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 1.37 by root, Sat Nov 30 18:13:53 2013 UTC

9#include <limits.h> 9#include <limits.h>
10#include <float.h> 10#include <float.h>
11 11
12#include "ecb.h" 12#include "ecb.h"
13 13
14// 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
45 104
46#define INIT_SIZE 32 // initial scalar size to be allocated 105#define INIT_SIZE 32 // initial scalar size to be allocated
47 106
48#define SB do { 107#define SB do {
49#define SE } while (0) 108#define SE } while (0)
61# define CBOR_SLOW 0 120# define CBOR_SLOW 0
62# define CBOR_STASH cbor_stash 121# define CBOR_STASH cbor_stash
63#endif 122#endif
64 123
65static HV *cbor_stash, *types_boolean_stash, *types_error_stash, *cbor_tagged_stash; // CBOR::XS:: 124static HV *cbor_stash, *types_boolean_stash, *types_error_stash, *cbor_tagged_stash; // CBOR::XS::
66static SV *types_true, *types_false, *types_error; 125static SV *types_true, *types_false, *types_error, *sv_cbor, *default_filter;
67 126
68typedef struct { 127typedef struct {
69 U32 flags; 128 U32 flags;
70 U32 max_depth; 129 U32 max_depth;
71 STRLEN max_size; 130 STRLEN max_size;
131 SV *filter;
72} CBOR; 132} CBOR;
73 133
74ecb_inline void 134ecb_inline void
75cbor_init (CBOR *cbor) 135cbor_init (CBOR *cbor)
76{ 136{
77 Zero (cbor, 1, CBOR); 137 Zero (cbor, 1, CBOR);
78 cbor->max_depth = 512; 138 cbor->max_depth = 512;
139}
140
141ecb_inline void
142cbor_free (CBOR *cbor)
143{
144 SvREFCNT_dec (cbor->filter);
79} 145}
80 146
81///////////////////////////////////////////////////////////////////////////// 147/////////////////////////////////////////////////////////////////////////////
82// utility functions 148// utility functions
83 149
105 SvPV_renew (sv, SvCUR (sv) + 1); 171 SvPV_renew (sv, SvCUR (sv) + 1);
106#endif 172#endif
107 } 173 }
108} 174}
109 175
110///////////////////////////////////////////////////////////////////////////// 176// minimum length of a string to be registered for stringref
111// fp hell 177ecb_inline int
112 178minimum_string_length (UV idx)
113//TODO 179{
180 return idx > 23
181 ? idx > 0xffU
182 ? idx > 0xffffU
183 ? idx > 0xffffffffU
184 ? 11
185 : 7
186 : 5
187 : 4
188 : 3;
189}
114 190
115///////////////////////////////////////////////////////////////////////////// 191/////////////////////////////////////////////////////////////////////////////
116// encoder 192// encoder
117 193
118// structure used for encoding CBOR 194// structure used for encoding CBOR
121 char *cur; // SvPVX (sv) + current output position 197 char *cur; // SvPVX (sv) + current output position
122 char *end; // SvEND (sv) 198 char *end; // SvEND (sv)
123 SV *sv; // result scalar 199 SV *sv; // result scalar
124 CBOR cbor; 200 CBOR cbor;
125 U32 depth; // recursion level 201 U32 depth; // recursion level
202 HV *stringref[2]; // string => index, or 0 ([0] = bytes, [1] = utf-8)
203 UV stringref_idx;
204 HV *shareable; // ptr => index, or 0
205 UV shareable_idx;
126} enc_t; 206} enc_t;
127 207
128ecb_inline void 208ecb_inline void
129need (enc_t *enc, STRLEN len) 209need (enc_t *enc, STRLEN len)
130{ 210{
147static void 227static void
148encode_uint (enc_t *enc, int major, UV len) 228encode_uint (enc_t *enc, int major, UV len)
149{ 229{
150 need (enc, 9); 230 need (enc, 9);
151 231
152 if (len < 24) 232 if (ecb_expect_true (len < LENGTH_EXT1))
153 *enc->cur++ = major | len; 233 *enc->cur++ = major | len;
154 else if (len <= 0xff) 234 else if (ecb_expect_true (len <= 0xffU))
155 { 235 {
156 *enc->cur++ = major | 24; 236 *enc->cur++ = major | LENGTH_EXT1;
157 *enc->cur++ = len; 237 *enc->cur++ = len;
158 } 238 }
159 else if (len <= 0xffff) 239 else if (len <= 0xffffU)
160 { 240 {
161 *enc->cur++ = major | 25; 241 *enc->cur++ = major | LENGTH_EXT2;
162 *enc->cur++ = len >> 8; 242 *enc->cur++ = len >> 8;
163 *enc->cur++ = len; 243 *enc->cur++ = len;
164 } 244 }
165 else if (len <= 0xffffffff) 245 else if (len <= 0xffffffffU)
166 { 246 {
167 *enc->cur++ = major | 26; 247 *enc->cur++ = major | LENGTH_EXT4;
168 *enc->cur++ = len >> 24; 248 *enc->cur++ = len >> 24;
169 *enc->cur++ = len >> 16; 249 *enc->cur++ = len >> 16;
170 *enc->cur++ = len >> 8; 250 *enc->cur++ = len >> 8;
171 *enc->cur++ = len; 251 *enc->cur++ = len;
172 } 252 }
173 else 253 else
174 { 254 {
175 *enc->cur++ = major | 27; 255 *enc->cur++ = major | LENGTH_EXT8;
176 *enc->cur++ = len >> 56; 256 *enc->cur++ = len >> 56;
177 *enc->cur++ = len >> 48; 257 *enc->cur++ = len >> 48;
178 *enc->cur++ = len >> 40; 258 *enc->cur++ = len >> 40;
179 *enc->cur++ = len >> 32; 259 *enc->cur++ = len >> 32;
180 *enc->cur++ = len >> 24; 260 *enc->cur++ = len >> 24;
182 *enc->cur++ = len >> 8; 262 *enc->cur++ = len >> 8;
183 *enc->cur++ = len; 263 *enc->cur++ = len;
184 } 264 }
185} 265}
186 266
187static void 267ecb_inline void
268encode_tag (enc_t *enc, UV tag)
269{
270 encode_uint (enc, MAJOR_TAG, tag);
271}
272
273ecb_inline void
188encode_str (enc_t *enc, int utf8, char *str, STRLEN len) 274encode_str (enc_t *enc, int utf8, char *str, STRLEN len)
189{ 275{
190 encode_uint (enc, utf8 ? 0x60 : 0x40, len); 276 encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len);
191 need (enc, len); 277 need (enc, len);
192 memcpy (enc->cur, str, len); 278 memcpy (enc->cur, str, len);
193 enc->cur += len; 279 enc->cur += len;
194} 280}
195 281
282static void
283encode_strref (enc_t *enc, int utf8, char *str, STRLEN len)
284{
285 if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS))
286 {
287 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1);
288
289 if (SvOK (*svp))
290 {
291 // already registered, use stringref
292 encode_tag (enc, CBOR_TAG_STRINGREF);
293 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
294 return;
295 }
296 else if (len >= minimum_string_length (enc->stringref_idx))
297 {
298 // register only
299 sv_setuv (*svp, enc->stringref_idx);
300 ++enc->stringref_idx;
301 }
302 }
303
304 encode_str (enc, utf8, str, len);
305}
306
196static void encode_sv (enc_t *enc, SV *sv); 307static void encode_sv (enc_t *enc, SV *sv);
197 308
198static void 309static void
199encode_av (enc_t *enc, AV *av) 310encode_av (enc_t *enc, AV *av)
200{ 311{
203 if (enc->depth >= enc->cbor.max_depth) 314 if (enc->depth >= enc->cbor.max_depth)
204 croak (ERR_NESTING_EXCEEDED); 315 croak (ERR_NESTING_EXCEEDED);
205 316
206 ++enc->depth; 317 ++enc->depth;
207 318
208 encode_uint (enc, 0x80, len + 1); 319 encode_uint (enc, MAJOR_ARRAY, len + 1);
209 320
210 for (i = 0; i <= len; ++i) 321 for (i = 0; i <= len; ++i)
211 { 322 {
212 SV **svp = av_fetch (av, i, 0); 323 SV **svp = av_fetch (av, i, 0);
213 encode_sv (enc, svp ? *svp : &PL_sv_undef); 324 encode_sv (enc, svp ? *svp : &PL_sv_undef);
228 339
229 int pairs = hv_iterinit (hv); 340 int pairs = hv_iterinit (hv);
230 int mg = SvMAGICAL (hv); 341 int mg = SvMAGICAL (hv);
231 342
232 if (mg) 343 if (mg)
233 encode_ch (enc, 0xa0 | 31); 344 encode_ch (enc, MAJOR_MAP | MINOR_INDEF);
234 else 345 else
235 encode_uint (enc, 0xa0, pairs); 346 encode_uint (enc, MAJOR_MAP, pairs);
236 347
237 while ((he = hv_iternext (hv))) 348 while ((he = hv_iternext (hv)))
238 { 349 {
239 if (HeKLEN (he) == HEf_SVKEY) 350 if (HeKLEN (he) == HEf_SVKEY)
240 encode_sv (enc, HeSVKEY (he)); 351 encode_sv (enc, HeSVKEY (he));
241 else 352 else
242 encode_str (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he)); 353 encode_strref (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he));
243 354
244 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he)); 355 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he));
245 } 356 }
246 357
247 if (mg) 358 if (mg)
248 encode_ch (enc, 0xe0 | 31); 359 encode_ch (enc, MAJOR_MISC | MINOR_INDEF);
249 360
250 --enc->depth; 361 --enc->depth;
251} 362}
252 363
253// encode objects, arrays and special \0=false and \1=true values. 364// encode objects, arrays and special \0=false and \1=true values.
254static void 365static void
255encode_rv (enc_t *enc, SV *sv) 366encode_rv (enc_t *enc, SV *sv)
256{ 367{
257 svtype svt;
258
259 SvGETMAGIC (sv); 368 SvGETMAGIC (sv);
369
260 svt = SvTYPE (sv); 370 svtype svt = SvTYPE (sv);
261 371
262 if (ecb_expect_false (SvOBJECT (sv))) 372 if (ecb_expect_false (SvOBJECT (sv)))
263 { 373 {
264 HV *boolean_stash = !CBOR_SLOW || types_boolean_stash 374 HV *boolean_stash = !CBOR_SLOW || types_boolean_stash
265 ? types_boolean_stash 375 ? types_boolean_stash
269 : gv_stashpv ("Types::Serialiser::Error", 1); 379 : gv_stashpv ("Types::Serialiser::Error", 1);
270 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 380 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
271 ? cbor_tagged_stash 381 ? cbor_tagged_stash
272 : gv_stashpv ("CBOR::XS::Tagged" , 1); 382 : gv_stashpv ("CBOR::XS::Tagged" , 1);
273 383
384 HV *stash = SvSTASH (sv);
385
274 if (SvSTASH (sv) == boolean_stash) 386 if (stash == boolean_stash)
275 encode_ch (enc, SvIV (sv) ? 0xe0 | 21 : 0xe0 | 20); 387 {
388 encode_ch (enc, SvIV (sv) ? MAJOR_MISC | SIMPLE_TRUE : MAJOR_MISC | SIMPLE_FALSE);
389 return;
390 }
276 else if (SvSTASH (sv) == error_stash) 391 else if (stash == error_stash)
277 encode_ch (enc, 0xe0 | 23); 392 {
393 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
394 return;
395 }
278 else if (SvSTASH (sv) == tagged_stash) 396 else if (stash == tagged_stash)
279 { 397 {
280 if (svt != SVt_PVAV) 398 if (svt != SVt_PVAV)
281 croak ("encountered CBOR::XS::Tagged object that isn't an array"); 399 croak ("encountered CBOR::XS::Tagged object that isn't an array");
282 400
283 encode_uint (enc, 0xc0, SvUV (*av_fetch ((AV *)sv, 0, 1))); 401 encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1)));
284 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1)); 402 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1));
403
404 return;
405 }
406 }
407
408 if (ecb_expect_false (SvREFCNT (sv) > 1)
409 && ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING))
410 {
411 if (!enc->shareable)
412 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
413
414 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
415
416 if (SvOK (*svp))
417 {
418 encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF);
419 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
420 return;
285 } 421 }
286 else 422 else
287 { 423 {
424 sv_setuv (*svp, enc->shareable_idx);
425 ++enc->shareable_idx;
426 encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE);
427 }
428 }
429
430 if (ecb_expect_false (SvOBJECT (sv)))
431 {
432 HV *stash = SvSTASH (sv);
433 GV *method;
434
435 if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
436 {
437 dSP;
438
439 ENTER; SAVETMPS; PUSHMARK (SP);
288 // we re-bless the reference to get overload and other niceties right 440 // we re-bless the reference to get overload and other niceties right
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))); 441 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
297 442
298 // calling with G_SCALAR ensures that we always get a 1 return value
299 PUTBACK; 443 PUTBACK;
444 // G_SCALAR ensures that return value is 1
300 call_sv ((SV *)GvCV (to_cbor), G_SCALAR); 445 call_sv ((SV *)GvCV (method), G_SCALAR);
301 SPAGAIN; 446 SPAGAIN;
302 447
303 // catch this surprisingly common error 448 // catch this surprisingly common error
304 if (SvROK (TOPs) && SvRV (TOPs) == sv) 449 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))); 450 croak ("%s::TO_CBOR method returned same object as was passed instead of a new one", HvNAME (stash));
306 451
307 sv = POPs; 452 encode_sv (enc, POPs);
453
308 PUTBACK; 454 PUTBACK;
309 455
310 encode_sv (enc, sv);
311
312 FREETMPS; LEAVE; 456 FREETMPS; LEAVE;
313 } 457 }
458 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)
459 {
460 dSP;
461
462 ENTER; SAVETMPS; PUSHMARK (SP);
463 EXTEND (SP, 2);
464 // we re-bless the reference to get overload and other niceties right
465 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
466 PUSHs (sv_cbor);
467
468 PUTBACK;
469 int count = call_sv ((SV *)GvCV (method), G_ARRAY);
470 SPAGAIN;
471
472 // catch this surprisingly common error
473 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)
474 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));
475
476 encode_tag (enc, CBOR_TAG_PERL_OBJECT);
477 encode_uint (enc, MAJOR_ARRAY, count + 1);
478 encode_strref (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
479
480 while (count)
481 encode_sv (enc, SP[1 - count--]);
482
483 PUTBACK;
484
485 FREETMPS; LEAVE;
486 }
314 else 487 else
315 croak ("encountered object '%s', but no TO_CBOR method available on it", 488 croak ("encountered object '%s', but no TO_CBOR or FREEZE methods available on it",
316 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 489 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
317 }
318 } 490 }
319 else if (svt == SVt_PVHV) 491 else if (svt == SVt_PVHV)
320 encode_hv (enc, (HV *)sv); 492 encode_hv (enc, (HV *)sv);
321 else if (svt == SVt_PVAV) 493 else if (svt == SVt_PVAV)
322 encode_av (enc, (AV *)sv); 494 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 495 else
341 croak ("encountered %s, but CBOR can only represent references to arrays or hashes", 496 {
342 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 497 encode_tag (enc, CBOR_TAG_INDIRECTION);
498 encode_sv (enc, sv);
499 }
343} 500}
344 501
345static void 502static void
346encode_nv (enc_t *enc, SV *sv) 503encode_nv (enc_t *enc, SV *sv)
347{ 504{
348 double nv = SvNVX (sv); 505 double nv = SvNVX (sv);
349 506
350 need (enc, 9); 507 need (enc, 9);
351 508
352 if (ecb_expect_false (nv == (U32)nv)) 509 if (ecb_expect_false (nv == (NV)(U32)nv))
353 encode_uint (enc, 0x00, (U32)nv); 510 encode_uint (enc, MAJOR_POS_INT, (U32)nv);
354 //TODO: maybe I32? 511 //TODO: maybe I32?
355 else if (ecb_expect_false (nv == (float)nv)) 512 else if (ecb_expect_false (nv == (float)nv))
356 { 513 {
357 uint32_t fp = ecb_float_to_binary32 (nv); 514 uint32_t fp = ecb_float_to_binary32 (nv);
358 515
359 *enc->cur++ = 0xe0 | 26; 516 *enc->cur++ = MAJOR_MISC | MISC_FLOAT32;
360 517
361 if (!ecb_big_endian ()) 518 if (!ecb_big_endian ())
362 fp = ecb_bswap32 (fp); 519 fp = ecb_bswap32 (fp);
363 520
364 memcpy (enc->cur, &fp, 4); 521 memcpy (enc->cur, &fp, 4);
366 } 523 }
367 else 524 else
368 { 525 {
369 uint64_t fp = ecb_double_to_binary64 (nv); 526 uint64_t fp = ecb_double_to_binary64 (nv);
370 527
371 *enc->cur++ = 0xe0 | 27; 528 *enc->cur++ = MAJOR_MISC | MISC_FLOAT64;
372 529
373 if (!ecb_big_endian ()) 530 if (!ecb_big_endian ())
374 fp = ecb_bswap64 (fp); 531 fp = ecb_bswap64 (fp);
375 532
376 memcpy (enc->cur, &fp, 8); 533 memcpy (enc->cur, &fp, 8);
385 542
386 if (SvPOKp (sv)) 543 if (SvPOKp (sv))
387 { 544 {
388 STRLEN len; 545 STRLEN len;
389 char *str = SvPV (sv, len); 546 char *str = SvPV (sv, len);
390 encode_str (enc, SvUTF8 (sv), str, len); 547 encode_strref (enc, SvUTF8 (sv), str, len);
391 } 548 }
392 else if (SvNOKp (sv)) 549 else if (SvNOKp (sv))
393 encode_nv (enc, sv); 550 encode_nv (enc, sv);
394 else if (SvIOKp (sv)) 551 else if (SvIOKp (sv))
395 { 552 {
396 if (SvIsUV (sv)) 553 if (SvIsUV (sv))
397 encode_uint (enc, 0x00, SvUVX (sv)); 554 encode_uint (enc, MAJOR_POS_INT, SvUVX (sv));
398 else if (SvIVX (sv) >= 0) 555 else if (SvIVX (sv) >= 0)
399 encode_uint (enc, 0x00, SvIVX (sv)); 556 encode_uint (enc, MAJOR_POS_INT, SvIVX (sv));
400 else 557 else
401 encode_uint (enc, 0x20, -(SvIVX (sv) + 1)); 558 encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1));
402 } 559 }
403 else if (SvROK (sv)) 560 else if (SvROK (sv))
404 encode_rv (enc, SvRV (sv)); 561 encode_rv (enc, SvRV (sv));
405 else if (!SvOK (sv)) 562 else if (!SvOK (sv))
406 encode_ch (enc, 0xe0 | 22); 563 encode_ch (enc, MAJOR_MISC | SIMPLE_NULL);
407 else if (enc->cbor.flags & F_ALLOW_UNKNOWN) 564 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
408 encode_ch (enc, 0xe0 | 23); 565 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
409 else 566 else
410 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data", 567 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data",
411 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv)); 568 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
412} 569}
413 570
414static SV * 571static SV *
415encode_cbor (SV *scalar, CBOR *cbor) 572encode_cbor (SV *scalar, CBOR *cbor)
416{ 573{
417 enc_t enc; 574 enc_t enc = { };
418 575
419 enc.cbor = *cbor; 576 enc.cbor = *cbor;
420 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 577 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
421 enc.cur = SvPVX (enc.sv); 578 enc.cur = SvPVX (enc.sv);
422 enc.end = SvEND (enc.sv); 579 enc.end = SvEND (enc.sv);
423 enc.depth = 0;
424 580
425 SvPOK_only (enc.sv); 581 SvPOK_only (enc.sv);
582
583 if (cbor->flags & F_PACK_STRINGS)
584 {
585 encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE);
586 enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ());
587 enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ());
588 }
589
426 encode_sv (&enc, scalar); 590 encode_sv (&enc, scalar);
427 591
428 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 592 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
429 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings 593 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
430 594
444 U8 *end; // end of input string 608 U8 *end; // end of input string
445 const char *err; // parse error, if != 0 609 const char *err; // parse error, if != 0
446 CBOR cbor; 610 CBOR cbor;
447 U32 depth; // recursion depth 611 U32 depth; // recursion depth
448 U32 maxdepth; // recursion depth limit 612 U32 maxdepth; // recursion depth limit
613 AV *shareable;
614 AV *stringref;
615 SV *decode_tagged;
449} dec_t; 616} dec_t;
450 617
451#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE 618#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE
452 619
453#define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data") 620#define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data")
456#define DEC_DEC_DEPTH --dec->depth 623#define DEC_DEC_DEPTH --dec->depth
457 624
458static UV 625static UV
459decode_uint (dec_t *dec) 626decode_uint (dec_t *dec)
460{ 627{
461 switch (*dec->cur & 31) 628 U8 m = *dec->cur & MINOR_MASK;
462 { 629 ++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;
467 630
468 case 24: 631 if (ecb_expect_true (m < LENGTH_EXT1))
632 return m;
633 else if (ecb_expect_true (m == LENGTH_EXT1))
634 {
469 WANT (2); 635 WANT (1);
470 dec->cur += 2; 636 dec->cur += 1;
471 return dec->cur[-1]; 637 return dec->cur[-1];
472 638 }
473 case 25: 639 else if (ecb_expect_true (m == LENGTH_EXT2))
640 {
474 WANT (3); 641 WANT (2);
475 dec->cur += 3; 642 dec->cur += 2;
476 return (((UV)dec->cur[-2]) << 8) 643 return (((UV)dec->cur[-2]) << 8)
477 | ((UV)dec->cur[-1]); 644 | ((UV)dec->cur[-1]);
478 645 }
479 case 26: 646 else if (ecb_expect_true (m == LENGTH_EXT4))
647 {
480 WANT (5); 648 WANT (4);
481 dec->cur += 5; 649 dec->cur += 4;
482 return (((UV)dec->cur[-4]) << 24) 650 return (((UV)dec->cur[-4]) << 24)
483 | (((UV)dec->cur[-3]) << 16) 651 | (((UV)dec->cur[-3]) << 16)
484 | (((UV)dec->cur[-2]) << 8) 652 | (((UV)dec->cur[-2]) << 8)
485 | ((UV)dec->cur[-1]); 653 | ((UV)dec->cur[-1]);
486 654 }
487 case 27: 655 else if (ecb_expect_true (m == LENGTH_EXT8))
656 {
488 WANT (9); 657 WANT (8);
489 dec->cur += 9; 658 dec->cur += 8;
659
660 return
661#if UVSIZE < 8
662 0
663#else
490 return (((UV)dec->cur[-8]) << 56) 664 (((UV)dec->cur[-8]) << 56)
491 | (((UV)dec->cur[-7]) << 48) 665 | (((UV)dec->cur[-7]) << 48)
492 | (((UV)dec->cur[-6]) << 40) 666 | (((UV)dec->cur[-6]) << 40)
493 | (((UV)dec->cur[-5]) << 32) 667 | (((UV)dec->cur[-5]) << 32)
668#endif
494 | (((UV)dec->cur[-4]) << 24) 669 | (((UV)dec->cur[-4]) << 24)
495 | (((UV)dec->cur[-3]) << 16) 670 | (((UV)dec->cur[-3]) << 16)
496 | (((UV)dec->cur[-2]) << 8) 671 | (((UV)dec->cur[-2]) << 8)
497 | ((UV)dec->cur[-1]); 672 | ((UV)dec->cur[-1]);
498 673 }
499 default: 674 else
500 ERR ("corrupted CBOR data (unsupported integer minor encoding)"); 675 ERR ("corrupted CBOR data (unsupported integer minor encoding)");
501 }
502 676
503fail: 677fail:
504 return 0; 678 return 0;
505} 679}
506 680
511{ 685{
512 AV *av = newAV (); 686 AV *av = newAV ();
513 687
514 DEC_INC_DEPTH; 688 DEC_INC_DEPTH;
515 689
516 if ((*dec->cur & 31) == 31) 690 if (*dec->cur == (MAJOR_ARRAY | MINOR_INDEF))
517 { 691 {
518 ++dec->cur; 692 ++dec->cur;
519 693
520 for (;;) 694 for (;;)
521 { 695 {
522 WANT (1); 696 WANT (1);
523 697
524 if (*dec->cur == (0xe0 | 31)) 698 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
525 { 699 {
526 ++dec->cur; 700 ++dec->cur;
527 break; 701 break;
528 } 702 }
529 703
532 } 706 }
533 else 707 else
534 { 708 {
535 int i, len = decode_uint (dec); 709 int i, len = decode_uint (dec);
536 710
711 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); 712 av_fill (av, len - 1);
538 713
539 for (i = 0; i < len; ++i) 714 for (i = 0; i < len; ++i)
540 AvARRAY (av)[i] = decode_sv (dec); 715 AvARRAY (av)[i] = decode_sv (dec);
541 } 716 }
547 SvREFCNT_dec (av); 722 SvREFCNT_dec (av);
548 DEC_DEC_DEPTH; 723 DEC_DEC_DEPTH;
549 return &PL_sv_undef; 724 return &PL_sv_undef;
550} 725}
551 726
727static void
728decode_he (dec_t *dec, HV *hv)
729{
730 // for speed reasons, we specialcase single-string
731 // byte or utf-8 strings as keys, but only when !stringref
732
733 if (ecb_expect_true (!dec->stringref))
734 if (ecb_expect_true ((*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8))
735 {
736 I32 len = decode_uint (dec);
737 char *key = (char *)dec->cur;
738
739 dec->cur += len;
740
741 if (ecb_expect_false (dec->stringref))
742 av_push (dec->stringref, newSVpvn (key, len));
743
744 hv_store (hv, key, len, decode_sv (dec), 0);
745
746 return;
747 }
748 else if (ecb_expect_true ((*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8))
749 {
750 I32 len = decode_uint (dec);
751 char *key = (char *)dec->cur;
752
753 dec->cur += len;
754
755 if (ecb_expect_false (dec->stringref))
756 av_push (dec->stringref, newSVpvn_utf8 (key, len, 1));
757
758 hv_store (hv, key, -len, decode_sv (dec), 0);
759
760 return;
761 }
762
763 SV *k = decode_sv (dec);
764 SV *v = decode_sv (dec);
765
766 hv_store_ent (hv, k, v, 0);
767 SvREFCNT_dec (k);
768}
769
552static SV * 770static SV *
553decode_hv (dec_t *dec) 771decode_hv (dec_t *dec)
554{ 772{
555 HV *hv = newHV (); 773 HV *hv = newHV ();
556 774
557 DEC_INC_DEPTH; 775 DEC_INC_DEPTH;
558 776
559 if ((*dec->cur & 31) == 31) 777 if (*dec->cur == (MAJOR_MAP | MINOR_INDEF))
560 { 778 {
561 ++dec->cur; 779 ++dec->cur;
562 780
563 for (;;) 781 for (;;)
564 { 782 {
565 WANT (1); 783 WANT (1);
566 784
567 if (*dec->cur == (0xe0 | 31)) 785 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
568 { 786 {
569 ++dec->cur; 787 ++dec->cur;
570 break; 788 break;
571 } 789 }
572 790
573 SV *k = decode_sv (dec); 791 decode_he (dec, hv);
574 SV *v = decode_sv (dec);
575
576 hv_store_ent (hv, k, v, 0);
577 } 792 }
578 } 793 }
579 else 794 else
580 { 795 {
581 int len = decode_uint (dec); 796 int pairs = decode_uint (dec);
582 797
583 while (len--) 798 while (pairs--)
584 { 799 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 } 800 }
591 801
592 DEC_DEC_DEPTH; 802 DEC_DEC_DEPTH;
593 return newRV_noinc ((SV *)hv); 803 return newRV_noinc ((SV *)hv);
594 804
601static SV * 811static SV *
602decode_str (dec_t *dec, int utf8) 812decode_str (dec_t *dec, int utf8)
603{ 813{
604 SV *sv = 0; 814 SV *sv = 0;
605 815
606 if ((*dec->cur & 31) == 31) 816 if ((*dec->cur & MINOR_MASK) == MINOR_INDEF)
607 { 817 {
818 // indefinite length strings
608 ++dec->cur; 819 ++dec->cur;
609 820
821 U8 major = *dec->cur & MAJOR_MISC;
822
610 sv = newSVpvn ("", 0); 823 sv = newSVpvn ("", 0);
611 824
612 // not very fast, and certainly not robust against illegal input
613 for (;;) 825 for (;;)
614 { 826 {
615 WANT (1); 827 WANT (1);
616 828
617 if (*dec->cur == (0xe0 | 31)) 829 if ((*dec->cur - major) > LENGTH_EXT8)
830 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
618 { 831 {
619 ++dec->cur; 832 ++dec->cur;
620 break; 833 break;
621 } 834 }
835 else
836 ERR ("corrupted CBOR data (invalid chunks in indefinite length string)");
622 837
623 sv_catsv (sv, decode_sv (dec)); 838 STRLEN len = decode_uint (dec);
839
840 WANT (len);
841 sv_catpvn (sv, dec->cur, len);
842 dec->cur += len;
624 } 843 }
625 } 844 }
626 else 845 else
627 { 846 {
628 STRLEN len = decode_uint (dec); 847 STRLEN len = decode_uint (dec);
629 848
630 WANT (len); 849 WANT (len);
631 sv = newSVpvn (dec->cur, len); 850 sv = newSVpvn (dec->cur, len);
632 dec->cur += len; 851 dec->cur += len;
852
853 if (ecb_expect_false (dec->stringref)
854 && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1))
855 av_push (dec->stringref, SvREFCNT_inc_NN (sv));
633 } 856 }
634 857
635 if (utf8) 858 if (utf8)
636 SvUTF8_on (sv); 859 SvUTF8_on (sv);
637 860
643} 866}
644 867
645static SV * 868static SV *
646decode_tagged (dec_t *dec) 869decode_tagged (dec_t *dec)
647{ 870{
871 SV *sv = 0;
648 UV tag = decode_uint (dec); 872 UV tag = decode_uint (dec);
873
874 WANT (1);
875
876 switch (tag)
877 {
878 case CBOR_TAG_MAGIC:
649 SV *sv = decode_sv (dec); 879 sv = decode_sv (dec);
880 break;
650 881
651 if (tag == CBOR_TAG_MAGIC) 882 case CBOR_TAG_INDIRECTION:
652 return sv; 883 sv = newRV_noinc (decode_sv (dec));
884 break;
653 885
654 if (tag == CBOR_TAG_PERL_OBJECT) 886 case CBOR_TAG_STRINGREF_NAMESPACE:
655 { 887 {
888 ENTER; SAVETMPS;
889
890 SAVESPTR (dec->stringref);
891 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ());
892
893 sv = decode_sv (dec);
894
895 FREETMPS; LEAVE;
896 }
897 break;
898
899 case CBOR_TAG_STRINGREF:
900 {
901 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
902 ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
903
904 UV idx = decode_uint (dec);
905
906 if (!dec->stringref || (int)idx > AvFILLp (dec->stringref))
907 ERR ("corrupted CBOR data (stringref index out of bounds or outside namespace)");
908
909 sv = newSVsv (AvARRAY (dec->stringref)[idx]);
910 }
911 break;
912
913 case CBOR_TAG_VALUE_SHAREABLE:
914 {
915 if (ecb_expect_false (!dec->shareable))
916 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ());
917
918 if (dec->cbor.flags & F_ALLOW_CYCLES)
919 {
920 sv = newSV (0);
921 av_push (dec->shareable, SvREFCNT_inc_NN (sv));
922
923 SV *osv = decode_sv (dec);
924 sv_setsv (sv, osv);
925 SvREFCNT_dec_NN (osv);
926 }
927 else
928 {
929 av_push (dec->shareable, &PL_sv_undef);
930 int idx = AvFILLp (dec->shareable);
931 sv = decode_sv (dec);
932 av_store (dec->shareable, idx, SvREFCNT_inc_NN (sv));
933 }
934 }
935 break;
936
937 case CBOR_TAG_VALUE_SHAREDREF:
938 {
939 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
940 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
941
942 UV idx = decode_uint (dec);
943
944 if (!dec->shareable || (int)idx > AvFILLp (dec->shareable))
945 ERR ("corrupted CBOR data (sharedref index out of bounds)");
946
947 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]);
948
949 if (sv == &PL_sv_undef)
950 ERR ("cyclic CBOR data structure found, but allow_cycles is not enabled");
951 }
952 break;
953
954 case CBOR_TAG_PERL_OBJECT:
955 {
956 sv = decode_sv (dec);
957
656 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV) 958 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
657 ERR ("corrupted CBOR data (non-array perl object)"); 959 ERR ("corrupted CBOR data (non-array perl object)");
960
961 AV *av = (AV *)SvRV (sv);
962 int len = av_len (av) + 1;
963 HV *stash = gv_stashsv (*av_fetch (av, 0, 1), 0);
964
965 if (!stash)
966 ERR ("cannot decode perl-object (package does not exist)");
967
968 GV *method = gv_fetchmethod_autoload (stash, "THAW", 0);
969
970 if (!method)
971 ERR ("cannot decode perl-object (package does not have a THAW method)");
972
973 dSP;
974
975 ENTER; SAVETMPS; PUSHMARK (SP);
976 EXTEND (SP, len + 1);
977 // we re-bless the reference to get overload and other niceties right
978 PUSHs (*av_fetch (av, 0, 1));
979 PUSHs (sv_cbor);
980
981 int i;
982
983 for (i = 1; i < len; ++i)
984 PUSHs (*av_fetch (av, i, 1));
985
986 PUTBACK;
987 call_sv ((SV *)GvCV (method), G_SCALAR | G_EVAL);
988 SPAGAIN;
989
990 if (SvTRUE (ERRSV))
991 {
992 FREETMPS; LEAVE;
993 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV))));
994 }
995
996 SvREFCNT_dec (sv);
997 sv = SvREFCNT_inc (POPs);
998
999 PUTBACK;
1000
1001 FREETMPS; LEAVE;
658 1002 }
659 // TODO 1003 break;
660 }
661 1004
1005 default:
1006 {
1007 sv = decode_sv (dec);
1008
1009 dSP;
1010 ENTER; SAVETMPS; PUSHMARK (SP);
1011 EXTEND (SP, 2);
1012 PUSHs (newSVuv (tag));
1013 PUSHs (sv);
1014
1015 PUTBACK;
1016 int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL);
1017 SPAGAIN;
1018
1019 if (SvTRUE (ERRSV))
1020 {
1021 FREETMPS; LEAVE;
1022 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV))));
1023 }
1024
1025 if (count)
1026 {
1027 SvREFCNT_dec (sv);
1028 sv = SvREFCNT_inc (POPs);
1029 }
1030 else
1031 {
662 AV *av = newAV (); 1032 AV *av = newAV ();
663 av_push (av, newSVuv (tag)); 1033 av_push (av, newSVuv (tag));
664 av_push (av, sv); 1034 av_push (av, sv);
665 1035
666 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 1036 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
667 ? cbor_tagged_stash 1037 ? cbor_tagged_stash
668 : gv_stashpv ("CBOR::XS::Tagged" , 1); 1038 : gv_stashpv ("CBOR::XS::Tagged" , 1);
669
670 return sv_bless (newRV_noinc ((SV *)av), tagged_stash); 1039 sv = sv_bless (newRV_noinc ((SV *)av), tagged_stash);
1040 }
1041
1042 PUTBACK;
1043
1044 FREETMPS; LEAVE;
1045 }
1046 break;
1047 }
1048
1049 return sv;
671 1050
672fail: 1051fail:
673 SvREFCNT_dec (sv); 1052 SvREFCNT_dec (sv);
674 return &PL_sv_undef; 1053 return &PL_sv_undef;
675} 1054}
677static SV * 1056static SV *
678decode_sv (dec_t *dec) 1057decode_sv (dec_t *dec)
679{ 1058{
680 WANT (1); 1059 WANT (1);
681 1060
682 switch (*dec->cur >> 5) 1061 switch (*dec->cur >> MAJOR_SHIFT)
683 { 1062 {
684 case 0: // unsigned int 1063 case MAJOR_POS_INT >> MAJOR_SHIFT: return newSVuv (decode_uint (dec));
685 return newSVuv (decode_uint (dec)); 1064 case MAJOR_NEG_INT >> MAJOR_SHIFT: return newSViv (-1 - (IV)decode_uint (dec));
686 case 1: // negative int 1065 case MAJOR_BYTES >> MAJOR_SHIFT: return decode_str (dec, 0);
687 return newSViv (-1 - (IV)decode_uint (dec)); 1066 case MAJOR_TEXT >> MAJOR_SHIFT: return decode_str (dec, 1);
688 case 2: // octet string 1067 case MAJOR_ARRAY >> MAJOR_SHIFT: return decode_av (dec);
689 return decode_str (dec, 0); 1068 case MAJOR_MAP >> MAJOR_SHIFT: return decode_hv (dec);
690 case 3: // utf-8 string 1069 case MAJOR_TAG >> MAJOR_SHIFT: return decode_tagged (dec);
691 return decode_str (dec, 1); 1070
692 case 4: // array 1071 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) 1072 switch (*dec->cur++ & MINOR_MASK)
700 { 1073 {
701 case 20: 1074 case SIMPLE_FALSE:
702#if CBOR_SLOW 1075#if CBOR_SLOW
703 types_false = get_bool ("Types::Serialiser::false"); 1076 types_false = get_bool ("Types::Serialiser::false");
704#endif 1077#endif
705 return newSVsv (types_false); 1078 return newSVsv (types_false);
706 case 21: 1079 case SIMPLE_TRUE:
707#if CBOR_SLOW 1080#if CBOR_SLOW
708 types_true = get_bool ("Types::Serialiser::true"); 1081 types_true = get_bool ("Types::Serialiser::true");
709#endif 1082#endif
710 return newSVsv (types_true); 1083 return newSVsv (types_true);
711 case 22: 1084 case SIMPLE_NULL:
712 return newSVsv (&PL_sv_undef); 1085 return newSVsv (&PL_sv_undef);
713 case 23: 1086 case SIMPLE_UNDEF:
714#if CBOR_SLOW 1087#if CBOR_SLOW
715 types_error = get_bool ("Types::Serialiser::error"); 1088 types_error = get_bool ("Types::Serialiser::error");
716#endif 1089#endif
717 return newSVsv (types_error); 1090 return newSVsv (types_error);
718 1091
719 case 25: 1092 case MISC_FLOAT16:
720 { 1093 {
721 WANT (2); 1094 WANT (2);
722 1095
723 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1]; 1096 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1];
724 dec->cur += 2; 1097 dec->cur += 2;
725 1098
726 return newSVnv (ecb_binary16_to_float (fp)); 1099 return newSVnv (ecb_binary16_to_float (fp));
727 } 1100 }
728 1101
729 case 26: 1102 case MISC_FLOAT32:
730 { 1103 {
731 uint32_t fp; 1104 uint32_t fp;
732 WANT (4); 1105 WANT (4);
733 memcpy (&fp, dec->cur, 4); 1106 memcpy (&fp, dec->cur, 4);
734 dec->cur += 4; 1107 dec->cur += 4;
737 fp = ecb_bswap32 (fp); 1110 fp = ecb_bswap32 (fp);
738 1111
739 return newSVnv (ecb_binary32_to_float (fp)); 1112 return newSVnv (ecb_binary32_to_float (fp));
740 } 1113 }
741 1114
742 case 27: 1115 case MISC_FLOAT64:
743 { 1116 {
744 uint64_t fp; 1117 uint64_t fp;
745 WANT (8); 1118 WANT (8);
746 memcpy (&fp, dec->cur, 8); 1119 memcpy (&fp, dec->cur, 8);
747 dec->cur += 8; 1120 dec->cur += 8;
750 fp = ecb_bswap64 (fp); 1123 fp = ecb_bswap64 (fp);
751 1124
752 return newSVnv (ecb_binary64_to_double (fp)); 1125 return newSVnv (ecb_binary64_to_double (fp));
753 } 1126 }
754 1127
755 // 0..19 unassigned 1128 // 0..19 unassigned simple
756 // 24 reserved + unassigned (reserved values are not encodable) 1129 // 24 reserved + unassigned (reserved values are not encodable)
757 default: 1130 default:
758 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)"); 1131 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)");
759 } 1132 }
760 1133
766} 1139}
767 1140
768static SV * 1141static SV *
769decode_cbor (SV *string, CBOR *cbor, char **offset_return) 1142decode_cbor (SV *string, CBOR *cbor, char **offset_return)
770{ 1143{
771 dec_t dec; 1144 dec_t dec = { };
772 SV *sv; 1145 SV *sv;
1146 STRLEN len;
1147 char *data = SvPVbyte (string, len);
773 1148
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) 1149 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", 1150 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); 1151 (unsigned long)len, (unsigned long)cbor->max_size);
805 }
806
807 sv_utf8_downgrade (string, 0);
808 1152
809 dec.cbor = *cbor; 1153 dec.cbor = *cbor;
810 dec.cur = (U8 *)SvPVX (string); 1154 dec.cur = (U8 *)data;
811 dec.end = (U8 *)SvEND (string); 1155 dec.end = (U8 *)data + len;
812 dec.err = 0;
813 dec.depth = 0;
814 1156
815 sv = decode_sv (&dec); 1157 sv = decode_sv (&dec);
816 1158
817 if (offset_return) 1159 if (offset_return)
818 *offset_return = dec.cur; 1160 *offset_return = dec.cur;
822 dec.err = "garbage after CBOR object"; 1164 dec.err = "garbage after CBOR object";
823 1165
824 if (dec.err) 1166 if (dec.err)
825 { 1167 {
826 SvREFCNT_dec (sv); 1168 SvREFCNT_dec (sv);
827 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)SvPVX (string), (int)(uint8_t)*dec.cur); 1169 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur);
828 } 1170 }
829 1171
830 sv = sv_2mortal (sv); 1172 sv = sv_2mortal (sv);
831 1173
832 return sv; 1174 return sv;
846 types_error_stash = gv_stashpv ("Types::Serialiser::Error" , 1); 1188 types_error_stash = gv_stashpv ("Types::Serialiser::Error" , 1);
847 1189
848 types_true = get_bool ("Types::Serialiser::true" ); 1190 types_true = get_bool ("Types::Serialiser::true" );
849 types_false = get_bool ("Types::Serialiser::false"); 1191 types_false = get_bool ("Types::Serialiser::false");
850 types_error = get_bool ("Types::Serialiser::error"); 1192 types_error = get_bool ("Types::Serialiser::error");
1193
1194 default_filter = newSVpv ("CBOR::XS::default_filter", 0);
1195
1196 sv_cbor = newSVpv ("CBOR", 0);
1197 SvREADONLY_on (sv_cbor);
851} 1198}
852 1199
853PROTOTYPES: DISABLE 1200PROTOTYPES: DISABLE
854 1201
855void CLONE (...) 1202void CLONE (...)
873 1220
874void shrink (CBOR *self, int enable = 1) 1221void shrink (CBOR *self, int enable = 1)
875 ALIAS: 1222 ALIAS:
876 shrink = F_SHRINK 1223 shrink = F_SHRINK
877 allow_unknown = F_ALLOW_UNKNOWN 1224 allow_unknown = F_ALLOW_UNKNOWN
1225 allow_sharing = F_ALLOW_SHARING
1226 allow_cycles = F_ALLOW_CYCLES
1227 pack_strings = F_PACK_STRINGS
878 PPCODE: 1228 PPCODE:
879{ 1229{
880 if (enable) 1230 if (enable)
881 self->flags |= ix; 1231 self->flags |= ix;
882 else 1232 else
887 1237
888void get_shrink (CBOR *self) 1238void get_shrink (CBOR *self)
889 ALIAS: 1239 ALIAS:
890 get_shrink = F_SHRINK 1240 get_shrink = F_SHRINK
891 get_allow_unknown = F_ALLOW_UNKNOWN 1241 get_allow_unknown = F_ALLOW_UNKNOWN
1242 get_allow_sharing = F_ALLOW_SHARING
1243 get_allow_cycles = F_ALLOW_CYCLES
1244 get_pack_strings = F_PACK_STRINGS
892 PPCODE: 1245 PPCODE:
893 XPUSHs (boolSV (self->flags & ix)); 1246 XPUSHs (boolSV (self->flags & ix));
894 1247
895void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) 1248void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
896 PPCODE: 1249 PPCODE:
912 CODE: 1265 CODE:
913 RETVAL = self->max_size; 1266 RETVAL = self->max_size;
914 OUTPUT: 1267 OUTPUT:
915 RETVAL 1268 RETVAL
916 1269
917#if 0 //TODO 1270void filter (CBOR *self, SV *filter = 0)
918
919void filter_cbor_object (CBOR *self, SV *cb = &PL_sv_undef)
920 PPCODE: 1271 PPCODE:
921{
922 SvREFCNT_dec (self->cb_object); 1272 SvREFCNT_dec (self->filter);
923 self->cb_object = SvOK (cb) ? newSVsv (cb) : 0; 1273 self->filter = filter ? newSVsv (filter) : filter;
924
925 XPUSHs (ST (0)); 1274 XPUSHs (ST (0));
926}
927 1275
928void filter_cbor_single_key_object (CBOR *self, SV *key, SV *cb = &PL_sv_undef) 1276SV *get_filter (CBOR *self)
929 PPCODE: 1277 CODE:
930{ 1278 RETVAL = self->filter ? self->filter : NEWSV (0, 0);
931 if (!self->cb_sk_object) 1279 OUTPUT:
932 self->cb_sk_object = newHV (); 1280 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 1281
952void encode (CBOR *self, SV *scalar) 1282void encode (CBOR *self, SV *scalar)
953 PPCODE: 1283 PPCODE:
954 PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN; 1284 PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN;
955 XPUSHs (scalar); 1285 XPUSHs (scalar);
968 EXTEND (SP, 2); 1298 EXTEND (SP, 2);
969 PUSHs (sv); 1299 PUSHs (sv);
970 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); 1300 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
971} 1301}
972 1302
973#if 0
974
975void DESTROY (CBOR *self) 1303void DESTROY (CBOR *self)
976 CODE: 1304 PPCODE:
977 SvREFCNT_dec (self->cb_sk_object); 1305 cbor_free (self);
978 SvREFCNT_dec (self->cb_object);
979
980#endif
981 1306
982PROTOTYPES: ENABLE 1307PROTOTYPES: ENABLE
983 1308
984void encode_cbor (SV *scalar) 1309void encode_cbor (SV *scalar)
1310 ALIAS:
1311 encode_cbor = 0
1312 encode_cbor_sharing = F_ALLOW_SHARING
985 PPCODE: 1313 PPCODE:
986{ 1314{
987 CBOR cbor; 1315 CBOR cbor;
988 cbor_init (&cbor); 1316 cbor_init (&cbor);
1317 cbor.flags |= ix;
989 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN; 1318 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;
990 XPUSHs (scalar); 1319 XPUSHs (scalar);
991} 1320}
992 1321
993void decode_cbor (SV *cborstr) 1322void decode_cbor (SV *cborstr)

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines