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 1.38 by root, Sun Dec 1 14:30:52 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
104#define F_VALIDATE_UTF8 0x00000020UL
40 105
41#define INIT_SIZE 32 // initial scalar size to be allocated 106#define INIT_SIZE 32 // initial scalar size to be allocated
42 107
43#define SB do { 108#define SB do {
44#define SE } while (0) 109#define SE } while (0)
55#else 120#else
56# define CBOR_SLOW 0 121# define CBOR_SLOW 0
57# define CBOR_STASH cbor_stash 122# define CBOR_STASH cbor_stash
58#endif 123#endif
59 124
60static HV *cbor_stash, *cbor_boolean_stash, *cbor_tagged_stash; // CBOR::XS:: 125static HV *cbor_stash, *types_boolean_stash, *types_error_stash, *cbor_tagged_stash; // CBOR::XS::
61static SV *cbor_true, *cbor_false; 126static SV *types_true, *types_false, *types_error, *sv_cbor, *default_filter;
62 127
63typedef struct { 128typedef struct {
64 U32 flags; 129 U32 flags;
65 U32 max_depth; 130 U32 max_depth;
66 STRLEN max_size; 131 STRLEN max_size;
132 SV *filter;
67} CBOR; 133} CBOR;
68 134
69ecb_inline void 135ecb_inline void
70cbor_init (CBOR *cbor) 136cbor_init (CBOR *cbor)
71{ 137{
72 Zero (cbor, 1, CBOR); 138 Zero (cbor, 1, CBOR);
73 cbor->max_depth = 512; 139 cbor->max_depth = 512;
140}
141
142ecb_inline void
143cbor_free (CBOR *cbor)
144{
145 SvREFCNT_dec (cbor->filter);
74} 146}
75 147
76///////////////////////////////////////////////////////////////////////////// 148/////////////////////////////////////////////////////////////////////////////
77// utility functions 149// utility functions
78 150
100 SvPV_renew (sv, SvCUR (sv) + 1); 172 SvPV_renew (sv, SvCUR (sv) + 1);
101#endif 173#endif
102 } 174 }
103} 175}
104 176
105///////////////////////////////////////////////////////////////////////////// 177// minimum length of a string to be registered for stringref
106// fp hell 178ecb_inline int
107 179minimum_string_length (UV idx)
108//TODO 180{
181 return idx > 23
182 ? idx > 0xffU
183 ? idx > 0xffffU
184 ? idx > 0xffffffffU
185 ? 11
186 : 7
187 : 5
188 : 4
189 : 3;
190}
109 191
110///////////////////////////////////////////////////////////////////////////// 192/////////////////////////////////////////////////////////////////////////////
111// encoder 193// encoder
112 194
113// structure used for encoding CBOR 195// structure used for encoding CBOR
116 char *cur; // SvPVX (sv) + current output position 198 char *cur; // SvPVX (sv) + current output position
117 char *end; // SvEND (sv) 199 char *end; // SvEND (sv)
118 SV *sv; // result scalar 200 SV *sv; // result scalar
119 CBOR cbor; 201 CBOR cbor;
120 U32 depth; // recursion level 202 U32 depth; // recursion level
203 HV *stringref[2]; // string => index, or 0 ([0] = bytes, [1] = utf-8)
204 UV stringref_idx;
205 HV *shareable; // ptr => index, or 0
206 UV shareable_idx;
121} enc_t; 207} enc_t;
122 208
123ecb_inline void 209ecb_inline void
124need (enc_t *enc, STRLEN len) 210need (enc_t *enc, STRLEN len)
125{ 211{
142static void 228static void
143encode_uint (enc_t *enc, int major, UV len) 229encode_uint (enc_t *enc, int major, UV len)
144{ 230{
145 need (enc, 9); 231 need (enc, 9);
146 232
147 if (len < 24) 233 if (ecb_expect_true (len < LENGTH_EXT1))
148 *enc->cur++ = major | len; 234 *enc->cur++ = major | len;
149 else if (len <= 0xff) 235 else if (ecb_expect_true (len <= 0xffU))
150 { 236 {
151 *enc->cur++ = major | 24; 237 *enc->cur++ = major | LENGTH_EXT1;
152 *enc->cur++ = len; 238 *enc->cur++ = len;
153 } 239 }
154 else if (len <= 0xffff) 240 else if (len <= 0xffffU)
155 { 241 {
156 *enc->cur++ = major | 25; 242 *enc->cur++ = major | LENGTH_EXT2;
157 *enc->cur++ = len >> 8; 243 *enc->cur++ = len >> 8;
158 *enc->cur++ = len; 244 *enc->cur++ = len;
159 } 245 }
160 else if (len <= 0xffffffff) 246 else if (len <= 0xffffffffU)
161 { 247 {
162 *enc->cur++ = major | 26; 248 *enc->cur++ = major | LENGTH_EXT4;
163 *enc->cur++ = len >> 24; 249 *enc->cur++ = len >> 24;
164 *enc->cur++ = len >> 16; 250 *enc->cur++ = len >> 16;
165 *enc->cur++ = len >> 8; 251 *enc->cur++ = len >> 8;
166 *enc->cur++ = len; 252 *enc->cur++ = len;
167 } 253 }
168 else 254 else
169 { 255 {
170 *enc->cur++ = major | 27; 256 *enc->cur++ = major | LENGTH_EXT8;
171 *enc->cur++ = len >> 56; 257 *enc->cur++ = len >> 56;
172 *enc->cur++ = len >> 48; 258 *enc->cur++ = len >> 48;
173 *enc->cur++ = len >> 40; 259 *enc->cur++ = len >> 40;
174 *enc->cur++ = len >> 32; 260 *enc->cur++ = len >> 32;
175 *enc->cur++ = len >> 24; 261 *enc->cur++ = len >> 24;
177 *enc->cur++ = len >> 8; 263 *enc->cur++ = len >> 8;
178 *enc->cur++ = len; 264 *enc->cur++ = len;
179 } 265 }
180} 266}
181 267
182static void 268ecb_inline void
269encode_tag (enc_t *enc, UV tag)
270{
271 encode_uint (enc, MAJOR_TAG, tag);
272}
273
274ecb_inline void
183encode_str (enc_t *enc, int utf8, char *str, STRLEN len) 275encode_str (enc_t *enc, int utf8, char *str, STRLEN len)
184{ 276{
185 encode_uint (enc, utf8 ? 0x60 : 0x40, len); 277 encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len);
186 need (enc, len); 278 need (enc, len);
187 memcpy (enc->cur, str, len); 279 memcpy (enc->cur, str, len);
188 enc->cur += len; 280 enc->cur += len;
189} 281}
190 282
283static void
284encode_strref (enc_t *enc, int utf8, char *str, STRLEN len)
285{
286 if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS))
287 {
288 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1);
289
290 if (SvOK (*svp))
291 {
292 // already registered, use stringref
293 encode_tag (enc, CBOR_TAG_STRINGREF);
294 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
295 return;
296 }
297 else if (len >= minimum_string_length (enc->stringref_idx))
298 {
299 // register only
300 sv_setuv (*svp, enc->stringref_idx);
301 ++enc->stringref_idx;
302 }
303 }
304
305 encode_str (enc, utf8, str, len);
306}
307
191static void encode_sv (enc_t *enc, SV *sv); 308static void encode_sv (enc_t *enc, SV *sv);
192 309
193static void 310static void
194encode_av (enc_t *enc, AV *av) 311encode_av (enc_t *enc, AV *av)
195{ 312{
198 if (enc->depth >= enc->cbor.max_depth) 315 if (enc->depth >= enc->cbor.max_depth)
199 croak (ERR_NESTING_EXCEEDED); 316 croak (ERR_NESTING_EXCEEDED);
200 317
201 ++enc->depth; 318 ++enc->depth;
202 319
203 encode_uint (enc, 0x80, len + 1); 320 encode_uint (enc, MAJOR_ARRAY, len + 1);
204 321
205 for (i = 0; i <= len; ++i) 322 for (i = 0; i <= len; ++i)
206 { 323 {
207 SV **svp = av_fetch (av, i, 0); 324 SV **svp = av_fetch (av, i, 0);
208 encode_sv (enc, svp ? *svp : &PL_sv_undef); 325 encode_sv (enc, svp ? *svp : &PL_sv_undef);
223 340
224 int pairs = hv_iterinit (hv); 341 int pairs = hv_iterinit (hv);
225 int mg = SvMAGICAL (hv); 342 int mg = SvMAGICAL (hv);
226 343
227 if (mg) 344 if (mg)
228 encode_ch (enc, 0xa0 | 31); 345 encode_ch (enc, MAJOR_MAP | MINOR_INDEF);
229 else 346 else
230 encode_uint (enc, 0xa0, pairs); 347 encode_uint (enc, MAJOR_MAP, pairs);
231 348
232 while ((he = hv_iternext (hv))) 349 while ((he = hv_iternext (hv)))
233 { 350 {
234 if (HeKLEN (he) == HEf_SVKEY) 351 if (HeKLEN (he) == HEf_SVKEY)
235 encode_sv (enc, HeSVKEY (he)); 352 encode_sv (enc, HeSVKEY (he));
236 else 353 else
237 encode_str (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he)); 354 encode_strref (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he));
238 355
239 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he)); 356 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he));
240 } 357 }
241 358
242 if (mg) 359 if (mg)
243 encode_ch (enc, 0xe0 | 31); 360 encode_ch (enc, MAJOR_MISC | MINOR_INDEF);
244 361
245 --enc->depth; 362 --enc->depth;
246} 363}
247 364
248// encode objects, arrays and special \0=false and \1=true values. 365// encode objects, arrays and special \0=false and \1=true values.
249static void 366static void
250encode_rv (enc_t *enc, SV *sv) 367encode_rv (enc_t *enc, SV *sv)
251{ 368{
252 svtype svt;
253
254 SvGETMAGIC (sv); 369 SvGETMAGIC (sv);
370
255 svt = SvTYPE (sv); 371 svtype svt = SvTYPE (sv);
256 372
257 if (ecb_expect_false (SvOBJECT (sv))) 373 if (ecb_expect_false (SvOBJECT (sv)))
258 { 374 {
259 HV *boolean_stash = !CBOR_SLOW || cbor_boolean_stash 375 HV *boolean_stash = !CBOR_SLOW || types_boolean_stash
260 ? cbor_boolean_stash 376 ? types_boolean_stash
261 : gv_stashpv ("CBOR::XS::Boolean", 1); 377 : gv_stashpv ("Types::Serialiser::Boolean", 1);
378 HV *error_stash = !CBOR_SLOW || types_error_stash
379 ? types_error_stash
380 : gv_stashpv ("Types::Serialiser::Error", 1);
262 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 381 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
263 ? cbor_tagged_stash 382 ? cbor_tagged_stash
264 : gv_stashpv ("CBOR::XS::Tagged" , 1); 383 : gv_stashpv ("CBOR::XS::Tagged" , 1);
265 384
385 HV *stash = SvSTASH (sv);
386
266 if (SvSTASH (sv) == boolean_stash) 387 if (stash == boolean_stash)
267 encode_ch (enc, SvIV (sv) ? 0xe0 | 21 : 0xe0 | 20); 388 {
389 encode_ch (enc, SvIV (sv) ? MAJOR_MISC | SIMPLE_TRUE : MAJOR_MISC | SIMPLE_FALSE);
390 return;
391 }
392 else if (stash == error_stash)
393 {
394 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
395 return;
396 }
268 else if (SvSTASH (sv) == tagged_stash) 397 else if (stash == tagged_stash)
269 { 398 {
270 if (svt != SVt_PVAV) 399 if (svt != SVt_PVAV)
271 croak ("encountered CBOR::XS::Tagged object that isn't an array"); 400 croak ("encountered CBOR::XS::Tagged object that isn't an array");
272 401
273 encode_uint (enc, 0xc0, SvUV (*av_fetch ((AV *)sv, 0, 1))); 402 encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1)));
274 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1)); 403 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1));
404
405 return;
406 }
407 }
408
409 if (ecb_expect_false (SvREFCNT (sv) > 1)
410 && ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING))
411 {
412 if (!enc->shareable)
413 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
414
415 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
416
417 if (SvOK (*svp))
418 {
419 encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF);
420 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
421 return;
275 } 422 }
276 else 423 else
277 { 424 {
425 sv_setuv (*svp, enc->shareable_idx);
426 ++enc->shareable_idx;
427 encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE);
428 }
429 }
430
431 if (ecb_expect_false (SvOBJECT (sv)))
432 {
433 HV *stash = SvSTASH (sv);
434 GV *method;
435
436 if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
437 {
438 dSP;
439
440 ENTER; SAVETMPS; PUSHMARK (SP);
278 // we re-bless the reference to get overload and other niceties right 441 // we re-bless the reference to get overload and other niceties right
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))); 442 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
287 443
288 // calling with G_SCALAR ensures that we always get a 1 return value
289 PUTBACK; 444 PUTBACK;
445 // G_SCALAR ensures that return value is 1
290 call_sv ((SV *)GvCV (to_cbor), G_SCALAR); 446 call_sv ((SV *)GvCV (method), G_SCALAR);
291 SPAGAIN; 447 SPAGAIN;
292 448
293 // catch this surprisingly common error 449 // catch this surprisingly common error
294 if (SvROK (TOPs) && SvRV (TOPs) == sv) 450 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))); 451 croak ("%s::TO_CBOR method returned same object as was passed instead of a new one", HvNAME (stash));
296 452
297 sv = POPs; 453 encode_sv (enc, POPs);
454
298 PUTBACK; 455 PUTBACK;
299 456
300 encode_sv (enc, sv);
301
302 FREETMPS; LEAVE; 457 FREETMPS; LEAVE;
303 } 458 }
459 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)
460 {
461 dSP;
462
463 ENTER; SAVETMPS; PUSHMARK (SP);
464 EXTEND (SP, 2);
465 // we re-bless the reference to get overload and other niceties right
466 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
467 PUSHs (sv_cbor);
468
469 PUTBACK;
470 int count = call_sv ((SV *)GvCV (method), G_ARRAY);
471 SPAGAIN;
472
473 // catch this surprisingly common error
474 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)
475 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));
476
477 encode_tag (enc, CBOR_TAG_PERL_OBJECT);
478 encode_uint (enc, MAJOR_ARRAY, count + 1);
479 encode_strref (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
480
481 while (count)
482 encode_sv (enc, SP[1 - count--]);
483
484 PUTBACK;
485
486 FREETMPS; LEAVE;
487 }
304 else 488 else
305 croak ("encountered object '%s', but no TO_CBOR method available on it", 489 croak ("encountered object '%s', but no TO_CBOR or FREEZE methods available on it",
306 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 490 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
307 }
308 } 491 }
309 else if (svt == SVt_PVHV) 492 else if (svt == SVt_PVHV)
310 encode_hv (enc, (HV *)sv); 493 encode_hv (enc, (HV *)sv);
311 else if (svt == SVt_PVAV) 494 else if (svt == SVt_PVAV)
312 encode_av (enc, (AV *)sv); 495 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 496 else
331 croak ("encountered %s, but CBOR can only represent references to arrays or hashes", 497 {
332 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 498 encode_tag (enc, CBOR_TAG_INDIRECTION);
499 encode_sv (enc, sv);
500 }
333} 501}
334 502
335static void 503static void
336encode_nv (enc_t *enc, SV *sv) 504encode_nv (enc_t *enc, SV *sv)
337{ 505{
338 double nv = SvNVX (sv); 506 double nv = SvNVX (sv);
339 507
340 need (enc, 9); 508 need (enc, 9);
341 509
342 if (ecb_expect_false (nv == (U32)nv)) 510 if (ecb_expect_false (nv == (NV)(U32)nv))
343 encode_uint (enc, 0x00, (U32)nv); 511 encode_uint (enc, MAJOR_POS_INT, (U32)nv);
344 //TODO: maybe I32? 512 //TODO: maybe I32?
345 else if (ecb_expect_false (nv == (float)nv)) 513 else if (ecb_expect_false (nv == (float)nv))
346 { 514 {
347 uint32_t fp = ecb_float_to_binary32 (nv); 515 uint32_t fp = ecb_float_to_binary32 (nv);
348 516
349 *enc->cur++ = 0xe0 | 26; 517 *enc->cur++ = MAJOR_MISC | MISC_FLOAT32;
350 518
351 if (!ecb_big_endian ()) 519 if (!ecb_big_endian ())
352 fp = ecb_bswap32 (fp); 520 fp = ecb_bswap32 (fp);
353 521
354 memcpy (enc->cur, &fp, 4); 522 memcpy (enc->cur, &fp, 4);
356 } 524 }
357 else 525 else
358 { 526 {
359 uint64_t fp = ecb_double_to_binary64 (nv); 527 uint64_t fp = ecb_double_to_binary64 (nv);
360 528
361 *enc->cur++ = 0xe0 | 27; 529 *enc->cur++ = MAJOR_MISC | MISC_FLOAT64;
362 530
363 if (!ecb_big_endian ()) 531 if (!ecb_big_endian ())
364 fp = ecb_bswap64 (fp); 532 fp = ecb_bswap64 (fp);
365 533
366 memcpy (enc->cur, &fp, 8); 534 memcpy (enc->cur, &fp, 8);
375 543
376 if (SvPOKp (sv)) 544 if (SvPOKp (sv))
377 { 545 {
378 STRLEN len; 546 STRLEN len;
379 char *str = SvPV (sv, len); 547 char *str = SvPV (sv, len);
380 encode_str (enc, SvUTF8 (sv), str, len); 548 encode_strref (enc, SvUTF8 (sv), str, len);
381 } 549 }
382 else if (SvNOKp (sv)) 550 else if (SvNOKp (sv))
383 encode_nv (enc, sv); 551 encode_nv (enc, sv);
384 else if (SvIOKp (sv)) 552 else if (SvIOKp (sv))
385 { 553 {
386 if (SvIsUV (sv)) 554 if (SvIsUV (sv))
387 encode_uint (enc, 0x00, SvUVX (sv)); 555 encode_uint (enc, MAJOR_POS_INT, SvUVX (sv));
388 else if (SvIVX (sv) >= 0) 556 else if (SvIVX (sv) >= 0)
389 encode_uint (enc, 0x00, SvIVX (sv)); 557 encode_uint (enc, MAJOR_POS_INT, SvIVX (sv));
390 else 558 else
391 encode_uint (enc, 0x20, -(SvIVX (sv) + 1)); 559 encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1));
392 } 560 }
393 else if (SvROK (sv)) 561 else if (SvROK (sv))
394 encode_rv (enc, SvRV (sv)); 562 encode_rv (enc, SvRV (sv));
395 else if (!SvOK (sv)) 563 else if (!SvOK (sv))
396 encode_ch (enc, 0xe0 | 22); 564 encode_ch (enc, MAJOR_MISC | SIMPLE_NULL);
397 else if (enc->cbor.flags & F_ALLOW_UNKNOWN) 565 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
398 encode_ch (enc, 0xe0 | 23); 566 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
399 else 567 else
400 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data", 568 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data",
401 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv)); 569 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
402} 570}
403 571
404static SV * 572static SV *
405encode_cbor (SV *scalar, CBOR *cbor) 573encode_cbor (SV *scalar, CBOR *cbor)
406{ 574{
407 enc_t enc; 575 enc_t enc = { };
408 576
409 enc.cbor = *cbor; 577 enc.cbor = *cbor;
410 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 578 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
411 enc.cur = SvPVX (enc.sv); 579 enc.cur = SvPVX (enc.sv);
412 enc.end = SvEND (enc.sv); 580 enc.end = SvEND (enc.sv);
413 enc.depth = 0;
414 581
415 SvPOK_only (enc.sv); 582 SvPOK_only (enc.sv);
583
584 if (cbor->flags & F_PACK_STRINGS)
585 {
586 encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE);
587 enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ());
588 enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ());
589 }
590
416 encode_sv (&enc, scalar); 591 encode_sv (&enc, scalar);
417 592
418 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 593 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
419 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings 594 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
420 595
434 U8 *end; // end of input string 609 U8 *end; // end of input string
435 const char *err; // parse error, if != 0 610 const char *err; // parse error, if != 0
436 CBOR cbor; 611 CBOR cbor;
437 U32 depth; // recursion depth 612 U32 depth; // recursion depth
438 U32 maxdepth; // recursion depth limit 613 U32 maxdepth; // recursion depth limit
614 AV *shareable;
615 AV *stringref;
616 SV *decode_tagged;
439} dec_t; 617} dec_t;
440 618
441#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE 619#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE
442 620
443#define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data") 621#define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data")
446#define DEC_DEC_DEPTH --dec->depth 624#define DEC_DEC_DEPTH --dec->depth
447 625
448static UV 626static UV
449decode_uint (dec_t *dec) 627decode_uint (dec_t *dec)
450{ 628{
451 switch (*dec->cur & 31) 629 U8 m = *dec->cur & MINOR_MASK;
452 { 630 ++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;
457 631
458 case 24: 632 if (ecb_expect_true (m < LENGTH_EXT1))
633 return m;
634 else if (ecb_expect_true (m == LENGTH_EXT1))
635 {
459 WANT (2); 636 WANT (1);
460 dec->cur += 2; 637 dec->cur += 1;
461 return dec->cur[-1]; 638 return dec->cur[-1];
462 639 }
463 case 25: 640 else if (ecb_expect_true (m == LENGTH_EXT2))
641 {
464 WANT (3); 642 WANT (2);
465 dec->cur += 3; 643 dec->cur += 2;
466 return (((UV)dec->cur[-2]) << 8) 644 return (((UV)dec->cur[-2]) << 8)
467 | ((UV)dec->cur[-1]); 645 | ((UV)dec->cur[-1]);
468 646 }
469 case 26: 647 else if (ecb_expect_true (m == LENGTH_EXT4))
648 {
470 WANT (5); 649 WANT (4);
471 dec->cur += 5; 650 dec->cur += 4;
472 return (((UV)dec->cur[-4]) << 24) 651 return (((UV)dec->cur[-4]) << 24)
473 | (((UV)dec->cur[-3]) << 16) 652 | (((UV)dec->cur[-3]) << 16)
474 | (((UV)dec->cur[-2]) << 8) 653 | (((UV)dec->cur[-2]) << 8)
475 | ((UV)dec->cur[-1]); 654 | ((UV)dec->cur[-1]);
476 655 }
477 case 27: 656 else if (ecb_expect_true (m == LENGTH_EXT8))
657 {
478 WANT (9); 658 WANT (8);
479 dec->cur += 9; 659 dec->cur += 8;
660
661 return
662#if UVSIZE < 8
663 0
664#else
480 return (((UV)dec->cur[-8]) << 56) 665 (((UV)dec->cur[-8]) << 56)
481 | (((UV)dec->cur[-7]) << 48) 666 | (((UV)dec->cur[-7]) << 48)
482 | (((UV)dec->cur[-6]) << 40) 667 | (((UV)dec->cur[-6]) << 40)
483 | (((UV)dec->cur[-5]) << 32) 668 | (((UV)dec->cur[-5]) << 32)
669#endif
484 | (((UV)dec->cur[-4]) << 24) 670 | (((UV)dec->cur[-4]) << 24)
485 | (((UV)dec->cur[-3]) << 16) 671 | (((UV)dec->cur[-3]) << 16)
486 | (((UV)dec->cur[-2]) << 8) 672 | (((UV)dec->cur[-2]) << 8)
487 | ((UV)dec->cur[-1]); 673 | ((UV)dec->cur[-1]);
488 674 }
489 default: 675 else
490 ERR ("corrupted CBOR data (unsupported integer minor encoding)"); 676 ERR ("corrupted CBOR data (unsupported integer minor encoding)");
491 }
492 677
493fail: 678fail:
494 return 0; 679 return 0;
495} 680}
496 681
501{ 686{
502 AV *av = newAV (); 687 AV *av = newAV ();
503 688
504 DEC_INC_DEPTH; 689 DEC_INC_DEPTH;
505 690
506 if ((*dec->cur & 31) == 31) 691 if (*dec->cur == (MAJOR_ARRAY | MINOR_INDEF))
507 { 692 {
508 ++dec->cur; 693 ++dec->cur;
509 694
510 for (;;) 695 for (;;)
511 { 696 {
512 WANT (1); 697 WANT (1);
513 698
514 if (*dec->cur == (0xe0 | 31)) 699 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
515 { 700 {
516 ++dec->cur; 701 ++dec->cur;
517 break; 702 break;
518 } 703 }
519 704
522 } 707 }
523 else 708 else
524 { 709 {
525 int i, len = decode_uint (dec); 710 int i, len = decode_uint (dec);
526 711
712 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); 713 av_fill (av, len - 1);
528 714
529 for (i = 0; i < len; ++i) 715 for (i = 0; i < len; ++i)
530 AvARRAY (av)[i] = decode_sv (dec); 716 AvARRAY (av)[i] = decode_sv (dec);
531 } 717 }
537 SvREFCNT_dec (av); 723 SvREFCNT_dec (av);
538 DEC_DEC_DEPTH; 724 DEC_DEC_DEPTH;
539 return &PL_sv_undef; 725 return &PL_sv_undef;
540} 726}
541 727
728static void
729decode_he (dec_t *dec, HV *hv)
730{
731 // for speed reasons, we specialcase single-string
732 // byte or utf-8 strings as keys, but only when !stringref
733
734 if (ecb_expect_true (!dec->stringref))
735 if (ecb_expect_true ((*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8))
736 {
737 I32 len = decode_uint (dec);
738 char *key = (char *)dec->cur;
739
740 dec->cur += len;
741
742 hv_store (hv, key, len, decode_sv (dec), 0);
743
744 return;
745 }
746 else if (ecb_expect_true ((*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8))
747 {
748 I32 len = decode_uint (dec);
749 char *key = (char *)dec->cur;
750
751 dec->cur += len;
752
753 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
754 if (!is_utf8_string (key, len))
755 ERR ("corrupted CBOR data (invalid UTF-8 in map key)");
756
757 hv_store (hv, key, -len, decode_sv (dec), 0);
758
759 return;
760 }
761
762 SV *k = decode_sv (dec);
763 SV *v = decode_sv (dec);
764
765 hv_store_ent (hv, k, v, 0);
766 SvREFCNT_dec (k);
767
768fail:
769 ;
770}
771
542static SV * 772static SV *
543decode_hv (dec_t *dec) 773decode_hv (dec_t *dec)
544{ 774{
545 HV *hv = newHV (); 775 HV *hv = newHV ();
546 776
547 DEC_INC_DEPTH; 777 DEC_INC_DEPTH;
548 778
549 if ((*dec->cur & 31) == 31) 779 if (*dec->cur == (MAJOR_MAP | MINOR_INDEF))
550 { 780 {
551 ++dec->cur; 781 ++dec->cur;
552 782
553 for (;;) 783 for (;;)
554 { 784 {
555 WANT (1); 785 WANT (1);
556 786
557 if (*dec->cur == (0xe0 | 31)) 787 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
558 { 788 {
559 ++dec->cur; 789 ++dec->cur;
560 break; 790 break;
561 } 791 }
562 792
563 SV *k = decode_sv (dec); 793 decode_he (dec, hv);
564 SV *v = decode_sv (dec);
565
566 hv_store_ent (hv, k, v, 0);
567 } 794 }
568 } 795 }
569 else 796 else
570 { 797 {
571 int len = decode_uint (dec); 798 int pairs = decode_uint (dec);
572 799
573 while (len--) 800 while (pairs--)
574 { 801 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 } 802 }
581 803
582 DEC_DEC_DEPTH; 804 DEC_DEC_DEPTH;
583 return newRV_noinc ((SV *)hv); 805 return newRV_noinc ((SV *)hv);
584 806
591static SV * 813static SV *
592decode_str (dec_t *dec, int utf8) 814decode_str (dec_t *dec, int utf8)
593{ 815{
594 SV *sv = 0; 816 SV *sv = 0;
595 817
596 if ((*dec->cur & 31) == 31) 818 if ((*dec->cur & MINOR_MASK) == MINOR_INDEF)
597 { 819 {
820 // indefinite length strings
598 ++dec->cur; 821 ++dec->cur;
599 822
823 U8 major = *dec->cur & MAJOR_MISC;
824
600 sv = newSVpvn ("", 0); 825 sv = newSVpvn ("", 0);
601 826
602 // not very fast, and certainly not robust against illegal input
603 for (;;) 827 for (;;)
604 { 828 {
605 WANT (1); 829 WANT (1);
606 830
607 if (*dec->cur == (0xe0 | 31)) 831 if ((*dec->cur - major) > LENGTH_EXT8)
832 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
608 { 833 {
609 ++dec->cur; 834 ++dec->cur;
610 break; 835 break;
611 } 836 }
837 else
838 ERR ("corrupted CBOR data (invalid chunks in indefinite length string)");
612 839
613 sv_catsv (sv, decode_sv (dec)); 840 STRLEN len = decode_uint (dec);
841
842 WANT (len);
843 sv_catpvn (sv, dec->cur, len);
844 dec->cur += len;
614 } 845 }
615 } 846 }
616 else 847 else
617 { 848 {
618 STRLEN len = decode_uint (dec); 849 STRLEN len = decode_uint (dec);
619 850
620 WANT (len); 851 WANT (len);
621 sv = newSVpvn (dec->cur, len); 852 sv = newSVpvn (dec->cur, len);
622 dec->cur += len; 853 dec->cur += len;
854
855 if (ecb_expect_false (dec->stringref)
856 && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1))
857 av_push (dec->stringref, SvREFCNT_inc_NN (sv));
623 } 858 }
624 859
625 if (utf8) 860 if (utf8)
861 {
862 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
863 if (!is_utf8_string (SvPVX (sv), SvCUR (sv)))
864 ERR ("corrupted CBOR data (invalid UTF-8 in text string)");
865
626 SvUTF8_on (sv); 866 SvUTF8_on (sv);
867 }
627 868
628 return sv; 869 return sv;
629 870
630fail: 871fail:
631 SvREFCNT_dec (sv); 872 SvREFCNT_dec (sv);
633} 874}
634 875
635static SV * 876static SV *
636decode_tagged (dec_t *dec) 877decode_tagged (dec_t *dec)
637{ 878{
879 SV *sv = 0;
638 UV tag = decode_uint (dec); 880 UV tag = decode_uint (dec);
881
882 WANT (1);
883
884 switch (tag)
885 {
886 case CBOR_TAG_MAGIC:
639 SV *sv = decode_sv (dec); 887 sv = decode_sv (dec);
888 break;
640 889
641 if (tag == CBOR_TAG_MAGIC) // 2.4.5 Self-Describe CBOR 890 case CBOR_TAG_INDIRECTION:
891 sv = newRV_noinc (decode_sv (dec));
892 break;
893
894 case CBOR_TAG_STRINGREF_NAMESPACE:
895 {
896 ENTER; SAVETMPS;
897
898 SAVESPTR (dec->stringref);
899 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ());
900
901 sv = decode_sv (dec);
902
903 FREETMPS; LEAVE;
904 }
905 break;
906
907 case CBOR_TAG_STRINGREF:
908 {
909 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
910 ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
911
912 UV idx = decode_uint (dec);
913
914 if (!dec->stringref || (int)idx > AvFILLp (dec->stringref))
915 ERR ("corrupted CBOR data (stringref index out of bounds or outside namespace)");
916
917 sv = newSVsv (AvARRAY (dec->stringref)[idx]);
918 }
919 break;
920
921 case CBOR_TAG_VALUE_SHAREABLE:
922 {
923 if (ecb_expect_false (!dec->shareable))
924 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ());
925
926 if (dec->cbor.flags & F_ALLOW_CYCLES)
927 {
928 sv = newSV (0);
929 av_push (dec->shareable, SvREFCNT_inc_NN (sv));
930
931 SV *osv = decode_sv (dec);
932 sv_setsv (sv, osv);
933 SvREFCNT_dec_NN (osv);
934 }
935 else
936 {
937 av_push (dec->shareable, &PL_sv_undef);
938 int idx = AvFILLp (dec->shareable);
939 sv = decode_sv (dec);
940 av_store (dec->shareable, idx, SvREFCNT_inc_NN (sv));
941 }
942 }
943 break;
944
945 case CBOR_TAG_VALUE_SHAREDREF:
946 {
947 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
948 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
949
950 UV idx = decode_uint (dec);
951
952 if (!dec->shareable || (int)idx > AvFILLp (dec->shareable))
953 ERR ("corrupted CBOR data (sharedref index out of bounds)");
954
955 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]);
956
957 if (sv == &PL_sv_undef)
958 ERR ("cyclic CBOR data structure found, but allow_cycles is not enabled");
959 }
960 break;
961
962 case CBOR_TAG_PERL_OBJECT:
963 {
964 sv = decode_sv (dec);
965
966 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
967 ERR ("corrupted CBOR data (non-array perl object)");
968
969 AV *av = (AV *)SvRV (sv);
970 int len = av_len (av) + 1;
971 HV *stash = gv_stashsv (*av_fetch (av, 0, 1), 0);
972
973 if (!stash)
974 ERR ("cannot decode perl-object (package does not exist)");
975
976 GV *method = gv_fetchmethod_autoload (stash, "THAW", 0);
977
978 if (!method)
979 ERR ("cannot decode perl-object (package does not have a THAW method)");
980
981 dSP;
982
983 ENTER; SAVETMPS; PUSHMARK (SP);
984 EXTEND (SP, len + 1);
985 // we re-bless the reference to get overload and other niceties right
986 PUSHs (*av_fetch (av, 0, 1));
987 PUSHs (sv_cbor);
988
989 int i;
990
991 for (i = 1; i < len; ++i)
992 PUSHs (*av_fetch (av, i, 1));
993
994 PUTBACK;
995 call_sv ((SV *)GvCV (method), G_SCALAR | G_EVAL);
996 SPAGAIN;
997
998 if (SvTRUE (ERRSV))
999 {
1000 FREETMPS; LEAVE;
1001 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV))));
1002 }
1003
1004 SvREFCNT_dec (sv);
1005 sv = SvREFCNT_inc (POPs);
1006
1007 PUTBACK;
1008
1009 FREETMPS; LEAVE;
1010 }
1011 break;
1012
1013 default:
1014 {
1015 sv = decode_sv (dec);
1016
1017 dSP;
1018 ENTER; SAVETMPS; PUSHMARK (SP);
1019 EXTEND (SP, 2);
1020 PUSHs (newSVuv (tag));
1021 PUSHs (sv);
1022
1023 PUTBACK;
1024 int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL);
1025 SPAGAIN;
1026
1027 if (SvTRUE (ERRSV))
1028 {
1029 FREETMPS; LEAVE;
1030 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV))));
1031 }
1032
1033 if (count)
1034 {
1035 SvREFCNT_dec (sv);
1036 sv = SvREFCNT_inc (POPs);
1037 }
1038 else
1039 {
1040 AV *av = newAV ();
1041 av_push (av, newSVuv (tag));
1042 av_push (av, sv);
1043
1044 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
1045 ? cbor_tagged_stash
1046 : gv_stashpv ("CBOR::XS::Tagged" , 1);
1047 sv = sv_bless (newRV_noinc ((SV *)av), tagged_stash);
1048 }
1049
1050 PUTBACK;
1051
1052 FREETMPS; LEAVE;
1053 }
1054 break;
1055 }
1056
642 return sv; 1057 return sv;
643 1058
644 AV *av = newAV (); 1059fail:
645 av_push (av, newSVuv (tag)); 1060 SvREFCNT_dec (sv);
646 av_push (av, sv); 1061 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} 1062}
654 1063
655static SV * 1064static SV *
656decode_sv (dec_t *dec) 1065decode_sv (dec_t *dec)
657{ 1066{
658 WANT (1); 1067 WANT (1);
659 1068
660 switch (*dec->cur >> 5) 1069 switch (*dec->cur >> MAJOR_SHIFT)
661 { 1070 {
662 case 0: // unsigned int 1071 case MAJOR_POS_INT >> MAJOR_SHIFT: return newSVuv (decode_uint (dec));
663 return newSVuv (decode_uint (dec)); 1072 case MAJOR_NEG_INT >> MAJOR_SHIFT: return newSViv (-1 - (IV)decode_uint (dec));
664 case 1: // negative int 1073 case MAJOR_BYTES >> MAJOR_SHIFT: return decode_str (dec, 0);
665 return newSViv (-1 - (IV)decode_uint (dec)); 1074 case MAJOR_TEXT >> MAJOR_SHIFT: return decode_str (dec, 1);
666 case 2: // octet string 1075 case MAJOR_ARRAY >> MAJOR_SHIFT: return decode_av (dec);
667 return decode_str (dec, 0); 1076 case MAJOR_MAP >> MAJOR_SHIFT: return decode_hv (dec);
668 case 3: // utf-8 string 1077 case MAJOR_TAG >> MAJOR_SHIFT: return decode_tagged (dec);
669 return decode_str (dec, 1); 1078
670 case 4: // array 1079 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) 1080 switch (*dec->cur++ & MINOR_MASK)
678 { 1081 {
679 case 20: 1082 case SIMPLE_FALSE:
680#if CBOR_SLOW 1083#if CBOR_SLOW
681 cbor_false = get_bool ("CBOR::XS::false"); 1084 types_false = get_bool ("Types::Serialiser::false");
682#endif 1085#endif
683 return newSVsv (cbor_false); 1086 return newSVsv (types_false);
684 case 21: 1087 case SIMPLE_TRUE:
685#if CBOR_SLOW 1088#if CBOR_SLOW
686 cbor_true = get_bool ("CBOR::XS::true"); 1089 types_true = get_bool ("Types::Serialiser::true");
687#endif 1090#endif
688 return newSVsv (cbor_true); 1091 return newSVsv (types_true);
689 case 22: 1092 case SIMPLE_NULL:
690 return newSVsv (&PL_sv_undef); 1093 return newSVsv (&PL_sv_undef);
1094 case SIMPLE_UNDEF:
1095#if CBOR_SLOW
1096 types_error = get_bool ("Types::Serialiser::error");
1097#endif
1098 return newSVsv (types_error);
691 1099
692 case 25: 1100 case MISC_FLOAT16:
693 { 1101 {
694 WANT (2); 1102 WANT (2);
695 1103
696 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1]; 1104 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1];
697 dec->cur += 2; 1105 dec->cur += 2;
698 1106
699 return newSVnv (ecb_binary16_to_float (fp)); 1107 return newSVnv (ecb_binary16_to_float (fp));
700 } 1108 }
701 1109
702 case 26: 1110 case MISC_FLOAT32:
703 { 1111 {
704 uint32_t fp; 1112 uint32_t fp;
705 WANT (4); 1113 WANT (4);
706 memcpy (&fp, dec->cur, 4); 1114 memcpy (&fp, dec->cur, 4);
707 dec->cur += 4; 1115 dec->cur += 4;
710 fp = ecb_bswap32 (fp); 1118 fp = ecb_bswap32 (fp);
711 1119
712 return newSVnv (ecb_binary32_to_float (fp)); 1120 return newSVnv (ecb_binary32_to_float (fp));
713 } 1121 }
714 1122
715 case 27: 1123 case MISC_FLOAT64:
716 { 1124 {
717 uint64_t fp; 1125 uint64_t fp;
718 WANT (8); 1126 WANT (8);
719 memcpy (&fp, dec->cur, 8); 1127 memcpy (&fp, dec->cur, 8);
720 dec->cur += 8; 1128 dec->cur += 8;
723 fp = ecb_bswap64 (fp); 1131 fp = ecb_bswap64 (fp);
724 1132
725 return newSVnv (ecb_binary64_to_double (fp)); 1133 return newSVnv (ecb_binary64_to_double (fp));
726 } 1134 }
727 1135
728 // 0..19 unassigned 1136 // 0..19 unassigned simple
729 // 24 reserved + unassigned (reserved values are not encodable) 1137 // 24 reserved + unassigned (reserved values are not encodable)
730 default: 1138 default:
731 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)"); 1139 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)");
732 } 1140 }
733 1141
739} 1147}
740 1148
741static SV * 1149static SV *
742decode_cbor (SV *string, CBOR *cbor, char **offset_return) 1150decode_cbor (SV *string, CBOR *cbor, char **offset_return)
743{ 1151{
744 dec_t dec; 1152 dec_t dec = { };
745 SV *sv; 1153 SV *sv;
1154 STRLEN len;
1155 char *data = SvPVbyte (string, len);
746 1156
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) 1157 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", 1158 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); 1159 (unsigned long)len, (unsigned long)cbor->max_size);
778 }
779
780 sv_utf8_downgrade (string, 0);
781 1160
782 dec.cbor = *cbor; 1161 dec.cbor = *cbor;
783 dec.cur = (U8 *)SvPVX (string); 1162 dec.cur = (U8 *)data;
784 dec.end = (U8 *)SvEND (string); 1163 dec.end = (U8 *)data + len;
785 dec.err = 0;
786 dec.depth = 0;
787 1164
788 sv = decode_sv (&dec); 1165 sv = decode_sv (&dec);
789 1166
790 if (offset_return) 1167 if (offset_return)
791 *offset_return = dec.cur; 1168 *offset_return = dec.cur;
795 dec.err = "garbage after CBOR object"; 1172 dec.err = "garbage after CBOR object";
796 1173
797 if (dec.err) 1174 if (dec.err)
798 { 1175 {
799 SvREFCNT_dec (sv); 1176 SvREFCNT_dec (sv);
800 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)SvPVX (string), (int)(uint8_t)*dec.cur); 1177 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur);
801 } 1178 }
802 1179
803 sv = sv_2mortal (sv); 1180 sv = sv_2mortal (sv);
804 1181
805 return sv; 1182 return sv;
811MODULE = CBOR::XS PACKAGE = CBOR::XS 1188MODULE = CBOR::XS PACKAGE = CBOR::XS
812 1189
813BOOT: 1190BOOT:
814{ 1191{
815 cbor_stash = gv_stashpv ("CBOR::XS" , 1); 1192 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); 1193 cbor_tagged_stash = gv_stashpv ("CBOR::XS::Tagged" , 1);
818 1194
819 cbor_true = get_bool ("CBOR::XS::true"); 1195 types_boolean_stash = gv_stashpv ("Types::Serialiser::Boolean", 1);
820 cbor_false = get_bool ("CBOR::XS::false"); 1196 types_error_stash = gv_stashpv ("Types::Serialiser::Error" , 1);
1197
1198 types_true = get_bool ("Types::Serialiser::true" );
1199 types_false = get_bool ("Types::Serialiser::false");
1200 types_error = get_bool ("Types::Serialiser::error");
1201
1202 default_filter = newSVpv ("CBOR::XS::default_filter", 0);
1203
1204 sv_cbor = newSVpv ("CBOR", 0);
1205 SvREADONLY_on (sv_cbor);
821} 1206}
822 1207
823PROTOTYPES: DISABLE 1208PROTOTYPES: DISABLE
824 1209
825void CLONE (...) 1210void CLONE (...)
826 CODE: 1211 CODE:
827 cbor_stash = 0; 1212 cbor_stash = 0;
828 cbor_boolean_stash = 0;
829 cbor_tagged_stash = 0; 1213 cbor_tagged_stash = 0;
1214 types_error_stash = 0;
1215 types_boolean_stash = 0;
830 1216
831void new (char *klass) 1217void new (char *klass)
832 PPCODE: 1218 PPCODE:
833{ 1219{
834 SV *pv = NEWSV (0, sizeof (CBOR)); 1220 SV *pv = NEWSV (0, sizeof (CBOR));
842 1228
843void shrink (CBOR *self, int enable = 1) 1229void shrink (CBOR *self, int enable = 1)
844 ALIAS: 1230 ALIAS:
845 shrink = F_SHRINK 1231 shrink = F_SHRINK
846 allow_unknown = F_ALLOW_UNKNOWN 1232 allow_unknown = F_ALLOW_UNKNOWN
1233 allow_sharing = F_ALLOW_SHARING
1234 allow_cycles = F_ALLOW_CYCLES
1235 pack_strings = F_PACK_STRINGS
1236 validate_utf8 = F_VALIDATE_UTF8
847 PPCODE: 1237 PPCODE:
848{ 1238{
849 if (enable) 1239 if (enable)
850 self->flags |= ix; 1240 self->flags |= ix;
851 else 1241 else
856 1246
857void get_shrink (CBOR *self) 1247void get_shrink (CBOR *self)
858 ALIAS: 1248 ALIAS:
859 get_shrink = F_SHRINK 1249 get_shrink = F_SHRINK
860 get_allow_unknown = F_ALLOW_UNKNOWN 1250 get_allow_unknown = F_ALLOW_UNKNOWN
1251 get_allow_sharing = F_ALLOW_SHARING
1252 get_allow_cycles = F_ALLOW_CYCLES
1253 get_pack_strings = F_PACK_STRINGS
1254 get_validate_utf8 = F_VALIDATE_UTF8
861 PPCODE: 1255 PPCODE:
862 XPUSHs (boolSV (self->flags & ix)); 1256 XPUSHs (boolSV (self->flags & ix));
863 1257
864void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) 1258void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
865 PPCODE: 1259 PPCODE:
881 CODE: 1275 CODE:
882 RETVAL = self->max_size; 1276 RETVAL = self->max_size;
883 OUTPUT: 1277 OUTPUT:
884 RETVAL 1278 RETVAL
885 1279
886#if 0 //TODO 1280void filter (CBOR *self, SV *filter = 0)
887
888void filter_cbor_object (CBOR *self, SV *cb = &PL_sv_undef)
889 PPCODE: 1281 PPCODE:
890{
891 SvREFCNT_dec (self->cb_object); 1282 SvREFCNT_dec (self->filter);
892 self->cb_object = SvOK (cb) ? newSVsv (cb) : 0; 1283 self->filter = filter ? newSVsv (filter) : filter;
893
894 XPUSHs (ST (0)); 1284 XPUSHs (ST (0));
895}
896 1285
897void filter_cbor_single_key_object (CBOR *self, SV *key, SV *cb = &PL_sv_undef) 1286SV *get_filter (CBOR *self)
898 PPCODE: 1287 CODE:
899{ 1288 RETVAL = self->filter ? self->filter : NEWSV (0, 0);
900 if (!self->cb_sk_object) 1289 OUTPUT:
901 self->cb_sk_object = newHV (); 1290 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 1291
921void encode (CBOR *self, SV *scalar) 1292void encode (CBOR *self, SV *scalar)
922 PPCODE: 1293 PPCODE:
923 PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN; 1294 PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN;
924 XPUSHs (scalar); 1295 XPUSHs (scalar);
937 EXTEND (SP, 2); 1308 EXTEND (SP, 2);
938 PUSHs (sv); 1309 PUSHs (sv);
939 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); 1310 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
940} 1311}
941 1312
942#if 0
943
944void DESTROY (CBOR *self) 1313void DESTROY (CBOR *self)
945 CODE: 1314 PPCODE:
946 SvREFCNT_dec (self->cb_sk_object); 1315 cbor_free (self);
947 SvREFCNT_dec (self->cb_object);
948
949#endif
950 1316
951PROTOTYPES: ENABLE 1317PROTOTYPES: ENABLE
952 1318
953void encode_cbor (SV *scalar) 1319void encode_cbor (SV *scalar)
1320 ALIAS:
1321 encode_cbor = 0
1322 encode_cbor_sharing = F_ALLOW_SHARING
954 PPCODE: 1323 PPCODE:
955{ 1324{
956 CBOR cbor; 1325 CBOR cbor;
957 cbor_init (&cbor); 1326 cbor_init (&cbor);
1327 cbor.flags |= ix;
958 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN; 1328 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;
959 XPUSHs (scalar); 1329 XPUSHs (scalar);
960} 1330}
961 1331
962void decode_cbor (SV *cborstr) 1332void decode_cbor (SV *cborstr)

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines