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 1.10 by root, Sun Oct 27 20:40:25 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
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)
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, *types_boolean_stash, *types_error_stash, *cbor_tagged_stash; // CBOR::XS:: 123static HV *cbor_stash, *types_boolean_stash, *types_error_stash, *cbor_tagged_stash; // CBOR::XS::
66static SV *types_true, *types_false, *types_error; 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 <= 0xffU))
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 <= 0xffffU)
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 <= 0xffffffffU)
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 || types_boolean_stash 373 HV *boolean_stash = !CBOR_SLOW || types_boolean_stash
265 ? types_boolean_stash 374 ? types_boolean_stash
269 : gv_stashpv ("Types::Serialiser::Error", 1); 378 : gv_stashpv ("Types::Serialiser::Error", 1);
270 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 379 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
271 ? cbor_tagged_stash 380 ? cbor_tagged_stash
272 : gv_stashpv ("CBOR::XS::Tagged" , 1); 381 : gv_stashpv ("CBOR::XS::Tagged" , 1);
273 382
383 HV *stash = SvSTASH (sv);
384
274 if (SvSTASH (sv) == boolean_stash) 385 if (stash == boolean_stash)
275 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 }
276 else if (SvSTASH (sv) == error_stash) 390 else if (stash == error_stash)
277 encode_ch (enc, 0xe0 | 23); 391 {
392 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
393 return;
394 }
278 else if (SvSTASH (sv) == tagged_stash) 395 else if (stash == tagged_stash)
279 { 396 {
280 if (svt != SVt_PVAV) 397 if (svt != SVt_PVAV)
281 croak ("encountered CBOR::XS::Tagged object that isn't an array"); 398 croak ("encountered CBOR::XS::Tagged object that isn't an array");
282 399
283 encode_uint (enc, 0xc0, SvUV (*av_fetch ((AV *)sv, 0, 1))); 400 encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1)));
284 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;
285 } 420 }
286 else 421 else
287 { 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);
288 // 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
289 GV *to_cbor = gv_fetchmethod_autoload (SvSTASH (sv), "TO_CBOR", 0);
290
291 if (to_cbor)
292 {
293 dSP;
294
295 ENTER; SAVETMPS; PUSHMARK (SP);
296 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv))); 440 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
297 441
298 // calling with G_SCALAR ensures that we always get a 1 return value
299 PUTBACK; 442 PUTBACK;
443 // G_SCALAR ensures that return value is 1
300 call_sv ((SV *)GvCV (to_cbor), G_SCALAR); 444 call_sv ((SV *)GvCV (method), G_SCALAR);
301 SPAGAIN; 445 SPAGAIN;
302 446
303 // catch this surprisingly common error 447 // catch this surprisingly common error
304 if (SvROK (TOPs) && SvRV (TOPs) == sv) 448 if (SvROK (TOPs) && SvRV (TOPs) == sv)
305 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));
306 450
307 sv = POPs; 451 encode_sv (enc, POPs);
452
308 PUTBACK; 453 PUTBACK;
309 454
310 encode_sv (enc, sv);
311
312 FREETMPS; LEAVE; 455 FREETMPS; LEAVE;
313 } 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 }
314 else 486 else
315 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",
316 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 488 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
317 }
318 } 489 }
319 else if (svt == SVt_PVHV) 490 else if (svt == SVt_PVHV)
320 encode_hv (enc, (HV *)sv); 491 encode_hv (enc, (HV *)sv);
321 else if (svt == SVt_PVAV) 492 else if (svt == SVt_PVAV)
322 encode_av (enc, (AV *)sv); 493 encode_av (enc, (AV *)sv);
323 else if (svt < SVt_PVAV)
324 {
325 STRLEN len = 0;
326 char *pv = svt ? SvPV (sv, len) : 0;
327
328 if (len == 1 && *pv == '1')
329 encode_ch (enc, 0xe0 | 21);
330 else if (len == 1 && *pv == '0')
331 encode_ch (enc, 0xe0 | 20);
332 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
333 encode_ch (enc, 0xe0 | 23);
334 else
335 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
336 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
337 }
338 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
339 encode_ch (enc, 0xe0 | 23);
340 else 494 else
341 croak ("encountered %s, but CBOR can only represent references to arrays or hashes", 495 {
342 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 496 encode_tag (enc, CBOR_TAG_INDIRECTION);
497 encode_sv (enc, sv);
498 }
343} 499}
344 500
345static void 501static void
346encode_nv (enc_t *enc, SV *sv) 502encode_nv (enc_t *enc, SV *sv)
347{ 503{
348 double nv = SvNVX (sv); 504 double nv = SvNVX (sv);
349 505
350 need (enc, 9); 506 need (enc, 9);
351 507
352 if (ecb_expect_false (nv == (U32)nv)) 508 if (ecb_expect_false (nv == (NV)(U32)nv))
353 encode_uint (enc, 0x00, (U32)nv); 509 encode_uint (enc, MAJOR_POS_INT, (U32)nv);
354 //TODO: maybe I32? 510 //TODO: maybe I32?
355 else if (ecb_expect_false (nv == (float)nv)) 511 else if (ecb_expect_false (nv == (float)nv))
356 { 512 {
357 uint32_t fp = ecb_float_to_binary32 (nv); 513 uint32_t fp = ecb_float_to_binary32 (nv);
358 514
359 *enc->cur++ = 0xe0 | 26; 515 *enc->cur++ = MAJOR_MISC | MISC_FLOAT32;
360 516
361 if (!ecb_big_endian ()) 517 if (!ecb_big_endian ())
362 fp = ecb_bswap32 (fp); 518 fp = ecb_bswap32 (fp);
363 519
364 memcpy (enc->cur, &fp, 4); 520 memcpy (enc->cur, &fp, 4);
366 } 522 }
367 else 523 else
368 { 524 {
369 uint64_t fp = ecb_double_to_binary64 (nv); 525 uint64_t fp = ecb_double_to_binary64 (nv);
370 526
371 *enc->cur++ = 0xe0 | 27; 527 *enc->cur++ = MAJOR_MISC | MISC_FLOAT64;
372 528
373 if (!ecb_big_endian ()) 529 if (!ecb_big_endian ())
374 fp = ecb_bswap64 (fp); 530 fp = ecb_bswap64 (fp);
375 531
376 memcpy (enc->cur, &fp, 8); 532 memcpy (enc->cur, &fp, 8);
385 541
386 if (SvPOKp (sv)) 542 if (SvPOKp (sv))
387 { 543 {
388 STRLEN len; 544 STRLEN len;
389 char *str = SvPV (sv, len); 545 char *str = SvPV (sv, len);
390 encode_str (enc, SvUTF8 (sv), str, len); 546 encode_strref (enc, SvUTF8 (sv), str, len);
391 } 547 }
392 else if (SvNOKp (sv)) 548 else if (SvNOKp (sv))
393 encode_nv (enc, sv); 549 encode_nv (enc, sv);
394 else if (SvIOKp (sv)) 550 else if (SvIOKp (sv))
395 { 551 {
396 if (SvIsUV (sv)) 552 if (SvIsUV (sv))
397 encode_uint (enc, 0x00, SvUVX (sv)); 553 encode_uint (enc, MAJOR_POS_INT, SvUVX (sv));
398 else if (SvIVX (sv) >= 0) 554 else if (SvIVX (sv) >= 0)
399 encode_uint (enc, 0x00, SvIVX (sv)); 555 encode_uint (enc, MAJOR_POS_INT, SvIVX (sv));
400 else 556 else
401 encode_uint (enc, 0x20, -(SvIVX (sv) + 1)); 557 encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1));
402 } 558 }
403 else if (SvROK (sv)) 559 else if (SvROK (sv))
404 encode_rv (enc, SvRV (sv)); 560 encode_rv (enc, SvRV (sv));
405 else if (!SvOK (sv)) 561 else if (!SvOK (sv))
406 encode_ch (enc, 0xe0 | 22); 562 encode_ch (enc, MAJOR_MISC | SIMPLE_NULL);
407 else if (enc->cbor.flags & F_ALLOW_UNKNOWN) 563 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
408 encode_ch (enc, 0xe0 | 23); 564 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
409 else 565 else
410 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",
411 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv)); 567 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
412} 568}
413 569
414static SV * 570static SV *
415encode_cbor (SV *scalar, CBOR *cbor) 571encode_cbor (SV *scalar, CBOR *cbor)
416{ 572{
417 enc_t enc; 573 enc_t enc = { };
418 574
419 enc.cbor = *cbor; 575 enc.cbor = *cbor;
420 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 576 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
421 enc.cur = SvPVX (enc.sv); 577 enc.cur = SvPVX (enc.sv);
422 enc.end = SvEND (enc.sv); 578 enc.end = SvEND (enc.sv);
423 enc.depth = 0;
424 579
425 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
426 encode_sv (&enc, scalar); 589 encode_sv (&enc, scalar);
427 590
428 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 591 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
429 *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
430 593
444 U8 *end; // end of input string 607 U8 *end; // end of input string
445 const char *err; // parse error, if != 0 608 const char *err; // parse error, if != 0
446 CBOR cbor; 609 CBOR cbor;
447 U32 depth; // recursion depth 610 U32 depth; // recursion depth
448 U32 maxdepth; // recursion depth limit 611 U32 maxdepth; // recursion depth limit
612 AV *shareable;
613 AV *stringref;
614 SV *decode_tagged;
449} dec_t; 615} dec_t;
450 616
451#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
452 618
453#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")
456#define DEC_DEC_DEPTH --dec->depth 622#define DEC_DEC_DEPTH --dec->depth
457 623
458static UV 624static UV
459decode_uint (dec_t *dec) 625decode_uint (dec_t *dec)
460{ 626{
461 switch (*dec->cur & 31) 627 U8 m = *dec->cur & MINOR_MASK;
462 { 628 ++dec->cur;
463 case 0: case 1: case 2: case 3: case 4: case 5: case 6: case 7:
464 case 8: case 9: case 10: case 11: case 12: case 13: case 14: case 15:
465 case 16: case 17: case 18: case 19: case 20: case 21: case 22: case 23:
466 return *dec->cur++ & 31;
467 629
468 case 24: 630 if (ecb_expect_true (m < LENGTH_EXT1))
631 return m;
632 else if (ecb_expect_true (m == LENGTH_EXT1))
633 {
469 WANT (2); 634 WANT (1);
470 dec->cur += 2; 635 dec->cur += 1;
471 return dec->cur[-1]; 636 return dec->cur[-1];
472 637 }
473 case 25: 638 else if (ecb_expect_true (m == LENGTH_EXT2))
639 {
474 WANT (3); 640 WANT (2);
475 dec->cur += 3; 641 dec->cur += 2;
476 return (((UV)dec->cur[-2]) << 8) 642 return (((UV)dec->cur[-2]) << 8)
477 | ((UV)dec->cur[-1]); 643 | ((UV)dec->cur[-1]);
478 644 }
479 case 26: 645 else if (ecb_expect_true (m == LENGTH_EXT4))
646 {
480 WANT (5); 647 WANT (4);
481 dec->cur += 5; 648 dec->cur += 4;
482 return (((UV)dec->cur[-4]) << 24) 649 return (((UV)dec->cur[-4]) << 24)
483 | (((UV)dec->cur[-3]) << 16) 650 | (((UV)dec->cur[-3]) << 16)
484 | (((UV)dec->cur[-2]) << 8) 651 | (((UV)dec->cur[-2]) << 8)
485 | ((UV)dec->cur[-1]); 652 | ((UV)dec->cur[-1]);
486 653 }
487 case 27: 654 else if (ecb_expect_true (m == LENGTH_EXT8))
655 {
488 WANT (9); 656 WANT (8);
489 dec->cur += 9; 657 dec->cur += 8;
658
659 return
660#if UVSIZE < 8
661 0
662#else
490 return (((UV)dec->cur[-8]) << 56) 663 (((UV)dec->cur[-8]) << 56)
491 | (((UV)dec->cur[-7]) << 48) 664 | (((UV)dec->cur[-7]) << 48)
492 | (((UV)dec->cur[-6]) << 40) 665 | (((UV)dec->cur[-6]) << 40)
493 | (((UV)dec->cur[-5]) << 32) 666 | (((UV)dec->cur[-5]) << 32)
667#endif
494 | (((UV)dec->cur[-4]) << 24) 668 | (((UV)dec->cur[-4]) << 24)
495 | (((UV)dec->cur[-3]) << 16) 669 | (((UV)dec->cur[-3]) << 16)
496 | (((UV)dec->cur[-2]) << 8) 670 | (((UV)dec->cur[-2]) << 8)
497 | ((UV)dec->cur[-1]); 671 | ((UV)dec->cur[-1]);
498 672 }
499 default: 673 else
500 ERR ("corrupted CBOR data (unsupported integer minor encoding)"); 674 ERR ("corrupted CBOR data (unsupported integer minor encoding)");
501 }
502 675
503fail: 676fail:
504 return 0; 677 return 0;
505} 678}
506 679
511{ 684{
512 AV *av = newAV (); 685 AV *av = newAV ();
513 686
514 DEC_INC_DEPTH; 687 DEC_INC_DEPTH;
515 688
516 if ((*dec->cur & 31) == 31) 689 if (*dec->cur == (MAJOR_ARRAY | MINOR_INDEF))
517 { 690 {
518 ++dec->cur; 691 ++dec->cur;
519 692
520 for (;;) 693 for (;;)
521 { 694 {
522 WANT (1); 695 WANT (1);
523 696
524 if (*dec->cur == (0xe0 | 31)) 697 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
525 { 698 {
526 ++dec->cur; 699 ++dec->cur;
527 break; 700 break;
528 } 701 }
529 702
532 } 705 }
533 else 706 else
534 { 707 {
535 int i, len = decode_uint (dec); 708 int i, len = decode_uint (dec);
536 709
710 WANT (len); // complexity check for av_fill - need at least one byte per value, do not allow supersize arrays
537 av_fill (av, len - 1); 711 av_fill (av, len - 1);
538 712
539 for (i = 0; i < len; ++i) 713 for (i = 0; i < len; ++i)
540 AvARRAY (av)[i] = decode_sv (dec); 714 AvARRAY (av)[i] = decode_sv (dec);
541 } 715 }
547 SvREFCNT_dec (av); 721 SvREFCNT_dec (av);
548 DEC_DEC_DEPTH; 722 DEC_DEC_DEPTH;
549 return &PL_sv_undef; 723 return &PL_sv_undef;
550} 724}
551 725
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
552static SV * 769static SV *
553decode_hv (dec_t *dec) 770decode_hv (dec_t *dec)
554{ 771{
555 HV *hv = newHV (); 772 HV *hv = newHV ();
556 773
557 DEC_INC_DEPTH; 774 DEC_INC_DEPTH;
558 775
559 if ((*dec->cur & 31) == 31) 776 if (*dec->cur == (MAJOR_MAP | MINOR_INDEF))
560 { 777 {
561 ++dec->cur; 778 ++dec->cur;
562 779
563 for (;;) 780 for (;;)
564 { 781 {
565 WANT (1); 782 WANT (1);
566 783
567 if (*dec->cur == (0xe0 | 31)) 784 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
568 { 785 {
569 ++dec->cur; 786 ++dec->cur;
570 break; 787 break;
571 } 788 }
572 789
573 SV *k = decode_sv (dec); 790 decode_he (dec, hv);
574 SV *v = decode_sv (dec);
575
576 hv_store_ent (hv, k, v, 0);
577 } 791 }
578 } 792 }
579 else 793 else
580 { 794 {
581 int len = decode_uint (dec); 795 int pairs = decode_uint (dec);
582 796
583 while (len--) 797 while (pairs--)
584 { 798 decode_he (dec, hv);
585 SV *k = decode_sv (dec);
586 SV *v = decode_sv (dec);
587
588 hv_store_ent (hv, k, v, 0);
589 }
590 } 799 }
591 800
592 DEC_DEC_DEPTH; 801 DEC_DEC_DEPTH;
593 return newRV_noinc ((SV *)hv); 802 return newRV_noinc ((SV *)hv);
594 803
601static SV * 810static SV *
602decode_str (dec_t *dec, int utf8) 811decode_str (dec_t *dec, int utf8)
603{ 812{
604 SV *sv = 0; 813 SV *sv = 0;
605 814
606 if ((*dec->cur & 31) == 31) 815 if ((*dec->cur & MINOR_MASK) == MINOR_INDEF)
607 { 816 {
817 // indefinite length strings
608 ++dec->cur; 818 ++dec->cur;
609 819
820 U8 major = *dec->cur & MAJOR_MISC;
821
610 sv = newSVpvn ("", 0); 822 sv = newSVpvn ("", 0);
611 823
612 // not very fast, and certainly not robust against illegal input
613 for (;;) 824 for (;;)
614 { 825 {
615 WANT (1); 826 WANT (1);
616 827
617 if (*dec->cur == (0xe0 | 31)) 828 if ((*dec->cur - major) > LENGTH_EXT8)
829 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
618 { 830 {
619 ++dec->cur; 831 ++dec->cur;
620 break; 832 break;
621 } 833 }
834 else
835 ERR ("corrupted CBOR data (invalid chunks in indefinite length string)");
622 836
623 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;
624 } 842 }
625 } 843 }
626 else 844 else
627 { 845 {
628 STRLEN len = decode_uint (dec); 846 STRLEN len = decode_uint (dec);
629 847
630 WANT (len); 848 WANT (len);
631 sv = newSVpvn (dec->cur, len); 849 sv = newSVpvn (dec->cur, len);
632 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));
633 } 855 }
634 856
635 if (utf8) 857 if (utf8)
636 SvUTF8_on (sv); 858 SvUTF8_on (sv);
637 859
643} 865}
644 866
645static SV * 867static SV *
646decode_tagged (dec_t *dec) 868decode_tagged (dec_t *dec)
647{ 869{
870 SV *sv = 0;
648 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
649 SV *sv = decode_sv (dec); 920 SV *osv = decode_sv (dec);
921 sv_setsv (sv, osv);
922 SvREFCNT_dec_NN (osv);
923 }
924 break;
650 925
651 if (tag == CBOR_TAG_MAGIC) 926 case CBOR_TAG_VALUE_SHAREDREF:
652 return sv; 927 {
928 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
929 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
653 930
654 if (tag == CBOR_TAG_PERL_OBJECT) 931 UV idx = decode_uint (dec);
655 { 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
656 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV) 944 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
657 ERR ("corrupted CBOR data (non-array perl object)"); 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;
658 988 }
659 // TODO 989 break;
660 }
661 990
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 {
662 AV *av = newAV (); 1018 AV *av = newAV ();
663 av_push (av, newSVuv (tag)); 1019 av_push (av, newSVuv (tag));
664 av_push (av, sv); 1020 av_push (av, sv);
665 1021
666 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 1022 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
667 ? cbor_tagged_stash 1023 ? cbor_tagged_stash
668 : gv_stashpv ("CBOR::XS::Tagged" , 1); 1024 : gv_stashpv ("CBOR::XS::Tagged" , 1);
669
670 return sv_bless (newRV_noinc ((SV *)av), tagged_stash); 1025 sv = sv_bless (newRV_noinc ((SV *)av), tagged_stash);
1026 }
1027
1028 PUTBACK;
1029
1030 FREETMPS; LEAVE;
1031 }
1032 break;
1033 }
1034
1035 return sv;
671 1036
672fail: 1037fail:
673 SvREFCNT_dec (sv); 1038 SvREFCNT_dec (sv);
674 return &PL_sv_undef; 1039 return &PL_sv_undef;
675} 1040}
677static SV * 1042static SV *
678decode_sv (dec_t *dec) 1043decode_sv (dec_t *dec)
679{ 1044{
680 WANT (1); 1045 WANT (1);
681 1046
682 switch (*dec->cur >> 5) 1047 switch (*dec->cur >> MAJOR_SHIFT)
683 { 1048 {
684 case 0: // unsigned int 1049 case MAJOR_POS_INT >> MAJOR_SHIFT: return newSVuv (decode_uint (dec));
685 return newSVuv (decode_uint (dec)); 1050 case MAJOR_NEG_INT >> MAJOR_SHIFT: return newSViv (-1 - (IV)decode_uint (dec));
686 case 1: // negative int 1051 case MAJOR_BYTES >> MAJOR_SHIFT: return decode_str (dec, 0);
687 return newSViv (-1 - (IV)decode_uint (dec)); 1052 case MAJOR_TEXT >> MAJOR_SHIFT: return decode_str (dec, 1);
688 case 2: // octet string 1053 case MAJOR_ARRAY >> MAJOR_SHIFT: return decode_av (dec);
689 return decode_str (dec, 0); 1054 case MAJOR_MAP >> MAJOR_SHIFT: return decode_hv (dec);
690 case 3: // utf-8 string 1055 case MAJOR_TAG >> MAJOR_SHIFT: return decode_tagged (dec);
691 return decode_str (dec, 1); 1056
692 case 4: // array 1057 case MAJOR_MISC >> MAJOR_SHIFT:
693 return decode_av (dec);
694 case 5: // map
695 return decode_hv (dec);
696 case 6: // tag
697 return decode_tagged (dec);
698 case 7: // misc
699 switch (*dec->cur++ & 31) 1058 switch (*dec->cur++ & MINOR_MASK)
700 { 1059 {
701 case 20: 1060 case SIMPLE_FALSE:
702#if CBOR_SLOW 1061#if CBOR_SLOW
703 types_false = get_bool ("Types::Serialiser::false"); 1062 types_false = get_bool ("Types::Serialiser::false");
704#endif 1063#endif
705 return newSVsv (types_false); 1064 return newSVsv (types_false);
706 case 21: 1065 case SIMPLE_TRUE:
707#if CBOR_SLOW 1066#if CBOR_SLOW
708 types_true = get_bool ("Types::Serialiser::true"); 1067 types_true = get_bool ("Types::Serialiser::true");
709#endif 1068#endif
710 return newSVsv (types_true); 1069 return newSVsv (types_true);
711 case 22: 1070 case SIMPLE_NULL:
712 return newSVsv (&PL_sv_undef); 1071 return newSVsv (&PL_sv_undef);
713 case 23: 1072 case SIMPLE_UNDEF:
714#if CBOR_SLOW 1073#if CBOR_SLOW
715 types_error = get_bool ("Types::Serialiser::error"); 1074 types_error = get_bool ("Types::Serialiser::error");
716#endif 1075#endif
717 return newSVsv (types_error); 1076 return newSVsv (types_error);
718 1077
719 case 25: 1078 case MISC_FLOAT16:
720 { 1079 {
721 WANT (2); 1080 WANT (2);
722 1081
723 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1]; 1082 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1];
724 dec->cur += 2; 1083 dec->cur += 2;
725 1084
726 return newSVnv (ecb_binary16_to_float (fp)); 1085 return newSVnv (ecb_binary16_to_float (fp));
727 } 1086 }
728 1087
729 case 26: 1088 case MISC_FLOAT32:
730 { 1089 {
731 uint32_t fp; 1090 uint32_t fp;
732 WANT (4); 1091 WANT (4);
733 memcpy (&fp, dec->cur, 4); 1092 memcpy (&fp, dec->cur, 4);
734 dec->cur += 4; 1093 dec->cur += 4;
737 fp = ecb_bswap32 (fp); 1096 fp = ecb_bswap32 (fp);
738 1097
739 return newSVnv (ecb_binary32_to_float (fp)); 1098 return newSVnv (ecb_binary32_to_float (fp));
740 } 1099 }
741 1100
742 case 27: 1101 case MISC_FLOAT64:
743 { 1102 {
744 uint64_t fp; 1103 uint64_t fp;
745 WANT (8); 1104 WANT (8);
746 memcpy (&fp, dec->cur, 8); 1105 memcpy (&fp, dec->cur, 8);
747 dec->cur += 8; 1106 dec->cur += 8;
750 fp = ecb_bswap64 (fp); 1109 fp = ecb_bswap64 (fp);
751 1110
752 return newSVnv (ecb_binary64_to_double (fp)); 1111 return newSVnv (ecb_binary64_to_double (fp));
753 } 1112 }
754 1113
755 // 0..19 unassigned 1114 // 0..19 unassigned simple
756 // 24 reserved + unassigned (reserved values are not encodable) 1115 // 24 reserved + unassigned (reserved values are not encodable)
757 default: 1116 default:
758 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)"); 1117 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)");
759 } 1118 }
760 1119
766} 1125}
767 1126
768static SV * 1127static SV *
769decode_cbor (SV *string, CBOR *cbor, char **offset_return) 1128decode_cbor (SV *string, CBOR *cbor, char **offset_return)
770{ 1129{
771 dec_t dec; 1130 dec_t dec = { };
772 SV *sv; 1131 SV *sv;
1132 STRLEN len;
1133 char *data = SvPVbyte (string, len);
773 1134
774 /* work around bugs in 5.10 where manipulating magic values
775 * makes perl ignore the magic in subsequent accesses.
776 * also make a copy of non-PV values, to get them into a clean
777 * state (SvPV should do that, but it's buggy, see below).
778 */
779 /*SvGETMAGIC (string);*/
780 if (SvMAGICAL (string) || !SvPOK (string))
781 string = sv_2mortal (newSVsv (string));
782
783 SvUPGRADE (string, SVt_PV);
784
785 /* work around a bug in perl 5.10, which causes SvCUR to fail an
786 * assertion with -DDEBUGGING, although SvCUR is documented to
787 * return the xpv_cur field which certainly exists after upgrading.
788 * according to nicholas clark, calling SvPOK fixes this.
789 * But it doesn't fix it, so try another workaround, call SvPV_nolen
790 * and hope for the best.
791 * Damnit, SvPV_nolen still trips over yet another assertion. This
792 * assertion business is seriously broken, try yet another workaround
793 * for the broken -DDEBUGGING.
794 */
795 {
796#ifdef DEBUGGING
797 STRLEN offset = SvOK (string) ? sv_len (string) : 0;
798#else
799 STRLEN offset = SvCUR (string);
800#endif
801
802 if (offset > cbor->max_size && cbor->max_size) 1135 if (len > cbor->max_size && cbor->max_size)
803 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",
804 (unsigned long)SvCUR (string), (unsigned long)cbor->max_size); 1137 (unsigned long)len, (unsigned long)cbor->max_size);
805 }
806
807 sv_utf8_downgrade (string, 0);
808 1138
809 dec.cbor = *cbor; 1139 dec.cbor = *cbor;
810 dec.cur = (U8 *)SvPVX (string); 1140 dec.cur = (U8 *)data;
811 dec.end = (U8 *)SvEND (string); 1141 dec.end = (U8 *)data + len;
812 dec.err = 0;
813 dec.depth = 0;
814 1142
815 sv = decode_sv (&dec); 1143 sv = decode_sv (&dec);
816 1144
817 if (offset_return) 1145 if (offset_return)
818 *offset_return = dec.cur; 1146 *offset_return = dec.cur;
822 dec.err = "garbage after CBOR object"; 1150 dec.err = "garbage after CBOR object";
823 1151
824 if (dec.err) 1152 if (dec.err)
825 { 1153 {
826 SvREFCNT_dec (sv); 1154 SvREFCNT_dec (sv);
827 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);
828 } 1156 }
829 1157
830 sv = sv_2mortal (sv); 1158 sv = sv_2mortal (sv);
831 1159
832 return sv; 1160 return sv;
846 types_error_stash = gv_stashpv ("Types::Serialiser::Error" , 1); 1174 types_error_stash = gv_stashpv ("Types::Serialiser::Error" , 1);
847 1175
848 types_true = get_bool ("Types::Serialiser::true" ); 1176 types_true = get_bool ("Types::Serialiser::true" );
849 types_false = get_bool ("Types::Serialiser::false"); 1177 types_false = get_bool ("Types::Serialiser::false");
850 types_error = get_bool ("Types::Serialiser::error"); 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);
851} 1184}
852 1185
853PROTOTYPES: DISABLE 1186PROTOTYPES: DISABLE
854 1187
855void CLONE (...) 1188void CLONE (...)
873 1206
874void shrink (CBOR *self, int enable = 1) 1207void shrink (CBOR *self, int enable = 1)
875 ALIAS: 1208 ALIAS:
876 shrink = F_SHRINK 1209 shrink = F_SHRINK
877 allow_unknown = F_ALLOW_UNKNOWN 1210 allow_unknown = F_ALLOW_UNKNOWN
1211 allow_sharing = F_ALLOW_SHARING
1212 pack_strings = F_PACK_STRINGS
878 PPCODE: 1213 PPCODE:
879{ 1214{
880 if (enable) 1215 if (enable)
881 self->flags |= ix; 1216 self->flags |= ix;
882 else 1217 else
887 1222
888void get_shrink (CBOR *self) 1223void get_shrink (CBOR *self)
889 ALIAS: 1224 ALIAS:
890 get_shrink = F_SHRINK 1225 get_shrink = F_SHRINK
891 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
892 PPCODE: 1229 PPCODE:
893 XPUSHs (boolSV (self->flags & ix)); 1230 XPUSHs (boolSV (self->flags & ix));
894 1231
895void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) 1232void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
896 PPCODE: 1233 PPCODE:
912 CODE: 1249 CODE:
913 RETVAL = self->max_size; 1250 RETVAL = self->max_size;
914 OUTPUT: 1251 OUTPUT:
915 RETVAL 1252 RETVAL
916 1253
917#if 0 //TODO 1254void filter (CBOR *self, SV *filter = 0)
918
919void filter_cbor_object (CBOR *self, SV *cb = &PL_sv_undef)
920 PPCODE: 1255 PPCODE:
921{
922 SvREFCNT_dec (self->cb_object); 1256 SvREFCNT_dec (self->filter);
923 self->cb_object = SvOK (cb) ? newSVsv (cb) : 0; 1257 self->filter = filter ? newSVsv (filter) : filter;
924
925 XPUSHs (ST (0)); 1258 XPUSHs (ST (0));
926}
927 1259
928void filter_cbor_single_key_object (CBOR *self, SV *key, SV *cb = &PL_sv_undef) 1260SV *get_filter (CBOR *self)
929 PPCODE: 1261 CODE:
930{ 1262 RETVAL = self->filter ? self->filter : NEWSV (0, 0);
931 if (!self->cb_sk_object) 1263 OUTPUT:
932 self->cb_sk_object = newHV (); 1264 RETVAL
933
934 if (SvOK (cb))
935 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);
936 else
937 {
938 hv_delete_ent (self->cb_sk_object, key, G_DISCARD, 0);
939
940 if (!HvKEYS (self->cb_sk_object))
941 {
942 SvREFCNT_dec (self->cb_sk_object);
943 self->cb_sk_object = 0;
944 }
945 }
946
947 XPUSHs (ST (0));
948}
949
950#endif
951 1265
952void encode (CBOR *self, SV *scalar) 1266void encode (CBOR *self, SV *scalar)
953 PPCODE: 1267 PPCODE:
954 PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN; 1268 PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN;
955 XPUSHs (scalar); 1269 XPUSHs (scalar);
968 EXTEND (SP, 2); 1282 EXTEND (SP, 2);
969 PUSHs (sv); 1283 PUSHs (sv);
970 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); 1284 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
971} 1285}
972 1286
973#if 0
974
975void DESTROY (CBOR *self) 1287void DESTROY (CBOR *self)
976 CODE: 1288 PPCODE:
977 SvREFCNT_dec (self->cb_sk_object); 1289 cbor_free (self);
978 SvREFCNT_dec (self->cb_object);
979
980#endif
981 1290
982PROTOTYPES: ENABLE 1291PROTOTYPES: ENABLE
983 1292
984void encode_cbor (SV *scalar) 1293void encode_cbor (SV *scalar)
1294 ALIAS:
1295 encode_cbor = 0
1296 encode_cbor_sharing = F_ALLOW_SHARING
985 PPCODE: 1297 PPCODE:
986{ 1298{
987 CBOR cbor; 1299 CBOR cbor;
988 cbor_init (&cbor); 1300 cbor_init (&cbor);
1301 cbor.flags |= ix;
989 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN; 1302 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;
990 XPUSHs (scalar); 1303 XPUSHs (scalar);
991} 1304}
992 1305
993void decode_cbor (SV *cborstr) 1306void decode_cbor (SV *cborstr)

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines