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 1.6 by root, Sat Oct 26 22:25:47 2013 UTC vs.
Revision 1.36 by root, Sat Nov 30 17:37:45 2013 UTC

9#include <limits.h> 9#include <limits.h>
10#include <float.h> 10#include <float.h>
11 11
12#include "ecb.h" 12#include "ecb.h"
13 13
14// compatibility with perl <5.18
15#ifndef HvNAMELEN_get
16# define HvNAMELEN_get(hv) strlen (HvNAME (hv))
17#endif
18#ifndef HvNAMELEN
19# define HvNAMELEN(hv) HvNAMELEN_get (hv)
20#endif
21#ifndef HvNAMEUTF8
22# define HvNAMEUTF8(hv) 0
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};
64
65// known tags
66enum cbor_tag
67{
68 // extensions
69 CBOR_TAG_STRINGREF = 25, // http://cbor.schmorp.de/stringref
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
76
77 // rfc7049
78 CBOR_TAG_DATETIME = 0, // rfc4287, utf-8
79 CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any
80 CBOR_TAG_POS_BIGNUM = 2, // byte string
81 CBOR_TAG_NEG_BIGNUM = 3, // byte string
82 CBOR_TAG_DECIMAL = 4, // decimal fraction, array
83 CBOR_TAG_BIGFLOAT = 5, // array
84
85 CBOR_TAG_CONV_B64U = 21, // base64url, any
86 CBOR_TAG_CONV_B64 = 22, // base64, any
87 CBOR_TAG_CONV_HEX = 23, // base16, any
88 CBOR_TAG_CBOR = 24, // embedded cbor, byte string
89
90 CBOR_TAG_URI = 32, // URI rfc3986, utf-8
91 CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8
92 CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8
93 CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8
94 CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8
95
96 CBOR_TAG_MAGIC = 55799, // self-describe cbor
97};
98
14#define F_SHRINK 0x00000200UL 99#define F_SHRINK 0x00000001UL
15#define F_ALLOW_UNKNOWN 0x00002000UL 100#define F_ALLOW_UNKNOWN 0x00000002UL
101#define F_ALLOW_SHARING 0x00000004UL
102#define F_PACK_STRINGS 0x00000008UL
16 103
17#define INIT_SIZE 32 // initial scalar size to be allocated 104#define INIT_SIZE 32 // initial scalar size to be allocated
18 105
19#define SB do { 106#define SB do {
20#define SE } while (0) 107#define SE } while (0)
31#else 118#else
32# define CBOR_SLOW 0 119# define CBOR_SLOW 0
33# define CBOR_STASH cbor_stash 120# define CBOR_STASH cbor_stash
34#endif 121#endif
35 122
36static HV *cbor_stash, *cbor_boolean_stash, *cbor_tagged_stash; // CBOR::XS:: 123static HV *cbor_stash, *types_boolean_stash, *types_error_stash, *cbor_tagged_stash; // CBOR::XS::
37static SV *cbor_true, *cbor_false; 124static SV *types_true, *types_false, *types_error, *sv_cbor, *default_filter;
38 125
39typedef struct { 126typedef struct {
40 U32 flags; 127 U32 flags;
41 U32 max_depth; 128 U32 max_depth;
42 STRLEN max_size; 129 STRLEN max_size;
130 SV *filter;
43} CBOR; 131} CBOR;
44 132
45ecb_inline void 133ecb_inline void
46cbor_init (CBOR *cbor) 134cbor_init (CBOR *cbor)
47{ 135{
48 Zero (cbor, 1, CBOR); 136 Zero (cbor, 1, CBOR);
49 cbor->max_depth = 512; 137 cbor->max_depth = 512;
138}
139
140ecb_inline void
141cbor_free (CBOR *cbor)
142{
143 SvREFCNT_dec (cbor->filter);
50} 144}
51 145
52///////////////////////////////////////////////////////////////////////////// 146/////////////////////////////////////////////////////////////////////////////
53// utility functions 147// utility functions
54 148
76 SvPV_renew (sv, SvCUR (sv) + 1); 170 SvPV_renew (sv, SvCUR (sv) + 1);
77#endif 171#endif
78 } 172 }
79} 173}
80 174
81///////////////////////////////////////////////////////////////////////////// 175// minimum length of a string to be registered for stringref
82// fp hell 176ecb_inline int
83 177minimum_string_length (UV idx)
84//TODO 178{
179 return idx > 23
180 ? idx > 0xffU
181 ? idx > 0xffffU
182 ? idx > 0xffffffffU
183 ? 11
184 : 7
185 : 5
186 : 4
187 : 3;
188}
85 189
86///////////////////////////////////////////////////////////////////////////// 190/////////////////////////////////////////////////////////////////////////////
87// encoder 191// encoder
88 192
89// structure used for encoding CBOR 193// structure used for encoding CBOR
92 char *cur; // SvPVX (sv) + current output position 196 char *cur; // SvPVX (sv) + current output position
93 char *end; // SvEND (sv) 197 char *end; // SvEND (sv)
94 SV *sv; // result scalar 198 SV *sv; // result scalar
95 CBOR cbor; 199 CBOR cbor;
96 U32 depth; // recursion level 200 U32 depth; // recursion level
201 HV *stringref[2]; // string => index, or 0 ([0] = bytes, [1] = utf-8)
202 UV stringref_idx;
203 HV *shareable; // ptr => index, or 0
204 UV shareable_idx;
97} enc_t; 205} enc_t;
98 206
99ecb_inline void 207ecb_inline void
100need (enc_t *enc, STRLEN len) 208need (enc_t *enc, STRLEN len)
101{ 209{
118static void 226static void
119encode_uint (enc_t *enc, int major, UV len) 227encode_uint (enc_t *enc, int major, UV len)
120{ 228{
121 need (enc, 9); 229 need (enc, 9);
122 230
123 if (len < 24) 231 if (ecb_expect_true (len < LENGTH_EXT1))
124 *enc->cur++ = major | len; 232 *enc->cur++ = major | len;
125 else if (len <= 0xff) 233 else if (ecb_expect_true (len <= 0xffU))
126 { 234 {
127 *enc->cur++ = major | 24; 235 *enc->cur++ = major | LENGTH_EXT1;
128 *enc->cur++ = len; 236 *enc->cur++ = len;
129 } 237 }
130 else if (len <= 0xffff) 238 else if (len <= 0xffffU)
131 { 239 {
132 *enc->cur++ = major | 25; 240 *enc->cur++ = major | LENGTH_EXT2;
133 *enc->cur++ = len >> 8; 241 *enc->cur++ = len >> 8;
134 *enc->cur++ = len; 242 *enc->cur++ = len;
135 } 243 }
136 else if (len <= 0xffffffff) 244 else if (len <= 0xffffffffU)
137 { 245 {
138 *enc->cur++ = major | 26; 246 *enc->cur++ = major | LENGTH_EXT4;
139 *enc->cur++ = len >> 24; 247 *enc->cur++ = len >> 24;
140 *enc->cur++ = len >> 16; 248 *enc->cur++ = len >> 16;
141 *enc->cur++ = len >> 8; 249 *enc->cur++ = len >> 8;
142 *enc->cur++ = len; 250 *enc->cur++ = len;
143 } 251 }
144 else 252 else
145 { 253 {
146 *enc->cur++ = major | 27; 254 *enc->cur++ = major | LENGTH_EXT8;
147 *enc->cur++ = len >> 56; 255 *enc->cur++ = len >> 56;
148 *enc->cur++ = len >> 48; 256 *enc->cur++ = len >> 48;
149 *enc->cur++ = len >> 40; 257 *enc->cur++ = len >> 40;
150 *enc->cur++ = len >> 32; 258 *enc->cur++ = len >> 32;
151 *enc->cur++ = len >> 24; 259 *enc->cur++ = len >> 24;
153 *enc->cur++ = len >> 8; 261 *enc->cur++ = len >> 8;
154 *enc->cur++ = len; 262 *enc->cur++ = len;
155 } 263 }
156} 264}
157 265
158static void 266ecb_inline void
267encode_tag (enc_t *enc, UV tag)
268{
269 encode_uint (enc, MAJOR_TAG, tag);
270}
271
272ecb_inline void
159encode_str (enc_t *enc, int utf8, char *str, STRLEN len) 273encode_str (enc_t *enc, int utf8, char *str, STRLEN len)
160{ 274{
161 encode_uint (enc, utf8 ? 0x60 : 0x40, len); 275 encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len);
162 need (enc, len); 276 need (enc, len);
163 memcpy (enc->cur, str, len); 277 memcpy (enc->cur, str, len);
164 enc->cur += len; 278 enc->cur += len;
165} 279}
166 280
281static void
282encode_strref (enc_t *enc, int utf8, char *str, STRLEN len)
283{
284 if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS))
285 {
286 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1);
287
288 if (SvOK (*svp))
289 {
290 // already registered, use stringref
291 encode_tag (enc, CBOR_TAG_STRINGREF);
292 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
293 return;
294 }
295 else if (len >= minimum_string_length (enc->stringref_idx))
296 {
297 // register only
298 sv_setuv (*svp, enc->stringref_idx);
299 ++enc->stringref_idx;
300 }
301 }
302
303 encode_str (enc, utf8, str, len);
304}
305
167static void encode_sv (enc_t *enc, SV *sv); 306static void encode_sv (enc_t *enc, SV *sv);
168 307
169static void 308static void
170encode_av (enc_t *enc, AV *av) 309encode_av (enc_t *enc, AV *av)
171{ 310{
174 if (enc->depth >= enc->cbor.max_depth) 313 if (enc->depth >= enc->cbor.max_depth)
175 croak (ERR_NESTING_EXCEEDED); 314 croak (ERR_NESTING_EXCEEDED);
176 315
177 ++enc->depth; 316 ++enc->depth;
178 317
179 encode_uint (enc, 0x80, len + 1); 318 encode_uint (enc, MAJOR_ARRAY, len + 1);
180 319
181 for (i = 0; i <= len; ++i) 320 for (i = 0; i <= len; ++i)
182 { 321 {
183 SV **svp = av_fetch (av, i, 0); 322 SV **svp = av_fetch (av, i, 0);
184 encode_sv (enc, svp ? *svp : &PL_sv_undef); 323 encode_sv (enc, svp ? *svp : &PL_sv_undef);
199 338
200 int pairs = hv_iterinit (hv); 339 int pairs = hv_iterinit (hv);
201 int mg = SvMAGICAL (hv); 340 int mg = SvMAGICAL (hv);
202 341
203 if (mg) 342 if (mg)
204 encode_ch (enc, 0xa0 | 31); 343 encode_ch (enc, MAJOR_MAP | MINOR_INDEF);
205 else 344 else
206 encode_uint (enc, 0xa0, pairs); 345 encode_uint (enc, MAJOR_MAP, pairs);
207 346
208 while ((he = hv_iternext (hv))) 347 while ((he = hv_iternext (hv)))
209 { 348 {
210 if (HeKLEN (he) == HEf_SVKEY) 349 if (HeKLEN (he) == HEf_SVKEY)
211 encode_sv (enc, HeSVKEY (he)); 350 encode_sv (enc, HeSVKEY (he));
212 else 351 else
213 encode_str (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he)); 352 encode_strref (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he));
214 353
215 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he)); 354 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he));
216 } 355 }
217 356
218 if (mg) 357 if (mg)
219 encode_ch (enc, 0xe0 | 31); 358 encode_ch (enc, MAJOR_MISC | MINOR_INDEF);
220 359
221 --enc->depth; 360 --enc->depth;
222} 361}
223 362
224// encode objects, arrays and special \0=false and \1=true values. 363// encode objects, arrays and special \0=false and \1=true values.
225static void 364static void
226encode_rv (enc_t *enc, SV *sv) 365encode_rv (enc_t *enc, SV *sv)
227{ 366{
228 svtype svt;
229
230 SvGETMAGIC (sv); 367 SvGETMAGIC (sv);
368
231 svt = SvTYPE (sv); 369 svtype svt = SvTYPE (sv);
232 370
233 if (ecb_expect_false (SvOBJECT (sv))) 371 if (ecb_expect_false (SvOBJECT (sv)))
234 { 372 {
235 HV *boolean_stash = !CBOR_SLOW || cbor_boolean_stash 373 HV *boolean_stash = !CBOR_SLOW || types_boolean_stash
236 ? cbor_boolean_stash 374 ? types_boolean_stash
237 : gv_stashpv ("CBOR::XS::Boolean", 1); 375 : gv_stashpv ("Types::Serialiser::Boolean", 1);
376 HV *error_stash = !CBOR_SLOW || types_error_stash
377 ? types_error_stash
378 : gv_stashpv ("Types::Serialiser::Error", 1);
238 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 379 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
239 ? cbor_tagged_stash 380 ? cbor_tagged_stash
240 : gv_stashpv ("CBOR::XS::Tagged" , 1); 381 : gv_stashpv ("CBOR::XS::Tagged" , 1);
241 382
383 HV *stash = SvSTASH (sv);
384
242 if (SvSTASH (sv) == boolean_stash) 385 if (stash == boolean_stash)
243 encode_ch (enc, SvIV (sv) ? 0xe0 | 21 : 0xe0 | 20); 386 {
387 encode_ch (enc, SvIV (sv) ? MAJOR_MISC | SIMPLE_TRUE : MAJOR_MISC | SIMPLE_FALSE);
388 return;
389 }
390 else if (stash == error_stash)
391 {
392 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
393 return;
394 }
244 else if (SvSTASH (sv) == tagged_stash) 395 else if (stash == tagged_stash)
245 { 396 {
246 if (svt != SVt_PVAV) 397 if (svt != SVt_PVAV)
247 croak ("encountered CBOR::XS::Tagged object that isn't an array"); 398 croak ("encountered CBOR::XS::Tagged object that isn't an array");
248 399
249 encode_uint (enc, 0xc0, SvUV (*av_fetch ((AV *)sv, 0, 1))); 400 encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1)));
250 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1)); 401 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1));
402
403 return;
404 }
405 }
406
407 if (ecb_expect_false (SvREFCNT (sv) > 1)
408 && ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING))
409 {
410 if (!enc->shareable)
411 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
412
413 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
414
415 if (SvOK (*svp))
416 {
417 encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF);
418 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
419 return;
251 } 420 }
252 else 421 else
253 { 422 {
423 sv_setuv (*svp, enc->shareable_idx);
424 ++enc->shareable_idx;
425 encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE);
426 }
427 }
428
429 if (ecb_expect_false (SvOBJECT (sv)))
430 {
431 HV *stash = SvSTASH (sv);
432 GV *method;
433
434 if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
435 {
436 dSP;
437
438 ENTER; SAVETMPS; PUSHMARK (SP);
254 // we re-bless the reference to get overload and other niceties right 439 // we re-bless the reference to get overload and other niceties right
255 GV *to_cbor = gv_fetchmethod_autoload (SvSTASH (sv), "TO_CBOR", 0);
256
257 if (to_cbor)
258 {
259 dSP;
260
261 ENTER; SAVETMPS; PUSHMARK (SP);
262 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv))); 440 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
263 441
264 // calling with G_SCALAR ensures that we always get a 1 return value
265 PUTBACK; 442 PUTBACK;
443 // G_SCALAR ensures that return value is 1
266 call_sv ((SV *)GvCV (to_cbor), G_SCALAR); 444 call_sv ((SV *)GvCV (method), G_SCALAR);
267 SPAGAIN; 445 SPAGAIN;
268 446
269 // catch this surprisingly common error 447 // catch this surprisingly common error
270 if (SvROK (TOPs) && SvRV (TOPs) == sv) 448 if (SvROK (TOPs) && SvRV (TOPs) == sv)
271 croak ("%s::TO_CBOR method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv))); 449 croak ("%s::TO_CBOR method returned same object as was passed instead of a new one", HvNAME (stash));
272 450
273 sv = POPs; 451 encode_sv (enc, POPs);
452
274 PUTBACK; 453 PUTBACK;
275 454
276 encode_sv (enc, sv);
277
278 FREETMPS; LEAVE; 455 FREETMPS; LEAVE;
279 } 456 }
457 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)
458 {
459 dSP;
460
461 ENTER; SAVETMPS; PUSHMARK (SP);
462 EXTEND (SP, 2);
463 // we re-bless the reference to get overload and other niceties right
464 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
465 PUSHs (sv_cbor);
466
467 PUTBACK;
468 int count = call_sv ((SV *)GvCV (method), G_ARRAY);
469 SPAGAIN;
470
471 // catch this surprisingly common error
472 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)
473 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));
474
475 encode_tag (enc, CBOR_TAG_PERL_OBJECT);
476 encode_uint (enc, MAJOR_ARRAY, count + 1);
477 encode_strref (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
478
479 while (count)
480 encode_sv (enc, SP[1 - count--]);
481
482 PUTBACK;
483
484 FREETMPS; LEAVE;
485 }
280 else 486 else
281 croak ("encountered object '%s', but no TO_CBOR method available on it", 487 croak ("encountered object '%s', but no TO_CBOR or FREEZE methods available on it",
282 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 488 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
283 }
284 } 489 }
285 else if (svt == SVt_PVHV) 490 else if (svt == SVt_PVHV)
286 encode_hv (enc, (HV *)sv); 491 encode_hv (enc, (HV *)sv);
287 else if (svt == SVt_PVAV) 492 else if (svt == SVt_PVAV)
288 encode_av (enc, (AV *)sv); 493 encode_av (enc, (AV *)sv);
289 else if (svt < SVt_PVAV)
290 {
291 STRLEN len = 0;
292 char *pv = svt ? SvPV (sv, len) : 0;
293
294 if (len == 1 && *pv == '1')
295 encode_ch (enc, 0xe0 | 21);
296 else if (len == 1 && *pv == '0')
297 encode_ch (enc, 0xe0 | 20);
298 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
299 encode_ch (enc, 0xe0 | 23);
300 else
301 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
302 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
303 }
304 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
305 encode_ch (enc, 0xe0 | 23);
306 else 494 else
307 croak ("encountered %s, but CBOR can only represent references to arrays or hashes", 495 {
308 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 496 encode_tag (enc, CBOR_TAG_INDIRECTION);
497 encode_sv (enc, sv);
498 }
309} 499}
310 500
311static void 501static void
312encode_nv (enc_t *enc, SV *sv) 502encode_nv (enc_t *enc, SV *sv)
313{ 503{
314 double nv = SvNVX (sv); 504 double nv = SvNVX (sv);
315 505
316 need (enc, 9); 506 need (enc, 9);
317 507
318 if (ecb_expect_false (nv == (U32)nv)) 508 if (ecb_expect_false (nv == (NV)(U32)nv))
319 encode_uint (enc, 0x00, (U32)nv); 509 encode_uint (enc, MAJOR_POS_INT, (U32)nv);
320 //TODO: maybe I32? 510 //TODO: maybe I32?
321 else if (ecb_expect_false (nv == (float)nv)) 511 else if (ecb_expect_false (nv == (float)nv))
322 { 512 {
323 uint32_t fp = ecb_float_to_binary32 (nv); 513 uint32_t fp = ecb_float_to_binary32 (nv);
324 514
325 *enc->cur++ = 0xe0 | 26; 515 *enc->cur++ = MAJOR_MISC | MISC_FLOAT32;
326 516
327 if (!ecb_big_endian ()) 517 if (!ecb_big_endian ())
328 fp = ecb_bswap32 (fp); 518 fp = ecb_bswap32 (fp);
329 519
330 memcpy (enc->cur, &fp, 4); 520 memcpy (enc->cur, &fp, 4);
332 } 522 }
333 else 523 else
334 { 524 {
335 uint64_t fp = ecb_double_to_binary64 (nv); 525 uint64_t fp = ecb_double_to_binary64 (nv);
336 526
337 *enc->cur++ = 0xe0 | 27; 527 *enc->cur++ = MAJOR_MISC | MISC_FLOAT64;
338 528
339 if (!ecb_big_endian ()) 529 if (!ecb_big_endian ())
340 fp = ecb_bswap64 (fp); 530 fp = ecb_bswap64 (fp);
341 531
342 memcpy (enc->cur, &fp, 8); 532 memcpy (enc->cur, &fp, 8);
351 541
352 if (SvPOKp (sv)) 542 if (SvPOKp (sv))
353 { 543 {
354 STRLEN len; 544 STRLEN len;
355 char *str = SvPV (sv, len); 545 char *str = SvPV (sv, len);
356 encode_str (enc, SvUTF8 (sv), str, len); 546 encode_strref (enc, SvUTF8 (sv), str, len);
357 } 547 }
358 else if (SvNOKp (sv)) 548 else if (SvNOKp (sv))
359 encode_nv (enc, sv); 549 encode_nv (enc, sv);
360 else if (SvIOKp (sv)) 550 else if (SvIOKp (sv))
361 { 551 {
362 if (SvIsUV (sv)) 552 if (SvIsUV (sv))
363 encode_uint (enc, 0x00, SvUVX (sv)); 553 encode_uint (enc, MAJOR_POS_INT, SvUVX (sv));
364 else if (SvIVX (sv) >= 0) 554 else if (SvIVX (sv) >= 0)
365 encode_uint (enc, 0x00, SvIVX (sv)); 555 encode_uint (enc, MAJOR_POS_INT, SvIVX (sv));
366 else 556 else
367 encode_uint (enc, 0x20, -(SvIVX (sv) + 1)); 557 encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1));
368 } 558 }
369 else if (SvROK (sv)) 559 else if (SvROK (sv))
370 encode_rv (enc, SvRV (sv)); 560 encode_rv (enc, SvRV (sv));
371 else if (!SvOK (sv)) 561 else if (!SvOK (sv))
372 encode_ch (enc, 0xe0 | 22); 562 encode_ch (enc, MAJOR_MISC | SIMPLE_NULL);
373 else if (enc->cbor.flags & F_ALLOW_UNKNOWN) 563 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
374 encode_ch (enc, 0xe0 | 23); 564 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
375 else 565 else
376 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data", 566 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data",
377 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv)); 567 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
378} 568}
379 569
380static SV * 570static SV *
381encode_cbor (SV *scalar, CBOR *cbor) 571encode_cbor (SV *scalar, CBOR *cbor)
382{ 572{
383 enc_t enc; 573 enc_t enc = { };
384 574
385 enc.cbor = *cbor; 575 enc.cbor = *cbor;
386 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 576 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
387 enc.cur = SvPVX (enc.sv); 577 enc.cur = SvPVX (enc.sv);
388 enc.end = SvEND (enc.sv); 578 enc.end = SvEND (enc.sv);
389 enc.depth = 0;
390 579
391 SvPOK_only (enc.sv); 580 SvPOK_only (enc.sv);
581
582 if (cbor->flags & F_PACK_STRINGS)
583 {
584 encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE);
585 enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ());
586 enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ());
587 }
588
392 encode_sv (&enc, scalar); 589 encode_sv (&enc, scalar);
393 590
394 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 591 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
395 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings 592 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
396 593
410 U8 *end; // end of input string 607 U8 *end; // end of input string
411 const char *err; // parse error, if != 0 608 const char *err; // parse error, if != 0
412 CBOR cbor; 609 CBOR cbor;
413 U32 depth; // recursion depth 610 U32 depth; // recursion depth
414 U32 maxdepth; // recursion depth limit 611 U32 maxdepth; // recursion depth limit
612 AV *shareable;
613 AV *stringref;
614 SV *decode_tagged;
415} dec_t; 615} dec_t;
416 616
417#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE 617#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE
418 618
419#define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data") 619#define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data")
422#define DEC_DEC_DEPTH --dec->depth 622#define DEC_DEC_DEPTH --dec->depth
423 623
424static UV 624static UV
425decode_uint (dec_t *dec) 625decode_uint (dec_t *dec)
426{ 626{
427 switch (*dec->cur & 31) 627 U8 m = *dec->cur & MINOR_MASK;
428 { 628 ++dec->cur;
429 case 0: case 1: case 2: case 3: case 4: case 5: case 6: case 7:
430 case 8: case 9: case 10: case 11: case 12: case 13: case 14: case 15:
431 case 16: case 17: case 18: case 19: case 20: case 21: case 22: case 23:
432 return *dec->cur++ & 31;
433 629
434 case 24: 630 if (ecb_expect_true (m < LENGTH_EXT1))
631 return m;
632 else if (ecb_expect_true (m == LENGTH_EXT1))
633 {
435 WANT (2); 634 WANT (1);
436 dec->cur += 2; 635 dec->cur += 1;
437 return dec->cur[-1]; 636 return dec->cur[-1];
438 637 }
439 case 25: 638 else if (ecb_expect_true (m == LENGTH_EXT2))
639 {
440 WANT (3); 640 WANT (2);
441 dec->cur += 3; 641 dec->cur += 2;
442 return (((UV)dec->cur[-2]) << 8) 642 return (((UV)dec->cur[-2]) << 8)
443 | ((UV)dec->cur[-1]); 643 | ((UV)dec->cur[-1]);
444 644 }
445 case 26: 645 else if (ecb_expect_true (m == LENGTH_EXT4))
646 {
446 WANT (5); 647 WANT (4);
447 dec->cur += 5; 648 dec->cur += 4;
448 return (((UV)dec->cur[-4]) << 24) 649 return (((UV)dec->cur[-4]) << 24)
449 | (((UV)dec->cur[-3]) << 16) 650 | (((UV)dec->cur[-3]) << 16)
450 | (((UV)dec->cur[-2]) << 8) 651 | (((UV)dec->cur[-2]) << 8)
451 | ((UV)dec->cur[-1]); 652 | ((UV)dec->cur[-1]);
452 653 }
453 case 27: 654 else if (ecb_expect_true (m == LENGTH_EXT8))
655 {
454 WANT (9); 656 WANT (8);
455 dec->cur += 9; 657 dec->cur += 8;
658
659 return
660#if UVSIZE < 8
661 0
662#else
456 return (((UV)dec->cur[-8]) << 56) 663 (((UV)dec->cur[-8]) << 56)
457 | (((UV)dec->cur[-7]) << 48) 664 | (((UV)dec->cur[-7]) << 48)
458 | (((UV)dec->cur[-6]) << 40) 665 | (((UV)dec->cur[-6]) << 40)
459 | (((UV)dec->cur[-5]) << 32) 666 | (((UV)dec->cur[-5]) << 32)
667#endif
460 | (((UV)dec->cur[-4]) << 24) 668 | (((UV)dec->cur[-4]) << 24)
461 | (((UV)dec->cur[-3]) << 16) 669 | (((UV)dec->cur[-3]) << 16)
462 | (((UV)dec->cur[-2]) << 8) 670 | (((UV)dec->cur[-2]) << 8)
463 | ((UV)dec->cur[-1]); 671 | ((UV)dec->cur[-1]);
464 672 }
465 default: 673 else
466 ERR ("corrupted CBOR data (unsupported integer minor encoding)"); 674 ERR ("corrupted CBOR data (unsupported integer minor encoding)");
467 }
468 675
469fail: 676fail:
470 return 0; 677 return 0;
471} 678}
472 679
477{ 684{
478 AV *av = newAV (); 685 AV *av = newAV ();
479 686
480 DEC_INC_DEPTH; 687 DEC_INC_DEPTH;
481 688
482 if ((*dec->cur & 31) == 31) 689 if (*dec->cur == (MAJOR_ARRAY | MINOR_INDEF))
483 { 690 {
484 ++dec->cur; 691 ++dec->cur;
485 692
486 for (;;) 693 for (;;)
487 { 694 {
488 WANT (1); 695 WANT (1);
489 696
490 if (*dec->cur == (0xe0 | 31)) 697 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
491 { 698 {
492 ++dec->cur; 699 ++dec->cur;
493 break; 700 break;
494 } 701 }
495 702
498 } 705 }
499 else 706 else
500 { 707 {
501 int i, len = decode_uint (dec); 708 int i, len = decode_uint (dec);
502 709
710 WANT (len); // complexity check for av_fill - need at least one byte per value, do not allow supersize arrays
503 av_fill (av, len - 1); 711 av_fill (av, len - 1);
504 712
505 for (i = 0; i < len; ++i) 713 for (i = 0; i < len; ++i)
506 AvARRAY (av)[i] = decode_sv (dec); 714 AvARRAY (av)[i] = decode_sv (dec);
507 } 715 }
513 SvREFCNT_dec (av); 721 SvREFCNT_dec (av);
514 DEC_DEC_DEPTH; 722 DEC_DEC_DEPTH;
515 return &PL_sv_undef; 723 return &PL_sv_undef;
516} 724}
517 725
726static void
727decode_he (dec_t *dec, HV *hv)
728{
729 // for speed reasons, we specialcase single-string
730 // byte or utf-8 strings as keys, but only when !stringref
731
732 if (ecb_expect_true (!dec->stringref))
733 if (ecb_expect_true ((*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8))
734 {
735 I32 len = decode_uint (dec);
736 char *key = (char *)dec->cur;
737
738 dec->cur += len;
739
740 if (ecb_expect_false (dec->stringref))
741 av_push (dec->stringref, newSVpvn (key, len));
742
743 hv_store (hv, key, len, decode_sv (dec), 0);
744
745 return;
746 }
747 else if (ecb_expect_true ((*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8))
748 {
749 I32 len = decode_uint (dec);
750 char *key = (char *)dec->cur;
751
752 dec->cur += len;
753
754 if (ecb_expect_false (dec->stringref))
755 av_push (dec->stringref, newSVpvn_utf8 (key, len, 1));
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}
768
518static SV * 769static SV *
519decode_hv (dec_t *dec) 770decode_hv (dec_t *dec)
520{ 771{
521 HV *hv = newHV (); 772 HV *hv = newHV ();
522 773
523 DEC_INC_DEPTH; 774 DEC_INC_DEPTH;
524 775
525 if ((*dec->cur & 31) == 31) 776 if (*dec->cur == (MAJOR_MAP | MINOR_INDEF))
526 { 777 {
527 ++dec->cur; 778 ++dec->cur;
528 779
529 for (;;) 780 for (;;)
530 { 781 {
531 WANT (1); 782 WANT (1);
532 783
533 if (*dec->cur == (0xe0 | 31)) 784 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
534 { 785 {
535 ++dec->cur; 786 ++dec->cur;
536 break; 787 break;
537 } 788 }
538 789
539 SV *k = decode_sv (dec); 790 decode_he (dec, hv);
540 SV *v = decode_sv (dec);
541
542 hv_store_ent (hv, k, v, 0);
543 } 791 }
544 } 792 }
545 else 793 else
546 { 794 {
547 int len = decode_uint (dec); 795 int pairs = decode_uint (dec);
548 796
549 while (len--) 797 while (pairs--)
550 { 798 decode_he (dec, hv);
551 SV *k = decode_sv (dec);
552 SV *v = decode_sv (dec);
553
554 hv_store_ent (hv, k, v, 0);
555 }
556 } 799 }
557 800
558 DEC_DEC_DEPTH; 801 DEC_DEC_DEPTH;
559 return newRV_noinc ((SV *)hv); 802 return newRV_noinc ((SV *)hv);
560 803
567static SV * 810static SV *
568decode_str (dec_t *dec, int utf8) 811decode_str (dec_t *dec, int utf8)
569{ 812{
570 SV *sv = 0; 813 SV *sv = 0;
571 814
572 if ((*dec->cur & 31) == 31) 815 if ((*dec->cur & MINOR_MASK) == MINOR_INDEF)
573 { 816 {
817 // indefinite length strings
574 ++dec->cur; 818 ++dec->cur;
575 819
820 U8 major = *dec->cur & MAJOR_MISC;
821
576 sv = newSVpvn ("", 0); 822 sv = newSVpvn ("", 0);
577 823
578 // not very fast, and certainly not robust against illegal input
579 for (;;) 824 for (;;)
580 { 825 {
581 WANT (1); 826 WANT (1);
582 827
583 if (*dec->cur == (0xe0 | 31)) 828 if ((*dec->cur - major) > LENGTH_EXT8)
829 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
584 { 830 {
585 ++dec->cur; 831 ++dec->cur;
586 break; 832 break;
587 } 833 }
834 else
835 ERR ("corrupted CBOR data (invalid chunks in indefinite length string)");
588 836
589 sv_catsv (sv, decode_sv (dec)); 837 STRLEN len = decode_uint (dec);
838
839 WANT (len);
840 sv_catpvn (sv, dec->cur, len);
841 dec->cur += len;
590 } 842 }
591 } 843 }
592 else 844 else
593 { 845 {
594 STRLEN len = decode_uint (dec); 846 STRLEN len = decode_uint (dec);
595 847
596 WANT (len); 848 WANT (len);
597 sv = newSVpvn (dec->cur, len); 849 sv = newSVpvn (dec->cur, len);
598 dec->cur += len; 850 dec->cur += len;
851
852 if (ecb_expect_false (dec->stringref)
853 && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1))
854 av_push (dec->stringref, SvREFCNT_inc_NN (sv));
599 } 855 }
600 856
601 if (utf8) 857 if (utf8)
602 SvUTF8_on (sv); 858 SvUTF8_on (sv);
603 859
609} 865}
610 866
611static SV * 867static SV *
612decode_tagged (dec_t *dec) 868decode_tagged (dec_t *dec)
613{ 869{
870 SV *sv = 0;
614 UV tag = decode_uint (dec); 871 UV tag = decode_uint (dec);
872
873 WANT (1);
874
875 switch (tag)
876 {
877 case CBOR_TAG_MAGIC:
878 sv = decode_sv (dec);
879 break;
880
881 case CBOR_TAG_INDIRECTION:
882 sv = newRV_noinc (decode_sv (dec));
883 break;
884
885 case CBOR_TAG_STRINGREF_NAMESPACE:
886 {
887 ENTER; SAVETMPS;
888
889 SAVESPTR (dec->stringref);
890 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ());
891
892 sv = decode_sv (dec);
893
894 FREETMPS; LEAVE;
895 }
896 break;
897
898 case CBOR_TAG_STRINGREF:
899 {
900 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
901 ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
902
903 UV idx = decode_uint (dec);
904
905 if (!dec->stringref || (int)idx > AvFILLp (dec->stringref))
906 ERR ("corrupted CBOR data (stringref index out of bounds or outside namespace)");
907
908 sv = newSVsv (AvARRAY (dec->stringref)[idx]);
909 }
910 break;
911
912 case CBOR_TAG_VALUE_SHAREABLE:
913 {
914 if (ecb_expect_false (!dec->shareable))
915 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ());
916
917 sv = newSV (0);
918 av_push (dec->shareable, SvREFCNT_inc_NN (sv));
919
615 SV *sv = decode_sv (dec); 920 SV *osv = decode_sv (dec);
921 sv_setsv (sv, osv);
922 SvREFCNT_dec_NN (osv);
923 }
924 break;
616 925
617 if (tag == 55799) // 2.4.5 Self-Describe CBOR 926 case CBOR_TAG_VALUE_SHAREDREF:
927 {
928 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
929 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
930
931 UV idx = decode_uint (dec);
932
933 if (!dec->shareable || (int)idx > AvFILLp (dec->shareable))
934 ERR ("corrupted CBOR data (sharedref index out of bounds)");
935
936 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]);
937 }
938 break;
939
940 case CBOR_TAG_PERL_OBJECT:
941 {
942 sv = decode_sv (dec);
943
944 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
945 ERR ("corrupted CBOR data (non-array perl object)");
946
947 AV *av = (AV *)SvRV (sv);
948 int len = av_len (av) + 1;
949 HV *stash = gv_stashsv (*av_fetch (av, 0, 1), 0);
950
951 if (!stash)
952 ERR ("cannot decode perl-object (package does not exist)");
953
954 GV *method = gv_fetchmethod_autoload (stash, "THAW", 0);
955
956 if (!method)
957 ERR ("cannot decode perl-object (package does not have a THAW method)");
958
959 dSP;
960
961 ENTER; SAVETMPS; PUSHMARK (SP);
962 EXTEND (SP, len + 1);
963 // we re-bless the reference to get overload and other niceties right
964 PUSHs (*av_fetch (av, 0, 1));
965 PUSHs (sv_cbor);
966
967 int i;
968
969 for (i = 1; i < len; ++i)
970 PUSHs (*av_fetch (av, i, 1));
971
972 PUTBACK;
973 call_sv ((SV *)GvCV (method), G_SCALAR | G_EVAL);
974 SPAGAIN;
975
976 if (SvTRUE (ERRSV))
977 {
978 FREETMPS; LEAVE;
979 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV))));
980 }
981
982 SvREFCNT_dec (sv);
983 sv = SvREFCNT_inc (POPs);
984
985 PUTBACK;
986
987 FREETMPS; LEAVE;
988 }
989 break;
990
991 default:
992 {
993 sv = decode_sv (dec);
994
995 dSP;
996 ENTER; SAVETMPS; PUSHMARK (SP);
997 EXTEND (SP, 2);
998 PUSHs (newSVuv (tag));
999 PUSHs (sv);
1000
1001 PUTBACK;
1002 int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL);
1003 SPAGAIN;
1004
1005 if (SvTRUE (ERRSV))
1006 {
1007 FREETMPS; LEAVE;
1008 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV))));
1009 }
1010
1011 if (count)
1012 {
1013 SvREFCNT_dec (sv);
1014 sv = SvREFCNT_inc (POPs);
1015 }
1016 else
1017 {
1018 AV *av = newAV ();
1019 av_push (av, newSVuv (tag));
1020 av_push (av, sv);
1021
1022 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
1023 ? cbor_tagged_stash
1024 : gv_stashpv ("CBOR::XS::Tagged" , 1);
1025 sv = sv_bless (newRV_noinc ((SV *)av), tagged_stash);
1026 }
1027
1028 PUTBACK;
1029
1030 FREETMPS; LEAVE;
1031 }
1032 break;
1033 }
1034
618 return sv; 1035 return sv;
619 1036
620 AV *av = newAV (); 1037fail:
621 av_push (av, newSVuv (tag)); 1038 SvREFCNT_dec (sv);
622 av_push (av, sv); 1039 return &PL_sv_undef;
623 return newRV_noinc ((SV *)av);
624} 1040}
625 1041
626static SV * 1042static SV *
627decode_sv (dec_t *dec) 1043decode_sv (dec_t *dec)
628{ 1044{
629 WANT (1); 1045 WANT (1);
630 1046
631 switch (*dec->cur >> 5) 1047 switch (*dec->cur >> MAJOR_SHIFT)
632 { 1048 {
633 case 0: // unsigned int 1049 case MAJOR_POS_INT >> MAJOR_SHIFT: return newSVuv (decode_uint (dec));
634 return newSVuv (decode_uint (dec)); 1050 case MAJOR_NEG_INT >> MAJOR_SHIFT: return newSViv (-1 - (IV)decode_uint (dec));
635 case 1: // negative int 1051 case MAJOR_BYTES >> MAJOR_SHIFT: return decode_str (dec, 0);
636 return newSViv (-1 - (IV)decode_uint (dec)); 1052 case MAJOR_TEXT >> MAJOR_SHIFT: return decode_str (dec, 1);
637 case 2: // octet string 1053 case MAJOR_ARRAY >> MAJOR_SHIFT: return decode_av (dec);
638 return decode_str (dec, 0); 1054 case MAJOR_MAP >> MAJOR_SHIFT: return decode_hv (dec);
639 case 3: // utf-8 string 1055 case MAJOR_TAG >> MAJOR_SHIFT: return decode_tagged (dec);
640 return decode_str (dec, 1); 1056
641 case 4: // array 1057 case MAJOR_MISC >> MAJOR_SHIFT:
642 return decode_av (dec);
643 case 5: // map
644 return decode_hv (dec);
645 case 6: // tag
646 return decode_tagged (dec);
647 case 7: // misc
648 switch (*dec->cur++ & 31) 1058 switch (*dec->cur++ & MINOR_MASK)
649 { 1059 {
650 case 20: 1060 case SIMPLE_FALSE:
651#if CBOR_SLOW 1061#if CBOR_SLOW
652 cbor_false = get_bool ("CBOR::XS::false"); 1062 types_false = get_bool ("Types::Serialiser::false");
653#endif 1063#endif
654 return newSVsv (cbor_false); 1064 return newSVsv (types_false);
655 case 21: 1065 case SIMPLE_TRUE:
656#if CBOR_SLOW 1066#if CBOR_SLOW
657 cbor_true = get_bool ("CBOR::XS::true"); 1067 types_true = get_bool ("Types::Serialiser::true");
658#endif 1068#endif
659 return newSVsv (cbor_true); 1069 return newSVsv (types_true);
660 case 22: 1070 case SIMPLE_NULL:
661 return newSVsv (&PL_sv_undef); 1071 return newSVsv (&PL_sv_undef);
1072 case SIMPLE_UNDEF:
1073#if CBOR_SLOW
1074 types_error = get_bool ("Types::Serialiser::error");
1075#endif
1076 return newSVsv (types_error);
662 1077
663 case 25: 1078 case MISC_FLOAT16:
664 { 1079 {
665 WANT (2); 1080 WANT (2);
666 1081
667 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1]; 1082 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1];
668 dec->cur += 2; 1083 dec->cur += 2;
669 1084
670 return newSVnv (ecb_binary16_to_float (fp)); 1085 return newSVnv (ecb_binary16_to_float (fp));
671 } 1086 }
672 1087
673 case 26: 1088 case MISC_FLOAT32:
674 { 1089 {
675 uint32_t fp; 1090 uint32_t fp;
676 WANT (4); 1091 WANT (4);
677 memcpy (&fp, dec->cur, 4); 1092 memcpy (&fp, dec->cur, 4);
678 dec->cur += 4; 1093 dec->cur += 4;
681 fp = ecb_bswap32 (fp); 1096 fp = ecb_bswap32 (fp);
682 1097
683 return newSVnv (ecb_binary32_to_float (fp)); 1098 return newSVnv (ecb_binary32_to_float (fp));
684 } 1099 }
685 1100
686 case 27: 1101 case MISC_FLOAT64:
687 { 1102 {
688 uint64_t fp; 1103 uint64_t fp;
689 WANT (8); 1104 WANT (8);
690 memcpy (&fp, dec->cur, 8); 1105 memcpy (&fp, dec->cur, 8);
691 dec->cur += 8; 1106 dec->cur += 8;
694 fp = ecb_bswap64 (fp); 1109 fp = ecb_bswap64 (fp);
695 1110
696 return newSVnv (ecb_binary64_to_double (fp)); 1111 return newSVnv (ecb_binary64_to_double (fp));
697 } 1112 }
698 1113
699 // 0..19 unassigned 1114 // 0..19 unassigned simple
700 // 24 reserved + unassigned (reserved values are not encodable) 1115 // 24 reserved + unassigned (reserved values are not encodable)
701 default: 1116 default:
702 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)"); 1117 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)");
703 } 1118 }
704 1119
710} 1125}
711 1126
712static SV * 1127static SV *
713decode_cbor (SV *string, CBOR *cbor, char **offset_return) 1128decode_cbor (SV *string, CBOR *cbor, char **offset_return)
714{ 1129{
715 dec_t dec; 1130 dec_t dec = { };
716 SV *sv; 1131 SV *sv;
1132 STRLEN len;
1133 char *data = SvPVbyte (string, len);
717 1134
718 /* work around bugs in 5.10 where manipulating magic values
719 * makes perl ignore the magic in subsequent accesses.
720 * also make a copy of non-PV values, to get them into a clean
721 * state (SvPV should do that, but it's buggy, see below).
722 */
723 /*SvGETMAGIC (string);*/
724 if (SvMAGICAL (string) || !SvPOK (string))
725 string = sv_2mortal (newSVsv (string));
726
727 SvUPGRADE (string, SVt_PV);
728
729 /* work around a bug in perl 5.10, which causes SvCUR to fail an
730 * assertion with -DDEBUGGING, although SvCUR is documented to
731 * return the xpv_cur field which certainly exists after upgrading.
732 * according to nicholas clark, calling SvPOK fixes this.
733 * But it doesn't fix it, so try another workaround, call SvPV_nolen
734 * and hope for the best.
735 * Damnit, SvPV_nolen still trips over yet another assertion. This
736 * assertion business is seriously broken, try yet another workaround
737 * for the broken -DDEBUGGING.
738 */
739 {
740#ifdef DEBUGGING
741 STRLEN offset = SvOK (string) ? sv_len (string) : 0;
742#else
743 STRLEN offset = SvCUR (string);
744#endif
745
746 if (offset > cbor->max_size && cbor->max_size) 1135 if (len > cbor->max_size && cbor->max_size)
747 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu", 1136 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
748 (unsigned long)SvCUR (string), (unsigned long)cbor->max_size); 1137 (unsigned long)len, (unsigned long)cbor->max_size);
749 }
750
751 sv_utf8_downgrade (string, 0);
752 1138
753 dec.cbor = *cbor; 1139 dec.cbor = *cbor;
754 dec.cur = (U8 *)SvPVX (string); 1140 dec.cur = (U8 *)data;
755 dec.end = (U8 *)SvEND (string); 1141 dec.end = (U8 *)data + len;
756 dec.err = 0;
757 dec.depth = 0;
758 1142
759 sv = decode_sv (&dec); 1143 sv = decode_sv (&dec);
760 1144
761 if (offset_return) 1145 if (offset_return)
762 *offset_return = dec.cur; 1146 *offset_return = dec.cur;
766 dec.err = "garbage after CBOR object"; 1150 dec.err = "garbage after CBOR object";
767 1151
768 if (dec.err) 1152 if (dec.err)
769 { 1153 {
770 SvREFCNT_dec (sv); 1154 SvREFCNT_dec (sv);
771 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)SvPVX (string), (int)(uint8_t)*dec.cur); 1155 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur);
772 } 1156 }
773 1157
774 sv = sv_2mortal (sv); 1158 sv = sv_2mortal (sv);
775 1159
776 return sv; 1160 return sv;
782MODULE = CBOR::XS PACKAGE = CBOR::XS 1166MODULE = CBOR::XS PACKAGE = CBOR::XS
783 1167
784BOOT: 1168BOOT:
785{ 1169{
786 cbor_stash = gv_stashpv ("CBOR::XS" , 1); 1170 cbor_stash = gv_stashpv ("CBOR::XS" , 1);
787 cbor_boolean_stash = gv_stashpv ("CBOR::XS::Boolean", 1);
788 cbor_tagged_stash = gv_stashpv ("CBOR::XS::Tagged" , 1); 1171 cbor_tagged_stash = gv_stashpv ("CBOR::XS::Tagged" , 1);
789 1172
790 cbor_true = get_bool ("CBOR::XS::true"); 1173 types_boolean_stash = gv_stashpv ("Types::Serialiser::Boolean", 1);
791 cbor_false = get_bool ("CBOR::XS::false"); 1174 types_error_stash = gv_stashpv ("Types::Serialiser::Error" , 1);
1175
1176 types_true = get_bool ("Types::Serialiser::true" );
1177 types_false = get_bool ("Types::Serialiser::false");
1178 types_error = get_bool ("Types::Serialiser::error");
1179
1180 default_filter = newSVpv ("CBOR::XS::default_filter", 0);
1181
1182 sv_cbor = newSVpv ("CBOR", 0);
1183 SvREADONLY_on (sv_cbor);
792} 1184}
793 1185
794PROTOTYPES: DISABLE 1186PROTOTYPES: DISABLE
795 1187
796void CLONE (...) 1188void CLONE (...)
797 CODE: 1189 CODE:
798 cbor_stash = 0; 1190 cbor_stash = 0;
799 cbor_boolean_stash = 0;
800 cbor_tagged_stash = 0; 1191 cbor_tagged_stash = 0;
1192 types_error_stash = 0;
1193 types_boolean_stash = 0;
801 1194
802void new (char *klass) 1195void new (char *klass)
803 PPCODE: 1196 PPCODE:
804{ 1197{
805 SV *pv = NEWSV (0, sizeof (CBOR)); 1198 SV *pv = NEWSV (0, sizeof (CBOR));
813 1206
814void shrink (CBOR *self, int enable = 1) 1207void shrink (CBOR *self, int enable = 1)
815 ALIAS: 1208 ALIAS:
816 shrink = F_SHRINK 1209 shrink = F_SHRINK
817 allow_unknown = F_ALLOW_UNKNOWN 1210 allow_unknown = F_ALLOW_UNKNOWN
1211 allow_sharing = F_ALLOW_SHARING
1212 pack_strings = F_PACK_STRINGS
818 PPCODE: 1213 PPCODE:
819{ 1214{
820 if (enable) 1215 if (enable)
821 self->flags |= ix; 1216 self->flags |= ix;
822 else 1217 else
827 1222
828void get_shrink (CBOR *self) 1223void get_shrink (CBOR *self)
829 ALIAS: 1224 ALIAS:
830 get_shrink = F_SHRINK 1225 get_shrink = F_SHRINK
831 get_allow_unknown = F_ALLOW_UNKNOWN 1226 get_allow_unknown = F_ALLOW_UNKNOWN
1227 get_allow_sharing = F_ALLOW_SHARING
1228 get_pack_strings = F_PACK_STRINGS
832 PPCODE: 1229 PPCODE:
833 XPUSHs (boolSV (self->flags & ix)); 1230 XPUSHs (boolSV (self->flags & ix));
834 1231
835void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) 1232void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
836 PPCODE: 1233 PPCODE:
852 CODE: 1249 CODE:
853 RETVAL = self->max_size; 1250 RETVAL = self->max_size;
854 OUTPUT: 1251 OUTPUT:
855 RETVAL 1252 RETVAL
856 1253
857#if 0 //TODO 1254void filter (CBOR *self, SV *filter = 0)
858
859void filter_cbor_object (CBOR *self, SV *cb = &PL_sv_undef)
860 PPCODE: 1255 PPCODE:
861{
862 SvREFCNT_dec (self->cb_object); 1256 SvREFCNT_dec (self->filter);
863 self->cb_object = SvOK (cb) ? newSVsv (cb) : 0; 1257 self->filter = filter ? newSVsv (filter) : filter;
864
865 XPUSHs (ST (0)); 1258 XPUSHs (ST (0));
866}
867 1259
868void filter_cbor_single_key_object (CBOR *self, SV *key, SV *cb = &PL_sv_undef) 1260SV *get_filter (CBOR *self)
869 PPCODE: 1261 CODE:
870{ 1262 RETVAL = self->filter ? self->filter : NEWSV (0, 0);
871 if (!self->cb_sk_object) 1263 OUTPUT:
872 self->cb_sk_object = newHV (); 1264 RETVAL
873
874 if (SvOK (cb))
875 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);
876 else
877 {
878 hv_delete_ent (self->cb_sk_object, key, G_DISCARD, 0);
879
880 if (!HvKEYS (self->cb_sk_object))
881 {
882 SvREFCNT_dec (self->cb_sk_object);
883 self->cb_sk_object = 0;
884 }
885 }
886
887 XPUSHs (ST (0));
888}
889
890#endif
891 1265
892void encode (CBOR *self, SV *scalar) 1266void encode (CBOR *self, SV *scalar)
893 PPCODE: 1267 PPCODE:
894 PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN; 1268 PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN;
895 XPUSHs (scalar); 1269 XPUSHs (scalar);
908 EXTEND (SP, 2); 1282 EXTEND (SP, 2);
909 PUSHs (sv); 1283 PUSHs (sv);
910 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); 1284 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
911} 1285}
912 1286
913#if 0
914
915void DESTROY (CBOR *self) 1287void DESTROY (CBOR *self)
916 CODE: 1288 PPCODE:
917 SvREFCNT_dec (self->cb_sk_object); 1289 cbor_free (self);
918 SvREFCNT_dec (self->cb_object);
919
920#endif
921 1290
922PROTOTYPES: ENABLE 1291PROTOTYPES: ENABLE
923 1292
924void encode_cbor (SV *scalar) 1293void encode_cbor (SV *scalar)
1294 ALIAS:
1295 encode_cbor = 0
1296 encode_cbor_sharing = F_ALLOW_SHARING
925 PPCODE: 1297 PPCODE:
926{ 1298{
927 CBOR cbor; 1299 CBOR cbor;
928 cbor_init (&cbor); 1300 cbor_init (&cbor);
1301 cbor.flags |= ix;
929 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN; 1302 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;
930 XPUSHs (scalar); 1303 XPUSHs (scalar);
931} 1304}
932 1305
933void decode_cbor (SV *cborstr) 1306void decode_cbor (SV *cborstr)

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines