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

Comparing cvsroot/CBOR-XS/XS.xs (file contents):
Revision 1.27 by root, Fri Nov 22 15:28:38 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
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)
… …
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;
90 SV *filter; 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.
91} CBOR; 139} CBOR;
92 140
93ecb_inline void 141ecb_inline void
94cbor_init (CBOR *cbor) 142cbor_init (CBOR *cbor)
95{ 143{
… …
99 147
100ecb_inline void 148ecb_inline void
101cbor_free (CBOR *cbor) 149cbor_free (CBOR *cbor)
102{ 150{
103 SvREFCNT_dec (cbor->filter); 151 SvREFCNT_dec (cbor->filter);
152 SvREFCNT_dec (cbor->incr_count);
104} 153}
105 154
106///////////////////////////////////////////////////////////////////////////// 155/////////////////////////////////////////////////////////////////////////////
107// utility functions 156// utility functions
108 157
… …
138{ 187{
139 return idx > 23 188 return idx > 23
140 ? idx > 0xffU 189 ? idx > 0xffU
141 ? idx > 0xffffU 190 ? idx > 0xffffU
142 ? idx > 0xffffffffU 191 ? idx > 0xffffffffU
192 ? 11
143 ? 7 193 : 7
144 : 6
145 : 5 194 : 5
146 : 4 195 : 4
147 : 3; 196 : 3;
148} 197}
149 198
… …
186static void 235static void
187encode_uint (enc_t *enc, int major, UV len) 236encode_uint (enc_t *enc, int major, UV len)
188{ 237{
189 need (enc, 9); 238 need (enc, 9);
190 239
191 if (len < 24) 240 if (ecb_expect_true (len < LENGTH_EXT1))
192 *enc->cur++ = major | len; 241 *enc->cur++ = major | len;
193 else if (len <= 0xff) 242 else if (ecb_expect_true (len <= 0xffU))
194 { 243 {
195 *enc->cur++ = major | 24; 244 *enc->cur++ = major | LENGTH_EXT1;
196 *enc->cur++ = len; 245 *enc->cur++ = len;
197 } 246 }
198 else if (len <= 0xffff) 247 else if (len <= 0xffffU)
199 { 248 {
200 *enc->cur++ = major | 25; 249 *enc->cur++ = major | LENGTH_EXT2;
201 *enc->cur++ = len >> 8; 250 *enc->cur++ = len >> 8;
202 *enc->cur++ = len; 251 *enc->cur++ = len;
203 } 252 }
204 else if (len <= 0xffffffff) 253 else if (len <= 0xffffffffU)
205 { 254 {
206 *enc->cur++ = major | 26; 255 *enc->cur++ = major | LENGTH_EXT4;
207 *enc->cur++ = len >> 24; 256 *enc->cur++ = len >> 24;
208 *enc->cur++ = len >> 16; 257 *enc->cur++ = len >> 16;
209 *enc->cur++ = len >> 8; 258 *enc->cur++ = len >> 8;
210 *enc->cur++ = len; 259 *enc->cur++ = len;
211 } 260 }
212 else 261 else
213 { 262 {
214 *enc->cur++ = major | 27; 263 *enc->cur++ = major | LENGTH_EXT8;
215 *enc->cur++ = len >> 56; 264 *enc->cur++ = len >> 56;
216 *enc->cur++ = len >> 48; 265 *enc->cur++ = len >> 48;
217 *enc->cur++ = len >> 40; 266 *enc->cur++ = len >> 40;
218 *enc->cur++ = len >> 32; 267 *enc->cur++ = len >> 32;
219 *enc->cur++ = len >> 24; 268 *enc->cur++ = len >> 24;
… …
224} 273}
225 274
226ecb_inline void 275ecb_inline void
227encode_tag (enc_t *enc, UV tag) 276encode_tag (enc_t *enc, UV tag)
228{ 277{
229 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;
230} 288}
231 289
232static void 290static void
233encode_str (enc_t *enc, int utf8, char *str, STRLEN len) 291encode_strref (enc_t *enc, int utf8, char *str, STRLEN len)
234{ 292{
235 if (ecb_expect_false (enc->cbor.flags & F_ALLOW_STRINGREF)) 293 if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS))
236 { 294 {
237 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1); 295 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1);
238 296
239 if (SvOK (*svp)) 297 if (SvOK (*svp))
240 { 298 {
241 // already registered, use stringref 299 // already registered, use stringref
242 encode_tag (enc, CBOR_TAG_STRINGREF); 300 encode_tag (enc, CBOR_TAG_STRINGREF);
243 encode_uint (enc, 0x00, SvUV (*svp)); 301 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
244 return; 302 return;
245 } 303 }
246 else if (len >= minimum_string_length (enc->stringref_idx)) 304 else if (len >= minimum_string_length (enc->stringref_idx))
247 { 305 {
248 // register only 306 // register only
249 sv_setuv (*svp, enc->stringref_idx); 307 sv_setuv (*svp, enc->stringref_idx);
250 ++enc->stringref_idx; 308 ++enc->stringref_idx;
251 } 309 }
252 } 310 }
253 311
254 encode_uint (enc, utf8 ? 0x60 : 0x40, len); 312 encode_str (enc, utf8, str, len);
255 need (enc, len);
256 memcpy (enc->cur, str, len);
257 enc->cur += len;
258} 313}
259 314
260static void encode_sv (enc_t *enc, SV *sv); 315static void encode_sv (enc_t *enc, SV *sv);
261 316
262static void 317static void
… …
267 if (enc->depth >= enc->cbor.max_depth) 322 if (enc->depth >= enc->cbor.max_depth)
268 croak (ERR_NESTING_EXCEEDED); 323 croak (ERR_NESTING_EXCEEDED);
269 324
270 ++enc->depth; 325 ++enc->depth;
271 326
272 encode_uint (enc, 0x80, len + 1); 327 encode_uint (enc, MAJOR_ARRAY, len + 1);
273 328
329 if (SvMAGICAL (av))
274 for (i = 0; i <= len; ++i) 330 for (i = 0; i <= len; ++i)
275 { 331 {
276 SV **svp = av_fetch (av, i, 0); 332 SV **svp = av_fetch (av, i, 0);
277 encode_sv (enc, svp ? *svp : &PL_sv_undef); 333 encode_sv (enc, svp ? *svp : &PL_sv_undef);
278 } 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 }
279 341
280 --enc->depth; 342 --enc->depth;
281} 343}
282 344
283static void 345static void
… …
292 354
293 int pairs = hv_iterinit (hv); 355 int pairs = hv_iterinit (hv);
294 int mg = SvMAGICAL (hv); 356 int mg = SvMAGICAL (hv);
295 357
296 if (mg) 358 if (mg)
297 encode_ch (enc, 0xa0 | 31); 359 encode_ch (enc, MAJOR_MAP | MINOR_INDEF);
298 else 360 else
299 encode_uint (enc, 0xa0, pairs); 361 encode_uint (enc, MAJOR_MAP, pairs);
300 362
301 while ((he = hv_iternext (hv))) 363 while ((he = hv_iternext (hv)))
302 { 364 {
303 if (HeKLEN (he) == HEf_SVKEY) 365 if (HeKLEN (he) == HEf_SVKEY)
304 encode_sv (enc, HeSVKEY (he)); 366 encode_sv (enc, HeSVKEY (he));
305 else 367 else
306 encode_str (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he)); 368 encode_strref (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he));
307 369
308 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));
309 } 371 }
310 372
311 if (mg) 373 if (mg)
312 encode_ch (enc, 0xe0 | 31); 374 encode_ch (enc, MAJOR_MISC | MINOR_INDEF);
313 375
314 --enc->depth; 376 --enc->depth;
315} 377}
316 378
317// encode objects, arrays and special \0=false and \1=true values. 379// encode objects, arrays and special \0=false and \1=true values.
318static void 380static void
319encode_rv (enc_t *enc, SV *sv) 381encode_rv (enc_t *enc, SV *sv)
320{ 382{
321 SvGETMAGIC (sv); 383 SvGETMAGIC (sv);
322
323 if (ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING)
324 && ecb_expect_false (SvREFCNT (sv) > 1))
325 {
326 if (!enc->shareable)
327 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
328
329 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
330
331 if (SvOK (*svp))
332 {
333 encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF);
334 encode_uint (enc, 0x00, SvUV (*svp));
335 return;
336 }
337 else
338 {
339 sv_setuv (*svp, enc->shareable_idx);
340 ++enc->shareable_idx;
341 encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE);
342 }
343 }
344 384
345 svtype svt = SvTYPE (sv); 385 svtype svt = SvTYPE (sv);
346 386
347 if (ecb_expect_false (SvOBJECT (sv))) 387 if (ecb_expect_false (SvOBJECT (sv)))
348 { 388 {
… …
355 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 395 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
356 ? cbor_tagged_stash 396 ? cbor_tagged_stash
357 : gv_stashpv ("CBOR::XS::Tagged" , 1); 397 : gv_stashpv ("CBOR::XS::Tagged" , 1);
358 398
359 HV *stash = SvSTASH (sv); 399 HV *stash = SvSTASH (sv);
360 GV *method;
361 400
362 if (stash == boolean_stash) 401 if (stash == boolean_stash)
363 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 }
364 else if (stash == error_stash) 406 else if (stash == error_stash)
365 encode_ch (enc, 0xe0 | 23); 407 {
408 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
409 return;
410 }
366 else if (stash == tagged_stash) 411 else if (stash == tagged_stash)
367 { 412 {
368 if (svt != SVt_PVAV) 413 if (svt != SVt_PVAV)
369 croak ("encountered CBOR::XS::Tagged object that isn't an array"); 414 croak ("encountered CBOR::XS::Tagged object that isn't an array");
370 415
371 encode_uint (enc, 0xc0, SvUV (*av_fetch ((AV *)sv, 0, 1))); 416 encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1)));
372 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))
373 } 432 {
433 encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF);
434 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
435 return;
436 }
437 else
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
374 else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0))) 450 if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
375 { 451 {
376 dSP; 452 dSP;
377 453
378 ENTER; SAVETMPS; PUSHMARK (SP); 454 ENTER; SAVETMPS;
455 PUSHMARK (SP);
379 // 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
380 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); 457 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
381 458
382 PUTBACK; 459 PUTBACK;
383 // G_SCALAR ensures that return value is 1 460 // G_SCALAR ensures that return value is 1
… …
396 } 473 }
397 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0) 474 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)
398 { 475 {
399 dSP; 476 dSP;
400 477
401 ENTER; SAVETMPS; PUSHMARK (SP); 478 ENTER; SAVETMPS;
479 SAVESTACK_POS ();
480 PUSHMARK (SP);
402 EXTEND (SP, 2); 481 EXTEND (SP, 2);
403 // we re-bless the reference to get overload and other niceties right 482 // we re-bless the reference to get overload and other niceties right
404 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); 483 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
405 PUSHs (sv_cbor); 484 PUSHs (sv_cbor);
406 485
… …
411 // catch this surprisingly common error 490 // catch this surprisingly common error
412 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv) 491 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)
413 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash)); 492 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));
414 493
415 encode_tag (enc, CBOR_TAG_PERL_OBJECT); 494 encode_tag (enc, CBOR_TAG_PERL_OBJECT);
416 encode_uint (enc, 0x80, count + 1); 495 encode_uint (enc, MAJOR_ARRAY, count + 1);
417 encode_str (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash)); 496 encode_strref (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
418 497
419 while (count) 498 while (count)
420 encode_sv (enc, SP[1 - count--]); 499 encode_sv (enc, SP[1 - count--]);
421 500
422 PUTBACK; 501 PUTBACK;
… …
443{ 522{
444 double nv = SvNVX (sv); 523 double nv = SvNVX (sv);
445 524
446 need (enc, 9); 525 need (enc, 9);
447 526
448 if (ecb_expect_false (nv == (U32)nv)) 527 if (ecb_expect_false (nv == (NV)(U32)nv))
449 encode_uint (enc, 0x00, (U32)nv); 528 encode_uint (enc, MAJOR_POS_INT, (U32)nv);
450 //TODO: maybe I32? 529 //TODO: maybe I32?
451 else if (ecb_expect_false (nv == (float)nv)) 530 else if (ecb_expect_false (nv == (float)nv))
452 { 531 {
453 uint32_t fp = ecb_float_to_binary32 (nv); 532 uint32_t fp = ecb_float_to_binary32 (nv);
454 533
455 *enc->cur++ = 0xe0 | 26; 534 *enc->cur++ = MAJOR_MISC | MISC_FLOAT32;
456 535
457 if (!ecb_big_endian ()) 536 if (!ecb_big_endian ())
458 fp = ecb_bswap32 (fp); 537 fp = ecb_bswap32 (fp);
459 538
460 memcpy (enc->cur, &fp, 4); 539 memcpy (enc->cur, &fp, 4);
… …
462 } 541 }
463 else 542 else
464 { 543 {
465 uint64_t fp = ecb_double_to_binary64 (nv); 544 uint64_t fp = ecb_double_to_binary64 (nv);
466 545
467 *enc->cur++ = 0xe0 | 27; 546 *enc->cur++ = MAJOR_MISC | MISC_FLOAT64;
468 547
469 if (!ecb_big_endian ()) 548 if (!ecb_big_endian ())
470 fp = ecb_bswap64 (fp); 549 fp = ecb_bswap64 (fp);
471 550
472 memcpy (enc->cur, &fp, 8); 551 memcpy (enc->cur, &fp, 8);
… …
481 560
482 if (SvPOKp (sv)) 561 if (SvPOKp (sv))
483 { 562 {
484 STRLEN len; 563 STRLEN len;
485 char *str = SvPV (sv, len); 564 char *str = SvPV (sv, len);
486 encode_str (enc, SvUTF8 (sv), str, len); 565 encode_strref (enc, SvUTF8 (sv), str, len);
487 } 566 }
488 else if (SvNOKp (sv)) 567 else if (SvNOKp (sv))
489 encode_nv (enc, sv); 568 encode_nv (enc, sv);
490 else if (SvIOKp (sv)) 569 else if (SvIOKp (sv))
491 { 570 {
492 if (SvIsUV (sv)) 571 if (SvIsUV (sv))
493 encode_uint (enc, 0x00, SvUVX (sv)); 572 encode_uint (enc, MAJOR_POS_INT, SvUVX (sv));
494 else if (SvIVX (sv) >= 0) 573 else if (SvIVX (sv) >= 0)
495 encode_uint (enc, 0x00, SvIVX (sv)); 574 encode_uint (enc, MAJOR_POS_INT, SvIVX (sv));
496 else 575 else
497 encode_uint (enc, 0x20, -(SvIVX (sv) + 1)); 576 encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1));
498 } 577 }
499 else if (SvROK (sv)) 578 else if (SvROK (sv))
500 encode_rv (enc, SvRV (sv)); 579 encode_rv (enc, SvRV (sv));
501 else if (!SvOK (sv)) 580 else if (!SvOK (sv))
502 encode_ch (enc, 0xe0 | 22); 581 encode_ch (enc, MAJOR_MISC | SIMPLE_NULL);
503 else if (enc->cbor.flags & F_ALLOW_UNKNOWN) 582 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
504 encode_ch (enc, 0xe0 | 23); 583 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
505 else 584 else
506 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",
507 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv)); 586 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
508} 587}
509 588
510static SV * 589static SV *
511encode_cbor (SV *scalar, CBOR *cbor) 590encode_cbor (SV *scalar, CBOR *cbor)
512{ 591{
513 enc_t enc = { }; 592 enc_t enc = { 0 };
514 593
515 enc.cbor = *cbor; 594 enc.cbor = *cbor;
516 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 595 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
517 enc.cur = SvPVX (enc.sv); 596 enc.cur = SvPVX (enc.sv);
518 enc.end = SvEND (enc.sv); 597 enc.end = SvEND (enc.sv);
519 598
520 SvPOK_only (enc.sv); 599 SvPOK_only (enc.sv);
521 600
522 if (cbor->flags & F_ALLOW_STRINGREF) 601 if (cbor->flags & F_PACK_STRINGS)
523 { 602 {
524 encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE); 603 encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE);
525 enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ()); 604 enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ());
526 enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ()); 605 enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ());
527 } 606 }
… …
562#define DEC_DEC_DEPTH --dec->depth 641#define DEC_DEC_DEPTH --dec->depth
563 642
564static UV 643static UV
565decode_uint (dec_t *dec) 644decode_uint (dec_t *dec)
566{ 645{
567 switch (*dec->cur & 31) 646 U8 m = *dec->cur & MINOR_MASK;
568 { 647 ++dec->cur;
569 case 0: case 1: case 2: case 3: case 4: case 5: case 6: case 7:
570 case 8: case 9: case 10: case 11: case 12: case 13: case 14: case 15:
571 case 16: case 17: case 18: case 19: case 20: case 21: case 22: case 23:
572 return *dec->cur++ & 31;
573 648
574 case 24: 649 if (ecb_expect_true (m < LENGTH_EXT1))
650 return m;
651 else if (ecb_expect_true (m == LENGTH_EXT1))
652 {
575 WANT (2); 653 WANT (1);
576 dec->cur += 2; 654 dec->cur += 1;
577 return dec->cur[-1]; 655 return dec->cur[-1];
578 656 }
579 case 25: 657 else if (ecb_expect_true (m == LENGTH_EXT2))
658 {
580 WANT (3); 659 WANT (2);
581 dec->cur += 3; 660 dec->cur += 2;
582 return (((UV)dec->cur[-2]) << 8) 661 return (((UV)dec->cur[-2]) << 8)
583 | ((UV)dec->cur[-1]); 662 | ((UV)dec->cur[-1]);
584 663 }
585 case 26: 664 else if (ecb_expect_true (m == LENGTH_EXT4))
665 {
586 WANT (5); 666 WANT (4);
587 dec->cur += 5; 667 dec->cur += 4;
588 return (((UV)dec->cur[-4]) << 24) 668 return (((UV)dec->cur[-4]) << 24)
589 | (((UV)dec->cur[-3]) << 16) 669 | (((UV)dec->cur[-3]) << 16)
590 | (((UV)dec->cur[-2]) << 8) 670 | (((UV)dec->cur[-2]) << 8)
591 | ((UV)dec->cur[-1]); 671 | ((UV)dec->cur[-1]);
592 672 }
593 case 27: 673 else if (ecb_expect_true (m == LENGTH_EXT8))
674 {
594 WANT (9); 675 WANT (8);
595 dec->cur += 9; 676 dec->cur += 8;
677
678 return
679#if UVSIZE < 8
680 0
681#else
596 return (((UV)dec->cur[-8]) << 56) 682 (((UV)dec->cur[-8]) << 56)
597 | (((UV)dec->cur[-7]) << 48) 683 | (((UV)dec->cur[-7]) << 48)
598 | (((UV)dec->cur[-6]) << 40) 684 | (((UV)dec->cur[-6]) << 40)
599 | (((UV)dec->cur[-5]) << 32) 685 | (((UV)dec->cur[-5]) << 32)
686#endif
600 | (((UV)dec->cur[-4]) << 24) 687 | (((UV)dec->cur[-4]) << 24)
601 | (((UV)dec->cur[-3]) << 16) 688 | (((UV)dec->cur[-3]) << 16)
602 | (((UV)dec->cur[-2]) << 8) 689 | (((UV)dec->cur[-2]) << 8)
603 | ((UV)dec->cur[-1]); 690 | ((UV)dec->cur[-1]);
604 691 }
605 default: 692 else
606 ERR ("corrupted CBOR data (unsupported integer minor encoding)"); 693 ERR ("corrupted CBOR data (unsupported integer minor encoding)");
607 }
608 694
609fail: 695fail:
610 return 0; 696 return 0;
611} 697}
612 698
… …
617{ 703{
618 AV *av = newAV (); 704 AV *av = newAV ();
619 705
620 DEC_INC_DEPTH; 706 DEC_INC_DEPTH;
621 707
622 if ((*dec->cur & 31) == 31) 708 if (*dec->cur == (MAJOR_ARRAY | MINOR_INDEF))
623 { 709 {
624 ++dec->cur; 710 ++dec->cur;
625 711
626 for (;;) 712 for (;;)
627 { 713 {
628 WANT (1); 714 WANT (1);
629 715
630 if (*dec->cur == (0xe0 | 31)) 716 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
631 { 717 {
632 ++dec->cur; 718 ++dec->cur;
633 break; 719 break;
634 } 720 }
635 721
… …
638 } 724 }
639 else 725 else
640 { 726 {
641 int i, len = decode_uint (dec); 727 int i, len = decode_uint (dec);
642 728
729 WANT (len); // complexity check for av_fill - need at least one byte per value, do not allow supersize arrays
643 av_fill (av, len - 1); 730 av_fill (av, len - 1);
644 731
645 for (i = 0; i < len; ++i) 732 for (i = 0; i < len; ++i)
646 AvARRAY (av)[i] = decode_sv (dec); 733 AvARRAY (av)[i] = decode_sv (dec);
647 } 734 }
… …
660{ 747{
661 // for speed reasons, we specialcase single-string 748 // for speed reasons, we specialcase single-string
662 // byte or utf-8 strings as keys, but only when !stringref 749 // byte or utf-8 strings as keys, but only when !stringref
663 750
664 if (ecb_expect_true (!dec->stringref)) 751 if (ecb_expect_true (!dec->stringref))
665 if (*dec->cur >= 0x40 && *dec->cur <= 0x40 + 27) 752 if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8))
666 { 753 {
667 I32 len = decode_uint (dec); 754 I32 len = decode_uint (dec);
668 char *key = (char *)dec->cur; 755 char *key = (char *)dec->cur;
669 756
757 WANT (len);
670 dec->cur += len; 758 dec->cur += len;
671 759
672 if (ecb_expect_false (dec->stringref))
673 av_push (dec->stringref, newSVpvn (key, len));
674
675 hv_store (hv, key, len, decode_sv (dec), 0); 760 hv_store (hv, key, len, decode_sv (dec), 0);
676 761
677 return; 762 return;
678 } 763 }
679 else if (*dec->cur >= 0x60 && *dec->cur <= 0x60 + 27) 764 else if (ecb_expect_true ((U8)(*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8))
680 { 765 {
681 I32 len = decode_uint (dec); 766 I32 len = decode_uint (dec);
682 char *key = (char *)dec->cur; 767 char *key = (char *)dec->cur;
683 768
769 WANT (len);
684 dec->cur += len; 770 dec->cur += len;
685 771
686 if (ecb_expect_false (dec->stringref)) 772 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
687 av_push (dec->stringref, newSVpvn_utf8 (key, len, 1)); 773 if (!is_utf8_string (key, len))
774 ERR ("corrupted CBOR data (invalid UTF-8 in map key)");
688 775
689 hv_store (hv, key, -len, decode_sv (dec), 0); 776 hv_store (hv, key, -len, decode_sv (dec), 0);
690 777
691 return; 778 return;
692 } 779 }
… …
694 SV *k = decode_sv (dec); 781 SV *k = decode_sv (dec);
695 SV *v = decode_sv (dec); 782 SV *v = decode_sv (dec);
696 783
697 hv_store_ent (hv, k, v, 0); 784 hv_store_ent (hv, k, v, 0);
698 SvREFCNT_dec (k); 785 SvREFCNT_dec (k);
786
787fail:
788 ;
699} 789}
700 790
701static SV * 791static SV *
702decode_hv (dec_t *dec) 792decode_hv (dec_t *dec)
703{ 793{
704 HV *hv = newHV (); 794 HV *hv = newHV ();
705 795
706 DEC_INC_DEPTH; 796 DEC_INC_DEPTH;
707 797
708 if ((*dec->cur & 31) == 31) 798 if (*dec->cur == (MAJOR_MAP | MINOR_INDEF))
709 { 799 {
710 ++dec->cur; 800 ++dec->cur;
711 801
712 for (;;) 802 for (;;)
713 { 803 {
714 WANT (1); 804 WANT (1);
715 805
716 if (*dec->cur == (0xe0 | 31)) 806 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
717 { 807 {
718 ++dec->cur; 808 ++dec->cur;
719 break; 809 break;
720 } 810 }
721 811
… …
742static SV * 832static SV *
743decode_str (dec_t *dec, int utf8) 833decode_str (dec_t *dec, int utf8)
744{ 834{
745 SV *sv = 0; 835 SV *sv = 0;
746 836
747 if ((*dec->cur & 31) == 31) 837 if ((*dec->cur & MINOR_MASK) == MINOR_INDEF)
748 { 838 {
839 // indefinite length strings
749 ++dec->cur; 840 ++dec->cur;
750 841
842 U8 major = *dec->cur & MAJOR_MISC;
843
751 sv = newSVpvn ("", 0); 844 sv = newSVpvn ("", 0);
752 845
753 // not very fast, and certainly not robust against illegal input
754 for (;;) 846 for (;;)
755 { 847 {
756 WANT (1); 848 WANT (1);
757 849
758 if (*dec->cur == (0xe0 | 31)) 850 if ((*dec->cur - major) > LENGTH_EXT8)
851 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
759 { 852 {
760 ++dec->cur; 853 ++dec->cur;
761 break; 854 break;
762 } 855 }
856 else
857 ERR ("corrupted CBOR data (invalid chunks in indefinite length string)");
763 858
764 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;
765 } 864 }
766 } 865 }
767 else 866 else
768 { 867 {
769 STRLEN len = decode_uint (dec); 868 STRLEN len = decode_uint (dec);
… …
776 && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1)) 875 && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1))
777 av_push (dec->stringref, SvREFCNT_inc_NN (sv)); 876 av_push (dec->stringref, SvREFCNT_inc_NN (sv));
778 } 877 }
779 878
780 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
781 SvUTF8_on (sv); 885 SvUTF8_on (sv);
886 }
782 887
783 return sv; 888 return sv;
784 889
785fail: 890fail:
786 SvREFCNT_dec (sv); 891 SvREFCNT_dec (sv);
… …
818 } 923 }
819 break; 924 break;
820 925
821 case CBOR_TAG_STRINGREF: 926 case CBOR_TAG_STRINGREF:
822 { 927 {
823 if ((*dec->cur >> 5) != 0) 928 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
824 ERR ("corrupted CBOR data (stringref index not an unsigned integer)"); 929 ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
825 930
826 UV idx = decode_uint (dec); 931 UV idx = decode_uint (dec);
827 932
828 if (!dec->stringref || (int)idx > AvFILLp (dec->stringref)) 933 if (!dec->stringref || (int)idx > AvFILLp (dec->stringref))
… …
835 case CBOR_TAG_VALUE_SHAREABLE: 940 case CBOR_TAG_VALUE_SHAREABLE:
836 { 941 {
837 if (ecb_expect_false (!dec->shareable)) 942 if (ecb_expect_false (!dec->shareable))
838 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ()); 943 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ());
839 944
945 if (dec->cbor.flags & F_ALLOW_CYCLES)
946 {
840 sv = newSV (0); 947 sv = newSV (0);
841 av_push (dec->shareable, SvREFCNT_inc_NN (sv)); 948 av_push (dec->shareable, SvREFCNT_inc_NN (sv));
842 949
843 SV *osv = decode_sv (dec); 950 SV *osv = decode_sv (dec);
844 sv_setsv (sv, osv); 951 sv_setsv (sv, osv);
845 SvREFCNT_dec_NN (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 }
846 } 961 }
847 break; 962 break;
848 963
849 case CBOR_TAG_VALUE_SHAREDREF: 964 case CBOR_TAG_VALUE_SHAREDREF:
850 { 965 {
851 if ((*dec->cur >> 5) != 0) 966 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
852 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)"); 967 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
853 968
854 UV idx = decode_uint (dec); 969 UV idx = decode_uint (dec);
855 970
856 if (!dec->shareable || (int)idx > AvFILLp (dec->shareable)) 971 if (!dec->shareable || (int)idx > AvFILLp (dec->shareable))
857 ERR ("corrupted CBOR data (sharedref index out of bounds)"); 972 ERR ("corrupted CBOR data (sharedref index out of bounds)");
858 973
859 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]); 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");
860 } 978 }
861 break; 979 break;
862 980
863 case CBOR_TAG_PERL_OBJECT: 981 case CBOR_TAG_PERL_OBJECT:
864 { 982 {
… …
879 if (!method) 997 if (!method)
880 ERR ("cannot decode perl-object (package does not have a THAW method)"); 998 ERR ("cannot decode perl-object (package does not have a THAW method)");
881 999
882 dSP; 1000 dSP;
883 1001
884 ENTER; SAVETMPS; PUSHMARK (SP); 1002 ENTER; SAVETMPS;
1003 PUSHMARK (SP);
885 EXTEND (SP, len + 1); 1004 EXTEND (SP, len + 1);
886 // we re-bless the reference to get overload and other niceties right 1005 // we re-bless the reference to get overload and other niceties right
887 PUSHs (*av_fetch (av, 0, 1)); 1006 PUSHs (*av_fetch (av, 0, 1));
888 PUSHs (sv_cbor); 1007 PUSHs (sv_cbor);
889 1008
… …
914 default: 1033 default:
915 { 1034 {
916 sv = decode_sv (dec); 1035 sv = decode_sv (dec);
917 1036
918 dSP; 1037 dSP;
919 ENTER; SAVETMPS; PUSHMARK (SP); 1038 ENTER; SAVETMPS;
1039 SAVESTACK_POS ();
1040 PUSHMARK (SP);
920 EXTEND (SP, 2); 1041 EXTEND (SP, 2);
921 PUSHs (newSVuv (tag)); 1042 PUSHs (newSVuv (tag));
922 PUSHs (sv); 1043 PUSHs (sv);
923 1044
924 PUTBACK; 1045 PUTBACK;
… …
965static SV * 1086static SV *
966decode_sv (dec_t *dec) 1087decode_sv (dec_t *dec)
967{ 1088{
968 WANT (1); 1089 WANT (1);
969 1090
970 switch (*dec->cur >> 5) 1091 switch (*dec->cur >> MAJOR_SHIFT)
971 { 1092 {
972 case 0: // unsigned int 1093 case MAJOR_POS_INT >> MAJOR_SHIFT: return newSVuv (decode_uint (dec));
973 return newSVuv (decode_uint (dec)); 1094 case MAJOR_NEG_INT >> MAJOR_SHIFT: return newSViv (-1 - (IV)decode_uint (dec));
974 case 1: // negative int 1095 case MAJOR_BYTES >> MAJOR_SHIFT: return decode_str (dec, 0);
975 return newSViv (-1 - (IV)decode_uint (dec)); 1096 case MAJOR_TEXT >> MAJOR_SHIFT: return decode_str (dec, 1);
976 case 2: // octet string 1097 case MAJOR_ARRAY >> MAJOR_SHIFT: return decode_av (dec);
977 return decode_str (dec, 0); 1098 case MAJOR_MAP >> MAJOR_SHIFT: return decode_hv (dec);
978 case 3: // utf-8 string 1099 case MAJOR_TAG >> MAJOR_SHIFT: return decode_tagged (dec);
979 return decode_str (dec, 1); 1100
980 case 4: // array 1101 case MAJOR_MISC >> MAJOR_SHIFT:
981 return decode_av (dec);
982 case 5: // map
983 return decode_hv (dec);
984 case 6: // tag
985 return decode_tagged (dec);
986 case 7: // misc
987 switch (*dec->cur++ & 31) 1102 switch (*dec->cur++ & MINOR_MASK)
988 { 1103 {
989 case 20: 1104 case SIMPLE_FALSE:
990#if CBOR_SLOW 1105#if CBOR_SLOW
991 types_false = get_bool ("Types::Serialiser::false"); 1106 types_false = get_bool ("Types::Serialiser::false");
992#endif 1107#endif
993 return newSVsv (types_false); 1108 return newSVsv (types_false);
994 case 21: 1109 case SIMPLE_TRUE:
995#if CBOR_SLOW 1110#if CBOR_SLOW
996 types_true = get_bool ("Types::Serialiser::true"); 1111 types_true = get_bool ("Types::Serialiser::true");
997#endif 1112#endif
998 return newSVsv (types_true); 1113 return newSVsv (types_true);
999 case 22: 1114 case SIMPLE_NULL:
1000 return newSVsv (&PL_sv_undef); 1115 return newSVsv (&PL_sv_undef);
1001 case 23: 1116 case SIMPLE_UNDEF:
1002#if CBOR_SLOW 1117#if CBOR_SLOW
1003 types_error = get_bool ("Types::Serialiser::error"); 1118 types_error = get_bool ("Types::Serialiser::error");
1004#endif 1119#endif
1005 return newSVsv (types_error); 1120 return newSVsv (types_error);
1006 1121
1007 case 25: 1122 case MISC_FLOAT16:
1008 { 1123 {
1009 WANT (2); 1124 WANT (2);
1010 1125
1011 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1]; 1126 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1];
1012 dec->cur += 2; 1127 dec->cur += 2;
1013 1128
1014 return newSVnv (ecb_binary16_to_float (fp)); 1129 return newSVnv (ecb_binary16_to_float (fp));
1015 } 1130 }
1016 1131
1017 case 26: 1132 case MISC_FLOAT32:
1018 { 1133 {
1019 uint32_t fp; 1134 uint32_t fp;
1020 WANT (4); 1135 WANT (4);
1021 memcpy (&fp, dec->cur, 4); 1136 memcpy (&fp, dec->cur, 4);
1022 dec->cur += 4; 1137 dec->cur += 4;
… …
1025 fp = ecb_bswap32 (fp); 1140 fp = ecb_bswap32 (fp);
1026 1141
1027 return newSVnv (ecb_binary32_to_float (fp)); 1142 return newSVnv (ecb_binary32_to_float (fp));
1028 } 1143 }
1029 1144
1030 case 27: 1145 case MISC_FLOAT64:
1031 { 1146 {
1032 uint64_t fp; 1147 uint64_t fp;
1033 WANT (8); 1148 WANT (8);
1034 memcpy (&fp, dec->cur, 8); 1149 memcpy (&fp, dec->cur, 8);
1035 dec->cur += 8; 1150 dec->cur += 8;
… …
1038 fp = ecb_bswap64 (fp); 1153 fp = ecb_bswap64 (fp);
1039 1154
1040 return newSVnv (ecb_binary64_to_double (fp)); 1155 return newSVnv (ecb_binary64_to_double (fp));
1041 } 1156 }
1042 1157
1043 // 0..19 unassigned 1158 // 0..19 unassigned simple
1044 // 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
1045 default: 1162 default:
1046 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)"); 1163 ERR ("corrupted CBOR data (reserved/unassigned/unexpected major 7 value)");
1047 } 1164 }
1048 1165
1049 break; 1166 break;
1050 } 1167 }
1051 1168
… …
1054} 1171}
1055 1172
1056static SV * 1173static SV *
1057decode_cbor (SV *string, CBOR *cbor, char **offset_return) 1174decode_cbor (SV *string, CBOR *cbor, char **offset_return)
1058{ 1175{
1059 dec_t dec = { }; 1176 dec_t dec = { 0 };
1060 SV *sv; 1177 SV *sv;
1061 STRLEN len; 1178 STRLEN len;
1062 char *data = SvPVbyte (string, len); 1179 char *data = SvPVbyte (string, len);
1063 1180
1064 if (len > cbor->max_size && cbor->max_size) 1181 if (len > cbor->max_size && cbor->max_size)
… …
1078 if (dec.cur != dec.end && !dec.err) 1195 if (dec.cur != dec.end && !dec.err)
1079 dec.err = "garbage after CBOR object"; 1196 dec.err = "garbage after CBOR object";
1080 1197
1081 if (dec.err) 1198 if (dec.err)
1082 { 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
1083 SvREFCNT_dec (sv); 1211 SvREFCNT_dec (sv);
1084 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (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);
1085 } 1213 }
1086 1214
1087 sv = sv_2mortal (sv); 1215 sv = sv_2mortal (sv);
1088 1216
1089 return sv; 1217 return sv;
1090} 1218}
1091 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
1092///////////////////////////////////////////////////////////////////////////// 1344/////////////////////////////////////////////////////////////////////////////
1093// XS interface functions 1345// XS interface functions
1094 1346
1095MODULE = CBOR::XS PACKAGE = CBOR::XS 1347MODULE = CBOR::XS PACKAGE = CBOR::XS
1096 1348
… …
1136void shrink (CBOR *self, int enable = 1) 1388void shrink (CBOR *self, int enable = 1)
1137 ALIAS: 1389 ALIAS:
1138 shrink = F_SHRINK 1390 shrink = F_SHRINK
1139 allow_unknown = F_ALLOW_UNKNOWN 1391 allow_unknown = F_ALLOW_UNKNOWN
1140 allow_sharing = F_ALLOW_SHARING 1392 allow_sharing = F_ALLOW_SHARING
1141 allow_stringref = F_ALLOW_STRINGREF 1393 allow_cycles = F_ALLOW_CYCLES
1394 pack_strings = F_PACK_STRINGS
1395 validate_utf8 = F_VALIDATE_UTF8
1142 PPCODE: 1396 PPCODE:
1143{ 1397{
1144 if (enable) 1398 if (enable)
1145 self->flags |= ix; 1399 self->flags |= ix;
1146 else 1400 else
… …
1152void get_shrink (CBOR *self) 1406void get_shrink (CBOR *self)
1153 ALIAS: 1407 ALIAS:
1154 get_shrink = F_SHRINK 1408 get_shrink = F_SHRINK
1155 get_allow_unknown = F_ALLOW_UNKNOWN 1409 get_allow_unknown = F_ALLOW_UNKNOWN
1156 get_allow_sharing = F_ALLOW_SHARING 1410 get_allow_sharing = F_ALLOW_SHARING
1157 get_allow_stringref = F_ALLOW_STRINGREF 1411 get_allow_cycles = F_ALLOW_CYCLES
1412 get_pack_strings = F_PACK_STRINGS
1413 get_validate_utf8 = F_VALIDATE_UTF8
1158 PPCODE: 1414 PPCODE:
1159 XPUSHs (boolSV (self->flags & ix)); 1415 XPUSHs (boolSV (self->flags & ix));
1160 1416
1161void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) 1417void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
1162 PPCODE: 1418 PPCODE:
… …
1211 EXTEND (SP, 2); 1467 EXTEND (SP, 2);
1212 PUSHs (sv); 1468 PUSHs (sv);
1213 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); 1469 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
1214} 1470}
1215 1471
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}
1523
1216void DESTROY (CBOR *self) 1524void DESTROY (CBOR *self)
1217 PPCODE: 1525 PPCODE:
1218 cbor_free (self); 1526 cbor_free (self);
1219 1527
1220PROTOTYPES: ENABLE 1528PROTOTYPES: ENABLE
1221 1529
1222void encode_cbor (SV *scalar) 1530void encode_cbor (SV *scalar)
1531 ALIAS:
1532 encode_cbor = 0
1533 encode_cbor_sharing = F_ALLOW_SHARING
1223 PPCODE: 1534 PPCODE:
1224{ 1535{
1225 CBOR cbor; 1536 CBOR cbor;
1226 cbor_init (&cbor); 1537 cbor_init (&cbor);
1538 cbor.flags |= ix;
1227 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN; 1539 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;
1228 XPUSHs (scalar); 1540 XPUSHs (scalar);
1229} 1541}
1230 1542
1231void decode_cbor (SV *cborstr) 1543void decode_cbor (SV *cborstr)

Diff Legend

– Removed lines
+ Added lines
< Changed lines
> Changed lines