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 1.32 by root, Thu Nov 28 12:08:07 2013 UTC vs.
Revision 1.49 by root, Mon Feb 8 04:11:11 2016 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 102#define F_ALLOW_SHARING 0x00000004UL
103#define F_ALLOW_CYCLES 0x00000008UL
65#define F_PACK_STRINGS 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 (ecb_expect_true (len < 24)) 240 if (ecb_expect_true (len < LENGTH_EXT1))
195 *enc->cur++ = major | len; 241 *enc->cur++ = major | len;
196 else if (ecb_expect_true (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);
233} 279}
234 280
235ecb_inline void 281ecb_inline void
236encode_str (enc_t *enc, int utf8, char *str, STRLEN len) 282encode_str (enc_t *enc, int utf8, char *str, STRLEN len)
237{ 283{
238 encode_uint (enc, utf8 ? 0x60 : 0x40, len); 284 encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len);
239 need (enc, len); 285 need (enc, len);
240 memcpy (enc->cur, str, len); 286 memcpy (enc->cur, str, len);
241 enc->cur += len; 287 enc->cur += len;
242} 288}
243 289
250 296
251 if (SvOK (*svp)) 297 if (SvOK (*svp))
252 { 298 {
253 // already registered, use stringref 299 // already registered, use stringref
254 encode_tag (enc, CBOR_TAG_STRINGREF); 300 encode_tag (enc, CBOR_TAG_STRINGREF);
255 encode_uint (enc, 0x00, SvUV (*svp)); 301 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
256 return; 302 return;
257 } 303 }
258 else if (len >= minimum_string_length (enc->stringref_idx)) 304 else if (len >= minimum_string_length (enc->stringref_idx))
259 { 305 {
260 // register only 306 // register only
276 if (enc->depth >= enc->cbor.max_depth) 322 if (enc->depth >= enc->cbor.max_depth)
277 croak (ERR_NESTING_EXCEEDED); 323 croak (ERR_NESTING_EXCEEDED);
278 324
279 ++enc->depth; 325 ++enc->depth;
280 326
281 encode_uint (enc, 0x80, len + 1); 327 encode_uint (enc, MAJOR_ARRAY, len + 1);
282 328
329 if (SvMAGICAL (av))
283 for (i = 0; i <= len; ++i) 330 for (i = 0; i <= len; ++i)
284 { 331 {
285 SV **svp = av_fetch (av, i, 0); 332 SV **svp = av_fetch (av, i, 0);
286 encode_sv (enc, svp ? *svp : &PL_sv_undef); 333 encode_sv (enc, svp ? *svp : &PL_sv_undef);
287 } 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 }
288 341
289 --enc->depth; 342 --enc->depth;
290} 343}
291 344
292static void 345static void
301 354
302 int pairs = hv_iterinit (hv); 355 int pairs = hv_iterinit (hv);
303 int mg = SvMAGICAL (hv); 356 int mg = SvMAGICAL (hv);
304 357
305 if (mg) 358 if (mg)
306 encode_ch (enc, 0xa0 | 31); 359 encode_ch (enc, MAJOR_MAP | MINOR_INDEF);
307 else 360 else
308 encode_uint (enc, 0xa0, pairs); 361 encode_uint (enc, MAJOR_MAP, pairs);
309 362
310 while ((he = hv_iternext (hv))) 363 while ((he = hv_iternext (hv)))
311 { 364 {
312 if (HeKLEN (he) == HEf_SVKEY) 365 if (HeKLEN (he) == HEf_SVKEY)
313 encode_sv (enc, HeSVKEY (he)); 366 encode_sv (enc, HeSVKEY (he));
316 369
317 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));
318 } 371 }
319 372
320 if (mg) 373 if (mg)
321 encode_ch (enc, 0xe0 | 31); 374 encode_ch (enc, MAJOR_MISC | MINOR_INDEF);
322 375
323 --enc->depth; 376 --enc->depth;
324} 377}
325 378
326// encode objects, arrays and special \0=false and \1=true values. 379// encode objects, arrays and special \0=false and \1=true values.
327static void 380static void
328encode_rv (enc_t *enc, SV *sv) 381encode_rv (enc_t *enc, SV *sv)
329{ 382{
330 SvGETMAGIC (sv); 383 SvGETMAGIC (sv);
331
332 if (ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING)
333 && ecb_expect_false (SvREFCNT (sv) > 1))
334 {
335 if (!enc->shareable)
336 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
337
338 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
339
340 if (SvOK (*svp))
341 {
342 encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF);
343 encode_uint (enc, 0x00, SvUV (*svp));
344 return;
345 }
346 else
347 {
348 sv_setuv (*svp, enc->shareable_idx);
349 ++enc->shareable_idx;
350 encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE);
351 }
352 }
353 384
354 svtype svt = SvTYPE (sv); 385 svtype svt = SvTYPE (sv);
355 386
356 if (ecb_expect_false (SvOBJECT (sv))) 387 if (ecb_expect_false (SvOBJECT (sv)))
357 { 388 {
364 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 395 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
365 ? cbor_tagged_stash 396 ? cbor_tagged_stash
366 : gv_stashpv ("CBOR::XS::Tagged" , 1); 397 : gv_stashpv ("CBOR::XS::Tagged" , 1);
367 398
368 HV *stash = SvSTASH (sv); 399 HV *stash = SvSTASH (sv);
369 GV *method;
370 400
371 if (stash == boolean_stash) 401 if (stash == boolean_stash)
372 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 }
373 else if (stash == error_stash) 406 else if (stash == error_stash)
374 encode_ch (enc, 0xe0 | 23); 407 {
408 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
409 return;
410 }
375 else if (stash == tagged_stash) 411 else if (stash == tagged_stash)
376 { 412 {
377 if (svt != SVt_PVAV) 413 if (svt != SVt_PVAV)
378 croak ("encountered CBOR::XS::Tagged object that isn't an array"); 414 croak ("encountered CBOR::XS::Tagged object that isn't an array");
379 415
380 encode_uint (enc, 0xc0, SvUV (*av_fetch ((AV *)sv, 0, 1))); 416 encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1)));
381 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))
382 } 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
383 else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0))) 450 if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
384 { 451 {
385 dSP; 452 dSP;
386 453
387 ENTER; SAVETMPS; PUSHMARK (SP); 454 ENTER; SAVETMPS; PUSHMARK (SP);
388 // 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
420 // catch this surprisingly common error 487 // catch this surprisingly common error
421 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv) 488 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)
422 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash)); 489 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));
423 490
424 encode_tag (enc, CBOR_TAG_PERL_OBJECT); 491 encode_tag (enc, CBOR_TAG_PERL_OBJECT);
425 encode_uint (enc, 0x80, count + 1); 492 encode_uint (enc, MAJOR_ARRAY, count + 1);
426 encode_strref (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash)); 493 encode_strref (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
427 494
428 while (count) 495 while (count)
429 encode_sv (enc, SP[1 - count--]); 496 encode_sv (enc, SP[1 - count--]);
430 497
452{ 519{
453 double nv = SvNVX (sv); 520 double nv = SvNVX (sv);
454 521
455 need (enc, 9); 522 need (enc, 9);
456 523
457 if (ecb_expect_false (nv == (U32)nv)) 524 if (ecb_expect_false (nv == (NV)(U32)nv))
458 encode_uint (enc, 0x00, (U32)nv); 525 encode_uint (enc, MAJOR_POS_INT, (U32)nv);
459 //TODO: maybe I32? 526 //TODO: maybe I32?
460 else if (ecb_expect_false (nv == (float)nv)) 527 else if (ecb_expect_false (nv == (float)nv))
461 { 528 {
462 uint32_t fp = ecb_float_to_binary32 (nv); 529 uint32_t fp = ecb_float_to_binary32 (nv);
463 530
464 *enc->cur++ = 0xe0 | 26; 531 *enc->cur++ = MAJOR_MISC | MISC_FLOAT32;
465 532
466 if (!ecb_big_endian ()) 533 if (!ecb_big_endian ())
467 fp = ecb_bswap32 (fp); 534 fp = ecb_bswap32 (fp);
468 535
469 memcpy (enc->cur, &fp, 4); 536 memcpy (enc->cur, &fp, 4);
471 } 538 }
472 else 539 else
473 { 540 {
474 uint64_t fp = ecb_double_to_binary64 (nv); 541 uint64_t fp = ecb_double_to_binary64 (nv);
475 542
476 *enc->cur++ = 0xe0 | 27; 543 *enc->cur++ = MAJOR_MISC | MISC_FLOAT64;
477 544
478 if (!ecb_big_endian ()) 545 if (!ecb_big_endian ())
479 fp = ecb_bswap64 (fp); 546 fp = ecb_bswap64 (fp);
480 547
481 memcpy (enc->cur, &fp, 8); 548 memcpy (enc->cur, &fp, 8);
497 else if (SvNOKp (sv)) 564 else if (SvNOKp (sv))
498 encode_nv (enc, sv); 565 encode_nv (enc, sv);
499 else if (SvIOKp (sv)) 566 else if (SvIOKp (sv))
500 { 567 {
501 if (SvIsUV (sv)) 568 if (SvIsUV (sv))
502 encode_uint (enc, 0x00, SvUVX (sv)); 569 encode_uint (enc, MAJOR_POS_INT, SvUVX (sv));
503 else if (SvIVX (sv) >= 0) 570 else if (SvIVX (sv) >= 0)
504 encode_uint (enc, 0x00, SvIVX (sv)); 571 encode_uint (enc, MAJOR_POS_INT, SvIVX (sv));
505 else 572 else
506 encode_uint (enc, 0x20, -(SvIVX (sv) + 1)); 573 encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1));
507 } 574 }
508 else if (SvROK (sv)) 575 else if (SvROK (sv))
509 encode_rv (enc, SvRV (sv)); 576 encode_rv (enc, SvRV (sv));
510 else if (!SvOK (sv)) 577 else if (!SvOK (sv))
511 encode_ch (enc, 0xe0 | 22); 578 encode_ch (enc, MAJOR_MISC | SIMPLE_NULL);
512 else if (enc->cbor.flags & F_ALLOW_UNKNOWN) 579 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
513 encode_ch (enc, 0xe0 | 23); 580 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
514 else 581 else
515 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",
516 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv)); 583 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
517} 584}
518 585
519static SV * 586static SV *
520encode_cbor (SV *scalar, CBOR *cbor) 587encode_cbor (SV *scalar, CBOR *cbor)
521{ 588{
522 enc_t enc = { }; 589 enc_t enc = { 0 };
523 590
524 enc.cbor = *cbor; 591 enc.cbor = *cbor;
525 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 592 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
526 enc.cur = SvPVX (enc.sv); 593 enc.cur = SvPVX (enc.sv);
527 enc.end = SvEND (enc.sv); 594 enc.end = SvEND (enc.sv);
571#define DEC_DEC_DEPTH --dec->depth 638#define DEC_DEC_DEPTH --dec->depth
572 639
573static UV 640static UV
574decode_uint (dec_t *dec) 641decode_uint (dec_t *dec)
575{ 642{
576 switch (*dec->cur & 31) 643 U8 m = *dec->cur & MINOR_MASK;
577 { 644 ++dec->cur;
578 case 0: case 1: case 2: case 3: case 4: case 5: case 6: case 7:
579 case 8: case 9: case 10: case 11: case 12: case 13: case 14: case 15:
580 case 16: case 17: case 18: case 19: case 20: case 21: case 22: case 23:
581 return *dec->cur++ & 31;
582 645
583 case 24: 646 if (ecb_expect_true (m < LENGTH_EXT1))
647 return m;
648 else if (ecb_expect_true (m == LENGTH_EXT1))
649 {
584 WANT (2); 650 WANT (1);
585 dec->cur += 2; 651 dec->cur += 1;
586 return dec->cur[-1]; 652 return dec->cur[-1];
587 653 }
588 case 25: 654 else if (ecb_expect_true (m == LENGTH_EXT2))
655 {
589 WANT (3); 656 WANT (2);
590 dec->cur += 3; 657 dec->cur += 2;
591 return (((UV)dec->cur[-2]) << 8) 658 return (((UV)dec->cur[-2]) << 8)
592 | ((UV)dec->cur[-1]); 659 | ((UV)dec->cur[-1]);
593 660 }
594 case 26: 661 else if (ecb_expect_true (m == LENGTH_EXT4))
662 {
595 WANT (5); 663 WANT (4);
596 dec->cur += 5; 664 dec->cur += 4;
597 return (((UV)dec->cur[-4]) << 24) 665 return (((UV)dec->cur[-4]) << 24)
598 | (((UV)dec->cur[-3]) << 16) 666 | (((UV)dec->cur[-3]) << 16)
599 | (((UV)dec->cur[-2]) << 8) 667 | (((UV)dec->cur[-2]) << 8)
600 | ((UV)dec->cur[-1]); 668 | ((UV)dec->cur[-1]);
601 669 }
602 case 27: 670 else if (ecb_expect_true (m == LENGTH_EXT8))
671 {
603 WANT (9); 672 WANT (8);
604 dec->cur += 9; 673 dec->cur += 8;
674
675 return
676#if UVSIZE < 8
677 0
678#else
605 return (((UV)dec->cur[-8]) << 56) 679 (((UV)dec->cur[-8]) << 56)
606 | (((UV)dec->cur[-7]) << 48) 680 | (((UV)dec->cur[-7]) << 48)
607 | (((UV)dec->cur[-6]) << 40) 681 | (((UV)dec->cur[-6]) << 40)
608 | (((UV)dec->cur[-5]) << 32) 682 | (((UV)dec->cur[-5]) << 32)
683#endif
609 | (((UV)dec->cur[-4]) << 24) 684 | (((UV)dec->cur[-4]) << 24)
610 | (((UV)dec->cur[-3]) << 16) 685 | (((UV)dec->cur[-3]) << 16)
611 | (((UV)dec->cur[-2]) << 8) 686 | (((UV)dec->cur[-2]) << 8)
612 | ((UV)dec->cur[-1]); 687 | ((UV)dec->cur[-1]);
613 688 }
614 default: 689 else
615 ERR ("corrupted CBOR data (unsupported integer minor encoding)"); 690 ERR ("corrupted CBOR data (unsupported integer minor encoding)");
616 }
617 691
618fail: 692fail:
619 return 0; 693 return 0;
620} 694}
621 695
626{ 700{
627 AV *av = newAV (); 701 AV *av = newAV ();
628 702
629 DEC_INC_DEPTH; 703 DEC_INC_DEPTH;
630 704
631 if ((*dec->cur & 31) == 31) 705 if (*dec->cur == (MAJOR_ARRAY | MINOR_INDEF))
632 { 706 {
633 ++dec->cur; 707 ++dec->cur;
634 708
635 for (;;) 709 for (;;)
636 { 710 {
637 WANT (1); 711 WANT (1);
638 712
639 if (*dec->cur == (0xe0 | 31)) 713 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
640 { 714 {
641 ++dec->cur; 715 ++dec->cur;
642 break; 716 break;
643 } 717 }
644 718
647 } 721 }
648 else 722 else
649 { 723 {
650 int i, len = decode_uint (dec); 724 int i, len = decode_uint (dec);
651 725
726 WANT (len); // complexity check for av_fill - need at least one byte per value, do not allow supersize arrays
652 av_fill (av, len - 1); 727 av_fill (av, len - 1);
653 728
654 for (i = 0; i < len; ++i) 729 for (i = 0; i < len; ++i)
655 AvARRAY (av)[i] = decode_sv (dec); 730 AvARRAY (av)[i] = decode_sv (dec);
656 } 731 }
669{ 744{
670 // for speed reasons, we specialcase single-string 745 // for speed reasons, we specialcase single-string
671 // byte or utf-8 strings as keys, but only when !stringref 746 // byte or utf-8 strings as keys, but only when !stringref
672 747
673 if (ecb_expect_true (!dec->stringref)) 748 if (ecb_expect_true (!dec->stringref))
674 if (*dec->cur >= 0x40 && *dec->cur <= 0x40 + 27) 749 if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8))
675 { 750 {
676 I32 len = decode_uint (dec); 751 I32 len = decode_uint (dec);
677 char *key = (char *)dec->cur; 752 char *key = (char *)dec->cur;
678 753
754 WANT (len);
679 dec->cur += len; 755 dec->cur += len;
680 756
681 if (ecb_expect_false (dec->stringref))
682 av_push (dec->stringref, newSVpvn (key, len));
683
684 hv_store (hv, key, len, decode_sv (dec), 0); 757 hv_store (hv, key, len, decode_sv (dec), 0);
685 758
686 return; 759 return;
687 } 760 }
688 else if (*dec->cur >= 0x60 && *dec->cur <= 0x60 + 27) 761 else if (ecb_expect_true ((U8)(*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8))
689 { 762 {
690 I32 len = decode_uint (dec); 763 I32 len = decode_uint (dec);
691 char *key = (char *)dec->cur; 764 char *key = (char *)dec->cur;
692 765
766 WANT (len);
693 dec->cur += len; 767 dec->cur += len;
694 768
695 if (ecb_expect_false (dec->stringref)) 769 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
696 av_push (dec->stringref, newSVpvn_utf8 (key, len, 1)); 770 if (!is_utf8_string (key, len))
771 ERR ("corrupted CBOR data (invalid UTF-8 in map key)");
697 772
698 hv_store (hv, key, -len, decode_sv (dec), 0); 773 hv_store (hv, key, -len, decode_sv (dec), 0);
699 774
700 return; 775 return;
701 } 776 }
703 SV *k = decode_sv (dec); 778 SV *k = decode_sv (dec);
704 SV *v = decode_sv (dec); 779 SV *v = decode_sv (dec);
705 780
706 hv_store_ent (hv, k, v, 0); 781 hv_store_ent (hv, k, v, 0);
707 SvREFCNT_dec (k); 782 SvREFCNT_dec (k);
783
784fail:
785 ;
708} 786}
709 787
710static SV * 788static SV *
711decode_hv (dec_t *dec) 789decode_hv (dec_t *dec)
712{ 790{
713 HV *hv = newHV (); 791 HV *hv = newHV ();
714 792
715 DEC_INC_DEPTH; 793 DEC_INC_DEPTH;
716 794
717 if ((*dec->cur & 31) == 31) 795 if (*dec->cur == (MAJOR_MAP | MINOR_INDEF))
718 { 796 {
719 ++dec->cur; 797 ++dec->cur;
720 798
721 for (;;) 799 for (;;)
722 { 800 {
723 WANT (1); 801 WANT (1);
724 802
725 if (*dec->cur == (0xe0 | 31)) 803 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
726 { 804 {
727 ++dec->cur; 805 ++dec->cur;
728 break; 806 break;
729 } 807 }
730 808
751static SV * 829static SV *
752decode_str (dec_t *dec, int utf8) 830decode_str (dec_t *dec, int utf8)
753{ 831{
754 SV *sv = 0; 832 SV *sv = 0;
755 833
756 if ((*dec->cur & 31) == 31) 834 if ((*dec->cur & MINOR_MASK) == MINOR_INDEF)
757 { 835 {
836 // indefinite length strings
758 ++dec->cur; 837 ++dec->cur;
759 838
839 U8 major = *dec->cur & MAJOR_MISC;
840
760 sv = newSVpvn ("", 0); 841 sv = newSVpvn ("", 0);
761 842
762 // not very fast, and certainly not robust against illegal input
763 for (;;) 843 for (;;)
764 { 844 {
765 WANT (1); 845 WANT (1);
766 846
767 if (*dec->cur == (0xe0 | 31)) 847 if ((*dec->cur - major) > LENGTH_EXT8)
848 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
768 { 849 {
769 ++dec->cur; 850 ++dec->cur;
770 break; 851 break;
771 } 852 }
853 else
854 ERR ("corrupted CBOR data (invalid chunks in indefinite length string)");
772 855
773 sv_catsv (sv, decode_sv (dec)); 856 STRLEN len = decode_uint (dec);
857
858 WANT (len);
859 sv_catpvn (sv, dec->cur, len);
860 dec->cur += len;
774 } 861 }
775 } 862 }
776 else 863 else
777 { 864 {
778 STRLEN len = decode_uint (dec); 865 STRLEN len = decode_uint (dec);
785 && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1)) 872 && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1))
786 av_push (dec->stringref, SvREFCNT_inc_NN (sv)); 873 av_push (dec->stringref, SvREFCNT_inc_NN (sv));
787 } 874 }
788 875
789 if (utf8) 876 if (utf8)
877 {
878 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
879 if (!is_utf8_string (SvPVX (sv), SvCUR (sv)))
880 ERR ("corrupted CBOR data (invalid UTF-8 in text string)");
881
790 SvUTF8_on (sv); 882 SvUTF8_on (sv);
883 }
791 884
792 return sv; 885 return sv;
793 886
794fail: 887fail:
795 SvREFCNT_dec (sv); 888 SvREFCNT_dec (sv);
827 } 920 }
828 break; 921 break;
829 922
830 case CBOR_TAG_STRINGREF: 923 case CBOR_TAG_STRINGREF:
831 { 924 {
832 if ((*dec->cur >> 5) != 0) 925 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
833 ERR ("corrupted CBOR data (stringref index not an unsigned integer)"); 926 ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
834 927
835 UV idx = decode_uint (dec); 928 UV idx = decode_uint (dec);
836 929
837 if (!dec->stringref || (int)idx > AvFILLp (dec->stringref)) 930 if (!dec->stringref || (int)idx > AvFILLp (dec->stringref))
844 case CBOR_TAG_VALUE_SHAREABLE: 937 case CBOR_TAG_VALUE_SHAREABLE:
845 { 938 {
846 if (ecb_expect_false (!dec->shareable)) 939 if (ecb_expect_false (!dec->shareable))
847 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ()); 940 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ());
848 941
942 if (dec->cbor.flags & F_ALLOW_CYCLES)
943 {
849 sv = newSV (0); 944 sv = newSV (0);
850 av_push (dec->shareable, SvREFCNT_inc_NN (sv)); 945 av_push (dec->shareable, SvREFCNT_inc_NN (sv));
851 946
852 SV *osv = decode_sv (dec); 947 SV *osv = decode_sv (dec);
853 sv_setsv (sv, osv); 948 sv_setsv (sv, osv);
854 SvREFCNT_dec_NN (osv); 949 SvREFCNT_dec_NN (osv);
950 }
951 else
952 {
953 av_push (dec->shareable, &PL_sv_undef);
954 int idx = AvFILLp (dec->shareable);
955 sv = decode_sv (dec);
956 av_store (dec->shareable, idx, SvREFCNT_inc_NN (sv));
957 }
855 } 958 }
856 break; 959 break;
857 960
858 case CBOR_TAG_VALUE_SHAREDREF: 961 case CBOR_TAG_VALUE_SHAREDREF:
859 { 962 {
860 if ((*dec->cur >> 5) != 0) 963 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
861 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)"); 964 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
862 965
863 UV idx = decode_uint (dec); 966 UV idx = decode_uint (dec);
864 967
865 if (!dec->shareable || (int)idx > AvFILLp (dec->shareable)) 968 if (!dec->shareable || (int)idx > AvFILLp (dec->shareable))
866 ERR ("corrupted CBOR data (sharedref index out of bounds)"); 969 ERR ("corrupted CBOR data (sharedref index out of bounds)");
867 970
868 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]); 971 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]);
972
973 if (sv == &PL_sv_undef)
974 ERR ("cyclic CBOR data structure found, but allow_cycles is not enabled");
869 } 975 }
870 break; 976 break;
871 977
872 case CBOR_TAG_PERL_OBJECT: 978 case CBOR_TAG_PERL_OBJECT:
873 { 979 {
974static SV * 1080static SV *
975decode_sv (dec_t *dec) 1081decode_sv (dec_t *dec)
976{ 1082{
977 WANT (1); 1083 WANT (1);
978 1084
979 switch (*dec->cur >> 5) 1085 switch (*dec->cur >> MAJOR_SHIFT)
980 { 1086 {
981 case 0: // unsigned int 1087 case MAJOR_POS_INT >> MAJOR_SHIFT: return newSVuv (decode_uint (dec));
982 return newSVuv (decode_uint (dec)); 1088 case MAJOR_NEG_INT >> MAJOR_SHIFT: return newSViv (-1 - (IV)decode_uint (dec));
983 case 1: // negative int 1089 case MAJOR_BYTES >> MAJOR_SHIFT: return decode_str (dec, 0);
984 return newSViv (-1 - (IV)decode_uint (dec)); 1090 case MAJOR_TEXT >> MAJOR_SHIFT: return decode_str (dec, 1);
985 case 2: // octet string 1091 case MAJOR_ARRAY >> MAJOR_SHIFT: return decode_av (dec);
986 return decode_str (dec, 0); 1092 case MAJOR_MAP >> MAJOR_SHIFT: return decode_hv (dec);
987 case 3: // utf-8 string 1093 case MAJOR_TAG >> MAJOR_SHIFT: return decode_tagged (dec);
988 return decode_str (dec, 1); 1094
989 case 4: // array 1095 case MAJOR_MISC >> MAJOR_SHIFT:
990 return decode_av (dec);
991 case 5: // map
992 return decode_hv (dec);
993 case 6: // tag
994 return decode_tagged (dec);
995 case 7: // misc
996 switch (*dec->cur++ & 31) 1096 switch (*dec->cur++ & MINOR_MASK)
997 { 1097 {
998 case 20: 1098 case SIMPLE_FALSE:
999#if CBOR_SLOW 1099#if CBOR_SLOW
1000 types_false = get_bool ("Types::Serialiser::false"); 1100 types_false = get_bool ("Types::Serialiser::false");
1001#endif 1101#endif
1002 return newSVsv (types_false); 1102 return newSVsv (types_false);
1003 case 21: 1103 case SIMPLE_TRUE:
1004#if CBOR_SLOW 1104#if CBOR_SLOW
1005 types_true = get_bool ("Types::Serialiser::true"); 1105 types_true = get_bool ("Types::Serialiser::true");
1006#endif 1106#endif
1007 return newSVsv (types_true); 1107 return newSVsv (types_true);
1008 case 22: 1108 case SIMPLE_NULL:
1009 return newSVsv (&PL_sv_undef); 1109 return newSVsv (&PL_sv_undef);
1010 case 23: 1110 case SIMPLE_UNDEF:
1011#if CBOR_SLOW 1111#if CBOR_SLOW
1012 types_error = get_bool ("Types::Serialiser::error"); 1112 types_error = get_bool ("Types::Serialiser::error");
1013#endif 1113#endif
1014 return newSVsv (types_error); 1114 return newSVsv (types_error);
1015 1115
1016 case 25: 1116 case MISC_FLOAT16:
1017 { 1117 {
1018 WANT (2); 1118 WANT (2);
1019 1119
1020 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1]; 1120 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1];
1021 dec->cur += 2; 1121 dec->cur += 2;
1022 1122
1023 return newSVnv (ecb_binary16_to_float (fp)); 1123 return newSVnv (ecb_binary16_to_float (fp));
1024 } 1124 }
1025 1125
1026 case 26: 1126 case MISC_FLOAT32:
1027 { 1127 {
1028 uint32_t fp; 1128 uint32_t fp;
1029 WANT (4); 1129 WANT (4);
1030 memcpy (&fp, dec->cur, 4); 1130 memcpy (&fp, dec->cur, 4);
1031 dec->cur += 4; 1131 dec->cur += 4;
1034 fp = ecb_bswap32 (fp); 1134 fp = ecb_bswap32 (fp);
1035 1135
1036 return newSVnv (ecb_binary32_to_float (fp)); 1136 return newSVnv (ecb_binary32_to_float (fp));
1037 } 1137 }
1038 1138
1039 case 27: 1139 case MISC_FLOAT64:
1040 { 1140 {
1041 uint64_t fp; 1141 uint64_t fp;
1042 WANT (8); 1142 WANT (8);
1043 memcpy (&fp, dec->cur, 8); 1143 memcpy (&fp, dec->cur, 8);
1044 dec->cur += 8; 1144 dec->cur += 8;
1047 fp = ecb_bswap64 (fp); 1147 fp = ecb_bswap64 (fp);
1048 1148
1049 return newSVnv (ecb_binary64_to_double (fp)); 1149 return newSVnv (ecb_binary64_to_double (fp));
1050 } 1150 }
1051 1151
1052 // 0..19 unassigned 1152 // 0..19 unassigned simple
1053 // 24 reserved + unassigned (reserved values are not encodable) 1153 // 24 reserved + unassigned simple (reserved values are not encodable)
1154 // 28-30 unassigned misc
1155 // 31 break code
1054 default: 1156 default:
1055 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)"); 1157 ERR ("corrupted CBOR data (reserved/unassigned/unexpected major 7 value)");
1056 } 1158 }
1057 1159
1058 break; 1160 break;
1059 } 1161 }
1060 1162
1063} 1165}
1064 1166
1065static SV * 1167static SV *
1066decode_cbor (SV *string, CBOR *cbor, char **offset_return) 1168decode_cbor (SV *string, CBOR *cbor, char **offset_return)
1067{ 1169{
1068 dec_t dec = { }; 1170 dec_t dec = { 0 };
1069 SV *sv; 1171 SV *sv;
1070 STRLEN len; 1172 STRLEN len;
1071 char *data = SvPVbyte (string, len); 1173 char *data = SvPVbyte (string, len);
1072 1174
1073 if (len > cbor->max_size && cbor->max_size) 1175 if (len > cbor->max_size && cbor->max_size)
1087 if (dec.cur != dec.end && !dec.err) 1189 if (dec.cur != dec.end && !dec.err)
1088 dec.err = "garbage after CBOR object"; 1190 dec.err = "garbage after CBOR object";
1089 1191
1090 if (dec.err) 1192 if (dec.err)
1091 { 1193 {
1194 if (dec.shareable)
1195 {
1196 // need to break cyclic links, which whould all be in shareable
1197 int i;
1198 SV **svp;
1199
1200 for (i = av_len (dec.shareable) + 1; i--; )
1201 if ((svp = av_fetch (dec.shareable, i, 0)))
1202 sv_setsv (*svp, &PL_sv_undef);
1203 }
1204
1092 SvREFCNT_dec (sv); 1205 SvREFCNT_dec (sv);
1093 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur); 1206 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur);
1094 } 1207 }
1095 1208
1096 sv = sv_2mortal (sv); 1209 sv = sv_2mortal (sv);
1097 1210
1098 return sv; 1211 return sv;
1099} 1212}
1100 1213
1214/////////////////////////////////////////////////////////////////////////////
1215// incremental parser
1216
1217#define INCR_DONE(cbor) (AvFILLp (cbor->incr_count) < 0)
1218
1219// returns 0 for notyet, 1 for success or error
1220static int
1221incr_parse (CBOR *self, SV *cborstr)
1222{
1223 STRLEN cur;
1224 SvPV (cborstr, cur);
1225
1226 while (ecb_expect_true (self->incr_need <= cur))
1227 {
1228 // table of integer count bytes
1229 static I8 incr_len[MINOR_MASK + 1] = {
1230 0, 0, 0, 0, 0, 0, 0, 0,
1231 0, 0, 0, 0, 0, 0, 0, 0,
1232 0, 0, 0, 0, 0, 0, 0, 0,
1233 1, 2, 4, 8,-1,-1,-1,-2
1234 };
1235
1236 const U8 *p = SvPVX (cborstr) + self->incr_pos;
1237 U8 m = *p & MINOR_MASK;
1238 IV count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1239 I8 ilen = incr_len[m];
1240
1241 self->incr_need = self->incr_pos + 1;
1242
1243 if (ecb_expect_false (ilen < 0))
1244 {
1245 if (m != MINOR_INDEF)
1246 return 1; // error
1247
1248 if (*p == (MAJOR_MISC | MINOR_INDEF))
1249 {
1250 if (count >= 0)
1251 return 1; // error
1252
1253 count = 1;
1254 }
1255 else
1256 {
1257 av_push (self->incr_count, newSViv (-1)); //TODO: nest
1258 count = -1;
1259 }
1260 }
1261 else
1262 {
1263 self->incr_need += ilen;
1264 if (ecb_expect_false (self->incr_need > cur))
1265 return 0;
1266
1267 int major = *p >> MAJOR_SHIFT;
1268
1269 switch (major)
1270 {
1271 case MAJOR_TAG >> MAJOR_SHIFT:
1272 ++count; // tags merely prefix another value
1273 break;
1274
1275 case MAJOR_BYTES >> MAJOR_SHIFT:
1276 case MAJOR_TEXT >> MAJOR_SHIFT:
1277 case MAJOR_ARRAY >> MAJOR_SHIFT:
1278 case MAJOR_MAP >> MAJOR_SHIFT:
1279 {
1280 UV len;
1281
1282 if (ecb_expect_false (ilen))
1283 {
1284 len = 0;
1285
1286 do {
1287 len = (len << 8) | *++p;
1288 } while (--ilen);
1289 }
1290 else
1291 len = m;
1292
1293 switch (major)
1294 {
1295 case MAJOR_BYTES >> MAJOR_SHIFT:
1296 case MAJOR_TEXT >> MAJOR_SHIFT:
1297 self->incr_need += len;
1298 if (ecb_expect_false (self->incr_need > cur))
1299 return 0;
1300
1301 break;
1302
1303 case MAJOR_MAP >> MAJOR_SHIFT:
1304 len <<= 1;
1305 case MAJOR_ARRAY >> MAJOR_SHIFT:
1306 if (len)
1307 {
1308 av_push (self->incr_count, newSViv (len + 1)); //TODO: nest
1309 count = len + 1;
1310 }
1311 break;
1312 }
1313 }
1314 }
1315 }
1316
1317 self->incr_pos = self->incr_need;
1318
1319 if (count > 0)
1320 {
1321 while (!--count)
1322 {
1323 if (!AvFILLp (self->incr_count))
1324 return 1; // done
1325
1326 SvREFCNT_dec_NN (av_pop (self->incr_count));
1327 count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1328 }
1329
1330 SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]) = count;
1331 }
1332 }
1333
1334 return 0;
1335}
1336
1337
1101///////////////////////////////////////////////////////////////////////////// 1338/////////////////////////////////////////////////////////////////////////////
1102// XS interface functions 1339// XS interface functions
1103 1340
1104MODULE = CBOR::XS PACKAGE = CBOR::XS 1341MODULE = CBOR::XS PACKAGE = CBOR::XS
1105 1342
1145void shrink (CBOR *self, int enable = 1) 1382void shrink (CBOR *self, int enable = 1)
1146 ALIAS: 1383 ALIAS:
1147 shrink = F_SHRINK 1384 shrink = F_SHRINK
1148 allow_unknown = F_ALLOW_UNKNOWN 1385 allow_unknown = F_ALLOW_UNKNOWN
1149 allow_sharing = F_ALLOW_SHARING 1386 allow_sharing = F_ALLOW_SHARING
1387 allow_cycles = F_ALLOW_CYCLES
1150 pack_strings = F_PACK_STRINGS 1388 pack_strings = F_PACK_STRINGS
1389 validate_utf8 = F_VALIDATE_UTF8
1151 PPCODE: 1390 PPCODE:
1152{ 1391{
1153 if (enable) 1392 if (enable)
1154 self->flags |= ix; 1393 self->flags |= ix;
1155 else 1394 else
1161void get_shrink (CBOR *self) 1400void get_shrink (CBOR *self)
1162 ALIAS: 1401 ALIAS:
1163 get_shrink = F_SHRINK 1402 get_shrink = F_SHRINK
1164 get_allow_unknown = F_ALLOW_UNKNOWN 1403 get_allow_unknown = F_ALLOW_UNKNOWN
1165 get_allow_sharing = F_ALLOW_SHARING 1404 get_allow_sharing = F_ALLOW_SHARING
1405 get_allow_cycles = F_ALLOW_CYCLES
1166 get_pack_strings = F_PACK_STRINGS 1406 get_pack_strings = F_PACK_STRINGS
1407 get_validate_utf8 = F_VALIDATE_UTF8
1167 PPCODE: 1408 PPCODE:
1168 XPUSHs (boolSV (self->flags & ix)); 1409 XPUSHs (boolSV (self->flags & ix));
1169 1410
1170void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) 1411void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
1171 PPCODE: 1412 PPCODE:
1220 EXTEND (SP, 2); 1461 EXTEND (SP, 2);
1221 PUSHs (sv); 1462 PUSHs (sv);
1222 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); 1463 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
1223} 1464}
1224 1465
1466void incr_parse (CBOR *self, SV *cborstr)
1467 ALIAS:
1468 incr_parse_multiple = 1
1469 PPCODE:
1470{
1471 if (SvUTF8 (cborstr))
1472 sv_utf8_downgrade (cborstr, 0);
1473
1474 if (!self->incr_count)
1475 {
1476 self->incr_count = newAV ();
1477 self->incr_pos = 0;
1478 self->incr_need = 1;
1479
1480 av_push (self->incr_count, newSViv (1));
1481 }
1482
1483 do
1484 {
1485 if (!incr_parse (self, cborstr))
1486 {
1487 if (self->incr_need > self->max_size && self->max_size)
1488 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
1489 (unsigned long)self->incr_need, (unsigned long)self->max_size);
1490
1491 break;
1492 }
1493
1494 SV *sv;
1495 char *offset;
1496
1497 PUTBACK; sv = decode_cbor (cborstr, self, &offset); SPAGAIN;
1498 XPUSHs (sv);
1499
1500 sv_chop (cborstr, offset);
1501
1502 av_clear (self->incr_count);
1503 av_push (self->incr_count, newSViv (1));
1504
1505 self->incr_pos = 0;
1506 self->incr_need = self->incr_pos + 1;
1507 }
1508 while (ix);
1509}
1510
1511void incr_reset (CBOR *self)
1512 CODE:
1513{
1514 SvREFCNT_dec (self->incr_count);
1515 self->incr_count = 0;
1516}
1517
1225void DESTROY (CBOR *self) 1518void DESTROY (CBOR *self)
1226 PPCODE: 1519 PPCODE:
1227 cbor_free (self); 1520 cbor_free (self);
1228 1521
1229PROTOTYPES: ENABLE 1522PROTOTYPES: ENABLE
1230 1523
1231void encode_cbor (SV *scalar) 1524void encode_cbor (SV *scalar)
1525 ALIAS:
1526 encode_cbor = 0
1527 encode_cbor_sharing = F_ALLOW_SHARING
1232 PPCODE: 1528 PPCODE:
1233{ 1529{
1234 CBOR cbor; 1530 CBOR cbor;
1235 cbor_init (&cbor); 1531 cbor_init (&cbor);
1532 cbor.flags |= ix;
1236 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN; 1533 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;
1237 XPUSHs (scalar); 1534 XPUSHs (scalar);
1238} 1535}
1239 1536
1240void decode_cbor (SV *cborstr) 1537void decode_cbor (SV *cborstr)

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines