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 1.29 by root, Thu Nov 28 09:13:12 2013 UTC vs.
Revision 1.40 by root, Tue Dec 10 13:54:40 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)
89typedef struct { 128typedef struct {
90 U32 flags; 129 U32 flags;
91 U32 max_depth; 130 U32 max_depth;
92 STRLEN max_size; 131 STRLEN max_size;
93 SV *filter; 132 SV *filter;
133
134 // for the incremental parser
135 STRLEN incr_pos; // the current offset into the text
136 STRLEN incr_need; // minimum bytes needed to decode
137 AV *incr_count; // for every nesting level, the number of outstanding values, or -1 for indef.
94} CBOR; 138} CBOR;
95 139
96ecb_inline void 140ecb_inline void
97cbor_init (CBOR *cbor) 141cbor_init (CBOR *cbor)
98{ 142{
102 146
103ecb_inline void 147ecb_inline void
104cbor_free (CBOR *cbor) 148cbor_free (CBOR *cbor)
105{ 149{
106 SvREFCNT_dec (cbor->filter); 150 SvREFCNT_dec (cbor->filter);
151 SvREFCNT_dec (cbor->incr_count);
107} 152}
108 153
109///////////////////////////////////////////////////////////////////////////// 154/////////////////////////////////////////////////////////////////////////////
110// utility functions 155// utility functions
111 156
189static void 234static void
190encode_uint (enc_t *enc, int major, UV len) 235encode_uint (enc_t *enc, int major, UV len)
191{ 236{
192 need (enc, 9); 237 need (enc, 9);
193 238
194 if (len < 24) 239 if (ecb_expect_true (len < LENGTH_EXT1))
195 *enc->cur++ = major | len; 240 *enc->cur++ = major | len;
196 else if (len <= 0xff) 241 else if (ecb_expect_true (len <= 0xffU))
197 { 242 {
198 *enc->cur++ = major | 24; 243 *enc->cur++ = major | LENGTH_EXT1;
199 *enc->cur++ = len; 244 *enc->cur++ = len;
200 } 245 }
201 else if (len <= 0xffff) 246 else if (len <= 0xffffU)
202 { 247 {
203 *enc->cur++ = major | 25; 248 *enc->cur++ = major | LENGTH_EXT2;
204 *enc->cur++ = len >> 8; 249 *enc->cur++ = len >> 8;
205 *enc->cur++ = len; 250 *enc->cur++ = len;
206 } 251 }
207 else if (len <= 0xffffffff) 252 else if (len <= 0xffffffffU)
208 { 253 {
209 *enc->cur++ = major | 26; 254 *enc->cur++ = major | LENGTH_EXT4;
210 *enc->cur++ = len >> 24; 255 *enc->cur++ = len >> 24;
211 *enc->cur++ = len >> 16; 256 *enc->cur++ = len >> 16;
212 *enc->cur++ = len >> 8; 257 *enc->cur++ = len >> 8;
213 *enc->cur++ = len; 258 *enc->cur++ = len;
214 } 259 }
215 else 260 else
216 { 261 {
217 *enc->cur++ = major | 27; 262 *enc->cur++ = major | LENGTH_EXT8;
218 *enc->cur++ = len >> 56; 263 *enc->cur++ = len >> 56;
219 *enc->cur++ = len >> 48; 264 *enc->cur++ = len >> 48;
220 *enc->cur++ = len >> 40; 265 *enc->cur++ = len >> 40;
221 *enc->cur++ = len >> 32; 266 *enc->cur++ = len >> 32;
222 *enc->cur++ = len >> 24; 267 *enc->cur++ = len >> 24;
227} 272}
228 273
229ecb_inline void 274ecb_inline void
230encode_tag (enc_t *enc, UV tag) 275encode_tag (enc_t *enc, UV tag)
231{ 276{
232 encode_uint (enc, 0xc0, tag); 277 encode_uint (enc, MAJOR_TAG, tag);
278}
279
280ecb_inline void
281encode_str (enc_t *enc, int utf8, char *str, STRLEN len)
282{
283 encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len);
284 need (enc, len);
285 memcpy (enc->cur, str, len);
286 enc->cur += len;
233} 287}
234 288
235static void 289static void
236encode_str (enc_t *enc, int utf8, char *str, STRLEN len) 290encode_strref (enc_t *enc, int utf8, char *str, STRLEN len)
237{ 291{
238 if (ecb_expect_false (enc->cbor.flags & F_ALLOW_STRINGREF)) 292 if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS))
239 { 293 {
240 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1); 294 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1);
241 295
242 if (SvOK (*svp)) 296 if (SvOK (*svp))
243 { 297 {
244 // already registered, use stringref 298 // already registered, use stringref
245 encode_tag (enc, CBOR_TAG_STRINGREF); 299 encode_tag (enc, CBOR_TAG_STRINGREF);
246 encode_uint (enc, 0x00, SvUV (*svp)); 300 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
247 return; 301 return;
248 } 302 }
249 else if (len >= minimum_string_length (enc->stringref_idx)) 303 else if (len >= minimum_string_length (enc->stringref_idx))
250 { 304 {
251 // register only 305 // register only
252 sv_setuv (*svp, enc->stringref_idx); 306 sv_setuv (*svp, enc->stringref_idx);
253 ++enc->stringref_idx; 307 ++enc->stringref_idx;
254 } 308 }
255 } 309 }
256 310
257 encode_uint (enc, utf8 ? 0x60 : 0x40, len); 311 encode_str (enc, utf8, str, len);
258 need (enc, len);
259 memcpy (enc->cur, str, len);
260 enc->cur += len;
261} 312}
262 313
263static void encode_sv (enc_t *enc, SV *sv); 314static void encode_sv (enc_t *enc, SV *sv);
264 315
265static void 316static void
270 if (enc->depth >= enc->cbor.max_depth) 321 if (enc->depth >= enc->cbor.max_depth)
271 croak (ERR_NESTING_EXCEEDED); 322 croak (ERR_NESTING_EXCEEDED);
272 323
273 ++enc->depth; 324 ++enc->depth;
274 325
275 encode_uint (enc, 0x80, len + 1); 326 encode_uint (enc, MAJOR_ARRAY, len + 1);
276 327
277 for (i = 0; i <= len; ++i) 328 for (i = 0; i <= len; ++i)
278 { 329 {
279 SV **svp = av_fetch (av, i, 0); 330 SV **svp = av_fetch (av, i, 0);
280 encode_sv (enc, svp ? *svp : &PL_sv_undef); 331 encode_sv (enc, svp ? *svp : &PL_sv_undef);
295 346
296 int pairs = hv_iterinit (hv); 347 int pairs = hv_iterinit (hv);
297 int mg = SvMAGICAL (hv); 348 int mg = SvMAGICAL (hv);
298 349
299 if (mg) 350 if (mg)
300 encode_ch (enc, 0xa0 | 31); 351 encode_ch (enc, MAJOR_MAP | MINOR_INDEF);
301 else 352 else
302 encode_uint (enc, 0xa0, pairs); 353 encode_uint (enc, MAJOR_MAP, pairs);
303 354
304 while ((he = hv_iternext (hv))) 355 while ((he = hv_iternext (hv)))
305 { 356 {
306 if (HeKLEN (he) == HEf_SVKEY) 357 if (HeKLEN (he) == HEf_SVKEY)
307 encode_sv (enc, HeSVKEY (he)); 358 encode_sv (enc, HeSVKEY (he));
308 else 359 else
309 encode_str (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he)); 360 encode_strref (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he));
310 361
311 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he)); 362 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he));
312 } 363 }
313 364
314 if (mg) 365 if (mg)
315 encode_ch (enc, 0xe0 | 31); 366 encode_ch (enc, MAJOR_MISC | MINOR_INDEF);
316 367
317 --enc->depth; 368 --enc->depth;
318} 369}
319 370
320// encode objects, arrays and special \0=false and \1=true values. 371// encode objects, arrays and special \0=false and \1=true values.
321static void 372static void
322encode_rv (enc_t *enc, SV *sv) 373encode_rv (enc_t *enc, SV *sv)
323{ 374{
324 SvGETMAGIC (sv); 375 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 }
347 376
348 svtype svt = SvTYPE (sv); 377 svtype svt = SvTYPE (sv);
349 378
350 if (ecb_expect_false (SvOBJECT (sv))) 379 if (ecb_expect_false (SvOBJECT (sv)))
351 { 380 {
358 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 387 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
359 ? cbor_tagged_stash 388 ? cbor_tagged_stash
360 : gv_stashpv ("CBOR::XS::Tagged" , 1); 389 : gv_stashpv ("CBOR::XS::Tagged" , 1);
361 390
362 HV *stash = SvSTASH (sv); 391 HV *stash = SvSTASH (sv);
363 GV *method;
364 392
365 if (stash == boolean_stash) 393 if (stash == boolean_stash)
366 encode_ch (enc, SvIV (sv) ? 0xe0 | 21 : 0xe0 | 20); 394 {
395 encode_ch (enc, SvIV (sv) ? MAJOR_MISC | SIMPLE_TRUE : MAJOR_MISC | SIMPLE_FALSE);
396 return;
397 }
367 else if (stash == error_stash) 398 else if (stash == error_stash)
368 encode_ch (enc, 0xe0 | 23); 399 {
400 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
401 return;
402 }
369 else if (stash == tagged_stash) 403 else if (stash == tagged_stash)
370 { 404 {
371 if (svt != SVt_PVAV) 405 if (svt != SVt_PVAV)
372 croak ("encountered CBOR::XS::Tagged object that isn't an array"); 406 croak ("encountered CBOR::XS::Tagged object that isn't an array");
373 407
374 encode_uint (enc, 0xc0, SvUV (*av_fetch ((AV *)sv, 0, 1))); 408 encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1)));
375 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1)); 409 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1));
410
411 return;
412 }
413 }
414
415 if (ecb_expect_false (SvREFCNT (sv) > 1)
416 && ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING))
417 {
418 if (!enc->shareable)
419 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
420
421 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
422
423 if (SvOK (*svp))
376 } 424 {
425 encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF);
426 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
427 return;
428 }
429 else
430 {
431 sv_setuv (*svp, enc->shareable_idx);
432 ++enc->shareable_idx;
433 encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE);
434 }
435 }
436
437 if (ecb_expect_false (SvOBJECT (sv)))
438 {
439 HV *stash = SvSTASH (sv);
440 GV *method;
441
377 else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0))) 442 if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
378 { 443 {
379 dSP; 444 dSP;
380 445
381 ENTER; SAVETMPS; PUSHMARK (SP); 446 ENTER; SAVETMPS; PUSHMARK (SP);
382 // we re-bless the reference to get overload and other niceties right 447 // we re-bless the reference to get overload and other niceties right
414 // catch this surprisingly common error 479 // catch this surprisingly common error
415 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv) 480 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)); 481 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));
417 482
418 encode_tag (enc, CBOR_TAG_PERL_OBJECT); 483 encode_tag (enc, CBOR_TAG_PERL_OBJECT);
419 encode_uint (enc, 0x80, count + 1); 484 encode_uint (enc, MAJOR_ARRAY, count + 1);
420 encode_str (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash)); 485 encode_strref (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
421 486
422 while (count) 487 while (count)
423 encode_sv (enc, SP[1 - count--]); 488 encode_sv (enc, SP[1 - count--]);
424 489
425 PUTBACK; 490 PUTBACK;
446{ 511{
447 double nv = SvNVX (sv); 512 double nv = SvNVX (sv);
448 513
449 need (enc, 9); 514 need (enc, 9);
450 515
451 if (ecb_expect_false (nv == (U32)nv)) 516 if (ecb_expect_false (nv == (NV)(U32)nv))
452 encode_uint (enc, 0x00, (U32)nv); 517 encode_uint (enc, MAJOR_POS_INT, (U32)nv);
453 //TODO: maybe I32? 518 //TODO: maybe I32?
454 else if (ecb_expect_false (nv == (float)nv)) 519 else if (ecb_expect_false (nv == (float)nv))
455 { 520 {
456 uint32_t fp = ecb_float_to_binary32 (nv); 521 uint32_t fp = ecb_float_to_binary32 (nv);
457 522
458 *enc->cur++ = 0xe0 | 26; 523 *enc->cur++ = MAJOR_MISC | MISC_FLOAT32;
459 524
460 if (!ecb_big_endian ()) 525 if (!ecb_big_endian ())
461 fp = ecb_bswap32 (fp); 526 fp = ecb_bswap32 (fp);
462 527
463 memcpy (enc->cur, &fp, 4); 528 memcpy (enc->cur, &fp, 4);
465 } 530 }
466 else 531 else
467 { 532 {
468 uint64_t fp = ecb_double_to_binary64 (nv); 533 uint64_t fp = ecb_double_to_binary64 (nv);
469 534
470 *enc->cur++ = 0xe0 | 27; 535 *enc->cur++ = MAJOR_MISC | MISC_FLOAT64;
471 536
472 if (!ecb_big_endian ()) 537 if (!ecb_big_endian ())
473 fp = ecb_bswap64 (fp); 538 fp = ecb_bswap64 (fp);
474 539
475 memcpy (enc->cur, &fp, 8); 540 memcpy (enc->cur, &fp, 8);
484 549
485 if (SvPOKp (sv)) 550 if (SvPOKp (sv))
486 { 551 {
487 STRLEN len; 552 STRLEN len;
488 char *str = SvPV (sv, len); 553 char *str = SvPV (sv, len);
489 encode_str (enc, SvUTF8 (sv), str, len); 554 encode_strref (enc, SvUTF8 (sv), str, len);
490 } 555 }
491 else if (SvNOKp (sv)) 556 else if (SvNOKp (sv))
492 encode_nv (enc, sv); 557 encode_nv (enc, sv);
493 else if (SvIOKp (sv)) 558 else if (SvIOKp (sv))
494 { 559 {
495 if (SvIsUV (sv)) 560 if (SvIsUV (sv))
496 encode_uint (enc, 0x00, SvUVX (sv)); 561 encode_uint (enc, MAJOR_POS_INT, SvUVX (sv));
497 else if (SvIVX (sv) >= 0) 562 else if (SvIVX (sv) >= 0)
498 encode_uint (enc, 0x00, SvIVX (sv)); 563 encode_uint (enc, MAJOR_POS_INT, SvIVX (sv));
499 else 564 else
500 encode_uint (enc, 0x20, -(SvIVX (sv) + 1)); 565 encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1));
501 } 566 }
502 else if (SvROK (sv)) 567 else if (SvROK (sv))
503 encode_rv (enc, SvRV (sv)); 568 encode_rv (enc, SvRV (sv));
504 else if (!SvOK (sv)) 569 else if (!SvOK (sv))
505 encode_ch (enc, 0xe0 | 22); 570 encode_ch (enc, MAJOR_MISC | SIMPLE_NULL);
506 else if (enc->cbor.flags & F_ALLOW_UNKNOWN) 571 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
507 encode_ch (enc, 0xe0 | 23); 572 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
508 else 573 else
509 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data", 574 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data",
510 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv)); 575 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
511} 576}
512 577
520 enc.cur = SvPVX (enc.sv); 585 enc.cur = SvPVX (enc.sv);
521 enc.end = SvEND (enc.sv); 586 enc.end = SvEND (enc.sv);
522 587
523 SvPOK_only (enc.sv); 588 SvPOK_only (enc.sv);
524 589
525 if (cbor->flags & F_ALLOW_STRINGREF) 590 if (cbor->flags & F_PACK_STRINGS)
526 { 591 {
527 encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE); 592 encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE);
528 enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ()); 593 enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ());
529 enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ()); 594 enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ());
530 } 595 }
565#define DEC_DEC_DEPTH --dec->depth 630#define DEC_DEC_DEPTH --dec->depth
566 631
567static UV 632static UV
568decode_uint (dec_t *dec) 633decode_uint (dec_t *dec)
569{ 634{
570 switch (*dec->cur & 31) 635 U8 m = *dec->cur & MINOR_MASK;
571 { 636 ++dec->cur;
572 case 0: case 1: case 2: case 3: case 4: case 5: case 6: case 7:
573 case 8: case 9: case 10: case 11: case 12: case 13: case 14: case 15:
574 case 16: case 17: case 18: case 19: case 20: case 21: case 22: case 23:
575 return *dec->cur++ & 31;
576 637
577 case 24: 638 if (ecb_expect_true (m < LENGTH_EXT1))
639 return m;
640 else if (ecb_expect_true (m == LENGTH_EXT1))
641 {
578 WANT (2); 642 WANT (1);
579 dec->cur += 2; 643 dec->cur += 1;
580 return dec->cur[-1]; 644 return dec->cur[-1];
581 645 }
582 case 25: 646 else if (ecb_expect_true (m == LENGTH_EXT2))
647 {
583 WANT (3); 648 WANT (2);
584 dec->cur += 3; 649 dec->cur += 2;
585 return (((UV)dec->cur[-2]) << 8) 650 return (((UV)dec->cur[-2]) << 8)
586 | ((UV)dec->cur[-1]); 651 | ((UV)dec->cur[-1]);
587 652 }
588 case 26: 653 else if (ecb_expect_true (m == LENGTH_EXT4))
654 {
589 WANT (5); 655 WANT (4);
590 dec->cur += 5; 656 dec->cur += 4;
591 return (((UV)dec->cur[-4]) << 24) 657 return (((UV)dec->cur[-4]) << 24)
592 | (((UV)dec->cur[-3]) << 16) 658 | (((UV)dec->cur[-3]) << 16)
593 | (((UV)dec->cur[-2]) << 8) 659 | (((UV)dec->cur[-2]) << 8)
594 | ((UV)dec->cur[-1]); 660 | ((UV)dec->cur[-1]);
595 661 }
596 case 27: 662 else if (ecb_expect_true (m == LENGTH_EXT8))
663 {
597 WANT (9); 664 WANT (8);
598 dec->cur += 9; 665 dec->cur += 8;
666
667 return
668#if UVSIZE < 8
669 0
670#else
599 return (((UV)dec->cur[-8]) << 56) 671 (((UV)dec->cur[-8]) << 56)
600 | (((UV)dec->cur[-7]) << 48) 672 | (((UV)dec->cur[-7]) << 48)
601 | (((UV)dec->cur[-6]) << 40) 673 | (((UV)dec->cur[-6]) << 40)
602 | (((UV)dec->cur[-5]) << 32) 674 | (((UV)dec->cur[-5]) << 32)
675#endif
603 | (((UV)dec->cur[-4]) << 24) 676 | (((UV)dec->cur[-4]) << 24)
604 | (((UV)dec->cur[-3]) << 16) 677 | (((UV)dec->cur[-3]) << 16)
605 | (((UV)dec->cur[-2]) << 8) 678 | (((UV)dec->cur[-2]) << 8)
606 | ((UV)dec->cur[-1]); 679 | ((UV)dec->cur[-1]);
607 680 }
608 default: 681 else
609 ERR ("corrupted CBOR data (unsupported integer minor encoding)"); 682 ERR ("corrupted CBOR data (unsupported integer minor encoding)");
610 }
611 683
612fail: 684fail:
613 return 0; 685 return 0;
614} 686}
615 687
620{ 692{
621 AV *av = newAV (); 693 AV *av = newAV ();
622 694
623 DEC_INC_DEPTH; 695 DEC_INC_DEPTH;
624 696
625 if ((*dec->cur & 31) == 31) 697 if (*dec->cur == (MAJOR_ARRAY | MINOR_INDEF))
626 { 698 {
627 ++dec->cur; 699 ++dec->cur;
628 700
629 for (;;) 701 for (;;)
630 { 702 {
631 WANT (1); 703 WANT (1);
632 704
633 if (*dec->cur == (0xe0 | 31)) 705 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
634 { 706 {
635 ++dec->cur; 707 ++dec->cur;
636 break; 708 break;
637 } 709 }
638 710
641 } 713 }
642 else 714 else
643 { 715 {
644 int i, len = decode_uint (dec); 716 int i, len = decode_uint (dec);
645 717
718 WANT (len); // complexity check for av_fill - need at least one byte per value, do not allow supersize arrays
646 av_fill (av, len - 1); 719 av_fill (av, len - 1);
647 720
648 for (i = 0; i < len; ++i) 721 for (i = 0; i < len; ++i)
649 AvARRAY (av)[i] = decode_sv (dec); 722 AvARRAY (av)[i] = decode_sv (dec);
650 } 723 }
663{ 736{
664 // for speed reasons, we specialcase single-string 737 // for speed reasons, we specialcase single-string
665 // byte or utf-8 strings as keys, but only when !stringref 738 // byte or utf-8 strings as keys, but only when !stringref
666 739
667 if (ecb_expect_true (!dec->stringref)) 740 if (ecb_expect_true (!dec->stringref))
668 if (*dec->cur >= 0x40 && *dec->cur <= 0x40 + 27) 741 if (ecb_expect_true ((*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8))
669 { 742 {
670 I32 len = decode_uint (dec); 743 I32 len = decode_uint (dec);
671 char *key = (char *)dec->cur; 744 char *key = (char *)dec->cur;
672 745
673 dec->cur += len; 746 dec->cur += len;
674 747
675 if (ecb_expect_false (dec->stringref))
676 av_push (dec->stringref, newSVpvn (key, len));
677
678 hv_store (hv, key, len, decode_sv (dec), 0); 748 hv_store (hv, key, len, decode_sv (dec), 0);
679 749
680 return; 750 return;
681 } 751 }
682 else if (*dec->cur >= 0x60 && *dec->cur <= 0x60 + 27) 752 else if (ecb_expect_true ((*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8))
683 { 753 {
684 I32 len = decode_uint (dec); 754 I32 len = decode_uint (dec);
685 char *key = (char *)dec->cur; 755 char *key = (char *)dec->cur;
686 756
687 dec->cur += len; 757 dec->cur += len;
688 758
689 if (ecb_expect_false (dec->stringref)) 759 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
690 av_push (dec->stringref, newSVpvn_utf8 (key, len, 1)); 760 if (!is_utf8_string (key, len))
761 ERR ("corrupted CBOR data (invalid UTF-8 in map key)");
691 762
692 hv_store (hv, key, -len, decode_sv (dec), 0); 763 hv_store (hv, key, -len, decode_sv (dec), 0);
693 764
694 return; 765 return;
695 } 766 }
697 SV *k = decode_sv (dec); 768 SV *k = decode_sv (dec);
698 SV *v = decode_sv (dec); 769 SV *v = decode_sv (dec);
699 770
700 hv_store_ent (hv, k, v, 0); 771 hv_store_ent (hv, k, v, 0);
701 SvREFCNT_dec (k); 772 SvREFCNT_dec (k);
773
774fail:
775 ;
702} 776}
703 777
704static SV * 778static SV *
705decode_hv (dec_t *dec) 779decode_hv (dec_t *dec)
706{ 780{
707 HV *hv = newHV (); 781 HV *hv = newHV ();
708 782
709 DEC_INC_DEPTH; 783 DEC_INC_DEPTH;
710 784
711 if ((*dec->cur & 31) == 31) 785 if (*dec->cur == (MAJOR_MAP | MINOR_INDEF))
712 { 786 {
713 ++dec->cur; 787 ++dec->cur;
714 788
715 for (;;) 789 for (;;)
716 { 790 {
717 WANT (1); 791 WANT (1);
718 792
719 if (*dec->cur == (0xe0 | 31)) 793 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
720 { 794 {
721 ++dec->cur; 795 ++dec->cur;
722 break; 796 break;
723 } 797 }
724 798
745static SV * 819static SV *
746decode_str (dec_t *dec, int utf8) 820decode_str (dec_t *dec, int utf8)
747{ 821{
748 SV *sv = 0; 822 SV *sv = 0;
749 823
750 if ((*dec->cur & 31) == 31) 824 if ((*dec->cur & MINOR_MASK) == MINOR_INDEF)
751 { 825 {
826 // indefinite length strings
752 ++dec->cur; 827 ++dec->cur;
753 828
829 U8 major = *dec->cur & MAJOR_MISC;
830
754 sv = newSVpvn ("", 0); 831 sv = newSVpvn ("", 0);
755 832
756 // not very fast, and certainly not robust against illegal input
757 for (;;) 833 for (;;)
758 { 834 {
759 WANT (1); 835 WANT (1);
760 836
761 if (*dec->cur == (0xe0 | 31)) 837 if ((*dec->cur - major) > LENGTH_EXT8)
838 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
762 { 839 {
763 ++dec->cur; 840 ++dec->cur;
764 break; 841 break;
765 } 842 }
843 else
844 ERR ("corrupted CBOR data (invalid chunks in indefinite length string)");
766 845
767 sv_catsv (sv, decode_sv (dec)); 846 STRLEN len = decode_uint (dec);
847
848 WANT (len);
849 sv_catpvn (sv, dec->cur, len);
850 dec->cur += len;
768 } 851 }
769 } 852 }
770 else 853 else
771 { 854 {
772 STRLEN len = decode_uint (dec); 855 STRLEN len = decode_uint (dec);
779 && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1)) 862 && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1))
780 av_push (dec->stringref, SvREFCNT_inc_NN (sv)); 863 av_push (dec->stringref, SvREFCNT_inc_NN (sv));
781 } 864 }
782 865
783 if (utf8) 866 if (utf8)
867 {
868 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
869 if (!is_utf8_string (SvPVX (sv), SvCUR (sv)))
870 ERR ("corrupted CBOR data (invalid UTF-8 in text string)");
871
784 SvUTF8_on (sv); 872 SvUTF8_on (sv);
873 }
785 874
786 return sv; 875 return sv;
787 876
788fail: 877fail:
789 SvREFCNT_dec (sv); 878 SvREFCNT_dec (sv);
821 } 910 }
822 break; 911 break;
823 912
824 case CBOR_TAG_STRINGREF: 913 case CBOR_TAG_STRINGREF:
825 { 914 {
826 if ((*dec->cur >> 5) != 0) 915 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
827 ERR ("corrupted CBOR data (stringref index not an unsigned integer)"); 916 ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
828 917
829 UV idx = decode_uint (dec); 918 UV idx = decode_uint (dec);
830 919
831 if (!dec->stringref || (int)idx > AvFILLp (dec->stringref)) 920 if (!dec->stringref || (int)idx > AvFILLp (dec->stringref))
838 case CBOR_TAG_VALUE_SHAREABLE: 927 case CBOR_TAG_VALUE_SHAREABLE:
839 { 928 {
840 if (ecb_expect_false (!dec->shareable)) 929 if (ecb_expect_false (!dec->shareable))
841 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ()); 930 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ());
842 931
932 if (dec->cbor.flags & F_ALLOW_CYCLES)
933 {
843 sv = newSV (0); 934 sv = newSV (0);
844 av_push (dec->shareable, SvREFCNT_inc_NN (sv)); 935 av_push (dec->shareable, SvREFCNT_inc_NN (sv));
845 936
846 SV *osv = decode_sv (dec); 937 SV *osv = decode_sv (dec);
847 sv_setsv (sv, osv); 938 sv_setsv (sv, osv);
848 SvREFCNT_dec_NN (osv); 939 SvREFCNT_dec_NN (osv);
940 }
941 else
942 {
943 av_push (dec->shareable, &PL_sv_undef);
944 int idx = AvFILLp (dec->shareable);
945 sv = decode_sv (dec);
946 av_store (dec->shareable, idx, SvREFCNT_inc_NN (sv));
947 }
849 } 948 }
850 break; 949 break;
851 950
852 case CBOR_TAG_VALUE_SHAREDREF: 951 case CBOR_TAG_VALUE_SHAREDREF:
853 { 952 {
854 if ((*dec->cur >> 5) != 0) 953 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
855 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)"); 954 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
856 955
857 UV idx = decode_uint (dec); 956 UV idx = decode_uint (dec);
858 957
859 if (!dec->shareable || (int)idx > AvFILLp (dec->shareable)) 958 if (!dec->shareable || (int)idx > AvFILLp (dec->shareable))
860 ERR ("corrupted CBOR data (sharedref index out of bounds)"); 959 ERR ("corrupted CBOR data (sharedref index out of bounds)");
861 960
862 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]); 961 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]);
962
963 if (sv == &PL_sv_undef)
964 ERR ("cyclic CBOR data structure found, but allow_cycles is not enabled");
863 } 965 }
864 break; 966 break;
865 967
866 case CBOR_TAG_PERL_OBJECT: 968 case CBOR_TAG_PERL_OBJECT:
867 { 969 {
968static SV * 1070static SV *
969decode_sv (dec_t *dec) 1071decode_sv (dec_t *dec)
970{ 1072{
971 WANT (1); 1073 WANT (1);
972 1074
973 switch (*dec->cur >> 5) 1075 switch (*dec->cur >> MAJOR_SHIFT)
974 { 1076 {
975 case 0: // unsigned int 1077 case MAJOR_POS_INT >> MAJOR_SHIFT: return newSVuv (decode_uint (dec));
976 return newSVuv (decode_uint (dec)); 1078 case MAJOR_NEG_INT >> MAJOR_SHIFT: return newSViv (-1 - (IV)decode_uint (dec));
977 case 1: // negative int 1079 case MAJOR_BYTES >> MAJOR_SHIFT: return decode_str (dec, 0);
978 return newSViv (-1 - (IV)decode_uint (dec)); 1080 case MAJOR_TEXT >> MAJOR_SHIFT: return decode_str (dec, 1);
979 case 2: // octet string 1081 case MAJOR_ARRAY >> MAJOR_SHIFT: return decode_av (dec);
980 return decode_str (dec, 0); 1082 case MAJOR_MAP >> MAJOR_SHIFT: return decode_hv (dec);
981 case 3: // utf-8 string 1083 case MAJOR_TAG >> MAJOR_SHIFT: return decode_tagged (dec);
982 return decode_str (dec, 1); 1084
983 case 4: // array 1085 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) 1086 switch (*dec->cur++ & MINOR_MASK)
991 { 1087 {
992 case 20: 1088 case SIMPLE_FALSE:
993#if CBOR_SLOW 1089#if CBOR_SLOW
994 types_false = get_bool ("Types::Serialiser::false"); 1090 types_false = get_bool ("Types::Serialiser::false");
995#endif 1091#endif
996 return newSVsv (types_false); 1092 return newSVsv (types_false);
997 case 21: 1093 case SIMPLE_TRUE:
998#if CBOR_SLOW 1094#if CBOR_SLOW
999 types_true = get_bool ("Types::Serialiser::true"); 1095 types_true = get_bool ("Types::Serialiser::true");
1000#endif 1096#endif
1001 return newSVsv (types_true); 1097 return newSVsv (types_true);
1002 case 22: 1098 case SIMPLE_NULL:
1003 return newSVsv (&PL_sv_undef); 1099 return newSVsv (&PL_sv_undef);
1004 case 23: 1100 case SIMPLE_UNDEF:
1005#if CBOR_SLOW 1101#if CBOR_SLOW
1006 types_error = get_bool ("Types::Serialiser::error"); 1102 types_error = get_bool ("Types::Serialiser::error");
1007#endif 1103#endif
1008 return newSVsv (types_error); 1104 return newSVsv (types_error);
1009 1105
1010 case 25: 1106 case MISC_FLOAT16:
1011 { 1107 {
1012 WANT (2); 1108 WANT (2);
1013 1109
1014 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1]; 1110 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1];
1015 dec->cur += 2; 1111 dec->cur += 2;
1016 1112
1017 return newSVnv (ecb_binary16_to_float (fp)); 1113 return newSVnv (ecb_binary16_to_float (fp));
1018 } 1114 }
1019 1115
1020 case 26: 1116 case MISC_FLOAT32:
1021 { 1117 {
1022 uint32_t fp; 1118 uint32_t fp;
1023 WANT (4); 1119 WANT (4);
1024 memcpy (&fp, dec->cur, 4); 1120 memcpy (&fp, dec->cur, 4);
1025 dec->cur += 4; 1121 dec->cur += 4;
1028 fp = ecb_bswap32 (fp); 1124 fp = ecb_bswap32 (fp);
1029 1125
1030 return newSVnv (ecb_binary32_to_float (fp)); 1126 return newSVnv (ecb_binary32_to_float (fp));
1031 } 1127 }
1032 1128
1033 case 27: 1129 case MISC_FLOAT64:
1034 { 1130 {
1035 uint64_t fp; 1131 uint64_t fp;
1036 WANT (8); 1132 WANT (8);
1037 memcpy (&fp, dec->cur, 8); 1133 memcpy (&fp, dec->cur, 8);
1038 dec->cur += 8; 1134 dec->cur += 8;
1041 fp = ecb_bswap64 (fp); 1137 fp = ecb_bswap64 (fp);
1042 1138
1043 return newSVnv (ecb_binary64_to_double (fp)); 1139 return newSVnv (ecb_binary64_to_double (fp));
1044 } 1140 }
1045 1141
1046 // 0..19 unassigned 1142 // 0..19 unassigned simple
1047 // 24 reserved + unassigned (reserved values are not encodable) 1143 // 24 reserved + unassigned simple (reserved values are not encodable)
1144 // 28-30 unassigned misc
1145 // 31 break code
1048 default: 1146 default:
1049 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)"); 1147 ERR ("corrupted CBOR data (reserved/unassigned/unexpected major 7 value)");
1050 } 1148 }
1051 1149
1052 break; 1150 break;
1053 } 1151 }
1054 1152
1081 if (dec.cur != dec.end && !dec.err) 1179 if (dec.cur != dec.end && !dec.err)
1082 dec.err = "garbage after CBOR object"; 1180 dec.err = "garbage after CBOR object";
1083 1181
1084 if (dec.err) 1182 if (dec.err)
1085 { 1183 {
1184 if (dec.shareable)
1185 {
1186 // need to break cyclic links, which whould all be in shareable
1187 int i;
1188 SV **svp;
1189
1190 for (i = av_len (dec.shareable) + 1; i--; )
1191 if ((svp = av_fetch (dec.shareable, i, 0)))
1192 sv_setsv (*svp, &PL_sv_undef);
1193 }
1194
1086 SvREFCNT_dec (sv); 1195 SvREFCNT_dec (sv);
1087 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur); 1196 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur);
1088 } 1197 }
1089 1198
1090 sv = sv_2mortal (sv); 1199 sv = sv_2mortal (sv);
1091 1200
1092 return sv; 1201 return sv;
1093} 1202}
1094 1203
1204/////////////////////////////////////////////////////////////////////////////
1205// incremental parser
1206
1207#define INCR_DONE(cbor) (AvFILLp (cbor->incr_count) < 0)
1208
1209// returns 0 for notyet, 1 for success or error
1210static int
1211incr_parse (CBOR *self, SV *cborstr)
1212{
1213 STRLEN cur;
1214 SvPV (cborstr, cur);
1215
1216 while (ecb_expect_true (self->incr_need <= cur))
1217 {
1218 // table of integer count bytes
1219 static I8 incr_len[MINOR_MASK + 1] = {
1220 0, 0, 0, 0, 0, 0, 0, 0,
1221 0, 0, 0, 0, 0, 0, 0, 0,
1222 0, 0, 0, 0, 0, 0, 0, 0,
1223 1, 2, 4, 8,-1,-1,-1,-2
1224 };
1225
1226 const U8 *p = SvPVX (cborstr) + self->incr_pos;
1227 U8 m = *p & MINOR_MASK;
1228 IV count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1229 I8 ilen = incr_len[m];
1230
1231 self->incr_need = self->incr_pos + 1;
1232
1233 if (ecb_expect_false (ilen < 0))
1234 {
1235 if (m != MINOR_INDEF)
1236 return 1; // error
1237
1238 if (*p == (MAJOR_MISC | MINOR_INDEF))
1239 {
1240 if (count >= 0)
1241 return 1; // error
1242
1243 count = 1;
1244 }
1245 else
1246 {
1247 av_push (self->incr_count, newSViv (-1)); //TODO: nest
1248 count = -1;
1249 }
1250 }
1251 else
1252 {
1253 self->incr_need += ilen;
1254 if (ecb_expect_false (self->incr_need > cur))
1255 return 0;
1256
1257 int major = *p >> MAJOR_SHIFT;
1258
1259 switch (major)
1260 {
1261 case MAJOR_BYTES >> MAJOR_SHIFT:
1262 case MAJOR_TEXT >> MAJOR_SHIFT:
1263 case MAJOR_ARRAY >> MAJOR_SHIFT:
1264 case MAJOR_MAP >> MAJOR_SHIFT:
1265 {
1266 UV len;
1267
1268 if (ecb_expect_false (ilen))
1269 {
1270 len = 0;
1271
1272 do {
1273 len = (len << 8) | *++p;
1274 } while (--ilen);
1275 }
1276 else
1277 len = m;
1278
1279 switch (major)
1280 {
1281 case MAJOR_BYTES >> MAJOR_SHIFT:
1282 case MAJOR_TEXT >> MAJOR_SHIFT:
1283 self->incr_need += len;
1284 if (ecb_expect_false (self->incr_need > cur))
1285 return 0;
1286
1287 break;
1288
1289 case MAJOR_MAP >> MAJOR_SHIFT:
1290 len <<= 1;
1291 case MAJOR_ARRAY >> MAJOR_SHIFT:
1292 if (len)
1293 {
1294 av_push (self->incr_count, newSViv (len + 1)); //TODO: nest
1295 count = len + 1;
1296 }
1297 break;
1298 }
1299 }
1300 }
1301 }
1302
1303 self->incr_pos = self->incr_need;
1304
1305 if (count > 0)
1306 {
1307 while (!--count)
1308 {
1309 if (!AvFILLp (self->incr_count))
1310 return 1; // done
1311
1312 SvREFCNT_dec_NN (av_pop (self->incr_count));
1313 count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1314 }
1315
1316 SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]) = count;
1317 }
1318 }
1319
1320 return 0;
1321}
1322
1323
1095///////////////////////////////////////////////////////////////////////////// 1324/////////////////////////////////////////////////////////////////////////////
1096// XS interface functions 1325// XS interface functions
1097 1326
1098MODULE = CBOR::XS PACKAGE = CBOR::XS 1327MODULE = CBOR::XS PACKAGE = CBOR::XS
1099 1328
1139void shrink (CBOR *self, int enable = 1) 1368void shrink (CBOR *self, int enable = 1)
1140 ALIAS: 1369 ALIAS:
1141 shrink = F_SHRINK 1370 shrink = F_SHRINK
1142 allow_unknown = F_ALLOW_UNKNOWN 1371 allow_unknown = F_ALLOW_UNKNOWN
1143 allow_sharing = F_ALLOW_SHARING 1372 allow_sharing = F_ALLOW_SHARING
1144 allow_stringref = F_ALLOW_STRINGREF 1373 allow_cycles = F_ALLOW_CYCLES
1374 pack_strings = F_PACK_STRINGS
1375 validate_utf8 = F_VALIDATE_UTF8
1145 PPCODE: 1376 PPCODE:
1146{ 1377{
1147 if (enable) 1378 if (enable)
1148 self->flags |= ix; 1379 self->flags |= ix;
1149 else 1380 else
1155void get_shrink (CBOR *self) 1386void get_shrink (CBOR *self)
1156 ALIAS: 1387 ALIAS:
1157 get_shrink = F_SHRINK 1388 get_shrink = F_SHRINK
1158 get_allow_unknown = F_ALLOW_UNKNOWN 1389 get_allow_unknown = F_ALLOW_UNKNOWN
1159 get_allow_sharing = F_ALLOW_SHARING 1390 get_allow_sharing = F_ALLOW_SHARING
1160 get_allow_stringref = F_ALLOW_STRINGREF 1391 get_allow_cycles = F_ALLOW_CYCLES
1392 get_pack_strings = F_PACK_STRINGS
1393 get_validate_utf8 = F_VALIDATE_UTF8
1161 PPCODE: 1394 PPCODE:
1162 XPUSHs (boolSV (self->flags & ix)); 1395 XPUSHs (boolSV (self->flags & ix));
1163 1396
1164void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) 1397void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
1165 PPCODE: 1398 PPCODE:
1214 EXTEND (SP, 2); 1447 EXTEND (SP, 2);
1215 PUSHs (sv); 1448 PUSHs (sv);
1216 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); 1449 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
1217} 1450}
1218 1451
1452void incr_parse (CBOR *self, SV *cborstr, int chop = 0)
1453 PPCODE:
1454{
1455 if (SvUTF8 (cborstr))
1456 sv_utf8_downgrade (cborstr, 0);
1457
1458 if (!self->incr_count)
1459 {
1460 self->incr_count = newAV ();
1461 self->incr_pos = 0;
1462 self->incr_need = 1;
1463
1464 av_push (self->incr_count, newSViv (1));
1465 }
1466
1467 for (;;)
1468 {
1469 if (!incr_parse (self, cborstr))
1470 {
1471 if (self->incr_need > self->max_size && self->max_size)
1472 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
1473 (unsigned long)self->incr_need, (unsigned long)self->max_size);
1474
1475 break;
1476 }
1477
1478 SV *sv;
1479 char *offset;
1480
1481 PUTBACK; sv = decode_cbor (cborstr, self, &offset); SPAGAIN;
1482 XPUSHs (sv);
1483
1484 av_clear (self->incr_count);
1485 av_push (self->incr_count, newSViv (1));
1486
1487 if (chop)
1488 {
1489 self->incr_pos = 0;
1490 sv_chop (cborstr, offset);
1491 }
1492 else
1493 self->incr_pos = offset - SvPVX (cborstr);
1494
1495 self->incr_need = self->incr_pos + 1;
1496 }
1497}
1498
1499void incr_reset (CBOR *self)
1500 CODE:
1501{
1502 SvREFCNT_dec (self->incr_count);
1503 self->incr_count = 0;
1504}
1505
1219void DESTROY (CBOR *self) 1506void DESTROY (CBOR *self)
1220 PPCODE: 1507 PPCODE:
1221 cbor_free (self); 1508 cbor_free (self);
1222 1509
1223PROTOTYPES: ENABLE 1510PROTOTYPES: ENABLE
1224 1511
1225void encode_cbor (SV *scalar) 1512void encode_cbor (SV *scalar)
1513 ALIAS:
1514 encode_cbor = 0
1515 encode_cbor_sharing = F_ALLOW_SHARING
1226 PPCODE: 1516 PPCODE:
1227{ 1517{
1228 CBOR cbor; 1518 CBOR cbor;
1229 cbor_init (&cbor); 1519 cbor_init (&cbor);
1520 cbor.flags |= ix;
1230 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN; 1521 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;
1231 XPUSHs (scalar); 1522 XPUSHs (scalar);
1232} 1523}
1233 1524
1234void decode_cbor (SV *cborstr) 1525void decode_cbor (SV *cborstr)

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines