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 1.27 by root, Fri Nov 22 15:28:38 2013 UTC vs.
Revision 1.58 by root, Fri Nov 25 13:27:29 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_TEXT_KEYS 0x00000020UL
106#define F_TEXT_STRINGS 0x00000040UL
107#define F_VALIDATE_UTF8 0x00000080UL
63 108
64#define INIT_SIZE 32 // initial scalar size to be allocated 109#define INIT_SIZE 32 // initial scalar size to be allocated
65 110
66#define SB do { 111#define SB do {
67#define SE } while (0) 112#define SE } while (0)
86typedef struct { 131typedef struct {
87 U32 flags; 132 U32 flags;
88 U32 max_depth; 133 U32 max_depth;
89 STRLEN max_size; 134 STRLEN max_size;
90 SV *filter; 135 SV *filter;
136
137 // for the incremental parser
138 STRLEN incr_pos; // the current offset into the text
139 STRLEN incr_need; // minimum bytes needed to decode
140 AV *incr_count; // for every nesting level, the number of outstanding values, or -1 for indef.
91} CBOR; 141} CBOR;
92 142
93ecb_inline void 143ecb_inline void
94cbor_init (CBOR *cbor) 144cbor_init (CBOR *cbor)
95{ 145{
99 149
100ecb_inline void 150ecb_inline void
101cbor_free (CBOR *cbor) 151cbor_free (CBOR *cbor)
102{ 152{
103 SvREFCNT_dec (cbor->filter); 153 SvREFCNT_dec (cbor->filter);
154 SvREFCNT_dec (cbor->incr_count);
104} 155}
105 156
106///////////////////////////////////////////////////////////////////////////// 157/////////////////////////////////////////////////////////////////////////////
107// utility functions 158// utility functions
108 159
138{ 189{
139 return idx > 23 190 return idx > 23
140 ? idx > 0xffU 191 ? idx > 0xffU
141 ? idx > 0xffffU 192 ? idx > 0xffffU
142 ? idx > 0xffffffffU 193 ? idx > 0xffffffffU
194 ? 11
143 ? 7 195 : 7
144 : 6
145 : 5 196 : 5
146 : 4 197 : 4
147 : 3; 198 : 3;
148} 199}
149 200
186static void 237static void
187encode_uint (enc_t *enc, int major, UV len) 238encode_uint (enc_t *enc, int major, UV len)
188{ 239{
189 need (enc, 9); 240 need (enc, 9);
190 241
191 if (len < 24) 242 if (ecb_expect_true (len < LENGTH_EXT1))
192 *enc->cur++ = major | len; 243 *enc->cur++ = major | len;
193 else if (len <= 0xff) 244 else if (ecb_expect_true (len <= 0xffU))
194 { 245 {
195 *enc->cur++ = major | 24; 246 *enc->cur++ = major | LENGTH_EXT1;
196 *enc->cur++ = len; 247 *enc->cur++ = len;
197 } 248 }
198 else if (len <= 0xffff) 249 else if (len <= 0xffffU)
199 { 250 {
200 *enc->cur++ = major | 25; 251 *enc->cur++ = major | LENGTH_EXT2;
201 *enc->cur++ = len >> 8; 252 *enc->cur++ = len >> 8;
202 *enc->cur++ = len; 253 *enc->cur++ = len;
203 } 254 }
204 else if (len <= 0xffffffff) 255 else if (len <= 0xffffffffU)
205 { 256 {
206 *enc->cur++ = major | 26; 257 *enc->cur++ = major | LENGTH_EXT4;
207 *enc->cur++ = len >> 24; 258 *enc->cur++ = len >> 24;
208 *enc->cur++ = len >> 16; 259 *enc->cur++ = len >> 16;
209 *enc->cur++ = len >> 8; 260 *enc->cur++ = len >> 8;
210 *enc->cur++ = len; 261 *enc->cur++ = len;
211 } 262 }
212 else 263 else
213 { 264 {
214 *enc->cur++ = major | 27; 265 *enc->cur++ = major | LENGTH_EXT8;
215 *enc->cur++ = len >> 56; 266 *enc->cur++ = len >> 56;
216 *enc->cur++ = len >> 48; 267 *enc->cur++ = len >> 48;
217 *enc->cur++ = len >> 40; 268 *enc->cur++ = len >> 40;
218 *enc->cur++ = len >> 32; 269 *enc->cur++ = len >> 32;
219 *enc->cur++ = len >> 24; 270 *enc->cur++ = len >> 24;
224} 275}
225 276
226ecb_inline void 277ecb_inline void
227encode_tag (enc_t *enc, UV tag) 278encode_tag (enc_t *enc, UV tag)
228{ 279{
229 encode_uint (enc, 0xc0, tag); 280 encode_uint (enc, MAJOR_TAG, tag);
230} 281}
231 282
232static void 283// exceptional (hopefully) slow path for byte strings that need to be utf8-encoded
284ecb_noinline static void
233encode_str (enc_t *enc, int utf8, char *str, STRLEN len) 285encode_str_utf8 (enc_t *enc, int utf8, char *str, STRLEN len)
234{ 286{
287 STRLEN ulen = len;
288 U8 *p, *pend = (U8 *)str + len;
289
290 for (p = (U8 *)str; p < pend; ++p)
291 ulen += *p >> 7; // count set high bits
292
293 encode_uint (enc, MAJOR_TEXT, ulen);
294
295 need (enc, ulen);
296 for (p = (U8 *)str; p < pend; ++p)
297 if (*p < 0x80)
298 *enc->cur++ = *p;
299 else
300 {
301 *enc->cur++ = 0xc0 + (*p >> 6);
302 *enc->cur++ = 0x80 + (*p & 63);
303 }
304}
305
306ecb_inline void
307encode_str (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len)
308{
309 if (ecb_expect_false (upgrade_utf8))
310 if (!utf8)
311 {
312 encode_str_utf8 (enc, utf8, str, len);
313 return;
314 }
315
316 encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len);
317 need (enc, len);
318 memcpy (enc->cur, str, len);
319 enc->cur += len;
320}
321
322ecb_inline void
323encode_strref (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len)
324{
235 if (ecb_expect_false (enc->cbor.flags & F_ALLOW_STRINGREF)) 325 if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS))
236 { 326 {
237 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1); 327 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1);
238 328
239 if (SvOK (*svp)) 329 if (SvOK (*svp))
240 { 330 {
241 // already registered, use stringref 331 // already registered, use stringref
242 encode_tag (enc, CBOR_TAG_STRINGREF); 332 encode_tag (enc, CBOR_TAG_STRINGREF);
243 encode_uint (enc, 0x00, SvUV (*svp)); 333 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
244 return; 334 return;
245 } 335 }
246 else if (len >= minimum_string_length (enc->stringref_idx)) 336 else if (len >= minimum_string_length (enc->stringref_idx))
247 { 337 {
248 // register only 338 // register only
249 sv_setuv (*svp, enc->stringref_idx); 339 sv_setuv (*svp, enc->stringref_idx);
250 ++enc->stringref_idx; 340 ++enc->stringref_idx;
251 } 341 }
252 } 342 }
253 343
254 encode_uint (enc, utf8 ? 0x60 : 0x40, len); 344 encode_str (enc, upgrade_utf8, utf8, str, len);
255 need (enc, len);
256 memcpy (enc->cur, str, len);
257 enc->cur += len;
258} 345}
259 346
260static void encode_sv (enc_t *enc, SV *sv); 347static void encode_sv (enc_t *enc, SV *sv);
261 348
262static void 349static void
267 if (enc->depth >= enc->cbor.max_depth) 354 if (enc->depth >= enc->cbor.max_depth)
268 croak (ERR_NESTING_EXCEEDED); 355 croak (ERR_NESTING_EXCEEDED);
269 356
270 ++enc->depth; 357 ++enc->depth;
271 358
272 encode_uint (enc, 0x80, len + 1); 359 encode_uint (enc, MAJOR_ARRAY, len + 1);
273 360
361 if (SvMAGICAL (av))
274 for (i = 0; i <= len; ++i) 362 for (i = 0; i <= len; ++i)
275 { 363 {
276 SV **svp = av_fetch (av, i, 0); 364 SV **svp = av_fetch (av, i, 0);
277 encode_sv (enc, svp ? *svp : &PL_sv_undef); 365 encode_sv (enc, svp ? *svp : &PL_sv_undef);
278 } 366 }
367 else
368 for (i = 0; i <= len; ++i)
369 {
370 SV *sv = AvARRAY (av)[i];
371 encode_sv (enc, sv ? sv : &PL_sv_undef);
372 }
279 373
280 --enc->depth; 374 --enc->depth;
281} 375}
282 376
283static void 377static void
292 386
293 int pairs = hv_iterinit (hv); 387 int pairs = hv_iterinit (hv);
294 int mg = SvMAGICAL (hv); 388 int mg = SvMAGICAL (hv);
295 389
296 if (mg) 390 if (mg)
297 encode_ch (enc, 0xa0 | 31); 391 encode_ch (enc, MAJOR_MAP | MINOR_INDEF);
298 else 392 else
299 encode_uint (enc, 0xa0, pairs); 393 encode_uint (enc, MAJOR_MAP, pairs);
300 394
301 while ((he = hv_iternext (hv))) 395 while ((he = hv_iternext (hv)))
302 { 396 {
303 if (HeKLEN (he) == HEf_SVKEY) 397 if (HeKLEN (he) == HEf_SVKEY)
304 encode_sv (enc, HeSVKEY (he)); 398 encode_sv (enc, HeSVKEY (he));
305 else 399 else
306 encode_str (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he)); 400 encode_strref (enc, enc->cbor.flags & (F_TEXT_KEYS | F_TEXT_STRINGS), HeKUTF8 (he), HeKEY (he), HeKLEN (he));
307 401
308 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he)); 402 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he));
309 } 403 }
310 404
311 if (mg) 405 if (mg)
312 encode_ch (enc, 0xe0 | 31); 406 encode_ch (enc, MAJOR_MISC | MINOR_INDEF);
313 407
314 --enc->depth; 408 --enc->depth;
315} 409}
316 410
317// encode objects, arrays and special \0=false and \1=true values. 411// encode objects, arrays and special \0=false and \1=true values.
318static void 412static void
319encode_rv (enc_t *enc, SV *sv) 413encode_rv (enc_t *enc, SV *sv)
320{ 414{
321 SvGETMAGIC (sv); 415 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 }
344 416
345 svtype svt = SvTYPE (sv); 417 svtype svt = SvTYPE (sv);
346 418
347 if (ecb_expect_false (SvOBJECT (sv))) 419 if (ecb_expect_false (SvOBJECT (sv)))
348 { 420 {
355 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 427 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
356 ? cbor_tagged_stash 428 ? cbor_tagged_stash
357 : gv_stashpv ("CBOR::XS::Tagged" , 1); 429 : gv_stashpv ("CBOR::XS::Tagged" , 1);
358 430
359 HV *stash = SvSTASH (sv); 431 HV *stash = SvSTASH (sv);
360 GV *method;
361 432
362 if (stash == boolean_stash) 433 if (stash == boolean_stash)
363 encode_ch (enc, SvIV (sv) ? 0xe0 | 21 : 0xe0 | 20); 434 {
435 encode_ch (enc, SvIV (sv) ? MAJOR_MISC | SIMPLE_TRUE : MAJOR_MISC | SIMPLE_FALSE);
436 return;
437 }
364 else if (stash == error_stash) 438 else if (stash == error_stash)
365 encode_ch (enc, 0xe0 | 23); 439 {
440 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
441 return;
442 }
366 else if (stash == tagged_stash) 443 else if (stash == tagged_stash)
367 { 444 {
368 if (svt != SVt_PVAV) 445 if (svt != SVt_PVAV)
369 croak ("encountered CBOR::XS::Tagged object that isn't an array"); 446 croak ("encountered CBOR::XS::Tagged object that isn't an array");
370 447
371 encode_uint (enc, 0xc0, SvUV (*av_fetch ((AV *)sv, 0, 1))); 448 encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1)));
372 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1)); 449 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1));
450
451 return;
452 }
453 }
454
455 if (ecb_expect_false (SvREFCNT (sv) > 1)
456 && ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING))
457 {
458 if (!enc->shareable)
459 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
460
461 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
462
463 if (SvOK (*svp))
373 } 464 {
465 encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF);
466 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
467 return;
468 }
469 else
470 {
471 sv_setuv (*svp, enc->shareable_idx);
472 ++enc->shareable_idx;
473 encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE);
474 }
475 }
476
477 if (ecb_expect_false (SvOBJECT (sv)))
478 {
479 HV *stash = SvSTASH (sv);
480 GV *method;
481
374 else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0))) 482 if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
375 { 483 {
376 dSP; 484 dSP;
377 485
378 ENTER; SAVETMPS; PUSHMARK (SP); 486 ENTER; SAVETMPS;
487 PUSHMARK (SP);
379 // we re-bless the reference to get overload and other niceties right 488 // we re-bless the reference to get overload and other niceties right
380 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); 489 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
381 490
382 PUTBACK; 491 PUTBACK;
383 // G_SCALAR ensures that return value is 1 492 // G_SCALAR ensures that return value is 1
396 } 505 }
397 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0) 506 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)
398 { 507 {
399 dSP; 508 dSP;
400 509
401 ENTER; SAVETMPS; PUSHMARK (SP); 510 ENTER; SAVETMPS;
511 SAVESTACK_POS ();
512 PUSHMARK (SP);
402 EXTEND (SP, 2); 513 EXTEND (SP, 2);
403 // we re-bless the reference to get overload and other niceties right 514 // we re-bless the reference to get overload and other niceties right
404 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); 515 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
405 PUSHs (sv_cbor); 516 PUSHs (sv_cbor);
406 517
411 // catch this surprisingly common error 522 // catch this surprisingly common error
412 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv) 523 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)); 524 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));
414 525
415 encode_tag (enc, CBOR_TAG_PERL_OBJECT); 526 encode_tag (enc, CBOR_TAG_PERL_OBJECT);
416 encode_uint (enc, 0x80, count + 1); 527 encode_uint (enc, MAJOR_ARRAY, count + 1);
417 encode_str (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash)); 528 encode_strref (enc, 0, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
418 529
419 while (count) 530 while (count)
420 encode_sv (enc, SP[1 - count--]); 531 encode_sv (enc, SP[1 - count--]);
421 532
422 PUTBACK; 533 PUTBACK;
443{ 554{
444 double nv = SvNVX (sv); 555 double nv = SvNVX (sv);
445 556
446 need (enc, 9); 557 need (enc, 9);
447 558
448 if (ecb_expect_false (nv == (U32)nv)) 559 if (ecb_expect_false (nv == (NV)(U32)nv))
449 encode_uint (enc, 0x00, (U32)nv); 560 encode_uint (enc, MAJOR_POS_INT, (U32)nv);
450 //TODO: maybe I32? 561 //TODO: maybe I32?
451 else if (ecb_expect_false (nv == (float)nv)) 562 else if (ecb_expect_false (nv == (float)nv))
452 { 563 {
453 uint32_t fp = ecb_float_to_binary32 (nv); 564 uint32_t fp = ecb_float_to_binary32 (nv);
454 565
455 *enc->cur++ = 0xe0 | 26; 566 *enc->cur++ = MAJOR_MISC | MISC_FLOAT32;
456 567
457 if (!ecb_big_endian ()) 568 if (!ecb_big_endian ())
458 fp = ecb_bswap32 (fp); 569 fp = ecb_bswap32 (fp);
459 570
460 memcpy (enc->cur, &fp, 4); 571 memcpy (enc->cur, &fp, 4);
462 } 573 }
463 else 574 else
464 { 575 {
465 uint64_t fp = ecb_double_to_binary64 (nv); 576 uint64_t fp = ecb_double_to_binary64 (nv);
466 577
467 *enc->cur++ = 0xe0 | 27; 578 *enc->cur++ = MAJOR_MISC | MISC_FLOAT64;
468 579
469 if (!ecb_big_endian ()) 580 if (!ecb_big_endian ())
470 fp = ecb_bswap64 (fp); 581 fp = ecb_bswap64 (fp);
471 582
472 memcpy (enc->cur, &fp, 8); 583 memcpy (enc->cur, &fp, 8);
481 592
482 if (SvPOKp (sv)) 593 if (SvPOKp (sv))
483 { 594 {
484 STRLEN len; 595 STRLEN len;
485 char *str = SvPV (sv, len); 596 char *str = SvPV (sv, len);
486 encode_str (enc, SvUTF8 (sv), str, len); 597 encode_strref (enc, enc->cbor.flags & F_TEXT_STRINGS, SvUTF8 (sv), str, len);
487 } 598 }
488 else if (SvNOKp (sv)) 599 else if (SvNOKp (sv))
489 encode_nv (enc, sv); 600 encode_nv (enc, sv);
490 else if (SvIOKp (sv)) 601 else if (SvIOKp (sv))
491 { 602 {
492 if (SvIsUV (sv)) 603 if (SvIsUV (sv))
493 encode_uint (enc, 0x00, SvUVX (sv)); 604 encode_uint (enc, MAJOR_POS_INT, SvUVX (sv));
494 else if (SvIVX (sv) >= 0) 605 else if (SvIVX (sv) >= 0)
495 encode_uint (enc, 0x00, SvIVX (sv)); 606 encode_uint (enc, MAJOR_POS_INT, SvIVX (sv));
496 else 607 else
497 encode_uint (enc, 0x20, -(SvIVX (sv) + 1)); 608 encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1));
498 } 609 }
499 else if (SvROK (sv)) 610 else if (SvROK (sv))
500 encode_rv (enc, SvRV (sv)); 611 encode_rv (enc, SvRV (sv));
501 else if (!SvOK (sv)) 612 else if (!SvOK (sv))
502 encode_ch (enc, 0xe0 | 22); 613 encode_ch (enc, MAJOR_MISC | SIMPLE_NULL);
503 else if (enc->cbor.flags & F_ALLOW_UNKNOWN) 614 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
504 encode_ch (enc, 0xe0 | 23); 615 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
505 else 616 else
506 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data", 617 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data",
507 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv)); 618 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
508} 619}
509 620
510static SV * 621static SV *
511encode_cbor (SV *scalar, CBOR *cbor) 622encode_cbor (SV *scalar, CBOR *cbor)
512{ 623{
513 enc_t enc = { }; 624 enc_t enc = { 0 };
514 625
515 enc.cbor = *cbor; 626 enc.cbor = *cbor;
516 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 627 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
517 enc.cur = SvPVX (enc.sv); 628 enc.cur = SvPVX (enc.sv);
518 enc.end = SvEND (enc.sv); 629 enc.end = SvEND (enc.sv);
519 630
520 SvPOK_only (enc.sv); 631 SvPOK_only (enc.sv);
521 632
522 if (cbor->flags & F_ALLOW_STRINGREF) 633 if (cbor->flags & F_PACK_STRINGS)
523 { 634 {
524 encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE); 635 encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE);
525 enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ()); 636 enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ());
526 enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ()); 637 enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ());
527 } 638 }
554 SV *decode_tagged; 665 SV *decode_tagged;
555} dec_t; 666} dec_t;
556 667
557#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE 668#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE
558 669
559#define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data") 670#define WANT(len) if (ecb_expect_false ((UV)(dec->end - dec->cur) < (UV)len)) ERR ("unexpected end of CBOR data")
560 671
561#define DEC_INC_DEPTH if (++dec->depth > dec->cbor.max_depth) ERR (ERR_NESTING_EXCEEDED) 672#define DEC_INC_DEPTH if (ecb_expect_false (++dec->depth > dec->cbor.max_depth)) ERR (ERR_NESTING_EXCEEDED)
562#define DEC_DEC_DEPTH --dec->depth 673#define DEC_DEC_DEPTH --dec->depth
563 674
564static UV 675static UV
565decode_uint (dec_t *dec) 676decode_uint (dec_t *dec)
566{ 677{
567 switch (*dec->cur & 31) 678 U8 m = *dec->cur & MINOR_MASK;
568 { 679 ++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;
573 680
574 case 24: 681 if (ecb_expect_true (m < LENGTH_EXT1))
682 return m;
683 else if (ecb_expect_true (m == LENGTH_EXT1))
684 {
575 WANT (2); 685 WANT (1);
576 dec->cur += 2; 686 dec->cur += 1;
577 return dec->cur[-1]; 687 return dec->cur[-1];
578 688 }
579 case 25: 689 else if (ecb_expect_true (m == LENGTH_EXT2))
690 {
580 WANT (3); 691 WANT (2);
581 dec->cur += 3; 692 dec->cur += 2;
582 return (((UV)dec->cur[-2]) << 8) 693 return (((UV)dec->cur[-2]) << 8)
583 | ((UV)dec->cur[-1]); 694 | ((UV)dec->cur[-1]);
584 695 }
585 case 26: 696 else if (ecb_expect_true (m == LENGTH_EXT4))
697 {
586 WANT (5); 698 WANT (4);
587 dec->cur += 5; 699 dec->cur += 4;
588 return (((UV)dec->cur[-4]) << 24) 700 return (((UV)dec->cur[-4]) << 24)
589 | (((UV)dec->cur[-3]) << 16) 701 | (((UV)dec->cur[-3]) << 16)
590 | (((UV)dec->cur[-2]) << 8) 702 | (((UV)dec->cur[-2]) << 8)
591 | ((UV)dec->cur[-1]); 703 | ((UV)dec->cur[-1]);
592 704 }
593 case 27: 705 else if (ecb_expect_true (m == LENGTH_EXT8))
706 {
594 WANT (9); 707 WANT (8);
595 dec->cur += 9; 708 dec->cur += 8;
709
710 return
711#if UVSIZE < 8
712 0
713#else
596 return (((UV)dec->cur[-8]) << 56) 714 (((UV)dec->cur[-8]) << 56)
597 | (((UV)dec->cur[-7]) << 48) 715 | (((UV)dec->cur[-7]) << 48)
598 | (((UV)dec->cur[-6]) << 40) 716 | (((UV)dec->cur[-6]) << 40)
599 | (((UV)dec->cur[-5]) << 32) 717 | (((UV)dec->cur[-5]) << 32)
718#endif
600 | (((UV)dec->cur[-4]) << 24) 719 | (((UV)dec->cur[-4]) << 24)
601 | (((UV)dec->cur[-3]) << 16) 720 | (((UV)dec->cur[-3]) << 16)
602 | (((UV)dec->cur[-2]) << 8) 721 | (((UV)dec->cur[-2]) << 8)
603 | ((UV)dec->cur[-1]); 722 | ((UV)dec->cur[-1]);
604 723 }
605 default: 724 else
606 ERR ("corrupted CBOR data (unsupported integer minor encoding)"); 725 ERR ("corrupted CBOR data (unsupported integer minor encoding)");
607 }
608 726
609fail: 727fail:
610 return 0; 728 return 0;
611} 729}
612 730
617{ 735{
618 AV *av = newAV (); 736 AV *av = newAV ();
619 737
620 DEC_INC_DEPTH; 738 DEC_INC_DEPTH;
621 739
622 if ((*dec->cur & 31) == 31) 740 if (*dec->cur == (MAJOR_ARRAY | MINOR_INDEF))
623 { 741 {
624 ++dec->cur; 742 ++dec->cur;
625 743
626 for (;;) 744 for (;;)
627 { 745 {
628 WANT (1); 746 WANT (1);
629 747
630 if (*dec->cur == (0xe0 | 31)) 748 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
631 { 749 {
632 ++dec->cur; 750 ++dec->cur;
633 break; 751 break;
634 } 752 }
635 753
636 av_push (av, decode_sv (dec)); 754 av_push (av, decode_sv (dec));
637 } 755 }
638 } 756 }
639 else 757 else
640 { 758 {
641 int i, len = decode_uint (dec); 759 UV i, len = decode_uint (dec);
642 760
761 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); 762 av_fill (av, len - 1);
644 763
645 for (i = 0; i < len; ++i) 764 for (i = 0; i < len; ++i)
646 AvARRAY (av)[i] = decode_sv (dec); 765 AvARRAY (av)[i] = decode_sv (dec);
647 } 766 }
660{ 779{
661 // for speed reasons, we specialcase single-string 780 // for speed reasons, we specialcase single-string
662 // byte or utf-8 strings as keys, but only when !stringref 781 // byte or utf-8 strings as keys, but only when !stringref
663 782
664 if (ecb_expect_true (!dec->stringref)) 783 if (ecb_expect_true (!dec->stringref))
665 if (*dec->cur >= 0x40 && *dec->cur <= 0x40 + 27) 784 if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8))
666 { 785 {
667 I32 len = decode_uint (dec); 786 STRLEN len = decode_uint (dec);
668 char *key = (char *)dec->cur; 787 char *key = (char *)dec->cur;
669 788
789 WANT (len);
670 dec->cur += len; 790 dec->cur += len;
671 791
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); 792 hv_store (hv, key, len, decode_sv (dec), 0);
676 793
677 return; 794 return;
678 } 795 }
679 else if (*dec->cur >= 0x60 && *dec->cur <= 0x60 + 27) 796 else if (ecb_expect_true ((U8)(*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8))
680 { 797 {
681 I32 len = decode_uint (dec); 798 STRLEN len = decode_uint (dec);
682 char *key = (char *)dec->cur; 799 char *key = (char *)dec->cur;
683 800
801 WANT (len);
684 dec->cur += len; 802 dec->cur += len;
685 803
686 if (ecb_expect_false (dec->stringref)) 804 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
687 av_push (dec->stringref, newSVpvn_utf8 (key, len, 1)); 805 if (!is_utf8_string (key, len))
806 ERR ("corrupted CBOR data (invalid UTF-8 in map key)");
688 807
689 hv_store (hv, key, -len, decode_sv (dec), 0); 808 hv_store (hv, key, -len, decode_sv (dec), 0);
690 809
691 return; 810 return;
692 } 811 }
694 SV *k = decode_sv (dec); 813 SV *k = decode_sv (dec);
695 SV *v = decode_sv (dec); 814 SV *v = decode_sv (dec);
696 815
697 hv_store_ent (hv, k, v, 0); 816 hv_store_ent (hv, k, v, 0);
698 SvREFCNT_dec (k); 817 SvREFCNT_dec (k);
818
819fail:
820 ;
699} 821}
700 822
701static SV * 823static SV *
702decode_hv (dec_t *dec) 824decode_hv (dec_t *dec)
703{ 825{
704 HV *hv = newHV (); 826 HV *hv = newHV ();
705 827
706 DEC_INC_DEPTH; 828 DEC_INC_DEPTH;
707 829
708 if ((*dec->cur & 31) == 31) 830 if (*dec->cur == (MAJOR_MAP | MINOR_INDEF))
709 { 831 {
710 ++dec->cur; 832 ++dec->cur;
711 833
712 for (;;) 834 for (;;)
713 { 835 {
714 WANT (1); 836 WANT (1);
715 837
716 if (*dec->cur == (0xe0 | 31)) 838 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
717 { 839 {
718 ++dec->cur; 840 ++dec->cur;
719 break; 841 break;
720 } 842 }
721 843
722 decode_he (dec, hv); 844 decode_he (dec, hv);
723 } 845 }
724 } 846 }
725 else 847 else
726 { 848 {
727 int pairs = decode_uint (dec); 849 UV pairs = decode_uint (dec);
850
851 WANT (pairs); // complexity check - need at least one byte per value, do not allow supersize hashes
728 852
729 while (pairs--) 853 while (pairs--)
730 decode_he (dec, hv); 854 decode_he (dec, hv);
731 } 855 }
732 856
742static SV * 866static SV *
743decode_str (dec_t *dec, int utf8) 867decode_str (dec_t *dec, int utf8)
744{ 868{
745 SV *sv = 0; 869 SV *sv = 0;
746 870
747 if ((*dec->cur & 31) == 31) 871 if ((*dec->cur & MINOR_MASK) == MINOR_INDEF)
748 { 872 {
873 // indefinite length strings
749 ++dec->cur; 874 ++dec->cur;
750 875
876 U8 major = *dec->cur & MAJOR_MISC;
877
751 sv = newSVpvn ("", 0); 878 sv = newSVpvn ("", 0);
752 879
753 // not very fast, and certainly not robust against illegal input
754 for (;;) 880 for (;;)
755 { 881 {
756 WANT (1); 882 WANT (1);
757 883
758 if (*dec->cur == (0xe0 | 31)) 884 if ((*dec->cur - major) > LENGTH_EXT8)
885 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
759 { 886 {
760 ++dec->cur; 887 ++dec->cur;
761 break; 888 break;
762 } 889 }
890 else
891 ERR ("corrupted CBOR data (invalid chunks in indefinite length string)");
763 892
764 sv_catsv (sv, decode_sv (dec)); 893 STRLEN len = decode_uint (dec);
894
895 WANT (len);
896 sv_catpvn (sv, dec->cur, len);
897 dec->cur += len;
765 } 898 }
766 } 899 }
767 else 900 else
768 { 901 {
769 STRLEN len = decode_uint (dec); 902 STRLEN len = decode_uint (dec);
776 && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1)) 909 && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1))
777 av_push (dec->stringref, SvREFCNT_inc_NN (sv)); 910 av_push (dec->stringref, SvREFCNT_inc_NN (sv));
778 } 911 }
779 912
780 if (utf8) 913 if (utf8)
914 {
915 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
916 if (!is_utf8_string (SvPVX (sv), SvCUR (sv)))
917 ERR ("corrupted CBOR data (invalid UTF-8 in text string)");
918
781 SvUTF8_on (sv); 919 SvUTF8_on (sv);
920 }
782 921
783 return sv; 922 return sv;
784 923
785fail: 924fail:
786 SvREFCNT_dec (sv); 925 SvREFCNT_dec (sv);
805 sv = newRV_noinc (decode_sv (dec)); 944 sv = newRV_noinc (decode_sv (dec));
806 break; 945 break;
807 946
808 case CBOR_TAG_STRINGREF_NAMESPACE: 947 case CBOR_TAG_STRINGREF_NAMESPACE:
809 { 948 {
949 // do nmot use SAVETMPS/FREETMPS, as these will
950 // erase mortalised caches, e.g. "shareable"
810 ENTER; SAVETMPS; 951 ENTER;
811 952
812 SAVESPTR (dec->stringref); 953 SAVESPTR (dec->stringref);
813 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ()); 954 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ());
814 955
815 sv = decode_sv (dec); 956 sv = decode_sv (dec);
816 957
817 FREETMPS; LEAVE; 958 LEAVE;
818 } 959 }
819 break; 960 break;
820 961
821 case CBOR_TAG_STRINGREF: 962 case CBOR_TAG_STRINGREF:
822 { 963 {
823 if ((*dec->cur >> 5) != 0) 964 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
824 ERR ("corrupted CBOR data (stringref index not an unsigned integer)"); 965 ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
825 966
826 UV idx = decode_uint (dec); 967 UV idx = decode_uint (dec);
827 968
828 if (!dec->stringref || (int)idx > AvFILLp (dec->stringref)) 969 if (!dec->stringref || (int)idx > AvFILLp (dec->stringref))
835 case CBOR_TAG_VALUE_SHAREABLE: 976 case CBOR_TAG_VALUE_SHAREABLE:
836 { 977 {
837 if (ecb_expect_false (!dec->shareable)) 978 if (ecb_expect_false (!dec->shareable))
838 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ()); 979 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ());
839 980
981 if (dec->cbor.flags & F_ALLOW_CYCLES)
982 {
840 sv = newSV (0); 983 sv = newSV (0);
841 av_push (dec->shareable, SvREFCNT_inc_NN (sv)); 984 av_push (dec->shareable, SvREFCNT_inc_NN (sv));
842 985
843 SV *osv = decode_sv (dec); 986 SV *osv = decode_sv (dec);
844 sv_setsv (sv, osv); 987 sv_setsv (sv, osv);
845 SvREFCNT_dec_NN (osv); 988 SvREFCNT_dec_NN (osv);
989 }
990 else
991 {
992 av_push (dec->shareable, &PL_sv_undef);
993 int idx = AvFILLp (dec->shareable);
994 sv = decode_sv (dec);
995 av_store (dec->shareable, idx, SvREFCNT_inc_NN (sv));
996 }
846 } 997 }
847 break; 998 break;
848 999
849 case CBOR_TAG_VALUE_SHAREDREF: 1000 case CBOR_TAG_VALUE_SHAREDREF:
850 { 1001 {
851 if ((*dec->cur >> 5) != 0) 1002 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
852 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)"); 1003 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
853 1004
854 UV idx = decode_uint (dec); 1005 UV idx = decode_uint (dec);
855 1006
856 if (!dec->shareable || (int)idx > AvFILLp (dec->shareable)) 1007 if (!dec->shareable || (int)idx > AvFILLp (dec->shareable))
857 ERR ("corrupted CBOR data (sharedref index out of bounds)"); 1008 ERR ("corrupted CBOR data (sharedref index out of bounds)");
858 1009
859 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]); 1010 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]);
1011
1012 if (sv == &PL_sv_undef)
1013 ERR ("cyclic CBOR data structure found, but allow_cycles is not enabled");
860 } 1014 }
861 break; 1015 break;
862 1016
863 case CBOR_TAG_PERL_OBJECT: 1017 case CBOR_TAG_PERL_OBJECT:
864 { 1018 {
879 if (!method) 1033 if (!method)
880 ERR ("cannot decode perl-object (package does not have a THAW method)"); 1034 ERR ("cannot decode perl-object (package does not have a THAW method)");
881 1035
882 dSP; 1036 dSP;
883 1037
884 ENTER; SAVETMPS; PUSHMARK (SP); 1038 ENTER; SAVETMPS;
1039 PUSHMARK (SP);
885 EXTEND (SP, len + 1); 1040 EXTEND (SP, len + 1);
886 // we re-bless the reference to get overload and other niceties right 1041 // we re-bless the reference to get overload and other niceties right
887 PUSHs (*av_fetch (av, 0, 1)); 1042 PUSHs (*av_fetch (av, 0, 1));
888 PUSHs (sv_cbor); 1043 PUSHs (sv_cbor);
889 1044
911 } 1066 }
912 break; 1067 break;
913 1068
914 default: 1069 default:
915 { 1070 {
1071 SV *tag_sv = newSVuv (tag);
1072
916 sv = decode_sv (dec); 1073 sv = decode_sv (dec);
917 1074
918 dSP; 1075 dSP;
919 ENTER; SAVETMPS; PUSHMARK (SP); 1076 ENTER; SAVETMPS;
1077 SAVESTACK_POS ();
1078 PUSHMARK (SP);
920 EXTEND (SP, 2); 1079 EXTEND (SP, 2);
921 PUSHs (newSVuv (tag)); 1080 PUSHs (tag_sv);
922 PUSHs (sv); 1081 PUSHs (sv);
923 1082
924 PUTBACK; 1083 PUTBACK;
925 int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL); 1084 int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL);
926 SPAGAIN; 1085 SPAGAIN;
927 1086
928 if (SvTRUE (ERRSV)) 1087 if (SvTRUE (ERRSV))
929 { 1088 {
1089 SvREFCNT_dec (tag_sv);
930 FREETMPS; LEAVE; 1090 FREETMPS; LEAVE;
931 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV)))); 1091 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV))));
932 } 1092 }
933 1093
934 if (count) 1094 if (count)
935 { 1095 {
1096 SvREFCNT_dec (tag_sv);
936 SvREFCNT_dec (sv); 1097 SvREFCNT_dec (sv);
937 sv = SvREFCNT_inc (POPs); 1098 sv = SvREFCNT_inc (POPs);
938 } 1099 }
939 else 1100 else
940 { 1101 {
941 AV *av = newAV (); 1102 AV *av = newAV ();
942 av_push (av, newSVuv (tag)); 1103 av_push (av, tag_sv);
943 av_push (av, sv); 1104 av_push (av, sv);
944 1105
945 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 1106 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
946 ? cbor_tagged_stash 1107 ? cbor_tagged_stash
947 : gv_stashpv ("CBOR::XS::Tagged" , 1); 1108 : gv_stashpv ("CBOR::XS::Tagged" , 1);
965static SV * 1126static SV *
966decode_sv (dec_t *dec) 1127decode_sv (dec_t *dec)
967{ 1128{
968 WANT (1); 1129 WANT (1);
969 1130
970 switch (*dec->cur >> 5) 1131 switch (*dec->cur >> MAJOR_SHIFT)
971 { 1132 {
972 case 0: // unsigned int 1133 case MAJOR_POS_INT >> MAJOR_SHIFT: return newSVuv (decode_uint (dec));
973 return newSVuv (decode_uint (dec)); 1134 case MAJOR_NEG_INT >> MAJOR_SHIFT: return newSViv (-1 - (IV)decode_uint (dec));
974 case 1: // negative int 1135 case MAJOR_BYTES >> MAJOR_SHIFT: return decode_str (dec, 0);
975 return newSViv (-1 - (IV)decode_uint (dec)); 1136 case MAJOR_TEXT >> MAJOR_SHIFT: return decode_str (dec, 1);
976 case 2: // octet string 1137 case MAJOR_ARRAY >> MAJOR_SHIFT: return decode_av (dec);
977 return decode_str (dec, 0); 1138 case MAJOR_MAP >> MAJOR_SHIFT: return decode_hv (dec);
978 case 3: // utf-8 string 1139 case MAJOR_TAG >> MAJOR_SHIFT: return decode_tagged (dec);
979 return decode_str (dec, 1); 1140
980 case 4: // array 1141 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) 1142 switch (*dec->cur++ & MINOR_MASK)
988 { 1143 {
989 case 20: 1144 case SIMPLE_FALSE:
990#if CBOR_SLOW 1145#if CBOR_SLOW
991 types_false = get_bool ("Types::Serialiser::false"); 1146 types_false = get_bool ("Types::Serialiser::false");
992#endif 1147#endif
993 return newSVsv (types_false); 1148 return newSVsv (types_false);
994 case 21: 1149 case SIMPLE_TRUE:
995#if CBOR_SLOW 1150#if CBOR_SLOW
996 types_true = get_bool ("Types::Serialiser::true"); 1151 types_true = get_bool ("Types::Serialiser::true");
997#endif 1152#endif
998 return newSVsv (types_true); 1153 return newSVsv (types_true);
999 case 22: 1154 case SIMPLE_NULL:
1000 return newSVsv (&PL_sv_undef); 1155 return newSVsv (&PL_sv_undef);
1001 case 23: 1156 case SIMPLE_UNDEF:
1002#if CBOR_SLOW 1157#if CBOR_SLOW
1003 types_error = get_bool ("Types::Serialiser::error"); 1158 types_error = get_bool ("Types::Serialiser::error");
1004#endif 1159#endif
1005 return newSVsv (types_error); 1160 return newSVsv (types_error);
1006 1161
1007 case 25: 1162 case MISC_FLOAT16:
1008 { 1163 {
1009 WANT (2); 1164 WANT (2);
1010 1165
1011 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1]; 1166 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1];
1012 dec->cur += 2; 1167 dec->cur += 2;
1013 1168
1014 return newSVnv (ecb_binary16_to_float (fp)); 1169 return newSVnv (ecb_binary16_to_float (fp));
1015 } 1170 }
1016 1171
1017 case 26: 1172 case MISC_FLOAT32:
1018 { 1173 {
1019 uint32_t fp; 1174 uint32_t fp;
1020 WANT (4); 1175 WANT (4);
1021 memcpy (&fp, dec->cur, 4); 1176 memcpy (&fp, dec->cur, 4);
1022 dec->cur += 4; 1177 dec->cur += 4;
1025 fp = ecb_bswap32 (fp); 1180 fp = ecb_bswap32 (fp);
1026 1181
1027 return newSVnv (ecb_binary32_to_float (fp)); 1182 return newSVnv (ecb_binary32_to_float (fp));
1028 } 1183 }
1029 1184
1030 case 27: 1185 case MISC_FLOAT64:
1031 { 1186 {
1032 uint64_t fp; 1187 uint64_t fp;
1033 WANT (8); 1188 WANT (8);
1034 memcpy (&fp, dec->cur, 8); 1189 memcpy (&fp, dec->cur, 8);
1035 dec->cur += 8; 1190 dec->cur += 8;
1038 fp = ecb_bswap64 (fp); 1193 fp = ecb_bswap64 (fp);
1039 1194
1040 return newSVnv (ecb_binary64_to_double (fp)); 1195 return newSVnv (ecb_binary64_to_double (fp));
1041 } 1196 }
1042 1197
1043 // 0..19 unassigned 1198 // 0..19 unassigned simple
1044 // 24 reserved + unassigned (reserved values are not encodable) 1199 // 24 reserved + unassigned simple (reserved values are not encodable)
1200 // 28-30 unassigned misc
1201 // 31 break code
1045 default: 1202 default:
1046 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)"); 1203 ERR ("corrupted CBOR data (reserved/unassigned/unexpected major 7 value)");
1047 } 1204 }
1048 1205
1049 break; 1206 break;
1050 } 1207 }
1051 1208
1054} 1211}
1055 1212
1056static SV * 1213static SV *
1057decode_cbor (SV *string, CBOR *cbor, char **offset_return) 1214decode_cbor (SV *string, CBOR *cbor, char **offset_return)
1058{ 1215{
1059 dec_t dec = { }; 1216 dec_t dec = { 0 };
1060 SV *sv; 1217 SV *sv;
1061 STRLEN len; 1218 STRLEN len;
1062 char *data = SvPVbyte (string, len); 1219 char *data = SvPVbyte (string, len);
1063 1220
1064 if (len > cbor->max_size && cbor->max_size) 1221 if (len > cbor->max_size && cbor->max_size)
1078 if (dec.cur != dec.end && !dec.err) 1235 if (dec.cur != dec.end && !dec.err)
1079 dec.err = "garbage after CBOR object"; 1236 dec.err = "garbage after CBOR object";
1080 1237
1081 if (dec.err) 1238 if (dec.err)
1082 { 1239 {
1240 if (dec.shareable)
1241 {
1242 // need to break cyclic links, which would all be in shareable
1243 int i;
1244 SV **svp;
1245
1246 for (i = av_len (dec.shareable) + 1; i--; )
1247 if ((svp = av_fetch (dec.shareable, i, 0)))
1248 sv_setsv (*svp, &PL_sv_undef);
1249 }
1250
1083 SvREFCNT_dec (sv); 1251 SvREFCNT_dec (sv);
1084 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur); 1252 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur);
1085 } 1253 }
1086 1254
1087 sv = sv_2mortal (sv); 1255 sv = sv_2mortal (sv);
1088 1256
1089 return sv; 1257 return sv;
1090} 1258}
1091 1259
1260/////////////////////////////////////////////////////////////////////////////
1261// incremental parser
1262
1263#define INCR_DONE(cbor) (AvFILLp (cbor->incr_count) < 0)
1264
1265// returns 0 for notyet, 1 for success or error
1266static int
1267incr_parse (CBOR *self, SV *cborstr)
1268{
1269 STRLEN cur;
1270 SvPV (cborstr, cur);
1271
1272 while (ecb_expect_true (self->incr_need <= cur))
1273 {
1274 // table of integer count bytes
1275 static I8 incr_len[MINOR_MASK + 1] = {
1276 0, 0, 0, 0, 0, 0, 0, 0,
1277 0, 0, 0, 0, 0, 0, 0, 0,
1278 0, 0, 0, 0, 0, 0, 0, 0,
1279 1, 2, 4, 8,-1,-1,-1,-2
1280 };
1281
1282 const U8 *p = SvPVX (cborstr) + self->incr_pos;
1283 U8 m = *p & MINOR_MASK;
1284 IV count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1285 I8 ilen = incr_len[m];
1286
1287 self->incr_need = self->incr_pos + 1;
1288
1289 if (ecb_expect_false (ilen < 0))
1290 {
1291 if (m != MINOR_INDEF)
1292 return 1; // error
1293
1294 if (*p == (MAJOR_MISC | MINOR_INDEF))
1295 {
1296 if (count >= 0)
1297 return 1; // error
1298
1299 count = 1;
1300 }
1301 else
1302 {
1303 av_push (self->incr_count, newSViv (-1)); //TODO: nest
1304 count = -1;
1305 }
1306 }
1307 else
1308 {
1309 self->incr_need += ilen;
1310 if (ecb_expect_false (self->incr_need > cur))
1311 return 0;
1312
1313 int major = *p >> MAJOR_SHIFT;
1314
1315 switch (major)
1316 {
1317 case MAJOR_TAG >> MAJOR_SHIFT:
1318 ++count; // tags merely prefix another value
1319 break;
1320
1321 case MAJOR_BYTES >> MAJOR_SHIFT:
1322 case MAJOR_TEXT >> MAJOR_SHIFT:
1323 case MAJOR_ARRAY >> MAJOR_SHIFT:
1324 case MAJOR_MAP >> MAJOR_SHIFT:
1325 {
1326 UV len;
1327
1328 if (ecb_expect_false (ilen))
1329 {
1330 len = 0;
1331
1332 do {
1333 len = (len << 8) | *++p;
1334 } while (--ilen);
1335 }
1336 else
1337 len = m;
1338
1339 switch (major)
1340 {
1341 case MAJOR_BYTES >> MAJOR_SHIFT:
1342 case MAJOR_TEXT >> MAJOR_SHIFT:
1343 self->incr_need += len;
1344 if (ecb_expect_false (self->incr_need > cur))
1345 return 0;
1346
1347 break;
1348
1349 case MAJOR_MAP >> MAJOR_SHIFT:
1350 len <<= 1;
1351 case MAJOR_ARRAY >> MAJOR_SHIFT:
1352 if (len)
1353 {
1354 av_push (self->incr_count, newSViv (len + 1)); //TODO: nest
1355 count = len + 1;
1356 }
1357 break;
1358 }
1359 }
1360 }
1361 }
1362
1363 self->incr_pos = self->incr_need;
1364
1365 if (count > 0)
1366 {
1367 while (!--count)
1368 {
1369 if (!AvFILLp (self->incr_count))
1370 return 1; // done
1371
1372 SvREFCNT_dec_NN (av_pop (self->incr_count));
1373 count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1374 }
1375
1376 SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]) = count;
1377 }
1378 }
1379
1380 return 0;
1381}
1382
1383
1092///////////////////////////////////////////////////////////////////////////// 1384/////////////////////////////////////////////////////////////////////////////
1093// XS interface functions 1385// XS interface functions
1094 1386
1095MODULE = CBOR::XS PACKAGE = CBOR::XS 1387MODULE = CBOR::XS PACKAGE = CBOR::XS
1096 1388
1108 1400
1109 default_filter = newSVpv ("CBOR::XS::default_filter", 0); 1401 default_filter = newSVpv ("CBOR::XS::default_filter", 0);
1110 1402
1111 sv_cbor = newSVpv ("CBOR", 0); 1403 sv_cbor = newSVpv ("CBOR", 0);
1112 SvREADONLY_on (sv_cbor); 1404 SvREADONLY_on (sv_cbor);
1405
1406 assert (("STRLEN must be an unsigned type", 0 <= (STRLEN)-1));
1113} 1407}
1114 1408
1115PROTOTYPES: DISABLE 1409PROTOTYPES: DISABLE
1116 1410
1117void CLONE (...) 1411void CLONE (...)
1136void shrink (CBOR *self, int enable = 1) 1430void shrink (CBOR *self, int enable = 1)
1137 ALIAS: 1431 ALIAS:
1138 shrink = F_SHRINK 1432 shrink = F_SHRINK
1139 allow_unknown = F_ALLOW_UNKNOWN 1433 allow_unknown = F_ALLOW_UNKNOWN
1140 allow_sharing = F_ALLOW_SHARING 1434 allow_sharing = F_ALLOW_SHARING
1141 allow_stringref = F_ALLOW_STRINGREF 1435 allow_cycles = F_ALLOW_CYCLES
1436 pack_strings = F_PACK_STRINGS
1437 text_keys = F_TEXT_KEYS
1438 text_strings = F_TEXT_STRINGS
1439 validate_utf8 = F_VALIDATE_UTF8
1142 PPCODE: 1440 PPCODE:
1143{ 1441{
1144 if (enable) 1442 if (enable)
1145 self->flags |= ix; 1443 self->flags |= ix;
1146 else 1444 else
1152void get_shrink (CBOR *self) 1450void get_shrink (CBOR *self)
1153 ALIAS: 1451 ALIAS:
1154 get_shrink = F_SHRINK 1452 get_shrink = F_SHRINK
1155 get_allow_unknown = F_ALLOW_UNKNOWN 1453 get_allow_unknown = F_ALLOW_UNKNOWN
1156 get_allow_sharing = F_ALLOW_SHARING 1454 get_allow_sharing = F_ALLOW_SHARING
1157 get_allow_stringref = F_ALLOW_STRINGREF 1455 get_allow_cycles = F_ALLOW_CYCLES
1456 get_pack_strings = F_PACK_STRINGS
1457 get_text_keys = F_TEXT_KEYS
1458 get_text_strings = F_TEXT_STRINGS
1459 get_validate_utf8 = F_VALIDATE_UTF8
1158 PPCODE: 1460 PPCODE:
1159 XPUSHs (boolSV (self->flags & ix)); 1461 XPUSHs (boolSV (self->flags & ix));
1160 1462
1161void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) 1463void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
1162 PPCODE: 1464 PPCODE:
1211 EXTEND (SP, 2); 1513 EXTEND (SP, 2);
1212 PUSHs (sv); 1514 PUSHs (sv);
1213 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); 1515 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
1214} 1516}
1215 1517
1518void incr_parse (CBOR *self, SV *cborstr)
1519 ALIAS:
1520 incr_parse_multiple = 1
1521 PPCODE:
1522{
1523 if (SvUTF8 (cborstr))
1524 sv_utf8_downgrade (cborstr, 0);
1525
1526 if (!self->incr_count)
1527 {
1528 self->incr_count = newAV ();
1529 self->incr_pos = 0;
1530 self->incr_need = 1;
1531
1532 av_push (self->incr_count, newSViv (1));
1533 }
1534
1535 do
1536 {
1537 if (!incr_parse (self, cborstr))
1538 {
1539 if (self->incr_need > self->max_size && self->max_size)
1540 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
1541 (unsigned long)self->incr_need, (unsigned long)self->max_size);
1542
1543 break;
1544 }
1545
1546 SV *sv;
1547 char *offset;
1548
1549 PUTBACK; sv = decode_cbor (cborstr, self, &offset); SPAGAIN;
1550 XPUSHs (sv);
1551
1552 sv_chop (cborstr, offset);
1553
1554 av_clear (self->incr_count);
1555 av_push (self->incr_count, newSViv (1));
1556
1557 self->incr_pos = 0;
1558 self->incr_need = self->incr_pos + 1;
1559 }
1560 while (ix);
1561}
1562
1563void incr_reset (CBOR *self)
1564 CODE:
1565{
1566 SvREFCNT_dec (self->incr_count);
1567 self->incr_count = 0;
1568}
1569
1216void DESTROY (CBOR *self) 1570void DESTROY (CBOR *self)
1217 PPCODE: 1571 PPCODE:
1218 cbor_free (self); 1572 cbor_free (self);
1219 1573
1220PROTOTYPES: ENABLE 1574PROTOTYPES: ENABLE
1221 1575
1222void encode_cbor (SV *scalar) 1576void encode_cbor (SV *scalar)
1577 ALIAS:
1578 encode_cbor = 0
1579 encode_cbor_sharing = F_ALLOW_SHARING
1223 PPCODE: 1580 PPCODE:
1224{ 1581{
1225 CBOR cbor; 1582 CBOR cbor;
1226 cbor_init (&cbor); 1583 cbor_init (&cbor);
1584 cbor.flags |= ix;
1227 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN; 1585 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;
1228 XPUSHs (scalar); 1586 XPUSHs (scalar);
1229} 1587}
1230 1588
1231void decode_cbor (SV *cborstr) 1589void decode_cbor (SV *cborstr)

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines