVsv (&PL_sv_undef);
957 case 23:
958#if CBOR_SLOW
959 types_error = get_bool ("Types::Serialiser::error");
960#endif
961 return newSVsv (types_error);
691 962
692 case 25: 963 case 25:
693 { 964 {
694 WANT (2); 965 WANT (2);
695 966
739} 1010}
740 1011
741static SV * 1012static SV *
742decode_cbor (SV *string, CBOR *cbor, char **offset_return) 1013decode_cbor (SV *string, CBOR *cbor, char **offset_return)
743{ 1014{
744 dec_t dec; 1015 dec_t dec = { };
745 SV *sv; 1016 SV *sv;
1017 STRLEN len;
1018 char *data = SvPVbyte (string, len);
746 1019
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) 1020 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", 1021 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); 1022 (unsigned long)len, (unsigned long)cbor->max_size);
778 }
779
780 sv_utf8_downgrade (string, 0);
781 1023
782 dec.cbor = *cbor; 1024 dec.cbor = *cbor;
783 dec.cur = (U8 *)SvPVX (string); 1025 dec.cur = (U8 *)data;
784 dec.end = (U8 *)SvEND (string); 1026 dec.end = (U8 *)data + len;
785 dec.err = 0;
786 dec.depth = 0;
787 1027
788 sv = decode_sv (&dec); 1028 sv = decode_sv (&dec);
789 1029
790 if (offset_return) 1030 if (offset_return)
791 *offset_return = dec.cur; 1031 *offset_return = dec.cur;
795 dec.err = "garbage after CBOR object"; 1035 dec.err = "garbage after CBOR object";
796 1036
797 if (dec.err) 1037 if (dec.err)
798 { 1038 {
799 SvREFCNT_dec (sv); 1039 SvREFCNT_dec (sv);
800 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)SvPVX (string), (int)(uint8_t)*dec.cur); 1040 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur);
801 } 1041 }
802 1042
803 sv = sv_2mortal (sv); 1043 sv = sv_2mortal (sv);
804 1044
805 return sv; 1045 return sv;
811MODULE = CBOR::XS PACKAGE = CBOR::XS 1051MODULE = CBOR::XS PACKAGE = CBOR::XS
812 1052
813BOOT: 1053BOOT:
814{ 1054{
815 cbor_stash = gv_stashpv ("CBOR::XS" , 1); 1055 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); 1056 cbor_tagged_stash = gv_stashpv ("CBOR::XS::Tagged" , 1);
818 1057
819 cbor_true = get_bool ("CBOR::XS::true"); 1058 types_boolean_stash = gv_stashpv ("Types::Serialiser::Boolean", 1);
820 cbor_false = get_bool ("CBOR::XS::false"); 1059 types_error_stash = gv_stashpv ("Types::Serialiser::Error" , 1);
1060
1061 types_true = get_bool ("Types::Serialiser::true" );
1062 types_false = get_bool ("Types::Serialiser::false");
1063 types_error = get_bool ("Types::Serialiser::error");
1064
1065 sv_cbor = newSVpv ("CBOR", 0);
1066 SvREADONLY_on (sv_cbor);
821} 1067}
822 1068
823PROTOTYPES: DISABLE 1069PROTOTYPES: DISABLE
824 1070
825void CLONE (...) 1071void CLONE (...)
826 CODE: 1072 CODE:
827 cbor_stash = 0; 1073 cbor_stash = 0;
828 cbor_boolean_stash = 0;
829 cbor_tagged_stash = 0; 1074 cbor_tagged_stash = 0;
1075 types_error_stash = 0;
1076 types_boolean_stash = 0;
830 1077
831void new (char *klass) 1078void new (char *klass)
832 PPCODE: 1079 PPCODE:
833{ 1080{
834 SV *pv = NEWSV (0, sizeof (CBOR)); 1081 SV *pv = NEWSV (0, sizeof (CBOR));
842 1089
843void shrink (CBOR *self, int enable = 1) 1090void shrink (CBOR *self, int enable = 1)
844 ALIAS: 1091 ALIAS:
845 shrink = F_SHRINK 1092 shrink = F_SHRINK
846 allow_unknown = F_ALLOW_UNKNOWN 1093 allow_unknown = F_ALLOW_UNKNOWN
1094 allow_sharing = F_ALLOW_SHARING
1095 dedup_keys = F_DEDUP_KEYS
1096 dedup_strings = F_DEDUP_STRINGS
847 PPCODE: 1097 PPCODE:
848{ 1098{
849 if (enable) 1099 if (enable)
850 self->flags |= ix; 1100 self->flags |= ix;
851 else 1101 else
856 1106
857void get_shrink (CBOR *self) 1107void get_shrink (CBOR *self)
858 ALIAS: 1108 ALIAS:
859 get_shrink = F_SHRINK 1109 get_shrink = F_SHRINK
860 get_allow_unknown = F_ALLOW_UNKNOWN 1110 get_allow_unknown = F_ALLOW_UNKNOWN
1111 get_allow_sharing = F_ALLOW_SHARING
1112 get_dedup_keys = F_DEDUP_KEYS
1113 get_dedup_strings = F_DEDUP_STRINGS
861 PPCODE: 1114 PPCODE:
862 XPUSHs (boolSV (self->flags & ix)); 1115 XPUSHs (boolSV (self->flags & ix));
863 1116
864void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) 1117void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
865 PPCODE: 1118 PPCODE:
880int get_max_size (CBOR *self) 1133int get_max_size (CBOR *self)
881 CODE: 1134 CODE:
882 RETVAL = self->max_size; 1135 RETVAL = self->max_size;
883 OUTPUT: 1136 OUTPUT:
884 RETVAL 1137 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 1138
921void encode (CBOR *self, SV *scalar) 1139void encode (CBOR *self, SV *scalar)
922 PPCODE: 1140 PPCODE:
923 PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN; 1141 PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN;
924 XPUSHs (scalar); 1142 XPUSHs (scalar);
937 EXTEND (SP, 2); 1155 EXTEND (SP, 2);
938 PUSHs (sv); 1156 PUSHs (sv);
939 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); 1157 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
940} 1158}
941 1159
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 1160PROTOTYPES: ENABLE
952 1161
953void encode_cbor (SV *scalar) 1162void encode_cbor (SV *scalar)
954 PPCODE: 1163 PPCODE:
955{ 1164{

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines