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 1.9 by root, Sun Oct 27 10:17:12 2013 UTC vs.
Revision 1.35 by root, Sat Nov 30 17:19:34 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
14// known tags 65// known tags
15enum cbor_tag 66enum cbor_tag
16{ 67{
17 // inofficial extensions (pending iana registration) 68 // extensions
18 CBOR_TAG_PERL_OBJECT = 256, 69 CBOR_TAG_STRINGREF = 25, // http://cbor.schmorp.de/stringref
19 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
20 76
21 // rfc7049 77 // rfc7049
22 CBOR_TAG_DATETIME = 0, // rfc4287, utf-8 78 CBOR_TAG_DATETIME = 0, // rfc4287, utf-8
23 CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any 79 CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any
24 CBOR_TAG_POS_BIGNUM = 2, // byte string 80 CBOR_TAG_POS_BIGNUM = 2, // byte string
25 CBOR_TAG_NEG_BIGNUM = 3, // byte string 81 CBOR_TAG_NEG_BIGNUM = 3, // byte string
26 CBOR_TAG_DECIMAL = 4, // decimal fraction, array 82 CBOR_TAG_DECIMAL = 4, // decimal fraction, array
27 CBOR_TAG_BIGFLOAT = 5, // array 83 CBOR_TAG_BIGFLOAT = 5, // array
28 84
29 CBOR_TAG_CONV_B64U = 21, // base64url, any 85 CBOR_TAG_CONV_B64U = 21, // base64url, any
30 CBOR_TAG_CONV_B64 = 22, // base64, any 86 CBOR_TAG_CONV_B64 = 22, // base64, any
31 CBOR_TAG_CONV_HEX = 23, // base16, any 87 CBOR_TAG_CONV_HEX = 23, // base16, any
32 CBOR_TAG_CBOR = 24, // embedded cbor, byte string 88 CBOR_TAG_CBOR = 24, // embedded cbor, byte string
33 89
34 CBOR_TAG_URI = 32, // URI rfc3986, utf-8 90 CBOR_TAG_URI = 32, // URI rfc3986, utf-8
35 CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8 91 CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8
36 CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8 92 CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8
37 CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8 93 CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8
38 CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8 94 CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8
39 95
40 CBOR_TAG_MAGIC = 55799 // self-describe cbor 96 CBOR_TAG_MAGIC = 55799, // self-describe cbor
41}; 97};
42 98
43#define F_SHRINK 0x00000200UL 99#define F_SHRINK 0x00000001UL
44#define F_ALLOW_UNKNOWN 0x00002000UL 100#define F_ALLOW_UNKNOWN 0x00000002UL
101#define F_ALLOW_SHARING 0x00000004UL
102#define F_PACK_STRINGS 0x00000008UL
45 103
46#define INIT_SIZE 32 // initial scalar size to be allocated 104#define INIT_SIZE 32 // initial scalar size to be allocated
47 105
48#define SB do { 106#define SB do {
49#define SE } while (0) 107#define SE } while (0)
60#else 118#else
61# define CBOR_SLOW 0 119# define CBOR_SLOW 0
62# define CBOR_STASH cbor_stash 120# define CBOR_STASH cbor_stash
63#endif 121#endif
64 122
65static 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::
66static SV *cbor_true, *cbor_false; 124static SV *types_true, *types_false, *types_error, *sv_cbor, *default_filter;
67 125
68typedef struct { 126typedef struct {
69 U32 flags; 127 U32 flags;
70 U32 max_depth; 128 U32 max_depth;
71 STRLEN max_size; 129 STRLEN max_size;
130 SV *filter;
72} CBOR; 131} CBOR;
73 132
74ecb_inline void 133ecb_inline void
75cbor_init (CBOR *cbor) 134cbor_init (CBOR *cbor)
76{ 135{
77 Zero (cbor, 1, CBOR); 136 Zero (cbor, 1, CBOR);
78 cbor->max_depth = 512; 137 cbor->max_depth = 512;
138}
139
140ecb_inline void
141cbor_free (CBOR *cbor)
142{
143 SvREFCNT_dec (cbor->filter);
79} 144}
80 145
81///////////////////////////////////////////////////////////////////////////// 146/////////////////////////////////////////////////////////////////////////////
82// utility functions 147// utility functions
83 148
105 SvPV_renew (sv, SvCUR (sv) + 1); 170 SvPV_renew (sv, SvCUR (sv) + 1);
106#endif 171#endif
107 } 172 }
108} 173}
109 174
110///////////////////////////////////////////////////////////////////////////// 175// minimum length of a string to be registered for stringref
111// fp hell 176ecb_inline int
112 177minimum_string_length (UV idx)
113//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}
114 189
115///////////////////////////////////////////////////////////////////////////// 190/////////////////////////////////////////////////////////////////////////////
116// encoder 191// encoder
117 192
118// structure used for encoding CBOR 193// structure used for encoding CBOR
121 char *cur; // SvPVX (sv) + current output position 196 char *cur; // SvPVX (sv) + current output position
122 char *end; // SvEND (sv) 197 char *end; // SvEND (sv)
123 SV *sv; // result scalar 198 SV *sv; // result scalar
124 CBOR cbor; 199 CBOR cbor;
125 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;
126} enc_t; 205} enc_t;
127 206
128ecb_inline void 207ecb_inline void
129need (enc_t *enc, STRLEN len) 208need (enc_t *enc, STRLEN len)
130{ 209{
147static void 226static void
148encode_uint (enc_t *enc, int major, UV len) 227encode_uint (enc_t *enc, int major, UV len)
149{ 228{
150 need (enc, 9); 229 need (enc, 9);
151 230
152 if (len < 24) 231 if (ecb_expect_true (len < LENGTH_EXT1))
153 *enc->cur++ = major | len; 232 *enc->cur++ = major | len;
154 else if (len <= 0xff) 233 else if (ecb_expect_true (len <= 0xff))
155 { 234 {
156 *enc->cur++ = major | 24; 235 *enc->cur++ = major | LENGTH_EXT1;
157 *enc->cur++ = len; 236 *enc->cur++ = len;
158 } 237 }
159 else if (len <= 0xffff) 238 else if (len <= 0xffff)
160 { 239 {
161 *enc->cur++ = major | 25; 240 *enc->cur++ = major | LENGTH_EXT2;
162 *enc->cur++ = len >> 8; 241 *enc->cur++ = len >> 8;
163 *enc->cur++ = len; 242 *enc->cur++ = len;
164 } 243 }
165 else if (len <= 0xffffffff) 244 else if (len <= 0xffffffff)
166 { 245 {
167 *enc->cur++ = major | 26; 246 *enc->cur++ = major | LENGTH_EXT4;
168 *enc->cur++ = len >> 24; 247 *enc->cur++ = len >> 24;
169 *enc->cur++ = len >> 16; 248 *enc->cur++ = len >> 16;
170 *enc->cur++ = len >> 8; 249 *enc->cur++ = len >> 8;
171 *enc->cur++ = len; 250 *enc->cur++ = len;
172 } 251 }
173 else 252 else
174 { 253 {
175 *enc->cur++ = major | 27; 254 *enc->cur++ = major | LENGTH_EXT8;
176 *enc->cur++ = len >> 56; 255 *enc->cur++ = len >> 56;
177 *enc->cur++ = len >> 48; 256 *enc->cur++ = len >> 48;
178 *enc->cur++ = len >> 40; 257 *enc->cur++ = len >> 40;
179 *enc->cur++ = len >> 32; 258 *enc->cur++ = len >> 32;
180 *enc->cur++ = len >> 24; 259 *enc->cur++ = len >> 24;
182 *enc->cur++ = len >> 8; 261 *enc->cur++ = len >> 8;
183 *enc->cur++ = len; 262 *enc->cur++ = len;
184 } 263 }
185} 264}
186 265
187static void 266ecb_inline void
267encode_tag (enc_t *enc, UV tag)
268{
269 encode_uint (enc, MAJOR_TAG, tag);
270}
271
272ecb_inline void
188encode_str (enc_t *enc, int utf8, char *str, STRLEN len) 273encode_str (enc_t *enc, int utf8, char *str, STRLEN len)
189{ 274{
190 encode_uint (enc, utf8 ? 0x60 : 0x40, len); 275 encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len);
191 need (enc, len); 276 need (enc, len);
192 memcpy (enc->cur, str, len); 277 memcpy (enc->cur, str, len);
193 enc->cur += len; 278 enc->cur += len;
194} 279}
195 280
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
196static void encode_sv (enc_t *enc, SV *sv); 306static void encode_sv (enc_t *enc, SV *sv);
197 307
198static void 308static void
199encode_av (enc_t *enc, AV *av) 309encode_av (enc_t *enc, AV *av)
200{ 310{
203 if (enc->depth >= enc->cbor.max_depth) 313 if (enc->depth >= enc->cbor.max_depth)
204 croak (ERR_NESTING_EXCEEDED); 314 croak (ERR_NESTING_EXCEEDED);
205 315
206 ++enc->depth; 316 ++enc->depth;
207 317
208 encode_uint (enc, 0x80, len + 1); 318 encode_uint (enc, MAJOR_ARRAY, len + 1);
209 319
210 for (i = 0; i <= len; ++i) 320 for (i = 0; i <= len; ++i)
211 { 321 {
212 SV **svp = av_fetch (av, i, 0); 322 SV **svp = av_fetch (av, i, 0);
213 encode_sv (enc, svp ? *svp : &PL_sv_undef); 323 encode_sv (enc, svp ? *svp : &PL_sv_undef);
228 338
229 int pairs = hv_iterinit (hv); 339 int pairs = hv_iterinit (hv);
230 int mg = SvMAGICAL (hv); 340 int mg = SvMAGICAL (hv);
231 341
232 if (mg) 342 if (mg)
233 encode_ch (enc, 0xa0 | 31); 343 encode_ch (enc, MAJOR_MAP | MINOR_INDEF);
234 else 344 else
235 encode_uint (enc, 0xa0, pairs); 345 encode_uint (enc, MAJOR_MAP, pairs);
236 346
237 while ((he = hv_iternext (hv))) 347 while ((he = hv_iternext (hv)))
238 { 348 {
239 if (HeKLEN (he) == HEf_SVKEY) 349 if (HeKLEN (he) == HEf_SVKEY)
240 encode_sv (enc, HeSVKEY (he)); 350 encode_sv (enc, HeSVKEY (he));
241 else 351 else
242 encode_str (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he)); 352 encode_strref (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he));
243 353
244 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));
245 } 355 }
246 356
247 if (mg) 357 if (mg)
248 encode_ch (enc, 0xe0 | 31); 358 encode_ch (enc, MAJOR_MISC | MINOR_INDEF);
249 359
250 --enc->depth; 360 --enc->depth;
251} 361}
252 362
253// encode objects, arrays and special \0=false and \1=true values. 363// encode objects, arrays and special \0=false and \1=true values.
254static void 364static void
255encode_rv (enc_t *enc, SV *sv) 365encode_rv (enc_t *enc, SV *sv)
256{ 366{
257 svtype svt;
258
259 SvGETMAGIC (sv); 367 SvGETMAGIC (sv);
368
260 svt = SvTYPE (sv); 369 svtype svt = SvTYPE (sv);
261 370
262 if (ecb_expect_false (SvOBJECT (sv))) 371 if (ecb_expect_false (SvOBJECT (sv)))
263 { 372 {
264 HV *boolean_stash = !CBOR_SLOW || cbor_boolean_stash 373 HV *boolean_stash = !CBOR_SLOW || types_boolean_stash
265 ? cbor_boolean_stash 374 ? types_boolean_stash
266 : 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);
267 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 379 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
268 ? cbor_tagged_stash 380 ? cbor_tagged_stash
269 : gv_stashpv ("CBOR::XS::Tagged" , 1); 381 : gv_stashpv ("CBOR::XS::Tagged" , 1);
270 382
383 HV *stash = SvSTASH (sv);
384
271 if (SvSTASH (sv) == boolean_stash) 385 if (stash == boolean_stash)
272 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 }
273 else if (SvSTASH (sv) == tagged_stash) 395 else if (stash == tagged_stash)
274 { 396 {
275 if (svt != SVt_PVAV) 397 if (svt != SVt_PVAV)
276 croak ("encountered CBOR::XS::Tagged object that isn't an array"); 398 croak ("encountered CBOR::XS::Tagged object that isn't an array");
277 399
278 encode_uint (enc, 0xc0, SvUV (*av_fetch ((AV *)sv, 0, 1))); 400 encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1)));
279 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;
280 } 420 }
281 else 421 else
282 { 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);
283 // 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
284 GV *to_cbor = gv_fetchmethod_autoload (SvSTASH (sv), "TO_CBOR", 0);
285
286 if (to_cbor)
287 {
288 dSP;
289
290 ENTER; SAVETMPS; PUSHMARK (SP);
291 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv))); 440 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
292 441
293 // calling with G_SCALAR ensures that we always get a 1 return value
294 PUTBACK; 442 PUTBACK;
443 // G_SCALAR ensures that return value is 1
295 call_sv ((SV *)GvCV (to_cbor), G_SCALAR); 444 call_sv ((SV *)GvCV (method), G_SCALAR);
296 SPAGAIN; 445 SPAGAIN;
297 446
298 // catch this surprisingly common error 447 // catch this surprisingly common error
299 if (SvROK (TOPs) && SvRV (TOPs) == sv) 448 if (SvROK (TOPs) && SvRV (TOPs) == sv)
300 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));
301 450
302 sv = POPs; 451 encode_sv (enc, POPs);
452
303 PUTBACK; 453 PUTBACK;
304 454
305 encode_sv (enc, sv);
306
307 FREETMPS; LEAVE; 455 FREETMPS; LEAVE;
308 } 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 }
309 else 486 else
310 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",
311 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 488 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
312 }
313 } 489 }
314 else if (svt == SVt_PVHV) 490 else if (svt == SVt_PVHV)
315 encode_hv (enc, (HV *)sv); 491 encode_hv (enc, (HV *)sv);
316 else if (svt == SVt_PVAV) 492 else if (svt == SVt_PVAV)
317 encode_av (enc, (AV *)sv); 493 encode_av (enc, (AV *)sv);
318 else if (svt < SVt_PVAV)
319 {
320 STRLEN len = 0;
321 char *pv = svt ? SvPV (sv, len) : 0;
322
323 if (len == 1 && *pv == '1')
324 encode_ch (enc, 0xe0 | 21);
325 else if (len == 1 && *pv == '0')
326 encode_ch (enc, 0xe0 | 20);
327 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
328 encode_ch (enc, 0xe0 | 23);
329 else
330 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
331 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
332 }
333 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
334 encode_ch (enc, 0xe0 | 23);
335 else 494 else
336 croak ("encountered %s, but CBOR can only represent references to arrays or hashes", 495 {
337 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 496 encode_tag (enc, CBOR_TAG_INDIRECTION);
497 encode_sv (enc, sv);
498 }
338} 499}
339 500
340static void 501static void
341encode_nv (enc_t *enc, SV *sv) 502encode_nv (enc_t *enc, SV *sv)
342{ 503{
343 double nv = SvNVX (sv); 504 double nv = SvNVX (sv);
344 505
345 need (enc, 9); 506 need (enc, 9);
346 507
347 if (ecb_expect_false (nv == (U32)nv)) 508 if (ecb_expect_false (nv == (NV)(U32)nv))
348 encode_uint (enc, 0x00, (U32)nv); 509 encode_uint (enc, MAJOR_POS_INT, (U32)nv);
349 //TODO: maybe I32? 510 //TODO: maybe I32?
350 else if (ecb_expect_false (nv == (float)nv)) 511 else if (ecb_expect_false (nv == (float)nv))
351 { 512 {
352 uint32_t fp = ecb_float_to_binary32 (nv); 513 uint32_t fp = ecb_float_to_binary32 (nv);
353 514
354 *enc->cur++ = 0xe0 | 26; 515 *enc->cur++ = MAJOR_MISC | MISC_FLOAT32;
355 516
356 if (!ecb_big_endian ()) 517 if (!ecb_big_endian ())
357 fp = ecb_bswap32 (fp); 518 fp = ecb_bswap32 (fp);
358 519
359 memcpy (enc->cur, &fp, 4); 520 memcpy (enc->cur, &fp, 4);
361 } 522 }
362 else 523 else
363 { 524 {
364 uint64_t fp = ecb_double_to_binary64 (nv); 525 uint64_t fp = ecb_double_to_binary64 (nv);
365 526
366 *enc->cur++ = 0xe0 | 27; 527 *enc->cur++ = MAJOR_MISC | MISC_FLOAT64;
367 528
368 if (!ecb_big_endian ()) 529 if (!ecb_big_endian ())
369 fp = ecb_bswap64 (fp); 530 fp = ecb_bswap64 (fp);
370 531
371 memcpy (enc->cur, &fp, 8); 532 memcpy (enc->cur, &fp, 8);
380 541
381 if (SvPOKp (sv)) 542 if (SvPOKp (sv))
382 { 543 {
383 STRLEN len; 544 STRLEN len;
384 char *str = SvPV (sv, len); 545 char *str = SvPV (sv, len);
385 encode_str (enc, SvUTF8 (sv), str, len); 546 encode_strref (enc, SvUTF8 (sv), str, len);
386 } 547 }
387 else if (SvNOKp (sv)) 548 else if (SvNOKp (sv))
388 encode_nv (enc, sv); 549 encode_nv (enc, sv);
389 else if (SvIOKp (sv)) 550 else if (SvIOKp (sv))
390 { 551 {
391 if (SvIsUV (sv)) 552 if (SvIsUV (sv))
392 encode_uint (enc, 0x00, SvUVX (sv)); 553 encode_uint (enc, MAJOR_POS_INT, SvUVX (sv));
393 else if (SvIVX (sv) >= 0) 554 else if (SvIVX (sv) >= 0)
394 encode_uint (enc, 0x00, SvIVX (sv)); 555 encode_uint (enc, MAJOR_POS_INT, SvIVX (sv));
395 else 556 else
396 encode_uint (enc, 0x20, -(SvIVX (sv) + 1)); 557 encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1));
397 } 558 }
398 else if (SvROK (sv)) 559 else if (SvROK (sv))
399 encode_rv (enc, SvRV (sv)); 560 encode_rv (enc, SvRV (sv));
400 else if (!SvOK (sv)) 561 else if (!SvOK (sv))
401 encode_ch (enc, 0xe0 | 22); 562 encode_ch (enc, MAJOR_MISC | SIMPLE_NULL);
402 else if (enc->cbor.flags & F_ALLOW_UNKNOWN) 563 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
403 encode_ch (enc, 0xe0 | 23); 564 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
404 else 565 else
405 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",
406 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv)); 567 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
407} 568}
408 569
409static SV * 570static SV *
410encode_cbor (SV *scalar, CBOR *cbor) 571encode_cbor (SV *scalar, CBOR *cbor)
411{ 572{
412 enc_t enc; 573 enc_t enc = { };
413 574
414 enc.cbor = *cbor; 575 enc.cbor = *cbor;
415 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 576 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
416 enc.cur = SvPVX (enc.sv); 577 enc.cur = SvPVX (enc.sv);
417 enc.end = SvEND (enc.sv); 578 enc.end = SvEND (enc.sv);
418 enc.depth = 0;
419 579
420 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
421 encode_sv (&enc, scalar); 589 encode_sv (&enc, scalar);
422 590
423 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 591 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
424 *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
425 593
439 U8 *end; // end of input string 607 U8 *end; // end of input string
440 const char *err; // parse error, if != 0 608 const char *err; // parse error, if != 0
441 CBOR cbor; 609 CBOR cbor;
442 U32 depth; // recursion depth 610 U32 depth; // recursion depth
443 U32 maxdepth; // recursion depth limit 611 U32 maxdepth; // recursion depth limit
612 AV *shareable;
613 AV *stringref;
614 SV *decode_tagged;
444} dec_t; 615} dec_t;
445 616
446#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
447 618
448#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")
451#define DEC_DEC_DEPTH --dec->depth 622#define DEC_DEC_DEPTH --dec->depth
452 623
453static UV 624static UV
454decode_uint (dec_t *dec) 625decode_uint (dec_t *dec)
455{ 626{
456 switch (*dec->cur & 31) 627 U8 m = *dec->cur & MINOR_MASK;
628 ++dec->cur;
629
630 if (ecb_expect_true (m < LENGTH_EXT1))
631 return m;
632
633 switch (m)
457 { 634 {
458 case 0: case 1: case 2: case 3: case 4: case 5: case 6: case 7: 635 case LENGTH_EXT1:
459 case 8: case 9: case 10: case 11: case 12: case 13: case 14: case 15: 636 WANT (1);
460 case 16: case 17: case 18: case 19: case 20: case 21: case 22: case 23: 637 dec->cur += 1;
461 return *dec->cur++ & 31; 638 return dec->cur[-1];
462 639
463 case 24: 640 case LENGTH_EXT2:
464 WANT (2); 641 WANT (2);
465 dec->cur += 2; 642 dec->cur += 2;
466 return dec->cur[-1];
467
468 case 25:
469 WANT (3);
470 dec->cur += 3;
471 return (((UV)dec->cur[-2]) << 8) 643 return (((UV)dec->cur[-2]) << 8)
472 | ((UV)dec->cur[-1]); 644 | ((UV)dec->cur[-1]);
473 645
474 case 26: 646 case LENGTH_EXT4:
475 WANT (5); 647 WANT (4);
476 dec->cur += 5; 648 dec->cur += 4;
477 return (((UV)dec->cur[-4]) << 24) 649 return (((UV)dec->cur[-4]) << 24)
478 | (((UV)dec->cur[-3]) << 16) 650 | (((UV)dec->cur[-3]) << 16)
479 | (((UV)dec->cur[-2]) << 8) 651 | (((UV)dec->cur[-2]) << 8)
480 | ((UV)dec->cur[-1]); 652 | ((UV)dec->cur[-1]);
481 653
482 case 27: 654 case LENGTH_EXT8:
483 WANT (9); 655 WANT (8);
484 dec->cur += 9; 656 dec->cur += 8;
657
658 return
659#if UVSIZE < 8
660 0
661#else
485 return (((UV)dec->cur[-8]) << 56) 662 (((UV)dec->cur[-8]) << 56)
486 | (((UV)dec->cur[-7]) << 48) 663 | (((UV)dec->cur[-7]) << 48)
487 | (((UV)dec->cur[-6]) << 40) 664 | (((UV)dec->cur[-6]) << 40)
488 | (((UV)dec->cur[-5]) << 32) 665 | (((UV)dec->cur[-5]) << 32)
666#endif
489 | (((UV)dec->cur[-4]) << 24) 667 | (((UV)dec->cur[-4]) << 24)
490 | (((UV)dec->cur[-3]) << 16) 668 | (((UV)dec->cur[-3]) << 16)
491 | (((UV)dec->cur[-2]) << 8) 669 | (((UV)dec->cur[-2]) << 8)
492 | ((UV)dec->cur[-1]); 670 | ((UV)dec->cur[-1]);
493 671
506{ 684{
507 AV *av = newAV (); 685 AV *av = newAV ();
508 686
509 DEC_INC_DEPTH; 687 DEC_INC_DEPTH;
510 688
511 if ((*dec->cur & 31) == 31) 689 if (*dec->cur == (MAJOR_ARRAY | MINOR_INDEF))
512 { 690 {
513 ++dec->cur; 691 ++dec->cur;
514 692
515 for (;;) 693 for (;;)
516 { 694 {
517 WANT (1); 695 WANT (1);
518 696
519 if (*dec->cur == (0xe0 | 31)) 697 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
520 { 698 {
521 ++dec->cur; 699 ++dec->cur;
522 break; 700 break;
523 } 701 }
524 702
542 SvREFCNT_dec (av); 720 SvREFCNT_dec (av);
543 DEC_DEC_DEPTH; 721 DEC_DEC_DEPTH;
544 return &PL_sv_undef; 722 return &PL_sv_undef;
545} 723}
546 724
725static void
726decode_he (dec_t *dec, HV *hv)
727{
728 // for speed reasons, we specialcase single-string
729 // byte or utf-8 strings as keys, but only when !stringref
730
731 if (ecb_expect_true (!dec->stringref))
732 if ((*dec->cur - MAJOR_BYTES) <= 27)
733 {
734 I32 len = decode_uint (dec);
735 char *key = (char *)dec->cur;
736
737 dec->cur += len;
738
739 if (ecb_expect_false (dec->stringref))
740 av_push (dec->stringref, newSVpvn (key, len));
741
742 hv_store (hv, key, len, decode_sv (dec), 0);
743
744 return;
745 }
746 else if ((*dec->cur - MAJOR_TEXT) <= 27)
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->stringref))
754 av_push (dec->stringref, newSVpvn_utf8 (key, len, 1));
755
756 hv_store (hv, key, -len, decode_sv (dec), 0);
757
758 return;
759 }
760
761 SV *k = decode_sv (dec);
762 SV *v = decode_sv (dec);
763
764 hv_store_ent (hv, k, v, 0);
765 SvREFCNT_dec (k);
766}
767
547static SV * 768static SV *
548decode_hv (dec_t *dec) 769decode_hv (dec_t *dec)
549{ 770{
550 HV *hv = newHV (); 771 HV *hv = newHV ();
551 772
552 DEC_INC_DEPTH; 773 DEC_INC_DEPTH;
553 774
554 if ((*dec->cur & 31) == 31) 775 if (*dec->cur == (MAJOR_MAP | MINOR_INDEF))
555 { 776 {
556 ++dec->cur; 777 ++dec->cur;
557 778
558 for (;;) 779 for (;;)
559 { 780 {
560 WANT (1); 781 WANT (1);
561 782
562 if (*dec->cur == (0xe0 | 31)) 783 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
563 { 784 {
564 ++dec->cur; 785 ++dec->cur;
565 break; 786 break;
566 } 787 }
567 788
568 SV *k = decode_sv (dec); 789 decode_he (dec, hv);
569 SV *v = decode_sv (dec);
570
571 hv_store_ent (hv, k, v, 0);
572 } 790 }
573 } 791 }
574 else 792 else
575 { 793 {
576 int len = decode_uint (dec); 794 int pairs = decode_uint (dec);
577 795
578 while (len--) 796 while (pairs--)
579 { 797 decode_he (dec, hv);
580 SV *k = decode_sv (dec);
581 SV *v = decode_sv (dec);
582
583 hv_store_ent (hv, k, v, 0);
584 }
585 } 798 }
586 799
587 DEC_DEC_DEPTH; 800 DEC_DEC_DEPTH;
588 return newRV_noinc ((SV *)hv); 801 return newRV_noinc ((SV *)hv);
589 802
596static SV * 809static SV *
597decode_str (dec_t *dec, int utf8) 810decode_str (dec_t *dec, int utf8)
598{ 811{
599 SV *sv = 0; 812 SV *sv = 0;
600 813
601 if ((*dec->cur & 31) == 31) 814 if ((*dec->cur & MINOR_MASK) == MINOR_INDEF)
602 { 815 {
816 // indefinite length strings
603 ++dec->cur; 817 ++dec->cur;
604 818
819 U8 major = *dec->cur & MAJOR_MISC;
820
605 sv = newSVpvn ("", 0); 821 sv = newSVpvn ("", 0);
606 822
607 // not very fast, and certainly not robust against illegal input
608 for (;;) 823 for (;;)
609 { 824 {
610 WANT (1); 825 WANT (1);
611 826
612 if (*dec->cur == (0xe0 | 31)) 827 if ((*dec->cur - major) > LENGTH_EXT8)
828 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
613 { 829 {
614 ++dec->cur; 830 ++dec->cur;
615 break; 831 break;
616 } 832 }
833 else
834 ERR ("corrupted CBOR data (invalid chunks in indefinite length string)");
617 835
618 sv_catsv (sv, decode_sv (dec)); 836 STRLEN len = decode_uint (dec);
837
838 WANT (len);
839 sv_catpvn (sv, dec->cur, len);
840 dec->cur += len;
619 } 841 }
620 } 842 }
621 else 843 else
622 { 844 {
623 STRLEN len = decode_uint (dec); 845 STRLEN len = decode_uint (dec);
624 846
625 WANT (len); 847 WANT (len);
626 sv = newSVpvn (dec->cur, len); 848 sv = newSVpvn (dec->cur, len);
627 dec->cur += len; 849 dec->cur += len;
850
851 if (ecb_expect_false (dec->stringref)
852 && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1))
853 av_push (dec->stringref, SvREFCNT_inc_NN (sv));
628 } 854 }
629 855
630 if (utf8) 856 if (utf8)
631 SvUTF8_on (sv); 857 SvUTF8_on (sv);
632 858
638} 864}
639 865
640static SV * 866static SV *
641decode_tagged (dec_t *dec) 867decode_tagged (dec_t *dec)
642{ 868{
869 SV *sv = 0;
643 UV tag = decode_uint (dec); 870 UV tag = decode_uint (dec);
871
872 WANT (1);
873
874 switch (tag)
875 {
876 case CBOR_TAG_MAGIC:
877 sv = decode_sv (dec);
878 break;
879
880 case CBOR_TAG_INDIRECTION:
881 sv = newRV_noinc (decode_sv (dec));
882 break;
883
884 case CBOR_TAG_STRINGREF_NAMESPACE:
885 {
886 ENTER; SAVETMPS;
887
888 SAVESPTR (dec->stringref);
889 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ());
890
891 sv = decode_sv (dec);
892
893 FREETMPS; LEAVE;
894 }
895 break;
896
897 case CBOR_TAG_STRINGREF:
898 {
899 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
900 ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
901
902 UV idx = decode_uint (dec);
903
904 if (!dec->stringref || (int)idx > AvFILLp (dec->stringref))
905 ERR ("corrupted CBOR data (stringref index out of bounds or outside namespace)");
906
907 sv = newSVsv (AvARRAY (dec->stringref)[idx]);
908 }
909 break;
910
911 case CBOR_TAG_VALUE_SHAREABLE:
912 {
913 if (ecb_expect_false (!dec->shareable))
914 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ());
915
916 sv = newSV (0);
917 av_push (dec->shareable, SvREFCNT_inc_NN (sv));
918
644 SV *sv = decode_sv (dec); 919 SV *osv = decode_sv (dec);
920 sv_setsv (sv, osv);
921 SvREFCNT_dec_NN (osv);
922 }
923 break;
645 924
646 if (tag == CBOR_TAG_MAGIC) 925 case CBOR_TAG_VALUE_SHAREDREF:
647 return sv; 926 {
927 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
928 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
648 929
649 if (tag == CBOR_TAG_PERL_OBJECT) 930 UV idx = decode_uint (dec);
650 { 931
932 if (!dec->shareable || (int)idx > AvFILLp (dec->shareable))
933 ERR ("corrupted CBOR data (sharedref index out of bounds)");
934
935 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]);
936 }
937 break;
938
939 case CBOR_TAG_PERL_OBJECT:
940 {
941 sv = decode_sv (dec);
942
651 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV) 943 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
652 ERR ("corrupted CBOR data (non-array perl object)"); 944 ERR ("corrupted CBOR data (non-array perl object)");
945
946 AV *av = (AV *)SvRV (sv);
947 int len = av_len (av) + 1;
948 HV *stash = gv_stashsv (*av_fetch (av, 0, 1), 0);
949
950 if (!stash)
951 ERR ("cannot decode perl-object (package does not exist)");
952
953 GV *method = gv_fetchmethod_autoload (stash, "THAW", 0);
954
955 if (!method)
956 ERR ("cannot decode perl-object (package does not have a THAW method)");
957
958 dSP;
959
960 ENTER; SAVETMPS; PUSHMARK (SP);
961 EXTEND (SP, len + 1);
962 // we re-bless the reference to get overload and other niceties right
963 PUSHs (*av_fetch (av, 0, 1));
964 PUSHs (sv_cbor);
965
966 int i;
967
968 for (i = 1; i < len; ++i)
969 PUSHs (*av_fetch (av, i, 1));
970
971 PUTBACK;
972 call_sv ((SV *)GvCV (method), G_SCALAR | G_EVAL);
973 SPAGAIN;
974
975 if (SvTRUE (ERRSV))
976 {
977 FREETMPS; LEAVE;
978 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV))));
979 }
980
981 SvREFCNT_dec (sv);
982 sv = SvREFCNT_inc (POPs);
983
984 PUTBACK;
985
986 FREETMPS; LEAVE;
653 987 }
654 // TODO 988 break;
655 }
656 989
990 default:
991 {
992 sv = decode_sv (dec);
993
994 dSP;
995 ENTER; SAVETMPS; PUSHMARK (SP);
996 EXTEND (SP, 2);
997 PUSHs (newSVuv (tag));
998 PUSHs (sv);
999
1000 PUTBACK;
1001 int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL);
1002 SPAGAIN;
1003
1004 if (SvTRUE (ERRSV))
1005 {
1006 FREETMPS; LEAVE;
1007 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV))));
1008 }
1009
1010 if (count)
1011 {
1012 SvREFCNT_dec (sv);
1013 sv = SvREFCNT_inc (POPs);
1014 }
1015 else
1016 {
657 AV *av = newAV (); 1017 AV *av = newAV ();
658 av_push (av, newSVuv (tag)); 1018 av_push (av, newSVuv (tag));
659 av_push (av, sv); 1019 av_push (av, sv);
660 1020
661 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 1021 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
662 ? cbor_tagged_stash 1022 ? cbor_tagged_stash
663 : gv_stashpv ("CBOR::XS::Tagged" , 1); 1023 : gv_stashpv ("CBOR::XS::Tagged" , 1);
664
665 return sv_bless (newRV_noinc ((SV *)av), tagged_stash); 1024 sv = sv_bless (newRV_noinc ((SV *)av), tagged_stash);
1025 }
1026
1027 PUTBACK;
1028
1029 FREETMPS; LEAVE;
1030 }
1031 break;
1032 }
1033
1034 return sv;
666 1035
667fail: 1036fail:
668 SvREFCNT_dec (sv); 1037 SvREFCNT_dec (sv);
669 return &PL_sv_undef; 1038 return &PL_sv_undef;
670} 1039}
672static SV * 1041static SV *
673decode_sv (dec_t *dec) 1042decode_sv (dec_t *dec)
674{ 1043{
675 WANT (1); 1044 WANT (1);
676 1045
677 switch (*dec->cur >> 5) 1046 switch (*dec->cur >> MAJOR_SHIFT)
678 { 1047 {
679 case 0: // unsigned int 1048 case MAJOR_POS_INT >> MAJOR_SHIFT: return newSVuv (decode_uint (dec));
680 return newSVuv (decode_uint (dec)); 1049 case MAJOR_NEG_INT >> MAJOR_SHIFT: return newSViv (-1 - (IV)decode_uint (dec));
681 case 1: // negative int 1050 case MAJOR_BYTES >> MAJOR_SHIFT: return decode_str (dec, 0);
682 return newSViv (-1 - (IV)decode_uint (dec)); 1051 case MAJOR_TEXT >> MAJOR_SHIFT: return decode_str (dec, 1);
683 case 2: // octet string 1052 case MAJOR_ARRAY >> MAJOR_SHIFT: return decode_av (dec);
684 return decode_str (dec, 0); 1053 case MAJOR_MAP >> MAJOR_SHIFT: return decode_hv (dec);
685 case 3: // utf-8 string 1054 case MAJOR_TAG >> MAJOR_SHIFT: return decode_tagged (dec);
686 return decode_str (dec, 1); 1055
687 case 4: // array 1056 case MAJOR_MISC >> MAJOR_SHIFT:
688 return decode_av (dec);
689 case 5: // map
690 return decode_hv (dec);
691 case 6: // tag
692 return decode_tagged (dec);
693 case 7: // misc
694 switch (*dec->cur++ & 31) 1057 switch (*dec->cur++ & MINOR_MASK)
695 { 1058 {
696 case 20: 1059 case SIMPLE_FALSE:
697#if CBOR_SLOW 1060#if CBOR_SLOW
698 cbor_false = get_bool ("CBOR::XS::false"); 1061 types_false = get_bool ("Types::Serialiser::false");
699#endif 1062#endif
700 return newSVsv (cbor_false); 1063 return newSVsv (types_false);
701 case 21: 1064 case SIMPLE_TRUE:
702#if CBOR_SLOW 1065#if CBOR_SLOW
703 cbor_true = get_bool ("CBOR::XS::true"); 1066 types_true = get_bool ("Types::Serialiser::true");
704#endif 1067#endif
705 return newSVsv (cbor_true); 1068 return newSVsv (types_true);
706 case 22: 1069 case SIMPLE_NULL:
707 return newSVsv (&PL_sv_undef); 1070 return newSVsv (&PL_sv_undef);
1071 case SIMPLE_UNDEF:
1072#if CBOR_SLOW
1073 types_error = get_bool ("Types::Serialiser::error");
1074#endif
1075 return newSVsv (types_error);
708 1076
709 case 25: 1077 case MISC_FLOAT16:
710 { 1078 {
711 WANT (2); 1079 WANT (2);
712 1080
713 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1]; 1081 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1];
714 dec->cur += 2; 1082 dec->cur += 2;
715 1083
716 return newSVnv (ecb_binary16_to_float (fp)); 1084 return newSVnv (ecb_binary16_to_float (fp));
717 } 1085 }
718 1086
719 case 26: 1087 case MISC_FLOAT32:
720 { 1088 {
721 uint32_t fp; 1089 uint32_t fp;
722 WANT (4); 1090 WANT (4);
723 memcpy (&fp, dec->cur, 4); 1091 memcpy (&fp, dec->cur, 4);
724 dec->cur += 4; 1092 dec->cur += 4;
727 fp = ecb_bswap32 (fp); 1095 fp = ecb_bswap32 (fp);
728 1096
729 return newSVnv (ecb_binary32_to_float (fp)); 1097 return newSVnv (ecb_binary32_to_float (fp));
730 } 1098 }
731 1099
732 case 27: 1100 case MISC_FLOAT64:
733 { 1101 {
734 uint64_t fp; 1102 uint64_t fp;
735 WANT (8); 1103 WANT (8);
736 memcpy (&fp, dec->cur, 8); 1104 memcpy (&fp, dec->cur, 8);
737 dec->cur += 8; 1105 dec->cur += 8;
740 fp = ecb_bswap64 (fp); 1108 fp = ecb_bswap64 (fp);
741 1109
742 return newSVnv (ecb_binary64_to_double (fp)); 1110 return newSVnv (ecb_binary64_to_double (fp));
743 } 1111 }
744 1112
745 // 0..19 unassigned 1113 // 0..19 unassigned simple
746 // 24 reserved + unassigned (reserved values are not encodable) 1114 // 24 reserved + unassigned (reserved values are not encodable)
747 default: 1115 default:
748 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)"); 1116 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)");
749 } 1117 }
750 1118
756} 1124}
757 1125
758static SV * 1126static SV *
759decode_cbor (SV *string, CBOR *cbor, char **offset_return) 1127decode_cbor (SV *string, CBOR *cbor, char **offset_return)
760{ 1128{
761 dec_t dec; 1129 dec_t dec = { };
762 SV *sv; 1130 SV *sv;
1131 STRLEN len;
1132 char *data = SvPVbyte (string, len);
763 1133
764 /* work around bugs in 5.10 where manipulating magic values
765 * makes perl ignore the magic in subsequent accesses.
766 * also make a copy of non-PV values, to get them into a clean
767 * state (SvPV should do that, but it's buggy, see below).
768 */
769 /*SvGETMAGIC (string);*/
770 if (SvMAGICAL (string) || !SvPOK (string))
771 string = sv_2mortal (newSVsv (string));
772
773 SvUPGRADE (string, SVt_PV);
774
775 /* work around a bug in perl 5.10, which causes SvCUR to fail an
776 * assertion with -DDEBUGGING, although SvCUR is documented to
777 * return the xpv_cur field which certainly exists after upgrading.
778 * according to nicholas clark, calling SvPOK fixes this.
779 * But it doesn't fix it, so try another workaround, call SvPV_nolen
780 * and hope for the best.
781 * Damnit, SvPV_nolen still trips over yet another assertion. This
782 * assertion business is seriously broken, try yet another workaround
783 * for the broken -DDEBUGGING.
784 */
785 {
786#ifdef DEBUGGING
787 STRLEN offset = SvOK (string) ? sv_len (string) : 0;
788#else
789 STRLEN offset = SvCUR (string);
790#endif
791
792 if (offset > cbor->max_size && cbor->max_size) 1134 if (len > cbor->max_size && cbor->max_size)
793 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu", 1135 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
794 (unsigned long)SvCUR (string), (unsigned long)cbor->max_size); 1136 (unsigned long)len, (unsigned long)cbor->max_size);
795 }
796
797 sv_utf8_downgrade (string, 0);
798 1137
799 dec.cbor = *cbor; 1138 dec.cbor = *cbor;
800 dec.cur = (U8 *)SvPVX (string); 1139 dec.cur = (U8 *)data;
801 dec.end = (U8 *)SvEND (string); 1140 dec.end = (U8 *)data + len;
802 dec.err = 0;
803 dec.depth = 0;
804 1141
805 sv = decode_sv (&dec); 1142 sv = decode_sv (&dec);
806 1143
807 if (offset_return) 1144 if (offset_return)
808 *offset_return = dec.cur; 1145 *offset_return = dec.cur;
812 dec.err = "garbage after CBOR object"; 1149 dec.err = "garbage after CBOR object";
813 1150
814 if (dec.err) 1151 if (dec.err)
815 { 1152 {
816 SvREFCNT_dec (sv); 1153 SvREFCNT_dec (sv);
817 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)SvPVX (string), (int)(uint8_t)*dec.cur); 1154 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur);
818 } 1155 }
819 1156
820 sv = sv_2mortal (sv); 1157 sv = sv_2mortal (sv);
821 1158
822 return sv; 1159 return sv;
828MODULE = CBOR::XS PACKAGE = CBOR::XS 1165MODULE = CBOR::XS PACKAGE = CBOR::XS
829 1166
830BOOT: 1167BOOT:
831{ 1168{
832 cbor_stash = gv_stashpv ("CBOR::XS" , 1); 1169 cbor_stash = gv_stashpv ("CBOR::XS" , 1);
833 cbor_boolean_stash = gv_stashpv ("CBOR::XS::Boolean", 1);
834 cbor_tagged_stash = gv_stashpv ("CBOR::XS::Tagged" , 1); 1170 cbor_tagged_stash = gv_stashpv ("CBOR::XS::Tagged" , 1);
835 1171
836 cbor_true = get_bool ("CBOR::XS::true"); 1172 types_boolean_stash = gv_stashpv ("Types::Serialiser::Boolean", 1);
837 cbor_false = get_bool ("CBOR::XS::false"); 1173 types_error_stash = gv_stashpv ("Types::Serialiser::Error" , 1);
1174
1175 types_true = get_bool ("Types::Serialiser::true" );
1176 types_false = get_bool ("Types::Serialiser::false");
1177 types_error = get_bool ("Types::Serialiser::error");
1178
1179 default_filter = newSVpv ("CBOR::XS::default_filter", 0);
1180
1181 sv_cbor = newSVpv ("CBOR", 0);
1182 SvREADONLY_on (sv_cbor);
838} 1183}
839 1184
840PROTOTYPES: DISABLE 1185PROTOTYPES: DISABLE
841 1186
842void CLONE (...) 1187void CLONE (...)
843 CODE: 1188 CODE:
844 cbor_stash = 0; 1189 cbor_stash = 0;
845 cbor_boolean_stash = 0;
846 cbor_tagged_stash = 0; 1190 cbor_tagged_stash = 0;
1191 types_error_stash = 0;
1192 types_boolean_stash = 0;
847 1193
848void new (char *klass) 1194void new (char *klass)
849 PPCODE: 1195 PPCODE:
850{ 1196{
851 SV *pv = NEWSV (0, sizeof (CBOR)); 1197 SV *pv = NEWSV (0, sizeof (CBOR));
859 1205
860void shrink (CBOR *self, int enable = 1) 1206void shrink (CBOR *self, int enable = 1)
861 ALIAS: 1207 ALIAS:
862 shrink = F_SHRINK 1208 shrink = F_SHRINK
863 allow_unknown = F_ALLOW_UNKNOWN 1209 allow_unknown = F_ALLOW_UNKNOWN
1210 allow_sharing = F_ALLOW_SHARING
1211 pack_strings = F_PACK_STRINGS
864 PPCODE: 1212 PPCODE:
865{ 1213{
866 if (enable) 1214 if (enable)
867 self->flags |= ix; 1215 self->flags |= ix;
868 else 1216 else
873 1221
874void get_shrink (CBOR *self) 1222void get_shrink (CBOR *self)
875 ALIAS: 1223 ALIAS:
876 get_shrink = F_SHRINK 1224 get_shrink = F_SHRINK
877 get_allow_unknown = F_ALLOW_UNKNOWN 1225 get_allow_unknown = F_ALLOW_UNKNOWN
1226 get_allow_sharing = F_ALLOW_SHARING
1227 get_pack_strings = F_PACK_STRINGS
878 PPCODE: 1228 PPCODE:
879 XPUSHs (boolSV (self->flags & ix)); 1229 XPUSHs (boolSV (self->flags & ix));
880 1230
881void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) 1231void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
882 PPCODE: 1232 PPCODE:
898 CODE: 1248 CODE:
899 RETVAL = self->max_size; 1249 RETVAL = self->max_size;
900 OUTPUT: 1250 OUTPUT:
901 RETVAL 1251 RETVAL
902 1252
903#if 0 //TODO 1253void filter (CBOR *self, SV *filter = 0)
904
905void filter_cbor_object (CBOR *self, SV *cb = &PL_sv_undef)
906 PPCODE: 1254 PPCODE:
907{
908 SvREFCNT_dec (self->cb_object); 1255 SvREFCNT_dec (self->filter);
909 self->cb_object = SvOK (cb) ? newSVsv (cb) : 0; 1256 self->filter = filter ? newSVsv (filter) : filter;
910
911 XPUSHs (ST (0)); 1257 XPUSHs (ST (0));
912}
913 1258
914void filter_cbor_single_key_object (CBOR *self, SV *key, SV *cb = &PL_sv_undef) 1259SV *get_filter (CBOR *self)
915 PPCODE: 1260 CODE:
916{ 1261 RETVAL = self->filter ? self->filter : NEWSV (0, 0);
917 if (!self->cb_sk_object) 1262 OUTPUT:
918 self->cb_sk_object = newHV (); 1263 RETVAL
919
920 if (SvOK (cb))
921 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);
922 else
923 {
924 hv_delete_ent (self->cb_sk_object, key, G_DISCARD, 0);
925
926 if (!HvKEYS (self->cb_sk_object))
927 {
928 SvREFCNT_dec (self->cb_sk_object);
929 self->cb_sk_object = 0;
930 }
931 }
932
933 XPUSHs (ST (0));
934}
935
936#endif
937 1264
938void encode (CBOR *self, SV *scalar) 1265void encode (CBOR *self, SV *scalar)
939 PPCODE: 1266 PPCODE:
940 PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN; 1267 PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN;
941 XPUSHs (scalar); 1268 XPUSHs (scalar);
954 EXTEND (SP, 2); 1281 EXTEND (SP, 2);
955 PUSHs (sv); 1282 PUSHs (sv);
956 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); 1283 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
957} 1284}
958 1285
959#if 0
960
961void DESTROY (CBOR *self) 1286void DESTROY (CBOR *self)
962 CODE: 1287 PPCODE:
963 SvREFCNT_dec (self->cb_sk_object); 1288 cbor_free (self);
964 SvREFCNT_dec (self->cb_object);
965
966#endif
967 1289
968PROTOTYPES: ENABLE 1290PROTOTYPES: ENABLE
969 1291
970void encode_cbor (SV *scalar) 1292void encode_cbor (SV *scalar)
971 PPCODE: 1293 PPCODE:

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines