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 1.19 by root, Wed Nov 20 02:03:09 2013 UTC vs.
Revision 1.43 by root, Sun Jan 5 14:24:54 2014 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
69 CBOR_TAG_STRINGREF = 25, // http://cbor.schmorp.de/stringref
29 CBOR_TAG_PERL_OBJECT = 24, // http://cbor.schmorp.de/perl-object 70 CBOR_TAG_PERL_OBJECT = 26, // http://cbor.schmorp.de/perl-object
30 CBOR_TAG_GENERIC_OBJECT = 25, // http://cbor.schmorp.de/generic-object 71 CBOR_TAG_GENERIC_OBJECT = 27, // http://cbor.schmorp.de/generic-object
31 CBOR_TAG_VALUE_SHAREABLE = 26, // http://cbor.schmorp.de/value-sharing 72 CBOR_TAG_VALUE_SHAREABLE = 28, // http://cbor.schmorp.de/value-sharing
32 CBOR_TAG_VALUE_SHAREDREF = 27, // http://cbor.schmorp.de/value-sharing 73 CBOR_TAG_VALUE_SHAREDREF = 29, // http://cbor.schmorp.de/value-sharing
33 CBOR_TAG_STRINGREF_NAMESPACE = 65537, // http://cbor.schmorp.de/stringref 74 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 75 CBOR_TAG_INDIRECTION = 22098, // http://cbor.schmorp.de/indirection
36 76
37 // rfc7049 77 // rfc7049
38 CBOR_TAG_DATETIME = 0, // rfc4287, utf-8 78 CBOR_TAG_DATETIME = 0, // rfc4287, utf-8
39 CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any 79 CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any
40 CBOR_TAG_POS_BIGNUM = 2, // byte string 80 CBOR_TAG_POS_BIGNUM = 2, // byte string
41 CBOR_TAG_NEG_BIGNUM = 3, // byte string 81 CBOR_TAG_NEG_BIGNUM = 3, // byte string
42 CBOR_TAG_DECIMAL = 4, // decimal fraction, array 82 CBOR_TAG_DECIMAL = 4, // decimal fraction, array
43 CBOR_TAG_BIGFLOAT = 5, // array 83 CBOR_TAG_BIGFLOAT = 5, // array
44 84
45 CBOR_TAG_CONV_B64U = 21, // base64url, any 85 CBOR_TAG_CONV_B64U = 21, // base64url, any
46 CBOR_TAG_CONV_B64 = 22, // base64, any 86 CBOR_TAG_CONV_B64 = 22, // base64, any
47 CBOR_TAG_CONV_HEX = 23, // base16, any 87 CBOR_TAG_CONV_HEX = 23, // base16, any
48 CBOR_TAG_CBOR = 24, // embedded cbor, byte string 88 CBOR_TAG_CBOR = 24, // embedded cbor, byte string
49 89
50 CBOR_TAG_URI = 32, // URI rfc3986, utf-8 90 CBOR_TAG_URI = 32, // URI rfc3986, utf-8
51 CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8 91 CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8
52 CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8 92 CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8
53 CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8 93 CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8
54 CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8 94 CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8
55 95
56 CBOR_TAG_MAGIC = 55799 // self-describe cbor 96 CBOR_TAG_MAGIC = 55799, // self-describe cbor
57}; 97};
58 98
59#define F_SHRINK 0x00000001UL 99#define F_SHRINK 0x00000001UL
60#define F_ALLOW_UNKNOWN 0x00000002UL 100#define F_ALLOW_UNKNOWN 0x00000002UL
61#define F_ALLOW_SHARING 0x00000004UL //TODO 101#define F_ALLOW_SHARING 0x00000004UL
102#define F_ALLOW_CYCLES 0x00000008UL
62#define F_DEDUP_STRINGS 0x00000008UL //TODO 103#define F_PACK_STRINGS 0x00000010UL
63#define F_DEDUP_KEYS 0x00000010UL //TODO 104#define F_VALIDATE_UTF8 0x00000020UL
64 105
65#define INIT_SIZE 32 // initial scalar size to be allocated 106#define INIT_SIZE 32 // initial scalar size to be allocated
66 107
67#define SB do { 108#define SB do {
68#define SE } while (0) 109#define SE } while (0)
80# define CBOR_SLOW 0 121# define CBOR_SLOW 0
81# define CBOR_STASH cbor_stash 122# define CBOR_STASH cbor_stash
82#endif 123#endif
83 124
84static 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::
85static SV *types_true, *types_false, *types_error, *sv_cbor; 126static SV *types_true, *types_false, *types_error, *sv_cbor, *default_filter;
86 127
87typedef struct { 128typedef struct {
88 U32 flags; 129 U32 flags;
89 U32 max_depth; 130 U32 max_depth;
90 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.
91} CBOR; 138} CBOR;
92 139
93ecb_inline void 140ecb_inline void
94cbor_init (CBOR *cbor) 141cbor_init (CBOR *cbor)
95{ 142{
96 Zero (cbor, 1, CBOR); 143 Zero (cbor, 1, CBOR);
97 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);
98} 152}
99 153
100///////////////////////////////////////////////////////////////////////////// 154/////////////////////////////////////////////////////////////////////////////
101// utility functions 155// utility functions
102 156
124 SvPV_renew (sv, SvCUR (sv) + 1); 178 SvPV_renew (sv, SvCUR (sv) + 1);
125#endif 179#endif
126 } 180 }
127} 181}
128 182
183// minimum length of a string to be registered for stringref
184ecb_inline int
185minimum_string_length (UV idx)
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}
197
129///////////////////////////////////////////////////////////////////////////// 198/////////////////////////////////////////////////////////////////////////////
130// encoder 199// encoder
131 200
132// structure used for encoding CBOR 201// structure used for encoding CBOR
133typedef struct 202typedef struct
135 char *cur; // SvPVX (sv) + current output position 204 char *cur; // SvPVX (sv) + current output position
136 char *end; // SvEND (sv) 205 char *end; // SvEND (sv)
137 SV *sv; // result scalar 206 SV *sv; // result scalar
138 CBOR cbor; 207 CBOR cbor;
139 U32 depth; // recursion level 208 U32 depth; // recursion level
140 HV *stringref; // string => index, or 0 209 HV *stringref[2]; // string => index, or 0 ([0] = bytes, [1] = utf-8)
210 UV stringref_idx;
141 HV *shareable; // ptr => index, or 0 211 HV *shareable; // ptr => index, or 0
142 UV shareable_idx; 212 UV shareable_idx;
143} enc_t; 213} enc_t;
144 214
145ecb_inline void 215ecb_inline void
164static void 234static void
165encode_uint (enc_t *enc, int major, UV len) 235encode_uint (enc_t *enc, int major, UV len)
166{ 236{
167 need (enc, 9); 237 need (enc, 9);
168 238
169 if (len < 24) 239 if (ecb_expect_true (len < LENGTH_EXT1))
170 *enc->cur++ = major | len; 240 *enc->cur++ = major | len;
171 else if (len <= 0xff) 241 else if (ecb_expect_true (len <= 0xffU))
172 { 242 {
173 *enc->cur++ = major | 24; 243 *enc->cur++ = major | LENGTH_EXT1;
174 *enc->cur++ = len; 244 *enc->cur++ = len;
175 } 245 }
176 else if (len <= 0xffff) 246 else if (len <= 0xffffU)
177 { 247 {
178 *enc->cur++ = major | 25; 248 *enc->cur++ = major | LENGTH_EXT2;
179 *enc->cur++ = len >> 8; 249 *enc->cur++ = len >> 8;
180 *enc->cur++ = len; 250 *enc->cur++ = len;
181 } 251 }
182 else if (len <= 0xffffffff) 252 else if (len <= 0xffffffffU)
183 { 253 {
184 *enc->cur++ = major | 26; 254 *enc->cur++ = major | LENGTH_EXT4;
185 *enc->cur++ = len >> 24; 255 *enc->cur++ = len >> 24;
186 *enc->cur++ = len >> 16; 256 *enc->cur++ = len >> 16;
187 *enc->cur++ = len >> 8; 257 *enc->cur++ = len >> 8;
188 *enc->cur++ = len; 258 *enc->cur++ = len;
189 } 259 }
190 else 260 else
191 { 261 {
192 *enc->cur++ = major | 27; 262 *enc->cur++ = major | LENGTH_EXT8;
193 *enc->cur++ = len >> 56; 263 *enc->cur++ = len >> 56;
194 *enc->cur++ = len >> 48; 264 *enc->cur++ = len >> 48;
195 *enc->cur++ = len >> 40; 265 *enc->cur++ = len >> 40;
196 *enc->cur++ = len >> 32; 266 *enc->cur++ = len >> 32;
197 *enc->cur++ = len >> 24; 267 *enc->cur++ = len >> 24;
199 *enc->cur++ = len >> 8; 269 *enc->cur++ = len >> 8;
200 *enc->cur++ = len; 270 *enc->cur++ = len;
201 } 271 }
202} 272}
203 273
204static void 274ecb_inline void
275encode_tag (enc_t *enc, UV tag)
276{
277 encode_uint (enc, MAJOR_TAG, tag);
278}
279
280ecb_inline void
205encode_str (enc_t *enc, int utf8, char *str, STRLEN len) 281encode_str (enc_t *enc, int utf8, char *str, STRLEN len)
206{ 282{
207 encode_uint (enc, utf8 ? 0x60 : 0x40, len); 283 encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len);
208 need (enc, len); 284 need (enc, len);
209 memcpy (enc->cur, str, len); 285 memcpy (enc->cur, str, len);
210 enc->cur += len; 286 enc->cur += len;
211} 287}
212 288
213ecb_inline void 289static void
214encode_tag (enc_t *enc, UV tag) 290encode_strref (enc_t *enc, int utf8, char *str, STRLEN len)
215{ 291{
216 encode_uint (enc, 0xc0, tag); 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);
217} 312}
218 313
219static void encode_sv (enc_t *enc, SV *sv); 314static void encode_sv (enc_t *enc, SV *sv);
220 315
221static void 316static void
226 if (enc->depth >= enc->cbor.max_depth) 321 if (enc->depth >= enc->cbor.max_depth)
227 croak (ERR_NESTING_EXCEEDED); 322 croak (ERR_NESTING_EXCEEDED);
228 323
229 ++enc->depth; 324 ++enc->depth;
230 325
231 encode_uint (enc, 0x80, len + 1); 326 encode_uint (enc, MAJOR_ARRAY, len + 1);
232 327
233 for (i = 0; i <= len; ++i) 328 for (i = 0; i <= len; ++i)
234 { 329 {
235 SV **svp = av_fetch (av, i, 0); 330 SV **svp = av_fetch (av, i, 0);
236 encode_sv (enc, svp ? *svp : &PL_sv_undef); 331 encode_sv (enc, svp ? *svp : &PL_sv_undef);
251 346
252 int pairs = hv_iterinit (hv); 347 int pairs = hv_iterinit (hv);
253 int mg = SvMAGICAL (hv); 348 int mg = SvMAGICAL (hv);
254 349
255 if (mg) 350 if (mg)
256 encode_ch (enc, 0xa0 | 31); 351 encode_ch (enc, MAJOR_MAP | MINOR_INDEF);
257 else 352 else
258 encode_uint (enc, 0xa0, pairs); 353 encode_uint (enc, MAJOR_MAP, pairs);
259 354
260 while ((he = hv_iternext (hv))) 355 while ((he = hv_iternext (hv)))
261 { 356 {
262 if (HeKLEN (he) == HEf_SVKEY) 357 if (HeKLEN (he) == HEf_SVKEY)
263 encode_sv (enc, HeSVKEY (he)); 358 encode_sv (enc, HeSVKEY (he));
264 else 359 else
265 encode_str (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he)); 360 encode_strref (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he));
266 361
267 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));
268 } 363 }
269 364
270 if (mg) 365 if (mg)
271 encode_ch (enc, 0xe0 | 31); 366 encode_ch (enc, MAJOR_MISC | MINOR_INDEF);
272 367
273 --enc->depth; 368 --enc->depth;
274} 369}
275 370
276// encode objects, arrays and special \0=false and \1=true values. 371// encode objects, arrays and special \0=false and \1=true values.
277static void 372static void
278encode_rv (enc_t *enc, SV *sv) 373encode_rv (enc_t *enc, SV *sv)
279{ 374{
280 SvGETMAGIC (sv); 375 SvGETMAGIC (sv);
281
282 if (ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING)
283 && ecb_expect_false (SvREFCNT (sv) > 1))
284 {
285 if (!enc->shareable)
286 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
287
288 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
289
290 if (SvOK (*svp))
291 {
292 encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF);
293 encode_uint (enc, 0x00, SvUV (*svp));
294 return;
295 }
296 else
297 {
298 sv_setuv (*svp, enc->shareable_idx);
299 ++enc->shareable_idx;
300 encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE);
301 }
302 }
303 376
304 svtype svt = SvTYPE (sv); 377 svtype svt = SvTYPE (sv);
305 378
306 if (ecb_expect_false (SvOBJECT (sv))) 379 if (ecb_expect_false (SvOBJECT (sv)))
307 { 380 {
314 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 387 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
315 ? cbor_tagged_stash 388 ? cbor_tagged_stash
316 : gv_stashpv ("CBOR::XS::Tagged" , 1); 389 : gv_stashpv ("CBOR::XS::Tagged" , 1);
317 390
318 HV *stash = SvSTASH (sv); 391 HV *stash = SvSTASH (sv);
319 GV *method;
320 392
321 if (stash == boolean_stash) 393 if (stash == boolean_stash)
322 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 }
323 else if (stash == error_stash) 398 else if (stash == error_stash)
324 encode_ch (enc, 0xe0 | 23); 399 {
400 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
401 return;
402 }
325 else if (stash == tagged_stash) 403 else if (stash == tagged_stash)
326 { 404 {
327 if (svt != SVt_PVAV) 405 if (svt != SVt_PVAV)
328 croak ("encountered CBOR::XS::Tagged object that isn't an array"); 406 croak ("encountered CBOR::XS::Tagged object that isn't an array");
329 407
330 encode_uint (enc, 0xc0, SvUV (*av_fetch ((AV *)sv, 0, 1))); 408 encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1)));
331 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))
332 } 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
333 else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0))) 442 if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
334 { 443 {
335 dSP; 444 dSP;
336 445
337 ENTER; SAVETMPS; PUSHMARK (SP); 446 ENTER; SAVETMPS; PUSHMARK (SP);
338 // 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
370 // catch this surprisingly common error 479 // catch this surprisingly common error
371 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv) 480 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)
372 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));
373 482
374 encode_tag (enc, CBOR_TAG_PERL_OBJECT); 483 encode_tag (enc, CBOR_TAG_PERL_OBJECT);
375 encode_uint (enc, 0x80, count + 1); 484 encode_uint (enc, MAJOR_ARRAY, count + 1);
376 encode_str (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash)); 485 encode_strref (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
377 486
378 while (count) 487 while (count)
379 encode_sv (enc, SP[1 - count--]); 488 encode_sv (enc, SP[1 - count--]);
380 489
381 PUTBACK; 490 PUTBACK;
402{ 511{
403 double nv = SvNVX (sv); 512 double nv = SvNVX (sv);
404 513
405 need (enc, 9); 514 need (enc, 9);
406 515
407 if (ecb_expect_false (nv == (U32)nv)) 516 if (ecb_expect_false (nv == (NV)(U32)nv))
408 encode_uint (enc, 0x00, (U32)nv); 517 encode_uint (enc, MAJOR_POS_INT, (U32)nv);
409 //TODO: maybe I32? 518 //TODO: maybe I32?
410 else if (ecb_expect_false (nv == (float)nv)) 519 else if (ecb_expect_false (nv == (float)nv))
411 { 520 {
412 uint32_t fp = ecb_float_to_binary32 (nv); 521 uint32_t fp = ecb_float_to_binary32 (nv);
413 522
414 *enc->cur++ = 0xe0 | 26; 523 *enc->cur++ = MAJOR_MISC | MISC_FLOAT32;
415 524
416 if (!ecb_big_endian ()) 525 if (!ecb_big_endian ())
417 fp = ecb_bswap32 (fp); 526 fp = ecb_bswap32 (fp);
418 527
419 memcpy (enc->cur, &fp, 4); 528 memcpy (enc->cur, &fp, 4);
421 } 530 }
422 else 531 else
423 { 532 {
424 uint64_t fp = ecb_double_to_binary64 (nv); 533 uint64_t fp = ecb_double_to_binary64 (nv);
425 534
426 *enc->cur++ = 0xe0 | 27; 535 *enc->cur++ = MAJOR_MISC | MISC_FLOAT64;
427 536
428 if (!ecb_big_endian ()) 537 if (!ecb_big_endian ())
429 fp = ecb_bswap64 (fp); 538 fp = ecb_bswap64 (fp);
430 539
431 memcpy (enc->cur, &fp, 8); 540 memcpy (enc->cur, &fp, 8);
440 549
441 if (SvPOKp (sv)) 550 if (SvPOKp (sv))
442 { 551 {
443 STRLEN len; 552 STRLEN len;
444 char *str = SvPV (sv, len); 553 char *str = SvPV (sv, len);
445 encode_str (enc, SvUTF8 (sv), str, len); 554 encode_strref (enc, SvUTF8 (sv), str, len);
446 } 555 }
447 else if (SvNOKp (sv)) 556 else if (SvNOKp (sv))
448 encode_nv (enc, sv); 557 encode_nv (enc, sv);
449 else if (SvIOKp (sv)) 558 else if (SvIOKp (sv))
450 { 559 {
451 if (SvIsUV (sv)) 560 if (SvIsUV (sv))
452 encode_uint (enc, 0x00, SvUVX (sv)); 561 encode_uint (enc, MAJOR_POS_INT, SvUVX (sv));
453 else if (SvIVX (sv) >= 0) 562 else if (SvIVX (sv) >= 0)
454 encode_uint (enc, 0x00, SvIVX (sv)); 563 encode_uint (enc, MAJOR_POS_INT, SvIVX (sv));
455 else 564 else
456 encode_uint (enc, 0x20, -(SvIVX (sv) + 1)); 565 encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1));
457 } 566 }
458 else if (SvROK (sv)) 567 else if (SvROK (sv))
459 encode_rv (enc, SvRV (sv)); 568 encode_rv (enc, SvRV (sv));
460 else if (!SvOK (sv)) 569 else if (!SvOK (sv))
461 encode_ch (enc, 0xe0 | 22); 570 encode_ch (enc, MAJOR_MISC | SIMPLE_NULL);
462 else if (enc->cbor.flags & F_ALLOW_UNKNOWN) 571 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
463 encode_ch (enc, 0xe0 | 23); 572 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
464 else 573 else
465 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",
466 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv)); 575 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
467} 576}
468 577
475 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 584 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
476 enc.cur = SvPVX (enc.sv); 585 enc.cur = SvPVX (enc.sv);
477 enc.end = SvEND (enc.sv); 586 enc.end = SvEND (enc.sv);
478 587
479 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
480 encode_sv (&enc, scalar); 597 encode_sv (&enc, scalar);
481 598
482 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 599 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
483 *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
484 601
499 const char *err; // parse error, if != 0 616 const char *err; // parse error, if != 0
500 CBOR cbor; 617 CBOR cbor;
501 U32 depth; // recursion depth 618 U32 depth; // recursion depth
502 U32 maxdepth; // recursion depth limit 619 U32 maxdepth; // recursion depth limit
503 AV *shareable; 620 AV *shareable;
621 AV *stringref;
622 SV *decode_tagged;
504} dec_t; 623} dec_t;
505 624
506#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
507 626
508#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")
511#define DEC_DEC_DEPTH --dec->depth 630#define DEC_DEC_DEPTH --dec->depth
512 631
513static UV 632static UV
514decode_uint (dec_t *dec) 633decode_uint (dec_t *dec)
515{ 634{
516 switch (*dec->cur & 31) 635 U8 m = *dec->cur & MINOR_MASK;
517 { 636 ++dec->cur;
518 case 0: case 1: case 2: case 3: case 4: case 5: case 6: case 7:
519 case 8: case 9: case 10: case 11: case 12: case 13: case 14: case 15:
520 case 16: case 17: case 18: case 19: case 20: case 21: case 22: case 23:
521 return *dec->cur++ & 31;
522 637
523 case 24: 638 if (ecb_expect_true (m < LENGTH_EXT1))
639 return m;
640 else if (ecb_expect_true (m == LENGTH_EXT1))
641 {
524 WANT (2); 642 WANT (1);
525 dec->cur += 2; 643 dec->cur += 1;
526 return dec->cur[-1]; 644 return dec->cur[-1];
527 645 }
528 case 25: 646 else if (ecb_expect_true (m == LENGTH_EXT2))
647 {
529 WANT (3); 648 WANT (2);
530 dec->cur += 3; 649 dec->cur += 2;
531 return (((UV)dec->cur[-2]) << 8) 650 return (((UV)dec->cur[-2]) << 8)
532 | ((UV)dec->cur[-1]); 651 | ((UV)dec->cur[-1]);
533 652 }
534 case 26: 653 else if (ecb_expect_true (m == LENGTH_EXT4))
654 {
535 WANT (5); 655 WANT (4);
536 dec->cur += 5; 656 dec->cur += 4;
537 return (((UV)dec->cur[-4]) << 24) 657 return (((UV)dec->cur[-4]) << 24)
538 | (((UV)dec->cur[-3]) << 16) 658 | (((UV)dec->cur[-3]) << 16)
539 | (((UV)dec->cur[-2]) << 8) 659 | (((UV)dec->cur[-2]) << 8)
540 | ((UV)dec->cur[-1]); 660 | ((UV)dec->cur[-1]);
541 661 }
542 case 27: 662 else if (ecb_expect_true (m == LENGTH_EXT8))
663 {
543 WANT (9); 664 WANT (8);
544 dec->cur += 9; 665 dec->cur += 8;
666
667 return
668#if UVSIZE < 8
669 0
670#else
545 return (((UV)dec->cur[-8]) << 56) 671 (((UV)dec->cur[-8]) << 56)
546 | (((UV)dec->cur[-7]) << 48) 672 | (((UV)dec->cur[-7]) << 48)
547 | (((UV)dec->cur[-6]) << 40) 673 | (((UV)dec->cur[-6]) << 40)
548 | (((UV)dec->cur[-5]) << 32) 674 | (((UV)dec->cur[-5]) << 32)
675#endif
549 | (((UV)dec->cur[-4]) << 24) 676 | (((UV)dec->cur[-4]) << 24)
550 | (((UV)dec->cur[-3]) << 16) 677 | (((UV)dec->cur[-3]) << 16)
551 | (((UV)dec->cur[-2]) << 8) 678 | (((UV)dec->cur[-2]) << 8)
552 | ((UV)dec->cur[-1]); 679 | ((UV)dec->cur[-1]);
553 680 }
554 default: 681 else
555 ERR ("corrupted CBOR data (unsupported integer minor encoding)"); 682 ERR ("corrupted CBOR data (unsupported integer minor encoding)");
556 }
557 683
558fail: 684fail:
559 return 0; 685 return 0;
560} 686}
561 687
566{ 692{
567 AV *av = newAV (); 693 AV *av = newAV ();
568 694
569 DEC_INC_DEPTH; 695 DEC_INC_DEPTH;
570 696
571 if ((*dec->cur & 31) == 31) 697 if (*dec->cur == (MAJOR_ARRAY | MINOR_INDEF))
572 { 698 {
573 ++dec->cur; 699 ++dec->cur;
574 700
575 for (;;) 701 for (;;)
576 { 702 {
577 WANT (1); 703 WANT (1);
578 704
579 if (*dec->cur == (0xe0 | 31)) 705 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
580 { 706 {
581 ++dec->cur; 707 ++dec->cur;
582 break; 708 break;
583 } 709 }
584 710
587 } 713 }
588 else 714 else
589 { 715 {
590 int i, len = decode_uint (dec); 716 int i, len = decode_uint (dec);
591 717
718 WANT (len); // complexity check for av_fill - need at least one byte per value, do not allow supersize arrays
592 av_fill (av, len - 1); 719 av_fill (av, len - 1);
593 720
594 for (i = 0; i < len; ++i) 721 for (i = 0; i < len; ++i)
595 AvARRAY (av)[i] = decode_sv (dec); 722 AvARRAY (av)[i] = decode_sv (dec);
596 } 723 }
606 733
607static void 734static void
608decode_he (dec_t *dec, HV *hv) 735decode_he (dec_t *dec, HV *hv)
609{ 736{
610 // for speed reasons, we specialcase single-string 737 // for speed reasons, we specialcase single-string
611 // byte or utf-8 strings as keys. 738 // byte or utf-8 strings as keys, but only when !stringref
612 739
613 if (*dec->cur >= 0x40 && *dec->cur <= 0x40 + 27) 740 if (ecb_expect_true (!dec->stringref))
741 if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8))
614 { 742 {
615 I32 len = decode_uint (dec); 743 I32 len = decode_uint (dec);
616 char *key = (char *)dec->cur; 744 char *key = (char *)dec->cur;
617 745
618 dec->cur += len; 746 dec->cur += len;
619 747
620 hv_store (hv, key, len, decode_sv (dec), 0); 748 hv_store (hv, key, len, decode_sv (dec), 0);
749
750 return;
621 } 751 }
622 else if (*dec->cur >= 0x60 && *dec->cur <= 0x60 + 27) 752 else if (ecb_expect_true ((U8)(*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8))
623 { 753 {
624 I32 len = decode_uint (dec); 754 I32 len = decode_uint (dec);
625 char *key = (char *)dec->cur; 755 char *key = (char *)dec->cur;
626 756
627 dec->cur += len; 757 dec->cur += len;
628 758
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
629 hv_store (hv, key, -len, decode_sv (dec), 0); 763 hv_store (hv, key, -len, decode_sv (dec), 0);
764
765 return;
630 } 766 }
631 else 767
632 {
633 SV *k = decode_sv (dec); 768 SV *k = decode_sv (dec);
634 SV *v = decode_sv (dec); 769 SV *v = decode_sv (dec);
635 770
636 hv_store_ent (hv, k, v, 0); 771 hv_store_ent (hv, k, v, 0);
637 SvREFCNT_dec (k); 772 SvREFCNT_dec (k);
638 } 773
774fail:
775 ;
639} 776}
640 777
641static SV * 778static SV *
642decode_hv (dec_t *dec) 779decode_hv (dec_t *dec)
643{ 780{
644 HV *hv = newHV (); 781 HV *hv = newHV ();
645 782
646 DEC_INC_DEPTH; 783 DEC_INC_DEPTH;
647 784
648 if ((*dec->cur & 31) == 31) 785 if (*dec->cur == (MAJOR_MAP | MINOR_INDEF))
649 { 786 {
650 ++dec->cur; 787 ++dec->cur;
651 788
652 for (;;) 789 for (;;)
653 { 790 {
654 WANT (1); 791 WANT (1);
655 792
656 if (*dec->cur == (0xe0 | 31)) 793 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
657 { 794 {
658 ++dec->cur; 795 ++dec->cur;
659 break; 796 break;
660 } 797 }
661 798
682static SV * 819static SV *
683decode_str (dec_t *dec, int utf8) 820decode_str (dec_t *dec, int utf8)
684{ 821{
685 SV *sv = 0; 822 SV *sv = 0;
686 823
687 if ((*dec->cur & 31) == 31) 824 if ((*dec->cur & MINOR_MASK) == MINOR_INDEF)
688 { 825 {
826 // indefinite length strings
689 ++dec->cur; 827 ++dec->cur;
690 828
829 U8 major = *dec->cur & MAJOR_MISC;
830
691 sv = newSVpvn ("", 0); 831 sv = newSVpvn ("", 0);
692 832
693 // not very fast, and certainly not robust against illegal input
694 for (;;) 833 for (;;)
695 { 834 {
696 WANT (1); 835 WANT (1);
697 836
698 if (*dec->cur == (0xe0 | 31)) 837 if ((*dec->cur - major) > LENGTH_EXT8)
838 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
699 { 839 {
700 ++dec->cur; 840 ++dec->cur;
701 break; 841 break;
702 } 842 }
843 else
844 ERR ("corrupted CBOR data (invalid chunks in indefinite length string)");
703 845
704 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;
705 } 851 }
706 } 852 }
707 else 853 else
708 { 854 {
709 STRLEN len = decode_uint (dec); 855 STRLEN len = decode_uint (dec);
710 856
711 WANT (len); 857 WANT (len);
712 sv = newSVpvn (dec->cur, len); 858 sv = newSVpvn (dec->cur, len);
713 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));
714 } 864 }
715 865
716 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
717 SvUTF8_on (sv); 872 SvUTF8_on (sv);
873 }
718 874
719 return sv; 875 return sv;
720 876
721fail: 877fail:
722 SvREFCNT_dec (sv); 878 SvREFCNT_dec (sv);
732 WANT (1); 888 WANT (1);
733 889
734 switch (tag) 890 switch (tag)
735 { 891 {
736 case CBOR_TAG_MAGIC: 892 case CBOR_TAG_MAGIC:
737 return decode_sv (dec); 893 sv = decode_sv (dec);
894 break;
738 895
739 case CBOR_TAG_INDIRECTION: 896 case CBOR_TAG_INDIRECTION:
740 return newRV_noinc (decode_sv (dec)); 897 sv = newRV_noinc (decode_sv (dec));
898 break;
899
900 case CBOR_TAG_STRINGREF_NAMESPACE:
901 {
902 ENTER; SAVETMPS;
903
904 SAVESPTR (dec->stringref);
905 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ());
906
907 sv = decode_sv (dec);
908
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;
741 926
742 case CBOR_TAG_VALUE_SHAREABLE: 927 case CBOR_TAG_VALUE_SHAREABLE:
743 { 928 {
744 if (ecb_expect_false (!dec->shareable)) 929 if (ecb_expect_false (!dec->shareable))
745 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ()); 930 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ());
746 931
932 if (dec->cbor.flags & F_ALLOW_CYCLES)
933 {
747 sv = newSV (0); 934 sv = newSV (0);
748 av_push (dec->shareable, SvREFCNT_inc_NN (sv)); 935 av_push (dec->shareable, SvREFCNT_inc_NN (sv));
749 936
750 SV *osv = decode_sv (dec); 937 SV *osv = decode_sv (dec);
751 sv_setsv (sv, osv); 938 sv_setsv (sv, osv);
752 SvREFCNT_dec_NN (osv); 939 SvREFCNT_dec_NN (osv);
753 } 940 }
754 941 else
755 return sv; 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;
756 950
757 case CBOR_TAG_VALUE_SHAREDREF: 951 case CBOR_TAG_VALUE_SHAREDREF:
758 { 952 {
759 if ((*dec->cur >> 5) != 0) 953 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
760 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)"); 954 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
761 955
762 UV idx = decode_uint (dec); 956 UV idx = decode_uint (dec);
763 957
764 if (!dec->shareable || idx > AvFILLp (dec->shareable)) 958 if (!dec->shareable || (int)idx > AvFILLp (dec->shareable))
765 ERR ("corrupted CBOR data (sharedref index out of bounds)"); 959 ERR ("corrupted CBOR data (sharedref index out of bounds)");
766 960
767 return SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]); 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");
768 } 965 }
966 break;
769 967
770 case CBOR_TAG_PERL_OBJECT: 968 case CBOR_TAG_PERL_OBJECT:
771 { 969 {
772 sv = decode_sv (dec); 970 sv = decode_sv (dec);
773 971
813 sv = SvREFCNT_inc (POPs); 1011 sv = SvREFCNT_inc (POPs);
814 1012
815 PUTBACK; 1013 PUTBACK;
816 1014
817 FREETMPS; LEAVE; 1015 FREETMPS; LEAVE;
818
819 return sv;
820 } 1016 }
1017 break;
821 1018
822 default: 1019 default:
823 { 1020 {
824 sv = decode_sv (dec); 1021 sv = decode_sv (dec);
825 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 {
826 AV *av = newAV (); 1046 AV *av = newAV ();
827 av_push (av, newSVuv (tag)); 1047 av_push (av, newSVuv (tag));
828 av_push (av, sv); 1048 av_push (av, sv);
829 1049
830 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 1050 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
831 ? cbor_tagged_stash 1051 ? cbor_tagged_stash
832 : gv_stashpv ("CBOR::XS::Tagged" , 1); 1052 : gv_stashpv ("CBOR::XS::Tagged" , 1);
833
834 return sv_bless (newRV_noinc ((SV *)av), tagged_stash); 1053 sv = sv_bless (newRV_noinc ((SV *)av), tagged_stash);
835 } 1054 }
1055
1056 PUTBACK;
1057
1058 FREETMPS; LEAVE;
1059 }
1060 break;
836 } 1061 }
1062
1063 return sv;
837 1064
838fail: 1065fail:
839 SvREFCNT_dec (sv); 1066 SvREFCNT_dec (sv);
840 return &PL_sv_undef; 1067 return &PL_sv_undef;
841} 1068}
843static SV * 1070static SV *
844decode_sv (dec_t *dec) 1071decode_sv (dec_t *dec)
845{ 1072{
846 WANT (1); 1073 WANT (1);
847 1074
848 switch (*dec->cur >> 5) 1075 switch (*dec->cur >> MAJOR_SHIFT)
849 { 1076 {
850 case 0: // unsigned int 1077 case MAJOR_POS_INT >> MAJOR_SHIFT: return newSVuv (decode_uint (dec));
851 return newSVuv (decode_uint (dec)); 1078 case MAJOR_NEG_INT >> MAJOR_SHIFT: return newSViv (-1 - (IV)decode_uint (dec));
852 case 1: // negative int 1079 case MAJOR_BYTES >> MAJOR_SHIFT: return decode_str (dec, 0);
853 return newSViv (-1 - (IV)decode_uint (dec)); 1080 case MAJOR_TEXT >> MAJOR_SHIFT: return decode_str (dec, 1);
854 case 2: // octet string 1081 case MAJOR_ARRAY >> MAJOR_SHIFT: return decode_av (dec);
855 return decode_str (dec, 0); 1082 case MAJOR_MAP >> MAJOR_SHIFT: return decode_hv (dec);
856 case 3: // utf-8 string 1083 case MAJOR_TAG >> MAJOR_SHIFT: return decode_tagged (dec);
857 return decode_str (dec, 1); 1084
858 case 4: // array 1085 case MAJOR_MISC >> MAJOR_SHIFT:
859 return decode_av (dec);
860 case 5: // map
861 return decode_hv (dec);
862 case 6: // tag
863 return decode_tagged (dec);
864 case 7: // misc
865 switch (*dec->cur++ & 31) 1086 switch (*dec->cur++ & MINOR_MASK)
866 { 1087 {
867 case 20: 1088 case SIMPLE_FALSE:
868#if CBOR_SLOW 1089#if CBOR_SLOW
869 types_false = get_bool ("Types::Serialiser::false"); 1090 types_false = get_bool ("Types::Serialiser::false");
870#endif 1091#endif
871 return newSVsv (types_false); 1092 return newSVsv (types_false);
872 case 21: 1093 case SIMPLE_TRUE:
873#if CBOR_SLOW 1094#if CBOR_SLOW
874 types_true = get_bool ("Types::Serialiser::true"); 1095 types_true = get_bool ("Types::Serialiser::true");
875#endif 1096#endif
876 return newSVsv (types_true); 1097 return newSVsv (types_true);
877 case 22: 1098 case SIMPLE_NULL:
878 return newSVsv (&PL_sv_undef); 1099 return newSVsv (&PL_sv_undef);
879 case 23: 1100 case SIMPLE_UNDEF:
880#if CBOR_SLOW 1101#if CBOR_SLOW
881 types_error = get_bool ("Types::Serialiser::error"); 1102 types_error = get_bool ("Types::Serialiser::error");
882#endif 1103#endif
883 return newSVsv (types_error); 1104 return newSVsv (types_error);
884 1105
885 case 25: 1106 case MISC_FLOAT16:
886 { 1107 {
887 WANT (2); 1108 WANT (2);
888 1109
889 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1]; 1110 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1];
890 dec->cur += 2; 1111 dec->cur += 2;
891 1112
892 return newSVnv (ecb_binary16_to_float (fp)); 1113 return newSVnv (ecb_binary16_to_float (fp));
893 } 1114 }
894 1115
895 case 26: 1116 case MISC_FLOAT32:
896 { 1117 {
897 uint32_t fp; 1118 uint32_t fp;
898 WANT (4); 1119 WANT (4);
899 memcpy (&fp, dec->cur, 4); 1120 memcpy (&fp, dec->cur, 4);
900 dec->cur += 4; 1121 dec->cur += 4;
903 fp = ecb_bswap32 (fp); 1124 fp = ecb_bswap32 (fp);
904 1125
905 return newSVnv (ecb_binary32_to_float (fp)); 1126 return newSVnv (ecb_binary32_to_float (fp));
906 } 1127 }
907 1128
908 case 27: 1129 case MISC_FLOAT64:
909 { 1130 {
910 uint64_t fp; 1131 uint64_t fp;
911 WANT (8); 1132 WANT (8);
912 memcpy (&fp, dec->cur, 8); 1133 memcpy (&fp, dec->cur, 8);
913 dec->cur += 8; 1134 dec->cur += 8;
916 fp = ecb_bswap64 (fp); 1137 fp = ecb_bswap64 (fp);
917 1138
918 return newSVnv (ecb_binary64_to_double (fp)); 1139 return newSVnv (ecb_binary64_to_double (fp));
919 } 1140 }
920 1141
921 // 0..19 unassigned 1142 // 0..19 unassigned simple
922 // 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
923 default: 1146 default:
924 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)"); 1147 ERR ("corrupted CBOR data (reserved/unassigned/unexpected major 7 value)");
925 } 1148 }
926 1149
927 break; 1150 break;
928 } 1151 }
929 1152
956 if (dec.cur != dec.end && !dec.err) 1179 if (dec.cur != dec.end && !dec.err)
957 dec.err = "garbage after CBOR object"; 1180 dec.err = "garbage after CBOR object";
958 1181
959 if (dec.err) 1182 if (dec.err)
960 { 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
961 SvREFCNT_dec (sv); 1195 SvREFCNT_dec (sv);
962 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);
963 } 1197 }
964 1198
965 sv = sv_2mortal (sv); 1199 sv = sv_2mortal (sv);
966 1200
967 return sv; 1201 return sv;
968} 1202}
969 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
970///////////////////////////////////////////////////////////////////////////// 1324/////////////////////////////////////////////////////////////////////////////
971// XS interface functions 1325// XS interface functions
972 1326
973MODULE = CBOR::XS PACKAGE = CBOR::XS 1327MODULE = CBOR::XS PACKAGE = CBOR::XS
974 1328
981 types_error_stash = gv_stashpv ("Types::Serialiser::Error" , 1); 1335 types_error_stash = gv_stashpv ("Types::Serialiser::Error" , 1);
982 1336
983 types_true = get_bool ("Types::Serialiser::true" ); 1337 types_true = get_bool ("Types::Serialiser::true" );
984 types_false = get_bool ("Types::Serialiser::false"); 1338 types_false = get_bool ("Types::Serialiser::false");
985 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);
986 1342
987 sv_cbor = newSVpv ("CBOR", 0); 1343 sv_cbor = newSVpv ("CBOR", 0);
988 SvREADONLY_on (sv_cbor); 1344 SvREADONLY_on (sv_cbor);
989} 1345}
990 1346
1012void shrink (CBOR *self, int enable = 1) 1368void shrink (CBOR *self, int enable = 1)
1013 ALIAS: 1369 ALIAS:
1014 shrink = F_SHRINK 1370 shrink = F_SHRINK
1015 allow_unknown = F_ALLOW_UNKNOWN 1371 allow_unknown = F_ALLOW_UNKNOWN
1016 allow_sharing = F_ALLOW_SHARING 1372 allow_sharing = F_ALLOW_SHARING
1017 dedup_keys = F_DEDUP_KEYS 1373 allow_cycles = F_ALLOW_CYCLES
1018 dedup_strings = F_DEDUP_STRINGS 1374 pack_strings = F_PACK_STRINGS
1375 validate_utf8 = F_VALIDATE_UTF8
1019 PPCODE: 1376 PPCODE:
1020{ 1377{
1021 if (enable) 1378 if (enable)
1022 self->flags |= ix; 1379 self->flags |= ix;
1023 else 1380 else
1029void get_shrink (CBOR *self) 1386void get_shrink (CBOR *self)
1030 ALIAS: 1387 ALIAS:
1031 get_shrink = F_SHRINK 1388 get_shrink = F_SHRINK
1032 get_allow_unknown = F_ALLOW_UNKNOWN 1389 get_allow_unknown = F_ALLOW_UNKNOWN
1033 get_allow_sharing = F_ALLOW_SHARING 1390 get_allow_sharing = F_ALLOW_SHARING
1034 get_dedup_keys = F_DEDUP_KEYS 1391 get_allow_cycles = F_ALLOW_CYCLES
1035 get_dedup_strings = F_DEDUP_STRINGS 1392 get_pack_strings = F_PACK_STRINGS
1393 get_validate_utf8 = F_VALIDATE_UTF8
1036 PPCODE: 1394 PPCODE:
1037 XPUSHs (boolSV (self->flags & ix)); 1395 XPUSHs (boolSV (self->flags & ix));
1038 1396
1039void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) 1397void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
1040 PPCODE: 1398 PPCODE:
1053 XPUSHs (ST (0)); 1411 XPUSHs (ST (0));
1054 1412
1055int get_max_size (CBOR *self) 1413int get_max_size (CBOR *self)
1056 CODE: 1414 CODE:
1057 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);
1058 OUTPUT: 1428 OUTPUT:
1059 RETVAL 1429 RETVAL
1060 1430
1061void encode (CBOR *self, SV *scalar) 1431void encode (CBOR *self, SV *scalar)
1062 PPCODE: 1432 PPCODE:
1077 EXTEND (SP, 2); 1447 EXTEND (SP, 2);
1078 PUSHs (sv); 1448 PUSHs (sv);
1079 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); 1449 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
1080} 1450}
1081 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
1082PROTOTYPES: ENABLE 1508PROTOTYPES: ENABLE
1083 1509
1084void encode_cbor (SV *scalar) 1510void encode_cbor (SV *scalar)
1511 ALIAS:
1512 encode_cbor = 0
1513 encode_cbor_sharing = F_ALLOW_SHARING
1085 PPCODE: 1514 PPCODE:
1086{ 1515{
1087 CBOR cbor; 1516 CBOR cbor;
1088 cbor_init (&cbor); 1517 cbor_init (&cbor);
1518 cbor.flags |= ix;
1089 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN; 1519 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;
1090 XPUSHs (scalar); 1520 XPUSHs (scalar);
1091} 1521}
1092 1522
1093void decode_cbor (SV *cborstr) 1523void decode_cbor (SV *cborstr)

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines