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 1.30 by root, Thu Nov 28 11:36:53 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 //TODO 101#define F_ALLOW_SHARING 0x00000004UL
65#define F_ALLOW_STRINGREF 0x00000008UL //TODO 102#define F_ALLOW_CYCLES 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 (len < 24) 233 if (ecb_expect_true (len < LENGTH_EXT1))
195 *enc->cur++ = major | len; 234 *enc->cur++ = major | len;
196 else if (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
244static void 283static void
245encode_strref (enc_t *enc, int utf8, char *str, STRLEN len) 284encode_strref (enc_t *enc, int utf8, char *str, STRLEN len)
246{ 285{
247 if (ecb_expect_false (enc->cbor.flags & F_ALLOW_STRINGREF)) 286 if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS))
248 { 287 {
249 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1); 288 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1);
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.
327static void 366static void
328encode_rv (enc_t *enc, SV *sv) 367encode_rv (enc_t *enc, SV *sv)
329{ 368{
330 SvGETMAGIC (sv); 369 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 }
353 370
354 svtype svt = SvTYPE (sv); 371 svtype svt = SvTYPE (sv);
355 372
356 if (ecb_expect_false (SvOBJECT (sv))) 373 if (ecb_expect_false (SvOBJECT (sv)))
357 { 374 {
… …
364 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 381 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
365 ? cbor_tagged_stash 382 ? cbor_tagged_stash
366 : gv_stashpv ("CBOR::XS::Tagged" , 1); 383 : gv_stashpv ("CBOR::XS::Tagged" , 1);
367 384
368 HV *stash = SvSTASH (sv); 385 HV *stash = SvSTASH (sv);
369 GV *method;
370 386
371 if (stash == boolean_stash) 387 if (stash == boolean_stash)
372 encode_ch (enc, SvIV (sv) ? 0xe0 | 21 : 0xe0 | 20); 388 {
389 encode_ch (enc, SvIV (sv) ? MAJOR_MISC | SIMPLE_TRUE : MAJOR_MISC | SIMPLE_FALSE);
390 return;
391 }
373 else if (stash == error_stash) 392 else if (stash == error_stash)
374 encode_ch (enc, 0xe0 | 23); 393 {
394 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
395 return;
396 }
375 else if (stash == tagged_stash) 397 else if (stash == tagged_stash)
376 { 398 {
377 if (svt != SVt_PVAV) 399 if (svt != SVt_PVAV)
378 croak ("encountered CBOR::XS::Tagged object that isn't an array"); 400 croak ("encountered CBOR::XS::Tagged object that isn't an array");
379 401
380 encode_uint (enc, 0xc0, SvUV (*av_fetch ((AV *)sv, 0, 1))); 402 encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1)));
381 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1)); 403 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1));
404
405 return;
406 }
407 }
408
409 if (ecb_expect_false (SvREFCNT (sv) > 1)
410 && ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING))
411 {
412 if (!enc->shareable)
413 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
414
415 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
416
417 if (SvOK (*svp))
382 } 418 {
419 encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF);
420 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
421 return;
422 }
423 else
424 {
425 sv_setuv (*svp, enc->shareable_idx);
426 ++enc->shareable_idx;
427 encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE);
428 }
429 }
430
431 if (ecb_expect_false (SvOBJECT (sv)))
432 {
433 HV *stash = SvSTASH (sv);
434 GV *method;
435
383 else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0))) 436 if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
384 { 437 {
385 dSP; 438 dSP;
386 439
387 ENTER; SAVETMPS; PUSHMARK (SP); 440 ENTER; SAVETMPS; PUSHMARK (SP);
388 // we re-bless the reference to get overload and other niceties right 441 // we re-bless the reference to get overload and other niceties right
… …
420 // catch this surprisingly common error 473 // catch this surprisingly common error
421 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv) 474 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)); 475 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));
423 476
424 encode_tag (enc, CBOR_TAG_PERL_OBJECT); 477 encode_tag (enc, CBOR_TAG_PERL_OBJECT);
425 encode_uint (enc, 0x80, count + 1); 478 encode_uint (enc, MAJOR_ARRAY, count + 1);
426 encode_strref (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash)); 479 encode_strref (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
427 480
428 while (count) 481 while (count)
429 encode_sv (enc, SP[1 - count--]); 482 encode_sv (enc, SP[1 - count--]);
430 483
… …
452{ 505{
453 double nv = SvNVX (sv); 506 double nv = SvNVX (sv);
454 507
455 need (enc, 9); 508 need (enc, 9);
456 509
457 if (ecb_expect_false (nv == (U32)nv)) 510 if (ecb_expect_false (nv == (NV)(U32)nv))
458 encode_uint (enc, 0x00, (U32)nv); 511 encode_uint (enc, MAJOR_POS_INT, (U32)nv);
459 //TODO: maybe I32? 512 //TODO: maybe I32?
460 else if (ecb_expect_false (nv == (float)nv)) 513 else if (ecb_expect_false (nv == (float)nv))
461 { 514 {
462 uint32_t fp = ecb_float_to_binary32 (nv); 515 uint32_t fp = ecb_float_to_binary32 (nv);
463 516
464 *enc->cur++ = 0xe0 | 26; 517 *enc->cur++ = MAJOR_MISC | MISC_FLOAT32;
465 518
466 if (!ecb_big_endian ()) 519 if (!ecb_big_endian ())
467 fp = ecb_bswap32 (fp); 520 fp = ecb_bswap32 (fp);
468 521
469 memcpy (enc->cur, &fp, 4); 522 memcpy (enc->cur, &fp, 4);
… …
471 } 524 }
472 else 525 else
473 { 526 {
474 uint64_t fp = ecb_double_to_binary64 (nv); 527 uint64_t fp = ecb_double_to_binary64 (nv);
475 528
476 *enc->cur++ = 0xe0 | 27; 529 *enc->cur++ = MAJOR_MISC | MISC_FLOAT64;
477 530
478 if (!ecb_big_endian ()) 531 if (!ecb_big_endian ())
479 fp = ecb_bswap64 (fp); 532 fp = ecb_bswap64 (fp);
480 533
481 memcpy (enc->cur, &fp, 8); 534 memcpy (enc->cur, &fp, 8);
… …
497 else if (SvNOKp (sv)) 550 else if (SvNOKp (sv))
498 encode_nv (enc, sv); 551 encode_nv (enc, sv);
499 else if (SvIOKp (sv)) 552 else if (SvIOKp (sv))
500 { 553 {
501 if (SvIsUV (sv)) 554 if (SvIsUV (sv))
502 encode_uint (enc, 0x00, SvUVX (sv)); 555 encode_uint (enc, MAJOR_POS_INT, SvUVX (sv));
503 else if (SvIVX (sv) >= 0) 556 else if (SvIVX (sv) >= 0)
504 encode_uint (enc, 0x00, SvIVX (sv)); 557 encode_uint (enc, MAJOR_POS_INT, SvIVX (sv));
505 else 558 else
506 encode_uint (enc, 0x20, -(SvIVX (sv) + 1)); 559 encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1));
507 } 560 }
508 else if (SvROK (sv)) 561 else if (SvROK (sv))
509 encode_rv (enc, SvRV (sv)); 562 encode_rv (enc, SvRV (sv));
510 else if (!SvOK (sv)) 563 else if (!SvOK (sv))
511 encode_ch (enc, 0xe0 | 22); 564 encode_ch (enc, MAJOR_MISC | SIMPLE_NULL);
512 else if (enc->cbor.flags & F_ALLOW_UNKNOWN) 565 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
513 encode_ch (enc, 0xe0 | 23); 566 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
514 else 567 else
515 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",
516 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv)); 569 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
517} 570}
518 571
… …
526 enc.cur = SvPVX (enc.sv); 579 enc.cur = SvPVX (enc.sv);
527 enc.end = SvEND (enc.sv); 580 enc.end = SvEND (enc.sv);
528 581
529 SvPOK_only (enc.sv); 582 SvPOK_only (enc.sv);
530 583
531 if (cbor->flags & F_ALLOW_STRINGREF) 584 if (cbor->flags & F_PACK_STRINGS)
532 { 585 {
533 encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE); 586 encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE);
534 enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ()); 587 enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ());
535 enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ()); 588 enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ());
536 } 589 }
… …
571#define DEC_DEC_DEPTH --dec->depth 624#define DEC_DEC_DEPTH --dec->depth
572 625
573static UV 626static UV
574decode_uint (dec_t *dec) 627decode_uint (dec_t *dec)
575{ 628{
576 switch (*dec->cur & 31) 629 U8 m = *dec->cur & MINOR_MASK;
577 { 630 ++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;
582 631
583 case 24: 632 if (ecb_expect_true (m < LENGTH_EXT1))
633 return m;
634 else if (ecb_expect_true (m == LENGTH_EXT1))
635 {
584 WANT (2); 636 WANT (1);
585 dec->cur += 2; 637 dec->cur += 1;
586 return dec->cur[-1]; 638 return dec->cur[-1];
587 639 }
588 case 25: 640 else if (ecb_expect_true (m == LENGTH_EXT2))
641 {
589 WANT (3); 642 WANT (2);
590 dec->cur += 3; 643 dec->cur += 2;
591 return (((UV)dec->cur[-2]) << 8) 644 return (((UV)dec->cur[-2]) << 8)
592 | ((UV)dec->cur[-1]); 645 | ((UV)dec->cur[-1]);
593 646 }
594 case 26: 647 else if (ecb_expect_true (m == LENGTH_EXT4))
648 {
595 WANT (5); 649 WANT (4);
596 dec->cur += 5; 650 dec->cur += 4;
597 return (((UV)dec->cur[-4]) << 24) 651 return (((UV)dec->cur[-4]) << 24)
598 | (((UV)dec->cur[-3]) << 16) 652 | (((UV)dec->cur[-3]) << 16)
599 | (((UV)dec->cur[-2]) << 8) 653 | (((UV)dec->cur[-2]) << 8)
600 | ((UV)dec->cur[-1]); 654 | ((UV)dec->cur[-1]);
601 655 }
602 case 27: 656 else if (ecb_expect_true (m == LENGTH_EXT8))
657 {
603 WANT (9); 658 WANT (8);
604 dec->cur += 9; 659 dec->cur += 8;
660
661 return
662#if UVSIZE < 8
663 0
664#else
605 return (((UV)dec->cur[-8]) << 56) 665 (((UV)dec->cur[-8]) << 56)
606 | (((UV)dec->cur[-7]) << 48) 666 | (((UV)dec->cur[-7]) << 48)
607 | (((UV)dec->cur[-6]) << 40) 667 | (((UV)dec->cur[-6]) << 40)
608 | (((UV)dec->cur[-5]) << 32) 668 | (((UV)dec->cur[-5]) << 32)
669#endif
609 | (((UV)dec->cur[-4]) << 24) 670 | (((UV)dec->cur[-4]) << 24)
610 | (((UV)dec->cur[-3]) << 16) 671 | (((UV)dec->cur[-3]) << 16)
611 | (((UV)dec->cur[-2]) << 8) 672 | (((UV)dec->cur[-2]) << 8)
612 | ((UV)dec->cur[-1]); 673 | ((UV)dec->cur[-1]);
613 674 }
614 default: 675 else
615 ERR ("corrupted CBOR data (unsupported integer minor encoding)"); 676 ERR ("corrupted CBOR data (unsupported integer minor encoding)");
616 }
617 677
618fail: 678fail:
619 return 0; 679 return 0;
620} 680}
621 681
… …
626{ 686{
627 AV *av = newAV (); 687 AV *av = newAV ();
628 688
629 DEC_INC_DEPTH; 689 DEC_INC_DEPTH;
630 690
631 if ((*dec->cur & 31) == 31) 691 if (*dec->cur == (MAJOR_ARRAY | MINOR_INDEF))
632 { 692 {
633 ++dec->cur; 693 ++dec->cur;
634 694
635 for (;;) 695 for (;;)
636 { 696 {
637 WANT (1); 697 WANT (1);
638 698
639 if (*dec->cur == (0xe0 | 31)) 699 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
640 { 700 {
641 ++dec->cur; 701 ++dec->cur;
642 break; 702 break;
643 } 703 }
644 704
… …
647 } 707 }
648 else 708 else
649 { 709 {
650 int i, len = decode_uint (dec); 710 int i, len = decode_uint (dec);
651 711
712 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); 713 av_fill (av, len - 1);
653 714
654 for (i = 0; i < len; ++i) 715 for (i = 0; i < len; ++i)
655 AvARRAY (av)[i] = decode_sv (dec); 716 AvARRAY (av)[i] = decode_sv (dec);
656 } 717 }
… …
669{ 730{
670 // for speed reasons, we specialcase single-string 731 // for speed reasons, we specialcase single-string
671 // byte or utf-8 strings as keys, but only when !stringref 732 // byte or utf-8 strings as keys, but only when !stringref
672 733
673 if (ecb_expect_true (!dec->stringref)) 734 if (ecb_expect_true (!dec->stringref))
674 if (*dec->cur >= 0x40 && *dec->cur <= 0x40 + 27) 735 if (ecb_expect_true ((*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8))
675 { 736 {
676 I32 len = decode_uint (dec); 737 I32 len = decode_uint (dec);
677 char *key = (char *)dec->cur; 738 char *key = (char *)dec->cur;
678 739
679 dec->cur += len; 740 dec->cur += len;
680 741
681 if (ecb_expect_false (dec->stringref))
682 av_push (dec->stringref, newSVpvn (key, len));
683
684 hv_store (hv, key, len, decode_sv (dec), 0); 742 hv_store (hv, key, len, decode_sv (dec), 0);
685 743
686 return; 744 return;
687 } 745 }
688 else if (*dec->cur >= 0x60 && *dec->cur <= 0x60 + 27) 746 else if (ecb_expect_true ((*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8))
689 { 747 {
690 I32 len = decode_uint (dec); 748 I32 len = decode_uint (dec);
691 char *key = (char *)dec->cur; 749 char *key = (char *)dec->cur;
692 750
693 dec->cur += len; 751 dec->cur += len;
694 752
695 if (ecb_expect_false (dec->stringref)) 753 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
696 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)");
697 756
698 hv_store (hv, key, -len, decode_sv (dec), 0); 757 hv_store (hv, key, -len, decode_sv (dec), 0);
699 758
700 return; 759 return;
701 } 760 }
… …
703 SV *k = decode_sv (dec); 762 SV *k = decode_sv (dec);
704 SV *v = decode_sv (dec); 763 SV *v = decode_sv (dec);
705 764
706 hv_store_ent (hv, k, v, 0); 765 hv_store_ent (hv, k, v, 0);
707 SvREFCNT_dec (k); 766 SvREFCNT_dec (k);
767
768fail:
769 ;
708} 770}
709 771
710static SV * 772static SV *
711decode_hv (dec_t *dec) 773decode_hv (dec_t *dec)
712{ 774{
713 HV *hv = newHV (); 775 HV *hv = newHV ();
714 776
715 DEC_INC_DEPTH; 777 DEC_INC_DEPTH;
716 778
717 if ((*dec->cur & 31) == 31) 779 if (*dec->cur == (MAJOR_MAP | MINOR_INDEF))
718 { 780 {
719 ++dec->cur; 781 ++dec->cur;
720 782
721 for (;;) 783 for (;;)
722 { 784 {
723 WANT (1); 785 WANT (1);
724 786
725 if (*dec->cur == (0xe0 | 31)) 787 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
726 { 788 {
727 ++dec->cur; 789 ++dec->cur;
728 break; 790 break;
729 } 791 }
730 792
… …
751static SV * 813static SV *
752decode_str (dec_t *dec, int utf8) 814decode_str (dec_t *dec, int utf8)
753{ 815{
754 SV *sv = 0; 816 SV *sv = 0;
755 817
756 if ((*dec->cur & 31) == 31) 818 if ((*dec->cur & MINOR_MASK) == MINOR_INDEF)
757 { 819 {
820 // indefinite length strings
758 ++dec->cur; 821 ++dec->cur;
759 822
823 U8 major = *dec->cur & MAJOR_MISC;
824
760 sv = newSVpvn ("", 0); 825 sv = newSVpvn ("", 0);
761 826
762 // not very fast, and certainly not robust against illegal input
763 for (;;) 827 for (;;)
764 { 828 {
765 WANT (1); 829 WANT (1);
766 830
767 if (*dec->cur == (0xe0 | 31)) 831 if ((*dec->cur - major) > LENGTH_EXT8)
832 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
768 { 833 {
769 ++dec->cur; 834 ++dec->cur;
770 break; 835 break;
771 } 836 }
837 else
838 ERR ("corrupted CBOR data (invalid chunks in indefinite length string)");
772 839
773 sv_catsv (sv, decode_sv (dec)); 840 STRLEN len = decode_uint (dec);
841
842 WANT (len);
843 sv_catpvn (sv, dec->cur, len);
844 dec->cur += len;
774 } 845 }
775 } 846 }
776 else 847 else
777 { 848 {
778 STRLEN len = decode_uint (dec); 849 STRLEN len = decode_uint (dec);
… …
785 && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1)) 856 && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1))
786 av_push (dec->stringref, SvREFCNT_inc_NN (sv)); 857 av_push (dec->stringref, SvREFCNT_inc_NN (sv));
787 } 858 }
788 859
789 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
790 SvUTF8_on (sv); 866 SvUTF8_on (sv);
867 }
791 868
792 return sv; 869 return sv;
793 870
794fail: 871fail:
795 SvREFCNT_dec (sv); 872 SvREFCNT_dec (sv);
… …
827 } 904 }
828 break; 905 break;
829 906
830 case CBOR_TAG_STRINGREF: 907 case CBOR_TAG_STRINGREF:
831 { 908 {
832 if ((*dec->cur >> 5) != 0) 909 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
833 ERR ("corrupted CBOR data (stringref index not an unsigned integer)"); 910 ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
834 911
835 UV idx = decode_uint (dec); 912 UV idx = decode_uint (dec);
836 913
837 if (!dec->stringref || (int)idx > AvFILLp (dec->stringref)) 914 if (!dec->stringref || (int)idx > AvFILLp (dec->stringref))
… …
844 case CBOR_TAG_VALUE_SHAREABLE: 921 case CBOR_TAG_VALUE_SHAREABLE:
845 { 922 {
846 if (ecb_expect_false (!dec->shareable)) 923 if (ecb_expect_false (!dec->shareable))
847 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ()); 924 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ());
848 925
926 if (dec->cbor.flags & F_ALLOW_CYCLES)
927 {
849 sv = newSV (0); 928 sv = newSV (0);
850 av_push (dec->shareable, SvREFCNT_inc_NN (sv)); 929 av_push (dec->shareable, SvREFCNT_inc_NN (sv));
851 930
852 SV *osv = decode_sv (dec); 931 SV *osv = decode_sv (dec);
853 sv_setsv (sv, osv); 932 sv_setsv (sv, osv);
854 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 }
855 } 942 }
856 break; 943 break;
857 944
858 case CBOR_TAG_VALUE_SHAREDREF: 945 case CBOR_TAG_VALUE_SHAREDREF:
859 { 946 {
860 if ((*dec->cur >> 5) != 0) 947 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
861 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)"); 948 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
862 949
863 UV idx = decode_uint (dec); 950 UV idx = decode_uint (dec);
864 951
865 if (!dec->shareable || (int)idx > AvFILLp (dec->shareable)) 952 if (!dec->shareable || (int)idx > AvFILLp (dec->shareable))
866 ERR ("corrupted CBOR data (sharedref index out of bounds)"); 953 ERR ("corrupted CBOR data (sharedref index out of bounds)");
867 954
868 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");
869 } 959 }
870 break; 960 break;
871 961
872 case CBOR_TAG_PERL_OBJECT: 962 case CBOR_TAG_PERL_OBJECT:
873 { 963 {
… …
974static SV * 1064static SV *
975decode_sv (dec_t *dec) 1065decode_sv (dec_t *dec)
976{ 1066{
977 WANT (1); 1067 WANT (1);
978 1068
979 switch (*dec->cur >> 5) 1069 switch (*dec->cur >> MAJOR_SHIFT)
980 { 1070 {
981 case 0: // unsigned int 1071 case MAJOR_POS_INT >> MAJOR_SHIFT: return newSVuv (decode_uint (dec));
982 return newSVuv (decode_uint (dec)); 1072 case MAJOR_NEG_INT >> MAJOR_SHIFT: return newSViv (-1 - (IV)decode_uint (dec));
983 case 1: // negative int 1073 case MAJOR_BYTES >> MAJOR_SHIFT: return decode_str (dec, 0);
984 return newSViv (-1 - (IV)decode_uint (dec)); 1074 case MAJOR_TEXT >> MAJOR_SHIFT: return decode_str (dec, 1);
985 case 2: // octet string 1075 case MAJOR_ARRAY >> MAJOR_SHIFT: return decode_av (dec);
986 return decode_str (dec, 0); 1076 case MAJOR_MAP >> MAJOR_SHIFT: return decode_hv (dec);
987 case 3: // utf-8 string 1077 case MAJOR_TAG >> MAJOR_SHIFT: return decode_tagged (dec);
988 return decode_str (dec, 1); 1078
989 case 4: // array 1079 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) 1080 switch (*dec->cur++ & MINOR_MASK)
997 { 1081 {
998 case 20: 1082 case SIMPLE_FALSE:
999#if CBOR_SLOW 1083#if CBOR_SLOW
1000 types_false = get_bool ("Types::Serialiser::false"); 1084 types_false = get_bool ("Types::Serialiser::false");
1001#endif 1085#endif
1002 return newSVsv (types_false); 1086 return newSVsv (types_false);
1003 case 21: 1087 case SIMPLE_TRUE:
1004#if CBOR_SLOW 1088#if CBOR_SLOW
1005 types_true = get_bool ("Types::Serialiser::true"); 1089 types_true = get_bool ("Types::Serialiser::true");
1006#endif 1090#endif
1007 return newSVsv (types_true); 1091 return newSVsv (types_true);
1008 case 22: 1092 case SIMPLE_NULL:
1009 return newSVsv (&PL_sv_undef); 1093 return newSVsv (&PL_sv_undef);
1010 case 23: 1094 case SIMPLE_UNDEF:
1011#if CBOR_SLOW 1095#if CBOR_SLOW
1012 types_error = get_bool ("Types::Serialiser::error"); 1096 types_error = get_bool ("Types::Serialiser::error");
1013#endif 1097#endif
1014 return newSVsv (types_error); 1098 return newSVsv (types_error);
1015 1099
1016 case 25: 1100 case MISC_FLOAT16:
1017 { 1101 {
1018 WANT (2); 1102 WANT (2);
1019 1103
1020 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1]; 1104 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1];
1021 dec->cur += 2; 1105 dec->cur += 2;
1022 1106
1023 return newSVnv (ecb_binary16_to_float (fp)); 1107 return newSVnv (ecb_binary16_to_float (fp));
1024 } 1108 }
1025 1109
1026 case 26: 1110 case MISC_FLOAT32:
1027 { 1111 {
1028 uint32_t fp; 1112 uint32_t fp;
1029 WANT (4); 1113 WANT (4);
1030 memcpy (&fp, dec->cur, 4); 1114 memcpy (&fp, dec->cur, 4);
1031 dec->cur += 4; 1115 dec->cur += 4;
… …
1034 fp = ecb_bswap32 (fp); 1118 fp = ecb_bswap32 (fp);
1035 1119
1036 return newSVnv (ecb_binary32_to_float (fp)); 1120 return newSVnv (ecb_binary32_to_float (fp));
1037 } 1121 }
1038 1122
1039 case 27: 1123 case MISC_FLOAT64:
1040 { 1124 {
1041 uint64_t fp; 1125 uint64_t fp;
1042 WANT (8); 1126 WANT (8);
1043 memcpy (&fp, dec->cur, 8); 1127 memcpy (&fp, dec->cur, 8);
1044 dec->cur += 8; 1128 dec->cur += 8;
… …
1047 fp = ecb_bswap64 (fp); 1131 fp = ecb_bswap64 (fp);
1048 1132
1049 return newSVnv (ecb_binary64_to_double (fp)); 1133 return newSVnv (ecb_binary64_to_double (fp));
1050 } 1134 }
1051 1135
1052 // 0..19 unassigned 1136 // 0..19 unassigned simple
1053 // 24 reserved + unassigned (reserved values are not encodable) 1137 // 24 reserved + unassigned (reserved values are not encodable)
1054 default: 1138 default:
1055 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)"); 1139 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)");
1056 } 1140 }
1057 1141
… …
1087 if (dec.cur != dec.end && !dec.err) 1171 if (dec.cur != dec.end && !dec.err)
1088 dec.err = "garbage after CBOR object"; 1172 dec.err = "garbage after CBOR object";
1089 1173
1090 if (dec.err) 1174 if (dec.err)
1091 { 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
1092 SvREFCNT_dec (sv); 1187 SvREFCNT_dec (sv);
1093 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);
1094 } 1189 }
1095 1190
1096 sv = sv_2mortal (sv); 1191 sv = sv_2mortal (sv);
… …
1145void shrink (CBOR *self, int enable = 1) 1240void shrink (CBOR *self, int enable = 1)
1146 ALIAS: 1241 ALIAS:
1147 shrink = F_SHRINK 1242 shrink = F_SHRINK
1148 allow_unknown = F_ALLOW_UNKNOWN 1243 allow_unknown = F_ALLOW_UNKNOWN
1149 allow_sharing = F_ALLOW_SHARING 1244 allow_sharing = F_ALLOW_SHARING
1150 allow_stringref = F_ALLOW_STRINGREF 1245 allow_cycles = F_ALLOW_CYCLES
1246 pack_strings = F_PACK_STRINGS
1247 validate_utf8 = F_VALIDATE_UTF8
1151 PPCODE: 1248 PPCODE:
1152{ 1249{
1153 if (enable) 1250 if (enable)
1154 self->flags |= ix; 1251 self->flags |= ix;
1155 else 1252 else
… …
1161void get_shrink (CBOR *self) 1258void get_shrink (CBOR *self)
1162 ALIAS: 1259 ALIAS:
1163 get_shrink = F_SHRINK 1260 get_shrink = F_SHRINK
1164 get_allow_unknown = F_ALLOW_UNKNOWN 1261 get_allow_unknown = F_ALLOW_UNKNOWN
1165 get_allow_sharing = F_ALLOW_SHARING 1262 get_allow_sharing = F_ALLOW_SHARING
1166 get_allow_stringref = F_ALLOW_STRINGREF 1263 get_allow_cycles = F_ALLOW_CYCLES
1264 get_pack_strings = F_PACK_STRINGS
1265 get_validate_utf8 = F_VALIDATE_UTF8
1167 PPCODE: 1266 PPCODE:
1168 XPUSHs (boolSV (self->flags & ix)); 1267 XPUSHs (boolSV (self->flags & ix));
1169 1268
1170void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) 1269void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
1171 PPCODE: 1270 PPCODE:
… …
1227 cbor_free (self); 1326 cbor_free (self);
1228 1327
1229PROTOTYPES: ENABLE 1328PROTOTYPES: ENABLE
1230 1329
1231void encode_cbor (SV *scalar) 1330void encode_cbor (SV *scalar)
1331 ALIAS:
1332 encode_cbor = 0
1333 encode_cbor_sharing = F_ALLOW_SHARING
1232 PPCODE: 1334 PPCODE:
1233{ 1335{
1234 CBOR cbor; 1336 CBOR cbor;
1235 cbor_init (&cbor); 1337 cbor_init (&cbor);
1338 cbor.flags |= ix;
1236 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN; 1339 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;
1237 XPUSHs (scalar); 1340 XPUSHs (scalar);
1238} 1341}
1239 1342
1240void decode_cbor (SV *cborstr) 1343void decode_cbor (SV *cborstr)

Diff Legend

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