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 1.44 by root, Tue Feb 18 22:12:12 2014 UTC

… …
7#include <stdlib.h> 7#include <stdlib.h>
8#include <stdio.h> 8#include <stdio.h>
9#include <limits.h> 9#include <limits.h>
10#include <float.h> 10#include <float.h>
11 11
12#define ECB_NO_THREADS 1
12#include "ecb.h" 13#include "ecb.h"
13 14
14// compatibility with perl <5.18 15// compatibility with perl <5.18
15#ifndef HvNAMELEN_get 16#ifndef HvNAMELEN_get
16# define HvNAMELEN_get(hv) strlen (HvNAME (hv)) 17# define HvNAMELEN_get(hv) strlen (HvNAME (hv))
… …
23#endif 24#endif
24#ifndef SvREFCNT_dec_NN 25#ifndef SvREFCNT_dec_NN
25# define SvREFCNT_dec_NN(sv) SvREFCNT_dec (sv) 26# define SvREFCNT_dec_NN(sv) SvREFCNT_dec (sv)
26#endif 27#endif
27 28
29// known major and minor types
30enum cbor_type
31{
32 MAJOR_SHIFT = 5,
33 MINOR_MASK = 0x1f,
34
35 MAJOR_POS_INT = 0 << MAJOR_SHIFT,
36 MAJOR_NEG_INT = 1 << MAJOR_SHIFT,
37 MAJOR_BYTES = 2 << MAJOR_SHIFT,
38 MAJOR_TEXT = 3 << MAJOR_SHIFT,
39 MAJOR_ARRAY = 4 << MAJOR_SHIFT,
40 MAJOR_MAP = 5 << MAJOR_SHIFT,
41 MAJOR_TAG = 6 << MAJOR_SHIFT,
42 MAJOR_MISC = 7 << MAJOR_SHIFT,
43
44 // INT/STRING/ARRAY/MAP subtypes
45 LENGTH_EXT1 = 24,
46 LENGTH_EXT2 = 25,
47 LENGTH_EXT4 = 26,
48 LENGTH_EXT8 = 27,
49
50 // SIMPLE types (effectively MISC subtypes)
51 SIMPLE_FALSE = 20,
52 SIMPLE_TRUE = 21,
53 SIMPLE_NULL = 22,
54 SIMPLE_UNDEF = 23,
55
56 // MISC subtype (unused)
57 MISC_EXT1 = 24,
58 MISC_FLOAT16 = 25,
59 MISC_FLOAT32 = 26,
60 MISC_FLOAT64 = 27,
61
62 // BYTES/TEXT/ARRAY/MAP
63 MINOR_INDEF = 31,
64};
65
28// known tags 66// known tags
29enum cbor_tag 67enum cbor_tag
30{ 68{
31 // extensions 69 // extensions
32 CBOR_TAG_STRINGREF = 25, // http://cbor.schmorp.de/stringref 70 CBOR_TAG_STRINGREF = 25, // http://cbor.schmorp.de/stringref
33 CBOR_TAG_PERL_OBJECT = 26, // http://cbor.schmorp.de/perl-object 71 CBOR_TAG_PERL_OBJECT = 26, // http://cbor.schmorp.de/perl-object
34 CBOR_TAG_GENERIC_OBJECT = 27, // http://cbor.schmorp.de/generic-object 72 CBOR_TAG_GENERIC_OBJECT = 27, // http://cbor.schmorp.de/generic-object
35 CBOR_TAG_VALUE_SHAREABLE = 28, // http://cbor.schmorp.de/value-sharing 73 CBOR_TAG_VALUE_SHAREABLE = 28, // http://cbor.schmorp.de/value-sharing
36 CBOR_TAG_VALUE_SHAREDREF = 29, // http://cbor.schmorp.de/value-sharing 74 CBOR_TAG_VALUE_SHAREDREF = 29, // http://cbor.schmorp.de/value-sharing
37 CBOR_TAG_STRINGREF_NAMESPACE = 256, // http://cbor.schmorp.de/stringref 75 CBOR_TAG_STRINGREF_NAMESPACE = 256, // http://cbor.schmorp.de/stringref
38 CBOR_TAG_INDIRECTION = 22098, // http://cbor.schmorp.de/indirection 76 CBOR_TAG_INDIRECTION = 22098, // http://cbor.schmorp.de/indirection
39 77
40 // rfc7049 78 // rfc7049
41 CBOR_TAG_DATETIME = 0, // rfc4287, utf-8 79 CBOR_TAG_DATETIME = 0, // rfc4287, utf-8
42 CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any 80 CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any
43 CBOR_TAG_POS_BIGNUM = 2, // byte string 81 CBOR_TAG_POS_BIGNUM = 2, // byte string
44 CBOR_TAG_NEG_BIGNUM = 3, // byte string 82 CBOR_TAG_NEG_BIGNUM = 3, // byte string
45 CBOR_TAG_DECIMAL = 4, // decimal fraction, array 83 CBOR_TAG_DECIMAL = 4, // decimal fraction, array
46 CBOR_TAG_BIGFLOAT = 5, // array 84 CBOR_TAG_BIGFLOAT = 5, // array
47 85
48 CBOR_TAG_CONV_B64U = 21, // base64url, any 86 CBOR_TAG_CONV_B64U = 21, // base64url, any
49 CBOR_TAG_CONV_B64 = 22, // base64, any 87 CBOR_TAG_CONV_B64 = 22, // base64, any
50 CBOR_TAG_CONV_HEX = 23, // base16, any 88 CBOR_TAG_CONV_HEX = 23, // base16, any
51 CBOR_TAG_CBOR = 24, // embedded cbor, byte string 89 CBOR_TAG_CBOR = 24, // embedded cbor, byte string
52 90
53 CBOR_TAG_URI = 32, // URI rfc3986, utf-8 91 CBOR_TAG_URI = 32, // URI rfc3986, utf-8
54 CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8 92 CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8
55 CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8 93 CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8
56 CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8 94 CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8
57 CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8 95 CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8
58 96
59 CBOR_TAG_MAGIC = 55799 // self-describe cbor 97 CBOR_TAG_MAGIC = 55799, // self-describe cbor
60}; 98};
61 99
62#define F_SHRINK 0x00000001UL 100#define F_SHRINK 0x00000001UL
63#define F_ALLOW_UNKNOWN 0x00000002UL 101#define F_ALLOW_UNKNOWN 0x00000002UL
64#define F_ALLOW_SHARING 0x00000004UL //TODO 102#define F_ALLOW_SHARING 0x00000004UL
65#define F_ALLOW_STRINGREF 0x00000008UL //TODO 103#define F_ALLOW_CYCLES 0x00000008UL
104#define F_PACK_STRINGS 0x00000010UL
105#define F_VALIDATE_UTF8 0x00000020UL
66 106
67#define INIT_SIZE 32 // initial scalar size to be allocated 107#define INIT_SIZE 32 // initial scalar size to be allocated
68 108
69#define SB do { 109#define SB do {
70#define SE } while (0) 110#define SE } while (0)
… …
89typedef struct { 129typedef struct {
90 U32 flags; 130 U32 flags;
91 U32 max_depth; 131 U32 max_depth;
92 STRLEN max_size; 132 STRLEN max_size;
93 SV *filter; 133 SV *filter;
134
135 // for the incremental parser
136 STRLEN incr_pos; // the current offset into the text
137 STRLEN incr_need; // minimum bytes needed to decode
138 AV *incr_count; // for every nesting level, the number of outstanding values, or -1 for indef.
94} CBOR; 139} CBOR;
95 140
96ecb_inline void 141ecb_inline void
97cbor_init (CBOR *cbor) 142cbor_init (CBOR *cbor)
98{ 143{
… …
102 147
103ecb_inline void 148ecb_inline void
104cbor_free (CBOR *cbor) 149cbor_free (CBOR *cbor)
105{ 150{
106 SvREFCNT_dec (cbor->filter); 151 SvREFCNT_dec (cbor->filter);
152 SvREFCNT_dec (cbor->incr_count);
107} 153}
108 154
109///////////////////////////////////////////////////////////////////////////// 155/////////////////////////////////////////////////////////////////////////////
110// utility functions 156// utility functions
111 157
… …
189static void 235static void
190encode_uint (enc_t *enc, int major, UV len) 236encode_uint (enc_t *enc, int major, UV len)
191{ 237{
192 need (enc, 9); 238 need (enc, 9);
193 239
194 if (len < 24) 240 if (ecb_expect_true (len < LENGTH_EXT1))
195 *enc->cur++ = major | len; 241 *enc->cur++ = major | len;
196 else if (len <= 0xff) 242 else if (ecb_expect_true (len <= 0xffU))
197 { 243 {
198 *enc->cur++ = major | 24; 244 *enc->cur++ = major | LENGTH_EXT1;
199 *enc->cur++ = len; 245 *enc->cur++ = len;
200 } 246 }
201 else if (len <= 0xffff) 247 else if (len <= 0xffffU)
202 { 248 {
203 *enc->cur++ = major | 25; 249 *enc->cur++ = major | LENGTH_EXT2;
204 *enc->cur++ = len >> 8; 250 *enc->cur++ = len >> 8;
205 *enc->cur++ = len; 251 *enc->cur++ = len;
206 } 252 }
207 else if (len <= 0xffffffff) 253 else if (len <= 0xffffffffU)
208 { 254 {
209 *enc->cur++ = major | 26; 255 *enc->cur++ = major | LENGTH_EXT4;
210 *enc->cur++ = len >> 24; 256 *enc->cur++ = len >> 24;
211 *enc->cur++ = len >> 16; 257 *enc->cur++ = len >> 16;
212 *enc->cur++ = len >> 8; 258 *enc->cur++ = len >> 8;
213 *enc->cur++ = len; 259 *enc->cur++ = len;
214 } 260 }
215 else 261 else
216 { 262 {
217 *enc->cur++ = major | 27; 263 *enc->cur++ = major | LENGTH_EXT8;
218 *enc->cur++ = len >> 56; 264 *enc->cur++ = len >> 56;
219 *enc->cur++ = len >> 48; 265 *enc->cur++ = len >> 48;
220 *enc->cur++ = len >> 40; 266 *enc->cur++ = len >> 40;
221 *enc->cur++ = len >> 32; 267 *enc->cur++ = len >> 32;
222 *enc->cur++ = len >> 24; 268 *enc->cur++ = len >> 24;
… …
227} 273}
228 274
229ecb_inline void 275ecb_inline void
230encode_tag (enc_t *enc, UV tag) 276encode_tag (enc_t *enc, UV tag)
231{ 277{
232 encode_uint (enc, 0xc0, tag); 278 encode_uint (enc, MAJOR_TAG, tag);
279}
280
281ecb_inline void
282encode_str (enc_t *enc, int utf8, char *str, STRLEN len)
283{
284 encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len);
285 need (enc, len);
286 memcpy (enc->cur, str, len);
287 enc->cur += len;
233} 288}
234 289
235static void 290static void
236encode_str (enc_t *enc, int utf8, char *str, STRLEN len) 291encode_strref (enc_t *enc, int utf8, char *str, STRLEN len)
237{ 292{
238 if (ecb_expect_false (enc->cbor.flags & F_ALLOW_STRINGREF)) 293 if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS))
239 { 294 {
240 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1); 295 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1);
241 296
242 if (SvOK (*svp)) 297 if (SvOK (*svp))
243 { 298 {
244 // already registered, use stringref 299 // already registered, use stringref
245 encode_tag (enc, CBOR_TAG_STRINGREF); 300 encode_tag (enc, CBOR_TAG_STRINGREF);
246 encode_uint (enc, 0x00, SvUV (*svp)); 301 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
247 return; 302 return;
248 } 303 }
249 else if (len >= minimum_string_length (enc->stringref_idx)) 304 else if (len >= minimum_string_length (enc->stringref_idx))
250 { 305 {
251 // register only 306 // register only
252 sv_setuv (*svp, enc->stringref_idx); 307 sv_setuv (*svp, enc->stringref_idx);
253 ++enc->stringref_idx; 308 ++enc->stringref_idx;
254 } 309 }
255 } 310 }
256 311
257 encode_uint (enc, utf8 ? 0x60 : 0x40, len); 312 encode_str (enc, utf8, str, len);
258 need (enc, len);
259 memcpy (enc->cur, str, len);
260 enc->cur += len;
261} 313}
262 314
263static void encode_sv (enc_t *enc, SV *sv); 315static void encode_sv (enc_t *enc, SV *sv);
264 316
265static void 317static void
… …
270 if (enc->depth >= enc->cbor.max_depth) 322 if (enc->depth >= enc->cbor.max_depth)
271 croak (ERR_NESTING_EXCEEDED); 323 croak (ERR_NESTING_EXCEEDED);
272 324
273 ++enc->depth; 325 ++enc->depth;
274 326
275 encode_uint (enc, 0x80, len + 1); 327 encode_uint (enc, MAJOR_ARRAY, len + 1);
276 328
277 for (i = 0; i <= len; ++i) 329 for (i = 0; i <= len; ++i)
278 { 330 {
279 SV **svp = av_fetch (av, i, 0); 331 SV **svp = av_fetch (av, i, 0);
280 encode_sv (enc, svp ? *svp : &PL_sv_undef); 332 encode_sv (enc, svp ? *svp : &PL_sv_undef);
… …
295 347
296 int pairs = hv_iterinit (hv); 348 int pairs = hv_iterinit (hv);
297 int mg = SvMAGICAL (hv); 349 int mg = SvMAGICAL (hv);
298 350
299 if (mg) 351 if (mg)
300 encode_ch (enc, 0xa0 | 31); 352 encode_ch (enc, MAJOR_MAP | MINOR_INDEF);
301 else 353 else
302 encode_uint (enc, 0xa0, pairs); 354 encode_uint (enc, MAJOR_MAP, pairs);
303 355
304 while ((he = hv_iternext (hv))) 356 while ((he = hv_iternext (hv)))
305 { 357 {
306 if (HeKLEN (he) == HEf_SVKEY) 358 if (HeKLEN (he) == HEf_SVKEY)
307 encode_sv (enc, HeSVKEY (he)); 359 encode_sv (enc, HeSVKEY (he));
308 else 360 else
309 encode_str (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he)); 361 encode_strref (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he));
310 362
311 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he)); 363 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he));
312 } 364 }
313 365
314 if (mg) 366 if (mg)
315 encode_ch (enc, 0xe0 | 31); 367 encode_ch (enc, MAJOR_MISC | MINOR_INDEF);
316 368
317 --enc->depth; 369 --enc->depth;
318} 370}
319 371
320// encode objects, arrays and special \0=false and \1=true values. 372// encode objects, arrays and special \0=false and \1=true values.
321static void 373static void
322encode_rv (enc_t *enc, SV *sv) 374encode_rv (enc_t *enc, SV *sv)
323{ 375{
324 SvGETMAGIC (sv); 376 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 }
347 377
348 svtype svt = SvTYPE (sv); 378 svtype svt = SvTYPE (sv);
349 379
350 if (ecb_expect_false (SvOBJECT (sv))) 380 if (ecb_expect_false (SvOBJECT (sv)))
351 { 381 {
… …
358 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 388 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
359 ? cbor_tagged_stash 389 ? cbor_tagged_stash
360 : gv_stashpv ("CBOR::XS::Tagged" , 1); 390 : gv_stashpv ("CBOR::XS::Tagged" , 1);
361 391
362 HV *stash = SvSTASH (sv); 392 HV *stash = SvSTASH (sv);
363 GV *method;
364 393
365 if (stash == boolean_stash) 394 if (stash == boolean_stash)
366 encode_ch (enc, SvIV (sv) ? 0xe0 | 21 : 0xe0 | 20); 395 {
396 encode_ch (enc, SvIV (sv) ? MAJOR_MISC | SIMPLE_TRUE : MAJOR_MISC | SIMPLE_FALSE);
397 return;
398 }
367 else if (stash == error_stash) 399 else if (stash == error_stash)
368 encode_ch (enc, 0xe0 | 23); 400 {
401 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
402 return;
403 }
369 else if (stash == tagged_stash) 404 else if (stash == tagged_stash)
370 { 405 {
371 if (svt != SVt_PVAV) 406 if (svt != SVt_PVAV)
372 croak ("encountered CBOR::XS::Tagged object that isn't an array"); 407 croak ("encountered CBOR::XS::Tagged object that isn't an array");
373 408
374 encode_uint (enc, 0xc0, SvUV (*av_fetch ((AV *)sv, 0, 1))); 409 encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1)));
375 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1)); 410 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1));
411
412 return;
413 }
414 }
415
416 if (ecb_expect_false (SvREFCNT (sv) > 1)
417 && ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING))
418 {
419 if (!enc->shareable)
420 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
421
422 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
423
424 if (SvOK (*svp))
376 } 425 {
426 encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF);
427 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
428 return;
429 }
430 else
431 {
432 sv_setuv (*svp, enc->shareable_idx);
433 ++enc->shareable_idx;
434 encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE);
435 }
436 }
437
438 if (ecb_expect_false (SvOBJECT (sv)))
439 {
440 HV *stash = SvSTASH (sv);
441 GV *method;
442
377 else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0))) 443 if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
378 { 444 {
379 dSP; 445 dSP;
380 446
381 ENTER; SAVETMPS; PUSHMARK (SP); 447 ENTER; SAVETMPS; PUSHMARK (SP);
382 // we re-bless the reference to get overload and other niceties right 448 // we re-bless the reference to get overload and other niceties right
… …
414 // catch this surprisingly common error 480 // catch this surprisingly common error
415 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv) 481 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)); 482 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));
417 483
418 encode_tag (enc, CBOR_TAG_PERL_OBJECT); 484 encode_tag (enc, CBOR_TAG_PERL_OBJECT);
419 encode_uint (enc, 0x80, count + 1); 485 encode_uint (enc, MAJOR_ARRAY, count + 1);
420 encode_str (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash)); 486 encode_strref (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
421 487
422 while (count) 488 while (count)
423 encode_sv (enc, SP[1 - count--]); 489 encode_sv (enc, SP[1 - count--]);
424 490
425 PUTBACK; 491 PUTBACK;
… …
446{ 512{
447 double nv = SvNVX (sv); 513 double nv = SvNVX (sv);
448 514
449 need (enc, 9); 515 need (enc, 9);
450 516
451 if (ecb_expect_false (nv == (U32)nv)) 517 if (ecb_expect_false (nv == (NV)(U32)nv))
452 encode_uint (enc, 0x00, (U32)nv); 518 encode_uint (enc, MAJOR_POS_INT, (U32)nv);
453 //TODO: maybe I32? 519 //TODO: maybe I32?
454 else if (ecb_expect_false (nv == (float)nv)) 520 else if (ecb_expect_false (nv == (float)nv))
455 { 521 {
456 uint32_t fp = ecb_float_to_binary32 (nv); 522 uint32_t fp = ecb_float_to_binary32 (nv);
457 523
458 *enc->cur++ = 0xe0 | 26; 524 *enc->cur++ = MAJOR_MISC | MISC_FLOAT32;
459 525
460 if (!ecb_big_endian ()) 526 if (!ecb_big_endian ())
461 fp = ecb_bswap32 (fp); 527 fp = ecb_bswap32 (fp);
462 528
463 memcpy (enc->cur, &fp, 4); 529 memcpy (enc->cur, &fp, 4);
… …
465 } 531 }
466 else 532 else
467 { 533 {
468 uint64_t fp = ecb_double_to_binary64 (nv); 534 uint64_t fp = ecb_double_to_binary64 (nv);
469 535
470 *enc->cur++ = 0xe0 | 27; 536 *enc->cur++ = MAJOR_MISC | MISC_FLOAT64;
471 537
472 if (!ecb_big_endian ()) 538 if (!ecb_big_endian ())
473 fp = ecb_bswap64 (fp); 539 fp = ecb_bswap64 (fp);
474 540
475 memcpy (enc->cur, &fp, 8); 541 memcpy (enc->cur, &fp, 8);
… …
484 550
485 if (SvPOKp (sv)) 551 if (SvPOKp (sv))
486 { 552 {
487 STRLEN len; 553 STRLEN len;
488 char *str = SvPV (sv, len); 554 char *str = SvPV (sv, len);
489 encode_str (enc, SvUTF8 (sv), str, len); 555 encode_strref (enc, SvUTF8 (sv), str, len);
490 } 556 }
491 else if (SvNOKp (sv)) 557 else if (SvNOKp (sv))
492 encode_nv (enc, sv); 558 encode_nv (enc, sv);
493 else if (SvIOKp (sv)) 559 else if (SvIOKp (sv))
494 { 560 {
495 if (SvIsUV (sv)) 561 if (SvIsUV (sv))
496 encode_uint (enc, 0x00, SvUVX (sv)); 562 encode_uint (enc, MAJOR_POS_INT, SvUVX (sv));
497 else if (SvIVX (sv) >= 0) 563 else if (SvIVX (sv) >= 0)
498 encode_uint (enc, 0x00, SvIVX (sv)); 564 encode_uint (enc, MAJOR_POS_INT, SvIVX (sv));
499 else 565 else
500 encode_uint (enc, 0x20, -(SvIVX (sv) + 1)); 566 encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1));
501 } 567 }
502 else if (SvROK (sv)) 568 else if (SvROK (sv))
503 encode_rv (enc, SvRV (sv)); 569 encode_rv (enc, SvRV (sv));
504 else if (!SvOK (sv)) 570 else if (!SvOK (sv))
505 encode_ch (enc, 0xe0 | 22); 571 encode_ch (enc, MAJOR_MISC | SIMPLE_NULL);
506 else if (enc->cbor.flags & F_ALLOW_UNKNOWN) 572 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
507 encode_ch (enc, 0xe0 | 23); 573 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
508 else 574 else
509 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data", 575 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data",
510 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv)); 576 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
511} 577}
512 578
… …
520 enc.cur = SvPVX (enc.sv); 586 enc.cur = SvPVX (enc.sv);
521 enc.end = SvEND (enc.sv); 587 enc.end = SvEND (enc.sv);
522 588
523 SvPOK_only (enc.sv); 589 SvPOK_only (enc.sv);
524 590
525 if (cbor->flags & F_ALLOW_STRINGREF) 591 if (cbor->flags & F_PACK_STRINGS)
526 { 592 {
527 encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE); 593 encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE);
528 enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ()); 594 enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ());
529 enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ()); 595 enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ());
530 } 596 }
… …
565#define DEC_DEC_DEPTH --dec->depth 631#define DEC_DEC_DEPTH --dec->depth
566 632
567static UV 633static UV
568decode_uint (dec_t *dec) 634decode_uint (dec_t *dec)
569{ 635{
570 switch (*dec->cur & 31) 636 U8 m = *dec->cur & MINOR_MASK;
571 { 637 ++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;
576 638
577 case 24: 639 if (ecb_expect_true (m < LENGTH_EXT1))
640 return m;
641 else if (ecb_expect_true (m == LENGTH_EXT1))
642 {
578 WANT (2); 643 WANT (1);
579 dec->cur += 2; 644 dec->cur += 1;
580 return dec->cur[-1]; 645 return dec->cur[-1];
581 646 }
582 case 25: 647 else if (ecb_expect_true (m == LENGTH_EXT2))
648 {
583 WANT (3); 649 WANT (2);
584 dec->cur += 3; 650 dec->cur += 2;
585 return (((UV)dec->cur[-2]) << 8) 651 return (((UV)dec->cur[-2]) << 8)
586 | ((UV)dec->cur[-1]); 652 | ((UV)dec->cur[-1]);
587 653 }
588 case 26: 654 else if (ecb_expect_true (m == LENGTH_EXT4))
655 {
589 WANT (5); 656 WANT (4);
590 dec->cur += 5; 657 dec->cur += 4;
591 return (((UV)dec->cur[-4]) << 24) 658 return (((UV)dec->cur[-4]) << 24)
592 | (((UV)dec->cur[-3]) << 16) 659 | (((UV)dec->cur[-3]) << 16)
593 | (((UV)dec->cur[-2]) << 8) 660 | (((UV)dec->cur[-2]) << 8)
594 | ((UV)dec->cur[-1]); 661 | ((UV)dec->cur[-1]);
595 662 }
596 case 27: 663 else if (ecb_expect_true (m == LENGTH_EXT8))
664 {
597 WANT (9); 665 WANT (8);
598 dec->cur += 9; 666 dec->cur += 8;
667
668 return
669#if UVSIZE < 8
670 0
671#else
599 return (((UV)dec->cur[-8]) << 56) 672 (((UV)dec->cur[-8]) << 56)
600 | (((UV)dec->cur[-7]) << 48) 673 | (((UV)dec->cur[-7]) << 48)
601 | (((UV)dec->cur[-6]) << 40) 674 | (((UV)dec->cur[-6]) << 40)
602 | (((UV)dec->cur[-5]) << 32) 675 | (((UV)dec->cur[-5]) << 32)
676#endif
603 | (((UV)dec->cur[-4]) << 24) 677 | (((UV)dec->cur[-4]) << 24)
604 | (((UV)dec->cur[-3]) << 16) 678 | (((UV)dec->cur[-3]) << 16)
605 | (((UV)dec->cur[-2]) << 8) 679 | (((UV)dec->cur[-2]) << 8)
606 | ((UV)dec->cur[-1]); 680 | ((UV)dec->cur[-1]);
607 681 }
608 default: 682 else
609 ERR ("corrupted CBOR data (unsupported integer minor encoding)"); 683 ERR ("corrupted CBOR data (unsupported integer minor encoding)");
610 }
611 684
612fail: 685fail:
613 return 0; 686 return 0;
614} 687}
615 688
… …
620{ 693{
621 AV *av = newAV (); 694 AV *av = newAV ();
622 695
623 DEC_INC_DEPTH; 696 DEC_INC_DEPTH;
624 697
625 if ((*dec->cur & 31) == 31) 698 if (*dec->cur == (MAJOR_ARRAY | MINOR_INDEF))
626 { 699 {
627 ++dec->cur; 700 ++dec->cur;
628 701
629 for (;;) 702 for (;;)
630 { 703 {
631 WANT (1); 704 WANT (1);
632 705
633 if (*dec->cur == (0xe0 | 31)) 706 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
634 { 707 {
635 ++dec->cur; 708 ++dec->cur;
636 break; 709 break;
637 } 710 }
638 711
… …
641 } 714 }
642 else 715 else
643 { 716 {
644 int i, len = decode_uint (dec); 717 int i, len = decode_uint (dec);
645 718
719 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); 720 av_fill (av, len - 1);
647 721
648 for (i = 0; i < len; ++i) 722 for (i = 0; i < len; ++i)
649 AvARRAY (av)[i] = decode_sv (dec); 723 AvARRAY (av)[i] = decode_sv (dec);
650 } 724 }
… …
663{ 737{
664 // for speed reasons, we specialcase single-string 738 // for speed reasons, we specialcase single-string
665 // byte or utf-8 strings as keys, but only when !stringref 739 // byte or utf-8 strings as keys, but only when !stringref
666 740
667 if (ecb_expect_true (!dec->stringref)) 741 if (ecb_expect_true (!dec->stringref))
668 if (*dec->cur >= 0x40 && *dec->cur <= 0x40 + 27) 742 if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8))
669 { 743 {
670 I32 len = decode_uint (dec); 744 I32 len = decode_uint (dec);
671 char *key = (char *)dec->cur; 745 char *key = (char *)dec->cur;
672 746
673 dec->cur += len; 747 dec->cur += len;
674 748
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); 749 hv_store (hv, key, len, decode_sv (dec), 0);
679 750
680 return; 751 return;
681 } 752 }
682 else if (*dec->cur >= 0x60 && *dec->cur <= 0x60 + 27) 753 else if (ecb_expect_true ((U8)(*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8))
683 { 754 {
684 I32 len = decode_uint (dec); 755 I32 len = decode_uint (dec);
685 char *key = (char *)dec->cur; 756 char *key = (char *)dec->cur;
686 757
687 dec->cur += len; 758 dec->cur += len;
688 759
689 if (ecb_expect_false (dec->stringref)) 760 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
690 av_push (dec->stringref, newSVpvn_utf8 (key, len, 1)); 761 if (!is_utf8_string (key, len))
762 ERR ("corrupted CBOR data (invalid UTF-8 in map key)");
691 763
692 hv_store (hv, key, -len, decode_sv (dec), 0); 764 hv_store (hv, key, -len, decode_sv (dec), 0);
693 765
694 return; 766 return;
695 } 767 }
… …
697 SV *k = decode_sv (dec); 769 SV *k = decode_sv (dec);
698 SV *v = decode_sv (dec); 770 SV *v = decode_sv (dec);
699 771
700 hv_store_ent (hv, k, v, 0); 772 hv_store_ent (hv, k, v, 0);
701 SvREFCNT_dec (k); 773 SvREFCNT_dec (k);
774
775fail:
776 ;
702} 777}
703 778
704static SV * 779static SV *
705decode_hv (dec_t *dec) 780decode_hv (dec_t *dec)
706{ 781{
707 HV *hv = newHV (); 782 HV *hv = newHV ();
708 783
709 DEC_INC_DEPTH; 784 DEC_INC_DEPTH;
710 785
711 if ((*dec->cur & 31) == 31) 786 if (*dec->cur == (MAJOR_MAP | MINOR_INDEF))
712 { 787 {
713 ++dec->cur; 788 ++dec->cur;
714 789
715 for (;;) 790 for (;;)
716 { 791 {
717 WANT (1); 792 WANT (1);
718 793
719 if (*dec->cur == (0xe0 | 31)) 794 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
720 { 795 {
721 ++dec->cur; 796 ++dec->cur;
722 break; 797 break;
723 } 798 }
724 799
… …
745static SV * 820static SV *
746decode_str (dec_t *dec, int utf8) 821decode_str (dec_t *dec, int utf8)
747{ 822{
748 SV *sv = 0; 823 SV *sv = 0;
749 824
750 if ((*dec->cur & 31) == 31) 825 if ((*dec->cur & MINOR_MASK) == MINOR_INDEF)
751 { 826 {
827 // indefinite length strings
752 ++dec->cur; 828 ++dec->cur;
753 829
830 U8 major = *dec->cur & MAJOR_MISC;
831
754 sv = newSVpvn ("", 0); 832 sv = newSVpvn ("", 0);
755 833
756 // not very fast, and certainly not robust against illegal input
757 for (;;) 834 for (;;)
758 { 835 {
759 WANT (1); 836 WANT (1);
760 837
761 if (*dec->cur == (0xe0 | 31)) 838 if ((*dec->cur - major) > LENGTH_EXT8)
839 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
762 { 840 {
763 ++dec->cur; 841 ++dec->cur;
764 break; 842 break;
765 } 843 }
844 else
845 ERR ("corrupted CBOR data (invalid chunks in indefinite length string)");
766 846
767 sv_catsv (sv, decode_sv (dec)); 847 STRLEN len = decode_uint (dec);
848
849 WANT (len);
850 sv_catpvn (sv, dec->cur, len);
851 dec->cur += len;
768 } 852 }
769 } 853 }
770 else 854 else
771 { 855 {
772 STRLEN len = decode_uint (dec); 856 STRLEN len = decode_uint (dec);
… …
779 && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1)) 863 && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1))
780 av_push (dec->stringref, SvREFCNT_inc_NN (sv)); 864 av_push (dec->stringref, SvREFCNT_inc_NN (sv));
781 } 865 }
782 866
783 if (utf8) 867 if (utf8)
868 {
869 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
870 if (!is_utf8_string (SvPVX (sv), SvCUR (sv)))
871 ERR ("corrupted CBOR data (invalid UTF-8 in text string)");
872
784 SvUTF8_on (sv); 873 SvUTF8_on (sv);
874 }
785 875
786 return sv; 876 return sv;
787 877
788fail: 878fail:
789 SvREFCNT_dec (sv); 879 SvREFCNT_dec (sv);
… …
821 } 911 }
822 break; 912 break;
823 913
824 case CBOR_TAG_STRINGREF: 914 case CBOR_TAG_STRINGREF:
825 { 915 {
826 if ((*dec->cur >> 5) != 0) 916 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
827 ERR ("corrupted CBOR data (stringref index not an unsigned integer)"); 917 ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
828 918
829 UV idx = decode_uint (dec); 919 UV idx = decode_uint (dec);
830 920
831 if (!dec->stringref || (int)idx > AvFILLp (dec->stringref)) 921 if (!dec->stringref || (int)idx > AvFILLp (dec->stringref))
… …
838 case CBOR_TAG_VALUE_SHAREABLE: 928 case CBOR_TAG_VALUE_SHAREABLE:
839 { 929 {
840 if (ecb_expect_false (!dec->shareable)) 930 if (ecb_expect_false (!dec->shareable))
841 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ()); 931 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ());
842 932
933 if (dec->cbor.flags & F_ALLOW_CYCLES)
934 {
843 sv = newSV (0); 935 sv = newSV (0);
844 av_push (dec->shareable, SvREFCNT_inc_NN (sv)); 936 av_push (dec->shareable, SvREFCNT_inc_NN (sv));
845 937
846 SV *osv = decode_sv (dec); 938 SV *osv = decode_sv (dec);
847 sv_setsv (sv, osv); 939 sv_setsv (sv, osv);
848 SvREFCNT_dec_NN (osv); 940 SvREFCNT_dec_NN (osv);
941 }
942 else
943 {
944 av_push (dec->shareable, &PL_sv_undef);
945 int idx = AvFILLp (dec->shareable);
946 sv = decode_sv (dec);
947 av_store (dec->shareable, idx, SvREFCNT_inc_NN (sv));
948 }
849 } 949 }
850 break; 950 break;
851 951
852 case CBOR_TAG_VALUE_SHAREDREF: 952 case CBOR_TAG_VALUE_SHAREDREF:
853 { 953 {
854 if ((*dec->cur >> 5) != 0) 954 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
855 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)"); 955 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
856 956
857 UV idx = decode_uint (dec); 957 UV idx = decode_uint (dec);
858 958
859 if (!dec->shareable || (int)idx > AvFILLp (dec->shareable)) 959 if (!dec->shareable || (int)idx > AvFILLp (dec->shareable))
860 ERR ("corrupted CBOR data (sharedref index out of bounds)"); 960 ERR ("corrupted CBOR data (sharedref index out of bounds)");
861 961
862 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]); 962 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]);
963
964 if (sv == &PL_sv_undef)
965 ERR ("cyclic CBOR data structure found, but allow_cycles is not enabled");
863 } 966 }
864 break; 967 break;
865 968
866 case CBOR_TAG_PERL_OBJECT: 969 case CBOR_TAG_PERL_OBJECT:
867 { 970 {
… …
968static SV * 1071static SV *
969decode_sv (dec_t *dec) 1072decode_sv (dec_t *dec)
970{ 1073{
971 WANT (1); 1074 WANT (1);
972 1075
973 switch (*dec->cur >> 5) 1076 switch (*dec->cur >> MAJOR_SHIFT)
974 { 1077 {
975 case 0: // unsigned int 1078 case MAJOR_POS_INT >> MAJOR_SHIFT: return newSVuv (decode_uint (dec));
976 return newSVuv (decode_uint (dec)); 1079 case MAJOR_NEG_INT >> MAJOR_SHIFT: return newSViv (-1 - (IV)decode_uint (dec));
977 case 1: // negative int 1080 case MAJOR_BYTES >> MAJOR_SHIFT: return decode_str (dec, 0);
978 return newSViv (-1 - (IV)decode_uint (dec)); 1081 case MAJOR_TEXT >> MAJOR_SHIFT: return decode_str (dec, 1);
979 case 2: // octet string 1082 case MAJOR_ARRAY >> MAJOR_SHIFT: return decode_av (dec);
980 return decode_str (dec, 0); 1083 case MAJOR_MAP >> MAJOR_SHIFT: return decode_hv (dec);
981 case 3: // utf-8 string 1084 case MAJOR_TAG >> MAJOR_SHIFT: return decode_tagged (dec);
982 return decode_str (dec, 1); 1085
983 case 4: // array 1086 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) 1087 switch (*dec->cur++ & MINOR_MASK)
991 { 1088 {
992 case 20: 1089 case SIMPLE_FALSE:
993#if CBOR_SLOW 1090#if CBOR_SLOW
994 types_false = get_bool ("Types::Serialiser::false"); 1091 types_false = get_bool ("Types::Serialiser::false");
995#endif 1092#endif
996 return newSVsv (types_false); 1093 return newSVsv (types_false);
997 case 21: 1094 case SIMPLE_TRUE:
998#if CBOR_SLOW 1095#if CBOR_SLOW
999 types_true = get_bool ("Types::Serialiser::true"); 1096 types_true = get_bool ("Types::Serialiser::true");
1000#endif 1097#endif
1001 return newSVsv (types_true); 1098 return newSVsv (types_true);
1002 case 22: 1099 case SIMPLE_NULL:
1003 return newSVsv (&PL_sv_undef); 1100 return newSVsv (&PL_sv_undef);
1004 case 23: 1101 case SIMPLE_UNDEF:
1005#if CBOR_SLOW 1102#if CBOR_SLOW
1006 types_error = get_bool ("Types::Serialiser::error"); 1103 types_error = get_bool ("Types::Serialiser::error");
1007#endif 1104#endif
1008 return newSVsv (types_error); 1105 return newSVsv (types_error);
1009 1106
1010 case 25: 1107 case MISC_FLOAT16:
1011 { 1108 {
1012 WANT (2); 1109 WANT (2);
1013 1110
1014 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1]; 1111 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1];
1015 dec->cur += 2; 1112 dec->cur += 2;
1016 1113
1017 return newSVnv (ecb_binary16_to_float (fp)); 1114 return newSVnv (ecb_binary16_to_float (fp));
1018 } 1115 }
1019 1116
1020 case 26: 1117 case MISC_FLOAT32:
1021 { 1118 {
1022 uint32_t fp; 1119 uint32_t fp;
1023 WANT (4); 1120 WANT (4);
1024 memcpy (&fp, dec->cur, 4); 1121 memcpy (&fp, dec->cur, 4);
1025 dec->cur += 4; 1122 dec->cur += 4;
… …
1028 fp = ecb_bswap32 (fp); 1125 fp = ecb_bswap32 (fp);
1029 1126
1030 return newSVnv (ecb_binary32_to_float (fp)); 1127 return newSVnv (ecb_binary32_to_float (fp));
1031 } 1128 }
1032 1129
1033 case 27: 1130 case MISC_FLOAT64:
1034 { 1131 {
1035 uint64_t fp; 1132 uint64_t fp;
1036 WANT (8); 1133 WANT (8);
1037 memcpy (&fp, dec->cur, 8); 1134 memcpy (&fp, dec->cur, 8);
1038 dec->cur += 8; 1135 dec->cur += 8;
… …
1041 fp = ecb_bswap64 (fp); 1138 fp = ecb_bswap64 (fp);
1042 1139
1043 return newSVnv (ecb_binary64_to_double (fp)); 1140 return newSVnv (ecb_binary64_to_double (fp));
1044 } 1141 }
1045 1142
1046 // 0..19 unassigned 1143 // 0..19 unassigned simple
1047 // 24 reserved + unassigned (reserved values are not encodable) 1144 // 24 reserved + unassigned simple (reserved values are not encodable)
1145 // 28-30 unassigned misc
1146 // 31 break code
1048 default: 1147 default:
1049 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)"); 1148 ERR ("corrupted CBOR data (reserved/unassigned/unexpected major 7 value)");
1050 } 1149 }
1051 1150
1052 break; 1151 break;
1053 } 1152 }
1054 1153
… …
1081 if (dec.cur != dec.end && !dec.err) 1180 if (dec.cur != dec.end && !dec.err)
1082 dec.err = "garbage after CBOR object"; 1181 dec.err = "garbage after CBOR object";
1083 1182
1084 if (dec.err) 1183 if (dec.err)
1085 { 1184 {
1185 if (dec.shareable)
1186 {
1187 // need to break cyclic links, which whould all be in shareable
1188 int i;
1189 SV **svp;
1190
1191 for (i = av_len (dec.shareable) + 1; i--; )
1192 if ((svp = av_fetch (dec.shareable, i, 0)))
1193 sv_setsv (*svp, &PL_sv_undef);
1194 }
1195
1086 SvREFCNT_dec (sv); 1196 SvREFCNT_dec (sv);
1087 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur); 1197 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur);
1088 } 1198 }
1089 1199
1090 sv = sv_2mortal (sv); 1200 sv = sv_2mortal (sv);
1091 1201
1092 return sv; 1202 return sv;
1093} 1203}
1094 1204
1205/////////////////////////////////////////////////////////////////////////////
1206// incremental parser
1207
1208#define INCR_DONE(cbor) (AvFILLp (cbor->incr_count) < 0)
1209
1210// returns 0 for notyet, 1 for success or error
1211static int
1212incr_parse (CBOR *self, SV *cborstr)
1213{
1214 STRLEN cur;
1215 SvPV (cborstr, cur);
1216
1217 while (ecb_expect_true (self->incr_need <= cur))
1218 {
1219 // table of integer count bytes
1220 static I8 incr_len[MINOR_MASK + 1] = {
1221 0, 0, 0, 0, 0, 0, 0, 0,
1222 0, 0, 0, 0, 0, 0, 0, 0,
1223 0, 0, 0, 0, 0, 0, 0, 0,
1224 1, 2, 4, 8,-1,-1,-1,-2
1225 };
1226
1227 const U8 *p = SvPVX (cborstr) + self->incr_pos;
1228 U8 m = *p & MINOR_MASK;
1229 IV count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1230 I8 ilen = incr_len[m];
1231
1232 self->incr_need = self->incr_pos + 1;
1233
1234 if (ecb_expect_false (ilen < 0))
1235 {
1236 if (m != MINOR_INDEF)
1237 return 1; // error
1238
1239 if (*p == (MAJOR_MISC | MINOR_INDEF))
1240 {
1241 if (count >= 0)
1242 return 1; // error
1243
1244 count = 1;
1245 }
1246 else
1247 {
1248 av_push (self->incr_count, newSViv (-1)); //TODO: nest
1249 count = -1;
1250 }
1251 }
1252 else
1253 {
1254 self->incr_need += ilen;
1255 if (ecb_expect_false (self->incr_need > cur))
1256 return 0;
1257
1258 int major = *p >> MAJOR_SHIFT;
1259
1260 switch (major)
1261 {
1262 case MAJOR_BYTES >> MAJOR_SHIFT:
1263 case MAJOR_TEXT >> MAJOR_SHIFT:
1264 case MAJOR_ARRAY >> MAJOR_SHIFT:
1265 case MAJOR_MAP >> MAJOR_SHIFT:
1266 {
1267 UV len;
1268
1269 if (ecb_expect_false (ilen))
1270 {
1271 len = 0;
1272
1273 do {
1274 len = (len << 8) | *++p;
1275 } while (--ilen);
1276 }
1277 else
1278 len = m;
1279
1280 switch (major)
1281 {
1282 case MAJOR_BYTES >> MAJOR_SHIFT:
1283 case MAJOR_TEXT >> MAJOR_SHIFT:
1284 self->incr_need += len;
1285 if (ecb_expect_false (self->incr_need > cur))
1286 return 0;
1287
1288 break;
1289
1290 case MAJOR_MAP >> MAJOR_SHIFT:
1291 len <<= 1;
1292 case MAJOR_ARRAY >> MAJOR_SHIFT:
1293 if (len)
1294 {
1295 av_push (self->incr_count, newSViv (len + 1)); //TODO: nest
1296 count = len + 1;
1297 }
1298 break;
1299 }
1300 }
1301 }
1302 }
1303
1304 self->incr_pos = self->incr_need;
1305
1306 if (count > 0)
1307 {
1308 while (!--count)
1309 {
1310 if (!AvFILLp (self->incr_count))
1311 return 1; // done
1312
1313 SvREFCNT_dec_NN (av_pop (self->incr_count));
1314 count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1315 }
1316
1317 SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]) = count;
1318 }
1319 }
1320
1321 return 0;
1322}
1323
1324
1095///////////////////////////////////////////////////////////////////////////// 1325/////////////////////////////////////////////////////////////////////////////
1096// XS interface functions 1326// XS interface functions
1097 1327
1098MODULE = CBOR::XS PACKAGE = CBOR::XS 1328MODULE = CBOR::XS PACKAGE = CBOR::XS
1099 1329
… …
1139void shrink (CBOR *self, int enable = 1) 1369void shrink (CBOR *self, int enable = 1)
1140 ALIAS: 1370 ALIAS:
1141 shrink = F_SHRINK 1371 shrink = F_SHRINK
1142 allow_unknown = F_ALLOW_UNKNOWN 1372 allow_unknown = F_ALLOW_UNKNOWN
1143 allow_sharing = F_ALLOW_SHARING 1373 allow_sharing = F_ALLOW_SHARING
1144 allow_stringref = F_ALLOW_STRINGREF 1374 allow_cycles = F_ALLOW_CYCLES
1375 pack_strings = F_PACK_STRINGS
1376 validate_utf8 = F_VALIDATE_UTF8
1145 PPCODE: 1377 PPCODE:
1146{ 1378{
1147 if (enable) 1379 if (enable)
1148 self->flags |= ix; 1380 self->flags |= ix;
1149 else 1381 else
… …
1155void get_shrink (CBOR *self) 1387void get_shrink (CBOR *self)
1156 ALIAS: 1388 ALIAS:
1157 get_shrink = F_SHRINK 1389 get_shrink = F_SHRINK
1158 get_allow_unknown = F_ALLOW_UNKNOWN 1390 get_allow_unknown = F_ALLOW_UNKNOWN
1159 get_allow_sharing = F_ALLOW_SHARING 1391 get_allow_sharing = F_ALLOW_SHARING
1160 get_allow_stringref = F_ALLOW_STRINGREF 1392 get_allow_cycles = F_ALLOW_CYCLES
1393 get_pack_strings = F_PACK_STRINGS
1394 get_validate_utf8 = F_VALIDATE_UTF8
1161 PPCODE: 1395 PPCODE:
1162 XPUSHs (boolSV (self->flags & ix)); 1396 XPUSHs (boolSV (self->flags & ix));
1163 1397
1164void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) 1398void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
1165 PPCODE: 1399 PPCODE:
… …
1214 EXTEND (SP, 2); 1448 EXTEND (SP, 2);
1215 PUSHs (sv); 1449 PUSHs (sv);
1216 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); 1450 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
1217} 1451}
1218 1452
1453void incr_parse (CBOR *self, SV *cborstr)
1454 ALIAS:
1455 incr_parse_multiple = 1
1456 PPCODE:
1457{
1458 if (SvUTF8 (cborstr))
1459 sv_utf8_downgrade (cborstr, 0);
1460
1461 if (!self->incr_count)
1462 {
1463 self->incr_count = newAV ();
1464 self->incr_pos = 0;
1465 self->incr_need = 1;
1466
1467 av_push (self->incr_count, newSViv (1));
1468 }
1469
1470 do
1471 {
1472 if (!incr_parse (self, cborstr))
1473 {
1474 if (self->incr_need > self->max_size && self->max_size)
1475 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
1476 (unsigned long)self->incr_need, (unsigned long)self->max_size);
1477
1478 break;
1479 }
1480
1481 SV *sv;
1482 char *offset;
1483
1484 PUTBACK; sv = decode_cbor (cborstr, self, &offset); SPAGAIN;
1485 XPUSHs (sv);
1486
1487 sv_chop (cborstr, offset);
1488
1489 av_clear (self->incr_count);
1490 av_push (self->incr_count, newSViv (1));
1491
1492 self->incr_pos = 0;
1493 self->incr_need = self->incr_pos + 1;
1494 }
1495 while (ix);
1496}
1497
1498void incr_reset (CBOR *self)
1499 CODE:
1500{
1501 SvREFCNT_dec (self->incr_count);
1502 self->incr_count = 0;
1503}
1504
1219void DESTROY (CBOR *self) 1505void DESTROY (CBOR *self)
1220 PPCODE: 1506 PPCODE:
1221 cbor_free (self); 1507 cbor_free (self);
1222 1508
1223PROTOTYPES: ENABLE 1509PROTOTYPES: ENABLE
1224 1510
1225void encode_cbor (SV *scalar) 1511void encode_cbor (SV *scalar)
1512 ALIAS:
1513 encode_cbor = 0
1514 encode_cbor_sharing = F_ALLOW_SHARING
1226 PPCODE: 1515 PPCODE:
1227{ 1516{
1228 CBOR cbor; 1517 CBOR cbor;
1229 cbor_init (&cbor); 1518 cbor_init (&cbor);
1519 cbor.flags |= ix;
1230 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN; 1520 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;
1231 XPUSHs (scalar); 1521 XPUSHs (scalar);
1232} 1522}
1233 1523
1234void decode_cbor (SV *cborstr) 1524void decode_cbor (SV *cborstr)

Diff Legend

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