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 1.38 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// 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_ALLOW_CYCLES 0x00000008UL
103#define F_PACK_STRINGS 0x00000010UL
104#define F_VALIDATE_UTF8 0x00000020UL
45 105
46#define INIT_SIZE 32 // initial scalar size to be allocated 106#define INIT_SIZE 32 // initial scalar size to be allocated
47 107
48#define SB do { 108#define SB do {
49#define SE } while (0) 109#define SE } while (0)
61# define CBOR_SLOW 0 121# define CBOR_SLOW 0
62# define CBOR_STASH cbor_stash 122# define CBOR_STASH cbor_stash
63#endif 123#endif
64 124
65static HV *cbor_stash, *types_boolean_stash, *types_error_stash, *cbor_tagged_stash; // CBOR::XS:: 125static HV *cbor_stash, *types_boolean_stash, *types_error_stash, *cbor_tagged_stash; // CBOR::XS::
66static SV *types_true, *types_false, *types_error; 126static SV *types_true, *types_false, *types_error, *sv_cbor, *default_filter;
67 127
68typedef struct { 128typedef struct {
69 U32 flags; 129 U32 flags;
70 U32 max_depth; 130 U32 max_depth;
71 STRLEN max_size; 131 STRLEN max_size;
132 SV *filter;
72} CBOR; 133} CBOR;
73 134
74ecb_inline void 135ecb_inline void
75cbor_init (CBOR *cbor) 136cbor_init (CBOR *cbor)
76{ 137{
77 Zero (cbor, 1, CBOR); 138 Zero (cbor, 1, CBOR);
78 cbor->max_depth = 512; 139 cbor->max_depth = 512;
140}
141
142ecb_inline void
143cbor_free (CBOR *cbor)
144{
145 SvREFCNT_dec (cbor->filter);
79} 146}
80 147
81///////////////////////////////////////////////////////////////////////////// 148/////////////////////////////////////////////////////////////////////////////
82// utility functions 149// utility functions
83 150
105 SvPV_renew (sv, SvCUR (sv) + 1); 172 SvPV_renew (sv, SvCUR (sv) + 1);
106#endif 173#endif
107 } 174 }
108} 175}
109 176
110///////////////////////////////////////////////////////////////////////////// 177// minimum length of a string to be registered for stringref
111// fp hell 178ecb_inline int
112 179minimum_string_length (UV idx)
113//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}
114 191
115///////////////////////////////////////////////////////////////////////////// 192/////////////////////////////////////////////////////////////////////////////
116// encoder 193// encoder
117 194
118// structure used for encoding CBOR 195// structure used for encoding CBOR
121 char *cur; // SvPVX (sv) + current output position 198 char *cur; // SvPVX (sv) + current output position
122 char *end; // SvEND (sv) 199 char *end; // SvEND (sv)
123 SV *sv; // result scalar 200 SV *sv; // result scalar
124 CBOR cbor; 201 CBOR cbor;
125 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;
126} enc_t; 207} enc_t;
127 208
128ecb_inline void 209ecb_inline void
129need (enc_t *enc, STRLEN len) 210need (enc_t *enc, STRLEN len)
130{ 211{
147static void 228static void
148encode_uint (enc_t *enc, int major, UV len) 229encode_uint (enc_t *enc, int major, UV len)
149{ 230{
150 need (enc, 9); 231 need (enc, 9);
151 232
152 if (len < 24) 233 if (ecb_expect_true (len < LENGTH_EXT1))
153 *enc->cur++ = major | len; 234 *enc->cur++ = major | len;
154 else if (len <= 0xff) 235 else if (ecb_expect_true (len <= 0xffU))
155 { 236 {
156 *enc->cur++ = major | 24; 237 *enc->cur++ = major | LENGTH_EXT1;
157 *enc->cur++ = len; 238 *enc->cur++ = len;
158 } 239 }
159 else if (len <= 0xffff) 240 else if (len <= 0xffffU)
160 { 241 {
161 *enc->cur++ = major | 25; 242 *enc->cur++ = major | LENGTH_EXT2;
162 *enc->cur++ = len >> 8; 243 *enc->cur++ = len >> 8;
163 *enc->cur++ = len; 244 *enc->cur++ = len;
164 } 245 }
165 else if (len <= 0xffffffff) 246 else if (len <= 0xffffffffU)
166 { 247 {
167 *enc->cur++ = major | 26; 248 *enc->cur++ = major | LENGTH_EXT4;
168 *enc->cur++ = len >> 24; 249 *enc->cur++ = len >> 24;
169 *enc->cur++ = len >> 16; 250 *enc->cur++ = len >> 16;
170 *enc->cur++ = len >> 8; 251 *enc->cur++ = len >> 8;
171 *enc->cur++ = len; 252 *enc->cur++ = len;
172 } 253 }
173 else 254 else
174 { 255 {
175 *enc->cur++ = major | 27; 256 *enc->cur++ = major | LENGTH_EXT8;
176 *enc->cur++ = len >> 56; 257 *enc->cur++ = len >> 56;
177 *enc->cur++ = len >> 48; 258 *enc->cur++ = len >> 48;
178 *enc->cur++ = len >> 40; 259 *enc->cur++ = len >> 40;
179 *enc->cur++ = len >> 32; 260 *enc->cur++ = len >> 32;
180 *enc->cur++ = len >> 24; 261 *enc->cur++ = len >> 24;
182 *enc->cur++ = len >> 8; 263 *enc->cur++ = len >> 8;
183 *enc->cur++ = len; 264 *enc->cur++ = len;
184 } 265 }
185} 266}
186 267
187static void 268ecb_inline void
269encode_tag (enc_t *enc, UV tag)
270{
271 encode_uint (enc, MAJOR_TAG, tag);
272}
273
274ecb_inline void
188encode_str (enc_t *enc, int utf8, char *str, STRLEN len) 275encode_str (enc_t *enc, int utf8, char *str, STRLEN len)
189{ 276{
190 encode_uint (enc, utf8 ? 0x60 : 0x40, len); 277 encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len);
191 need (enc, len); 278 need (enc, len);
192 memcpy (enc->cur, str, len); 279 memcpy (enc->cur, str, len);
193 enc->cur += len; 280 enc->cur += len;
194} 281}
195 282
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
196static void encode_sv (enc_t *enc, SV *sv); 308static void encode_sv (enc_t *enc, SV *sv);
197 309
198static void 310static void
199encode_av (enc_t *enc, AV *av) 311encode_av (enc_t *enc, AV *av)
200{ 312{
203 if (enc->depth >= enc->cbor.max_depth) 315 if (enc->depth >= enc->cbor.max_depth)
204 croak (ERR_NESTING_EXCEEDED); 316 croak (ERR_NESTING_EXCEEDED);
205 317
206 ++enc->depth; 318 ++enc->depth;
207 319
208 encode_uint (enc, 0x80, len + 1); 320 encode_uint (enc, MAJOR_ARRAY, len + 1);
209 321
210 for (i = 0; i <= len; ++i) 322 for (i = 0; i <= len; ++i)
211 { 323 {
212 SV **svp = av_fetch (av, i, 0); 324 SV **svp = av_fetch (av, i, 0);
213 encode_sv (enc, svp ? *svp : &PL_sv_undef); 325 encode_sv (enc, svp ? *svp : &PL_sv_undef);
228 340
229 int pairs = hv_iterinit (hv); 341 int pairs = hv_iterinit (hv);
230 int mg = SvMAGICAL (hv); 342 int mg = SvMAGICAL (hv);
231 343
232 if (mg) 344 if (mg)
233 encode_ch (enc, 0xa0 | 31); 345 encode_ch (enc, MAJOR_MAP | MINOR_INDEF);
234 else 346 else
235 encode_uint (enc, 0xa0, pairs); 347 encode_uint (enc, MAJOR_MAP, pairs);
236 348
237 while ((he = hv_iternext (hv))) 349 while ((he = hv_iternext (hv)))
238 { 350 {
239 if (HeKLEN (he) == HEf_SVKEY) 351 if (HeKLEN (he) == HEf_SVKEY)
240 encode_sv (enc, HeSVKEY (he)); 352 encode_sv (enc, HeSVKEY (he));
241 else 353 else
242 encode_str (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he)); 354 encode_strref (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he));
243 355
244 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));
245 } 357 }
246 358
247 if (mg) 359 if (mg)
248 encode_ch (enc, 0xe0 | 31); 360 encode_ch (enc, MAJOR_MISC | MINOR_INDEF);
249 361
250 --enc->depth; 362 --enc->depth;
251} 363}
252 364
253// encode objects, arrays and special \0=false and \1=true values. 365// encode objects, arrays and special \0=false and \1=true values.
254static void 366static void
255encode_rv (enc_t *enc, SV *sv) 367encode_rv (enc_t *enc, SV *sv)
256{ 368{
257 svtype svt;
258
259 SvGETMAGIC (sv); 369 SvGETMAGIC (sv);
370
260 svt = SvTYPE (sv); 371 svtype svt = SvTYPE (sv);
261 372
262 if (ecb_expect_false (SvOBJECT (sv))) 373 if (ecb_expect_false (SvOBJECT (sv)))
263 { 374 {
264 HV *boolean_stash = !CBOR_SLOW || types_boolean_stash 375 HV *boolean_stash = !CBOR_SLOW || types_boolean_stash
265 ? types_boolean_stash 376 ? types_boolean_stash
269 : gv_stashpv ("Types::Serialiser::Error", 1); 380 : gv_stashpv ("Types::Serialiser::Error", 1);
270 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 381 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
271 ? cbor_tagged_stash 382 ? cbor_tagged_stash
272 : gv_stashpv ("CBOR::XS::Tagged" , 1); 383 : gv_stashpv ("CBOR::XS::Tagged" , 1);
273 384
385 HV *stash = SvSTASH (sv);
386
274 if (SvSTASH (sv) == boolean_stash) 387 if (stash == boolean_stash)
275 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 }
276 else if (SvSTASH (sv) == error_stash) 392 else if (stash == error_stash)
277 encode_ch (enc, 0xe0 | 23); 393 {
394 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
395 return;
396 }
278 else if (SvSTASH (sv) == tagged_stash) 397 else if (stash == tagged_stash)
279 { 398 {
280 if (svt != SVt_PVAV) 399 if (svt != SVt_PVAV)
281 croak ("encountered CBOR::XS::Tagged object that isn't an array"); 400 croak ("encountered CBOR::XS::Tagged object that isn't an array");
282 401
283 encode_uint (enc, 0xc0, SvUV (*av_fetch ((AV *)sv, 0, 1))); 402 encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1)));
284 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;
285 } 422 }
286 else 423 else
287 { 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);
288 // 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
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))); 442 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
297 443
298 // calling with G_SCALAR ensures that we always get a 1 return value
299 PUTBACK; 444 PUTBACK;
445 // G_SCALAR ensures that return value is 1
300 call_sv ((SV *)GvCV (to_cbor), G_SCALAR); 446 call_sv ((SV *)GvCV (method), G_SCALAR);
301 SPAGAIN; 447 SPAGAIN;
302 448
303 // catch this surprisingly common error 449 // catch this surprisingly common error
304 if (SvROK (TOPs) && SvRV (TOPs) == sv) 450 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))); 451 croak ("%s::TO_CBOR method returned same object as was passed instead of a new one", HvNAME (stash));
306 452
307 sv = POPs; 453 encode_sv (enc, POPs);
454
308 PUTBACK; 455 PUTBACK;
309 456
310 encode_sv (enc, sv);
311
312 FREETMPS; LEAVE; 457 FREETMPS; LEAVE;
313 } 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 }
314 else 488 else
315 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",
316 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 490 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
317 }
318 } 491 }
319 else if (svt == SVt_PVHV) 492 else if (svt == SVt_PVHV)
320 encode_hv (enc, (HV *)sv); 493 encode_hv (enc, (HV *)sv);
321 else if (svt == SVt_PVAV) 494 else if (svt == SVt_PVAV)
322 encode_av (enc, (AV *)sv); 495 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 496 else
341 croak ("encountered %s, but CBOR can only represent references to arrays or hashes", 497 {
342 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 498 encode_tag (enc, CBOR_TAG_INDIRECTION);
499 encode_sv (enc, sv);
500 }
343} 501}
344 502
345static void 503static void
346encode_nv (enc_t *enc, SV *sv) 504encode_nv (enc_t *enc, SV *sv)
347{ 505{
348 double nv = SvNVX (sv); 506 double nv = SvNVX (sv);
349 507
350 need (enc, 9); 508 need (enc, 9);
351 509
352 if (ecb_expect_false (nv == (U32)nv)) 510 if (ecb_expect_false (nv == (NV)(U32)nv))
353 encode_uint (enc, 0x00, (U32)nv); 511 encode_uint (enc, MAJOR_POS_INT, (U32)nv);
354 //TODO: maybe I32? 512 //TODO: maybe I32?
355 else if (ecb_expect_false (nv == (float)nv)) 513 else if (ecb_expect_false (nv == (float)nv))
356 { 514 {
357 uint32_t fp = ecb_float_to_binary32 (nv); 515 uint32_t fp = ecb_float_to_binary32 (nv);
358 516
359 *enc->cur++ = 0xe0 | 26; 517 *enc->cur++ = MAJOR_MISC | MISC_FLOAT32;
360 518
361 if (!ecb_big_endian ()) 519 if (!ecb_big_endian ())
362 fp = ecb_bswap32 (fp); 520 fp = ecb_bswap32 (fp);
363 521
364 memcpy (enc->cur, &fp, 4); 522 memcpy (enc->cur, &fp, 4);
366 } 524 }
367 else 525 else
368 { 526 {
369 uint64_t fp = ecb_double_to_binary64 (nv); 527 uint64_t fp = ecb_double_to_binary64 (nv);
370 528
371 *enc->cur++ = 0xe0 | 27; 529 *enc->cur++ = MAJOR_MISC | MISC_FLOAT64;
372 530
373 if (!ecb_big_endian ()) 531 if (!ecb_big_endian ())
374 fp = ecb_bswap64 (fp); 532 fp = ecb_bswap64 (fp);
375 533
376 memcpy (enc->cur, &fp, 8); 534 memcpy (enc->cur, &fp, 8);
385 543
386 if (SvPOKp (sv)) 544 if (SvPOKp (sv))
387 { 545 {
388 STRLEN len; 546 STRLEN len;
389 char *str = SvPV (sv, len); 547 char *str = SvPV (sv, len);
390 encode_str (enc, SvUTF8 (sv), str, len); 548 encode_strref (enc, SvUTF8 (sv), str, len);
391 } 549 }
392 else if (SvNOKp (sv)) 550 else if (SvNOKp (sv))
393 encode_nv (enc, sv); 551 encode_nv (enc, sv);
394 else if (SvIOKp (sv)) 552 else if (SvIOKp (sv))
395 { 553 {
396 if (SvIsUV (sv)) 554 if (SvIsUV (sv))
397 encode_uint (enc, 0x00, SvUVX (sv)); 555 encode_uint (enc, MAJOR_POS_INT, SvUVX (sv));
398 else if (SvIVX (sv) >= 0) 556 else if (SvIVX (sv) >= 0)
399 encode_uint (enc, 0x00, SvIVX (sv)); 557 encode_uint (enc, MAJOR_POS_INT, SvIVX (sv));
400 else 558 else
401 encode_uint (enc, 0x20, -(SvIVX (sv) + 1)); 559 encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1));
402 } 560 }
403 else if (SvROK (sv)) 561 else if (SvROK (sv))
404 encode_rv (enc, SvRV (sv)); 562 encode_rv (enc, SvRV (sv));
405 else if (!SvOK (sv)) 563 else if (!SvOK (sv))
406 encode_ch (enc, 0xe0 | 22); 564 encode_ch (enc, MAJOR_MISC | SIMPLE_NULL);
407 else if (enc->cbor.flags & F_ALLOW_UNKNOWN) 565 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
408 encode_ch (enc, 0xe0 | 23); 566 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
409 else 567 else
410 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",
411 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv)); 569 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
412} 570}
413 571
414static SV * 572static SV *
415encode_cbor (SV *scalar, CBOR *cbor) 573encode_cbor (SV *scalar, CBOR *cbor)
416{ 574{
417 enc_t enc; 575 enc_t enc = { };
418 576
419 enc.cbor = *cbor; 577 enc.cbor = *cbor;
420 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 578 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
421 enc.cur = SvPVX (enc.sv); 579 enc.cur = SvPVX (enc.sv);
422 enc.end = SvEND (enc.sv); 580 enc.end = SvEND (enc.sv);
423 enc.depth = 0;
424 581
425 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
426 encode_sv (&enc, scalar); 591 encode_sv (&enc, scalar);
427 592
428 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 593 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
429 *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
430 595
444 U8 *end; // end of input string 609 U8 *end; // end of input string
445 const char *err; // parse error, if != 0 610 const char *err; // parse error, if != 0
446 CBOR cbor; 611 CBOR cbor;
447 U32 depth; // recursion depth 612 U32 depth; // recursion depth
448 U32 maxdepth; // recursion depth limit 613 U32 maxdepth; // recursion depth limit
614 AV *shareable;
615 AV *stringref;
616 SV *decode_tagged;
449} dec_t; 617} dec_t;
450 618
451#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
452 620
453#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")
456#define DEC_DEC_DEPTH --dec->depth 624#define DEC_DEC_DEPTH --dec->depth
457 625
458static UV 626static UV
459decode_uint (dec_t *dec) 627decode_uint (dec_t *dec)
460{ 628{
461 switch (*dec->cur & 31) 629 U8 m = *dec->cur & MINOR_MASK;
462 { 630 ++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;
467 631
468 case 24: 632 if (ecb_expect_true (m < LENGTH_EXT1))
633 return m;
634 else if (ecb_expect_true (m == LENGTH_EXT1))
635 {
469 WANT (2); 636 WANT (1);
470 dec->cur += 2; 637 dec->cur += 1;
471 return dec->cur[-1]; 638 return dec->cur[-1];
472 639 }
473 case 25: 640 else if (ecb_expect_true (m == LENGTH_EXT2))
641 {
474 WANT (3); 642 WANT (2);
475 dec->cur += 3; 643 dec->cur += 2;
476 return (((UV)dec->cur[-2]) << 8) 644 return (((UV)dec->cur[-2]) << 8)
477 | ((UV)dec->cur[-1]); 645 | ((UV)dec->cur[-1]);
478 646 }
479 case 26: 647 else if (ecb_expect_true (m == LENGTH_EXT4))
648 {
480 WANT (5); 649 WANT (4);
481 dec->cur += 5; 650 dec->cur += 4;
482 return (((UV)dec->cur[-4]) << 24) 651 return (((UV)dec->cur[-4]) << 24)
483 | (((UV)dec->cur[-3]) << 16) 652 | (((UV)dec->cur[-3]) << 16)
484 | (((UV)dec->cur[-2]) << 8) 653 | (((UV)dec->cur[-2]) << 8)
485 | ((UV)dec->cur[-1]); 654 | ((UV)dec->cur[-1]);
486 655 }
487 case 27: 656 else if (ecb_expect_true (m == LENGTH_EXT8))
657 {
488 WANT (9); 658 WANT (8);
489 dec->cur += 9; 659 dec->cur += 8;
660
661 return
662#if UVSIZE < 8
663 0
664#else
490 return (((UV)dec->cur[-8]) << 56) 665 (((UV)dec->cur[-8]) << 56)
491 | (((UV)dec->cur[-7]) << 48) 666 | (((UV)dec->cur[-7]) << 48)
492 | (((UV)dec->cur[-6]) << 40) 667 | (((UV)dec->cur[-6]) << 40)
493 | (((UV)dec->cur[-5]) << 32) 668 | (((UV)dec->cur[-5]) << 32)
669#endif
494 | (((UV)dec->cur[-4]) << 24) 670 | (((UV)dec->cur[-4]) << 24)
495 | (((UV)dec->cur[-3]) << 16) 671 | (((UV)dec->cur[-3]) << 16)
496 | (((UV)dec->cur[-2]) << 8) 672 | (((UV)dec->cur[-2]) << 8)
497 | ((UV)dec->cur[-1]); 673 | ((UV)dec->cur[-1]);
498 674 }
499 default: 675 else
500 ERR ("corrupted CBOR data (unsupported integer minor encoding)"); 676 ERR ("corrupted CBOR data (unsupported integer minor encoding)");
501 }
502 677
503fail: 678fail:
504 return 0; 679 return 0;
505} 680}
506 681
511{ 686{
512 AV *av = newAV (); 687 AV *av = newAV ();
513 688
514 DEC_INC_DEPTH; 689 DEC_INC_DEPTH;
515 690
516 if ((*dec->cur & 31) == 31) 691 if (*dec->cur == (MAJOR_ARRAY | MINOR_INDEF))
517 { 692 {
518 ++dec->cur; 693 ++dec->cur;
519 694
520 for (;;) 695 for (;;)
521 { 696 {
522 WANT (1); 697 WANT (1);
523 698
524 if (*dec->cur == (0xe0 | 31)) 699 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
525 { 700 {
526 ++dec->cur; 701 ++dec->cur;
527 break; 702 break;
528 } 703 }
529 704
532 } 707 }
533 else 708 else
534 { 709 {
535 int i, len = decode_uint (dec); 710 int i, len = decode_uint (dec);
536 711
712 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); 713 av_fill (av, len - 1);
538 714
539 for (i = 0; i < len; ++i) 715 for (i = 0; i < len; ++i)
540 AvARRAY (av)[i] = decode_sv (dec); 716 AvARRAY (av)[i] = decode_sv (dec);
541 } 717 }
547 SvREFCNT_dec (av); 723 SvREFCNT_dec (av);
548 DEC_DEC_DEPTH; 724 DEC_DEC_DEPTH;
549 return &PL_sv_undef; 725 return &PL_sv_undef;
550} 726}
551 727
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
552static SV * 772static SV *
553decode_hv (dec_t *dec) 773decode_hv (dec_t *dec)
554{ 774{
555 HV *hv = newHV (); 775 HV *hv = newHV ();
556 776
557 DEC_INC_DEPTH; 777 DEC_INC_DEPTH;
558 778
559 if ((*dec->cur & 31) == 31) 779 if (*dec->cur == (MAJOR_MAP | MINOR_INDEF))
560 { 780 {
561 ++dec->cur; 781 ++dec->cur;
562 782
563 for (;;) 783 for (;;)
564 { 784 {
565 WANT (1); 785 WANT (1);
566 786
567 if (*dec->cur == (0xe0 | 31)) 787 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
568 { 788 {
569 ++dec->cur; 789 ++dec->cur;
570 break; 790 break;
571 } 791 }
572 792
573 SV *k = decode_sv (dec); 793 decode_he (dec, hv);
574 SV *v = decode_sv (dec);
575
576 hv_store_ent (hv, k, v, 0);
577 } 794 }
578 } 795 }
579 else 796 else
580 { 797 {
581 int len = decode_uint (dec); 798 int pairs = decode_uint (dec);
582 799
583 while (len--) 800 while (pairs--)
584 { 801 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 } 802 }
591 803
592 DEC_DEC_DEPTH; 804 DEC_DEC_DEPTH;
593 return newRV_noinc ((SV *)hv); 805 return newRV_noinc ((SV *)hv);
594 806
601static SV * 813static SV *
602decode_str (dec_t *dec, int utf8) 814decode_str (dec_t *dec, int utf8)
603{ 815{
604 SV *sv = 0; 816 SV *sv = 0;
605 817
606 if ((*dec->cur & 31) == 31) 818 if ((*dec->cur & MINOR_MASK) == MINOR_INDEF)
607 { 819 {
820 // indefinite length strings
608 ++dec->cur; 821 ++dec->cur;
609 822
823 U8 major = *dec->cur & MAJOR_MISC;
824
610 sv = newSVpvn ("", 0); 825 sv = newSVpvn ("", 0);
611 826
612 // not very fast, and certainly not robust against illegal input
613 for (;;) 827 for (;;)
614 { 828 {
615 WANT (1); 829 WANT (1);
616 830
617 if (*dec->cur == (0xe0 | 31)) 831 if ((*dec->cur - major) > LENGTH_EXT8)
832 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
618 { 833 {
619 ++dec->cur; 834 ++dec->cur;
620 break; 835 break;
621 } 836 }
837 else
838 ERR ("corrupted CBOR data (invalid chunks in indefinite length string)");
622 839
623 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;
624 } 845 }
625 } 846 }
626 else 847 else
627 { 848 {
628 STRLEN len = decode_uint (dec); 849 STRLEN len = decode_uint (dec);
629 850
630 WANT (len); 851 WANT (len);
631 sv = newSVpvn (dec->cur, len); 852 sv = newSVpvn (dec->cur, len);
632 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));
633 } 858 }
634 859
635 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
636 SvUTF8_on (sv); 866 SvUTF8_on (sv);
867 }
637 868
638 return sv; 869 return sv;
639 870
640fail: 871fail:
641 SvREFCNT_dec (sv); 872 SvREFCNT_dec (sv);
643} 874}
644 875
645static SV * 876static SV *
646decode_tagged (dec_t *dec) 877decode_tagged (dec_t *dec)
647{ 878{
879 SV *sv = 0;
648 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:
649 SV *sv = decode_sv (dec); 887 sv = decode_sv (dec);
888 break;
650 889
651 if (tag == CBOR_TAG_MAGIC) 890 case CBOR_TAG_INDIRECTION:
652 return sv; 891 sv = newRV_noinc (decode_sv (dec));
892 break;
653 893
654 if (tag == CBOR_TAG_PERL_OBJECT) 894 case CBOR_TAG_STRINGREF_NAMESPACE:
655 { 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
656 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV) 966 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
657 ERR ("corrupted CBOR data (non-array perl object)"); 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;
658 1010 }
659 // TODO 1011 break;
660 }
661 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 {
662 AV *av = newAV (); 1040 AV *av = newAV ();
663 av_push (av, newSVuv (tag)); 1041 av_push (av, newSVuv (tag));
664 av_push (av, sv); 1042 av_push (av, sv);
665 1043
666 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 1044 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
667 ? cbor_tagged_stash 1045 ? cbor_tagged_stash
668 : gv_stashpv ("CBOR::XS::Tagged" , 1); 1046 : gv_stashpv ("CBOR::XS::Tagged" , 1);
669
670 return sv_bless (newRV_noinc ((SV *)av), tagged_stash); 1047 sv = sv_bless (newRV_noinc ((SV *)av), tagged_stash);
1048 }
1049
1050 PUTBACK;
1051
1052 FREETMPS; LEAVE;
1053 }
1054 break;
1055 }
1056
1057 return sv;
671 1058
672fail: 1059fail:
673 SvREFCNT_dec (sv); 1060 SvREFCNT_dec (sv);
674 return &PL_sv_undef; 1061 return &PL_sv_undef;
675} 1062}
677static SV * 1064static SV *
678decode_sv (dec_t *dec) 1065decode_sv (dec_t *dec)
679{ 1066{
680 WANT (1); 1067 WANT (1);
681 1068
682 switch (*dec->cur >> 5) 1069 switch (*dec->cur >> MAJOR_SHIFT)
683 { 1070 {
684 case 0: // unsigned int 1071 case MAJOR_POS_INT >> MAJOR_SHIFT: return newSVuv (decode_uint (dec));
685 return newSVuv (decode_uint (dec)); 1072 case MAJOR_NEG_INT >> MAJOR_SHIFT: return newSViv (-1 - (IV)decode_uint (dec));
686 case 1: // negative int 1073 case MAJOR_BYTES >> MAJOR_SHIFT: return decode_str (dec, 0);
687 return newSViv (-1 - (IV)decode_uint (dec)); 1074 case MAJOR_TEXT >> MAJOR_SHIFT: return decode_str (dec, 1);
688 case 2: // octet string 1075 case MAJOR_ARRAY >> MAJOR_SHIFT: return decode_av (dec);
689 return decode_str (dec, 0); 1076 case MAJOR_MAP >> MAJOR_SHIFT: return decode_hv (dec);
690 case 3: // utf-8 string 1077 case MAJOR_TAG >> MAJOR_SHIFT: return decode_tagged (dec);
691 return decode_str (dec, 1); 1078
692 case 4: // array 1079 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) 1080 switch (*dec->cur++ & MINOR_MASK)
700 { 1081 {
701 case 20: 1082 case SIMPLE_FALSE:
702#if CBOR_SLOW 1083#if CBOR_SLOW
703 types_false = get_bool ("Types::Serialiser::false"); 1084 types_false = get_bool ("Types::Serialiser::false");
704#endif 1085#endif
705 return newSVsv (types_false); 1086 return newSVsv (types_false);
706 case 21: 1087 case SIMPLE_TRUE:
707#if CBOR_SLOW 1088#if CBOR_SLOW
708 types_true = get_bool ("Types::Serialiser::true"); 1089 types_true = get_bool ("Types::Serialiser::true");
709#endif 1090#endif
710 return newSVsv (types_true); 1091 return newSVsv (types_true);
711 case 22: 1092 case SIMPLE_NULL:
712 return newSVsv (&PL_sv_undef); 1093 return newSVsv (&PL_sv_undef);
713 case 23: 1094 case SIMPLE_UNDEF:
714#if CBOR_SLOW 1095#if CBOR_SLOW
715 types_error = get_bool ("Types::Serialiser::error"); 1096 types_error = get_bool ("Types::Serialiser::error");
716#endif 1097#endif
717 return newSVsv (types_error); 1098 return newSVsv (types_error);
718 1099
719 case 25: 1100 case MISC_FLOAT16:
720 { 1101 {
721 WANT (2); 1102 WANT (2);
722 1103
723 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1]; 1104 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1];
724 dec->cur += 2; 1105 dec->cur += 2;
725 1106
726 return newSVnv (ecb_binary16_to_float (fp)); 1107 return newSVnv (ecb_binary16_to_float (fp));
727 } 1108 }
728 1109
729 case 26: 1110 case MISC_FLOAT32:
730 { 1111 {
731 uint32_t fp; 1112 uint32_t fp;
732 WANT (4); 1113 WANT (4);
733 memcpy (&fp, dec->cur, 4); 1114 memcpy (&fp, dec->cur, 4);
734 dec->cur += 4; 1115 dec->cur += 4;
737 fp = ecb_bswap32 (fp); 1118 fp = ecb_bswap32 (fp);
738 1119
739 return newSVnv (ecb_binary32_to_float (fp)); 1120 return newSVnv (ecb_binary32_to_float (fp));
740 } 1121 }
741 1122
742 case 27: 1123 case MISC_FLOAT64:
743 { 1124 {
744 uint64_t fp; 1125 uint64_t fp;
745 WANT (8); 1126 WANT (8);
746 memcpy (&fp, dec->cur, 8); 1127 memcpy (&fp, dec->cur, 8);
747 dec->cur += 8; 1128 dec->cur += 8;
750 fp = ecb_bswap64 (fp); 1131 fp = ecb_bswap64 (fp);
751 1132
752 return newSVnv (ecb_binary64_to_double (fp)); 1133 return newSVnv (ecb_binary64_to_double (fp));
753 } 1134 }
754 1135
755 // 0..19 unassigned 1136 // 0..19 unassigned simple
756 // 24 reserved + unassigned (reserved values are not encodable) 1137 // 24 reserved + unassigned (reserved values are not encodable)
757 default: 1138 default:
758 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)"); 1139 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)");
759 } 1140 }
760 1141
766} 1147}
767 1148
768static SV * 1149static SV *
769decode_cbor (SV *string, CBOR *cbor, char **offset_return) 1150decode_cbor (SV *string, CBOR *cbor, char **offset_return)
770{ 1151{
771 dec_t dec; 1152 dec_t dec = { };
772 SV *sv; 1153 SV *sv;
1154 STRLEN len;
1155 char *data = SvPVbyte (string, len);
773 1156
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) 1157 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", 1158 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); 1159 (unsigned long)len, (unsigned long)cbor->max_size);
805 }
806
807 sv_utf8_downgrade (string, 0);
808 1160
809 dec.cbor = *cbor; 1161 dec.cbor = *cbor;
810 dec.cur = (U8 *)SvPVX (string); 1162 dec.cur = (U8 *)data;
811 dec.end = (U8 *)SvEND (string); 1163 dec.end = (U8 *)data + len;
812 dec.err = 0;
813 dec.depth = 0;
814 1164
815 sv = decode_sv (&dec); 1165 sv = decode_sv (&dec);
816 1166
817 if (offset_return) 1167 if (offset_return)
818 *offset_return = dec.cur; 1168 *offset_return = dec.cur;
822 dec.err = "garbage after CBOR object"; 1172 dec.err = "garbage after CBOR object";
823 1173
824 if (dec.err) 1174 if (dec.err)
825 { 1175 {
826 SvREFCNT_dec (sv); 1176 SvREFCNT_dec (sv);
827 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);
828 } 1178 }
829 1179
830 sv = sv_2mortal (sv); 1180 sv = sv_2mortal (sv);
831 1181
832 return sv; 1182 return sv;
846 types_error_stash = gv_stashpv ("Types::Serialiser::Error" , 1); 1196 types_error_stash = gv_stashpv ("Types::Serialiser::Error" , 1);
847 1197
848 types_true = get_bool ("Types::Serialiser::true" ); 1198 types_true = get_bool ("Types::Serialiser::true" );
849 types_false = get_bool ("Types::Serialiser::false"); 1199 types_false = get_bool ("Types::Serialiser::false");
850 types_error = get_bool ("Types::Serialiser::error"); 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);
851} 1206}
852 1207
853PROTOTYPES: DISABLE 1208PROTOTYPES: DISABLE
854 1209
855void CLONE (...) 1210void CLONE (...)
873 1228
874void shrink (CBOR *self, int enable = 1) 1229void shrink (CBOR *self, int enable = 1)
875 ALIAS: 1230 ALIAS:
876 shrink = F_SHRINK 1231 shrink = F_SHRINK
877 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
878 PPCODE: 1237 PPCODE:
879{ 1238{
880 if (enable) 1239 if (enable)
881 self->flags |= ix; 1240 self->flags |= ix;
882 else 1241 else
887 1246
888void get_shrink (CBOR *self) 1247void get_shrink (CBOR *self)
889 ALIAS: 1248 ALIAS:
890 get_shrink = F_SHRINK 1249 get_shrink = F_SHRINK
891 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
892 PPCODE: 1255 PPCODE:
893 XPUSHs (boolSV (self->flags & ix)); 1256 XPUSHs (boolSV (self->flags & ix));
894 1257
895void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) 1258void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
896 PPCODE: 1259 PPCODE:
912 CODE: 1275 CODE:
913 RETVAL = self->max_size; 1276 RETVAL = self->max_size;
914 OUTPUT: 1277 OUTPUT:
915 RETVAL 1278 RETVAL
916 1279
917#if 0 //TODO 1280void filter (CBOR *self, SV *filter = 0)
918
919void filter_cbor_object (CBOR *self, SV *cb = &PL_sv_undef)
920 PPCODE: 1281 PPCODE:
921{
922 SvREFCNT_dec (self->cb_object); 1282 SvREFCNT_dec (self->filter);
923 self->cb_object = SvOK (cb) ? newSVsv (cb) : 0; 1283 self->filter = filter ? newSVsv (filter) : filter;
924
925 XPUSHs (ST (0)); 1284 XPUSHs (ST (0));
926}
927 1285
928void filter_cbor_single_key_object (CBOR *self, SV *key, SV *cb = &PL_sv_undef) 1286SV *get_filter (CBOR *self)
929 PPCODE: 1287 CODE:
930{ 1288 RETVAL = self->filter ? self->filter : NEWSV (0, 0);
931 if (!self->cb_sk_object) 1289 OUTPUT:
932 self->cb_sk_object = newHV (); 1290 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 1291
952void encode (CBOR *self, SV *scalar) 1292void encode (CBOR *self, SV *scalar)
953 PPCODE: 1293 PPCODE:
954 PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN; 1294 PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN;
955 XPUSHs (scalar); 1295 XPUSHs (scalar);
968 EXTEND (SP, 2); 1308 EXTEND (SP, 2);
969 PUSHs (sv); 1309 PUSHs (sv);
970 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); 1310 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
971} 1311}
972 1312
973#if 0
974
975void DESTROY (CBOR *self) 1313void DESTROY (CBOR *self)
976 CODE: 1314 PPCODE:
977 SvREFCNT_dec (self->cb_sk_object); 1315 cbor_free (self);
978 SvREFCNT_dec (self->cb_object);
979
980#endif
981 1316
982PROTOTYPES: ENABLE 1317PROTOTYPES: ENABLE
983 1318
984void encode_cbor (SV *scalar) 1319void encode_cbor (SV *scalar)
1320 ALIAS:
1321 encode_cbor = 0
1322 encode_cbor_sharing = F_ALLOW_SHARING
985 PPCODE: 1323 PPCODE:
986{ 1324{
987 CBOR cbor; 1325 CBOR cbor;
988 cbor_init (&cbor); 1326 cbor_init (&cbor);
1327 cbor.flags |= ix;
989 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN; 1328 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;
990 XPUSHs (scalar); 1329 XPUSHs (scalar);
991} 1330}
992 1331
993void decode_cbor (SV *cborstr) 1332void decode_cbor (SV *cborstr)

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines