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 1.27 by root, Fri Nov 22 15:28:38 2013 UTC vs.
Revision 1.36 by root, Sat Nov 30 17:37:45 2013 UTC

… …
19# define HvNAMELEN(hv) HvNAMELEN_get (hv) 19# define HvNAMELEN(hv) HvNAMELEN_get (hv)
20#endif 20#endif
21#ifndef HvNAMEUTF8 21#ifndef HvNAMEUTF8
22# define HvNAMEUTF8(hv) 0 22# define HvNAMEUTF8(hv) 0
23#endif 23#endif
24#ifndef SvREFCNT_dec_NN
25# define SvREFCNT_dec_NN(sv) SvREFCNT_dec (sv)
26#endif
27
28// known major and minor types
29enum cbor_type
30{
31 MAJOR_SHIFT = 5,
32 MINOR_MASK = 0x1f,
33
34 MAJOR_POS_INT = 0 << MAJOR_SHIFT,
35 MAJOR_NEG_INT = 1 << MAJOR_SHIFT,
36 MAJOR_BYTES = 2 << MAJOR_SHIFT,
37 MAJOR_TEXT = 3 << MAJOR_SHIFT,
38 MAJOR_ARRAY = 4 << MAJOR_SHIFT,
39 MAJOR_MAP = 5 << MAJOR_SHIFT,
40 MAJOR_TAG = 6 << MAJOR_SHIFT,
41 MAJOR_MISC = 7 << MAJOR_SHIFT,
42
43 // INT/STRING/ARRAY/MAP subtypes
44 LENGTH_EXT1 = 24,
45 LENGTH_EXT2 = 25,
46 LENGTH_EXT4 = 26,
47 LENGTH_EXT8 = 27,
48
49 // SIMPLE types (effectively MISC subtypes)
50 SIMPLE_FALSE = 20,
51 SIMPLE_TRUE = 21,
52 SIMPLE_NULL = 22,
53 SIMPLE_UNDEF = 23,
54
55 // MISC subtype (unused)
56 MISC_EXT1 = 24,
57 MISC_FLOAT16 = 25,
58 MISC_FLOAT32 = 26,
59 MISC_FLOAT64 = 27,
60
61 // BYTES/TEXT/ARRAY/MAP
62 MINOR_INDEF = 31,
63};
24 64
25// known tags 65// known tags
26enum cbor_tag 66enum cbor_tag
27{ 67{
28 // inofficial extensions (pending iana registration) 68 // extensions
69 CBOR_TAG_STRINGREF = 25, // http://cbor.schmorp.de/stringref
29 CBOR_TAG_PERL_OBJECT = 24, // http://cbor.schmorp.de/perl-object 70 CBOR_TAG_PERL_OBJECT = 26, // http://cbor.schmorp.de/perl-object
30 CBOR_TAG_GENERIC_OBJECT = 25, // http://cbor.schmorp.de/generic-object 71 CBOR_TAG_GENERIC_OBJECT = 27, // http://cbor.schmorp.de/generic-object
31 CBOR_TAG_VALUE_SHAREABLE = 26, // http://cbor.schmorp.de/value-sharing 72 CBOR_TAG_VALUE_SHAREABLE = 28, // http://cbor.schmorp.de/value-sharing
32 CBOR_TAG_VALUE_SHAREDREF = 27, // http://cbor.schmorp.de/value-sharing 73 CBOR_TAG_VALUE_SHAREDREF = 29, // http://cbor.schmorp.de/value-sharing
33 CBOR_TAG_STRINGREF_NAMESPACE = 65537, // http://cbor.schmorp.de/stringref 74 CBOR_TAG_STRINGREF_NAMESPACE = 256, // http://cbor.schmorp.de/stringref
34 CBOR_TAG_STRINGREF = 28, // http://cbor.schmorp.de/stringref
35 CBOR_TAG_INDIRECTION = 22098, // http://cbor.schmorp.de/indirection 75 CBOR_TAG_INDIRECTION = 22098, // http://cbor.schmorp.de/indirection
36 76
37 // rfc7049 77 // rfc7049
38 CBOR_TAG_DATETIME = 0, // rfc4287, utf-8 78 CBOR_TAG_DATETIME = 0, // rfc4287, utf-8
39 CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any 79 CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any
40 CBOR_TAG_POS_BIGNUM = 2, // byte string 80 CBOR_TAG_POS_BIGNUM = 2, // byte string
41 CBOR_TAG_NEG_BIGNUM = 3, // byte string 81 CBOR_TAG_NEG_BIGNUM = 3, // byte string
42 CBOR_TAG_DECIMAL = 4, // decimal fraction, array 82 CBOR_TAG_DECIMAL = 4, // decimal fraction, array
43 CBOR_TAG_BIGFLOAT = 5, // array 83 CBOR_TAG_BIGFLOAT = 5, // array
44 84
45 CBOR_TAG_CONV_B64U = 21, // base64url, any 85 CBOR_TAG_CONV_B64U = 21, // base64url, any
46 CBOR_TAG_CONV_B64 = 22, // base64, any 86 CBOR_TAG_CONV_B64 = 22, // base64, any
47 CBOR_TAG_CONV_HEX = 23, // base16, any 87 CBOR_TAG_CONV_HEX = 23, // base16, any
48 CBOR_TAG_CBOR = 24, // embedded cbor, byte string 88 CBOR_TAG_CBOR = 24, // embedded cbor, byte string
49 89
50 CBOR_TAG_URI = 32, // URI rfc3986, utf-8 90 CBOR_TAG_URI = 32, // URI rfc3986, utf-8
51 CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8 91 CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8
52 CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8 92 CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8
53 CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8 93 CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8
54 CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8 94 CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8
55 95
56 CBOR_TAG_MAGIC = 55799 // self-describe cbor 96 CBOR_TAG_MAGIC = 55799, // self-describe cbor
57}; 97};
58 98
59#define F_SHRINK 0x00000001UL 99#define F_SHRINK 0x00000001UL
60#define F_ALLOW_UNKNOWN 0x00000002UL 100#define F_ALLOW_UNKNOWN 0x00000002UL
61#define F_ALLOW_SHARING 0x00000004UL //TODO 101#define F_ALLOW_SHARING 0x00000004UL
62#define F_ALLOW_STRINGREF 0x00000008UL //TODO 102#define F_PACK_STRINGS 0x00000008UL
63 103
64#define INIT_SIZE 32 // initial scalar size to be allocated 104#define INIT_SIZE 32 // initial scalar size to be allocated
65 105
66#define SB do { 106#define SB do {
67#define SE } while (0) 107#define SE } while (0)
… …
138{ 178{
139 return idx > 23 179 return idx > 23
140 ? idx > 0xffU 180 ? idx > 0xffU
141 ? idx > 0xffffU 181 ? idx > 0xffffU
142 ? idx > 0xffffffffU 182 ? idx > 0xffffffffU
183 ? 11
143 ? 7 184 : 7
144 : 6
145 : 5 185 : 5
146 : 4 186 : 4
147 : 3; 187 : 3;
148} 188}
149 189
… …
186static void 226static void
187encode_uint (enc_t *enc, int major, UV len) 227encode_uint (enc_t *enc, int major, UV len)
188{ 228{
189 need (enc, 9); 229 need (enc, 9);
190 230
191 if (len < 24) 231 if (ecb_expect_true (len < LENGTH_EXT1))
192 *enc->cur++ = major | len; 232 *enc->cur++ = major | len;
193 else if (len <= 0xff) 233 else if (ecb_expect_true (len <= 0xffU))
194 { 234 {
195 *enc->cur++ = major | 24; 235 *enc->cur++ = major | LENGTH_EXT1;
196 *enc->cur++ = len; 236 *enc->cur++ = len;
197 } 237 }
198 else if (len <= 0xffff) 238 else if (len <= 0xffffU)
199 { 239 {
200 *enc->cur++ = major | 25; 240 *enc->cur++ = major | LENGTH_EXT2;
201 *enc->cur++ = len >> 8; 241 *enc->cur++ = len >> 8;
202 *enc->cur++ = len; 242 *enc->cur++ = len;
203 } 243 }
204 else if (len <= 0xffffffff) 244 else if (len <= 0xffffffffU)
205 { 245 {
206 *enc->cur++ = major | 26; 246 *enc->cur++ = major | LENGTH_EXT4;
207 *enc->cur++ = len >> 24; 247 *enc->cur++ = len >> 24;
208 *enc->cur++ = len >> 16; 248 *enc->cur++ = len >> 16;
209 *enc->cur++ = len >> 8; 249 *enc->cur++ = len >> 8;
210 *enc->cur++ = len; 250 *enc->cur++ = len;
211 } 251 }
212 else 252 else
213 { 253 {
214 *enc->cur++ = major | 27; 254 *enc->cur++ = major | LENGTH_EXT8;
215 *enc->cur++ = len >> 56; 255 *enc->cur++ = len >> 56;
216 *enc->cur++ = len >> 48; 256 *enc->cur++ = len >> 48;
217 *enc->cur++ = len >> 40; 257 *enc->cur++ = len >> 40;
218 *enc->cur++ = len >> 32; 258 *enc->cur++ = len >> 32;
219 *enc->cur++ = len >> 24; 259 *enc->cur++ = len >> 24;
… …
224} 264}
225 265
226ecb_inline void 266ecb_inline void
227encode_tag (enc_t *enc, UV tag) 267encode_tag (enc_t *enc, UV tag)
228{ 268{
229 encode_uint (enc, 0xc0, tag); 269 encode_uint (enc, MAJOR_TAG, tag);
270}
271
272ecb_inline void
273encode_str (enc_t *enc, int utf8, char *str, STRLEN len)
274{
275 encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len);
276 need (enc, len);
277 memcpy (enc->cur, str, len);
278 enc->cur += len;
230} 279}
231 280
232static void 281static void
233encode_str (enc_t *enc, int utf8, char *str, STRLEN len) 282encode_strref (enc_t *enc, int utf8, char *str, STRLEN len)
234{ 283{
235 if (ecb_expect_false (enc->cbor.flags & F_ALLOW_STRINGREF)) 284 if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS))
236 { 285 {
237 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1); 286 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1);
238 287
239 if (SvOK (*svp)) 288 if (SvOK (*svp))
240 { 289 {
241 // already registered, use stringref 290 // already registered, use stringref
242 encode_tag (enc, CBOR_TAG_STRINGREF); 291 encode_tag (enc, CBOR_TAG_STRINGREF);
243 encode_uint (enc, 0x00, SvUV (*svp)); 292 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
244 return; 293 return;
245 } 294 }
246 else if (len >= minimum_string_length (enc->stringref_idx)) 295 else if (len >= minimum_string_length (enc->stringref_idx))
247 { 296 {
248 // register only 297 // register only
249 sv_setuv (*svp, enc->stringref_idx); 298 sv_setuv (*svp, enc->stringref_idx);
250 ++enc->stringref_idx; 299 ++enc->stringref_idx;
251 } 300 }
252 } 301 }
253 302
254 encode_uint (enc, utf8 ? 0x60 : 0x40, len); 303 encode_str (enc, utf8, str, len);
255 need (enc, len);
256 memcpy (enc->cur, str, len);
257 enc->cur += len;
258} 304}
259 305
260static void encode_sv (enc_t *enc, SV *sv); 306static void encode_sv (enc_t *enc, SV *sv);
261 307
262static void 308static void
… …
267 if (enc->depth >= enc->cbor.max_depth) 313 if (enc->depth >= enc->cbor.max_depth)
268 croak (ERR_NESTING_EXCEEDED); 314 croak (ERR_NESTING_EXCEEDED);
269 315
270 ++enc->depth; 316 ++enc->depth;
271 317
272 encode_uint (enc, 0x80, len + 1); 318 encode_uint (enc, MAJOR_ARRAY, len + 1);
273 319
274 for (i = 0; i <= len; ++i) 320 for (i = 0; i <= len; ++i)
275 { 321 {
276 SV **svp = av_fetch (av, i, 0); 322 SV **svp = av_fetch (av, i, 0);
277 encode_sv (enc, svp ? *svp : &PL_sv_undef); 323 encode_sv (enc, svp ? *svp : &PL_sv_undef);
… …
292 338
293 int pairs = hv_iterinit (hv); 339 int pairs = hv_iterinit (hv);
294 int mg = SvMAGICAL (hv); 340 int mg = SvMAGICAL (hv);
295 341
296 if (mg) 342 if (mg)
297 encode_ch (enc, 0xa0 | 31); 343 encode_ch (enc, MAJOR_MAP | MINOR_INDEF);
298 else 344 else
299 encode_uint (enc, 0xa0, pairs); 345 encode_uint (enc, MAJOR_MAP, pairs);
300 346
301 while ((he = hv_iternext (hv))) 347 while ((he = hv_iternext (hv)))
302 { 348 {
303 if (HeKLEN (he) == HEf_SVKEY) 349 if (HeKLEN (he) == HEf_SVKEY)
304 encode_sv (enc, HeSVKEY (he)); 350 encode_sv (enc, HeSVKEY (he));
305 else 351 else
306 encode_str (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he)); 352 encode_strref (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he));
307 353
308 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he)); 354 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he));
309 } 355 }
310 356
311 if (mg) 357 if (mg)
312 encode_ch (enc, 0xe0 | 31); 358 encode_ch (enc, MAJOR_MISC | MINOR_INDEF);
313 359
314 --enc->depth; 360 --enc->depth;
315} 361}
316 362
317// encode objects, arrays and special \0=false and \1=true values. 363// encode objects, arrays and special \0=false and \1=true values.
318static void 364static void
319encode_rv (enc_t *enc, SV *sv) 365encode_rv (enc_t *enc, SV *sv)
320{ 366{
321 SvGETMAGIC (sv); 367 SvGETMAGIC (sv);
322
323 if (ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING)
324 && ecb_expect_false (SvREFCNT (sv) > 1))
325 {
326 if (!enc->shareable)
327 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
328
329 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
330
331 if (SvOK (*svp))
332 {
333 encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF);
334 encode_uint (enc, 0x00, SvUV (*svp));
335 return;
336 }
337 else
338 {
339 sv_setuv (*svp, enc->shareable_idx);
340 ++enc->shareable_idx;
341 encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE);
342 }
343 }
344 368
345 svtype svt = SvTYPE (sv); 369 svtype svt = SvTYPE (sv);
346 370
347 if (ecb_expect_false (SvOBJECT (sv))) 371 if (ecb_expect_false (SvOBJECT (sv)))
348 { 372 {
… …
355 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 379 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
356 ? cbor_tagged_stash 380 ? cbor_tagged_stash
357 : gv_stashpv ("CBOR::XS::Tagged" , 1); 381 : gv_stashpv ("CBOR::XS::Tagged" , 1);
358 382
359 HV *stash = SvSTASH (sv); 383 HV *stash = SvSTASH (sv);
360 GV *method;
361 384
362 if (stash == boolean_stash) 385 if (stash == boolean_stash)
363 encode_ch (enc, SvIV (sv) ? 0xe0 | 21 : 0xe0 | 20); 386 {
387 encode_ch (enc, SvIV (sv) ? MAJOR_MISC | SIMPLE_TRUE : MAJOR_MISC | SIMPLE_FALSE);
388 return;
389 }
364 else if (stash == error_stash) 390 else if (stash == error_stash)
365 encode_ch (enc, 0xe0 | 23); 391 {
392 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
393 return;
394 }
366 else if (stash == tagged_stash) 395 else if (stash == tagged_stash)
367 { 396 {
368 if (svt != SVt_PVAV) 397 if (svt != SVt_PVAV)
369 croak ("encountered CBOR::XS::Tagged object that isn't an array"); 398 croak ("encountered CBOR::XS::Tagged object that isn't an array");
370 399
371 encode_uint (enc, 0xc0, SvUV (*av_fetch ((AV *)sv, 0, 1))); 400 encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1)));
372 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1)); 401 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1));
402
403 return;
404 }
405 }
406
407 if (ecb_expect_false (SvREFCNT (sv) > 1)
408 && ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING))
409 {
410 if (!enc->shareable)
411 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
412
413 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
414
415 if (SvOK (*svp))
373 } 416 {
417 encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF);
418 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
419 return;
420 }
421 else
422 {
423 sv_setuv (*svp, enc->shareable_idx);
424 ++enc->shareable_idx;
425 encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE);
426 }
427 }
428
429 if (ecb_expect_false (SvOBJECT (sv)))
430 {
431 HV *stash = SvSTASH (sv);
432 GV *method;
433
374 else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0))) 434 if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
375 { 435 {
376 dSP; 436 dSP;
377 437
378 ENTER; SAVETMPS; PUSHMARK (SP); 438 ENTER; SAVETMPS; PUSHMARK (SP);
379 // we re-bless the reference to get overload and other niceties right 439 // we re-bless the reference to get overload and other niceties right
… …
411 // catch this surprisingly common error 471 // catch this surprisingly common error
412 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv) 472 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)
413 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash)); 473 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));
414 474
415 encode_tag (enc, CBOR_TAG_PERL_OBJECT); 475 encode_tag (enc, CBOR_TAG_PERL_OBJECT);
416 encode_uint (enc, 0x80, count + 1); 476 encode_uint (enc, MAJOR_ARRAY, count + 1);
417 encode_str (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash)); 477 encode_strref (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
418 478
419 while (count) 479 while (count)
420 encode_sv (enc, SP[1 - count--]); 480 encode_sv (enc, SP[1 - count--]);
421 481
422 PUTBACK; 482 PUTBACK;
… …
443{ 503{
444 double nv = SvNVX (sv); 504 double nv = SvNVX (sv);
445 505
446 need (enc, 9); 506 need (enc, 9);
447 507
448 if (ecb_expect_false (nv == (U32)nv)) 508 if (ecb_expect_false (nv == (NV)(U32)nv))
449 encode_uint (enc, 0x00, (U32)nv); 509 encode_uint (enc, MAJOR_POS_INT, (U32)nv);
450 //TODO: maybe I32? 510 //TODO: maybe I32?
451 else if (ecb_expect_false (nv == (float)nv)) 511 else if (ecb_expect_false (nv == (float)nv))
452 { 512 {
453 uint32_t fp = ecb_float_to_binary32 (nv); 513 uint32_t fp = ecb_float_to_binary32 (nv);
454 514
455 *enc->cur++ = 0xe0 | 26; 515 *enc->cur++ = MAJOR_MISC | MISC_FLOAT32;
456 516
457 if (!ecb_big_endian ()) 517 if (!ecb_big_endian ())
458 fp = ecb_bswap32 (fp); 518 fp = ecb_bswap32 (fp);
459 519
460 memcpy (enc->cur, &fp, 4); 520 memcpy (enc->cur, &fp, 4);
… …
462 } 522 }
463 else 523 else
464 { 524 {
465 uint64_t fp = ecb_double_to_binary64 (nv); 525 uint64_t fp = ecb_double_to_binary64 (nv);
466 526
467 *enc->cur++ = 0xe0 | 27; 527 *enc->cur++ = MAJOR_MISC | MISC_FLOAT64;
468 528
469 if (!ecb_big_endian ()) 529 if (!ecb_big_endian ())
470 fp = ecb_bswap64 (fp); 530 fp = ecb_bswap64 (fp);
471 531
472 memcpy (enc->cur, &fp, 8); 532 memcpy (enc->cur, &fp, 8);
… …
481 541
482 if (SvPOKp (sv)) 542 if (SvPOKp (sv))
483 { 543 {
484 STRLEN len; 544 STRLEN len;
485 char *str = SvPV (sv, len); 545 char *str = SvPV (sv, len);
486 encode_str (enc, SvUTF8 (sv), str, len); 546 encode_strref (enc, SvUTF8 (sv), str, len);
487 } 547 }
488 else if (SvNOKp (sv)) 548 else if (SvNOKp (sv))
489 encode_nv (enc, sv); 549 encode_nv (enc, sv);
490 else if (SvIOKp (sv)) 550 else if (SvIOKp (sv))
491 { 551 {
492 if (SvIsUV (sv)) 552 if (SvIsUV (sv))
493 encode_uint (enc, 0x00, SvUVX (sv)); 553 encode_uint (enc, MAJOR_POS_INT, SvUVX (sv));
494 else if (SvIVX (sv) >= 0) 554 else if (SvIVX (sv) >= 0)
495 encode_uint (enc, 0x00, SvIVX (sv)); 555 encode_uint (enc, MAJOR_POS_INT, SvIVX (sv));
496 else 556 else
497 encode_uint (enc, 0x20, -(SvIVX (sv) + 1)); 557 encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1));
498 } 558 }
499 else if (SvROK (sv)) 559 else if (SvROK (sv))
500 encode_rv (enc, SvRV (sv)); 560 encode_rv (enc, SvRV (sv));
501 else if (!SvOK (sv)) 561 else if (!SvOK (sv))
502 encode_ch (enc, 0xe0 | 22); 562 encode_ch (enc, MAJOR_MISC | SIMPLE_NULL);
503 else if (enc->cbor.flags & F_ALLOW_UNKNOWN) 563 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
504 encode_ch (enc, 0xe0 | 23); 564 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
505 else 565 else
506 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data", 566 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data",
507 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv)); 567 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
508} 568}
509 569
… …
517 enc.cur = SvPVX (enc.sv); 577 enc.cur = SvPVX (enc.sv);
518 enc.end = SvEND (enc.sv); 578 enc.end = SvEND (enc.sv);
519 579
520 SvPOK_only (enc.sv); 580 SvPOK_only (enc.sv);
521 581
522 if (cbor->flags & F_ALLOW_STRINGREF) 582 if (cbor->flags & F_PACK_STRINGS)
523 { 583 {
524 encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE); 584 encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE);
525 enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ()); 585 enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ());
526 enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ()); 586 enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ());
527 } 587 }
… …
562#define DEC_DEC_DEPTH --dec->depth 622#define DEC_DEC_DEPTH --dec->depth
563 623
564static UV 624static UV
565decode_uint (dec_t *dec) 625decode_uint (dec_t *dec)
566{ 626{
567 switch (*dec->cur & 31) 627 U8 m = *dec->cur & MINOR_MASK;
568 { 628 ++dec->cur;
569 case 0: case 1: case 2: case 3: case 4: case 5: case 6: case 7:
570 case 8: case 9: case 10: case 11: case 12: case 13: case 14: case 15:
571 case 16: case 17: case 18: case 19: case 20: case 21: case 22: case 23:
572 return *dec->cur++ & 31;
573 629
574 case 24: 630 if (ecb_expect_true (m < LENGTH_EXT1))
631 return m;
632 else if (ecb_expect_true (m == LENGTH_EXT1))
633 {
575 WANT (2); 634 WANT (1);
576 dec->cur += 2; 635 dec->cur += 1;
577 return dec->cur[-1]; 636 return dec->cur[-1];
578 637 }
579 case 25: 638 else if (ecb_expect_true (m == LENGTH_EXT2))
639 {
580 WANT (3); 640 WANT (2);
581 dec->cur += 3; 641 dec->cur += 2;
582 return (((UV)dec->cur[-2]) << 8) 642 return (((UV)dec->cur[-2]) << 8)
583 | ((UV)dec->cur[-1]); 643 | ((UV)dec->cur[-1]);
584 644 }
585 case 26: 645 else if (ecb_expect_true (m == LENGTH_EXT4))
646 {
586 WANT (5); 647 WANT (4);
587 dec->cur += 5; 648 dec->cur += 4;
588 return (((UV)dec->cur[-4]) << 24) 649 return (((UV)dec->cur[-4]) << 24)
589 | (((UV)dec->cur[-3]) << 16) 650 | (((UV)dec->cur[-3]) << 16)
590 | (((UV)dec->cur[-2]) << 8) 651 | (((UV)dec->cur[-2]) << 8)
591 | ((UV)dec->cur[-1]); 652 | ((UV)dec->cur[-1]);
592 653 }
593 case 27: 654 else if (ecb_expect_true (m == LENGTH_EXT8))
655 {
594 WANT (9); 656 WANT (8);
595 dec->cur += 9; 657 dec->cur += 8;
658
659 return
660#if UVSIZE < 8
661 0
662#else
596 return (((UV)dec->cur[-8]) << 56) 663 (((UV)dec->cur[-8]) << 56)
597 | (((UV)dec->cur[-7]) << 48) 664 | (((UV)dec->cur[-7]) << 48)
598 | (((UV)dec->cur[-6]) << 40) 665 | (((UV)dec->cur[-6]) << 40)
599 | (((UV)dec->cur[-5]) << 32) 666 | (((UV)dec->cur[-5]) << 32)
667#endif
600 | (((UV)dec->cur[-4]) << 24) 668 | (((UV)dec->cur[-4]) << 24)
601 | (((UV)dec->cur[-3]) << 16) 669 | (((UV)dec->cur[-3]) << 16)
602 | (((UV)dec->cur[-2]) << 8) 670 | (((UV)dec->cur[-2]) << 8)
603 | ((UV)dec->cur[-1]); 671 | ((UV)dec->cur[-1]);
604 672 }
605 default: 673 else
606 ERR ("corrupted CBOR data (unsupported integer minor encoding)"); 674 ERR ("corrupted CBOR data (unsupported integer minor encoding)");
607 }
608 675
609fail: 676fail:
610 return 0; 677 return 0;
611} 678}
612 679
… …
617{ 684{
618 AV *av = newAV (); 685 AV *av = newAV ();
619 686
620 DEC_INC_DEPTH; 687 DEC_INC_DEPTH;
621 688
622 if ((*dec->cur & 31) == 31) 689 if (*dec->cur == (MAJOR_ARRAY | MINOR_INDEF))
623 { 690 {
624 ++dec->cur; 691 ++dec->cur;
625 692
626 for (;;) 693 for (;;)
627 { 694 {
628 WANT (1); 695 WANT (1);
629 696
630 if (*dec->cur == (0xe0 | 31)) 697 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
631 { 698 {
632 ++dec->cur; 699 ++dec->cur;
633 break; 700 break;
634 } 701 }
635 702
… …
638 } 705 }
639 else 706 else
640 { 707 {
641 int i, len = decode_uint (dec); 708 int i, len = decode_uint (dec);
642 709
710 WANT (len); // complexity check for av_fill - need at least one byte per value, do not allow supersize arrays
643 av_fill (av, len - 1); 711 av_fill (av, len - 1);
644 712
645 for (i = 0; i < len; ++i) 713 for (i = 0; i < len; ++i)
646 AvARRAY (av)[i] = decode_sv (dec); 714 AvARRAY (av)[i] = decode_sv (dec);
647 } 715 }
… …
660{ 728{
661 // for speed reasons, we specialcase single-string 729 // for speed reasons, we specialcase single-string
662 // byte or utf-8 strings as keys, but only when !stringref 730 // byte or utf-8 strings as keys, but only when !stringref
663 731
664 if (ecb_expect_true (!dec->stringref)) 732 if (ecb_expect_true (!dec->stringref))
665 if (*dec->cur >= 0x40 && *dec->cur <= 0x40 + 27) 733 if (ecb_expect_true ((*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8))
666 { 734 {
667 I32 len = decode_uint (dec); 735 I32 len = decode_uint (dec);
668 char *key = (char *)dec->cur; 736 char *key = (char *)dec->cur;
669 737
670 dec->cur += len; 738 dec->cur += len;
… …
674 742
675 hv_store (hv, key, len, decode_sv (dec), 0); 743 hv_store (hv, key, len, decode_sv (dec), 0);
676 744
677 return; 745 return;
678 } 746 }
679 else if (*dec->cur >= 0x60 && *dec->cur <= 0x60 + 27) 747 else if (ecb_expect_true ((*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8))
680 { 748 {
681 I32 len = decode_uint (dec); 749 I32 len = decode_uint (dec);
682 char *key = (char *)dec->cur; 750 char *key = (char *)dec->cur;
683 751
684 dec->cur += len; 752 dec->cur += len;
… …
703{ 771{
704 HV *hv = newHV (); 772 HV *hv = newHV ();
705 773
706 DEC_INC_DEPTH; 774 DEC_INC_DEPTH;
707 775
708 if ((*dec->cur & 31) == 31) 776 if (*dec->cur == (MAJOR_MAP | MINOR_INDEF))
709 { 777 {
710 ++dec->cur; 778 ++dec->cur;
711 779
712 for (;;) 780 for (;;)
713 { 781 {
714 WANT (1); 782 WANT (1);
715 783
716 if (*dec->cur == (0xe0 | 31)) 784 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
717 { 785 {
718 ++dec->cur; 786 ++dec->cur;
719 break; 787 break;
720 } 788 }
721 789
… …
742static SV * 810static SV *
743decode_str (dec_t *dec, int utf8) 811decode_str (dec_t *dec, int utf8)
744{ 812{
745 SV *sv = 0; 813 SV *sv = 0;
746 814
747 if ((*dec->cur & 31) == 31) 815 if ((*dec->cur & MINOR_MASK) == MINOR_INDEF)
748 { 816 {
817 // indefinite length strings
749 ++dec->cur; 818 ++dec->cur;
750 819
820 U8 major = *dec->cur & MAJOR_MISC;
821
751 sv = newSVpvn ("", 0); 822 sv = newSVpvn ("", 0);
752 823
753 // not very fast, and certainly not robust against illegal input
754 for (;;) 824 for (;;)
755 { 825 {
756 WANT (1); 826 WANT (1);
757 827
758 if (*dec->cur == (0xe0 | 31)) 828 if ((*dec->cur - major) > LENGTH_EXT8)
829 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
759 { 830 {
760 ++dec->cur; 831 ++dec->cur;
761 break; 832 break;
762 } 833 }
834 else
835 ERR ("corrupted CBOR data (invalid chunks in indefinite length string)");
763 836
764 sv_catsv (sv, decode_sv (dec)); 837 STRLEN len = decode_uint (dec);
838
839 WANT (len);
840 sv_catpvn (sv, dec->cur, len);
841 dec->cur += len;
765 } 842 }
766 } 843 }
767 else 844 else
768 { 845 {
769 STRLEN len = decode_uint (dec); 846 STRLEN len = decode_uint (dec);
… …
818 } 895 }
819 break; 896 break;
820 897
821 case CBOR_TAG_STRINGREF: 898 case CBOR_TAG_STRINGREF:
822 { 899 {
823 if ((*dec->cur >> 5) != 0) 900 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
824 ERR ("corrupted CBOR data (stringref index not an unsigned integer)"); 901 ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
825 902
826 UV idx = decode_uint (dec); 903 UV idx = decode_uint (dec);
827 904
828 if (!dec->stringref || (int)idx > AvFILLp (dec->stringref)) 905 if (!dec->stringref || (int)idx > AvFILLp (dec->stringref))
… …
846 } 923 }
847 break; 924 break;
848 925
849 case CBOR_TAG_VALUE_SHAREDREF: 926 case CBOR_TAG_VALUE_SHAREDREF:
850 { 927 {
851 if ((*dec->cur >> 5) != 0) 928 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
852 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)"); 929 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
853 930
854 UV idx = decode_uint (dec); 931 UV idx = decode_uint (dec);
855 932
856 if (!dec->shareable || (int)idx > AvFILLp (dec->shareable)) 933 if (!dec->shareable || (int)idx > AvFILLp (dec->shareable))
… …
965static SV * 1042static SV *
966decode_sv (dec_t *dec) 1043decode_sv (dec_t *dec)
967{ 1044{
968 WANT (1); 1045 WANT (1);
969 1046
970 switch (*dec->cur >> 5) 1047 switch (*dec->cur >> MAJOR_SHIFT)
971 { 1048 {
972 case 0: // unsigned int 1049 case MAJOR_POS_INT >> MAJOR_SHIFT: return newSVuv (decode_uint (dec));
973 return newSVuv (decode_uint (dec)); 1050 case MAJOR_NEG_INT >> MAJOR_SHIFT: return newSViv (-1 - (IV)decode_uint (dec));
974 case 1: // negative int 1051 case MAJOR_BYTES >> MAJOR_SHIFT: return decode_str (dec, 0);
975 return newSViv (-1 - (IV)decode_uint (dec)); 1052 case MAJOR_TEXT >> MAJOR_SHIFT: return decode_str (dec, 1);
976 case 2: // octet string 1053 case MAJOR_ARRAY >> MAJOR_SHIFT: return decode_av (dec);
977 return decode_str (dec, 0); 1054 case MAJOR_MAP >> MAJOR_SHIFT: return decode_hv (dec);
978 case 3: // utf-8 string 1055 case MAJOR_TAG >> MAJOR_SHIFT: return decode_tagged (dec);
979 return decode_str (dec, 1); 1056
980 case 4: // array 1057 case MAJOR_MISC >> MAJOR_SHIFT:
981 return decode_av (dec);
982 case 5: // map
983 return decode_hv (dec);
984 case 6: // tag
985 return decode_tagged (dec);
986 case 7: // misc
987 switch (*dec->cur++ & 31) 1058 switch (*dec->cur++ & MINOR_MASK)
988 { 1059 {
989 case 20: 1060 case SIMPLE_FALSE:
990#if CBOR_SLOW 1061#if CBOR_SLOW
991 types_false = get_bool ("Types::Serialiser::false"); 1062 types_false = get_bool ("Types::Serialiser::false");
992#endif 1063#endif
993 return newSVsv (types_false); 1064 return newSVsv (types_false);
994 case 21: 1065 case SIMPLE_TRUE:
995#if CBOR_SLOW 1066#if CBOR_SLOW
996 types_true = get_bool ("Types::Serialiser::true"); 1067 types_true = get_bool ("Types::Serialiser::true");
997#endif 1068#endif
998 return newSVsv (types_true); 1069 return newSVsv (types_true);
999 case 22: 1070 case SIMPLE_NULL:
1000 return newSVsv (&PL_sv_undef); 1071 return newSVsv (&PL_sv_undef);
1001 case 23: 1072 case SIMPLE_UNDEF:
1002#if CBOR_SLOW 1073#if CBOR_SLOW
1003 types_error = get_bool ("Types::Serialiser::error"); 1074 types_error = get_bool ("Types::Serialiser::error");
1004#endif 1075#endif
1005 return newSVsv (types_error); 1076 return newSVsv (types_error);
1006 1077
1007 case 25: 1078 case MISC_FLOAT16:
1008 { 1079 {
1009 WANT (2); 1080 WANT (2);
1010 1081
1011 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1]; 1082 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1];
1012 dec->cur += 2; 1083 dec->cur += 2;
1013 1084
1014 return newSVnv (ecb_binary16_to_float (fp)); 1085 return newSVnv (ecb_binary16_to_float (fp));
1015 } 1086 }
1016 1087
1017 case 26: 1088 case MISC_FLOAT32:
1018 { 1089 {
1019 uint32_t fp; 1090 uint32_t fp;
1020 WANT (4); 1091 WANT (4);
1021 memcpy (&fp, dec->cur, 4); 1092 memcpy (&fp, dec->cur, 4);
1022 dec->cur += 4; 1093 dec->cur += 4;
… …
1025 fp = ecb_bswap32 (fp); 1096 fp = ecb_bswap32 (fp);
1026 1097
1027 return newSVnv (ecb_binary32_to_float (fp)); 1098 return newSVnv (ecb_binary32_to_float (fp));
1028 } 1099 }
1029 1100
1030 case 27: 1101 case MISC_FLOAT64:
1031 { 1102 {
1032 uint64_t fp; 1103 uint64_t fp;
1033 WANT (8); 1104 WANT (8);
1034 memcpy (&fp, dec->cur, 8); 1105 memcpy (&fp, dec->cur, 8);
1035 dec->cur += 8; 1106 dec->cur += 8;
… …
1038 fp = ecb_bswap64 (fp); 1109 fp = ecb_bswap64 (fp);
1039 1110
1040 return newSVnv (ecb_binary64_to_double (fp)); 1111 return newSVnv (ecb_binary64_to_double (fp));
1041 } 1112 }
1042 1113
1043 // 0..19 unassigned 1114 // 0..19 unassigned simple
1044 // 24 reserved + unassigned (reserved values are not encodable) 1115 // 24 reserved + unassigned (reserved values are not encodable)
1045 default: 1116 default:
1046 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)"); 1117 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)");
1047 } 1118 }
1048 1119
… …
1136void shrink (CBOR *self, int enable = 1) 1207void shrink (CBOR *self, int enable = 1)
1137 ALIAS: 1208 ALIAS:
1138 shrink = F_SHRINK 1209 shrink = F_SHRINK
1139 allow_unknown = F_ALLOW_UNKNOWN 1210 allow_unknown = F_ALLOW_UNKNOWN
1140 allow_sharing = F_ALLOW_SHARING 1211 allow_sharing = F_ALLOW_SHARING
1141 allow_stringref = F_ALLOW_STRINGREF 1212 pack_strings = F_PACK_STRINGS
1142 PPCODE: 1213 PPCODE:
1143{ 1214{
1144 if (enable) 1215 if (enable)
1145 self->flags |= ix; 1216 self->flags |= ix;
1146 else 1217 else
… …
1152void get_shrink (CBOR *self) 1223void get_shrink (CBOR *self)
1153 ALIAS: 1224 ALIAS:
1154 get_shrink = F_SHRINK 1225 get_shrink = F_SHRINK
1155 get_allow_unknown = F_ALLOW_UNKNOWN 1226 get_allow_unknown = F_ALLOW_UNKNOWN
1156 get_allow_sharing = F_ALLOW_SHARING 1227 get_allow_sharing = F_ALLOW_SHARING
1157 get_allow_stringref = F_ALLOW_STRINGREF 1228 get_pack_strings = F_PACK_STRINGS
1158 PPCODE: 1229 PPCODE:
1159 XPUSHs (boolSV (self->flags & ix)); 1230 XPUSHs (boolSV (self->flags & ix));
1160 1231
1161void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) 1232void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
1162 PPCODE: 1233 PPCODE:
… …
1218 cbor_free (self); 1289 cbor_free (self);
1219 1290
1220PROTOTYPES: ENABLE 1291PROTOTYPES: ENABLE
1221 1292
1222void encode_cbor (SV *scalar) 1293void encode_cbor (SV *scalar)
1294 ALIAS:
1295 encode_cbor = 0
1296 encode_cbor_sharing = F_ALLOW_SHARING
1223 PPCODE: 1297 PPCODE:
1224{ 1298{
1225 CBOR cbor; 1299 CBOR cbor;
1226 cbor_init (&cbor); 1300 cbor_init (&cbor);
1301 cbor.flags |= ix;
1227 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN; 1302 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;
1228 XPUSHs (scalar); 1303 XPUSHs (scalar);
1229} 1304}
1230 1305
1231void decode_cbor (SV *cborstr) 1306void decode_cbor (SV *cborstr)

Diff Legend

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