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 1.33 by root, Sat Nov 30 15:23:59 2013 UTC vs.
Revision 1.51 by root, Sun Apr 24 13:15:19 2016 UTC

… …
7#include <stdlib.h> 7#include <stdlib.h>
8#include <stdio.h> 8#include <stdio.h>
9#include <limits.h> 9#include <limits.h>
10#include <float.h> 10#include <float.h>
11 11
12#define ECB_NO_THREADS 1
12#include "ecb.h" 13#include "ecb.h"
13 14
14// 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))
… …
23#endif 24#endif
24#ifndef SvREFCNT_dec_NN 25#ifndef SvREFCNT_dec_NN
25# define SvREFCNT_dec_NN(sv) SvREFCNT_dec (sv) 26# define SvREFCNT_dec_NN(sv) SvREFCNT_dec (sv)
26#endif 27#endif
27 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
28// known tags 66// known tags
29enum cbor_tag 67enum cbor_tag
30{ 68{
31 // extensions 69 // extensions
32 CBOR_TAG_STRINGREF = 25, // http://cbor.schmorp.de/stringref 70 CBOR_TAG_STRINGREF = 25, // http://cbor.schmorp.de/stringref
33 CBOR_TAG_PERL_OBJECT = 26, // http://cbor.schmorp.de/perl-object 71 CBOR_TAG_PERL_OBJECT = 26, // http://cbor.schmorp.de/perl-object
34 CBOR_TAG_GENERIC_OBJECT = 27, // http://cbor.schmorp.de/generic-object 72 CBOR_TAG_GENERIC_OBJECT = 27, // http://cbor.schmorp.de/generic-object
35 CBOR_TAG_VALUE_SHAREABLE = 28, // http://cbor.schmorp.de/value-sharing 73 CBOR_TAG_VALUE_SHAREABLE = 28, // http://cbor.schmorp.de/value-sharing
36 CBOR_TAG_VALUE_SHAREDREF = 29, // http://cbor.schmorp.de/value-sharing 74 CBOR_TAG_VALUE_SHAREDREF = 29, // http://cbor.schmorp.de/value-sharing
37 CBOR_TAG_STRINGREF_NAMESPACE = 256, // http://cbor.schmorp.de/stringref 75 CBOR_TAG_STRINGREF_NAMESPACE = 256, // http://cbor.schmorp.de/stringref
38 CBOR_TAG_INDIRECTION = 22098, // http://cbor.schmorp.de/indirection 76 CBOR_TAG_INDIRECTION = 22098, // http://cbor.schmorp.de/indirection
39 77
40 // rfc7049 78 // rfc7049
41 CBOR_TAG_DATETIME = 0, // rfc4287, utf-8 79 CBOR_TAG_DATETIME = 0, // rfc4287, utf-8
42 CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any 80 CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any
43 CBOR_TAG_POS_BIGNUM = 2, // byte string 81 CBOR_TAG_POS_BIGNUM = 2, // byte string
44 CBOR_TAG_NEG_BIGNUM = 3, // byte string 82 CBOR_TAG_NEG_BIGNUM = 3, // byte string
45 CBOR_TAG_DECIMAL = 4, // decimal fraction, array 83 CBOR_TAG_DECIMAL = 4, // decimal fraction, array
46 CBOR_TAG_BIGFLOAT = 5, // array 84 CBOR_TAG_BIGFLOAT = 5, // array
47 85
48 CBOR_TAG_CONV_B64U = 21, // base64url, any 86 CBOR_TAG_CONV_B64U = 21, // base64url, any
49 CBOR_TAG_CONV_B64 = 22, // base64, any 87 CBOR_TAG_CONV_B64 = 22, // base64, any
50 CBOR_TAG_CONV_HEX = 23, // base16, any 88 CBOR_TAG_CONV_HEX = 23, // base16, any
51 CBOR_TAG_CBOR = 24, // embedded cbor, byte string 89 CBOR_TAG_CBOR = 24, // embedded cbor, byte string
52 90
53 CBOR_TAG_URI = 32, // URI rfc3986, utf-8 91 CBOR_TAG_URI = 32, // URI rfc3986, utf-8
54 CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8 92 CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8
55 CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8 93 CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8
56 CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8 94 CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8
57 CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8 95 CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8
58 96
59 CBOR_TAG_MAGIC = 55799 // self-describe cbor 97 CBOR_TAG_MAGIC = 55799, // self-describe cbor
60}; 98};
61 99
62#define F_SHRINK 0x00000001UL 100#define F_SHRINK 0x00000001UL
63#define F_ALLOW_UNKNOWN 0x00000002UL 101#define F_ALLOW_UNKNOWN 0x00000002UL
64#define F_ALLOW_SHARING 0x00000004UL 102#define F_ALLOW_SHARING 0x00000004UL
103#define F_ALLOW_CYCLES 0x00000008UL
65#define F_PACK_STRINGS 0x00000008UL 104#define F_PACK_STRINGS 0x00000010UL
105#define F_UTF8_STRINGS 0x00000020UL
106#define F_VALIDATE_UTF8 0x00000040UL
66 107
67#define INIT_SIZE 32 // initial scalar size to be allocated 108#define INIT_SIZE 32 // initial scalar size to be allocated
68 109
69#define SB do { 110#define SB do {
70#define SE } while (0) 111#define SE } while (0)
… …
89typedef struct { 130typedef struct {
90 U32 flags; 131 U32 flags;
91 U32 max_depth; 132 U32 max_depth;
92 STRLEN max_size; 133 STRLEN max_size;
93 SV *filter; 134 SV *filter;
135
136 // for the incremental parser
137 STRLEN incr_pos; // the current offset into the text
138 STRLEN incr_need; // minimum bytes needed to decode
139 AV *incr_count; // for every nesting level, the number of outstanding values, or -1 for indef.
94} CBOR; 140} CBOR;
95 141
96ecb_inline void 142ecb_inline void
97cbor_init (CBOR *cbor) 143cbor_init (CBOR *cbor)
98{ 144{
… …
102 148
103ecb_inline void 149ecb_inline void
104cbor_free (CBOR *cbor) 150cbor_free (CBOR *cbor)
105{ 151{
106 SvREFCNT_dec (cbor->filter); 152 SvREFCNT_dec (cbor->filter);
153 SvREFCNT_dec (cbor->incr_count);
107} 154}
108 155
109///////////////////////////////////////////////////////////////////////////// 156/////////////////////////////////////////////////////////////////////////////
110// utility functions 157// utility functions
111 158
… …
189static void 236static void
190encode_uint (enc_t *enc, int major, UV len) 237encode_uint (enc_t *enc, int major, UV len)
191{ 238{
192 need (enc, 9); 239 need (enc, 9);
193 240
194 if (ecb_expect_true (len < 24)) 241 if (ecb_expect_true (len < LENGTH_EXT1))
195 *enc->cur++ = major | len; 242 *enc->cur++ = major | len;
196 else if (ecb_expect_true (len <= 0xff)) 243 else if (ecb_expect_true (len <= 0xffU))
197 { 244 {
198 *enc->cur++ = major | 24; 245 *enc->cur++ = major | LENGTH_EXT1;
199 *enc->cur++ = len; 246 *enc->cur++ = len;
200 } 247 }
201 else if (len <= 0xffff) 248 else if (len <= 0xffffU)
202 { 249 {
203 *enc->cur++ = major | 25; 250 *enc->cur++ = major | LENGTH_EXT2;
204 *enc->cur++ = len >> 8; 251 *enc->cur++ = len >> 8;
205 *enc->cur++ = len; 252 *enc->cur++ = len;
206 } 253 }
207 else if (len <= 0xffffffff) 254 else if (len <= 0xffffffffU)
208 { 255 {
209 *enc->cur++ = major | 26; 256 *enc->cur++ = major | LENGTH_EXT4;
210 *enc->cur++ = len >> 24; 257 *enc->cur++ = len >> 24;
211 *enc->cur++ = len >> 16; 258 *enc->cur++ = len >> 16;
212 *enc->cur++ = len >> 8; 259 *enc->cur++ = len >> 8;
213 *enc->cur++ = len; 260 *enc->cur++ = len;
214 } 261 }
215 else 262 else
216 { 263 {
217 *enc->cur++ = major | 27; 264 *enc->cur++ = major | LENGTH_EXT8;
218 *enc->cur++ = len >> 56; 265 *enc->cur++ = len >> 56;
219 *enc->cur++ = len >> 48; 266 *enc->cur++ = len >> 48;
220 *enc->cur++ = len >> 40; 267 *enc->cur++ = len >> 40;
221 *enc->cur++ = len >> 32; 268 *enc->cur++ = len >> 32;
222 *enc->cur++ = len >> 24; 269 *enc->cur++ = len >> 24;
… …
227} 274}
228 275
229ecb_inline void 276ecb_inline void
230encode_tag (enc_t *enc, UV tag) 277encode_tag (enc_t *enc, UV tag)
231{ 278{
232 encode_uint (enc, 0xc0, tag); 279 encode_uint (enc, MAJOR_TAG, tag);
233} 280}
234 281
235ecb_inline void 282ecb_inline void
236encode_str (enc_t *enc, int utf8, char *str, STRLEN len) 283encode_str (enc_t *enc, int utf8, char *str, STRLEN len)
237{ 284{
238 encode_uint (enc, utf8 ? 0x60 : 0x40, len); 285 if (ecb_expect_false (enc->cbor.flags & F_UTF8_STRINGS))
286 if (!utf8)
287 {
288 SV *sv = sv_newmortal ();
289 char *s; STRLEN l;
290
291 sv_setpvn (sv, str, len);
292
293 s = SvPVutf8 (sv, l);
294 encode_str (enc, 1, s, l);
295 return;
296 }
297
298 encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len);
239 need (enc, len); 299 need (enc, len);
240 memcpy (enc->cur, str, len); 300 memcpy (enc->cur, str, len);
241 enc->cur += len; 301 enc->cur += len;
242} 302}
243 303
… …
250 310
251 if (SvOK (*svp)) 311 if (SvOK (*svp))
252 { 312 {
253 // already registered, use stringref 313 // already registered, use stringref
254 encode_tag (enc, CBOR_TAG_STRINGREF); 314 encode_tag (enc, CBOR_TAG_STRINGREF);
255 encode_uint (enc, 0x00, SvUV (*svp)); 315 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
256 return; 316 return;
257 } 317 }
258 else if (len >= minimum_string_length (enc->stringref_idx)) 318 else if (len >= minimum_string_length (enc->stringref_idx))
259 { 319 {
260 // register only 320 // register only
… …
276 if (enc->depth >= enc->cbor.max_depth) 336 if (enc->depth >= enc->cbor.max_depth)
277 croak (ERR_NESTING_EXCEEDED); 337 croak (ERR_NESTING_EXCEEDED);
278 338
279 ++enc->depth; 339 ++enc->depth;
280 340
281 encode_uint (enc, 0x80, len + 1); 341 encode_uint (enc, MAJOR_ARRAY, len + 1);
282 342
343 if (SvMAGICAL (av))
283 for (i = 0; i <= len; ++i) 344 for (i = 0; i <= len; ++i)
284 { 345 {
285 SV **svp = av_fetch (av, i, 0); 346 SV **svp = av_fetch (av, i, 0);
286 encode_sv (enc, svp ? *svp : &PL_sv_undef); 347 encode_sv (enc, svp ? *svp : &PL_sv_undef);
287 } 348 }
349 else
350 for (i = 0; i <= len; ++i)
351 {
352 SV *sv = AvARRAY (av)[i];
353 encode_sv (enc, sv ? sv : &PL_sv_undef);
354 }
288 355
289 --enc->depth; 356 --enc->depth;
290} 357}
291 358
292static void 359static void
… …
301 368
302 int pairs = hv_iterinit (hv); 369 int pairs = hv_iterinit (hv);
303 int mg = SvMAGICAL (hv); 370 int mg = SvMAGICAL (hv);
304 371
305 if (mg) 372 if (mg)
306 encode_ch (enc, 0xa0 | 31); 373 encode_ch (enc, MAJOR_MAP | MINOR_INDEF);
307 else 374 else
308 encode_uint (enc, 0xa0, pairs); 375 encode_uint (enc, MAJOR_MAP, pairs);
309 376
310 while ((he = hv_iternext (hv))) 377 while ((he = hv_iternext (hv)))
311 { 378 {
312 if (HeKLEN (he) == HEf_SVKEY) 379 if (HeKLEN (he) == HEf_SVKEY)
313 encode_sv (enc, HeSVKEY (he)); 380 encode_sv (enc, HeSVKEY (he));
… …
316 383
317 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he)); 384 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he));
318 } 385 }
319 386
320 if (mg) 387 if (mg)
321 encode_ch (enc, 0xe0 | 31); 388 encode_ch (enc, MAJOR_MISC | MINOR_INDEF);
322 389
323 --enc->depth; 390 --enc->depth;
324} 391}
325 392
326// encode objects, arrays and special \0=false and \1=true values. 393// encode objects, arrays and special \0=false and \1=true values.
… …
342 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 409 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
343 ? cbor_tagged_stash 410 ? cbor_tagged_stash
344 : gv_stashpv ("CBOR::XS::Tagged" , 1); 411 : gv_stashpv ("CBOR::XS::Tagged" , 1);
345 412
346 HV *stash = SvSTASH (sv); 413 HV *stash = SvSTASH (sv);
347 GV *method;
348 414
349 if (stash == boolean_stash) 415 if (stash == boolean_stash)
350 encode_ch (enc, SvIV (sv) ? 0xe0 | 21 : 0xe0 | 20); 416 {
417 encode_ch (enc, SvIV (sv) ? MAJOR_MISC | SIMPLE_TRUE : MAJOR_MISC | SIMPLE_FALSE);
418 return;
419 }
351 else if (stash == error_stash) 420 else if (stash == error_stash)
352 encode_ch (enc, 0xe0 | 23); 421 {
422 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
423 return;
424 }
353 else if (stash == tagged_stash) 425 else if (stash == tagged_stash)
354 { 426 {
355 if (svt != SVt_PVAV) 427 if (svt != SVt_PVAV)
356 croak ("encountered CBOR::XS::Tagged object that isn't an array"); 428 croak ("encountered CBOR::XS::Tagged object that isn't an array");
357 429
358 encode_uint (enc, 0xc0, SvUV (*av_fetch ((AV *)sv, 0, 1))); 430 encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1)));
359 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1)); 431 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1));
432
433 return;
434 }
435 }
436
437 if (ecb_expect_false (SvREFCNT (sv) > 1)
438 && ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING))
439 {
440 if (!enc->shareable)
441 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
442
443 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
444
445 if (SvOK (*svp))
360 } 446 {
447 encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF);
448 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
449 return;
450 }
451 else
452 {
453 sv_setuv (*svp, enc->shareable_idx);
454 ++enc->shareable_idx;
455 encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE);
456 }
457 }
458
459 if (ecb_expect_false (SvOBJECT (sv)))
460 {
461 HV *stash = SvSTASH (sv);
462 GV *method;
463
361 else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0))) 464 if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
362 { 465 {
363 dSP; 466 dSP;
364 467
365 ENTER; SAVETMPS; PUSHMARK (SP); 468 ENTER; SAVETMPS;
469 PUSHMARK (SP);
366 // we re-bless the reference to get overload and other niceties right 470 // we re-bless the reference to get overload and other niceties right
367 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); 471 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
368 472
369 PUTBACK; 473 PUTBACK;
370 // G_SCALAR ensures that return value is 1 474 // G_SCALAR ensures that return value is 1
… …
383 } 487 }
384 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0) 488 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)
385 { 489 {
386 dSP; 490 dSP;
387 491
388 ENTER; SAVETMPS; PUSHMARK (SP); 492 ENTER; SAVETMPS;
493 SAVESTACK_POS ();
494 PUSHMARK (SP);
389 EXTEND (SP, 2); 495 EXTEND (SP, 2);
390 // we re-bless the reference to get overload and other niceties right 496 // we re-bless the reference to get overload and other niceties right
391 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); 497 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
392 PUSHs (sv_cbor); 498 PUSHs (sv_cbor);
393 499
… …
398 // catch this surprisingly common error 504 // catch this surprisingly common error
399 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv) 505 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)
400 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash)); 506 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));
401 507
402 encode_tag (enc, CBOR_TAG_PERL_OBJECT); 508 encode_tag (enc, CBOR_TAG_PERL_OBJECT);
403 encode_uint (enc, 0x80, count + 1); 509 encode_uint (enc, MAJOR_ARRAY, count + 1);
404 encode_strref (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash)); 510 encode_strref (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
405 511
406 while (count) 512 while (count)
407 encode_sv (enc, SP[1 - count--]); 513 encode_sv (enc, SP[1 - count--]);
408 514
… …
418 encode_hv (enc, (HV *)sv); 524 encode_hv (enc, (HV *)sv);
419 else if (svt == SVt_PVAV) 525 else if (svt == SVt_PVAV)
420 encode_av (enc, (AV *)sv); 526 encode_av (enc, (AV *)sv);
421 else 527 else
422 { 528 {
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, 0x00, 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 encode_tag (enc, CBOR_TAG_INDIRECTION); 529 encode_tag (enc, CBOR_TAG_INDIRECTION);
446 encode_sv (enc, sv); 530 encode_sv (enc, sv);
447 } 531 }
448} 532}
449 533
… …
452{ 536{
453 double nv = SvNVX (sv); 537 double nv = SvNVX (sv);
454 538
455 need (enc, 9); 539 need (enc, 9);
456 540
457 if (ecb_expect_false (nv == (U32)nv)) 541 if (ecb_expect_false (nv == (NV)(U32)nv))
458 encode_uint (enc, 0x00, (U32)nv); 542 encode_uint (enc, MAJOR_POS_INT, (U32)nv);
459 //TODO: maybe I32? 543 //TODO: maybe I32?
460 else if (ecb_expect_false (nv == (float)nv)) 544 else if (ecb_expect_false (nv == (float)nv))
461 { 545 {
462 uint32_t fp = ecb_float_to_binary32 (nv); 546 uint32_t fp = ecb_float_to_binary32 (nv);
463 547
464 *enc->cur++ = 0xe0 | 26; 548 *enc->cur++ = MAJOR_MISC | MISC_FLOAT32;
465 549
466 if (!ecb_big_endian ()) 550 if (!ecb_big_endian ())
467 fp = ecb_bswap32 (fp); 551 fp = ecb_bswap32 (fp);
468 552
469 memcpy (enc->cur, &fp, 4); 553 memcpy (enc->cur, &fp, 4);
… …
471 } 555 }
472 else 556 else
473 { 557 {
474 uint64_t fp = ecb_double_to_binary64 (nv); 558 uint64_t fp = ecb_double_to_binary64 (nv);
475 559
476 *enc->cur++ = 0xe0 | 27; 560 *enc->cur++ = MAJOR_MISC | MISC_FLOAT64;
477 561
478 if (!ecb_big_endian ()) 562 if (!ecb_big_endian ())
479 fp = ecb_bswap64 (fp); 563 fp = ecb_bswap64 (fp);
480 564
481 memcpy (enc->cur, &fp, 8); 565 memcpy (enc->cur, &fp, 8);
… …
497 else if (SvNOKp (sv)) 581 else if (SvNOKp (sv))
498 encode_nv (enc, sv); 582 encode_nv (enc, sv);
499 else if (SvIOKp (sv)) 583 else if (SvIOKp (sv))
500 { 584 {
501 if (SvIsUV (sv)) 585 if (SvIsUV (sv))
502 encode_uint (enc, 0x00, SvUVX (sv)); 586 encode_uint (enc, MAJOR_POS_INT, SvUVX (sv));
503 else if (SvIVX (sv) >= 0) 587 else if (SvIVX (sv) >= 0)
504 encode_uint (enc, 0x00, SvIVX (sv)); 588 encode_uint (enc, MAJOR_POS_INT, SvIVX (sv));
505 else 589 else
506 encode_uint (enc, 0x20, -(SvIVX (sv) + 1)); 590 encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1));
507 } 591 }
508 else if (SvROK (sv)) 592 else if (SvROK (sv))
509 encode_rv (enc, SvRV (sv)); 593 encode_rv (enc, SvRV (sv));
510 else if (!SvOK (sv)) 594 else if (!SvOK (sv))
511 encode_ch (enc, 0xe0 | 22); 595 encode_ch (enc, MAJOR_MISC | SIMPLE_NULL);
512 else if (enc->cbor.flags & F_ALLOW_UNKNOWN) 596 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
513 encode_ch (enc, 0xe0 | 23); 597 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
514 else 598 else
515 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data", 599 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data",
516 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv)); 600 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
517} 601}
518 602
519static SV * 603static SV *
520encode_cbor (SV *scalar, CBOR *cbor) 604encode_cbor (SV *scalar, CBOR *cbor)
521{ 605{
522 enc_t enc = { }; 606 enc_t enc = { 0 };
523 607
524 enc.cbor = *cbor; 608 enc.cbor = *cbor;
525 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 609 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
526 enc.cur = SvPVX (enc.sv); 610 enc.cur = SvPVX (enc.sv);
527 enc.end = SvEND (enc.sv); 611 enc.end = SvEND (enc.sv);
… …
571#define DEC_DEC_DEPTH --dec->depth 655#define DEC_DEC_DEPTH --dec->depth
572 656
573static UV 657static UV
574decode_uint (dec_t *dec) 658decode_uint (dec_t *dec)
575{ 659{
576 switch (*dec->cur & 31) 660 U8 m = *dec->cur & MINOR_MASK;
577 { 661 ++dec->cur;
578 case 0: case 1: case 2: case 3: case 4: case 5: case 6: case 7:
579 case 8: case 9: case 10: case 11: case 12: case 13: case 14: case 15:
580 case 16: case 17: case 18: case 19: case 20: case 21: case 22: case 23:
581 return *dec->cur++ & 31;
582 662
583 case 24: 663 if (ecb_expect_true (m < LENGTH_EXT1))
664 return m;
665 else if (ecb_expect_true (m == LENGTH_EXT1))
666 {
584 WANT (2); 667 WANT (1);
585 dec->cur += 2; 668 dec->cur += 1;
586 return dec->cur[-1]; 669 return dec->cur[-1];
587 670 }
588 case 25: 671 else if (ecb_expect_true (m == LENGTH_EXT2))
672 {
589 WANT (3); 673 WANT (2);
590 dec->cur += 3; 674 dec->cur += 2;
591 return (((UV)dec->cur[-2]) << 8) 675 return (((UV)dec->cur[-2]) << 8)
592 | ((UV)dec->cur[-1]); 676 | ((UV)dec->cur[-1]);
593 677 }
594 case 26: 678 else if (ecb_expect_true (m == LENGTH_EXT4))
679 {
595 WANT (5); 680 WANT (4);
596 dec->cur += 5; 681 dec->cur += 4;
597 return (((UV)dec->cur[-4]) << 24) 682 return (((UV)dec->cur[-4]) << 24)
598 | (((UV)dec->cur[-3]) << 16) 683 | (((UV)dec->cur[-3]) << 16)
599 | (((UV)dec->cur[-2]) << 8) 684 | (((UV)dec->cur[-2]) << 8)
600 | ((UV)dec->cur[-1]); 685 | ((UV)dec->cur[-1]);
601 686 }
602 case 27: 687 else if (ecb_expect_true (m == LENGTH_EXT8))
688 {
603 WANT (9); 689 WANT (8);
604 dec->cur += 9; 690 dec->cur += 8;
691
692 return
693#if UVSIZE < 8
694 0
695#else
605 return (((UV)dec->cur[-8]) << 56) 696 (((UV)dec->cur[-8]) << 56)
606 | (((UV)dec->cur[-7]) << 48) 697 | (((UV)dec->cur[-7]) << 48)
607 | (((UV)dec->cur[-6]) << 40) 698 | (((UV)dec->cur[-6]) << 40)
608 | (((UV)dec->cur[-5]) << 32) 699 | (((UV)dec->cur[-5]) << 32)
700#endif
609 | (((UV)dec->cur[-4]) << 24) 701 | (((UV)dec->cur[-4]) << 24)
610 | (((UV)dec->cur[-3]) << 16) 702 | (((UV)dec->cur[-3]) << 16)
611 | (((UV)dec->cur[-2]) << 8) 703 | (((UV)dec->cur[-2]) << 8)
612 | ((UV)dec->cur[-1]); 704 | ((UV)dec->cur[-1]);
613 705 }
614 default: 706 else
615 ERR ("corrupted CBOR data (unsupported integer minor encoding)"); 707 ERR ("corrupted CBOR data (unsupported integer minor encoding)");
616 }
617 708
618fail: 709fail:
619 return 0; 710 return 0;
620} 711}
621 712
… …
626{ 717{
627 AV *av = newAV (); 718 AV *av = newAV ();
628 719
629 DEC_INC_DEPTH; 720 DEC_INC_DEPTH;
630 721
631 if ((*dec->cur & 31) == 31) 722 if (*dec->cur == (MAJOR_ARRAY | MINOR_INDEF))
632 { 723 {
633 ++dec->cur; 724 ++dec->cur;
634 725
635 for (;;) 726 for (;;)
636 { 727 {
637 WANT (1); 728 WANT (1);
638 729
639 if (*dec->cur == (0xe0 | 31)) 730 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
640 { 731 {
641 ++dec->cur; 732 ++dec->cur;
642 break; 733 break;
643 } 734 }
644 735
… …
647 } 738 }
648 else 739 else
649 { 740 {
650 int i, len = decode_uint (dec); 741 int i, len = decode_uint (dec);
651 742
743 WANT (len); // complexity check for av_fill - need at least one byte per value, do not allow supersize arrays
652 av_fill (av, len - 1); 744 av_fill (av, len - 1);
653 745
654 for (i = 0; i < len; ++i) 746 for (i = 0; i < len; ++i)
655 AvARRAY (av)[i] = decode_sv (dec); 747 AvARRAY (av)[i] = decode_sv (dec);
656 } 748 }
… …
669{ 761{
670 // for speed reasons, we specialcase single-string 762 // for speed reasons, we specialcase single-string
671 // byte or utf-8 strings as keys, but only when !stringref 763 // byte or utf-8 strings as keys, but only when !stringref
672 764
673 if (ecb_expect_true (!dec->stringref)) 765 if (ecb_expect_true (!dec->stringref))
674 if (*dec->cur >= 0x40 && *dec->cur <= 0x40 + 27) 766 if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8))
675 { 767 {
676 I32 len = decode_uint (dec); 768 I32 len = decode_uint (dec);
677 char *key = (char *)dec->cur; 769 char *key = (char *)dec->cur;
678 770
771 WANT (len);
679 dec->cur += len; 772 dec->cur += len;
680 773
681 if (ecb_expect_false (dec->stringref))
682 av_push (dec->stringref, newSVpvn (key, len));
683
684 hv_store (hv, key, len, decode_sv (dec), 0); 774 hv_store (hv, key, len, decode_sv (dec), 0);
685 775
686 return; 776 return;
687 } 777 }
688 else if (*dec->cur >= 0x60 && *dec->cur <= 0x60 + 27) 778 else if (ecb_expect_true ((U8)(*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8))
689 { 779 {
690 I32 len = decode_uint (dec); 780 I32 len = decode_uint (dec);
691 char *key = (char *)dec->cur; 781 char *key = (char *)dec->cur;
692 782
783 WANT (len);
693 dec->cur += len; 784 dec->cur += len;
694 785
695 if (ecb_expect_false (dec->stringref)) 786 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
696 av_push (dec->stringref, newSVpvn_utf8 (key, len, 1)); 787 if (!is_utf8_string (key, len))
788 ERR ("corrupted CBOR data (invalid UTF-8 in map key)");
697 789
698 hv_store (hv, key, -len, decode_sv (dec), 0); 790 hv_store (hv, key, -len, decode_sv (dec), 0);
699 791
700 return; 792 return;
701 } 793 }
… …
703 SV *k = decode_sv (dec); 795 SV *k = decode_sv (dec);
704 SV *v = decode_sv (dec); 796 SV *v = decode_sv (dec);
705 797
706 hv_store_ent (hv, k, v, 0); 798 hv_store_ent (hv, k, v, 0);
707 SvREFCNT_dec (k); 799 SvREFCNT_dec (k);
800
801fail:
802 ;
708} 803}
709 804
710static SV * 805static SV *
711decode_hv (dec_t *dec) 806decode_hv (dec_t *dec)
712{ 807{
713 HV *hv = newHV (); 808 HV *hv = newHV ();
714 809
715 DEC_INC_DEPTH; 810 DEC_INC_DEPTH;
716 811
717 if ((*dec->cur & 31) == 31) 812 if (*dec->cur == (MAJOR_MAP | MINOR_INDEF))
718 { 813 {
719 ++dec->cur; 814 ++dec->cur;
720 815
721 for (;;) 816 for (;;)
722 { 817 {
723 WANT (1); 818 WANT (1);
724 819
725 if (*dec->cur == (0xe0 | 31)) 820 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
726 { 821 {
727 ++dec->cur; 822 ++dec->cur;
728 break; 823 break;
729 } 824 }
730 825
… …
751static SV * 846static SV *
752decode_str (dec_t *dec, int utf8) 847decode_str (dec_t *dec, int utf8)
753{ 848{
754 SV *sv = 0; 849 SV *sv = 0;
755 850
756 if ((*dec->cur & 31) == 31) 851 if ((*dec->cur & MINOR_MASK) == MINOR_INDEF)
757 { 852 {
758 // indefinite length strings 853 // indefinite length strings
759 ++dec->cur; 854 ++dec->cur;
760 855
761 unsigned char major = *dec->cur & 0xe0; 856 U8 major = *dec->cur & MAJOR_MISC;
762 857
763 sv = newSVpvn ("", 0); 858 sv = newSVpvn ("", 0);
764 859
765 for (;;) 860 for (;;)
766 { 861 {
767 WANT (1); 862 WANT (1);
768 863
769 if ((*dec->cur ^ major) >= 31) 864 if ((*dec->cur - major) > LENGTH_EXT8)
770 if (*dec->cur == (0xe0 | 31)) 865 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
771 { 866 {
772 ++dec->cur; 867 ++dec->cur;
773 break; 868 break;
774 } 869 }
775 else 870 else
… …
794 && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1)) 889 && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1))
795 av_push (dec->stringref, SvREFCNT_inc_NN (sv)); 890 av_push (dec->stringref, SvREFCNT_inc_NN (sv));
796 } 891 }
797 892
798 if (utf8) 893 if (utf8)
894 {
895 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
896 if (!is_utf8_string (SvPVX (sv), SvCUR (sv)))
897 ERR ("corrupted CBOR data (invalid UTF-8 in text string)");
898
799 SvUTF8_on (sv); 899 SvUTF8_on (sv);
900 }
800 901
801 return sv; 902 return sv;
802 903
803fail: 904fail:
804 SvREFCNT_dec (sv); 905 SvREFCNT_dec (sv);
… …
836 } 937 }
837 break; 938 break;
838 939
839 case CBOR_TAG_STRINGREF: 940 case CBOR_TAG_STRINGREF:
840 { 941 {
841 if ((*dec->cur >> 5) != 0) 942 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
842 ERR ("corrupted CBOR data (stringref index not an unsigned integer)"); 943 ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
843 944
844 UV idx = decode_uint (dec); 945 UV idx = decode_uint (dec);
845 946
846 if (!dec->stringref || (int)idx > AvFILLp (dec->stringref)) 947 if (!dec->stringref || (int)idx > AvFILLp (dec->stringref))
… …
853 case CBOR_TAG_VALUE_SHAREABLE: 954 case CBOR_TAG_VALUE_SHAREABLE:
854 { 955 {
855 if (ecb_expect_false (!dec->shareable)) 956 if (ecb_expect_false (!dec->shareable))
856 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ()); 957 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ());
857 958
959 if (dec->cbor.flags & F_ALLOW_CYCLES)
960 {
858 sv = newSV (0); 961 sv = newSV (0);
859 av_push (dec->shareable, SvREFCNT_inc_NN (sv)); 962 av_push (dec->shareable, SvREFCNT_inc_NN (sv));
860 963
861 SV *osv = decode_sv (dec); 964 SV *osv = decode_sv (dec);
862 sv_setsv (sv, osv); 965 sv_setsv (sv, osv);
863 SvREFCNT_dec_NN (osv); 966 SvREFCNT_dec_NN (osv);
967 }
968 else
969 {
970 av_push (dec->shareable, &PL_sv_undef);
971 int idx = AvFILLp (dec->shareable);
972 sv = decode_sv (dec);
973 av_store (dec->shareable, idx, SvREFCNT_inc_NN (sv));
974 }
864 } 975 }
865 break; 976 break;
866 977
867 case CBOR_TAG_VALUE_SHAREDREF: 978 case CBOR_TAG_VALUE_SHAREDREF:
868 { 979 {
869 if ((*dec->cur >> 5) != 0) 980 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
870 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)"); 981 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
871 982
872 UV idx = decode_uint (dec); 983 UV idx = decode_uint (dec);
873 984
874 if (!dec->shareable || (int)idx > AvFILLp (dec->shareable)) 985 if (!dec->shareable || (int)idx > AvFILLp (dec->shareable))
875 ERR ("corrupted CBOR data (sharedref index out of bounds)"); 986 ERR ("corrupted CBOR data (sharedref index out of bounds)");
876 987
877 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]); 988 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]);
989
990 if (sv == &PL_sv_undef)
991 ERR ("cyclic CBOR data structure found, but allow_cycles is not enabled");
878 } 992 }
879 break; 993 break;
880 994
881 case CBOR_TAG_PERL_OBJECT: 995 case CBOR_TAG_PERL_OBJECT:
882 { 996 {
… …
897 if (!method) 1011 if (!method)
898 ERR ("cannot decode perl-object (package does not have a THAW method)"); 1012 ERR ("cannot decode perl-object (package does not have a THAW method)");
899 1013
900 dSP; 1014 dSP;
901 1015
902 ENTER; SAVETMPS; PUSHMARK (SP); 1016 ENTER; SAVETMPS;
1017 PUSHMARK (SP);
903 EXTEND (SP, len + 1); 1018 EXTEND (SP, len + 1);
904 // we re-bless the reference to get overload and other niceties right 1019 // we re-bless the reference to get overload and other niceties right
905 PUSHs (*av_fetch (av, 0, 1)); 1020 PUSHs (*av_fetch (av, 0, 1));
906 PUSHs (sv_cbor); 1021 PUSHs (sv_cbor);
907 1022
… …
932 default: 1047 default:
933 { 1048 {
934 sv = decode_sv (dec); 1049 sv = decode_sv (dec);
935 1050
936 dSP; 1051 dSP;
937 ENTER; SAVETMPS; PUSHMARK (SP); 1052 ENTER; SAVETMPS;
1053 SAVESTACK_POS ();
1054 PUSHMARK (SP);
938 EXTEND (SP, 2); 1055 EXTEND (SP, 2);
939 PUSHs (newSVuv (tag)); 1056 PUSHs (newSVuv (tag));
940 PUSHs (sv); 1057 PUSHs (sv);
941 1058
942 PUTBACK; 1059 PUTBACK;
… …
983static SV * 1100static SV *
984decode_sv (dec_t *dec) 1101decode_sv (dec_t *dec)
985{ 1102{
986 WANT (1); 1103 WANT (1);
987 1104
988 switch (*dec->cur >> 5) 1105 switch (*dec->cur >> MAJOR_SHIFT)
989 { 1106 {
990 case 0: // unsigned int 1107 case MAJOR_POS_INT >> MAJOR_SHIFT: return newSVuv (decode_uint (dec));
991 return newSVuv (decode_uint (dec)); 1108 case MAJOR_NEG_INT >> MAJOR_SHIFT: return newSViv (-1 - (IV)decode_uint (dec));
992 case 1: // negative int 1109 case MAJOR_BYTES >> MAJOR_SHIFT: return decode_str (dec, 0);
993 return newSViv (-1 - (IV)decode_uint (dec)); 1110 case MAJOR_TEXT >> MAJOR_SHIFT: return decode_str (dec, 1);
994 case 2: // octet string 1111 case MAJOR_ARRAY >> MAJOR_SHIFT: return decode_av (dec);
995 return decode_str (dec, 0); 1112 case MAJOR_MAP >> MAJOR_SHIFT: return decode_hv (dec);
996 case 3: // utf-8 string 1113 case MAJOR_TAG >> MAJOR_SHIFT: return decode_tagged (dec);
997 return decode_str (dec, 1); 1114
998 case 4: // array 1115 case MAJOR_MISC >> MAJOR_SHIFT:
999 return decode_av (dec);
1000 case 5: // map
1001 return decode_hv (dec);
1002 case 6: // tag
1003 return decode_tagged (dec);
1004 case 7: // misc
1005 switch (*dec->cur++ & 31) 1116 switch (*dec->cur++ & MINOR_MASK)
1006 { 1117 {
1007 case 20: 1118 case SIMPLE_FALSE:
1008#if CBOR_SLOW 1119#if CBOR_SLOW
1009 types_false = get_bool ("Types::Serialiser::false"); 1120 types_false = get_bool ("Types::Serialiser::false");
1010#endif 1121#endif
1011 return newSVsv (types_false); 1122 return newSVsv (types_false);
1012 case 21: 1123 case SIMPLE_TRUE:
1013#if CBOR_SLOW 1124#if CBOR_SLOW
1014 types_true = get_bool ("Types::Serialiser::true"); 1125 types_true = get_bool ("Types::Serialiser::true");
1015#endif 1126#endif
1016 return newSVsv (types_true); 1127 return newSVsv (types_true);
1017 case 22: 1128 case SIMPLE_NULL:
1018 return newSVsv (&PL_sv_undef); 1129 return newSVsv (&PL_sv_undef);
1019 case 23: 1130 case SIMPLE_UNDEF:
1020#if CBOR_SLOW 1131#if CBOR_SLOW
1021 types_error = get_bool ("Types::Serialiser::error"); 1132 types_error = get_bool ("Types::Serialiser::error");
1022#endif 1133#endif
1023 return newSVsv (types_error); 1134 return newSVsv (types_error);
1024 1135
1025 case 25: 1136 case MISC_FLOAT16:
1026 { 1137 {
1027 WANT (2); 1138 WANT (2);
1028 1139
1029 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1]; 1140 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1];
1030 dec->cur += 2; 1141 dec->cur += 2;
1031 1142
1032 return newSVnv (ecb_binary16_to_float (fp)); 1143 return newSVnv (ecb_binary16_to_float (fp));
1033 } 1144 }
1034 1145
1035 case 26: 1146 case MISC_FLOAT32:
1036 { 1147 {
1037 uint32_t fp; 1148 uint32_t fp;
1038 WANT (4); 1149 WANT (4);
1039 memcpy (&fp, dec->cur, 4); 1150 memcpy (&fp, dec->cur, 4);
1040 dec->cur += 4; 1151 dec->cur += 4;
… …
1043 fp = ecb_bswap32 (fp); 1154 fp = ecb_bswap32 (fp);
1044 1155
1045 return newSVnv (ecb_binary32_to_float (fp)); 1156 return newSVnv (ecb_binary32_to_float (fp));
1046 } 1157 }
1047 1158
1048 case 27: 1159 case MISC_FLOAT64:
1049 { 1160 {
1050 uint64_t fp; 1161 uint64_t fp;
1051 WANT (8); 1162 WANT (8);
1052 memcpy (&fp, dec->cur, 8); 1163 memcpy (&fp, dec->cur, 8);
1053 dec->cur += 8; 1164 dec->cur += 8;
… …
1056 fp = ecb_bswap64 (fp); 1167 fp = ecb_bswap64 (fp);
1057 1168
1058 return newSVnv (ecb_binary64_to_double (fp)); 1169 return newSVnv (ecb_binary64_to_double (fp));
1059 } 1170 }
1060 1171
1061 // 0..19 unassigned 1172 // 0..19 unassigned simple
1062 // 24 reserved + unassigned (reserved values are not encodable) 1173 // 24 reserved + unassigned simple (reserved values are not encodable)
1174 // 28-30 unassigned misc
1175 // 31 break code
1063 default: 1176 default:
1064 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)"); 1177 ERR ("corrupted CBOR data (reserved/unassigned/unexpected major 7 value)");
1065 } 1178 }
1066 1179
1067 break; 1180 break;
1068 } 1181 }
1069 1182
… …
1072} 1185}
1073 1186
1074static SV * 1187static SV *
1075decode_cbor (SV *string, CBOR *cbor, char **offset_return) 1188decode_cbor (SV *string, CBOR *cbor, char **offset_return)
1076{ 1189{
1077 dec_t dec = { }; 1190 dec_t dec = { 0 };
1078 SV *sv; 1191 SV *sv;
1079 STRLEN len; 1192 STRLEN len;
1080 char *data = SvPVbyte (string, len); 1193 char *data = SvPVbyte (string, len);
1081 1194
1082 if (len > cbor->max_size && cbor->max_size) 1195 if (len > cbor->max_size && cbor->max_size)
… …
1096 if (dec.cur != dec.end && !dec.err) 1209 if (dec.cur != dec.end && !dec.err)
1097 dec.err = "garbage after CBOR object"; 1210 dec.err = "garbage after CBOR object";
1098 1211
1099 if (dec.err) 1212 if (dec.err)
1100 { 1213 {
1214 if (dec.shareable)
1215 {
1216 // need to break cyclic links, which whould all be in shareable
1217 int i;
1218 SV **svp;
1219
1220 for (i = av_len (dec.shareable) + 1; i--; )
1221 if ((svp = av_fetch (dec.shareable, i, 0)))
1222 sv_setsv (*svp, &PL_sv_undef);
1223 }
1224
1101 SvREFCNT_dec (sv); 1225 SvREFCNT_dec (sv);
1102 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur); 1226 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur);
1103 } 1227 }
1104 1228
1105 sv = sv_2mortal (sv); 1229 sv = sv_2mortal (sv);
1106 1230
1107 return sv; 1231 return sv;
1108} 1232}
1109 1233
1234/////////////////////////////////////////////////////////////////////////////
1235// incremental parser
1236
1237#define INCR_DONE(cbor) (AvFILLp (cbor->incr_count) < 0)
1238
1239// returns 0 for notyet, 1 for success or error
1240static int
1241incr_parse (CBOR *self, SV *cborstr)
1242{
1243 STRLEN cur;
1244 SvPV (cborstr, cur);
1245
1246 while (ecb_expect_true (self->incr_need <= cur))
1247 {
1248 // table of integer count bytes
1249 static I8 incr_len[MINOR_MASK + 1] = {
1250 0, 0, 0, 0, 0, 0, 0, 0,
1251 0, 0, 0, 0, 0, 0, 0, 0,
1252 0, 0, 0, 0, 0, 0, 0, 0,
1253 1, 2, 4, 8,-1,-1,-1,-2
1254 };
1255
1256 const U8 *p = SvPVX (cborstr) + self->incr_pos;
1257 U8 m = *p & MINOR_MASK;
1258 IV count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1259 I8 ilen = incr_len[m];
1260
1261 self->incr_need = self->incr_pos + 1;
1262
1263 if (ecb_expect_false (ilen < 0))
1264 {
1265 if (m != MINOR_INDEF)
1266 return 1; // error
1267
1268 if (*p == (MAJOR_MISC | MINOR_INDEF))
1269 {
1270 if (count >= 0)
1271 return 1; // error
1272
1273 count = 1;
1274 }
1275 else
1276 {
1277 av_push (self->incr_count, newSViv (-1)); //TODO: nest
1278 count = -1;
1279 }
1280 }
1281 else
1282 {
1283 self->incr_need += ilen;
1284 if (ecb_expect_false (self->incr_need > cur))
1285 return 0;
1286
1287 int major = *p >> MAJOR_SHIFT;
1288
1289 switch (major)
1290 {
1291 case MAJOR_TAG >> MAJOR_SHIFT:
1292 ++count; // tags merely prefix another value
1293 break;
1294
1295 case MAJOR_BYTES >> MAJOR_SHIFT:
1296 case MAJOR_TEXT >> MAJOR_SHIFT:
1297 case MAJOR_ARRAY >> MAJOR_SHIFT:
1298 case MAJOR_MAP >> MAJOR_SHIFT:
1299 {
1300 UV len;
1301
1302 if (ecb_expect_false (ilen))
1303 {
1304 len = 0;
1305
1306 do {
1307 len = (len << 8) | *++p;
1308 } while (--ilen);
1309 }
1310 else
1311 len = m;
1312
1313 switch (major)
1314 {
1315 case MAJOR_BYTES >> MAJOR_SHIFT:
1316 case MAJOR_TEXT >> MAJOR_SHIFT:
1317 self->incr_need += len;
1318 if (ecb_expect_false (self->incr_need > cur))
1319 return 0;
1320
1321 break;
1322
1323 case MAJOR_MAP >> MAJOR_SHIFT:
1324 len <<= 1;
1325 case MAJOR_ARRAY >> MAJOR_SHIFT:
1326 if (len)
1327 {
1328 av_push (self->incr_count, newSViv (len + 1)); //TODO: nest
1329 count = len + 1;
1330 }
1331 break;
1332 }
1333 }
1334 }
1335 }
1336
1337 self->incr_pos = self->incr_need;
1338
1339 if (count > 0)
1340 {
1341 while (!--count)
1342 {
1343 if (!AvFILLp (self->incr_count))
1344 return 1; // done
1345
1346 SvREFCNT_dec_NN (av_pop (self->incr_count));
1347 count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1348 }
1349
1350 SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]) = count;
1351 }
1352 }
1353
1354 return 0;
1355}
1356
1357
1110///////////////////////////////////////////////////////////////////////////// 1358/////////////////////////////////////////////////////////////////////////////
1111// XS interface functions 1359// XS interface functions
1112 1360
1113MODULE = CBOR::XS PACKAGE = CBOR::XS 1361MODULE = CBOR::XS PACKAGE = CBOR::XS
1114 1362
… …
1154void shrink (CBOR *self, int enable = 1) 1402void shrink (CBOR *self, int enable = 1)
1155 ALIAS: 1403 ALIAS:
1156 shrink = F_SHRINK 1404 shrink = F_SHRINK
1157 allow_unknown = F_ALLOW_UNKNOWN 1405 allow_unknown = F_ALLOW_UNKNOWN
1158 allow_sharing = F_ALLOW_SHARING 1406 allow_sharing = F_ALLOW_SHARING
1407 allow_cycles = F_ALLOW_CYCLES
1159 pack_strings = F_PACK_STRINGS 1408 pack_strings = F_PACK_STRINGS
1409 utf8_strings = F_UTF8_STRINGS
1410 validate_utf8 = F_VALIDATE_UTF8
1160 PPCODE: 1411 PPCODE:
1161{ 1412{
1162 if (enable) 1413 if (enable)
1163 self->flags |= ix; 1414 self->flags |= ix;
1164 else 1415 else
… …
1170void get_shrink (CBOR *self) 1421void get_shrink (CBOR *self)
1171 ALIAS: 1422 ALIAS:
1172 get_shrink = F_SHRINK 1423 get_shrink = F_SHRINK
1173 get_allow_unknown = F_ALLOW_UNKNOWN 1424 get_allow_unknown = F_ALLOW_UNKNOWN
1174 get_allow_sharing = F_ALLOW_SHARING 1425 get_allow_sharing = F_ALLOW_SHARING
1426 get_allow_cycles = F_ALLOW_CYCLES
1175 get_pack_strings = F_PACK_STRINGS 1427 get_pack_strings = F_PACK_STRINGS
1428 get_validate_utf8 = F_VALIDATE_UTF8
1176 PPCODE: 1429 PPCODE:
1177 XPUSHs (boolSV (self->flags & ix)); 1430 XPUSHs (boolSV (self->flags & ix));
1178 1431
1179void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) 1432void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
1180 PPCODE: 1433 PPCODE:
… …
1229 EXTEND (SP, 2); 1482 EXTEND (SP, 2);
1230 PUSHs (sv); 1483 PUSHs (sv);
1231 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); 1484 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
1232} 1485}
1233 1486
1487void incr_parse (CBOR *self, SV *cborstr)
1488 ALIAS:
1489 incr_parse_multiple = 1
1490 PPCODE:
1491{
1492 if (SvUTF8 (cborstr))
1493 sv_utf8_downgrade (cborstr, 0);
1494
1495 if (!self->incr_count)
1496 {
1497 self->incr_count = newAV ();
1498 self->incr_pos = 0;
1499 self->incr_need = 1;
1500
1501 av_push (self->incr_count, newSViv (1));
1502 }
1503
1504 do
1505 {
1506 if (!incr_parse (self, cborstr))
1507 {
1508 if (self->incr_need > self->max_size && self->max_size)
1509 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
1510 (unsigned long)self->incr_need, (unsigned long)self->max_size);
1511
1512 break;
1513 }
1514
1515 SV *sv;
1516 char *offset;
1517
1518 PUTBACK; sv = decode_cbor (cborstr, self, &offset); SPAGAIN;
1519 XPUSHs (sv);
1520
1521 sv_chop (cborstr, offset);
1522
1523 av_clear (self->incr_count);
1524 av_push (self->incr_count, newSViv (1));
1525
1526 self->incr_pos = 0;
1527 self->incr_need = self->incr_pos + 1;
1528 }
1529 while (ix);
1530}
1531
1532void incr_reset (CBOR *self)
1533 CODE:
1534{
1535 SvREFCNT_dec (self->incr_count);
1536 self->incr_count = 0;
1537}
1538
1234void DESTROY (CBOR *self) 1539void DESTROY (CBOR *self)
1235 PPCODE: 1540 PPCODE:
1236 cbor_free (self); 1541 cbor_free (self);
1237 1542
1238PROTOTYPES: ENABLE 1543PROTOTYPES: ENABLE
1239 1544
1240void encode_cbor (SV *scalar) 1545void encode_cbor (SV *scalar)
1546 ALIAS:
1547 encode_cbor = 0
1548 encode_cbor_sharing = F_ALLOW_SHARING
1241 PPCODE: 1549 PPCODE:
1242{ 1550{
1243 CBOR cbor; 1551 CBOR cbor;
1244 cbor_init (&cbor); 1552 cbor_init (&cbor);
1553 cbor.flags |= ix;
1245 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN; 1554 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;
1246 XPUSHs (scalar); 1555 XPUSHs (scalar);
1247} 1556}
1248 1557
1249void decode_cbor (SV *cborstr) 1558void decode_cbor (SV *cborstr)

Diff Legend

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