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 1.47 by root, Mon Apr 27 20:21:53 2015 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
329 if (SvMAGICAL (av))
277 for (i = 0; i <= len; ++i) 330 for (i = 0; i <= len; ++i)
278 { 331 {
279 SV **svp = av_fetch (av, i, 0); 332 SV **svp = av_fetch (av, i, 0);
280 encode_sv (enc, svp ? *svp : &PL_sv_undef); 333 encode_sv (enc, svp ? *svp : &PL_sv_undef);
281 } 334 }
335 else
336 for (i = 0; i <= len; ++i)
337 {
338 SV *sv = AvARRAY (av)[i];
339 encode_sv (enc, sv ? sv : &PL_sv_undef);
340 }
282 341
283 --enc->depth; 342 --enc->depth;
284} 343}
285 344
286static void 345static void
… …
295 354
296 int pairs = hv_iterinit (hv); 355 int pairs = hv_iterinit (hv);
297 int mg = SvMAGICAL (hv); 356 int mg = SvMAGICAL (hv);
298 357
299 if (mg) 358 if (mg)
300 encode_ch (enc, 0xa0 | 31); 359 encode_ch (enc, MAJOR_MAP | MINOR_INDEF);
301 else 360 else
302 encode_uint (enc, 0xa0, pairs); 361 encode_uint (enc, MAJOR_MAP, pairs);
303 362
304 while ((he = hv_iternext (hv))) 363 while ((he = hv_iternext (hv)))
305 { 364 {
306 if (HeKLEN (he) == HEf_SVKEY) 365 if (HeKLEN (he) == HEf_SVKEY)
307 encode_sv (enc, HeSVKEY (he)); 366 encode_sv (enc, HeSVKEY (he));
308 else 367 else
309 encode_str (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he)); 368 encode_strref (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he));
310 369
311 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he)); 370 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he));
312 } 371 }
313 372
314 if (mg) 373 if (mg)
315 encode_ch (enc, 0xe0 | 31); 374 encode_ch (enc, MAJOR_MISC | MINOR_INDEF);
316 375
317 --enc->depth; 376 --enc->depth;
318} 377}
319 378
320// encode objects, arrays and special \0=false and \1=true values. 379// encode objects, arrays and special \0=false and \1=true values.
321static void 380static void
322encode_rv (enc_t *enc, SV *sv) 381encode_rv (enc_t *enc, SV *sv)
323{ 382{
324 SvGETMAGIC (sv); 383 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 }
347 384
348 svtype svt = SvTYPE (sv); 385 svtype svt = SvTYPE (sv);
349 386
350 if (ecb_expect_false (SvOBJECT (sv))) 387 if (ecb_expect_false (SvOBJECT (sv)))
351 { 388 {
… …
358 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 395 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
359 ? cbor_tagged_stash 396 ? cbor_tagged_stash
360 : gv_stashpv ("CBOR::XS::Tagged" , 1); 397 : gv_stashpv ("CBOR::XS::Tagged" , 1);
361 398
362 HV *stash = SvSTASH (sv); 399 HV *stash = SvSTASH (sv);
363 GV *method;
364 400
365 if (stash == boolean_stash) 401 if (stash == boolean_stash)
366 encode_ch (enc, SvIV (sv) ? 0xe0 | 21 : 0xe0 | 20); 402 {
403 encode_ch (enc, SvIV (sv) ? MAJOR_MISC | SIMPLE_TRUE : MAJOR_MISC | SIMPLE_FALSE);
404 return;
405 }
367 else if (stash == error_stash) 406 else if (stash == error_stash)
368 encode_ch (enc, 0xe0 | 23); 407 {
408 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
409 return;
410 }
369 else if (stash == tagged_stash) 411 else if (stash == tagged_stash)
370 { 412 {
371 if (svt != SVt_PVAV) 413 if (svt != SVt_PVAV)
372 croak ("encountered CBOR::XS::Tagged object that isn't an array"); 414 croak ("encountered CBOR::XS::Tagged object that isn't an array");
373 415
374 encode_uint (enc, 0xc0, SvUV (*av_fetch ((AV *)sv, 0, 1))); 416 encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1)));
375 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1)); 417 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1));
418
419 return;
420 }
421 }
422
423 if (ecb_expect_false (SvREFCNT (sv) > 1)
424 && ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING))
425 {
426 if (!enc->shareable)
427 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
428
429 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
430
431 if (SvOK (*svp))
376 } 432 {
433 encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF);
434 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
435 return;
436 }
437 else
438 {
439 sv_setuv (*svp, enc->shareable_idx);
440 ++enc->shareable_idx;
441 encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE);
442 }
443 }
444
445 if (ecb_expect_false (SvOBJECT (sv)))
446 {
447 HV *stash = SvSTASH (sv);
448 GV *method;
449
377 else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0))) 450 if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
378 { 451 {
379 dSP; 452 dSP;
380 453
381 ENTER; SAVETMPS; PUSHMARK (SP); 454 ENTER; SAVETMPS; PUSHMARK (SP);
382 // we re-bless the reference to get overload and other niceties right 455 // we re-bless the reference to get overload and other niceties right
… …
414 // catch this surprisingly common error 487 // catch this surprisingly common error
415 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv) 488 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)); 489 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));
417 490
418 encode_tag (enc, CBOR_TAG_PERL_OBJECT); 491 encode_tag (enc, CBOR_TAG_PERL_OBJECT);
419 encode_uint (enc, 0x80, count + 1); 492 encode_uint (enc, MAJOR_ARRAY, count + 1);
420 encode_str (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash)); 493 encode_strref (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
421 494
422 while (count) 495 while (count)
423 encode_sv (enc, SP[1 - count--]); 496 encode_sv (enc, SP[1 - count--]);
424 497
425 PUTBACK; 498 PUTBACK;
… …
446{ 519{
447 double nv = SvNVX (sv); 520 double nv = SvNVX (sv);
448 521
449 need (enc, 9); 522 need (enc, 9);
450 523
451 if (ecb_expect_false (nv == (U32)nv)) 524 if (ecb_expect_false (nv == (NV)(U32)nv))
452 encode_uint (enc, 0x00, (U32)nv); 525 encode_uint (enc, MAJOR_POS_INT, (U32)nv);
453 //TODO: maybe I32? 526 //TODO: maybe I32?
454 else if (ecb_expect_false (nv == (float)nv)) 527 else if (ecb_expect_false (nv == (float)nv))
455 { 528 {
456 uint32_t fp = ecb_float_to_binary32 (nv); 529 uint32_t fp = ecb_float_to_binary32 (nv);
457 530
458 *enc->cur++ = 0xe0 | 26; 531 *enc->cur++ = MAJOR_MISC | MISC_FLOAT32;
459 532
460 if (!ecb_big_endian ()) 533 if (!ecb_big_endian ())
461 fp = ecb_bswap32 (fp); 534 fp = ecb_bswap32 (fp);
462 535
463 memcpy (enc->cur, &fp, 4); 536 memcpy (enc->cur, &fp, 4);
… …
465 } 538 }
466 else 539 else
467 { 540 {
468 uint64_t fp = ecb_double_to_binary64 (nv); 541 uint64_t fp = ecb_double_to_binary64 (nv);
469 542
470 *enc->cur++ = 0xe0 | 27; 543 *enc->cur++ = MAJOR_MISC | MISC_FLOAT64;
471 544
472 if (!ecb_big_endian ()) 545 if (!ecb_big_endian ())
473 fp = ecb_bswap64 (fp); 546 fp = ecb_bswap64 (fp);
474 547
475 memcpy (enc->cur, &fp, 8); 548 memcpy (enc->cur, &fp, 8);
… …
484 557
485 if (SvPOKp (sv)) 558 if (SvPOKp (sv))
486 { 559 {
487 STRLEN len; 560 STRLEN len;
488 char *str = SvPV (sv, len); 561 char *str = SvPV (sv, len);
489 encode_str (enc, SvUTF8 (sv), str, len); 562 encode_strref (enc, SvUTF8 (sv), str, len);
490 } 563 }
491 else if (SvNOKp (sv)) 564 else if (SvNOKp (sv))
492 encode_nv (enc, sv); 565 encode_nv (enc, sv);
493 else if (SvIOKp (sv)) 566 else if (SvIOKp (sv))
494 { 567 {
495 if (SvIsUV (sv)) 568 if (SvIsUV (sv))
496 encode_uint (enc, 0x00, SvUVX (sv)); 569 encode_uint (enc, MAJOR_POS_INT, SvUVX (sv));
497 else if (SvIVX (sv) >= 0) 570 else if (SvIVX (sv) >= 0)
498 encode_uint (enc, 0x00, SvIVX (sv)); 571 encode_uint (enc, MAJOR_POS_INT, SvIVX (sv));
499 else 572 else
500 encode_uint (enc, 0x20, -(SvIVX (sv) + 1)); 573 encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1));
501 } 574 }
502 else if (SvROK (sv)) 575 else if (SvROK (sv))
503 encode_rv (enc, SvRV (sv)); 576 encode_rv (enc, SvRV (sv));
504 else if (!SvOK (sv)) 577 else if (!SvOK (sv))
505 encode_ch (enc, 0xe0 | 22); 578 encode_ch (enc, MAJOR_MISC | SIMPLE_NULL);
506 else if (enc->cbor.flags & F_ALLOW_UNKNOWN) 579 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
507 encode_ch (enc, 0xe0 | 23); 580 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
508 else 581 else
509 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data", 582 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data",
510 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv)); 583 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
511} 584}
512 585
… …
520 enc.cur = SvPVX (enc.sv); 593 enc.cur = SvPVX (enc.sv);
521 enc.end = SvEND (enc.sv); 594 enc.end = SvEND (enc.sv);
522 595
523 SvPOK_only (enc.sv); 596 SvPOK_only (enc.sv);
524 597
525 if (cbor->flags & F_ALLOW_STRINGREF) 598 if (cbor->flags & F_PACK_STRINGS)
526 { 599 {
527 encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE); 600 encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE);
528 enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ()); 601 enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ());
529 enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ()); 602 enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ());
530 } 603 }
… …
565#define DEC_DEC_DEPTH --dec->depth 638#define DEC_DEC_DEPTH --dec->depth
566 639
567static UV 640static UV
568decode_uint (dec_t *dec) 641decode_uint (dec_t *dec)
569{ 642{
570 switch (*dec->cur & 31) 643 U8 m = *dec->cur & MINOR_MASK;
571 { 644 ++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;
576 645
577 case 24: 646 if (ecb_expect_true (m < LENGTH_EXT1))
647 return m;
648 else if (ecb_expect_true (m == LENGTH_EXT1))
649 {
578 WANT (2); 650 WANT (1);
579 dec->cur += 2; 651 dec->cur += 1;
580 return dec->cur[-1]; 652 return dec->cur[-1];
581 653 }
582 case 25: 654 else if (ecb_expect_true (m == LENGTH_EXT2))
655 {
583 WANT (3); 656 WANT (2);
584 dec->cur += 3; 657 dec->cur += 2;
585 return (((UV)dec->cur[-2]) << 8) 658 return (((UV)dec->cur[-2]) << 8)
586 | ((UV)dec->cur[-1]); 659 | ((UV)dec->cur[-1]);
587 660 }
588 case 26: 661 else if (ecb_expect_true (m == LENGTH_EXT4))
662 {
589 WANT (5); 663 WANT (4);
590 dec->cur += 5; 664 dec->cur += 4;
591 return (((UV)dec->cur[-4]) << 24) 665 return (((UV)dec->cur[-4]) << 24)
592 | (((UV)dec->cur[-3]) << 16) 666 | (((UV)dec->cur[-3]) << 16)
593 | (((UV)dec->cur[-2]) << 8) 667 | (((UV)dec->cur[-2]) << 8)
594 | ((UV)dec->cur[-1]); 668 | ((UV)dec->cur[-1]);
595 669 }
596 case 27: 670 else if (ecb_expect_true (m == LENGTH_EXT8))
671 {
597 WANT (9); 672 WANT (8);
598 dec->cur += 9; 673 dec->cur += 8;
674
675 return
676#if UVSIZE < 8
677 0
678#else
599 return (((UV)dec->cur[-8]) << 56) 679 (((UV)dec->cur[-8]) << 56)
600 | (((UV)dec->cur[-7]) << 48) 680 | (((UV)dec->cur[-7]) << 48)
601 | (((UV)dec->cur[-6]) << 40) 681 | (((UV)dec->cur[-6]) << 40)
602 | (((UV)dec->cur[-5]) << 32) 682 | (((UV)dec->cur[-5]) << 32)
683#endif
603 | (((UV)dec->cur[-4]) << 24) 684 | (((UV)dec->cur[-4]) << 24)
604 | (((UV)dec->cur[-3]) << 16) 685 | (((UV)dec->cur[-3]) << 16)
605 | (((UV)dec->cur[-2]) << 8) 686 | (((UV)dec->cur[-2]) << 8)
606 | ((UV)dec->cur[-1]); 687 | ((UV)dec->cur[-1]);
607 688 }
608 default: 689 else
609 ERR ("corrupted CBOR data (unsupported integer minor encoding)"); 690 ERR ("corrupted CBOR data (unsupported integer minor encoding)");
610 }
611 691
612fail: 692fail:
613 return 0; 693 return 0;
614} 694}
615 695
… …
620{ 700{
621 AV *av = newAV (); 701 AV *av = newAV ();
622 702
623 DEC_INC_DEPTH; 703 DEC_INC_DEPTH;
624 704
625 if ((*dec->cur & 31) == 31) 705 if (*dec->cur == (MAJOR_ARRAY | MINOR_INDEF))
626 { 706 {
627 ++dec->cur; 707 ++dec->cur;
628 708
629 for (;;) 709 for (;;)
630 { 710 {
631 WANT (1); 711 WANT (1);
632 712
633 if (*dec->cur == (0xe0 | 31)) 713 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
634 { 714 {
635 ++dec->cur; 715 ++dec->cur;
636 break; 716 break;
637 } 717 }
638 718
… …
641 } 721 }
642 else 722 else
643 { 723 {
644 int i, len = decode_uint (dec); 724 int i, len = decode_uint (dec);
645 725
726 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); 727 av_fill (av, len - 1);
647 728
648 for (i = 0; i < len; ++i) 729 for (i = 0; i < len; ++i)
649 AvARRAY (av)[i] = decode_sv (dec); 730 AvARRAY (av)[i] = decode_sv (dec);
650 } 731 }
… …
663{ 744{
664 // for speed reasons, we specialcase single-string 745 // for speed reasons, we specialcase single-string
665 // byte or utf-8 strings as keys, but only when !stringref 746 // byte or utf-8 strings as keys, but only when !stringref
666 747
667 if (ecb_expect_true (!dec->stringref)) 748 if (ecb_expect_true (!dec->stringref))
668 if (*dec->cur >= 0x40 && *dec->cur <= 0x40 + 27) 749 if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8))
669 { 750 {
670 I32 len = decode_uint (dec); 751 I32 len = decode_uint (dec);
671 char *key = (char *)dec->cur; 752 char *key = (char *)dec->cur;
672 753
673 dec->cur += len; 754 dec->cur += len;
674 755
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); 756 hv_store (hv, key, len, decode_sv (dec), 0);
679 757
680 return; 758 return;
681 } 759 }
682 else if (*dec->cur >= 0x60 && *dec->cur <= 0x60 + 27) 760 else if (ecb_expect_true ((U8)(*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8))
683 { 761 {
684 I32 len = decode_uint (dec); 762 I32 len = decode_uint (dec);
685 char *key = (char *)dec->cur; 763 char *key = (char *)dec->cur;
686 764
687 dec->cur += len; 765 dec->cur += len;
688 766
689 if (ecb_expect_false (dec->stringref)) 767 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
690 av_push (dec->stringref, newSVpvn_utf8 (key, len, 1)); 768 if (!is_utf8_string (key, len))
769 ERR ("corrupted CBOR data (invalid UTF-8 in map key)");
691 770
692 hv_store (hv, key, -len, decode_sv (dec), 0); 771 hv_store (hv, key, -len, decode_sv (dec), 0);
693 772
694 return; 773 return;
695 } 774 }
… …
697 SV *k = decode_sv (dec); 776 SV *k = decode_sv (dec);
698 SV *v = decode_sv (dec); 777 SV *v = decode_sv (dec);
699 778
700 hv_store_ent (hv, k, v, 0); 779 hv_store_ent (hv, k, v, 0);
701 SvREFCNT_dec (k); 780 SvREFCNT_dec (k);
781
782fail:
783 ;
702} 784}
703 785
704static SV * 786static SV *
705decode_hv (dec_t *dec) 787decode_hv (dec_t *dec)
706{ 788{
707 HV *hv = newHV (); 789 HV *hv = newHV ();
708 790
709 DEC_INC_DEPTH; 791 DEC_INC_DEPTH;
710 792
711 if ((*dec->cur & 31) == 31) 793 if (*dec->cur == (MAJOR_MAP | MINOR_INDEF))
712 { 794 {
713 ++dec->cur; 795 ++dec->cur;
714 796
715 for (;;) 797 for (;;)
716 { 798 {
717 WANT (1); 799 WANT (1);
718 800
719 if (*dec->cur == (0xe0 | 31)) 801 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
720 { 802 {
721 ++dec->cur; 803 ++dec->cur;
722 break; 804 break;
723 } 805 }
724 806
… …
745static SV * 827static SV *
746decode_str (dec_t *dec, int utf8) 828decode_str (dec_t *dec, int utf8)
747{ 829{
748 SV *sv = 0; 830 SV *sv = 0;
749 831
750 if ((*dec->cur & 31) == 31) 832 if ((*dec->cur & MINOR_MASK) == MINOR_INDEF)
751 { 833 {
834 // indefinite length strings
752 ++dec->cur; 835 ++dec->cur;
753 836
837 U8 major = *dec->cur & MAJOR_MISC;
838
754 sv = newSVpvn ("", 0); 839 sv = newSVpvn ("", 0);
755 840
756 // not very fast, and certainly not robust against illegal input
757 for (;;) 841 for (;;)
758 { 842 {
759 WANT (1); 843 WANT (1);
760 844
761 if (*dec->cur == (0xe0 | 31)) 845 if ((*dec->cur - major) > LENGTH_EXT8)
846 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
762 { 847 {
763 ++dec->cur; 848 ++dec->cur;
764 break; 849 break;
765 } 850 }
851 else
852 ERR ("corrupted CBOR data (invalid chunks in indefinite length string)");
766 853
767 sv_catsv (sv, decode_sv (dec)); 854 STRLEN len = decode_uint (dec);
855
856 WANT (len);
857 sv_catpvn (sv, dec->cur, len);
858 dec->cur += len;
768 } 859 }
769 } 860 }
770 else 861 else
771 { 862 {
772 STRLEN len = decode_uint (dec); 863 STRLEN len = decode_uint (dec);
… …
779 && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1)) 870 && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1))
780 av_push (dec->stringref, SvREFCNT_inc_NN (sv)); 871 av_push (dec->stringref, SvREFCNT_inc_NN (sv));
781 } 872 }
782 873
783 if (utf8) 874 if (utf8)
875 {
876 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
877 if (!is_utf8_string (SvPVX (sv), SvCUR (sv)))
878 ERR ("corrupted CBOR data (invalid UTF-8 in text string)");
879
784 SvUTF8_on (sv); 880 SvUTF8_on (sv);
881 }
785 882
786 return sv; 883 return sv;
787 884
788fail: 885fail:
789 SvREFCNT_dec (sv); 886 SvREFCNT_dec (sv);
… …
821 } 918 }
822 break; 919 break;
823 920
824 case CBOR_TAG_STRINGREF: 921 case CBOR_TAG_STRINGREF:
825 { 922 {
826 if ((*dec->cur >> 5) != 0) 923 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
827 ERR ("corrupted CBOR data (stringref index not an unsigned integer)"); 924 ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
828 925
829 UV idx = decode_uint (dec); 926 UV idx = decode_uint (dec);
830 927
831 if (!dec->stringref || (int)idx > AvFILLp (dec->stringref)) 928 if (!dec->stringref || (int)idx > AvFILLp (dec->stringref))
… …
838 case CBOR_TAG_VALUE_SHAREABLE: 935 case CBOR_TAG_VALUE_SHAREABLE:
839 { 936 {
840 if (ecb_expect_false (!dec->shareable)) 937 if (ecb_expect_false (!dec->shareable))
841 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ()); 938 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ());
842 939
940 if (dec->cbor.flags & F_ALLOW_CYCLES)
941 {
843 sv = newSV (0); 942 sv = newSV (0);
844 av_push (dec->shareable, SvREFCNT_inc_NN (sv)); 943 av_push (dec->shareable, SvREFCNT_inc_NN (sv));
845 944
846 SV *osv = decode_sv (dec); 945 SV *osv = decode_sv (dec);
847 sv_setsv (sv, osv); 946 sv_setsv (sv, osv);
848 SvREFCNT_dec_NN (osv); 947 SvREFCNT_dec_NN (osv);
948 }
949 else
950 {
951 av_push (dec->shareable, &PL_sv_undef);
952 int idx = AvFILLp (dec->shareable);
953 sv = decode_sv (dec);
954 av_store (dec->shareable, idx, SvREFCNT_inc_NN (sv));
955 }
849 } 956 }
850 break; 957 break;
851 958
852 case CBOR_TAG_VALUE_SHAREDREF: 959 case CBOR_TAG_VALUE_SHAREDREF:
853 { 960 {
854 if ((*dec->cur >> 5) != 0) 961 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
855 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)"); 962 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
856 963
857 UV idx = decode_uint (dec); 964 UV idx = decode_uint (dec);
858 965
859 if (!dec->shareable || (int)idx > AvFILLp (dec->shareable)) 966 if (!dec->shareable || (int)idx > AvFILLp (dec->shareable))
860 ERR ("corrupted CBOR data (sharedref index out of bounds)"); 967 ERR ("corrupted CBOR data (sharedref index out of bounds)");
861 968
862 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]); 969 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]);
970
971 if (sv == &PL_sv_undef)
972 ERR ("cyclic CBOR data structure found, but allow_cycles is not enabled");
863 } 973 }
864 break; 974 break;
865 975
866 case CBOR_TAG_PERL_OBJECT: 976 case CBOR_TAG_PERL_OBJECT:
867 { 977 {
… …
968static SV * 1078static SV *
969decode_sv (dec_t *dec) 1079decode_sv (dec_t *dec)
970{ 1080{
971 WANT (1); 1081 WANT (1);
972 1082
973 switch (*dec->cur >> 5) 1083 switch (*dec->cur >> MAJOR_SHIFT)
974 { 1084 {
975 case 0: // unsigned int 1085 case MAJOR_POS_INT >> MAJOR_SHIFT: return newSVuv (decode_uint (dec));
976 return newSVuv (decode_uint (dec)); 1086 case MAJOR_NEG_INT >> MAJOR_SHIFT: return newSViv (-1 - (IV)decode_uint (dec));
977 case 1: // negative int 1087 case MAJOR_BYTES >> MAJOR_SHIFT: return decode_str (dec, 0);
978 return newSViv (-1 - (IV)decode_uint (dec)); 1088 case MAJOR_TEXT >> MAJOR_SHIFT: return decode_str (dec, 1);
979 case 2: // octet string 1089 case MAJOR_ARRAY >> MAJOR_SHIFT: return decode_av (dec);
980 return decode_str (dec, 0); 1090 case MAJOR_MAP >> MAJOR_SHIFT: return decode_hv (dec);
981 case 3: // utf-8 string 1091 case MAJOR_TAG >> MAJOR_SHIFT: return decode_tagged (dec);
982 return decode_str (dec, 1); 1092
983 case 4: // array 1093 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) 1094 switch (*dec->cur++ & MINOR_MASK)
991 { 1095 {
992 case 20: 1096 case SIMPLE_FALSE:
993#if CBOR_SLOW 1097#if CBOR_SLOW
994 types_false = get_bool ("Types::Serialiser::false"); 1098 types_false = get_bool ("Types::Serialiser::false");
995#endif 1099#endif
996 return newSVsv (types_false); 1100 return newSVsv (types_false);
997 case 21: 1101 case SIMPLE_TRUE:
998#if CBOR_SLOW 1102#if CBOR_SLOW
999 types_true = get_bool ("Types::Serialiser::true"); 1103 types_true = get_bool ("Types::Serialiser::true");
1000#endif 1104#endif
1001 return newSVsv (types_true); 1105 return newSVsv (types_true);
1002 case 22: 1106 case SIMPLE_NULL:
1003 return newSVsv (&PL_sv_undef); 1107 return newSVsv (&PL_sv_undef);
1004 case 23: 1108 case SIMPLE_UNDEF:
1005#if CBOR_SLOW 1109#if CBOR_SLOW
1006 types_error = get_bool ("Types::Serialiser::error"); 1110 types_error = get_bool ("Types::Serialiser::error");
1007#endif 1111#endif
1008 return newSVsv (types_error); 1112 return newSVsv (types_error);
1009 1113
1010 case 25: 1114 case MISC_FLOAT16:
1011 { 1115 {
1012 WANT (2); 1116 WANT (2);
1013 1117
1014 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1]; 1118 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1];
1015 dec->cur += 2; 1119 dec->cur += 2;
1016 1120
1017 return newSVnv (ecb_binary16_to_float (fp)); 1121 return newSVnv (ecb_binary16_to_float (fp));
1018 } 1122 }
1019 1123
1020 case 26: 1124 case MISC_FLOAT32:
1021 { 1125 {
1022 uint32_t fp; 1126 uint32_t fp;
1023 WANT (4); 1127 WANT (4);
1024 memcpy (&fp, dec->cur, 4); 1128 memcpy (&fp, dec->cur, 4);
1025 dec->cur += 4; 1129 dec->cur += 4;
… …
1028 fp = ecb_bswap32 (fp); 1132 fp = ecb_bswap32 (fp);
1029 1133
1030 return newSVnv (ecb_binary32_to_float (fp)); 1134 return newSVnv (ecb_binary32_to_float (fp));
1031 } 1135 }
1032 1136
1033 case 27: 1137 case MISC_FLOAT64:
1034 { 1138 {
1035 uint64_t fp; 1139 uint64_t fp;
1036 WANT (8); 1140 WANT (8);
1037 memcpy (&fp, dec->cur, 8); 1141 memcpy (&fp, dec->cur, 8);
1038 dec->cur += 8; 1142 dec->cur += 8;
… …
1041 fp = ecb_bswap64 (fp); 1145 fp = ecb_bswap64 (fp);
1042 1146
1043 return newSVnv (ecb_binary64_to_double (fp)); 1147 return newSVnv (ecb_binary64_to_double (fp));
1044 } 1148 }
1045 1149
1046 // 0..19 unassigned 1150 // 0..19 unassigned simple
1047 // 24 reserved + unassigned (reserved values are not encodable) 1151 // 24 reserved + unassigned simple (reserved values are not encodable)
1152 // 28-30 unassigned misc
1153 // 31 break code
1048 default: 1154 default:
1049 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)"); 1155 ERR ("corrupted CBOR data (reserved/unassigned/unexpected major 7 value)");
1050 } 1156 }
1051 1157
1052 break; 1158 break;
1053 } 1159 }
1054 1160
… …
1081 if (dec.cur != dec.end && !dec.err) 1187 if (dec.cur != dec.end && !dec.err)
1082 dec.err = "garbage after CBOR object"; 1188 dec.err = "garbage after CBOR object";
1083 1189
1084 if (dec.err) 1190 if (dec.err)
1085 { 1191 {
1192 if (dec.shareable)
1193 {
1194 // need to break cyclic links, which whould all be in shareable
1195 int i;
1196 SV **svp;
1197
1198 for (i = av_len (dec.shareable) + 1; i--; )
1199 if ((svp = av_fetch (dec.shareable, i, 0)))
1200 sv_setsv (*svp, &PL_sv_undef);
1201 }
1202
1086 SvREFCNT_dec (sv); 1203 SvREFCNT_dec (sv);
1087 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur); 1204 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur);
1088 } 1205 }
1089 1206
1090 sv = sv_2mortal (sv); 1207 sv = sv_2mortal (sv);
1091 1208
1092 return sv; 1209 return sv;
1093} 1210}
1094 1211
1212/////////////////////////////////////////////////////////////////////////////
1213// incremental parser
1214
1215#define INCR_DONE(cbor) (AvFILLp (cbor->incr_count) < 0)
1216
1217// returns 0 for notyet, 1 for success or error
1218static int
1219incr_parse (CBOR *self, SV *cborstr)
1220{
1221 STRLEN cur;
1222 SvPV (cborstr, cur);
1223
1224 while (ecb_expect_true (self->incr_need <= cur))
1225 {
1226 // table of integer count bytes
1227 static I8 incr_len[MINOR_MASK + 1] = {
1228 0, 0, 0, 0, 0, 0, 0, 0,
1229 0, 0, 0, 0, 0, 0, 0, 0,
1230 0, 0, 0, 0, 0, 0, 0, 0,
1231 1, 2, 4, 8,-1,-1,-1,-2
1232 };
1233
1234 const U8 *p = SvPVX (cborstr) + self->incr_pos;
1235 U8 m = *p & MINOR_MASK;
1236 IV count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1237 I8 ilen = incr_len[m];
1238
1239 self->incr_need = self->incr_pos + 1;
1240
1241 if (ecb_expect_false (ilen < 0))
1242 {
1243 if (m != MINOR_INDEF)
1244 return 1; // error
1245
1246 if (*p == (MAJOR_MISC | MINOR_INDEF))
1247 {
1248 if (count >= 0)
1249 return 1; // error
1250
1251 count = 1;
1252 }
1253 else
1254 {
1255 av_push (self->incr_count, newSViv (-1)); //TODO: nest
1256 count = -1;
1257 }
1258 }
1259 else
1260 {
1261 self->incr_need += ilen;
1262 if (ecb_expect_false (self->incr_need > cur))
1263 return 0;
1264
1265 int major = *p >> MAJOR_SHIFT;
1266
1267 switch (major)
1268 {
1269 case MAJOR_TAG >> MAJOR_SHIFT:
1270 ++count; // tags merely prefix another value
1271 break;
1272
1273 case MAJOR_BYTES >> MAJOR_SHIFT:
1274 case MAJOR_TEXT >> MAJOR_SHIFT:
1275 case MAJOR_ARRAY >> MAJOR_SHIFT:
1276 case MAJOR_MAP >> MAJOR_SHIFT:
1277 {
1278 UV len;
1279
1280 if (ecb_expect_false (ilen))
1281 {
1282 len = 0;
1283
1284 do {
1285 len = (len << 8) | *++p;
1286 } while (--ilen);
1287 }
1288 else
1289 len = m;
1290
1291 switch (major)
1292 {
1293 case MAJOR_BYTES >> MAJOR_SHIFT:
1294 case MAJOR_TEXT >> MAJOR_SHIFT:
1295 self->incr_need += len;
1296 if (ecb_expect_false (self->incr_need > cur))
1297 return 0;
1298
1299 break;
1300
1301 case MAJOR_MAP >> MAJOR_SHIFT:
1302 len <<= 1;
1303 case MAJOR_ARRAY >> MAJOR_SHIFT:
1304 if (len)
1305 {
1306 av_push (self->incr_count, newSViv (len + 1)); //TODO: nest
1307 count = len + 1;
1308 }
1309 break;
1310 }
1311 }
1312 }
1313 }
1314
1315 self->incr_pos = self->incr_need;
1316
1317 if (count > 0)
1318 {
1319 while (!--count)
1320 {
1321 if (!AvFILLp (self->incr_count))
1322 return 1; // done
1323
1324 SvREFCNT_dec_NN (av_pop (self->incr_count));
1325 count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1326 }
1327
1328 SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]) = count;
1329 }
1330 }
1331
1332 return 0;
1333}
1334
1335
1095///////////////////////////////////////////////////////////////////////////// 1336/////////////////////////////////////////////////////////////////////////////
1096// XS interface functions 1337// XS interface functions
1097 1338
1098MODULE = CBOR::XS PACKAGE = CBOR::XS 1339MODULE = CBOR::XS PACKAGE = CBOR::XS
1099 1340
… …
1139void shrink (CBOR *self, int enable = 1) 1380void shrink (CBOR *self, int enable = 1)
1140 ALIAS: 1381 ALIAS:
1141 shrink = F_SHRINK 1382 shrink = F_SHRINK
1142 allow_unknown = F_ALLOW_UNKNOWN 1383 allow_unknown = F_ALLOW_UNKNOWN
1143 allow_sharing = F_ALLOW_SHARING 1384 allow_sharing = F_ALLOW_SHARING
1144 allow_stringref = F_ALLOW_STRINGREF 1385 allow_cycles = F_ALLOW_CYCLES
1386 pack_strings = F_PACK_STRINGS
1387 validate_utf8 = F_VALIDATE_UTF8
1145 PPCODE: 1388 PPCODE:
1146{ 1389{
1147 if (enable) 1390 if (enable)
1148 self->flags |= ix; 1391 self->flags |= ix;
1149 else 1392 else
… …
1155void get_shrink (CBOR *self) 1398void get_shrink (CBOR *self)
1156 ALIAS: 1399 ALIAS:
1157 get_shrink = F_SHRINK 1400 get_shrink = F_SHRINK
1158 get_allow_unknown = F_ALLOW_UNKNOWN 1401 get_allow_unknown = F_ALLOW_UNKNOWN
1159 get_allow_sharing = F_ALLOW_SHARING 1402 get_allow_sharing = F_ALLOW_SHARING
1160 get_allow_stringref = F_ALLOW_STRINGREF 1403 get_allow_cycles = F_ALLOW_CYCLES
1404 get_pack_strings = F_PACK_STRINGS
1405 get_validate_utf8 = F_VALIDATE_UTF8
1161 PPCODE: 1406 PPCODE:
1162 XPUSHs (boolSV (self->flags & ix)); 1407 XPUSHs (boolSV (self->flags & ix));
1163 1408
1164void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) 1409void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
1165 PPCODE: 1410 PPCODE:
… …
1214 EXTEND (SP, 2); 1459 EXTEND (SP, 2);
1215 PUSHs (sv); 1460 PUSHs (sv);
1216 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); 1461 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
1217} 1462}
1218 1463
1464void incr_parse (CBOR *self, SV *cborstr)
1465 ALIAS:
1466 incr_parse_multiple = 1
1467 PPCODE:
1468{
1469 if (SvUTF8 (cborstr))
1470 sv_utf8_downgrade (cborstr, 0);
1471
1472 if (!self->incr_count)
1473 {
1474 self->incr_count = newAV ();
1475 self->incr_pos = 0;
1476 self->incr_need = 1;
1477
1478 av_push (self->incr_count, newSViv (1));
1479 }
1480
1481 do
1482 {
1483 if (!incr_parse (self, cborstr))
1484 {
1485 if (self->incr_need > self->max_size && self->max_size)
1486 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
1487 (unsigned long)self->incr_need, (unsigned long)self->max_size);
1488
1489 break;
1490 }
1491
1492 SV *sv;
1493 char *offset;
1494
1495 PUTBACK; sv = decode_cbor (cborstr, self, &offset); SPAGAIN;
1496 XPUSHs (sv);
1497
1498 sv_chop (cborstr, offset);
1499
1500 av_clear (self->incr_count);
1501 av_push (self->incr_count, newSViv (1));
1502
1503 self->incr_pos = 0;
1504 self->incr_need = self->incr_pos + 1;
1505 }
1506 while (ix);
1507}
1508
1509void incr_reset (CBOR *self)
1510 CODE:
1511{
1512 SvREFCNT_dec (self->incr_count);
1513 self->incr_count = 0;
1514}
1515
1219void DESTROY (CBOR *self) 1516void DESTROY (CBOR *self)
1220 PPCODE: 1517 PPCODE:
1221 cbor_free (self); 1518 cbor_free (self);
1222 1519
1223PROTOTYPES: ENABLE 1520PROTOTYPES: ENABLE
1224 1521
1225void encode_cbor (SV *scalar) 1522void encode_cbor (SV *scalar)
1523 ALIAS:
1524 encode_cbor = 0
1525 encode_cbor_sharing = F_ALLOW_SHARING
1226 PPCODE: 1526 PPCODE:
1227{ 1527{
1228 CBOR cbor; 1528 CBOR cbor;
1229 cbor_init (&cbor); 1529 cbor_init (&cbor);
1530 cbor.flags |= ix;
1230 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN; 1531 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;
1231 XPUSHs (scalar); 1532 XPUSHs (scalar);
1232} 1533}
1233 1534
1234void decode_cbor (SV *cborstr) 1535void decode_cbor (SV *cborstr)

Diff Legend

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