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 1.28 by root, Sat Nov 23 18:30:59 2013 UTC vs.
Revision 1.35 by root, Sat Nov 30 17:19:34 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 // inofficial extensions (pending iana registration) 68 // extensions
69 CBOR_TAG_STRINGREF = 25, // http://cbor.schmorp.de/stringref
32 CBOR_TAG_PERL_OBJECT = 24, // http://cbor.schmorp.de/perl-object 70 CBOR_TAG_PERL_OBJECT = 26, // http://cbor.schmorp.de/perl-object
33 CBOR_TAG_GENERIC_OBJECT = 25, // http://cbor.schmorp.de/generic-object 71 CBOR_TAG_GENERIC_OBJECT = 27, // http://cbor.schmorp.de/generic-object
34 CBOR_TAG_VALUE_SHAREABLE = 26, // http://cbor.schmorp.de/value-sharing 72 CBOR_TAG_VALUE_SHAREABLE = 28, // http://cbor.schmorp.de/value-sharing
35 CBOR_TAG_VALUE_SHAREDREF = 27, // http://cbor.schmorp.de/value-sharing 73 CBOR_TAG_VALUE_SHAREDREF = 29, // http://cbor.schmorp.de/value-sharing
36 CBOR_TAG_STRINGREF_NAMESPACE = 65537, // http://cbor.schmorp.de/stringref 74 CBOR_TAG_STRINGREF_NAMESPACE = 256, // http://cbor.schmorp.de/stringref
37 CBOR_TAG_STRINGREF = 28, // 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_PACK_STRINGS 0x00000008UL
66 103
67#define INIT_SIZE 32 // initial scalar size to be allocated 104#define INIT_SIZE 32 // initial scalar size to be allocated
68 105
69#define SB do { 106#define SB do {
70#define SE } while (0) 107#define SE } while (0)
… …
141{ 178{
142 return idx > 23 179 return idx > 23
143 ? idx > 0xffU 180 ? idx > 0xffU
144 ? idx > 0xffffU 181 ? idx > 0xffffU
145 ? idx > 0xffffffffU 182 ? idx > 0xffffffffU
183 ? 11
146 ? 7 184 : 7
147 : 6
148 : 5 185 : 5
149 : 4 186 : 4
150 : 3; 187 : 3;
151} 188}
152 189
… …
189static void 226static void
190encode_uint (enc_t *enc, int major, UV len) 227encode_uint (enc_t *enc, int major, UV len)
191{ 228{
192 need (enc, 9); 229 need (enc, 9);
193 230
194 if (len < 24) 231 if (ecb_expect_true (len < LENGTH_EXT1))
195 *enc->cur++ = major | len; 232 *enc->cur++ = major | len;
196 else if (len <= 0xff) 233 else if (ecb_expect_true (len <= 0xff))
197 { 234 {
198 *enc->cur++ = major | 24; 235 *enc->cur++ = major | LENGTH_EXT1;
199 *enc->cur++ = len; 236 *enc->cur++ = len;
200 } 237 }
201 else if (len <= 0xffff) 238 else if (len <= 0xffff)
202 { 239 {
203 *enc->cur++ = major | 25; 240 *enc->cur++ = major | LENGTH_EXT2;
204 *enc->cur++ = len >> 8; 241 *enc->cur++ = len >> 8;
205 *enc->cur++ = len; 242 *enc->cur++ = len;
206 } 243 }
207 else if (len <= 0xffffffff) 244 else if (len <= 0xffffffff)
208 { 245 {
209 *enc->cur++ = major | 26; 246 *enc->cur++ = major | LENGTH_EXT4;
210 *enc->cur++ = len >> 24; 247 *enc->cur++ = len >> 24;
211 *enc->cur++ = len >> 16; 248 *enc->cur++ = len >> 16;
212 *enc->cur++ = len >> 8; 249 *enc->cur++ = len >> 8;
213 *enc->cur++ = len; 250 *enc->cur++ = len;
214 } 251 }
215 else 252 else
216 { 253 {
217 *enc->cur++ = major | 27; 254 *enc->cur++ = major | LENGTH_EXT8;
218 *enc->cur++ = len >> 56; 255 *enc->cur++ = len >> 56;
219 *enc->cur++ = len >> 48; 256 *enc->cur++ = len >> 48;
220 *enc->cur++ = len >> 40; 257 *enc->cur++ = len >> 40;
221 *enc->cur++ = len >> 32; 258 *enc->cur++ = len >> 32;
222 *enc->cur++ = len >> 24; 259 *enc->cur++ = len >> 24;
… …
227} 264}
228 265
229ecb_inline void 266ecb_inline void
230encode_tag (enc_t *enc, UV tag) 267encode_tag (enc_t *enc, UV tag)
231{ 268{
232 encode_uint (enc, 0xc0, tag); 269 encode_uint (enc, MAJOR_TAG, tag);
270}
271
272ecb_inline void
273encode_str (enc_t *enc, int utf8, char *str, STRLEN len)
274{
275 encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len);
276 need (enc, len);
277 memcpy (enc->cur, str, len);
278 enc->cur += len;
233} 279}
234 280
235static void 281static void
236encode_str (enc_t *enc, int utf8, char *str, STRLEN len) 282encode_strref (enc_t *enc, int utf8, char *str, STRLEN len)
237{ 283{
238 if (ecb_expect_false (enc->cbor.flags & F_ALLOW_STRINGREF)) 284 if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS))
239 { 285 {
240 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1); 286 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1);
241 287
242 if (SvOK (*svp)) 288 if (SvOK (*svp))
243 { 289 {
244 // already registered, use stringref 290 // already registered, use stringref
245 encode_tag (enc, CBOR_TAG_STRINGREF); 291 encode_tag (enc, CBOR_TAG_STRINGREF);
246 encode_uint (enc, 0x00, SvUV (*svp)); 292 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
247 return; 293 return;
248 } 294 }
249 else if (len >= minimum_string_length (enc->stringref_idx)) 295 else if (len >= minimum_string_length (enc->stringref_idx))
250 { 296 {
251 // register only 297 // register only
252 sv_setuv (*svp, enc->stringref_idx); 298 sv_setuv (*svp, enc->stringref_idx);
253 ++enc->stringref_idx; 299 ++enc->stringref_idx;
254 } 300 }
255 } 301 }
256 302
257 encode_uint (enc, utf8 ? 0x60 : 0x40, len); 303 encode_str (enc, utf8, str, len);
258 need (enc, len);
259 memcpy (enc->cur, str, len);
260 enc->cur += len;
261} 304}
262 305
263static void encode_sv (enc_t *enc, SV *sv); 306static void encode_sv (enc_t *enc, SV *sv);
264 307
265static void 308static void
… …
270 if (enc->depth >= enc->cbor.max_depth) 313 if (enc->depth >= enc->cbor.max_depth)
271 croak (ERR_NESTING_EXCEEDED); 314 croak (ERR_NESTING_EXCEEDED);
272 315
273 ++enc->depth; 316 ++enc->depth;
274 317
275 encode_uint (enc, 0x80, len + 1); 318 encode_uint (enc, MAJOR_ARRAY, len + 1);
276 319
277 for (i = 0; i <= len; ++i) 320 for (i = 0; i <= len; ++i)
278 { 321 {
279 SV **svp = av_fetch (av, i, 0); 322 SV **svp = av_fetch (av, i, 0);
280 encode_sv (enc, svp ? *svp : &PL_sv_undef); 323 encode_sv (enc, svp ? *svp : &PL_sv_undef);
… …
295 338
296 int pairs = hv_iterinit (hv); 339 int pairs = hv_iterinit (hv);
297 int mg = SvMAGICAL (hv); 340 int mg = SvMAGICAL (hv);
298 341
299 if (mg) 342 if (mg)
300 encode_ch (enc, 0xa0 | 31); 343 encode_ch (enc, MAJOR_MAP | MINOR_INDEF);
301 else 344 else
302 encode_uint (enc, 0xa0, pairs); 345 encode_uint (enc, MAJOR_MAP, pairs);
303 346
304 while ((he = hv_iternext (hv))) 347 while ((he = hv_iternext (hv)))
305 { 348 {
306 if (HeKLEN (he) == HEf_SVKEY) 349 if (HeKLEN (he) == HEf_SVKEY)
307 encode_sv (enc, HeSVKEY (he)); 350 encode_sv (enc, HeSVKEY (he));
308 else 351 else
309 encode_str (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he)); 352 encode_strref (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he));
310 353
311 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he)); 354 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he));
312 } 355 }
313 356
314 if (mg) 357 if (mg)
315 encode_ch (enc, 0xe0 | 31); 358 encode_ch (enc, MAJOR_MISC | MINOR_INDEF);
316 359
317 --enc->depth; 360 --enc->depth;
318} 361}
319 362
320// encode objects, arrays and special \0=false and \1=true values. 363// encode objects, arrays and special \0=false and \1=true values.
321static void 364static void
322encode_rv (enc_t *enc, SV *sv) 365encode_rv (enc_t *enc, SV *sv)
323{ 366{
324 SvGETMAGIC (sv); 367 SvGETMAGIC (sv);
325
326 if (ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING)
327 && ecb_expect_false (SvREFCNT (sv) > 1))
328 {
329 if (!enc->shareable)
330 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
331
332 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
333
334 if (SvOK (*svp))
335 {
336 encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF);
337 encode_uint (enc, 0x00, SvUV (*svp));
338 return;
339 }
340 else
341 {
342 sv_setuv (*svp, enc->shareable_idx);
343 ++enc->shareable_idx;
344 encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE);
345 }
346 }
347 368
348 svtype svt = SvTYPE (sv); 369 svtype svt = SvTYPE (sv);
349 370
350 if (ecb_expect_false (SvOBJECT (sv))) 371 if (ecb_expect_false (SvOBJECT (sv)))
351 { 372 {
… …
358 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 379 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
359 ? cbor_tagged_stash 380 ? cbor_tagged_stash
360 : gv_stashpv ("CBOR::XS::Tagged" , 1); 381 : gv_stashpv ("CBOR::XS::Tagged" , 1);
361 382
362 HV *stash = SvSTASH (sv); 383 HV *stash = SvSTASH (sv);
363 GV *method;
364 384
365 if (stash == boolean_stash) 385 if (stash == boolean_stash)
366 encode_ch (enc, SvIV (sv) ? 0xe0 | 21 : 0xe0 | 20); 386 {
387 encode_ch (enc, SvIV (sv) ? MAJOR_MISC | SIMPLE_TRUE : MAJOR_MISC | SIMPLE_FALSE);
388 return;
389 }
367 else if (stash == error_stash) 390 else if (stash == error_stash)
368 encode_ch (enc, 0xe0 | 23); 391 {
392 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
393 return;
394 }
369 else if (stash == tagged_stash) 395 else if (stash == tagged_stash)
370 { 396 {
371 if (svt != SVt_PVAV) 397 if (svt != SVt_PVAV)
372 croak ("encountered CBOR::XS::Tagged object that isn't an array"); 398 croak ("encountered CBOR::XS::Tagged object that isn't an array");
373 399
374 encode_uint (enc, 0xc0, SvUV (*av_fetch ((AV *)sv, 0, 1))); 400 encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1)));
375 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1)); 401 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1));
402
403 return;
404 }
405 }
406
407 if (ecb_expect_false (SvREFCNT (sv) > 1)
408 && ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING))
409 {
410 if (!enc->shareable)
411 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
412
413 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
414
415 if (SvOK (*svp))
376 } 416 {
417 encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF);
418 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
419 return;
420 }
421 else
422 {
423 sv_setuv (*svp, enc->shareable_idx);
424 ++enc->shareable_idx;
425 encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE);
426 }
427 }
428
429 if (ecb_expect_false (SvOBJECT (sv)))
430 {
431 HV *stash = SvSTASH (sv);
432 GV *method;
433
377 else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0))) 434 if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
378 { 435 {
379 dSP; 436 dSP;
380 437
381 ENTER; SAVETMPS; PUSHMARK (SP); 438 ENTER; SAVETMPS; PUSHMARK (SP);
382 // we re-bless the reference to get overload and other niceties right 439 // we re-bless the reference to get overload and other niceties right
… …
414 // catch this surprisingly common error 471 // catch this surprisingly common error
415 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv) 472 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)
416 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash)); 473 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));
417 474
418 encode_tag (enc, CBOR_TAG_PERL_OBJECT); 475 encode_tag (enc, CBOR_TAG_PERL_OBJECT);
419 encode_uint (enc, 0x80, count + 1); 476 encode_uint (enc, MAJOR_ARRAY, count + 1);
420 encode_str (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash)); 477 encode_strref (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
421 478
422 while (count) 479 while (count)
423 encode_sv (enc, SP[1 - count--]); 480 encode_sv (enc, SP[1 - count--]);
424 481
425 PUTBACK; 482 PUTBACK;
… …
446{ 503{
447 double nv = SvNVX (sv); 504 double nv = SvNVX (sv);
448 505
449 need (enc, 9); 506 need (enc, 9);
450 507
451 if (ecb_expect_false (nv == (U32)nv)) 508 if (ecb_expect_false (nv == (NV)(U32)nv))
452 encode_uint (enc, 0x00, (U32)nv); 509 encode_uint (enc, MAJOR_POS_INT, (U32)nv);
453 //TODO: maybe I32? 510 //TODO: maybe I32?
454 else if (ecb_expect_false (nv == (float)nv)) 511 else if (ecb_expect_false (nv == (float)nv))
455 { 512 {
456 uint32_t fp = ecb_float_to_binary32 (nv); 513 uint32_t fp = ecb_float_to_binary32 (nv);
457 514
458 *enc->cur++ = 0xe0 | 26; 515 *enc->cur++ = MAJOR_MISC | MISC_FLOAT32;
459 516
460 if (!ecb_big_endian ()) 517 if (!ecb_big_endian ())
461 fp = ecb_bswap32 (fp); 518 fp = ecb_bswap32 (fp);
462 519
463 memcpy (enc->cur, &fp, 4); 520 memcpy (enc->cur, &fp, 4);
… …
465 } 522 }
466 else 523 else
467 { 524 {
468 uint64_t fp = ecb_double_to_binary64 (nv); 525 uint64_t fp = ecb_double_to_binary64 (nv);
469 526
470 *enc->cur++ = 0xe0 | 27; 527 *enc->cur++ = MAJOR_MISC | MISC_FLOAT64;
471 528
472 if (!ecb_big_endian ()) 529 if (!ecb_big_endian ())
473 fp = ecb_bswap64 (fp); 530 fp = ecb_bswap64 (fp);
474 531
475 memcpy (enc->cur, &fp, 8); 532 memcpy (enc->cur, &fp, 8);
… …
484 541
485 if (SvPOKp (sv)) 542 if (SvPOKp (sv))
486 { 543 {
487 STRLEN len; 544 STRLEN len;
488 char *str = SvPV (sv, len); 545 char *str = SvPV (sv, len);
489 encode_str (enc, SvUTF8 (sv), str, len); 546 encode_strref (enc, SvUTF8 (sv), str, len);
490 } 547 }
491 else if (SvNOKp (sv)) 548 else if (SvNOKp (sv))
492 encode_nv (enc, sv); 549 encode_nv (enc, sv);
493 else if (SvIOKp (sv)) 550 else if (SvIOKp (sv))
494 { 551 {
495 if (SvIsUV (sv)) 552 if (SvIsUV (sv))
496 encode_uint (enc, 0x00, SvUVX (sv)); 553 encode_uint (enc, MAJOR_POS_INT, SvUVX (sv));
497 else if (SvIVX (sv) >= 0) 554 else if (SvIVX (sv) >= 0)
498 encode_uint (enc, 0x00, SvIVX (sv)); 555 encode_uint (enc, MAJOR_POS_INT, SvIVX (sv));
499 else 556 else
500 encode_uint (enc, 0x20, -(SvIVX (sv) + 1)); 557 encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1));
501 } 558 }
502 else if (SvROK (sv)) 559 else if (SvROK (sv))
503 encode_rv (enc, SvRV (sv)); 560 encode_rv (enc, SvRV (sv));
504 else if (!SvOK (sv)) 561 else if (!SvOK (sv))
505 encode_ch (enc, 0xe0 | 22); 562 encode_ch (enc, MAJOR_MISC | SIMPLE_NULL);
506 else if (enc->cbor.flags & F_ALLOW_UNKNOWN) 563 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
507 encode_ch (enc, 0xe0 | 23); 564 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
508 else 565 else
509 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data", 566 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data",
510 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv)); 567 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
511} 568}
512 569
… …
520 enc.cur = SvPVX (enc.sv); 577 enc.cur = SvPVX (enc.sv);
521 enc.end = SvEND (enc.sv); 578 enc.end = SvEND (enc.sv);
522 579
523 SvPOK_only (enc.sv); 580 SvPOK_only (enc.sv);
524 581
525 if (cbor->flags & F_ALLOW_STRINGREF) 582 if (cbor->flags & F_PACK_STRINGS)
526 { 583 {
527 encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE); 584 encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE);
528 enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ()); 585 enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ());
529 enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ()); 586 enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ());
530 } 587 }
… …
565#define DEC_DEC_DEPTH --dec->depth 622#define DEC_DEC_DEPTH --dec->depth
566 623
567static UV 624static UV
568decode_uint (dec_t *dec) 625decode_uint (dec_t *dec)
569{ 626{
570 switch (*dec->cur & 31) 627 U8 m = *dec->cur & MINOR_MASK;
628 ++dec->cur;
629
630 if (ecb_expect_true (m < LENGTH_EXT1))
631 return m;
632
633 switch (m)
571 { 634 {
572 case 0: case 1: case 2: case 3: case 4: case 5: case 6: case 7: 635 case LENGTH_EXT1:
573 case 8: case 9: case 10: case 11: case 12: case 13: case 14: case 15: 636 WANT (1);
574 case 16: case 17: case 18: case 19: case 20: case 21: case 22: case 23: 637 dec->cur += 1;
575 return *dec->cur++ & 31; 638 return dec->cur[-1];
576 639
577 case 24: 640 case LENGTH_EXT2:
578 WANT (2); 641 WANT (2);
579 dec->cur += 2; 642 dec->cur += 2;
580 return dec->cur[-1];
581
582 case 25:
583 WANT (3);
584 dec->cur += 3;
585 return (((UV)dec->cur[-2]) << 8) 643 return (((UV)dec->cur[-2]) << 8)
586 | ((UV)dec->cur[-1]); 644 | ((UV)dec->cur[-1]);
587 645
588 case 26: 646 case LENGTH_EXT4:
589 WANT (5); 647 WANT (4);
590 dec->cur += 5; 648 dec->cur += 4;
591 return (((UV)dec->cur[-4]) << 24) 649 return (((UV)dec->cur[-4]) << 24)
592 | (((UV)dec->cur[-3]) << 16) 650 | (((UV)dec->cur[-3]) << 16)
593 | (((UV)dec->cur[-2]) << 8) 651 | (((UV)dec->cur[-2]) << 8)
594 | ((UV)dec->cur[-1]); 652 | ((UV)dec->cur[-1]);
595 653
596 case 27: 654 case LENGTH_EXT8:
597 WANT (9); 655 WANT (8);
598 dec->cur += 9; 656 dec->cur += 8;
657
658 return
659#if UVSIZE < 8
660 0
661#else
599 return (((UV)dec->cur[-8]) << 56) 662 (((UV)dec->cur[-8]) << 56)
600 | (((UV)dec->cur[-7]) << 48) 663 | (((UV)dec->cur[-7]) << 48)
601 | (((UV)dec->cur[-6]) << 40) 664 | (((UV)dec->cur[-6]) << 40)
602 | (((UV)dec->cur[-5]) << 32) 665 | (((UV)dec->cur[-5]) << 32)
666#endif
603 | (((UV)dec->cur[-4]) << 24) 667 | (((UV)dec->cur[-4]) << 24)
604 | (((UV)dec->cur[-3]) << 16) 668 | (((UV)dec->cur[-3]) << 16)
605 | (((UV)dec->cur[-2]) << 8) 669 | (((UV)dec->cur[-2]) << 8)
606 | ((UV)dec->cur[-1]); 670 | ((UV)dec->cur[-1]);
607 671
… …
620{ 684{
621 AV *av = newAV (); 685 AV *av = newAV ();
622 686
623 DEC_INC_DEPTH; 687 DEC_INC_DEPTH;
624 688
625 if ((*dec->cur & 31) == 31) 689 if (*dec->cur == (MAJOR_ARRAY | MINOR_INDEF))
626 { 690 {
627 ++dec->cur; 691 ++dec->cur;
628 692
629 for (;;) 693 for (;;)
630 { 694 {
631 WANT (1); 695 WANT (1);
632 696
633 if (*dec->cur == (0xe0 | 31)) 697 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
634 { 698 {
635 ++dec->cur; 699 ++dec->cur;
636 break; 700 break;
637 } 701 }
638 702
… …
663{ 727{
664 // for speed reasons, we specialcase single-string 728 // for speed reasons, we specialcase single-string
665 // byte or utf-8 strings as keys, but only when !stringref 729 // byte or utf-8 strings as keys, but only when !stringref
666 730
667 if (ecb_expect_true (!dec->stringref)) 731 if (ecb_expect_true (!dec->stringref))
668 if (*dec->cur >= 0x40 && *dec->cur <= 0x40 + 27) 732 if ((*dec->cur - MAJOR_BYTES) <= 27)
669 { 733 {
670 I32 len = decode_uint (dec); 734 I32 len = decode_uint (dec);
671 char *key = (char *)dec->cur; 735 char *key = (char *)dec->cur;
672 736
673 dec->cur += len; 737 dec->cur += len;
… …
677 741
678 hv_store (hv, key, len, decode_sv (dec), 0); 742 hv_store (hv, key, len, decode_sv (dec), 0);
679 743
680 return; 744 return;
681 } 745 }
682 else if (*dec->cur >= 0x60 && *dec->cur <= 0x60 + 27) 746 else if ((*dec->cur - MAJOR_TEXT) <= 27)
683 { 747 {
684 I32 len = decode_uint (dec); 748 I32 len = decode_uint (dec);
685 char *key = (char *)dec->cur; 749 char *key = (char *)dec->cur;
686 750
687 dec->cur += len; 751 dec->cur += len;
… …
706{ 770{
707 HV *hv = newHV (); 771 HV *hv = newHV ();
708 772
709 DEC_INC_DEPTH; 773 DEC_INC_DEPTH;
710 774
711 if ((*dec->cur & 31) == 31) 775 if (*dec->cur == (MAJOR_MAP | MINOR_INDEF))
712 { 776 {
713 ++dec->cur; 777 ++dec->cur;
714 778
715 for (;;) 779 for (;;)
716 { 780 {
717 WANT (1); 781 WANT (1);
718 782
719 if (*dec->cur == (0xe0 | 31)) 783 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
720 { 784 {
721 ++dec->cur; 785 ++dec->cur;
722 break; 786 break;
723 } 787 }
724 788
… …
745static SV * 809static SV *
746decode_str (dec_t *dec, int utf8) 810decode_str (dec_t *dec, int utf8)
747{ 811{
748 SV *sv = 0; 812 SV *sv = 0;
749 813
750 if ((*dec->cur & 31) == 31) 814 if ((*dec->cur & MINOR_MASK) == MINOR_INDEF)
751 { 815 {
816 // indefinite length strings
752 ++dec->cur; 817 ++dec->cur;
753 818
819 U8 major = *dec->cur & MAJOR_MISC;
820
754 sv = newSVpvn ("", 0); 821 sv = newSVpvn ("", 0);
755 822
756 // not very fast, and certainly not robust against illegal input
757 for (;;) 823 for (;;)
758 { 824 {
759 WANT (1); 825 WANT (1);
760 826
761 if (*dec->cur == (0xe0 | 31)) 827 if ((*dec->cur - major) > LENGTH_EXT8)
828 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
762 { 829 {
763 ++dec->cur; 830 ++dec->cur;
764 break; 831 break;
765 } 832 }
833 else
834 ERR ("corrupted CBOR data (invalid chunks in indefinite length string)");
766 835
767 sv_catsv (sv, decode_sv (dec)); 836 STRLEN len = decode_uint (dec);
837
838 WANT (len);
839 sv_catpvn (sv, dec->cur, len);
840 dec->cur += len;
768 } 841 }
769 } 842 }
770 else 843 else
771 { 844 {
772 STRLEN len = decode_uint (dec); 845 STRLEN len = decode_uint (dec);
… …
821 } 894 }
822 break; 895 break;
823 896
824 case CBOR_TAG_STRINGREF: 897 case CBOR_TAG_STRINGREF:
825 { 898 {
826 if ((*dec->cur >> 5) != 0) 899 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
827 ERR ("corrupted CBOR data (stringref index not an unsigned integer)"); 900 ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
828 901
829 UV idx = decode_uint (dec); 902 UV idx = decode_uint (dec);
830 903
831 if (!dec->stringref || (int)idx > AvFILLp (dec->stringref)) 904 if (!dec->stringref || (int)idx > AvFILLp (dec->stringref))
… …
849 } 922 }
850 break; 923 break;
851 924
852 case CBOR_TAG_VALUE_SHAREDREF: 925 case CBOR_TAG_VALUE_SHAREDREF:
853 { 926 {
854 if ((*dec->cur >> 5) != 0) 927 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
855 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)"); 928 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
856 929
857 UV idx = decode_uint (dec); 930 UV idx = decode_uint (dec);
858 931
859 if (!dec->shareable || (int)idx > AvFILLp (dec->shareable)) 932 if (!dec->shareable || (int)idx > AvFILLp (dec->shareable))
… …
968static SV * 1041static SV *
969decode_sv (dec_t *dec) 1042decode_sv (dec_t *dec)
970{ 1043{
971 WANT (1); 1044 WANT (1);
972 1045
973 switch (*dec->cur >> 5) 1046 switch (*dec->cur >> MAJOR_SHIFT)
974 { 1047 {
975 case 0: // unsigned int 1048 case MAJOR_POS_INT >> MAJOR_SHIFT: return newSVuv (decode_uint (dec));
976 return newSVuv (decode_uint (dec)); 1049 case MAJOR_NEG_INT >> MAJOR_SHIFT: return newSViv (-1 - (IV)decode_uint (dec));
977 case 1: // negative int 1050 case MAJOR_BYTES >> MAJOR_SHIFT: return decode_str (dec, 0);
978 return newSViv (-1 - (IV)decode_uint (dec)); 1051 case MAJOR_TEXT >> MAJOR_SHIFT: return decode_str (dec, 1);
979 case 2: // octet string 1052 case MAJOR_ARRAY >> MAJOR_SHIFT: return decode_av (dec);
980 return decode_str (dec, 0); 1053 case MAJOR_MAP >> MAJOR_SHIFT: return decode_hv (dec);
981 case 3: // utf-8 string 1054 case MAJOR_TAG >> MAJOR_SHIFT: return decode_tagged (dec);
982 return decode_str (dec, 1); 1055
983 case 4: // array 1056 case MAJOR_MISC >> MAJOR_SHIFT:
984 return decode_av (dec);
985 case 5: // map
986 return decode_hv (dec);
987 case 6: // tag
988 return decode_tagged (dec);
989 case 7: // misc
990 switch (*dec->cur++ & 31) 1057 switch (*dec->cur++ & MINOR_MASK)
991 { 1058 {
992 case 20: 1059 case SIMPLE_FALSE:
993#if CBOR_SLOW 1060#if CBOR_SLOW
994 types_false = get_bool ("Types::Serialiser::false"); 1061 types_false = get_bool ("Types::Serialiser::false");
995#endif 1062#endif
996 return newSVsv (types_false); 1063 return newSVsv (types_false);
997 case 21: 1064 case SIMPLE_TRUE:
998#if CBOR_SLOW 1065#if CBOR_SLOW
999 types_true = get_bool ("Types::Serialiser::true"); 1066 types_true = get_bool ("Types::Serialiser::true");
1000#endif 1067#endif
1001 return newSVsv (types_true); 1068 return newSVsv (types_true);
1002 case 22: 1069 case SIMPLE_NULL:
1003 return newSVsv (&PL_sv_undef); 1070 return newSVsv (&PL_sv_undef);
1004 case 23: 1071 case SIMPLE_UNDEF:
1005#if CBOR_SLOW 1072#if CBOR_SLOW
1006 types_error = get_bool ("Types::Serialiser::error"); 1073 types_error = get_bool ("Types::Serialiser::error");
1007#endif 1074#endif
1008 return newSVsv (types_error); 1075 return newSVsv (types_error);
1009 1076
1010 case 25: 1077 case MISC_FLOAT16:
1011 { 1078 {
1012 WANT (2); 1079 WANT (2);
1013 1080
1014 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1]; 1081 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1];
1015 dec->cur += 2; 1082 dec->cur += 2;
1016 1083
1017 return newSVnv (ecb_binary16_to_float (fp)); 1084 return newSVnv (ecb_binary16_to_float (fp));
1018 } 1085 }
1019 1086
1020 case 26: 1087 case MISC_FLOAT32:
1021 { 1088 {
1022 uint32_t fp; 1089 uint32_t fp;
1023 WANT (4); 1090 WANT (4);
1024 memcpy (&fp, dec->cur, 4); 1091 memcpy (&fp, dec->cur, 4);
1025 dec->cur += 4; 1092 dec->cur += 4;
… …
1028 fp = ecb_bswap32 (fp); 1095 fp = ecb_bswap32 (fp);
1029 1096
1030 return newSVnv (ecb_binary32_to_float (fp)); 1097 return newSVnv (ecb_binary32_to_float (fp));
1031 } 1098 }
1032 1099
1033 case 27: 1100 case MISC_FLOAT64:
1034 { 1101 {
1035 uint64_t fp; 1102 uint64_t fp;
1036 WANT (8); 1103 WANT (8);
1037 memcpy (&fp, dec->cur, 8); 1104 memcpy (&fp, dec->cur, 8);
1038 dec->cur += 8; 1105 dec->cur += 8;
… …
1041 fp = ecb_bswap64 (fp); 1108 fp = ecb_bswap64 (fp);
1042 1109
1043 return newSVnv (ecb_binary64_to_double (fp)); 1110 return newSVnv (ecb_binary64_to_double (fp));
1044 } 1111 }
1045 1112
1046 // 0..19 unassigned 1113 // 0..19 unassigned simple
1047 // 24 reserved + unassigned (reserved values are not encodable) 1114 // 24 reserved + unassigned (reserved values are not encodable)
1048 default: 1115 default:
1049 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)"); 1116 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)");
1050 } 1117 }
1051 1118
… …
1139void shrink (CBOR *self, int enable = 1) 1206void shrink (CBOR *self, int enable = 1)
1140 ALIAS: 1207 ALIAS:
1141 shrink = F_SHRINK 1208 shrink = F_SHRINK
1142 allow_unknown = F_ALLOW_UNKNOWN 1209 allow_unknown = F_ALLOW_UNKNOWN
1143 allow_sharing = F_ALLOW_SHARING 1210 allow_sharing = F_ALLOW_SHARING
1144 allow_stringref = F_ALLOW_STRINGREF 1211 pack_strings = F_PACK_STRINGS
1145 PPCODE: 1212 PPCODE:
1146{ 1213{
1147 if (enable) 1214 if (enable)
1148 self->flags |= ix; 1215 self->flags |= ix;
1149 else 1216 else
… …
1155void get_shrink (CBOR *self) 1222void get_shrink (CBOR *self)
1156 ALIAS: 1223 ALIAS:
1157 get_shrink = F_SHRINK 1224 get_shrink = F_SHRINK
1158 get_allow_unknown = F_ALLOW_UNKNOWN 1225 get_allow_unknown = F_ALLOW_UNKNOWN
1159 get_allow_sharing = F_ALLOW_SHARING 1226 get_allow_sharing = F_ALLOW_SHARING
1160 get_allow_stringref = F_ALLOW_STRINGREF 1227 get_pack_strings = F_PACK_STRINGS
1161 PPCODE: 1228 PPCODE:
1162 XPUSHs (boolSV (self->flags & ix)); 1229 XPUSHs (boolSV (self->flags & ix));
1163 1230
1164void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) 1231void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
1165 PPCODE: 1232 PPCODE:

Diff Legend

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