ViewVC Help
View File | Revision Log | Show Annotations | Download File
/cvs/cvsroot/CBOR-XS/XS.xs
(Generate patch)

Comparing cvsroot/CBOR-XS/XS.xs (file contents):
Revision 1.9 by root, Sun Oct 27 10:17:12 2013 UTC vs.
Revision 1.33 by root, Sat Nov 30 15:23:59 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)
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
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 || cbor_boolean_stash 336 HV *boolean_stash = !CBOR_SLOW || types_boolean_stash
265 ? cbor_boolean_stash 337 ? types_boolean_stash
266 : gv_stashpv ("CBOR::XS::Boolean", 1); 338 : gv_stashpv ("Types::Serialiser::Boolean", 1);
339 HV *error_stash = !CBOR_SLOW || types_error_stash
340 ? types_error_stash
341 : gv_stashpv ("Types::Serialiser::Error", 1);
267 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 342 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
268 ? cbor_tagged_stash 343 ? cbor_tagged_stash
269 : gv_stashpv ("CBOR::XS::Tagged" , 1); 344 : gv_stashpv ("CBOR::XS::Tagged" , 1);
270 345
346 HV *stash = SvSTASH (sv);
347 GV *method;
348
271 if (SvSTASH (sv) == boolean_stash) 349 if (stash == boolean_stash)
272 encode_ch (enc, SvIV (sv) ? 0xe0 | 21 : 0xe0 | 20); 350 encode_ch (enc, SvIV (sv) ? 0xe0 | 21 : 0xe0 | 20);
351 else if (stash == error_stash)
352 encode_ch (enc, 0xe0 | 23);
273 else if (SvSTASH (sv) == tagged_stash) 353 else if (stash == tagged_stash)
274 { 354 {
275 if (svt != SVt_PVAV) 355 if (svt != SVt_PVAV)
276 croak ("encountered CBOR::XS::Tagged object that isn't an array"); 356 croak ("encountered CBOR::XS::Tagged object that isn't an array");
277 357
278 encode_uint (enc, 0xc0, SvUV (*av_fetch ((AV *)sv, 0, 1))); 358 encode_uint (enc, 0xc0, SvUV (*av_fetch ((AV *)sv, 0, 1)));
279 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1)); 359 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1));
280 } 360 }
361 else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
362 {
363 dSP;
364
365 ENTER; SAVETMPS; PUSHMARK (SP);
366 // we re-bless the reference to get overload and other niceties right
367 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
368
369 PUTBACK;
370 // G_SCALAR ensures that return value is 1
371 call_sv ((SV *)GvCV (method), G_SCALAR);
372 SPAGAIN;
373
374 // catch this surprisingly common error
375 if (SvROK (TOPs) && SvRV (TOPs) == sv)
376 croak ("%s::TO_CBOR method returned same object as was passed instead of a new one", HvNAME (stash));
377
378 encode_sv (enc, POPs);
379
380 PUTBACK;
381
382 FREETMPS; LEAVE;
383 }
384 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)
385 {
386 dSP;
387
388 ENTER; SAVETMPS; PUSHMARK (SP);
389 EXTEND (SP, 2);
390 // we re-bless the reference to get overload and other niceties right
391 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
392 PUSHs (sv_cbor);
393
394 PUTBACK;
395 int count = call_sv ((SV *)GvCV (method), G_ARRAY);
396 SPAGAIN;
397
398 // catch this surprisingly common error
399 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)
400 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));
401
402 encode_tag (enc, CBOR_TAG_PERL_OBJECT);
403 encode_uint (enc, 0x80, count + 1);
404 encode_strref (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
405
406 while (count)
407 encode_sv (enc, SP[1 - count--]);
408
409 PUTBACK;
410
411 FREETMPS; LEAVE;
412 }
281 else 413 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", 414 croak ("encountered object '%s', but no TO_CBOR or FREEZE methods available on it",
311 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 415 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
312 }
313 } 416 }
314 else if (svt == SVt_PVHV) 417 else if (svt == SVt_PVHV)
315 encode_hv (enc, (HV *)sv); 418 encode_hv (enc, (HV *)sv);
316 else if (svt == SVt_PVAV) 419 else if (svt == SVt_PVAV)
317 encode_av (enc, (AV *)sv); 420 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 421 else
336 croak ("encountered %s, but CBOR can only represent references to arrays or hashes", 422 {
337 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 423 if (ecb_expect_false (SvREFCNT (sv) > 1)
424 && ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING))
425 {
426 if (!enc->shareable)
427 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
428
429 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
430
431 if (SvOK (*svp))
432 {
433 encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF);
434 encode_uint (enc, 0x00, SvUV (*svp));
435 return;
436 }
437 else
438 {
439 sv_setuv (*svp, enc->shareable_idx);
440 ++enc->shareable_idx;
441 encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE);
442 }
443 }
444
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
598{ 753{
599 SV *sv = 0; 754 SV *sv = 0;
600 755
601 if ((*dec->cur & 31) == 31) 756 if ((*dec->cur & 31) == 31)
602 { 757 {
758 // indefinite length strings
603 ++dec->cur; 759 ++dec->cur;
604 760
761 unsigned char major = *dec->cur & 0xe0;
762
605 sv = newSVpvn ("", 0); 763 sv = newSVpvn ("", 0);
606 764
607 // not very fast, and certainly not robust against illegal input
608 for (;;) 765 for (;;)
609 { 766 {
610 WANT (1); 767 WANT (1);
611 768
769 if ((*dec->cur ^ major) >= 31)
612 if (*dec->cur == (0xe0 | 31)) 770 if (*dec->cur == (0xe0 | 31))
613 { 771 {
614 ++dec->cur; 772 ++dec->cur;
615 break; 773 break;
616 } 774 }
775 else
776 ERR ("corrupted CBOR data (invalid chunks in indefinite length string)");
617 777
618 sv_catsv (sv, decode_sv (dec)); 778 STRLEN len = decode_uint (dec);
779
780 WANT (len);
781 sv_catpvn (sv, dec->cur, len);
782 dec->cur += len;
619 } 783 }
620 } 784 }
621 else 785 else
622 { 786 {
623 STRLEN len = decode_uint (dec); 787 STRLEN len = decode_uint (dec);
624 788
625 WANT (len); 789 WANT (len);
626 sv = newSVpvn (dec->cur, len); 790 sv = newSVpvn (dec->cur, len);
627 dec->cur += len; 791 dec->cur += len;
792
793 if (ecb_expect_false (dec->stringref)
794 && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1))
795 av_push (dec->stringref, SvREFCNT_inc_NN (sv));
628 } 796 }
629 797
630 if (utf8) 798 if (utf8)
631 SvUTF8_on (sv); 799 SvUTF8_on (sv);
632 800
638} 806}
639 807
640static SV * 808static SV *
641decode_tagged (dec_t *dec) 809decode_tagged (dec_t *dec)
642{ 810{
811 SV *sv = 0;
643 UV tag = decode_uint (dec); 812 UV tag = decode_uint (dec);
813
814 WANT (1);
815
816 switch (tag)
817 {
818 case CBOR_TAG_MAGIC:
819 sv = decode_sv (dec);
820 break;
821
822 case CBOR_TAG_INDIRECTION:
823 sv = newRV_noinc (decode_sv (dec));
824 break;
825
826 case CBOR_TAG_STRINGREF_NAMESPACE:
827 {
828 ENTER; SAVETMPS;
829
830 SAVESPTR (dec->stringref);
831 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ());
832
833 sv = decode_sv (dec);
834
835 FREETMPS; LEAVE;
836 }
837 break;
838
839 case CBOR_TAG_STRINGREF:
840 {
841 if ((*dec->cur >> 5) != 0)
842 ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
843
844 UV idx = decode_uint (dec);
845
846 if (!dec->stringref || (int)idx > AvFILLp (dec->stringref))
847 ERR ("corrupted CBOR data (stringref index out of bounds or outside namespace)");
848
849 sv = newSVsv (AvARRAY (dec->stringref)[idx]);
850 }
851 break;
852
853 case CBOR_TAG_VALUE_SHAREABLE:
854 {
855 if (ecb_expect_false (!dec->shareable))
856 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ());
857
858 sv = newSV (0);
859 av_push (dec->shareable, SvREFCNT_inc_NN (sv));
860
644 SV *sv = decode_sv (dec); 861 SV *osv = decode_sv (dec);
862 sv_setsv (sv, osv);
863 SvREFCNT_dec_NN (osv);
864 }
865 break;
645 866
646 if (tag == CBOR_TAG_MAGIC) 867 case CBOR_TAG_VALUE_SHAREDREF:
647 return sv; 868 {
869 if ((*dec->cur >> 5) != 0)
870 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
648 871
649 if (tag == CBOR_TAG_PERL_OBJECT) 872 UV idx = decode_uint (dec);
650 { 873
874 if (!dec->shareable || (int)idx > AvFILLp (dec->shareable))
875 ERR ("corrupted CBOR data (sharedref index out of bounds)");
876
877 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]);
878 }
879 break;
880
881 case CBOR_TAG_PERL_OBJECT:
882 {
883 sv = decode_sv (dec);
884
651 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV) 885 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
652 ERR ("corrupted CBOR data (non-array perl object)"); 886 ERR ("corrupted CBOR data (non-array perl object)");
887
888 AV *av = (AV *)SvRV (sv);
889 int len = av_len (av) + 1;
890 HV *stash = gv_stashsv (*av_fetch (av, 0, 1), 0);
891
892 if (!stash)
893 ERR ("cannot decode perl-object (package does not exist)");
894
895 GV *method = gv_fetchmethod_autoload (stash, "THAW", 0);
896
897 if (!method)
898 ERR ("cannot decode perl-object (package does not have a THAW method)");
899
900 dSP;
901
902 ENTER; SAVETMPS; PUSHMARK (SP);
903 EXTEND (SP, len + 1);
904 // we re-bless the reference to get overload and other niceties right
905 PUSHs (*av_fetch (av, 0, 1));
906 PUSHs (sv_cbor);
907
908 int i;
909
910 for (i = 1; i < len; ++i)
911 PUSHs (*av_fetch (av, i, 1));
912
913 PUTBACK;
914 call_sv ((SV *)GvCV (method), G_SCALAR | G_EVAL);
915 SPAGAIN;
916
917 if (SvTRUE (ERRSV))
918 {
919 FREETMPS; LEAVE;
920 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV))));
921 }
922
923 SvREFCNT_dec (sv);
924 sv = SvREFCNT_inc (POPs);
925
926 PUTBACK;
927
928 FREETMPS; LEAVE;
653 929 }
654 // TODO 930 break;
655 }
656 931
932 default:
933 {
934 sv = decode_sv (dec);
935
936 dSP;
937 ENTER; SAVETMPS; PUSHMARK (SP);
938 EXTEND (SP, 2);
939 PUSHs (newSVuv (tag));
940 PUSHs (sv);
941
942 PUTBACK;
943 int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL);
944 SPAGAIN;
945
946 if (SvTRUE (ERRSV))
947 {
948 FREETMPS; LEAVE;
949 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV))));
950 }
951
952 if (count)
953 {
954 SvREFCNT_dec (sv);
955 sv = SvREFCNT_inc (POPs);
956 }
957 else
958 {
657 AV *av = newAV (); 959 AV *av = newAV ();
658 av_push (av, newSVuv (tag)); 960 av_push (av, newSVuv (tag));
659 av_push (av, sv); 961 av_push (av, sv);
660 962
661 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 963 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
662 ? cbor_tagged_stash 964 ? cbor_tagged_stash
663 : gv_stashpv ("CBOR::XS::Tagged" , 1); 965 : gv_stashpv ("CBOR::XS::Tagged" , 1);
664
665 return sv_bless (newRV_noinc ((SV *)av), tagged_stash); 966 sv = sv_bless (newRV_noinc ((SV *)av), tagged_stash);
967 }
968
969 PUTBACK;
970
971 FREETMPS; LEAVE;
972 }
973 break;
974 }
975
976 return sv;
666 977
667fail: 978fail:
668 SvREFCNT_dec (sv); 979 SvREFCNT_dec (sv);
669 return &PL_sv_undef; 980 return &PL_sv_undef;
670} 981}
693 case 7: // misc 1004 case 7: // misc
694 switch (*dec->cur++ & 31) 1005 switch (*dec->cur++ & 31)
695 { 1006 {
696 case 20: 1007 case 20:
697#if CBOR_SLOW 1008#if CBOR_SLOW
698 cbor_false = get_bool ("CBOR::XS::false"); 1009 types_false = get_bool ("Types::Serialiser::false");
699#endif 1010#endif
700 return newSVsv (cbor_false); 1011 return newSVsv (types_false);
701 case 21: 1012 case 21:
702#if CBOR_SLOW 1013#if CBOR_SLOW
703 cbor_true = get_bool ("CBOR::XS::true"); 1014 types_true = get_bool ("Types::Serialiser::true");
704#endif 1015#endif
705 return newSVsv (cbor_true); 1016 return newSVsv (types_true);
706 case 22: 1017 case 22:
707 return newSVsv (&PL_sv_undef); 1018 return newSVsv (&PL_sv_undef);
1019 case 23:
1020#if CBOR_SLOW
1021 types_error = get_bool ("Types::Serialiser::error");
1022#endif
1023 return newSVsv (types_error);
708 1024
709 case 25: 1025 case 25:
710 { 1026 {
711 WANT (2); 1027 WANT (2);
712 1028
756} 1072}
757 1073
758static SV * 1074static SV *
759decode_cbor (SV *string, CBOR *cbor, char **offset_return) 1075decode_cbor (SV *string, CBOR *cbor, char **offset_return)
760{ 1076{
761 dec_t dec; 1077 dec_t dec = { };
762 SV *sv; 1078 SV *sv;
1079 STRLEN len;
1080 char *data = SvPVbyte (string, len);
763 1081
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) 1082 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", 1083 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); 1084 (unsigned long)len, (unsigned long)cbor->max_size);
795 }
796
797 sv_utf8_downgrade (string, 0);
798 1085
799 dec.cbor = *cbor; 1086 dec.cbor = *cbor;
800 dec.cur = (U8 *)SvPVX (string); 1087 dec.cur = (U8 *)data;
801 dec.end = (U8 *)SvEND (string); 1088 dec.end = (U8 *)data + len;
802 dec.err = 0;
803 dec.depth = 0;
804 1089
805 sv = decode_sv (&dec); 1090 sv = decode_sv (&dec);
806 1091
807 if (offset_return) 1092 if (offset_return)
808 *offset_return = dec.cur; 1093 *offset_return = dec.cur;
812 dec.err = "garbage after CBOR object"; 1097 dec.err = "garbage after CBOR object";
813 1098
814 if (dec.err) 1099 if (dec.err)
815 { 1100 {
816 SvREFCNT_dec (sv); 1101 SvREFCNT_dec (sv);
817 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)SvPVX (string), (int)(uint8_t)*dec.cur); 1102 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur);
818 } 1103 }
819 1104
820 sv = sv_2mortal (sv); 1105 sv = sv_2mortal (sv);
821 1106
822 return sv; 1107 return sv;
828MODULE = CBOR::XS PACKAGE = CBOR::XS 1113MODULE = CBOR::XS PACKAGE = CBOR::XS
829 1114
830BOOT: 1115BOOT:
831{ 1116{
832 cbor_stash = gv_stashpv ("CBOR::XS" , 1); 1117 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); 1118 cbor_tagged_stash = gv_stashpv ("CBOR::XS::Tagged" , 1);
835 1119
836 cbor_true = get_bool ("CBOR::XS::true"); 1120 types_boolean_stash = gv_stashpv ("Types::Serialiser::Boolean", 1);
837 cbor_false = get_bool ("CBOR::XS::false"); 1121 types_error_stash = gv_stashpv ("Types::Serialiser::Error" , 1);
1122
1123 types_true = get_bool ("Types::Serialiser::true" );
1124 types_false = get_bool ("Types::Serialiser::false");
1125 types_error = get_bool ("Types::Serialiser::error");
1126
1127 default_filter = newSVpv ("CBOR::XS::default_filter", 0);
1128
1129 sv_cbor = newSVpv ("CBOR", 0);
1130 SvREADONLY_on (sv_cbor);
838} 1131}
839 1132
840PROTOTYPES: DISABLE 1133PROTOTYPES: DISABLE
841 1134
842void CLONE (...) 1135void CLONE (...)
843 CODE: 1136 CODE:
844 cbor_stash = 0; 1137 cbor_stash = 0;
845 cbor_boolean_stash = 0;
846 cbor_tagged_stash = 0; 1138 cbor_tagged_stash = 0;
1139 types_error_stash = 0;
1140 types_boolean_stash = 0;
847 1141
848void new (char *klass) 1142void new (char *klass)
849 PPCODE: 1143 PPCODE:
850{ 1144{
851 SV *pv = NEWSV (0, sizeof (CBOR)); 1145 SV *pv = NEWSV (0, sizeof (CBOR));
859 1153
860void shrink (CBOR *self, int enable = 1) 1154void shrink (CBOR *self, int enable = 1)
861 ALIAS: 1155 ALIAS:
862 shrink = F_SHRINK 1156 shrink = F_SHRINK
863 allow_unknown = F_ALLOW_UNKNOWN 1157 allow_unknown = F_ALLOW_UNKNOWN
1158 allow_sharing = F_ALLOW_SHARING
1159 pack_strings = F_PACK_STRINGS
864 PPCODE: 1160 PPCODE:
865{ 1161{
866 if (enable) 1162 if (enable)
867 self->flags |= ix; 1163 self->flags |= ix;
868 else 1164 else
873 1169
874void get_shrink (CBOR *self) 1170void get_shrink (CBOR *self)
875 ALIAS: 1171 ALIAS:
876 get_shrink = F_SHRINK 1172 get_shrink = F_SHRINK
877 get_allow_unknown = F_ALLOW_UNKNOWN 1173 get_allow_unknown = F_ALLOW_UNKNOWN
1174 get_allow_sharing = F_ALLOW_SHARING
1175 get_pack_strings = F_PACK_STRINGS
878 PPCODE: 1176 PPCODE:
879 XPUSHs (boolSV (self->flags & ix)); 1177 XPUSHs (boolSV (self->flags & ix));
880 1178
881void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) 1179void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
882 PPCODE: 1180 PPCODE:
898 CODE: 1196 CODE:
899 RETVAL = self->max_size; 1197 RETVAL = self->max_size;
900 OUTPUT: 1198 OUTPUT:
901 RETVAL 1199 RETVAL
902 1200
903#if 0 //TODO 1201void filter (CBOR *self, SV *filter = 0)
904
905void filter_cbor_object (CBOR *self, SV *cb = &PL_sv_undef)
906 PPCODE: 1202 PPCODE:
907{
908 SvREFCNT_dec (self->cb_object); 1203 SvREFCNT_dec (self->filter);
909 self->cb_object = SvOK (cb) ? newSVsv (cb) : 0; 1204 self->filter = filter ? newSVsv (filter) : filter;
910
911 XPUSHs (ST (0)); 1205 XPUSHs (ST (0));
912}
913 1206
914void filter_cbor_single_key_object (CBOR *self, SV *key, SV *cb = &PL_sv_undef) 1207SV *get_filter (CBOR *self)
915 PPCODE: 1208 CODE:
916{ 1209 RETVAL = self->filter ? self->filter : NEWSV (0, 0);
917 if (!self->cb_sk_object) 1210 OUTPUT:
918 self->cb_sk_object = newHV (); 1211 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 1212
938void encode (CBOR *self, SV *scalar) 1213void encode (CBOR *self, SV *scalar)
939 PPCODE: 1214 PPCODE:
940 PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN; 1215 PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN;
941 XPUSHs (scalar); 1216 XPUSHs (scalar);
954 EXTEND (SP, 2); 1229 EXTEND (SP, 2);
955 PUSHs (sv); 1230 PUSHs (sv);
956 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); 1231 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
957} 1232}
958 1233
959#if 0
960
961void DESTROY (CBOR *self) 1234void DESTROY (CBOR *self)
962 CODE: 1235 PPCODE:
963 SvREFCNT_dec (self->cb_sk_object); 1236 cbor_free (self);
964 SvREFCNT_dec (self->cb_object);
965
966#endif
967 1237
968PROTOTYPES: ENABLE 1238PROTOTYPES: ENABLE
969 1239
970void encode_cbor (SV *scalar) 1240void encode_cbor (SV *scalar)
971 PPCODE: 1241 PPCODE:

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines