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 1.6 by root, Sat Oct 26 22:25:47 2013 UTC vs.
Revision 1.50 by root, Thu Feb 25 02:29:22 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_VALIDATE_UTF8 0x00000020UL
16 106
17#define INIT_SIZE 32 // initial scalar size to be allocated 107#define INIT_SIZE 32 // initial scalar size to be allocated
18 108
19#define SB do { 109#define SB do {
20#define SE } while (0) 110#define SE } while (0)
31#else 121#else
32# define CBOR_SLOW 0 122# define CBOR_SLOW 0
33# define CBOR_STASH cbor_stash 123# define CBOR_STASH cbor_stash
34#endif 124#endif
35 125
36static HV *cbor_stash, *cbor_boolean_stash, *cbor_tagged_stash; // CBOR::XS:: 126static HV *cbor_stash, *types_boolean_stash, *types_error_stash, *cbor_tagged_stash; // CBOR::XS::
37static SV *cbor_true, *cbor_false; 127static SV *types_true, *types_false, *types_error, *sv_cbor, *default_filter;
38 128
39typedef struct { 129typedef struct {
40 U32 flags; 130 U32 flags;
41 U32 max_depth; 131 U32 max_depth;
42 STRLEN max_size; 132 STRLEN max_size;
133 SV *filter;
134
135 // for the incremental parser
136 STRLEN incr_pos; // the current offset into the text
137 STRLEN incr_need; // minimum bytes needed to decode
138 AV *incr_count; // for every nesting level, the number of outstanding values, or -1 for indef.
43} CBOR; 139} CBOR;
44 140
45ecb_inline void 141ecb_inline void
46cbor_init (CBOR *cbor) 142cbor_init (CBOR *cbor)
47{ 143{
48 Zero (cbor, 1, CBOR); 144 Zero (cbor, 1, CBOR);
49 cbor->max_depth = 512; 145 cbor->max_depth = 512;
146}
147
148ecb_inline void
149cbor_free (CBOR *cbor)
150{
151 SvREFCNT_dec (cbor->filter);
152 SvREFCNT_dec (cbor->incr_count);
50} 153}
51 154
52///////////////////////////////////////////////////////////////////////////// 155/////////////////////////////////////////////////////////////////////////////
53// utility functions 156// utility functions
54 157
76 SvPV_renew (sv, SvCUR (sv) + 1); 179 SvPV_renew (sv, SvCUR (sv) + 1);
77#endif 180#endif
78 } 181 }
79} 182}
80 183
81///////////////////////////////////////////////////////////////////////////// 184// minimum length of a string to be registered for stringref
82// fp hell 185ecb_inline int
83 186minimum_string_length (UV idx)
84//TODO 187{
188 return idx > 23
189 ? idx > 0xffU
190 ? idx > 0xffffU
191 ? idx > 0xffffffffU
192 ? 11
193 : 7
194 : 5
195 : 4
196 : 3;
197}
85 198
86///////////////////////////////////////////////////////////////////////////// 199/////////////////////////////////////////////////////////////////////////////
87// encoder 200// encoder
88 201
89// structure used for encoding CBOR 202// structure used for encoding CBOR
92 char *cur; // SvPVX (sv) + current output position 205 char *cur; // SvPVX (sv) + current output position
93 char *end; // SvEND (sv) 206 char *end; // SvEND (sv)
94 SV *sv; // result scalar 207 SV *sv; // result scalar
95 CBOR cbor; 208 CBOR cbor;
96 U32 depth; // recursion level 209 U32 depth; // recursion level
210 HV *stringref[2]; // string => index, or 0 ([0] = bytes, [1] = utf-8)
211 UV stringref_idx;
212 HV *shareable; // ptr => index, or 0
213 UV shareable_idx;
97} enc_t; 214} enc_t;
98 215
99ecb_inline void 216ecb_inline void
100need (enc_t *enc, STRLEN len) 217need (enc_t *enc, STRLEN len)
101{ 218{
118static void 235static void
119encode_uint (enc_t *enc, int major, UV len) 236encode_uint (enc_t *enc, int major, UV len)
120{ 237{
121 need (enc, 9); 238 need (enc, 9);
122 239
123 if (len < 24) 240 if (ecb_expect_true (len < LENGTH_EXT1))
124 *enc->cur++ = major | len; 241 *enc->cur++ = major | len;
125 else if (len <= 0xff) 242 else if (ecb_expect_true (len <= 0xffU))
126 { 243 {
127 *enc->cur++ = major | 24; 244 *enc->cur++ = major | LENGTH_EXT1;
128 *enc->cur++ = len; 245 *enc->cur++ = len;
129 } 246 }
130 else if (len <= 0xffff) 247 else if (len <= 0xffffU)
131 { 248 {
132 *enc->cur++ = major | 25; 249 *enc->cur++ = major | LENGTH_EXT2;
133 *enc->cur++ = len >> 8; 250 *enc->cur++ = len >> 8;
134 *enc->cur++ = len; 251 *enc->cur++ = len;
135 } 252 }
136 else if (len <= 0xffffffff) 253 else if (len <= 0xffffffffU)
137 { 254 {
138 *enc->cur++ = major | 26; 255 *enc->cur++ = major | LENGTH_EXT4;
139 *enc->cur++ = len >> 24; 256 *enc->cur++ = len >> 24;
140 *enc->cur++ = len >> 16; 257 *enc->cur++ = len >> 16;
141 *enc->cur++ = len >> 8; 258 *enc->cur++ = len >> 8;
142 *enc->cur++ = len; 259 *enc->cur++ = len;
143 } 260 }
144 else 261 else
145 { 262 {
146 *enc->cur++ = major | 27; 263 *enc->cur++ = major | LENGTH_EXT8;
147 *enc->cur++ = len >> 56; 264 *enc->cur++ = len >> 56;
148 *enc->cur++ = len >> 48; 265 *enc->cur++ = len >> 48;
149 *enc->cur++ = len >> 40; 266 *enc->cur++ = len >> 40;
150 *enc->cur++ = len >> 32; 267 *enc->cur++ = len >> 32;
151 *enc->cur++ = len >> 24; 268 *enc->cur++ = len >> 24;
153 *enc->cur++ = len >> 8; 270 *enc->cur++ = len >> 8;
154 *enc->cur++ = len; 271 *enc->cur++ = len;
155 } 272 }
156} 273}
157 274
158static void 275ecb_inline void
276encode_tag (enc_t *enc, UV tag)
277{
278 encode_uint (enc, MAJOR_TAG, tag);
279}
280
281ecb_inline void
159encode_str (enc_t *enc, int utf8, char *str, STRLEN len) 282encode_str (enc_t *enc, int utf8, char *str, STRLEN len)
160{ 283{
161 encode_uint (enc, utf8 ? 0x60 : 0x40, len); 284 encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len);
162 need (enc, len); 285 need (enc, len);
163 memcpy (enc->cur, str, len); 286 memcpy (enc->cur, str, len);
164 enc->cur += len; 287 enc->cur += len;
165} 288}
166 289
290static void
291encode_strref (enc_t *enc, int utf8, char *str, STRLEN len)
292{
293 if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS))
294 {
295 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1);
296
297 if (SvOK (*svp))
298 {
299 // already registered, use stringref
300 encode_tag (enc, CBOR_TAG_STRINGREF);
301 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
302 return;
303 }
304 else if (len >= minimum_string_length (enc->stringref_idx))
305 {
306 // register only
307 sv_setuv (*svp, enc->stringref_idx);
308 ++enc->stringref_idx;
309 }
310 }
311
312 encode_str (enc, utf8, str, len);
313}
314
167static void encode_sv (enc_t *enc, SV *sv); 315static void encode_sv (enc_t *enc, SV *sv);
168 316
169static void 317static void
170encode_av (enc_t *enc, AV *av) 318encode_av (enc_t *enc, AV *av)
171{ 319{
174 if (enc->depth >= enc->cbor.max_depth) 322 if (enc->depth >= enc->cbor.max_depth)
175 croak (ERR_NESTING_EXCEEDED); 323 croak (ERR_NESTING_EXCEEDED);
176 324
177 ++enc->depth; 325 ++enc->depth;
178 326
179 encode_uint (enc, 0x80, len + 1); 327 encode_uint (enc, MAJOR_ARRAY, len + 1);
180 328
329 if (SvMAGICAL (av))
181 for (i = 0; i <= len; ++i) 330 for (i = 0; i <= len; ++i)
182 { 331 {
183 SV **svp = av_fetch (av, i, 0); 332 SV **svp = av_fetch (av, i, 0);
184 encode_sv (enc, svp ? *svp : &PL_sv_undef); 333 encode_sv (enc, svp ? *svp : &PL_sv_undef);
185 } 334 }
335 else
336 for (i = 0; i <= len; ++i)
337 {
338 SV *sv = AvARRAY (av)[i];
339 encode_sv (enc, sv ? sv : &PL_sv_undef);
340 }
186 341
187 --enc->depth; 342 --enc->depth;
188} 343}
189 344
190static void 345static void
199 354
200 int pairs = hv_iterinit (hv); 355 int pairs = hv_iterinit (hv);
201 int mg = SvMAGICAL (hv); 356 int mg = SvMAGICAL (hv);
202 357
203 if (mg) 358 if (mg)
204 encode_ch (enc, 0xa0 | 31); 359 encode_ch (enc, MAJOR_MAP | MINOR_INDEF);
205 else 360 else
206 encode_uint (enc, 0xa0, pairs); 361 encode_uint (enc, MAJOR_MAP, pairs);
207 362
208 while ((he = hv_iternext (hv))) 363 while ((he = hv_iternext (hv)))
209 { 364 {
210 if (HeKLEN (he) == HEf_SVKEY) 365 if (HeKLEN (he) == HEf_SVKEY)
211 encode_sv (enc, HeSVKEY (he)); 366 encode_sv (enc, HeSVKEY (he));
212 else 367 else
213 encode_str (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he)); 368 encode_strref (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he));
214 369
215 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he)); 370 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he));
216 } 371 }
217 372
218 if (mg) 373 if (mg)
219 encode_ch (enc, 0xe0 | 31); 374 encode_ch (enc, MAJOR_MISC | MINOR_INDEF);
220 375
221 --enc->depth; 376 --enc->depth;
222} 377}
223 378
224// encode objects, arrays and special \0=false and \1=true values. 379// encode objects, arrays and special \0=false and \1=true values.
225static void 380static void
226encode_rv (enc_t *enc, SV *sv) 381encode_rv (enc_t *enc, SV *sv)
227{ 382{
228 svtype svt;
229
230 SvGETMAGIC (sv); 383 SvGETMAGIC (sv);
384
231 svt = SvTYPE (sv); 385 svtype svt = SvTYPE (sv);
232 386
233 if (ecb_expect_false (SvOBJECT (sv))) 387 if (ecb_expect_false (SvOBJECT (sv)))
234 { 388 {
235 HV *boolean_stash = !CBOR_SLOW || cbor_boolean_stash 389 HV *boolean_stash = !CBOR_SLOW || types_boolean_stash
236 ? cbor_boolean_stash 390 ? types_boolean_stash
237 : gv_stashpv ("CBOR::XS::Boolean", 1); 391 : gv_stashpv ("Types::Serialiser::Boolean", 1);
392 HV *error_stash = !CBOR_SLOW || types_error_stash
393 ? types_error_stash
394 : gv_stashpv ("Types::Serialiser::Error", 1);
238 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 395 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
239 ? cbor_tagged_stash 396 ? cbor_tagged_stash
240 : gv_stashpv ("CBOR::XS::Tagged" , 1); 397 : gv_stashpv ("CBOR::XS::Tagged" , 1);
241 398
399 HV *stash = SvSTASH (sv);
400
242 if (SvSTASH (sv) == boolean_stash) 401 if (stash == boolean_stash)
243 encode_ch (enc, SvIV (sv) ? 0xe0 | 21 : 0xe0 | 20); 402 {
403 encode_ch (enc, SvIV (sv) ? MAJOR_MISC | SIMPLE_TRUE : MAJOR_MISC | SIMPLE_FALSE);
404 return;
405 }
406 else if (stash == error_stash)
407 {
408 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
409 return;
410 }
244 else if (SvSTASH (sv) == tagged_stash) 411 else if (stash == tagged_stash)
245 { 412 {
246 if (svt != SVt_PVAV) 413 if (svt != SVt_PVAV)
247 croak ("encountered CBOR::XS::Tagged object that isn't an array"); 414 croak ("encountered CBOR::XS::Tagged object that isn't an array");
248 415
249 encode_uint (enc, 0xc0, SvUV (*av_fetch ((AV *)sv, 0, 1))); 416 encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1)));
250 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1)); 417 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1));
418
419 return;
420 }
421 }
422
423 if (ecb_expect_false (SvREFCNT (sv) > 1)
424 && ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING))
425 {
426 if (!enc->shareable)
427 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
428
429 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
430
431 if (SvOK (*svp))
432 {
433 encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF);
434 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
435 return;
251 } 436 }
252 else 437 else
253 { 438 {
439 sv_setuv (*svp, enc->shareable_idx);
440 ++enc->shareable_idx;
441 encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE);
442 }
443 }
444
445 if (ecb_expect_false (SvOBJECT (sv)))
446 {
447 HV *stash = SvSTASH (sv);
448 GV *method;
449
450 if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
451 {
452 dSP;
453
454 ENTER; SAVETMPS;
455 PUSHMARK (SP);
254 // we re-bless the reference to get overload and other niceties right 456 // 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))); 457 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
263 458
264 // calling with G_SCALAR ensures that we always get a 1 return value
265 PUTBACK; 459 PUTBACK;
460 // G_SCALAR ensures that return value is 1
266 call_sv ((SV *)GvCV (to_cbor), G_SCALAR); 461 call_sv ((SV *)GvCV (method), G_SCALAR);
267 SPAGAIN; 462 SPAGAIN;
268 463
269 // catch this surprisingly common error 464 // catch this surprisingly common error
270 if (SvROK (TOPs) && SvRV (TOPs) == sv) 465 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))); 466 croak ("%s::TO_CBOR method returned same object as was passed instead of a new one", HvNAME (stash));
272 467
273 sv = POPs; 468 encode_sv (enc, POPs);
469
274 PUTBACK; 470 PUTBACK;
275 471
276 encode_sv (enc, sv);
277
278 FREETMPS; LEAVE; 472 FREETMPS; LEAVE;
279 } 473 }
474 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)
475 {
476 dSP;
477
478 ENTER; SAVETMPS;
479 SAVESTACK_POS ();
480 PUSHMARK (SP);
481 EXTEND (SP, 2);
482 // we re-bless the reference to get overload and other niceties right
483 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
484 PUSHs (sv_cbor);
485
486 PUTBACK;
487 int count = call_sv ((SV *)GvCV (method), G_ARRAY);
488 SPAGAIN;
489
490 // catch this surprisingly common error
491 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)
492 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));
493
494 encode_tag (enc, CBOR_TAG_PERL_OBJECT);
495 encode_uint (enc, MAJOR_ARRAY, count + 1);
496 encode_strref (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
497
498 while (count)
499 encode_sv (enc, SP[1 - count--]);
500
501 PUTBACK;
502
503 FREETMPS; LEAVE;
504 }
280 else 505 else
281 croak ("encountered object '%s', but no TO_CBOR method available on it", 506 croak ("encountered object '%s', but no TO_CBOR or FREEZE methods available on it",
282 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 507 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
283 }
284 } 508 }
285 else if (svt == SVt_PVHV) 509 else if (svt == SVt_PVHV)
286 encode_hv (enc, (HV *)sv); 510 encode_hv (enc, (HV *)sv);
287 else if (svt == SVt_PVAV) 511 else if (svt == SVt_PVAV)
288 encode_av (enc, (AV *)sv); 512 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 513 else
307 croak ("encountered %s, but CBOR can only represent references to arrays or hashes", 514 {
308 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 515 encode_tag (enc, CBOR_TAG_INDIRECTION);
516 encode_sv (enc, sv);
517 }
309} 518}
310 519
311static void 520static void
312encode_nv (enc_t *enc, SV *sv) 521encode_nv (enc_t *enc, SV *sv)
313{ 522{
314 double nv = SvNVX (sv); 523 double nv = SvNVX (sv);
315 524
316 need (enc, 9); 525 need (enc, 9);
317 526
318 if (ecb_expect_false (nv == (U32)nv)) 527 if (ecb_expect_false (nv == (NV)(U32)nv))
319 encode_uint (enc, 0x00, (U32)nv); 528 encode_uint (enc, MAJOR_POS_INT, (U32)nv);
320 //TODO: maybe I32? 529 //TODO: maybe I32?
321 else if (ecb_expect_false (nv == (float)nv)) 530 else if (ecb_expect_false (nv == (float)nv))
322 { 531 {
323 uint32_t fp = ecb_float_to_binary32 (nv); 532 uint32_t fp = ecb_float_to_binary32 (nv);
324 533
325 *enc->cur++ = 0xe0 | 26; 534 *enc->cur++ = MAJOR_MISC | MISC_FLOAT32;
326 535
327 if (!ecb_big_endian ()) 536 if (!ecb_big_endian ())
328 fp = ecb_bswap32 (fp); 537 fp = ecb_bswap32 (fp);
329 538
330 memcpy (enc->cur, &fp, 4); 539 memcpy (enc->cur, &fp, 4);
332 } 541 }
333 else 542 else
334 { 543 {
335 uint64_t fp = ecb_double_to_binary64 (nv); 544 uint64_t fp = ecb_double_to_binary64 (nv);
336 545
337 *enc->cur++ = 0xe0 | 27; 546 *enc->cur++ = MAJOR_MISC | MISC_FLOAT64;
338 547
339 if (!ecb_big_endian ()) 548 if (!ecb_big_endian ())
340 fp = ecb_bswap64 (fp); 549 fp = ecb_bswap64 (fp);
341 550
342 memcpy (enc->cur, &fp, 8); 551 memcpy (enc->cur, &fp, 8);
351 560
352 if (SvPOKp (sv)) 561 if (SvPOKp (sv))
353 { 562 {
354 STRLEN len; 563 STRLEN len;
355 char *str = SvPV (sv, len); 564 char *str = SvPV (sv, len);
356 encode_str (enc, SvUTF8 (sv), str, len); 565 encode_strref (enc, SvUTF8 (sv), str, len);
357 } 566 }
358 else if (SvNOKp (sv)) 567 else if (SvNOKp (sv))
359 encode_nv (enc, sv); 568 encode_nv (enc, sv);
360 else if (SvIOKp (sv)) 569 else if (SvIOKp (sv))
361 { 570 {
362 if (SvIsUV (sv)) 571 if (SvIsUV (sv))
363 encode_uint (enc, 0x00, SvUVX (sv)); 572 encode_uint (enc, MAJOR_POS_INT, SvUVX (sv));
364 else if (SvIVX (sv) >= 0) 573 else if (SvIVX (sv) >= 0)
365 encode_uint (enc, 0x00, SvIVX (sv)); 574 encode_uint (enc, MAJOR_POS_INT, SvIVX (sv));
366 else 575 else
367 encode_uint (enc, 0x20, -(SvIVX (sv) + 1)); 576 encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1));
368 } 577 }
369 else if (SvROK (sv)) 578 else if (SvROK (sv))
370 encode_rv (enc, SvRV (sv)); 579 encode_rv (enc, SvRV (sv));
371 else if (!SvOK (sv)) 580 else if (!SvOK (sv))
372 encode_ch (enc, 0xe0 | 22); 581 encode_ch (enc, MAJOR_MISC | SIMPLE_NULL);
373 else if (enc->cbor.flags & F_ALLOW_UNKNOWN) 582 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
374 encode_ch (enc, 0xe0 | 23); 583 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
375 else 584 else
376 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data", 585 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data",
377 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv)); 586 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
378} 587}
379 588
380static SV * 589static SV *
381encode_cbor (SV *scalar, CBOR *cbor) 590encode_cbor (SV *scalar, CBOR *cbor)
382{ 591{
383 enc_t enc; 592 enc_t enc = { 0 };
384 593
385 enc.cbor = *cbor; 594 enc.cbor = *cbor;
386 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 595 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
387 enc.cur = SvPVX (enc.sv); 596 enc.cur = SvPVX (enc.sv);
388 enc.end = SvEND (enc.sv); 597 enc.end = SvEND (enc.sv);
389 enc.depth = 0;
390 598
391 SvPOK_only (enc.sv); 599 SvPOK_only (enc.sv);
600
601 if (cbor->flags & F_PACK_STRINGS)
602 {
603 encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE);
604 enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ());
605 enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ());
606 }
607
392 encode_sv (&enc, scalar); 608 encode_sv (&enc, scalar);
393 609
394 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 610 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
395 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings 611 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
396 612
410 U8 *end; // end of input string 626 U8 *end; // end of input string
411 const char *err; // parse error, if != 0 627 const char *err; // parse error, if != 0
412 CBOR cbor; 628 CBOR cbor;
413 U32 depth; // recursion depth 629 U32 depth; // recursion depth
414 U32 maxdepth; // recursion depth limit 630 U32 maxdepth; // recursion depth limit
631 AV *shareable;
632 AV *stringref;
633 SV *decode_tagged;
415} dec_t; 634} dec_t;
416 635
417#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE 636#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE
418 637
419#define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data") 638#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 641#define DEC_DEC_DEPTH --dec->depth
423 642
424static UV 643static UV
425decode_uint (dec_t *dec) 644decode_uint (dec_t *dec)
426{ 645{
427 switch (*dec->cur & 31) 646 U8 m = *dec->cur & MINOR_MASK;
428 { 647 ++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;
433 648
434 case 24: 649 if (ecb_expect_true (m < LENGTH_EXT1))
650 return m;
651 else if (ecb_expect_true (m == LENGTH_EXT1))
652 {
435 WANT (2); 653 WANT (1);
436 dec->cur += 2; 654 dec->cur += 1;
437 return dec->cur[-1]; 655 return dec->cur[-1];
438 656 }
439 case 25: 657 else if (ecb_expect_true (m == LENGTH_EXT2))
658 {
440 WANT (3); 659 WANT (2);
441 dec->cur += 3; 660 dec->cur += 2;
442 return (((UV)dec->cur[-2]) << 8) 661 return (((UV)dec->cur[-2]) << 8)
443 | ((UV)dec->cur[-1]); 662 | ((UV)dec->cur[-1]);
444 663 }
445 case 26: 664 else if (ecb_expect_true (m == LENGTH_EXT4))
665 {
446 WANT (5); 666 WANT (4);
447 dec->cur += 5; 667 dec->cur += 4;
448 return (((UV)dec->cur[-4]) << 24) 668 return (((UV)dec->cur[-4]) << 24)
449 | (((UV)dec->cur[-3]) << 16) 669 | (((UV)dec->cur[-3]) << 16)
450 | (((UV)dec->cur[-2]) << 8) 670 | (((UV)dec->cur[-2]) << 8)
451 | ((UV)dec->cur[-1]); 671 | ((UV)dec->cur[-1]);
452 672 }
453 case 27: 673 else if (ecb_expect_true (m == LENGTH_EXT8))
674 {
454 WANT (9); 675 WANT (8);
455 dec->cur += 9; 676 dec->cur += 8;
677
678 return
679#if UVSIZE < 8
680 0
681#else
456 return (((UV)dec->cur[-8]) << 56) 682 (((UV)dec->cur[-8]) << 56)
457 | (((UV)dec->cur[-7]) << 48) 683 | (((UV)dec->cur[-7]) << 48)
458 | (((UV)dec->cur[-6]) << 40) 684 | (((UV)dec->cur[-6]) << 40)
459 | (((UV)dec->cur[-5]) << 32) 685 | (((UV)dec->cur[-5]) << 32)
686#endif
460 | (((UV)dec->cur[-4]) << 24) 687 | (((UV)dec->cur[-4]) << 24)
461 | (((UV)dec->cur[-3]) << 16) 688 | (((UV)dec->cur[-3]) << 16)
462 | (((UV)dec->cur[-2]) << 8) 689 | (((UV)dec->cur[-2]) << 8)
463 | ((UV)dec->cur[-1]); 690 | ((UV)dec->cur[-1]);
464 691 }
465 default: 692 else
466 ERR ("corrupted CBOR data (unsupported integer minor encoding)"); 693 ERR ("corrupted CBOR data (unsupported integer minor encoding)");
467 }
468 694
469fail: 695fail:
470 return 0; 696 return 0;
471} 697}
472 698
477{ 703{
478 AV *av = newAV (); 704 AV *av = newAV ();
479 705
480 DEC_INC_DEPTH; 706 DEC_INC_DEPTH;
481 707
482 if ((*dec->cur & 31) == 31) 708 if (*dec->cur == (MAJOR_ARRAY | MINOR_INDEF))
483 { 709 {
484 ++dec->cur; 710 ++dec->cur;
485 711
486 for (;;) 712 for (;;)
487 { 713 {
488 WANT (1); 714 WANT (1);
489 715
490 if (*dec->cur == (0xe0 | 31)) 716 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
491 { 717 {
492 ++dec->cur; 718 ++dec->cur;
493 break; 719 break;
494 } 720 }
495 721
498 } 724 }
499 else 725 else
500 { 726 {
501 int i, len = decode_uint (dec); 727 int i, len = decode_uint (dec);
502 728
729 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); 730 av_fill (av, len - 1);
504 731
505 for (i = 0; i < len; ++i) 732 for (i = 0; i < len; ++i)
506 AvARRAY (av)[i] = decode_sv (dec); 733 AvARRAY (av)[i] = decode_sv (dec);
507 } 734 }
513 SvREFCNT_dec (av); 740 SvREFCNT_dec (av);
514 DEC_DEC_DEPTH; 741 DEC_DEC_DEPTH;
515 return &PL_sv_undef; 742 return &PL_sv_undef;
516} 743}
517 744
745static void
746decode_he (dec_t *dec, HV *hv)
747{
748 // for speed reasons, we specialcase single-string
749 // byte or utf-8 strings as keys, but only when !stringref
750
751 if (ecb_expect_true (!dec->stringref))
752 if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8))
753 {
754 I32 len = decode_uint (dec);
755 char *key = (char *)dec->cur;
756
757 WANT (len);
758 dec->cur += len;
759
760 hv_store (hv, key, len, decode_sv (dec), 0);
761
762 return;
763 }
764 else if (ecb_expect_true ((U8)(*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8))
765 {
766 I32 len = decode_uint (dec);
767 char *key = (char *)dec->cur;
768
769 WANT (len);
770 dec->cur += len;
771
772 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
773 if (!is_utf8_string (key, len))
774 ERR ("corrupted CBOR data (invalid UTF-8 in map key)");
775
776 hv_store (hv, key, -len, decode_sv (dec), 0);
777
778 return;
779 }
780
781 SV *k = decode_sv (dec);
782 SV *v = decode_sv (dec);
783
784 hv_store_ent (hv, k, v, 0);
785 SvREFCNT_dec (k);
786
787fail:
788 ;
789}
790
518static SV * 791static SV *
519decode_hv (dec_t *dec) 792decode_hv (dec_t *dec)
520{ 793{
521 HV *hv = newHV (); 794 HV *hv = newHV ();
522 795
523 DEC_INC_DEPTH; 796 DEC_INC_DEPTH;
524 797
525 if ((*dec->cur & 31) == 31) 798 if (*dec->cur == (MAJOR_MAP | MINOR_INDEF))
526 { 799 {
527 ++dec->cur; 800 ++dec->cur;
528 801
529 for (;;) 802 for (;;)
530 { 803 {
531 WANT (1); 804 WANT (1);
532 805
533 if (*dec->cur == (0xe0 | 31)) 806 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
534 { 807 {
535 ++dec->cur; 808 ++dec->cur;
536 break; 809 break;
537 } 810 }
538 811
539 SV *k = decode_sv (dec); 812 decode_he (dec, hv);
540 SV *v = decode_sv (dec);
541
542 hv_store_ent (hv, k, v, 0);
543 } 813 }
544 } 814 }
545 else 815 else
546 { 816 {
547 int len = decode_uint (dec); 817 int pairs = decode_uint (dec);
548 818
549 while (len--) 819 while (pairs--)
550 { 820 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 } 821 }
557 822
558 DEC_DEC_DEPTH; 823 DEC_DEC_DEPTH;
559 return newRV_noinc ((SV *)hv); 824 return newRV_noinc ((SV *)hv);
560 825
567static SV * 832static SV *
568decode_str (dec_t *dec, int utf8) 833decode_str (dec_t *dec, int utf8)
569{ 834{
570 SV *sv = 0; 835 SV *sv = 0;
571 836
572 if ((*dec->cur & 31) == 31) 837 if ((*dec->cur & MINOR_MASK) == MINOR_INDEF)
573 { 838 {
839 // indefinite length strings
574 ++dec->cur; 840 ++dec->cur;
575 841
842 U8 major = *dec->cur & MAJOR_MISC;
843
576 sv = newSVpvn ("", 0); 844 sv = newSVpvn ("", 0);
577 845
578 // not very fast, and certainly not robust against illegal input
579 for (;;) 846 for (;;)
580 { 847 {
581 WANT (1); 848 WANT (1);
582 849
583 if (*dec->cur == (0xe0 | 31)) 850 if ((*dec->cur - major) > LENGTH_EXT8)
851 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
584 { 852 {
585 ++dec->cur; 853 ++dec->cur;
586 break; 854 break;
587 } 855 }
856 else
857 ERR ("corrupted CBOR data (invalid chunks in indefinite length string)");
588 858
589 sv_catsv (sv, decode_sv (dec)); 859 STRLEN len = decode_uint (dec);
860
861 WANT (len);
862 sv_catpvn (sv, dec->cur, len);
863 dec->cur += len;
590 } 864 }
591 } 865 }
592 else 866 else
593 { 867 {
594 STRLEN len = decode_uint (dec); 868 STRLEN len = decode_uint (dec);
595 869
596 WANT (len); 870 WANT (len);
597 sv = newSVpvn (dec->cur, len); 871 sv = newSVpvn (dec->cur, len);
598 dec->cur += len; 872 dec->cur += len;
873
874 if (ecb_expect_false (dec->stringref)
875 && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1))
876 av_push (dec->stringref, SvREFCNT_inc_NN (sv));
599 } 877 }
600 878
601 if (utf8) 879 if (utf8)
880 {
881 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
882 if (!is_utf8_string (SvPVX (sv), SvCUR (sv)))
883 ERR ("corrupted CBOR data (invalid UTF-8 in text string)");
884
602 SvUTF8_on (sv); 885 SvUTF8_on (sv);
886 }
603 887
604 return sv; 888 return sv;
605 889
606fail: 890fail:
607 SvREFCNT_dec (sv); 891 SvREFCNT_dec (sv);
609} 893}
610 894
611static SV * 895static SV *
612decode_tagged (dec_t *dec) 896decode_tagged (dec_t *dec)
613{ 897{
898 SV *sv = 0;
614 UV tag = decode_uint (dec); 899 UV tag = decode_uint (dec);
900
901 WANT (1);
902
903 switch (tag)
904 {
905 case CBOR_TAG_MAGIC:
615 SV *sv = decode_sv (dec); 906 sv = decode_sv (dec);
907 break;
616 908
617 if (tag == 55799) // 2.4.5 Self-Describe CBOR 909 case CBOR_TAG_INDIRECTION:
910 sv = newRV_noinc (decode_sv (dec));
911 break;
912
913 case CBOR_TAG_STRINGREF_NAMESPACE:
914 {
915 ENTER; SAVETMPS;
916
917 SAVESPTR (dec->stringref);
918 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ());
919
920 sv = decode_sv (dec);
921
922 FREETMPS; LEAVE;
923 }
924 break;
925
926 case CBOR_TAG_STRINGREF:
927 {
928 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
929 ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
930
931 UV idx = decode_uint (dec);
932
933 if (!dec->stringref || (int)idx > AvFILLp (dec->stringref))
934 ERR ("corrupted CBOR data (stringref index out of bounds or outside namespace)");
935
936 sv = newSVsv (AvARRAY (dec->stringref)[idx]);
937 }
938 break;
939
940 case CBOR_TAG_VALUE_SHAREABLE:
941 {
942 if (ecb_expect_false (!dec->shareable))
943 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ());
944
945 if (dec->cbor.flags & F_ALLOW_CYCLES)
946 {
947 sv = newSV (0);
948 av_push (dec->shareable, SvREFCNT_inc_NN (sv));
949
950 SV *osv = decode_sv (dec);
951 sv_setsv (sv, osv);
952 SvREFCNT_dec_NN (osv);
953 }
954 else
955 {
956 av_push (dec->shareable, &PL_sv_undef);
957 int idx = AvFILLp (dec->shareable);
958 sv = decode_sv (dec);
959 av_store (dec->shareable, idx, SvREFCNT_inc_NN (sv));
960 }
961 }
962 break;
963
964 case CBOR_TAG_VALUE_SHAREDREF:
965 {
966 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
967 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
968
969 UV idx = decode_uint (dec);
970
971 if (!dec->shareable || (int)idx > AvFILLp (dec->shareable))
972 ERR ("corrupted CBOR data (sharedref index out of bounds)");
973
974 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]);
975
976 if (sv == &PL_sv_undef)
977 ERR ("cyclic CBOR data structure found, but allow_cycles is not enabled");
978 }
979 break;
980
981 case CBOR_TAG_PERL_OBJECT:
982 {
983 sv = decode_sv (dec);
984
985 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
986 ERR ("corrupted CBOR data (non-array perl object)");
987
988 AV *av = (AV *)SvRV (sv);
989 int len = av_len (av) + 1;
990 HV *stash = gv_stashsv (*av_fetch (av, 0, 1), 0);
991
992 if (!stash)
993 ERR ("cannot decode perl-object (package does not exist)");
994
995 GV *method = gv_fetchmethod_autoload (stash, "THAW", 0);
996
997 if (!method)
998 ERR ("cannot decode perl-object (package does not have a THAW method)");
999
1000 dSP;
1001
1002 ENTER; SAVETMPS;
1003 PUSHMARK (SP);
1004 EXTEND (SP, len + 1);
1005 // we re-bless the reference to get overload and other niceties right
1006 PUSHs (*av_fetch (av, 0, 1));
1007 PUSHs (sv_cbor);
1008
1009 int i;
1010
1011 for (i = 1; i < len; ++i)
1012 PUSHs (*av_fetch (av, i, 1));
1013
1014 PUTBACK;
1015 call_sv ((SV *)GvCV (method), G_SCALAR | G_EVAL);
1016 SPAGAIN;
1017
1018 if (SvTRUE (ERRSV))
1019 {
1020 FREETMPS; LEAVE;
1021 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV))));
1022 }
1023
1024 SvREFCNT_dec (sv);
1025 sv = SvREFCNT_inc (POPs);
1026
1027 PUTBACK;
1028
1029 FREETMPS; LEAVE;
1030 }
1031 break;
1032
1033 default:
1034 {
1035 sv = decode_sv (dec);
1036
1037 dSP;
1038 ENTER; SAVETMPS;
1039 SAVESTACK_POS ();
1040 PUSHMARK (SP);
1041 EXTEND (SP, 2);
1042 PUSHs (newSVuv (tag));
1043 PUSHs (sv);
1044
1045 PUTBACK;
1046 int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL);
1047 SPAGAIN;
1048
1049 if (SvTRUE (ERRSV))
1050 {
1051 FREETMPS; LEAVE;
1052 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV))));
1053 }
1054
1055 if (count)
1056 {
1057 SvREFCNT_dec (sv);
1058 sv = SvREFCNT_inc (POPs);
1059 }
1060 else
1061 {
1062 AV *av = newAV ();
1063 av_push (av, newSVuv (tag));
1064 av_push (av, sv);
1065
1066 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
1067 ? cbor_tagged_stash
1068 : gv_stashpv ("CBOR::XS::Tagged" , 1);
1069 sv = sv_bless (newRV_noinc ((SV *)av), tagged_stash);
1070 }
1071
1072 PUTBACK;
1073
1074 FREETMPS; LEAVE;
1075 }
1076 break;
1077 }
1078
618 return sv; 1079 return sv;
619 1080
620 AV *av = newAV (); 1081fail:
621 av_push (av, newSVuv (tag)); 1082 SvREFCNT_dec (sv);
622 av_push (av, sv); 1083 return &PL_sv_undef;
623 return newRV_noinc ((SV *)av);
624} 1084}
625 1085
626static SV * 1086static SV *
627decode_sv (dec_t *dec) 1087decode_sv (dec_t *dec)
628{ 1088{
629 WANT (1); 1089 WANT (1);
630 1090
631 switch (*dec->cur >> 5) 1091 switch (*dec->cur >> MAJOR_SHIFT)
632 { 1092 {
633 case 0: // unsigned int 1093 case MAJOR_POS_INT >> MAJOR_SHIFT: return newSVuv (decode_uint (dec));
634 return newSVuv (decode_uint (dec)); 1094 case MAJOR_NEG_INT >> MAJOR_SHIFT: return newSViv (-1 - (IV)decode_uint (dec));
635 case 1: // negative int 1095 case MAJOR_BYTES >> MAJOR_SHIFT: return decode_str (dec, 0);
636 return newSViv (-1 - (IV)decode_uint (dec)); 1096 case MAJOR_TEXT >> MAJOR_SHIFT: return decode_str (dec, 1);
637 case 2: // octet string 1097 case MAJOR_ARRAY >> MAJOR_SHIFT: return decode_av (dec);
638 return decode_str (dec, 0); 1098 case MAJOR_MAP >> MAJOR_SHIFT: return decode_hv (dec);
639 case 3: // utf-8 string 1099 case MAJOR_TAG >> MAJOR_SHIFT: return decode_tagged (dec);
640 return decode_str (dec, 1); 1100
641 case 4: // array 1101 case MAJOR_MISC >> MAJOR_SHIFT:
642 return decode_av (dec);
643 case 5: // map
644 return decode_hv (dec);
645 case 6: // tag
646 return decode_tagged (dec);
647 case 7: // misc
648 switch (*dec->cur++ & 31) 1102 switch (*dec->cur++ & MINOR_MASK)
649 { 1103 {
650 case 20: 1104 case SIMPLE_FALSE:
651#if CBOR_SLOW 1105#if CBOR_SLOW
652 cbor_false = get_bool ("CBOR::XS::false"); 1106 types_false = get_bool ("Types::Serialiser::false");
653#endif 1107#endif
654 return newSVsv (cbor_false); 1108 return newSVsv (types_false);
655 case 21: 1109 case SIMPLE_TRUE:
656#if CBOR_SLOW 1110#if CBOR_SLOW
657 cbor_true = get_bool ("CBOR::XS::true"); 1111 types_true = get_bool ("Types::Serialiser::true");
658#endif 1112#endif
659 return newSVsv (cbor_true); 1113 return newSVsv (types_true);
660 case 22: 1114 case SIMPLE_NULL:
661 return newSVsv (&PL_sv_undef); 1115 return newSVsv (&PL_sv_undef);
1116 case SIMPLE_UNDEF:
1117#if CBOR_SLOW
1118 types_error = get_bool ("Types::Serialiser::error");
1119#endif
1120 return newSVsv (types_error);
662 1121
663 case 25: 1122 case MISC_FLOAT16:
664 { 1123 {
665 WANT (2); 1124 WANT (2);
666 1125
667 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1]; 1126 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1];
668 dec->cur += 2; 1127 dec->cur += 2;
669 1128
670 return newSVnv (ecb_binary16_to_float (fp)); 1129 return newSVnv (ecb_binary16_to_float (fp));
671 } 1130 }
672 1131
673 case 26: 1132 case MISC_FLOAT32:
674 { 1133 {
675 uint32_t fp; 1134 uint32_t fp;
676 WANT (4); 1135 WANT (4);
677 memcpy (&fp, dec->cur, 4); 1136 memcpy (&fp, dec->cur, 4);
678 dec->cur += 4; 1137 dec->cur += 4;
681 fp = ecb_bswap32 (fp); 1140 fp = ecb_bswap32 (fp);
682 1141
683 return newSVnv (ecb_binary32_to_float (fp)); 1142 return newSVnv (ecb_binary32_to_float (fp));
684 } 1143 }
685 1144
686 case 27: 1145 case MISC_FLOAT64:
687 { 1146 {
688 uint64_t fp; 1147 uint64_t fp;
689 WANT (8); 1148 WANT (8);
690 memcpy (&fp, dec->cur, 8); 1149 memcpy (&fp, dec->cur, 8);
691 dec->cur += 8; 1150 dec->cur += 8;
694 fp = ecb_bswap64 (fp); 1153 fp = ecb_bswap64 (fp);
695 1154
696 return newSVnv (ecb_binary64_to_double (fp)); 1155 return newSVnv (ecb_binary64_to_double (fp));
697 } 1156 }
698 1157
699 // 0..19 unassigned 1158 // 0..19 unassigned simple
700 // 24 reserved + unassigned (reserved values are not encodable) 1159 // 24 reserved + unassigned simple (reserved values are not encodable)
1160 // 28-30 unassigned misc
1161 // 31 break code
701 default: 1162 default:
702 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)"); 1163 ERR ("corrupted CBOR data (reserved/unassigned/unexpected major 7 value)");
703 } 1164 }
704 1165
705 break; 1166 break;
706 } 1167 }
707 1168
710} 1171}
711 1172
712static SV * 1173static SV *
713decode_cbor (SV *string, CBOR *cbor, char **offset_return) 1174decode_cbor (SV *string, CBOR *cbor, char **offset_return)
714{ 1175{
715 dec_t dec; 1176 dec_t dec = { 0 };
716 SV *sv; 1177 SV *sv;
1178 STRLEN len;
1179 char *data = SvPVbyte (string, len);
717 1180
718 /* work around bugs in 5.10 where manipulating magic values
719 * makes perl ignore the magic in subsequent accesses.
720 * also make a copy of non-PV values, to get them into a clean
721 * state (SvPV should do that, but it's buggy, see below).
722 */
723 /*SvGETMAGIC (string);*/
724 if (SvMAGICAL (string) || !SvPOK (string))
725 string = sv_2mortal (newSVsv (string));
726
727 SvUPGRADE (string, SVt_PV);
728
729 /* work around a bug in perl 5.10, which causes SvCUR to fail an
730 * assertion with -DDEBUGGING, although SvCUR is documented to
731 * return the xpv_cur field which certainly exists after upgrading.
732 * according to nicholas clark, calling SvPOK fixes this.
733 * But it doesn't fix it, so try another workaround, call SvPV_nolen
734 * and hope for the best.
735 * Damnit, SvPV_nolen still trips over yet another assertion. This
736 * assertion business is seriously broken, try yet another workaround
737 * for the broken -DDEBUGGING.
738 */
739 {
740#ifdef DEBUGGING
741 STRLEN offset = SvOK (string) ? sv_len (string) : 0;
742#else
743 STRLEN offset = SvCUR (string);
744#endif
745
746 if (offset > cbor->max_size && cbor->max_size) 1181 if (len > cbor->max_size && cbor->max_size)
747 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu", 1182 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
748 (unsigned long)SvCUR (string), (unsigned long)cbor->max_size); 1183 (unsigned long)len, (unsigned long)cbor->max_size);
749 }
750
751 sv_utf8_downgrade (string, 0);
752 1184
753 dec.cbor = *cbor; 1185 dec.cbor = *cbor;
754 dec.cur = (U8 *)SvPVX (string); 1186 dec.cur = (U8 *)data;
755 dec.end = (U8 *)SvEND (string); 1187 dec.end = (U8 *)data + len;
756 dec.err = 0;
757 dec.depth = 0;
758 1188
759 sv = decode_sv (&dec); 1189 sv = decode_sv (&dec);
760 1190
761 if (offset_return) 1191 if (offset_return)
762 *offset_return = dec.cur; 1192 *offset_return = dec.cur;
765 if (dec.cur != dec.end && !dec.err) 1195 if (dec.cur != dec.end && !dec.err)
766 dec.err = "garbage after CBOR object"; 1196 dec.err = "garbage after CBOR object";
767 1197
768 if (dec.err) 1198 if (dec.err)
769 { 1199 {
1200 if (dec.shareable)
1201 {
1202 // need to break cyclic links, which whould all be in shareable
1203 int i;
1204 SV **svp;
1205
1206 for (i = av_len (dec.shareable) + 1; i--; )
1207 if ((svp = av_fetch (dec.shareable, i, 0)))
1208 sv_setsv (*svp, &PL_sv_undef);
1209 }
1210
770 SvREFCNT_dec (sv); 1211 SvREFCNT_dec (sv);
771 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)SvPVX (string), (int)(uint8_t)*dec.cur); 1212 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur);
772 } 1213 }
773 1214
774 sv = sv_2mortal (sv); 1215 sv = sv_2mortal (sv);
775 1216
776 return sv; 1217 return sv;
777} 1218}
778 1219
1220/////////////////////////////////////////////////////////////////////////////
1221// incremental parser
1222
1223#define INCR_DONE(cbor) (AvFILLp (cbor->incr_count) < 0)
1224
1225// returns 0 for notyet, 1 for success or error
1226static int
1227incr_parse (CBOR *self, SV *cborstr)
1228{
1229 STRLEN cur;
1230 SvPV (cborstr, cur);
1231
1232 while (ecb_expect_true (self->incr_need <= cur))
1233 {
1234 // table of integer count bytes
1235 static I8 incr_len[MINOR_MASK + 1] = {
1236 0, 0, 0, 0, 0, 0, 0, 0,
1237 0, 0, 0, 0, 0, 0, 0, 0,
1238 0, 0, 0, 0, 0, 0, 0, 0,
1239 1, 2, 4, 8,-1,-1,-1,-2
1240 };
1241
1242 const U8 *p = SvPVX (cborstr) + self->incr_pos;
1243 U8 m = *p & MINOR_MASK;
1244 IV count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1245 I8 ilen = incr_len[m];
1246
1247 self->incr_need = self->incr_pos + 1;
1248
1249 if (ecb_expect_false (ilen < 0))
1250 {
1251 if (m != MINOR_INDEF)
1252 return 1; // error
1253
1254 if (*p == (MAJOR_MISC | MINOR_INDEF))
1255 {
1256 if (count >= 0)
1257 return 1; // error
1258
1259 count = 1;
1260 }
1261 else
1262 {
1263 av_push (self->incr_count, newSViv (-1)); //TODO: nest
1264 count = -1;
1265 }
1266 }
1267 else
1268 {
1269 self->incr_need += ilen;
1270 if (ecb_expect_false (self->incr_need > cur))
1271 return 0;
1272
1273 int major = *p >> MAJOR_SHIFT;
1274
1275 switch (major)
1276 {
1277 case MAJOR_TAG >> MAJOR_SHIFT:
1278 ++count; // tags merely prefix another value
1279 break;
1280
1281 case MAJOR_BYTES >> MAJOR_SHIFT:
1282 case MAJOR_TEXT >> MAJOR_SHIFT:
1283 case MAJOR_ARRAY >> MAJOR_SHIFT:
1284 case MAJOR_MAP >> MAJOR_SHIFT:
1285 {
1286 UV len;
1287
1288 if (ecb_expect_false (ilen))
1289 {
1290 len = 0;
1291
1292 do {
1293 len = (len << 8) | *++p;
1294 } while (--ilen);
1295 }
1296 else
1297 len = m;
1298
1299 switch (major)
1300 {
1301 case MAJOR_BYTES >> MAJOR_SHIFT:
1302 case MAJOR_TEXT >> MAJOR_SHIFT:
1303 self->incr_need += len;
1304 if (ecb_expect_false (self->incr_need > cur))
1305 return 0;
1306
1307 break;
1308
1309 case MAJOR_MAP >> MAJOR_SHIFT:
1310 len <<= 1;
1311 case MAJOR_ARRAY >> MAJOR_SHIFT:
1312 if (len)
1313 {
1314 av_push (self->incr_count, newSViv (len + 1)); //TODO: nest
1315 count = len + 1;
1316 }
1317 break;
1318 }
1319 }
1320 }
1321 }
1322
1323 self->incr_pos = self->incr_need;
1324
1325 if (count > 0)
1326 {
1327 while (!--count)
1328 {
1329 if (!AvFILLp (self->incr_count))
1330 return 1; // done
1331
1332 SvREFCNT_dec_NN (av_pop (self->incr_count));
1333 count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1334 }
1335
1336 SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]) = count;
1337 }
1338 }
1339
1340 return 0;
1341}
1342
1343
779///////////////////////////////////////////////////////////////////////////// 1344/////////////////////////////////////////////////////////////////////////////
780// XS interface functions 1345// XS interface functions
781 1346
782MODULE = CBOR::XS PACKAGE = CBOR::XS 1347MODULE = CBOR::XS PACKAGE = CBOR::XS
783 1348
784BOOT: 1349BOOT:
785{ 1350{
786 cbor_stash = gv_stashpv ("CBOR::XS" , 1); 1351 cbor_stash = gv_stashpv ("CBOR::XS" , 1);
787 cbor_boolean_stash = gv_stashpv ("CBOR::XS::Boolean", 1);
788 cbor_tagged_stash = gv_stashpv ("CBOR::XS::Tagged" , 1); 1352 cbor_tagged_stash = gv_stashpv ("CBOR::XS::Tagged" , 1);
789 1353
790 cbor_true = get_bool ("CBOR::XS::true"); 1354 types_boolean_stash = gv_stashpv ("Types::Serialiser::Boolean", 1);
791 cbor_false = get_bool ("CBOR::XS::false"); 1355 types_error_stash = gv_stashpv ("Types::Serialiser::Error" , 1);
1356
1357 types_true = get_bool ("Types::Serialiser::true" );
1358 types_false = get_bool ("Types::Serialiser::false");
1359 types_error = get_bool ("Types::Serialiser::error");
1360
1361 default_filter = newSVpv ("CBOR::XS::default_filter", 0);
1362
1363 sv_cbor = newSVpv ("CBOR", 0);
1364 SvREADONLY_on (sv_cbor);
792} 1365}
793 1366
794PROTOTYPES: DISABLE 1367PROTOTYPES: DISABLE
795 1368
796void CLONE (...) 1369void CLONE (...)
797 CODE: 1370 CODE:
798 cbor_stash = 0; 1371 cbor_stash = 0;
799 cbor_boolean_stash = 0;
800 cbor_tagged_stash = 0; 1372 cbor_tagged_stash = 0;
1373 types_error_stash = 0;
1374 types_boolean_stash = 0;
801 1375
802void new (char *klass) 1376void new (char *klass)
803 PPCODE: 1377 PPCODE:
804{ 1378{
805 SV *pv = NEWSV (0, sizeof (CBOR)); 1379 SV *pv = NEWSV (0, sizeof (CBOR));
813 1387
814void shrink (CBOR *self, int enable = 1) 1388void shrink (CBOR *self, int enable = 1)
815 ALIAS: 1389 ALIAS:
816 shrink = F_SHRINK 1390 shrink = F_SHRINK
817 allow_unknown = F_ALLOW_UNKNOWN 1391 allow_unknown = F_ALLOW_UNKNOWN
1392 allow_sharing = F_ALLOW_SHARING
1393 allow_cycles = F_ALLOW_CYCLES
1394 pack_strings = F_PACK_STRINGS
1395 validate_utf8 = F_VALIDATE_UTF8
818 PPCODE: 1396 PPCODE:
819{ 1397{
820 if (enable) 1398 if (enable)
821 self->flags |= ix; 1399 self->flags |= ix;
822 else 1400 else
827 1405
828void get_shrink (CBOR *self) 1406void get_shrink (CBOR *self)
829 ALIAS: 1407 ALIAS:
830 get_shrink = F_SHRINK 1408 get_shrink = F_SHRINK
831 get_allow_unknown = F_ALLOW_UNKNOWN 1409 get_allow_unknown = F_ALLOW_UNKNOWN
1410 get_allow_sharing = F_ALLOW_SHARING
1411 get_allow_cycles = F_ALLOW_CYCLES
1412 get_pack_strings = F_PACK_STRINGS
1413 get_validate_utf8 = F_VALIDATE_UTF8
832 PPCODE: 1414 PPCODE:
833 XPUSHs (boolSV (self->flags & ix)); 1415 XPUSHs (boolSV (self->flags & ix));
834 1416
835void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) 1417void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
836 PPCODE: 1418 PPCODE:
852 CODE: 1434 CODE:
853 RETVAL = self->max_size; 1435 RETVAL = self->max_size;
854 OUTPUT: 1436 OUTPUT:
855 RETVAL 1437 RETVAL
856 1438
857#if 0 //TODO 1439void filter (CBOR *self, SV *filter = 0)
858
859void filter_cbor_object (CBOR *self, SV *cb = &PL_sv_undef)
860 PPCODE: 1440 PPCODE:
861{
862 SvREFCNT_dec (self->cb_object); 1441 SvREFCNT_dec (self->filter);
863 self->cb_object = SvOK (cb) ? newSVsv (cb) : 0; 1442 self->filter = filter ? newSVsv (filter) : filter;
864
865 XPUSHs (ST (0)); 1443 XPUSHs (ST (0));
866}
867 1444
868void filter_cbor_single_key_object (CBOR *self, SV *key, SV *cb = &PL_sv_undef) 1445SV *get_filter (CBOR *self)
869 PPCODE: 1446 CODE:
870{ 1447 RETVAL = self->filter ? self->filter : NEWSV (0, 0);
871 if (!self->cb_sk_object) 1448 OUTPUT:
872 self->cb_sk_object = newHV (); 1449 RETVAL
873
874 if (SvOK (cb))
875 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);
876 else
877 {
878 hv_delete_ent (self->cb_sk_object, key, G_DISCARD, 0);
879
880 if (!HvKEYS (self->cb_sk_object))
881 {
882 SvREFCNT_dec (self->cb_sk_object);
883 self->cb_sk_object = 0;
884 }
885 }
886
887 XPUSHs (ST (0));
888}
889
890#endif
891 1450
892void encode (CBOR *self, SV *scalar) 1451void encode (CBOR *self, SV *scalar)
893 PPCODE: 1452 PPCODE:
894 PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN; 1453 PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN;
895 XPUSHs (scalar); 1454 XPUSHs (scalar);
908 EXTEND (SP, 2); 1467 EXTEND (SP, 2);
909 PUSHs (sv); 1468 PUSHs (sv);
910 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); 1469 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
911} 1470}
912 1471
913#if 0 1472void incr_parse (CBOR *self, SV *cborstr)
1473 ALIAS:
1474 incr_parse_multiple = 1
1475 PPCODE:
1476{
1477 if (SvUTF8 (cborstr))
1478 sv_utf8_downgrade (cborstr, 0);
1479
1480 if (!self->incr_count)
1481 {
1482 self->incr_count = newAV ();
1483 self->incr_pos = 0;
1484 self->incr_need = 1;
1485
1486 av_push (self->incr_count, newSViv (1));
1487 }
1488
1489 do
1490 {
1491 if (!incr_parse (self, cborstr))
1492 {
1493 if (self->incr_need > self->max_size && self->max_size)
1494 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
1495 (unsigned long)self->incr_need, (unsigned long)self->max_size);
1496
1497 break;
1498 }
1499
1500 SV *sv;
1501 char *offset;
1502
1503 PUTBACK; sv = decode_cbor (cborstr, self, &offset); SPAGAIN;
1504 XPUSHs (sv);
1505
1506 sv_chop (cborstr, offset);
1507
1508 av_clear (self->incr_count);
1509 av_push (self->incr_count, newSViv (1));
1510
1511 self->incr_pos = 0;
1512 self->incr_need = self->incr_pos + 1;
1513 }
1514 while (ix);
1515}
1516
1517void incr_reset (CBOR *self)
1518 CODE:
1519{
1520 SvREFCNT_dec (self->incr_count);
1521 self->incr_count = 0;
1522}
914 1523
915void DESTROY (CBOR *self) 1524void DESTROY (CBOR *self)
916 CODE: 1525 PPCODE:
917 SvREFCNT_dec (self->cb_sk_object); 1526 cbor_free (self);
918 SvREFCNT_dec (self->cb_object);
919
920#endif
921 1527
922PROTOTYPES: ENABLE 1528PROTOTYPES: ENABLE
923 1529
924void encode_cbor (SV *scalar) 1530void encode_cbor (SV *scalar)
1531 ALIAS:
1532 encode_cbor = 0
1533 encode_cbor_sharing = F_ALLOW_SHARING
925 PPCODE: 1534 PPCODE:
926{ 1535{
927 CBOR cbor; 1536 CBOR cbor;
928 cbor_init (&cbor); 1537 cbor_init (&cbor);
1538 cbor.flags |= ix;
929 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN; 1539 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;
930 XPUSHs (scalar); 1540 XPUSHs (scalar);
931} 1541}
932 1542
933void decode_cbor (SV *cborstr) 1543void decode_cbor (SV *cborstr)

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines