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 1.15 by root, Tue Oct 29 18:37:31 2013 UTC vs.
Revision 1.38 by root, Sun Dec 1 14:30:52 2013 UTC

… …
19# define HvNAMELEN(hv) HvNAMELEN_get (hv) 19# define HvNAMELEN(hv) HvNAMELEN_get (hv)
20#endif 20#endif
21#ifndef HvNAMEUTF8 21#ifndef HvNAMEUTF8
22# define HvNAMEUTF8(hv) 0 22# define HvNAMEUTF8(hv) 0
23#endif 23#endif
24#ifndef SvREFCNT_dec_NN
25# define SvREFCNT_dec_NN(sv) SvREFCNT_dec (sv)
26#endif
27
28// known major and minor types
29enum cbor_type
30{
31 MAJOR_SHIFT = 5,
32 MINOR_MASK = 0x1f,
33
34 MAJOR_POS_INT = 0 << MAJOR_SHIFT,
35 MAJOR_NEG_INT = 1 << MAJOR_SHIFT,
36 MAJOR_BYTES = 2 << MAJOR_SHIFT,
37 MAJOR_TEXT = 3 << MAJOR_SHIFT,
38 MAJOR_ARRAY = 4 << MAJOR_SHIFT,
39 MAJOR_MAP = 5 << MAJOR_SHIFT,
40 MAJOR_TAG = 6 << MAJOR_SHIFT,
41 MAJOR_MISC = 7 << MAJOR_SHIFT,
42
43 // INT/STRING/ARRAY/MAP subtypes
44 LENGTH_EXT1 = 24,
45 LENGTH_EXT2 = 25,
46 LENGTH_EXT4 = 26,
47 LENGTH_EXT8 = 27,
48
49 // SIMPLE types (effectively MISC subtypes)
50 SIMPLE_FALSE = 20,
51 SIMPLE_TRUE = 21,
52 SIMPLE_NULL = 22,
53 SIMPLE_UNDEF = 23,
54
55 // MISC subtype (unused)
56 MISC_EXT1 = 24,
57 MISC_FLOAT16 = 25,
58 MISC_FLOAT32 = 26,
59 MISC_FLOAT64 = 27,
60
61 // BYTES/TEXT/ARRAY/MAP
62 MINOR_INDEF = 31,
63};
24 64
25// known tags 65// known tags
26enum cbor_tag 66enum cbor_tag
27{ 67{
28 // inofficial extensions (pending iana registration) 68 // extensions
29 CBOR_TAG_PERL_OBJECT = 256, 69 CBOR_TAG_STRINGREF = 25, // http://cbor.schmorp.de/stringref
30 CBOR_TAG_GENERIC_OBJECT = 257, 70 CBOR_TAG_PERL_OBJECT = 26, // http://cbor.schmorp.de/perl-object
71 CBOR_TAG_GENERIC_OBJECT = 27, // http://cbor.schmorp.de/generic-object
72 CBOR_TAG_VALUE_SHAREABLE = 28, // http://cbor.schmorp.de/value-sharing
73 CBOR_TAG_VALUE_SHAREDREF = 29, // http://cbor.schmorp.de/value-sharing
74 CBOR_TAG_STRINGREF_NAMESPACE = 256, // http://cbor.schmorp.de/stringref
75 CBOR_TAG_INDIRECTION = 22098, // http://cbor.schmorp.de/indirection
31 76
32 // rfc7049 77 // rfc7049
33 CBOR_TAG_DATETIME = 0, // rfc4287, utf-8 78 CBOR_TAG_DATETIME = 0, // rfc4287, utf-8
34 CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any 79 CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any
35 CBOR_TAG_POS_BIGNUM = 2, // byte string 80 CBOR_TAG_POS_BIGNUM = 2, // byte string
36 CBOR_TAG_NEG_BIGNUM = 3, // byte string 81 CBOR_TAG_NEG_BIGNUM = 3, // byte string
37 CBOR_TAG_DECIMAL = 4, // decimal fraction, array 82 CBOR_TAG_DECIMAL = 4, // decimal fraction, array
38 CBOR_TAG_BIGFLOAT = 5, // array 83 CBOR_TAG_BIGFLOAT = 5, // array
39 84
40 CBOR_TAG_CONV_B64U = 21, // base64url, any 85 CBOR_TAG_CONV_B64U = 21, // base64url, any
41 CBOR_TAG_CONV_B64 = 22, // base64, any 86 CBOR_TAG_CONV_B64 = 22, // base64, any
42 CBOR_TAG_CONV_HEX = 23, // base16, any 87 CBOR_TAG_CONV_HEX = 23, // base16, any
43 CBOR_TAG_CBOR = 24, // embedded cbor, byte string 88 CBOR_TAG_CBOR = 24, // embedded cbor, byte string
44 89
45 CBOR_TAG_URI = 32, // URI rfc3986, utf-8 90 CBOR_TAG_URI = 32, // URI rfc3986, utf-8
46 CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8 91 CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8
47 CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8 92 CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8
48 CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8 93 CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8
49 CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8 94 CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8
50 95
51 CBOR_TAG_MAGIC = 55799 // self-describe cbor 96 CBOR_TAG_MAGIC = 55799, // self-describe cbor
52}; 97};
53 98
54#define F_SHRINK 0x00000200UL 99#define F_SHRINK 0x00000001UL
55#define F_ALLOW_UNKNOWN 0x00002000UL 100#define F_ALLOW_UNKNOWN 0x00000002UL
101#define F_ALLOW_SHARING 0x00000004UL
102#define F_ALLOW_CYCLES 0x00000008UL
103#define F_PACK_STRINGS 0x00000010UL
104#define F_VALIDATE_UTF8 0x00000020UL
56 105
57#define INIT_SIZE 32 // initial scalar size to be allocated 106#define INIT_SIZE 32 // initial scalar size to be allocated
58 107
59#define SB do { 108#define SB do {
60#define SE } while (0) 109#define SE } while (0)
… …
72# define CBOR_SLOW 0 121# define CBOR_SLOW 0
73# define CBOR_STASH cbor_stash 122# define CBOR_STASH cbor_stash
74#endif 123#endif
75 124
76static HV *cbor_stash, *types_boolean_stash, *types_error_stash, *cbor_tagged_stash; // CBOR::XS:: 125static HV *cbor_stash, *types_boolean_stash, *types_error_stash, *cbor_tagged_stash; // CBOR::XS::
77static SV *types_true, *types_false, *types_error, *sv_cbor; 126static SV *types_true, *types_false, *types_error, *sv_cbor, *default_filter;
78 127
79typedef struct { 128typedef struct {
80 U32 flags; 129 U32 flags;
81 U32 max_depth; 130 U32 max_depth;
82 STRLEN max_size; 131 STRLEN max_size;
132 SV *filter;
83} CBOR; 133} CBOR;
84 134
85ecb_inline void 135ecb_inline void
86cbor_init (CBOR *cbor) 136cbor_init (CBOR *cbor)
87{ 137{
88 Zero (cbor, 1, CBOR); 138 Zero (cbor, 1, CBOR);
89 cbor->max_depth = 512; 139 cbor->max_depth = 512;
140}
141
142ecb_inline void
143cbor_free (CBOR *cbor)
144{
145 SvREFCNT_dec (cbor->filter);
90} 146}
91 147
92///////////////////////////////////////////////////////////////////////////// 148/////////////////////////////////////////////////////////////////////////////
93// utility functions 149// utility functions
94 150
… …
116 SvPV_renew (sv, SvCUR (sv) + 1); 172 SvPV_renew (sv, SvCUR (sv) + 1);
117#endif 173#endif
118 } 174 }
119} 175}
120 176
121///////////////////////////////////////////////////////////////////////////// 177// minimum length of a string to be registered for stringref
122// fp hell 178ecb_inline int
123 179minimum_string_length (UV idx)
124//TODO 180{
181 return idx > 23
182 ? idx > 0xffU
183 ? idx > 0xffffU
184 ? idx > 0xffffffffU
185 ? 11
186 : 7
187 : 5
188 : 4
189 : 3;
190}
125 191
126///////////////////////////////////////////////////////////////////////////// 192/////////////////////////////////////////////////////////////////////////////
127// encoder 193// encoder
128 194
129// structure used for encoding CBOR 195// structure used for encoding CBOR
… …
132 char *cur; // SvPVX (sv) + current output position 198 char *cur; // SvPVX (sv) + current output position
133 char *end; // SvEND (sv) 199 char *end; // SvEND (sv)
134 SV *sv; // result scalar 200 SV *sv; // result scalar
135 CBOR cbor; 201 CBOR cbor;
136 U32 depth; // recursion level 202 U32 depth; // recursion level
203 HV *stringref[2]; // string => index, or 0 ([0] = bytes, [1] = utf-8)
204 UV stringref_idx;
205 HV *shareable; // ptr => index, or 0
206 UV shareable_idx;
137} enc_t; 207} enc_t;
138 208
139ecb_inline void 209ecb_inline void
140need (enc_t *enc, STRLEN len) 210need (enc_t *enc, STRLEN len)
141{ 211{
… …
158static void 228static void
159encode_uint (enc_t *enc, int major, UV len) 229encode_uint (enc_t *enc, int major, UV len)
160{ 230{
161 need (enc, 9); 231 need (enc, 9);
162 232
163 if (len < 24) 233 if (ecb_expect_true (len < LENGTH_EXT1))
164 *enc->cur++ = major | len; 234 *enc->cur++ = major | len;
165 else if (len <= 0xff) 235 else if (ecb_expect_true (len <= 0xffU))
166 { 236 {
167 *enc->cur++ = major | 24; 237 *enc->cur++ = major | LENGTH_EXT1;
168 *enc->cur++ = len; 238 *enc->cur++ = len;
169 } 239 }
170 else if (len <= 0xffff) 240 else if (len <= 0xffffU)
171 { 241 {
172 *enc->cur++ = major | 25; 242 *enc->cur++ = major | LENGTH_EXT2;
173 *enc->cur++ = len >> 8; 243 *enc->cur++ = len >> 8;
174 *enc->cur++ = len; 244 *enc->cur++ = len;
175 } 245 }
176 else if (len <= 0xffffffff) 246 else if (len <= 0xffffffffU)
177 { 247 {
178 *enc->cur++ = major | 26; 248 *enc->cur++ = major | LENGTH_EXT4;
179 *enc->cur++ = len >> 24; 249 *enc->cur++ = len >> 24;
180 *enc->cur++ = len >> 16; 250 *enc->cur++ = len >> 16;
181 *enc->cur++ = len >> 8; 251 *enc->cur++ = len >> 8;
182 *enc->cur++ = len; 252 *enc->cur++ = len;
183 } 253 }
184 else 254 else
185 { 255 {
186 *enc->cur++ = major | 27; 256 *enc->cur++ = major | LENGTH_EXT8;
187 *enc->cur++ = len >> 56; 257 *enc->cur++ = len >> 56;
188 *enc->cur++ = len >> 48; 258 *enc->cur++ = len >> 48;
189 *enc->cur++ = len >> 40; 259 *enc->cur++ = len >> 40;
190 *enc->cur++ = len >> 32; 260 *enc->cur++ = len >> 32;
191 *enc->cur++ = len >> 24; 261 *enc->cur++ = len >> 24;
… …
193 *enc->cur++ = len >> 8; 263 *enc->cur++ = len >> 8;
194 *enc->cur++ = len; 264 *enc->cur++ = len;
195 } 265 }
196} 266}
197 267
198static void 268ecb_inline void
269encode_tag (enc_t *enc, UV tag)
270{
271 encode_uint (enc, MAJOR_TAG, tag);
272}
273
274ecb_inline void
199encode_str (enc_t *enc, int utf8, char *str, STRLEN len) 275encode_str (enc_t *enc, int utf8, char *str, STRLEN len)
200{ 276{
201 encode_uint (enc, utf8 ? 0x60 : 0x40, len); 277 encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len);
202 need (enc, len); 278 need (enc, len);
203 memcpy (enc->cur, str, len); 279 memcpy (enc->cur, str, len);
204 enc->cur += len; 280 enc->cur += len;
205} 281}
206 282
283static void
284encode_strref (enc_t *enc, int utf8, char *str, STRLEN len)
285{
286 if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS))
287 {
288 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1);
289
290 if (SvOK (*svp))
291 {
292 // already registered, use stringref
293 encode_tag (enc, CBOR_TAG_STRINGREF);
294 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
295 return;
296 }
297 else if (len >= minimum_string_length (enc->stringref_idx))
298 {
299 // register only
300 sv_setuv (*svp, enc->stringref_idx);
301 ++enc->stringref_idx;
302 }
303 }
304
305 encode_str (enc, utf8, str, len);
306}
307
207static void encode_sv (enc_t *enc, SV *sv); 308static void encode_sv (enc_t *enc, SV *sv);
208 309
209static void 310static void
210encode_av (enc_t *enc, AV *av) 311encode_av (enc_t *enc, AV *av)
211{ 312{
… …
214 if (enc->depth >= enc->cbor.max_depth) 315 if (enc->depth >= enc->cbor.max_depth)
215 croak (ERR_NESTING_EXCEEDED); 316 croak (ERR_NESTING_EXCEEDED);
216 317
217 ++enc->depth; 318 ++enc->depth;
218 319
219 encode_uint (enc, 0x80, len + 1); 320 encode_uint (enc, MAJOR_ARRAY, len + 1);
220 321
221 for (i = 0; i <= len; ++i) 322 for (i = 0; i <= len; ++i)
222 { 323 {
223 SV **svp = av_fetch (av, i, 0); 324 SV **svp = av_fetch (av, i, 0);
224 encode_sv (enc, svp ? *svp : &PL_sv_undef); 325 encode_sv (enc, svp ? *svp : &PL_sv_undef);
… …
239 340
240 int pairs = hv_iterinit (hv); 341 int pairs = hv_iterinit (hv);
241 int mg = SvMAGICAL (hv); 342 int mg = SvMAGICAL (hv);
242 343
243 if (mg) 344 if (mg)
244 encode_ch (enc, 0xa0 | 31); 345 encode_ch (enc, MAJOR_MAP | MINOR_INDEF);
245 else 346 else
246 encode_uint (enc, 0xa0, pairs); 347 encode_uint (enc, MAJOR_MAP, pairs);
247 348
248 while ((he = hv_iternext (hv))) 349 while ((he = hv_iternext (hv)))
249 { 350 {
250 if (HeKLEN (he) == HEf_SVKEY) 351 if (HeKLEN (he) == HEf_SVKEY)
251 encode_sv (enc, HeSVKEY (he)); 352 encode_sv (enc, HeSVKEY (he));
252 else 353 else
253 encode_str (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he)); 354 encode_strref (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he));
254 355
255 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he)); 356 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he));
256 } 357 }
257 358
258 if (mg) 359 if (mg)
259 encode_ch (enc, 0xe0 | 31); 360 encode_ch (enc, MAJOR_MISC | MINOR_INDEF);
260 361
261 --enc->depth; 362 --enc->depth;
262} 363}
263 364
264// encode objects, arrays and special \0=false and \1=true values. 365// encode objects, arrays and special \0=false and \1=true values.
265static void 366static void
266encode_rv (enc_t *enc, SV *sv) 367encode_rv (enc_t *enc, SV *sv)
267{ 368{
268 svtype svt;
269
270 SvGETMAGIC (sv); 369 SvGETMAGIC (sv);
370
271 svt = SvTYPE (sv); 371 svtype svt = SvTYPE (sv);
272 372
273 if (ecb_expect_false (SvOBJECT (sv))) 373 if (ecb_expect_false (SvOBJECT (sv)))
274 { 374 {
275 HV *boolean_stash = !CBOR_SLOW || types_boolean_stash 375 HV *boolean_stash = !CBOR_SLOW || types_boolean_stash
276 ? types_boolean_stash 376 ? types_boolean_stash
… …
281 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 381 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
282 ? cbor_tagged_stash 382 ? cbor_tagged_stash
283 : gv_stashpv ("CBOR::XS::Tagged" , 1); 383 : gv_stashpv ("CBOR::XS::Tagged" , 1);
284 384
285 HV *stash = SvSTASH (sv); 385 HV *stash = SvSTASH (sv);
286 GV *method;
287 386
288 if (stash == boolean_stash) 387 if (stash == boolean_stash)
289 encode_ch (enc, SvIV (sv) ? 0xe0 | 21 : 0xe0 | 20); 388 {
389 encode_ch (enc, SvIV (sv) ? MAJOR_MISC | SIMPLE_TRUE : MAJOR_MISC | SIMPLE_FALSE);
390 return;
391 }
290 else if (stash == error_stash) 392 else if (stash == error_stash)
291 encode_ch (enc, 0xe0 | 23); 393 {
394 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
395 return;
396 }
292 else if (stash == tagged_stash) 397 else if (stash == tagged_stash)
293 { 398 {
294 if (svt != SVt_PVAV) 399 if (svt != SVt_PVAV)
295 croak ("encountered CBOR::XS::Tagged object that isn't an array"); 400 croak ("encountered CBOR::XS::Tagged object that isn't an array");
296 401
297 encode_uint (enc, 0xc0, SvUV (*av_fetch ((AV *)sv, 0, 1))); 402 encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1)));
298 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1)); 403 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1));
404
405 return;
406 }
407 }
408
409 if (ecb_expect_false (SvREFCNT (sv) > 1)
410 && ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING))
411 {
412 if (!enc->shareable)
413 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
414
415 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
416
417 if (SvOK (*svp))
299 } 418 {
419 encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF);
420 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
421 return;
422 }
423 else
424 {
425 sv_setuv (*svp, enc->shareable_idx);
426 ++enc->shareable_idx;
427 encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE);
428 }
429 }
430
431 if (ecb_expect_false (SvOBJECT (sv)))
432 {
433 HV *stash = SvSTASH (sv);
434 GV *method;
435
300 else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0))) 436 if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
301 { 437 {
302 dSP; 438 dSP;
303 439
304 ENTER; SAVETMPS; PUSHMARK (SP); 440 ENTER; SAVETMPS; PUSHMARK (SP);
305 // we re-bless the reference to get overload and other niceties right 441 // we re-bless the reference to get overload and other niceties right
… …
336 472
337 // catch this surprisingly common error 473 // catch this surprisingly common error
338 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv) 474 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)
339 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash)); 475 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));
340 476
341 encode_uint (enc, 0xc0, CBOR_TAG_PERL_OBJECT); 477 encode_tag (enc, CBOR_TAG_PERL_OBJECT);
342 encode_uint (enc, 0x80, count + 1); 478 encode_uint (enc, MAJOR_ARRAY, count + 1);
343 encode_str (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash)); 479 encode_strref (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
344 480
345 while (count) 481 while (count)
346 encode_sv (enc, SP[1 - count--]); 482 encode_sv (enc, SP[1 - count--]);
347 483
348 PUTBACK; 484 PUTBACK;
… …
355 } 491 }
356 else if (svt == SVt_PVHV) 492 else if (svt == SVt_PVHV)
357 encode_hv (enc, (HV *)sv); 493 encode_hv (enc, (HV *)sv);
358 else if (svt == SVt_PVAV) 494 else if (svt == SVt_PVAV)
359 encode_av (enc, (AV *)sv); 495 encode_av (enc, (AV *)sv);
360 else if (svt < SVt_PVAV)
361 {
362 STRLEN len = 0;
363 char *pv = svt ? SvPV (sv, len) : 0;
364
365 if (len == 1 && *pv == '1')
366 encode_ch (enc, 0xe0 | 21);
367 else if (len == 1 && *pv == '0')
368 encode_ch (enc, 0xe0 | 20);
369 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
370 encode_ch (enc, 0xe0 | 23);
371 else
372 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
373 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
374 }
375 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
376 encode_ch (enc, 0xe0 | 23);
377 else 496 else
378 croak ("encountered %s, but CBOR can only represent references to arrays or hashes", 497 {
379 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 498 encode_tag (enc, CBOR_TAG_INDIRECTION);
499 encode_sv (enc, sv);
500 }
380} 501}
381 502
382static void 503static void
383encode_nv (enc_t *enc, SV *sv) 504encode_nv (enc_t *enc, SV *sv)
384{ 505{
385 double nv = SvNVX (sv); 506 double nv = SvNVX (sv);
386 507
387 need (enc, 9); 508 need (enc, 9);
388 509
389 if (ecb_expect_false (nv == (U32)nv)) 510 if (ecb_expect_false (nv == (NV)(U32)nv))
390 encode_uint (enc, 0x00, (U32)nv); 511 encode_uint (enc, MAJOR_POS_INT, (U32)nv);
391 //TODO: maybe I32? 512 //TODO: maybe I32?
392 else if (ecb_expect_false (nv == (float)nv)) 513 else if (ecb_expect_false (nv == (float)nv))
393 { 514 {
394 uint32_t fp = ecb_float_to_binary32 (nv); 515 uint32_t fp = ecb_float_to_binary32 (nv);
395 516
396 *enc->cur++ = 0xe0 | 26; 517 *enc->cur++ = MAJOR_MISC | MISC_FLOAT32;
397 518
398 if (!ecb_big_endian ()) 519 if (!ecb_big_endian ())
399 fp = ecb_bswap32 (fp); 520 fp = ecb_bswap32 (fp);
400 521
401 memcpy (enc->cur, &fp, 4); 522 memcpy (enc->cur, &fp, 4);
… …
403 } 524 }
404 else 525 else
405 { 526 {
406 uint64_t fp = ecb_double_to_binary64 (nv); 527 uint64_t fp = ecb_double_to_binary64 (nv);
407 528
408 *enc->cur++ = 0xe0 | 27; 529 *enc->cur++ = MAJOR_MISC | MISC_FLOAT64;
409 530
410 if (!ecb_big_endian ()) 531 if (!ecb_big_endian ())
411 fp = ecb_bswap64 (fp); 532 fp = ecb_bswap64 (fp);
412 533
413 memcpy (enc->cur, &fp, 8); 534 memcpy (enc->cur, &fp, 8);
… …
422 543
423 if (SvPOKp (sv)) 544 if (SvPOKp (sv))
424 { 545 {
425 STRLEN len; 546 STRLEN len;
426 char *str = SvPV (sv, len); 547 char *str = SvPV (sv, len);
427 encode_str (enc, SvUTF8 (sv), str, len); 548 encode_strref (enc, SvUTF8 (sv), str, len);
428 } 549 }
429 else if (SvNOKp (sv)) 550 else if (SvNOKp (sv))
430 encode_nv (enc, sv); 551 encode_nv (enc, sv);
431 else if (SvIOKp (sv)) 552 else if (SvIOKp (sv))
432 { 553 {
433 if (SvIsUV (sv)) 554 if (SvIsUV (sv))
434 encode_uint (enc, 0x00, SvUVX (sv)); 555 encode_uint (enc, MAJOR_POS_INT, SvUVX (sv));
435 else if (SvIVX (sv) >= 0) 556 else if (SvIVX (sv) >= 0)
436 encode_uint (enc, 0x00, SvIVX (sv)); 557 encode_uint (enc, MAJOR_POS_INT, SvIVX (sv));
437 else 558 else
438 encode_uint (enc, 0x20, -(SvIVX (sv) + 1)); 559 encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1));
439 } 560 }
440 else if (SvROK (sv)) 561 else if (SvROK (sv))
441 encode_rv (enc, SvRV (sv)); 562 encode_rv (enc, SvRV (sv));
442 else if (!SvOK (sv)) 563 else if (!SvOK (sv))
443 encode_ch (enc, 0xe0 | 22); 564 encode_ch (enc, MAJOR_MISC | SIMPLE_NULL);
444 else if (enc->cbor.flags & F_ALLOW_UNKNOWN) 565 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
445 encode_ch (enc, 0xe0 | 23); 566 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
446 else 567 else
447 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data", 568 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data",
448 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv)); 569 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
449} 570}
450 571
451static SV * 572static SV *
452encode_cbor (SV *scalar, CBOR *cbor) 573encode_cbor (SV *scalar, CBOR *cbor)
453{ 574{
454 enc_t enc; 575 enc_t enc = { };
455 576
456 enc.cbor = *cbor; 577 enc.cbor = *cbor;
457 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 578 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
458 enc.cur = SvPVX (enc.sv); 579 enc.cur = SvPVX (enc.sv);
459 enc.end = SvEND (enc.sv); 580 enc.end = SvEND (enc.sv);
460 enc.depth = 0;
461 581
462 SvPOK_only (enc.sv); 582 SvPOK_only (enc.sv);
583
584 if (cbor->flags & F_PACK_STRINGS)
585 {
586 encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE);
587 enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ());
588 enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ());
589 }
590
463 encode_sv (&enc, scalar); 591 encode_sv (&enc, scalar);
464 592
465 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 593 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
466 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings 594 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
467 595
… …
481 U8 *end; // end of input string 609 U8 *end; // end of input string
482 const char *err; // parse error, if != 0 610 const char *err; // parse error, if != 0
483 CBOR cbor; 611 CBOR cbor;
484 U32 depth; // recursion depth 612 U32 depth; // recursion depth
485 U32 maxdepth; // recursion depth limit 613 U32 maxdepth; // recursion depth limit
614 AV *shareable;
615 AV *stringref;
616 SV *decode_tagged;
486} dec_t; 617} dec_t;
487 618
488#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE 619#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE
489 620
490#define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data") 621#define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data")
… …
493#define DEC_DEC_DEPTH --dec->depth 624#define DEC_DEC_DEPTH --dec->depth
494 625
495static UV 626static UV
496decode_uint (dec_t *dec) 627decode_uint (dec_t *dec)
497{ 628{
498 switch (*dec->cur & 31) 629 U8 m = *dec->cur & MINOR_MASK;
499 { 630 ++dec->cur;
500 case 0: case 1: case 2: case 3: case 4: case 5: case 6: case 7:
501 case 8: case 9: case 10: case 11: case 12: case 13: case 14: case 15:
502 case 16: case 17: case 18: case 19: case 20: case 21: case 22: case 23:
503 return *dec->cur++ & 31;
504 631
505 case 24: 632 if (ecb_expect_true (m < LENGTH_EXT1))
633 return m;
634 else if (ecb_expect_true (m == LENGTH_EXT1))
635 {
506 WANT (2); 636 WANT (1);
507 dec->cur += 2; 637 dec->cur += 1;
508 return dec->cur[-1]; 638 return dec->cur[-1];
509 639 }
510 case 25: 640 else if (ecb_expect_true (m == LENGTH_EXT2))
641 {
511 WANT (3); 642 WANT (2);
512 dec->cur += 3; 643 dec->cur += 2;
513 return (((UV)dec->cur[-2]) << 8) 644 return (((UV)dec->cur[-2]) << 8)
514 | ((UV)dec->cur[-1]); 645 | ((UV)dec->cur[-1]);
515 646 }
516 case 26: 647 else if (ecb_expect_true (m == LENGTH_EXT4))
648 {
517 WANT (5); 649 WANT (4);
518 dec->cur += 5; 650 dec->cur += 4;
519 return (((UV)dec->cur[-4]) << 24) 651 return (((UV)dec->cur[-4]) << 24)
520 | (((UV)dec->cur[-3]) << 16) 652 | (((UV)dec->cur[-3]) << 16)
521 | (((UV)dec->cur[-2]) << 8) 653 | (((UV)dec->cur[-2]) << 8)
522 | ((UV)dec->cur[-1]); 654 | ((UV)dec->cur[-1]);
523 655 }
524 case 27: 656 else if (ecb_expect_true (m == LENGTH_EXT8))
657 {
525 WANT (9); 658 WANT (8);
526 dec->cur += 9; 659 dec->cur += 8;
660
661 return
662#if UVSIZE < 8
663 0
664#else
527 return (((UV)dec->cur[-8]) << 56) 665 (((UV)dec->cur[-8]) << 56)
528 | (((UV)dec->cur[-7]) << 48) 666 | (((UV)dec->cur[-7]) << 48)
529 | (((UV)dec->cur[-6]) << 40) 667 | (((UV)dec->cur[-6]) << 40)
530 | (((UV)dec->cur[-5]) << 32) 668 | (((UV)dec->cur[-5]) << 32)
669#endif
531 | (((UV)dec->cur[-4]) << 24) 670 | (((UV)dec->cur[-4]) << 24)
532 | (((UV)dec->cur[-3]) << 16) 671 | (((UV)dec->cur[-3]) << 16)
533 | (((UV)dec->cur[-2]) << 8) 672 | (((UV)dec->cur[-2]) << 8)
534 | ((UV)dec->cur[-1]); 673 | ((UV)dec->cur[-1]);
535 674 }
536 default: 675 else
537 ERR ("corrupted CBOR data (unsupported integer minor encoding)"); 676 ERR ("corrupted CBOR data (unsupported integer minor encoding)");
538 }
539 677
540fail: 678fail:
541 return 0; 679 return 0;
542} 680}
543 681
… …
548{ 686{
549 AV *av = newAV (); 687 AV *av = newAV ();
550 688
551 DEC_INC_DEPTH; 689 DEC_INC_DEPTH;
552 690
553 if ((*dec->cur & 31) == 31) 691 if (*dec->cur == (MAJOR_ARRAY | MINOR_INDEF))
554 { 692 {
555 ++dec->cur; 693 ++dec->cur;
556 694
557 for (;;) 695 for (;;)
558 { 696 {
559 WANT (1); 697 WANT (1);
560 698
561 if (*dec->cur == (0xe0 | 31)) 699 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
562 { 700 {
563 ++dec->cur; 701 ++dec->cur;
564 break; 702 break;
565 } 703 }
566 704
… …
569 } 707 }
570 else 708 else
571 { 709 {
572 int i, len = decode_uint (dec); 710 int i, len = decode_uint (dec);
573 711
712 WANT (len); // complexity check for av_fill - need at least one byte per value, do not allow supersize arrays
574 av_fill (av, len - 1); 713 av_fill (av, len - 1);
575 714
576 for (i = 0; i < len; ++i) 715 for (i = 0; i < len; ++i)
577 AvARRAY (av)[i] = decode_sv (dec); 716 AvARRAY (av)[i] = decode_sv (dec);
578 } 717 }
… …
584 SvREFCNT_dec (av); 723 SvREFCNT_dec (av);
585 DEC_DEC_DEPTH; 724 DEC_DEC_DEPTH;
586 return &PL_sv_undef; 725 return &PL_sv_undef;
587} 726}
588 727
728static void
729decode_he (dec_t *dec, HV *hv)
730{
731 // for speed reasons, we specialcase single-string
732 // byte or utf-8 strings as keys, but only when !stringref
733
734 if (ecb_expect_true (!dec->stringref))
735 if (ecb_expect_true ((*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8))
736 {
737 I32 len = decode_uint (dec);
738 char *key = (char *)dec->cur;
739
740 dec->cur += len;
741
742 hv_store (hv, key, len, decode_sv (dec), 0);
743
744 return;
745 }
746 else if (ecb_expect_true ((*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8))
747 {
748 I32 len = decode_uint (dec);
749 char *key = (char *)dec->cur;
750
751 dec->cur += len;
752
753 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
754 if (!is_utf8_string (key, len))
755 ERR ("corrupted CBOR data (invalid UTF-8 in map key)");
756
757 hv_store (hv, key, -len, decode_sv (dec), 0);
758
759 return;
760 }
761
762 SV *k = decode_sv (dec);
763 SV *v = decode_sv (dec);
764
765 hv_store_ent (hv, k, v, 0);
766 SvREFCNT_dec (k);
767
768fail:
769 ;
770}
771
589static SV * 772static SV *
590decode_hv (dec_t *dec) 773decode_hv (dec_t *dec)
591{ 774{
592 HV *hv = newHV (); 775 HV *hv = newHV ();
593 776
594 DEC_INC_DEPTH; 777 DEC_INC_DEPTH;
595 778
596 if ((*dec->cur & 31) == 31) 779 if (*dec->cur == (MAJOR_MAP | MINOR_INDEF))
597 { 780 {
598 ++dec->cur; 781 ++dec->cur;
599 782
600 for (;;) 783 for (;;)
601 { 784 {
602 WANT (1); 785 WANT (1);
603 786
604 if (*dec->cur == (0xe0 | 31)) 787 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
605 { 788 {
606 ++dec->cur; 789 ++dec->cur;
607 break; 790 break;
608 } 791 }
609 792
610 SV *k = decode_sv (dec); 793 decode_he (dec, hv);
611 SV *v = decode_sv (dec);
612
613 hv_store_ent (hv, k, v, 0);
614 SvREFCNT_dec (k);
615 } 794 }
616 } 795 }
617 else 796 else
618 { 797 {
619 int len = decode_uint (dec); 798 int pairs = decode_uint (dec);
620 799
621 while (len--) 800 while (pairs--)
622 { 801 decode_he (dec, hv);
623 SV *k = decode_sv (dec);
624 SV *v = decode_sv (dec);
625
626 hv_store_ent (hv, k, v, 0);
627 SvREFCNT_dec (k);
628 }
629 } 802 }
630 803
631 DEC_DEC_DEPTH; 804 DEC_DEC_DEPTH;
632 return newRV_noinc ((SV *)hv); 805 return newRV_noinc ((SV *)hv);
633 806
… …
640static SV * 813static SV *
641decode_str (dec_t *dec, int utf8) 814decode_str (dec_t *dec, int utf8)
642{ 815{
643 SV *sv = 0; 816 SV *sv = 0;
644 817
645 if ((*dec->cur & 31) == 31) 818 if ((*dec->cur & MINOR_MASK) == MINOR_INDEF)
646 { 819 {
820 // indefinite length strings
647 ++dec->cur; 821 ++dec->cur;
648 822
823 U8 major = *dec->cur & MAJOR_MISC;
824
649 sv = newSVpvn ("", 0); 825 sv = newSVpvn ("", 0);
650 826
651 // not very fast, and certainly not robust against illegal input
652 for (;;) 827 for (;;)
653 { 828 {
654 WANT (1); 829 WANT (1);
655 830
656 if (*dec->cur == (0xe0 | 31)) 831 if ((*dec->cur - major) > LENGTH_EXT8)
832 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
657 { 833 {
658 ++dec->cur; 834 ++dec->cur;
659 break; 835 break;
660 } 836 }
837 else
838 ERR ("corrupted CBOR data (invalid chunks in indefinite length string)");
661 839
662 sv_catsv (sv, decode_sv (dec)); 840 STRLEN len = decode_uint (dec);
841
842 WANT (len);
843 sv_catpvn (sv, dec->cur, len);
844 dec->cur += len;
663 } 845 }
664 } 846 }
665 else 847 else
666 { 848 {
667 STRLEN len = decode_uint (dec); 849 STRLEN len = decode_uint (dec);
668 850
669 WANT (len); 851 WANT (len);
670 sv = newSVpvn (dec->cur, len); 852 sv = newSVpvn (dec->cur, len);
671 dec->cur += len; 853 dec->cur += len;
854
855 if (ecb_expect_false (dec->stringref)
856 && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1))
857 av_push (dec->stringref, SvREFCNT_inc_NN (sv));
672 } 858 }
673 859
674 if (utf8) 860 if (utf8)
861 {
862 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
863 if (!is_utf8_string (SvPVX (sv), SvCUR (sv)))
864 ERR ("corrupted CBOR data (invalid UTF-8 in text string)");
865
675 SvUTF8_on (sv); 866 SvUTF8_on (sv);
867 }
676 868
677 return sv; 869 return sv;
678 870
679fail: 871fail:
680 SvREFCNT_dec (sv); 872 SvREFCNT_dec (sv);
… …
682} 874}
683 875
684static SV * 876static SV *
685decode_tagged (dec_t *dec) 877decode_tagged (dec_t *dec)
686{ 878{
879 SV *sv = 0;
687 UV tag = decode_uint (dec); 880 UV tag = decode_uint (dec);
881
882 WANT (1);
883
884 switch (tag)
885 {
886 case CBOR_TAG_MAGIC:
688 SV *sv = decode_sv (dec); 887 sv = decode_sv (dec);
888 break;
689 889
690 if (tag == CBOR_TAG_MAGIC) 890 case CBOR_TAG_INDIRECTION:
691 return sv; 891 sv = newRV_noinc (decode_sv (dec));
692 else if (tag == CBOR_TAG_PERL_OBJECT) 892 break;
693 { 893
894 case CBOR_TAG_STRINGREF_NAMESPACE:
895 {
896 ENTER; SAVETMPS;
897
898 SAVESPTR (dec->stringref);
899 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ());
900
901 sv = decode_sv (dec);
902
903 FREETMPS; LEAVE;
904 }
905 break;
906
907 case CBOR_TAG_STRINGREF:
908 {
909 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
910 ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
911
912 UV idx = decode_uint (dec);
913
914 if (!dec->stringref || (int)idx > AvFILLp (dec->stringref))
915 ERR ("corrupted CBOR data (stringref index out of bounds or outside namespace)");
916
917 sv = newSVsv (AvARRAY (dec->stringref)[idx]);
918 }
919 break;
920
921 case CBOR_TAG_VALUE_SHAREABLE:
922 {
923 if (ecb_expect_false (!dec->shareable))
924 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ());
925
926 if (dec->cbor.flags & F_ALLOW_CYCLES)
927 {
928 sv = newSV (0);
929 av_push (dec->shareable, SvREFCNT_inc_NN (sv));
930
931 SV *osv = decode_sv (dec);
932 sv_setsv (sv, osv);
933 SvREFCNT_dec_NN (osv);
934 }
935 else
936 {
937 av_push (dec->shareable, &PL_sv_undef);
938 int idx = AvFILLp (dec->shareable);
939 sv = decode_sv (dec);
940 av_store (dec->shareable, idx, SvREFCNT_inc_NN (sv));
941 }
942 }
943 break;
944
945 case CBOR_TAG_VALUE_SHAREDREF:
946 {
947 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
948 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
949
950 UV idx = decode_uint (dec);
951
952 if (!dec->shareable || (int)idx > AvFILLp (dec->shareable))
953 ERR ("corrupted CBOR data (sharedref index out of bounds)");
954
955 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]);
956
957 if (sv == &PL_sv_undef)
958 ERR ("cyclic CBOR data structure found, but allow_cycles is not enabled");
959 }
960 break;
961
962 case CBOR_TAG_PERL_OBJECT:
963 {
964 sv = decode_sv (dec);
965
694 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV) 966 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
695 ERR ("corrupted CBOR data (non-array perl object)"); 967 ERR ("corrupted CBOR data (non-array perl object)");
696 968
697 AV *av = (AV *)SvRV (sv); 969 AV *av = (AV *)SvRV (sv);
698 int len = av_len (av) + 1; 970 int len = av_len (av) + 1;
699 HV *stash = gv_stashsv (*av_fetch (av, 0, 1), 0); 971 HV *stash = gv_stashsv (*av_fetch (av, 0, 1), 0);
700 972
701 if (!stash) 973 if (!stash)
702 ERR ("cannot decode perl-object (package does not exist)"); 974 ERR ("cannot decode perl-object (package does not exist)");
703 975
704 GV *method = gv_fetchmethod_autoload (stash, "THAW", 0); 976 GV *method = gv_fetchmethod_autoload (stash, "THAW", 0);
705 977
706 if (!method) 978 if (!method)
707 ERR ("cannot decode perl-object (package does not have a THAW method)"); 979 ERR ("cannot decode perl-object (package does not have a THAW method)");
708 980
709 dSP; 981 dSP;
710 982
711 ENTER; SAVETMPS; PUSHMARK (SP); 983 ENTER; SAVETMPS; PUSHMARK (SP);
712 EXTEND (SP, len + 1); 984 EXTEND (SP, len + 1);
713 // we re-bless the reference to get overload and other niceties right 985 // we re-bless the reference to get overload and other niceties right
714 PUSHs (*av_fetch (av, 0, 1)); 986 PUSHs (*av_fetch (av, 0, 1));
715 PUSHs (sv_cbor); 987 PUSHs (sv_cbor);
716 988
717 int i; 989 int i;
718 990
719 for (i = 1; i < len; ++i) 991 for (i = 1; i < len; ++i)
720 PUSHs (*av_fetch (av, i, 1)); 992 PUSHs (*av_fetch (av, i, 1));
721 993
722 PUTBACK; 994 PUTBACK;
723 call_sv ((SV *)GvCV (method), G_SCALAR); 995 call_sv ((SV *)GvCV (method), G_SCALAR | G_EVAL);
724 SPAGAIN; 996 SPAGAIN;
725 997
998 if (SvTRUE (ERRSV))
999 {
1000 FREETMPS; LEAVE;
1001 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV))));
1002 }
1003
726 SvREFCNT_dec (sv); 1004 SvREFCNT_dec (sv);
727 sv = SvREFCNT_inc (POPs); 1005 sv = SvREFCNT_inc (POPs);
728 1006
729 PUTBACK; 1007 PUTBACK;
730 1008
731 FREETMPS; LEAVE; 1009 FREETMPS; LEAVE;
1010 }
1011 break;
732 1012
733 return sv; 1013 default:
734 } 1014 {
735 else 1015 sv = decode_sv (dec);
736 { 1016
1017 dSP;
1018 ENTER; SAVETMPS; PUSHMARK (SP);
1019 EXTEND (SP, 2);
1020 PUSHs (newSVuv (tag));
1021 PUSHs (sv);
1022
1023 PUTBACK;
1024 int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL);
1025 SPAGAIN;
1026
1027 if (SvTRUE (ERRSV))
1028 {
1029 FREETMPS; LEAVE;
1030 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV))));
1031 }
1032
1033 if (count)
1034 {
1035 SvREFCNT_dec (sv);
1036 sv = SvREFCNT_inc (POPs);
1037 }
1038 else
1039 {
737 AV *av = newAV (); 1040 AV *av = newAV ();
738 av_push (av, newSVuv (tag)); 1041 av_push (av, newSVuv (tag));
739 av_push (av, sv); 1042 av_push (av, sv);
740 1043
741 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 1044 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
742 ? cbor_tagged_stash 1045 ? cbor_tagged_stash
743 : gv_stashpv ("CBOR::XS::Tagged" , 1); 1046 : gv_stashpv ("CBOR::XS::Tagged" , 1);
744
745 return sv_bless (newRV_noinc ((SV *)av), tagged_stash); 1047 sv = sv_bless (newRV_noinc ((SV *)av), tagged_stash);
1048 }
1049
1050 PUTBACK;
1051
1052 FREETMPS; LEAVE;
1053 }
1054 break;
746 } 1055 }
1056
1057 return sv;
747 1058
748fail: 1059fail:
749 SvREFCNT_dec (sv); 1060 SvREFCNT_dec (sv);
750 return &PL_sv_undef; 1061 return &PL_sv_undef;
751} 1062}
… …
753static SV * 1064static SV *
754decode_sv (dec_t *dec) 1065decode_sv (dec_t *dec)
755{ 1066{
756 WANT (1); 1067 WANT (1);
757 1068
758 switch (*dec->cur >> 5) 1069 switch (*dec->cur >> MAJOR_SHIFT)
759 { 1070 {
760 case 0: // unsigned int 1071 case MAJOR_POS_INT >> MAJOR_SHIFT: return newSVuv (decode_uint (dec));
761 return newSVuv (decode_uint (dec)); 1072 case MAJOR_NEG_INT >> MAJOR_SHIFT: return newSViv (-1 - (IV)decode_uint (dec));
762 case 1: // negative int 1073 case MAJOR_BYTES >> MAJOR_SHIFT: return decode_str (dec, 0);
763 return newSViv (-1 - (IV)decode_uint (dec)); 1074 case MAJOR_TEXT >> MAJOR_SHIFT: return decode_str (dec, 1);
764 case 2: // octet string 1075 case MAJOR_ARRAY >> MAJOR_SHIFT: return decode_av (dec);
765 return decode_str (dec, 0); 1076 case MAJOR_MAP >> MAJOR_SHIFT: return decode_hv (dec);
766 case 3: // utf-8 string 1077 case MAJOR_TAG >> MAJOR_SHIFT: return decode_tagged (dec);
767 return decode_str (dec, 1); 1078
768 case 4: // array 1079 case MAJOR_MISC >> MAJOR_SHIFT:
769 return decode_av (dec);
770 case 5: // map
771 return decode_hv (dec);
772 case 6: // tag
773 return decode_tagged (dec);
774 case 7: // misc
775 switch (*dec->cur++ & 31) 1080 switch (*dec->cur++ & MINOR_MASK)
776 { 1081 {
777 case 20: 1082 case SIMPLE_FALSE:
778#if CBOR_SLOW 1083#if CBOR_SLOW
779 types_false = get_bool ("Types::Serialiser::false"); 1084 types_false = get_bool ("Types::Serialiser::false");
780#endif 1085#endif
781 return newSVsv (types_false); 1086 return newSVsv (types_false);
782 case 21: 1087 case SIMPLE_TRUE:
783#if CBOR_SLOW 1088#if CBOR_SLOW
784 types_true = get_bool ("Types::Serialiser::true"); 1089 types_true = get_bool ("Types::Serialiser::true");
785#endif 1090#endif
786 return newSVsv (types_true); 1091 return newSVsv (types_true);
787 case 22: 1092 case SIMPLE_NULL:
788 return newSVsv (&PL_sv_undef); 1093 return newSVsv (&PL_sv_undef);
789 case 23: 1094 case SIMPLE_UNDEF:
790#if CBOR_SLOW 1095#if CBOR_SLOW
791 types_error = get_bool ("Types::Serialiser::error"); 1096 types_error = get_bool ("Types::Serialiser::error");
792#endif 1097#endif
793 return newSVsv (types_error); 1098 return newSVsv (types_error);
794 1099
795 case 25: 1100 case MISC_FLOAT16:
796 { 1101 {
797 WANT (2); 1102 WANT (2);
798 1103
799 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1]; 1104 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1];
800 dec->cur += 2; 1105 dec->cur += 2;
801 1106
802 return newSVnv (ecb_binary16_to_float (fp)); 1107 return newSVnv (ecb_binary16_to_float (fp));
803 } 1108 }
804 1109
805 case 26: 1110 case MISC_FLOAT32:
806 { 1111 {
807 uint32_t fp; 1112 uint32_t fp;
808 WANT (4); 1113 WANT (4);
809 memcpy (&fp, dec->cur, 4); 1114 memcpy (&fp, dec->cur, 4);
810 dec->cur += 4; 1115 dec->cur += 4;
… …
813 fp = ecb_bswap32 (fp); 1118 fp = ecb_bswap32 (fp);
814 1119
815 return newSVnv (ecb_binary32_to_float (fp)); 1120 return newSVnv (ecb_binary32_to_float (fp));
816 } 1121 }
817 1122
818 case 27: 1123 case MISC_FLOAT64:
819 { 1124 {
820 uint64_t fp; 1125 uint64_t fp;
821 WANT (8); 1126 WANT (8);
822 memcpy (&fp, dec->cur, 8); 1127 memcpy (&fp, dec->cur, 8);
823 dec->cur += 8; 1128 dec->cur += 8;
… …
826 fp = ecb_bswap64 (fp); 1131 fp = ecb_bswap64 (fp);
827 1132
828 return newSVnv (ecb_binary64_to_double (fp)); 1133 return newSVnv (ecb_binary64_to_double (fp));
829 } 1134 }
830 1135
831 // 0..19 unassigned 1136 // 0..19 unassigned simple
832 // 24 reserved + unassigned (reserved values are not encodable) 1137 // 24 reserved + unassigned (reserved values are not encodable)
833 default: 1138 default:
834 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)"); 1139 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)");
835 } 1140 }
836 1141
… …
842} 1147}
843 1148
844static SV * 1149static SV *
845decode_cbor (SV *string, CBOR *cbor, char **offset_return) 1150decode_cbor (SV *string, CBOR *cbor, char **offset_return)
846{ 1151{
847 dec_t dec; 1152 dec_t dec = { };
848 SV *sv; 1153 SV *sv;
1154 STRLEN len;
1155 char *data = SvPVbyte (string, len);
849 1156
850 /* work around bugs in 5.10 where manipulating magic values
851 * makes perl ignore the magic in subsequent accesses.
852 * also make a copy of non-PV values, to get them into a clean
853 * state (SvPV should do that, but it's buggy, see below).
854 */
855 /*SvGETMAGIC (string);*/
856 if (SvMAGICAL (string) || !SvPOK (string))
857 string = sv_2mortal (newSVsv (string));
858
859 SvUPGRADE (string, SVt_PV);
860
861 /* work around a bug in perl 5.10, which causes SvCUR to fail an
862 * assertion with -DDEBUGGING, although SvCUR is documented to
863 * return the xpv_cur field which certainly exists after upgrading.
864 * according to nicholas clark, calling SvPOK fixes this.
865 * But it doesn't fix it, so try another workaround, call SvPV_nolen
866 * and hope for the best.
867 * Damnit, SvPV_nolen still trips over yet another assertion. This
868 * assertion business is seriously broken, try yet another workaround
869 * for the broken -DDEBUGGING.
870 */
871 {
872#ifdef DEBUGGING
873 STRLEN offset = SvOK (string) ? sv_len (string) : 0;
874#else
875 STRLEN offset = SvCUR (string);
876#endif
877
878 if (offset > cbor->max_size && cbor->max_size) 1157 if (len > cbor->max_size && cbor->max_size)
879 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu", 1158 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
880 (unsigned long)SvCUR (string), (unsigned long)cbor->max_size); 1159 (unsigned long)len, (unsigned long)cbor->max_size);
881 }
882
883 sv_utf8_downgrade (string, 0);
884 1160
885 dec.cbor = *cbor; 1161 dec.cbor = *cbor;
886 dec.cur = (U8 *)SvPVX (string); 1162 dec.cur = (U8 *)data;
887 dec.end = (U8 *)SvEND (string); 1163 dec.end = (U8 *)data + len;
888 dec.err = 0;
889 dec.depth = 0;
890 1164
891 sv = decode_sv (&dec); 1165 sv = decode_sv (&dec);
892 1166
893 if (offset_return) 1167 if (offset_return)
894 *offset_return = dec.cur; 1168 *offset_return = dec.cur;
… …
898 dec.err = "garbage after CBOR object"; 1172 dec.err = "garbage after CBOR object";
899 1173
900 if (dec.err) 1174 if (dec.err)
901 { 1175 {
902 SvREFCNT_dec (sv); 1176 SvREFCNT_dec (sv);
903 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)SvPVX (string), (int)(uint8_t)*dec.cur); 1177 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur);
904 } 1178 }
905 1179
906 sv = sv_2mortal (sv); 1180 sv = sv_2mortal (sv);
907 1181
908 return sv; 1182 return sv;
… …
922 types_error_stash = gv_stashpv ("Types::Serialiser::Error" , 1); 1196 types_error_stash = gv_stashpv ("Types::Serialiser::Error" , 1);
923 1197
924 types_true = get_bool ("Types::Serialiser::true" ); 1198 types_true = get_bool ("Types::Serialiser::true" );
925 types_false = get_bool ("Types::Serialiser::false"); 1199 types_false = get_bool ("Types::Serialiser::false");
926 types_error = get_bool ("Types::Serialiser::error"); 1200 types_error = get_bool ("Types::Serialiser::error");
1201
1202 default_filter = newSVpv ("CBOR::XS::default_filter", 0);
927 1203
928 sv_cbor = newSVpv ("CBOR", 0); 1204 sv_cbor = newSVpv ("CBOR", 0);
929 SvREADONLY_on (sv_cbor); 1205 SvREADONLY_on (sv_cbor);
930} 1206}
931 1207
… …
952 1228
953void shrink (CBOR *self, int enable = 1) 1229void shrink (CBOR *self, int enable = 1)
954 ALIAS: 1230 ALIAS:
955 shrink = F_SHRINK 1231 shrink = F_SHRINK
956 allow_unknown = F_ALLOW_UNKNOWN 1232 allow_unknown = F_ALLOW_UNKNOWN
1233 allow_sharing = F_ALLOW_SHARING
1234 allow_cycles = F_ALLOW_CYCLES
1235 pack_strings = F_PACK_STRINGS
1236 validate_utf8 = F_VALIDATE_UTF8
957 PPCODE: 1237 PPCODE:
958{ 1238{
959 if (enable) 1239 if (enable)
960 self->flags |= ix; 1240 self->flags |= ix;
961 else 1241 else
… …
966 1246
967void get_shrink (CBOR *self) 1247void get_shrink (CBOR *self)
968 ALIAS: 1248 ALIAS:
969 get_shrink = F_SHRINK 1249 get_shrink = F_SHRINK
970 get_allow_unknown = F_ALLOW_UNKNOWN 1250 get_allow_unknown = F_ALLOW_UNKNOWN
1251 get_allow_sharing = F_ALLOW_SHARING
1252 get_allow_cycles = F_ALLOW_CYCLES
1253 get_pack_strings = F_PACK_STRINGS
1254 get_validate_utf8 = F_VALIDATE_UTF8
971 PPCODE: 1255 PPCODE:
972 XPUSHs (boolSV (self->flags & ix)); 1256 XPUSHs (boolSV (self->flags & ix));
973 1257
974void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) 1258void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
975 PPCODE: 1259 PPCODE:
… …
988 XPUSHs (ST (0)); 1272 XPUSHs (ST (0));
989 1273
990int get_max_size (CBOR *self) 1274int get_max_size (CBOR *self)
991 CODE: 1275 CODE:
992 RETVAL = self->max_size; 1276 RETVAL = self->max_size;
1277 OUTPUT:
1278 RETVAL
1279
1280void filter (CBOR *self, SV *filter = 0)
1281 PPCODE:
1282 SvREFCNT_dec (self->filter);
1283 self->filter = filter ? newSVsv (filter) : filter;
1284 XPUSHs (ST (0));
1285
1286SV *get_filter (CBOR *self)
1287 CODE:
1288 RETVAL = self->filter ? self->filter : NEWSV (0, 0);
993 OUTPUT: 1289 OUTPUT:
994 RETVAL 1290 RETVAL
995 1291
996void encode (CBOR *self, SV *scalar) 1292void encode (CBOR *self, SV *scalar)
997 PPCODE: 1293 PPCODE:
… …
1012 EXTEND (SP, 2); 1308 EXTEND (SP, 2);
1013 PUSHs (sv); 1309 PUSHs (sv);
1014 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); 1310 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
1015} 1311}
1016 1312
1313void DESTROY (CBOR *self)
1314 PPCODE:
1315 cbor_free (self);
1316
1017PROTOTYPES: ENABLE 1317PROTOTYPES: ENABLE
1018 1318
1019void encode_cbor (SV *scalar) 1319void encode_cbor (SV *scalar)
1320 ALIAS:
1321 encode_cbor = 0
1322 encode_cbor_sharing = F_ALLOW_SHARING
1020 PPCODE: 1323 PPCODE:
1021{ 1324{
1022 CBOR cbor; 1325 CBOR cbor;
1023 cbor_init (&cbor); 1326 cbor_init (&cbor);
1327 cbor.flags |= ix;
1024 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN; 1328 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;
1025 XPUSHs (scalar); 1329 XPUSHs (scalar);
1026} 1330}
1027 1331
1028void decode_cbor (SV *cborstr) 1332void decode_cbor (SV *cborstr)

Diff Legend

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