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 1.48 by root, Sun Sep 20 23:25:53 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))
… …
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 {
338 SV *sv = AvARRAY (av)[i];
339 encode_sv (enc, sv ? sv : &PL_sv_undef);
340 }
279 341
280 --enc->depth; 342 --enc->depth;
281} 343}
282 344
283static void 345static void
… …
292 354
293 int pairs = hv_iterinit (hv); 355 int pairs = hv_iterinit (hv);
294 int mg = SvMAGICAL (hv); 356 int mg = SvMAGICAL (hv);
295 357
296 if (mg) 358 if (mg)
297 encode_ch (enc, 0xa0 | 31); 359 encode_ch (enc, MAJOR_MAP | MINOR_INDEF);
298 else 360 else
299 encode_uint (enc, 0xa0, pairs); 361 encode_uint (enc, MAJOR_MAP, pairs);
300 362
301 while ((he = hv_iternext (hv))) 363 while ((he = hv_iternext (hv)))
302 { 364 {
303 if (HeKLEN (he) == HEf_SVKEY) 365 if (HeKLEN (he) == HEf_SVKEY)
304 encode_sv (enc, HeSVKEY (he)); 366 encode_sv (enc, HeSVKEY (he));
305 else 367 else
306 encode_str (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he)); 368 encode_strref (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he));
307 369
308 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));
309 } 371 }
310 372
311 if (mg) 373 if (mg)
312 encode_ch (enc, 0xe0 | 31); 374 encode_ch (enc, MAJOR_MISC | MINOR_INDEF);
313 375
314 --enc->depth; 376 --enc->depth;
315} 377}
316 378
317// encode objects, arrays and special \0=false and \1=true values. 379// encode objects, arrays and special \0=false and \1=true values.
318static void 380static void
319encode_rv (enc_t *enc, SV *sv) 381encode_rv (enc_t *enc, SV *sv)
320{ 382{
321 SvGETMAGIC (sv); 383 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 }
344 384
345 svtype svt = SvTYPE (sv); 385 svtype svt = SvTYPE (sv);
346 386
347 if (ecb_expect_false (SvOBJECT (sv))) 387 if (ecb_expect_false (SvOBJECT (sv)))
348 { 388 {
… …
355 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 395 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
356 ? cbor_tagged_stash 396 ? cbor_tagged_stash
357 : gv_stashpv ("CBOR::XS::Tagged" , 1); 397 : gv_stashpv ("CBOR::XS::Tagged" , 1);
358 398
359 HV *stash = SvSTASH (sv); 399 HV *stash = SvSTASH (sv);
360 GV *method;
361 400
362 if (stash == boolean_stash) 401 if (stash == boolean_stash)
363 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 }
364 else if (stash == error_stash) 406 else if (stash == error_stash)
365 encode_ch (enc, 0xe0 | 23); 407 {
408 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
409 return;
410 }
366 else if (stash == tagged_stash) 411 else if (stash == tagged_stash)
367 { 412 {
368 if (svt != SVt_PVAV) 413 if (svt != SVt_PVAV)
369 croak ("encountered CBOR::XS::Tagged object that isn't an array"); 414 croak ("encountered CBOR::XS::Tagged object that isn't an array");
370 415
371 encode_uint (enc, 0xc0, SvUV (*av_fetch ((AV *)sv, 0, 1))); 416 encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1)));
372 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))
373 } 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
374 else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0))) 450 if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
375 { 451 {
376 dSP; 452 dSP;
377 453
378 ENTER; SAVETMPS; PUSHMARK (SP); 454 ENTER; SAVETMPS; PUSHMARK (SP);
379 // 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
… …
411 // catch this surprisingly common error 487 // catch this surprisingly common error
412 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv) 488 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)); 489 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));
414 490
415 encode_tag (enc, CBOR_TAG_PERL_OBJECT); 491 encode_tag (enc, CBOR_TAG_PERL_OBJECT);
416 encode_uint (enc, 0x80, count + 1); 492 encode_uint (enc, MAJOR_ARRAY, count + 1);
417 encode_str (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash)); 493 encode_strref (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
418 494
419 while (count) 495 while (count)
420 encode_sv (enc, SP[1 - count--]); 496 encode_sv (enc, SP[1 - count--]);
421 497
422 PUTBACK; 498 PUTBACK;
… …
443{ 519{
444 double nv = SvNVX (sv); 520 double nv = SvNVX (sv);
445 521
446 need (enc, 9); 522 need (enc, 9);
447 523
448 if (ecb_expect_false (nv == (U32)nv)) 524 if (ecb_expect_false (nv == (NV)(U32)nv))
449 encode_uint (enc, 0x00, (U32)nv); 525 encode_uint (enc, MAJOR_POS_INT, (U32)nv);
450 //TODO: maybe I32? 526 //TODO: maybe I32?
451 else if (ecb_expect_false (nv == (float)nv)) 527 else if (ecb_expect_false (nv == (float)nv))
452 { 528 {
453 uint32_t fp = ecb_float_to_binary32 (nv); 529 uint32_t fp = ecb_float_to_binary32 (nv);
454 530
455 *enc->cur++ = 0xe0 | 26; 531 *enc->cur++ = MAJOR_MISC | MISC_FLOAT32;
456 532
457 if (!ecb_big_endian ()) 533 if (!ecb_big_endian ())
458 fp = ecb_bswap32 (fp); 534 fp = ecb_bswap32 (fp);
459 535
460 memcpy (enc->cur, &fp, 4); 536 memcpy (enc->cur, &fp, 4);
… …
462 } 538 }
463 else 539 else
464 { 540 {
465 uint64_t fp = ecb_double_to_binary64 (nv); 541 uint64_t fp = ecb_double_to_binary64 (nv);
466 542
467 *enc->cur++ = 0xe0 | 27; 543 *enc->cur++ = MAJOR_MISC | MISC_FLOAT64;
468 544
469 if (!ecb_big_endian ()) 545 if (!ecb_big_endian ())
470 fp = ecb_bswap64 (fp); 546 fp = ecb_bswap64 (fp);
471 547
472 memcpy (enc->cur, &fp, 8); 548 memcpy (enc->cur, &fp, 8);
… …
481 557
482 if (SvPOKp (sv)) 558 if (SvPOKp (sv))
483 { 559 {
484 STRLEN len; 560 STRLEN len;
485 char *str = SvPV (sv, len); 561 char *str = SvPV (sv, len);
486 encode_str (enc, SvUTF8 (sv), str, len); 562 encode_strref (enc, SvUTF8 (sv), str, len);
487 } 563 }
488 else if (SvNOKp (sv)) 564 else if (SvNOKp (sv))
489 encode_nv (enc, sv); 565 encode_nv (enc, sv);
490 else if (SvIOKp (sv)) 566 else if (SvIOKp (sv))
491 { 567 {
492 if (SvIsUV (sv)) 568 if (SvIsUV (sv))
493 encode_uint (enc, 0x00, SvUVX (sv)); 569 encode_uint (enc, MAJOR_POS_INT, SvUVX (sv));
494 else if (SvIVX (sv) >= 0) 570 else if (SvIVX (sv) >= 0)
495 encode_uint (enc, 0x00, SvIVX (sv)); 571 encode_uint (enc, MAJOR_POS_INT, SvIVX (sv));
496 else 572 else
497 encode_uint (enc, 0x20, -(SvIVX (sv) + 1)); 573 encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1));
498 } 574 }
499 else if (SvROK (sv)) 575 else if (SvROK (sv))
500 encode_rv (enc, SvRV (sv)); 576 encode_rv (enc, SvRV (sv));
501 else if (!SvOK (sv)) 577 else if (!SvOK (sv))
502 encode_ch (enc, 0xe0 | 22); 578 encode_ch (enc, MAJOR_MISC | SIMPLE_NULL);
503 else if (enc->cbor.flags & F_ALLOW_UNKNOWN) 579 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
504 encode_ch (enc, 0xe0 | 23); 580 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
505 else 581 else
506 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",
507 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv)); 583 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
508} 584}
509 585
510static SV * 586static SV *
511encode_cbor (SV *scalar, CBOR *cbor) 587encode_cbor (SV *scalar, CBOR *cbor)
512{ 588{
513 enc_t enc = { }; 589 enc_t enc = { 0 };
514 590
515 enc.cbor = *cbor; 591 enc.cbor = *cbor;
516 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 592 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
517 enc.cur = SvPVX (enc.sv); 593 enc.cur = SvPVX (enc.sv);
518 enc.end = SvEND (enc.sv); 594 enc.end = SvEND (enc.sv);
519 595
520 SvPOK_only (enc.sv); 596 SvPOK_only (enc.sv);
521 597
522 if (cbor->flags & F_ALLOW_STRINGREF) 598 if (cbor->flags & F_PACK_STRINGS)
523 { 599 {
524 encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE); 600 encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE);
525 enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ()); 601 enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ());
526 enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ()); 602 enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ());
527 } 603 }
… …
562#define DEC_DEC_DEPTH --dec->depth 638#define DEC_DEC_DEPTH --dec->depth
563 639
564static UV 640static UV
565decode_uint (dec_t *dec) 641decode_uint (dec_t *dec)
566{ 642{
567 switch (*dec->cur & 31) 643 U8 m = *dec->cur & MINOR_MASK;
568 { 644 ++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;
573 645
574 case 24: 646 if (ecb_expect_true (m < LENGTH_EXT1))
647 return m;
648 else if (ecb_expect_true (m == LENGTH_EXT1))
649 {
575 WANT (2); 650 WANT (1);
576 dec->cur += 2; 651 dec->cur += 1;
577 return dec->cur[-1]; 652 return dec->cur[-1];
578 653 }
579 case 25: 654 else if (ecb_expect_true (m == LENGTH_EXT2))
655 {
580 WANT (3); 656 WANT (2);
581 dec->cur += 3; 657 dec->cur += 2;
582 return (((UV)dec->cur[-2]) << 8) 658 return (((UV)dec->cur[-2]) << 8)
583 | ((UV)dec->cur[-1]); 659 | ((UV)dec->cur[-1]);
584 660 }
585 case 26: 661 else if (ecb_expect_true (m == LENGTH_EXT4))
662 {
586 WANT (5); 663 WANT (4);
587 dec->cur += 5; 664 dec->cur += 4;
588 return (((UV)dec->cur[-4]) << 24) 665 return (((UV)dec->cur[-4]) << 24)
589 | (((UV)dec->cur[-3]) << 16) 666 | (((UV)dec->cur[-3]) << 16)
590 | (((UV)dec->cur[-2]) << 8) 667 | (((UV)dec->cur[-2]) << 8)
591 | ((UV)dec->cur[-1]); 668 | ((UV)dec->cur[-1]);
592 669 }
593 case 27: 670 else if (ecb_expect_true (m == LENGTH_EXT8))
671 {
594 WANT (9); 672 WANT (8);
595 dec->cur += 9; 673 dec->cur += 8;
674
675 return
676#if UVSIZE < 8
677 0
678#else
596 return (((UV)dec->cur[-8]) << 56) 679 (((UV)dec->cur[-8]) << 56)
597 | (((UV)dec->cur[-7]) << 48) 680 | (((UV)dec->cur[-7]) << 48)
598 | (((UV)dec->cur[-6]) << 40) 681 | (((UV)dec->cur[-6]) << 40)
599 | (((UV)dec->cur[-5]) << 32) 682 | (((UV)dec->cur[-5]) << 32)
683#endif
600 | (((UV)dec->cur[-4]) << 24) 684 | (((UV)dec->cur[-4]) << 24)
601 | (((UV)dec->cur[-3]) << 16) 685 | (((UV)dec->cur[-3]) << 16)
602 | (((UV)dec->cur[-2]) << 8) 686 | (((UV)dec->cur[-2]) << 8)
603 | ((UV)dec->cur[-1]); 687 | ((UV)dec->cur[-1]);
604 688 }
605 default: 689 else
606 ERR ("corrupted CBOR data (unsupported integer minor encoding)"); 690 ERR ("corrupted CBOR data (unsupported integer minor encoding)");
607 }
608 691
609fail: 692fail:
610 return 0; 693 return 0;
611} 694}
612 695
… …
617{ 700{
618 AV *av = newAV (); 701 AV *av = newAV ();
619 702
620 DEC_INC_DEPTH; 703 DEC_INC_DEPTH;
621 704
622 if ((*dec->cur & 31) == 31) 705 if (*dec->cur == (MAJOR_ARRAY | MINOR_INDEF))
623 { 706 {
624 ++dec->cur; 707 ++dec->cur;
625 708
626 for (;;) 709 for (;;)
627 { 710 {
628 WANT (1); 711 WANT (1);
629 712
630 if (*dec->cur == (0xe0 | 31)) 713 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
631 { 714 {
632 ++dec->cur; 715 ++dec->cur;
633 break; 716 break;
634 } 717 }
635 718
… …
638 } 721 }
639 else 722 else
640 { 723 {
641 int i, len = decode_uint (dec); 724 int i, len = decode_uint (dec);
642 725
726 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); 727 av_fill (av, len - 1);
644 728
645 for (i = 0; i < len; ++i) 729 for (i = 0; i < len; ++i)
646 AvARRAY (av)[i] = decode_sv (dec); 730 AvARRAY (av)[i] = decode_sv (dec);
647 } 731 }
… …
660{ 744{
661 // for speed reasons, we specialcase single-string 745 // for speed reasons, we specialcase single-string
662 // byte or utf-8 strings as keys, but only when !stringref 746 // byte or utf-8 strings as keys, but only when !stringref
663 747
664 if (ecb_expect_true (!dec->stringref)) 748 if (ecb_expect_true (!dec->stringref))
665 if (*dec->cur >= 0x40 && *dec->cur <= 0x40 + 27) 749 if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8))
666 { 750 {
667 I32 len = decode_uint (dec); 751 I32 len = decode_uint (dec);
668 char *key = (char *)dec->cur; 752 char *key = (char *)dec->cur;
669 753
670 dec->cur += len; 754 dec->cur += len;
671 755
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); 756 hv_store (hv, key, len, decode_sv (dec), 0);
676 757
677 return; 758 return;
678 } 759 }
679 else if (*dec->cur >= 0x60 && *dec->cur <= 0x60 + 27) 760 else if (ecb_expect_true ((U8)(*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8))
680 { 761 {
681 I32 len = decode_uint (dec); 762 I32 len = decode_uint (dec);
682 char *key = (char *)dec->cur; 763 char *key = (char *)dec->cur;
683 764
684 dec->cur += len; 765 dec->cur += len;
685 766
686 if (ecb_expect_false (dec->stringref)) 767 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
687 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)");
688 770
689 hv_store (hv, key, -len, decode_sv (dec), 0); 771 hv_store (hv, key, -len, decode_sv (dec), 0);
690 772
691 return; 773 return;
692 } 774 }
… …
694 SV *k = decode_sv (dec); 776 SV *k = decode_sv (dec);
695 SV *v = decode_sv (dec); 777 SV *v = decode_sv (dec);
696 778
697 hv_store_ent (hv, k, v, 0); 779 hv_store_ent (hv, k, v, 0);
698 SvREFCNT_dec (k); 780 SvREFCNT_dec (k);
781
782fail:
783 ;
699} 784}
700 785
701static SV * 786static SV *
702decode_hv (dec_t *dec) 787decode_hv (dec_t *dec)
703{ 788{
704 HV *hv = newHV (); 789 HV *hv = newHV ();
705 790
706 DEC_INC_DEPTH; 791 DEC_INC_DEPTH;
707 792
708 if ((*dec->cur & 31) == 31) 793 if (*dec->cur == (MAJOR_MAP | MINOR_INDEF))
709 { 794 {
710 ++dec->cur; 795 ++dec->cur;
711 796
712 for (;;) 797 for (;;)
713 { 798 {
714 WANT (1); 799 WANT (1);
715 800
716 if (*dec->cur == (0xe0 | 31)) 801 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
717 { 802 {
718 ++dec->cur; 803 ++dec->cur;
719 break; 804 break;
720 } 805 }
721 806
… …
742static SV * 827static SV *
743decode_str (dec_t *dec, int utf8) 828decode_str (dec_t *dec, int utf8)
744{ 829{
745 SV *sv = 0; 830 SV *sv = 0;
746 831
747 if ((*dec->cur & 31) == 31) 832 if ((*dec->cur & MINOR_MASK) == MINOR_INDEF)
748 { 833 {
834 // indefinite length strings
749 ++dec->cur; 835 ++dec->cur;
750 836
837 U8 major = *dec->cur & MAJOR_MISC;
838
751 sv = newSVpvn ("", 0); 839 sv = newSVpvn ("", 0);
752 840
753 // not very fast, and certainly not robust against illegal input
754 for (;;) 841 for (;;)
755 { 842 {
756 WANT (1); 843 WANT (1);
757 844
758 if (*dec->cur == (0xe0 | 31)) 845 if ((*dec->cur - major) > LENGTH_EXT8)
846 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
759 { 847 {
760 ++dec->cur; 848 ++dec->cur;
761 break; 849 break;
762 } 850 }
851 else
852 ERR ("corrupted CBOR data (invalid chunks in indefinite length string)");
763 853
764 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;
765 } 859 }
766 } 860 }
767 else 861 else
768 { 862 {
769 STRLEN len = decode_uint (dec); 863 STRLEN len = decode_uint (dec);
… …
776 && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1)) 870 && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1))
777 av_push (dec->stringref, SvREFCNT_inc_NN (sv)); 871 av_push (dec->stringref, SvREFCNT_inc_NN (sv));
778 } 872 }
779 873
780 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
781 SvUTF8_on (sv); 880 SvUTF8_on (sv);
881 }
782 882
783 return sv; 883 return sv;
784 884
785fail: 885fail:
786 SvREFCNT_dec (sv); 886 SvREFCNT_dec (sv);
… …
818 } 918 }
819 break; 919 break;
820 920
821 case CBOR_TAG_STRINGREF: 921 case CBOR_TAG_STRINGREF:
822 { 922 {
823 if ((*dec->cur >> 5) != 0) 923 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
824 ERR ("corrupted CBOR data (stringref index not an unsigned integer)"); 924 ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
825 925
826 UV idx = decode_uint (dec); 926 UV idx = decode_uint (dec);
827 927
828 if (!dec->stringref || (int)idx > AvFILLp (dec->stringref)) 928 if (!dec->stringref || (int)idx > AvFILLp (dec->stringref))
… …
835 case CBOR_TAG_VALUE_SHAREABLE: 935 case CBOR_TAG_VALUE_SHAREABLE:
836 { 936 {
837 if (ecb_expect_false (!dec->shareable)) 937 if (ecb_expect_false (!dec->shareable))
838 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ()); 938 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ());
839 939
940 if (dec->cbor.flags & F_ALLOW_CYCLES)
941 {
840 sv = newSV (0); 942 sv = newSV (0);
841 av_push (dec->shareable, SvREFCNT_inc_NN (sv)); 943 av_push (dec->shareable, SvREFCNT_inc_NN (sv));
842 944
843 SV *osv = decode_sv (dec); 945 SV *osv = decode_sv (dec);
844 sv_setsv (sv, osv); 946 sv_setsv (sv, osv);
845 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 }
846 } 956 }
847 break; 957 break;
848 958
849 case CBOR_TAG_VALUE_SHAREDREF: 959 case CBOR_TAG_VALUE_SHAREDREF:
850 { 960 {
851 if ((*dec->cur >> 5) != 0) 961 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
852 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)"); 962 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
853 963
854 UV idx = decode_uint (dec); 964 UV idx = decode_uint (dec);
855 965
856 if (!dec->shareable || (int)idx > AvFILLp (dec->shareable)) 966 if (!dec->shareable || (int)idx > AvFILLp (dec->shareable))
857 ERR ("corrupted CBOR data (sharedref index out of bounds)"); 967 ERR ("corrupted CBOR data (sharedref index out of bounds)");
858 968
859 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");
860 } 973 }
861 break; 974 break;
862 975
863 case CBOR_TAG_PERL_OBJECT: 976 case CBOR_TAG_PERL_OBJECT:
864 { 977 {
… …
965static SV * 1078static SV *
966decode_sv (dec_t *dec) 1079decode_sv (dec_t *dec)
967{ 1080{
968 WANT (1); 1081 WANT (1);
969 1082
970 switch (*dec->cur >> 5) 1083 switch (*dec->cur >> MAJOR_SHIFT)
971 { 1084 {
972 case 0: // unsigned int 1085 case MAJOR_POS_INT >> MAJOR_SHIFT: return newSVuv (decode_uint (dec));
973 return newSVuv (decode_uint (dec)); 1086 case MAJOR_NEG_INT >> MAJOR_SHIFT: return newSViv (-1 - (IV)decode_uint (dec));
974 case 1: // negative int 1087 case MAJOR_BYTES >> MAJOR_SHIFT: return decode_str (dec, 0);
975 return newSViv (-1 - (IV)decode_uint (dec)); 1088 case MAJOR_TEXT >> MAJOR_SHIFT: return decode_str (dec, 1);
976 case 2: // octet string 1089 case MAJOR_ARRAY >> MAJOR_SHIFT: return decode_av (dec);
977 return decode_str (dec, 0); 1090 case MAJOR_MAP >> MAJOR_SHIFT: return decode_hv (dec);
978 case 3: // utf-8 string 1091 case MAJOR_TAG >> MAJOR_SHIFT: return decode_tagged (dec);
979 return decode_str (dec, 1); 1092
980 case 4: // array 1093 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) 1094 switch (*dec->cur++ & MINOR_MASK)
988 { 1095 {
989 case 20: 1096 case SIMPLE_FALSE:
990#if CBOR_SLOW 1097#if CBOR_SLOW
991 types_false = get_bool ("Types::Serialiser::false"); 1098 types_false = get_bool ("Types::Serialiser::false");
992#endif 1099#endif
993 return newSVsv (types_false); 1100 return newSVsv (types_false);
994 case 21: 1101 case SIMPLE_TRUE:
995#if CBOR_SLOW 1102#if CBOR_SLOW
996 types_true = get_bool ("Types::Serialiser::true"); 1103 types_true = get_bool ("Types::Serialiser::true");
997#endif 1104#endif
998 return newSVsv (types_true); 1105 return newSVsv (types_true);
999 case 22: 1106 case SIMPLE_NULL:
1000 return newSVsv (&PL_sv_undef); 1107 return newSVsv (&PL_sv_undef);
1001 case 23: 1108 case SIMPLE_UNDEF:
1002#if CBOR_SLOW 1109#if CBOR_SLOW
1003 types_error = get_bool ("Types::Serialiser::error"); 1110 types_error = get_bool ("Types::Serialiser::error");
1004#endif 1111#endif
1005 return newSVsv (types_error); 1112 return newSVsv (types_error);
1006 1113
1007 case 25: 1114 case MISC_FLOAT16:
1008 { 1115 {
1009 WANT (2); 1116 WANT (2);
1010 1117
1011 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1]; 1118 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1];
1012 dec->cur += 2; 1119 dec->cur += 2;
1013 1120
1014 return newSVnv (ecb_binary16_to_float (fp)); 1121 return newSVnv (ecb_binary16_to_float (fp));
1015 } 1122 }
1016 1123
1017 case 26: 1124 case MISC_FLOAT32:
1018 { 1125 {
1019 uint32_t fp; 1126 uint32_t fp;
1020 WANT (4); 1127 WANT (4);
1021 memcpy (&fp, dec->cur, 4); 1128 memcpy (&fp, dec->cur, 4);
1022 dec->cur += 4; 1129 dec->cur += 4;
… …
1025 fp = ecb_bswap32 (fp); 1132 fp = ecb_bswap32 (fp);
1026 1133
1027 return newSVnv (ecb_binary32_to_float (fp)); 1134 return newSVnv (ecb_binary32_to_float (fp));
1028 } 1135 }
1029 1136
1030 case 27: 1137 case MISC_FLOAT64:
1031 { 1138 {
1032 uint64_t fp; 1139 uint64_t fp;
1033 WANT (8); 1140 WANT (8);
1034 memcpy (&fp, dec->cur, 8); 1141 memcpy (&fp, dec->cur, 8);
1035 dec->cur += 8; 1142 dec->cur += 8;
… …
1038 fp = ecb_bswap64 (fp); 1145 fp = ecb_bswap64 (fp);
1039 1146
1040 return newSVnv (ecb_binary64_to_double (fp)); 1147 return newSVnv (ecb_binary64_to_double (fp));
1041 } 1148 }
1042 1149
1043 // 0..19 unassigned 1150 // 0..19 unassigned simple
1044 // 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
1045 default: 1154 default:
1046 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)"); 1155 ERR ("corrupted CBOR data (reserved/unassigned/unexpected major 7 value)");
1047 } 1156 }
1048 1157
1049 break; 1158 break;
1050 } 1159 }
1051 1160
… …
1054} 1163}
1055 1164
1056static SV * 1165static SV *
1057decode_cbor (SV *string, CBOR *cbor, char **offset_return) 1166decode_cbor (SV *string, CBOR *cbor, char **offset_return)
1058{ 1167{
1059 dec_t dec = { }; 1168 dec_t dec = { 0 };
1060 SV *sv; 1169 SV *sv;
1061 STRLEN len; 1170 STRLEN len;
1062 char *data = SvPVbyte (string, len); 1171 char *data = SvPVbyte (string, len);
1063 1172
1064 if (len > cbor->max_size && cbor->max_size) 1173 if (len > cbor->max_size && cbor->max_size)
… …
1078 if (dec.cur != dec.end && !dec.err) 1187 if (dec.cur != dec.end && !dec.err)
1079 dec.err = "garbage after CBOR object"; 1188 dec.err = "garbage after CBOR object";
1080 1189
1081 if (dec.err) 1190 if (dec.err)
1082 { 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
1083 SvREFCNT_dec (sv); 1203 SvREFCNT_dec (sv);
1084 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);
1085 } 1205 }
1086 1206
1087 sv = sv_2mortal (sv); 1207 sv = sv_2mortal (sv);
1088 1208
1089 return sv; 1209 return sv;
1090} 1210}
1091 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
1092///////////////////////////////////////////////////////////////////////////// 1336/////////////////////////////////////////////////////////////////////////////
1093// XS interface functions 1337// XS interface functions
1094 1338
1095MODULE = CBOR::XS PACKAGE = CBOR::XS 1339MODULE = CBOR::XS PACKAGE = CBOR::XS
1096 1340
… …
1136void shrink (CBOR *self, int enable = 1) 1380void shrink (CBOR *self, int enable = 1)
1137 ALIAS: 1381 ALIAS:
1138 shrink = F_SHRINK 1382 shrink = F_SHRINK
1139 allow_unknown = F_ALLOW_UNKNOWN 1383 allow_unknown = F_ALLOW_UNKNOWN
1140 allow_sharing = F_ALLOW_SHARING 1384 allow_sharing = F_ALLOW_SHARING
1141 allow_stringref = F_ALLOW_STRINGREF 1385 allow_cycles = F_ALLOW_CYCLES
1386 pack_strings = F_PACK_STRINGS
1387 validate_utf8 = F_VALIDATE_UTF8
1142 PPCODE: 1388 PPCODE:
1143{ 1389{
1144 if (enable) 1390 if (enable)
1145 self->flags |= ix; 1391 self->flags |= ix;
1146 else 1392 else
… …
1152void get_shrink (CBOR *self) 1398void get_shrink (CBOR *self)
1153 ALIAS: 1399 ALIAS:
1154 get_shrink = F_SHRINK 1400 get_shrink = F_SHRINK
1155 get_allow_unknown = F_ALLOW_UNKNOWN 1401 get_allow_unknown = F_ALLOW_UNKNOWN
1156 get_allow_sharing = F_ALLOW_SHARING 1402 get_allow_sharing = F_ALLOW_SHARING
1157 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
1158 PPCODE: 1406 PPCODE:
1159 XPUSHs (boolSV (self->flags & ix)); 1407 XPUSHs (boolSV (self->flags & ix));
1160 1408
1161void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) 1409void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
1162 PPCODE: 1410 PPCODE:
… …
1211 EXTEND (SP, 2); 1459 EXTEND (SP, 2);
1212 PUSHs (sv); 1460 PUSHs (sv);
1213 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); 1461 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
1214} 1462}
1215 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
1216void DESTROY (CBOR *self) 1516void DESTROY (CBOR *self)
1217 PPCODE: 1517 PPCODE:
1218 cbor_free (self); 1518 cbor_free (self);
1219 1519
1220PROTOTYPES: ENABLE 1520PROTOTYPES: ENABLE
1221 1521
1222void encode_cbor (SV *scalar) 1522void encode_cbor (SV *scalar)
1523 ALIAS:
1524 encode_cbor = 0
1525 encode_cbor_sharing = F_ALLOW_SHARING
1223 PPCODE: 1526 PPCODE:
1224{ 1527{
1225 CBOR cbor; 1528 CBOR cbor;
1226 cbor_init (&cbor); 1529 cbor_init (&cbor);
1530 cbor.flags |= ix;
1227 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN; 1531 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;
1228 XPUSHs (scalar); 1532 XPUSHs (scalar);
1229} 1533}
1230 1534
1231void decode_cbor (SV *cborstr) 1535void decode_cbor (SV *cborstr)

Diff Legend

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