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 1.9 by root, Sun Oct 27 10:17:12 2013 UTC vs.
Revision 1.32 by root, Thu Nov 28 12:08:07 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#ifndef SvREFCNT_dec_NN
25# define SvREFCNT_dec_NN(sv) SvREFCNT_dec (sv)
26#endif
27
14// known tags 28// known tags
15enum cbor_tag 29enum cbor_tag
16{ 30{
17 // inofficial extensions (pending iana registration) 31 // extensions
18 CBOR_TAG_PERL_OBJECT = 256, 32 CBOR_TAG_STRINGREF = 25, // http://cbor.schmorp.de/stringref
19 CBOR_TAG_GENERIC_OBJECT = 257, 33 CBOR_TAG_PERL_OBJECT = 26, // http://cbor.schmorp.de/perl-object
34 CBOR_TAG_GENERIC_OBJECT = 27, // http://cbor.schmorp.de/generic-object
35 CBOR_TAG_VALUE_SHAREABLE = 28, // http://cbor.schmorp.de/value-sharing
36 CBOR_TAG_VALUE_SHAREDREF = 29, // http://cbor.schmorp.de/value-sharing
37 CBOR_TAG_STRINGREF_NAMESPACE = 256, // http://cbor.schmorp.de/stringref
38 CBOR_TAG_INDIRECTION = 22098, // http://cbor.schmorp.de/indirection
20 39
21 // rfc7049 40 // rfc7049
22 CBOR_TAG_DATETIME = 0, // rfc4287, utf-8 41 CBOR_TAG_DATETIME = 0, // rfc4287, utf-8
23 CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any 42 CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any
24 CBOR_TAG_POS_BIGNUM = 2, // byte string 43 CBOR_TAG_POS_BIGNUM = 2, // byte string
38 CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8 57 CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8
39 58
40 CBOR_TAG_MAGIC = 55799 // self-describe cbor 59 CBOR_TAG_MAGIC = 55799 // self-describe cbor
41}; 60};
42 61
43#define F_SHRINK 0x00000200UL 62#define F_SHRINK 0x00000001UL
44#define F_ALLOW_UNKNOWN 0x00002000UL 63#define F_ALLOW_UNKNOWN 0x00000002UL
64#define F_ALLOW_SHARING 0x00000004UL
65#define F_PACK_STRINGS 0x00000008UL
45 66
46#define INIT_SIZE 32 // initial scalar size to be allocated 67#define INIT_SIZE 32 // initial scalar size to be allocated
47 68
48#define SB do { 69#define SB do {
49#define SE } while (0) 70#define SE } while (0)
60#else 81#else
61# define CBOR_SLOW 0 82# define CBOR_SLOW 0
62# define CBOR_STASH cbor_stash 83# define CBOR_STASH cbor_stash
63#endif 84#endif
64 85
65static 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::
66static SV *cbor_true, *cbor_false; 87static SV *types_true, *types_false, *types_error, *sv_cbor, *default_filter;
67 88
68typedef struct { 89typedef struct {
69 U32 flags; 90 U32 flags;
70 U32 max_depth; 91 U32 max_depth;
71 STRLEN max_size; 92 STRLEN max_size;
93 SV *filter;
72} CBOR; 94} CBOR;
73 95
74ecb_inline void 96ecb_inline void
75cbor_init (CBOR *cbor) 97cbor_init (CBOR *cbor)
76{ 98{
77 Zero (cbor, 1, CBOR); 99 Zero (cbor, 1, CBOR);
78 cbor->max_depth = 512; 100 cbor->max_depth = 512;
101}
102
103ecb_inline void
104cbor_free (CBOR *cbor)
105{
106 SvREFCNT_dec (cbor->filter);
79} 107}
80 108
81///////////////////////////////////////////////////////////////////////////// 109/////////////////////////////////////////////////////////////////////////////
82// utility functions 110// utility functions
83 111
105 SvPV_renew (sv, SvCUR (sv) + 1); 133 SvPV_renew (sv, SvCUR (sv) + 1);
106#endif 134#endif
107 } 135 }
108} 136}
109 137
110///////////////////////////////////////////////////////////////////////////// 138// minimum length of a string to be registered for stringref
111// fp hell 139ecb_inline int
112 140minimum_string_length (UV idx)
113//TODO 141{
142 return idx > 23
143 ? idx > 0xffU
144 ? idx > 0xffffU
145 ? idx > 0xffffffffU
146 ? 11
147 : 7
148 : 5
149 : 4
150 : 3;
151}
114 152
115///////////////////////////////////////////////////////////////////////////// 153/////////////////////////////////////////////////////////////////////////////
116// encoder 154// encoder
117 155
118// structure used for encoding CBOR 156// structure used for encoding CBOR
121 char *cur; // SvPVX (sv) + current output position 159 char *cur; // SvPVX (sv) + current output position
122 char *end; // SvEND (sv) 160 char *end; // SvEND (sv)
123 SV *sv; // result scalar 161 SV *sv; // result scalar
124 CBOR cbor; 162 CBOR cbor;
125 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;
126} enc_t; 168} enc_t;
127 169
128ecb_inline void 170ecb_inline void
129need (enc_t *enc, STRLEN len) 171need (enc_t *enc, STRLEN len)
130{ 172{
147static void 189static void
148encode_uint (enc_t *enc, int major, UV len) 190encode_uint (enc_t *enc, int major, UV len)
149{ 191{
150 need (enc, 9); 192 need (enc, 9);
151 193
152 if (len < 24) 194 if (ecb_expect_true (len < 24))
153 *enc->cur++ = major | len; 195 *enc->cur++ = major | len;
154 else if (len <= 0xff) 196 else if (ecb_expect_true (len <= 0xff))
155 { 197 {
156 *enc->cur++ = major | 24; 198 *enc->cur++ = major | 24;
157 *enc->cur++ = len; 199 *enc->cur++ = len;
158 } 200 }
159 else if (len <= 0xffff) 201 else if (len <= 0xffff)
182 *enc->cur++ = len >> 8; 224 *enc->cur++ = len >> 8;
183 *enc->cur++ = len; 225 *enc->cur++ = len;
184 } 226 }
185} 227}
186 228
187static void 229ecb_inline void
230encode_tag (enc_t *enc, UV tag)
231{
232 encode_uint (enc, 0xc0, tag);
233}
234
235ecb_inline void
188encode_str (enc_t *enc, int utf8, char *str, STRLEN len) 236encode_str (enc_t *enc, int utf8, char *str, STRLEN len)
189{ 237{
190 encode_uint (enc, utf8 ? 0x60 : 0x40, len); 238 encode_uint (enc, utf8 ? 0x60 : 0x40, len);
191 need (enc, len); 239 need (enc, len);
192 memcpy (enc->cur, str, len); 240 memcpy (enc->cur, str, len);
193 enc->cur += len; 241 enc->cur += len;
194} 242}
195 243
244static void
245encode_strref (enc_t *enc, int utf8, char *str, STRLEN len)
246{
247 if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS))
248 {
249 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1);
250
251 if (SvOK (*svp))
252 {
253 // already registered, use stringref
254 encode_tag (enc, CBOR_TAG_STRINGREF);
255 encode_uint (enc, 0x00, SvUV (*svp));
256 return;
257 }
258 else if (len >= minimum_string_length (enc->stringref_idx))
259 {
260 // register only
261 sv_setuv (*svp, enc->stringref_idx);
262 ++enc->stringref_idx;
263 }
264 }
265
266 encode_str (enc, utf8, str, len);
267}
268
196static void encode_sv (enc_t *enc, SV *sv); 269static void encode_sv (enc_t *enc, SV *sv);
197 270
198static void 271static void
199encode_av (enc_t *enc, AV *av) 272encode_av (enc_t *enc, AV *av)
200{ 273{
237 while ((he = hv_iternext (hv))) 310 while ((he = hv_iternext (hv)))
238 { 311 {
239 if (HeKLEN (he) == HEf_SVKEY) 312 if (HeKLEN (he) == HEf_SVKEY)
240 encode_sv (enc, HeSVKEY (he)); 313 encode_sv (enc, HeSVKEY (he));
241 else 314 else
242 encode_str (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he)); 315 encode_strref (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he));
243 316
244 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he)); 317 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he));
245 } 318 }
246 319
247 if (mg) 320 if (mg)
252 325
253// encode objects, arrays and special \0=false and \1=true values. 326// encode objects, arrays and special \0=false and \1=true values.
254static void 327static void
255encode_rv (enc_t *enc, SV *sv) 328encode_rv (enc_t *enc, SV *sv)
256{ 329{
257 svtype svt;
258
259 SvGETMAGIC (sv); 330 SvGETMAGIC (sv);
331
332 if (ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING)
333 && ecb_expect_false (SvREFCNT (sv) > 1))
334 {
335 if (!enc->shareable)
336 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
337
338 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
339
340 if (SvOK (*svp))
341 {
342 encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF);
343 encode_uint (enc, 0x00, SvUV (*svp));
344 return;
345 }
346 else
347 {
348 sv_setuv (*svp, enc->shareable_idx);
349 ++enc->shareable_idx;
350 encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE);
351 }
352 }
353
260 svt = SvTYPE (sv); 354 svtype svt = SvTYPE (sv);
261 355
262 if (ecb_expect_false (SvOBJECT (sv))) 356 if (ecb_expect_false (SvOBJECT (sv)))
263 { 357 {
264 HV *boolean_stash = !CBOR_SLOW || cbor_boolean_stash 358 HV *boolean_stash = !CBOR_SLOW || types_boolean_stash
265 ? cbor_boolean_stash 359 ? types_boolean_stash
266 : gv_stashpv ("CBOR::XS::Boolean", 1); 360 : gv_stashpv ("Types::Serialiser::Boolean", 1);
361 HV *error_stash = !CBOR_SLOW || types_error_stash
362 ? types_error_stash
363 : gv_stashpv ("Types::Serialiser::Error", 1);
267 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 364 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
268 ? cbor_tagged_stash 365 ? cbor_tagged_stash
269 : gv_stashpv ("CBOR::XS::Tagged" , 1); 366 : gv_stashpv ("CBOR::XS::Tagged" , 1);
270 367
368 HV *stash = SvSTASH (sv);
369 GV *method;
370
271 if (SvSTASH (sv) == boolean_stash) 371 if (stash == boolean_stash)
272 encode_ch (enc, SvIV (sv) ? 0xe0 | 21 : 0xe0 | 20); 372 encode_ch (enc, SvIV (sv) ? 0xe0 | 21 : 0xe0 | 20);
373 else if (stash == error_stash)
374 encode_ch (enc, 0xe0 | 23);
273 else if (SvSTASH (sv) == tagged_stash) 375 else if (stash == tagged_stash)
274 { 376 {
275 if (svt != SVt_PVAV) 377 if (svt != SVt_PVAV)
276 croak ("encountered CBOR::XS::Tagged object that isn't an array"); 378 croak ("encountered CBOR::XS::Tagged object that isn't an array");
277 379
278 encode_uint (enc, 0xc0, SvUV (*av_fetch ((AV *)sv, 0, 1))); 380 encode_uint (enc, 0xc0, SvUV (*av_fetch ((AV *)sv, 0, 1)));
279 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1)); 381 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1));
280 } 382 }
383 else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
384 {
385 dSP;
386
387 ENTER; SAVETMPS; PUSHMARK (SP);
388 // we re-bless the reference to get overload and other niceties right
389 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
390
391 PUTBACK;
392 // G_SCALAR ensures that return value is 1
393 call_sv ((SV *)GvCV (method), G_SCALAR);
394 SPAGAIN;
395
396 // catch this surprisingly common error
397 if (SvROK (TOPs) && SvRV (TOPs) == sv)
398 croak ("%s::TO_CBOR method returned same object as was passed instead of a new one", HvNAME (stash));
399
400 encode_sv (enc, POPs);
401
402 PUTBACK;
403
404 FREETMPS; LEAVE;
405 }
406 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)
407 {
408 dSP;
409
410 ENTER; SAVETMPS; PUSHMARK (SP);
411 EXTEND (SP, 2);
412 // we re-bless the reference to get overload and other niceties right
413 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
414 PUSHs (sv_cbor);
415
416 PUTBACK;
417 int count = call_sv ((SV *)GvCV (method), G_ARRAY);
418 SPAGAIN;
419
420 // catch this surprisingly common error
421 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)
422 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));
423
424 encode_tag (enc, CBOR_TAG_PERL_OBJECT);
425 encode_uint (enc, 0x80, count + 1);
426 encode_strref (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
427
428 while (count)
429 encode_sv (enc, SP[1 - count--]);
430
431 PUTBACK;
432
433 FREETMPS; LEAVE;
434 }
281 else 435 else
282 {
283 // we re-bless the reference to get overload and other niceties right
284 GV *to_cbor = gv_fetchmethod_autoload (SvSTASH (sv), "TO_CBOR", 0);
285
286 if (to_cbor)
287 {
288 dSP;
289
290 ENTER; SAVETMPS; PUSHMARK (SP);
291 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv)));
292
293 // calling with G_SCALAR ensures that we always get a 1 return value
294 PUTBACK;
295 call_sv ((SV *)GvCV (to_cbor), G_SCALAR);
296 SPAGAIN;
297
298 // catch this surprisingly common error
299 if (SvROK (TOPs) && SvRV (TOPs) == sv)
300 croak ("%s::TO_CBOR method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv)));
301
302 sv = POPs;
303 PUTBACK;
304
305 encode_sv (enc, sv);
306
307 FREETMPS; LEAVE;
308 }
309 else
310 croak ("encountered object '%s', but no TO_CBOR method available on it", 436 croak ("encountered object '%s', but no TO_CBOR or FREEZE methods available on it",
311 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 437 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
312 }
313 } 438 }
314 else if (svt == SVt_PVHV) 439 else if (svt == SVt_PVHV)
315 encode_hv (enc, (HV *)sv); 440 encode_hv (enc, (HV *)sv);
316 else if (svt == SVt_PVAV) 441 else if (svt == SVt_PVAV)
317 encode_av (enc, (AV *)sv); 442 encode_av (enc, (AV *)sv);
318 else if (svt < SVt_PVAV)
319 {
320 STRLEN len = 0;
321 char *pv = svt ? SvPV (sv, len) : 0;
322
323 if (len == 1 && *pv == '1')
324 encode_ch (enc, 0xe0 | 21);
325 else if (len == 1 && *pv == '0')
326 encode_ch (enc, 0xe0 | 20);
327 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
328 encode_ch (enc, 0xe0 | 23);
329 else
330 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
331 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
332 }
333 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
334 encode_ch (enc, 0xe0 | 23);
335 else 443 else
336 croak ("encountered %s, but CBOR can only represent references to arrays or hashes", 444 {
337 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 445 encode_tag (enc, CBOR_TAG_INDIRECTION);
446 encode_sv (enc, sv);
447 }
338} 448}
339 449
340static void 450static void
341encode_nv (enc_t *enc, SV *sv) 451encode_nv (enc_t *enc, SV *sv)
342{ 452{
380 490
381 if (SvPOKp (sv)) 491 if (SvPOKp (sv))
382 { 492 {
383 STRLEN len; 493 STRLEN len;
384 char *str = SvPV (sv, len); 494 char *str = SvPV (sv, len);
385 encode_str (enc, SvUTF8 (sv), str, len); 495 encode_strref (enc, SvUTF8 (sv), str, len);
386 } 496 }
387 else if (SvNOKp (sv)) 497 else if (SvNOKp (sv))
388 encode_nv (enc, sv); 498 encode_nv (enc, sv);
389 else if (SvIOKp (sv)) 499 else if (SvIOKp (sv))
390 { 500 {
407} 517}
408 518
409static SV * 519static SV *
410encode_cbor (SV *scalar, CBOR *cbor) 520encode_cbor (SV *scalar, CBOR *cbor)
411{ 521{
412 enc_t enc; 522 enc_t enc = { };
413 523
414 enc.cbor = *cbor; 524 enc.cbor = *cbor;
415 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 525 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
416 enc.cur = SvPVX (enc.sv); 526 enc.cur = SvPVX (enc.sv);
417 enc.end = SvEND (enc.sv); 527 enc.end = SvEND (enc.sv);
418 enc.depth = 0;
419 528
420 SvPOK_only (enc.sv); 529 SvPOK_only (enc.sv);
530
531 if (cbor->flags & F_PACK_STRINGS)
532 {
533 encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE);
534 enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ());
535 enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ());
536 }
537
421 encode_sv (&enc, scalar); 538 encode_sv (&enc, scalar);
422 539
423 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 540 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
424 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings 541 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
425 542
439 U8 *end; // end of input string 556 U8 *end; // end of input string
440 const char *err; // parse error, if != 0 557 const char *err; // parse error, if != 0
441 CBOR cbor; 558 CBOR cbor;
442 U32 depth; // recursion depth 559 U32 depth; // recursion depth
443 U32 maxdepth; // recursion depth limit 560 U32 maxdepth; // recursion depth limit
561 AV *shareable;
562 AV *stringref;
563 SV *decode_tagged;
444} dec_t; 564} dec_t;
445 565
446#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE 566#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE
447 567
448#define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data") 568#define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data")
542 SvREFCNT_dec (av); 662 SvREFCNT_dec (av);
543 DEC_DEC_DEPTH; 663 DEC_DEC_DEPTH;
544 return &PL_sv_undef; 664 return &PL_sv_undef;
545} 665}
546 666
667static void
668decode_he (dec_t *dec, HV *hv)
669{
670 // for speed reasons, we specialcase single-string
671 // byte or utf-8 strings as keys, but only when !stringref
672
673 if (ecb_expect_true (!dec->stringref))
674 if (*dec->cur >= 0x40 && *dec->cur <= 0x40 + 27)
675 {
676 I32 len = decode_uint (dec);
677 char *key = (char *)dec->cur;
678
679 dec->cur += len;
680
681 if (ecb_expect_false (dec->stringref))
682 av_push (dec->stringref, newSVpvn (key, len));
683
684 hv_store (hv, key, len, decode_sv (dec), 0);
685
686 return;
687 }
688 else if (*dec->cur >= 0x60 && *dec->cur <= 0x60 + 27)
689 {
690 I32 len = decode_uint (dec);
691 char *key = (char *)dec->cur;
692
693 dec->cur += len;
694
695 if (ecb_expect_false (dec->stringref))
696 av_push (dec->stringref, newSVpvn_utf8 (key, len, 1));
697
698 hv_store (hv, key, -len, decode_sv (dec), 0);
699
700 return;
701 }
702
703 SV *k = decode_sv (dec);
704 SV *v = decode_sv (dec);
705
706 hv_store_ent (hv, k, v, 0);
707 SvREFCNT_dec (k);
708}
709
547static SV * 710static SV *
548decode_hv (dec_t *dec) 711decode_hv (dec_t *dec)
549{ 712{
550 HV *hv = newHV (); 713 HV *hv = newHV ();
551 714
563 { 726 {
564 ++dec->cur; 727 ++dec->cur;
565 break; 728 break;
566 } 729 }
567 730
568 SV *k = decode_sv (dec); 731 decode_he (dec, hv);
569 SV *v = decode_sv (dec);
570
571 hv_store_ent (hv, k, v, 0);
572 } 732 }
573 } 733 }
574 else 734 else
575 { 735 {
576 int len = decode_uint (dec); 736 int pairs = decode_uint (dec);
577 737
578 while (len--) 738 while (pairs--)
579 { 739 decode_he (dec, hv);
580 SV *k = decode_sv (dec);
581 SV *v = decode_sv (dec);
582
583 hv_store_ent (hv, k, v, 0);
584 }
585 } 740 }
586 741
587 DEC_DEC_DEPTH; 742 DEC_DEC_DEPTH;
588 return newRV_noinc ((SV *)hv); 743 return newRV_noinc ((SV *)hv);
589 744
623 STRLEN len = decode_uint (dec); 778 STRLEN len = decode_uint (dec);
624 779
625 WANT (len); 780 WANT (len);
626 sv = newSVpvn (dec->cur, len); 781 sv = newSVpvn (dec->cur, len);
627 dec->cur += len; 782 dec->cur += len;
783
784 if (ecb_expect_false (dec->stringref)
785 && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1))
786 av_push (dec->stringref, SvREFCNT_inc_NN (sv));
628 } 787 }
629 788
630 if (utf8) 789 if (utf8)
631 SvUTF8_on (sv); 790 SvUTF8_on (sv);
632 791
638} 797}
639 798
640static SV * 799static SV *
641decode_tagged (dec_t *dec) 800decode_tagged (dec_t *dec)
642{ 801{
802 SV *sv = 0;
643 UV tag = decode_uint (dec); 803 UV tag = decode_uint (dec);
804
805 WANT (1);
806
807 switch (tag)
808 {
809 case CBOR_TAG_MAGIC:
810 sv = decode_sv (dec);
811 break;
812
813 case CBOR_TAG_INDIRECTION:
814 sv = newRV_noinc (decode_sv (dec));
815 break;
816
817 case CBOR_TAG_STRINGREF_NAMESPACE:
818 {
819 ENTER; SAVETMPS;
820
821 SAVESPTR (dec->stringref);
822 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ());
823
824 sv = decode_sv (dec);
825
826 FREETMPS; LEAVE;
827 }
828 break;
829
830 case CBOR_TAG_STRINGREF:
831 {
832 if ((*dec->cur >> 5) != 0)
833 ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
834
835 UV idx = decode_uint (dec);
836
837 if (!dec->stringref || (int)idx > AvFILLp (dec->stringref))
838 ERR ("corrupted CBOR data (stringref index out of bounds or outside namespace)");
839
840 sv = newSVsv (AvARRAY (dec->stringref)[idx]);
841 }
842 break;
843
844 case CBOR_TAG_VALUE_SHAREABLE:
845 {
846 if (ecb_expect_false (!dec->shareable))
847 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ());
848
849 sv = newSV (0);
850 av_push (dec->shareable, SvREFCNT_inc_NN (sv));
851
644 SV *sv = decode_sv (dec); 852 SV *osv = decode_sv (dec);
853 sv_setsv (sv, osv);
854 SvREFCNT_dec_NN (osv);
855 }
856 break;
645 857
646 if (tag == CBOR_TAG_MAGIC) 858 case CBOR_TAG_VALUE_SHAREDREF:
647 return sv; 859 {
860 if ((*dec->cur >> 5) != 0)
861 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
648 862
649 if (tag == CBOR_TAG_PERL_OBJECT) 863 UV idx = decode_uint (dec);
650 { 864
865 if (!dec->shareable || (int)idx > AvFILLp (dec->shareable))
866 ERR ("corrupted CBOR data (sharedref index out of bounds)");
867
868 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]);
869 }
870 break;
871
872 case CBOR_TAG_PERL_OBJECT:
873 {
874 sv = decode_sv (dec);
875
651 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV) 876 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
652 ERR ("corrupted CBOR data (non-array perl object)"); 877 ERR ("corrupted CBOR data (non-array perl object)");
878
879 AV *av = (AV *)SvRV (sv);
880 int len = av_len (av) + 1;
881 HV *stash = gv_stashsv (*av_fetch (av, 0, 1), 0);
882
883 if (!stash)
884 ERR ("cannot decode perl-object (package does not exist)");
885
886 GV *method = gv_fetchmethod_autoload (stash, "THAW", 0);
887
888 if (!method)
889 ERR ("cannot decode perl-object (package does not have a THAW method)");
890
891 dSP;
892
893 ENTER; SAVETMPS; PUSHMARK (SP);
894 EXTEND (SP, len + 1);
895 // we re-bless the reference to get overload and other niceties right
896 PUSHs (*av_fetch (av, 0, 1));
897 PUSHs (sv_cbor);
898
899 int i;
900
901 for (i = 1; i < len; ++i)
902 PUSHs (*av_fetch (av, i, 1));
903
904 PUTBACK;
905 call_sv ((SV *)GvCV (method), G_SCALAR | G_EVAL);
906 SPAGAIN;
907
908 if (SvTRUE (ERRSV))
909 {
910 FREETMPS; LEAVE;
911 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV))));
912 }
913
914 SvREFCNT_dec (sv);
915 sv = SvREFCNT_inc (POPs);
916
917 PUTBACK;
918
919 FREETMPS; LEAVE;
653 920 }
654 // TODO 921 break;
655 }
656 922
923 default:
924 {
925 sv = decode_sv (dec);
926
927 dSP;
928 ENTER; SAVETMPS; PUSHMARK (SP);
929 EXTEND (SP, 2);
930 PUSHs (newSVuv (tag));
931 PUSHs (sv);
932
933 PUTBACK;
934 int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL);
935 SPAGAIN;
936
937 if (SvTRUE (ERRSV))
938 {
939 FREETMPS; LEAVE;
940 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV))));
941 }
942
943 if (count)
944 {
945 SvREFCNT_dec (sv);
946 sv = SvREFCNT_inc (POPs);
947 }
948 else
949 {
657 AV *av = newAV (); 950 AV *av = newAV ();
658 av_push (av, newSVuv (tag)); 951 av_push (av, newSVuv (tag));
659 av_push (av, sv); 952 av_push (av, sv);
660 953
661 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 954 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
662 ? cbor_tagged_stash 955 ? cbor_tagged_stash
663 : gv_stashpv ("CBOR::XS::Tagged" , 1); 956 : gv_stashpv ("CBOR::XS::Tagged" , 1);
664
665 return sv_bless (newRV_noinc ((SV *)av), tagged_stash); 957 sv = sv_bless (newRV_noinc ((SV *)av), tagged_stash);
958 }
959
960 PUTBACK;
961
962 FREETMPS; LEAVE;
963 }
964 break;
965 }
966
967 return sv;
666 968
667fail: 969fail:
668 SvREFCNT_dec (sv); 970 SvREFCNT_dec (sv);
669 return &PL_sv_undef; 971 return &PL_sv_undef;
670} 972}
693 case 7: // misc 995 case 7: // misc
694 switch (*dec->cur++ & 31) 996 switch (*dec->cur++ & 31)
695 { 997 {
696 case 20: 998 case 20:
697#if CBOR_SLOW 999#if CBOR_SLOW
698 cbor_false = get_bool ("CBOR::XS::false"); 1000 types_false = get_bool ("Types::Serialiser::false");
699#endif 1001#endif
700 return newSVsv (cbor_false); 1002 return newSVsv (types_false);
701 case 21: 1003 case 21:
702#if CBOR_SLOW 1004#if CBOR_SLOW
703 cbor_true = get_bool ("CBOR::XS::true"); 1005 types_true = get_bool ("Types::Serialiser::true");
704#endif 1006#endif
705 return newSVsv (cbor_true); 1007 return newSVsv (types_true);
706 case 22: 1008 case 22:
707 return newSVsv (&PL_sv_undef); 1009 return newSVsv (&PL_sv_undef);
1010 case 23:
1011#if CBOR_SLOW
1012 types_error = get_bool ("Types::Serialiser::error");
1013#endif
1014 return newSVsv (types_error);
708 1015
709 case 25: 1016 case 25:
710 { 1017 {
711 WANT (2); 1018 WANT (2);
712 1019
756} 1063}
757 1064
758static SV * 1065static SV *
759decode_cbor (SV *string, CBOR *cbor, char **offset_return) 1066decode_cbor (SV *string, CBOR *cbor, char **offset_return)
760{ 1067{
761 dec_t dec; 1068 dec_t dec = { };
762 SV *sv; 1069 SV *sv;
1070 STRLEN len;
1071 char *data = SvPVbyte (string, len);
763 1072
764 /* work around bugs in 5.10 where manipulating magic values
765 * makes perl ignore the magic in subsequent accesses.
766 * also make a copy of non-PV values, to get them into a clean
767 * state (SvPV should do that, but it's buggy, see below).
768 */
769 /*SvGETMAGIC (string);*/
770 if (SvMAGICAL (string) || !SvPOK (string))
771 string = sv_2mortal (newSVsv (string));
772
773 SvUPGRADE (string, SVt_PV);
774
775 /* work around a bug in perl 5.10, which causes SvCUR to fail an
776 * assertion with -DDEBUGGING, although SvCUR is documented to
777 * return the xpv_cur field which certainly exists after upgrading.
778 * according to nicholas clark, calling SvPOK fixes this.
779 * But it doesn't fix it, so try another workaround, call SvPV_nolen
780 * and hope for the best.
781 * Damnit, SvPV_nolen still trips over yet another assertion. This
782 * assertion business is seriously broken, try yet another workaround
783 * for the broken -DDEBUGGING.
784 */
785 {
786#ifdef DEBUGGING
787 STRLEN offset = SvOK (string) ? sv_len (string) : 0;
788#else
789 STRLEN offset = SvCUR (string);
790#endif
791
792 if (offset > cbor->max_size && cbor->max_size) 1073 if (len > cbor->max_size && cbor->max_size)
793 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu", 1074 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
794 (unsigned long)SvCUR (string), (unsigned long)cbor->max_size); 1075 (unsigned long)len, (unsigned long)cbor->max_size);
795 }
796
797 sv_utf8_downgrade (string, 0);
798 1076
799 dec.cbor = *cbor; 1077 dec.cbor = *cbor;
800 dec.cur = (U8 *)SvPVX (string); 1078 dec.cur = (U8 *)data;
801 dec.end = (U8 *)SvEND (string); 1079 dec.end = (U8 *)data + len;
802 dec.err = 0;
803 dec.depth = 0;
804 1080
805 sv = decode_sv (&dec); 1081 sv = decode_sv (&dec);
806 1082
807 if (offset_return) 1083 if (offset_return)
808 *offset_return = dec.cur; 1084 *offset_return = dec.cur;
812 dec.err = "garbage after CBOR object"; 1088 dec.err = "garbage after CBOR object";
813 1089
814 if (dec.err) 1090 if (dec.err)
815 { 1091 {
816 SvREFCNT_dec (sv); 1092 SvREFCNT_dec (sv);
817 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)SvPVX (string), (int)(uint8_t)*dec.cur); 1093 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur);
818 } 1094 }
819 1095
820 sv = sv_2mortal (sv); 1096 sv = sv_2mortal (sv);
821 1097
822 return sv; 1098 return sv;
828MODULE = CBOR::XS PACKAGE = CBOR::XS 1104MODULE = CBOR::XS PACKAGE = CBOR::XS
829 1105
830BOOT: 1106BOOT:
831{ 1107{
832 cbor_stash = gv_stashpv ("CBOR::XS" , 1); 1108 cbor_stash = gv_stashpv ("CBOR::XS" , 1);
833 cbor_boolean_stash = gv_stashpv ("CBOR::XS::Boolean", 1);
834 cbor_tagged_stash = gv_stashpv ("CBOR::XS::Tagged" , 1); 1109 cbor_tagged_stash = gv_stashpv ("CBOR::XS::Tagged" , 1);
835 1110
836 cbor_true = get_bool ("CBOR::XS::true"); 1111 types_boolean_stash = gv_stashpv ("Types::Serialiser::Boolean", 1);
837 cbor_false = get_bool ("CBOR::XS::false"); 1112 types_error_stash = gv_stashpv ("Types::Serialiser::Error" , 1);
1113
1114 types_true = get_bool ("Types::Serialiser::true" );
1115 types_false = get_bool ("Types::Serialiser::false");
1116 types_error = get_bool ("Types::Serialiser::error");
1117
1118 default_filter = newSVpv ("CBOR::XS::default_filter", 0);
1119
1120 sv_cbor = newSVpv ("CBOR", 0);
1121 SvREADONLY_on (sv_cbor);
838} 1122}
839 1123
840PROTOTYPES: DISABLE 1124PROTOTYPES: DISABLE
841 1125
842void CLONE (...) 1126void CLONE (...)
843 CODE: 1127 CODE:
844 cbor_stash = 0; 1128 cbor_stash = 0;
845 cbor_boolean_stash = 0;
846 cbor_tagged_stash = 0; 1129 cbor_tagged_stash = 0;
1130 types_error_stash = 0;
1131 types_boolean_stash = 0;
847 1132
848void new (char *klass) 1133void new (char *klass)
849 PPCODE: 1134 PPCODE:
850{ 1135{
851 SV *pv = NEWSV (0, sizeof (CBOR)); 1136 SV *pv = NEWSV (0, sizeof (CBOR));
859 1144
860void shrink (CBOR *self, int enable = 1) 1145void shrink (CBOR *self, int enable = 1)
861 ALIAS: 1146 ALIAS:
862 shrink = F_SHRINK 1147 shrink = F_SHRINK
863 allow_unknown = F_ALLOW_UNKNOWN 1148 allow_unknown = F_ALLOW_UNKNOWN
1149 allow_sharing = F_ALLOW_SHARING
1150 pack_strings = F_PACK_STRINGS
864 PPCODE: 1151 PPCODE:
865{ 1152{
866 if (enable) 1153 if (enable)
867 self->flags |= ix; 1154 self->flags |= ix;
868 else 1155 else
873 1160
874void get_shrink (CBOR *self) 1161void get_shrink (CBOR *self)
875 ALIAS: 1162 ALIAS:
876 get_shrink = F_SHRINK 1163 get_shrink = F_SHRINK
877 get_allow_unknown = F_ALLOW_UNKNOWN 1164 get_allow_unknown = F_ALLOW_UNKNOWN
1165 get_allow_sharing = F_ALLOW_SHARING
1166 get_pack_strings = F_PACK_STRINGS
878 PPCODE: 1167 PPCODE:
879 XPUSHs (boolSV (self->flags & ix)); 1168 XPUSHs (boolSV (self->flags & ix));
880 1169
881void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) 1170void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
882 PPCODE: 1171 PPCODE:
898 CODE: 1187 CODE:
899 RETVAL = self->max_size; 1188 RETVAL = self->max_size;
900 OUTPUT: 1189 OUTPUT:
901 RETVAL 1190 RETVAL
902 1191
903#if 0 //TODO 1192void filter (CBOR *self, SV *filter = 0)
904
905void filter_cbor_object (CBOR *self, SV *cb = &PL_sv_undef)
906 PPCODE: 1193 PPCODE:
907{
908 SvREFCNT_dec (self->cb_object); 1194 SvREFCNT_dec (self->filter);
909 self->cb_object = SvOK (cb) ? newSVsv (cb) : 0; 1195 self->filter = filter ? newSVsv (filter) : filter;
910
911 XPUSHs (ST (0)); 1196 XPUSHs (ST (0));
912}
913 1197
914void filter_cbor_single_key_object (CBOR *self, SV *key, SV *cb = &PL_sv_undef) 1198SV *get_filter (CBOR *self)
915 PPCODE: 1199 CODE:
916{ 1200 RETVAL = self->filter ? self->filter : NEWSV (0, 0);
917 if (!self->cb_sk_object) 1201 OUTPUT:
918 self->cb_sk_object = newHV (); 1202 RETVAL
919
920 if (SvOK (cb))
921 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);
922 else
923 {
924 hv_delete_ent (self->cb_sk_object, key, G_DISCARD, 0);
925
926 if (!HvKEYS (self->cb_sk_object))
927 {
928 SvREFCNT_dec (self->cb_sk_object);
929 self->cb_sk_object = 0;
930 }
931 }
932
933 XPUSHs (ST (0));
934}
935
936#endif
937 1203
938void encode (CBOR *self, SV *scalar) 1204void encode (CBOR *self, SV *scalar)
939 PPCODE: 1205 PPCODE:
940 PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN; 1206 PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN;
941 XPUSHs (scalar); 1207 XPUSHs (scalar);
954 EXTEND (SP, 2); 1220 EXTEND (SP, 2);
955 PUSHs (sv); 1221 PUSHs (sv);
956 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); 1222 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
957} 1223}
958 1224
959#if 0
960
961void DESTROY (CBOR *self) 1225void DESTROY (CBOR *self)
962 CODE: 1226 PPCODE:
963 SvREFCNT_dec (self->cb_sk_object); 1227 cbor_free (self);
964 SvREFCNT_dec (self->cb_object);
965
966#endif
967 1228
968PROTOTYPES: ENABLE 1229PROTOTYPES: ENABLE
969 1230
970void encode_cbor (SV *scalar) 1231void encode_cbor (SV *scalar)
971 PPCODE: 1232 PPCODE:

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines