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 1.32 by root, Thu Nov 28 12:08:07 2013 UTC vs.
Revision 1.36 by root, Sat Nov 30 17:37:45 2013 UTC

23#endif 23#endif
24#ifndef SvREFCNT_dec_NN 24#ifndef SvREFCNT_dec_NN
25# define SvREFCNT_dec_NN(sv) SvREFCNT_dec (sv) 25# define SvREFCNT_dec_NN(sv) SvREFCNT_dec (sv)
26#endif 26#endif
27 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};
64
28// known tags 65// known tags
29enum cbor_tag 66enum cbor_tag
30{ 67{
31 // extensions 68 // extensions
32 CBOR_TAG_STRINGREF = 25, // http://cbor.schmorp.de/stringref 69 CBOR_TAG_STRINGREF = 25, // http://cbor.schmorp.de/stringref
33 CBOR_TAG_PERL_OBJECT = 26, // http://cbor.schmorp.de/perl-object 70 CBOR_TAG_PERL_OBJECT = 26, // http://cbor.schmorp.de/perl-object
34 CBOR_TAG_GENERIC_OBJECT = 27, // http://cbor.schmorp.de/generic-object 71 CBOR_TAG_GENERIC_OBJECT = 27, // http://cbor.schmorp.de/generic-object
35 CBOR_TAG_VALUE_SHAREABLE = 28, // http://cbor.schmorp.de/value-sharing 72 CBOR_TAG_VALUE_SHAREABLE = 28, // http://cbor.schmorp.de/value-sharing
36 CBOR_TAG_VALUE_SHAREDREF = 29, // http://cbor.schmorp.de/value-sharing 73 CBOR_TAG_VALUE_SHAREDREF = 29, // http://cbor.schmorp.de/value-sharing
37 CBOR_TAG_STRINGREF_NAMESPACE = 256, // http://cbor.schmorp.de/stringref 74 CBOR_TAG_STRINGREF_NAMESPACE = 256, // http://cbor.schmorp.de/stringref
38 CBOR_TAG_INDIRECTION = 22098, // http://cbor.schmorp.de/indirection 75 CBOR_TAG_INDIRECTION = 22098, // http://cbor.schmorp.de/indirection
39 76
40 // rfc7049 77 // rfc7049
41 CBOR_TAG_DATETIME = 0, // rfc4287, utf-8 78 CBOR_TAG_DATETIME = 0, // rfc4287, utf-8
42 CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any 79 CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any
43 CBOR_TAG_POS_BIGNUM = 2, // byte string 80 CBOR_TAG_POS_BIGNUM = 2, // byte string
44 CBOR_TAG_NEG_BIGNUM = 3, // byte string 81 CBOR_TAG_NEG_BIGNUM = 3, // byte string
45 CBOR_TAG_DECIMAL = 4, // decimal fraction, array 82 CBOR_TAG_DECIMAL = 4, // decimal fraction, array
46 CBOR_TAG_BIGFLOAT = 5, // array 83 CBOR_TAG_BIGFLOAT = 5, // array
47 84
48 CBOR_TAG_CONV_B64U = 21, // base64url, any 85 CBOR_TAG_CONV_B64U = 21, // base64url, any
49 CBOR_TAG_CONV_B64 = 22, // base64, any 86 CBOR_TAG_CONV_B64 = 22, // base64, any
50 CBOR_TAG_CONV_HEX = 23, // base16, any 87 CBOR_TAG_CONV_HEX = 23, // base16, any
51 CBOR_TAG_CBOR = 24, // embedded cbor, byte string 88 CBOR_TAG_CBOR = 24, // embedded cbor, byte string
52 89
53 CBOR_TAG_URI = 32, // URI rfc3986, utf-8 90 CBOR_TAG_URI = 32, // URI rfc3986, utf-8
54 CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8 91 CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8
55 CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8 92 CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8
56 CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8 93 CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8
57 CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8 94 CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8
58 95
59 CBOR_TAG_MAGIC = 55799 // self-describe cbor 96 CBOR_TAG_MAGIC = 55799, // self-describe cbor
60}; 97};
61 98
62#define F_SHRINK 0x00000001UL 99#define F_SHRINK 0x00000001UL
63#define F_ALLOW_UNKNOWN 0x00000002UL 100#define F_ALLOW_UNKNOWN 0x00000002UL
64#define F_ALLOW_SHARING 0x00000004UL 101#define F_ALLOW_SHARING 0x00000004UL
189static void 226static void
190encode_uint (enc_t *enc, int major, UV len) 227encode_uint (enc_t *enc, int major, UV len)
191{ 228{
192 need (enc, 9); 229 need (enc, 9);
193 230
194 if (ecb_expect_true (len < 24)) 231 if (ecb_expect_true (len < LENGTH_EXT1))
195 *enc->cur++ = major | len; 232 *enc->cur++ = major | len;
196 else if (ecb_expect_true (len <= 0xff)) 233 else if (ecb_expect_true (len <= 0xffU))
197 { 234 {
198 *enc->cur++ = major | 24; 235 *enc->cur++ = major | LENGTH_EXT1;
199 *enc->cur++ = len; 236 *enc->cur++ = len;
200 } 237 }
201 else if (len <= 0xffff) 238 else if (len <= 0xffffU)
202 { 239 {
203 *enc->cur++ = major | 25; 240 *enc->cur++ = major | LENGTH_EXT2;
204 *enc->cur++ = len >> 8; 241 *enc->cur++ = len >> 8;
205 *enc->cur++ = len; 242 *enc->cur++ = len;
206 } 243 }
207 else if (len <= 0xffffffff) 244 else if (len <= 0xffffffffU)
208 { 245 {
209 *enc->cur++ = major | 26; 246 *enc->cur++ = major | LENGTH_EXT4;
210 *enc->cur++ = len >> 24; 247 *enc->cur++ = len >> 24;
211 *enc->cur++ = len >> 16; 248 *enc->cur++ = len >> 16;
212 *enc->cur++ = len >> 8; 249 *enc->cur++ = len >> 8;
213 *enc->cur++ = len; 250 *enc->cur++ = len;
214 } 251 }
215 else 252 else
216 { 253 {
217 *enc->cur++ = major | 27; 254 *enc->cur++ = major | LENGTH_EXT8;
218 *enc->cur++ = len >> 56; 255 *enc->cur++ = len >> 56;
219 *enc->cur++ = len >> 48; 256 *enc->cur++ = len >> 48;
220 *enc->cur++ = len >> 40; 257 *enc->cur++ = len >> 40;
221 *enc->cur++ = len >> 32; 258 *enc->cur++ = len >> 32;
222 *enc->cur++ = len >> 24; 259 *enc->cur++ = len >> 24;
227} 264}
228 265
229ecb_inline void 266ecb_inline void
230encode_tag (enc_t *enc, UV tag) 267encode_tag (enc_t *enc, UV tag)
231{ 268{
232 encode_uint (enc, 0xc0, tag); 269 encode_uint (enc, MAJOR_TAG, tag);
233} 270}
234 271
235ecb_inline void 272ecb_inline void
236encode_str (enc_t *enc, int utf8, char *str, STRLEN len) 273encode_str (enc_t *enc, int utf8, char *str, STRLEN len)
237{ 274{
238 encode_uint (enc, utf8 ? 0x60 : 0x40, len); 275 encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len);
239 need (enc, len); 276 need (enc, len);
240 memcpy (enc->cur, str, len); 277 memcpy (enc->cur, str, len);
241 enc->cur += len; 278 enc->cur += len;
242} 279}
243 280
250 287
251 if (SvOK (*svp)) 288 if (SvOK (*svp))
252 { 289 {
253 // already registered, use stringref 290 // already registered, use stringref
254 encode_tag (enc, CBOR_TAG_STRINGREF); 291 encode_tag (enc, CBOR_TAG_STRINGREF);
255 encode_uint (enc, 0x00, SvUV (*svp)); 292 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
256 return; 293 return;
257 } 294 }
258 else if (len >= minimum_string_length (enc->stringref_idx)) 295 else if (len >= minimum_string_length (enc->stringref_idx))
259 { 296 {
260 // register only 297 // register only
276 if (enc->depth >= enc->cbor.max_depth) 313 if (enc->depth >= enc->cbor.max_depth)
277 croak (ERR_NESTING_EXCEEDED); 314 croak (ERR_NESTING_EXCEEDED);
278 315
279 ++enc->depth; 316 ++enc->depth;
280 317
281 encode_uint (enc, 0x80, len + 1); 318 encode_uint (enc, MAJOR_ARRAY, len + 1);
282 319
283 for (i = 0; i <= len; ++i) 320 for (i = 0; i <= len; ++i)
284 { 321 {
285 SV **svp = av_fetch (av, i, 0); 322 SV **svp = av_fetch (av, i, 0);
286 encode_sv (enc, svp ? *svp : &PL_sv_undef); 323 encode_sv (enc, svp ? *svp : &PL_sv_undef);
301 338
302 int pairs = hv_iterinit (hv); 339 int pairs = hv_iterinit (hv);
303 int mg = SvMAGICAL (hv); 340 int mg = SvMAGICAL (hv);
304 341
305 if (mg) 342 if (mg)
306 encode_ch (enc, 0xa0 | 31); 343 encode_ch (enc, MAJOR_MAP | MINOR_INDEF);
307 else 344 else
308 encode_uint (enc, 0xa0, pairs); 345 encode_uint (enc, MAJOR_MAP, pairs);
309 346
310 while ((he = hv_iternext (hv))) 347 while ((he = hv_iternext (hv)))
311 { 348 {
312 if (HeKLEN (he) == HEf_SVKEY) 349 if (HeKLEN (he) == HEf_SVKEY)
313 encode_sv (enc, HeSVKEY (he)); 350 encode_sv (enc, HeSVKEY (he));
316 353
317 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));
318 } 355 }
319 356
320 if (mg) 357 if (mg)
321 encode_ch (enc, 0xe0 | 31); 358 encode_ch (enc, MAJOR_MISC | MINOR_INDEF);
322 359
323 --enc->depth; 360 --enc->depth;
324} 361}
325 362
326// encode objects, arrays and special \0=false and \1=true values. 363// encode objects, arrays and special \0=false and \1=true values.
327static void 364static void
328encode_rv (enc_t *enc, SV *sv) 365encode_rv (enc_t *enc, SV *sv)
329{ 366{
330 SvGETMAGIC (sv); 367 SvGETMAGIC (sv);
331
332 if (ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING)
333 && ecb_expect_false (SvREFCNT (sv) > 1))
334 {
335 if (!enc->shareable)
336 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
337
338 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
339
340 if (SvOK (*svp))
341 {
342 encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF);
343 encode_uint (enc, 0x00, SvUV (*svp));
344 return;
345 }
346 else
347 {
348 sv_setuv (*svp, enc->shareable_idx);
349 ++enc->shareable_idx;
350 encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE);
351 }
352 }
353 368
354 svtype svt = SvTYPE (sv); 369 svtype svt = SvTYPE (sv);
355 370
356 if (ecb_expect_false (SvOBJECT (sv))) 371 if (ecb_expect_false (SvOBJECT (sv)))
357 { 372 {
364 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 379 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
365 ? cbor_tagged_stash 380 ? cbor_tagged_stash
366 : gv_stashpv ("CBOR::XS::Tagged" , 1); 381 : gv_stashpv ("CBOR::XS::Tagged" , 1);
367 382
368 HV *stash = SvSTASH (sv); 383 HV *stash = SvSTASH (sv);
369 GV *method;
370 384
371 if (stash == boolean_stash) 385 if (stash == boolean_stash)
372 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 }
373 else if (stash == error_stash) 390 else if (stash == error_stash)
374 encode_ch (enc, 0xe0 | 23); 391 {
392 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
393 return;
394 }
375 else if (stash == tagged_stash) 395 else if (stash == tagged_stash)
376 { 396 {
377 if (svt != SVt_PVAV) 397 if (svt != SVt_PVAV)
378 croak ("encountered CBOR::XS::Tagged object that isn't an array"); 398 croak ("encountered CBOR::XS::Tagged object that isn't an array");
379 399
380 encode_uint (enc, 0xc0, SvUV (*av_fetch ((AV *)sv, 0, 1))); 400 encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1)));
381 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))
382 } 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
383 else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0))) 434 if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
384 { 435 {
385 dSP; 436 dSP;
386 437
387 ENTER; SAVETMPS; PUSHMARK (SP); 438 ENTER; SAVETMPS; PUSHMARK (SP);
388 // 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
420 // catch this surprisingly common error 471 // catch this surprisingly common error
421 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv) 472 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)
422 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash)); 473 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));
423 474
424 encode_tag (enc, CBOR_TAG_PERL_OBJECT); 475 encode_tag (enc, CBOR_TAG_PERL_OBJECT);
425 encode_uint (enc, 0x80, count + 1); 476 encode_uint (enc, MAJOR_ARRAY, count + 1);
426 encode_strref (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash)); 477 encode_strref (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
427 478
428 while (count) 479 while (count)
429 encode_sv (enc, SP[1 - count--]); 480 encode_sv (enc, SP[1 - count--]);
430 481
452{ 503{
453 double nv = SvNVX (sv); 504 double nv = SvNVX (sv);
454 505
455 need (enc, 9); 506 need (enc, 9);
456 507
457 if (ecb_expect_false (nv == (U32)nv)) 508 if (ecb_expect_false (nv == (NV)(U32)nv))
458 encode_uint (enc, 0x00, (U32)nv); 509 encode_uint (enc, MAJOR_POS_INT, (U32)nv);
459 //TODO: maybe I32? 510 //TODO: maybe I32?
460 else if (ecb_expect_false (nv == (float)nv)) 511 else if (ecb_expect_false (nv == (float)nv))
461 { 512 {
462 uint32_t fp = ecb_float_to_binary32 (nv); 513 uint32_t fp = ecb_float_to_binary32 (nv);
463 514
464 *enc->cur++ = 0xe0 | 26; 515 *enc->cur++ = MAJOR_MISC | MISC_FLOAT32;
465 516
466 if (!ecb_big_endian ()) 517 if (!ecb_big_endian ())
467 fp = ecb_bswap32 (fp); 518 fp = ecb_bswap32 (fp);
468 519
469 memcpy (enc->cur, &fp, 4); 520 memcpy (enc->cur, &fp, 4);
471 } 522 }
472 else 523 else
473 { 524 {
474 uint64_t fp = ecb_double_to_binary64 (nv); 525 uint64_t fp = ecb_double_to_binary64 (nv);
475 526
476 *enc->cur++ = 0xe0 | 27; 527 *enc->cur++ = MAJOR_MISC | MISC_FLOAT64;
477 528
478 if (!ecb_big_endian ()) 529 if (!ecb_big_endian ())
479 fp = ecb_bswap64 (fp); 530 fp = ecb_bswap64 (fp);
480 531
481 memcpy (enc->cur, &fp, 8); 532 memcpy (enc->cur, &fp, 8);
497 else if (SvNOKp (sv)) 548 else if (SvNOKp (sv))
498 encode_nv (enc, sv); 549 encode_nv (enc, sv);
499 else if (SvIOKp (sv)) 550 else if (SvIOKp (sv))
500 { 551 {
501 if (SvIsUV (sv)) 552 if (SvIsUV (sv))
502 encode_uint (enc, 0x00, SvUVX (sv)); 553 encode_uint (enc, MAJOR_POS_INT, SvUVX (sv));
503 else if (SvIVX (sv) >= 0) 554 else if (SvIVX (sv) >= 0)
504 encode_uint (enc, 0x00, SvIVX (sv)); 555 encode_uint (enc, MAJOR_POS_INT, SvIVX (sv));
505 else 556 else
506 encode_uint (enc, 0x20, -(SvIVX (sv) + 1)); 557 encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1));
507 } 558 }
508 else if (SvROK (sv)) 559 else if (SvROK (sv))
509 encode_rv (enc, SvRV (sv)); 560 encode_rv (enc, SvRV (sv));
510 else if (!SvOK (sv)) 561 else if (!SvOK (sv))
511 encode_ch (enc, 0xe0 | 22); 562 encode_ch (enc, MAJOR_MISC | SIMPLE_NULL);
512 else if (enc->cbor.flags & F_ALLOW_UNKNOWN) 563 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
513 encode_ch (enc, 0xe0 | 23); 564 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
514 else 565 else
515 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",
516 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv)); 567 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
517} 568}
518 569
571#define DEC_DEC_DEPTH --dec->depth 622#define DEC_DEC_DEPTH --dec->depth
572 623
573static UV 624static UV
574decode_uint (dec_t *dec) 625decode_uint (dec_t *dec)
575{ 626{
576 switch (*dec->cur & 31) 627 U8 m = *dec->cur & MINOR_MASK;
577 { 628 ++dec->cur;
578 case 0: case 1: case 2: case 3: case 4: case 5: case 6: case 7:
579 case 8: case 9: case 10: case 11: case 12: case 13: case 14: case 15:
580 case 16: case 17: case 18: case 19: case 20: case 21: case 22: case 23:
581 return *dec->cur++ & 31;
582 629
583 case 24: 630 if (ecb_expect_true (m < LENGTH_EXT1))
631 return m;
632 else if (ecb_expect_true (m == LENGTH_EXT1))
633 {
584 WANT (2); 634 WANT (1);
585 dec->cur += 2; 635 dec->cur += 1;
586 return dec->cur[-1]; 636 return dec->cur[-1];
587 637 }
588 case 25: 638 else if (ecb_expect_true (m == LENGTH_EXT2))
639 {
589 WANT (3); 640 WANT (2);
590 dec->cur += 3; 641 dec->cur += 2;
591 return (((UV)dec->cur[-2]) << 8) 642 return (((UV)dec->cur[-2]) << 8)
592 | ((UV)dec->cur[-1]); 643 | ((UV)dec->cur[-1]);
593 644 }
594 case 26: 645 else if (ecb_expect_true (m == LENGTH_EXT4))
646 {
595 WANT (5); 647 WANT (4);
596 dec->cur += 5; 648 dec->cur += 4;
597 return (((UV)dec->cur[-4]) << 24) 649 return (((UV)dec->cur[-4]) << 24)
598 | (((UV)dec->cur[-3]) << 16) 650 | (((UV)dec->cur[-3]) << 16)
599 | (((UV)dec->cur[-2]) << 8) 651 | (((UV)dec->cur[-2]) << 8)
600 | ((UV)dec->cur[-1]); 652 | ((UV)dec->cur[-1]);
601 653 }
602 case 27: 654 else if (ecb_expect_true (m == LENGTH_EXT8))
655 {
603 WANT (9); 656 WANT (8);
604 dec->cur += 9; 657 dec->cur += 8;
658
659 return
660#if UVSIZE < 8
661 0
662#else
605 return (((UV)dec->cur[-8]) << 56) 663 (((UV)dec->cur[-8]) << 56)
606 | (((UV)dec->cur[-7]) << 48) 664 | (((UV)dec->cur[-7]) << 48)
607 | (((UV)dec->cur[-6]) << 40) 665 | (((UV)dec->cur[-6]) << 40)
608 | (((UV)dec->cur[-5]) << 32) 666 | (((UV)dec->cur[-5]) << 32)
667#endif
609 | (((UV)dec->cur[-4]) << 24) 668 | (((UV)dec->cur[-4]) << 24)
610 | (((UV)dec->cur[-3]) << 16) 669 | (((UV)dec->cur[-3]) << 16)
611 | (((UV)dec->cur[-2]) << 8) 670 | (((UV)dec->cur[-2]) << 8)
612 | ((UV)dec->cur[-1]); 671 | ((UV)dec->cur[-1]);
613 672 }
614 default: 673 else
615 ERR ("corrupted CBOR data (unsupported integer minor encoding)"); 674 ERR ("corrupted CBOR data (unsupported integer minor encoding)");
616 }
617 675
618fail: 676fail:
619 return 0; 677 return 0;
620} 678}
621 679
626{ 684{
627 AV *av = newAV (); 685 AV *av = newAV ();
628 686
629 DEC_INC_DEPTH; 687 DEC_INC_DEPTH;
630 688
631 if ((*dec->cur & 31) == 31) 689 if (*dec->cur == (MAJOR_ARRAY | MINOR_INDEF))
632 { 690 {
633 ++dec->cur; 691 ++dec->cur;
634 692
635 for (;;) 693 for (;;)
636 { 694 {
637 WANT (1); 695 WANT (1);
638 696
639 if (*dec->cur == (0xe0 | 31)) 697 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
640 { 698 {
641 ++dec->cur; 699 ++dec->cur;
642 break; 700 break;
643 } 701 }
644 702
647 } 705 }
648 else 706 else
649 { 707 {
650 int i, len = decode_uint (dec); 708 int i, len = decode_uint (dec);
651 709
710 WANT (len); // complexity check for av_fill - need at least one byte per value, do not allow supersize arrays
652 av_fill (av, len - 1); 711 av_fill (av, len - 1);
653 712
654 for (i = 0; i < len; ++i) 713 for (i = 0; i < len; ++i)
655 AvARRAY (av)[i] = decode_sv (dec); 714 AvARRAY (av)[i] = decode_sv (dec);
656 } 715 }
669{ 728{
670 // for speed reasons, we specialcase single-string 729 // for speed reasons, we specialcase single-string
671 // byte or utf-8 strings as keys, but only when !stringref 730 // byte or utf-8 strings as keys, but only when !stringref
672 731
673 if (ecb_expect_true (!dec->stringref)) 732 if (ecb_expect_true (!dec->stringref))
674 if (*dec->cur >= 0x40 && *dec->cur <= 0x40 + 27) 733 if (ecb_expect_true ((*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8))
675 { 734 {
676 I32 len = decode_uint (dec); 735 I32 len = decode_uint (dec);
677 char *key = (char *)dec->cur; 736 char *key = (char *)dec->cur;
678 737
679 dec->cur += len; 738 dec->cur += len;
683 742
684 hv_store (hv, key, len, decode_sv (dec), 0); 743 hv_store (hv, key, len, decode_sv (dec), 0);
685 744
686 return; 745 return;
687 } 746 }
688 else if (*dec->cur >= 0x60 && *dec->cur <= 0x60 + 27) 747 else if (ecb_expect_true ((*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8))
689 { 748 {
690 I32 len = decode_uint (dec); 749 I32 len = decode_uint (dec);
691 char *key = (char *)dec->cur; 750 char *key = (char *)dec->cur;
692 751
693 dec->cur += len; 752 dec->cur += len;
712{ 771{
713 HV *hv = newHV (); 772 HV *hv = newHV ();
714 773
715 DEC_INC_DEPTH; 774 DEC_INC_DEPTH;
716 775
717 if ((*dec->cur & 31) == 31) 776 if (*dec->cur == (MAJOR_MAP | MINOR_INDEF))
718 { 777 {
719 ++dec->cur; 778 ++dec->cur;
720 779
721 for (;;) 780 for (;;)
722 { 781 {
723 WANT (1); 782 WANT (1);
724 783
725 if (*dec->cur == (0xe0 | 31)) 784 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
726 { 785 {
727 ++dec->cur; 786 ++dec->cur;
728 break; 787 break;
729 } 788 }
730 789
751static SV * 810static SV *
752decode_str (dec_t *dec, int utf8) 811decode_str (dec_t *dec, int utf8)
753{ 812{
754 SV *sv = 0; 813 SV *sv = 0;
755 814
756 if ((*dec->cur & 31) == 31) 815 if ((*dec->cur & MINOR_MASK) == MINOR_INDEF)
757 { 816 {
817 // indefinite length strings
758 ++dec->cur; 818 ++dec->cur;
759 819
820 U8 major = *dec->cur & MAJOR_MISC;
821
760 sv = newSVpvn ("", 0); 822 sv = newSVpvn ("", 0);
761 823
762 // not very fast, and certainly not robust against illegal input
763 for (;;) 824 for (;;)
764 { 825 {
765 WANT (1); 826 WANT (1);
766 827
767 if (*dec->cur == (0xe0 | 31)) 828 if ((*dec->cur - major) > LENGTH_EXT8)
829 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
768 { 830 {
769 ++dec->cur; 831 ++dec->cur;
770 break; 832 break;
771 } 833 }
834 else
835 ERR ("corrupted CBOR data (invalid chunks in indefinite length string)");
772 836
773 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;
774 } 842 }
775 } 843 }
776 else 844 else
777 { 845 {
778 STRLEN len = decode_uint (dec); 846 STRLEN len = decode_uint (dec);
827 } 895 }
828 break; 896 break;
829 897
830 case CBOR_TAG_STRINGREF: 898 case CBOR_TAG_STRINGREF:
831 { 899 {
832 if ((*dec->cur >> 5) != 0) 900 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
833 ERR ("corrupted CBOR data (stringref index not an unsigned integer)"); 901 ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
834 902
835 UV idx = decode_uint (dec); 903 UV idx = decode_uint (dec);
836 904
837 if (!dec->stringref || (int)idx > AvFILLp (dec->stringref)) 905 if (!dec->stringref || (int)idx > AvFILLp (dec->stringref))
855 } 923 }
856 break; 924 break;
857 925
858 case CBOR_TAG_VALUE_SHAREDREF: 926 case CBOR_TAG_VALUE_SHAREDREF:
859 { 927 {
860 if ((*dec->cur >> 5) != 0) 928 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
861 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)"); 929 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
862 930
863 UV idx = decode_uint (dec); 931 UV idx = decode_uint (dec);
864 932
865 if (!dec->shareable || (int)idx > AvFILLp (dec->shareable)) 933 if (!dec->shareable || (int)idx > AvFILLp (dec->shareable))
974static SV * 1042static SV *
975decode_sv (dec_t *dec) 1043decode_sv (dec_t *dec)
976{ 1044{
977 WANT (1); 1045 WANT (1);
978 1046
979 switch (*dec->cur >> 5) 1047 switch (*dec->cur >> MAJOR_SHIFT)
980 { 1048 {
981 case 0: // unsigned int 1049 case MAJOR_POS_INT >> MAJOR_SHIFT: return newSVuv (decode_uint (dec));
982 return newSVuv (decode_uint (dec)); 1050 case MAJOR_NEG_INT >> MAJOR_SHIFT: return newSViv (-1 - (IV)decode_uint (dec));
983 case 1: // negative int 1051 case MAJOR_BYTES >> MAJOR_SHIFT: return decode_str (dec, 0);
984 return newSViv (-1 - (IV)decode_uint (dec)); 1052 case MAJOR_TEXT >> MAJOR_SHIFT: return decode_str (dec, 1);
985 case 2: // octet string 1053 case MAJOR_ARRAY >> MAJOR_SHIFT: return decode_av (dec);
986 return decode_str (dec, 0); 1054 case MAJOR_MAP >> MAJOR_SHIFT: return decode_hv (dec);
987 case 3: // utf-8 string 1055 case MAJOR_TAG >> MAJOR_SHIFT: return decode_tagged (dec);
988 return decode_str (dec, 1); 1056
989 case 4: // array 1057 case MAJOR_MISC >> MAJOR_SHIFT:
990 return decode_av (dec);
991 case 5: // map
992 return decode_hv (dec);
993 case 6: // tag
994 return decode_tagged (dec);
995 case 7: // misc
996 switch (*dec->cur++ & 31) 1058 switch (*dec->cur++ & MINOR_MASK)
997 { 1059 {
998 case 20: 1060 case SIMPLE_FALSE:
999#if CBOR_SLOW 1061#if CBOR_SLOW
1000 types_false = get_bool ("Types::Serialiser::false"); 1062 types_false = get_bool ("Types::Serialiser::false");
1001#endif 1063#endif
1002 return newSVsv (types_false); 1064 return newSVsv (types_false);
1003 case 21: 1065 case SIMPLE_TRUE:
1004#if CBOR_SLOW 1066#if CBOR_SLOW
1005 types_true = get_bool ("Types::Serialiser::true"); 1067 types_true = get_bool ("Types::Serialiser::true");
1006#endif 1068#endif
1007 return newSVsv (types_true); 1069 return newSVsv (types_true);
1008 case 22: 1070 case SIMPLE_NULL:
1009 return newSVsv (&PL_sv_undef); 1071 return newSVsv (&PL_sv_undef);
1010 case 23: 1072 case SIMPLE_UNDEF:
1011#if CBOR_SLOW 1073#if CBOR_SLOW
1012 types_error = get_bool ("Types::Serialiser::error"); 1074 types_error = get_bool ("Types::Serialiser::error");
1013#endif 1075#endif
1014 return newSVsv (types_error); 1076 return newSVsv (types_error);
1015 1077
1016 case 25: 1078 case MISC_FLOAT16:
1017 { 1079 {
1018 WANT (2); 1080 WANT (2);
1019 1081
1020 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1]; 1082 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1];
1021 dec->cur += 2; 1083 dec->cur += 2;
1022 1084
1023 return newSVnv (ecb_binary16_to_float (fp)); 1085 return newSVnv (ecb_binary16_to_float (fp));
1024 } 1086 }
1025 1087
1026 case 26: 1088 case MISC_FLOAT32:
1027 { 1089 {
1028 uint32_t fp; 1090 uint32_t fp;
1029 WANT (4); 1091 WANT (4);
1030 memcpy (&fp, dec->cur, 4); 1092 memcpy (&fp, dec->cur, 4);
1031 dec->cur += 4; 1093 dec->cur += 4;
1034 fp = ecb_bswap32 (fp); 1096 fp = ecb_bswap32 (fp);
1035 1097
1036 return newSVnv (ecb_binary32_to_float (fp)); 1098 return newSVnv (ecb_binary32_to_float (fp));
1037 } 1099 }
1038 1100
1039 case 27: 1101 case MISC_FLOAT64:
1040 { 1102 {
1041 uint64_t fp; 1103 uint64_t fp;
1042 WANT (8); 1104 WANT (8);
1043 memcpy (&fp, dec->cur, 8); 1105 memcpy (&fp, dec->cur, 8);
1044 dec->cur += 8; 1106 dec->cur += 8;
1047 fp = ecb_bswap64 (fp); 1109 fp = ecb_bswap64 (fp);
1048 1110
1049 return newSVnv (ecb_binary64_to_double (fp)); 1111 return newSVnv (ecb_binary64_to_double (fp));
1050 } 1112 }
1051 1113
1052 // 0..19 unassigned 1114 // 0..19 unassigned simple
1053 // 24 reserved + unassigned (reserved values are not encodable) 1115 // 24 reserved + unassigned (reserved values are not encodable)
1054 default: 1116 default:
1055 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)"); 1117 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)");
1056 } 1118 }
1057 1119
1227 cbor_free (self); 1289 cbor_free (self);
1228 1290
1229PROTOTYPES: ENABLE 1291PROTOTYPES: ENABLE
1230 1292
1231void encode_cbor (SV *scalar) 1293void encode_cbor (SV *scalar)
1294 ALIAS:
1295 encode_cbor = 0
1296 encode_cbor_sharing = F_ALLOW_SHARING
1232 PPCODE: 1297 PPCODE:
1233{ 1298{
1234 CBOR cbor; 1299 CBOR cbor;
1235 cbor_init (&cbor); 1300 cbor_init (&cbor);
1301 cbor.flags |= ix;
1236 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN; 1302 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;
1237 XPUSHs (scalar); 1303 XPUSHs (scalar);
1238} 1304}
1239 1305
1240void decode_cbor (SV *cborstr) 1306void decode_cbor (SV *cborstr)

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines