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 1.18 by root, Wed Nov 20 01:09:46 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
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_SHARABLE = 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 *sharable; // ptr => index, or 0 211 HV *shareable; // ptr => index, or 0
142 HV *sharable_idx; 212 UV shareable_idx;
143} enc_t; 213} enc_t;
144 214
145ecb_inline void 215ecb_inline void
146need (enc_t *enc, STRLEN len) 216need (enc_t *enc, STRLEN len)
147{ 217{
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
214encode_tag (enc_t *enc, UV tag)
215{
216 encode_uint (enc, 0xc0, tag);
217}
218
219static int 289static void
220encode_sharable2 (enc_t *enc, SV *sv) 290encode_strref (enc_t *enc, int utf8, char *str, STRLEN len)
221{ 291{
222 if (!enc->sharable) 292 if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS))
223 enc->sharable = (HV *)sv_2mortal ((SV *)newHV ()); 293 {
294 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1);
224 295
225 SV **svp = hv_fetch (enc->sharable, &sv, sizeof (sv), 1);
226
227 if (SvOK (*svp)) 296 if (SvOK (*svp))
228 { 297 {
298 // already registered, use stringref
229 encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF); 299 encode_tag (enc, CBOR_TAG_STRINGREF);
230 encode_uint (enc, 0x00, SvUV (*svp)); 300 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
231
232 return 1; 301 return;
233 } 302 }
234 else 303 else if (len >= minimum_string_length (enc->stringref_idx))
235 { 304 {
305 // register only
236 sv_setuv (*svp, enc->sharable_idx++); 306 sv_setuv (*svp, enc->stringref_idx);
237 encode_tag (enc, CBOR_TAG_VALUE_SHARABLE); 307 ++enc->stringref_idx;
238 308 }
239 return 0;
240 } 309 }
241}
242 310
243ecb_inline int 311 encode_str (enc, utf8, str, len);
244encode_sharable (enc_t *enc, SV *sv)
245{
246 if (ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING)
247 && ecb_expect_false (SvREFCNT (sv) > 1))
248 return encode_sharable2 (enc, sv);
249
250 return 0;
251} 312}
252 313
253static void encode_sv (enc_t *enc, SV *sv); 314static void encode_sv (enc_t *enc, SV *sv);
254 315
255static void 316static void
260 if (enc->depth >= enc->cbor.max_depth) 321 if (enc->depth >= enc->cbor.max_depth)
261 croak (ERR_NESTING_EXCEEDED); 322 croak (ERR_NESTING_EXCEEDED);
262 323
263 ++enc->depth; 324 ++enc->depth;
264 325
265 encode_uint (enc, 0x80, len + 1); 326 encode_uint (enc, MAJOR_ARRAY, len + 1);
266 327
267 for (i = 0; i <= len; ++i) 328 for (i = 0; i <= len; ++i)
268 { 329 {
269 SV **svp = av_fetch (av, i, 0); 330 SV **svp = av_fetch (av, i, 0);
270 encode_sv (enc, svp ? *svp : &PL_sv_undef); 331 encode_sv (enc, svp ? *svp : &PL_sv_undef);
285 346
286 int pairs = hv_iterinit (hv); 347 int pairs = hv_iterinit (hv);
287 int mg = SvMAGICAL (hv); 348 int mg = SvMAGICAL (hv);
288 349
289 if (mg) 350 if (mg)
290 encode_ch (enc, 0xa0 | 31); 351 encode_ch (enc, MAJOR_MAP | MINOR_INDEF);
291 else 352 else
292 encode_uint (enc, 0xa0, pairs); 353 encode_uint (enc, MAJOR_MAP, pairs);
293 354
294 while ((he = hv_iternext (hv))) 355 while ((he = hv_iternext (hv)))
295 { 356 {
296 if (HeKLEN (he) == HEf_SVKEY) 357 if (HeKLEN (he) == HEf_SVKEY)
297 encode_sv (enc, HeSVKEY (he)); 358 encode_sv (enc, HeSVKEY (he));
298 else 359 else
299 encode_str (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he)); 360 encode_strref (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he));
300 361
301 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));
302 } 363 }
303 364
304 if (mg) 365 if (mg)
305 encode_ch (enc, 0xe0 | 31); 366 encode_ch (enc, MAJOR_MISC | MINOR_INDEF);
306 367
307 --enc->depth; 368 --enc->depth;
308} 369}
309 370
310// encode objects, arrays and special \0=false and \1=true values. 371// encode objects, arrays and special \0=false and \1=true values.
311static void 372static void
312encode_rv (enc_t *enc, SV *sv) 373encode_rv (enc_t *enc, SV *sv)
313{ 374{
314 svtype svt;
315
316 SvGETMAGIC (sv); 375 SvGETMAGIC (sv);
376
317 svt = SvTYPE (sv); 377 svtype svt = SvTYPE (sv);
318
319 if (encode_sharable (enc, sv))
320 return;
321 378
322 if (ecb_expect_false (SvOBJECT (sv))) 379 if (ecb_expect_false (SvOBJECT (sv)))
323 { 380 {
324 HV *boolean_stash = !CBOR_SLOW || types_boolean_stash 381 HV *boolean_stash = !CBOR_SLOW || types_boolean_stash
325 ? types_boolean_stash 382 ? types_boolean_stash
330 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 387 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
331 ? cbor_tagged_stash 388 ? cbor_tagged_stash
332 : gv_stashpv ("CBOR::XS::Tagged" , 1); 389 : gv_stashpv ("CBOR::XS::Tagged" , 1);
333 390
334 HV *stash = SvSTASH (sv); 391 HV *stash = SvSTASH (sv);
335 GV *method;
336 392
337 if (stash == boolean_stash) 393 if (stash == boolean_stash)
338 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 }
339 else if (stash == error_stash) 398 else if (stash == error_stash)
340 encode_ch (enc, 0xe0 | 23); 399 {
400 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
401 return;
402 }
341 else if (stash == tagged_stash) 403 else if (stash == tagged_stash)
342 { 404 {
343 if (svt != SVt_PVAV) 405 if (svt != SVt_PVAV)
344 croak ("encountered CBOR::XS::Tagged object that isn't an array"); 406 croak ("encountered CBOR::XS::Tagged object that isn't an array");
345 407
346 encode_uint (enc, 0xc0, SvUV (*av_fetch ((AV *)sv, 0, 1))); 408 encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1)));
347 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))
348 } 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
349 else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0))) 442 if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
350 { 443 {
351 dSP; 444 dSP;
352 445
353 ENTER; SAVETMPS; PUSHMARK (SP); 446 ENTER; SAVETMPS; PUSHMARK (SP);
354 // 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
386 // catch this surprisingly common error 479 // catch this surprisingly common error
387 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv) 480 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)
388 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));
389 482
390 encode_tag (enc, CBOR_TAG_PERL_OBJECT); 483 encode_tag (enc, CBOR_TAG_PERL_OBJECT);
391 encode_uint (enc, 0x80, count + 1); 484 encode_uint (enc, MAJOR_ARRAY, count + 1);
392 encode_str (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash)); 485 encode_strref (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
393 486
394 while (count) 487 while (count)
395 encode_sv (enc, SP[1 - count--]); 488 encode_sv (enc, SP[1 - count--]);
396 489
397 PUTBACK; 490 PUTBACK;
418{ 511{
419 double nv = SvNVX (sv); 512 double nv = SvNVX (sv);
420 513
421 need (enc, 9); 514 need (enc, 9);
422 515
423 if (ecb_expect_false (nv == (U32)nv)) 516 if (ecb_expect_false (nv == (NV)(U32)nv))
424 encode_uint (enc, 0x00, (U32)nv); 517 encode_uint (enc, MAJOR_POS_INT, (U32)nv);
425 //TODO: maybe I32? 518 //TODO: maybe I32?
426 else if (ecb_expect_false (nv == (float)nv)) 519 else if (ecb_expect_false (nv == (float)nv))
427 { 520 {
428 uint32_t fp = ecb_float_to_binary32 (nv); 521 uint32_t fp = ecb_float_to_binary32 (nv);
429 522
430 *enc->cur++ = 0xe0 | 26; 523 *enc->cur++ = MAJOR_MISC | MISC_FLOAT32;
431 524
432 if (!ecb_big_endian ()) 525 if (!ecb_big_endian ())
433 fp = ecb_bswap32 (fp); 526 fp = ecb_bswap32 (fp);
434 527
435 memcpy (enc->cur, &fp, 4); 528 memcpy (enc->cur, &fp, 4);
437 } 530 }
438 else 531 else
439 { 532 {
440 uint64_t fp = ecb_double_to_binary64 (nv); 533 uint64_t fp = ecb_double_to_binary64 (nv);
441 534
442 *enc->cur++ = 0xe0 | 27; 535 *enc->cur++ = MAJOR_MISC | MISC_FLOAT64;
443 536
444 if (!ecb_big_endian ()) 537 if (!ecb_big_endian ())
445 fp = ecb_bswap64 (fp); 538 fp = ecb_bswap64 (fp);
446 539
447 memcpy (enc->cur, &fp, 8); 540 memcpy (enc->cur, &fp, 8);
452static void 545static void
453encode_sv (enc_t *enc, SV *sv) 546encode_sv (enc_t *enc, SV *sv)
454{ 547{
455 SvGETMAGIC (sv); 548 SvGETMAGIC (sv);
456 549
457 if (encode_sharable (enc, sv))
458 return;
459
460 if (SvPOKp (sv)) 550 if (SvPOKp (sv))
461 { 551 {
462 STRLEN len; 552 STRLEN len;
463 char *str = SvPV (sv, len); 553 char *str = SvPV (sv, len);
464 encode_str (enc, SvUTF8 (sv), str, len); 554 encode_strref (enc, SvUTF8 (sv), str, len);
465 } 555 }
466 else if (SvNOKp (sv)) 556 else if (SvNOKp (sv))
467 encode_nv (enc, sv); 557 encode_nv (enc, sv);
468 else if (SvIOKp (sv)) 558 else if (SvIOKp (sv))
469 { 559 {
470 if (SvIsUV (sv)) 560 if (SvIsUV (sv))
471 encode_uint (enc, 0x00, SvUVX (sv)); 561 encode_uint (enc, MAJOR_POS_INT, SvUVX (sv));
472 else if (SvIVX (sv) >= 0) 562 else if (SvIVX (sv) >= 0)
473 encode_uint (enc, 0x00, SvIVX (sv)); 563 encode_uint (enc, MAJOR_POS_INT, SvIVX (sv));
474 else 564 else
475 encode_uint (enc, 0x20, -(SvIVX (sv) + 1)); 565 encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1));
476 } 566 }
477 else if (SvROK (sv)) 567 else if (SvROK (sv))
478 encode_rv (enc, SvRV (sv)); 568 encode_rv (enc, SvRV (sv));
479 else if (!SvOK (sv)) 569 else if (!SvOK (sv))
480 encode_ch (enc, 0xe0 | 22); 570 encode_ch (enc, MAJOR_MISC | SIMPLE_NULL);
481 else if (enc->cbor.flags & F_ALLOW_UNKNOWN) 571 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
482 encode_ch (enc, 0xe0 | 23); 572 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
483 else 573 else
484 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",
485 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv)); 575 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
486} 576}
487 577
494 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 584 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
495 enc.cur = SvPVX (enc.sv); 585 enc.cur = SvPVX (enc.sv);
496 enc.end = SvEND (enc.sv); 586 enc.end = SvEND (enc.sv);
497 587
498 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
499 encode_sv (&enc, scalar); 597 encode_sv (&enc, scalar);
500 598
501 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 599 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
502 *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
503 601
517 U8 *end; // end of input string 615 U8 *end; // end of input string
518 const char *err; // parse error, if != 0 616 const char *err; // parse error, if != 0
519 CBOR cbor; 617 CBOR cbor;
520 U32 depth; // recursion depth 618 U32 depth; // recursion depth
521 U32 maxdepth; // recursion depth limit 619 U32 maxdepth; // recursion depth limit
522 AV *sharable; 620 AV *shareable;
621 AV *stringref;
622 SV *decode_tagged;
523} dec_t; 623} dec_t;
524 624
525#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
526 626
527#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")
530#define DEC_DEC_DEPTH --dec->depth 630#define DEC_DEC_DEPTH --dec->depth
531 631
532static UV 632static UV
533decode_uint (dec_t *dec) 633decode_uint (dec_t *dec)
534{ 634{
535 switch (*dec->cur & 31) 635 U8 m = *dec->cur & MINOR_MASK;
536 { 636 ++dec->cur;
537 case 0: case 1: case 2: case 3: case 4: case 5: case 6: case 7:
538 case 8: case 9: case 10: case 11: case 12: case 13: case 14: case 15:
539 case 16: case 17: case 18: case 19: case 20: case 21: case 22: case 23:
540 return *dec->cur++ & 31;
541 637
542 case 24: 638 if (ecb_expect_true (m < LENGTH_EXT1))
639 return m;
640 else if (ecb_expect_true (m == LENGTH_EXT1))
641 {
543 WANT (2); 642 WANT (1);
544 dec->cur += 2; 643 dec->cur += 1;
545 return dec->cur[-1]; 644 return dec->cur[-1];
546 645 }
547 case 25: 646 else if (ecb_expect_true (m == LENGTH_EXT2))
647 {
548 WANT (3); 648 WANT (2);
549 dec->cur += 3; 649 dec->cur += 2;
550 return (((UV)dec->cur[-2]) << 8) 650 return (((UV)dec->cur[-2]) << 8)
551 | ((UV)dec->cur[-1]); 651 | ((UV)dec->cur[-1]);
552 652 }
553 case 26: 653 else if (ecb_expect_true (m == LENGTH_EXT4))
654 {
554 WANT (5); 655 WANT (4);
555 dec->cur += 5; 656 dec->cur += 4;
556 return (((UV)dec->cur[-4]) << 24) 657 return (((UV)dec->cur[-4]) << 24)
557 | (((UV)dec->cur[-3]) << 16) 658 | (((UV)dec->cur[-3]) << 16)
558 | (((UV)dec->cur[-2]) << 8) 659 | (((UV)dec->cur[-2]) << 8)
559 | ((UV)dec->cur[-1]); 660 | ((UV)dec->cur[-1]);
560 661 }
561 case 27: 662 else if (ecb_expect_true (m == LENGTH_EXT8))
663 {
562 WANT (9); 664 WANT (8);
563 dec->cur += 9; 665 dec->cur += 8;
666
667 return
668#if UVSIZE < 8
669 0
670#else
564 return (((UV)dec->cur[-8]) << 56) 671 (((UV)dec->cur[-8]) << 56)
565 | (((UV)dec->cur[-7]) << 48) 672 | (((UV)dec->cur[-7]) << 48)
566 | (((UV)dec->cur[-6]) << 40) 673 | (((UV)dec->cur[-6]) << 40)
567 | (((UV)dec->cur[-5]) << 32) 674 | (((UV)dec->cur[-5]) << 32)
675#endif
568 | (((UV)dec->cur[-4]) << 24) 676 | (((UV)dec->cur[-4]) << 24)
569 | (((UV)dec->cur[-3]) << 16) 677 | (((UV)dec->cur[-3]) << 16)
570 | (((UV)dec->cur[-2]) << 8) 678 | (((UV)dec->cur[-2]) << 8)
571 | ((UV)dec->cur[-1]); 679 | ((UV)dec->cur[-1]);
572 680 }
573 default: 681 else
574 ERR ("corrupted CBOR data (unsupported integer minor encoding)"); 682 ERR ("corrupted CBOR data (unsupported integer minor encoding)");
575 }
576 683
577fail: 684fail:
578 return 0; 685 return 0;
579} 686}
580 687
585{ 692{
586 AV *av = newAV (); 693 AV *av = newAV ();
587 694
588 DEC_INC_DEPTH; 695 DEC_INC_DEPTH;
589 696
590 if ((*dec->cur & 31) == 31) 697 if (*dec->cur == (MAJOR_ARRAY | MINOR_INDEF))
591 { 698 {
592 ++dec->cur; 699 ++dec->cur;
593 700
594 for (;;) 701 for (;;)
595 { 702 {
596 WANT (1); 703 WANT (1);
597 704
598 if (*dec->cur == (0xe0 | 31)) 705 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
599 { 706 {
600 ++dec->cur; 707 ++dec->cur;
601 break; 708 break;
602 } 709 }
603 710
606 } 713 }
607 else 714 else
608 { 715 {
609 int i, len = decode_uint (dec); 716 int i, len = decode_uint (dec);
610 717
718 WANT (len); // complexity check for av_fill - need at least one byte per value, do not allow supersize arrays
611 av_fill (av, len - 1); 719 av_fill (av, len - 1);
612 720
613 for (i = 0; i < len; ++i) 721 for (i = 0; i < len; ++i)
614 AvARRAY (av)[i] = decode_sv (dec); 722 AvARRAY (av)[i] = decode_sv (dec);
615 } 723 }
625 733
626static void 734static void
627decode_he (dec_t *dec, HV *hv) 735decode_he (dec_t *dec, HV *hv)
628{ 736{
629 // for speed reasons, we specialcase single-string 737 // for speed reasons, we specialcase single-string
630 // byte or utf-8 strings as keys. 738 // byte or utf-8 strings as keys, but only when !stringref
631 739
632 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))
633 { 742 {
634 I32 len = decode_uint (dec); 743 I32 len = decode_uint (dec);
635 char *key = (char *)dec->cur; 744 char *key = (char *)dec->cur;
636 745
637 dec->cur += len; 746 dec->cur += len;
638 747
639 hv_store (hv, key, len, decode_sv (dec), 0); 748 hv_store (hv, key, len, decode_sv (dec), 0);
749
750 return;
640 } 751 }
641 else if (*dec->cur >= 0x60 && *dec->cur <= 0x60 + 27) 752 else if (ecb_expect_true ((*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8))
642 { 753 {
643 I32 len = decode_uint (dec); 754 I32 len = decode_uint (dec);
644 char *key = (char *)dec->cur; 755 char *key = (char *)dec->cur;
645 756
646 dec->cur += len; 757 dec->cur += len;
647 758
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
648 hv_store (hv, key, -len, decode_sv (dec), 0); 763 hv_store (hv, key, -len, decode_sv (dec), 0);
764
765 return;
649 } 766 }
650 else 767
651 {
652 SV *k = decode_sv (dec); 768 SV *k = decode_sv (dec);
653 SV *v = decode_sv (dec); 769 SV *v = decode_sv (dec);
654 770
655 hv_store_ent (hv, k, v, 0); 771 hv_store_ent (hv, k, v, 0);
656 SvREFCNT_dec (k); 772 SvREFCNT_dec (k);
657 } 773
774fail:
775 ;
658} 776}
659 777
660static SV * 778static SV *
661decode_hv (dec_t *dec) 779decode_hv (dec_t *dec)
662{ 780{
663 HV *hv = newHV (); 781 HV *hv = newHV ();
664 782
665 DEC_INC_DEPTH; 783 DEC_INC_DEPTH;
666 784
667 if ((*dec->cur & 31) == 31) 785 if (*dec->cur == (MAJOR_MAP | MINOR_INDEF))
668 { 786 {
669 ++dec->cur; 787 ++dec->cur;
670 788
671 for (;;) 789 for (;;)
672 { 790 {
673 WANT (1); 791 WANT (1);
674 792
675 if (*dec->cur == (0xe0 | 31)) 793 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
676 { 794 {
677 ++dec->cur; 795 ++dec->cur;
678 break; 796 break;
679 } 797 }
680 798
701static SV * 819static SV *
702decode_str (dec_t *dec, int utf8) 820decode_str (dec_t *dec, int utf8)
703{ 821{
704 SV *sv = 0; 822 SV *sv = 0;
705 823
706 if ((*dec->cur & 31) == 31) 824 if ((*dec->cur & MINOR_MASK) == MINOR_INDEF)
707 { 825 {
826 // indefinite length strings
708 ++dec->cur; 827 ++dec->cur;
709 828
829 U8 major = *dec->cur & MAJOR_MISC;
830
710 sv = newSVpvn ("", 0); 831 sv = newSVpvn ("", 0);
711 832
712 // not very fast, and certainly not robust against illegal input
713 for (;;) 833 for (;;)
714 { 834 {
715 WANT (1); 835 WANT (1);
716 836
717 if (*dec->cur == (0xe0 | 31)) 837 if ((*dec->cur - major) > LENGTH_EXT8)
838 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
718 { 839 {
719 ++dec->cur; 840 ++dec->cur;
720 break; 841 break;
721 } 842 }
843 else
844 ERR ("corrupted CBOR data (invalid chunks in indefinite length string)");
722 845
723 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;
724 } 851 }
725 } 852 }
726 else 853 else
727 { 854 {
728 STRLEN len = decode_uint (dec); 855 STRLEN len = decode_uint (dec);
729 856
730 WANT (len); 857 WANT (len);
731 sv = newSVpvn (dec->cur, len); 858 sv = newSVpvn (dec->cur, len);
732 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));
733 } 864 }
734 865
735 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
736 SvUTF8_on (sv); 872 SvUTF8_on (sv);
873 }
737 874
738 return sv; 875 return sv;
739 876
740fail: 877fail:
741 SvREFCNT_dec (sv); 878 SvREFCNT_dec (sv);
743} 880}
744 881
745static SV * 882static SV *
746decode_tagged (dec_t *dec) 883decode_tagged (dec_t *dec)
747{ 884{
885 SV *sv = 0;
748 UV tag = decode_uint (dec); 886 UV tag = decode_uint (dec);
749 SV *sv = decode_sv (dec); 887
888 WANT (1);
750 889
751 switch (tag) 890 switch (tag)
752 { 891 {
753 case CBOR_TAG_MAGIC: 892 case CBOR_TAG_MAGIC:
754 return sv; 893 sv = decode_sv (dec);
894 break;
755 895
756 case CBOR_TAG_INDIRECTION: 896 case CBOR_TAG_INDIRECTION:
757 return newRV_noinc (sv); 897 sv = newRV_noinc (decode_sv (dec));
898 break;
758 899
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;
926
759 case CBOR_TAG_VALUE_SHARABLE: 927 case CBOR_TAG_VALUE_SHAREABLE:
928 {
760 if (ecb_expect_false (!dec->sharable)) 929 if (ecb_expect_false (!dec->shareable))
761 dec->sharable = (AV *)sv_2mortal ((SV *)newAV ()); 930 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ());
762 931
932 if (dec->cbor.flags & F_ALLOW_CYCLES)
933 {
934 sv = newSV (0);
763 av_push (dec->sharable, SvREFCNT_inc_NN (sv)); 935 av_push (dec->shareable, SvREFCNT_inc_NN (sv));
764 936
765 return sv; 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;
766 950
767 case CBOR_TAG_VALUE_SHAREDREF: 951 case CBOR_TAG_VALUE_SHAREDREF:
768 { 952 {
769 // TODO: should verify that the sv atcually was a CBOR unsigned integer 953 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
770 UV idx = SvUV (sv); 954 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
771 955
956 UV idx = decode_uint (dec);
957
772 if (!dec->sharable || idx > AvFILLp (dec->sharable)) 958 if (!dec->shareable || (int)idx > AvFILLp (dec->shareable))
773 ERR ("corrupted CBOR data (sharedref index out of bounds)"); 959 ERR ("corrupted CBOR data (sharedref index out of bounds)");
774 960
775 SvREFCNT_dec (sv);
776
777 return SvREFCNT_inc_NN (AvARRAY (dec->sharable)[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");
778 } 965 }
966 break;
779 967
780 case CBOR_TAG_PERL_OBJECT: 968 case CBOR_TAG_PERL_OBJECT:
781 { 969 {
970 sv = decode_sv (dec);
971
782 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV) 972 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
783 ERR ("corrupted CBOR data (non-array perl object)"); 973 ERR ("corrupted CBOR data (non-array perl object)");
784 974
785 AV *av = (AV *)SvRV (sv); 975 AV *av = (AV *)SvRV (sv);
786 int len = av_len (av) + 1; 976 int len = av_len (av) + 1;
821 sv = SvREFCNT_inc (POPs); 1011 sv = SvREFCNT_inc (POPs);
822 1012
823 PUTBACK; 1013 PUTBACK;
824 1014
825 FREETMPS; LEAVE; 1015 FREETMPS; LEAVE;
826
827 return sv;
828 } 1016 }
1017 break;
829 1018
830 default: 1019 default:
831 { 1020 {
1021 sv = decode_sv (dec);
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 {
832 AV *av = newAV (); 1046 AV *av = newAV ();
833 av_push (av, newSVuv (tag)); 1047 av_push (av, newSVuv (tag));
834 av_push (av, sv); 1048 av_push (av, sv);
835 1049
836 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 1050 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
837 ? cbor_tagged_stash 1051 ? cbor_tagged_stash
838 : gv_stashpv ("CBOR::XS::Tagged" , 1); 1052 : gv_stashpv ("CBOR::XS::Tagged" , 1);
839
840 return sv_bless (newRV_noinc ((SV *)av), tagged_stash); 1053 sv = sv_bless (newRV_noinc ((SV *)av), tagged_stash);
841 } 1054 }
1055
1056 PUTBACK;
1057
1058 FREETMPS; LEAVE;
1059 }
1060 break;
842 } 1061 }
1062
1063 return sv;
843 1064
844fail: 1065fail:
845 SvREFCNT_dec (sv); 1066 SvREFCNT_dec (sv);
846 return &PL_sv_undef; 1067 return &PL_sv_undef;
847} 1068}
849static SV * 1070static SV *
850decode_sv (dec_t *dec) 1071decode_sv (dec_t *dec)
851{ 1072{
852 WANT (1); 1073 WANT (1);
853 1074
854 switch (*dec->cur >> 5) 1075 switch (*dec->cur >> MAJOR_SHIFT)
855 { 1076 {
856 case 0: // unsigned int 1077 case MAJOR_POS_INT >> MAJOR_SHIFT: return newSVuv (decode_uint (dec));
857 return newSVuv (decode_uint (dec)); 1078 case MAJOR_NEG_INT >> MAJOR_SHIFT: return newSViv (-1 - (IV)decode_uint (dec));
858 case 1: // negative int 1079 case MAJOR_BYTES >> MAJOR_SHIFT: return decode_str (dec, 0);
859 return newSViv (-1 - (IV)decode_uint (dec)); 1080 case MAJOR_TEXT >> MAJOR_SHIFT: return decode_str (dec, 1);
860 case 2: // octet string 1081 case MAJOR_ARRAY >> MAJOR_SHIFT: return decode_av (dec);
861 return decode_str (dec, 0); 1082 case MAJOR_MAP >> MAJOR_SHIFT: return decode_hv (dec);
862 case 3: // utf-8 string 1083 case MAJOR_TAG >> MAJOR_SHIFT: return decode_tagged (dec);
863 return decode_str (dec, 1); 1084
864 case 4: // array 1085 case MAJOR_MISC >> MAJOR_SHIFT:
865 return decode_av (dec);
866 case 5: // map
867 return decode_hv (dec);
868 case 6: // tag
869 return decode_tagged (dec);
870 case 7: // misc
871 switch (*dec->cur++ & 31) 1086 switch (*dec->cur++ & MINOR_MASK)
872 { 1087 {
873 case 20: 1088 case SIMPLE_FALSE:
874#if CBOR_SLOW 1089#if CBOR_SLOW
875 types_false = get_bool ("Types::Serialiser::false"); 1090 types_false = get_bool ("Types::Serialiser::false");
876#endif 1091#endif
877 return newSVsv (types_false); 1092 return newSVsv (types_false);
878 case 21: 1093 case SIMPLE_TRUE:
879#if CBOR_SLOW 1094#if CBOR_SLOW
880 types_true = get_bool ("Types::Serialiser::true"); 1095 types_true = get_bool ("Types::Serialiser::true");
881#endif 1096#endif
882 return newSVsv (types_true); 1097 return newSVsv (types_true);
883 case 22: 1098 case SIMPLE_NULL:
884 return newSVsv (&PL_sv_undef); 1099 return newSVsv (&PL_sv_undef);
885 case 23: 1100 case SIMPLE_UNDEF:
886#if CBOR_SLOW 1101#if CBOR_SLOW
887 types_error = get_bool ("Types::Serialiser::error"); 1102 types_error = get_bool ("Types::Serialiser::error");
888#endif 1103#endif
889 return newSVsv (types_error); 1104 return newSVsv (types_error);
890 1105
891 case 25: 1106 case MISC_FLOAT16:
892 { 1107 {
893 WANT (2); 1108 WANT (2);
894 1109
895 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1]; 1110 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1];
896 dec->cur += 2; 1111 dec->cur += 2;
897 1112
898 return newSVnv (ecb_binary16_to_float (fp)); 1113 return newSVnv (ecb_binary16_to_float (fp));
899 } 1114 }
900 1115
901 case 26: 1116 case MISC_FLOAT32:
902 { 1117 {
903 uint32_t fp; 1118 uint32_t fp;
904 WANT (4); 1119 WANT (4);
905 memcpy (&fp, dec->cur, 4); 1120 memcpy (&fp, dec->cur, 4);
906 dec->cur += 4; 1121 dec->cur += 4;
909 fp = ecb_bswap32 (fp); 1124 fp = ecb_bswap32 (fp);
910 1125
911 return newSVnv (ecb_binary32_to_float (fp)); 1126 return newSVnv (ecb_binary32_to_float (fp));
912 } 1127 }
913 1128
914 case 27: 1129 case MISC_FLOAT64:
915 { 1130 {
916 uint64_t fp; 1131 uint64_t fp;
917 WANT (8); 1132 WANT (8);
918 memcpy (&fp, dec->cur, 8); 1133 memcpy (&fp, dec->cur, 8);
919 dec->cur += 8; 1134 dec->cur += 8;
922 fp = ecb_bswap64 (fp); 1137 fp = ecb_bswap64 (fp);
923 1138
924 return newSVnv (ecb_binary64_to_double (fp)); 1139 return newSVnv (ecb_binary64_to_double (fp));
925 } 1140 }
926 1141
927 // 0..19 unassigned 1142 // 0..19 unassigned simple
928 // 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
929 default: 1146 default:
930 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)"); 1147 ERR ("corrupted CBOR data (reserved/unassigned/unexpected major 7 value)");
931 } 1148 }
932 1149
933 break; 1150 break;
934 } 1151 }
935 1152
962 if (dec.cur != dec.end && !dec.err) 1179 if (dec.cur != dec.end && !dec.err)
963 dec.err = "garbage after CBOR object"; 1180 dec.err = "garbage after CBOR object";
964 1181
965 if (dec.err) 1182 if (dec.err)
966 { 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
967 SvREFCNT_dec (sv); 1195 SvREFCNT_dec (sv);
968 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);
969 } 1197 }
970 1198
971 sv = sv_2mortal (sv); 1199 sv = sv_2mortal (sv);
972 1200
973 return sv; 1201 return sv;
974} 1202}
975 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
976///////////////////////////////////////////////////////////////////////////// 1324/////////////////////////////////////////////////////////////////////////////
977// XS interface functions 1325// XS interface functions
978 1326
979MODULE = CBOR::XS PACKAGE = CBOR::XS 1327MODULE = CBOR::XS PACKAGE = CBOR::XS
980 1328
987 types_error_stash = gv_stashpv ("Types::Serialiser::Error" , 1); 1335 types_error_stash = gv_stashpv ("Types::Serialiser::Error" , 1);
988 1336
989 types_true = get_bool ("Types::Serialiser::true" ); 1337 types_true = get_bool ("Types::Serialiser::true" );
990 types_false = get_bool ("Types::Serialiser::false"); 1338 types_false = get_bool ("Types::Serialiser::false");
991 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);
992 1342
993 sv_cbor = newSVpv ("CBOR", 0); 1343 sv_cbor = newSVpv ("CBOR", 0);
994 SvREADONLY_on (sv_cbor); 1344 SvREADONLY_on (sv_cbor);
995} 1345}
996 1346
1018void shrink (CBOR *self, int enable = 1) 1368void shrink (CBOR *self, int enable = 1)
1019 ALIAS: 1369 ALIAS:
1020 shrink = F_SHRINK 1370 shrink = F_SHRINK
1021 allow_unknown = F_ALLOW_UNKNOWN 1371 allow_unknown = F_ALLOW_UNKNOWN
1022 allow_sharing = F_ALLOW_SHARING 1372 allow_sharing = F_ALLOW_SHARING
1023 dedup_keys = F_DEDUP_KEYS 1373 allow_cycles = F_ALLOW_CYCLES
1024 dedup_strings = F_DEDUP_STRINGS 1374 pack_strings = F_PACK_STRINGS
1375 validate_utf8 = F_VALIDATE_UTF8
1025 PPCODE: 1376 PPCODE:
1026{ 1377{
1027 if (enable) 1378 if (enable)
1028 self->flags |= ix; 1379 self->flags |= ix;
1029 else 1380 else
1035void get_shrink (CBOR *self) 1386void get_shrink (CBOR *self)
1036 ALIAS: 1387 ALIAS:
1037 get_shrink = F_SHRINK 1388 get_shrink = F_SHRINK
1038 get_allow_unknown = F_ALLOW_UNKNOWN 1389 get_allow_unknown = F_ALLOW_UNKNOWN
1039 get_allow_sharing = F_ALLOW_SHARING 1390 get_allow_sharing = F_ALLOW_SHARING
1040 get_dedup_keys = F_DEDUP_KEYS 1391 get_allow_cycles = F_ALLOW_CYCLES
1041 get_dedup_strings = F_DEDUP_STRINGS 1392 get_pack_strings = F_PACK_STRINGS
1393 get_validate_utf8 = F_VALIDATE_UTF8
1042 PPCODE: 1394 PPCODE:
1043 XPUSHs (boolSV (self->flags & ix)); 1395 XPUSHs (boolSV (self->flags & ix));
1044 1396
1045void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) 1397void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
1046 PPCODE: 1398 PPCODE:
1059 XPUSHs (ST (0)); 1411 XPUSHs (ST (0));
1060 1412
1061int get_max_size (CBOR *self) 1413int get_max_size (CBOR *self)
1062 CODE: 1414 CODE:
1063 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);
1064 OUTPUT: 1428 OUTPUT:
1065 RETVAL 1429 RETVAL
1066 1430
1067void encode (CBOR *self, SV *scalar) 1431void encode (CBOR *self, SV *scalar)
1068 PPCODE: 1432 PPCODE:
1083 EXTEND (SP, 2); 1447 EXTEND (SP, 2);
1084 PUSHs (sv); 1448 PUSHs (sv);
1085 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); 1449 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
1086} 1450}
1087 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
1088PROTOTYPES: ENABLE 1508PROTOTYPES: ENABLE
1089 1509
1090void encode_cbor (SV *scalar) 1510void encode_cbor (SV *scalar)
1511 ALIAS:
1512 encode_cbor = 0
1513 encode_cbor_sharing = F_ALLOW_SHARING
1091 PPCODE: 1514 PPCODE:
1092{ 1515{
1093 CBOR cbor; 1516 CBOR cbor;
1094 cbor_init (&cbor); 1517 cbor_init (&cbor);
1518 cbor.flags |= ix;
1095 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN; 1519 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;
1096 XPUSHs (scalar); 1520 XPUSHs (scalar);
1097} 1521}
1098 1522
1099void decode_cbor (SV *cborstr) 1523void decode_cbor (SV *cborstr)

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines