ViewVC Help
View File | Revision Log | Show Annotations | Download File
/cvs/CBOR-XS/XS.xs
(Generate patch)

Comparing CBOR-XS/XS.xs (file contents):
Revision 1.8 by root, Sun Oct 27 10:12:01 2013 UTC vs.
Revision 1.51 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
14// known tags, rfc7049 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
15enum cbor_tag 67enum cbor_tag
16{ 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
17 CBOR_TAG_DATETIME = 0, // rfc4287, utf-8 79 CBOR_TAG_DATETIME = 0, // rfc4287, utf-8
18 CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any 80 CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any
19 CBOR_TAG_POS_BIGNUM = 2, // byte string 81 CBOR_TAG_POS_BIGNUM = 2, // byte string
20 CBOR_TAG_NEG_BIGNUM = 3, // byte string 82 CBOR_TAG_NEG_BIGNUM = 3, // byte string
21 CBOR_TAG_DECIMAL = 4, // decimal fraction, array 83 CBOR_TAG_DECIMAL = 4, // decimal fraction, array
22 CBOR_TAG_BIGFLOAT = 5, // array 84 CBOR_TAG_BIGFLOAT = 5, // array
23 85
24 CBOR_TAG_CONV_B64U = 21, // base64url, any 86 CBOR_TAG_CONV_B64U = 21, // base64url, any
25 CBOR_TAG_CONV_B64 = 22, // base64, any 87 CBOR_TAG_CONV_B64 = 22, // base64, any
26 CBOR_TAG_CONV_HEX = 23, // base16, any 88 CBOR_TAG_CONV_HEX = 23, // base16, any
27 CBOR_TAG_CBOR = 24, // embedded cbor, byte string 89 CBOR_TAG_CBOR = 24, // embedded cbor, byte string
28 90
29 CBOR_TAG_URI = 32, // URI rfc3986, utf-8 91 CBOR_TAG_URI = 32, // URI rfc3986, utf-8
30 CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8 92 CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8
31 CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8 93 CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8
32 CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8 94 CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8
33 CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8 95 CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8
34 96
35 CBOR_TAG_MAGIC = 55799 97 CBOR_TAG_MAGIC = 55799, // self-describe cbor
36}; 98};
37 99
38#define F_SHRINK 0x00000200UL 100#define F_SHRINK 0x00000001UL
39#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
40 107
41#define INIT_SIZE 32 // initial scalar size to be allocated 108#define INIT_SIZE 32 // initial scalar size to be allocated
42 109
43#define SB do { 110#define SB do {
44#define SE } while (0) 111#define SE } while (0)
55#else 122#else
56# define CBOR_SLOW 0 123# define CBOR_SLOW 0
57# define CBOR_STASH cbor_stash 124# define CBOR_STASH cbor_stash
58#endif 125#endif
59 126
60static 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::
61static SV *cbor_true, *cbor_false; 128static SV *types_true, *types_false, *types_error, *sv_cbor, *default_filter;
62 129
63typedef struct { 130typedef struct {
64 U32 flags; 131 U32 flags;
65 U32 max_depth; 132 U32 max_depth;
66 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.
67} CBOR; 140} CBOR;
68 141
69ecb_inline void 142ecb_inline void
70cbor_init (CBOR *cbor) 143cbor_init (CBOR *cbor)
71{ 144{
72 Zero (cbor, 1, CBOR); 145 Zero (cbor, 1, CBOR);
73 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);
74} 154}
75 155
76///////////////////////////////////////////////////////////////////////////// 156/////////////////////////////////////////////////////////////////////////////
77// utility functions 157// utility functions
78 158
100 SvPV_renew (sv, SvCUR (sv) + 1); 180 SvPV_renew (sv, SvCUR (sv) + 1);
101#endif 181#endif
102 } 182 }
103} 183}
104 184
105///////////////////////////////////////////////////////////////////////////// 185// minimum length of a string to be registered for stringref
106// fp hell 186ecb_inline int
107 187minimum_string_length (UV idx)
108//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}
109 199
110///////////////////////////////////////////////////////////////////////////// 200/////////////////////////////////////////////////////////////////////////////
111// encoder 201// encoder
112 202
113// structure used for encoding CBOR 203// structure used for encoding CBOR
116 char *cur; // SvPVX (sv) + current output position 206 char *cur; // SvPVX (sv) + current output position
117 char *end; // SvEND (sv) 207 char *end; // SvEND (sv)
118 SV *sv; // result scalar 208 SV *sv; // result scalar
119 CBOR cbor; 209 CBOR cbor;
120 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;
121} enc_t; 215} enc_t;
122 216
123ecb_inline void 217ecb_inline void
124need (enc_t *enc, STRLEN len) 218need (enc_t *enc, STRLEN len)
125{ 219{
142static void 236static void
143encode_uint (enc_t *enc, int major, UV len) 237encode_uint (enc_t *enc, int major, UV len)
144{ 238{
145 need (enc, 9); 239 need (enc, 9);
146 240
147 if (len < 24) 241 if (ecb_expect_true (len < LENGTH_EXT1))
148 *enc->cur++ = major | len; 242 *enc->cur++ = major | len;
149 else if (len <= 0xff) 243 else if (ecb_expect_true (len <= 0xffU))
150 { 244 {
151 *enc->cur++ = major | 24; 245 *enc->cur++ = major | LENGTH_EXT1;
152 *enc->cur++ = len; 246 *enc->cur++ = len;
153 } 247 }
154 else if (len <= 0xffff) 248 else if (len <= 0xffffU)
155 { 249 {
156 *enc->cur++ = major | 25; 250 *enc->cur++ = major | LENGTH_EXT2;
157 *enc->cur++ = len >> 8; 251 *enc->cur++ = len >> 8;
158 *enc->cur++ = len; 252 *enc->cur++ = len;
159 } 253 }
160 else if (len <= 0xffffffff) 254 else if (len <= 0xffffffffU)
161 { 255 {
162 *enc->cur++ = major | 26; 256 *enc->cur++ = major | LENGTH_EXT4;
163 *enc->cur++ = len >> 24; 257 *enc->cur++ = len >> 24;
164 *enc->cur++ = len >> 16; 258 *enc->cur++ = len >> 16;
165 *enc->cur++ = len >> 8; 259 *enc->cur++ = len >> 8;
166 *enc->cur++ = len; 260 *enc->cur++ = len;
167 } 261 }
168 else 262 else
169 { 263 {
170 *enc->cur++ = major | 27; 264 *enc->cur++ = major | LENGTH_EXT8;
171 *enc->cur++ = len >> 56; 265 *enc->cur++ = len >> 56;
172 *enc->cur++ = len >> 48; 266 *enc->cur++ = len >> 48;
173 *enc->cur++ = len >> 40; 267 *enc->cur++ = len >> 40;
174 *enc->cur++ = len >> 32; 268 *enc->cur++ = len >> 32;
175 *enc->cur++ = len >> 24; 269 *enc->cur++ = len >> 24;
177 *enc->cur++ = len >> 8; 271 *enc->cur++ = len >> 8;
178 *enc->cur++ = len; 272 *enc->cur++ = len;
179 } 273 }
180} 274}
181 275
182static void 276ecb_inline void
277encode_tag (enc_t *enc, UV tag)
278{
279 encode_uint (enc, MAJOR_TAG, tag);
280}
281
282ecb_inline void
183encode_str (enc_t *enc, int utf8, char *str, STRLEN len) 283encode_str (enc_t *enc, int utf8, char *str, STRLEN len)
184{ 284{
185 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);
186 need (enc, len); 299 need (enc, len);
187 memcpy (enc->cur, str, len); 300 memcpy (enc->cur, str, len);
188 enc->cur += len; 301 enc->cur += len;
189} 302}
190 303
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
191static void encode_sv (enc_t *enc, SV *sv); 329static void encode_sv (enc_t *enc, SV *sv);
192 330
193static void 331static void
194encode_av (enc_t *enc, AV *av) 332encode_av (enc_t *enc, AV *av)
195{ 333{
198 if (enc->depth >= enc->cbor.max_depth) 336 if (enc->depth >= enc->cbor.max_depth)
199 croak (ERR_NESTING_EXCEEDED); 337 croak (ERR_NESTING_EXCEEDED);
200 338
201 ++enc->depth; 339 ++enc->depth;
202 340
203 encode_uint (enc, 0x80, len + 1); 341 encode_uint (enc, MAJOR_ARRAY, len + 1);
204 342
343 if (SvMAGICAL (av))
205 for (i = 0; i <= len; ++i) 344 for (i = 0; i <= len; ++i)
206 { 345 {
207 SV **svp = av_fetch (av, i, 0); 346 SV **svp = av_fetch (av, i, 0);
208 encode_sv (enc, svp ? *svp : &PL_sv_undef); 347 encode_sv (enc, svp ? *svp : &PL_sv_undef);
209 } 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 }
210 355
211 --enc->depth; 356 --enc->depth;
212} 357}
213 358
214static void 359static void
223 368
224 int pairs = hv_iterinit (hv); 369 int pairs = hv_iterinit (hv);
225 int mg = SvMAGICAL (hv); 370 int mg = SvMAGICAL (hv);
226 371
227 if (mg) 372 if (mg)
228 encode_ch (enc, 0xa0 | 31); 373 encode_ch (enc, MAJOR_MAP | MINOR_INDEF);
229 else 374 else
230 encode_uint (enc, 0xa0, pairs); 375 encode_uint (enc, MAJOR_MAP, pairs);
231 376
232 while ((he = hv_iternext (hv))) 377 while ((he = hv_iternext (hv)))
233 { 378 {
234 if (HeKLEN (he) == HEf_SVKEY) 379 if (HeKLEN (he) == HEf_SVKEY)
235 encode_sv (enc, HeSVKEY (he)); 380 encode_sv (enc, HeSVKEY (he));
236 else 381 else
237 encode_str (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he)); 382 encode_strref (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he));
238 383
239 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));
240 } 385 }
241 386
242 if (mg) 387 if (mg)
243 encode_ch (enc, 0xe0 | 31); 388 encode_ch (enc, MAJOR_MISC | MINOR_INDEF);
244 389
245 --enc->depth; 390 --enc->depth;
246} 391}
247 392
248// encode objects, arrays and special \0=false and \1=true values. 393// encode objects, arrays and special \0=false and \1=true values.
249static void 394static void
250encode_rv (enc_t *enc, SV *sv) 395encode_rv (enc_t *enc, SV *sv)
251{ 396{
252 svtype svt;
253
254 SvGETMAGIC (sv); 397 SvGETMAGIC (sv);
398
255 svt = SvTYPE (sv); 399 svtype svt = SvTYPE (sv);
256 400
257 if (ecb_expect_false (SvOBJECT (sv))) 401 if (ecb_expect_false (SvOBJECT (sv)))
258 { 402 {
259 HV *boolean_stash = !CBOR_SLOW || cbor_boolean_stash 403 HV *boolean_stash = !CBOR_SLOW || types_boolean_stash
260 ? cbor_boolean_stash 404 ? types_boolean_stash
261 : 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);
262 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 409 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
263 ? cbor_tagged_stash 410 ? cbor_tagged_stash
264 : gv_stashpv ("CBOR::XS::Tagged" , 1); 411 : gv_stashpv ("CBOR::XS::Tagged" , 1);
265 412
413 HV *stash = SvSTASH (sv);
414
266 if (SvSTASH (sv) == boolean_stash) 415 if (stash == boolean_stash)
267 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 }
268 else if (SvSTASH (sv) == tagged_stash) 425 else if (stash == tagged_stash)
269 { 426 {
270 if (svt != SVt_PVAV) 427 if (svt != SVt_PVAV)
271 croak ("encountered CBOR::XS::Tagged object that isn't an array"); 428 croak ("encountered CBOR::XS::Tagged object that isn't an array");
272 429
273 encode_uint (enc, 0xc0, SvUV (*av_fetch ((AV *)sv, 0, 1))); 430 encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1)));
274 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;
275 } 450 }
276 else 451 else
277 { 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);
278 // 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
279 GV *to_cbor = gv_fetchmethod_autoload (SvSTASH (sv), "TO_CBOR", 0);
280
281 if (to_cbor)
282 {
283 dSP;
284
285 ENTER; SAVETMPS; PUSHMARK (SP);
286 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv))); 471 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
287 472
288 // calling with G_SCALAR ensures that we always get a 1 return value
289 PUTBACK; 473 PUTBACK;
474 // G_SCALAR ensures that return value is 1
290 call_sv ((SV *)GvCV (to_cbor), G_SCALAR); 475 call_sv ((SV *)GvCV (method), G_SCALAR);
291 SPAGAIN; 476 SPAGAIN;
292 477
293 // catch this surprisingly common error 478 // catch this surprisingly common error
294 if (SvROK (TOPs) && SvRV (TOPs) == sv) 479 if (SvROK (TOPs) && SvRV (TOPs) == sv)
295 croak ("%s::TO_CBOR method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv))); 480 croak ("%s::TO_CBOR method returned same object as was passed instead of a new one", HvNAME (stash));
296 481
297 sv = POPs; 482 encode_sv (enc, POPs);
483
298 PUTBACK; 484 PUTBACK;
299 485
300 encode_sv (enc, sv);
301
302 FREETMPS; LEAVE; 486 FREETMPS; LEAVE;
303 } 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 }
304 else 519 else
305 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",
306 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 521 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
307 }
308 } 522 }
309 else if (svt == SVt_PVHV) 523 else if (svt == SVt_PVHV)
310 encode_hv (enc, (HV *)sv); 524 encode_hv (enc, (HV *)sv);
311 else if (svt == SVt_PVAV) 525 else if (svt == SVt_PVAV)
312 encode_av (enc, (AV *)sv); 526 encode_av (enc, (AV *)sv);
313 else if (svt < SVt_PVAV)
314 {
315 STRLEN len = 0;
316 char *pv = svt ? SvPV (sv, len) : 0;
317
318 if (len == 1 && *pv == '1')
319 encode_ch (enc, 0xe0 | 21);
320 else if (len == 1 && *pv == '0')
321 encode_ch (enc, 0xe0 | 20);
322 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
323 encode_ch (enc, 0xe0 | 23);
324 else
325 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
326 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
327 }
328 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
329 encode_ch (enc, 0xe0 | 23);
330 else 527 else
331 croak ("encountered %s, but CBOR can only represent references to arrays or hashes", 528 {
332 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 529 encode_tag (enc, CBOR_TAG_INDIRECTION);
530 encode_sv (enc, sv);
531 }
333} 532}
334 533
335static void 534static void
336encode_nv (enc_t *enc, SV *sv) 535encode_nv (enc_t *enc, SV *sv)
337{ 536{
338 double nv = SvNVX (sv); 537 double nv = SvNVX (sv);
339 538
340 need (enc, 9); 539 need (enc, 9);
341 540
342 if (ecb_expect_false (nv == (U32)nv)) 541 if (ecb_expect_false (nv == (NV)(U32)nv))
343 encode_uint (enc, 0x00, (U32)nv); 542 encode_uint (enc, MAJOR_POS_INT, (U32)nv);
344 //TODO: maybe I32? 543 //TODO: maybe I32?
345 else if (ecb_expect_false (nv == (float)nv)) 544 else if (ecb_expect_false (nv == (float)nv))
346 { 545 {
347 uint32_t fp = ecb_float_to_binary32 (nv); 546 uint32_t fp = ecb_float_to_binary32 (nv);
348 547
349 *enc->cur++ = 0xe0 | 26; 548 *enc->cur++ = MAJOR_MISC | MISC_FLOAT32;
350 549
351 if (!ecb_big_endian ()) 550 if (!ecb_big_endian ())
352 fp = ecb_bswap32 (fp); 551 fp = ecb_bswap32 (fp);
353 552
354 memcpy (enc->cur, &fp, 4); 553 memcpy (enc->cur, &fp, 4);
356 } 555 }
357 else 556 else
358 { 557 {
359 uint64_t fp = ecb_double_to_binary64 (nv); 558 uint64_t fp = ecb_double_to_binary64 (nv);
360 559
361 *enc->cur++ = 0xe0 | 27; 560 *enc->cur++ = MAJOR_MISC | MISC_FLOAT64;
362 561
363 if (!ecb_big_endian ()) 562 if (!ecb_big_endian ())
364 fp = ecb_bswap64 (fp); 563 fp = ecb_bswap64 (fp);
365 564
366 memcpy (enc->cur, &fp, 8); 565 memcpy (enc->cur, &fp, 8);
375 574
376 if (SvPOKp (sv)) 575 if (SvPOKp (sv))
377 { 576 {
378 STRLEN len; 577 STRLEN len;
379 char *str = SvPV (sv, len); 578 char *str = SvPV (sv, len);
380 encode_str (enc, SvUTF8 (sv), str, len); 579 encode_strref (enc, SvUTF8 (sv), str, len);
381 } 580 }
382 else if (SvNOKp (sv)) 581 else if (SvNOKp (sv))
383 encode_nv (enc, sv); 582 encode_nv (enc, sv);
384 else if (SvIOKp (sv)) 583 else if (SvIOKp (sv))
385 { 584 {
386 if (SvIsUV (sv)) 585 if (SvIsUV (sv))
387 encode_uint (enc, 0x00, SvUVX (sv)); 586 encode_uint (enc, MAJOR_POS_INT, SvUVX (sv));
388 else if (SvIVX (sv) >= 0) 587 else if (SvIVX (sv) >= 0)
389 encode_uint (enc, 0x00, SvIVX (sv)); 588 encode_uint (enc, MAJOR_POS_INT, SvIVX (sv));
390 else 589 else
391 encode_uint (enc, 0x20, -(SvIVX (sv) + 1)); 590 encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1));
392 } 591 }
393 else if (SvROK (sv)) 592 else if (SvROK (sv))
394 encode_rv (enc, SvRV (sv)); 593 encode_rv (enc, SvRV (sv));
395 else if (!SvOK (sv)) 594 else if (!SvOK (sv))
396 encode_ch (enc, 0xe0 | 22); 595 encode_ch (enc, MAJOR_MISC | SIMPLE_NULL);
397 else if (enc->cbor.flags & F_ALLOW_UNKNOWN) 596 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
398 encode_ch (enc, 0xe0 | 23); 597 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
399 else 598 else
400 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",
401 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv)); 600 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
402} 601}
403 602
404static SV * 603static SV *
405encode_cbor (SV *scalar, CBOR *cbor) 604encode_cbor (SV *scalar, CBOR *cbor)
406{ 605{
407 enc_t enc; 606 enc_t enc = { 0 };
408 607
409 enc.cbor = *cbor; 608 enc.cbor = *cbor;
410 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 609 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
411 enc.cur = SvPVX (enc.sv); 610 enc.cur = SvPVX (enc.sv);
412 enc.end = SvEND (enc.sv); 611 enc.end = SvEND (enc.sv);
413 enc.depth = 0;
414 612
415 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
416 encode_sv (&enc, scalar); 622 encode_sv (&enc, scalar);
417 623
418 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 624 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
419 *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
420 626
434 U8 *end; // end of input string 640 U8 *end; // end of input string
435 const char *err; // parse error, if != 0 641 const char *err; // parse error, if != 0
436 CBOR cbor; 642 CBOR cbor;
437 U32 depth; // recursion depth 643 U32 depth; // recursion depth
438 U32 maxdepth; // recursion depth limit 644 U32 maxdepth; // recursion depth limit
645 AV *shareable;
646 AV *stringref;
647 SV *decode_tagged;
439} dec_t; 648} dec_t;
440 649
441#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
442 651
443#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")
446#define DEC_DEC_DEPTH --dec->depth 655#define DEC_DEC_DEPTH --dec->depth
447 656
448static UV 657static UV
449decode_uint (dec_t *dec) 658decode_uint (dec_t *dec)
450{ 659{
451 switch (*dec->cur & 31) 660 U8 m = *dec->cur & MINOR_MASK;
452 { 661 ++dec->cur;
453 case 0: case 1: case 2: case 3: case 4: case 5: case 6: case 7:
454 case 8: case 9: case 10: case 11: case 12: case 13: case 14: case 15:
455 case 16: case 17: case 18: case 19: case 20: case 21: case 22: case 23:
456 return *dec->cur++ & 31;
457 662
458 case 24: 663 if (ecb_expect_true (m < LENGTH_EXT1))
664 return m;
665 else if (ecb_expect_true (m == LENGTH_EXT1))
666 {
459 WANT (2); 667 WANT (1);
460 dec->cur += 2; 668 dec->cur += 1;
461 return dec->cur[-1]; 669 return dec->cur[-1];
462 670 }
463 case 25: 671 else if (ecb_expect_true (m == LENGTH_EXT2))
672 {
464 WANT (3); 673 WANT (2);
465 dec->cur += 3; 674 dec->cur += 2;
466 return (((UV)dec->cur[-2]) << 8) 675 return (((UV)dec->cur[-2]) << 8)
467 | ((UV)dec->cur[-1]); 676 | ((UV)dec->cur[-1]);
468 677 }
469 case 26: 678 else if (ecb_expect_true (m == LENGTH_EXT4))
679 {
470 WANT (5); 680 WANT (4);
471 dec->cur += 5; 681 dec->cur += 4;
472 return (((UV)dec->cur[-4]) << 24) 682 return (((UV)dec->cur[-4]) << 24)
473 | (((UV)dec->cur[-3]) << 16) 683 | (((UV)dec->cur[-3]) << 16)
474 | (((UV)dec->cur[-2]) << 8) 684 | (((UV)dec->cur[-2]) << 8)
475 | ((UV)dec->cur[-1]); 685 | ((UV)dec->cur[-1]);
476 686 }
477 case 27: 687 else if (ecb_expect_true (m == LENGTH_EXT8))
688 {
478 WANT (9); 689 WANT (8);
479 dec->cur += 9; 690 dec->cur += 8;
691
692 return
693#if UVSIZE < 8
694 0
695#else
480 return (((UV)dec->cur[-8]) << 56) 696 (((UV)dec->cur[-8]) << 56)
481 | (((UV)dec->cur[-7]) << 48) 697 | (((UV)dec->cur[-7]) << 48)
482 | (((UV)dec->cur[-6]) << 40) 698 | (((UV)dec->cur[-6]) << 40)
483 | (((UV)dec->cur[-5]) << 32) 699 | (((UV)dec->cur[-5]) << 32)
700#endif
484 | (((UV)dec->cur[-4]) << 24) 701 | (((UV)dec->cur[-4]) << 24)
485 | (((UV)dec->cur[-3]) << 16) 702 | (((UV)dec->cur[-3]) << 16)
486 | (((UV)dec->cur[-2]) << 8) 703 | (((UV)dec->cur[-2]) << 8)
487 | ((UV)dec->cur[-1]); 704 | ((UV)dec->cur[-1]);
488 705 }
489 default: 706 else
490 ERR ("corrupted CBOR data (unsupported integer minor encoding)"); 707 ERR ("corrupted CBOR data (unsupported integer minor encoding)");
491 }
492 708
493fail: 709fail:
494 return 0; 710 return 0;
495} 711}
496 712
501{ 717{
502 AV *av = newAV (); 718 AV *av = newAV ();
503 719
504 DEC_INC_DEPTH; 720 DEC_INC_DEPTH;
505 721
506 if ((*dec->cur & 31) == 31) 722 if (*dec->cur == (MAJOR_ARRAY | MINOR_INDEF))
507 { 723 {
508 ++dec->cur; 724 ++dec->cur;
509 725
510 for (;;) 726 for (;;)
511 { 727 {
512 WANT (1); 728 WANT (1);
513 729
514 if (*dec->cur == (0xe0 | 31)) 730 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
515 { 731 {
516 ++dec->cur; 732 ++dec->cur;
517 break; 733 break;
518 } 734 }
519 735
522 } 738 }
523 else 739 else
524 { 740 {
525 int i, len = decode_uint (dec); 741 int i, len = decode_uint (dec);
526 742
743 WANT (len); // complexity check for av_fill - need at least one byte per value, do not allow supersize arrays
527 av_fill (av, len - 1); 744 av_fill (av, len - 1);
528 745
529 for (i = 0; i < len; ++i) 746 for (i = 0; i < len; ++i)
530 AvARRAY (av)[i] = decode_sv (dec); 747 AvARRAY (av)[i] = decode_sv (dec);
531 } 748 }
537 SvREFCNT_dec (av); 754 SvREFCNT_dec (av);
538 DEC_DEC_DEPTH; 755 DEC_DEC_DEPTH;
539 return &PL_sv_undef; 756 return &PL_sv_undef;
540} 757}
541 758
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
542static SV * 805static SV *
543decode_hv (dec_t *dec) 806decode_hv (dec_t *dec)
544{ 807{
545 HV *hv = newHV (); 808 HV *hv = newHV ();
546 809
547 DEC_INC_DEPTH; 810 DEC_INC_DEPTH;
548 811
549 if ((*dec->cur & 31) == 31) 812 if (*dec->cur == (MAJOR_MAP | MINOR_INDEF))
550 { 813 {
551 ++dec->cur; 814 ++dec->cur;
552 815
553 for (;;) 816 for (;;)
554 { 817 {
555 WANT (1); 818 WANT (1);
556 819
557 if (*dec->cur == (0xe0 | 31)) 820 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
558 { 821 {
559 ++dec->cur; 822 ++dec->cur;
560 break; 823 break;
561 } 824 }
562 825
563 SV *k = decode_sv (dec); 826 decode_he (dec, hv);
564 SV *v = decode_sv (dec);
565
566 hv_store_ent (hv, k, v, 0);
567 } 827 }
568 } 828 }
569 else 829 else
570 { 830 {
571 int len = decode_uint (dec); 831 int pairs = decode_uint (dec);
572 832
573 while (len--) 833 while (pairs--)
574 { 834 decode_he (dec, hv);
575 SV *k = decode_sv (dec);
576 SV *v = decode_sv (dec);
577
578 hv_store_ent (hv, k, v, 0);
579 }
580 } 835 }
581 836
582 DEC_DEC_DEPTH; 837 DEC_DEC_DEPTH;
583 return newRV_noinc ((SV *)hv); 838 return newRV_noinc ((SV *)hv);
584 839
591static SV * 846static SV *
592decode_str (dec_t *dec, int utf8) 847decode_str (dec_t *dec, int utf8)
593{ 848{
594 SV *sv = 0; 849 SV *sv = 0;
595 850
596 if ((*dec->cur & 31) == 31) 851 if ((*dec->cur & MINOR_MASK) == MINOR_INDEF)
597 { 852 {
853 // indefinite length strings
598 ++dec->cur; 854 ++dec->cur;
599 855
856 U8 major = *dec->cur & MAJOR_MISC;
857
600 sv = newSVpvn ("", 0); 858 sv = newSVpvn ("", 0);
601 859
602 // not very fast, and certainly not robust against illegal input
603 for (;;) 860 for (;;)
604 { 861 {
605 WANT (1); 862 WANT (1);
606 863
607 if (*dec->cur == (0xe0 | 31)) 864 if ((*dec->cur - major) > LENGTH_EXT8)
865 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
608 { 866 {
609 ++dec->cur; 867 ++dec->cur;
610 break; 868 break;
611 } 869 }
870 else
871 ERR ("corrupted CBOR data (invalid chunks in indefinite length string)");
612 872
613 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;
614 } 878 }
615 } 879 }
616 else 880 else
617 { 881 {
618 STRLEN len = decode_uint (dec); 882 STRLEN len = decode_uint (dec);
619 883
620 WANT (len); 884 WANT (len);
621 sv = newSVpvn (dec->cur, len); 885 sv = newSVpvn (dec->cur, len);
622 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));
623 } 891 }
624 892
625 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
626 SvUTF8_on (sv); 899 SvUTF8_on (sv);
900 }
627 901
628 return sv; 902 return sv;
629 903
630fail: 904fail:
631 SvREFCNT_dec (sv); 905 SvREFCNT_dec (sv);
633} 907}
634 908
635static SV * 909static SV *
636decode_tagged (dec_t *dec) 910decode_tagged (dec_t *dec)
637{ 911{
912 SV *sv = 0;
638 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:
639 SV *sv = decode_sv (dec); 920 sv = decode_sv (dec);
921 break;
640 922
641 if (tag == CBOR_TAG_MAGIC) // 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
642 return sv; 1093 return sv;
643 1094
644 AV *av = newAV (); 1095fail:
645 av_push (av, newSVuv (tag)); 1096 SvREFCNT_dec (sv);
646 av_push (av, sv); 1097 return &PL_sv_undef;
647
648 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
649 ? cbor_tagged_stash
650 : gv_stashpv ("CBOR::XS::Tagged" , 1);
651
652 return sv_bless (newRV_noinc ((SV *)av), tagged_stash);
653} 1098}
654 1099
655static SV * 1100static SV *
656decode_sv (dec_t *dec) 1101decode_sv (dec_t *dec)
657{ 1102{
658 WANT (1); 1103 WANT (1);
659 1104
660 switch (*dec->cur >> 5) 1105 switch (*dec->cur >> MAJOR_SHIFT)
661 { 1106 {
662 case 0: // unsigned int 1107 case MAJOR_POS_INT >> MAJOR_SHIFT: return newSVuv (decode_uint (dec));
663 return newSVuv (decode_uint (dec)); 1108 case MAJOR_NEG_INT >> MAJOR_SHIFT: return newSViv (-1 - (IV)decode_uint (dec));
664 case 1: // negative int 1109 case MAJOR_BYTES >> MAJOR_SHIFT: return decode_str (dec, 0);
665 return newSViv (-1 - (IV)decode_uint (dec)); 1110 case MAJOR_TEXT >> MAJOR_SHIFT: return decode_str (dec, 1);
666 case 2: // octet string 1111 case MAJOR_ARRAY >> MAJOR_SHIFT: return decode_av (dec);
667 return decode_str (dec, 0); 1112 case MAJOR_MAP >> MAJOR_SHIFT: return decode_hv (dec);
668 case 3: // utf-8 string 1113 case MAJOR_TAG >> MAJOR_SHIFT: return decode_tagged (dec);
669 return decode_str (dec, 1); 1114
670 case 4: // array 1115 case MAJOR_MISC >> MAJOR_SHIFT:
671 return decode_av (dec);
672 case 5: // map
673 return decode_hv (dec);
674 case 6: // tag
675 return decode_tagged (dec);
676 case 7: // misc
677 switch (*dec->cur++ & 31) 1116 switch (*dec->cur++ & MINOR_MASK)
678 { 1117 {
679 case 20: 1118 case SIMPLE_FALSE:
680#if CBOR_SLOW 1119#if CBOR_SLOW
681 cbor_false = get_bool ("CBOR::XS::false"); 1120 types_false = get_bool ("Types::Serialiser::false");
682#endif 1121#endif
683 return newSVsv (cbor_false); 1122 return newSVsv (types_false);
684 case 21: 1123 case SIMPLE_TRUE:
685#if CBOR_SLOW 1124#if CBOR_SLOW
686 cbor_true = get_bool ("CBOR::XS::true"); 1125 types_true = get_bool ("Types::Serialiser::true");
687#endif 1126#endif
688 return newSVsv (cbor_true); 1127 return newSVsv (types_true);
689 case 22: 1128 case SIMPLE_NULL:
690 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);
691 1135
692 case 25: 1136 case MISC_FLOAT16:
693 { 1137 {
694 WANT (2); 1138 WANT (2);
695 1139
696 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1]; 1140 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1];
697 dec->cur += 2; 1141 dec->cur += 2;
698 1142
699 return newSVnv (ecb_binary16_to_float (fp)); 1143 return newSVnv (ecb_binary16_to_float (fp));
700 } 1144 }
701 1145
702 case 26: 1146 case MISC_FLOAT32:
703 { 1147 {
704 uint32_t fp; 1148 uint32_t fp;
705 WANT (4); 1149 WANT (4);
706 memcpy (&fp, dec->cur, 4); 1150 memcpy (&fp, dec->cur, 4);
707 dec->cur += 4; 1151 dec->cur += 4;
710 fp = ecb_bswap32 (fp); 1154 fp = ecb_bswap32 (fp);
711 1155
712 return newSVnv (ecb_binary32_to_float (fp)); 1156 return newSVnv (ecb_binary32_to_float (fp));
713 } 1157 }
714 1158
715 case 27: 1159 case MISC_FLOAT64:
716 { 1160 {
717 uint64_t fp; 1161 uint64_t fp;
718 WANT (8); 1162 WANT (8);
719 memcpy (&fp, dec->cur, 8); 1163 memcpy (&fp, dec->cur, 8);
720 dec->cur += 8; 1164 dec->cur += 8;
723 fp = ecb_bswap64 (fp); 1167 fp = ecb_bswap64 (fp);
724 1168
725 return newSVnv (ecb_binary64_to_double (fp)); 1169 return newSVnv (ecb_binary64_to_double (fp));
726 } 1170 }
727 1171
728 // 0..19 unassigned 1172 // 0..19 unassigned simple
729 // 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
730 default: 1176 default:
731 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)"); 1177 ERR ("corrupted CBOR data (reserved/unassigned/unexpected major 7 value)");
732 } 1178 }
733 1179
734 break; 1180 break;
735 } 1181 }
736 1182
739} 1185}
740 1186
741static SV * 1187static SV *
742decode_cbor (SV *string, CBOR *cbor, char **offset_return) 1188decode_cbor (SV *string, CBOR *cbor, char **offset_return)
743{ 1189{
744 dec_t dec; 1190 dec_t dec = { 0 };
745 SV *sv; 1191 SV *sv;
1192 STRLEN len;
1193 char *data = SvPVbyte (string, len);
746 1194
747 /* work around bugs in 5.10 where manipulating magic values
748 * makes perl ignore the magic in subsequent accesses.
749 * also make a copy of non-PV values, to get them into a clean
750 * state (SvPV should do that, but it's buggy, see below).
751 */
752 /*SvGETMAGIC (string);*/
753 if (SvMAGICAL (string) || !SvPOK (string))
754 string = sv_2mortal (newSVsv (string));
755
756 SvUPGRADE (string, SVt_PV);
757
758 /* work around a bug in perl 5.10, which causes SvCUR to fail an
759 * assertion with -DDEBUGGING, although SvCUR is documented to
760 * return the xpv_cur field which certainly exists after upgrading.
761 * according to nicholas clark, calling SvPOK fixes this.
762 * But it doesn't fix it, so try another workaround, call SvPV_nolen
763 * and hope for the best.
764 * Damnit, SvPV_nolen still trips over yet another assertion. This
765 * assertion business is seriously broken, try yet another workaround
766 * for the broken -DDEBUGGING.
767 */
768 {
769#ifdef DEBUGGING
770 STRLEN offset = SvOK (string) ? sv_len (string) : 0;
771#else
772 STRLEN offset = SvCUR (string);
773#endif
774
775 if (offset > cbor->max_size && cbor->max_size) 1195 if (len > cbor->max_size && cbor->max_size)
776 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu", 1196 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
777 (unsigned long)SvCUR (string), (unsigned long)cbor->max_size); 1197 (unsigned long)len, (unsigned long)cbor->max_size);
778 }
779
780 sv_utf8_downgrade (string, 0);
781 1198
782 dec.cbor = *cbor; 1199 dec.cbor = *cbor;
783 dec.cur = (U8 *)SvPVX (string); 1200 dec.cur = (U8 *)data;
784 dec.end = (U8 *)SvEND (string); 1201 dec.end = (U8 *)data + len;
785 dec.err = 0;
786 dec.depth = 0;
787 1202
788 sv = decode_sv (&dec); 1203 sv = decode_sv (&dec);
789 1204
790 if (offset_return) 1205 if (offset_return)
791 *offset_return = dec.cur; 1206 *offset_return = dec.cur;
794 if (dec.cur != dec.end && !dec.err) 1209 if (dec.cur != dec.end && !dec.err)
795 dec.err = "garbage after CBOR object"; 1210 dec.err = "garbage after CBOR object";
796 1211
797 if (dec.err) 1212 if (dec.err)
798 { 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
799 SvREFCNT_dec (sv); 1225 SvREFCNT_dec (sv);
800 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);
801 } 1227 }
802 1228
803 sv = sv_2mortal (sv); 1229 sv = sv_2mortal (sv);
804 1230
805 return sv; 1231 return sv;
806} 1232}
807 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
808///////////////////////////////////////////////////////////////////////////// 1358/////////////////////////////////////////////////////////////////////////////
809// XS interface functions 1359// XS interface functions
810 1360
811MODULE = CBOR::XS PACKAGE = CBOR::XS 1361MODULE = CBOR::XS PACKAGE = CBOR::XS
812 1362
813BOOT: 1363BOOT:
814{ 1364{
815 cbor_stash = gv_stashpv ("CBOR::XS" , 1); 1365 cbor_stash = gv_stashpv ("CBOR::XS" , 1);
816 cbor_boolean_stash = gv_stashpv ("CBOR::XS::Boolean", 1);
817 cbor_tagged_stash = gv_stashpv ("CBOR::XS::Tagged" , 1); 1366 cbor_tagged_stash = gv_stashpv ("CBOR::XS::Tagged" , 1);
818 1367
819 cbor_true = get_bool ("CBOR::XS::true"); 1368 types_boolean_stash = gv_stashpv ("Types::Serialiser::Boolean", 1);
820 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);
821} 1379}
822 1380
823PROTOTYPES: DISABLE 1381PROTOTYPES: DISABLE
824 1382
825void CLONE (...) 1383void CLONE (...)
826 CODE: 1384 CODE:
827 cbor_stash = 0; 1385 cbor_stash = 0;
828 cbor_boolean_stash = 0;
829 cbor_tagged_stash = 0; 1386 cbor_tagged_stash = 0;
1387 types_error_stash = 0;
1388 types_boolean_stash = 0;
830 1389
831void new (char *klass) 1390void new (char *klass)
832 PPCODE: 1391 PPCODE:
833{ 1392{
834 SV *pv = NEWSV (0, sizeof (CBOR)); 1393 SV *pv = NEWSV (0, sizeof (CBOR));
842 1401
843void shrink (CBOR *self, int enable = 1) 1402void shrink (CBOR *self, int enable = 1)
844 ALIAS: 1403 ALIAS:
845 shrink = F_SHRINK 1404 shrink = F_SHRINK
846 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
847 PPCODE: 1411 PPCODE:
848{ 1412{
849 if (enable) 1413 if (enable)
850 self->flags |= ix; 1414 self->flags |= ix;
851 else 1415 else
856 1420
857void get_shrink (CBOR *self) 1421void get_shrink (CBOR *self)
858 ALIAS: 1422 ALIAS:
859 get_shrink = F_SHRINK 1423 get_shrink = F_SHRINK
860 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
861 PPCODE: 1429 PPCODE:
862 XPUSHs (boolSV (self->flags & ix)); 1430 XPUSHs (boolSV (self->flags & ix));
863 1431
864void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) 1432void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
865 PPCODE: 1433 PPCODE:
881 CODE: 1449 CODE:
882 RETVAL = self->max_size; 1450 RETVAL = self->max_size;
883 OUTPUT: 1451 OUTPUT:
884 RETVAL 1452 RETVAL
885 1453
886#if 0 //TODO 1454void filter (CBOR *self, SV *filter = 0)
887
888void filter_cbor_object (CBOR *self, SV *cb = &PL_sv_undef)
889 PPCODE: 1455 PPCODE:
890{
891 SvREFCNT_dec (self->cb_object); 1456 SvREFCNT_dec (self->filter);
892 self->cb_object = SvOK (cb) ? newSVsv (cb) : 0; 1457 self->filter = filter ? newSVsv (filter) : filter;
893
894 XPUSHs (ST (0)); 1458 XPUSHs (ST (0));
895}
896 1459
897void filter_cbor_single_key_object (CBOR *self, SV *key, SV *cb = &PL_sv_undef) 1460SV *get_filter (CBOR *self)
898 PPCODE: 1461 CODE:
899{ 1462 RETVAL = self->filter ? self->filter : NEWSV (0, 0);
900 if (!self->cb_sk_object) 1463 OUTPUT:
901 self->cb_sk_object = newHV (); 1464 RETVAL
902
903 if (SvOK (cb))
904 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);
905 else
906 {
907 hv_delete_ent (self->cb_sk_object, key, G_DISCARD, 0);
908
909 if (!HvKEYS (self->cb_sk_object))
910 {
911 SvREFCNT_dec (self->cb_sk_object);
912 self->cb_sk_object = 0;
913 }
914 }
915
916 XPUSHs (ST (0));
917}
918
919#endif
920 1465
921void encode (CBOR *self, SV *scalar) 1466void encode (CBOR *self, SV *scalar)
922 PPCODE: 1467 PPCODE:
923 PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN; 1468 PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN;
924 XPUSHs (scalar); 1469 XPUSHs (scalar);
937 EXTEND (SP, 2); 1482 EXTEND (SP, 2);
938 PUSHs (sv); 1483 PUSHs (sv);
939 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); 1484 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
940} 1485}
941 1486
942#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}
943 1538
944void DESTROY (CBOR *self) 1539void DESTROY (CBOR *self)
945 CODE: 1540 PPCODE:
946 SvREFCNT_dec (self->cb_sk_object); 1541 cbor_free (self);
947 SvREFCNT_dec (self->cb_object);
948
949#endif
950 1542
951PROTOTYPES: ENABLE 1543PROTOTYPES: ENABLE
952 1544
953void encode_cbor (SV *scalar) 1545void encode_cbor (SV *scalar)
1546 ALIAS:
1547 encode_cbor = 0
1548 encode_cbor_sharing = F_ALLOW_SHARING
954 PPCODE: 1549 PPCODE:
955{ 1550{
956 CBOR cbor; 1551 CBOR cbor;
957 cbor_init (&cbor); 1552 cbor_init (&cbor);
1553 cbor.flags |= ix;
958 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN; 1554 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;
959 XPUSHs (scalar); 1555 XPUSHs (scalar);
960} 1556}
961 1557
962void decode_cbor (SV *cborstr) 1558void decode_cbor (SV *cborstr)

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines