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 1.34 by root, Sat Nov 30 16:19:59 2013 UTC vs.
Revision 1.39 by root, Sun Dec 1 14:45:03 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
102#define F_ALLOW_CYCLES 0x00000008UL
65#define F_PACK_STRINGS 0x00000008UL 103#define F_PACK_STRINGS 0x00000010UL
104#define F_VALIDATE_UTF8 0x00000020UL
66 105
67#define INIT_SIZE 32 // initial scalar size to be allocated 106#define INIT_SIZE 32 // initial scalar size to be allocated
68 107
69#define SB do { 108#define SB do {
70#define SE } while (0) 109#define SE } while (0)
189static void 228static void
190encode_uint (enc_t *enc, int major, UV len) 229encode_uint (enc_t *enc, int major, UV len)
191{ 230{
192 need (enc, 9); 231 need (enc, 9);
193 232
194 if (ecb_expect_true (len < 24)) 233 if (ecb_expect_true (len < LENGTH_EXT1))
195 *enc->cur++ = major | len; 234 *enc->cur++ = major | len;
196 else if (ecb_expect_true (len <= 0xff)) 235 else if (ecb_expect_true (len <= 0xffU))
197 { 236 {
198 *enc->cur++ = major | 24; 237 *enc->cur++ = major | LENGTH_EXT1;
199 *enc->cur++ = len; 238 *enc->cur++ = len;
200 } 239 }
201 else if (len <= 0xffff) 240 else if (len <= 0xffffU)
202 { 241 {
203 *enc->cur++ = major | 25; 242 *enc->cur++ = major | LENGTH_EXT2;
204 *enc->cur++ = len >> 8; 243 *enc->cur++ = len >> 8;
205 *enc->cur++ = len; 244 *enc->cur++ = len;
206 } 245 }
207 else if (len <= 0xffffffff) 246 else if (len <= 0xffffffffU)
208 { 247 {
209 *enc->cur++ = major | 26; 248 *enc->cur++ = major | LENGTH_EXT4;
210 *enc->cur++ = len >> 24; 249 *enc->cur++ = len >> 24;
211 *enc->cur++ = len >> 16; 250 *enc->cur++ = len >> 16;
212 *enc->cur++ = len >> 8; 251 *enc->cur++ = len >> 8;
213 *enc->cur++ = len; 252 *enc->cur++ = len;
214 } 253 }
215 else 254 else
216 { 255 {
217 *enc->cur++ = major | 27; 256 *enc->cur++ = major | LENGTH_EXT8;
218 *enc->cur++ = len >> 56; 257 *enc->cur++ = len >> 56;
219 *enc->cur++ = len >> 48; 258 *enc->cur++ = len >> 48;
220 *enc->cur++ = len >> 40; 259 *enc->cur++ = len >> 40;
221 *enc->cur++ = len >> 32; 260 *enc->cur++ = len >> 32;
222 *enc->cur++ = len >> 24; 261 *enc->cur++ = len >> 24;
227} 266}
228 267
229ecb_inline void 268ecb_inline void
230encode_tag (enc_t *enc, UV tag) 269encode_tag (enc_t *enc, UV tag)
231{ 270{
232 encode_uint (enc, 0xc0, tag); 271 encode_uint (enc, MAJOR_TAG, tag);
233} 272}
234 273
235ecb_inline void 274ecb_inline void
236encode_str (enc_t *enc, int utf8, char *str, STRLEN len) 275encode_str (enc_t *enc, int utf8, char *str, STRLEN len)
237{ 276{
238 encode_uint (enc, utf8 ? 0x60 : 0x40, len); 277 encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len);
239 need (enc, len); 278 need (enc, len);
240 memcpy (enc->cur, str, len); 279 memcpy (enc->cur, str, len);
241 enc->cur += len; 280 enc->cur += len;
242} 281}
243 282
250 289
251 if (SvOK (*svp)) 290 if (SvOK (*svp))
252 { 291 {
253 // already registered, use stringref 292 // already registered, use stringref
254 encode_tag (enc, CBOR_TAG_STRINGREF); 293 encode_tag (enc, CBOR_TAG_STRINGREF);
255 encode_uint (enc, 0x00, SvUV (*svp)); 294 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
256 return; 295 return;
257 } 296 }
258 else if (len >= minimum_string_length (enc->stringref_idx)) 297 else if (len >= minimum_string_length (enc->stringref_idx))
259 { 298 {
260 // register only 299 // register only
276 if (enc->depth >= enc->cbor.max_depth) 315 if (enc->depth >= enc->cbor.max_depth)
277 croak (ERR_NESTING_EXCEEDED); 316 croak (ERR_NESTING_EXCEEDED);
278 317
279 ++enc->depth; 318 ++enc->depth;
280 319
281 encode_uint (enc, 0x80, len + 1); 320 encode_uint (enc, MAJOR_ARRAY, len + 1);
282 321
283 for (i = 0; i <= len; ++i) 322 for (i = 0; i <= len; ++i)
284 { 323 {
285 SV **svp = av_fetch (av, i, 0); 324 SV **svp = av_fetch (av, i, 0);
286 encode_sv (enc, svp ? *svp : &PL_sv_undef); 325 encode_sv (enc, svp ? *svp : &PL_sv_undef);
301 340
302 int pairs = hv_iterinit (hv); 341 int pairs = hv_iterinit (hv);
303 int mg = SvMAGICAL (hv); 342 int mg = SvMAGICAL (hv);
304 343
305 if (mg) 344 if (mg)
306 encode_ch (enc, 0xa0 | 31); 345 encode_ch (enc, MAJOR_MAP | MINOR_INDEF);
307 else 346 else
308 encode_uint (enc, 0xa0, pairs); 347 encode_uint (enc, MAJOR_MAP, pairs);
309 348
310 while ((he = hv_iternext (hv))) 349 while ((he = hv_iternext (hv)))
311 { 350 {
312 if (HeKLEN (he) == HEf_SVKEY) 351 if (HeKLEN (he) == HEf_SVKEY)
313 encode_sv (enc, HeSVKEY (he)); 352 encode_sv (enc, HeSVKEY (he));
316 355
317 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he)); 356 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he));
318 } 357 }
319 358
320 if (mg) 359 if (mg)
321 encode_ch (enc, 0xe0 | 31); 360 encode_ch (enc, MAJOR_MISC | MINOR_INDEF);
322 361
323 --enc->depth; 362 --enc->depth;
324} 363}
325 364
326// encode objects, arrays and special \0=false and \1=true values. 365// encode objects, arrays and special \0=false and \1=true values.
345 384
346 HV *stash = SvSTASH (sv); 385 HV *stash = SvSTASH (sv);
347 386
348 if (stash == boolean_stash) 387 if (stash == boolean_stash)
349 { 388 {
350 encode_ch (enc, SvIV (sv) ? 0xe0 | 21 : 0xe0 | 20); 389 encode_ch (enc, SvIV (sv) ? MAJOR_MISC | SIMPLE_TRUE : MAJOR_MISC | SIMPLE_FALSE);
351 return; 390 return;
352 } 391 }
353 else if (stash == error_stash) 392 else if (stash == error_stash)
354 { 393 {
355 encode_ch (enc, 0xe0 | 23); 394 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
356 return; 395 return;
357 } 396 }
358 else if (stash == tagged_stash) 397 else if (stash == tagged_stash)
359 { 398 {
360 if (svt != SVt_PVAV) 399 if (svt != SVt_PVAV)
361 croak ("encountered CBOR::XS::Tagged object that isn't an array"); 400 croak ("encountered CBOR::XS::Tagged object that isn't an array");
362 401
363 encode_uint (enc, 0xc0, SvUV (*av_fetch ((AV *)sv, 0, 1))); 402 encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1)));
364 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1)); 403 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1));
365 404
366 return; 405 return;
367 } 406 }
368 } 407 }
376 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1); 415 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
377 416
378 if (SvOK (*svp)) 417 if (SvOK (*svp))
379 { 418 {
380 encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF); 419 encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF);
381 encode_uint (enc, 0x00, SvUV (*svp)); 420 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
382 return; 421 return;
383 } 422 }
384 else 423 else
385 { 424 {
386 sv_setuv (*svp, enc->shareable_idx); 425 sv_setuv (*svp, enc->shareable_idx);
434 // catch this surprisingly common error 473 // catch this surprisingly common error
435 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv) 474 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)
436 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash)); 475 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));
437 476
438 encode_tag (enc, CBOR_TAG_PERL_OBJECT); 477 encode_tag (enc, CBOR_TAG_PERL_OBJECT);
439 encode_uint (enc, 0x80, count + 1); 478 encode_uint (enc, MAJOR_ARRAY, count + 1);
440 encode_strref (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash)); 479 encode_strref (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
441 480
442 while (count) 481 while (count)
443 encode_sv (enc, SP[1 - count--]); 482 encode_sv (enc, SP[1 - count--]);
444 483
466{ 505{
467 double nv = SvNVX (sv); 506 double nv = SvNVX (sv);
468 507
469 need (enc, 9); 508 need (enc, 9);
470 509
471 if (ecb_expect_false (nv == (U32)nv)) 510 if (ecb_expect_false (nv == (NV)(U32)nv))
472 encode_uint (enc, 0x00, (U32)nv); 511 encode_uint (enc, MAJOR_POS_INT, (U32)nv);
473 //TODO: maybe I32? 512 //TODO: maybe I32?
474 else if (ecb_expect_false (nv == (float)nv)) 513 else if (ecb_expect_false (nv == (float)nv))
475 { 514 {
476 uint32_t fp = ecb_float_to_binary32 (nv); 515 uint32_t fp = ecb_float_to_binary32 (nv);
477 516
478 *enc->cur++ = 0xe0 | 26; 517 *enc->cur++ = MAJOR_MISC | MISC_FLOAT32;
479 518
480 if (!ecb_big_endian ()) 519 if (!ecb_big_endian ())
481 fp = ecb_bswap32 (fp); 520 fp = ecb_bswap32 (fp);
482 521
483 memcpy (enc->cur, &fp, 4); 522 memcpy (enc->cur, &fp, 4);
485 } 524 }
486 else 525 else
487 { 526 {
488 uint64_t fp = ecb_double_to_binary64 (nv); 527 uint64_t fp = ecb_double_to_binary64 (nv);
489 528
490 *enc->cur++ = 0xe0 | 27; 529 *enc->cur++ = MAJOR_MISC | MISC_FLOAT64;
491 530
492 if (!ecb_big_endian ()) 531 if (!ecb_big_endian ())
493 fp = ecb_bswap64 (fp); 532 fp = ecb_bswap64 (fp);
494 533
495 memcpy (enc->cur, &fp, 8); 534 memcpy (enc->cur, &fp, 8);
511 else if (SvNOKp (sv)) 550 else if (SvNOKp (sv))
512 encode_nv (enc, sv); 551 encode_nv (enc, sv);
513 else if (SvIOKp (sv)) 552 else if (SvIOKp (sv))
514 { 553 {
515 if (SvIsUV (sv)) 554 if (SvIsUV (sv))
516 encode_uint (enc, 0x00, SvUVX (sv)); 555 encode_uint (enc, MAJOR_POS_INT, SvUVX (sv));
517 else if (SvIVX (sv) >= 0) 556 else if (SvIVX (sv) >= 0)
518 encode_uint (enc, 0x00, SvIVX (sv)); 557 encode_uint (enc, MAJOR_POS_INT, SvIVX (sv));
519 else 558 else
520 encode_uint (enc, 0x20, -(SvIVX (sv) + 1)); 559 encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1));
521 } 560 }
522 else if (SvROK (sv)) 561 else if (SvROK (sv))
523 encode_rv (enc, SvRV (sv)); 562 encode_rv (enc, SvRV (sv));
524 else if (!SvOK (sv)) 563 else if (!SvOK (sv))
525 encode_ch (enc, 0xe0 | 22); 564 encode_ch (enc, MAJOR_MISC | SIMPLE_NULL);
526 else if (enc->cbor.flags & F_ALLOW_UNKNOWN) 565 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
527 encode_ch (enc, 0xe0 | 23); 566 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
528 else 567 else
529 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data", 568 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data",
530 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv)); 569 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
531} 570}
532 571
585#define DEC_DEC_DEPTH --dec->depth 624#define DEC_DEC_DEPTH --dec->depth
586 625
587static UV 626static UV
588decode_uint (dec_t *dec) 627decode_uint (dec_t *dec)
589{ 628{
590 switch (*dec->cur & 31) 629 U8 m = *dec->cur & MINOR_MASK;
591 { 630 ++dec->cur;
592 case 0: case 1: case 2: case 3: case 4: case 5: case 6: case 7:
593 case 8: case 9: case 10: case 11: case 12: case 13: case 14: case 15:
594 case 16: case 17: case 18: case 19: case 20: case 21: case 22: case 23:
595 return *dec->cur++ & 31;
596 631
597 case 24: 632 if (ecb_expect_true (m < LENGTH_EXT1))
633 return m;
634 else if (ecb_expect_true (m == LENGTH_EXT1))
635 {
598 WANT (2); 636 WANT (1);
599 dec->cur += 2; 637 dec->cur += 1;
600 return dec->cur[-1]; 638 return dec->cur[-1];
601 639 }
602 case 25: 640 else if (ecb_expect_true (m == LENGTH_EXT2))
641 {
603 WANT (3); 642 WANT (2);
604 dec->cur += 3; 643 dec->cur += 2;
605 return (((UV)dec->cur[-2]) << 8) 644 return (((UV)dec->cur[-2]) << 8)
606 | ((UV)dec->cur[-1]); 645 | ((UV)dec->cur[-1]);
607 646 }
608 case 26: 647 else if (ecb_expect_true (m == LENGTH_EXT4))
648 {
609 WANT (5); 649 WANT (4);
610 dec->cur += 5; 650 dec->cur += 4;
611 return (((UV)dec->cur[-4]) << 24) 651 return (((UV)dec->cur[-4]) << 24)
612 | (((UV)dec->cur[-3]) << 16) 652 | (((UV)dec->cur[-3]) << 16)
613 | (((UV)dec->cur[-2]) << 8) 653 | (((UV)dec->cur[-2]) << 8)
614 | ((UV)dec->cur[-1]); 654 | ((UV)dec->cur[-1]);
615 655 }
616 case 27: 656 else if (ecb_expect_true (m == LENGTH_EXT8))
657 {
617 WANT (9); 658 WANT (8);
618 dec->cur += 9; 659 dec->cur += 8;
619 660
620 return 661 return
621#if UVSIZE < 8 662#if UVSIZE < 8
622 0 663 0
623#else 664#else
624 (((UV)dec->cur[-8]) << 56) 665 (((UV)dec->cur[-8]) << 56)
625 | (((UV)dec->cur[-7]) << 48) 666 | (((UV)dec->cur[-7]) << 48)
626 | (((UV)dec->cur[-6]) << 40) 667 | (((UV)dec->cur[-6]) << 40)
627 | (((UV)dec->cur[-5]) << 32) 668 | (((UV)dec->cur[-5]) << 32)
628#endif 669#endif
629 | (((UV)dec->cur[-4]) << 24) 670 | (((UV)dec->cur[-4]) << 24)
630 | (((UV)dec->cur[-3]) << 16) 671 | (((UV)dec->cur[-3]) << 16)
631 | (((UV)dec->cur[-2]) << 8) 672 | (((UV)dec->cur[-2]) << 8)
632 | ((UV)dec->cur[-1]); 673 | ((UV)dec->cur[-1]);
633 674 }
634 default: 675 else
635 ERR ("corrupted CBOR data (unsupported integer minor encoding)"); 676 ERR ("corrupted CBOR data (unsupported integer minor encoding)");
636 }
637 677
638fail: 678fail:
639 return 0; 679 return 0;
640} 680}
641 681
646{ 686{
647 AV *av = newAV (); 687 AV *av = newAV ();
648 688
649 DEC_INC_DEPTH; 689 DEC_INC_DEPTH;
650 690
651 if ((*dec->cur & 31) == 31) 691 if (*dec->cur == (MAJOR_ARRAY | MINOR_INDEF))
652 { 692 {
653 ++dec->cur; 693 ++dec->cur;
654 694
655 for (;;) 695 for (;;)
656 { 696 {
657 WANT (1); 697 WANT (1);
658 698
659 if (*dec->cur == (0xe0 | 31)) 699 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
660 { 700 {
661 ++dec->cur; 701 ++dec->cur;
662 break; 702 break;
663 } 703 }
664 704
667 } 707 }
668 else 708 else
669 { 709 {
670 int i, len = decode_uint (dec); 710 int i, len = decode_uint (dec);
671 711
712 WANT (len); // complexity check for av_fill - need at least one byte per value, do not allow supersize arrays
672 av_fill (av, len - 1); 713 av_fill (av, len - 1);
673 714
674 for (i = 0; i < len; ++i) 715 for (i = 0; i < len; ++i)
675 AvARRAY (av)[i] = decode_sv (dec); 716 AvARRAY (av)[i] = decode_sv (dec);
676 } 717 }
689{ 730{
690 // for speed reasons, we specialcase single-string 731 // for speed reasons, we specialcase single-string
691 // byte or utf-8 strings as keys, but only when !stringref 732 // byte or utf-8 strings as keys, but only when !stringref
692 733
693 if (ecb_expect_true (!dec->stringref)) 734 if (ecb_expect_true (!dec->stringref))
694 if (*dec->cur >= 0x40 && *dec->cur <= 0x40 + 27) 735 if (ecb_expect_true ((*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8))
695 { 736 {
696 I32 len = decode_uint (dec); 737 I32 len = decode_uint (dec);
697 char *key = (char *)dec->cur; 738 char *key = (char *)dec->cur;
698 739
699 dec->cur += len; 740 dec->cur += len;
700 741
701 if (ecb_expect_false (dec->stringref))
702 av_push (dec->stringref, newSVpvn (key, len));
703
704 hv_store (hv, key, len, decode_sv (dec), 0); 742 hv_store (hv, key, len, decode_sv (dec), 0);
705 743
706 return; 744 return;
707 } 745 }
708 else if (*dec->cur >= 0x60 && *dec->cur <= 0x60 + 27) 746 else if (ecb_expect_true ((*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8))
709 { 747 {
710 I32 len = decode_uint (dec); 748 I32 len = decode_uint (dec);
711 char *key = (char *)dec->cur; 749 char *key = (char *)dec->cur;
712 750
713 dec->cur += len; 751 dec->cur += len;
714 752
715 if (ecb_expect_false (dec->stringref)) 753 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
716 av_push (dec->stringref, newSVpvn_utf8 (key, len, 1)); 754 if (!is_utf8_string (key, len))
755 ERR ("corrupted CBOR data (invalid UTF-8 in map key)");
717 756
718 hv_store (hv, key, -len, decode_sv (dec), 0); 757 hv_store (hv, key, -len, decode_sv (dec), 0);
719 758
720 return; 759 return;
721 } 760 }
723 SV *k = decode_sv (dec); 762 SV *k = decode_sv (dec);
724 SV *v = decode_sv (dec); 763 SV *v = decode_sv (dec);
725 764
726 hv_store_ent (hv, k, v, 0); 765 hv_store_ent (hv, k, v, 0);
727 SvREFCNT_dec (k); 766 SvREFCNT_dec (k);
767
768fail:
769 ;
728} 770}
729 771
730static SV * 772static SV *
731decode_hv (dec_t *dec) 773decode_hv (dec_t *dec)
732{ 774{
733 HV *hv = newHV (); 775 HV *hv = newHV ();
734 776
735 DEC_INC_DEPTH; 777 DEC_INC_DEPTH;
736 778
737 if ((*dec->cur & 31) == 31) 779 if (*dec->cur == (MAJOR_MAP | MINOR_INDEF))
738 { 780 {
739 ++dec->cur; 781 ++dec->cur;
740 782
741 for (;;) 783 for (;;)
742 { 784 {
743 WANT (1); 785 WANT (1);
744 786
745 if (*dec->cur == (0xe0 | 31)) 787 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
746 { 788 {
747 ++dec->cur; 789 ++dec->cur;
748 break; 790 break;
749 } 791 }
750 792
771static SV * 813static SV *
772decode_str (dec_t *dec, int utf8) 814decode_str (dec_t *dec, int utf8)
773{ 815{
774 SV *sv = 0; 816 SV *sv = 0;
775 817
776 if ((*dec->cur & 31) == 31) 818 if ((*dec->cur & MINOR_MASK) == MINOR_INDEF)
777 { 819 {
778 // indefinite length strings 820 // indefinite length strings
779 ++dec->cur; 821 ++dec->cur;
780 822
781 unsigned char major = *dec->cur & 0xe0; 823 U8 major = *dec->cur & MAJOR_MISC;
782 824
783 sv = newSVpvn ("", 0); 825 sv = newSVpvn ("", 0);
784 826
785 for (;;) 827 for (;;)
786 { 828 {
787 WANT (1); 829 WANT (1);
788 830
789 if ((*dec->cur ^ major) >= 31) 831 if ((*dec->cur - major) > LENGTH_EXT8)
790 if (*dec->cur == (0xe0 | 31)) 832 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
791 { 833 {
792 ++dec->cur; 834 ++dec->cur;
793 break; 835 break;
794 } 836 }
795 else 837 else
814 && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1)) 856 && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1))
815 av_push (dec->stringref, SvREFCNT_inc_NN (sv)); 857 av_push (dec->stringref, SvREFCNT_inc_NN (sv));
816 } 858 }
817 859
818 if (utf8) 860 if (utf8)
861 {
862 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
863 if (!is_utf8_string (SvPVX (sv), SvCUR (sv)))
864 ERR ("corrupted CBOR data (invalid UTF-8 in text string)");
865
819 SvUTF8_on (sv); 866 SvUTF8_on (sv);
867 }
820 868
821 return sv; 869 return sv;
822 870
823fail: 871fail:
824 SvREFCNT_dec (sv); 872 SvREFCNT_dec (sv);
856 } 904 }
857 break; 905 break;
858 906
859 case CBOR_TAG_STRINGREF: 907 case CBOR_TAG_STRINGREF:
860 { 908 {
861 if ((*dec->cur >> 5) != 0) 909 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
862 ERR ("corrupted CBOR data (stringref index not an unsigned integer)"); 910 ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
863 911
864 UV idx = decode_uint (dec); 912 UV idx = decode_uint (dec);
865 913
866 if (!dec->stringref || (int)idx > AvFILLp (dec->stringref)) 914 if (!dec->stringref || (int)idx > AvFILLp (dec->stringref))
873 case CBOR_TAG_VALUE_SHAREABLE: 921 case CBOR_TAG_VALUE_SHAREABLE:
874 { 922 {
875 if (ecb_expect_false (!dec->shareable)) 923 if (ecb_expect_false (!dec->shareable))
876 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ()); 924 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ());
877 925
926 if (dec->cbor.flags & F_ALLOW_CYCLES)
927 {
878 sv = newSV (0); 928 sv = newSV (0);
879 av_push (dec->shareable, SvREFCNT_inc_NN (sv)); 929 av_push (dec->shareable, SvREFCNT_inc_NN (sv));
880 930
881 SV *osv = decode_sv (dec); 931 SV *osv = decode_sv (dec);
882 sv_setsv (sv, osv); 932 sv_setsv (sv, osv);
883 SvREFCNT_dec_NN (osv); 933 SvREFCNT_dec_NN (osv);
934 }
935 else
936 {
937 av_push (dec->shareable, &PL_sv_undef);
938 int idx = AvFILLp (dec->shareable);
939 sv = decode_sv (dec);
940 av_store (dec->shareable, idx, SvREFCNT_inc_NN (sv));
941 }
884 } 942 }
885 break; 943 break;
886 944
887 case CBOR_TAG_VALUE_SHAREDREF: 945 case CBOR_TAG_VALUE_SHAREDREF:
888 { 946 {
889 if ((*dec->cur >> 5) != 0) 947 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
890 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)"); 948 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
891 949
892 UV idx = decode_uint (dec); 950 UV idx = decode_uint (dec);
893 951
894 if (!dec->shareable || (int)idx > AvFILLp (dec->shareable)) 952 if (!dec->shareable || (int)idx > AvFILLp (dec->shareable))
895 ERR ("corrupted CBOR data (sharedref index out of bounds)"); 953 ERR ("corrupted CBOR data (sharedref index out of bounds)");
896 954
897 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]); 955 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]);
956
957 if (sv == &PL_sv_undef)
958 ERR ("cyclic CBOR data structure found, but allow_cycles is not enabled");
898 } 959 }
899 break; 960 break;
900 961
901 case CBOR_TAG_PERL_OBJECT: 962 case CBOR_TAG_PERL_OBJECT:
902 { 963 {
1003static SV * 1064static SV *
1004decode_sv (dec_t *dec) 1065decode_sv (dec_t *dec)
1005{ 1066{
1006 WANT (1); 1067 WANT (1);
1007 1068
1008 switch (*dec->cur >> 5) 1069 switch (*dec->cur >> MAJOR_SHIFT)
1009 { 1070 {
1010 case 0: // unsigned int 1071 case MAJOR_POS_INT >> MAJOR_SHIFT: return newSVuv (decode_uint (dec));
1011 return newSVuv (decode_uint (dec)); 1072 case MAJOR_NEG_INT >> MAJOR_SHIFT: return newSViv (-1 - (IV)decode_uint (dec));
1012 case 1: // negative int 1073 case MAJOR_BYTES >> MAJOR_SHIFT: return decode_str (dec, 0);
1013 return newSViv (-1 - (IV)decode_uint (dec)); 1074 case MAJOR_TEXT >> MAJOR_SHIFT: return decode_str (dec, 1);
1014 case 2: // octet string 1075 case MAJOR_ARRAY >> MAJOR_SHIFT: return decode_av (dec);
1015 return decode_str (dec, 0); 1076 case MAJOR_MAP >> MAJOR_SHIFT: return decode_hv (dec);
1016 case 3: // utf-8 string 1077 case MAJOR_TAG >> MAJOR_SHIFT: return decode_tagged (dec);
1017 return decode_str (dec, 1); 1078
1018 case 4: // array 1079 case MAJOR_MISC >> MAJOR_SHIFT:
1019 return decode_av (dec);
1020 case 5: // map
1021 return decode_hv (dec);
1022 case 6: // tag
1023 return decode_tagged (dec);
1024 case 7: // misc
1025 switch (*dec->cur++ & 31) 1080 switch (*dec->cur++ & MINOR_MASK)
1026 { 1081 {
1027 case 20: 1082 case SIMPLE_FALSE:
1028#if CBOR_SLOW 1083#if CBOR_SLOW
1029 types_false = get_bool ("Types::Serialiser::false"); 1084 types_false = get_bool ("Types::Serialiser::false");
1030#endif 1085#endif
1031 return newSVsv (types_false); 1086 return newSVsv (types_false);
1032 case 21: 1087 case SIMPLE_TRUE:
1033#if CBOR_SLOW 1088#if CBOR_SLOW
1034 types_true = get_bool ("Types::Serialiser::true"); 1089 types_true = get_bool ("Types::Serialiser::true");
1035#endif 1090#endif
1036 return newSVsv (types_true); 1091 return newSVsv (types_true);
1037 case 22: 1092 case SIMPLE_NULL:
1038 return newSVsv (&PL_sv_undef); 1093 return newSVsv (&PL_sv_undef);
1039 case 23: 1094 case SIMPLE_UNDEF:
1040#if CBOR_SLOW 1095#if CBOR_SLOW
1041 types_error = get_bool ("Types::Serialiser::error"); 1096 types_error = get_bool ("Types::Serialiser::error");
1042#endif 1097#endif
1043 return newSVsv (types_error); 1098 return newSVsv (types_error);
1044 1099
1045 case 25: 1100 case MISC_FLOAT16:
1046 { 1101 {
1047 WANT (2); 1102 WANT (2);
1048 1103
1049 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1]; 1104 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1];
1050 dec->cur += 2; 1105 dec->cur += 2;
1051 1106
1052 return newSVnv (ecb_binary16_to_float (fp)); 1107 return newSVnv (ecb_binary16_to_float (fp));
1053 } 1108 }
1054 1109
1055 case 26: 1110 case MISC_FLOAT32:
1056 { 1111 {
1057 uint32_t fp; 1112 uint32_t fp;
1058 WANT (4); 1113 WANT (4);
1059 memcpy (&fp, dec->cur, 4); 1114 memcpy (&fp, dec->cur, 4);
1060 dec->cur += 4; 1115 dec->cur += 4;
1063 fp = ecb_bswap32 (fp); 1118 fp = ecb_bswap32 (fp);
1064 1119
1065 return newSVnv (ecb_binary32_to_float (fp)); 1120 return newSVnv (ecb_binary32_to_float (fp));
1066 } 1121 }
1067 1122
1068 case 27: 1123 case MISC_FLOAT64:
1069 { 1124 {
1070 uint64_t fp; 1125 uint64_t fp;
1071 WANT (8); 1126 WANT (8);
1072 memcpy (&fp, dec->cur, 8); 1127 memcpy (&fp, dec->cur, 8);
1073 dec->cur += 8; 1128 dec->cur += 8;
1076 fp = ecb_bswap64 (fp); 1131 fp = ecb_bswap64 (fp);
1077 1132
1078 return newSVnv (ecb_binary64_to_double (fp)); 1133 return newSVnv (ecb_binary64_to_double (fp));
1079 } 1134 }
1080 1135
1081 // 0..19 unassigned 1136 // 0..19 unassigned simple
1082 // 24 reserved + unassigned (reserved values are not encodable) 1137 // 24 reserved + unassigned (reserved values are not encodable)
1083 default: 1138 default:
1084 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)"); 1139 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)");
1085 } 1140 }
1086 1141
1116 if (dec.cur != dec.end && !dec.err) 1171 if (dec.cur != dec.end && !dec.err)
1117 dec.err = "garbage after CBOR object"; 1172 dec.err = "garbage after CBOR object";
1118 1173
1119 if (dec.err) 1174 if (dec.err)
1120 { 1175 {
1176 if (dec.shareable)
1177 {
1178 // need to break cyclic links, which whould all be in shareable
1179 int i;
1180 SV **svp;
1181
1182 for (i = av_len (dec.shareable) + 1; i--; )
1183 if ((svp = av_fetch (dec.shareable, i, 0)))
1184 sv_setsv (*svp, &PL_sv_undef);
1185 }
1186
1121 SvREFCNT_dec (sv); 1187 SvREFCNT_dec (sv);
1122 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur); 1188 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur);
1123 } 1189 }
1124 1190
1125 sv = sv_2mortal (sv); 1191 sv = sv_2mortal (sv);
1174void shrink (CBOR *self, int enable = 1) 1240void shrink (CBOR *self, int enable = 1)
1175 ALIAS: 1241 ALIAS:
1176 shrink = F_SHRINK 1242 shrink = F_SHRINK
1177 allow_unknown = F_ALLOW_UNKNOWN 1243 allow_unknown = F_ALLOW_UNKNOWN
1178 allow_sharing = F_ALLOW_SHARING 1244 allow_sharing = F_ALLOW_SHARING
1245 allow_cycles = F_ALLOW_CYCLES
1179 pack_strings = F_PACK_STRINGS 1246 pack_strings = F_PACK_STRINGS
1247 validate_utf8 = F_VALIDATE_UTF8
1180 PPCODE: 1248 PPCODE:
1181{ 1249{
1182 if (enable) 1250 if (enable)
1183 self->flags |= ix; 1251 self->flags |= ix;
1184 else 1252 else
1190void get_shrink (CBOR *self) 1258void get_shrink (CBOR *self)
1191 ALIAS: 1259 ALIAS:
1192 get_shrink = F_SHRINK 1260 get_shrink = F_SHRINK
1193 get_allow_unknown = F_ALLOW_UNKNOWN 1261 get_allow_unknown = F_ALLOW_UNKNOWN
1194 get_allow_sharing = F_ALLOW_SHARING 1262 get_allow_sharing = F_ALLOW_SHARING
1263 get_allow_cycles = F_ALLOW_CYCLES
1195 get_pack_strings = F_PACK_STRINGS 1264 get_pack_strings = F_PACK_STRINGS
1265 get_validate_utf8 = F_VALIDATE_UTF8
1196 PPCODE: 1266 PPCODE:
1197 XPUSHs (boolSV (self->flags & ix)); 1267 XPUSHs (boolSV (self->flags & ix));
1198 1268
1199void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) 1269void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
1200 PPCODE: 1270 PPCODE:
1256 cbor_free (self); 1326 cbor_free (self);
1257 1327
1258PROTOTYPES: ENABLE 1328PROTOTYPES: ENABLE
1259 1329
1260void encode_cbor (SV *scalar) 1330void encode_cbor (SV *scalar)
1331 ALIAS:
1332 encode_cbor = 0
1333 encode_cbor_sharing = F_ALLOW_SHARING
1261 PPCODE: 1334 PPCODE:
1262{ 1335{
1263 CBOR cbor; 1336 CBOR cbor;
1264 cbor_init (&cbor); 1337 cbor_init (&cbor);
1338 cbor.flags |= ix;
1265 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN; 1339 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;
1266 XPUSHs (scalar); 1340 XPUSHs (scalar);
1267} 1341}
1268 1342
1269void decode_cbor (SV *cborstr) 1343void decode_cbor (SV *cborstr)

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines