ViewVC Help
View File | Revision Log | Show Annotations | Download File
/cvs/cvsroot/CBOR-XS/XS.xs
(Generate patch)

Comparing cvsroot/CBOR-XS/XS.xs (file contents):
Revision 1.26 by root, Fri Nov 22 09:40:13 2013 UTC vs.
Revision 1.44 by root, Tue Feb 18 22:12:12 2014 UTC

7#include <stdlib.h> 7#include <stdlib.h>
8#include <stdio.h> 8#include <stdio.h>
9#include <limits.h> 9#include <limits.h>
10#include <float.h> 10#include <float.h>
11 11
12#define ECB_NO_THREADS 1
12#include "ecb.h" 13#include "ecb.h"
13 14
14// compatibility with perl <5.18 15// compatibility with perl <5.18
15#ifndef HvNAMELEN_get 16#ifndef HvNAMELEN_get
16# define HvNAMELEN_get(hv) strlen (HvNAME (hv)) 17# define HvNAMELEN_get(hv) strlen (HvNAME (hv))
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)
79# define CBOR_SLOW 0 122# define CBOR_SLOW 0
80# define CBOR_STASH cbor_stash 123# define CBOR_STASH cbor_stash
81#endif 124#endif
82 125
83static HV *cbor_stash, *types_boolean_stash, *types_error_stash, *cbor_tagged_stash; // CBOR::XS:: 126static HV *cbor_stash, *types_boolean_stash, *types_error_stash, *cbor_tagged_stash; // CBOR::XS::
84static SV *types_true, *types_false, *types_error, *sv_cbor; 127static SV *types_true, *types_false, *types_error, *sv_cbor, *default_filter;
85 128
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;
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.
90} CBOR; 139} CBOR;
91 140
92ecb_inline void 141ecb_inline void
93cbor_init (CBOR *cbor) 142cbor_init (CBOR *cbor)
94{ 143{
95 Zero (cbor, 1, CBOR); 144 Zero (cbor, 1, CBOR);
96 cbor->max_depth = 512; 145 cbor->max_depth = 512;
146}
147
148ecb_inline void
149cbor_free (CBOR *cbor)
150{
151 SvREFCNT_dec (cbor->filter);
152 SvREFCNT_dec (cbor->incr_count);
97} 153}
98 154
99///////////////////////////////////////////////////////////////////////////// 155/////////////////////////////////////////////////////////////////////////////
100// utility functions 156// utility functions
101 157
131{ 187{
132 return idx > 23 188 return idx > 23
133 ? idx > 0xffU 189 ? idx > 0xffU
134 ? idx > 0xffffU 190 ? idx > 0xffffU
135 ? idx > 0xffffffffU 191 ? idx > 0xffffffffU
192 ? 11
136 ? 7 193 : 7
137 : 6
138 : 5 194 : 5
139 : 4 195 : 4
140 : 3; 196 : 3;
141} 197}
142 198
179static void 235static void
180encode_uint (enc_t *enc, int major, UV len) 236encode_uint (enc_t *enc, int major, UV len)
181{ 237{
182 need (enc, 9); 238 need (enc, 9);
183 239
184 if (len < 24) 240 if (ecb_expect_true (len < LENGTH_EXT1))
185 *enc->cur++ = major | len; 241 *enc->cur++ = major | len;
186 else if (len <= 0xff) 242 else if (ecb_expect_true (len <= 0xffU))
187 { 243 {
188 *enc->cur++ = major | 24; 244 *enc->cur++ = major | LENGTH_EXT1;
189 *enc->cur++ = len; 245 *enc->cur++ = len;
190 } 246 }
191 else if (len <= 0xffff) 247 else if (len <= 0xffffU)
192 { 248 {
193 *enc->cur++ = major | 25; 249 *enc->cur++ = major | LENGTH_EXT2;
194 *enc->cur++ = len >> 8; 250 *enc->cur++ = len >> 8;
195 *enc->cur++ = len; 251 *enc->cur++ = len;
196 } 252 }
197 else if (len <= 0xffffffff) 253 else if (len <= 0xffffffffU)
198 { 254 {
199 *enc->cur++ = major | 26; 255 *enc->cur++ = major | LENGTH_EXT4;
200 *enc->cur++ = len >> 24; 256 *enc->cur++ = len >> 24;
201 *enc->cur++ = len >> 16; 257 *enc->cur++ = len >> 16;
202 *enc->cur++ = len >> 8; 258 *enc->cur++ = len >> 8;
203 *enc->cur++ = len; 259 *enc->cur++ = len;
204 } 260 }
205 else 261 else
206 { 262 {
207 *enc->cur++ = major | 27; 263 *enc->cur++ = major | LENGTH_EXT8;
208 *enc->cur++ = len >> 56; 264 *enc->cur++ = len >> 56;
209 *enc->cur++ = len >> 48; 265 *enc->cur++ = len >> 48;
210 *enc->cur++ = len >> 40; 266 *enc->cur++ = len >> 40;
211 *enc->cur++ = len >> 32; 267 *enc->cur++ = len >> 32;
212 *enc->cur++ = len >> 24; 268 *enc->cur++ = len >> 24;
217} 273}
218 274
219ecb_inline void 275ecb_inline void
220encode_tag (enc_t *enc, UV tag) 276encode_tag (enc_t *enc, UV tag)
221{ 277{
222 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;
223} 288}
224 289
225static void 290static void
226encode_str (enc_t *enc, int utf8, char *str, STRLEN len) 291encode_strref (enc_t *enc, int utf8, char *str, STRLEN len)
227{ 292{
228 if (ecb_expect_false (enc->cbor.flags & F_ALLOW_STRINGREF)) 293 if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS))
229 { 294 {
230 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1); 295 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1);
231 296
232 if (SvOK (*svp)) 297 if (SvOK (*svp))
233 { 298 {
234 // already registered, use stringref 299 // already registered, use stringref
235 encode_tag (enc, CBOR_TAG_STRINGREF); 300 encode_tag (enc, CBOR_TAG_STRINGREF);
236 encode_uint (enc, 0x00, SvUV (*svp)); 301 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
237 return; 302 return;
238 } 303 }
239 else if (len >= minimum_string_length (enc->stringref_idx)) 304 else if (len >= minimum_string_length (enc->stringref_idx))
240 { 305 {
241 // register only 306 // register only
242 sv_setuv (*svp, enc->stringref_idx); 307 sv_setuv (*svp, enc->stringref_idx);
243 ++enc->stringref_idx; 308 ++enc->stringref_idx;
244 } 309 }
245 } 310 }
246 311
247 encode_uint (enc, utf8 ? 0x60 : 0x40, len); 312 encode_str (enc, utf8, str, len);
248 need (enc, len);
249 memcpy (enc->cur, str, len);
250 enc->cur += len;
251} 313}
252 314
253static void encode_sv (enc_t *enc, SV *sv); 315static void encode_sv (enc_t *enc, SV *sv);
254 316
255static void 317static void
260 if (enc->depth >= enc->cbor.max_depth) 322 if (enc->depth >= enc->cbor.max_depth)
261 croak (ERR_NESTING_EXCEEDED); 323 croak (ERR_NESTING_EXCEEDED);
262 324
263 ++enc->depth; 325 ++enc->depth;
264 326
265 encode_uint (enc, 0x80, len + 1); 327 encode_uint (enc, MAJOR_ARRAY, len + 1);
266 328
267 for (i = 0; i <= len; ++i) 329 for (i = 0; i <= len; ++i)
268 { 330 {
269 SV **svp = av_fetch (av, i, 0); 331 SV **svp = av_fetch (av, i, 0);
270 encode_sv (enc, svp ? *svp : &PL_sv_undef); 332 encode_sv (enc, svp ? *svp : &PL_sv_undef);
285 347
286 int pairs = hv_iterinit (hv); 348 int pairs = hv_iterinit (hv);
287 int mg = SvMAGICAL (hv); 349 int mg = SvMAGICAL (hv);
288 350
289 if (mg) 351 if (mg)
290 encode_ch (enc, 0xa0 | 31); 352 encode_ch (enc, MAJOR_MAP | MINOR_INDEF);
291 else 353 else
292 encode_uint (enc, 0xa0, pairs); 354 encode_uint (enc, MAJOR_MAP, pairs);
293 355
294 while ((he = hv_iternext (hv))) 356 while ((he = hv_iternext (hv)))
295 { 357 {
296 if (HeKLEN (he) == HEf_SVKEY) 358 if (HeKLEN (he) == HEf_SVKEY)
297 encode_sv (enc, HeSVKEY (he)); 359 encode_sv (enc, HeSVKEY (he));
298 else 360 else
299 encode_str (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he)); 361 encode_strref (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he));
300 362
301 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he)); 363 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he));
302 } 364 }
303 365
304 if (mg) 366 if (mg)
305 encode_ch (enc, 0xe0 | 31); 367 encode_ch (enc, MAJOR_MISC | MINOR_INDEF);
306 368
307 --enc->depth; 369 --enc->depth;
308} 370}
309 371
310// encode objects, arrays and special \0=false and \1=true values. 372// encode objects, arrays and special \0=false and \1=true values.
311static void 373static void
312encode_rv (enc_t *enc, SV *sv) 374encode_rv (enc_t *enc, SV *sv)
313{ 375{
314 SvGETMAGIC (sv); 376 SvGETMAGIC (sv);
315
316 if (ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING)
317 && ecb_expect_false (SvREFCNT (sv) > 1))
318 {
319 if (!enc->shareable)
320 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
321
322 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
323
324 if (SvOK (*svp))
325 {
326 encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF);
327 encode_uint (enc, 0x00, SvUV (*svp));
328 return;
329 }
330 else
331 {
332 sv_setuv (*svp, enc->shareable_idx);
333 ++enc->shareable_idx;
334 encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE);
335 }
336 }
337 377
338 svtype svt = SvTYPE (sv); 378 svtype svt = SvTYPE (sv);
339 379
340 if (ecb_expect_false (SvOBJECT (sv))) 380 if (ecb_expect_false (SvOBJECT (sv)))
341 { 381 {
348 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 388 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
349 ? cbor_tagged_stash 389 ? cbor_tagged_stash
350 : gv_stashpv ("CBOR::XS::Tagged" , 1); 390 : gv_stashpv ("CBOR::XS::Tagged" , 1);
351 391
352 HV *stash = SvSTASH (sv); 392 HV *stash = SvSTASH (sv);
353 GV *method;
354 393
355 if (stash == boolean_stash) 394 if (stash == boolean_stash)
356 encode_ch (enc, SvIV (sv) ? 0xe0 | 21 : 0xe0 | 20); 395 {
396 encode_ch (enc, SvIV (sv) ? MAJOR_MISC | SIMPLE_TRUE : MAJOR_MISC | SIMPLE_FALSE);
397 return;
398 }
357 else if (stash == error_stash) 399 else if (stash == error_stash)
358 encode_ch (enc, 0xe0 | 23); 400 {
401 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
402 return;
403 }
359 else if (stash == tagged_stash) 404 else if (stash == tagged_stash)
360 { 405 {
361 if (svt != SVt_PVAV) 406 if (svt != SVt_PVAV)
362 croak ("encountered CBOR::XS::Tagged object that isn't an array"); 407 croak ("encountered CBOR::XS::Tagged object that isn't an array");
363 408
364 encode_uint (enc, 0xc0, SvUV (*av_fetch ((AV *)sv, 0, 1))); 409 encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1)));
365 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1)); 410 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1));
411
412 return;
413 }
414 }
415
416 if (ecb_expect_false (SvREFCNT (sv) > 1)
417 && ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING))
418 {
419 if (!enc->shareable)
420 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
421
422 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
423
424 if (SvOK (*svp))
366 } 425 {
426 encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF);
427 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
428 return;
429 }
430 else
431 {
432 sv_setuv (*svp, enc->shareable_idx);
433 ++enc->shareable_idx;
434 encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE);
435 }
436 }
437
438 if (ecb_expect_false (SvOBJECT (sv)))
439 {
440 HV *stash = SvSTASH (sv);
441 GV *method;
442
367 else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0))) 443 if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
368 { 444 {
369 dSP; 445 dSP;
370 446
371 ENTER; SAVETMPS; PUSHMARK (SP); 447 ENTER; SAVETMPS; PUSHMARK (SP);
372 // we re-bless the reference to get overload and other niceties right 448 // we re-bless the reference to get overload and other niceties right
404 // catch this surprisingly common error 480 // catch this surprisingly common error
405 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv) 481 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)
406 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash)); 482 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));
407 483
408 encode_tag (enc, CBOR_TAG_PERL_OBJECT); 484 encode_tag (enc, CBOR_TAG_PERL_OBJECT);
409 encode_uint (enc, 0x80, count + 1); 485 encode_uint (enc, MAJOR_ARRAY, count + 1);
410 encode_str (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash)); 486 encode_strref (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
411 487
412 while (count) 488 while (count)
413 encode_sv (enc, SP[1 - count--]); 489 encode_sv (enc, SP[1 - count--]);
414 490
415 PUTBACK; 491 PUTBACK;
436{ 512{
437 double nv = SvNVX (sv); 513 double nv = SvNVX (sv);
438 514
439 need (enc, 9); 515 need (enc, 9);
440 516
441 if (ecb_expect_false (nv == (U32)nv)) 517 if (ecb_expect_false (nv == (NV)(U32)nv))
442 encode_uint (enc, 0x00, (U32)nv); 518 encode_uint (enc, MAJOR_POS_INT, (U32)nv);
443 //TODO: maybe I32? 519 //TODO: maybe I32?
444 else if (ecb_expect_false (nv == (float)nv)) 520 else if (ecb_expect_false (nv == (float)nv))
445 { 521 {
446 uint32_t fp = ecb_float_to_binary32 (nv); 522 uint32_t fp = ecb_float_to_binary32 (nv);
447 523
448 *enc->cur++ = 0xe0 | 26; 524 *enc->cur++ = MAJOR_MISC | MISC_FLOAT32;
449 525
450 if (!ecb_big_endian ()) 526 if (!ecb_big_endian ())
451 fp = ecb_bswap32 (fp); 527 fp = ecb_bswap32 (fp);
452 528
453 memcpy (enc->cur, &fp, 4); 529 memcpy (enc->cur, &fp, 4);
455 } 531 }
456 else 532 else
457 { 533 {
458 uint64_t fp = ecb_double_to_binary64 (nv); 534 uint64_t fp = ecb_double_to_binary64 (nv);
459 535
460 *enc->cur++ = 0xe0 | 27; 536 *enc->cur++ = MAJOR_MISC | MISC_FLOAT64;
461 537
462 if (!ecb_big_endian ()) 538 if (!ecb_big_endian ())
463 fp = ecb_bswap64 (fp); 539 fp = ecb_bswap64 (fp);
464 540
465 memcpy (enc->cur, &fp, 8); 541 memcpy (enc->cur, &fp, 8);
474 550
475 if (SvPOKp (sv)) 551 if (SvPOKp (sv))
476 { 552 {
477 STRLEN len; 553 STRLEN len;
478 char *str = SvPV (sv, len); 554 char *str = SvPV (sv, len);
479 encode_str (enc, SvUTF8 (sv), str, len); 555 encode_strref (enc, SvUTF8 (sv), str, len);
480 } 556 }
481 else if (SvNOKp (sv)) 557 else if (SvNOKp (sv))
482 encode_nv (enc, sv); 558 encode_nv (enc, sv);
483 else if (SvIOKp (sv)) 559 else if (SvIOKp (sv))
484 { 560 {
485 if (SvIsUV (sv)) 561 if (SvIsUV (sv))
486 encode_uint (enc, 0x00, SvUVX (sv)); 562 encode_uint (enc, MAJOR_POS_INT, SvUVX (sv));
487 else if (SvIVX (sv) >= 0) 563 else if (SvIVX (sv) >= 0)
488 encode_uint (enc, 0x00, SvIVX (sv)); 564 encode_uint (enc, MAJOR_POS_INT, SvIVX (sv));
489 else 565 else
490 encode_uint (enc, 0x20, -(SvIVX (sv) + 1)); 566 encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1));
491 } 567 }
492 else if (SvROK (sv)) 568 else if (SvROK (sv))
493 encode_rv (enc, SvRV (sv)); 569 encode_rv (enc, SvRV (sv));
494 else if (!SvOK (sv)) 570 else if (!SvOK (sv))
495 encode_ch (enc, 0xe0 | 22); 571 encode_ch (enc, MAJOR_MISC | SIMPLE_NULL);
496 else if (enc->cbor.flags & F_ALLOW_UNKNOWN) 572 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
497 encode_ch (enc, 0xe0 | 23); 573 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
498 else 574 else
499 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data", 575 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data",
500 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv)); 576 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
501} 577}
502 578
510 enc.cur = SvPVX (enc.sv); 586 enc.cur = SvPVX (enc.sv);
511 enc.end = SvEND (enc.sv); 587 enc.end = SvEND (enc.sv);
512 588
513 SvPOK_only (enc.sv); 589 SvPOK_only (enc.sv);
514 590
515 if (cbor->flags & F_ALLOW_STRINGREF) 591 if (cbor->flags & F_PACK_STRINGS)
516 { 592 {
517 encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE); 593 encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE);
518 enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ()); 594 enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ());
519 enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ()); 595 enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ());
520 } 596 }
542 CBOR cbor; 618 CBOR cbor;
543 U32 depth; // recursion depth 619 U32 depth; // recursion depth
544 U32 maxdepth; // recursion depth limit 620 U32 maxdepth; // recursion depth limit
545 AV *shareable; 621 AV *shareable;
546 AV *stringref; 622 AV *stringref;
623 SV *decode_tagged;
547} dec_t; 624} dec_t;
548 625
549#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE 626#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE
550 627
551#define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data") 628#define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data")
554#define DEC_DEC_DEPTH --dec->depth 631#define DEC_DEC_DEPTH --dec->depth
555 632
556static UV 633static UV
557decode_uint (dec_t *dec) 634decode_uint (dec_t *dec)
558{ 635{
559 switch (*dec->cur & 31) 636 U8 m = *dec->cur & MINOR_MASK;
560 { 637 ++dec->cur;
561 case 0: case 1: case 2: case 3: case 4: case 5: case 6: case 7:
562 case 8: case 9: case 10: case 11: case 12: case 13: case 14: case 15:
563 case 16: case 17: case 18: case 19: case 20: case 21: case 22: case 23:
564 return *dec->cur++ & 31;
565 638
566 case 24: 639 if (ecb_expect_true (m < LENGTH_EXT1))
640 return m;
641 else if (ecb_expect_true (m == LENGTH_EXT1))
642 {
567 WANT (2); 643 WANT (1);
568 dec->cur += 2; 644 dec->cur += 1;
569 return dec->cur[-1]; 645 return dec->cur[-1];
570 646 }
571 case 25: 647 else if (ecb_expect_true (m == LENGTH_EXT2))
648 {
572 WANT (3); 649 WANT (2);
573 dec->cur += 3; 650 dec->cur += 2;
574 return (((UV)dec->cur[-2]) << 8) 651 return (((UV)dec->cur[-2]) << 8)
575 | ((UV)dec->cur[-1]); 652 | ((UV)dec->cur[-1]);
576 653 }
577 case 26: 654 else if (ecb_expect_true (m == LENGTH_EXT4))
655 {
578 WANT (5); 656 WANT (4);
579 dec->cur += 5; 657 dec->cur += 4;
580 return (((UV)dec->cur[-4]) << 24) 658 return (((UV)dec->cur[-4]) << 24)
581 | (((UV)dec->cur[-3]) << 16) 659 | (((UV)dec->cur[-3]) << 16)
582 | (((UV)dec->cur[-2]) << 8) 660 | (((UV)dec->cur[-2]) << 8)
583 | ((UV)dec->cur[-1]); 661 | ((UV)dec->cur[-1]);
584 662 }
585 case 27: 663 else if (ecb_expect_true (m == LENGTH_EXT8))
664 {
586 WANT (9); 665 WANT (8);
587 dec->cur += 9; 666 dec->cur += 8;
667
668 return
669#if UVSIZE < 8
670 0
671#else
588 return (((UV)dec->cur[-8]) << 56) 672 (((UV)dec->cur[-8]) << 56)
589 | (((UV)dec->cur[-7]) << 48) 673 | (((UV)dec->cur[-7]) << 48)
590 | (((UV)dec->cur[-6]) << 40) 674 | (((UV)dec->cur[-6]) << 40)
591 | (((UV)dec->cur[-5]) << 32) 675 | (((UV)dec->cur[-5]) << 32)
676#endif
592 | (((UV)dec->cur[-4]) << 24) 677 | (((UV)dec->cur[-4]) << 24)
593 | (((UV)dec->cur[-3]) << 16) 678 | (((UV)dec->cur[-3]) << 16)
594 | (((UV)dec->cur[-2]) << 8) 679 | (((UV)dec->cur[-2]) << 8)
595 | ((UV)dec->cur[-1]); 680 | ((UV)dec->cur[-1]);
596 681 }
597 default: 682 else
598 ERR ("corrupted CBOR data (unsupported integer minor encoding)"); 683 ERR ("corrupted CBOR data (unsupported integer minor encoding)");
599 }
600 684
601fail: 685fail:
602 return 0; 686 return 0;
603} 687}
604 688
609{ 693{
610 AV *av = newAV (); 694 AV *av = newAV ();
611 695
612 DEC_INC_DEPTH; 696 DEC_INC_DEPTH;
613 697
614 if ((*dec->cur & 31) == 31) 698 if (*dec->cur == (MAJOR_ARRAY | MINOR_INDEF))
615 { 699 {
616 ++dec->cur; 700 ++dec->cur;
617 701
618 for (;;) 702 for (;;)
619 { 703 {
620 WANT (1); 704 WANT (1);
621 705
622 if (*dec->cur == (0xe0 | 31)) 706 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
623 { 707 {
624 ++dec->cur; 708 ++dec->cur;
625 break; 709 break;
626 } 710 }
627 711
630 } 714 }
631 else 715 else
632 { 716 {
633 int i, len = decode_uint (dec); 717 int i, len = decode_uint (dec);
634 718
719 WANT (len); // complexity check for av_fill - need at least one byte per value, do not allow supersize arrays
635 av_fill (av, len - 1); 720 av_fill (av, len - 1);
636 721
637 for (i = 0; i < len; ++i) 722 for (i = 0; i < len; ++i)
638 AvARRAY (av)[i] = decode_sv (dec); 723 AvARRAY (av)[i] = decode_sv (dec);
639 } 724 }
652{ 737{
653 // for speed reasons, we specialcase single-string 738 // for speed reasons, we specialcase single-string
654 // byte or utf-8 strings as keys, but only when !stringref 739 // byte or utf-8 strings as keys, but only when !stringref
655 740
656 if (ecb_expect_true (!dec->stringref)) 741 if (ecb_expect_true (!dec->stringref))
657 if (*dec->cur >= 0x40 && *dec->cur <= 0x40 + 27) 742 if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8))
658 { 743 {
659 I32 len = decode_uint (dec); 744 I32 len = decode_uint (dec);
660 char *key = (char *)dec->cur; 745 char *key = (char *)dec->cur;
661 746
662 dec->cur += len; 747 dec->cur += len;
663 748
664 if (ecb_expect_false (dec->stringref))
665 av_push (dec->stringref, newSVpvn (key, len));
666
667 hv_store (hv, key, len, decode_sv (dec), 0); 749 hv_store (hv, key, len, decode_sv (dec), 0);
668 750
669 return; 751 return;
670 } 752 }
671 else if (*dec->cur >= 0x60 && *dec->cur <= 0x60 + 27) 753 else if (ecb_expect_true ((U8)(*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8))
672 { 754 {
673 I32 len = decode_uint (dec); 755 I32 len = decode_uint (dec);
674 char *key = (char *)dec->cur; 756 char *key = (char *)dec->cur;
675 757
676 dec->cur += len; 758 dec->cur += len;
677 759
678 if (ecb_expect_false (dec->stringref)) 760 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
679 av_push (dec->stringref, newSVpvn_utf8 (key, len, 1)); 761 if (!is_utf8_string (key, len))
762 ERR ("corrupted CBOR data (invalid UTF-8 in map key)");
680 763
681 hv_store (hv, key, -len, decode_sv (dec), 0); 764 hv_store (hv, key, -len, decode_sv (dec), 0);
682 765
683 return; 766 return;
684 } 767 }
686 SV *k = decode_sv (dec); 769 SV *k = decode_sv (dec);
687 SV *v = decode_sv (dec); 770 SV *v = decode_sv (dec);
688 771
689 hv_store_ent (hv, k, v, 0); 772 hv_store_ent (hv, k, v, 0);
690 SvREFCNT_dec (k); 773 SvREFCNT_dec (k);
774
775fail:
776 ;
691} 777}
692 778
693static SV * 779static SV *
694decode_hv (dec_t *dec) 780decode_hv (dec_t *dec)
695{ 781{
696 HV *hv = newHV (); 782 HV *hv = newHV ();
697 783
698 DEC_INC_DEPTH; 784 DEC_INC_DEPTH;
699 785
700 if ((*dec->cur & 31) == 31) 786 if (*dec->cur == (MAJOR_MAP | MINOR_INDEF))
701 { 787 {
702 ++dec->cur; 788 ++dec->cur;
703 789
704 for (;;) 790 for (;;)
705 { 791 {
706 WANT (1); 792 WANT (1);
707 793
708 if (*dec->cur == (0xe0 | 31)) 794 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
709 { 795 {
710 ++dec->cur; 796 ++dec->cur;
711 break; 797 break;
712 } 798 }
713 799
734static SV * 820static SV *
735decode_str (dec_t *dec, int utf8) 821decode_str (dec_t *dec, int utf8)
736{ 822{
737 SV *sv = 0; 823 SV *sv = 0;
738 824
739 if ((*dec->cur & 31) == 31) 825 if ((*dec->cur & MINOR_MASK) == MINOR_INDEF)
740 { 826 {
827 // indefinite length strings
741 ++dec->cur; 828 ++dec->cur;
742 829
830 U8 major = *dec->cur & MAJOR_MISC;
831
743 sv = newSVpvn ("", 0); 832 sv = newSVpvn ("", 0);
744 833
745 // not very fast, and certainly not robust against illegal input
746 for (;;) 834 for (;;)
747 { 835 {
748 WANT (1); 836 WANT (1);
749 837
750 if (*dec->cur == (0xe0 | 31)) 838 if ((*dec->cur - major) > LENGTH_EXT8)
839 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
751 { 840 {
752 ++dec->cur; 841 ++dec->cur;
753 break; 842 break;
754 } 843 }
844 else
845 ERR ("corrupted CBOR data (invalid chunks in indefinite length string)");
755 846
756 sv_catsv (sv, decode_sv (dec)); 847 STRLEN len = decode_uint (dec);
848
849 WANT (len);
850 sv_catpvn (sv, dec->cur, len);
851 dec->cur += len;
757 } 852 }
758 } 853 }
759 else 854 else
760 { 855 {
761 STRLEN len = decode_uint (dec); 856 STRLEN len = decode_uint (dec);
768 && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1)) 863 && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1))
769 av_push (dec->stringref, SvREFCNT_inc_NN (sv)); 864 av_push (dec->stringref, SvREFCNT_inc_NN (sv));
770 } 865 }
771 866
772 if (utf8) 867 if (utf8)
868 {
869 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
870 if (!is_utf8_string (SvPVX (sv), SvCUR (sv)))
871 ERR ("corrupted CBOR data (invalid UTF-8 in text string)");
872
773 SvUTF8_on (sv); 873 SvUTF8_on (sv);
874 }
774 875
775 return sv; 876 return sv;
776 877
777fail: 878fail:
778 SvREFCNT_dec (sv); 879 SvREFCNT_dec (sv);
810 } 911 }
811 break; 912 break;
812 913
813 case CBOR_TAG_STRINGREF: 914 case CBOR_TAG_STRINGREF:
814 { 915 {
815 if ((*dec->cur >> 5) != 0) 916 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
816 ERR ("corrupted CBOR data (stringref index not an unsigned integer)"); 917 ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
817 918
818 UV idx = decode_uint (dec); 919 UV idx = decode_uint (dec);
819 920
820 if (!dec->stringref || (int)idx > AvFILLp (dec->stringref)) 921 if (!dec->stringref || (int)idx > AvFILLp (dec->stringref))
827 case CBOR_TAG_VALUE_SHAREABLE: 928 case CBOR_TAG_VALUE_SHAREABLE:
828 { 929 {
829 if (ecb_expect_false (!dec->shareable)) 930 if (ecb_expect_false (!dec->shareable))
830 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ()); 931 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ());
831 932
933 if (dec->cbor.flags & F_ALLOW_CYCLES)
934 {
832 sv = newSV (0); 935 sv = newSV (0);
833 av_push (dec->shareable, SvREFCNT_inc_NN (sv)); 936 av_push (dec->shareable, SvREFCNT_inc_NN (sv));
834 937
835 SV *osv = decode_sv (dec); 938 SV *osv = decode_sv (dec);
836 sv_setsv (sv, osv); 939 sv_setsv (sv, osv);
837 SvREFCNT_dec_NN (osv); 940 SvREFCNT_dec_NN (osv);
941 }
942 else
943 {
944 av_push (dec->shareable, &PL_sv_undef);
945 int idx = AvFILLp (dec->shareable);
946 sv = decode_sv (dec);
947 av_store (dec->shareable, idx, SvREFCNT_inc_NN (sv));
948 }
838 } 949 }
839 break; 950 break;
840 951
841 case CBOR_TAG_VALUE_SHAREDREF: 952 case CBOR_TAG_VALUE_SHAREDREF:
842 { 953 {
843 if ((*dec->cur >> 5) != 0) 954 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
844 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)"); 955 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
845 956
846 UV idx = decode_uint (dec); 957 UV idx = decode_uint (dec);
847 958
848 if (!dec->shareable || (int)idx > AvFILLp (dec->shareable)) 959 if (!dec->shareable || (int)idx > AvFILLp (dec->shareable))
849 ERR ("corrupted CBOR data (sharedref index out of bounds)"); 960 ERR ("corrupted CBOR data (sharedref index out of bounds)");
850 961
851 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]); 962 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]);
963
964 if (sv == &PL_sv_undef)
965 ERR ("cyclic CBOR data structure found, but allow_cycles is not enabled");
852 } 966 }
853 break; 967 break;
854 968
855 case CBOR_TAG_PERL_OBJECT: 969 case CBOR_TAG_PERL_OBJECT:
856 { 970 {
905 1019
906 default: 1020 default:
907 { 1021 {
908 sv = decode_sv (dec); 1022 sv = decode_sv (dec);
909 1023
1024 dSP;
1025 ENTER; SAVETMPS; PUSHMARK (SP);
1026 EXTEND (SP, 2);
1027 PUSHs (newSVuv (tag));
1028 PUSHs (sv);
1029
1030 PUTBACK;
1031 int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL);
1032 SPAGAIN;
1033
1034 if (SvTRUE (ERRSV))
1035 {
1036 FREETMPS; LEAVE;
1037 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV))));
1038 }
1039
1040 if (count)
1041 {
1042 SvREFCNT_dec (sv);
1043 sv = SvREFCNT_inc (POPs);
1044 }
1045 else
1046 {
910 AV *av = newAV (); 1047 AV *av = newAV ();
911 av_push (av, newSVuv (tag)); 1048 av_push (av, newSVuv (tag));
912 av_push (av, sv); 1049 av_push (av, sv);
913 1050
914 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 1051 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
915 ? cbor_tagged_stash 1052 ? cbor_tagged_stash
916 : gv_stashpv ("CBOR::XS::Tagged" , 1); 1053 : gv_stashpv ("CBOR::XS::Tagged" , 1);
917
918 sv = sv_bless (newRV_noinc ((SV *)av), tagged_stash); 1054 sv = sv_bless (newRV_noinc ((SV *)av), tagged_stash);
1055 }
1056
1057 PUTBACK;
1058
1059 FREETMPS; LEAVE;
919 } 1060 }
920 break; 1061 break;
921 } 1062 }
922 1063
923 return sv; 1064 return sv;
930static SV * 1071static SV *
931decode_sv (dec_t *dec) 1072decode_sv (dec_t *dec)
932{ 1073{
933 WANT (1); 1074 WANT (1);
934 1075
935 switch (*dec->cur >> 5) 1076 switch (*dec->cur >> MAJOR_SHIFT)
936 { 1077 {
937 case 0: // unsigned int 1078 case MAJOR_POS_INT >> MAJOR_SHIFT: return newSVuv (decode_uint (dec));
938 return newSVuv (decode_uint (dec)); 1079 case MAJOR_NEG_INT >> MAJOR_SHIFT: return newSViv (-1 - (IV)decode_uint (dec));
939 case 1: // negative int 1080 case MAJOR_BYTES >> MAJOR_SHIFT: return decode_str (dec, 0);
940 return newSViv (-1 - (IV)decode_uint (dec)); 1081 case MAJOR_TEXT >> MAJOR_SHIFT: return decode_str (dec, 1);
941 case 2: // octet string 1082 case MAJOR_ARRAY >> MAJOR_SHIFT: return decode_av (dec);
942 return decode_str (dec, 0); 1083 case MAJOR_MAP >> MAJOR_SHIFT: return decode_hv (dec);
943 case 3: // utf-8 string 1084 case MAJOR_TAG >> MAJOR_SHIFT: return decode_tagged (dec);
944 return decode_str (dec, 1); 1085
945 case 4: // array 1086 case MAJOR_MISC >> MAJOR_SHIFT:
946 return decode_av (dec);
947 case 5: // map
948 return decode_hv (dec);
949 case 6: // tag
950 return decode_tagged (dec);
951 case 7: // misc
952 switch (*dec->cur++ & 31) 1087 switch (*dec->cur++ & MINOR_MASK)
953 { 1088 {
954 case 20: 1089 case SIMPLE_FALSE:
955#if CBOR_SLOW 1090#if CBOR_SLOW
956 types_false = get_bool ("Types::Serialiser::false"); 1091 types_false = get_bool ("Types::Serialiser::false");
957#endif 1092#endif
958 return newSVsv (types_false); 1093 return newSVsv (types_false);
959 case 21: 1094 case SIMPLE_TRUE:
960#if CBOR_SLOW 1095#if CBOR_SLOW
961 types_true = get_bool ("Types::Serialiser::true"); 1096 types_true = get_bool ("Types::Serialiser::true");
962#endif 1097#endif
963 return newSVsv (types_true); 1098 return newSVsv (types_true);
964 case 22: 1099 case SIMPLE_NULL:
965 return newSVsv (&PL_sv_undef); 1100 return newSVsv (&PL_sv_undef);
966 case 23: 1101 case SIMPLE_UNDEF:
967#if CBOR_SLOW 1102#if CBOR_SLOW
968 types_error = get_bool ("Types::Serialiser::error"); 1103 types_error = get_bool ("Types::Serialiser::error");
969#endif 1104#endif
970 return newSVsv (types_error); 1105 return newSVsv (types_error);
971 1106
972 case 25: 1107 case MISC_FLOAT16:
973 { 1108 {
974 WANT (2); 1109 WANT (2);
975 1110
976 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1]; 1111 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1];
977 dec->cur += 2; 1112 dec->cur += 2;
978 1113
979 return newSVnv (ecb_binary16_to_float (fp)); 1114 return newSVnv (ecb_binary16_to_float (fp));
980 } 1115 }
981 1116
982 case 26: 1117 case MISC_FLOAT32:
983 { 1118 {
984 uint32_t fp; 1119 uint32_t fp;
985 WANT (4); 1120 WANT (4);
986 memcpy (&fp, dec->cur, 4); 1121 memcpy (&fp, dec->cur, 4);
987 dec->cur += 4; 1122 dec->cur += 4;
990 fp = ecb_bswap32 (fp); 1125 fp = ecb_bswap32 (fp);
991 1126
992 return newSVnv (ecb_binary32_to_float (fp)); 1127 return newSVnv (ecb_binary32_to_float (fp));
993 } 1128 }
994 1129
995 case 27: 1130 case MISC_FLOAT64:
996 { 1131 {
997 uint64_t fp; 1132 uint64_t fp;
998 WANT (8); 1133 WANT (8);
999 memcpy (&fp, dec->cur, 8); 1134 memcpy (&fp, dec->cur, 8);
1000 dec->cur += 8; 1135 dec->cur += 8;
1003 fp = ecb_bswap64 (fp); 1138 fp = ecb_bswap64 (fp);
1004 1139
1005 return newSVnv (ecb_binary64_to_double (fp)); 1140 return newSVnv (ecb_binary64_to_double (fp));
1006 } 1141 }
1007 1142
1008 // 0..19 unassigned 1143 // 0..19 unassigned simple
1009 // 24 reserved + unassigned (reserved values are not encodable) 1144 // 24 reserved + unassigned simple (reserved values are not encodable)
1145 // 28-30 unassigned misc
1146 // 31 break code
1010 default: 1147 default:
1011 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)"); 1148 ERR ("corrupted CBOR data (reserved/unassigned/unexpected major 7 value)");
1012 } 1149 }
1013 1150
1014 break; 1151 break;
1015 } 1152 }
1016 1153
1043 if (dec.cur != dec.end && !dec.err) 1180 if (dec.cur != dec.end && !dec.err)
1044 dec.err = "garbage after CBOR object"; 1181 dec.err = "garbage after CBOR object";
1045 1182
1046 if (dec.err) 1183 if (dec.err)
1047 { 1184 {
1185 if (dec.shareable)
1186 {
1187 // need to break cyclic links, which whould all be in shareable
1188 int i;
1189 SV **svp;
1190
1191 for (i = av_len (dec.shareable) + 1; i--; )
1192 if ((svp = av_fetch (dec.shareable, i, 0)))
1193 sv_setsv (*svp, &PL_sv_undef);
1194 }
1195
1048 SvREFCNT_dec (sv); 1196 SvREFCNT_dec (sv);
1049 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur); 1197 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur);
1050 } 1198 }
1051 1199
1052 sv = sv_2mortal (sv); 1200 sv = sv_2mortal (sv);
1053 1201
1054 return sv; 1202 return sv;
1055} 1203}
1056 1204
1205/////////////////////////////////////////////////////////////////////////////
1206// incremental parser
1207
1208#define INCR_DONE(cbor) (AvFILLp (cbor->incr_count) < 0)
1209
1210// returns 0 for notyet, 1 for success or error
1211static int
1212incr_parse (CBOR *self, SV *cborstr)
1213{
1214 STRLEN cur;
1215 SvPV (cborstr, cur);
1216
1217 while (ecb_expect_true (self->incr_need <= cur))
1218 {
1219 // table of integer count bytes
1220 static I8 incr_len[MINOR_MASK + 1] = {
1221 0, 0, 0, 0, 0, 0, 0, 0,
1222 0, 0, 0, 0, 0, 0, 0, 0,
1223 0, 0, 0, 0, 0, 0, 0, 0,
1224 1, 2, 4, 8,-1,-1,-1,-2
1225 };
1226
1227 const U8 *p = SvPVX (cborstr) + self->incr_pos;
1228 U8 m = *p & MINOR_MASK;
1229 IV count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1230 I8 ilen = incr_len[m];
1231
1232 self->incr_need = self->incr_pos + 1;
1233
1234 if (ecb_expect_false (ilen < 0))
1235 {
1236 if (m != MINOR_INDEF)
1237 return 1; // error
1238
1239 if (*p == (MAJOR_MISC | MINOR_INDEF))
1240 {
1241 if (count >= 0)
1242 return 1; // error
1243
1244 count = 1;
1245 }
1246 else
1247 {
1248 av_push (self->incr_count, newSViv (-1)); //TODO: nest
1249 count = -1;
1250 }
1251 }
1252 else
1253 {
1254 self->incr_need += ilen;
1255 if (ecb_expect_false (self->incr_need > cur))
1256 return 0;
1257
1258 int major = *p >> MAJOR_SHIFT;
1259
1260 switch (major)
1261 {
1262 case MAJOR_BYTES >> MAJOR_SHIFT:
1263 case MAJOR_TEXT >> MAJOR_SHIFT:
1264 case MAJOR_ARRAY >> MAJOR_SHIFT:
1265 case MAJOR_MAP >> MAJOR_SHIFT:
1266 {
1267 UV len;
1268
1269 if (ecb_expect_false (ilen))
1270 {
1271 len = 0;
1272
1273 do {
1274 len = (len << 8) | *++p;
1275 } while (--ilen);
1276 }
1277 else
1278 len = m;
1279
1280 switch (major)
1281 {
1282 case MAJOR_BYTES >> MAJOR_SHIFT:
1283 case MAJOR_TEXT >> MAJOR_SHIFT:
1284 self->incr_need += len;
1285 if (ecb_expect_false (self->incr_need > cur))
1286 return 0;
1287
1288 break;
1289
1290 case MAJOR_MAP >> MAJOR_SHIFT:
1291 len <<= 1;
1292 case MAJOR_ARRAY >> MAJOR_SHIFT:
1293 if (len)
1294 {
1295 av_push (self->incr_count, newSViv (len + 1)); //TODO: nest
1296 count = len + 1;
1297 }
1298 break;
1299 }
1300 }
1301 }
1302 }
1303
1304 self->incr_pos = self->incr_need;
1305
1306 if (count > 0)
1307 {
1308 while (!--count)
1309 {
1310 if (!AvFILLp (self->incr_count))
1311 return 1; // done
1312
1313 SvREFCNT_dec_NN (av_pop (self->incr_count));
1314 count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1315 }
1316
1317 SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]) = count;
1318 }
1319 }
1320
1321 return 0;
1322}
1323
1324
1057///////////////////////////////////////////////////////////////////////////// 1325/////////////////////////////////////////////////////////////////////////////
1058// XS interface functions 1326// XS interface functions
1059 1327
1060MODULE = CBOR::XS PACKAGE = CBOR::XS 1328MODULE = CBOR::XS PACKAGE = CBOR::XS
1061 1329
1068 types_error_stash = gv_stashpv ("Types::Serialiser::Error" , 1); 1336 types_error_stash = gv_stashpv ("Types::Serialiser::Error" , 1);
1069 1337
1070 types_true = get_bool ("Types::Serialiser::true" ); 1338 types_true = get_bool ("Types::Serialiser::true" );
1071 types_false = get_bool ("Types::Serialiser::false"); 1339 types_false = get_bool ("Types::Serialiser::false");
1072 types_error = get_bool ("Types::Serialiser::error"); 1340 types_error = get_bool ("Types::Serialiser::error");
1341
1342 default_filter = newSVpv ("CBOR::XS::default_filter", 0);
1073 1343
1074 sv_cbor = newSVpv ("CBOR", 0); 1344 sv_cbor = newSVpv ("CBOR", 0);
1075 SvREADONLY_on (sv_cbor); 1345 SvREADONLY_on (sv_cbor);
1076} 1346}
1077 1347
1099void shrink (CBOR *self, int enable = 1) 1369void shrink (CBOR *self, int enable = 1)
1100 ALIAS: 1370 ALIAS:
1101 shrink = F_SHRINK 1371 shrink = F_SHRINK
1102 allow_unknown = F_ALLOW_UNKNOWN 1372 allow_unknown = F_ALLOW_UNKNOWN
1103 allow_sharing = F_ALLOW_SHARING 1373 allow_sharing = F_ALLOW_SHARING
1104 allow_stringref = F_ALLOW_STRINGREF 1374 allow_cycles = F_ALLOW_CYCLES
1375 pack_strings = F_PACK_STRINGS
1376 validate_utf8 = F_VALIDATE_UTF8
1105 PPCODE: 1377 PPCODE:
1106{ 1378{
1107 if (enable) 1379 if (enable)
1108 self->flags |= ix; 1380 self->flags |= ix;
1109 else 1381 else
1115void get_shrink (CBOR *self) 1387void get_shrink (CBOR *self)
1116 ALIAS: 1388 ALIAS:
1117 get_shrink = F_SHRINK 1389 get_shrink = F_SHRINK
1118 get_allow_unknown = F_ALLOW_UNKNOWN 1390 get_allow_unknown = F_ALLOW_UNKNOWN
1119 get_allow_sharing = F_ALLOW_SHARING 1391 get_allow_sharing = F_ALLOW_SHARING
1120 get_allow_stringref = F_ALLOW_STRINGREF 1392 get_allow_cycles = F_ALLOW_CYCLES
1393 get_pack_strings = F_PACK_STRINGS
1394 get_validate_utf8 = F_VALIDATE_UTF8
1121 PPCODE: 1395 PPCODE:
1122 XPUSHs (boolSV (self->flags & ix)); 1396 XPUSHs (boolSV (self->flags & ix));
1123 1397
1124void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) 1398void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
1125 PPCODE: 1399 PPCODE:
1138 XPUSHs (ST (0)); 1412 XPUSHs (ST (0));
1139 1413
1140int get_max_size (CBOR *self) 1414int get_max_size (CBOR *self)
1141 CODE: 1415 CODE:
1142 RETVAL = self->max_size; 1416 RETVAL = self->max_size;
1417 OUTPUT:
1418 RETVAL
1419
1420void filter (CBOR *self, SV *filter = 0)
1421 PPCODE:
1422 SvREFCNT_dec (self->filter);
1423 self->filter = filter ? newSVsv (filter) : filter;
1424 XPUSHs (ST (0));
1425
1426SV *get_filter (CBOR *self)
1427 CODE:
1428 RETVAL = self->filter ? self->filter : NEWSV (0, 0);
1143 OUTPUT: 1429 OUTPUT:
1144 RETVAL 1430 RETVAL
1145 1431
1146void encode (CBOR *self, SV *scalar) 1432void encode (CBOR *self, SV *scalar)
1147 PPCODE: 1433 PPCODE:
1162 EXTEND (SP, 2); 1448 EXTEND (SP, 2);
1163 PUSHs (sv); 1449 PUSHs (sv);
1164 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); 1450 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
1165} 1451}
1166 1452
1453void incr_parse (CBOR *self, SV *cborstr)
1454 ALIAS:
1455 incr_parse_multiple = 1
1456 PPCODE:
1457{
1458 if (SvUTF8 (cborstr))
1459 sv_utf8_downgrade (cborstr, 0);
1460
1461 if (!self->incr_count)
1462 {
1463 self->incr_count = newAV ();
1464 self->incr_pos = 0;
1465 self->incr_need = 1;
1466
1467 av_push (self->incr_count, newSViv (1));
1468 }
1469
1470 do
1471 {
1472 if (!incr_parse (self, cborstr))
1473 {
1474 if (self->incr_need > self->max_size && self->max_size)
1475 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
1476 (unsigned long)self->incr_need, (unsigned long)self->max_size);
1477
1478 break;
1479 }
1480
1481 SV *sv;
1482 char *offset;
1483
1484 PUTBACK; sv = decode_cbor (cborstr, self, &offset); SPAGAIN;
1485 XPUSHs (sv);
1486
1487 sv_chop (cborstr, offset);
1488
1489 av_clear (self->incr_count);
1490 av_push (self->incr_count, newSViv (1));
1491
1492 self->incr_pos = 0;
1493 self->incr_need = self->incr_pos + 1;
1494 }
1495 while (ix);
1496}
1497
1498void incr_reset (CBOR *self)
1499 CODE:
1500{
1501 SvREFCNT_dec (self->incr_count);
1502 self->incr_count = 0;
1503}
1504
1505void DESTROY (CBOR *self)
1506 PPCODE:
1507 cbor_free (self);
1508
1167PROTOTYPES: ENABLE 1509PROTOTYPES: ENABLE
1168 1510
1169void encode_cbor (SV *scalar) 1511void encode_cbor (SV *scalar)
1512 ALIAS:
1513 encode_cbor = 0
1514 encode_cbor_sharing = F_ALLOW_SHARING
1170 PPCODE: 1515 PPCODE:
1171{ 1516{
1172 CBOR cbor; 1517 CBOR cbor;
1173 cbor_init (&cbor); 1518 cbor_init (&cbor);
1519 cbor.flags |= ix;
1174 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN; 1520 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;
1175 XPUSHs (scalar); 1521 XPUSHs (scalar);
1176} 1522}
1177 1523
1178void decode_cbor (SV *cborstr) 1524void decode_cbor (SV *cborstr)

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines