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 1.27 by root, Fri Nov 22 15:28:38 2013 UTC vs.
Revision 1.45 by root, Sun Dec 14 05:48:39 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))
… …
19# define HvNAMELEN(hv) HvNAMELEN_get (hv) 20# define HvNAMELEN(hv) HvNAMELEN_get (hv)
20#endif 21#endif
21#ifndef HvNAMEUTF8 22#ifndef HvNAMEUTF8
22# define HvNAMEUTF8(hv) 0 23# define HvNAMEUTF8(hv) 0
23#endif 24#endif
25#ifndef SvREFCNT_dec_NN
26# define SvREFCNT_dec_NN(sv) SvREFCNT_dec (sv)
27#endif
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};
24 65
25// known tags 66// known tags
26enum cbor_tag 67enum cbor_tag
27{ 68{
28 // inofficial extensions (pending iana registration) 69 // extensions
70 CBOR_TAG_STRINGREF = 25, // http://cbor.schmorp.de/stringref
29 CBOR_TAG_PERL_OBJECT = 24, // http://cbor.schmorp.de/perl-object 71 CBOR_TAG_PERL_OBJECT = 26, // http://cbor.schmorp.de/perl-object
30 CBOR_TAG_GENERIC_OBJECT = 25, // http://cbor.schmorp.de/generic-object 72 CBOR_TAG_GENERIC_OBJECT = 27, // http://cbor.schmorp.de/generic-object
31 CBOR_TAG_VALUE_SHAREABLE = 26, // http://cbor.schmorp.de/value-sharing 73 CBOR_TAG_VALUE_SHAREABLE = 28, // http://cbor.schmorp.de/value-sharing
32 CBOR_TAG_VALUE_SHAREDREF = 27, // http://cbor.schmorp.de/value-sharing 74 CBOR_TAG_VALUE_SHAREDREF = 29, // http://cbor.schmorp.de/value-sharing
33 CBOR_TAG_STRINGREF_NAMESPACE = 65537, // http://cbor.schmorp.de/stringref 75 CBOR_TAG_STRINGREF_NAMESPACE = 256, // http://cbor.schmorp.de/stringref
34 CBOR_TAG_STRINGREF = 28, // http://cbor.schmorp.de/stringref
35 CBOR_TAG_INDIRECTION = 22098, // http://cbor.schmorp.de/indirection 76 CBOR_TAG_INDIRECTION = 22098, // http://cbor.schmorp.de/indirection
36 77
37 // rfc7049 78 // rfc7049
38 CBOR_TAG_DATETIME = 0, // rfc4287, utf-8 79 CBOR_TAG_DATETIME = 0, // rfc4287, utf-8
39 CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any 80 CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any
40 CBOR_TAG_POS_BIGNUM = 2, // byte string 81 CBOR_TAG_POS_BIGNUM = 2, // byte string
41 CBOR_TAG_NEG_BIGNUM = 3, // byte string 82 CBOR_TAG_NEG_BIGNUM = 3, // byte string
42 CBOR_TAG_DECIMAL = 4, // decimal fraction, array 83 CBOR_TAG_DECIMAL = 4, // decimal fraction, array
43 CBOR_TAG_BIGFLOAT = 5, // array 84 CBOR_TAG_BIGFLOAT = 5, // array
44 85
45 CBOR_TAG_CONV_B64U = 21, // base64url, any 86 CBOR_TAG_CONV_B64U = 21, // base64url, any
46 CBOR_TAG_CONV_B64 = 22, // base64, any 87 CBOR_TAG_CONV_B64 = 22, // base64, any
47 CBOR_TAG_CONV_HEX = 23, // base16, any 88 CBOR_TAG_CONV_HEX = 23, // base16, any
48 CBOR_TAG_CBOR = 24, // embedded cbor, byte string 89 CBOR_TAG_CBOR = 24, // embedded cbor, byte string
49 90
50 CBOR_TAG_URI = 32, // URI rfc3986, utf-8 91 CBOR_TAG_URI = 32, // URI rfc3986, utf-8
51 CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8 92 CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8
52 CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8 93 CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8
53 CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8 94 CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8
54 CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8 95 CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8
55 96
56 CBOR_TAG_MAGIC = 55799 // self-describe cbor 97 CBOR_TAG_MAGIC = 55799, // self-describe cbor
57}; 98};
58 99
59#define F_SHRINK 0x00000001UL 100#define F_SHRINK 0x00000001UL
60#define F_ALLOW_UNKNOWN 0x00000002UL 101#define F_ALLOW_UNKNOWN 0x00000002UL
61#define F_ALLOW_SHARING 0x00000004UL //TODO 102#define F_ALLOW_SHARING 0x00000004UL
62#define F_ALLOW_STRINGREF 0x00000008UL //TODO 103#define F_ALLOW_CYCLES 0x00000008UL
104#define F_PACK_STRINGS 0x00000010UL
105#define F_VALIDATE_UTF8 0x00000020UL
63 106
64#define INIT_SIZE 32 // initial scalar size to be allocated 107#define INIT_SIZE 32 // initial scalar size to be allocated
65 108
66#define SB do { 109#define SB do {
67#define SE } while (0) 110#define SE } while (0)
… …
86typedef struct { 129typedef struct {
87 U32 flags; 130 U32 flags;
88 U32 max_depth; 131 U32 max_depth;
89 STRLEN max_size; 132 STRLEN max_size;
90 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.
91} CBOR; 139} CBOR;
92 140
93ecb_inline void 141ecb_inline void
94cbor_init (CBOR *cbor) 142cbor_init (CBOR *cbor)
95{ 143{
… …
99 147
100ecb_inline void 148ecb_inline void
101cbor_free (CBOR *cbor) 149cbor_free (CBOR *cbor)
102{ 150{
103 SvREFCNT_dec (cbor->filter); 151 SvREFCNT_dec (cbor->filter);
152 SvREFCNT_dec (cbor->incr_count);
104} 153}
105 154
106///////////////////////////////////////////////////////////////////////////// 155/////////////////////////////////////////////////////////////////////////////
107// utility functions 156// utility functions
108 157
… …
138{ 187{
139 return idx > 23 188 return idx > 23
140 ? idx > 0xffU 189 ? idx > 0xffU
141 ? idx > 0xffffU 190 ? idx > 0xffffU
142 ? idx > 0xffffffffU 191 ? idx > 0xffffffffU
192 ? 11
143 ? 7 193 : 7
144 : 6
145 : 5 194 : 5
146 : 4 195 : 4
147 : 3; 196 : 3;
148} 197}
149 198
… …
186static void 235static void
187encode_uint (enc_t *enc, int major, UV len) 236encode_uint (enc_t *enc, int major, UV len)
188{ 237{
189 need (enc, 9); 238 need (enc, 9);
190 239
191 if (len < 24) 240 if (ecb_expect_true (len < LENGTH_EXT1))
192 *enc->cur++ = major | len; 241 *enc->cur++ = major | len;
193 else if (len <= 0xff) 242 else if (ecb_expect_true (len <= 0xffU))
194 { 243 {
195 *enc->cur++ = major | 24; 244 *enc->cur++ = major | LENGTH_EXT1;
196 *enc->cur++ = len; 245 *enc->cur++ = len;
197 } 246 }
198 else if (len <= 0xffff) 247 else if (len <= 0xffffU)
199 { 248 {
200 *enc->cur++ = major | 25; 249 *enc->cur++ = major | LENGTH_EXT2;
201 *enc->cur++ = len >> 8; 250 *enc->cur++ = len >> 8;
202 *enc->cur++ = len; 251 *enc->cur++ = len;
203 } 252 }
204 else if (len <= 0xffffffff) 253 else if (len <= 0xffffffffU)
205 { 254 {
206 *enc->cur++ = major | 26; 255 *enc->cur++ = major | LENGTH_EXT4;
207 *enc->cur++ = len >> 24; 256 *enc->cur++ = len >> 24;
208 *enc->cur++ = len >> 16; 257 *enc->cur++ = len >> 16;
209 *enc->cur++ = len >> 8; 258 *enc->cur++ = len >> 8;
210 *enc->cur++ = len; 259 *enc->cur++ = len;
211 } 260 }
212 else 261 else
213 { 262 {
214 *enc->cur++ = major | 27; 263 *enc->cur++ = major | LENGTH_EXT8;
215 *enc->cur++ = len >> 56; 264 *enc->cur++ = len >> 56;
216 *enc->cur++ = len >> 48; 265 *enc->cur++ = len >> 48;
217 *enc->cur++ = len >> 40; 266 *enc->cur++ = len >> 40;
218 *enc->cur++ = len >> 32; 267 *enc->cur++ = len >> 32;
219 *enc->cur++ = len >> 24; 268 *enc->cur++ = len >> 24;
… …
224} 273}
225 274
226ecb_inline void 275ecb_inline void
227encode_tag (enc_t *enc, UV tag) 276encode_tag (enc_t *enc, UV tag)
228{ 277{
229 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;
230} 288}
231 289
232static void 290static void
233encode_str (enc_t *enc, int utf8, char *str, STRLEN len) 291encode_strref (enc_t *enc, int utf8, char *str, STRLEN len)
234{ 292{
235 if (ecb_expect_false (enc->cbor.flags & F_ALLOW_STRINGREF)) 293 if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS))
236 { 294 {
237 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1); 295 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1);
238 296
239 if (SvOK (*svp)) 297 if (SvOK (*svp))
240 { 298 {
241 // already registered, use stringref 299 // already registered, use stringref
242 encode_tag (enc, CBOR_TAG_STRINGREF); 300 encode_tag (enc, CBOR_TAG_STRINGREF);
243 encode_uint (enc, 0x00, SvUV (*svp)); 301 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
244 return; 302 return;
245 } 303 }
246 else if (len >= minimum_string_length (enc->stringref_idx)) 304 else if (len >= minimum_string_length (enc->stringref_idx))
247 { 305 {
248 // register only 306 // register only
249 sv_setuv (*svp, enc->stringref_idx); 307 sv_setuv (*svp, enc->stringref_idx);
250 ++enc->stringref_idx; 308 ++enc->stringref_idx;
251 } 309 }
252 } 310 }
253 311
254 encode_uint (enc, utf8 ? 0x60 : 0x40, len); 312 encode_str (enc, utf8, str, len);
255 need (enc, len);
256 memcpy (enc->cur, str, len);
257 enc->cur += len;
258} 313}
259 314
260static void encode_sv (enc_t *enc, SV *sv); 315static void encode_sv (enc_t *enc, SV *sv);
261 316
262static void 317static void
… …
267 if (enc->depth >= enc->cbor.max_depth) 322 if (enc->depth >= enc->cbor.max_depth)
268 croak (ERR_NESTING_EXCEEDED); 323 croak (ERR_NESTING_EXCEEDED);
269 324
270 ++enc->depth; 325 ++enc->depth;
271 326
272 encode_uint (enc, 0x80, len + 1); 327 encode_uint (enc, MAJOR_ARRAY, len + 1);
273 328
329 if (SvMAGICAL (av))
274 for (i = 0; i <= len; ++i) 330 for (i = 0; i <= len; ++i)
275 { 331 {
276 SV **svp = av_fetch (av, i, 0); 332 SV **svp = av_fetch (av, i, 0);
277 encode_sv (enc, svp ? *svp : &PL_sv_undef); 333 encode_sv (enc, svp ? *svp : &PL_sv_undef);
278 } 334 }
335 else
336 for (i = 0; i <= len; ++i)
337 encode_sv (enc, AvARRAY (av)[i]);
279 338
280 --enc->depth; 339 --enc->depth;
281} 340}
282 341
283static void 342static void
… …
292 351
293 int pairs = hv_iterinit (hv); 352 int pairs = hv_iterinit (hv);
294 int mg = SvMAGICAL (hv); 353 int mg = SvMAGICAL (hv);
295 354
296 if (mg) 355 if (mg)
297 encode_ch (enc, 0xa0 | 31); 356 encode_ch (enc, MAJOR_MAP | MINOR_INDEF);
298 else 357 else
299 encode_uint (enc, 0xa0, pairs); 358 encode_uint (enc, MAJOR_MAP, pairs);
300 359
301 while ((he = hv_iternext (hv))) 360 while ((he = hv_iternext (hv)))
302 { 361 {
303 if (HeKLEN (he) == HEf_SVKEY) 362 if (HeKLEN (he) == HEf_SVKEY)
304 encode_sv (enc, HeSVKEY (he)); 363 encode_sv (enc, HeSVKEY (he));
305 else 364 else
306 encode_str (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he)); 365 encode_strref (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he));
307 366
308 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he)); 367 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he));
309 } 368 }
310 369
311 if (mg) 370 if (mg)
312 encode_ch (enc, 0xe0 | 31); 371 encode_ch (enc, MAJOR_MISC | MINOR_INDEF);
313 372
314 --enc->depth; 373 --enc->depth;
315} 374}
316 375
317// encode objects, arrays and special \0=false and \1=true values. 376// encode objects, arrays and special \0=false and \1=true values.
318static void 377static void
319encode_rv (enc_t *enc, SV *sv) 378encode_rv (enc_t *enc, SV *sv)
320{ 379{
321 SvGETMAGIC (sv); 380 SvGETMAGIC (sv);
322
323 if (ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING)
324 && ecb_expect_false (SvREFCNT (sv) > 1))
325 {
326 if (!enc->shareable)
327 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
328
329 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
330
331 if (SvOK (*svp))
332 {
333 encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF);
334 encode_uint (enc, 0x00, SvUV (*svp));
335 return;
336 }
337 else
338 {
339 sv_setuv (*svp, enc->shareable_idx);
340 ++enc->shareable_idx;
341 encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE);
342 }
343 }
344 381
345 svtype svt = SvTYPE (sv); 382 svtype svt = SvTYPE (sv);
346 383
347 if (ecb_expect_false (SvOBJECT (sv))) 384 if (ecb_expect_false (SvOBJECT (sv)))
348 { 385 {
… …
355 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 392 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
356 ? cbor_tagged_stash 393 ? cbor_tagged_stash
357 : gv_stashpv ("CBOR::XS::Tagged" , 1); 394 : gv_stashpv ("CBOR::XS::Tagged" , 1);
358 395
359 HV *stash = SvSTASH (sv); 396 HV *stash = SvSTASH (sv);
360 GV *method;
361 397
362 if (stash == boolean_stash) 398 if (stash == boolean_stash)
363 encode_ch (enc, SvIV (sv) ? 0xe0 | 21 : 0xe0 | 20); 399 {
400 encode_ch (enc, SvIV (sv) ? MAJOR_MISC | SIMPLE_TRUE : MAJOR_MISC | SIMPLE_FALSE);
401 return;
402 }
364 else if (stash == error_stash) 403 else if (stash == error_stash)
365 encode_ch (enc, 0xe0 | 23); 404 {
405 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
406 return;
407 }
366 else if (stash == tagged_stash) 408 else if (stash == tagged_stash)
367 { 409 {
368 if (svt != SVt_PVAV) 410 if (svt != SVt_PVAV)
369 croak ("encountered CBOR::XS::Tagged object that isn't an array"); 411 croak ("encountered CBOR::XS::Tagged object that isn't an array");
370 412
371 encode_uint (enc, 0xc0, SvUV (*av_fetch ((AV *)sv, 0, 1))); 413 encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1)));
372 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1)); 414 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1));
415
416 return;
417 }
418 }
419
420 if (ecb_expect_false (SvREFCNT (sv) > 1)
421 && ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING))
422 {
423 if (!enc->shareable)
424 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
425
426 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
427
428 if (SvOK (*svp))
373 } 429 {
430 encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF);
431 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
432 return;
433 }
434 else
435 {
436 sv_setuv (*svp, enc->shareable_idx);
437 ++enc->shareable_idx;
438 encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE);
439 }
440 }
441
442 if (ecb_expect_false (SvOBJECT (sv)))
443 {
444 HV *stash = SvSTASH (sv);
445 GV *method;
446
374 else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0))) 447 if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
375 { 448 {
376 dSP; 449 dSP;
377 450
378 ENTER; SAVETMPS; PUSHMARK (SP); 451 ENTER; SAVETMPS; PUSHMARK (SP);
379 // we re-bless the reference to get overload and other niceties right 452 // we re-bless the reference to get overload and other niceties right
… …
411 // catch this surprisingly common error 484 // catch this surprisingly common error
412 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv) 485 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)
413 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash)); 486 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));
414 487
415 encode_tag (enc, CBOR_TAG_PERL_OBJECT); 488 encode_tag (enc, CBOR_TAG_PERL_OBJECT);
416 encode_uint (enc, 0x80, count + 1); 489 encode_uint (enc, MAJOR_ARRAY, count + 1);
417 encode_str (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash)); 490 encode_strref (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
418 491
419 while (count) 492 while (count)
420 encode_sv (enc, SP[1 - count--]); 493 encode_sv (enc, SP[1 - count--]);
421 494
422 PUTBACK; 495 PUTBACK;
… …
443{ 516{
444 double nv = SvNVX (sv); 517 double nv = SvNVX (sv);
445 518
446 need (enc, 9); 519 need (enc, 9);
447 520
448 if (ecb_expect_false (nv == (U32)nv)) 521 if (ecb_expect_false (nv == (NV)(U32)nv))
449 encode_uint (enc, 0x00, (U32)nv); 522 encode_uint (enc, MAJOR_POS_INT, (U32)nv);
450 //TODO: maybe I32? 523 //TODO: maybe I32?
451 else if (ecb_expect_false (nv == (float)nv)) 524 else if (ecb_expect_false (nv == (float)nv))
452 { 525 {
453 uint32_t fp = ecb_float_to_binary32 (nv); 526 uint32_t fp = ecb_float_to_binary32 (nv);
454 527
455 *enc->cur++ = 0xe0 | 26; 528 *enc->cur++ = MAJOR_MISC | MISC_FLOAT32;
456 529
457 if (!ecb_big_endian ()) 530 if (!ecb_big_endian ())
458 fp = ecb_bswap32 (fp); 531 fp = ecb_bswap32 (fp);
459 532
460 memcpy (enc->cur, &fp, 4); 533 memcpy (enc->cur, &fp, 4);
… …
462 } 535 }
463 else 536 else
464 { 537 {
465 uint64_t fp = ecb_double_to_binary64 (nv); 538 uint64_t fp = ecb_double_to_binary64 (nv);
466 539
467 *enc->cur++ = 0xe0 | 27; 540 *enc->cur++ = MAJOR_MISC | MISC_FLOAT64;
468 541
469 if (!ecb_big_endian ()) 542 if (!ecb_big_endian ())
470 fp = ecb_bswap64 (fp); 543 fp = ecb_bswap64 (fp);
471 544
472 memcpy (enc->cur, &fp, 8); 545 memcpy (enc->cur, &fp, 8);
… …
481 554
482 if (SvPOKp (sv)) 555 if (SvPOKp (sv))
483 { 556 {
484 STRLEN len; 557 STRLEN len;
485 char *str = SvPV (sv, len); 558 char *str = SvPV (sv, len);
486 encode_str (enc, SvUTF8 (sv), str, len); 559 encode_strref (enc, SvUTF8 (sv), str, len);
487 } 560 }
488 else if (SvNOKp (sv)) 561 else if (SvNOKp (sv))
489 encode_nv (enc, sv); 562 encode_nv (enc, sv);
490 else if (SvIOKp (sv)) 563 else if (SvIOKp (sv))
491 { 564 {
492 if (SvIsUV (sv)) 565 if (SvIsUV (sv))
493 encode_uint (enc, 0x00, SvUVX (sv)); 566 encode_uint (enc, MAJOR_POS_INT, SvUVX (sv));
494 else if (SvIVX (sv) >= 0) 567 else if (SvIVX (sv) >= 0)
495 encode_uint (enc, 0x00, SvIVX (sv)); 568 encode_uint (enc, MAJOR_POS_INT, SvIVX (sv));
496 else 569 else
497 encode_uint (enc, 0x20, -(SvIVX (sv) + 1)); 570 encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1));
498 } 571 }
499 else if (SvROK (sv)) 572 else if (SvROK (sv))
500 encode_rv (enc, SvRV (sv)); 573 encode_rv (enc, SvRV (sv));
501 else if (!SvOK (sv)) 574 else if (!SvOK (sv))
502 encode_ch (enc, 0xe0 | 22); 575 encode_ch (enc, MAJOR_MISC | SIMPLE_NULL);
503 else if (enc->cbor.flags & F_ALLOW_UNKNOWN) 576 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
504 encode_ch (enc, 0xe0 | 23); 577 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
505 else 578 else
506 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data", 579 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data",
507 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv)); 580 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
508} 581}
509 582
… …
517 enc.cur = SvPVX (enc.sv); 590 enc.cur = SvPVX (enc.sv);
518 enc.end = SvEND (enc.sv); 591 enc.end = SvEND (enc.sv);
519 592
520 SvPOK_only (enc.sv); 593 SvPOK_only (enc.sv);
521 594
522 if (cbor->flags & F_ALLOW_STRINGREF) 595 if (cbor->flags & F_PACK_STRINGS)
523 { 596 {
524 encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE); 597 encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE);
525 enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ()); 598 enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ());
526 enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ()); 599 enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ());
527 } 600 }
… …
562#define DEC_DEC_DEPTH --dec->depth 635#define DEC_DEC_DEPTH --dec->depth
563 636
564static UV 637static UV
565decode_uint (dec_t *dec) 638decode_uint (dec_t *dec)
566{ 639{
567 switch (*dec->cur & 31) 640 U8 m = *dec->cur & MINOR_MASK;
568 { 641 ++dec->cur;
569 case 0: case 1: case 2: case 3: case 4: case 5: case 6: case 7:
570 case 8: case 9: case 10: case 11: case 12: case 13: case 14: case 15:
571 case 16: case 17: case 18: case 19: case 20: case 21: case 22: case 23:
572 return *dec->cur++ & 31;
573 642
574 case 24: 643 if (ecb_expect_true (m < LENGTH_EXT1))
644 return m;
645 else if (ecb_expect_true (m == LENGTH_EXT1))
646 {
575 WANT (2); 647 WANT (1);
576 dec->cur += 2; 648 dec->cur += 1;
577 return dec->cur[-1]; 649 return dec->cur[-1];
578 650 }
579 case 25: 651 else if (ecb_expect_true (m == LENGTH_EXT2))
652 {
580 WANT (3); 653 WANT (2);
581 dec->cur += 3; 654 dec->cur += 2;
582 return (((UV)dec->cur[-2]) << 8) 655 return (((UV)dec->cur[-2]) << 8)
583 | ((UV)dec->cur[-1]); 656 | ((UV)dec->cur[-1]);
584 657 }
585 case 26: 658 else if (ecb_expect_true (m == LENGTH_EXT4))
659 {
586 WANT (5); 660 WANT (4);
587 dec->cur += 5; 661 dec->cur += 4;
588 return (((UV)dec->cur[-4]) << 24) 662 return (((UV)dec->cur[-4]) << 24)
589 | (((UV)dec->cur[-3]) << 16) 663 | (((UV)dec->cur[-3]) << 16)
590 | (((UV)dec->cur[-2]) << 8) 664 | (((UV)dec->cur[-2]) << 8)
591 | ((UV)dec->cur[-1]); 665 | ((UV)dec->cur[-1]);
592 666 }
593 case 27: 667 else if (ecb_expect_true (m == LENGTH_EXT8))
668 {
594 WANT (9); 669 WANT (8);
595 dec->cur += 9; 670 dec->cur += 8;
671
672 return
673#if UVSIZE < 8
674 0
675#else
596 return (((UV)dec->cur[-8]) << 56) 676 (((UV)dec->cur[-8]) << 56)
597 | (((UV)dec->cur[-7]) << 48) 677 | (((UV)dec->cur[-7]) << 48)
598 | (((UV)dec->cur[-6]) << 40) 678 | (((UV)dec->cur[-6]) << 40)
599 | (((UV)dec->cur[-5]) << 32) 679 | (((UV)dec->cur[-5]) << 32)
680#endif
600 | (((UV)dec->cur[-4]) << 24) 681 | (((UV)dec->cur[-4]) << 24)
601 | (((UV)dec->cur[-3]) << 16) 682 | (((UV)dec->cur[-3]) << 16)
602 | (((UV)dec->cur[-2]) << 8) 683 | (((UV)dec->cur[-2]) << 8)
603 | ((UV)dec->cur[-1]); 684 | ((UV)dec->cur[-1]);
604 685 }
605 default: 686 else
606 ERR ("corrupted CBOR data (unsupported integer minor encoding)"); 687 ERR ("corrupted CBOR data (unsupported integer minor encoding)");
607 }
608 688
609fail: 689fail:
610 return 0; 690 return 0;
611} 691}
612 692
… …
617{ 697{
618 AV *av = newAV (); 698 AV *av = newAV ();
619 699
620 DEC_INC_DEPTH; 700 DEC_INC_DEPTH;
621 701
622 if ((*dec->cur & 31) == 31) 702 if (*dec->cur == (MAJOR_ARRAY | MINOR_INDEF))
623 { 703 {
624 ++dec->cur; 704 ++dec->cur;
625 705
626 for (;;) 706 for (;;)
627 { 707 {
628 WANT (1); 708 WANT (1);
629 709
630 if (*dec->cur == (0xe0 | 31)) 710 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
631 { 711 {
632 ++dec->cur; 712 ++dec->cur;
633 break; 713 break;
634 } 714 }
635 715
… …
638 } 718 }
639 else 719 else
640 { 720 {
641 int i, len = decode_uint (dec); 721 int i, len = decode_uint (dec);
642 722
723 WANT (len); // complexity check for av_fill - need at least one byte per value, do not allow supersize arrays
643 av_fill (av, len - 1); 724 av_fill (av, len - 1);
644 725
645 for (i = 0; i < len; ++i) 726 for (i = 0; i < len; ++i)
646 AvARRAY (av)[i] = decode_sv (dec); 727 AvARRAY (av)[i] = decode_sv (dec);
647 } 728 }
… …
660{ 741{
661 // for speed reasons, we specialcase single-string 742 // for speed reasons, we specialcase single-string
662 // byte or utf-8 strings as keys, but only when !stringref 743 // byte or utf-8 strings as keys, but only when !stringref
663 744
664 if (ecb_expect_true (!dec->stringref)) 745 if (ecb_expect_true (!dec->stringref))
665 if (*dec->cur >= 0x40 && *dec->cur <= 0x40 + 27) 746 if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8))
666 { 747 {
667 I32 len = decode_uint (dec); 748 I32 len = decode_uint (dec);
668 char *key = (char *)dec->cur; 749 char *key = (char *)dec->cur;
669 750
670 dec->cur += len; 751 dec->cur += len;
671 752
672 if (ecb_expect_false (dec->stringref))
673 av_push (dec->stringref, newSVpvn (key, len));
674
675 hv_store (hv, key, len, decode_sv (dec), 0); 753 hv_store (hv, key, len, decode_sv (dec), 0);
676 754
677 return; 755 return;
678 } 756 }
679 else if (*dec->cur >= 0x60 && *dec->cur <= 0x60 + 27) 757 else if (ecb_expect_true ((U8)(*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8))
680 { 758 {
681 I32 len = decode_uint (dec); 759 I32 len = decode_uint (dec);
682 char *key = (char *)dec->cur; 760 char *key = (char *)dec->cur;
683 761
684 dec->cur += len; 762 dec->cur += len;
685 763
686 if (ecb_expect_false (dec->stringref)) 764 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
687 av_push (dec->stringref, newSVpvn_utf8 (key, len, 1)); 765 if (!is_utf8_string (key, len))
766 ERR ("corrupted CBOR data (invalid UTF-8 in map key)");
688 767
689 hv_store (hv, key, -len, decode_sv (dec), 0); 768 hv_store (hv, key, -len, decode_sv (dec), 0);
690 769
691 return; 770 return;
692 } 771 }
… …
694 SV *k = decode_sv (dec); 773 SV *k = decode_sv (dec);
695 SV *v = decode_sv (dec); 774 SV *v = decode_sv (dec);
696 775
697 hv_store_ent (hv, k, v, 0); 776 hv_store_ent (hv, k, v, 0);
698 SvREFCNT_dec (k); 777 SvREFCNT_dec (k);
778
779fail:
780 ;
699} 781}
700 782
701static SV * 783static SV *
702decode_hv (dec_t *dec) 784decode_hv (dec_t *dec)
703{ 785{
704 HV *hv = newHV (); 786 HV *hv = newHV ();
705 787
706 DEC_INC_DEPTH; 788 DEC_INC_DEPTH;
707 789
708 if ((*dec->cur & 31) == 31) 790 if (*dec->cur == (MAJOR_MAP | MINOR_INDEF))
709 { 791 {
710 ++dec->cur; 792 ++dec->cur;
711 793
712 for (;;) 794 for (;;)
713 { 795 {
714 WANT (1); 796 WANT (1);
715 797
716 if (*dec->cur == (0xe0 | 31)) 798 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
717 { 799 {
718 ++dec->cur; 800 ++dec->cur;
719 break; 801 break;
720 } 802 }
721 803
… …
742static SV * 824static SV *
743decode_str (dec_t *dec, int utf8) 825decode_str (dec_t *dec, int utf8)
744{ 826{
745 SV *sv = 0; 827 SV *sv = 0;
746 828
747 if ((*dec->cur & 31) == 31) 829 if ((*dec->cur & MINOR_MASK) == MINOR_INDEF)
748 { 830 {
831 // indefinite length strings
749 ++dec->cur; 832 ++dec->cur;
750 833
834 U8 major = *dec->cur & MAJOR_MISC;
835
751 sv = newSVpvn ("", 0); 836 sv = newSVpvn ("", 0);
752 837
753 // not very fast, and certainly not robust against illegal input
754 for (;;) 838 for (;;)
755 { 839 {
756 WANT (1); 840 WANT (1);
757 841
758 if (*dec->cur == (0xe0 | 31)) 842 if ((*dec->cur - major) > LENGTH_EXT8)
843 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
759 { 844 {
760 ++dec->cur; 845 ++dec->cur;
761 break; 846 break;
762 } 847 }
848 else
849 ERR ("corrupted CBOR data (invalid chunks in indefinite length string)");
763 850
764 sv_catsv (sv, decode_sv (dec)); 851 STRLEN len = decode_uint (dec);
852
853 WANT (len);
854 sv_catpvn (sv, dec->cur, len);
855 dec->cur += len;
765 } 856 }
766 } 857 }
767 else 858 else
768 { 859 {
769 STRLEN len = decode_uint (dec); 860 STRLEN len = decode_uint (dec);
… …
776 && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1)) 867 && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1))
777 av_push (dec->stringref, SvREFCNT_inc_NN (sv)); 868 av_push (dec->stringref, SvREFCNT_inc_NN (sv));
778 } 869 }
779 870
780 if (utf8) 871 if (utf8)
872 {
873 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
874 if (!is_utf8_string (SvPVX (sv), SvCUR (sv)))
875 ERR ("corrupted CBOR data (invalid UTF-8 in text string)");
876
781 SvUTF8_on (sv); 877 SvUTF8_on (sv);
878 }
782 879
783 return sv; 880 return sv;
784 881
785fail: 882fail:
786 SvREFCNT_dec (sv); 883 SvREFCNT_dec (sv);
… …
818 } 915 }
819 break; 916 break;
820 917
821 case CBOR_TAG_STRINGREF: 918 case CBOR_TAG_STRINGREF:
822 { 919 {
823 if ((*dec->cur >> 5) != 0) 920 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
824 ERR ("corrupted CBOR data (stringref index not an unsigned integer)"); 921 ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
825 922
826 UV idx = decode_uint (dec); 923 UV idx = decode_uint (dec);
827 924
828 if (!dec->stringref || (int)idx > AvFILLp (dec->stringref)) 925 if (!dec->stringref || (int)idx > AvFILLp (dec->stringref))
… …
835 case CBOR_TAG_VALUE_SHAREABLE: 932 case CBOR_TAG_VALUE_SHAREABLE:
836 { 933 {
837 if (ecb_expect_false (!dec->shareable)) 934 if (ecb_expect_false (!dec->shareable))
838 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ()); 935 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ());
839 936
937 if (dec->cbor.flags & F_ALLOW_CYCLES)
938 {
840 sv = newSV (0); 939 sv = newSV (0);
841 av_push (dec->shareable, SvREFCNT_inc_NN (sv)); 940 av_push (dec->shareable, SvREFCNT_inc_NN (sv));
842 941
843 SV *osv = decode_sv (dec); 942 SV *osv = decode_sv (dec);
844 sv_setsv (sv, osv); 943 sv_setsv (sv, osv);
845 SvREFCNT_dec_NN (osv); 944 SvREFCNT_dec_NN (osv);
945 }
946 else
947 {
948 av_push (dec->shareable, &PL_sv_undef);
949 int idx = AvFILLp (dec->shareable);
950 sv = decode_sv (dec);
951 av_store (dec->shareable, idx, SvREFCNT_inc_NN (sv));
952 }
846 } 953 }
847 break; 954 break;
848 955
849 case CBOR_TAG_VALUE_SHAREDREF: 956 case CBOR_TAG_VALUE_SHAREDREF:
850 { 957 {
851 if ((*dec->cur >> 5) != 0) 958 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
852 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)"); 959 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
853 960
854 UV idx = decode_uint (dec); 961 UV idx = decode_uint (dec);
855 962
856 if (!dec->shareable || (int)idx > AvFILLp (dec->shareable)) 963 if (!dec->shareable || (int)idx > AvFILLp (dec->shareable))
857 ERR ("corrupted CBOR data (sharedref index out of bounds)"); 964 ERR ("corrupted CBOR data (sharedref index out of bounds)");
858 965
859 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]); 966 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]);
967
968 if (sv == &PL_sv_undef)
969 ERR ("cyclic CBOR data structure found, but allow_cycles is not enabled");
860 } 970 }
861 break; 971 break;
862 972
863 case CBOR_TAG_PERL_OBJECT: 973 case CBOR_TAG_PERL_OBJECT:
864 { 974 {
… …
965static SV * 1075static SV *
966decode_sv (dec_t *dec) 1076decode_sv (dec_t *dec)
967{ 1077{
968 WANT (1); 1078 WANT (1);
969 1079
970 switch (*dec->cur >> 5) 1080 switch (*dec->cur >> MAJOR_SHIFT)
971 { 1081 {
972 case 0: // unsigned int 1082 case MAJOR_POS_INT >> MAJOR_SHIFT: return newSVuv (decode_uint (dec));
973 return newSVuv (decode_uint (dec)); 1083 case MAJOR_NEG_INT >> MAJOR_SHIFT: return newSViv (-1 - (IV)decode_uint (dec));
974 case 1: // negative int 1084 case MAJOR_BYTES >> MAJOR_SHIFT: return decode_str (dec, 0);
975 return newSViv (-1 - (IV)decode_uint (dec)); 1085 case MAJOR_TEXT >> MAJOR_SHIFT: return decode_str (dec, 1);
976 case 2: // octet string 1086 case MAJOR_ARRAY >> MAJOR_SHIFT: return decode_av (dec);
977 return decode_str (dec, 0); 1087 case MAJOR_MAP >> MAJOR_SHIFT: return decode_hv (dec);
978 case 3: // utf-8 string 1088 case MAJOR_TAG >> MAJOR_SHIFT: return decode_tagged (dec);
979 return decode_str (dec, 1); 1089
980 case 4: // array 1090 case MAJOR_MISC >> MAJOR_SHIFT:
981 return decode_av (dec);
982 case 5: // map
983 return decode_hv (dec);
984 case 6: // tag
985 return decode_tagged (dec);
986 case 7: // misc
987 switch (*dec->cur++ & 31) 1091 switch (*dec->cur++ & MINOR_MASK)
988 { 1092 {
989 case 20: 1093 case SIMPLE_FALSE:
990#if CBOR_SLOW 1094#if CBOR_SLOW
991 types_false = get_bool ("Types::Serialiser::false"); 1095 types_false = get_bool ("Types::Serialiser::false");
992#endif 1096#endif
993 return newSVsv (types_false); 1097 return newSVsv (types_false);
994 case 21: 1098 case SIMPLE_TRUE:
995#if CBOR_SLOW 1099#if CBOR_SLOW
996 types_true = get_bool ("Types::Serialiser::true"); 1100 types_true = get_bool ("Types::Serialiser::true");
997#endif 1101#endif
998 return newSVsv (types_true); 1102 return newSVsv (types_true);
999 case 22: 1103 case SIMPLE_NULL:
1000 return newSVsv (&PL_sv_undef); 1104 return newSVsv (&PL_sv_undef);
1001 case 23: 1105 case SIMPLE_UNDEF:
1002#if CBOR_SLOW 1106#if CBOR_SLOW
1003 types_error = get_bool ("Types::Serialiser::error"); 1107 types_error = get_bool ("Types::Serialiser::error");
1004#endif 1108#endif
1005 return newSVsv (types_error); 1109 return newSVsv (types_error);
1006 1110
1007 case 25: 1111 case MISC_FLOAT16:
1008 { 1112 {
1009 WANT (2); 1113 WANT (2);
1010 1114
1011 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1]; 1115 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1];
1012 dec->cur += 2; 1116 dec->cur += 2;
1013 1117
1014 return newSVnv (ecb_binary16_to_float (fp)); 1118 return newSVnv (ecb_binary16_to_float (fp));
1015 } 1119 }
1016 1120
1017 case 26: 1121 case MISC_FLOAT32:
1018 { 1122 {
1019 uint32_t fp; 1123 uint32_t fp;
1020 WANT (4); 1124 WANT (4);
1021 memcpy (&fp, dec->cur, 4); 1125 memcpy (&fp, dec->cur, 4);
1022 dec->cur += 4; 1126 dec->cur += 4;
… …
1025 fp = ecb_bswap32 (fp); 1129 fp = ecb_bswap32 (fp);
1026 1130
1027 return newSVnv (ecb_binary32_to_float (fp)); 1131 return newSVnv (ecb_binary32_to_float (fp));
1028 } 1132 }
1029 1133
1030 case 27: 1134 case MISC_FLOAT64:
1031 { 1135 {
1032 uint64_t fp; 1136 uint64_t fp;
1033 WANT (8); 1137 WANT (8);
1034 memcpy (&fp, dec->cur, 8); 1138 memcpy (&fp, dec->cur, 8);
1035 dec->cur += 8; 1139 dec->cur += 8;
… …
1038 fp = ecb_bswap64 (fp); 1142 fp = ecb_bswap64 (fp);
1039 1143
1040 return newSVnv (ecb_binary64_to_double (fp)); 1144 return newSVnv (ecb_binary64_to_double (fp));
1041 } 1145 }
1042 1146
1043 // 0..19 unassigned 1147 // 0..19 unassigned simple
1044 // 24 reserved + unassigned (reserved values are not encodable) 1148 // 24 reserved + unassigned simple (reserved values are not encodable)
1149 // 28-30 unassigned misc
1150 // 31 break code
1045 default: 1151 default:
1046 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)"); 1152 ERR ("corrupted CBOR data (reserved/unassigned/unexpected major 7 value)");
1047 } 1153 }
1048 1154
1049 break; 1155 break;
1050 } 1156 }
1051 1157
… …
1078 if (dec.cur != dec.end && !dec.err) 1184 if (dec.cur != dec.end && !dec.err)
1079 dec.err = "garbage after CBOR object"; 1185 dec.err = "garbage after CBOR object";
1080 1186
1081 if (dec.err) 1187 if (dec.err)
1082 { 1188 {
1189 if (dec.shareable)
1190 {
1191 // need to break cyclic links, which whould all be in shareable
1192 int i;
1193 SV **svp;
1194
1195 for (i = av_len (dec.shareable) + 1; i--; )
1196 if ((svp = av_fetch (dec.shareable, i, 0)))
1197 sv_setsv (*svp, &PL_sv_undef);
1198 }
1199
1083 SvREFCNT_dec (sv); 1200 SvREFCNT_dec (sv);
1084 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur); 1201 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur);
1085 } 1202 }
1086 1203
1087 sv = sv_2mortal (sv); 1204 sv = sv_2mortal (sv);
1088 1205
1089 return sv; 1206 return sv;
1090} 1207}
1091 1208
1209/////////////////////////////////////////////////////////////////////////////
1210// incremental parser
1211
1212#define INCR_DONE(cbor) (AvFILLp (cbor->incr_count) < 0)
1213
1214// returns 0 for notyet, 1 for success or error
1215static int
1216incr_parse (CBOR *self, SV *cborstr)
1217{
1218 STRLEN cur;
1219 SvPV (cborstr, cur);
1220
1221 while (ecb_expect_true (self->incr_need <= cur))
1222 {
1223 // table of integer count bytes
1224 static I8 incr_len[MINOR_MASK + 1] = {
1225 0, 0, 0, 0, 0, 0, 0, 0,
1226 0, 0, 0, 0, 0, 0, 0, 0,
1227 0, 0, 0, 0, 0, 0, 0, 0,
1228 1, 2, 4, 8,-1,-1,-1,-2
1229 };
1230
1231 const U8 *p = SvPVX (cborstr) + self->incr_pos;
1232 U8 m = *p & MINOR_MASK;
1233 IV count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1234 I8 ilen = incr_len[m];
1235
1236 self->incr_need = self->incr_pos + 1;
1237
1238 if (ecb_expect_false (ilen < 0))
1239 {
1240 if (m != MINOR_INDEF)
1241 return 1; // error
1242
1243 if (*p == (MAJOR_MISC | MINOR_INDEF))
1244 {
1245 if (count >= 0)
1246 return 1; // error
1247
1248 count = 1;
1249 }
1250 else
1251 {
1252 av_push (self->incr_count, newSViv (-1)); //TODO: nest
1253 count = -1;
1254 }
1255 }
1256 else
1257 {
1258 self->incr_need += ilen;
1259 if (ecb_expect_false (self->incr_need > cur))
1260 return 0;
1261
1262 int major = *p >> MAJOR_SHIFT;
1263
1264 switch (major)
1265 {
1266 case MAJOR_BYTES >> MAJOR_SHIFT:
1267 case MAJOR_TEXT >> MAJOR_SHIFT:
1268 case MAJOR_ARRAY >> MAJOR_SHIFT:
1269 case MAJOR_MAP >> MAJOR_SHIFT:
1270 {
1271 UV len;
1272
1273 if (ecb_expect_false (ilen))
1274 {
1275 len = 0;
1276
1277 do {
1278 len = (len << 8) | *++p;
1279 } while (--ilen);
1280 }
1281 else
1282 len = m;
1283
1284 switch (major)
1285 {
1286 case MAJOR_BYTES >> MAJOR_SHIFT:
1287 case MAJOR_TEXT >> MAJOR_SHIFT:
1288 self->incr_need += len;
1289 if (ecb_expect_false (self->incr_need > cur))
1290 return 0;
1291
1292 break;
1293
1294 case MAJOR_MAP >> MAJOR_SHIFT:
1295 len <<= 1;
1296 case MAJOR_ARRAY >> MAJOR_SHIFT:
1297 if (len)
1298 {
1299 av_push (self->incr_count, newSViv (len + 1)); //TODO: nest
1300 count = len + 1;
1301 }
1302 break;
1303 }
1304 }
1305 }
1306 }
1307
1308 self->incr_pos = self->incr_need;
1309
1310 if (count > 0)
1311 {
1312 while (!--count)
1313 {
1314 if (!AvFILLp (self->incr_count))
1315 return 1; // done
1316
1317 SvREFCNT_dec_NN (av_pop (self->incr_count));
1318 count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1319 }
1320
1321 SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]) = count;
1322 }
1323 }
1324
1325 return 0;
1326}
1327
1328
1092///////////////////////////////////////////////////////////////////////////// 1329/////////////////////////////////////////////////////////////////////////////
1093// XS interface functions 1330// XS interface functions
1094 1331
1095MODULE = CBOR::XS PACKAGE = CBOR::XS 1332MODULE = CBOR::XS PACKAGE = CBOR::XS
1096 1333
… …
1136void shrink (CBOR *self, int enable = 1) 1373void shrink (CBOR *self, int enable = 1)
1137 ALIAS: 1374 ALIAS:
1138 shrink = F_SHRINK 1375 shrink = F_SHRINK
1139 allow_unknown = F_ALLOW_UNKNOWN 1376 allow_unknown = F_ALLOW_UNKNOWN
1140 allow_sharing = F_ALLOW_SHARING 1377 allow_sharing = F_ALLOW_SHARING
1141 allow_stringref = F_ALLOW_STRINGREF 1378 allow_cycles = F_ALLOW_CYCLES
1379 pack_strings = F_PACK_STRINGS
1380 validate_utf8 = F_VALIDATE_UTF8
1142 PPCODE: 1381 PPCODE:
1143{ 1382{
1144 if (enable) 1383 if (enable)
1145 self->flags |= ix; 1384 self->flags |= ix;
1146 else 1385 else
… …
1152void get_shrink (CBOR *self) 1391void get_shrink (CBOR *self)
1153 ALIAS: 1392 ALIAS:
1154 get_shrink = F_SHRINK 1393 get_shrink = F_SHRINK
1155 get_allow_unknown = F_ALLOW_UNKNOWN 1394 get_allow_unknown = F_ALLOW_UNKNOWN
1156 get_allow_sharing = F_ALLOW_SHARING 1395 get_allow_sharing = F_ALLOW_SHARING
1157 get_allow_stringref = F_ALLOW_STRINGREF 1396 get_allow_cycles = F_ALLOW_CYCLES
1397 get_pack_strings = F_PACK_STRINGS
1398 get_validate_utf8 = F_VALIDATE_UTF8
1158 PPCODE: 1399 PPCODE:
1159 XPUSHs (boolSV (self->flags & ix)); 1400 XPUSHs (boolSV (self->flags & ix));
1160 1401
1161void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) 1402void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
1162 PPCODE: 1403 PPCODE:
… …
1211 EXTEND (SP, 2); 1452 EXTEND (SP, 2);
1212 PUSHs (sv); 1453 PUSHs (sv);
1213 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); 1454 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
1214} 1455}
1215 1456
1457void incr_parse (CBOR *self, SV *cborstr)
1458 ALIAS:
1459 incr_parse_multiple = 1
1460 PPCODE:
1461{
1462 if (SvUTF8 (cborstr))
1463 sv_utf8_downgrade (cborstr, 0);
1464
1465 if (!self->incr_count)
1466 {
1467 self->incr_count = newAV ();
1468 self->incr_pos = 0;
1469 self->incr_need = 1;
1470
1471 av_push (self->incr_count, newSViv (1));
1472 }
1473
1474 do
1475 {
1476 if (!incr_parse (self, cborstr))
1477 {
1478 if (self->incr_need > self->max_size && self->max_size)
1479 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
1480 (unsigned long)self->incr_need, (unsigned long)self->max_size);
1481
1482 break;
1483 }
1484
1485 SV *sv;
1486 char *offset;
1487
1488 PUTBACK; sv = decode_cbor (cborstr, self, &offset); SPAGAIN;
1489 XPUSHs (sv);
1490
1491 sv_chop (cborstr, offset);
1492
1493 av_clear (self->incr_count);
1494 av_push (self->incr_count, newSViv (1));
1495
1496 self->incr_pos = 0;
1497 self->incr_need = self->incr_pos + 1;
1498 }
1499 while (ix);
1500}
1501
1502void incr_reset (CBOR *self)
1503 CODE:
1504{
1505 SvREFCNT_dec (self->incr_count);
1506 self->incr_count = 0;
1507}
1508
1216void DESTROY (CBOR *self) 1509void DESTROY (CBOR *self)
1217 PPCODE: 1510 PPCODE:
1218 cbor_free (self); 1511 cbor_free (self);
1219 1512
1220PROTOTYPES: ENABLE 1513PROTOTYPES: ENABLE
1221 1514
1222void encode_cbor (SV *scalar) 1515void encode_cbor (SV *scalar)
1516 ALIAS:
1517 encode_cbor = 0
1518 encode_cbor_sharing = F_ALLOW_SHARING
1223 PPCODE: 1519 PPCODE:
1224{ 1520{
1225 CBOR cbor; 1521 CBOR cbor;
1226 cbor_init (&cbor); 1522 cbor_init (&cbor);
1523 cbor.flags |= ix;
1227 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN; 1524 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;
1228 XPUSHs (scalar); 1525 XPUSHs (scalar);
1229} 1526}
1230 1527
1231void decode_cbor (SV *cborstr) 1528void decode_cbor (SV *cborstr)

Diff Legend

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