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 1.24 by root, Wed Nov 20 16:29:02 2013 UTC vs.
Revision 1.45 by root, Sun Dec 14 05:48:39 2014 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// compatibility with perl <5.18 15// compatibility with perl <5.18
15#ifndef HvNAMELEN_get 16#ifndef HvNAMELEN_get
16# define HvNAMELEN_get(hv) strlen (HvNAME (hv)) 17# define HvNAMELEN_get(hv) strlen (HvNAME (hv))
19# define HvNAMELEN(hv) HvNAMELEN_get (hv) 20# define HvNAMELEN(hv) HvNAMELEN_get (hv)
20#endif 21#endif
21#ifndef HvNAMEUTF8 22#ifndef HvNAMEUTF8
22# define HvNAMEUTF8(hv) 0 23# define HvNAMEUTF8(hv) 0
23#endif 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};
24 65
25// known tags 66// known tags
26enum cbor_tag 67enum cbor_tag
27{ 68{
28 // inofficial extensions (pending iana registration) 69 // extensions
70 CBOR_TAG_STRINGREF = 25, // http://cbor.schmorp.de/stringref
29 CBOR_TAG_PERL_OBJECT = 24, // http://cbor.schmorp.de/perl-object 71 CBOR_TAG_PERL_OBJECT = 26, // http://cbor.schmorp.de/perl-object
30 CBOR_TAG_GENERIC_OBJECT = 25, // http://cbor.schmorp.de/generic-object 72 CBOR_TAG_GENERIC_OBJECT = 27, // http://cbor.schmorp.de/generic-object
31 CBOR_TAG_VALUE_SHAREABLE = 26, // http://cbor.schmorp.de/value-sharing 73 CBOR_TAG_VALUE_SHAREABLE = 28, // http://cbor.schmorp.de/value-sharing
32 CBOR_TAG_VALUE_SHAREDREF = 27, // http://cbor.schmorp.de/value-sharing 74 CBOR_TAG_VALUE_SHAREDREF = 29, // http://cbor.schmorp.de/value-sharing
33 CBOR_TAG_STRINGREF_NAMESPACE = 65537, // http://cbor.schmorp.de/stringref 75 CBOR_TAG_STRINGREF_NAMESPACE = 256, // http://cbor.schmorp.de/stringref
34 CBOR_TAG_STRINGREF = 28, // http://cbor.schmorp.de/stringref
35 CBOR_TAG_INDIRECTION = 22098, // http://cbor.schmorp.de/indirection 76 CBOR_TAG_INDIRECTION = 22098, // http://cbor.schmorp.de/indirection
36 77
37 // rfc7049 78 // rfc7049
38 CBOR_TAG_DATETIME = 0, // rfc4287, utf-8 79 CBOR_TAG_DATETIME = 0, // rfc4287, utf-8
39 CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any 80 CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any
40 CBOR_TAG_POS_BIGNUM = 2, // byte string 81 CBOR_TAG_POS_BIGNUM = 2, // byte string
41 CBOR_TAG_NEG_BIGNUM = 3, // byte string 82 CBOR_TAG_NEG_BIGNUM = 3, // byte string
42 CBOR_TAG_DECIMAL = 4, // decimal fraction, array 83 CBOR_TAG_DECIMAL = 4, // decimal fraction, array
43 CBOR_TAG_BIGFLOAT = 5, // array 84 CBOR_TAG_BIGFLOAT = 5, // array
44 85
45 CBOR_TAG_CONV_B64U = 21, // base64url, any 86 CBOR_TAG_CONV_B64U = 21, // base64url, any
46 CBOR_TAG_CONV_B64 = 22, // base64, any 87 CBOR_TAG_CONV_B64 = 22, // base64, any
47 CBOR_TAG_CONV_HEX = 23, // base16, any 88 CBOR_TAG_CONV_HEX = 23, // base16, any
48 CBOR_TAG_CBOR = 24, // embedded cbor, byte string 89 CBOR_TAG_CBOR = 24, // embedded cbor, byte string
49 90
50 CBOR_TAG_URI = 32, // URI rfc3986, utf-8 91 CBOR_TAG_URI = 32, // URI rfc3986, utf-8
51 CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8 92 CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8
52 CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8 93 CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8
53 CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8 94 CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8
54 CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8 95 CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8
55 96
56 CBOR_TAG_MAGIC = 55799 // self-describe cbor 97 CBOR_TAG_MAGIC = 55799, // self-describe cbor
57}; 98};
58 99
59#define F_SHRINK 0x00000001UL 100#define F_SHRINK 0x00000001UL
60#define F_ALLOW_UNKNOWN 0x00000002UL 101#define F_ALLOW_UNKNOWN 0x00000002UL
61#define F_ALLOW_SHARING 0x00000004UL //TODO 102#define F_ALLOW_SHARING 0x00000004UL
62#define F_ALLOW_STRINGREF 0x00000008UL //TODO 103#define F_ALLOW_CYCLES 0x00000008UL
104#define F_PACK_STRINGS 0x00000010UL
105#define F_VALIDATE_UTF8 0x00000020UL
63 106
64#define INIT_SIZE 32 // initial scalar size to be allocated 107#define INIT_SIZE 32 // initial scalar size to be allocated
65 108
66#define SB do { 109#define SB do {
67#define SE } while (0) 110#define SE } while (0)
79# define CBOR_SLOW 0 122# define CBOR_SLOW 0
80# define CBOR_STASH cbor_stash 123# define CBOR_STASH cbor_stash
81#endif 124#endif
82 125
83static HV *cbor_stash, *types_boolean_stash, *types_error_stash, *cbor_tagged_stash; // CBOR::XS:: 126static HV *cbor_stash, *types_boolean_stash, *types_error_stash, *cbor_tagged_stash; // CBOR::XS::
84static SV *types_true, *types_false, *types_error, *sv_cbor; 127static SV *types_true, *types_false, *types_error, *sv_cbor, *default_filter;
85 128
86typedef struct { 129typedef struct {
87 U32 flags; 130 U32 flags;
88 U32 max_depth; 131 U32 max_depth;
89 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.
90} CBOR; 139} CBOR;
91 140
92ecb_inline void 141ecb_inline void
93cbor_init (CBOR *cbor) 142cbor_init (CBOR *cbor)
94{ 143{
95 Zero (cbor, 1, CBOR); 144 Zero (cbor, 1, CBOR);
96 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);
97} 153}
98 154
99///////////////////////////////////////////////////////////////////////////// 155/////////////////////////////////////////////////////////////////////////////
100// utility functions 156// utility functions
101 157
131{ 187{
132 return idx > 23 188 return idx > 23
133 ? idx > 0xffU 189 ? idx > 0xffU
134 ? idx > 0xffffU 190 ? idx > 0xffffU
135 ? idx > 0xffffffffU 191 ? idx > 0xffffffffU
192 ? 11
136 ? 7 193 : 7
137 : 6
138 : 5 194 : 5
139 : 4 195 : 4
140 : 3; 196 : 3;
141} 197}
142 198
179static void 235static void
180encode_uint (enc_t *enc, int major, UV len) 236encode_uint (enc_t *enc, int major, UV len)
181{ 237{
182 need (enc, 9); 238 need (enc, 9);
183 239
184 if (len < 24) 240 if (ecb_expect_true (len < LENGTH_EXT1))
185 *enc->cur++ = major | len; 241 *enc->cur++ = major | len;
186 else if (len <= 0xff) 242 else if (ecb_expect_true (len <= 0xffU))
187 { 243 {
188 *enc->cur++ = major | 24; 244 *enc->cur++ = major | LENGTH_EXT1;
189 *enc->cur++ = len; 245 *enc->cur++ = len;
190 } 246 }
191 else if (len <= 0xffff) 247 else if (len <= 0xffffU)
192 { 248 {
193 *enc->cur++ = major | 25; 249 *enc->cur++ = major | LENGTH_EXT2;
194 *enc->cur++ = len >> 8; 250 *enc->cur++ = len >> 8;
195 *enc->cur++ = len; 251 *enc->cur++ = len;
196 } 252 }
197 else if (len <= 0xffffffff) 253 else if (len <= 0xffffffffU)
198 { 254 {
199 *enc->cur++ = major | 26; 255 *enc->cur++ = major | LENGTH_EXT4;
200 *enc->cur++ = len >> 24; 256 *enc->cur++ = len >> 24;
201 *enc->cur++ = len >> 16; 257 *enc->cur++ = len >> 16;
202 *enc->cur++ = len >> 8; 258 *enc->cur++ = len >> 8;
203 *enc->cur++ = len; 259 *enc->cur++ = len;
204 } 260 }
205 else 261 else
206 { 262 {
207 *enc->cur++ = major | 27; 263 *enc->cur++ = major | LENGTH_EXT8;
208 *enc->cur++ = len >> 56; 264 *enc->cur++ = len >> 56;
209 *enc->cur++ = len >> 48; 265 *enc->cur++ = len >> 48;
210 *enc->cur++ = len >> 40; 266 *enc->cur++ = len >> 40;
211 *enc->cur++ = len >> 32; 267 *enc->cur++ = len >> 32;
212 *enc->cur++ = len >> 24; 268 *enc->cur++ = len >> 24;
217} 273}
218 274
219ecb_inline void 275ecb_inline void
220encode_tag (enc_t *enc, UV tag) 276encode_tag (enc_t *enc, UV tag)
221{ 277{
222 encode_uint (enc, 0xc0, tag); 278 encode_uint (enc, MAJOR_TAG, tag);
279}
280
281ecb_inline void
282encode_str (enc_t *enc, int utf8, char *str, STRLEN len)
283{
284 encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len);
285 need (enc, len);
286 memcpy (enc->cur, str, len);
287 enc->cur += len;
223} 288}
224 289
225static void 290static void
226encode_str (enc_t *enc, int utf8, char *str, STRLEN len) 291encode_strref (enc_t *enc, int utf8, char *str, STRLEN len)
227{ 292{
228 if (ecb_expect_false (enc->cbor.flags & F_ALLOW_STRINGREF)) 293 if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS))
229 { 294 {
230 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1); 295 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1);
231 296
232 if (SvOK (*svp)) 297 if (SvOK (*svp))
233 { 298 {
234 // already registered, use stringref 299 // already registered, use stringref
235 encode_tag (enc, CBOR_TAG_STRINGREF); 300 encode_tag (enc, CBOR_TAG_STRINGREF);
236 encode_uint (enc, 0x00, SvUV (*svp)); 301 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
237 return; 302 return;
238 } 303 }
239 else if (len >= minimum_string_length (enc->stringref_idx)) 304 else if (len >= minimum_string_length (enc->stringref_idx))
240 { 305 {
241 // register only 306 // register only
242 sv_setuv (*svp, enc->stringref_idx); 307 sv_setuv (*svp, enc->stringref_idx);
243 ++enc->stringref_idx; 308 ++enc->stringref_idx;
244 } 309 }
245 } 310 }
246 311
247 encode_uint (enc, utf8 ? 0x60 : 0x40, len); 312 encode_str (enc, utf8, str, len);
248 need (enc, len);
249 memcpy (enc->cur, str, len);
250 enc->cur += len;
251} 313}
252 314
253static void encode_sv (enc_t *enc, SV *sv); 315static void encode_sv (enc_t *enc, SV *sv);
254 316
255static void 317static void
260 if (enc->depth >= enc->cbor.max_depth) 322 if (enc->depth >= enc->cbor.max_depth)
261 croak (ERR_NESTING_EXCEEDED); 323 croak (ERR_NESTING_EXCEEDED);
262 324
263 ++enc->depth; 325 ++enc->depth;
264 326
265 encode_uint (enc, 0x80, len + 1); 327 encode_uint (enc, MAJOR_ARRAY, len + 1);
266 328
329 if (SvMAGICAL (av))
267 for (i = 0; i <= len; ++i) 330 for (i = 0; i <= len; ++i)
268 { 331 {
269 SV **svp = av_fetch (av, i, 0); 332 SV **svp = av_fetch (av, i, 0);
270 encode_sv (enc, svp ? *svp : &PL_sv_undef); 333 encode_sv (enc, svp ? *svp : &PL_sv_undef);
271 } 334 }
335 else
336 for (i = 0; i <= len; ++i)
337 encode_sv (enc, AvARRAY (av)[i]);
272 338
273 --enc->depth; 339 --enc->depth;
274}
275
276ecb_inline void
277encode_he (enc_t *enc, HE *he)
278{
279} 340}
280 341
281static void 342static void
282encode_hv (enc_t *enc, HV *hv) 343encode_hv (enc_t *enc, HV *hv)
283{ 344{
290 351
291 int pairs = hv_iterinit (hv); 352 int pairs = hv_iterinit (hv);
292 int mg = SvMAGICAL (hv); 353 int mg = SvMAGICAL (hv);
293 354
294 if (mg) 355 if (mg)
295 encode_ch (enc, 0xa0 | 31); 356 encode_ch (enc, MAJOR_MAP | MINOR_INDEF);
296 else 357 else
297 encode_uint (enc, 0xa0, pairs); 358 encode_uint (enc, MAJOR_MAP, pairs);
298 359
299 while ((he = hv_iternext (hv))) 360 while ((he = hv_iternext (hv)))
300 { 361 {
301 if (HeKLEN (he) == HEf_SVKEY) 362 if (HeKLEN (he) == HEf_SVKEY)
302 encode_sv (enc, HeSVKEY (he)); 363 encode_sv (enc, HeSVKEY (he));
303 else 364 else
304 encode_str (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he)); 365 encode_strref (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he));
305 366
306 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he)); 367 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he));
307 } 368 }
308 369
309 if (mg) 370 if (mg)
310 encode_ch (enc, 0xe0 | 31); 371 encode_ch (enc, MAJOR_MISC | MINOR_INDEF);
311 372
312 --enc->depth; 373 --enc->depth;
313} 374}
314 375
315// encode objects, arrays and special \0=false and \1=true values. 376// encode objects, arrays and special \0=false and \1=true values.
316static void 377static void
317encode_rv (enc_t *enc, SV *sv) 378encode_rv (enc_t *enc, SV *sv)
318{ 379{
319 SvGETMAGIC (sv); 380 SvGETMAGIC (sv);
320
321 if (ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING)
322 && ecb_expect_false (SvREFCNT (sv) > 1))
323 {
324 if (!enc->shareable)
325 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
326
327 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
328
329 if (SvOK (*svp))
330 {
331 encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF);
332 encode_uint (enc, 0x00, SvUV (*svp));
333 return;
334 }
335 else
336 {
337 sv_setuv (*svp, enc->shareable_idx);
338 ++enc->shareable_idx;
339 encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE);
340 }
341 }
342 381
343 svtype svt = SvTYPE (sv); 382 svtype svt = SvTYPE (sv);
344 383
345 if (ecb_expect_false (SvOBJECT (sv))) 384 if (ecb_expect_false (SvOBJECT (sv)))
346 { 385 {
353 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 392 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
354 ? cbor_tagged_stash 393 ? cbor_tagged_stash
355 : gv_stashpv ("CBOR::XS::Tagged" , 1); 394 : gv_stashpv ("CBOR::XS::Tagged" , 1);
356 395
357 HV *stash = SvSTASH (sv); 396 HV *stash = SvSTASH (sv);
358 GV *method;
359 397
360 if (stash == boolean_stash) 398 if (stash == boolean_stash)
361 encode_ch (enc, SvIV (sv) ? 0xe0 | 21 : 0xe0 | 20); 399 {
400 encode_ch (enc, SvIV (sv) ? MAJOR_MISC | SIMPLE_TRUE : MAJOR_MISC | SIMPLE_FALSE);
401 return;
402 }
362 else if (stash == error_stash) 403 else if (stash == error_stash)
363 encode_ch (enc, 0xe0 | 23); 404 {
405 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
406 return;
407 }
364 else if (stash == tagged_stash) 408 else if (stash == tagged_stash)
365 { 409 {
366 if (svt != SVt_PVAV) 410 if (svt != SVt_PVAV)
367 croak ("encountered CBOR::XS::Tagged object that isn't an array"); 411 croak ("encountered CBOR::XS::Tagged object that isn't an array");
368 412
369 encode_uint (enc, 0xc0, SvUV (*av_fetch ((AV *)sv, 0, 1))); 413 encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1)));
370 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1)); 414 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1));
415
416 return;
417 }
418 }
419
420 if (ecb_expect_false (SvREFCNT (sv) > 1)
421 && ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING))
422 {
423 if (!enc->shareable)
424 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
425
426 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
427
428 if (SvOK (*svp))
371 } 429 {
430 encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF);
431 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
432 return;
433 }
434 else
435 {
436 sv_setuv (*svp, enc->shareable_idx);
437 ++enc->shareable_idx;
438 encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE);
439 }
440 }
441
442 if (ecb_expect_false (SvOBJECT (sv)))
443 {
444 HV *stash = SvSTASH (sv);
445 GV *method;
446
372 else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0))) 447 if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
373 { 448 {
374 dSP; 449 dSP;
375 450
376 ENTER; SAVETMPS; PUSHMARK (SP); 451 ENTER; SAVETMPS; PUSHMARK (SP);
377 // we re-bless the reference to get overload and other niceties right 452 // we re-bless the reference to get overload and other niceties right
409 // catch this surprisingly common error 484 // catch this surprisingly common error
410 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv) 485 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)
411 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash)); 486 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));
412 487
413 encode_tag (enc, CBOR_TAG_PERL_OBJECT); 488 encode_tag (enc, CBOR_TAG_PERL_OBJECT);
414 encode_uint (enc, 0x80, count + 1); 489 encode_uint (enc, MAJOR_ARRAY, count + 1);
415 encode_str (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash)); 490 encode_strref (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
416 491
417 while (count) 492 while (count)
418 encode_sv (enc, SP[1 - count--]); 493 encode_sv (enc, SP[1 - count--]);
419 494
420 PUTBACK; 495 PUTBACK;
441{ 516{
442 double nv = SvNVX (sv); 517 double nv = SvNVX (sv);
443 518
444 need (enc, 9); 519 need (enc, 9);
445 520
446 if (ecb_expect_false (nv == (U32)nv)) 521 if (ecb_expect_false (nv == (NV)(U32)nv))
447 encode_uint (enc, 0x00, (U32)nv); 522 encode_uint (enc, MAJOR_POS_INT, (U32)nv);
448 //TODO: maybe I32? 523 //TODO: maybe I32?
449 else if (ecb_expect_false (nv == (float)nv)) 524 else if (ecb_expect_false (nv == (float)nv))
450 { 525 {
451 uint32_t fp = ecb_float_to_binary32 (nv); 526 uint32_t fp = ecb_float_to_binary32 (nv);
452 527
453 *enc->cur++ = 0xe0 | 26; 528 *enc->cur++ = MAJOR_MISC | MISC_FLOAT32;
454 529
455 if (!ecb_big_endian ()) 530 if (!ecb_big_endian ())
456 fp = ecb_bswap32 (fp); 531 fp = ecb_bswap32 (fp);
457 532
458 memcpy (enc->cur, &fp, 4); 533 memcpy (enc->cur, &fp, 4);
460 } 535 }
461 else 536 else
462 { 537 {
463 uint64_t fp = ecb_double_to_binary64 (nv); 538 uint64_t fp = ecb_double_to_binary64 (nv);
464 539
465 *enc->cur++ = 0xe0 | 27; 540 *enc->cur++ = MAJOR_MISC | MISC_FLOAT64;
466 541
467 if (!ecb_big_endian ()) 542 if (!ecb_big_endian ())
468 fp = ecb_bswap64 (fp); 543 fp = ecb_bswap64 (fp);
469 544
470 memcpy (enc->cur, &fp, 8); 545 memcpy (enc->cur, &fp, 8);
479 554
480 if (SvPOKp (sv)) 555 if (SvPOKp (sv))
481 { 556 {
482 STRLEN len; 557 STRLEN len;
483 char *str = SvPV (sv, len); 558 char *str = SvPV (sv, len);
484 encode_str (enc, SvUTF8 (sv), str, len); 559 encode_strref (enc, SvUTF8 (sv), str, len);
485 } 560 }
486 else if (SvNOKp (sv)) 561 else if (SvNOKp (sv))
487 encode_nv (enc, sv); 562 encode_nv (enc, sv);
488 else if (SvIOKp (sv)) 563 else if (SvIOKp (sv))
489 { 564 {
490 if (SvIsUV (sv)) 565 if (SvIsUV (sv))
491 encode_uint (enc, 0x00, SvUVX (sv)); 566 encode_uint (enc, MAJOR_POS_INT, SvUVX (sv));
492 else if (SvIVX (sv) >= 0) 567 else if (SvIVX (sv) >= 0)
493 encode_uint (enc, 0x00, SvIVX (sv)); 568 encode_uint (enc, MAJOR_POS_INT, SvIVX (sv));
494 else 569 else
495 encode_uint (enc, 0x20, -(SvIVX (sv) + 1)); 570 encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1));
496 } 571 }
497 else if (SvROK (sv)) 572 else if (SvROK (sv))
498 encode_rv (enc, SvRV (sv)); 573 encode_rv (enc, SvRV (sv));
499 else if (!SvOK (sv)) 574 else if (!SvOK (sv))
500 encode_ch (enc, 0xe0 | 22); 575 encode_ch (enc, MAJOR_MISC | SIMPLE_NULL);
501 else if (enc->cbor.flags & F_ALLOW_UNKNOWN) 576 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
502 encode_ch (enc, 0xe0 | 23); 577 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
503 else 578 else
504 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data", 579 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data",
505 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv)); 580 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
506} 581}
507 582
515 enc.cur = SvPVX (enc.sv); 590 enc.cur = SvPVX (enc.sv);
516 enc.end = SvEND (enc.sv); 591 enc.end = SvEND (enc.sv);
517 592
518 SvPOK_only (enc.sv); 593 SvPOK_only (enc.sv);
519 594
520 if (cbor->flags & F_ALLOW_STRINGREF) 595 if (cbor->flags & F_PACK_STRINGS)
521 { 596 {
522 encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE); 597 encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE);
523 enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ()); 598 enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ());
524 enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ()); 599 enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ());
525 } 600 }
547 CBOR cbor; 622 CBOR cbor;
548 U32 depth; // recursion depth 623 U32 depth; // recursion depth
549 U32 maxdepth; // recursion depth limit 624 U32 maxdepth; // recursion depth limit
550 AV *shareable; 625 AV *shareable;
551 AV *stringref; 626 AV *stringref;
627 SV *decode_tagged;
552} dec_t; 628} dec_t;
553 629
554#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE 630#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE
555 631
556#define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data") 632#define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data")
559#define DEC_DEC_DEPTH --dec->depth 635#define DEC_DEC_DEPTH --dec->depth
560 636
561static UV 637static UV
562decode_uint (dec_t *dec) 638decode_uint (dec_t *dec)
563{ 639{
564 switch (*dec->cur & 31) 640 U8 m = *dec->cur & MINOR_MASK;
565 { 641 ++dec->cur;
566 case 0: case 1: case 2: case 3: case 4: case 5: case 6: case 7:
567 case 8: case 9: case 10: case 11: case 12: case 13: case 14: case 15:
568 case 16: case 17: case 18: case 19: case 20: case 21: case 22: case 23:
569 return *dec->cur++ & 31;
570 642
571 case 24: 643 if (ecb_expect_true (m < LENGTH_EXT1))
644 return m;
645 else if (ecb_expect_true (m == LENGTH_EXT1))
646 {
572 WANT (2); 647 WANT (1);
573 dec->cur += 2; 648 dec->cur += 1;
574 return dec->cur[-1]; 649 return dec->cur[-1];
575 650 }
576 case 25: 651 else if (ecb_expect_true (m == LENGTH_EXT2))
652 {
577 WANT (3); 653 WANT (2);
578 dec->cur += 3; 654 dec->cur += 2;
579 return (((UV)dec->cur[-2]) << 8) 655 return (((UV)dec->cur[-2]) << 8)
580 | ((UV)dec->cur[-1]); 656 | ((UV)dec->cur[-1]);
581 657 }
582 case 26: 658 else if (ecb_expect_true (m == LENGTH_EXT4))
659 {
583 WANT (5); 660 WANT (4);
584 dec->cur += 5; 661 dec->cur += 4;
585 return (((UV)dec->cur[-4]) << 24) 662 return (((UV)dec->cur[-4]) << 24)
586 | (((UV)dec->cur[-3]) << 16) 663 | (((UV)dec->cur[-3]) << 16)
587 | (((UV)dec->cur[-2]) << 8) 664 | (((UV)dec->cur[-2]) << 8)
588 | ((UV)dec->cur[-1]); 665 | ((UV)dec->cur[-1]);
589 666 }
590 case 27: 667 else if (ecb_expect_true (m == LENGTH_EXT8))
668 {
591 WANT (9); 669 WANT (8);
592 dec->cur += 9; 670 dec->cur += 8;
671
672 return
673#if UVSIZE < 8
674 0
675#else
593 return (((UV)dec->cur[-8]) << 56) 676 (((UV)dec->cur[-8]) << 56)
594 | (((UV)dec->cur[-7]) << 48) 677 | (((UV)dec->cur[-7]) << 48)
595 | (((UV)dec->cur[-6]) << 40) 678 | (((UV)dec->cur[-6]) << 40)
596 | (((UV)dec->cur[-5]) << 32) 679 | (((UV)dec->cur[-5]) << 32)
680#endif
597 | (((UV)dec->cur[-4]) << 24) 681 | (((UV)dec->cur[-4]) << 24)
598 | (((UV)dec->cur[-3]) << 16) 682 | (((UV)dec->cur[-3]) << 16)
599 | (((UV)dec->cur[-2]) << 8) 683 | (((UV)dec->cur[-2]) << 8)
600 | ((UV)dec->cur[-1]); 684 | ((UV)dec->cur[-1]);
601 685 }
602 default: 686 else
603 ERR ("corrupted CBOR data (unsupported integer minor encoding)"); 687 ERR ("corrupted CBOR data (unsupported integer minor encoding)");
604 }
605 688
606fail: 689fail:
607 return 0; 690 return 0;
608} 691}
609 692
614{ 697{
615 AV *av = newAV (); 698 AV *av = newAV ();
616 699
617 DEC_INC_DEPTH; 700 DEC_INC_DEPTH;
618 701
619 if ((*dec->cur & 31) == 31) 702 if (*dec->cur == (MAJOR_ARRAY | MINOR_INDEF))
620 { 703 {
621 ++dec->cur; 704 ++dec->cur;
622 705
623 for (;;) 706 for (;;)
624 { 707 {
625 WANT (1); 708 WANT (1);
626 709
627 if (*dec->cur == (0xe0 | 31)) 710 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
628 { 711 {
629 ++dec->cur; 712 ++dec->cur;
630 break; 713 break;
631 } 714 }
632 715
635 } 718 }
636 else 719 else
637 { 720 {
638 int i, len = decode_uint (dec); 721 int i, len = decode_uint (dec);
639 722
723 WANT (len); // complexity check for av_fill - need at least one byte per value, do not allow supersize arrays
640 av_fill (av, len - 1); 724 av_fill (av, len - 1);
641 725
642 for (i = 0; i < len; ++i) 726 for (i = 0; i < len; ++i)
643 AvARRAY (av)[i] = decode_sv (dec); 727 AvARRAY (av)[i] = decode_sv (dec);
644 } 728 }
657{ 741{
658 // for speed reasons, we specialcase single-string 742 // for speed reasons, we specialcase single-string
659 // byte or utf-8 strings as keys, but only when !stringref 743 // byte or utf-8 strings as keys, but only when !stringref
660 744
661 if (ecb_expect_true (!dec->stringref)) 745 if (ecb_expect_true (!dec->stringref))
662 if (*dec->cur >= 0x40 && *dec->cur <= 0x40 + 27) 746 if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8))
663 { 747 {
664 I32 len = decode_uint (dec); 748 I32 len = decode_uint (dec);
665 char *key = (char *)dec->cur; 749 char *key = (char *)dec->cur;
666 750
667 dec->cur += len; 751 dec->cur += len;
668 752
669 if (ecb_expect_false (dec->stringref))
670 av_push (dec->stringref, newSVpvn (key, len));
671
672 hv_store (hv, key, len, decode_sv (dec), 0); 753 hv_store (hv, key, len, decode_sv (dec), 0);
673 754
674 return; 755 return;
675 } 756 }
676 else if (*dec->cur >= 0x60 && *dec->cur <= 0x60 + 27) 757 else if (ecb_expect_true ((U8)(*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8))
677 { 758 {
678 I32 len = decode_uint (dec); 759 I32 len = decode_uint (dec);
679 char *key = (char *)dec->cur; 760 char *key = (char *)dec->cur;
680 761
681 dec->cur += len; 762 dec->cur += len;
682 763
683 if (ecb_expect_false (dec->stringref)) 764 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
684 av_push (dec->stringref, newSVpvn_utf8 (key, len, 1)); 765 if (!is_utf8_string (key, len))
766 ERR ("corrupted CBOR data (invalid UTF-8 in map key)");
685 767
686 hv_store (hv, key, -len, decode_sv (dec), 0); 768 hv_store (hv, key, -len, decode_sv (dec), 0);
687 769
688 return; 770 return;
689 } 771 }
691 SV *k = decode_sv (dec); 773 SV *k = decode_sv (dec);
692 SV *v = decode_sv (dec); 774 SV *v = decode_sv (dec);
693 775
694 hv_store_ent (hv, k, v, 0); 776 hv_store_ent (hv, k, v, 0);
695 SvREFCNT_dec (k); 777 SvREFCNT_dec (k);
778
779fail:
780 ;
696} 781}
697 782
698static SV * 783static SV *
699decode_hv (dec_t *dec) 784decode_hv (dec_t *dec)
700{ 785{
701 HV *hv = newHV (); 786 HV *hv = newHV ();
702 787
703 DEC_INC_DEPTH; 788 DEC_INC_DEPTH;
704 789
705 if ((*dec->cur & 31) == 31) 790 if (*dec->cur == (MAJOR_MAP | MINOR_INDEF))
706 { 791 {
707 ++dec->cur; 792 ++dec->cur;
708 793
709 for (;;) 794 for (;;)
710 { 795 {
711 WANT (1); 796 WANT (1);
712 797
713 if (*dec->cur == (0xe0 | 31)) 798 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
714 { 799 {
715 ++dec->cur; 800 ++dec->cur;
716 break; 801 break;
717 } 802 }
718 803
739static SV * 824static SV *
740decode_str (dec_t *dec, int utf8) 825decode_str (dec_t *dec, int utf8)
741{ 826{
742 SV *sv = 0; 827 SV *sv = 0;
743 828
744 if ((*dec->cur & 31) == 31) 829 if ((*dec->cur & MINOR_MASK) == MINOR_INDEF)
745 { 830 {
831 // indefinite length strings
746 ++dec->cur; 832 ++dec->cur;
747 833
834 U8 major = *dec->cur & MAJOR_MISC;
835
748 sv = newSVpvn ("", 0); 836 sv = newSVpvn ("", 0);
749 837
750 // not very fast, and certainly not robust against illegal input
751 for (;;) 838 for (;;)
752 { 839 {
753 WANT (1); 840 WANT (1);
754 841
755 if (*dec->cur == (0xe0 | 31)) 842 if ((*dec->cur - major) > LENGTH_EXT8)
843 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
756 { 844 {
757 ++dec->cur; 845 ++dec->cur;
758 break; 846 break;
759 } 847 }
848 else
849 ERR ("corrupted CBOR data (invalid chunks in indefinite length string)");
760 850
761 sv_catsv (sv, decode_sv (dec)); 851 STRLEN len = decode_uint (dec);
852
853 WANT (len);
854 sv_catpvn (sv, dec->cur, len);
855 dec->cur += len;
762 } 856 }
763 } 857 }
764 else 858 else
765 { 859 {
766 STRLEN len = decode_uint (dec); 860 STRLEN len = decode_uint (dec);
767 861
768 WANT (len); 862 WANT (len);
769 sv = newSVpvn (dec->cur, len); 863 sv = newSVpvn (dec->cur, len);
770 dec->cur += len; 864 dec->cur += len;
865
866 if (ecb_expect_false (dec->stringref)
867 && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1))
868 av_push (dec->stringref, SvREFCNT_inc_NN (sv));
771 } 869 }
772 870
773 if (utf8) 871 if (utf8)
872 {
873 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
874 if (!is_utf8_string (SvPVX (sv), SvCUR (sv)))
875 ERR ("corrupted CBOR data (invalid UTF-8 in text string)");
876
774 SvUTF8_on (sv); 877 SvUTF8_on (sv);
775 878 }
776 if (ecb_expect_false (dec->stringref)
777 && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1))
778 av_push (dec->stringref, SvREFCNT_inc_NN (sv));
779 879
780 return sv; 880 return sv;
781 881
782fail: 882fail:
783 SvREFCNT_dec (sv); 883 SvREFCNT_dec (sv);
815 } 915 }
816 break; 916 break;
817 917
818 case CBOR_TAG_STRINGREF: 918 case CBOR_TAG_STRINGREF:
819 { 919 {
820 if ((*dec->cur >> 5) != 0) 920 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
821 ERR ("corrupted CBOR data (stringref index not an unsigned integer)"); 921 ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
822 922
823 UV idx = decode_uint (dec); 923 UV idx = decode_uint (dec);
824 924
825 if (!dec->stringref || (int)idx > AvFILLp (dec->stringref)) 925 if (!dec->stringref || (int)idx > AvFILLp (dec->stringref))
832 case CBOR_TAG_VALUE_SHAREABLE: 932 case CBOR_TAG_VALUE_SHAREABLE:
833 { 933 {
834 if (ecb_expect_false (!dec->shareable)) 934 if (ecb_expect_false (!dec->shareable))
835 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ()); 935 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ());
836 936
937 if (dec->cbor.flags & F_ALLOW_CYCLES)
938 {
837 sv = newSV (0); 939 sv = newSV (0);
838 av_push (dec->shareable, SvREFCNT_inc_NN (sv)); 940 av_push (dec->shareable, SvREFCNT_inc_NN (sv));
839 941
840 SV *osv = decode_sv (dec); 942 SV *osv = decode_sv (dec);
841 sv_setsv (sv, osv); 943 sv_setsv (sv, osv);
842 SvREFCNT_dec_NN (osv); 944 SvREFCNT_dec_NN (osv);
945 }
946 else
947 {
948 av_push (dec->shareable, &PL_sv_undef);
949 int idx = AvFILLp (dec->shareable);
950 sv = decode_sv (dec);
951 av_store (dec->shareable, idx, SvREFCNT_inc_NN (sv));
952 }
843 } 953 }
844 break; 954 break;
845 955
846 case CBOR_TAG_VALUE_SHAREDREF: 956 case CBOR_TAG_VALUE_SHAREDREF:
847 { 957 {
848 if ((*dec->cur >> 5) != 0) 958 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
849 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)"); 959 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
850 960
851 UV idx = decode_uint (dec); 961 UV idx = decode_uint (dec);
852 962
853 if (!dec->shareable || (int)idx > AvFILLp (dec->shareable)) 963 if (!dec->shareable || (int)idx > AvFILLp (dec->shareable))
854 ERR ("corrupted CBOR data (sharedref index out of bounds)"); 964 ERR ("corrupted CBOR data (sharedref index out of bounds)");
855 965
856 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]); 966 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]);
967
968 if (sv == &PL_sv_undef)
969 ERR ("cyclic CBOR data structure found, but allow_cycles is not enabled");
857 } 970 }
858 break; 971 break;
859 972
860 case CBOR_TAG_PERL_OBJECT: 973 case CBOR_TAG_PERL_OBJECT:
861 { 974 {
910 1023
911 default: 1024 default:
912 { 1025 {
913 sv = decode_sv (dec); 1026 sv = decode_sv (dec);
914 1027
1028 dSP;
1029 ENTER; SAVETMPS; PUSHMARK (SP);
1030 EXTEND (SP, 2);
1031 PUSHs (newSVuv (tag));
1032 PUSHs (sv);
1033
1034 PUTBACK;
1035 int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL);
1036 SPAGAIN;
1037
1038 if (SvTRUE (ERRSV))
1039 {
1040 FREETMPS; LEAVE;
1041 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV))));
1042 }
1043
1044 if (count)
1045 {
1046 SvREFCNT_dec (sv);
1047 sv = SvREFCNT_inc (POPs);
1048 }
1049 else
1050 {
915 AV *av = newAV (); 1051 AV *av = newAV ();
916 av_push (av, newSVuv (tag)); 1052 av_push (av, newSVuv (tag));
917 av_push (av, sv); 1053 av_push (av, sv);
918 1054
919 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 1055 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
920 ? cbor_tagged_stash 1056 ? cbor_tagged_stash
921 : gv_stashpv ("CBOR::XS::Tagged" , 1); 1057 : gv_stashpv ("CBOR::XS::Tagged" , 1);
922
923 sv = sv_bless (newRV_noinc ((SV *)av), tagged_stash); 1058 sv = sv_bless (newRV_noinc ((SV *)av), tagged_stash);
1059 }
1060
1061 PUTBACK;
1062
1063 FREETMPS; LEAVE;
924 } 1064 }
925 break; 1065 break;
926 } 1066 }
927 1067
928 return sv; 1068 return sv;
935static SV * 1075static SV *
936decode_sv (dec_t *dec) 1076decode_sv (dec_t *dec)
937{ 1077{
938 WANT (1); 1078 WANT (1);
939 1079
940 switch (*dec->cur >> 5) 1080 switch (*dec->cur >> MAJOR_SHIFT)
941 { 1081 {
942 case 0: // unsigned int 1082 case MAJOR_POS_INT >> MAJOR_SHIFT: return newSVuv (decode_uint (dec));
943 return newSVuv (decode_uint (dec)); 1083 case MAJOR_NEG_INT >> MAJOR_SHIFT: return newSViv (-1 - (IV)decode_uint (dec));
944 case 1: // negative int 1084 case MAJOR_BYTES >> MAJOR_SHIFT: return decode_str (dec, 0);
945 return newSViv (-1 - (IV)decode_uint (dec)); 1085 case MAJOR_TEXT >> MAJOR_SHIFT: return decode_str (dec, 1);
946 case 2: // octet string 1086 case MAJOR_ARRAY >> MAJOR_SHIFT: return decode_av (dec);
947 return decode_str (dec, 0); 1087 case MAJOR_MAP >> MAJOR_SHIFT: return decode_hv (dec);
948 case 3: // utf-8 string 1088 case MAJOR_TAG >> MAJOR_SHIFT: return decode_tagged (dec);
949 return decode_str (dec, 1); 1089
950 case 4: // array 1090 case MAJOR_MISC >> MAJOR_SHIFT:
951 return decode_av (dec);
952 case 5: // map
953 return decode_hv (dec);
954 case 6: // tag
955 return decode_tagged (dec);
956 case 7: // misc
957 switch (*dec->cur++ & 31) 1091 switch (*dec->cur++ & MINOR_MASK)
958 { 1092 {
959 case 20: 1093 case SIMPLE_FALSE:
960#if CBOR_SLOW 1094#if CBOR_SLOW
961 types_false = get_bool ("Types::Serialiser::false"); 1095 types_false = get_bool ("Types::Serialiser::false");
962#endif 1096#endif
963 return newSVsv (types_false); 1097 return newSVsv (types_false);
964 case 21: 1098 case SIMPLE_TRUE:
965#if CBOR_SLOW 1099#if CBOR_SLOW
966 types_true = get_bool ("Types::Serialiser::true"); 1100 types_true = get_bool ("Types::Serialiser::true");
967#endif 1101#endif
968 return newSVsv (types_true); 1102 return newSVsv (types_true);
969 case 22: 1103 case SIMPLE_NULL:
970 return newSVsv (&PL_sv_undef); 1104 return newSVsv (&PL_sv_undef);
971 case 23: 1105 case SIMPLE_UNDEF:
972#if CBOR_SLOW 1106#if CBOR_SLOW
973 types_error = get_bool ("Types::Serialiser::error"); 1107 types_error = get_bool ("Types::Serialiser::error");
974#endif 1108#endif
975 return newSVsv (types_error); 1109 return newSVsv (types_error);
976 1110
977 case 25: 1111 case MISC_FLOAT16:
978 { 1112 {
979 WANT (2); 1113 WANT (2);
980 1114
981 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1]; 1115 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1];
982 dec->cur += 2; 1116 dec->cur += 2;
983 1117
984 return newSVnv (ecb_binary16_to_float (fp)); 1118 return newSVnv (ecb_binary16_to_float (fp));
985 } 1119 }
986 1120
987 case 26: 1121 case MISC_FLOAT32:
988 { 1122 {
989 uint32_t fp; 1123 uint32_t fp;
990 WANT (4); 1124 WANT (4);
991 memcpy (&fp, dec->cur, 4); 1125 memcpy (&fp, dec->cur, 4);
992 dec->cur += 4; 1126 dec->cur += 4;
995 fp = ecb_bswap32 (fp); 1129 fp = ecb_bswap32 (fp);
996 1130
997 return newSVnv (ecb_binary32_to_float (fp)); 1131 return newSVnv (ecb_binary32_to_float (fp));
998 } 1132 }
999 1133
1000 case 27: 1134 case MISC_FLOAT64:
1001 { 1135 {
1002 uint64_t fp; 1136 uint64_t fp;
1003 WANT (8); 1137 WANT (8);
1004 memcpy (&fp, dec->cur, 8); 1138 memcpy (&fp, dec->cur, 8);
1005 dec->cur += 8; 1139 dec->cur += 8;
1008 fp = ecb_bswap64 (fp); 1142 fp = ecb_bswap64 (fp);
1009 1143
1010 return newSVnv (ecb_binary64_to_double (fp)); 1144 return newSVnv (ecb_binary64_to_double (fp));
1011 } 1145 }
1012 1146
1013 // 0..19 unassigned 1147 // 0..19 unassigned simple
1014 // 24 reserved + unassigned (reserved values are not encodable) 1148 // 24 reserved + unassigned simple (reserved values are not encodable)
1149 // 28-30 unassigned misc
1150 // 31 break code
1015 default: 1151 default:
1016 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)"); 1152 ERR ("corrupted CBOR data (reserved/unassigned/unexpected major 7 value)");
1017 } 1153 }
1018 1154
1019 break; 1155 break;
1020 } 1156 }
1021 1157
1048 if (dec.cur != dec.end && !dec.err) 1184 if (dec.cur != dec.end && !dec.err)
1049 dec.err = "garbage after CBOR object"; 1185 dec.err = "garbage after CBOR object";
1050 1186
1051 if (dec.err) 1187 if (dec.err)
1052 { 1188 {
1189 if (dec.shareable)
1190 {
1191 // need to break cyclic links, which whould all be in shareable
1192 int i;
1193 SV **svp;
1194
1195 for (i = av_len (dec.shareable) + 1; i--; )
1196 if ((svp = av_fetch (dec.shareable, i, 0)))
1197 sv_setsv (*svp, &PL_sv_undef);
1198 }
1199
1053 SvREFCNT_dec (sv); 1200 SvREFCNT_dec (sv);
1054 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur); 1201 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur);
1055 } 1202 }
1056 1203
1057 sv = sv_2mortal (sv); 1204 sv = sv_2mortal (sv);
1058 1205
1059 return sv; 1206 return sv;
1060} 1207}
1061 1208
1209/////////////////////////////////////////////////////////////////////////////
1210// incremental parser
1211
1212#define INCR_DONE(cbor) (AvFILLp (cbor->incr_count) < 0)
1213
1214// returns 0 for notyet, 1 for success or error
1215static int
1216incr_parse (CBOR *self, SV *cborstr)
1217{
1218 STRLEN cur;
1219 SvPV (cborstr, cur);
1220
1221 while (ecb_expect_true (self->incr_need <= cur))
1222 {
1223 // table of integer count bytes
1224 static I8 incr_len[MINOR_MASK + 1] = {
1225 0, 0, 0, 0, 0, 0, 0, 0,
1226 0, 0, 0, 0, 0, 0, 0, 0,
1227 0, 0, 0, 0, 0, 0, 0, 0,
1228 1, 2, 4, 8,-1,-1,-1,-2
1229 };
1230
1231 const U8 *p = SvPVX (cborstr) + self->incr_pos;
1232 U8 m = *p & MINOR_MASK;
1233 IV count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1234 I8 ilen = incr_len[m];
1235
1236 self->incr_need = self->incr_pos + 1;
1237
1238 if (ecb_expect_false (ilen < 0))
1239 {
1240 if (m != MINOR_INDEF)
1241 return 1; // error
1242
1243 if (*p == (MAJOR_MISC | MINOR_INDEF))
1244 {
1245 if (count >= 0)
1246 return 1; // error
1247
1248 count = 1;
1249 }
1250 else
1251 {
1252 av_push (self->incr_count, newSViv (-1)); //TODO: nest
1253 count = -1;
1254 }
1255 }
1256 else
1257 {
1258 self->incr_need += ilen;
1259 if (ecb_expect_false (self->incr_need > cur))
1260 return 0;
1261
1262 int major = *p >> MAJOR_SHIFT;
1263
1264 switch (major)
1265 {
1266 case MAJOR_BYTES >> MAJOR_SHIFT:
1267 case MAJOR_TEXT >> MAJOR_SHIFT:
1268 case MAJOR_ARRAY >> MAJOR_SHIFT:
1269 case MAJOR_MAP >> MAJOR_SHIFT:
1270 {
1271 UV len;
1272
1273 if (ecb_expect_false (ilen))
1274 {
1275 len = 0;
1276
1277 do {
1278 len = (len << 8) | *++p;
1279 } while (--ilen);
1280 }
1281 else
1282 len = m;
1283
1284 switch (major)
1285 {
1286 case MAJOR_BYTES >> MAJOR_SHIFT:
1287 case MAJOR_TEXT >> MAJOR_SHIFT:
1288 self->incr_need += len;
1289 if (ecb_expect_false (self->incr_need > cur))
1290 return 0;
1291
1292 break;
1293
1294 case MAJOR_MAP >> MAJOR_SHIFT:
1295 len <<= 1;
1296 case MAJOR_ARRAY >> MAJOR_SHIFT:
1297 if (len)
1298 {
1299 av_push (self->incr_count, newSViv (len + 1)); //TODO: nest
1300 count = len + 1;
1301 }
1302 break;
1303 }
1304 }
1305 }
1306 }
1307
1308 self->incr_pos = self->incr_need;
1309
1310 if (count > 0)
1311 {
1312 while (!--count)
1313 {
1314 if (!AvFILLp (self->incr_count))
1315 return 1; // done
1316
1317 SvREFCNT_dec_NN (av_pop (self->incr_count));
1318 count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1319 }
1320
1321 SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]) = count;
1322 }
1323 }
1324
1325 return 0;
1326}
1327
1328
1062///////////////////////////////////////////////////////////////////////////// 1329/////////////////////////////////////////////////////////////////////////////
1063// XS interface functions 1330// XS interface functions
1064 1331
1065MODULE = CBOR::XS PACKAGE = CBOR::XS 1332MODULE = CBOR::XS PACKAGE = CBOR::XS
1066 1333
1073 types_error_stash = gv_stashpv ("Types::Serialiser::Error" , 1); 1340 types_error_stash = gv_stashpv ("Types::Serialiser::Error" , 1);
1074 1341
1075 types_true = get_bool ("Types::Serialiser::true" ); 1342 types_true = get_bool ("Types::Serialiser::true" );
1076 types_false = get_bool ("Types::Serialiser::false"); 1343 types_false = get_bool ("Types::Serialiser::false");
1077 types_error = get_bool ("Types::Serialiser::error"); 1344 types_error = get_bool ("Types::Serialiser::error");
1345
1346 default_filter = newSVpv ("CBOR::XS::default_filter", 0);
1078 1347
1079 sv_cbor = newSVpv ("CBOR", 0); 1348 sv_cbor = newSVpv ("CBOR", 0);
1080 SvREADONLY_on (sv_cbor); 1349 SvREADONLY_on (sv_cbor);
1081} 1350}
1082 1351
1104void shrink (CBOR *self, int enable = 1) 1373void shrink (CBOR *self, int enable = 1)
1105 ALIAS: 1374 ALIAS:
1106 shrink = F_SHRINK 1375 shrink = F_SHRINK
1107 allow_unknown = F_ALLOW_UNKNOWN 1376 allow_unknown = F_ALLOW_UNKNOWN
1108 allow_sharing = F_ALLOW_SHARING 1377 allow_sharing = F_ALLOW_SHARING
1109 allow_stringref = F_ALLOW_STRINGREF 1378 allow_cycles = F_ALLOW_CYCLES
1379 pack_strings = F_PACK_STRINGS
1380 validate_utf8 = F_VALIDATE_UTF8
1110 PPCODE: 1381 PPCODE:
1111{ 1382{
1112 if (enable) 1383 if (enable)
1113 self->flags |= ix; 1384 self->flags |= ix;
1114 else 1385 else
1120void get_shrink (CBOR *self) 1391void get_shrink (CBOR *self)
1121 ALIAS: 1392 ALIAS:
1122 get_shrink = F_SHRINK 1393 get_shrink = F_SHRINK
1123 get_allow_unknown = F_ALLOW_UNKNOWN 1394 get_allow_unknown = F_ALLOW_UNKNOWN
1124 get_allow_sharing = F_ALLOW_SHARING 1395 get_allow_sharing = F_ALLOW_SHARING
1125 get_allow_stringref = F_ALLOW_STRINGREF 1396 get_allow_cycles = F_ALLOW_CYCLES
1397 get_pack_strings = F_PACK_STRINGS
1398 get_validate_utf8 = F_VALIDATE_UTF8
1126 PPCODE: 1399 PPCODE:
1127 XPUSHs (boolSV (self->flags & ix)); 1400 XPUSHs (boolSV (self->flags & ix));
1128 1401
1129void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) 1402void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
1130 PPCODE: 1403 PPCODE:
1143 XPUSHs (ST (0)); 1416 XPUSHs (ST (0));
1144 1417
1145int get_max_size (CBOR *self) 1418int get_max_size (CBOR *self)
1146 CODE: 1419 CODE:
1147 RETVAL = self->max_size; 1420 RETVAL = self->max_size;
1421 OUTPUT:
1422 RETVAL
1423
1424void filter (CBOR *self, SV *filter = 0)
1425 PPCODE:
1426 SvREFCNT_dec (self->filter);
1427 self->filter = filter ? newSVsv (filter) : filter;
1428 XPUSHs (ST (0));
1429
1430SV *get_filter (CBOR *self)
1431 CODE:
1432 RETVAL = self->filter ? self->filter : NEWSV (0, 0);
1148 OUTPUT: 1433 OUTPUT:
1149 RETVAL 1434 RETVAL
1150 1435
1151void encode (CBOR *self, SV *scalar) 1436void encode (CBOR *self, SV *scalar)
1152 PPCODE: 1437 PPCODE:
1167 EXTEND (SP, 2); 1452 EXTEND (SP, 2);
1168 PUSHs (sv); 1453 PUSHs (sv);
1169 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); 1454 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
1170} 1455}
1171 1456
1457void incr_parse (CBOR *self, SV *cborstr)
1458 ALIAS:
1459 incr_parse_multiple = 1
1460 PPCODE:
1461{
1462 if (SvUTF8 (cborstr))
1463 sv_utf8_downgrade (cborstr, 0);
1464
1465 if (!self->incr_count)
1466 {
1467 self->incr_count = newAV ();
1468 self->incr_pos = 0;
1469 self->incr_need = 1;
1470
1471 av_push (self->incr_count, newSViv (1));
1472 }
1473
1474 do
1475 {
1476 if (!incr_parse (self, cborstr))
1477 {
1478 if (self->incr_need > self->max_size && self->max_size)
1479 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
1480 (unsigned long)self->incr_need, (unsigned long)self->max_size);
1481
1482 break;
1483 }
1484
1485 SV *sv;
1486 char *offset;
1487
1488 PUTBACK; sv = decode_cbor (cborstr, self, &offset); SPAGAIN;
1489 XPUSHs (sv);
1490
1491 sv_chop (cborstr, offset);
1492
1493 av_clear (self->incr_count);
1494 av_push (self->incr_count, newSViv (1));
1495
1496 self->incr_pos = 0;
1497 self->incr_need = self->incr_pos + 1;
1498 }
1499 while (ix);
1500}
1501
1502void incr_reset (CBOR *self)
1503 CODE:
1504{
1505 SvREFCNT_dec (self->incr_count);
1506 self->incr_count = 0;
1507}
1508
1509void DESTROY (CBOR *self)
1510 PPCODE:
1511 cbor_free (self);
1512
1172PROTOTYPES: ENABLE 1513PROTOTYPES: ENABLE
1173 1514
1174void encode_cbor (SV *scalar) 1515void encode_cbor (SV *scalar)
1516 ALIAS:
1517 encode_cbor = 0
1518 encode_cbor_sharing = F_ALLOW_SHARING
1175 PPCODE: 1519 PPCODE:
1176{ 1520{
1177 CBOR cbor; 1521 CBOR cbor;
1178 cbor_init (&cbor); 1522 cbor_init (&cbor);
1523 cbor.flags |= ix;
1179 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN; 1524 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;
1180 XPUSHs (scalar); 1525 XPUSHs (scalar);
1181} 1526}
1182 1527
1183void decode_cbor (SV *cborstr) 1528void decode_cbor (SV *cborstr)

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines