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 1.8 by root, Sun Oct 27 10:12:01 2013 UTC vs.
Revision 1.37 by root, Sat Nov 30 18:13:53 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// known tags, rfc7049 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
15enum cbor_tag 66enum cbor_tag
16{ 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
17 CBOR_TAG_DATETIME = 0, // rfc4287, utf-8 78 CBOR_TAG_DATETIME = 0, // rfc4287, utf-8
18 CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any 79 CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any
19 CBOR_TAG_POS_BIGNUM = 2, // byte string 80 CBOR_TAG_POS_BIGNUM = 2, // byte string
20 CBOR_TAG_NEG_BIGNUM = 3, // byte string 81 CBOR_TAG_NEG_BIGNUM = 3, // byte string
21 CBOR_TAG_DECIMAL = 4, // decimal fraction, array 82 CBOR_TAG_DECIMAL = 4, // decimal fraction, array
22 CBOR_TAG_BIGFLOAT = 5, // array 83 CBOR_TAG_BIGFLOAT = 5, // array
23 84
24 CBOR_TAG_CONV_B64U = 21, // base64url, any 85 CBOR_TAG_CONV_B64U = 21, // base64url, any
25 CBOR_TAG_CONV_B64 = 22, // base64, any 86 CBOR_TAG_CONV_B64 = 22, // base64, any
26 CBOR_TAG_CONV_HEX = 23, // base16, any 87 CBOR_TAG_CONV_HEX = 23, // base16, any
27 CBOR_TAG_CBOR = 24, // embedded cbor, byte string 88 CBOR_TAG_CBOR = 24, // embedded cbor, byte string
28 89
29 CBOR_TAG_URI = 32, // URI rfc3986, utf-8 90 CBOR_TAG_URI = 32, // URI rfc3986, utf-8
30 CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8 91 CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8
31 CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8 92 CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8
32 CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8 93 CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8
33 CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8 94 CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8
34 95
35 CBOR_TAG_MAGIC = 55799 96 CBOR_TAG_MAGIC = 55799, // self-describe cbor
36}; 97};
37 98
38#define F_SHRINK 0x00000200UL 99#define F_SHRINK 0x00000001UL
39#define F_ALLOW_UNKNOWN 0x00002000UL 100#define F_ALLOW_UNKNOWN 0x00000002UL
101#define F_ALLOW_SHARING 0x00000004UL
102#define F_ALLOW_CYCLES 0x00000008UL
103#define F_PACK_STRINGS 0x00000010UL
40 104
41#define INIT_SIZE 32 // initial scalar size to be allocated 105#define INIT_SIZE 32 // initial scalar size to be allocated
42 106
43#define SB do { 107#define SB do {
44#define SE } while (0) 108#define SE } while (0)
55#else 119#else
56# define CBOR_SLOW 0 120# define CBOR_SLOW 0
57# define CBOR_STASH cbor_stash 121# define CBOR_STASH cbor_stash
58#endif 122#endif
59 123
60static HV *cbor_stash, *cbor_boolean_stash, *cbor_tagged_stash; // CBOR::XS:: 124static HV *cbor_stash, *types_boolean_stash, *types_error_stash, *cbor_tagged_stash; // CBOR::XS::
61static SV *cbor_true, *cbor_false; 125static SV *types_true, *types_false, *types_error, *sv_cbor, *default_filter;
62 126
63typedef struct { 127typedef struct {
64 U32 flags; 128 U32 flags;
65 U32 max_depth; 129 U32 max_depth;
66 STRLEN max_size; 130 STRLEN max_size;
131 SV *filter;
67} CBOR; 132} CBOR;
68 133
69ecb_inline void 134ecb_inline void
70cbor_init (CBOR *cbor) 135cbor_init (CBOR *cbor)
71{ 136{
72 Zero (cbor, 1, CBOR); 137 Zero (cbor, 1, CBOR);
73 cbor->max_depth = 512; 138 cbor->max_depth = 512;
139}
140
141ecb_inline void
142cbor_free (CBOR *cbor)
143{
144 SvREFCNT_dec (cbor->filter);
74} 145}
75 146
76///////////////////////////////////////////////////////////////////////////// 147/////////////////////////////////////////////////////////////////////////////
77// utility functions 148// utility functions
78 149
100 SvPV_renew (sv, SvCUR (sv) + 1); 171 SvPV_renew (sv, SvCUR (sv) + 1);
101#endif 172#endif
102 } 173 }
103} 174}
104 175
105///////////////////////////////////////////////////////////////////////////// 176// minimum length of a string to be registered for stringref
106// fp hell 177ecb_inline int
107 178minimum_string_length (UV idx)
108//TODO 179{
180 return idx > 23
181 ? idx > 0xffU
182 ? idx > 0xffffU
183 ? idx > 0xffffffffU
184 ? 11
185 : 7
186 : 5
187 : 4
188 : 3;
189}
109 190
110///////////////////////////////////////////////////////////////////////////// 191/////////////////////////////////////////////////////////////////////////////
111// encoder 192// encoder
112 193
113// structure used for encoding CBOR 194// structure used for encoding CBOR
116 char *cur; // SvPVX (sv) + current output position 197 char *cur; // SvPVX (sv) + current output position
117 char *end; // SvEND (sv) 198 char *end; // SvEND (sv)
118 SV *sv; // result scalar 199 SV *sv; // result scalar
119 CBOR cbor; 200 CBOR cbor;
120 U32 depth; // recursion level 201 U32 depth; // recursion level
202 HV *stringref[2]; // string => index, or 0 ([0] = bytes, [1] = utf-8)
203 UV stringref_idx;
204 HV *shareable; // ptr => index, or 0
205 UV shareable_idx;
121} enc_t; 206} enc_t;
122 207
123ecb_inline void 208ecb_inline void
124need (enc_t *enc, STRLEN len) 209need (enc_t *enc, STRLEN len)
125{ 210{
142static void 227static void
143encode_uint (enc_t *enc, int major, UV len) 228encode_uint (enc_t *enc, int major, UV len)
144{ 229{
145 need (enc, 9); 230 need (enc, 9);
146 231
147 if (len < 24) 232 if (ecb_expect_true (len < LENGTH_EXT1))
148 *enc->cur++ = major | len; 233 *enc->cur++ = major | len;
149 else if (len <= 0xff) 234 else if (ecb_expect_true (len <= 0xffU))
150 { 235 {
151 *enc->cur++ = major | 24; 236 *enc->cur++ = major | LENGTH_EXT1;
152 *enc->cur++ = len; 237 *enc->cur++ = len;
153 } 238 }
154 else if (len <= 0xffff) 239 else if (len <= 0xffffU)
155 { 240 {
156 *enc->cur++ = major | 25; 241 *enc->cur++ = major | LENGTH_EXT2;
157 *enc->cur++ = len >> 8; 242 *enc->cur++ = len >> 8;
158 *enc->cur++ = len; 243 *enc->cur++ = len;
159 } 244 }
160 else if (len <= 0xffffffff) 245 else if (len <= 0xffffffffU)
161 { 246 {
162 *enc->cur++ = major | 26; 247 *enc->cur++ = major | LENGTH_EXT4;
163 *enc->cur++ = len >> 24; 248 *enc->cur++ = len >> 24;
164 *enc->cur++ = len >> 16; 249 *enc->cur++ = len >> 16;
165 *enc->cur++ = len >> 8; 250 *enc->cur++ = len >> 8;
166 *enc->cur++ = len; 251 *enc->cur++ = len;
167 } 252 }
168 else 253 else
169 { 254 {
170 *enc->cur++ = major | 27; 255 *enc->cur++ = major | LENGTH_EXT8;
171 *enc->cur++ = len >> 56; 256 *enc->cur++ = len >> 56;
172 *enc->cur++ = len >> 48; 257 *enc->cur++ = len >> 48;
173 *enc->cur++ = len >> 40; 258 *enc->cur++ = len >> 40;
174 *enc->cur++ = len >> 32; 259 *enc->cur++ = len >> 32;
175 *enc->cur++ = len >> 24; 260 *enc->cur++ = len >> 24;
177 *enc->cur++ = len >> 8; 262 *enc->cur++ = len >> 8;
178 *enc->cur++ = len; 263 *enc->cur++ = len;
179 } 264 }
180} 265}
181 266
182static void 267ecb_inline void
268encode_tag (enc_t *enc, UV tag)
269{
270 encode_uint (enc, MAJOR_TAG, tag);
271}
272
273ecb_inline void
183encode_str (enc_t *enc, int utf8, char *str, STRLEN len) 274encode_str (enc_t *enc, int utf8, char *str, STRLEN len)
184{ 275{
185 encode_uint (enc, utf8 ? 0x60 : 0x40, len); 276 encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len);
186 need (enc, len); 277 need (enc, len);
187 memcpy (enc->cur, str, len); 278 memcpy (enc->cur, str, len);
188 enc->cur += len; 279 enc->cur += len;
189} 280}
190 281
282static void
283encode_strref (enc_t *enc, int utf8, char *str, STRLEN len)
284{
285 if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS))
286 {
287 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1);
288
289 if (SvOK (*svp))
290 {
291 // already registered, use stringref
292 encode_tag (enc, CBOR_TAG_STRINGREF);
293 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
294 return;
295 }
296 else if (len >= minimum_string_length (enc->stringref_idx))
297 {
298 // register only
299 sv_setuv (*svp, enc->stringref_idx);
300 ++enc->stringref_idx;
301 }
302 }
303
304 encode_str (enc, utf8, str, len);
305}
306
191static void encode_sv (enc_t *enc, SV *sv); 307static void encode_sv (enc_t *enc, SV *sv);
192 308
193static void 309static void
194encode_av (enc_t *enc, AV *av) 310encode_av (enc_t *enc, AV *av)
195{ 311{
198 if (enc->depth >= enc->cbor.max_depth) 314 if (enc->depth >= enc->cbor.max_depth)
199 croak (ERR_NESTING_EXCEEDED); 315 croak (ERR_NESTING_EXCEEDED);
200 316
201 ++enc->depth; 317 ++enc->depth;
202 318
203 encode_uint (enc, 0x80, len + 1); 319 encode_uint (enc, MAJOR_ARRAY, len + 1);
204 320
205 for (i = 0; i <= len; ++i) 321 for (i = 0; i <= len; ++i)
206 { 322 {
207 SV **svp = av_fetch (av, i, 0); 323 SV **svp = av_fetch (av, i, 0);
208 encode_sv (enc, svp ? *svp : &PL_sv_undef); 324 encode_sv (enc, svp ? *svp : &PL_sv_undef);
223 339
224 int pairs = hv_iterinit (hv); 340 int pairs = hv_iterinit (hv);
225 int mg = SvMAGICAL (hv); 341 int mg = SvMAGICAL (hv);
226 342
227 if (mg) 343 if (mg)
228 encode_ch (enc, 0xa0 | 31); 344 encode_ch (enc, MAJOR_MAP | MINOR_INDEF);
229 else 345 else
230 encode_uint (enc, 0xa0, pairs); 346 encode_uint (enc, MAJOR_MAP, pairs);
231 347
232 while ((he = hv_iternext (hv))) 348 while ((he = hv_iternext (hv)))
233 { 349 {
234 if (HeKLEN (he) == HEf_SVKEY) 350 if (HeKLEN (he) == HEf_SVKEY)
235 encode_sv (enc, HeSVKEY (he)); 351 encode_sv (enc, HeSVKEY (he));
236 else 352 else
237 encode_str (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he)); 353 encode_strref (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he));
238 354
239 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he)); 355 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he));
240 } 356 }
241 357
242 if (mg) 358 if (mg)
243 encode_ch (enc, 0xe0 | 31); 359 encode_ch (enc, MAJOR_MISC | MINOR_INDEF);
244 360
245 --enc->depth; 361 --enc->depth;
246} 362}
247 363
248// encode objects, arrays and special \0=false and \1=true values. 364// encode objects, arrays and special \0=false and \1=true values.
249static void 365static void
250encode_rv (enc_t *enc, SV *sv) 366encode_rv (enc_t *enc, SV *sv)
251{ 367{
252 svtype svt;
253
254 SvGETMAGIC (sv); 368 SvGETMAGIC (sv);
369
255 svt = SvTYPE (sv); 370 svtype svt = SvTYPE (sv);
256 371
257 if (ecb_expect_false (SvOBJECT (sv))) 372 if (ecb_expect_false (SvOBJECT (sv)))
258 { 373 {
259 HV *boolean_stash = !CBOR_SLOW || cbor_boolean_stash 374 HV *boolean_stash = !CBOR_SLOW || types_boolean_stash
260 ? cbor_boolean_stash 375 ? types_boolean_stash
261 : gv_stashpv ("CBOR::XS::Boolean", 1); 376 : gv_stashpv ("Types::Serialiser::Boolean", 1);
377 HV *error_stash = !CBOR_SLOW || types_error_stash
378 ? types_error_stash
379 : gv_stashpv ("Types::Serialiser::Error", 1);
262 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 380 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
263 ? cbor_tagged_stash 381 ? cbor_tagged_stash
264 : gv_stashpv ("CBOR::XS::Tagged" , 1); 382 : gv_stashpv ("CBOR::XS::Tagged" , 1);
265 383
384 HV *stash = SvSTASH (sv);
385
266 if (SvSTASH (sv) == boolean_stash) 386 if (stash == boolean_stash)
267 encode_ch (enc, SvIV (sv) ? 0xe0 | 21 : 0xe0 | 20); 387 {
388 encode_ch (enc, SvIV (sv) ? MAJOR_MISC | SIMPLE_TRUE : MAJOR_MISC | SIMPLE_FALSE);
389 return;
390 }
391 else if (stash == error_stash)
392 {
393 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
394 return;
395 }
268 else if (SvSTASH (sv) == tagged_stash) 396 else if (stash == tagged_stash)
269 { 397 {
270 if (svt != SVt_PVAV) 398 if (svt != SVt_PVAV)
271 croak ("encountered CBOR::XS::Tagged object that isn't an array"); 399 croak ("encountered CBOR::XS::Tagged object that isn't an array");
272 400
273 encode_uint (enc, 0xc0, SvUV (*av_fetch ((AV *)sv, 0, 1))); 401 encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1)));
274 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1)); 402 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1));
403
404 return;
405 }
406 }
407
408 if (ecb_expect_false (SvREFCNT (sv) > 1)
409 && ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING))
410 {
411 if (!enc->shareable)
412 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
413
414 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
415
416 if (SvOK (*svp))
417 {
418 encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF);
419 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
420 return;
275 } 421 }
276 else 422 else
277 { 423 {
424 sv_setuv (*svp, enc->shareable_idx);
425 ++enc->shareable_idx;
426 encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE);
427 }
428 }
429
430 if (ecb_expect_false (SvOBJECT (sv)))
431 {
432 HV *stash = SvSTASH (sv);
433 GV *method;
434
435 if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
436 {
437 dSP;
438
439 ENTER; SAVETMPS; PUSHMARK (SP);
278 // we re-bless the reference to get overload and other niceties right 440 // we re-bless the reference to get overload and other niceties right
279 GV *to_cbor = gv_fetchmethod_autoload (SvSTASH (sv), "TO_CBOR", 0);
280
281 if (to_cbor)
282 {
283 dSP;
284
285 ENTER; SAVETMPS; PUSHMARK (SP);
286 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv))); 441 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
287 442
288 // calling with G_SCALAR ensures that we always get a 1 return value
289 PUTBACK; 443 PUTBACK;
444 // G_SCALAR ensures that return value is 1
290 call_sv ((SV *)GvCV (to_cbor), G_SCALAR); 445 call_sv ((SV *)GvCV (method), G_SCALAR);
291 SPAGAIN; 446 SPAGAIN;
292 447
293 // catch this surprisingly common error 448 // catch this surprisingly common error
294 if (SvROK (TOPs) && SvRV (TOPs) == sv) 449 if (SvROK (TOPs) && SvRV (TOPs) == sv)
295 croak ("%s::TO_CBOR method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv))); 450 croak ("%s::TO_CBOR method returned same object as was passed instead of a new one", HvNAME (stash));
296 451
297 sv = POPs; 452 encode_sv (enc, POPs);
453
298 PUTBACK; 454 PUTBACK;
299 455
300 encode_sv (enc, sv);
301
302 FREETMPS; LEAVE; 456 FREETMPS; LEAVE;
303 } 457 }
458 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)
459 {
460 dSP;
461
462 ENTER; SAVETMPS; PUSHMARK (SP);
463 EXTEND (SP, 2);
464 // we re-bless the reference to get overload and other niceties right
465 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
466 PUSHs (sv_cbor);
467
468 PUTBACK;
469 int count = call_sv ((SV *)GvCV (method), G_ARRAY);
470 SPAGAIN;
471
472 // catch this surprisingly common error
473 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)
474 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));
475
476 encode_tag (enc, CBOR_TAG_PERL_OBJECT);
477 encode_uint (enc, MAJOR_ARRAY, count + 1);
478 encode_strref (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
479
480 while (count)
481 encode_sv (enc, SP[1 - count--]);
482
483 PUTBACK;
484
485 FREETMPS; LEAVE;
486 }
304 else 487 else
305 croak ("encountered object '%s', but no TO_CBOR method available on it", 488 croak ("encountered object '%s', but no TO_CBOR or FREEZE methods available on it",
306 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 489 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
307 }
308 } 490 }
309 else if (svt == SVt_PVHV) 491 else if (svt == SVt_PVHV)
310 encode_hv (enc, (HV *)sv); 492 encode_hv (enc, (HV *)sv);
311 else if (svt == SVt_PVAV) 493 else if (svt == SVt_PVAV)
312 encode_av (enc, (AV *)sv); 494 encode_av (enc, (AV *)sv);
313 else if (svt < SVt_PVAV)
314 {
315 STRLEN len = 0;
316 char *pv = svt ? SvPV (sv, len) : 0;
317
318 if (len == 1 && *pv == '1')
319 encode_ch (enc, 0xe0 | 21);
320 else if (len == 1 && *pv == '0')
321 encode_ch (enc, 0xe0 | 20);
322 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
323 encode_ch (enc, 0xe0 | 23);
324 else
325 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
326 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
327 }
328 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
329 encode_ch (enc, 0xe0 | 23);
330 else 495 else
331 croak ("encountered %s, but CBOR can only represent references to arrays or hashes", 496 {
332 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 497 encode_tag (enc, CBOR_TAG_INDIRECTION);
498 encode_sv (enc, sv);
499 }
333} 500}
334 501
335static void 502static void
336encode_nv (enc_t *enc, SV *sv) 503encode_nv (enc_t *enc, SV *sv)
337{ 504{
338 double nv = SvNVX (sv); 505 double nv = SvNVX (sv);
339 506
340 need (enc, 9); 507 need (enc, 9);
341 508
342 if (ecb_expect_false (nv == (U32)nv)) 509 if (ecb_expect_false (nv == (NV)(U32)nv))
343 encode_uint (enc, 0x00, (U32)nv); 510 encode_uint (enc, MAJOR_POS_INT, (U32)nv);
344 //TODO: maybe I32? 511 //TODO: maybe I32?
345 else if (ecb_expect_false (nv == (float)nv)) 512 else if (ecb_expect_false (nv == (float)nv))
346 { 513 {
347 uint32_t fp = ecb_float_to_binary32 (nv); 514 uint32_t fp = ecb_float_to_binary32 (nv);
348 515
349 *enc->cur++ = 0xe0 | 26; 516 *enc->cur++ = MAJOR_MISC | MISC_FLOAT32;
350 517
351 if (!ecb_big_endian ()) 518 if (!ecb_big_endian ())
352 fp = ecb_bswap32 (fp); 519 fp = ecb_bswap32 (fp);
353 520
354 memcpy (enc->cur, &fp, 4); 521 memcpy (enc->cur, &fp, 4);
356 } 523 }
357 else 524 else
358 { 525 {
359 uint64_t fp = ecb_double_to_binary64 (nv); 526 uint64_t fp = ecb_double_to_binary64 (nv);
360 527
361 *enc->cur++ = 0xe0 | 27; 528 *enc->cur++ = MAJOR_MISC | MISC_FLOAT64;
362 529
363 if (!ecb_big_endian ()) 530 if (!ecb_big_endian ())
364 fp = ecb_bswap64 (fp); 531 fp = ecb_bswap64 (fp);
365 532
366 memcpy (enc->cur, &fp, 8); 533 memcpy (enc->cur, &fp, 8);
375 542
376 if (SvPOKp (sv)) 543 if (SvPOKp (sv))
377 { 544 {
378 STRLEN len; 545 STRLEN len;
379 char *str = SvPV (sv, len); 546 char *str = SvPV (sv, len);
380 encode_str (enc, SvUTF8 (sv), str, len); 547 encode_strref (enc, SvUTF8 (sv), str, len);
381 } 548 }
382 else if (SvNOKp (sv)) 549 else if (SvNOKp (sv))
383 encode_nv (enc, sv); 550 encode_nv (enc, sv);
384 else if (SvIOKp (sv)) 551 else if (SvIOKp (sv))
385 { 552 {
386 if (SvIsUV (sv)) 553 if (SvIsUV (sv))
387 encode_uint (enc, 0x00, SvUVX (sv)); 554 encode_uint (enc, MAJOR_POS_INT, SvUVX (sv));
388 else if (SvIVX (sv) >= 0) 555 else if (SvIVX (sv) >= 0)
389 encode_uint (enc, 0x00, SvIVX (sv)); 556 encode_uint (enc, MAJOR_POS_INT, SvIVX (sv));
390 else 557 else
391 encode_uint (enc, 0x20, -(SvIVX (sv) + 1)); 558 encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1));
392 } 559 }
393 else if (SvROK (sv)) 560 else if (SvROK (sv))
394 encode_rv (enc, SvRV (sv)); 561 encode_rv (enc, SvRV (sv));
395 else if (!SvOK (sv)) 562 else if (!SvOK (sv))
396 encode_ch (enc, 0xe0 | 22); 563 encode_ch (enc, MAJOR_MISC | SIMPLE_NULL);
397 else if (enc->cbor.flags & F_ALLOW_UNKNOWN) 564 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
398 encode_ch (enc, 0xe0 | 23); 565 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
399 else 566 else
400 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data", 567 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data",
401 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv)); 568 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
402} 569}
403 570
404static SV * 571static SV *
405encode_cbor (SV *scalar, CBOR *cbor) 572encode_cbor (SV *scalar, CBOR *cbor)
406{ 573{
407 enc_t enc; 574 enc_t enc = { };
408 575
409 enc.cbor = *cbor; 576 enc.cbor = *cbor;
410 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 577 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
411 enc.cur = SvPVX (enc.sv); 578 enc.cur = SvPVX (enc.sv);
412 enc.end = SvEND (enc.sv); 579 enc.end = SvEND (enc.sv);
413 enc.depth = 0;
414 580
415 SvPOK_only (enc.sv); 581 SvPOK_only (enc.sv);
582
583 if (cbor->flags & F_PACK_STRINGS)
584 {
585 encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE);
586 enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ());
587 enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ());
588 }
589
416 encode_sv (&enc, scalar); 590 encode_sv (&enc, scalar);
417 591
418 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 592 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
419 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings 593 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
420 594
434 U8 *end; // end of input string 608 U8 *end; // end of input string
435 const char *err; // parse error, if != 0 609 const char *err; // parse error, if != 0
436 CBOR cbor; 610 CBOR cbor;
437 U32 depth; // recursion depth 611 U32 depth; // recursion depth
438 U32 maxdepth; // recursion depth limit 612 U32 maxdepth; // recursion depth limit
613 AV *shareable;
614 AV *stringref;
615 SV *decode_tagged;
439} dec_t; 616} dec_t;
440 617
441#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE 618#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE
442 619
443#define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data") 620#define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data")
446#define DEC_DEC_DEPTH --dec->depth 623#define DEC_DEC_DEPTH --dec->depth
447 624
448static UV 625static UV
449decode_uint (dec_t *dec) 626decode_uint (dec_t *dec)
450{ 627{
451 switch (*dec->cur & 31) 628 U8 m = *dec->cur & MINOR_MASK;
452 { 629 ++dec->cur;
453 case 0: case 1: case 2: case 3: case 4: case 5: case 6: case 7:
454 case 8: case 9: case 10: case 11: case 12: case 13: case 14: case 15:
455 case 16: case 17: case 18: case 19: case 20: case 21: case 22: case 23:
456 return *dec->cur++ & 31;
457 630
458 case 24: 631 if (ecb_expect_true (m < LENGTH_EXT1))
632 return m;
633 else if (ecb_expect_true (m == LENGTH_EXT1))
634 {
459 WANT (2); 635 WANT (1);
460 dec->cur += 2; 636 dec->cur += 1;
461 return dec->cur[-1]; 637 return dec->cur[-1];
462 638 }
463 case 25: 639 else if (ecb_expect_true (m == LENGTH_EXT2))
640 {
464 WANT (3); 641 WANT (2);
465 dec->cur += 3; 642 dec->cur += 2;
466 return (((UV)dec->cur[-2]) << 8) 643 return (((UV)dec->cur[-2]) << 8)
467 | ((UV)dec->cur[-1]); 644 | ((UV)dec->cur[-1]);
468 645 }
469 case 26: 646 else if (ecb_expect_true (m == LENGTH_EXT4))
647 {
470 WANT (5); 648 WANT (4);
471 dec->cur += 5; 649 dec->cur += 4;
472 return (((UV)dec->cur[-4]) << 24) 650 return (((UV)dec->cur[-4]) << 24)
473 | (((UV)dec->cur[-3]) << 16) 651 | (((UV)dec->cur[-3]) << 16)
474 | (((UV)dec->cur[-2]) << 8) 652 | (((UV)dec->cur[-2]) << 8)
475 | ((UV)dec->cur[-1]); 653 | ((UV)dec->cur[-1]);
476 654 }
477 case 27: 655 else if (ecb_expect_true (m == LENGTH_EXT8))
656 {
478 WANT (9); 657 WANT (8);
479 dec->cur += 9; 658 dec->cur += 8;
659
660 return
661#if UVSIZE < 8
662 0
663#else
480 return (((UV)dec->cur[-8]) << 56) 664 (((UV)dec->cur[-8]) << 56)
481 | (((UV)dec->cur[-7]) << 48) 665 | (((UV)dec->cur[-7]) << 48)
482 | (((UV)dec->cur[-6]) << 40) 666 | (((UV)dec->cur[-6]) << 40)
483 | (((UV)dec->cur[-5]) << 32) 667 | (((UV)dec->cur[-5]) << 32)
668#endif
484 | (((UV)dec->cur[-4]) << 24) 669 | (((UV)dec->cur[-4]) << 24)
485 | (((UV)dec->cur[-3]) << 16) 670 | (((UV)dec->cur[-3]) << 16)
486 | (((UV)dec->cur[-2]) << 8) 671 | (((UV)dec->cur[-2]) << 8)
487 | ((UV)dec->cur[-1]); 672 | ((UV)dec->cur[-1]);
488 673 }
489 default: 674 else
490 ERR ("corrupted CBOR data (unsupported integer minor encoding)"); 675 ERR ("corrupted CBOR data (unsupported integer minor encoding)");
491 }
492 676
493fail: 677fail:
494 return 0; 678 return 0;
495} 679}
496 680
501{ 685{
502 AV *av = newAV (); 686 AV *av = newAV ();
503 687
504 DEC_INC_DEPTH; 688 DEC_INC_DEPTH;
505 689
506 if ((*dec->cur & 31) == 31) 690 if (*dec->cur == (MAJOR_ARRAY | MINOR_INDEF))
507 { 691 {
508 ++dec->cur; 692 ++dec->cur;
509 693
510 for (;;) 694 for (;;)
511 { 695 {
512 WANT (1); 696 WANT (1);
513 697
514 if (*dec->cur == (0xe0 | 31)) 698 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
515 { 699 {
516 ++dec->cur; 700 ++dec->cur;
517 break; 701 break;
518 } 702 }
519 703
522 } 706 }
523 else 707 else
524 { 708 {
525 int i, len = decode_uint (dec); 709 int i, len = decode_uint (dec);
526 710
711 WANT (len); // complexity check for av_fill - need at least one byte per value, do not allow supersize arrays
527 av_fill (av, len - 1); 712 av_fill (av, len - 1);
528 713
529 for (i = 0; i < len; ++i) 714 for (i = 0; i < len; ++i)
530 AvARRAY (av)[i] = decode_sv (dec); 715 AvARRAY (av)[i] = decode_sv (dec);
531 } 716 }
537 SvREFCNT_dec (av); 722 SvREFCNT_dec (av);
538 DEC_DEC_DEPTH; 723 DEC_DEC_DEPTH;
539 return &PL_sv_undef; 724 return &PL_sv_undef;
540} 725}
541 726
727static void
728decode_he (dec_t *dec, HV *hv)
729{
730 // for speed reasons, we specialcase single-string
731 // byte or utf-8 strings as keys, but only when !stringref
732
733 if (ecb_expect_true (!dec->stringref))
734 if (ecb_expect_true ((*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8))
735 {
736 I32 len = decode_uint (dec);
737 char *key = (char *)dec->cur;
738
739 dec->cur += len;
740
741 if (ecb_expect_false (dec->stringref))
742 av_push (dec->stringref, newSVpvn (key, len));
743
744 hv_store (hv, key, len, decode_sv (dec), 0);
745
746 return;
747 }
748 else if (ecb_expect_true ((*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8))
749 {
750 I32 len = decode_uint (dec);
751 char *key = (char *)dec->cur;
752
753 dec->cur += len;
754
755 if (ecb_expect_false (dec->stringref))
756 av_push (dec->stringref, newSVpvn_utf8 (key, len, 1));
757
758 hv_store (hv, key, -len, decode_sv (dec), 0);
759
760 return;
761 }
762
763 SV *k = decode_sv (dec);
764 SV *v = decode_sv (dec);
765
766 hv_store_ent (hv, k, v, 0);
767 SvREFCNT_dec (k);
768}
769
542static SV * 770static SV *
543decode_hv (dec_t *dec) 771decode_hv (dec_t *dec)
544{ 772{
545 HV *hv = newHV (); 773 HV *hv = newHV ();
546 774
547 DEC_INC_DEPTH; 775 DEC_INC_DEPTH;
548 776
549 if ((*dec->cur & 31) == 31) 777 if (*dec->cur == (MAJOR_MAP | MINOR_INDEF))
550 { 778 {
551 ++dec->cur; 779 ++dec->cur;
552 780
553 for (;;) 781 for (;;)
554 { 782 {
555 WANT (1); 783 WANT (1);
556 784
557 if (*dec->cur == (0xe0 | 31)) 785 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
558 { 786 {
559 ++dec->cur; 787 ++dec->cur;
560 break; 788 break;
561 } 789 }
562 790
563 SV *k = decode_sv (dec); 791 decode_he (dec, hv);
564 SV *v = decode_sv (dec);
565
566 hv_store_ent (hv, k, v, 0);
567 } 792 }
568 } 793 }
569 else 794 else
570 { 795 {
571 int len = decode_uint (dec); 796 int pairs = decode_uint (dec);
572 797
573 while (len--) 798 while (pairs--)
574 { 799 decode_he (dec, hv);
575 SV *k = decode_sv (dec);
576 SV *v = decode_sv (dec);
577
578 hv_store_ent (hv, k, v, 0);
579 }
580 } 800 }
581 801
582 DEC_DEC_DEPTH; 802 DEC_DEC_DEPTH;
583 return newRV_noinc ((SV *)hv); 803 return newRV_noinc ((SV *)hv);
584 804
591static SV * 811static SV *
592decode_str (dec_t *dec, int utf8) 812decode_str (dec_t *dec, int utf8)
593{ 813{
594 SV *sv = 0; 814 SV *sv = 0;
595 815
596 if ((*dec->cur & 31) == 31) 816 if ((*dec->cur & MINOR_MASK) == MINOR_INDEF)
597 { 817 {
818 // indefinite length strings
598 ++dec->cur; 819 ++dec->cur;
599 820
821 U8 major = *dec->cur & MAJOR_MISC;
822
600 sv = newSVpvn ("", 0); 823 sv = newSVpvn ("", 0);
601 824
602 // not very fast, and certainly not robust against illegal input
603 for (;;) 825 for (;;)
604 { 826 {
605 WANT (1); 827 WANT (1);
606 828
607 if (*dec->cur == (0xe0 | 31)) 829 if ((*dec->cur - major) > LENGTH_EXT8)
830 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
608 { 831 {
609 ++dec->cur; 832 ++dec->cur;
610 break; 833 break;
611 } 834 }
835 else
836 ERR ("corrupted CBOR data (invalid chunks in indefinite length string)");
612 837
613 sv_catsv (sv, decode_sv (dec)); 838 STRLEN len = decode_uint (dec);
839
840 WANT (len);
841 sv_catpvn (sv, dec->cur, len);
842 dec->cur += len;
614 } 843 }
615 } 844 }
616 else 845 else
617 { 846 {
618 STRLEN len = decode_uint (dec); 847 STRLEN len = decode_uint (dec);
619 848
620 WANT (len); 849 WANT (len);
621 sv = newSVpvn (dec->cur, len); 850 sv = newSVpvn (dec->cur, len);
622 dec->cur += len; 851 dec->cur += len;
852
853 if (ecb_expect_false (dec->stringref)
854 && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1))
855 av_push (dec->stringref, SvREFCNT_inc_NN (sv));
623 } 856 }
624 857
625 if (utf8) 858 if (utf8)
626 SvUTF8_on (sv); 859 SvUTF8_on (sv);
627 860
633} 866}
634 867
635static SV * 868static SV *
636decode_tagged (dec_t *dec) 869decode_tagged (dec_t *dec)
637{ 870{
871 SV *sv = 0;
638 UV tag = decode_uint (dec); 872 UV tag = decode_uint (dec);
873
874 WANT (1);
875
876 switch (tag)
877 {
878 case CBOR_TAG_MAGIC:
639 SV *sv = decode_sv (dec); 879 sv = decode_sv (dec);
880 break;
640 881
641 if (tag == CBOR_TAG_MAGIC) // 2.4.5 Self-Describe CBOR 882 case CBOR_TAG_INDIRECTION:
883 sv = newRV_noinc (decode_sv (dec));
884 break;
885
886 case CBOR_TAG_STRINGREF_NAMESPACE:
887 {
888 ENTER; SAVETMPS;
889
890 SAVESPTR (dec->stringref);
891 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ());
892
893 sv = decode_sv (dec);
894
895 FREETMPS; LEAVE;
896 }
897 break;
898
899 case CBOR_TAG_STRINGREF:
900 {
901 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
902 ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
903
904 UV idx = decode_uint (dec);
905
906 if (!dec->stringref || (int)idx > AvFILLp (dec->stringref))
907 ERR ("corrupted CBOR data (stringref index out of bounds or outside namespace)");
908
909 sv = newSVsv (AvARRAY (dec->stringref)[idx]);
910 }
911 break;
912
913 case CBOR_TAG_VALUE_SHAREABLE:
914 {
915 if (ecb_expect_false (!dec->shareable))
916 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ());
917
918 if (dec->cbor.flags & F_ALLOW_CYCLES)
919 {
920 sv = newSV (0);
921 av_push (dec->shareable, SvREFCNT_inc_NN (sv));
922
923 SV *osv = decode_sv (dec);
924 sv_setsv (sv, osv);
925 SvREFCNT_dec_NN (osv);
926 }
927 else
928 {
929 av_push (dec->shareable, &PL_sv_undef);
930 int idx = AvFILLp (dec->shareable);
931 sv = decode_sv (dec);
932 av_store (dec->shareable, idx, SvREFCNT_inc_NN (sv));
933 }
934 }
935 break;
936
937 case CBOR_TAG_VALUE_SHAREDREF:
938 {
939 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
940 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
941
942 UV idx = decode_uint (dec);
943
944 if (!dec->shareable || (int)idx > AvFILLp (dec->shareable))
945 ERR ("corrupted CBOR data (sharedref index out of bounds)");
946
947 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]);
948
949 if (sv == &PL_sv_undef)
950 ERR ("cyclic CBOR data structure found, but allow_cycles is not enabled");
951 }
952 break;
953
954 case CBOR_TAG_PERL_OBJECT:
955 {
956 sv = decode_sv (dec);
957
958 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
959 ERR ("corrupted CBOR data (non-array perl object)");
960
961 AV *av = (AV *)SvRV (sv);
962 int len = av_len (av) + 1;
963 HV *stash = gv_stashsv (*av_fetch (av, 0, 1), 0);
964
965 if (!stash)
966 ERR ("cannot decode perl-object (package does not exist)");
967
968 GV *method = gv_fetchmethod_autoload (stash, "THAW", 0);
969
970 if (!method)
971 ERR ("cannot decode perl-object (package does not have a THAW method)");
972
973 dSP;
974
975 ENTER; SAVETMPS; PUSHMARK (SP);
976 EXTEND (SP, len + 1);
977 // we re-bless the reference to get overload and other niceties right
978 PUSHs (*av_fetch (av, 0, 1));
979 PUSHs (sv_cbor);
980
981 int i;
982
983 for (i = 1; i < len; ++i)
984 PUSHs (*av_fetch (av, i, 1));
985
986 PUTBACK;
987 call_sv ((SV *)GvCV (method), G_SCALAR | G_EVAL);
988 SPAGAIN;
989
990 if (SvTRUE (ERRSV))
991 {
992 FREETMPS; LEAVE;
993 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV))));
994 }
995
996 SvREFCNT_dec (sv);
997 sv = SvREFCNT_inc (POPs);
998
999 PUTBACK;
1000
1001 FREETMPS; LEAVE;
1002 }
1003 break;
1004
1005 default:
1006 {
1007 sv = decode_sv (dec);
1008
1009 dSP;
1010 ENTER; SAVETMPS; PUSHMARK (SP);
1011 EXTEND (SP, 2);
1012 PUSHs (newSVuv (tag));
1013 PUSHs (sv);
1014
1015 PUTBACK;
1016 int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL);
1017 SPAGAIN;
1018
1019 if (SvTRUE (ERRSV))
1020 {
1021 FREETMPS; LEAVE;
1022 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV))));
1023 }
1024
1025 if (count)
1026 {
1027 SvREFCNT_dec (sv);
1028 sv = SvREFCNT_inc (POPs);
1029 }
1030 else
1031 {
1032 AV *av = newAV ();
1033 av_push (av, newSVuv (tag));
1034 av_push (av, sv);
1035
1036 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
1037 ? cbor_tagged_stash
1038 : gv_stashpv ("CBOR::XS::Tagged" , 1);
1039 sv = sv_bless (newRV_noinc ((SV *)av), tagged_stash);
1040 }
1041
1042 PUTBACK;
1043
1044 FREETMPS; LEAVE;
1045 }
1046 break;
1047 }
1048
642 return sv; 1049 return sv;
643 1050
644 AV *av = newAV (); 1051fail:
645 av_push (av, newSVuv (tag)); 1052 SvREFCNT_dec (sv);
646 av_push (av, sv); 1053 return &PL_sv_undef;
647
648 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
649 ? cbor_tagged_stash
650 : gv_stashpv ("CBOR::XS::Tagged" , 1);
651
652 return sv_bless (newRV_noinc ((SV *)av), tagged_stash);
653} 1054}
654 1055
655static SV * 1056static SV *
656decode_sv (dec_t *dec) 1057decode_sv (dec_t *dec)
657{ 1058{
658 WANT (1); 1059 WANT (1);
659 1060
660 switch (*dec->cur >> 5) 1061 switch (*dec->cur >> MAJOR_SHIFT)
661 { 1062 {
662 case 0: // unsigned int 1063 case MAJOR_POS_INT >> MAJOR_SHIFT: return newSVuv (decode_uint (dec));
663 return newSVuv (decode_uint (dec)); 1064 case MAJOR_NEG_INT >> MAJOR_SHIFT: return newSViv (-1 - (IV)decode_uint (dec));
664 case 1: // negative int 1065 case MAJOR_BYTES >> MAJOR_SHIFT: return decode_str (dec, 0);
665 return newSViv (-1 - (IV)decode_uint (dec)); 1066 case MAJOR_TEXT >> MAJOR_SHIFT: return decode_str (dec, 1);
666 case 2: // octet string 1067 case MAJOR_ARRAY >> MAJOR_SHIFT: return decode_av (dec);
667 return decode_str (dec, 0); 1068 case MAJOR_MAP >> MAJOR_SHIFT: return decode_hv (dec);
668 case 3: // utf-8 string 1069 case MAJOR_TAG >> MAJOR_SHIFT: return decode_tagged (dec);
669 return decode_str (dec, 1); 1070
670 case 4: // array 1071 case MAJOR_MISC >> MAJOR_SHIFT:
671 return decode_av (dec);
672 case 5: // map
673 return decode_hv (dec);
674 case 6: // tag
675 return decode_tagged (dec);
676 case 7: // misc
677 switch (*dec->cur++ & 31) 1072 switch (*dec->cur++ & MINOR_MASK)
678 { 1073 {
679 case 20: 1074 case SIMPLE_FALSE:
680#if CBOR_SLOW 1075#if CBOR_SLOW
681 cbor_false = get_bool ("CBOR::XS::false"); 1076 types_false = get_bool ("Types::Serialiser::false");
682#endif 1077#endif
683 return newSVsv (cbor_false); 1078 return newSVsv (types_false);
684 case 21: 1079 case SIMPLE_TRUE:
685#if CBOR_SLOW 1080#if CBOR_SLOW
686 cbor_true = get_bool ("CBOR::XS::true"); 1081 types_true = get_bool ("Types::Serialiser::true");
687#endif 1082#endif
688 return newSVsv (cbor_true); 1083 return newSVsv (types_true);
689 case 22: 1084 case SIMPLE_NULL:
690 return newSVsv (&PL_sv_undef); 1085 return newSVsv (&PL_sv_undef);
1086 case SIMPLE_UNDEF:
1087#if CBOR_SLOW
1088 types_error = get_bool ("Types::Serialiser::error");
1089#endif
1090 return newSVsv (types_error);
691 1091
692 case 25: 1092 case MISC_FLOAT16:
693 { 1093 {
694 WANT (2); 1094 WANT (2);
695 1095
696 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1]; 1096 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1];
697 dec->cur += 2; 1097 dec->cur += 2;
698 1098
699 return newSVnv (ecb_binary16_to_float (fp)); 1099 return newSVnv (ecb_binary16_to_float (fp));
700 } 1100 }
701 1101
702 case 26: 1102 case MISC_FLOAT32:
703 { 1103 {
704 uint32_t fp; 1104 uint32_t fp;
705 WANT (4); 1105 WANT (4);
706 memcpy (&fp, dec->cur, 4); 1106 memcpy (&fp, dec->cur, 4);
707 dec->cur += 4; 1107 dec->cur += 4;
710 fp = ecb_bswap32 (fp); 1110 fp = ecb_bswap32 (fp);
711 1111
712 return newSVnv (ecb_binary32_to_float (fp)); 1112 return newSVnv (ecb_binary32_to_float (fp));
713 } 1113 }
714 1114
715 case 27: 1115 case MISC_FLOAT64:
716 { 1116 {
717 uint64_t fp; 1117 uint64_t fp;
718 WANT (8); 1118 WANT (8);
719 memcpy (&fp, dec->cur, 8); 1119 memcpy (&fp, dec->cur, 8);
720 dec->cur += 8; 1120 dec->cur += 8;
723 fp = ecb_bswap64 (fp); 1123 fp = ecb_bswap64 (fp);
724 1124
725 return newSVnv (ecb_binary64_to_double (fp)); 1125 return newSVnv (ecb_binary64_to_double (fp));
726 } 1126 }
727 1127
728 // 0..19 unassigned 1128 // 0..19 unassigned simple
729 // 24 reserved + unassigned (reserved values are not encodable) 1129 // 24 reserved + unassigned (reserved values are not encodable)
730 default: 1130 default:
731 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)"); 1131 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)");
732 } 1132 }
733 1133
739} 1139}
740 1140
741static SV * 1141static SV *
742decode_cbor (SV *string, CBOR *cbor, char **offset_return) 1142decode_cbor (SV *string, CBOR *cbor, char **offset_return)
743{ 1143{
744 dec_t dec; 1144 dec_t dec = { };
745 SV *sv; 1145 SV *sv;
1146 STRLEN len;
1147 char *data = SvPVbyte (string, len);
746 1148
747 /* work around bugs in 5.10 where manipulating magic values
748 * makes perl ignore the magic in subsequent accesses.
749 * also make a copy of non-PV values, to get them into a clean
750 * state (SvPV should do that, but it's buggy, see below).
751 */
752 /*SvGETMAGIC (string);*/
753 if (SvMAGICAL (string) || !SvPOK (string))
754 string = sv_2mortal (newSVsv (string));
755
756 SvUPGRADE (string, SVt_PV);
757
758 /* work around a bug in perl 5.10, which causes SvCUR to fail an
759 * assertion with -DDEBUGGING, although SvCUR is documented to
760 * return the xpv_cur field which certainly exists after upgrading.
761 * according to nicholas clark, calling SvPOK fixes this.
762 * But it doesn't fix it, so try another workaround, call SvPV_nolen
763 * and hope for the best.
764 * Damnit, SvPV_nolen still trips over yet another assertion. This
765 * assertion business is seriously broken, try yet another workaround
766 * for the broken -DDEBUGGING.
767 */
768 {
769#ifdef DEBUGGING
770 STRLEN offset = SvOK (string) ? sv_len (string) : 0;
771#else
772 STRLEN offset = SvCUR (string);
773#endif
774
775 if (offset > cbor->max_size && cbor->max_size) 1149 if (len > cbor->max_size && cbor->max_size)
776 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu", 1150 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
777 (unsigned long)SvCUR (string), (unsigned long)cbor->max_size); 1151 (unsigned long)len, (unsigned long)cbor->max_size);
778 }
779
780 sv_utf8_downgrade (string, 0);
781 1152
782 dec.cbor = *cbor; 1153 dec.cbor = *cbor;
783 dec.cur = (U8 *)SvPVX (string); 1154 dec.cur = (U8 *)data;
784 dec.end = (U8 *)SvEND (string); 1155 dec.end = (U8 *)data + len;
785 dec.err = 0;
786 dec.depth = 0;
787 1156
788 sv = decode_sv (&dec); 1157 sv = decode_sv (&dec);
789 1158
790 if (offset_return) 1159 if (offset_return)
791 *offset_return = dec.cur; 1160 *offset_return = dec.cur;
795 dec.err = "garbage after CBOR object"; 1164 dec.err = "garbage after CBOR object";
796 1165
797 if (dec.err) 1166 if (dec.err)
798 { 1167 {
799 SvREFCNT_dec (sv); 1168 SvREFCNT_dec (sv);
800 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)SvPVX (string), (int)(uint8_t)*dec.cur); 1169 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur);
801 } 1170 }
802 1171
803 sv = sv_2mortal (sv); 1172 sv = sv_2mortal (sv);
804 1173
805 return sv; 1174 return sv;
811MODULE = CBOR::XS PACKAGE = CBOR::XS 1180MODULE = CBOR::XS PACKAGE = CBOR::XS
812 1181
813BOOT: 1182BOOT:
814{ 1183{
815 cbor_stash = gv_stashpv ("CBOR::XS" , 1); 1184 cbor_stash = gv_stashpv ("CBOR::XS" , 1);
816 cbor_boolean_stash = gv_stashpv ("CBOR::XS::Boolean", 1);
817 cbor_tagged_stash = gv_stashpv ("CBOR::XS::Tagged" , 1); 1185 cbor_tagged_stash = gv_stashpv ("CBOR::XS::Tagged" , 1);
818 1186
819 cbor_true = get_bool ("CBOR::XS::true"); 1187 types_boolean_stash = gv_stashpv ("Types::Serialiser::Boolean", 1);
820 cbor_false = get_bool ("CBOR::XS::false"); 1188 types_error_stash = gv_stashpv ("Types::Serialiser::Error" , 1);
1189
1190 types_true = get_bool ("Types::Serialiser::true" );
1191 types_false = get_bool ("Types::Serialiser::false");
1192 types_error = get_bool ("Types::Serialiser::error");
1193
1194 default_filter = newSVpv ("CBOR::XS::default_filter", 0);
1195
1196 sv_cbor = newSVpv ("CBOR", 0);
1197 SvREADONLY_on (sv_cbor);
821} 1198}
822 1199
823PROTOTYPES: DISABLE 1200PROTOTYPES: DISABLE
824 1201
825void CLONE (...) 1202void CLONE (...)
826 CODE: 1203 CODE:
827 cbor_stash = 0; 1204 cbor_stash = 0;
828 cbor_boolean_stash = 0;
829 cbor_tagged_stash = 0; 1205 cbor_tagged_stash = 0;
1206 types_error_stash = 0;
1207 types_boolean_stash = 0;
830 1208
831void new (char *klass) 1209void new (char *klass)
832 PPCODE: 1210 PPCODE:
833{ 1211{
834 SV *pv = NEWSV (0, sizeof (CBOR)); 1212 SV *pv = NEWSV (0, sizeof (CBOR));
842 1220
843void shrink (CBOR *self, int enable = 1) 1221void shrink (CBOR *self, int enable = 1)
844 ALIAS: 1222 ALIAS:
845 shrink = F_SHRINK 1223 shrink = F_SHRINK
846 allow_unknown = F_ALLOW_UNKNOWN 1224 allow_unknown = F_ALLOW_UNKNOWN
1225 allow_sharing = F_ALLOW_SHARING
1226 allow_cycles = F_ALLOW_CYCLES
1227 pack_strings = F_PACK_STRINGS
847 PPCODE: 1228 PPCODE:
848{ 1229{
849 if (enable) 1230 if (enable)
850 self->flags |= ix; 1231 self->flags |= ix;
851 else 1232 else
856 1237
857void get_shrink (CBOR *self) 1238void get_shrink (CBOR *self)
858 ALIAS: 1239 ALIAS:
859 get_shrink = F_SHRINK 1240 get_shrink = F_SHRINK
860 get_allow_unknown = F_ALLOW_UNKNOWN 1241 get_allow_unknown = F_ALLOW_UNKNOWN
1242 get_allow_sharing = F_ALLOW_SHARING
1243 get_allow_cycles = F_ALLOW_CYCLES
1244 get_pack_strings = F_PACK_STRINGS
861 PPCODE: 1245 PPCODE:
862 XPUSHs (boolSV (self->flags & ix)); 1246 XPUSHs (boolSV (self->flags & ix));
863 1247
864void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) 1248void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
865 PPCODE: 1249 PPCODE:
881 CODE: 1265 CODE:
882 RETVAL = self->max_size; 1266 RETVAL = self->max_size;
883 OUTPUT: 1267 OUTPUT:
884 RETVAL 1268 RETVAL
885 1269
886#if 0 //TODO 1270void filter (CBOR *self, SV *filter = 0)
887
888void filter_cbor_object (CBOR *self, SV *cb = &PL_sv_undef)
889 PPCODE: 1271 PPCODE:
890{
891 SvREFCNT_dec (self->cb_object); 1272 SvREFCNT_dec (self->filter);
892 self->cb_object = SvOK (cb) ? newSVsv (cb) : 0; 1273 self->filter = filter ? newSVsv (filter) : filter;
893
894 XPUSHs (ST (0)); 1274 XPUSHs (ST (0));
895}
896 1275
897void filter_cbor_single_key_object (CBOR *self, SV *key, SV *cb = &PL_sv_undef) 1276SV *get_filter (CBOR *self)
898 PPCODE: 1277 CODE:
899{ 1278 RETVAL = self->filter ? self->filter : NEWSV (0, 0);
900 if (!self->cb_sk_object) 1279 OUTPUT:
901 self->cb_sk_object = newHV (); 1280 RETVAL
902
903 if (SvOK (cb))
904 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);
905 else
906 {
907 hv_delete_ent (self->cb_sk_object, key, G_DISCARD, 0);
908
909 if (!HvKEYS (self->cb_sk_object))
910 {
911 SvREFCNT_dec (self->cb_sk_object);
912 self->cb_sk_object = 0;
913 }
914 }
915
916 XPUSHs (ST (0));
917}
918
919#endif
920 1281
921void encode (CBOR *self, SV *scalar) 1282void encode (CBOR *self, SV *scalar)
922 PPCODE: 1283 PPCODE:
923 PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN; 1284 PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN;
924 XPUSHs (scalar); 1285 XPUSHs (scalar);
937 EXTEND (SP, 2); 1298 EXTEND (SP, 2);
938 PUSHs (sv); 1299 PUSHs (sv);
939 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); 1300 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
940} 1301}
941 1302
942#if 0
943
944void DESTROY (CBOR *self) 1303void DESTROY (CBOR *self)
945 CODE: 1304 PPCODE:
946 SvREFCNT_dec (self->cb_sk_object); 1305 cbor_free (self);
947 SvREFCNT_dec (self->cb_object);
948
949#endif
950 1306
951PROTOTYPES: ENABLE 1307PROTOTYPES: ENABLE
952 1308
953void encode_cbor (SV *scalar) 1309void encode_cbor (SV *scalar)
1310 ALIAS:
1311 encode_cbor = 0
1312 encode_cbor_sharing = F_ALLOW_SHARING
954 PPCODE: 1313 PPCODE:
955{ 1314{
956 CBOR cbor; 1315 CBOR cbor;
957 cbor_init (&cbor); 1316 cbor_init (&cbor);
1317 cbor.flags |= ix;
958 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN; 1318 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;
959 XPUSHs (scalar); 1319 XPUSHs (scalar);
960} 1320}
961 1321
962void decode_cbor (SV *cborstr) 1322void decode_cbor (SV *cborstr)

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines