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 1.11 by root, Sun Oct 27 22:35:15 2013 UTC vs.
Revision 1.34 by root, Sat Nov 30 16:19:59 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// 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)
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, *types_boolean_stash, *types_error_stash, *cbor_tagged_stash; // CBOR::XS:: 86static HV *cbor_stash, *types_boolean_stash, *types_error_stash, *cbor_tagged_stash; // CBOR::XS::
66static SV *types_true, *types_false, *types_error, *sv_cbor; 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
260 svt = SvTYPE (sv); 332 svtype svt = SvTYPE (sv);
261 333
262 if (ecb_expect_false (SvOBJECT (sv))) 334 if (ecb_expect_false (SvOBJECT (sv)))
263 { 335 {
264 HV *boolean_stash = !CBOR_SLOW || types_boolean_stash 336 HV *boolean_stash = !CBOR_SLOW || types_boolean_stash
265 ? types_boolean_stash 337 ? types_boolean_stash
270 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 342 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
271 ? cbor_tagged_stash 343 ? cbor_tagged_stash
272 : gv_stashpv ("CBOR::XS::Tagged" , 1); 344 : gv_stashpv ("CBOR::XS::Tagged" , 1);
273 345
274 HV *stash = SvSTASH (sv); 346 HV *stash = SvSTASH (sv);
275 GV *method;
276 347
277 if (stash == boolean_stash) 348 if (stash == boolean_stash)
349 {
278 encode_ch (enc, SvIV (sv) ? 0xe0 | 21 : 0xe0 | 20); 350 encode_ch (enc, SvIV (sv) ? 0xe0 | 21 : 0xe0 | 20);
351 return;
352 }
279 else if (stash == error_stash) 353 else if (stash == error_stash)
354 {
280 encode_ch (enc, 0xe0 | 23); 355 encode_ch (enc, 0xe0 | 23);
356 return;
357 }
281 else if (stash == tagged_stash) 358 else if (stash == tagged_stash)
282 { 359 {
283 if (svt != SVt_PVAV) 360 if (svt != SVt_PVAV)
284 croak ("encountered CBOR::XS::Tagged object that isn't an array"); 361 croak ("encountered CBOR::XS::Tagged object that isn't an array");
285 362
286 encode_uint (enc, 0xc0, SvUV (*av_fetch ((AV *)sv, 0, 1))); 363 encode_uint (enc, 0xc0, SvUV (*av_fetch ((AV *)sv, 0, 1)));
287 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1)); 364 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1));
365
366 return;
367 }
368 }
369
370 if (ecb_expect_false (SvREFCNT (sv) > 1)
371 && ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING))
372 {
373 if (!enc->shareable)
374 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
375
376 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
377
378 if (SvOK (*svp))
288 } 379 {
380 encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF);
381 encode_uint (enc, 0x00, SvUV (*svp));
382 return;
383 }
384 else
385 {
386 sv_setuv (*svp, enc->shareable_idx);
387 ++enc->shareable_idx;
388 encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE);
389 }
390 }
391
392 if (ecb_expect_false (SvOBJECT (sv)))
393 {
394 HV *stash = SvSTASH (sv);
395 GV *method;
396
289 else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0))) 397 if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
290 { 398 {
291 dSP; 399 dSP;
292 400
293 ENTER; SAVETMPS; PUSHMARK (SP); 401 ENTER; SAVETMPS; PUSHMARK (SP);
294 // we re-bless the reference to get overload and other niceties right 402 // we re-bless the reference to get overload and other niceties right
325 433
326 // catch this surprisingly common error 434 // catch this surprisingly common error
327 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv) 435 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)
328 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash)); 436 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));
329 437
330 encode_uint (enc, 0xc0, CBOR_TAG_PERL_OBJECT); 438 encode_tag (enc, CBOR_TAG_PERL_OBJECT);
331 encode_uint (enc, 0x80, count + 1); 439 encode_uint (enc, 0x80, count + 1);
332 encode_str (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash)); 440 encode_strref (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
333 441
334 while (count) 442 while (count)
335 encode_sv (enc, SP[1 - count--]); 443 encode_sv (enc, SP[1 - count--]);
336 444
337 PUTBACK; 445 PUTBACK;
344 } 452 }
345 else if (svt == SVt_PVHV) 453 else if (svt == SVt_PVHV)
346 encode_hv (enc, (HV *)sv); 454 encode_hv (enc, (HV *)sv);
347 else if (svt == SVt_PVAV) 455 else if (svt == SVt_PVAV)
348 encode_av (enc, (AV *)sv); 456 encode_av (enc, (AV *)sv);
349 else if (svt < SVt_PVAV)
350 {
351 STRLEN len = 0;
352 char *pv = svt ? SvPV (sv, len) : 0;
353
354 if (len == 1 && *pv == '1')
355 encode_ch (enc, 0xe0 | 21);
356 else if (len == 1 && *pv == '0')
357 encode_ch (enc, 0xe0 | 20);
358 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
359 encode_ch (enc, 0xe0 | 23);
360 else
361 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
362 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
363 }
364 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
365 encode_ch (enc, 0xe0 | 23);
366 else 457 else
367 croak ("encountered %s, but CBOR can only represent references to arrays or hashes", 458 {
368 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 459 encode_tag (enc, CBOR_TAG_INDIRECTION);
460 encode_sv (enc, sv);
461 }
369} 462}
370 463
371static void 464static void
372encode_nv (enc_t *enc, SV *sv) 465encode_nv (enc_t *enc, SV *sv)
373{ 466{
411 504
412 if (SvPOKp (sv)) 505 if (SvPOKp (sv))
413 { 506 {
414 STRLEN len; 507 STRLEN len;
415 char *str = SvPV (sv, len); 508 char *str = SvPV (sv, len);
416 encode_str (enc, SvUTF8 (sv), str, len); 509 encode_strref (enc, SvUTF8 (sv), str, len);
417 } 510 }
418 else if (SvNOKp (sv)) 511 else if (SvNOKp (sv))
419 encode_nv (enc, sv); 512 encode_nv (enc, sv);
420 else if (SvIOKp (sv)) 513 else if (SvIOKp (sv))
421 { 514 {
438} 531}
439 532
440static SV * 533static SV *
441encode_cbor (SV *scalar, CBOR *cbor) 534encode_cbor (SV *scalar, CBOR *cbor)
442{ 535{
443 enc_t enc; 536 enc_t enc = { };
444 537
445 enc.cbor = *cbor; 538 enc.cbor = *cbor;
446 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 539 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
447 enc.cur = SvPVX (enc.sv); 540 enc.cur = SvPVX (enc.sv);
448 enc.end = SvEND (enc.sv); 541 enc.end = SvEND (enc.sv);
449 enc.depth = 0;
450 542
451 SvPOK_only (enc.sv); 543 SvPOK_only (enc.sv);
544
545 if (cbor->flags & F_PACK_STRINGS)
546 {
547 encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE);
548 enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ());
549 enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ());
550 }
551
452 encode_sv (&enc, scalar); 552 encode_sv (&enc, scalar);
453 553
454 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 554 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
455 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings 555 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
456 556
470 U8 *end; // end of input string 570 U8 *end; // end of input string
471 const char *err; // parse error, if != 0 571 const char *err; // parse error, if != 0
472 CBOR cbor; 572 CBOR cbor;
473 U32 depth; // recursion depth 573 U32 depth; // recursion depth
474 U32 maxdepth; // recursion depth limit 574 U32 maxdepth; // recursion depth limit
575 AV *shareable;
576 AV *stringref;
577 SV *decode_tagged;
475} dec_t; 578} dec_t;
476 579
477#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE 580#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE
478 581
479#define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data") 582#define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data")
511 | ((UV)dec->cur[-1]); 614 | ((UV)dec->cur[-1]);
512 615
513 case 27: 616 case 27:
514 WANT (9); 617 WANT (9);
515 dec->cur += 9; 618 dec->cur += 9;
619
620 return
621#if UVSIZE < 8
622 0
623#else
516 return (((UV)dec->cur[-8]) << 56) 624 (((UV)dec->cur[-8]) << 56)
517 | (((UV)dec->cur[-7]) << 48) 625 | (((UV)dec->cur[-7]) << 48)
518 | (((UV)dec->cur[-6]) << 40) 626 | (((UV)dec->cur[-6]) << 40)
519 | (((UV)dec->cur[-5]) << 32) 627 | (((UV)dec->cur[-5]) << 32)
628#endif
520 | (((UV)dec->cur[-4]) << 24) 629 | (((UV)dec->cur[-4]) << 24)
521 | (((UV)dec->cur[-3]) << 16) 630 | (((UV)dec->cur[-3]) << 16)
522 | (((UV)dec->cur[-2]) << 8) 631 | (((UV)dec->cur[-2]) << 8)
523 | ((UV)dec->cur[-1]); 632 | ((UV)dec->cur[-1]);
524 633
573 SvREFCNT_dec (av); 682 SvREFCNT_dec (av);
574 DEC_DEC_DEPTH; 683 DEC_DEC_DEPTH;
575 return &PL_sv_undef; 684 return &PL_sv_undef;
576} 685}
577 686
687static void
688decode_he (dec_t *dec, HV *hv)
689{
690 // for speed reasons, we specialcase single-string
691 // byte or utf-8 strings as keys, but only when !stringref
692
693 if (ecb_expect_true (!dec->stringref))
694 if (*dec->cur >= 0x40 && *dec->cur <= 0x40 + 27)
695 {
696 I32 len = decode_uint (dec);
697 char *key = (char *)dec->cur;
698
699 dec->cur += len;
700
701 if (ecb_expect_false (dec->stringref))
702 av_push (dec->stringref, newSVpvn (key, len));
703
704 hv_store (hv, key, len, decode_sv (dec), 0);
705
706 return;
707 }
708 else if (*dec->cur >= 0x60 && *dec->cur <= 0x60 + 27)
709 {
710 I32 len = decode_uint (dec);
711 char *key = (char *)dec->cur;
712
713 dec->cur += len;
714
715 if (ecb_expect_false (dec->stringref))
716 av_push (dec->stringref, newSVpvn_utf8 (key, len, 1));
717
718 hv_store (hv, key, -len, decode_sv (dec), 0);
719
720 return;
721 }
722
723 SV *k = decode_sv (dec);
724 SV *v = decode_sv (dec);
725
726 hv_store_ent (hv, k, v, 0);
727 SvREFCNT_dec (k);
728}
729
578static SV * 730static SV *
579decode_hv (dec_t *dec) 731decode_hv (dec_t *dec)
580{ 732{
581 HV *hv = newHV (); 733 HV *hv = newHV ();
582 734
594 { 746 {
595 ++dec->cur; 747 ++dec->cur;
596 break; 748 break;
597 } 749 }
598 750
599 SV *k = decode_sv (dec); 751 decode_he (dec, hv);
600 SV *v = decode_sv (dec);
601
602 hv_store_ent (hv, k, v, 0);
603 } 752 }
604 } 753 }
605 else 754 else
606 { 755 {
607 int len = decode_uint (dec); 756 int pairs = decode_uint (dec);
608 757
609 while (len--) 758 while (pairs--)
610 { 759 decode_he (dec, hv);
611 SV *k = decode_sv (dec);
612 SV *v = decode_sv (dec);
613
614 hv_store_ent (hv, k, v, 0);
615 }
616 } 760 }
617 761
618 DEC_DEC_DEPTH; 762 DEC_DEC_DEPTH;
619 return newRV_noinc ((SV *)hv); 763 return newRV_noinc ((SV *)hv);
620 764
629{ 773{
630 SV *sv = 0; 774 SV *sv = 0;
631 775
632 if ((*dec->cur & 31) == 31) 776 if ((*dec->cur & 31) == 31)
633 { 777 {
778 // indefinite length strings
634 ++dec->cur; 779 ++dec->cur;
635 780
781 unsigned char major = *dec->cur & 0xe0;
782
636 sv = newSVpvn ("", 0); 783 sv = newSVpvn ("", 0);
637 784
638 // not very fast, and certainly not robust against illegal input
639 for (;;) 785 for (;;)
640 { 786 {
641 WANT (1); 787 WANT (1);
642 788
789 if ((*dec->cur ^ major) >= 31)
643 if (*dec->cur == (0xe0 | 31)) 790 if (*dec->cur == (0xe0 | 31))
644 { 791 {
645 ++dec->cur; 792 ++dec->cur;
646 break; 793 break;
647 } 794 }
795 else
796 ERR ("corrupted CBOR data (invalid chunks in indefinite length string)");
648 797
649 sv_catsv (sv, decode_sv (dec)); 798 STRLEN len = decode_uint (dec);
799
800 WANT (len);
801 sv_catpvn (sv, dec->cur, len);
802 dec->cur += len;
650 } 803 }
651 } 804 }
652 else 805 else
653 { 806 {
654 STRLEN len = decode_uint (dec); 807 STRLEN len = decode_uint (dec);
655 808
656 WANT (len); 809 WANT (len);
657 sv = newSVpvn (dec->cur, len); 810 sv = newSVpvn (dec->cur, len);
658 dec->cur += len; 811 dec->cur += len;
812
813 if (ecb_expect_false (dec->stringref)
814 && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1))
815 av_push (dec->stringref, SvREFCNT_inc_NN (sv));
659 } 816 }
660 817
661 if (utf8) 818 if (utf8)
662 SvUTF8_on (sv); 819 SvUTF8_on (sv);
663 820
669} 826}
670 827
671static SV * 828static SV *
672decode_tagged (dec_t *dec) 829decode_tagged (dec_t *dec)
673{ 830{
831 SV *sv = 0;
674 UV tag = decode_uint (dec); 832 UV tag = decode_uint (dec);
833
834 WANT (1);
835
836 switch (tag)
837 {
838 case CBOR_TAG_MAGIC:
839 sv = decode_sv (dec);
840 break;
841
842 case CBOR_TAG_INDIRECTION:
843 sv = newRV_noinc (decode_sv (dec));
844 break;
845
846 case CBOR_TAG_STRINGREF_NAMESPACE:
847 {
848 ENTER; SAVETMPS;
849
850 SAVESPTR (dec->stringref);
851 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ());
852
853 sv = decode_sv (dec);
854
855 FREETMPS; LEAVE;
856 }
857 break;
858
859 case CBOR_TAG_STRINGREF:
860 {
861 if ((*dec->cur >> 5) != 0)
862 ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
863
864 UV idx = decode_uint (dec);
865
866 if (!dec->stringref || (int)idx > AvFILLp (dec->stringref))
867 ERR ("corrupted CBOR data (stringref index out of bounds or outside namespace)");
868
869 sv = newSVsv (AvARRAY (dec->stringref)[idx]);
870 }
871 break;
872
873 case CBOR_TAG_VALUE_SHAREABLE:
874 {
875 if (ecb_expect_false (!dec->shareable))
876 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ());
877
878 sv = newSV (0);
879 av_push (dec->shareable, SvREFCNT_inc_NN (sv));
880
675 SV *sv = decode_sv (dec); 881 SV *osv = decode_sv (dec);
882 sv_setsv (sv, osv);
883 SvREFCNT_dec_NN (osv);
884 }
885 break;
676 886
677 if (tag == CBOR_TAG_MAGIC) 887 case CBOR_TAG_VALUE_SHAREDREF:
678 return sv; 888 {
679 else if (tag == CBOR_TAG_PERL_OBJECT) 889 if ((*dec->cur >> 5) != 0)
680 { 890 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
891
892 UV idx = decode_uint (dec);
893
894 if (!dec->shareable || (int)idx > AvFILLp (dec->shareable))
895 ERR ("corrupted CBOR data (sharedref index out of bounds)");
896
897 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]);
898 }
899 break;
900
901 case CBOR_TAG_PERL_OBJECT:
902 {
903 sv = decode_sv (dec);
904
681 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV) 905 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
682 ERR ("corrupted CBOR data (non-array perl object)"); 906 ERR ("corrupted CBOR data (non-array perl object)");
683 907
684 AV *av = (AV *)SvRV (sv); 908 AV *av = (AV *)SvRV (sv);
685 int len = av_len (av) + 1; 909 int len = av_len (av) + 1;
686 HV *stash = gv_stashsv (*av_fetch (av, 0, 1), 0); 910 HV *stash = gv_stashsv (*av_fetch (av, 0, 1), 0);
687 911
688 if (!stash) 912 if (!stash)
689 ERR ("cannot decode perl-object (package does not exist)"); 913 ERR ("cannot decode perl-object (package does not exist)");
690 914
691 GV *method = gv_fetchmethod_autoload (stash, "THAW", 0); 915 GV *method = gv_fetchmethod_autoload (stash, "THAW", 0);
692 916
693 if (!method) 917 if (!method)
694 ERR ("cannot decode perl-object (package does not have a THAW method)"); 918 ERR ("cannot decode perl-object (package does not have a THAW method)");
695 919
696 dSP; 920 dSP;
697 921
698 ENTER; SAVETMPS; PUSHMARK (SP); 922 ENTER; SAVETMPS; PUSHMARK (SP);
699 EXTEND (SP, len + 1); 923 EXTEND (SP, len + 1);
700 // we re-bless the reference to get overload and other niceties right 924 // we re-bless the reference to get overload and other niceties right
701 PUSHs (*av_fetch (av, 0, 1)); 925 PUSHs (*av_fetch (av, 0, 1));
702 PUSHs (sv_cbor); 926 PUSHs (sv_cbor);
703 927
704 int i; 928 int i;
705 929
706 for (i = 1; i < len; ++i) 930 for (i = 1; i < len; ++i)
707 PUSHs (*av_fetch (av, i, 1)); 931 PUSHs (*av_fetch (av, i, 1));
708 932
709 PUTBACK; 933 PUTBACK;
710 call_sv ((SV *)GvCV (method), G_SCALAR); 934 call_sv ((SV *)GvCV (method), G_SCALAR | G_EVAL);
711 SPAGAIN; 935 SPAGAIN;
712 936
937 if (SvTRUE (ERRSV))
938 {
939 FREETMPS; LEAVE;
940 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV))));
941 }
942
943 SvREFCNT_dec (sv);
713 sv = SvREFCNT_inc (POPs); 944 sv = SvREFCNT_inc (POPs);
714 945
715 PUTBACK; 946 PUTBACK;
716 947
717 FREETMPS; LEAVE; 948 FREETMPS; LEAVE;
949 }
950 break;
718 951
719 return sv; 952 default:
720 } 953 {
721 else 954 sv = decode_sv (dec);
722 { 955
956 dSP;
957 ENTER; SAVETMPS; PUSHMARK (SP);
958 EXTEND (SP, 2);
959 PUSHs (newSVuv (tag));
960 PUSHs (sv);
961
962 PUTBACK;
963 int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL);
964 SPAGAIN;
965
966 if (SvTRUE (ERRSV))
967 {
968 FREETMPS; LEAVE;
969 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV))));
970 }
971
972 if (count)
973 {
974 SvREFCNT_dec (sv);
975 sv = SvREFCNT_inc (POPs);
976 }
977 else
978 {
723 AV *av = newAV (); 979 AV *av = newAV ();
724 av_push (av, newSVuv (tag)); 980 av_push (av, newSVuv (tag));
725 av_push (av, sv); 981 av_push (av, sv);
726 982
727 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 983 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
728 ? cbor_tagged_stash 984 ? cbor_tagged_stash
729 : gv_stashpv ("CBOR::XS::Tagged" , 1); 985 : gv_stashpv ("CBOR::XS::Tagged" , 1);
730
731 return sv_bless (newRV_noinc ((SV *)av), tagged_stash); 986 sv = sv_bless (newRV_noinc ((SV *)av), tagged_stash);
987 }
988
989 PUTBACK;
990
991 FREETMPS; LEAVE;
992 }
993 break;
732 } 994 }
995
996 return sv;
733 997
734fail: 998fail:
735 SvREFCNT_dec (sv); 999 SvREFCNT_dec (sv);
736 return &PL_sv_undef; 1000 return &PL_sv_undef;
737} 1001}
828} 1092}
829 1093
830static SV * 1094static SV *
831decode_cbor (SV *string, CBOR *cbor, char **offset_return) 1095decode_cbor (SV *string, CBOR *cbor, char **offset_return)
832{ 1096{
833 dec_t dec; 1097 dec_t dec = { };
834 SV *sv; 1098 SV *sv;
1099 STRLEN len;
1100 char *data = SvPVbyte (string, len);
835 1101
836 /* work around bugs in 5.10 where manipulating magic values
837 * makes perl ignore the magic in subsequent accesses.
838 * also make a copy of non-PV values, to get them into a clean
839 * state (SvPV should do that, but it's buggy, see below).
840 */
841 /*SvGETMAGIC (string);*/
842 if (SvMAGICAL (string) || !SvPOK (string))
843 string = sv_2mortal (newSVsv (string));
844
845 SvUPGRADE (string, SVt_PV);
846
847 /* work around a bug in perl 5.10, which causes SvCUR to fail an
848 * assertion with -DDEBUGGING, although SvCUR is documented to
849 * return the xpv_cur field which certainly exists after upgrading.
850 * according to nicholas clark, calling SvPOK fixes this.
851 * But it doesn't fix it, so try another workaround, call SvPV_nolen
852 * and hope for the best.
853 * Damnit, SvPV_nolen still trips over yet another assertion. This
854 * assertion business is seriously broken, try yet another workaround
855 * for the broken -DDEBUGGING.
856 */
857 {
858#ifdef DEBUGGING
859 STRLEN offset = SvOK (string) ? sv_len (string) : 0;
860#else
861 STRLEN offset = SvCUR (string);
862#endif
863
864 if (offset > cbor->max_size && cbor->max_size) 1102 if (len > cbor->max_size && cbor->max_size)
865 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu", 1103 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
866 (unsigned long)SvCUR (string), (unsigned long)cbor->max_size); 1104 (unsigned long)len, (unsigned long)cbor->max_size);
867 }
868
869 sv_utf8_downgrade (string, 0);
870 1105
871 dec.cbor = *cbor; 1106 dec.cbor = *cbor;
872 dec.cur = (U8 *)SvPVX (string); 1107 dec.cur = (U8 *)data;
873 dec.end = (U8 *)SvEND (string); 1108 dec.end = (U8 *)data + len;
874 dec.err = 0;
875 dec.depth = 0;
876 1109
877 sv = decode_sv (&dec); 1110 sv = decode_sv (&dec);
878 1111
879 if (offset_return) 1112 if (offset_return)
880 *offset_return = dec.cur; 1113 *offset_return = dec.cur;
884 dec.err = "garbage after CBOR object"; 1117 dec.err = "garbage after CBOR object";
885 1118
886 if (dec.err) 1119 if (dec.err)
887 { 1120 {
888 SvREFCNT_dec (sv); 1121 SvREFCNT_dec (sv);
889 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)SvPVX (string), (int)(uint8_t)*dec.cur); 1122 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur);
890 } 1123 }
891 1124
892 sv = sv_2mortal (sv); 1125 sv = sv_2mortal (sv);
893 1126
894 return sv; 1127 return sv;
908 types_error_stash = gv_stashpv ("Types::Serialiser::Error" , 1); 1141 types_error_stash = gv_stashpv ("Types::Serialiser::Error" , 1);
909 1142
910 types_true = get_bool ("Types::Serialiser::true" ); 1143 types_true = get_bool ("Types::Serialiser::true" );
911 types_false = get_bool ("Types::Serialiser::false"); 1144 types_false = get_bool ("Types::Serialiser::false");
912 types_error = get_bool ("Types::Serialiser::error"); 1145 types_error = get_bool ("Types::Serialiser::error");
1146
1147 default_filter = newSVpv ("CBOR::XS::default_filter", 0);
913 1148
914 sv_cbor = newSVpv ("CBOR", 0); 1149 sv_cbor = newSVpv ("CBOR", 0);
915 SvREADONLY_on (sv_cbor); 1150 SvREADONLY_on (sv_cbor);
916} 1151}
917 1152
938 1173
939void shrink (CBOR *self, int enable = 1) 1174void shrink (CBOR *self, int enable = 1)
940 ALIAS: 1175 ALIAS:
941 shrink = F_SHRINK 1176 shrink = F_SHRINK
942 allow_unknown = F_ALLOW_UNKNOWN 1177 allow_unknown = F_ALLOW_UNKNOWN
1178 allow_sharing = F_ALLOW_SHARING
1179 pack_strings = F_PACK_STRINGS
943 PPCODE: 1180 PPCODE:
944{ 1181{
945 if (enable) 1182 if (enable)
946 self->flags |= ix; 1183 self->flags |= ix;
947 else 1184 else
952 1189
953void get_shrink (CBOR *self) 1190void get_shrink (CBOR *self)
954 ALIAS: 1191 ALIAS:
955 get_shrink = F_SHRINK 1192 get_shrink = F_SHRINK
956 get_allow_unknown = F_ALLOW_UNKNOWN 1193 get_allow_unknown = F_ALLOW_UNKNOWN
1194 get_allow_sharing = F_ALLOW_SHARING
1195 get_pack_strings = F_PACK_STRINGS
957 PPCODE: 1196 PPCODE:
958 XPUSHs (boolSV (self->flags & ix)); 1197 XPUSHs (boolSV (self->flags & ix));
959 1198
960void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) 1199void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
961 PPCODE: 1200 PPCODE:
977 CODE: 1216 CODE:
978 RETVAL = self->max_size; 1217 RETVAL = self->max_size;
979 OUTPUT: 1218 OUTPUT:
980 RETVAL 1219 RETVAL
981 1220
982#if 0 //TODO 1221void filter (CBOR *self, SV *filter = 0)
983
984void filter_cbor_object (CBOR *self, SV *cb = &PL_sv_undef)
985 PPCODE: 1222 PPCODE:
986{
987 SvREFCNT_dec (self->cb_object); 1223 SvREFCNT_dec (self->filter);
988 self->cb_object = SvOK (cb) ? newSVsv (cb) : 0; 1224 self->filter = filter ? newSVsv (filter) : filter;
989
990 XPUSHs (ST (0)); 1225 XPUSHs (ST (0));
991}
992 1226
993void filter_cbor_single_key_object (CBOR *self, SV *key, SV *cb = &PL_sv_undef) 1227SV *get_filter (CBOR *self)
994 PPCODE: 1228 CODE:
995{ 1229 RETVAL = self->filter ? self->filter : NEWSV (0, 0);
996 if (!self->cb_sk_object) 1230 OUTPUT:
997 self->cb_sk_object = newHV (); 1231 RETVAL
998
999 if (SvOK (cb))
1000 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);
1001 else
1002 {
1003 hv_delete_ent (self->cb_sk_object, key, G_DISCARD, 0);
1004
1005 if (!HvKEYS (self->cb_sk_object))
1006 {
1007 SvREFCNT_dec (self->cb_sk_object);
1008 self->cb_sk_object = 0;
1009 }
1010 }
1011
1012 XPUSHs (ST (0));
1013}
1014
1015#endif
1016 1232
1017void encode (CBOR *self, SV *scalar) 1233void encode (CBOR *self, SV *scalar)
1018 PPCODE: 1234 PPCODE:
1019 PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN; 1235 PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN;
1020 XPUSHs (scalar); 1236 XPUSHs (scalar);
1033 EXTEND (SP, 2); 1249 EXTEND (SP, 2);
1034 PUSHs (sv); 1250 PUSHs (sv);
1035 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); 1251 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
1036} 1252}
1037 1253
1038#if 0
1039
1040void DESTROY (CBOR *self) 1254void DESTROY (CBOR *self)
1041 CODE: 1255 PPCODE:
1042 SvREFCNT_dec (self->cb_sk_object); 1256 cbor_free (self);
1043 SvREFCNT_dec (self->cb_object);
1044
1045#endif
1046 1257
1047PROTOTYPES: ENABLE 1258PROTOTYPES: ENABLE
1048 1259
1049void encode_cbor (SV *scalar) 1260void encode_cbor (SV *scalar)
1050 PPCODE: 1261 PPCODE:

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines