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 1.37 by root, Sat Nov 30 18:13:53 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
66 104
67#define INIT_SIZE 32 // initial scalar size to be allocated 105#define INIT_SIZE 32 // initial scalar size to be allocated
68 106
69#define SB do { 107#define SB do {
70#define SE } while (0) 108#define SE } while (0)
189static void 227static void
190encode_uint (enc_t *enc, int major, UV len) 228encode_uint (enc_t *enc, int major, UV len)
191{ 229{
192 need (enc, 9); 230 need (enc, 9);
193 231
194 if (ecb_expect_true (len < 24)) 232 if (ecb_expect_true (len < LENGTH_EXT1))
195 *enc->cur++ = major | len; 233 *enc->cur++ = major | len;
196 else if (ecb_expect_true (len <= 0xff)) 234 else if (ecb_expect_true (len <= 0xffU))
197 { 235 {
198 *enc->cur++ = major | 24; 236 *enc->cur++ = major | LENGTH_EXT1;
199 *enc->cur++ = len; 237 *enc->cur++ = len;
200 } 238 }
201 else if (len <= 0xffff) 239 else if (len <= 0xffffU)
202 { 240 {
203 *enc->cur++ = major | 25; 241 *enc->cur++ = major | LENGTH_EXT2;
204 *enc->cur++ = len >> 8; 242 *enc->cur++ = len >> 8;
205 *enc->cur++ = len; 243 *enc->cur++ = len;
206 } 244 }
207 else if (len <= 0xffffffff) 245 else if (len <= 0xffffffffU)
208 { 246 {
209 *enc->cur++ = major | 26; 247 *enc->cur++ = major | LENGTH_EXT4;
210 *enc->cur++ = len >> 24; 248 *enc->cur++ = len >> 24;
211 *enc->cur++ = len >> 16; 249 *enc->cur++ = len >> 16;
212 *enc->cur++ = len >> 8; 250 *enc->cur++ = len >> 8;
213 *enc->cur++ = len; 251 *enc->cur++ = len;
214 } 252 }
215 else 253 else
216 { 254 {
217 *enc->cur++ = major | 27; 255 *enc->cur++ = major | LENGTH_EXT8;
218 *enc->cur++ = len >> 56; 256 *enc->cur++ = len >> 56;
219 *enc->cur++ = len >> 48; 257 *enc->cur++ = len >> 48;
220 *enc->cur++ = len >> 40; 258 *enc->cur++ = len >> 40;
221 *enc->cur++ = len >> 32; 259 *enc->cur++ = len >> 32;
222 *enc->cur++ = len >> 24; 260 *enc->cur++ = len >> 24;
227} 265}
228 266
229ecb_inline void 267ecb_inline void
230encode_tag (enc_t *enc, UV tag) 268encode_tag (enc_t *enc, UV tag)
231{ 269{
232 encode_uint (enc, 0xc0, tag); 270 encode_uint (enc, MAJOR_TAG, tag);
233} 271}
234 272
235ecb_inline void 273ecb_inline void
236encode_str (enc_t *enc, int utf8, char *str, STRLEN len) 274encode_str (enc_t *enc, int utf8, char *str, STRLEN len)
237{ 275{
238 encode_uint (enc, utf8 ? 0x60 : 0x40, len); 276 encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len);
239 need (enc, len); 277 need (enc, len);
240 memcpy (enc->cur, str, len); 278 memcpy (enc->cur, str, len);
241 enc->cur += len; 279 enc->cur += len;
242} 280}
243 281
250 288
251 if (SvOK (*svp)) 289 if (SvOK (*svp))
252 { 290 {
253 // already registered, use stringref 291 // already registered, use stringref
254 encode_tag (enc, CBOR_TAG_STRINGREF); 292 encode_tag (enc, CBOR_TAG_STRINGREF);
255 encode_uint (enc, 0x00, SvUV (*svp)); 293 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
256 return; 294 return;
257 } 295 }
258 else if (len >= minimum_string_length (enc->stringref_idx)) 296 else if (len >= minimum_string_length (enc->stringref_idx))
259 { 297 {
260 // register only 298 // register only
276 if (enc->depth >= enc->cbor.max_depth) 314 if (enc->depth >= enc->cbor.max_depth)
277 croak (ERR_NESTING_EXCEEDED); 315 croak (ERR_NESTING_EXCEEDED);
278 316
279 ++enc->depth; 317 ++enc->depth;
280 318
281 encode_uint (enc, 0x80, len + 1); 319 encode_uint (enc, MAJOR_ARRAY, len + 1);
282 320
283 for (i = 0; i <= len; ++i) 321 for (i = 0; i <= len; ++i)
284 { 322 {
285 SV **svp = av_fetch (av, i, 0); 323 SV **svp = av_fetch (av, i, 0);
286 encode_sv (enc, svp ? *svp : &PL_sv_undef); 324 encode_sv (enc, svp ? *svp : &PL_sv_undef);
301 339
302 int pairs = hv_iterinit (hv); 340 int pairs = hv_iterinit (hv);
303 int mg = SvMAGICAL (hv); 341 int mg = SvMAGICAL (hv);
304 342
305 if (mg) 343 if (mg)
306 encode_ch (enc, 0xa0 | 31); 344 encode_ch (enc, MAJOR_MAP | MINOR_INDEF);
307 else 345 else
308 encode_uint (enc, 0xa0, pairs); 346 encode_uint (enc, MAJOR_MAP, pairs);
309 347
310 while ((he = hv_iternext (hv))) 348 while ((he = hv_iternext (hv)))
311 { 349 {
312 if (HeKLEN (he) == HEf_SVKEY) 350 if (HeKLEN (he) == HEf_SVKEY)
313 encode_sv (enc, HeSVKEY (he)); 351 encode_sv (enc, HeSVKEY (he));
316 354
317 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he)); 355 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he));
318 } 356 }
319 357
320 if (mg) 358 if (mg)
321 encode_ch (enc, 0xe0 | 31); 359 encode_ch (enc, MAJOR_MISC | MINOR_INDEF);
322 360
323 --enc->depth; 361 --enc->depth;
324} 362}
325 363
326// encode objects, arrays and special \0=false and \1=true values. 364// encode objects, arrays and special \0=false and \1=true values.
327static void 365static void
328encode_rv (enc_t *enc, SV *sv) 366encode_rv (enc_t *enc, SV *sv)
329{ 367{
330 SvGETMAGIC (sv); 368 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 }
353 369
354 svtype svt = SvTYPE (sv); 370 svtype svt = SvTYPE (sv);
355 371
356 if (ecb_expect_false (SvOBJECT (sv))) 372 if (ecb_expect_false (SvOBJECT (sv)))
357 { 373 {
364 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 380 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
365 ? cbor_tagged_stash 381 ? cbor_tagged_stash
366 : gv_stashpv ("CBOR::XS::Tagged" , 1); 382 : gv_stashpv ("CBOR::XS::Tagged" , 1);
367 383
368 HV *stash = SvSTASH (sv); 384 HV *stash = SvSTASH (sv);
369 GV *method;
370 385
371 if (stash == boolean_stash) 386 if (stash == boolean_stash)
372 encode_ch (enc, SvIV (sv) ? 0xe0 | 21 : 0xe0 | 20); 387 {
388 encode_ch (enc, SvIV (sv) ? MAJOR_MISC | SIMPLE_TRUE : MAJOR_MISC | SIMPLE_FALSE);
389 return;
390 }
373 else if (stash == error_stash) 391 else if (stash == error_stash)
374 encode_ch (enc, 0xe0 | 23); 392 {
393 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
394 return;
395 }
375 else if (stash == tagged_stash) 396 else if (stash == tagged_stash)
376 { 397 {
377 if (svt != SVt_PVAV) 398 if (svt != SVt_PVAV)
378 croak ("encountered CBOR::XS::Tagged object that isn't an array"); 399 croak ("encountered CBOR::XS::Tagged object that isn't an array");
379 400
380 encode_uint (enc, 0xc0, SvUV (*av_fetch ((AV *)sv, 0, 1))); 401 encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1)));
381 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1)); 402 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1));
403
404 return;
405 }
406 }
407
408 if (ecb_expect_false (SvREFCNT (sv) > 1)
409 && ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING))
410 {
411 if (!enc->shareable)
412 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
413
414 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
415
416 if (SvOK (*svp))
382 } 417 {
418 encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF);
419 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
420 return;
421 }
422 else
423 {
424 sv_setuv (*svp, enc->shareable_idx);
425 ++enc->shareable_idx;
426 encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE);
427 }
428 }
429
430 if (ecb_expect_false (SvOBJECT (sv)))
431 {
432 HV *stash = SvSTASH (sv);
433 GV *method;
434
383 else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0))) 435 if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
384 { 436 {
385 dSP; 437 dSP;
386 438
387 ENTER; SAVETMPS; PUSHMARK (SP); 439 ENTER; SAVETMPS; PUSHMARK (SP);
388 // we re-bless the reference to get overload and other niceties right 440 // we re-bless the reference to get overload and other niceties right
420 // catch this surprisingly common error 472 // catch this surprisingly common error
421 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv) 473 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)); 474 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));
423 475
424 encode_tag (enc, CBOR_TAG_PERL_OBJECT); 476 encode_tag (enc, CBOR_TAG_PERL_OBJECT);
425 encode_uint (enc, 0x80, count + 1); 477 encode_uint (enc, MAJOR_ARRAY, count + 1);
426 encode_strref (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash)); 478 encode_strref (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
427 479
428 while (count) 480 while (count)
429 encode_sv (enc, SP[1 - count--]); 481 encode_sv (enc, SP[1 - count--]);
430 482
452{ 504{
453 double nv = SvNVX (sv); 505 double nv = SvNVX (sv);
454 506
455 need (enc, 9); 507 need (enc, 9);
456 508
457 if (ecb_expect_false (nv == (U32)nv)) 509 if (ecb_expect_false (nv == (NV)(U32)nv))
458 encode_uint (enc, 0x00, (U32)nv); 510 encode_uint (enc, MAJOR_POS_INT, (U32)nv);
459 //TODO: maybe I32? 511 //TODO: maybe I32?
460 else if (ecb_expect_false (nv == (float)nv)) 512 else if (ecb_expect_false (nv == (float)nv))
461 { 513 {
462 uint32_t fp = ecb_float_to_binary32 (nv); 514 uint32_t fp = ecb_float_to_binary32 (nv);
463 515
464 *enc->cur++ = 0xe0 | 26; 516 *enc->cur++ = MAJOR_MISC | MISC_FLOAT32;
465 517
466 if (!ecb_big_endian ()) 518 if (!ecb_big_endian ())
467 fp = ecb_bswap32 (fp); 519 fp = ecb_bswap32 (fp);
468 520
469 memcpy (enc->cur, &fp, 4); 521 memcpy (enc->cur, &fp, 4);
471 } 523 }
472 else 524 else
473 { 525 {
474 uint64_t fp = ecb_double_to_binary64 (nv); 526 uint64_t fp = ecb_double_to_binary64 (nv);
475 527
476 *enc->cur++ = 0xe0 | 27; 528 *enc->cur++ = MAJOR_MISC | MISC_FLOAT64;
477 529
478 if (!ecb_big_endian ()) 530 if (!ecb_big_endian ())
479 fp = ecb_bswap64 (fp); 531 fp = ecb_bswap64 (fp);
480 532
481 memcpy (enc->cur, &fp, 8); 533 memcpy (enc->cur, &fp, 8);
497 else if (SvNOKp (sv)) 549 else if (SvNOKp (sv))
498 encode_nv (enc, sv); 550 encode_nv (enc, sv);
499 else if (SvIOKp (sv)) 551 else if (SvIOKp (sv))
500 { 552 {
501 if (SvIsUV (sv)) 553 if (SvIsUV (sv))
502 encode_uint (enc, 0x00, SvUVX (sv)); 554 encode_uint (enc, MAJOR_POS_INT, SvUVX (sv));
503 else if (SvIVX (sv) >= 0) 555 else if (SvIVX (sv) >= 0)
504 encode_uint (enc, 0x00, SvIVX (sv)); 556 encode_uint (enc, MAJOR_POS_INT, SvIVX (sv));
505 else 557 else
506 encode_uint (enc, 0x20, -(SvIVX (sv) + 1)); 558 encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1));
507 } 559 }
508 else if (SvROK (sv)) 560 else if (SvROK (sv))
509 encode_rv (enc, SvRV (sv)); 561 encode_rv (enc, SvRV (sv));
510 else if (!SvOK (sv)) 562 else if (!SvOK (sv))
511 encode_ch (enc, 0xe0 | 22); 563 encode_ch (enc, MAJOR_MISC | SIMPLE_NULL);
512 else if (enc->cbor.flags & F_ALLOW_UNKNOWN) 564 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
513 encode_ch (enc, 0xe0 | 23); 565 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
514 else 566 else
515 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data", 567 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data",
516 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv)); 568 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
517} 569}
518 570
571#define DEC_DEC_DEPTH --dec->depth 623#define DEC_DEC_DEPTH --dec->depth
572 624
573static UV 625static UV
574decode_uint (dec_t *dec) 626decode_uint (dec_t *dec)
575{ 627{
576 switch (*dec->cur & 31) 628 U8 m = *dec->cur & MINOR_MASK;
577 { 629 ++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;
582 630
583 case 24: 631 if (ecb_expect_true (m < LENGTH_EXT1))
632 return m;
633 else if (ecb_expect_true (m == LENGTH_EXT1))
634 {
584 WANT (2); 635 WANT (1);
585 dec->cur += 2; 636 dec->cur += 1;
586 return dec->cur[-1]; 637 return dec->cur[-1];
587 638 }
588 case 25: 639 else if (ecb_expect_true (m == LENGTH_EXT2))
640 {
589 WANT (3); 641 WANT (2);
590 dec->cur += 3; 642 dec->cur += 2;
591 return (((UV)dec->cur[-2]) << 8) 643 return (((UV)dec->cur[-2]) << 8)
592 | ((UV)dec->cur[-1]); 644 | ((UV)dec->cur[-1]);
593 645 }
594 case 26: 646 else if (ecb_expect_true (m == LENGTH_EXT4))
647 {
595 WANT (5); 648 WANT (4);
596 dec->cur += 5; 649 dec->cur += 4;
597 return (((UV)dec->cur[-4]) << 24) 650 return (((UV)dec->cur[-4]) << 24)
598 | (((UV)dec->cur[-3]) << 16) 651 | (((UV)dec->cur[-3]) << 16)
599 | (((UV)dec->cur[-2]) << 8) 652 | (((UV)dec->cur[-2]) << 8)
600 | ((UV)dec->cur[-1]); 653 | ((UV)dec->cur[-1]);
601 654 }
602 case 27: 655 else if (ecb_expect_true (m == LENGTH_EXT8))
656 {
603 WANT (9); 657 WANT (8);
604 dec->cur += 9; 658 dec->cur += 8;
659
660 return
661#if UVSIZE < 8
662 0
663#else
605 return (((UV)dec->cur[-8]) << 56) 664 (((UV)dec->cur[-8]) << 56)
606 | (((UV)dec->cur[-7]) << 48) 665 | (((UV)dec->cur[-7]) << 48)
607 | (((UV)dec->cur[-6]) << 40) 666 | (((UV)dec->cur[-6]) << 40)
608 | (((UV)dec->cur[-5]) << 32) 667 | (((UV)dec->cur[-5]) << 32)
668#endif
609 | (((UV)dec->cur[-4]) << 24) 669 | (((UV)dec->cur[-4]) << 24)
610 | (((UV)dec->cur[-3]) << 16) 670 | (((UV)dec->cur[-3]) << 16)
611 | (((UV)dec->cur[-2]) << 8) 671 | (((UV)dec->cur[-2]) << 8)
612 | ((UV)dec->cur[-1]); 672 | ((UV)dec->cur[-1]);
613 673 }
614 default: 674 else
615 ERR ("corrupted CBOR data (unsupported integer minor encoding)"); 675 ERR ("corrupted CBOR data (unsupported integer minor encoding)");
616 }
617 676
618fail: 677fail:
619 return 0; 678 return 0;
620} 679}
621 680
626{ 685{
627 AV *av = newAV (); 686 AV *av = newAV ();
628 687
629 DEC_INC_DEPTH; 688 DEC_INC_DEPTH;
630 689
631 if ((*dec->cur & 31) == 31) 690 if (*dec->cur == (MAJOR_ARRAY | MINOR_INDEF))
632 { 691 {
633 ++dec->cur; 692 ++dec->cur;
634 693
635 for (;;) 694 for (;;)
636 { 695 {
637 WANT (1); 696 WANT (1);
638 697
639 if (*dec->cur == (0xe0 | 31)) 698 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
640 { 699 {
641 ++dec->cur; 700 ++dec->cur;
642 break; 701 break;
643 } 702 }
644 703
647 } 706 }
648 else 707 else
649 { 708 {
650 int i, len = decode_uint (dec); 709 int i, len = decode_uint (dec);
651 710
711 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); 712 av_fill (av, len - 1);
653 713
654 for (i = 0; i < len; ++i) 714 for (i = 0; i < len; ++i)
655 AvARRAY (av)[i] = decode_sv (dec); 715 AvARRAY (av)[i] = decode_sv (dec);
656 } 716 }
669{ 729{
670 // for speed reasons, we specialcase single-string 730 // for speed reasons, we specialcase single-string
671 // byte or utf-8 strings as keys, but only when !stringref 731 // byte or utf-8 strings as keys, but only when !stringref
672 732
673 if (ecb_expect_true (!dec->stringref)) 733 if (ecb_expect_true (!dec->stringref))
674 if (*dec->cur >= 0x40 && *dec->cur <= 0x40 + 27) 734 if (ecb_expect_true ((*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8))
675 { 735 {
676 I32 len = decode_uint (dec); 736 I32 len = decode_uint (dec);
677 char *key = (char *)dec->cur; 737 char *key = (char *)dec->cur;
678 738
679 dec->cur += len; 739 dec->cur += len;
683 743
684 hv_store (hv, key, len, decode_sv (dec), 0); 744 hv_store (hv, key, len, decode_sv (dec), 0);
685 745
686 return; 746 return;
687 } 747 }
688 else if (*dec->cur >= 0x60 && *dec->cur <= 0x60 + 27) 748 else if (ecb_expect_true ((*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8))
689 { 749 {
690 I32 len = decode_uint (dec); 750 I32 len = decode_uint (dec);
691 char *key = (char *)dec->cur; 751 char *key = (char *)dec->cur;
692 752
693 dec->cur += len; 753 dec->cur += len;
712{ 772{
713 HV *hv = newHV (); 773 HV *hv = newHV ();
714 774
715 DEC_INC_DEPTH; 775 DEC_INC_DEPTH;
716 776
717 if ((*dec->cur & 31) == 31) 777 if (*dec->cur == (MAJOR_MAP | MINOR_INDEF))
718 { 778 {
719 ++dec->cur; 779 ++dec->cur;
720 780
721 for (;;) 781 for (;;)
722 { 782 {
723 WANT (1); 783 WANT (1);
724 784
725 if (*dec->cur == (0xe0 | 31)) 785 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
726 { 786 {
727 ++dec->cur; 787 ++dec->cur;
728 break; 788 break;
729 } 789 }
730 790
751static SV * 811static SV *
752decode_str (dec_t *dec, int utf8) 812decode_str (dec_t *dec, int utf8)
753{ 813{
754 SV *sv = 0; 814 SV *sv = 0;
755 815
756 if ((*dec->cur & 31) == 31) 816 if ((*dec->cur & MINOR_MASK) == MINOR_INDEF)
757 { 817 {
818 // indefinite length strings
758 ++dec->cur; 819 ++dec->cur;
759 820
821 U8 major = *dec->cur & MAJOR_MISC;
822
760 sv = newSVpvn ("", 0); 823 sv = newSVpvn ("", 0);
761 824
762 // not very fast, and certainly not robust against illegal input
763 for (;;) 825 for (;;)
764 { 826 {
765 WANT (1); 827 WANT (1);
766 828
767 if (*dec->cur == (0xe0 | 31)) 829 if ((*dec->cur - major) > LENGTH_EXT8)
830 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
768 { 831 {
769 ++dec->cur; 832 ++dec->cur;
770 break; 833 break;
771 } 834 }
835 else
836 ERR ("corrupted CBOR data (invalid chunks in indefinite length string)");
772 837
773 sv_catsv (sv, decode_sv (dec)); 838 STRLEN len = decode_uint (dec);
839
840 WANT (len);
841 sv_catpvn (sv, dec->cur, len);
842 dec->cur += len;
774 } 843 }
775 } 844 }
776 else 845 else
777 { 846 {
778 STRLEN len = decode_uint (dec); 847 STRLEN len = decode_uint (dec);
827 } 896 }
828 break; 897 break;
829 898
830 case CBOR_TAG_STRINGREF: 899 case CBOR_TAG_STRINGREF:
831 { 900 {
832 if ((*dec->cur >> 5) != 0) 901 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
833 ERR ("corrupted CBOR data (stringref index not an unsigned integer)"); 902 ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
834 903
835 UV idx = decode_uint (dec); 904 UV idx = decode_uint (dec);
836 905
837 if (!dec->stringref || (int)idx > AvFILLp (dec->stringref)) 906 if (!dec->stringref || (int)idx > AvFILLp (dec->stringref))
844 case CBOR_TAG_VALUE_SHAREABLE: 913 case CBOR_TAG_VALUE_SHAREABLE:
845 { 914 {
846 if (ecb_expect_false (!dec->shareable)) 915 if (ecb_expect_false (!dec->shareable))
847 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ()); 916 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ());
848 917
918 if (dec->cbor.flags & F_ALLOW_CYCLES)
919 {
849 sv = newSV (0); 920 sv = newSV (0);
850 av_push (dec->shareable, SvREFCNT_inc_NN (sv)); 921 av_push (dec->shareable, SvREFCNT_inc_NN (sv));
851 922
852 SV *osv = decode_sv (dec); 923 SV *osv = decode_sv (dec);
853 sv_setsv (sv, osv); 924 sv_setsv (sv, osv);
854 SvREFCNT_dec_NN (osv); 925 SvREFCNT_dec_NN (osv);
926 }
927 else
928 {
929 av_push (dec->shareable, &PL_sv_undef);
930 int idx = AvFILLp (dec->shareable);
931 sv = decode_sv (dec);
932 av_store (dec->shareable, idx, SvREFCNT_inc_NN (sv));
933 }
855 } 934 }
856 break; 935 break;
857 936
858 case CBOR_TAG_VALUE_SHAREDREF: 937 case CBOR_TAG_VALUE_SHAREDREF:
859 { 938 {
860 if ((*dec->cur >> 5) != 0) 939 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
861 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)"); 940 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
862 941
863 UV idx = decode_uint (dec); 942 UV idx = decode_uint (dec);
864 943
865 if (!dec->shareable || (int)idx > AvFILLp (dec->shareable)) 944 if (!dec->shareable || (int)idx > AvFILLp (dec->shareable))
866 ERR ("corrupted CBOR data (sharedref index out of bounds)"); 945 ERR ("corrupted CBOR data (sharedref index out of bounds)");
867 946
868 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]); 947 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]);
948
949 if (sv == &PL_sv_undef)
950 ERR ("cyclic CBOR data structure found, but allow_cycles is not enabled");
869 } 951 }
870 break; 952 break;
871 953
872 case CBOR_TAG_PERL_OBJECT: 954 case CBOR_TAG_PERL_OBJECT:
873 { 955 {
974static SV * 1056static SV *
975decode_sv (dec_t *dec) 1057decode_sv (dec_t *dec)
976{ 1058{
977 WANT (1); 1059 WANT (1);
978 1060
979 switch (*dec->cur >> 5) 1061 switch (*dec->cur >> MAJOR_SHIFT)
980 { 1062 {
981 case 0: // unsigned int 1063 case MAJOR_POS_INT >> MAJOR_SHIFT: return newSVuv (decode_uint (dec));
982 return newSVuv (decode_uint (dec)); 1064 case MAJOR_NEG_INT >> MAJOR_SHIFT: return newSViv (-1 - (IV)decode_uint (dec));
983 case 1: // negative int 1065 case MAJOR_BYTES >> MAJOR_SHIFT: return decode_str (dec, 0);
984 return newSViv (-1 - (IV)decode_uint (dec)); 1066 case MAJOR_TEXT >> MAJOR_SHIFT: return decode_str (dec, 1);
985 case 2: // octet string 1067 case MAJOR_ARRAY >> MAJOR_SHIFT: return decode_av (dec);
986 return decode_str (dec, 0); 1068 case MAJOR_MAP >> MAJOR_SHIFT: return decode_hv (dec);
987 case 3: // utf-8 string 1069 case MAJOR_TAG >> MAJOR_SHIFT: return decode_tagged (dec);
988 return decode_str (dec, 1); 1070
989 case 4: // array 1071 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) 1072 switch (*dec->cur++ & MINOR_MASK)
997 { 1073 {
998 case 20: 1074 case SIMPLE_FALSE:
999#if CBOR_SLOW 1075#if CBOR_SLOW
1000 types_false = get_bool ("Types::Serialiser::false"); 1076 types_false = get_bool ("Types::Serialiser::false");
1001#endif 1077#endif
1002 return newSVsv (types_false); 1078 return newSVsv (types_false);
1003 case 21: 1079 case SIMPLE_TRUE:
1004#if CBOR_SLOW 1080#if CBOR_SLOW
1005 types_true = get_bool ("Types::Serialiser::true"); 1081 types_true = get_bool ("Types::Serialiser::true");
1006#endif 1082#endif
1007 return newSVsv (types_true); 1083 return newSVsv (types_true);
1008 case 22: 1084 case SIMPLE_NULL:
1009 return newSVsv (&PL_sv_undef); 1085 return newSVsv (&PL_sv_undef);
1010 case 23: 1086 case SIMPLE_UNDEF:
1011#if CBOR_SLOW 1087#if CBOR_SLOW
1012 types_error = get_bool ("Types::Serialiser::error"); 1088 types_error = get_bool ("Types::Serialiser::error");
1013#endif 1089#endif
1014 return newSVsv (types_error); 1090 return newSVsv (types_error);
1015 1091
1016 case 25: 1092 case MISC_FLOAT16:
1017 { 1093 {
1018 WANT (2); 1094 WANT (2);
1019 1095
1020 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1]; 1096 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1];
1021 dec->cur += 2; 1097 dec->cur += 2;
1022 1098
1023 return newSVnv (ecb_binary16_to_float (fp)); 1099 return newSVnv (ecb_binary16_to_float (fp));
1024 } 1100 }
1025 1101
1026 case 26: 1102 case MISC_FLOAT32:
1027 { 1103 {
1028 uint32_t fp; 1104 uint32_t fp;
1029 WANT (4); 1105 WANT (4);
1030 memcpy (&fp, dec->cur, 4); 1106 memcpy (&fp, dec->cur, 4);
1031 dec->cur += 4; 1107 dec->cur += 4;
1034 fp = ecb_bswap32 (fp); 1110 fp = ecb_bswap32 (fp);
1035 1111
1036 return newSVnv (ecb_binary32_to_float (fp)); 1112 return newSVnv (ecb_binary32_to_float (fp));
1037 } 1113 }
1038 1114
1039 case 27: 1115 case MISC_FLOAT64:
1040 { 1116 {
1041 uint64_t fp; 1117 uint64_t fp;
1042 WANT (8); 1118 WANT (8);
1043 memcpy (&fp, dec->cur, 8); 1119 memcpy (&fp, dec->cur, 8);
1044 dec->cur += 8; 1120 dec->cur += 8;
1047 fp = ecb_bswap64 (fp); 1123 fp = ecb_bswap64 (fp);
1048 1124
1049 return newSVnv (ecb_binary64_to_double (fp)); 1125 return newSVnv (ecb_binary64_to_double (fp));
1050 } 1126 }
1051 1127
1052 // 0..19 unassigned 1128 // 0..19 unassigned simple
1053 // 24 reserved + unassigned (reserved values are not encodable) 1129 // 24 reserved + unassigned (reserved values are not encodable)
1054 default: 1130 default:
1055 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)"); 1131 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)");
1056 } 1132 }
1057 1133
1145void shrink (CBOR *self, int enable = 1) 1221void shrink (CBOR *self, int enable = 1)
1146 ALIAS: 1222 ALIAS:
1147 shrink = F_SHRINK 1223 shrink = F_SHRINK
1148 allow_unknown = F_ALLOW_UNKNOWN 1224 allow_unknown = F_ALLOW_UNKNOWN
1149 allow_sharing = F_ALLOW_SHARING 1225 allow_sharing = F_ALLOW_SHARING
1226 allow_cycles = F_ALLOW_CYCLES
1150 pack_strings = F_PACK_STRINGS 1227 pack_strings = F_PACK_STRINGS
1151 PPCODE: 1228 PPCODE:
1152{ 1229{
1153 if (enable) 1230 if (enable)
1154 self->flags |= ix; 1231 self->flags |= ix;
1161void get_shrink (CBOR *self) 1238void get_shrink (CBOR *self)
1162 ALIAS: 1239 ALIAS:
1163 get_shrink = F_SHRINK 1240 get_shrink = F_SHRINK
1164 get_allow_unknown = F_ALLOW_UNKNOWN 1241 get_allow_unknown = F_ALLOW_UNKNOWN
1165 get_allow_sharing = F_ALLOW_SHARING 1242 get_allow_sharing = F_ALLOW_SHARING
1243 get_allow_cycles = F_ALLOW_CYCLES
1166 get_pack_strings = F_PACK_STRINGS 1244 get_pack_strings = F_PACK_STRINGS
1167 PPCODE: 1245 PPCODE:
1168 XPUSHs (boolSV (self->flags & ix)); 1246 XPUSHs (boolSV (self->flags & ix));
1169 1247
1170void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) 1248void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
1227 cbor_free (self); 1305 cbor_free (self);
1228 1306
1229PROTOTYPES: ENABLE 1307PROTOTYPES: ENABLE
1230 1308
1231void encode_cbor (SV *scalar) 1309void encode_cbor (SV *scalar)
1310 ALIAS:
1311 encode_cbor = 0
1312 encode_cbor_sharing = F_ALLOW_SHARING
1232 PPCODE: 1313 PPCODE:
1233{ 1314{
1234 CBOR cbor; 1315 CBOR cbor;
1235 cbor_init (&cbor); 1316 cbor_init (&cbor);
1317 cbor.flags |= ix;
1236 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN; 1318 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;
1237 XPUSHs (scalar); 1319 XPUSHs (scalar);
1238} 1320}
1239 1321
1240void decode_cbor (SV *cborstr) 1322void decode_cbor (SV *cborstr)

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines