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 1.17 by root, Tue Oct 29 22:04:52 2013 UTC vs.
Revision 1.42 by root, Tue Dec 10 15:31:40 2013 UTC

19# define HvNAMELEN(hv) HvNAMELEN_get (hv) 19# define HvNAMELEN(hv) HvNAMELEN_get (hv)
20#endif 20#endif
21#ifndef HvNAMEUTF8 21#ifndef HvNAMEUTF8
22# define HvNAMEUTF8(hv) 0 22# define HvNAMEUTF8(hv) 0
23#endif 23#endif
24#ifndef SvREFCNT_dec_NN
25# define SvREFCNT_dec_NN(sv) SvREFCNT_dec (sv)
26#endif
27
28// known major and minor types
29enum cbor_type
30{
31 MAJOR_SHIFT = 5,
32 MINOR_MASK = 0x1f,
33
34 MAJOR_POS_INT = 0 << MAJOR_SHIFT,
35 MAJOR_NEG_INT = 1 << MAJOR_SHIFT,
36 MAJOR_BYTES = 2 << MAJOR_SHIFT,
37 MAJOR_TEXT = 3 << MAJOR_SHIFT,
38 MAJOR_ARRAY = 4 << MAJOR_SHIFT,
39 MAJOR_MAP = 5 << MAJOR_SHIFT,
40 MAJOR_TAG = 6 << MAJOR_SHIFT,
41 MAJOR_MISC = 7 << MAJOR_SHIFT,
42
43 // INT/STRING/ARRAY/MAP subtypes
44 LENGTH_EXT1 = 24,
45 LENGTH_EXT2 = 25,
46 LENGTH_EXT4 = 26,
47 LENGTH_EXT8 = 27,
48
49 // SIMPLE types (effectively MISC subtypes)
50 SIMPLE_FALSE = 20,
51 SIMPLE_TRUE = 21,
52 SIMPLE_NULL = 22,
53 SIMPLE_UNDEF = 23,
54
55 // MISC subtype (unused)
56 MISC_EXT1 = 24,
57 MISC_FLOAT16 = 25,
58 MISC_FLOAT32 = 26,
59 MISC_FLOAT64 = 27,
60
61 // BYTES/TEXT/ARRAY/MAP
62 MINOR_INDEF = 31,
63};
24 64
25// known tags 65// known tags
26enum cbor_tag 66enum cbor_tag
27{ 67{
28 // inofficial extensions (pending iana registration) 68 // extensions
29 CBOR_TAG_PERL_OBJECT = 256, 69 CBOR_TAG_STRINGREF = 25, // http://cbor.schmorp.de/stringref
30 CBOR_TAG_GENERIC_OBJECT = 257, 70 CBOR_TAG_PERL_OBJECT = 26, // http://cbor.schmorp.de/perl-object
71 CBOR_TAG_GENERIC_OBJECT = 27, // http://cbor.schmorp.de/generic-object
72 CBOR_TAG_VALUE_SHAREABLE = 28, // http://cbor.schmorp.de/value-sharing
73 CBOR_TAG_VALUE_SHAREDREF = 29, // http://cbor.schmorp.de/value-sharing
74 CBOR_TAG_STRINGREF_NAMESPACE = 256, // http://cbor.schmorp.de/stringref
75 CBOR_TAG_INDIRECTION = 22098, // http://cbor.schmorp.de/indirection
31 76
32 // rfc7049 77 // rfc7049
33 CBOR_TAG_DATETIME = 0, // rfc4287, utf-8 78 CBOR_TAG_DATETIME = 0, // rfc4287, utf-8
34 CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any 79 CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any
35 CBOR_TAG_POS_BIGNUM = 2, // byte string 80 CBOR_TAG_POS_BIGNUM = 2, // byte string
36 CBOR_TAG_NEG_BIGNUM = 3, // byte string 81 CBOR_TAG_NEG_BIGNUM = 3, // byte string
37 CBOR_TAG_DECIMAL = 4, // decimal fraction, array 82 CBOR_TAG_DECIMAL = 4, // decimal fraction, array
38 CBOR_TAG_BIGFLOAT = 5, // array 83 CBOR_TAG_BIGFLOAT = 5, // array
39 84
40 CBOR_TAG_CONV_B64U = 21, // base64url, any 85 CBOR_TAG_CONV_B64U = 21, // base64url, any
41 CBOR_TAG_CONV_B64 = 22, // base64, any 86 CBOR_TAG_CONV_B64 = 22, // base64, any
42 CBOR_TAG_CONV_HEX = 23, // base16, any 87 CBOR_TAG_CONV_HEX = 23, // base16, any
43 CBOR_TAG_CBOR = 24, // embedded cbor, byte string 88 CBOR_TAG_CBOR = 24, // embedded cbor, byte string
44 89
45 CBOR_TAG_URI = 32, // URI rfc3986, utf-8 90 CBOR_TAG_URI = 32, // URI rfc3986, utf-8
46 CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8 91 CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8
47 CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8 92 CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8
48 CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8 93 CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8
49 CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8 94 CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8
50 95
51 CBOR_TAG_MAGIC = 55799 // self-describe cbor 96 CBOR_TAG_MAGIC = 55799, // self-describe cbor
52}; 97};
53 98
54#define F_SHRINK 0x00000200UL 99#define F_SHRINK 0x00000001UL
55#define F_ALLOW_UNKNOWN 0x00002000UL 100#define F_ALLOW_UNKNOWN 0x00000002UL
101#define F_ALLOW_SHARING 0x00000004UL
102#define F_ALLOW_CYCLES 0x00000008UL
103#define F_PACK_STRINGS 0x00000010UL
104#define F_VALIDATE_UTF8 0x00000020UL
56 105
57#define INIT_SIZE 32 // initial scalar size to be allocated 106#define INIT_SIZE 32 // initial scalar size to be allocated
58 107
59#define SB do { 108#define SB do {
60#define SE } while (0) 109#define SE } while (0)
72# define CBOR_SLOW 0 121# define CBOR_SLOW 0
73# define CBOR_STASH cbor_stash 122# define CBOR_STASH cbor_stash
74#endif 123#endif
75 124
76static HV *cbor_stash, *types_boolean_stash, *types_error_stash, *cbor_tagged_stash; // CBOR::XS:: 125static HV *cbor_stash, *types_boolean_stash, *types_error_stash, *cbor_tagged_stash; // CBOR::XS::
77static SV *types_true, *types_false, *types_error, *sv_cbor; 126static SV *types_true, *types_false, *types_error, *sv_cbor, *default_filter;
78 127
79typedef struct { 128typedef struct {
80 U32 flags; 129 U32 flags;
81 U32 max_depth; 130 U32 max_depth;
82 STRLEN max_size; 131 STRLEN max_size;
132 SV *filter;
133
134 // for the incremental parser
135 STRLEN incr_pos; // the current offset into the text
136 STRLEN incr_need; // minimum bytes needed to decode
137 AV *incr_count; // for every nesting level, the number of outstanding values, or -1 for indef.
83} CBOR; 138} CBOR;
84 139
85ecb_inline void 140ecb_inline void
86cbor_init (CBOR *cbor) 141cbor_init (CBOR *cbor)
87{ 142{
88 Zero (cbor, 1, CBOR); 143 Zero (cbor, 1, CBOR);
89 cbor->max_depth = 512; 144 cbor->max_depth = 512;
145}
146
147ecb_inline void
148cbor_free (CBOR *cbor)
149{
150 SvREFCNT_dec (cbor->filter);
151 SvREFCNT_dec (cbor->incr_count);
90} 152}
91 153
92///////////////////////////////////////////////////////////////////////////// 154/////////////////////////////////////////////////////////////////////////////
93// utility functions 155// utility functions
94 156
116 SvPV_renew (sv, SvCUR (sv) + 1); 178 SvPV_renew (sv, SvCUR (sv) + 1);
117#endif 179#endif
118 } 180 }
119} 181}
120 182
121///////////////////////////////////////////////////////////////////////////// 183// minimum length of a string to be registered for stringref
122// fp hell 184ecb_inline int
123 185minimum_string_length (UV idx)
124//TODO 186{
187 return idx > 23
188 ? idx > 0xffU
189 ? idx > 0xffffU
190 ? idx > 0xffffffffU
191 ? 11
192 : 7
193 : 5
194 : 4
195 : 3;
196}
125 197
126///////////////////////////////////////////////////////////////////////////// 198/////////////////////////////////////////////////////////////////////////////
127// encoder 199// encoder
128 200
129// structure used for encoding CBOR 201// structure used for encoding CBOR
132 char *cur; // SvPVX (sv) + current output position 204 char *cur; // SvPVX (sv) + current output position
133 char *end; // SvEND (sv) 205 char *end; // SvEND (sv)
134 SV *sv; // result scalar 206 SV *sv; // result scalar
135 CBOR cbor; 207 CBOR cbor;
136 U32 depth; // recursion level 208 U32 depth; // recursion level
209 HV *stringref[2]; // string => index, or 0 ([0] = bytes, [1] = utf-8)
210 UV stringref_idx;
211 HV *shareable; // ptr => index, or 0
212 UV shareable_idx;
137} enc_t; 213} enc_t;
138 214
139ecb_inline void 215ecb_inline void
140need (enc_t *enc, STRLEN len) 216need (enc_t *enc, STRLEN len)
141{ 217{
158static void 234static void
159encode_uint (enc_t *enc, int major, UV len) 235encode_uint (enc_t *enc, int major, UV len)
160{ 236{
161 need (enc, 9); 237 need (enc, 9);
162 238
163 if (len < 24) 239 if (ecb_expect_true (len < LENGTH_EXT1))
164 *enc->cur++ = major | len; 240 *enc->cur++ = major | len;
165 else if (len <= 0xff) 241 else if (ecb_expect_true (len <= 0xffU))
166 { 242 {
167 *enc->cur++ = major | 24; 243 *enc->cur++ = major | LENGTH_EXT1;
168 *enc->cur++ = len; 244 *enc->cur++ = len;
169 } 245 }
170 else if (len <= 0xffff) 246 else if (len <= 0xffffU)
171 { 247 {
172 *enc->cur++ = major | 25; 248 *enc->cur++ = major | LENGTH_EXT2;
173 *enc->cur++ = len >> 8; 249 *enc->cur++ = len >> 8;
174 *enc->cur++ = len; 250 *enc->cur++ = len;
175 } 251 }
176 else if (len <= 0xffffffff) 252 else if (len <= 0xffffffffU)
177 { 253 {
178 *enc->cur++ = major | 26; 254 *enc->cur++ = major | LENGTH_EXT4;
179 *enc->cur++ = len >> 24; 255 *enc->cur++ = len >> 24;
180 *enc->cur++ = len >> 16; 256 *enc->cur++ = len >> 16;
181 *enc->cur++ = len >> 8; 257 *enc->cur++ = len >> 8;
182 *enc->cur++ = len; 258 *enc->cur++ = len;
183 } 259 }
184 else 260 else
185 { 261 {
186 *enc->cur++ = major | 27; 262 *enc->cur++ = major | LENGTH_EXT8;
187 *enc->cur++ = len >> 56; 263 *enc->cur++ = len >> 56;
188 *enc->cur++ = len >> 48; 264 *enc->cur++ = len >> 48;
189 *enc->cur++ = len >> 40; 265 *enc->cur++ = len >> 40;
190 *enc->cur++ = len >> 32; 266 *enc->cur++ = len >> 32;
191 *enc->cur++ = len >> 24; 267 *enc->cur++ = len >> 24;
193 *enc->cur++ = len >> 8; 269 *enc->cur++ = len >> 8;
194 *enc->cur++ = len; 270 *enc->cur++ = len;
195 } 271 }
196} 272}
197 273
198static void 274ecb_inline void
275encode_tag (enc_t *enc, UV tag)
276{
277 encode_uint (enc, MAJOR_TAG, tag);
278}
279
280ecb_inline void
199encode_str (enc_t *enc, int utf8, char *str, STRLEN len) 281encode_str (enc_t *enc, int utf8, char *str, STRLEN len)
200{ 282{
201 encode_uint (enc, utf8 ? 0x60 : 0x40, len); 283 encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len);
202 need (enc, len); 284 need (enc, len);
203 memcpy (enc->cur, str, len); 285 memcpy (enc->cur, str, len);
204 enc->cur += len; 286 enc->cur += len;
205} 287}
206 288
289static void
290encode_strref (enc_t *enc, int utf8, char *str, STRLEN len)
291{
292 if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS))
293 {
294 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1);
295
296 if (SvOK (*svp))
297 {
298 // already registered, use stringref
299 encode_tag (enc, CBOR_TAG_STRINGREF);
300 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
301 return;
302 }
303 else if (len >= minimum_string_length (enc->stringref_idx))
304 {
305 // register only
306 sv_setuv (*svp, enc->stringref_idx);
307 ++enc->stringref_idx;
308 }
309 }
310
311 encode_str (enc, utf8, str, len);
312}
313
207static void encode_sv (enc_t *enc, SV *sv); 314static void encode_sv (enc_t *enc, SV *sv);
208 315
209static void 316static void
210encode_av (enc_t *enc, AV *av) 317encode_av (enc_t *enc, AV *av)
211{ 318{
214 if (enc->depth >= enc->cbor.max_depth) 321 if (enc->depth >= enc->cbor.max_depth)
215 croak (ERR_NESTING_EXCEEDED); 322 croak (ERR_NESTING_EXCEEDED);
216 323
217 ++enc->depth; 324 ++enc->depth;
218 325
219 encode_uint (enc, 0x80, len + 1); 326 encode_uint (enc, MAJOR_ARRAY, len + 1);
220 327
221 for (i = 0; i <= len; ++i) 328 for (i = 0; i <= len; ++i)
222 { 329 {
223 SV **svp = av_fetch (av, i, 0); 330 SV **svp = av_fetch (av, i, 0);
224 encode_sv (enc, svp ? *svp : &PL_sv_undef); 331 encode_sv (enc, svp ? *svp : &PL_sv_undef);
239 346
240 int pairs = hv_iterinit (hv); 347 int pairs = hv_iterinit (hv);
241 int mg = SvMAGICAL (hv); 348 int mg = SvMAGICAL (hv);
242 349
243 if (mg) 350 if (mg)
244 encode_ch (enc, 0xa0 | 31); 351 encode_ch (enc, MAJOR_MAP | MINOR_INDEF);
245 else 352 else
246 encode_uint (enc, 0xa0, pairs); 353 encode_uint (enc, MAJOR_MAP, pairs);
247 354
248 while ((he = hv_iternext (hv))) 355 while ((he = hv_iternext (hv)))
249 { 356 {
250 if (HeKLEN (he) == HEf_SVKEY) 357 if (HeKLEN (he) == HEf_SVKEY)
251 encode_sv (enc, HeSVKEY (he)); 358 encode_sv (enc, HeSVKEY (he));
252 else 359 else
253 encode_str (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he)); 360 encode_strref (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he));
254 361
255 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he)); 362 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he));
256 } 363 }
257 364
258 if (mg) 365 if (mg)
259 encode_ch (enc, 0xe0 | 31); 366 encode_ch (enc, MAJOR_MISC | MINOR_INDEF);
260 367
261 --enc->depth; 368 --enc->depth;
262} 369}
263 370
264// encode objects, arrays and special \0=false and \1=true values. 371// encode objects, arrays and special \0=false and \1=true values.
265static void 372static void
266encode_rv (enc_t *enc, SV *sv) 373encode_rv (enc_t *enc, SV *sv)
267{ 374{
268 svtype svt;
269
270 SvGETMAGIC (sv); 375 SvGETMAGIC (sv);
376
271 svt = SvTYPE (sv); 377 svtype svt = SvTYPE (sv);
272 378
273 if (ecb_expect_false (SvOBJECT (sv))) 379 if (ecb_expect_false (SvOBJECT (sv)))
274 { 380 {
275 HV *boolean_stash = !CBOR_SLOW || types_boolean_stash 381 HV *boolean_stash = !CBOR_SLOW || types_boolean_stash
276 ? types_boolean_stash 382 ? types_boolean_stash
281 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 387 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
282 ? cbor_tagged_stash 388 ? cbor_tagged_stash
283 : gv_stashpv ("CBOR::XS::Tagged" , 1); 389 : gv_stashpv ("CBOR::XS::Tagged" , 1);
284 390
285 HV *stash = SvSTASH (sv); 391 HV *stash = SvSTASH (sv);
286 GV *method;
287 392
288 if (stash == boolean_stash) 393 if (stash == boolean_stash)
289 encode_ch (enc, SvIV (sv) ? 0xe0 | 21 : 0xe0 | 20); 394 {
395 encode_ch (enc, SvIV (sv) ? MAJOR_MISC | SIMPLE_TRUE : MAJOR_MISC | SIMPLE_FALSE);
396 return;
397 }
290 else if (stash == error_stash) 398 else if (stash == error_stash)
291 encode_ch (enc, 0xe0 | 23); 399 {
400 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
401 return;
402 }
292 else if (stash == tagged_stash) 403 else if (stash == tagged_stash)
293 { 404 {
294 if (svt != SVt_PVAV) 405 if (svt != SVt_PVAV)
295 croak ("encountered CBOR::XS::Tagged object that isn't an array"); 406 croak ("encountered CBOR::XS::Tagged object that isn't an array");
296 407
297 encode_uint (enc, 0xc0, SvUV (*av_fetch ((AV *)sv, 0, 1))); 408 encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1)));
298 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1)); 409 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1));
410
411 return;
412 }
413 }
414
415 if (ecb_expect_false (SvREFCNT (sv) > 1)
416 && ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING))
417 {
418 if (!enc->shareable)
419 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
420
421 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
422
423 if (SvOK (*svp))
299 } 424 {
425 encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF);
426 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
427 return;
428 }
429 else
430 {
431 sv_setuv (*svp, enc->shareable_idx);
432 ++enc->shareable_idx;
433 encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE);
434 }
435 }
436
437 if (ecb_expect_false (SvOBJECT (sv)))
438 {
439 HV *stash = SvSTASH (sv);
440 GV *method;
441
300 else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0))) 442 if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
301 { 443 {
302 dSP; 444 dSP;
303 445
304 ENTER; SAVETMPS; PUSHMARK (SP); 446 ENTER; SAVETMPS; PUSHMARK (SP);
305 // we re-bless the reference to get overload and other niceties right 447 // we re-bless the reference to get overload and other niceties right
336 478
337 // catch this surprisingly common error 479 // catch this surprisingly common error
338 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv) 480 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)
339 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash)); 481 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));
340 482
341 encode_uint (enc, 0xc0, CBOR_TAG_PERL_OBJECT); 483 encode_tag (enc, CBOR_TAG_PERL_OBJECT);
342 encode_uint (enc, 0x80, count + 1); 484 encode_uint (enc, MAJOR_ARRAY, count + 1);
343 encode_str (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash)); 485 encode_strref (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
344 486
345 while (count) 487 while (count)
346 encode_sv (enc, SP[1 - count--]); 488 encode_sv (enc, SP[1 - count--]);
347 489
348 PUTBACK; 490 PUTBACK;
355 } 497 }
356 else if (svt == SVt_PVHV) 498 else if (svt == SVt_PVHV)
357 encode_hv (enc, (HV *)sv); 499 encode_hv (enc, (HV *)sv);
358 else if (svt == SVt_PVAV) 500 else if (svt == SVt_PVAV)
359 encode_av (enc, (AV *)sv); 501 encode_av (enc, (AV *)sv);
360 else if (svt < SVt_PVAV)
361 {
362 STRLEN len = 0;
363 char *pv = svt ? SvPV (sv, len) : 0;
364
365 if (len == 1 && *pv == '1')
366 encode_ch (enc, 0xe0 | 21);
367 else if (len == 1 && *pv == '0')
368 encode_ch (enc, 0xe0 | 20);
369 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
370 encode_ch (enc, 0xe0 | 23);
371 else
372 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
373 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
374 }
375 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
376 encode_ch (enc, 0xe0 | 23);
377 else 502 else
378 croak ("encountered %s, but CBOR can only represent references to arrays or hashes", 503 {
379 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 504 encode_tag (enc, CBOR_TAG_INDIRECTION);
505 encode_sv (enc, sv);
506 }
380} 507}
381 508
382static void 509static void
383encode_nv (enc_t *enc, SV *sv) 510encode_nv (enc_t *enc, SV *sv)
384{ 511{
385 double nv = SvNVX (sv); 512 double nv = SvNVX (sv);
386 513
387 need (enc, 9); 514 need (enc, 9);
388 515
389 if (ecb_expect_false (nv == (U32)nv)) 516 if (ecb_expect_false (nv == (NV)(U32)nv))
390 encode_uint (enc, 0x00, (U32)nv); 517 encode_uint (enc, MAJOR_POS_INT, (U32)nv);
391 //TODO: maybe I32? 518 //TODO: maybe I32?
392 else if (ecb_expect_false (nv == (float)nv)) 519 else if (ecb_expect_false (nv == (float)nv))
393 { 520 {
394 uint32_t fp = ecb_float_to_binary32 (nv); 521 uint32_t fp = ecb_float_to_binary32 (nv);
395 522
396 *enc->cur++ = 0xe0 | 26; 523 *enc->cur++ = MAJOR_MISC | MISC_FLOAT32;
397 524
398 if (!ecb_big_endian ()) 525 if (!ecb_big_endian ())
399 fp = ecb_bswap32 (fp); 526 fp = ecb_bswap32 (fp);
400 527
401 memcpy (enc->cur, &fp, 4); 528 memcpy (enc->cur, &fp, 4);
403 } 530 }
404 else 531 else
405 { 532 {
406 uint64_t fp = ecb_double_to_binary64 (nv); 533 uint64_t fp = ecb_double_to_binary64 (nv);
407 534
408 *enc->cur++ = 0xe0 | 27; 535 *enc->cur++ = MAJOR_MISC | MISC_FLOAT64;
409 536
410 if (!ecb_big_endian ()) 537 if (!ecb_big_endian ())
411 fp = ecb_bswap64 (fp); 538 fp = ecb_bswap64 (fp);
412 539
413 memcpy (enc->cur, &fp, 8); 540 memcpy (enc->cur, &fp, 8);
422 549
423 if (SvPOKp (sv)) 550 if (SvPOKp (sv))
424 { 551 {
425 STRLEN len; 552 STRLEN len;
426 char *str = SvPV (sv, len); 553 char *str = SvPV (sv, len);
427 encode_str (enc, SvUTF8 (sv), str, len); 554 encode_strref (enc, SvUTF8 (sv), str, len);
428 } 555 }
429 else if (SvNOKp (sv)) 556 else if (SvNOKp (sv))
430 encode_nv (enc, sv); 557 encode_nv (enc, sv);
431 else if (SvIOKp (sv)) 558 else if (SvIOKp (sv))
432 { 559 {
433 if (SvIsUV (sv)) 560 if (SvIsUV (sv))
434 encode_uint (enc, 0x00, SvUVX (sv)); 561 encode_uint (enc, MAJOR_POS_INT, SvUVX (sv));
435 else if (SvIVX (sv) >= 0) 562 else if (SvIVX (sv) >= 0)
436 encode_uint (enc, 0x00, SvIVX (sv)); 563 encode_uint (enc, MAJOR_POS_INT, SvIVX (sv));
437 else 564 else
438 encode_uint (enc, 0x20, -(SvIVX (sv) + 1)); 565 encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1));
439 } 566 }
440 else if (SvROK (sv)) 567 else if (SvROK (sv))
441 encode_rv (enc, SvRV (sv)); 568 encode_rv (enc, SvRV (sv));
442 else if (!SvOK (sv)) 569 else if (!SvOK (sv))
443 encode_ch (enc, 0xe0 | 22); 570 encode_ch (enc, MAJOR_MISC | SIMPLE_NULL);
444 else if (enc->cbor.flags & F_ALLOW_UNKNOWN) 571 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
445 encode_ch (enc, 0xe0 | 23); 572 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
446 else 573 else
447 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data", 574 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data",
448 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv)); 575 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
449} 576}
450 577
451static SV * 578static SV *
452encode_cbor (SV *scalar, CBOR *cbor) 579encode_cbor (SV *scalar, CBOR *cbor)
453{ 580{
454 enc_t enc; 581 enc_t enc = { };
455 582
456 enc.cbor = *cbor; 583 enc.cbor = *cbor;
457 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 584 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
458 enc.cur = SvPVX (enc.sv); 585 enc.cur = SvPVX (enc.sv);
459 enc.end = SvEND (enc.sv); 586 enc.end = SvEND (enc.sv);
460 enc.depth = 0;
461 587
462 SvPOK_only (enc.sv); 588 SvPOK_only (enc.sv);
589
590 if (cbor->flags & F_PACK_STRINGS)
591 {
592 encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE);
593 enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ());
594 enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ());
595 }
596
463 encode_sv (&enc, scalar); 597 encode_sv (&enc, scalar);
464 598
465 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 599 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
466 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings 600 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
467 601
481 U8 *end; // end of input string 615 U8 *end; // end of input string
482 const char *err; // parse error, if != 0 616 const char *err; // parse error, if != 0
483 CBOR cbor; 617 CBOR cbor;
484 U32 depth; // recursion depth 618 U32 depth; // recursion depth
485 U32 maxdepth; // recursion depth limit 619 U32 maxdepth; // recursion depth limit
620 AV *shareable;
621 AV *stringref;
622 SV *decode_tagged;
486} dec_t; 623} dec_t;
487 624
488#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE 625#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE
489 626
490#define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data") 627#define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data")
493#define DEC_DEC_DEPTH --dec->depth 630#define DEC_DEC_DEPTH --dec->depth
494 631
495static UV 632static UV
496decode_uint (dec_t *dec) 633decode_uint (dec_t *dec)
497{ 634{
498 switch (*dec->cur & 31) 635 U8 m = *dec->cur & MINOR_MASK;
499 { 636 ++dec->cur;
500 case 0: case 1: case 2: case 3: case 4: case 5: case 6: case 7:
501 case 8: case 9: case 10: case 11: case 12: case 13: case 14: case 15:
502 case 16: case 17: case 18: case 19: case 20: case 21: case 22: case 23:
503 return *dec->cur++ & 31;
504 637
505 case 24: 638 if (ecb_expect_true (m < LENGTH_EXT1))
639 return m;
640 else if (ecb_expect_true (m == LENGTH_EXT1))
641 {
506 WANT (2); 642 WANT (1);
507 dec->cur += 2; 643 dec->cur += 1;
508 return dec->cur[-1]; 644 return dec->cur[-1];
509 645 }
510 case 25: 646 else if (ecb_expect_true (m == LENGTH_EXT2))
647 {
511 WANT (3); 648 WANT (2);
512 dec->cur += 3; 649 dec->cur += 2;
513 return (((UV)dec->cur[-2]) << 8) 650 return (((UV)dec->cur[-2]) << 8)
514 | ((UV)dec->cur[-1]); 651 | ((UV)dec->cur[-1]);
515 652 }
516 case 26: 653 else if (ecb_expect_true (m == LENGTH_EXT4))
654 {
517 WANT (5); 655 WANT (4);
518 dec->cur += 5; 656 dec->cur += 4;
519 return (((UV)dec->cur[-4]) << 24) 657 return (((UV)dec->cur[-4]) << 24)
520 | (((UV)dec->cur[-3]) << 16) 658 | (((UV)dec->cur[-3]) << 16)
521 | (((UV)dec->cur[-2]) << 8) 659 | (((UV)dec->cur[-2]) << 8)
522 | ((UV)dec->cur[-1]); 660 | ((UV)dec->cur[-1]);
523 661 }
524 case 27: 662 else if (ecb_expect_true (m == LENGTH_EXT8))
663 {
525 WANT (9); 664 WANT (8);
526 dec->cur += 9; 665 dec->cur += 8;
666
667 return
668#if UVSIZE < 8
669 0
670#else
527 return (((UV)dec->cur[-8]) << 56) 671 (((UV)dec->cur[-8]) << 56)
528 | (((UV)dec->cur[-7]) << 48) 672 | (((UV)dec->cur[-7]) << 48)
529 | (((UV)dec->cur[-6]) << 40) 673 | (((UV)dec->cur[-6]) << 40)
530 | (((UV)dec->cur[-5]) << 32) 674 | (((UV)dec->cur[-5]) << 32)
675#endif
531 | (((UV)dec->cur[-4]) << 24) 676 | (((UV)dec->cur[-4]) << 24)
532 | (((UV)dec->cur[-3]) << 16) 677 | (((UV)dec->cur[-3]) << 16)
533 | (((UV)dec->cur[-2]) << 8) 678 | (((UV)dec->cur[-2]) << 8)
534 | ((UV)dec->cur[-1]); 679 | ((UV)dec->cur[-1]);
535 680 }
536 default: 681 else
537 ERR ("corrupted CBOR data (unsupported integer minor encoding)"); 682 ERR ("corrupted CBOR data (unsupported integer minor encoding)");
538 }
539 683
540fail: 684fail:
541 return 0; 685 return 0;
542} 686}
543 687
548{ 692{
549 AV *av = newAV (); 693 AV *av = newAV ();
550 694
551 DEC_INC_DEPTH; 695 DEC_INC_DEPTH;
552 696
553 if ((*dec->cur & 31) == 31) 697 if (*dec->cur == (MAJOR_ARRAY | MINOR_INDEF))
554 { 698 {
555 ++dec->cur; 699 ++dec->cur;
556 700
557 for (;;) 701 for (;;)
558 { 702 {
559 WANT (1); 703 WANT (1);
560 704
561 if (*dec->cur == (0xe0 | 31)) 705 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
562 { 706 {
563 ++dec->cur; 707 ++dec->cur;
564 break; 708 break;
565 } 709 }
566 710
569 } 713 }
570 else 714 else
571 { 715 {
572 int i, len = decode_uint (dec); 716 int i, len = decode_uint (dec);
573 717
718 WANT (len); // complexity check for av_fill - need at least one byte per value, do not allow supersize arrays
574 av_fill (av, len - 1); 719 av_fill (av, len - 1);
575 720
576 for (i = 0; i < len; ++i) 721 for (i = 0; i < len; ++i)
577 AvARRAY (av)[i] = decode_sv (dec); 722 AvARRAY (av)[i] = decode_sv (dec);
578 } 723 }
588 733
589static void 734static void
590decode_he (dec_t *dec, HV *hv) 735decode_he (dec_t *dec, HV *hv)
591{ 736{
592 // for speed reasons, we specialcase single-string 737 // for speed reasons, we specialcase single-string
593 // byte or utf-8 strings as keys. 738 // byte or utf-8 strings as keys, but only when !stringref
594 739
595 if (*dec->cur >= 0x40 && *dec->cur <= 0x40 + 27) 740 if (ecb_expect_true (!dec->stringref))
741 if (ecb_expect_true ((*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8))
596 { 742 {
597 I32 len = decode_uint (dec); 743 I32 len = decode_uint (dec);
598 char *key = (char *)dec->cur; 744 char *key = (char *)dec->cur;
599 745
600 dec->cur += len; 746 dec->cur += len;
601 747
602 hv_store (hv, key, len, decode_sv (dec), 0); 748 hv_store (hv, key, len, decode_sv (dec), 0);
749
750 return;
603 } 751 }
604 else if (*dec->cur >= 0x60 && *dec->cur <= 0x60 + 27) 752 else if (ecb_expect_true ((*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8))
605 { 753 {
606 I32 len = decode_uint (dec); 754 I32 len = decode_uint (dec);
607 char *key = (char *)dec->cur; 755 char *key = (char *)dec->cur;
608 756
609 dec->cur += len; 757 dec->cur += len;
610 758
759 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
760 if (!is_utf8_string (key, len))
761 ERR ("corrupted CBOR data (invalid UTF-8 in map key)");
762
611 hv_store (hv, key, -len, decode_sv (dec), 0); 763 hv_store (hv, key, -len, decode_sv (dec), 0);
764
765 return;
612 } 766 }
613 else 767
614 {
615 SV *k = decode_sv (dec); 768 SV *k = decode_sv (dec);
616 SV *v = decode_sv (dec); 769 SV *v = decode_sv (dec);
617 770
618 hv_store_ent (hv, k, v, 0); 771 hv_store_ent (hv, k, v, 0);
619 SvREFCNT_dec (k); 772 SvREFCNT_dec (k);
620 } 773
774fail:
775 ;
621} 776}
622 777
623static SV * 778static SV *
624decode_hv (dec_t *dec) 779decode_hv (dec_t *dec)
625{ 780{
626 HV *hv = newHV (); 781 HV *hv = newHV ();
627 782
628 DEC_INC_DEPTH; 783 DEC_INC_DEPTH;
629 784
630 if ((*dec->cur & 31) == 31) 785 if (*dec->cur == (MAJOR_MAP | MINOR_INDEF))
631 { 786 {
632 ++dec->cur; 787 ++dec->cur;
633 788
634 for (;;) 789 for (;;)
635 { 790 {
636 WANT (1); 791 WANT (1);
637 792
638 if (*dec->cur == (0xe0 | 31)) 793 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
639 { 794 {
640 ++dec->cur; 795 ++dec->cur;
641 break; 796 break;
642 } 797 }
643 798
664static SV * 819static SV *
665decode_str (dec_t *dec, int utf8) 820decode_str (dec_t *dec, int utf8)
666{ 821{
667 SV *sv = 0; 822 SV *sv = 0;
668 823
669 if ((*dec->cur & 31) == 31) 824 if ((*dec->cur & MINOR_MASK) == MINOR_INDEF)
670 { 825 {
826 // indefinite length strings
671 ++dec->cur; 827 ++dec->cur;
672 828
829 U8 major = *dec->cur & MAJOR_MISC;
830
673 sv = newSVpvn ("", 0); 831 sv = newSVpvn ("", 0);
674 832
675 // not very fast, and certainly not robust against illegal input
676 for (;;) 833 for (;;)
677 { 834 {
678 WANT (1); 835 WANT (1);
679 836
680 if (*dec->cur == (0xe0 | 31)) 837 if ((*dec->cur - major) > LENGTH_EXT8)
838 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
681 { 839 {
682 ++dec->cur; 840 ++dec->cur;
683 break; 841 break;
684 } 842 }
843 else
844 ERR ("corrupted CBOR data (invalid chunks in indefinite length string)");
685 845
686 sv_catsv (sv, decode_sv (dec)); 846 STRLEN len = decode_uint (dec);
847
848 WANT (len);
849 sv_catpvn (sv, dec->cur, len);
850 dec->cur += len;
687 } 851 }
688 } 852 }
689 else 853 else
690 { 854 {
691 STRLEN len = decode_uint (dec); 855 STRLEN len = decode_uint (dec);
692 856
693 WANT (len); 857 WANT (len);
694 sv = newSVpvn (dec->cur, len); 858 sv = newSVpvn (dec->cur, len);
695 dec->cur += len; 859 dec->cur += len;
860
861 if (ecb_expect_false (dec->stringref)
862 && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1))
863 av_push (dec->stringref, SvREFCNT_inc_NN (sv));
696 } 864 }
697 865
698 if (utf8) 866 if (utf8)
867 {
868 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
869 if (!is_utf8_string (SvPVX (sv), SvCUR (sv)))
870 ERR ("corrupted CBOR data (invalid UTF-8 in text string)");
871
699 SvUTF8_on (sv); 872 SvUTF8_on (sv);
873 }
700 874
701 return sv; 875 return sv;
702 876
703fail: 877fail:
704 SvREFCNT_dec (sv); 878 SvREFCNT_dec (sv);
706} 880}
707 881
708static SV * 882static SV *
709decode_tagged (dec_t *dec) 883decode_tagged (dec_t *dec)
710{ 884{
885 SV *sv = 0;
711 UV tag = decode_uint (dec); 886 UV tag = decode_uint (dec);
887
888 WANT (1);
889
890 switch (tag)
891 {
892 case CBOR_TAG_MAGIC:
712 SV *sv = decode_sv (dec); 893 sv = decode_sv (dec);
894 break;
713 895
714 if (tag == CBOR_TAG_MAGIC) 896 case CBOR_TAG_INDIRECTION:
715 return sv; 897 sv = newRV_noinc (decode_sv (dec));
716 else if (tag == CBOR_TAG_PERL_OBJECT) 898 break;
717 {
718 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
719 ERR ("corrupted CBOR data (non-array perl object)");
720 899
721 AV *av = (AV *)SvRV (sv); 900 case CBOR_TAG_STRINGREF_NAMESPACE:
722 int len = av_len (av) + 1;
723 HV *stash = gv_stashsv (*av_fetch (av, 0, 1), 0);
724
725 if (!stash)
726 ERR ("cannot decode perl-object (package does not exist)");
727
728 GV *method = gv_fetchmethod_autoload (stash, "THAW", 0);
729 901 {
730 if (!method) 902 ENTER; SAVETMPS;
731 ERR ("cannot decode perl-object (package does not have a THAW method)");
732
733 dSP;
734 903
735 ENTER; SAVETMPS; PUSHMARK (SP); 904 SAVESPTR (dec->stringref);
736 EXTEND (SP, len + 1); 905 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ());
737 // we re-bless the reference to get overload and other niceties right
738 PUSHs (*av_fetch (av, 0, 1));
739 PUSHs (sv_cbor);
740 906
741 int i; 907 sv = decode_sv (dec);
742 908
743 for (i = 1; i < len; ++i)
744 PUSHs (*av_fetch (av, i, 1));
745
746 PUTBACK;
747 call_sv ((SV *)GvCV (method), G_SCALAR | G_EVAL);
748 SPAGAIN;
749
750 if (SvTRUE (ERRSV))
751 {
752 FREETMPS; LEAVE; 909 FREETMPS; LEAVE;
910 }
911 break;
912
913 case CBOR_TAG_STRINGREF:
914 {
915 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
916 ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
917
918 UV idx = decode_uint (dec);
919
920 if (!dec->stringref || (int)idx > AvFILLp (dec->stringref))
921 ERR ("corrupted CBOR data (stringref index out of bounds or outside namespace)");
922
923 sv = newSVsv (AvARRAY (dec->stringref)[idx]);
924 }
925 break;
926
927 case CBOR_TAG_VALUE_SHAREABLE:
928 {
929 if (ecb_expect_false (!dec->shareable))
930 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ());
931
932 if (dec->cbor.flags & F_ALLOW_CYCLES)
933 {
934 sv = newSV (0);
935 av_push (dec->shareable, SvREFCNT_inc_NN (sv));
936
937 SV *osv = decode_sv (dec);
938 sv_setsv (sv, osv);
939 SvREFCNT_dec_NN (osv);
940 }
941 else
942 {
943 av_push (dec->shareable, &PL_sv_undef);
944 int idx = AvFILLp (dec->shareable);
945 sv = decode_sv (dec);
946 av_store (dec->shareable, idx, SvREFCNT_inc_NN (sv));
947 }
948 }
949 break;
950
951 case CBOR_TAG_VALUE_SHAREDREF:
952 {
953 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
954 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
955
956 UV idx = decode_uint (dec);
957
958 if (!dec->shareable || (int)idx > AvFILLp (dec->shareable))
959 ERR ("corrupted CBOR data (sharedref index out of bounds)");
960
961 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]);
962
963 if (sv == &PL_sv_undef)
964 ERR ("cyclic CBOR data structure found, but allow_cycles is not enabled");
965 }
966 break;
967
968 case CBOR_TAG_PERL_OBJECT:
969 {
970 sv = decode_sv (dec);
971
972 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
973 ERR ("corrupted CBOR data (non-array perl object)");
974
975 AV *av = (AV *)SvRV (sv);
976 int len = av_len (av) + 1;
977 HV *stash = gv_stashsv (*av_fetch (av, 0, 1), 0);
978
979 if (!stash)
980 ERR ("cannot decode perl-object (package does not exist)");
981
982 GV *method = gv_fetchmethod_autoload (stash, "THAW", 0);
983
984 if (!method)
985 ERR ("cannot decode perl-object (package does not have a THAW method)");
986
987 dSP;
988
989 ENTER; SAVETMPS; PUSHMARK (SP);
990 EXTEND (SP, len + 1);
991 // we re-bless the reference to get overload and other niceties right
992 PUSHs (*av_fetch (av, 0, 1));
993 PUSHs (sv_cbor);
994
995 int i;
996
997 for (i = 1; i < len; ++i)
998 PUSHs (*av_fetch (av, i, 1));
999
1000 PUTBACK;
1001 call_sv ((SV *)GvCV (method), G_SCALAR | G_EVAL);
1002 SPAGAIN;
1003
1004 if (SvTRUE (ERRSV))
1005 {
1006 FREETMPS; LEAVE;
753 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV)))); 1007 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV))));
754 } 1008 }
755 1009
756 SvREFCNT_dec (sv); 1010 SvREFCNT_dec (sv);
757 sv = SvREFCNT_inc (POPs); 1011 sv = SvREFCNT_inc (POPs);
758 1012
759 PUTBACK; 1013 PUTBACK;
760 1014
761 FREETMPS; LEAVE; 1015 FREETMPS; LEAVE;
1016 }
1017 break;
762 1018
763 return sv; 1019 default:
764 } 1020 {
765 else 1021 sv = decode_sv (dec);
766 { 1022
1023 dSP;
1024 ENTER; SAVETMPS; PUSHMARK (SP);
1025 EXTEND (SP, 2);
1026 PUSHs (newSVuv (tag));
1027 PUSHs (sv);
1028
1029 PUTBACK;
1030 int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL);
1031 SPAGAIN;
1032
1033 if (SvTRUE (ERRSV))
1034 {
1035 FREETMPS; LEAVE;
1036 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV))));
1037 }
1038
1039 if (count)
1040 {
1041 SvREFCNT_dec (sv);
1042 sv = SvREFCNT_inc (POPs);
1043 }
1044 else
1045 {
767 AV *av = newAV (); 1046 AV *av = newAV ();
768 av_push (av, newSVuv (tag)); 1047 av_push (av, newSVuv (tag));
769 av_push (av, sv); 1048 av_push (av, sv);
770 1049
771 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 1050 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
772 ? cbor_tagged_stash 1051 ? cbor_tagged_stash
773 : gv_stashpv ("CBOR::XS::Tagged" , 1); 1052 : gv_stashpv ("CBOR::XS::Tagged" , 1);
774
775 return sv_bless (newRV_noinc ((SV *)av), tagged_stash); 1053 sv = sv_bless (newRV_noinc ((SV *)av), tagged_stash);
1054 }
1055
1056 PUTBACK;
1057
1058 FREETMPS; LEAVE;
1059 }
1060 break;
776 } 1061 }
1062
1063 return sv;
777 1064
778fail: 1065fail:
779 SvREFCNT_dec (sv); 1066 SvREFCNT_dec (sv);
780 return &PL_sv_undef; 1067 return &PL_sv_undef;
781} 1068}
783static SV * 1070static SV *
784decode_sv (dec_t *dec) 1071decode_sv (dec_t *dec)
785{ 1072{
786 WANT (1); 1073 WANT (1);
787 1074
788 switch (*dec->cur >> 5) 1075 switch (*dec->cur >> MAJOR_SHIFT)
789 { 1076 {
790 case 0: // unsigned int 1077 case MAJOR_POS_INT >> MAJOR_SHIFT: return newSVuv (decode_uint (dec));
791 return newSVuv (decode_uint (dec)); 1078 case MAJOR_NEG_INT >> MAJOR_SHIFT: return newSViv (-1 - (IV)decode_uint (dec));
792 case 1: // negative int 1079 case MAJOR_BYTES >> MAJOR_SHIFT: return decode_str (dec, 0);
793 return newSViv (-1 - (IV)decode_uint (dec)); 1080 case MAJOR_TEXT >> MAJOR_SHIFT: return decode_str (dec, 1);
794 case 2: // octet string 1081 case MAJOR_ARRAY >> MAJOR_SHIFT: return decode_av (dec);
795 return decode_str (dec, 0); 1082 case MAJOR_MAP >> MAJOR_SHIFT: return decode_hv (dec);
796 case 3: // utf-8 string 1083 case MAJOR_TAG >> MAJOR_SHIFT: return decode_tagged (dec);
797 return decode_str (dec, 1); 1084
798 case 4: // array 1085 case MAJOR_MISC >> MAJOR_SHIFT:
799 return decode_av (dec);
800 case 5: // map
801 return decode_hv (dec);
802 case 6: // tag
803 return decode_tagged (dec);
804 case 7: // misc
805 switch (*dec->cur++ & 31) 1086 switch (*dec->cur++ & MINOR_MASK)
806 { 1087 {
807 case 20: 1088 case SIMPLE_FALSE:
808#if CBOR_SLOW 1089#if CBOR_SLOW
809 types_false = get_bool ("Types::Serialiser::false"); 1090 types_false = get_bool ("Types::Serialiser::false");
810#endif 1091#endif
811 return newSVsv (types_false); 1092 return newSVsv (types_false);
812 case 21: 1093 case SIMPLE_TRUE:
813#if CBOR_SLOW 1094#if CBOR_SLOW
814 types_true = get_bool ("Types::Serialiser::true"); 1095 types_true = get_bool ("Types::Serialiser::true");
815#endif 1096#endif
816 return newSVsv (types_true); 1097 return newSVsv (types_true);
817 case 22: 1098 case SIMPLE_NULL:
818 return newSVsv (&PL_sv_undef); 1099 return newSVsv (&PL_sv_undef);
819 case 23: 1100 case SIMPLE_UNDEF:
820#if CBOR_SLOW 1101#if CBOR_SLOW
821 types_error = get_bool ("Types::Serialiser::error"); 1102 types_error = get_bool ("Types::Serialiser::error");
822#endif 1103#endif
823 return newSVsv (types_error); 1104 return newSVsv (types_error);
824 1105
825 case 25: 1106 case MISC_FLOAT16:
826 { 1107 {
827 WANT (2); 1108 WANT (2);
828 1109
829 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1]; 1110 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1];
830 dec->cur += 2; 1111 dec->cur += 2;
831 1112
832 return newSVnv (ecb_binary16_to_float (fp)); 1113 return newSVnv (ecb_binary16_to_float (fp));
833 } 1114 }
834 1115
835 case 26: 1116 case MISC_FLOAT32:
836 { 1117 {
837 uint32_t fp; 1118 uint32_t fp;
838 WANT (4); 1119 WANT (4);
839 memcpy (&fp, dec->cur, 4); 1120 memcpy (&fp, dec->cur, 4);
840 dec->cur += 4; 1121 dec->cur += 4;
843 fp = ecb_bswap32 (fp); 1124 fp = ecb_bswap32 (fp);
844 1125
845 return newSVnv (ecb_binary32_to_float (fp)); 1126 return newSVnv (ecb_binary32_to_float (fp));
846 } 1127 }
847 1128
848 case 27: 1129 case MISC_FLOAT64:
849 { 1130 {
850 uint64_t fp; 1131 uint64_t fp;
851 WANT (8); 1132 WANT (8);
852 memcpy (&fp, dec->cur, 8); 1133 memcpy (&fp, dec->cur, 8);
853 dec->cur += 8; 1134 dec->cur += 8;
856 fp = ecb_bswap64 (fp); 1137 fp = ecb_bswap64 (fp);
857 1138
858 return newSVnv (ecb_binary64_to_double (fp)); 1139 return newSVnv (ecb_binary64_to_double (fp));
859 } 1140 }
860 1141
861 // 0..19 unassigned 1142 // 0..19 unassigned simple
862 // 24 reserved + unassigned (reserved values are not encodable) 1143 // 24 reserved + unassigned simple (reserved values are not encodable)
1144 // 28-30 unassigned misc
1145 // 31 break code
863 default: 1146 default:
864 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)"); 1147 ERR ("corrupted CBOR data (reserved/unassigned/unexpected major 7 value)");
865 } 1148 }
866 1149
867 break; 1150 break;
868 } 1151 }
869 1152
872} 1155}
873 1156
874static SV * 1157static SV *
875decode_cbor (SV *string, CBOR *cbor, char **offset_return) 1158decode_cbor (SV *string, CBOR *cbor, char **offset_return)
876{ 1159{
877 dec_t dec; 1160 dec_t dec = { };
878 SV *sv; 1161 SV *sv;
879 STRLEN len; 1162 STRLEN len;
880 char *data = SvPVbyte (string, len); 1163 char *data = SvPVbyte (string, len);
881 1164
882 if (len > cbor->max_size && cbor->max_size) 1165 if (len > cbor->max_size && cbor->max_size)
884 (unsigned long)len, (unsigned long)cbor->max_size); 1167 (unsigned long)len, (unsigned long)cbor->max_size);
885 1168
886 dec.cbor = *cbor; 1169 dec.cbor = *cbor;
887 dec.cur = (U8 *)data; 1170 dec.cur = (U8 *)data;
888 dec.end = (U8 *)data + len; 1171 dec.end = (U8 *)data + len;
889 dec.err = 0;
890 dec.depth = 0;
891 1172
892 sv = decode_sv (&dec); 1173 sv = decode_sv (&dec);
893 1174
894 if (offset_return) 1175 if (offset_return)
895 *offset_return = dec.cur; 1176 *offset_return = dec.cur;
898 if (dec.cur != dec.end && !dec.err) 1179 if (dec.cur != dec.end && !dec.err)
899 dec.err = "garbage after CBOR object"; 1180 dec.err = "garbage after CBOR object";
900 1181
901 if (dec.err) 1182 if (dec.err)
902 { 1183 {
1184 if (dec.shareable)
1185 {
1186 // need to break cyclic links, which whould all be in shareable
1187 int i;
1188 SV **svp;
1189
1190 for (i = av_len (dec.shareable) + 1; i--; )
1191 if ((svp = av_fetch (dec.shareable, i, 0)))
1192 sv_setsv (*svp, &PL_sv_undef);
1193 }
1194
903 SvREFCNT_dec (sv); 1195 SvREFCNT_dec (sv);
904 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur); 1196 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur);
905 } 1197 }
906 1198
907 sv = sv_2mortal (sv); 1199 sv = sv_2mortal (sv);
908 1200
909 return sv; 1201 return sv;
910} 1202}
911 1203
1204/////////////////////////////////////////////////////////////////////////////
1205// incremental parser
1206
1207#define INCR_DONE(cbor) (AvFILLp (cbor->incr_count) < 0)
1208
1209// returns 0 for notyet, 1 for success or error
1210static int
1211incr_parse (CBOR *self, SV *cborstr)
1212{
1213 STRLEN cur;
1214 SvPV (cborstr, cur);
1215
1216 while (ecb_expect_true (self->incr_need <= cur))
1217 {
1218 // table of integer count bytes
1219 static I8 incr_len[MINOR_MASK + 1] = {
1220 0, 0, 0, 0, 0, 0, 0, 0,
1221 0, 0, 0, 0, 0, 0, 0, 0,
1222 0, 0, 0, 0, 0, 0, 0, 0,
1223 1, 2, 4, 8,-1,-1,-1,-2
1224 };
1225
1226 const U8 *p = SvPVX (cborstr) + self->incr_pos;
1227 U8 m = *p & MINOR_MASK;
1228 IV count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1229 I8 ilen = incr_len[m];
1230
1231 self->incr_need = self->incr_pos + 1;
1232
1233 if (ecb_expect_false (ilen < 0))
1234 {
1235 if (m != MINOR_INDEF)
1236 return 1; // error
1237
1238 if (*p == (MAJOR_MISC | MINOR_INDEF))
1239 {
1240 if (count >= 0)
1241 return 1; // error
1242
1243 count = 1;
1244 }
1245 else
1246 {
1247 av_push (self->incr_count, newSViv (-1)); //TODO: nest
1248 count = -1;
1249 }
1250 }
1251 else
1252 {
1253 self->incr_need += ilen;
1254 if (ecb_expect_false (self->incr_need > cur))
1255 return 0;
1256
1257 int major = *p >> MAJOR_SHIFT;
1258
1259 switch (major)
1260 {
1261 case MAJOR_BYTES >> MAJOR_SHIFT:
1262 case MAJOR_TEXT >> MAJOR_SHIFT:
1263 case MAJOR_ARRAY >> MAJOR_SHIFT:
1264 case MAJOR_MAP >> MAJOR_SHIFT:
1265 {
1266 UV len;
1267
1268 if (ecb_expect_false (ilen))
1269 {
1270 len = 0;
1271
1272 do {
1273 len = (len << 8) | *++p;
1274 } while (--ilen);
1275 }
1276 else
1277 len = m;
1278
1279 switch (major)
1280 {
1281 case MAJOR_BYTES >> MAJOR_SHIFT:
1282 case MAJOR_TEXT >> MAJOR_SHIFT:
1283 self->incr_need += len;
1284 if (ecb_expect_false (self->incr_need > cur))
1285 return 0;
1286
1287 break;
1288
1289 case MAJOR_MAP >> MAJOR_SHIFT:
1290 len <<= 1;
1291 case MAJOR_ARRAY >> MAJOR_SHIFT:
1292 if (len)
1293 {
1294 av_push (self->incr_count, newSViv (len + 1)); //TODO: nest
1295 count = len + 1;
1296 }
1297 break;
1298 }
1299 }
1300 }
1301 }
1302
1303 self->incr_pos = self->incr_need;
1304
1305 if (count > 0)
1306 {
1307 while (!--count)
1308 {
1309 if (!AvFILLp (self->incr_count))
1310 return 1; // done
1311
1312 SvREFCNT_dec_NN (av_pop (self->incr_count));
1313 count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1314 }
1315
1316 SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]) = count;
1317 }
1318 }
1319
1320 return 0;
1321}
1322
1323
912///////////////////////////////////////////////////////////////////////////// 1324/////////////////////////////////////////////////////////////////////////////
913// XS interface functions 1325// XS interface functions
914 1326
915MODULE = CBOR::XS PACKAGE = CBOR::XS 1327MODULE = CBOR::XS PACKAGE = CBOR::XS
916 1328
923 types_error_stash = gv_stashpv ("Types::Serialiser::Error" , 1); 1335 types_error_stash = gv_stashpv ("Types::Serialiser::Error" , 1);
924 1336
925 types_true = get_bool ("Types::Serialiser::true" ); 1337 types_true = get_bool ("Types::Serialiser::true" );
926 types_false = get_bool ("Types::Serialiser::false"); 1338 types_false = get_bool ("Types::Serialiser::false");
927 types_error = get_bool ("Types::Serialiser::error"); 1339 types_error = get_bool ("Types::Serialiser::error");
1340
1341 default_filter = newSVpv ("CBOR::XS::default_filter", 0);
928 1342
929 sv_cbor = newSVpv ("CBOR", 0); 1343 sv_cbor = newSVpv ("CBOR", 0);
930 SvREADONLY_on (sv_cbor); 1344 SvREADONLY_on (sv_cbor);
931} 1345}
932 1346
953 1367
954void shrink (CBOR *self, int enable = 1) 1368void shrink (CBOR *self, int enable = 1)
955 ALIAS: 1369 ALIAS:
956 shrink = F_SHRINK 1370 shrink = F_SHRINK
957 allow_unknown = F_ALLOW_UNKNOWN 1371 allow_unknown = F_ALLOW_UNKNOWN
1372 allow_sharing = F_ALLOW_SHARING
1373 allow_cycles = F_ALLOW_CYCLES
1374 pack_strings = F_PACK_STRINGS
1375 validate_utf8 = F_VALIDATE_UTF8
958 PPCODE: 1376 PPCODE:
959{ 1377{
960 if (enable) 1378 if (enable)
961 self->flags |= ix; 1379 self->flags |= ix;
962 else 1380 else
967 1385
968void get_shrink (CBOR *self) 1386void get_shrink (CBOR *self)
969 ALIAS: 1387 ALIAS:
970 get_shrink = F_SHRINK 1388 get_shrink = F_SHRINK
971 get_allow_unknown = F_ALLOW_UNKNOWN 1389 get_allow_unknown = F_ALLOW_UNKNOWN
1390 get_allow_sharing = F_ALLOW_SHARING
1391 get_allow_cycles = F_ALLOW_CYCLES
1392 get_pack_strings = F_PACK_STRINGS
1393 get_validate_utf8 = F_VALIDATE_UTF8
972 PPCODE: 1394 PPCODE:
973 XPUSHs (boolSV (self->flags & ix)); 1395 XPUSHs (boolSV (self->flags & ix));
974 1396
975void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) 1397void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
976 PPCODE: 1398 PPCODE:
989 XPUSHs (ST (0)); 1411 XPUSHs (ST (0));
990 1412
991int get_max_size (CBOR *self) 1413int get_max_size (CBOR *self)
992 CODE: 1414 CODE:
993 RETVAL = self->max_size; 1415 RETVAL = self->max_size;
1416 OUTPUT:
1417 RETVAL
1418
1419void filter (CBOR *self, SV *filter = 0)
1420 PPCODE:
1421 SvREFCNT_dec (self->filter);
1422 self->filter = filter ? newSVsv (filter) : filter;
1423 XPUSHs (ST (0));
1424
1425SV *get_filter (CBOR *self)
1426 CODE:
1427 RETVAL = self->filter ? self->filter : NEWSV (0, 0);
994 OUTPUT: 1428 OUTPUT:
995 RETVAL 1429 RETVAL
996 1430
997void encode (CBOR *self, SV *scalar) 1431void encode (CBOR *self, SV *scalar)
998 PPCODE: 1432 PPCODE:
1013 EXTEND (SP, 2); 1447 EXTEND (SP, 2);
1014 PUSHs (sv); 1448 PUSHs (sv);
1015 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); 1449 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
1016} 1450}
1017 1451
1452void incr_parse (CBOR *self, SV *cborstr)
1453 ALIAS:
1454 incr_parse_multiple = 1
1455 PPCODE:
1456{
1457 if (SvUTF8 (cborstr))
1458 sv_utf8_downgrade (cborstr, 0);
1459
1460 if (!self->incr_count)
1461 {
1462 self->incr_count = newAV ();
1463 self->incr_pos = 0;
1464 self->incr_need = 1;
1465
1466 av_push (self->incr_count, newSViv (1));
1467 }
1468
1469 do
1470 {
1471 if (!incr_parse (self, cborstr))
1472 {
1473 if (self->incr_need > self->max_size && self->max_size)
1474 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
1475 (unsigned long)self->incr_need, (unsigned long)self->max_size);
1476
1477 break;
1478 }
1479
1480 SV *sv;
1481 char *offset;
1482
1483 PUTBACK; sv = decode_cbor (cborstr, self, &offset); SPAGAIN;
1484 XPUSHs (sv);
1485
1486 sv_chop (cborstr, offset);
1487
1488 av_clear (self->incr_count);
1489 av_push (self->incr_count, newSViv (1));
1490
1491 self->incr_pos = 0;
1492 self->incr_need = self->incr_pos + 1;
1493 }
1494 while (ix);
1495}
1496
1497void incr_reset (CBOR *self)
1498 CODE:
1499{
1500 SvREFCNT_dec (self->incr_count);
1501 self->incr_count = 0;
1502}
1503
1504void DESTROY (CBOR *self)
1505 PPCODE:
1506 cbor_free (self);
1507
1018PROTOTYPES: ENABLE 1508PROTOTYPES: ENABLE
1019 1509
1020void encode_cbor (SV *scalar) 1510void encode_cbor (SV *scalar)
1511 ALIAS:
1512 encode_cbor = 0
1513 encode_cbor_sharing = F_ALLOW_SHARING
1021 PPCODE: 1514 PPCODE:
1022{ 1515{
1023 CBOR cbor; 1516 CBOR cbor;
1024 cbor_init (&cbor); 1517 cbor_init (&cbor);
1518 cbor.flags |= ix;
1025 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN; 1519 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;
1026 XPUSHs (scalar); 1520 XPUSHs (scalar);
1027} 1521}
1028 1522
1029void decode_cbor (SV *cborstr) 1523void decode_cbor (SV *cborstr)

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines