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 1.5 by root, Sat Oct 26 21:14:20 2013 UTC vs.
Revision 1.36 by root, Sat Nov 30 17:37:45 2013 UTC

9#include <limits.h> 9#include <limits.h>
10#include <float.h> 10#include <float.h>
11 11
12#include "ecb.h" 12#include "ecb.h"
13 13
14// compatibility with perl <5.18
15#ifndef HvNAMELEN_get
16# define HvNAMELEN_get(hv) strlen (HvNAME (hv))
17#endif
18#ifndef HvNAMELEN
19# define HvNAMELEN(hv) HvNAMELEN_get (hv)
20#endif
21#ifndef HvNAMEUTF8
22# define HvNAMEUTF8(hv) 0
23#endif
24#ifndef SvREFCNT_dec_NN
25# define SvREFCNT_dec_NN(sv) SvREFCNT_dec (sv)
26#endif
27
28// known major and minor types
29enum cbor_type
30{
31 MAJOR_SHIFT = 5,
32 MINOR_MASK = 0x1f,
33
34 MAJOR_POS_INT = 0 << MAJOR_SHIFT,
35 MAJOR_NEG_INT = 1 << MAJOR_SHIFT,
36 MAJOR_BYTES = 2 << MAJOR_SHIFT,
37 MAJOR_TEXT = 3 << MAJOR_SHIFT,
38 MAJOR_ARRAY = 4 << MAJOR_SHIFT,
39 MAJOR_MAP = 5 << MAJOR_SHIFT,
40 MAJOR_TAG = 6 << MAJOR_SHIFT,
41 MAJOR_MISC = 7 << MAJOR_SHIFT,
42
43 // INT/STRING/ARRAY/MAP subtypes
44 LENGTH_EXT1 = 24,
45 LENGTH_EXT2 = 25,
46 LENGTH_EXT4 = 26,
47 LENGTH_EXT8 = 27,
48
49 // SIMPLE types (effectively MISC subtypes)
50 SIMPLE_FALSE = 20,
51 SIMPLE_TRUE = 21,
52 SIMPLE_NULL = 22,
53 SIMPLE_UNDEF = 23,
54
55 // MISC subtype (unused)
56 MISC_EXT1 = 24,
57 MISC_FLOAT16 = 25,
58 MISC_FLOAT32 = 26,
59 MISC_FLOAT64 = 27,
60
61 // BYTES/TEXT/ARRAY/MAP
62 MINOR_INDEF = 31,
63};
64
65// known tags
66enum cbor_tag
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
78 CBOR_TAG_DATETIME = 0, // rfc4287, utf-8
79 CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any
80 CBOR_TAG_POS_BIGNUM = 2, // byte string
81 CBOR_TAG_NEG_BIGNUM = 3, // byte string
82 CBOR_TAG_DECIMAL = 4, // decimal fraction, array
83 CBOR_TAG_BIGFLOAT = 5, // array
84
85 CBOR_TAG_CONV_B64U = 21, // base64url, any
86 CBOR_TAG_CONV_B64 = 22, // base64, any
87 CBOR_TAG_CONV_HEX = 23, // base16, any
88 CBOR_TAG_CBOR = 24, // embedded cbor, byte string
89
90 CBOR_TAG_URI = 32, // URI rfc3986, utf-8
91 CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8
92 CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8
93 CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8
94 CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8
95
96 CBOR_TAG_MAGIC = 55799, // self-describe cbor
97};
98
14#define F_SHRINK 0x00000200UL 99#define F_SHRINK 0x00000001UL
15#define F_ALLOW_UNKNOWN 0x00002000UL 100#define F_ALLOW_UNKNOWN 0x00000002UL
101#define F_ALLOW_SHARING 0x00000004UL
102#define F_PACK_STRINGS 0x00000008UL
16 103
17#define INIT_SIZE 32 // initial scalar size to be allocated 104#define INIT_SIZE 32 // initial scalar size to be allocated
18 105
19#define SB do { 106#define SB do {
20#define SE } while (0) 107#define SE } while (0)
31#else 118#else
32# define CBOR_SLOW 0 119# define CBOR_SLOW 0
33# define CBOR_STASH cbor_stash 120# define CBOR_STASH cbor_stash
34#endif 121#endif
35 122
36static HV *cbor_stash, *cbor_boolean_stash; // CBOR::XS:: 123static HV *cbor_stash, *types_boolean_stash, *types_error_stash, *cbor_tagged_stash; // CBOR::XS::
37static SV *cbor_true, *cbor_false; 124static SV *types_true, *types_false, *types_error, *sv_cbor, *default_filter;
38 125
39typedef struct { 126typedef struct {
40 U32 flags; 127 U32 flags;
41 U32 max_depth; 128 U32 max_depth;
42 STRLEN max_size; 129 STRLEN max_size;
43 130 SV *filter;
44 SV *cb_object;
45 HV *cb_sk_object;
46} CBOR; 131} CBOR;
47 132
48ecb_inline void 133ecb_inline void
49cbor_init (CBOR *cbor) 134cbor_init (CBOR *cbor)
50{ 135{
51 Zero (cbor, 1, CBOR); 136 Zero (cbor, 1, CBOR);
52 cbor->max_depth = 512; 137 cbor->max_depth = 512;
138}
139
140ecb_inline void
141cbor_free (CBOR *cbor)
142{
143 SvREFCNT_dec (cbor->filter);
53} 144}
54 145
55///////////////////////////////////////////////////////////////////////////// 146/////////////////////////////////////////////////////////////////////////////
56// utility functions 147// utility functions
57 148
79 SvPV_renew (sv, SvCUR (sv) + 1); 170 SvPV_renew (sv, SvCUR (sv) + 1);
80#endif 171#endif
81 } 172 }
82} 173}
83 174
84///////////////////////////////////////////////////////////////////////////// 175// minimum length of a string to be registered for stringref
85// fp hell 176ecb_inline int
86 177minimum_string_length (UV idx)
87//TODO 178{
179 return idx > 23
180 ? idx > 0xffU
181 ? idx > 0xffffU
182 ? idx > 0xffffffffU
183 ? 11
184 : 7
185 : 5
186 : 4
187 : 3;
188}
88 189
89///////////////////////////////////////////////////////////////////////////// 190/////////////////////////////////////////////////////////////////////////////
90// encoder 191// encoder
91 192
92// structure used for encoding CBOR 193// structure used for encoding CBOR
95 char *cur; // SvPVX (sv) + current output position 196 char *cur; // SvPVX (sv) + current output position
96 char *end; // SvEND (sv) 197 char *end; // SvEND (sv)
97 SV *sv; // result scalar 198 SV *sv; // result scalar
98 CBOR cbor; 199 CBOR cbor;
99 U32 depth; // recursion level 200 U32 depth; // recursion level
201 HV *stringref[2]; // string => index, or 0 ([0] = bytes, [1] = utf-8)
202 UV stringref_idx;
203 HV *shareable; // ptr => index, or 0
204 UV shareable_idx;
100} enc_t; 205} enc_t;
101 206
102ecb_inline void 207ecb_inline void
103need (enc_t *enc, STRLEN len) 208need (enc_t *enc, STRLEN len)
104{ 209{
121static void 226static void
122encode_uint (enc_t *enc, int major, UV len) 227encode_uint (enc_t *enc, int major, UV len)
123{ 228{
124 need (enc, 9); 229 need (enc, 9);
125 230
126 if (len < 24) 231 if (ecb_expect_true (len < LENGTH_EXT1))
127 *enc->cur++ = major | len; 232 *enc->cur++ = major | len;
128 else if (len <= 0xff) 233 else if (ecb_expect_true (len <= 0xffU))
129 { 234 {
130 *enc->cur++ = major | 24; 235 *enc->cur++ = major | LENGTH_EXT1;
131 *enc->cur++ = len; 236 *enc->cur++ = len;
132 } 237 }
133 else if (len <= 0xffff) 238 else if (len <= 0xffffU)
134 { 239 {
135 *enc->cur++ = major | 25; 240 *enc->cur++ = major | LENGTH_EXT2;
136 *enc->cur++ = len >> 8; 241 *enc->cur++ = len >> 8;
137 *enc->cur++ = len; 242 *enc->cur++ = len;
138 } 243 }
139 else if (len <= 0xffffffff) 244 else if (len <= 0xffffffffU)
140 { 245 {
141 *enc->cur++ = major | 26; 246 *enc->cur++ = major | LENGTH_EXT4;
142 *enc->cur++ = len >> 24; 247 *enc->cur++ = len >> 24;
143 *enc->cur++ = len >> 16; 248 *enc->cur++ = len >> 16;
144 *enc->cur++ = len >> 8; 249 *enc->cur++ = len >> 8;
145 *enc->cur++ = len; 250 *enc->cur++ = len;
146 } 251 }
147 else 252 else
148 { 253 {
149 *enc->cur++ = major | 27; 254 *enc->cur++ = major | LENGTH_EXT8;
150 *enc->cur++ = len >> 56; 255 *enc->cur++ = len >> 56;
151 *enc->cur++ = len >> 48; 256 *enc->cur++ = len >> 48;
152 *enc->cur++ = len >> 40; 257 *enc->cur++ = len >> 40;
153 *enc->cur++ = len >> 32; 258 *enc->cur++ = len >> 32;
154 *enc->cur++ = len >> 24; 259 *enc->cur++ = len >> 24;
156 *enc->cur++ = len >> 8; 261 *enc->cur++ = len >> 8;
157 *enc->cur++ = len; 262 *enc->cur++ = len;
158 } 263 }
159} 264}
160 265
161static void 266ecb_inline void
267encode_tag (enc_t *enc, UV tag)
268{
269 encode_uint (enc, MAJOR_TAG, tag);
270}
271
272ecb_inline void
162encode_str (enc_t *enc, int utf8, char *str, STRLEN len) 273encode_str (enc_t *enc, int utf8, char *str, STRLEN len)
163{ 274{
164 encode_uint (enc, utf8 ? 0x60 : 0x40, len); 275 encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len);
165 need (enc, len); 276 need (enc, len);
166 memcpy (enc->cur, str, len); 277 memcpy (enc->cur, str, len);
167 enc->cur += len; 278 enc->cur += len;
168} 279}
169 280
281static void
282encode_strref (enc_t *enc, int utf8, char *str, STRLEN len)
283{
284 if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS))
285 {
286 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1);
287
288 if (SvOK (*svp))
289 {
290 // already registered, use stringref
291 encode_tag (enc, CBOR_TAG_STRINGREF);
292 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
293 return;
294 }
295 else if (len >= minimum_string_length (enc->stringref_idx))
296 {
297 // register only
298 sv_setuv (*svp, enc->stringref_idx);
299 ++enc->stringref_idx;
300 }
301 }
302
303 encode_str (enc, utf8, str, len);
304}
305
170static void encode_sv (enc_t *enc, SV *sv); 306static void encode_sv (enc_t *enc, SV *sv);
171 307
172static void 308static void
173encode_av (enc_t *enc, AV *av) 309encode_av (enc_t *enc, AV *av)
174{ 310{
177 if (enc->depth >= enc->cbor.max_depth) 313 if (enc->depth >= enc->cbor.max_depth)
178 croak (ERR_NESTING_EXCEEDED); 314 croak (ERR_NESTING_EXCEEDED);
179 315
180 ++enc->depth; 316 ++enc->depth;
181 317
182 encode_uint (enc, 0x80, len + 1); 318 encode_uint (enc, MAJOR_ARRAY, len + 1);
183 319
184 for (i = 0; i <= len; ++i) 320 for (i = 0; i <= len; ++i)
185 { 321 {
186 SV **svp = av_fetch (av, i, 0); 322 SV **svp = av_fetch (av, i, 0);
187 encode_sv (enc, svp ? *svp : &PL_sv_undef); 323 encode_sv (enc, svp ? *svp : &PL_sv_undef);
202 338
203 int pairs = hv_iterinit (hv); 339 int pairs = hv_iterinit (hv);
204 int mg = SvMAGICAL (hv); 340 int mg = SvMAGICAL (hv);
205 341
206 if (mg) 342 if (mg)
207 encode_ch (enc, 0xa0 | 31); 343 encode_ch (enc, MAJOR_MAP | MINOR_INDEF);
208 else 344 else
209 encode_uint (enc, 0xa0, pairs); 345 encode_uint (enc, MAJOR_MAP, pairs);
210 346
211 while ((he = hv_iternext (hv))) 347 while ((he = hv_iternext (hv)))
212 { 348 {
213 if (HeKLEN (he) == HEf_SVKEY) 349 if (HeKLEN (he) == HEf_SVKEY)
214 encode_sv (enc, HeSVKEY (he)); 350 encode_sv (enc, HeSVKEY (he));
215 else 351 else
216 encode_str (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he)); 352 encode_strref (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he));
217 353
218 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he)); 354 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he));
219 } 355 }
220 356
221 if (mg) 357 if (mg)
222 encode_ch (enc, 0xe0 | 31); 358 encode_ch (enc, MAJOR_MISC | MINOR_INDEF);
223 359
224 --enc->depth; 360 --enc->depth;
225} 361}
226 362
227// encode objects, arrays and special \0=false and \1=true values. 363// encode objects, arrays and special \0=false and \1=true values.
228static void 364static void
229encode_rv (enc_t *enc, SV *sv) 365encode_rv (enc_t *enc, SV *sv)
230{ 366{
231 svtype svt;
232
233 SvGETMAGIC (sv); 367 SvGETMAGIC (sv);
368
234 svt = SvTYPE (sv); 369 svtype svt = SvTYPE (sv);
235 370
236 if (ecb_expect_false (SvOBJECT (sv))) 371 if (ecb_expect_false (SvOBJECT (sv)))
237 { 372 {
238 HV *stash = !CBOR_SLOW || cbor_boolean_stash 373 HV *boolean_stash = !CBOR_SLOW || types_boolean_stash
239 ? cbor_boolean_stash 374 ? types_boolean_stash
375 : gv_stashpv ("Types::Serialiser::Boolean", 1);
376 HV *error_stash = !CBOR_SLOW || types_error_stash
377 ? types_error_stash
378 : gv_stashpv ("Types::Serialiser::Error", 1);
379 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
380 ? cbor_tagged_stash
240 : gv_stashpv ("CBOR::XS::Boolean", 1); 381 : gv_stashpv ("CBOR::XS::Tagged" , 1);
241 382
242 if (SvSTASH (sv) == stash) 383 HV *stash = SvSTASH (sv);
243 encode_ch (enc, SvIV (sv) ? 0xe0 | 21 : 0xe0 | 20); 384
385 if (stash == boolean_stash)
386 {
387 encode_ch (enc, SvIV (sv) ? MAJOR_MISC | SIMPLE_TRUE : MAJOR_MISC | SIMPLE_FALSE);
388 return;
389 }
390 else if (stash == error_stash)
391 {
392 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
393 return;
394 }
395 else if (stash == tagged_stash)
396 {
397 if (svt != SVt_PVAV)
398 croak ("encountered CBOR::XS::Tagged object that isn't an array");
399
400 encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1)));
401 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1));
402
403 return;
404 }
405 }
406
407 if (ecb_expect_false (SvREFCNT (sv) > 1)
408 && ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING))
409 {
410 if (!enc->shareable)
411 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
412
413 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
414
415 if (SvOK (*svp))
416 {
417 encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF);
418 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
419 return;
420 }
244 else 421 else
245 { 422 {
246#if 0 //TODO 423 sv_setuv (*svp, enc->shareable_idx);
247 if (enc->cbor.flags & F_CONV_BLESSED) 424 ++enc->shareable_idx;
425 encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE);
426 }
427 }
428
429 if (ecb_expect_false (SvOBJECT (sv)))
430 {
431 HV *stash = SvSTASH (sv);
432 GV *method;
433
434 if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
248 { 435 {
436 dSP;
437
438 ENTER; SAVETMPS; PUSHMARK (SP);
249 // we re-bless the reference to get overload and other niceties right 439 // we re-bless the reference to get overload and other niceties right
250 GV *to_cbor = gv_fetchmethod_autoload (SvSTASH (sv), "TO_CBOR", 0);
251
252 if (to_cbor)
253 {
254 dSP;
255
256 ENTER; SAVETMPS; PUSHMARK (SP);
257 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv))); 440 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
258 441
259 // calling with G_SCALAR ensures that we always get a 1 return value
260 PUTBACK; 442 PUTBACK;
443 // G_SCALAR ensures that return value is 1
261 call_sv ((SV *)GvCV (to_cbor), G_SCALAR); 444 call_sv ((SV *)GvCV (method), G_SCALAR);
262 SPAGAIN; 445 SPAGAIN;
263 446
264 // catch this surprisingly common error 447 // catch this surprisingly common error
265 if (SvROK (TOPs) && SvRV (TOPs) == sv) 448 if (SvROK (TOPs) && SvRV (TOPs) == sv)
266 croak ("%s::TO_CBOR method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv))); 449 croak ("%s::TO_CBOR method returned same object as was passed instead of a new one", HvNAME (stash));
267 450
268 sv = POPs;
269 PUTBACK;
270
271 encode_sv (enc, sv); 451 encode_sv (enc, POPs);
272 452
453 PUTBACK;
454
273 FREETMPS; LEAVE; 455 FREETMPS; LEAVE;
274 }
275 else if (enc->cbor.flags & F_ALLOW_BLESSED)
276 encode_str (enc, "null", 4, 0);
277 else
278 croak ("encountered object '%s', but neither allow_blessed enabled nor TO_CBOR method available on it",
279 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
280 } 456 }
281 else if (enc->cbor.flags & F_ALLOW_BLESSED) 457 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)
282 encode_str (enc, "null", 4, 0); 458 {
459 dSP;
460
461 ENTER; SAVETMPS; PUSHMARK (SP);
462 EXTEND (SP, 2);
463 // we re-bless the reference to get overload and other niceties right
464 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
465 PUSHs (sv_cbor);
466
467 PUTBACK;
468 int count = call_sv ((SV *)GvCV (method), G_ARRAY);
469 SPAGAIN;
470
471 // catch this surprisingly common error
472 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)
473 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));
474
475 encode_tag (enc, CBOR_TAG_PERL_OBJECT);
476 encode_uint (enc, MAJOR_ARRAY, count + 1);
477 encode_strref (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
478
479 while (count)
480 encode_sv (enc, SP[1 - count--]);
481
482 PUTBACK;
483
484 FREETMPS; LEAVE;
485 }
283 else 486 else
284 croak ("encountered object '%s', but neither allow_blessed nor convert_blessed settings are enabled", 487 croak ("encountered object '%s', but no TO_CBOR or FREEZE methods available on it",
285 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 488 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
286#endif
287 }
288 } 489 }
289 else if (svt == SVt_PVHV) 490 else if (svt == SVt_PVHV)
290 encode_hv (enc, (HV *)sv); 491 encode_hv (enc, (HV *)sv);
291 else if (svt == SVt_PVAV) 492 else if (svt == SVt_PVAV)
292 encode_av (enc, (AV *)sv); 493 encode_av (enc, (AV *)sv);
293 else if (svt < SVt_PVAV)
294 {
295 STRLEN len = 0;
296 char *pv = svt ? SvPV (sv, len) : 0;
297
298 if (len == 1 && *pv == '1')
299 encode_ch (enc, 0xe0 | 21);
300 else if (len == 1 && *pv == '0')
301 encode_ch (enc, 0xe0 | 20);
302 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
303 encode_ch (enc, 0xe0 | 23);
304 else
305 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
306 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
307 }
308 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
309 encode_ch (enc, 0xe0 | 23);
310 else 494 else
311 croak ("encountered %s, but CBOR can only represent references to arrays or hashes", 495 {
312 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 496 encode_tag (enc, CBOR_TAG_INDIRECTION);
497 encode_sv (enc, sv);
498 }
313} 499}
314 500
315static void 501static void
316encode_nv (enc_t *enc, SV *sv) 502encode_nv (enc_t *enc, SV *sv)
317{ 503{
318 double nv = SvNVX (sv); 504 double nv = SvNVX (sv);
319 505
320 need (enc, 9); 506 need (enc, 9);
321 507
322 if (ecb_expect_false (nv == (U32)nv)) 508 if (ecb_expect_false (nv == (NV)(U32)nv))
323 encode_uint (enc, 0x00, (U32)nv); 509 encode_uint (enc, MAJOR_POS_INT, (U32)nv);
324 //TODO: maybe I32? 510 //TODO: maybe I32?
325 else if (ecb_expect_false (nv == (float)nv)) 511 else if (ecb_expect_false (nv == (float)nv))
326 { 512 {
327 uint32_t fp = ecb_float_to_binary32 (nv); 513 uint32_t fp = ecb_float_to_binary32 (nv);
328 514
329 *enc->cur++ = 0xe0 | 26; 515 *enc->cur++ = MAJOR_MISC | MISC_FLOAT32;
330 516
331 if (!ecb_big_endian ()) 517 if (!ecb_big_endian ())
332 fp = ecb_bswap32 (fp); 518 fp = ecb_bswap32 (fp);
333 519
334 memcpy (enc->cur, &fp, 4); 520 memcpy (enc->cur, &fp, 4);
336 } 522 }
337 else 523 else
338 { 524 {
339 uint64_t fp = ecb_double_to_binary64 (nv); 525 uint64_t fp = ecb_double_to_binary64 (nv);
340 526
341 *enc->cur++ = 0xe0 | 27; 527 *enc->cur++ = MAJOR_MISC | MISC_FLOAT64;
342 528
343 if (!ecb_big_endian ()) 529 if (!ecb_big_endian ())
344 fp = ecb_bswap64 (fp); 530 fp = ecb_bswap64 (fp);
345 531
346 memcpy (enc->cur, &fp, 8); 532 memcpy (enc->cur, &fp, 8);
355 541
356 if (SvPOKp (sv)) 542 if (SvPOKp (sv))
357 { 543 {
358 STRLEN len; 544 STRLEN len;
359 char *str = SvPV (sv, len); 545 char *str = SvPV (sv, len);
360 encode_str (enc, SvUTF8 (sv), str, len); 546 encode_strref (enc, SvUTF8 (sv), str, len);
361 } 547 }
362 else if (SvNOKp (sv)) 548 else if (SvNOKp (sv))
363 encode_nv (enc, sv); 549 encode_nv (enc, sv);
364 else if (SvIOKp (sv)) 550 else if (SvIOKp (sv))
365 { 551 {
366 if (SvIsUV (sv)) 552 if (SvIsUV (sv))
367 encode_uint (enc, 0x00, SvUVX (sv)); 553 encode_uint (enc, MAJOR_POS_INT, SvUVX (sv));
368 else if (SvIVX (sv) >= 0) 554 else if (SvIVX (sv) >= 0)
369 encode_uint (enc, 0x00, SvIVX (sv)); 555 encode_uint (enc, MAJOR_POS_INT, SvIVX (sv));
370 else 556 else
371 encode_uint (enc, 0x20, -(SvIVX (sv) + 1)); 557 encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1));
372 } 558 }
373 else if (SvROK (sv)) 559 else if (SvROK (sv))
374 encode_rv (enc, SvRV (sv)); 560 encode_rv (enc, SvRV (sv));
375 else if (!SvOK (sv)) 561 else if (!SvOK (sv))
376 encode_ch (enc, 0xe0 | 22); 562 encode_ch (enc, MAJOR_MISC | SIMPLE_NULL);
377 else if (enc->cbor.flags & F_ALLOW_UNKNOWN) 563 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
378 encode_ch (enc, 0xe0 | 23); 564 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
379 else 565 else
380 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data", 566 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data",
381 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv)); 567 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
382} 568}
383 569
384static SV * 570static SV *
385encode_cbor (SV *scalar, CBOR *cbor) 571encode_cbor (SV *scalar, CBOR *cbor)
386{ 572{
387 enc_t enc; 573 enc_t enc = { };
388 574
389 enc.cbor = *cbor; 575 enc.cbor = *cbor;
390 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 576 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
391 enc.cur = SvPVX (enc.sv); 577 enc.cur = SvPVX (enc.sv);
392 enc.end = SvEND (enc.sv); 578 enc.end = SvEND (enc.sv);
393 enc.depth = 0;
394 579
395 SvPOK_only (enc.sv); 580 SvPOK_only (enc.sv);
581
582 if (cbor->flags & F_PACK_STRINGS)
583 {
584 encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE);
585 enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ());
586 enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ());
587 }
588
396 encode_sv (&enc, scalar); 589 encode_sv (&enc, scalar);
397 590
398 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 591 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
399 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings 592 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
400 593
414 U8 *end; // end of input string 607 U8 *end; // end of input string
415 const char *err; // parse error, if != 0 608 const char *err; // parse error, if != 0
416 CBOR cbor; 609 CBOR cbor;
417 U32 depth; // recursion depth 610 U32 depth; // recursion depth
418 U32 maxdepth; // recursion depth limit 611 U32 maxdepth; // recursion depth limit
612 AV *shareable;
613 AV *stringref;
614 SV *decode_tagged;
419} dec_t; 615} dec_t;
420 616
421#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE 617#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE
422 618
423#define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data") 619#define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data")
426#define DEC_DEC_DEPTH --dec->depth 622#define DEC_DEC_DEPTH --dec->depth
427 623
428static UV 624static UV
429decode_uint (dec_t *dec) 625decode_uint (dec_t *dec)
430{ 626{
431 switch (*dec->cur & 31) 627 U8 m = *dec->cur & MINOR_MASK;
432 { 628 ++dec->cur;
433 case 0: case 1: case 2: case 3: case 4: case 5: case 6: case 7:
434 case 8: case 9: case 10: case 11: case 12: case 13: case 14: case 15:
435 case 16: case 17: case 18: case 19: case 20: case 21: case 22: case 23:
436 return *dec->cur++ & 31;
437 629
438 case 24: 630 if (ecb_expect_true (m < LENGTH_EXT1))
631 return m;
632 else if (ecb_expect_true (m == LENGTH_EXT1))
633 {
439 WANT (2); 634 WANT (1);
440 dec->cur += 2; 635 dec->cur += 1;
441 return dec->cur[-1]; 636 return dec->cur[-1];
442 637 }
443 case 25: 638 else if (ecb_expect_true (m == LENGTH_EXT2))
639 {
444 WANT (3); 640 WANT (2);
445 dec->cur += 3; 641 dec->cur += 2;
446 return (((UV)dec->cur[-2]) << 8) 642 return (((UV)dec->cur[-2]) << 8)
447 | ((UV)dec->cur[-1]); 643 | ((UV)dec->cur[-1]);
448 644 }
449 case 26: 645 else if (ecb_expect_true (m == LENGTH_EXT4))
646 {
450 WANT (5); 647 WANT (4);
451 dec->cur += 5; 648 dec->cur += 4;
452 return (((UV)dec->cur[-4]) << 24) 649 return (((UV)dec->cur[-4]) << 24)
453 | (((UV)dec->cur[-3]) << 16) 650 | (((UV)dec->cur[-3]) << 16)
454 | (((UV)dec->cur[-2]) << 8) 651 | (((UV)dec->cur[-2]) << 8)
455 | ((UV)dec->cur[-1]); 652 | ((UV)dec->cur[-1]);
456 653 }
457 case 27: 654 else if (ecb_expect_true (m == LENGTH_EXT8))
655 {
458 WANT (9); 656 WANT (8);
459 dec->cur += 9; 657 dec->cur += 8;
658
659 return
660#if UVSIZE < 8
661 0
662#else
460 return (((UV)dec->cur[-8]) << 56) 663 (((UV)dec->cur[-8]) << 56)
461 | (((UV)dec->cur[-7]) << 48) 664 | (((UV)dec->cur[-7]) << 48)
462 | (((UV)dec->cur[-6]) << 40) 665 | (((UV)dec->cur[-6]) << 40)
463 | (((UV)dec->cur[-5]) << 32) 666 | (((UV)dec->cur[-5]) << 32)
667#endif
464 | (((UV)dec->cur[-4]) << 24) 668 | (((UV)dec->cur[-4]) << 24)
465 | (((UV)dec->cur[-3]) << 16) 669 | (((UV)dec->cur[-3]) << 16)
466 | (((UV)dec->cur[-2]) << 8) 670 | (((UV)dec->cur[-2]) << 8)
467 | ((UV)dec->cur[-1]); 671 | ((UV)dec->cur[-1]);
468 672 }
469 default: 673 else
470 ERR ("corrupted CBOR data (unsupported integer minor encoding)"); 674 ERR ("corrupted CBOR data (unsupported integer minor encoding)");
471 }
472 675
473fail: 676fail:
474 return 0; 677 return 0;
475} 678}
476 679
481{ 684{
482 AV *av = newAV (); 685 AV *av = newAV ();
483 686
484 DEC_INC_DEPTH; 687 DEC_INC_DEPTH;
485 688
486 if ((*dec->cur & 31) == 31) 689 if (*dec->cur == (MAJOR_ARRAY | MINOR_INDEF))
487 { 690 {
488 ++dec->cur; 691 ++dec->cur;
489 692
490 for (;;) 693 for (;;)
491 { 694 {
492 WANT (1); 695 WANT (1);
493 696
494 if (*dec->cur == (0xe0 | 31)) 697 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
495 { 698 {
496 ++dec->cur; 699 ++dec->cur;
497 break; 700 break;
498 } 701 }
499 702
502 } 705 }
503 else 706 else
504 { 707 {
505 int i, len = decode_uint (dec); 708 int i, len = decode_uint (dec);
506 709
710 WANT (len); // complexity check for av_fill - need at least one byte per value, do not allow supersize arrays
507 av_fill (av, len - 1); 711 av_fill (av, len - 1);
508 712
509 for (i = 0; i < len; ++i) 713 for (i = 0; i < len; ++i)
510 AvARRAY (av)[i] = decode_sv (dec); 714 AvARRAY (av)[i] = decode_sv (dec);
511 } 715 }
517 SvREFCNT_dec (av); 721 SvREFCNT_dec (av);
518 DEC_DEC_DEPTH; 722 DEC_DEC_DEPTH;
519 return &PL_sv_undef; 723 return &PL_sv_undef;
520} 724}
521 725
726static void
727decode_he (dec_t *dec, HV *hv)
728{
729 // for speed reasons, we specialcase single-string
730 // byte or utf-8 strings as keys, but only when !stringref
731
732 if (ecb_expect_true (!dec->stringref))
733 if (ecb_expect_true ((*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8))
734 {
735 I32 len = decode_uint (dec);
736 char *key = (char *)dec->cur;
737
738 dec->cur += len;
739
740 if (ecb_expect_false (dec->stringref))
741 av_push (dec->stringref, newSVpvn (key, len));
742
743 hv_store (hv, key, len, decode_sv (dec), 0);
744
745 return;
746 }
747 else if (ecb_expect_true ((*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8))
748 {
749 I32 len = decode_uint (dec);
750 char *key = (char *)dec->cur;
751
752 dec->cur += len;
753
754 if (ecb_expect_false (dec->stringref))
755 av_push (dec->stringref, newSVpvn_utf8 (key, len, 1));
756
757 hv_store (hv, key, -len, decode_sv (dec), 0);
758
759 return;
760 }
761
762 SV *k = decode_sv (dec);
763 SV *v = decode_sv (dec);
764
765 hv_store_ent (hv, k, v, 0);
766 SvREFCNT_dec (k);
767}
768
522static SV * 769static SV *
523decode_hv (dec_t *dec) 770decode_hv (dec_t *dec)
524{ 771{
525 HV *hv = newHV (); 772 HV *hv = newHV ();
526 773
527 DEC_INC_DEPTH; 774 DEC_INC_DEPTH;
528 775
529 if ((*dec->cur & 31) == 31) 776 if (*dec->cur == (MAJOR_MAP | MINOR_INDEF))
530 { 777 {
531 ++dec->cur; 778 ++dec->cur;
532 779
533 for (;;) 780 for (;;)
534 { 781 {
535 WANT (1); 782 WANT (1);
536 783
537 if (*dec->cur == (0xe0 | 31)) 784 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
538 { 785 {
539 ++dec->cur; 786 ++dec->cur;
540 break; 787 break;
541 } 788 }
542 789
543 SV *k = decode_sv (dec); 790 decode_he (dec, hv);
544 SV *v = decode_sv (dec);
545
546 hv_store_ent (hv, k, v, 0);
547 } 791 }
548 } 792 }
549 else 793 else
550 { 794 {
551 int len = decode_uint (dec); 795 int pairs = decode_uint (dec);
552 796
553 while (len--) 797 while (pairs--)
554 { 798 decode_he (dec, hv);
555 SV *k = decode_sv (dec);
556 SV *v = decode_sv (dec);
557
558 hv_store_ent (hv, k, v, 0);
559 }
560 } 799 }
561 800
562 DEC_DEC_DEPTH; 801 DEC_DEC_DEPTH;
563 return newRV_noinc ((SV *)hv); 802 return newRV_noinc ((SV *)hv);
564
565#if 0
566 SV *sv;
567 HV *hv = newHV ();
568
569 DEC_INC_DEPTH;
570 decode_ws (dec);
571
572 for (;;)
573 {
574 // heuristic: assume that
575 // a) decode_str + hv_store_ent are abysmally slow.
576 // b) most hash keys are short, simple ascii text.
577 // => try to "fast-match" such strings to avoid
578 // the overhead of decode_str + hv_store_ent.
579 {
580 SV *value;
581 char *p = dec->cur;
582 char *e = p + 24; // only try up to 24 bytes
583
584 for (;;)
585 {
586 // the >= 0x80 is false on most architectures
587 if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\')
588 {
589 // slow path, back up and use decode_str
590 SV *key = decode_str (dec);
591 if (!key)
592 goto fail;
593
594 decode_ws (dec); EXPECT_CH (':');
595
596 decode_ws (dec);
597 value = decode_sv (dec);
598 if (!value)
599 {
600 SvREFCNT_dec (key);
601 goto fail;
602 }
603
604 hv_store_ent (hv, key, value, 0);
605 SvREFCNT_dec (key);
606
607 break;
608 }
609 else if (*p == '"')
610 {
611 // fast path, got a simple key
612 char *key = dec->cur;
613 int len = p - key;
614 dec->cur = p + 1;
615
616 decode_ws (dec); EXPECT_CH (':');
617
618 decode_ws (dec);
619 value = decode_sv (dec);
620 if (!value)
621 goto fail;
622
623 hv_store (hv, key, len, value, 0);
624
625 break;
626 }
627
628 ++p;
629 }
630 }
631
632 decode_ws (dec);
633
634 if (*dec->cur == '}')
635 {
636 ++dec->cur;
637 break;
638 }
639
640 if (*dec->cur != ',')
641 ERR (", or } expected while parsing object/hash");
642
643 ++dec->cur;
644
645 decode_ws (dec);
646
647 if (*dec->cur == '}' && dec->cbor.flags & F_RELAXED)
648 {
649 ++dec->cur;
650 break;
651 }
652 }
653
654 DEC_DEC_DEPTH;
655 sv = newRV_noinc ((SV *)hv);
656
657 // check filter callbacks
658 if (dec->cbor.flags & F_HOOK)
659 {
660 if (dec->cbor.cb_sk_object && HvKEYS (hv) == 1)
661 {
662 HE *cb, *he;
663
664 hv_iterinit (hv);
665 he = hv_iternext (hv);
666 hv_iterinit (hv);
667
668 // the next line creates a mortal sv each time its called.
669 // might want to optimise this for common cases.
670 cb = hv_fetch_ent (dec->cbor.cb_sk_object, hv_iterkeysv (he), 0, 0);
671
672 if (cb)
673 {
674 dSP;
675 int count;
676
677 ENTER; SAVETMPS; PUSHMARK (SP);
678 XPUSHs (HeVAL (he));
679 sv_2mortal (sv);
680
681 PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN;
682
683 if (count == 1)
684 {
685 sv = newSVsv (POPs);
686 FREETMPS; LEAVE;
687 return sv;
688 }
689
690 SvREFCNT_inc (sv);
691 FREETMPS; LEAVE;
692 }
693 }
694
695 if (dec->cbor.cb_object)
696 {
697 dSP;
698 int count;
699
700 ENTER; SAVETMPS; PUSHMARK (SP);
701 XPUSHs (sv_2mortal (sv));
702
703 PUTBACK; count = call_sv (dec->cbor.cb_object, G_ARRAY); SPAGAIN;
704
705 if (count == 1)
706 {
707 sv = newSVsv (POPs);
708 FREETMPS; LEAVE;
709 return sv;
710 }
711
712 SvREFCNT_inc (sv);
713 FREETMPS; LEAVE;
714 }
715 }
716
717 return sv;
718#endif
719 803
720fail: 804fail:
721 SvREFCNT_dec (hv); 805 SvREFCNT_dec (hv);
722 DEC_DEC_DEPTH; 806 DEC_DEC_DEPTH;
723 return &PL_sv_undef; 807 return &PL_sv_undef;
724} 808}
725 809
726static SV * 810static SV *
727decode_str (dec_t *dec, int utf8) 811decode_str (dec_t *dec, int utf8)
728{ 812{
729 SV *sv; 813 SV *sv = 0;
730 814
731 if ((*dec->cur & 31) == 31) 815 if ((*dec->cur & MINOR_MASK) == MINOR_INDEF)
732 { 816 {
817 // indefinite length strings
733 ++dec->cur; 818 ++dec->cur;
734 819
820 U8 major = *dec->cur & MAJOR_MISC;
821
735 sv = newSVpvn ("", 0); 822 sv = newSVpvn ("", 0);
736 823
737 // not very fast, and certainly not robust against illegal input
738 for (;;) 824 for (;;)
739 { 825 {
740 WANT (1); 826 WANT (1);
741 827
742 if (*dec->cur == (0xe0 | 31)) 828 if ((*dec->cur - major) > LENGTH_EXT8)
829 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
743 { 830 {
744 ++dec->cur; 831 ++dec->cur;
745 break; 832 break;
746 } 833 }
834 else
835 ERR ("corrupted CBOR data (invalid chunks in indefinite length string)");
747 836
748 SV *sv2 = decode_sv (dec); 837 STRLEN len = decode_uint (dec);
749 sv_catsv (sv, sv2); 838
839 WANT (len);
840 sv_catpvn (sv, dec->cur, len);
841 dec->cur += len;
750 } 842 }
751 } 843 }
752 else 844 else
753 { 845 {
754 STRLEN len = decode_uint (dec); 846 STRLEN len = decode_uint (dec);
755 847
756 WANT (len); 848 WANT (len);
757 sv = newSVpvn (dec->cur, len); 849 sv = newSVpvn (dec->cur, len);
758 dec->cur += len; 850 dec->cur += len;
851
852 if (ecb_expect_false (dec->stringref)
853 && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1))
854 av_push (dec->stringref, SvREFCNT_inc_NN (sv));
759 } 855 }
760 856
761 if (utf8) 857 if (utf8)
762 SvUTF8_on (sv); 858 SvUTF8_on (sv);
763 859
764 return sv; 860 return sv;
765 861
766fail: 862fail:
863 SvREFCNT_dec (sv);
767 return &PL_sv_undef; 864 return &PL_sv_undef;
768} 865}
769 866
770static SV * 867static SV *
771decode_tagged (dec_t *dec) 868decode_tagged (dec_t *dec)
772{ 869{
870 SV *sv = 0;
773 UV tag = decode_uint (dec); 871 UV tag = decode_uint (dec);
872
873 WANT (1);
874
875 switch (tag)
876 {
877 case CBOR_TAG_MAGIC:
878 sv = decode_sv (dec);
879 break;
880
881 case CBOR_TAG_INDIRECTION:
882 sv = newRV_noinc (decode_sv (dec));
883 break;
884
885 case CBOR_TAG_STRINGREF_NAMESPACE:
886 {
887 ENTER; SAVETMPS;
888
889 SAVESPTR (dec->stringref);
890 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ());
891
892 sv = decode_sv (dec);
893
894 FREETMPS; LEAVE;
895 }
896 break;
897
898 case CBOR_TAG_STRINGREF:
899 {
900 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
901 ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
902
903 UV idx = decode_uint (dec);
904
905 if (!dec->stringref || (int)idx > AvFILLp (dec->stringref))
906 ERR ("corrupted CBOR data (stringref index out of bounds or outside namespace)");
907
908 sv = newSVsv (AvARRAY (dec->stringref)[idx]);
909 }
910 break;
911
912 case CBOR_TAG_VALUE_SHAREABLE:
913 {
914 if (ecb_expect_false (!dec->shareable))
915 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ());
916
917 sv = newSV (0);
918 av_push (dec->shareable, SvREFCNT_inc_NN (sv));
919
774 SV *sv = decode_sv (dec); 920 SV *osv = decode_sv (dec);
921 sv_setsv (sv, osv);
922 SvREFCNT_dec_NN (osv);
923 }
924 break;
775 925
776 if (tag == 55799) // 2.4.5 Self-Describe CBOR 926 case CBOR_TAG_VALUE_SHAREDREF:
927 {
928 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
929 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
930
931 UV idx = decode_uint (dec);
932
933 if (!dec->shareable || (int)idx > AvFILLp (dec->shareable))
934 ERR ("corrupted CBOR data (sharedref index out of bounds)");
935
936 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]);
937 }
938 break;
939
940 case CBOR_TAG_PERL_OBJECT:
941 {
942 sv = decode_sv (dec);
943
944 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
945 ERR ("corrupted CBOR data (non-array perl object)");
946
947 AV *av = (AV *)SvRV (sv);
948 int len = av_len (av) + 1;
949 HV *stash = gv_stashsv (*av_fetch (av, 0, 1), 0);
950
951 if (!stash)
952 ERR ("cannot decode perl-object (package does not exist)");
953
954 GV *method = gv_fetchmethod_autoload (stash, "THAW", 0);
955
956 if (!method)
957 ERR ("cannot decode perl-object (package does not have a THAW method)");
958
959 dSP;
960
961 ENTER; SAVETMPS; PUSHMARK (SP);
962 EXTEND (SP, len + 1);
963 // we re-bless the reference to get overload and other niceties right
964 PUSHs (*av_fetch (av, 0, 1));
965 PUSHs (sv_cbor);
966
967 int i;
968
969 for (i = 1; i < len; ++i)
970 PUSHs (*av_fetch (av, i, 1));
971
972 PUTBACK;
973 call_sv ((SV *)GvCV (method), G_SCALAR | G_EVAL);
974 SPAGAIN;
975
976 if (SvTRUE (ERRSV))
977 {
978 FREETMPS; LEAVE;
979 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV))));
980 }
981
982 SvREFCNT_dec (sv);
983 sv = SvREFCNT_inc (POPs);
984
985 PUTBACK;
986
987 FREETMPS; LEAVE;
988 }
989 break;
990
991 default:
992 {
993 sv = decode_sv (dec);
994
995 dSP;
996 ENTER; SAVETMPS; PUSHMARK (SP);
997 EXTEND (SP, 2);
998 PUSHs (newSVuv (tag));
999 PUSHs (sv);
1000
1001 PUTBACK;
1002 int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL);
1003 SPAGAIN;
1004
1005 if (SvTRUE (ERRSV))
1006 {
1007 FREETMPS; LEAVE;
1008 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV))));
1009 }
1010
1011 if (count)
1012 {
1013 SvREFCNT_dec (sv);
1014 sv = SvREFCNT_inc (POPs);
1015 }
1016 else
1017 {
1018 AV *av = newAV ();
1019 av_push (av, newSVuv (tag));
1020 av_push (av, sv);
1021
1022 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
1023 ? cbor_tagged_stash
1024 : gv_stashpv ("CBOR::XS::Tagged" , 1);
1025 sv = sv_bless (newRV_noinc ((SV *)av), tagged_stash);
1026 }
1027
1028 PUTBACK;
1029
1030 FREETMPS; LEAVE;
1031 }
1032 break;
1033 }
1034
777 return sv; 1035 return sv;
778 1036
779 AV *av = newAV (); 1037fail:
780 av_push (av, newSVuv (tag)); 1038 SvREFCNT_dec (sv);
781 av_push (av, sv); 1039 return &PL_sv_undef;
782 return newRV_noinc ((SV *)av);
783} 1040}
784 1041
785static SV * 1042static SV *
786decode_sv (dec_t *dec) 1043decode_sv (dec_t *dec)
787{ 1044{
788 WANT (1); 1045 WANT (1);
789 1046
790 switch (*dec->cur >> 5) 1047 switch (*dec->cur >> MAJOR_SHIFT)
791 { 1048 {
792 case 0: // unsigned int 1049 case MAJOR_POS_INT >> MAJOR_SHIFT: return newSVuv (decode_uint (dec));
793 //TODO: 64 bit values on 3 2bit perls 1050 case MAJOR_NEG_INT >> MAJOR_SHIFT: return newSViv (-1 - (IV)decode_uint (dec));
794 return newSVuv (decode_uint (dec)); 1051 case MAJOR_BYTES >> MAJOR_SHIFT: return decode_str (dec, 0);
795 case 1: // negative int 1052 case MAJOR_TEXT >> MAJOR_SHIFT: return decode_str (dec, 1);
796 return newSViv (-1 - (IV)decode_uint (dec)); 1053 case MAJOR_ARRAY >> MAJOR_SHIFT: return decode_av (dec);
797 case 2: // octet string 1054 case MAJOR_MAP >> MAJOR_SHIFT: return decode_hv (dec);
798 return decode_str (dec, 0); 1055 case MAJOR_TAG >> MAJOR_SHIFT: return decode_tagged (dec);
799 case 3: // utf-8 string 1056
800 return decode_str (dec, 1); 1057 case MAJOR_MISC >> MAJOR_SHIFT:
801 case 4: // array
802 return decode_av (dec);
803 case 5: // map
804 return decode_hv (dec);
805 case 6: // tag
806 return decode_tagged (dec);
807 case 7: // misc
808 switch (*dec->cur++ & 31) 1058 switch (*dec->cur++ & MINOR_MASK)
809 { 1059 {
810 case 20: 1060 case SIMPLE_FALSE:
811#if CBOR_SLOW 1061#if CBOR_SLOW
812 cbor_false = get_bool ("CBOR::XS::false"); 1062 types_false = get_bool ("Types::Serialiser::false");
813#endif 1063#endif
814 return newSVsv (cbor_false); 1064 return newSVsv (types_false);
815 case 21: 1065 case SIMPLE_TRUE:
816#if CBOR_SLOW 1066#if CBOR_SLOW
817 cbor_true = get_bool ("CBOR::XS::true"); 1067 types_true = get_bool ("Types::Serialiser::true");
818#endif 1068#endif
819 return newSVsv (cbor_true); 1069 return newSVsv (types_true);
820 case 22: 1070 case SIMPLE_NULL:
821 return newSVsv (&PL_sv_undef); 1071 return newSVsv (&PL_sv_undef);
1072 case SIMPLE_UNDEF:
1073#if CBOR_SLOW
1074 types_error = get_bool ("Types::Serialiser::error");
1075#endif
1076 return newSVsv (types_error);
822 1077
823 case 25: 1078 case MISC_FLOAT16:
824 { 1079 {
825 WANT (2); 1080 WANT (2);
826 1081
827 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1]; 1082 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1];
828 dec->cur += 2; 1083 dec->cur += 2;
829 1084
830 return newSVnv (ecb_binary16_to_float (fp)); 1085 return newSVnv (ecb_binary16_to_float (fp));
831 } 1086 }
832 1087
833 case 26: 1088 case MISC_FLOAT32:
834 { 1089 {
835 uint32_t fp; 1090 uint32_t fp;
836 WANT (4); 1091 WANT (4);
837 memcpy (&fp, dec->cur, 4); 1092 memcpy (&fp, dec->cur, 4);
838 dec->cur += 4; 1093 dec->cur += 4;
841 fp = ecb_bswap32 (fp); 1096 fp = ecb_bswap32 (fp);
842 1097
843 return newSVnv (ecb_binary32_to_float (fp)); 1098 return newSVnv (ecb_binary32_to_float (fp));
844 } 1099 }
845 1100
846 case 27: 1101 case MISC_FLOAT64:
847 { 1102 {
848 uint64_t fp; 1103 uint64_t fp;
849 WANT (8); 1104 WANT (8);
850 memcpy (&fp, dec->cur, 8); 1105 memcpy (&fp, dec->cur, 8);
851 dec->cur += 8; 1106 dec->cur += 8;
854 fp = ecb_bswap64 (fp); 1109 fp = ecb_bswap64 (fp);
855 1110
856 return newSVnv (ecb_binary64_to_double (fp)); 1111 return newSVnv (ecb_binary64_to_double (fp));
857 } 1112 }
858 1113
859 // 0..19 unassigned 1114 // 0..19 unassigned simple
860 // 24 reserved + unassigned (reserved values are not encodable) 1115 // 24 reserved + unassigned (reserved values are not encodable)
861 default: 1116 default:
862 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)"); 1117 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)");
863 } 1118 }
864 1119
865 break; 1120 break;
866 } 1121 }
867#if 0
868 switch (*dec->cur)
869 {
870 //case '"': ++dec->cur; return decode_str (dec);
871 case '[': ++dec->cur; return decode_av (dec);
872 case '{': ++dec->cur; return decode_hv (dec);
873
874 case '-':
875 case '0': case '1': case '2': case '3': case '4':
876 case '5': case '6': case '7': case '8': case '9':
877 //TODO return decode_num (dec);
878
879 case 't':
880 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
881 {
882 dec->cur += 4;
883#if CBOR_SLOW
884 cbor_true = get_bool ("CBOR::XS::true");
885#endif
886 return newSVsv (cbor_true);
887 }
888 else
889 ERR ("'true' expected");
890
891 break;
892
893 case 'f':
894 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
895 {
896 dec->cur += 5;
897#if CBOR_SLOW
898 cbor_false = get_bool ("CBOR::XS::false");
899#endif
900 return newSVsv (cbor_false);
901 }
902 else
903 ERR ("'false' expected");
904
905 break;
906
907 case 'n':
908 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "null", 4))
909 {
910 dec->cur += 4;
911 return newSVsv (&PL_sv_undef);
912 }
913 else
914 ERR ("'null' expected");
915
916 break;
917
918 default:
919 ERR ("malformed CBOR string, neither array, object, number, string or atom");
920 break;
921 }
922#endif
923 1122
924fail: 1123fail:
925 return &PL_sv_undef; 1124 return &PL_sv_undef;
926} 1125}
927 1126
928static SV * 1127static SV *
929decode_cbor (SV *string, CBOR *cbor, char **offset_return) 1128decode_cbor (SV *string, CBOR *cbor, char **offset_return)
930{ 1129{
931 dec_t dec; 1130 dec_t dec = { };
932 SV *sv; 1131 SV *sv;
1132 STRLEN len;
1133 char *data = SvPVbyte (string, len);
933 1134
934 /* work around bugs in 5.10 where manipulating magic values
935 * makes perl ignore the magic in subsequent accesses.
936 * also make a copy of non-PV values, to get them into a clean
937 * state (SvPV should do that, but it's buggy, see below).
938 */
939 /*SvGETMAGIC (string);*/
940 if (SvMAGICAL (string) || !SvPOK (string))
941 string = sv_2mortal (newSVsv (string));
942
943 SvUPGRADE (string, SVt_PV);
944
945 /* work around a bug in perl 5.10, which causes SvCUR to fail an
946 * assertion with -DDEBUGGING, although SvCUR is documented to
947 * return the xpv_cur field which certainly exists after upgrading.
948 * according to nicholas clark, calling SvPOK fixes this.
949 * But it doesn't fix it, so try another workaround, call SvPV_nolen
950 * and hope for the best.
951 * Damnit, SvPV_nolen still trips over yet another assertion. This
952 * assertion business is seriously broken, try yet another workaround
953 * for the broken -DDEBUGGING.
954 */
955 {
956#ifdef DEBUGGING
957 STRLEN offset = SvOK (string) ? sv_len (string) : 0;
958#else
959 STRLEN offset = SvCUR (string);
960#endif
961
962 if (offset > cbor->max_size && cbor->max_size) 1135 if (len > cbor->max_size && cbor->max_size)
963 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu", 1136 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
964 (unsigned long)SvCUR (string), (unsigned long)cbor->max_size); 1137 (unsigned long)len, (unsigned long)cbor->max_size);
965 }
966
967 sv_utf8_downgrade (string, 0);
968 1138
969 dec.cbor = *cbor; 1139 dec.cbor = *cbor;
970 dec.cur = (U8 *)SvPVX (string); 1140 dec.cur = (U8 *)data;
971 dec.end = (U8 *)SvEND (string); 1141 dec.end = (U8 *)data + len;
972 dec.err = 0;
973 dec.depth = 0;
974
975 if (dec.cbor.cb_object || dec.cbor.cb_sk_object)
976 ;//TODO dec.cbor.flags |= F_HOOK;
977 1142
978 sv = decode_sv (&dec); 1143 sv = decode_sv (&dec);
979 1144
980 if (offset_return) 1145 if (offset_return)
981 *offset_return = dec.cur; 1146 *offset_return = dec.cur;
985 dec.err = "garbage after CBOR object"; 1150 dec.err = "garbage after CBOR object";
986 1151
987 if (dec.err) 1152 if (dec.err)
988 { 1153 {
989 SvREFCNT_dec (sv); 1154 SvREFCNT_dec (sv);
990 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)SvPVX (string), (int)(uint8_t)*dec.cur); 1155 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur);
991 } 1156 }
992 1157
993 sv = sv_2mortal (sv); 1158 sv = sv_2mortal (sv);
994 1159
995 return sv; 1160 return sv;
1001MODULE = CBOR::XS PACKAGE = CBOR::XS 1166MODULE = CBOR::XS PACKAGE = CBOR::XS
1002 1167
1003BOOT: 1168BOOT:
1004{ 1169{
1005 cbor_stash = gv_stashpv ("CBOR::XS" , 1); 1170 cbor_stash = gv_stashpv ("CBOR::XS" , 1);
1006 cbor_boolean_stash = gv_stashpv ("CBOR::XS::Boolean", 1); 1171 cbor_tagged_stash = gv_stashpv ("CBOR::XS::Tagged" , 1);
1007 1172
1008 cbor_true = get_bool ("CBOR::XS::true"); 1173 types_boolean_stash = gv_stashpv ("Types::Serialiser::Boolean", 1);
1009 cbor_false = get_bool ("CBOR::XS::false"); 1174 types_error_stash = gv_stashpv ("Types::Serialiser::Error" , 1);
1175
1176 types_true = get_bool ("Types::Serialiser::true" );
1177 types_false = get_bool ("Types::Serialiser::false");
1178 types_error = get_bool ("Types::Serialiser::error");
1179
1180 default_filter = newSVpv ("CBOR::XS::default_filter", 0);
1181
1182 sv_cbor = newSVpv ("CBOR", 0);
1183 SvREADONLY_on (sv_cbor);
1010} 1184}
1011 1185
1012PROTOTYPES: DISABLE 1186PROTOTYPES: DISABLE
1013 1187
1014void CLONE (...) 1188void CLONE (...)
1015 CODE: 1189 CODE:
1016 cbor_stash = 0; 1190 cbor_stash = 0;
1191 cbor_tagged_stash = 0;
1192 types_error_stash = 0;
1017 cbor_boolean_stash = 0; 1193 types_boolean_stash = 0;
1018 1194
1019void new (char *klass) 1195void new (char *klass)
1020 PPCODE: 1196 PPCODE:
1021{ 1197{
1022 SV *pv = NEWSV (0, sizeof (CBOR)); 1198 SV *pv = NEWSV (0, sizeof (CBOR));
1030 1206
1031void shrink (CBOR *self, int enable = 1) 1207void shrink (CBOR *self, int enable = 1)
1032 ALIAS: 1208 ALIAS:
1033 shrink = F_SHRINK 1209 shrink = F_SHRINK
1034 allow_unknown = F_ALLOW_UNKNOWN 1210 allow_unknown = F_ALLOW_UNKNOWN
1211 allow_sharing = F_ALLOW_SHARING
1212 pack_strings = F_PACK_STRINGS
1035 PPCODE: 1213 PPCODE:
1036{ 1214{
1037 if (enable) 1215 if (enable)
1038 self->flags |= ix; 1216 self->flags |= ix;
1039 else 1217 else
1044 1222
1045void get_shrink (CBOR *self) 1223void get_shrink (CBOR *self)
1046 ALIAS: 1224 ALIAS:
1047 get_shrink = F_SHRINK 1225 get_shrink = F_SHRINK
1048 get_allow_unknown = F_ALLOW_UNKNOWN 1226 get_allow_unknown = F_ALLOW_UNKNOWN
1227 get_allow_sharing = F_ALLOW_SHARING
1228 get_pack_strings = F_PACK_STRINGS
1049 PPCODE: 1229 PPCODE:
1050 XPUSHs (boolSV (self->flags & ix)); 1230 XPUSHs (boolSV (self->flags & ix));
1051 1231
1052void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) 1232void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
1053 PPCODE: 1233 PPCODE:
1069 CODE: 1249 CODE:
1070 RETVAL = self->max_size; 1250 RETVAL = self->max_size;
1071 OUTPUT: 1251 OUTPUT:
1072 RETVAL 1252 RETVAL
1073 1253
1074#if 0 //TODO 1254void filter (CBOR *self, SV *filter = 0)
1075
1076void filter_cbor_object (CBOR *self, SV *cb = &PL_sv_undef)
1077 PPCODE: 1255 PPCODE:
1078{
1079 SvREFCNT_dec (self->cb_object); 1256 SvREFCNT_dec (self->filter);
1080 self->cb_object = SvOK (cb) ? newSVsv (cb) : 0; 1257 self->filter = filter ? newSVsv (filter) : filter;
1081
1082 XPUSHs (ST (0)); 1258 XPUSHs (ST (0));
1083}
1084 1259
1085void filter_cbor_single_key_object (CBOR *self, SV *key, SV *cb = &PL_sv_undef) 1260SV *get_filter (CBOR *self)
1086 PPCODE: 1261 CODE:
1087{ 1262 RETVAL = self->filter ? self->filter : NEWSV (0, 0);
1088 if (!self->cb_sk_object) 1263 OUTPUT:
1089 self->cb_sk_object = newHV (); 1264 RETVAL
1090
1091 if (SvOK (cb))
1092 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);
1093 else
1094 {
1095 hv_delete_ent (self->cb_sk_object, key, G_DISCARD, 0);
1096
1097 if (!HvKEYS (self->cb_sk_object))
1098 {
1099 SvREFCNT_dec (self->cb_sk_object);
1100 self->cb_sk_object = 0;
1101 }
1102 }
1103
1104 XPUSHs (ST (0));
1105}
1106
1107#endif
1108 1265
1109void encode (CBOR *self, SV *scalar) 1266void encode (CBOR *self, SV *scalar)
1110 PPCODE: 1267 PPCODE:
1111 PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN; 1268 PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN;
1112 XPUSHs (scalar); 1269 XPUSHs (scalar);
1126 PUSHs (sv); 1283 PUSHs (sv);
1127 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); 1284 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
1128} 1285}
1129 1286
1130void DESTROY (CBOR *self) 1287void DESTROY (CBOR *self)
1131 CODE: 1288 PPCODE:
1132 SvREFCNT_dec (self->cb_sk_object); 1289 cbor_free (self);
1133 SvREFCNT_dec (self->cb_object);
1134 1290
1135PROTOTYPES: ENABLE 1291PROTOTYPES: ENABLE
1136 1292
1137void encode_cbor (SV *scalar) 1293void encode_cbor (SV *scalar)
1294 ALIAS:
1295 encode_cbor = 0
1296 encode_cbor_sharing = F_ALLOW_SHARING
1138 PPCODE: 1297 PPCODE:
1139{ 1298{
1140 CBOR cbor; 1299 CBOR cbor;
1141 cbor_init (&cbor); 1300 cbor_init (&cbor);
1301 cbor.flags |= ix;
1142 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN; 1302 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;
1143 XPUSHs (scalar); 1303 XPUSHs (scalar);
1144} 1304}
1145 1305
1146void decode_cbor (SV *cborstr) 1306void decode_cbor (SV *cborstr)

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines