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 1.7 by root, Sat Oct 26 23:02:55 2013 UTC vs.
Revision 1.51 by root, Sun Apr 24 13:15:19 2016 UTC

7#include <stdlib.h> 7#include <stdlib.h>
8#include <stdio.h> 8#include <stdio.h>
9#include <limits.h> 9#include <limits.h>
10#include <float.h> 10#include <float.h>
11 11
12#define ECB_NO_THREADS 1
12#include "ecb.h" 13#include "ecb.h"
13 14
15// compatibility with perl <5.18
16#ifndef HvNAMELEN_get
17# define HvNAMELEN_get(hv) strlen (HvNAME (hv))
18#endif
19#ifndef HvNAMELEN
20# define HvNAMELEN(hv) HvNAMELEN_get (hv)
21#endif
22#ifndef HvNAMEUTF8
23# define HvNAMEUTF8(hv) 0
24#endif
25#ifndef SvREFCNT_dec_NN
26# define SvREFCNT_dec_NN(sv) SvREFCNT_dec (sv)
27#endif
28
29// known major and minor types
30enum cbor_type
31{
32 MAJOR_SHIFT = 5,
33 MINOR_MASK = 0x1f,
34
35 MAJOR_POS_INT = 0 << MAJOR_SHIFT,
36 MAJOR_NEG_INT = 1 << MAJOR_SHIFT,
37 MAJOR_BYTES = 2 << MAJOR_SHIFT,
38 MAJOR_TEXT = 3 << MAJOR_SHIFT,
39 MAJOR_ARRAY = 4 << MAJOR_SHIFT,
40 MAJOR_MAP = 5 << MAJOR_SHIFT,
41 MAJOR_TAG = 6 << MAJOR_SHIFT,
42 MAJOR_MISC = 7 << MAJOR_SHIFT,
43
44 // INT/STRING/ARRAY/MAP subtypes
45 LENGTH_EXT1 = 24,
46 LENGTH_EXT2 = 25,
47 LENGTH_EXT4 = 26,
48 LENGTH_EXT8 = 27,
49
50 // SIMPLE types (effectively MISC subtypes)
51 SIMPLE_FALSE = 20,
52 SIMPLE_TRUE = 21,
53 SIMPLE_NULL = 22,
54 SIMPLE_UNDEF = 23,
55
56 // MISC subtype (unused)
57 MISC_EXT1 = 24,
58 MISC_FLOAT16 = 25,
59 MISC_FLOAT32 = 26,
60 MISC_FLOAT64 = 27,
61
62 // BYTES/TEXT/ARRAY/MAP
63 MINOR_INDEF = 31,
64};
65
66// known tags
67enum cbor_tag
68{
69 // extensions
70 CBOR_TAG_STRINGREF = 25, // http://cbor.schmorp.de/stringref
71 CBOR_TAG_PERL_OBJECT = 26, // http://cbor.schmorp.de/perl-object
72 CBOR_TAG_GENERIC_OBJECT = 27, // http://cbor.schmorp.de/generic-object
73 CBOR_TAG_VALUE_SHAREABLE = 28, // http://cbor.schmorp.de/value-sharing
74 CBOR_TAG_VALUE_SHAREDREF = 29, // http://cbor.schmorp.de/value-sharing
75 CBOR_TAG_STRINGREF_NAMESPACE = 256, // http://cbor.schmorp.de/stringref
76 CBOR_TAG_INDIRECTION = 22098, // http://cbor.schmorp.de/indirection
77
78 // rfc7049
79 CBOR_TAG_DATETIME = 0, // rfc4287, utf-8
80 CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any
81 CBOR_TAG_POS_BIGNUM = 2, // byte string
82 CBOR_TAG_NEG_BIGNUM = 3, // byte string
83 CBOR_TAG_DECIMAL = 4, // decimal fraction, array
84 CBOR_TAG_BIGFLOAT = 5, // array
85
86 CBOR_TAG_CONV_B64U = 21, // base64url, any
87 CBOR_TAG_CONV_B64 = 22, // base64, any
88 CBOR_TAG_CONV_HEX = 23, // base16, any
89 CBOR_TAG_CBOR = 24, // embedded cbor, byte string
90
91 CBOR_TAG_URI = 32, // URI rfc3986, utf-8
92 CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8
93 CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8
94 CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8
95 CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8
96
97 CBOR_TAG_MAGIC = 55799, // self-describe cbor
98};
99
14#define F_SHRINK 0x00000200UL 100#define F_SHRINK 0x00000001UL
15#define F_ALLOW_UNKNOWN 0x00002000UL 101#define F_ALLOW_UNKNOWN 0x00000002UL
102#define F_ALLOW_SHARING 0x00000004UL
103#define F_ALLOW_CYCLES 0x00000008UL
104#define F_PACK_STRINGS 0x00000010UL
105#define F_UTF8_STRINGS 0x00000020UL
106#define F_VALIDATE_UTF8 0x00000040UL
16 107
17#define INIT_SIZE 32 // initial scalar size to be allocated 108#define INIT_SIZE 32 // initial scalar size to be allocated
18 109
19#define SB do { 110#define SB do {
20#define SE } while (0) 111#define SE } while (0)
31#else 122#else
32# define CBOR_SLOW 0 123# define CBOR_SLOW 0
33# define CBOR_STASH cbor_stash 124# define CBOR_STASH cbor_stash
34#endif 125#endif
35 126
36static HV *cbor_stash, *cbor_boolean_stash, *cbor_tagged_stash; // CBOR::XS:: 127static HV *cbor_stash, *types_boolean_stash, *types_error_stash, *cbor_tagged_stash; // CBOR::XS::
37static SV *cbor_true, *cbor_false; 128static SV *types_true, *types_false, *types_error, *sv_cbor, *default_filter;
38 129
39typedef struct { 130typedef struct {
40 U32 flags; 131 U32 flags;
41 U32 max_depth; 132 U32 max_depth;
42 STRLEN max_size; 133 STRLEN max_size;
134 SV *filter;
135
136 // for the incremental parser
137 STRLEN incr_pos; // the current offset into the text
138 STRLEN incr_need; // minimum bytes needed to decode
139 AV *incr_count; // for every nesting level, the number of outstanding values, or -1 for indef.
43} CBOR; 140} CBOR;
44 141
45ecb_inline void 142ecb_inline void
46cbor_init (CBOR *cbor) 143cbor_init (CBOR *cbor)
47{ 144{
48 Zero (cbor, 1, CBOR); 145 Zero (cbor, 1, CBOR);
49 cbor->max_depth = 512; 146 cbor->max_depth = 512;
147}
148
149ecb_inline void
150cbor_free (CBOR *cbor)
151{
152 SvREFCNT_dec (cbor->filter);
153 SvREFCNT_dec (cbor->incr_count);
50} 154}
51 155
52///////////////////////////////////////////////////////////////////////////// 156/////////////////////////////////////////////////////////////////////////////
53// utility functions 157// utility functions
54 158
76 SvPV_renew (sv, SvCUR (sv) + 1); 180 SvPV_renew (sv, SvCUR (sv) + 1);
77#endif 181#endif
78 } 182 }
79} 183}
80 184
81///////////////////////////////////////////////////////////////////////////// 185// minimum length of a string to be registered for stringref
82// fp hell 186ecb_inline int
83 187minimum_string_length (UV idx)
84//TODO 188{
189 return idx > 23
190 ? idx > 0xffU
191 ? idx > 0xffffU
192 ? idx > 0xffffffffU
193 ? 11
194 : 7
195 : 5
196 : 4
197 : 3;
198}
85 199
86///////////////////////////////////////////////////////////////////////////// 200/////////////////////////////////////////////////////////////////////////////
87// encoder 201// encoder
88 202
89// structure used for encoding CBOR 203// structure used for encoding CBOR
92 char *cur; // SvPVX (sv) + current output position 206 char *cur; // SvPVX (sv) + current output position
93 char *end; // SvEND (sv) 207 char *end; // SvEND (sv)
94 SV *sv; // result scalar 208 SV *sv; // result scalar
95 CBOR cbor; 209 CBOR cbor;
96 U32 depth; // recursion level 210 U32 depth; // recursion level
211 HV *stringref[2]; // string => index, or 0 ([0] = bytes, [1] = utf-8)
212 UV stringref_idx;
213 HV *shareable; // ptr => index, or 0
214 UV shareable_idx;
97} enc_t; 215} enc_t;
98 216
99ecb_inline void 217ecb_inline void
100need (enc_t *enc, STRLEN len) 218need (enc_t *enc, STRLEN len)
101{ 219{
118static void 236static void
119encode_uint (enc_t *enc, int major, UV len) 237encode_uint (enc_t *enc, int major, UV len)
120{ 238{
121 need (enc, 9); 239 need (enc, 9);
122 240
123 if (len < 24) 241 if (ecb_expect_true (len < LENGTH_EXT1))
124 *enc->cur++ = major | len; 242 *enc->cur++ = major | len;
125 else if (len <= 0xff) 243 else if (ecb_expect_true (len <= 0xffU))
126 { 244 {
127 *enc->cur++ = major | 24; 245 *enc->cur++ = major | LENGTH_EXT1;
128 *enc->cur++ = len; 246 *enc->cur++ = len;
129 } 247 }
130 else if (len <= 0xffff) 248 else if (len <= 0xffffU)
131 { 249 {
132 *enc->cur++ = major | 25; 250 *enc->cur++ = major | LENGTH_EXT2;
133 *enc->cur++ = len >> 8; 251 *enc->cur++ = len >> 8;
134 *enc->cur++ = len; 252 *enc->cur++ = len;
135 } 253 }
136 else if (len <= 0xffffffff) 254 else if (len <= 0xffffffffU)
137 { 255 {
138 *enc->cur++ = major | 26; 256 *enc->cur++ = major | LENGTH_EXT4;
139 *enc->cur++ = len >> 24; 257 *enc->cur++ = len >> 24;
140 *enc->cur++ = len >> 16; 258 *enc->cur++ = len >> 16;
141 *enc->cur++ = len >> 8; 259 *enc->cur++ = len >> 8;
142 *enc->cur++ = len; 260 *enc->cur++ = len;
143 } 261 }
144 else 262 else
145 { 263 {
146 *enc->cur++ = major | 27; 264 *enc->cur++ = major | LENGTH_EXT8;
147 *enc->cur++ = len >> 56; 265 *enc->cur++ = len >> 56;
148 *enc->cur++ = len >> 48; 266 *enc->cur++ = len >> 48;
149 *enc->cur++ = len >> 40; 267 *enc->cur++ = len >> 40;
150 *enc->cur++ = len >> 32; 268 *enc->cur++ = len >> 32;
151 *enc->cur++ = len >> 24; 269 *enc->cur++ = len >> 24;
153 *enc->cur++ = len >> 8; 271 *enc->cur++ = len >> 8;
154 *enc->cur++ = len; 272 *enc->cur++ = len;
155 } 273 }
156} 274}
157 275
158static void 276ecb_inline void
277encode_tag (enc_t *enc, UV tag)
278{
279 encode_uint (enc, MAJOR_TAG, tag);
280}
281
282ecb_inline void
159encode_str (enc_t *enc, int utf8, char *str, STRLEN len) 283encode_str (enc_t *enc, int utf8, char *str, STRLEN len)
160{ 284{
161 encode_uint (enc, utf8 ? 0x60 : 0x40, len); 285 if (ecb_expect_false (enc->cbor.flags & F_UTF8_STRINGS))
286 if (!utf8)
287 {
288 SV *sv = sv_newmortal ();
289 char *s; STRLEN l;
290
291 sv_setpvn (sv, str, len);
292
293 s = SvPVutf8 (sv, l);
294 encode_str (enc, 1, s, l);
295 return;
296 }
297
298 encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len);
162 need (enc, len); 299 need (enc, len);
163 memcpy (enc->cur, str, len); 300 memcpy (enc->cur, str, len);
164 enc->cur += len; 301 enc->cur += len;
165} 302}
166 303
304static void
305encode_strref (enc_t *enc, int utf8, char *str, STRLEN len)
306{
307 if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS))
308 {
309 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1);
310
311 if (SvOK (*svp))
312 {
313 // already registered, use stringref
314 encode_tag (enc, CBOR_TAG_STRINGREF);
315 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
316 return;
317 }
318 else if (len >= minimum_string_length (enc->stringref_idx))
319 {
320 // register only
321 sv_setuv (*svp, enc->stringref_idx);
322 ++enc->stringref_idx;
323 }
324 }
325
326 encode_str (enc, utf8, str, len);
327}
328
167static void encode_sv (enc_t *enc, SV *sv); 329static void encode_sv (enc_t *enc, SV *sv);
168 330
169static void 331static void
170encode_av (enc_t *enc, AV *av) 332encode_av (enc_t *enc, AV *av)
171{ 333{
174 if (enc->depth >= enc->cbor.max_depth) 336 if (enc->depth >= enc->cbor.max_depth)
175 croak (ERR_NESTING_EXCEEDED); 337 croak (ERR_NESTING_EXCEEDED);
176 338
177 ++enc->depth; 339 ++enc->depth;
178 340
179 encode_uint (enc, 0x80, len + 1); 341 encode_uint (enc, MAJOR_ARRAY, len + 1);
180 342
343 if (SvMAGICAL (av))
181 for (i = 0; i <= len; ++i) 344 for (i = 0; i <= len; ++i)
182 { 345 {
183 SV **svp = av_fetch (av, i, 0); 346 SV **svp = av_fetch (av, i, 0);
184 encode_sv (enc, svp ? *svp : &PL_sv_undef); 347 encode_sv (enc, svp ? *svp : &PL_sv_undef);
185 } 348 }
349 else
350 for (i = 0; i <= len; ++i)
351 {
352 SV *sv = AvARRAY (av)[i];
353 encode_sv (enc, sv ? sv : &PL_sv_undef);
354 }
186 355
187 --enc->depth; 356 --enc->depth;
188} 357}
189 358
190static void 359static void
199 368
200 int pairs = hv_iterinit (hv); 369 int pairs = hv_iterinit (hv);
201 int mg = SvMAGICAL (hv); 370 int mg = SvMAGICAL (hv);
202 371
203 if (mg) 372 if (mg)
204 encode_ch (enc, 0xa0 | 31); 373 encode_ch (enc, MAJOR_MAP | MINOR_INDEF);
205 else 374 else
206 encode_uint (enc, 0xa0, pairs); 375 encode_uint (enc, MAJOR_MAP, pairs);
207 376
208 while ((he = hv_iternext (hv))) 377 while ((he = hv_iternext (hv)))
209 { 378 {
210 if (HeKLEN (he) == HEf_SVKEY) 379 if (HeKLEN (he) == HEf_SVKEY)
211 encode_sv (enc, HeSVKEY (he)); 380 encode_sv (enc, HeSVKEY (he));
212 else 381 else
213 encode_str (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he)); 382 encode_strref (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he));
214 383
215 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he)); 384 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he));
216 } 385 }
217 386
218 if (mg) 387 if (mg)
219 encode_ch (enc, 0xe0 | 31); 388 encode_ch (enc, MAJOR_MISC | MINOR_INDEF);
220 389
221 --enc->depth; 390 --enc->depth;
222} 391}
223 392
224// encode objects, arrays and special \0=false and \1=true values. 393// encode objects, arrays and special \0=false and \1=true values.
225static void 394static void
226encode_rv (enc_t *enc, SV *sv) 395encode_rv (enc_t *enc, SV *sv)
227{ 396{
228 svtype svt;
229
230 SvGETMAGIC (sv); 397 SvGETMAGIC (sv);
398
231 svt = SvTYPE (sv); 399 svtype svt = SvTYPE (sv);
232 400
233 if (ecb_expect_false (SvOBJECT (sv))) 401 if (ecb_expect_false (SvOBJECT (sv)))
234 { 402 {
235 HV *boolean_stash = !CBOR_SLOW || cbor_boolean_stash 403 HV *boolean_stash = !CBOR_SLOW || types_boolean_stash
236 ? cbor_boolean_stash 404 ? types_boolean_stash
237 : gv_stashpv ("CBOR::XS::Boolean", 1); 405 : gv_stashpv ("Types::Serialiser::Boolean", 1);
406 HV *error_stash = !CBOR_SLOW || types_error_stash
407 ? types_error_stash
408 : gv_stashpv ("Types::Serialiser::Error", 1);
238 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 409 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
239 ? cbor_tagged_stash 410 ? cbor_tagged_stash
240 : gv_stashpv ("CBOR::XS::Tagged" , 1); 411 : gv_stashpv ("CBOR::XS::Tagged" , 1);
241 412
413 HV *stash = SvSTASH (sv);
414
242 if (SvSTASH (sv) == boolean_stash) 415 if (stash == boolean_stash)
243 encode_ch (enc, SvIV (sv) ? 0xe0 | 21 : 0xe0 | 20); 416 {
417 encode_ch (enc, SvIV (sv) ? MAJOR_MISC | SIMPLE_TRUE : MAJOR_MISC | SIMPLE_FALSE);
418 return;
419 }
420 else if (stash == error_stash)
421 {
422 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
423 return;
424 }
244 else if (SvSTASH (sv) == tagged_stash) 425 else if (stash == tagged_stash)
245 { 426 {
246 if (svt != SVt_PVAV) 427 if (svt != SVt_PVAV)
247 croak ("encountered CBOR::XS::Tagged object that isn't an array"); 428 croak ("encountered CBOR::XS::Tagged object that isn't an array");
248 429
249 encode_uint (enc, 0xc0, SvUV (*av_fetch ((AV *)sv, 0, 1))); 430 encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1)));
250 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1)); 431 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1));
432
433 return;
434 }
435 }
436
437 if (ecb_expect_false (SvREFCNT (sv) > 1)
438 && ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING))
439 {
440 if (!enc->shareable)
441 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
442
443 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
444
445 if (SvOK (*svp))
446 {
447 encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF);
448 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
449 return;
251 } 450 }
252 else 451 else
253 { 452 {
453 sv_setuv (*svp, enc->shareable_idx);
454 ++enc->shareable_idx;
455 encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE);
456 }
457 }
458
459 if (ecb_expect_false (SvOBJECT (sv)))
460 {
461 HV *stash = SvSTASH (sv);
462 GV *method;
463
464 if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
465 {
466 dSP;
467
468 ENTER; SAVETMPS;
469 PUSHMARK (SP);
254 // we re-bless the reference to get overload and other niceties right 470 // we re-bless the reference to get overload and other niceties right
255 GV *to_cbor = gv_fetchmethod_autoload (SvSTASH (sv), "TO_CBOR", 0);
256
257 if (to_cbor)
258 {
259 dSP;
260
261 ENTER; SAVETMPS; PUSHMARK (SP);
262 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv))); 471 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
263 472
264 // calling with G_SCALAR ensures that we always get a 1 return value
265 PUTBACK; 473 PUTBACK;
474 // G_SCALAR ensures that return value is 1
266 call_sv ((SV *)GvCV (to_cbor), G_SCALAR); 475 call_sv ((SV *)GvCV (method), G_SCALAR);
267 SPAGAIN; 476 SPAGAIN;
268 477
269 // catch this surprisingly common error 478 // catch this surprisingly common error
270 if (SvROK (TOPs) && SvRV (TOPs) == sv) 479 if (SvROK (TOPs) && SvRV (TOPs) == sv)
271 croak ("%s::TO_CBOR method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv))); 480 croak ("%s::TO_CBOR method returned same object as was passed instead of a new one", HvNAME (stash));
272 481
273 sv = POPs; 482 encode_sv (enc, POPs);
483
274 PUTBACK; 484 PUTBACK;
275 485
276 encode_sv (enc, sv);
277
278 FREETMPS; LEAVE; 486 FREETMPS; LEAVE;
279 } 487 }
488 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)
489 {
490 dSP;
491
492 ENTER; SAVETMPS;
493 SAVESTACK_POS ();
494 PUSHMARK (SP);
495 EXTEND (SP, 2);
496 // we re-bless the reference to get overload and other niceties right
497 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
498 PUSHs (sv_cbor);
499
500 PUTBACK;
501 int count = call_sv ((SV *)GvCV (method), G_ARRAY);
502 SPAGAIN;
503
504 // catch this surprisingly common error
505 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)
506 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));
507
508 encode_tag (enc, CBOR_TAG_PERL_OBJECT);
509 encode_uint (enc, MAJOR_ARRAY, count + 1);
510 encode_strref (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
511
512 while (count)
513 encode_sv (enc, SP[1 - count--]);
514
515 PUTBACK;
516
517 FREETMPS; LEAVE;
518 }
280 else 519 else
281 croak ("encountered object '%s', but no TO_CBOR method available on it", 520 croak ("encountered object '%s', but no TO_CBOR or FREEZE methods available on it",
282 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 521 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
283 }
284 } 522 }
285 else if (svt == SVt_PVHV) 523 else if (svt == SVt_PVHV)
286 encode_hv (enc, (HV *)sv); 524 encode_hv (enc, (HV *)sv);
287 else if (svt == SVt_PVAV) 525 else if (svt == SVt_PVAV)
288 encode_av (enc, (AV *)sv); 526 encode_av (enc, (AV *)sv);
289 else if (svt < SVt_PVAV)
290 {
291 STRLEN len = 0;
292 char *pv = svt ? SvPV (sv, len) : 0;
293
294 if (len == 1 && *pv == '1')
295 encode_ch (enc, 0xe0 | 21);
296 else if (len == 1 && *pv == '0')
297 encode_ch (enc, 0xe0 | 20);
298 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
299 encode_ch (enc, 0xe0 | 23);
300 else
301 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
302 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
303 }
304 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
305 encode_ch (enc, 0xe0 | 23);
306 else 527 else
307 croak ("encountered %s, but CBOR can only represent references to arrays or hashes", 528 {
308 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 529 encode_tag (enc, CBOR_TAG_INDIRECTION);
530 encode_sv (enc, sv);
531 }
309} 532}
310 533
311static void 534static void
312encode_nv (enc_t *enc, SV *sv) 535encode_nv (enc_t *enc, SV *sv)
313{ 536{
314 double nv = SvNVX (sv); 537 double nv = SvNVX (sv);
315 538
316 need (enc, 9); 539 need (enc, 9);
317 540
318 if (ecb_expect_false (nv == (U32)nv)) 541 if (ecb_expect_false (nv == (NV)(U32)nv))
319 encode_uint (enc, 0x00, (U32)nv); 542 encode_uint (enc, MAJOR_POS_INT, (U32)nv);
320 //TODO: maybe I32? 543 //TODO: maybe I32?
321 else if (ecb_expect_false (nv == (float)nv)) 544 else if (ecb_expect_false (nv == (float)nv))
322 { 545 {
323 uint32_t fp = ecb_float_to_binary32 (nv); 546 uint32_t fp = ecb_float_to_binary32 (nv);
324 547
325 *enc->cur++ = 0xe0 | 26; 548 *enc->cur++ = MAJOR_MISC | MISC_FLOAT32;
326 549
327 if (!ecb_big_endian ()) 550 if (!ecb_big_endian ())
328 fp = ecb_bswap32 (fp); 551 fp = ecb_bswap32 (fp);
329 552
330 memcpy (enc->cur, &fp, 4); 553 memcpy (enc->cur, &fp, 4);
332 } 555 }
333 else 556 else
334 { 557 {
335 uint64_t fp = ecb_double_to_binary64 (nv); 558 uint64_t fp = ecb_double_to_binary64 (nv);
336 559
337 *enc->cur++ = 0xe0 | 27; 560 *enc->cur++ = MAJOR_MISC | MISC_FLOAT64;
338 561
339 if (!ecb_big_endian ()) 562 if (!ecb_big_endian ())
340 fp = ecb_bswap64 (fp); 563 fp = ecb_bswap64 (fp);
341 564
342 memcpy (enc->cur, &fp, 8); 565 memcpy (enc->cur, &fp, 8);
351 574
352 if (SvPOKp (sv)) 575 if (SvPOKp (sv))
353 { 576 {
354 STRLEN len; 577 STRLEN len;
355 char *str = SvPV (sv, len); 578 char *str = SvPV (sv, len);
356 encode_str (enc, SvUTF8 (sv), str, len); 579 encode_strref (enc, SvUTF8 (sv), str, len);
357 } 580 }
358 else if (SvNOKp (sv)) 581 else if (SvNOKp (sv))
359 encode_nv (enc, sv); 582 encode_nv (enc, sv);
360 else if (SvIOKp (sv)) 583 else if (SvIOKp (sv))
361 { 584 {
362 if (SvIsUV (sv)) 585 if (SvIsUV (sv))
363 encode_uint (enc, 0x00, SvUVX (sv)); 586 encode_uint (enc, MAJOR_POS_INT, SvUVX (sv));
364 else if (SvIVX (sv) >= 0) 587 else if (SvIVX (sv) >= 0)
365 encode_uint (enc, 0x00, SvIVX (sv)); 588 encode_uint (enc, MAJOR_POS_INT, SvIVX (sv));
366 else 589 else
367 encode_uint (enc, 0x20, -(SvIVX (sv) + 1)); 590 encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1));
368 } 591 }
369 else if (SvROK (sv)) 592 else if (SvROK (sv))
370 encode_rv (enc, SvRV (sv)); 593 encode_rv (enc, SvRV (sv));
371 else if (!SvOK (sv)) 594 else if (!SvOK (sv))
372 encode_ch (enc, 0xe0 | 22); 595 encode_ch (enc, MAJOR_MISC | SIMPLE_NULL);
373 else if (enc->cbor.flags & F_ALLOW_UNKNOWN) 596 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
374 encode_ch (enc, 0xe0 | 23); 597 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
375 else 598 else
376 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data", 599 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data",
377 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv)); 600 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
378} 601}
379 602
380static SV * 603static SV *
381encode_cbor (SV *scalar, CBOR *cbor) 604encode_cbor (SV *scalar, CBOR *cbor)
382{ 605{
383 enc_t enc; 606 enc_t enc = { 0 };
384 607
385 enc.cbor = *cbor; 608 enc.cbor = *cbor;
386 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 609 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
387 enc.cur = SvPVX (enc.sv); 610 enc.cur = SvPVX (enc.sv);
388 enc.end = SvEND (enc.sv); 611 enc.end = SvEND (enc.sv);
389 enc.depth = 0;
390 612
391 SvPOK_only (enc.sv); 613 SvPOK_only (enc.sv);
614
615 if (cbor->flags & F_PACK_STRINGS)
616 {
617 encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE);
618 enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ());
619 enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ());
620 }
621
392 encode_sv (&enc, scalar); 622 encode_sv (&enc, scalar);
393 623
394 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 624 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
395 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings 625 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
396 626
410 U8 *end; // end of input string 640 U8 *end; // end of input string
411 const char *err; // parse error, if != 0 641 const char *err; // parse error, if != 0
412 CBOR cbor; 642 CBOR cbor;
413 U32 depth; // recursion depth 643 U32 depth; // recursion depth
414 U32 maxdepth; // recursion depth limit 644 U32 maxdepth; // recursion depth limit
645 AV *shareable;
646 AV *stringref;
647 SV *decode_tagged;
415} dec_t; 648} dec_t;
416 649
417#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE 650#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE
418 651
419#define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data") 652#define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data")
422#define DEC_DEC_DEPTH --dec->depth 655#define DEC_DEC_DEPTH --dec->depth
423 656
424static UV 657static UV
425decode_uint (dec_t *dec) 658decode_uint (dec_t *dec)
426{ 659{
427 switch (*dec->cur & 31) 660 U8 m = *dec->cur & MINOR_MASK;
428 { 661 ++dec->cur;
429 case 0: case 1: case 2: case 3: case 4: case 5: case 6: case 7:
430 case 8: case 9: case 10: case 11: case 12: case 13: case 14: case 15:
431 case 16: case 17: case 18: case 19: case 20: case 21: case 22: case 23:
432 return *dec->cur++ & 31;
433 662
434 case 24: 663 if (ecb_expect_true (m < LENGTH_EXT1))
664 return m;
665 else if (ecb_expect_true (m == LENGTH_EXT1))
666 {
435 WANT (2); 667 WANT (1);
436 dec->cur += 2; 668 dec->cur += 1;
437 return dec->cur[-1]; 669 return dec->cur[-1];
438 670 }
439 case 25: 671 else if (ecb_expect_true (m == LENGTH_EXT2))
672 {
440 WANT (3); 673 WANT (2);
441 dec->cur += 3; 674 dec->cur += 2;
442 return (((UV)dec->cur[-2]) << 8) 675 return (((UV)dec->cur[-2]) << 8)
443 | ((UV)dec->cur[-1]); 676 | ((UV)dec->cur[-1]);
444 677 }
445 case 26: 678 else if (ecb_expect_true (m == LENGTH_EXT4))
679 {
446 WANT (5); 680 WANT (4);
447 dec->cur += 5; 681 dec->cur += 4;
448 return (((UV)dec->cur[-4]) << 24) 682 return (((UV)dec->cur[-4]) << 24)
449 | (((UV)dec->cur[-3]) << 16) 683 | (((UV)dec->cur[-3]) << 16)
450 | (((UV)dec->cur[-2]) << 8) 684 | (((UV)dec->cur[-2]) << 8)
451 | ((UV)dec->cur[-1]); 685 | ((UV)dec->cur[-1]);
452 686 }
453 case 27: 687 else if (ecb_expect_true (m == LENGTH_EXT8))
688 {
454 WANT (9); 689 WANT (8);
455 dec->cur += 9; 690 dec->cur += 8;
691
692 return
693#if UVSIZE < 8
694 0
695#else
456 return (((UV)dec->cur[-8]) << 56) 696 (((UV)dec->cur[-8]) << 56)
457 | (((UV)dec->cur[-7]) << 48) 697 | (((UV)dec->cur[-7]) << 48)
458 | (((UV)dec->cur[-6]) << 40) 698 | (((UV)dec->cur[-6]) << 40)
459 | (((UV)dec->cur[-5]) << 32) 699 | (((UV)dec->cur[-5]) << 32)
700#endif
460 | (((UV)dec->cur[-4]) << 24) 701 | (((UV)dec->cur[-4]) << 24)
461 | (((UV)dec->cur[-3]) << 16) 702 | (((UV)dec->cur[-3]) << 16)
462 | (((UV)dec->cur[-2]) << 8) 703 | (((UV)dec->cur[-2]) << 8)
463 | ((UV)dec->cur[-1]); 704 | ((UV)dec->cur[-1]);
464 705 }
465 default: 706 else
466 ERR ("corrupted CBOR data (unsupported integer minor encoding)"); 707 ERR ("corrupted CBOR data (unsupported integer minor encoding)");
467 }
468 708
469fail: 709fail:
470 return 0; 710 return 0;
471} 711}
472 712
477{ 717{
478 AV *av = newAV (); 718 AV *av = newAV ();
479 719
480 DEC_INC_DEPTH; 720 DEC_INC_DEPTH;
481 721
482 if ((*dec->cur & 31) == 31) 722 if (*dec->cur == (MAJOR_ARRAY | MINOR_INDEF))
483 { 723 {
484 ++dec->cur; 724 ++dec->cur;
485 725
486 for (;;) 726 for (;;)
487 { 727 {
488 WANT (1); 728 WANT (1);
489 729
490 if (*dec->cur == (0xe0 | 31)) 730 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
491 { 731 {
492 ++dec->cur; 732 ++dec->cur;
493 break; 733 break;
494 } 734 }
495 735
498 } 738 }
499 else 739 else
500 { 740 {
501 int i, len = decode_uint (dec); 741 int i, len = decode_uint (dec);
502 742
743 WANT (len); // complexity check for av_fill - need at least one byte per value, do not allow supersize arrays
503 av_fill (av, len - 1); 744 av_fill (av, len - 1);
504 745
505 for (i = 0; i < len; ++i) 746 for (i = 0; i < len; ++i)
506 AvARRAY (av)[i] = decode_sv (dec); 747 AvARRAY (av)[i] = decode_sv (dec);
507 } 748 }
513 SvREFCNT_dec (av); 754 SvREFCNT_dec (av);
514 DEC_DEC_DEPTH; 755 DEC_DEC_DEPTH;
515 return &PL_sv_undef; 756 return &PL_sv_undef;
516} 757}
517 758
759static void
760decode_he (dec_t *dec, HV *hv)
761{
762 // for speed reasons, we specialcase single-string
763 // byte or utf-8 strings as keys, but only when !stringref
764
765 if (ecb_expect_true (!dec->stringref))
766 if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8))
767 {
768 I32 len = decode_uint (dec);
769 char *key = (char *)dec->cur;
770
771 WANT (len);
772 dec->cur += len;
773
774 hv_store (hv, key, len, decode_sv (dec), 0);
775
776 return;
777 }
778 else if (ecb_expect_true ((U8)(*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8))
779 {
780 I32 len = decode_uint (dec);
781 char *key = (char *)dec->cur;
782
783 WANT (len);
784 dec->cur += len;
785
786 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
787 if (!is_utf8_string (key, len))
788 ERR ("corrupted CBOR data (invalid UTF-8 in map key)");
789
790 hv_store (hv, key, -len, decode_sv (dec), 0);
791
792 return;
793 }
794
795 SV *k = decode_sv (dec);
796 SV *v = decode_sv (dec);
797
798 hv_store_ent (hv, k, v, 0);
799 SvREFCNT_dec (k);
800
801fail:
802 ;
803}
804
518static SV * 805static SV *
519decode_hv (dec_t *dec) 806decode_hv (dec_t *dec)
520{ 807{
521 HV *hv = newHV (); 808 HV *hv = newHV ();
522 809
523 DEC_INC_DEPTH; 810 DEC_INC_DEPTH;
524 811
525 if ((*dec->cur & 31) == 31) 812 if (*dec->cur == (MAJOR_MAP | MINOR_INDEF))
526 { 813 {
527 ++dec->cur; 814 ++dec->cur;
528 815
529 for (;;) 816 for (;;)
530 { 817 {
531 WANT (1); 818 WANT (1);
532 819
533 if (*dec->cur == (0xe0 | 31)) 820 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
534 { 821 {
535 ++dec->cur; 822 ++dec->cur;
536 break; 823 break;
537 } 824 }
538 825
539 SV *k = decode_sv (dec); 826 decode_he (dec, hv);
540 SV *v = decode_sv (dec);
541
542 hv_store_ent (hv, k, v, 0);
543 } 827 }
544 } 828 }
545 else 829 else
546 { 830 {
547 int len = decode_uint (dec); 831 int pairs = decode_uint (dec);
548 832
549 while (len--) 833 while (pairs--)
550 { 834 decode_he (dec, hv);
551 SV *k = decode_sv (dec);
552 SV *v = decode_sv (dec);
553
554 hv_store_ent (hv, k, v, 0);
555 }
556 } 835 }
557 836
558 DEC_DEC_DEPTH; 837 DEC_DEC_DEPTH;
559 return newRV_noinc ((SV *)hv); 838 return newRV_noinc ((SV *)hv);
560 839
567static SV * 846static SV *
568decode_str (dec_t *dec, int utf8) 847decode_str (dec_t *dec, int utf8)
569{ 848{
570 SV *sv = 0; 849 SV *sv = 0;
571 850
572 if ((*dec->cur & 31) == 31) 851 if ((*dec->cur & MINOR_MASK) == MINOR_INDEF)
573 { 852 {
853 // indefinite length strings
574 ++dec->cur; 854 ++dec->cur;
575 855
856 U8 major = *dec->cur & MAJOR_MISC;
857
576 sv = newSVpvn ("", 0); 858 sv = newSVpvn ("", 0);
577 859
578 // not very fast, and certainly not robust against illegal input
579 for (;;) 860 for (;;)
580 { 861 {
581 WANT (1); 862 WANT (1);
582 863
583 if (*dec->cur == (0xe0 | 31)) 864 if ((*dec->cur - major) > LENGTH_EXT8)
865 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
584 { 866 {
585 ++dec->cur; 867 ++dec->cur;
586 break; 868 break;
587 } 869 }
870 else
871 ERR ("corrupted CBOR data (invalid chunks in indefinite length string)");
588 872
589 sv_catsv (sv, decode_sv (dec)); 873 STRLEN len = decode_uint (dec);
874
875 WANT (len);
876 sv_catpvn (sv, dec->cur, len);
877 dec->cur += len;
590 } 878 }
591 } 879 }
592 else 880 else
593 { 881 {
594 STRLEN len = decode_uint (dec); 882 STRLEN len = decode_uint (dec);
595 883
596 WANT (len); 884 WANT (len);
597 sv = newSVpvn (dec->cur, len); 885 sv = newSVpvn (dec->cur, len);
598 dec->cur += len; 886 dec->cur += len;
887
888 if (ecb_expect_false (dec->stringref)
889 && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1))
890 av_push (dec->stringref, SvREFCNT_inc_NN (sv));
599 } 891 }
600 892
601 if (utf8) 893 if (utf8)
894 {
895 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
896 if (!is_utf8_string (SvPVX (sv), SvCUR (sv)))
897 ERR ("corrupted CBOR data (invalid UTF-8 in text string)");
898
602 SvUTF8_on (sv); 899 SvUTF8_on (sv);
900 }
603 901
604 return sv; 902 return sv;
605 903
606fail: 904fail:
607 SvREFCNT_dec (sv); 905 SvREFCNT_dec (sv);
609} 907}
610 908
611static SV * 909static SV *
612decode_tagged (dec_t *dec) 910decode_tagged (dec_t *dec)
613{ 911{
912 SV *sv = 0;
614 UV tag = decode_uint (dec); 913 UV tag = decode_uint (dec);
914
915 WANT (1);
916
917 switch (tag)
918 {
919 case CBOR_TAG_MAGIC:
615 SV *sv = decode_sv (dec); 920 sv = decode_sv (dec);
921 break;
616 922
617 if (tag == 55799) // 2.4.5 Self-Describe CBOR 923 case CBOR_TAG_INDIRECTION:
924 sv = newRV_noinc (decode_sv (dec));
925 break;
926
927 case CBOR_TAG_STRINGREF_NAMESPACE:
928 {
929 ENTER; SAVETMPS;
930
931 SAVESPTR (dec->stringref);
932 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ());
933
934 sv = decode_sv (dec);
935
936 FREETMPS; LEAVE;
937 }
938 break;
939
940 case CBOR_TAG_STRINGREF:
941 {
942 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
943 ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
944
945 UV idx = decode_uint (dec);
946
947 if (!dec->stringref || (int)idx > AvFILLp (dec->stringref))
948 ERR ("corrupted CBOR data (stringref index out of bounds or outside namespace)");
949
950 sv = newSVsv (AvARRAY (dec->stringref)[idx]);
951 }
952 break;
953
954 case CBOR_TAG_VALUE_SHAREABLE:
955 {
956 if (ecb_expect_false (!dec->shareable))
957 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ());
958
959 if (dec->cbor.flags & F_ALLOW_CYCLES)
960 {
961 sv = newSV (0);
962 av_push (dec->shareable, SvREFCNT_inc_NN (sv));
963
964 SV *osv = decode_sv (dec);
965 sv_setsv (sv, osv);
966 SvREFCNT_dec_NN (osv);
967 }
968 else
969 {
970 av_push (dec->shareable, &PL_sv_undef);
971 int idx = AvFILLp (dec->shareable);
972 sv = decode_sv (dec);
973 av_store (dec->shareable, idx, SvREFCNT_inc_NN (sv));
974 }
975 }
976 break;
977
978 case CBOR_TAG_VALUE_SHAREDREF:
979 {
980 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
981 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
982
983 UV idx = decode_uint (dec);
984
985 if (!dec->shareable || (int)idx > AvFILLp (dec->shareable))
986 ERR ("corrupted CBOR data (sharedref index out of bounds)");
987
988 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]);
989
990 if (sv == &PL_sv_undef)
991 ERR ("cyclic CBOR data structure found, but allow_cycles is not enabled");
992 }
993 break;
994
995 case CBOR_TAG_PERL_OBJECT:
996 {
997 sv = decode_sv (dec);
998
999 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
1000 ERR ("corrupted CBOR data (non-array perl object)");
1001
1002 AV *av = (AV *)SvRV (sv);
1003 int len = av_len (av) + 1;
1004 HV *stash = gv_stashsv (*av_fetch (av, 0, 1), 0);
1005
1006 if (!stash)
1007 ERR ("cannot decode perl-object (package does not exist)");
1008
1009 GV *method = gv_fetchmethod_autoload (stash, "THAW", 0);
1010
1011 if (!method)
1012 ERR ("cannot decode perl-object (package does not have a THAW method)");
1013
1014 dSP;
1015
1016 ENTER; SAVETMPS;
1017 PUSHMARK (SP);
1018 EXTEND (SP, len + 1);
1019 // we re-bless the reference to get overload and other niceties right
1020 PUSHs (*av_fetch (av, 0, 1));
1021 PUSHs (sv_cbor);
1022
1023 int i;
1024
1025 for (i = 1; i < len; ++i)
1026 PUSHs (*av_fetch (av, i, 1));
1027
1028 PUTBACK;
1029 call_sv ((SV *)GvCV (method), G_SCALAR | G_EVAL);
1030 SPAGAIN;
1031
1032 if (SvTRUE (ERRSV))
1033 {
1034 FREETMPS; LEAVE;
1035 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV))));
1036 }
1037
1038 SvREFCNT_dec (sv);
1039 sv = SvREFCNT_inc (POPs);
1040
1041 PUTBACK;
1042
1043 FREETMPS; LEAVE;
1044 }
1045 break;
1046
1047 default:
1048 {
1049 sv = decode_sv (dec);
1050
1051 dSP;
1052 ENTER; SAVETMPS;
1053 SAVESTACK_POS ();
1054 PUSHMARK (SP);
1055 EXTEND (SP, 2);
1056 PUSHs (newSVuv (tag));
1057 PUSHs (sv);
1058
1059 PUTBACK;
1060 int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL);
1061 SPAGAIN;
1062
1063 if (SvTRUE (ERRSV))
1064 {
1065 FREETMPS; LEAVE;
1066 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV))));
1067 }
1068
1069 if (count)
1070 {
1071 SvREFCNT_dec (sv);
1072 sv = SvREFCNT_inc (POPs);
1073 }
1074 else
1075 {
1076 AV *av = newAV ();
1077 av_push (av, newSVuv (tag));
1078 av_push (av, sv);
1079
1080 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
1081 ? cbor_tagged_stash
1082 : gv_stashpv ("CBOR::XS::Tagged" , 1);
1083 sv = sv_bless (newRV_noinc ((SV *)av), tagged_stash);
1084 }
1085
1086 PUTBACK;
1087
1088 FREETMPS; LEAVE;
1089 }
1090 break;
1091 }
1092
618 return sv; 1093 return sv;
619 1094
620 AV *av = newAV (); 1095fail:
621 av_push (av, newSVuv (tag)); 1096 SvREFCNT_dec (sv);
622 av_push (av, sv); 1097 return &PL_sv_undef;
623
624 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
625 ? cbor_tagged_stash
626 : gv_stashpv ("CBOR::XS::Tagged" , 1);
627
628 return sv_bless (newRV_noinc ((SV *)av), tagged_stash);
629} 1098}
630 1099
631static SV * 1100static SV *
632decode_sv (dec_t *dec) 1101decode_sv (dec_t *dec)
633{ 1102{
634 WANT (1); 1103 WANT (1);
635 1104
636 switch (*dec->cur >> 5) 1105 switch (*dec->cur >> MAJOR_SHIFT)
637 { 1106 {
638 case 0: // unsigned int 1107 case MAJOR_POS_INT >> MAJOR_SHIFT: return newSVuv (decode_uint (dec));
639 return newSVuv (decode_uint (dec)); 1108 case MAJOR_NEG_INT >> MAJOR_SHIFT: return newSViv (-1 - (IV)decode_uint (dec));
640 case 1: // negative int 1109 case MAJOR_BYTES >> MAJOR_SHIFT: return decode_str (dec, 0);
641 return newSViv (-1 - (IV)decode_uint (dec)); 1110 case MAJOR_TEXT >> MAJOR_SHIFT: return decode_str (dec, 1);
642 case 2: // octet string 1111 case MAJOR_ARRAY >> MAJOR_SHIFT: return decode_av (dec);
643 return decode_str (dec, 0); 1112 case MAJOR_MAP >> MAJOR_SHIFT: return decode_hv (dec);
644 case 3: // utf-8 string 1113 case MAJOR_TAG >> MAJOR_SHIFT: return decode_tagged (dec);
645 return decode_str (dec, 1); 1114
646 case 4: // array 1115 case MAJOR_MISC >> MAJOR_SHIFT:
647 return decode_av (dec);
648 case 5: // map
649 return decode_hv (dec);
650 case 6: // tag
651 return decode_tagged (dec);
652 case 7: // misc
653 switch (*dec->cur++ & 31) 1116 switch (*dec->cur++ & MINOR_MASK)
654 { 1117 {
655 case 20: 1118 case SIMPLE_FALSE:
656#if CBOR_SLOW 1119#if CBOR_SLOW
657 cbor_false = get_bool ("CBOR::XS::false"); 1120 types_false = get_bool ("Types::Serialiser::false");
658#endif 1121#endif
659 return newSVsv (cbor_false); 1122 return newSVsv (types_false);
660 case 21: 1123 case SIMPLE_TRUE:
661#if CBOR_SLOW 1124#if CBOR_SLOW
662 cbor_true = get_bool ("CBOR::XS::true"); 1125 types_true = get_bool ("Types::Serialiser::true");
663#endif 1126#endif
664 return newSVsv (cbor_true); 1127 return newSVsv (types_true);
665 case 22: 1128 case SIMPLE_NULL:
666 return newSVsv (&PL_sv_undef); 1129 return newSVsv (&PL_sv_undef);
1130 case SIMPLE_UNDEF:
1131#if CBOR_SLOW
1132 types_error = get_bool ("Types::Serialiser::error");
1133#endif
1134 return newSVsv (types_error);
667 1135
668 case 25: 1136 case MISC_FLOAT16:
669 { 1137 {
670 WANT (2); 1138 WANT (2);
671 1139
672 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1]; 1140 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1];
673 dec->cur += 2; 1141 dec->cur += 2;
674 1142
675 return newSVnv (ecb_binary16_to_float (fp)); 1143 return newSVnv (ecb_binary16_to_float (fp));
676 } 1144 }
677 1145
678 case 26: 1146 case MISC_FLOAT32:
679 { 1147 {
680 uint32_t fp; 1148 uint32_t fp;
681 WANT (4); 1149 WANT (4);
682 memcpy (&fp, dec->cur, 4); 1150 memcpy (&fp, dec->cur, 4);
683 dec->cur += 4; 1151 dec->cur += 4;
686 fp = ecb_bswap32 (fp); 1154 fp = ecb_bswap32 (fp);
687 1155
688 return newSVnv (ecb_binary32_to_float (fp)); 1156 return newSVnv (ecb_binary32_to_float (fp));
689 } 1157 }
690 1158
691 case 27: 1159 case MISC_FLOAT64:
692 { 1160 {
693 uint64_t fp; 1161 uint64_t fp;
694 WANT (8); 1162 WANT (8);
695 memcpy (&fp, dec->cur, 8); 1163 memcpy (&fp, dec->cur, 8);
696 dec->cur += 8; 1164 dec->cur += 8;
699 fp = ecb_bswap64 (fp); 1167 fp = ecb_bswap64 (fp);
700 1168
701 return newSVnv (ecb_binary64_to_double (fp)); 1169 return newSVnv (ecb_binary64_to_double (fp));
702 } 1170 }
703 1171
704 // 0..19 unassigned 1172 // 0..19 unassigned simple
705 // 24 reserved + unassigned (reserved values are not encodable) 1173 // 24 reserved + unassigned simple (reserved values are not encodable)
1174 // 28-30 unassigned misc
1175 // 31 break code
706 default: 1176 default:
707 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)"); 1177 ERR ("corrupted CBOR data (reserved/unassigned/unexpected major 7 value)");
708 } 1178 }
709 1179
710 break; 1180 break;
711 } 1181 }
712 1182
715} 1185}
716 1186
717static SV * 1187static SV *
718decode_cbor (SV *string, CBOR *cbor, char **offset_return) 1188decode_cbor (SV *string, CBOR *cbor, char **offset_return)
719{ 1189{
720 dec_t dec; 1190 dec_t dec = { 0 };
721 SV *sv; 1191 SV *sv;
1192 STRLEN len;
1193 char *data = SvPVbyte (string, len);
722 1194
723 /* work around bugs in 5.10 where manipulating magic values
724 * makes perl ignore the magic in subsequent accesses.
725 * also make a copy of non-PV values, to get them into a clean
726 * state (SvPV should do that, but it's buggy, see below).
727 */
728 /*SvGETMAGIC (string);*/
729 if (SvMAGICAL (string) || !SvPOK (string))
730 string = sv_2mortal (newSVsv (string));
731
732 SvUPGRADE (string, SVt_PV);
733
734 /* work around a bug in perl 5.10, which causes SvCUR to fail an
735 * assertion with -DDEBUGGING, although SvCUR is documented to
736 * return the xpv_cur field which certainly exists after upgrading.
737 * according to nicholas clark, calling SvPOK fixes this.
738 * But it doesn't fix it, so try another workaround, call SvPV_nolen
739 * and hope for the best.
740 * Damnit, SvPV_nolen still trips over yet another assertion. This
741 * assertion business is seriously broken, try yet another workaround
742 * for the broken -DDEBUGGING.
743 */
744 {
745#ifdef DEBUGGING
746 STRLEN offset = SvOK (string) ? sv_len (string) : 0;
747#else
748 STRLEN offset = SvCUR (string);
749#endif
750
751 if (offset > cbor->max_size && cbor->max_size) 1195 if (len > cbor->max_size && cbor->max_size)
752 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu", 1196 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
753 (unsigned long)SvCUR (string), (unsigned long)cbor->max_size); 1197 (unsigned long)len, (unsigned long)cbor->max_size);
754 }
755
756 sv_utf8_downgrade (string, 0);
757 1198
758 dec.cbor = *cbor; 1199 dec.cbor = *cbor;
759 dec.cur = (U8 *)SvPVX (string); 1200 dec.cur = (U8 *)data;
760 dec.end = (U8 *)SvEND (string); 1201 dec.end = (U8 *)data + len;
761 dec.err = 0;
762 dec.depth = 0;
763 1202
764 sv = decode_sv (&dec); 1203 sv = decode_sv (&dec);
765 1204
766 if (offset_return) 1205 if (offset_return)
767 *offset_return = dec.cur; 1206 *offset_return = dec.cur;
770 if (dec.cur != dec.end && !dec.err) 1209 if (dec.cur != dec.end && !dec.err)
771 dec.err = "garbage after CBOR object"; 1210 dec.err = "garbage after CBOR object";
772 1211
773 if (dec.err) 1212 if (dec.err)
774 { 1213 {
1214 if (dec.shareable)
1215 {
1216 // need to break cyclic links, which whould all be in shareable
1217 int i;
1218 SV **svp;
1219
1220 for (i = av_len (dec.shareable) + 1; i--; )
1221 if ((svp = av_fetch (dec.shareable, i, 0)))
1222 sv_setsv (*svp, &PL_sv_undef);
1223 }
1224
775 SvREFCNT_dec (sv); 1225 SvREFCNT_dec (sv);
776 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)SvPVX (string), (int)(uint8_t)*dec.cur); 1226 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur);
777 } 1227 }
778 1228
779 sv = sv_2mortal (sv); 1229 sv = sv_2mortal (sv);
780 1230
781 return sv; 1231 return sv;
782} 1232}
783 1233
1234/////////////////////////////////////////////////////////////////////////////
1235// incremental parser
1236
1237#define INCR_DONE(cbor) (AvFILLp (cbor->incr_count) < 0)
1238
1239// returns 0 for notyet, 1 for success or error
1240static int
1241incr_parse (CBOR *self, SV *cborstr)
1242{
1243 STRLEN cur;
1244 SvPV (cborstr, cur);
1245
1246 while (ecb_expect_true (self->incr_need <= cur))
1247 {
1248 // table of integer count bytes
1249 static I8 incr_len[MINOR_MASK + 1] = {
1250 0, 0, 0, 0, 0, 0, 0, 0,
1251 0, 0, 0, 0, 0, 0, 0, 0,
1252 0, 0, 0, 0, 0, 0, 0, 0,
1253 1, 2, 4, 8,-1,-1,-1,-2
1254 };
1255
1256 const U8 *p = SvPVX (cborstr) + self->incr_pos;
1257 U8 m = *p & MINOR_MASK;
1258 IV count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1259 I8 ilen = incr_len[m];
1260
1261 self->incr_need = self->incr_pos + 1;
1262
1263 if (ecb_expect_false (ilen < 0))
1264 {
1265 if (m != MINOR_INDEF)
1266 return 1; // error
1267
1268 if (*p == (MAJOR_MISC | MINOR_INDEF))
1269 {
1270 if (count >= 0)
1271 return 1; // error
1272
1273 count = 1;
1274 }
1275 else
1276 {
1277 av_push (self->incr_count, newSViv (-1)); //TODO: nest
1278 count = -1;
1279 }
1280 }
1281 else
1282 {
1283 self->incr_need += ilen;
1284 if (ecb_expect_false (self->incr_need > cur))
1285 return 0;
1286
1287 int major = *p >> MAJOR_SHIFT;
1288
1289 switch (major)
1290 {
1291 case MAJOR_TAG >> MAJOR_SHIFT:
1292 ++count; // tags merely prefix another value
1293 break;
1294
1295 case MAJOR_BYTES >> MAJOR_SHIFT:
1296 case MAJOR_TEXT >> MAJOR_SHIFT:
1297 case MAJOR_ARRAY >> MAJOR_SHIFT:
1298 case MAJOR_MAP >> MAJOR_SHIFT:
1299 {
1300 UV len;
1301
1302 if (ecb_expect_false (ilen))
1303 {
1304 len = 0;
1305
1306 do {
1307 len = (len << 8) | *++p;
1308 } while (--ilen);
1309 }
1310 else
1311 len = m;
1312
1313 switch (major)
1314 {
1315 case MAJOR_BYTES >> MAJOR_SHIFT:
1316 case MAJOR_TEXT >> MAJOR_SHIFT:
1317 self->incr_need += len;
1318 if (ecb_expect_false (self->incr_need > cur))
1319 return 0;
1320
1321 break;
1322
1323 case MAJOR_MAP >> MAJOR_SHIFT:
1324 len <<= 1;
1325 case MAJOR_ARRAY >> MAJOR_SHIFT:
1326 if (len)
1327 {
1328 av_push (self->incr_count, newSViv (len + 1)); //TODO: nest
1329 count = len + 1;
1330 }
1331 break;
1332 }
1333 }
1334 }
1335 }
1336
1337 self->incr_pos = self->incr_need;
1338
1339 if (count > 0)
1340 {
1341 while (!--count)
1342 {
1343 if (!AvFILLp (self->incr_count))
1344 return 1; // done
1345
1346 SvREFCNT_dec_NN (av_pop (self->incr_count));
1347 count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1348 }
1349
1350 SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]) = count;
1351 }
1352 }
1353
1354 return 0;
1355}
1356
1357
784///////////////////////////////////////////////////////////////////////////// 1358/////////////////////////////////////////////////////////////////////////////
785// XS interface functions 1359// XS interface functions
786 1360
787MODULE = CBOR::XS PACKAGE = CBOR::XS 1361MODULE = CBOR::XS PACKAGE = CBOR::XS
788 1362
789BOOT: 1363BOOT:
790{ 1364{
791 cbor_stash = gv_stashpv ("CBOR::XS" , 1); 1365 cbor_stash = gv_stashpv ("CBOR::XS" , 1);
792 cbor_boolean_stash = gv_stashpv ("CBOR::XS::Boolean", 1);
793 cbor_tagged_stash = gv_stashpv ("CBOR::XS::Tagged" , 1); 1366 cbor_tagged_stash = gv_stashpv ("CBOR::XS::Tagged" , 1);
794 1367
795 cbor_true = get_bool ("CBOR::XS::true"); 1368 types_boolean_stash = gv_stashpv ("Types::Serialiser::Boolean", 1);
796 cbor_false = get_bool ("CBOR::XS::false"); 1369 types_error_stash = gv_stashpv ("Types::Serialiser::Error" , 1);
1370
1371 types_true = get_bool ("Types::Serialiser::true" );
1372 types_false = get_bool ("Types::Serialiser::false");
1373 types_error = get_bool ("Types::Serialiser::error");
1374
1375 default_filter = newSVpv ("CBOR::XS::default_filter", 0);
1376
1377 sv_cbor = newSVpv ("CBOR", 0);
1378 SvREADONLY_on (sv_cbor);
797} 1379}
798 1380
799PROTOTYPES: DISABLE 1381PROTOTYPES: DISABLE
800 1382
801void CLONE (...) 1383void CLONE (...)
802 CODE: 1384 CODE:
803 cbor_stash = 0; 1385 cbor_stash = 0;
804 cbor_boolean_stash = 0;
805 cbor_tagged_stash = 0; 1386 cbor_tagged_stash = 0;
1387 types_error_stash = 0;
1388 types_boolean_stash = 0;
806 1389
807void new (char *klass) 1390void new (char *klass)
808 PPCODE: 1391 PPCODE:
809{ 1392{
810 SV *pv = NEWSV (0, sizeof (CBOR)); 1393 SV *pv = NEWSV (0, sizeof (CBOR));
818 1401
819void shrink (CBOR *self, int enable = 1) 1402void shrink (CBOR *self, int enable = 1)
820 ALIAS: 1403 ALIAS:
821 shrink = F_SHRINK 1404 shrink = F_SHRINK
822 allow_unknown = F_ALLOW_UNKNOWN 1405 allow_unknown = F_ALLOW_UNKNOWN
1406 allow_sharing = F_ALLOW_SHARING
1407 allow_cycles = F_ALLOW_CYCLES
1408 pack_strings = F_PACK_STRINGS
1409 utf8_strings = F_UTF8_STRINGS
1410 validate_utf8 = F_VALIDATE_UTF8
823 PPCODE: 1411 PPCODE:
824{ 1412{
825 if (enable) 1413 if (enable)
826 self->flags |= ix; 1414 self->flags |= ix;
827 else 1415 else
832 1420
833void get_shrink (CBOR *self) 1421void get_shrink (CBOR *self)
834 ALIAS: 1422 ALIAS:
835 get_shrink = F_SHRINK 1423 get_shrink = F_SHRINK
836 get_allow_unknown = F_ALLOW_UNKNOWN 1424 get_allow_unknown = F_ALLOW_UNKNOWN
1425 get_allow_sharing = F_ALLOW_SHARING
1426 get_allow_cycles = F_ALLOW_CYCLES
1427 get_pack_strings = F_PACK_STRINGS
1428 get_validate_utf8 = F_VALIDATE_UTF8
837 PPCODE: 1429 PPCODE:
838 XPUSHs (boolSV (self->flags & ix)); 1430 XPUSHs (boolSV (self->flags & ix));
839 1431
840void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) 1432void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
841 PPCODE: 1433 PPCODE:
857 CODE: 1449 CODE:
858 RETVAL = self->max_size; 1450 RETVAL = self->max_size;
859 OUTPUT: 1451 OUTPUT:
860 RETVAL 1452 RETVAL
861 1453
862#if 0 //TODO 1454void filter (CBOR *self, SV *filter = 0)
863
864void filter_cbor_object (CBOR *self, SV *cb = &PL_sv_undef)
865 PPCODE: 1455 PPCODE:
866{
867 SvREFCNT_dec (self->cb_object); 1456 SvREFCNT_dec (self->filter);
868 self->cb_object = SvOK (cb) ? newSVsv (cb) : 0; 1457 self->filter = filter ? newSVsv (filter) : filter;
869
870 XPUSHs (ST (0)); 1458 XPUSHs (ST (0));
871}
872 1459
873void filter_cbor_single_key_object (CBOR *self, SV *key, SV *cb = &PL_sv_undef) 1460SV *get_filter (CBOR *self)
874 PPCODE: 1461 CODE:
875{ 1462 RETVAL = self->filter ? self->filter : NEWSV (0, 0);
876 if (!self->cb_sk_object) 1463 OUTPUT:
877 self->cb_sk_object = newHV (); 1464 RETVAL
878
879 if (SvOK (cb))
880 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);
881 else
882 {
883 hv_delete_ent (self->cb_sk_object, key, G_DISCARD, 0);
884
885 if (!HvKEYS (self->cb_sk_object))
886 {
887 SvREFCNT_dec (self->cb_sk_object);
888 self->cb_sk_object = 0;
889 }
890 }
891
892 XPUSHs (ST (0));
893}
894
895#endif
896 1465
897void encode (CBOR *self, SV *scalar) 1466void encode (CBOR *self, SV *scalar)
898 PPCODE: 1467 PPCODE:
899 PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN; 1468 PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN;
900 XPUSHs (scalar); 1469 XPUSHs (scalar);
913 EXTEND (SP, 2); 1482 EXTEND (SP, 2);
914 PUSHs (sv); 1483 PUSHs (sv);
915 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); 1484 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
916} 1485}
917 1486
918#if 0 1487void incr_parse (CBOR *self, SV *cborstr)
1488 ALIAS:
1489 incr_parse_multiple = 1
1490 PPCODE:
1491{
1492 if (SvUTF8 (cborstr))
1493 sv_utf8_downgrade (cborstr, 0);
1494
1495 if (!self->incr_count)
1496 {
1497 self->incr_count = newAV ();
1498 self->incr_pos = 0;
1499 self->incr_need = 1;
1500
1501 av_push (self->incr_count, newSViv (1));
1502 }
1503
1504 do
1505 {
1506 if (!incr_parse (self, cborstr))
1507 {
1508 if (self->incr_need > self->max_size && self->max_size)
1509 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
1510 (unsigned long)self->incr_need, (unsigned long)self->max_size);
1511
1512 break;
1513 }
1514
1515 SV *sv;
1516 char *offset;
1517
1518 PUTBACK; sv = decode_cbor (cborstr, self, &offset); SPAGAIN;
1519 XPUSHs (sv);
1520
1521 sv_chop (cborstr, offset);
1522
1523 av_clear (self->incr_count);
1524 av_push (self->incr_count, newSViv (1));
1525
1526 self->incr_pos = 0;
1527 self->incr_need = self->incr_pos + 1;
1528 }
1529 while (ix);
1530}
1531
1532void incr_reset (CBOR *self)
1533 CODE:
1534{
1535 SvREFCNT_dec (self->incr_count);
1536 self->incr_count = 0;
1537}
919 1538
920void DESTROY (CBOR *self) 1539void DESTROY (CBOR *self)
921 CODE: 1540 PPCODE:
922 SvREFCNT_dec (self->cb_sk_object); 1541 cbor_free (self);
923 SvREFCNT_dec (self->cb_object);
924
925#endif
926 1542
927PROTOTYPES: ENABLE 1543PROTOTYPES: ENABLE
928 1544
929void encode_cbor (SV *scalar) 1545void encode_cbor (SV *scalar)
1546 ALIAS:
1547 encode_cbor = 0
1548 encode_cbor_sharing = F_ALLOW_SHARING
930 PPCODE: 1549 PPCODE:
931{ 1550{
932 CBOR cbor; 1551 CBOR cbor;
933 cbor_init (&cbor); 1552 cbor_init (&cbor);
1553 cbor.flags |= ix;
934 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN; 1554 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;
935 XPUSHs (scalar); 1555 XPUSHs (scalar);
936} 1556}
937 1557
938void decode_cbor (SV *cborstr) 1558void decode_cbor (SV *cborstr)

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines