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 1.6 by root, Sat Oct 26 22:25:47 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
28// known tags
29enum cbor_tag
30{
31 // extensions
32 CBOR_TAG_STRINGREF = 25, // http://cbor.schmorp.de/stringref
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
39
40 // rfc7049
41 CBOR_TAG_DATETIME = 0, // rfc4287, utf-8
42 CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any
43 CBOR_TAG_POS_BIGNUM = 2, // byte string
44 CBOR_TAG_NEG_BIGNUM = 3, // byte string
45 CBOR_TAG_DECIMAL = 4, // decimal fraction, array
46 CBOR_TAG_BIGFLOAT = 5, // array
47
48 CBOR_TAG_CONV_B64U = 21, // base64url, any
49 CBOR_TAG_CONV_B64 = 22, // base64, any
50 CBOR_TAG_CONV_HEX = 23, // base16, any
51 CBOR_TAG_CBOR = 24, // embedded cbor, byte string
52
53 CBOR_TAG_URI = 32, // URI rfc3986, utf-8
54 CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8
55 CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8
56 CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8
57 CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8
58
59 CBOR_TAG_MAGIC = 55799 // self-describe cbor
60};
61
14#define F_SHRINK 0x00000200UL 62#define F_SHRINK 0x00000001UL
15#define F_ALLOW_UNKNOWN 0x00002000UL 63#define F_ALLOW_UNKNOWN 0x00000002UL
64#define F_ALLOW_SHARING 0x00000004UL
65#define F_PACK_STRINGS 0x00000008UL
16 66
17#define INIT_SIZE 32 // initial scalar size to be allocated 67#define INIT_SIZE 32 // initial scalar size to be allocated
18 68
19#define SB do { 69#define SB do {
20#define SE } while (0) 70#define SE } while (0)
… …
31#else 81#else
32# define CBOR_SLOW 0 82# define CBOR_SLOW 0
33# define CBOR_STASH cbor_stash 83# define CBOR_STASH cbor_stash
34#endif 84#endif
35 85
36static 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::
37static SV *cbor_true, *cbor_false; 87static SV *types_true, *types_false, *types_error, *sv_cbor, *default_filter;
38 88
39typedef struct { 89typedef struct {
40 U32 flags; 90 U32 flags;
41 U32 max_depth; 91 U32 max_depth;
42 STRLEN max_size; 92 STRLEN max_size;
93 SV *filter;
43} CBOR; 94} CBOR;
44 95
45ecb_inline void 96ecb_inline void
46cbor_init (CBOR *cbor) 97cbor_init (CBOR *cbor)
47{ 98{
48 Zero (cbor, 1, CBOR); 99 Zero (cbor, 1, CBOR);
49 cbor->max_depth = 512; 100 cbor->max_depth = 512;
101}
102
103ecb_inline void
104cbor_free (CBOR *cbor)
105{
106 SvREFCNT_dec (cbor->filter);
50} 107}
51 108
52///////////////////////////////////////////////////////////////////////////// 109/////////////////////////////////////////////////////////////////////////////
53// utility functions 110// utility functions
54 111
… …
76 SvPV_renew (sv, SvCUR (sv) + 1); 133 SvPV_renew (sv, SvCUR (sv) + 1);
77#endif 134#endif
78 } 135 }
79} 136}
80 137
81///////////////////////////////////////////////////////////////////////////// 138// minimum length of a string to be registered for stringref
82// fp hell 139ecb_inline int
83 140minimum_string_length (UV idx)
84//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}
85 152
86///////////////////////////////////////////////////////////////////////////// 153/////////////////////////////////////////////////////////////////////////////
87// encoder 154// encoder
88 155
89// structure used for encoding CBOR 156// structure used for encoding CBOR
… …
92 char *cur; // SvPVX (sv) + current output position 159 char *cur; // SvPVX (sv) + current output position
93 char *end; // SvEND (sv) 160 char *end; // SvEND (sv)
94 SV *sv; // result scalar 161 SV *sv; // result scalar
95 CBOR cbor; 162 CBOR cbor;
96 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;
97} enc_t; 168} enc_t;
98 169
99ecb_inline void 170ecb_inline void
100need (enc_t *enc, STRLEN len) 171need (enc_t *enc, STRLEN len)
101{ 172{
… …
118static void 189static void
119encode_uint (enc_t *enc, int major, UV len) 190encode_uint (enc_t *enc, int major, UV len)
120{ 191{
121 need (enc, 9); 192 need (enc, 9);
122 193
123 if (len < 24) 194 if (ecb_expect_true (len < 24))
124 *enc->cur++ = major | len; 195 *enc->cur++ = major | len;
125 else if (len <= 0xff) 196 else if (ecb_expect_true (len <= 0xff))
126 { 197 {
127 *enc->cur++ = major | 24; 198 *enc->cur++ = major | 24;
128 *enc->cur++ = len; 199 *enc->cur++ = len;
129 } 200 }
130 else if (len <= 0xffff) 201 else if (len <= 0xffff)
… …
153 *enc->cur++ = len >> 8; 224 *enc->cur++ = len >> 8;
154 *enc->cur++ = len; 225 *enc->cur++ = len;
155 } 226 }
156} 227}
157 228
158static void 229ecb_inline void
230encode_tag (enc_t *enc, UV tag)
231{
232 encode_uint (enc, 0xc0, tag);
233}
234
235ecb_inline void
159encode_str (enc_t *enc, int utf8, char *str, STRLEN len) 236encode_str (enc_t *enc, int utf8, char *str, STRLEN len)
160{ 237{
161 encode_uint (enc, utf8 ? 0x60 : 0x40, len); 238 encode_uint (enc, utf8 ? 0x60 : 0x40, len);
162 need (enc, len); 239 need (enc, len);
163 memcpy (enc->cur, str, len); 240 memcpy (enc->cur, str, len);
164 enc->cur += len; 241 enc->cur += len;
165} 242}
166 243
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
167static void encode_sv (enc_t *enc, SV *sv); 269static void encode_sv (enc_t *enc, SV *sv);
168 270
169static void 271static void
170encode_av (enc_t *enc, AV *av) 272encode_av (enc_t *enc, AV *av)
171{ 273{
… …
208 while ((he = hv_iternext (hv))) 310 while ((he = hv_iternext (hv)))
209 { 311 {
210 if (HeKLEN (he) == HEf_SVKEY) 312 if (HeKLEN (he) == HEf_SVKEY)
211 encode_sv (enc, HeSVKEY (he)); 313 encode_sv (enc, HeSVKEY (he));
212 else 314 else
213 encode_str (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he)); 315 encode_strref (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he));
214 316
215 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));
216 } 318 }
217 319
218 if (mg) 320 if (mg)
… …
223 325
224// encode objects, arrays and special \0=false and \1=true values. 326// encode objects, arrays and special \0=false and \1=true values.
225static void 327static void
226encode_rv (enc_t *enc, SV *sv) 328encode_rv (enc_t *enc, SV *sv)
227{ 329{
228 svtype svt;
229
230 SvGETMAGIC (sv); 330 SvGETMAGIC (sv);
331
231 svt = SvTYPE (sv); 332 svtype svt = SvTYPE (sv);
232 333
233 if (ecb_expect_false (SvOBJECT (sv))) 334 if (ecb_expect_false (SvOBJECT (sv)))
234 { 335 {
235 HV *boolean_stash = !CBOR_SLOW || cbor_boolean_stash 336 HV *boolean_stash = !CBOR_SLOW || types_boolean_stash
236 ? cbor_boolean_stash 337 ? types_boolean_stash
237 : 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);
238 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 342 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
239 ? cbor_tagged_stash 343 ? cbor_tagged_stash
240 : gv_stashpv ("CBOR::XS::Tagged" , 1); 344 : gv_stashpv ("CBOR::XS::Tagged" , 1);
241 345
346 HV *stash = SvSTASH (sv);
347 GV *method;
348
242 if (SvSTASH (sv) == boolean_stash) 349 if (stash == boolean_stash)
243 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);
244 else if (SvSTASH (sv) == tagged_stash) 353 else if (stash == tagged_stash)
245 { 354 {
246 if (svt != SVt_PVAV) 355 if (svt != SVt_PVAV)
247 croak ("encountered CBOR::XS::Tagged object that isn't an array"); 356 croak ("encountered CBOR::XS::Tagged object that isn't an array");
248 357
249 encode_uint (enc, 0xc0, SvUV (*av_fetch ((AV *)sv, 0, 1))); 358 encode_uint (enc, 0xc0, SvUV (*av_fetch ((AV *)sv, 0, 1)));
250 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1)); 359 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1));
251 } 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 }
252 else 413 else
253 {
254 // we re-bless the reference to get overload and other niceties right
255 GV *to_cbor = gv_fetchmethod_autoload (SvSTASH (sv), "TO_CBOR", 0);
256
257 if (to_cbor)
258 {
259 dSP;
260
261 ENTER; SAVETMPS; PUSHMARK (SP);
262 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv)));
263
264 // calling with G_SCALAR ensures that we always get a 1 return value
265 PUTBACK;
266 call_sv ((SV *)GvCV (to_cbor), G_SCALAR);
267 SPAGAIN;
268
269 // catch this surprisingly common error
270 if (SvROK (TOPs) && SvRV (TOPs) == sv)
271 croak ("%s::TO_CBOR method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv)));
272
273 sv = POPs;
274 PUTBACK;
275
276 encode_sv (enc, sv);
277
278 FREETMPS; LEAVE;
279 }
280 else
281 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",
282 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 415 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
283 }
284 } 416 }
285 else if (svt == SVt_PVHV) 417 else if (svt == SVt_PVHV)
286 encode_hv (enc, (HV *)sv); 418 encode_hv (enc, (HV *)sv);
287 else if (svt == SVt_PVAV) 419 else if (svt == SVt_PVAV)
288 encode_av (enc, (AV *)sv); 420 encode_av (enc, (AV *)sv);
289 else if (svt < SVt_PVAV)
290 {
291 STRLEN len = 0;
292 char *pv = svt ? SvPV (sv, len) : 0;
293
294 if (len == 1 && *pv == '1')
295 encode_ch (enc, 0xe0 | 21);
296 else if (len == 1 && *pv == '0')
297 encode_ch (enc, 0xe0 | 20);
298 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
299 encode_ch (enc, 0xe0 | 23);
300 else
301 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
302 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
303 }
304 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
305 encode_ch (enc, 0xe0 | 23);
306 else 421 else
307 croak ("encountered %s, but CBOR can only represent references to arrays or hashes", 422 {
308 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 }
309} 448}
310 449
311static void 450static void
312encode_nv (enc_t *enc, SV *sv) 451encode_nv (enc_t *enc, SV *sv)
313{ 452{
… …
351 490
352 if (SvPOKp (sv)) 491 if (SvPOKp (sv))
353 { 492 {
354 STRLEN len; 493 STRLEN len;
355 char *str = SvPV (sv, len); 494 char *str = SvPV (sv, len);
356 encode_str (enc, SvUTF8 (sv), str, len); 495 encode_strref (enc, SvUTF8 (sv), str, len);
357 } 496 }
358 else if (SvNOKp (sv)) 497 else if (SvNOKp (sv))
359 encode_nv (enc, sv); 498 encode_nv (enc, sv);
360 else if (SvIOKp (sv)) 499 else if (SvIOKp (sv))
361 { 500 {
… …
378} 517}
379 518
380static SV * 519static SV *
381encode_cbor (SV *scalar, CBOR *cbor) 520encode_cbor (SV *scalar, CBOR *cbor)
382{ 521{
383 enc_t enc; 522 enc_t enc = { };
384 523
385 enc.cbor = *cbor; 524 enc.cbor = *cbor;
386 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 525 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
387 enc.cur = SvPVX (enc.sv); 526 enc.cur = SvPVX (enc.sv);
388 enc.end = SvEND (enc.sv); 527 enc.end = SvEND (enc.sv);
389 enc.depth = 0;
390 528
391 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
392 encode_sv (&enc, scalar); 538 encode_sv (&enc, scalar);
393 539
394 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 540 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
395 *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
396 542
… …
410 U8 *end; // end of input string 556 U8 *end; // end of input string
411 const char *err; // parse error, if != 0 557 const char *err; // parse error, if != 0
412 CBOR cbor; 558 CBOR cbor;
413 U32 depth; // recursion depth 559 U32 depth; // recursion depth
414 U32 maxdepth; // recursion depth limit 560 U32 maxdepth; // recursion depth limit
561 AV *shareable;
562 AV *stringref;
563 SV *decode_tagged;
415} dec_t; 564} dec_t;
416 565
417#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
418 567
419#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")
… …
513 SvREFCNT_dec (av); 662 SvREFCNT_dec (av);
514 DEC_DEC_DEPTH; 663 DEC_DEC_DEPTH;
515 return &PL_sv_undef; 664 return &PL_sv_undef;
516} 665}
517 666
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
518static SV * 710static SV *
519decode_hv (dec_t *dec) 711decode_hv (dec_t *dec)
520{ 712{
521 HV *hv = newHV (); 713 HV *hv = newHV ();
522 714
… …
534 { 726 {
535 ++dec->cur; 727 ++dec->cur;
536 break; 728 break;
537 } 729 }
538 730
539 SV *k = decode_sv (dec); 731 decode_he (dec, hv);
540 SV *v = decode_sv (dec);
541
542 hv_store_ent (hv, k, v, 0);
543 } 732 }
544 } 733 }
545 else 734 else
546 { 735 {
547 int len = decode_uint (dec); 736 int pairs = decode_uint (dec);
548 737
549 while (len--) 738 while (pairs--)
550 { 739 decode_he (dec, hv);
551 SV *k = decode_sv (dec);
552 SV *v = decode_sv (dec);
553
554 hv_store_ent (hv, k, v, 0);
555 }
556 } 740 }
557 741
558 DEC_DEC_DEPTH; 742 DEC_DEC_DEPTH;
559 return newRV_noinc ((SV *)hv); 743 return newRV_noinc ((SV *)hv);
560 744
… …
569{ 753{
570 SV *sv = 0; 754 SV *sv = 0;
571 755
572 if ((*dec->cur & 31) == 31) 756 if ((*dec->cur & 31) == 31)
573 { 757 {
758 // indefinite length strings
574 ++dec->cur; 759 ++dec->cur;
575 760
761 unsigned char major = *dec->cur & 0xe0;
762
576 sv = newSVpvn ("", 0); 763 sv = newSVpvn ("", 0);
577 764
578 // not very fast, and certainly not robust against illegal input
579 for (;;) 765 for (;;)
580 { 766 {
581 WANT (1); 767 WANT (1);
582 768
769 if ((*dec->cur ^ major) >= 31)
583 if (*dec->cur == (0xe0 | 31)) 770 if (*dec->cur == (0xe0 | 31))
584 { 771 {
585 ++dec->cur; 772 ++dec->cur;
586 break; 773 break;
587 } 774 }
775 else
776 ERR ("corrupted CBOR data (invalid chunks in indefinite length string)");
588 777
589 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;
590 } 783 }
591 } 784 }
592 else 785 else
593 { 786 {
594 STRLEN len = decode_uint (dec); 787 STRLEN len = decode_uint (dec);
595 788
596 WANT (len); 789 WANT (len);
597 sv = newSVpvn (dec->cur, len); 790 sv = newSVpvn (dec->cur, len);
598 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));
599 } 796 }
600 797
601 if (utf8) 798 if (utf8)
602 SvUTF8_on (sv); 799 SvUTF8_on (sv);
603 800
… …
609} 806}
610 807
611static SV * 808static SV *
612decode_tagged (dec_t *dec) 809decode_tagged (dec_t *dec)
613{ 810{
811 SV *sv = 0;
614 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
615 SV *sv = decode_sv (dec); 861 SV *osv = decode_sv (dec);
862 sv_setsv (sv, osv);
863 SvREFCNT_dec_NN (osv);
864 }
865 break;
616 866
617 if (tag == 55799) // 2.4.5 Self-Describe CBOR 867 case CBOR_TAG_VALUE_SHAREDREF:
868 {
869 if ((*dec->cur >> 5) != 0)
870 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
871
872 UV idx = decode_uint (dec);
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
885 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
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;
929 }
930 break;
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 {
959 AV *av = newAV ();
960 av_push (av, newSVuv (tag));
961 av_push (av, sv);
962
963 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
964 ? cbor_tagged_stash
965 : gv_stashpv ("CBOR::XS::Tagged" , 1);
966 sv = sv_bless (newRV_noinc ((SV *)av), tagged_stash);
967 }
968
969 PUTBACK;
970
971 FREETMPS; LEAVE;
972 }
973 break;
974 }
975
618 return sv; 976 return sv;
619 977
620 AV *av = newAV (); 978fail:
621 av_push (av, newSVuv (tag)); 979 SvREFCNT_dec (sv);
622 av_push (av, sv); 980 return &PL_sv_undef;
623 return newRV_noinc ((SV *)av);
624} 981}
625 982
626static SV * 983static SV *
627decode_sv (dec_t *dec) 984decode_sv (dec_t *dec)
628{ 985{
… …
647 case 7: // misc 1004 case 7: // misc
648 switch (*dec->cur++ & 31) 1005 switch (*dec->cur++ & 31)
649 { 1006 {
650 case 20: 1007 case 20:
651#if CBOR_SLOW 1008#if CBOR_SLOW
652 cbor_false = get_bool ("CBOR::XS::false"); 1009 types_false = get_bool ("Types::Serialiser::false");
653#endif 1010#endif
654 return newSVsv (cbor_false); 1011 return newSVsv (types_false);
655 case 21: 1012 case 21:
656#if CBOR_SLOW 1013#if CBOR_SLOW
657 cbor_true = get_bool ("CBOR::XS::true"); 1014 types_true = get_bool ("Types::Serialiser::true");
658#endif 1015#endif
659 return newSVsv (cbor_true); 1016 return newSVsv (types_true);
660 case 22: 1017 case 22:
661 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);
662 1024
663 case 25: 1025 case 25:
664 { 1026 {
665 WANT (2); 1027 WANT (2);
666 1028
… …
710} 1072}
711 1073
712static SV * 1074static SV *
713decode_cbor (SV *string, CBOR *cbor, char **offset_return) 1075decode_cbor (SV *string, CBOR *cbor, char **offset_return)
714{ 1076{
715 dec_t dec; 1077 dec_t dec = { };
716 SV *sv; 1078 SV *sv;
1079 STRLEN len;
1080 char *data = SvPVbyte (string, len);
717 1081
718 /* work around bugs in 5.10 where manipulating magic values
719 * makes perl ignore the magic in subsequent accesses.
720 * also make a copy of non-PV values, to get them into a clean
721 * state (SvPV should do that, but it's buggy, see below).
722 */
723 /*SvGETMAGIC (string);*/
724 if (SvMAGICAL (string) || !SvPOK (string))
725 string = sv_2mortal (newSVsv (string));
726
727 SvUPGRADE (string, SVt_PV);
728
729 /* work around a bug in perl 5.10, which causes SvCUR to fail an
730 * assertion with -DDEBUGGING, although SvCUR is documented to
731 * return the xpv_cur field which certainly exists after upgrading.
732 * according to nicholas clark, calling SvPOK fixes this.
733 * But it doesn't fix it, so try another workaround, call SvPV_nolen
734 * and hope for the best.
735 * Damnit, SvPV_nolen still trips over yet another assertion. This
736 * assertion business is seriously broken, try yet another workaround
737 * for the broken -DDEBUGGING.
738 */
739 {
740#ifdef DEBUGGING
741 STRLEN offset = SvOK (string) ? sv_len (string) : 0;
742#else
743 STRLEN offset = SvCUR (string);
744#endif
745
746 if (offset > cbor->max_size && cbor->max_size) 1082 if (len > cbor->max_size && cbor->max_size)
747 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",
748 (unsigned long)SvCUR (string), (unsigned long)cbor->max_size); 1084 (unsigned long)len, (unsigned long)cbor->max_size);
749 }
750
751 sv_utf8_downgrade (string, 0);
752 1085
753 dec.cbor = *cbor; 1086 dec.cbor = *cbor;
754 dec.cur = (U8 *)SvPVX (string); 1087 dec.cur = (U8 *)data;
755 dec.end = (U8 *)SvEND (string); 1088 dec.end = (U8 *)data + len;
756 dec.err = 0;
757 dec.depth = 0;
758 1089
759 sv = decode_sv (&dec); 1090 sv = decode_sv (&dec);
760 1091
761 if (offset_return) 1092 if (offset_return)
762 *offset_return = dec.cur; 1093 *offset_return = dec.cur;
… …
766 dec.err = "garbage after CBOR object"; 1097 dec.err = "garbage after CBOR object";
767 1098
768 if (dec.err) 1099 if (dec.err)
769 { 1100 {
770 SvREFCNT_dec (sv); 1101 SvREFCNT_dec (sv);
771 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);
772 } 1103 }
773 1104
774 sv = sv_2mortal (sv); 1105 sv = sv_2mortal (sv);
775 1106
776 return sv; 1107 return sv;
… …
782MODULE = CBOR::XS PACKAGE = CBOR::XS 1113MODULE = CBOR::XS PACKAGE = CBOR::XS
783 1114
784BOOT: 1115BOOT:
785{ 1116{
786 cbor_stash = gv_stashpv ("CBOR::XS" , 1); 1117 cbor_stash = gv_stashpv ("CBOR::XS" , 1);
787 cbor_boolean_stash = gv_stashpv ("CBOR::XS::Boolean", 1);
788 cbor_tagged_stash = gv_stashpv ("CBOR::XS::Tagged" , 1); 1118 cbor_tagged_stash = gv_stashpv ("CBOR::XS::Tagged" , 1);
789 1119
790 cbor_true = get_bool ("CBOR::XS::true"); 1120 types_boolean_stash = gv_stashpv ("Types::Serialiser::Boolean", 1);
791 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);
792} 1131}
793 1132
794PROTOTYPES: DISABLE 1133PROTOTYPES: DISABLE
795 1134
796void CLONE (...) 1135void CLONE (...)
797 CODE: 1136 CODE:
798 cbor_stash = 0; 1137 cbor_stash = 0;
799 cbor_boolean_stash = 0;
800 cbor_tagged_stash = 0; 1138 cbor_tagged_stash = 0;
1139 types_error_stash = 0;
1140 types_boolean_stash = 0;
801 1141
802void new (char *klass) 1142void new (char *klass)
803 PPCODE: 1143 PPCODE:
804{ 1144{
805 SV *pv = NEWSV (0, sizeof (CBOR)); 1145 SV *pv = NEWSV (0, sizeof (CBOR));
… …
813 1153
814void shrink (CBOR *self, int enable = 1) 1154void shrink (CBOR *self, int enable = 1)
815 ALIAS: 1155 ALIAS:
816 shrink = F_SHRINK 1156 shrink = F_SHRINK
817 allow_unknown = F_ALLOW_UNKNOWN 1157 allow_unknown = F_ALLOW_UNKNOWN
1158 allow_sharing = F_ALLOW_SHARING
1159 pack_strings = F_PACK_STRINGS
818 PPCODE: 1160 PPCODE:
819{ 1161{
820 if (enable) 1162 if (enable)
821 self->flags |= ix; 1163 self->flags |= ix;
822 else 1164 else
… …
827 1169
828void get_shrink (CBOR *self) 1170void get_shrink (CBOR *self)
829 ALIAS: 1171 ALIAS:
830 get_shrink = F_SHRINK 1172 get_shrink = F_SHRINK
831 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
832 PPCODE: 1176 PPCODE:
833 XPUSHs (boolSV (self->flags & ix)); 1177 XPUSHs (boolSV (self->flags & ix));
834 1178
835void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) 1179void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
836 PPCODE: 1180 PPCODE:
… …
852 CODE: 1196 CODE:
853 RETVAL = self->max_size; 1197 RETVAL = self->max_size;
854 OUTPUT: 1198 OUTPUT:
855 RETVAL 1199 RETVAL
856 1200
857#if 0 //TODO 1201void filter (CBOR *self, SV *filter = 0)
858
859void filter_cbor_object (CBOR *self, SV *cb = &PL_sv_undef)
860 PPCODE: 1202 PPCODE:
861{
862 SvREFCNT_dec (self->cb_object); 1203 SvREFCNT_dec (self->filter);
863 self->cb_object = SvOK (cb) ? newSVsv (cb) : 0; 1204 self->filter = filter ? newSVsv (filter) : filter;
864
865 XPUSHs (ST (0)); 1205 XPUSHs (ST (0));
866}
867 1206
868void filter_cbor_single_key_object (CBOR *self, SV *key, SV *cb = &PL_sv_undef) 1207SV *get_filter (CBOR *self)
869 PPCODE: 1208 CODE:
870{ 1209 RETVAL = self->filter ? self->filter : NEWSV (0, 0);
871 if (!self->cb_sk_object) 1210 OUTPUT:
872 self->cb_sk_object = newHV (); 1211 RETVAL
873
874 if (SvOK (cb))
875 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);
876 else
877 {
878 hv_delete_ent (self->cb_sk_object, key, G_DISCARD, 0);
879
880 if (!HvKEYS (self->cb_sk_object))
881 {
882 SvREFCNT_dec (self->cb_sk_object);
883 self->cb_sk_object = 0;
884 }
885 }
886
887 XPUSHs (ST (0));
888}
889
890#endif
891 1212
892void encode (CBOR *self, SV *scalar) 1213void encode (CBOR *self, SV *scalar)
893 PPCODE: 1214 PPCODE:
894 PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN; 1215 PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN;
895 XPUSHs (scalar); 1216 XPUSHs (scalar);
… …
908 EXTEND (SP, 2); 1229 EXTEND (SP, 2);
909 PUSHs (sv); 1230 PUSHs (sv);
910 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); 1231 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
911} 1232}
912 1233
913#if 0
914
915void DESTROY (CBOR *self) 1234void DESTROY (CBOR *self)
916 CODE: 1235 PPCODE:
917 SvREFCNT_dec (self->cb_sk_object); 1236 cbor_free (self);
918 SvREFCNT_dec (self->cb_object);
919
920#endif
921 1237
922PROTOTYPES: ENABLE 1238PROTOTYPES: ENABLE
923 1239
924void encode_cbor (SV *scalar) 1240void encode_cbor (SV *scalar)
925 PPCODE: 1241 PPCODE:

Diff Legend

– Removed lines
+ Added lines
< Changed lines
> Changed lines