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 1.28 by root, Sat Nov 23 18:30:59 2013 UTC vs.
Revision 1.60 by root, Sat Nov 26 00:47:02 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 // inofficial extensions (pending iana registration) 69 // extensions
70 CBOR_TAG_STRINGREF = 25, // http://cbor.schmorp.de/stringref
32 CBOR_TAG_PERL_OBJECT = 24, // http://cbor.schmorp.de/perl-object 71 CBOR_TAG_PERL_OBJECT = 26, // http://cbor.schmorp.de/perl-object
33 CBOR_TAG_GENERIC_OBJECT = 25, // http://cbor.schmorp.de/generic-object 72 CBOR_TAG_GENERIC_OBJECT = 27, // http://cbor.schmorp.de/generic-object
34 CBOR_TAG_VALUE_SHAREABLE = 26, // http://cbor.schmorp.de/value-sharing 73 CBOR_TAG_VALUE_SHAREABLE = 28, // http://cbor.schmorp.de/value-sharing
35 CBOR_TAG_VALUE_SHAREDREF = 27, // http://cbor.schmorp.de/value-sharing 74 CBOR_TAG_VALUE_SHAREDREF = 29, // http://cbor.schmorp.de/value-sharing
36 CBOR_TAG_STRINGREF_NAMESPACE = 65537, // http://cbor.schmorp.de/stringref 75 CBOR_TAG_STRINGREF_NAMESPACE = 256, // http://cbor.schmorp.de/stringref
37 CBOR_TAG_STRINGREF = 28, // 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 //TODO 102#define F_ALLOW_SHARING 0x00000004UL
65#define F_ALLOW_STRINGREF 0x00000008UL //TODO 103#define F_ALLOW_CYCLES 0x00000008UL
104#define F_FORBID_OBJECTS 0x00000010UL
105#define F_PACK_STRINGS 0x00000020UL
106#define F_TEXT_KEYS 0x00000040UL
107#define F_TEXT_STRINGS 0x00000080UL
108#define F_VALIDATE_UTF8 0x00000100UL
66 109
67#define INIT_SIZE 32 // initial scalar size to be allocated 110#define INIT_SIZE 32 // initial scalar size to be allocated
68 111
69#define SB do { 112#define SB do {
70#define SE } while (0) 113#define SE } while (0)
89typedef struct { 132typedef struct {
90 U32 flags; 133 U32 flags;
91 U32 max_depth; 134 U32 max_depth;
92 STRLEN max_size; 135 STRLEN max_size;
93 SV *filter; 136 SV *filter;
137
138 // for the incremental parser
139 STRLEN incr_pos; // the current offset into the text
140 STRLEN incr_need; // minimum bytes needed to decode
141 AV *incr_count; // for every nesting level, the number of outstanding values, or -1 for indef.
94} CBOR; 142} CBOR;
95 143
96ecb_inline void 144ecb_inline void
97cbor_init (CBOR *cbor) 145cbor_init (CBOR *cbor)
98{ 146{
102 150
103ecb_inline void 151ecb_inline void
104cbor_free (CBOR *cbor) 152cbor_free (CBOR *cbor)
105{ 153{
106 SvREFCNT_dec (cbor->filter); 154 SvREFCNT_dec (cbor->filter);
155 SvREFCNT_dec (cbor->incr_count);
107} 156}
108 157
109///////////////////////////////////////////////////////////////////////////// 158/////////////////////////////////////////////////////////////////////////////
110// utility functions 159// utility functions
111 160
141{ 190{
142 return idx > 23 191 return idx > 23
143 ? idx > 0xffU 192 ? idx > 0xffU
144 ? idx > 0xffffU 193 ? idx > 0xffffU
145 ? idx > 0xffffffffU 194 ? idx > 0xffffffffU
195 ? 11
146 ? 7 196 : 7
147 : 6
148 : 5 197 : 5
149 : 4 198 : 4
150 : 3; 199 : 3;
151} 200}
152 201
189static void 238static void
190encode_uint (enc_t *enc, int major, UV len) 239encode_uint (enc_t *enc, int major, UV len)
191{ 240{
192 need (enc, 9); 241 need (enc, 9);
193 242
194 if (len < 24) 243 if (ecb_expect_true (len < LENGTH_EXT1))
195 *enc->cur++ = major | len; 244 *enc->cur++ = major | len;
196 else if (len <= 0xff) 245 else if (ecb_expect_true (len <= 0xffU))
197 { 246 {
198 *enc->cur++ = major | 24; 247 *enc->cur++ = major | LENGTH_EXT1;
199 *enc->cur++ = len; 248 *enc->cur++ = len;
200 } 249 }
201 else if (len <= 0xffff) 250 else if (len <= 0xffffU)
202 { 251 {
203 *enc->cur++ = major | 25; 252 *enc->cur++ = major | LENGTH_EXT2;
204 *enc->cur++ = len >> 8; 253 *enc->cur++ = len >> 8;
205 *enc->cur++ = len; 254 *enc->cur++ = len;
206 } 255 }
207 else if (len <= 0xffffffff) 256 else if (len <= 0xffffffffU)
208 { 257 {
209 *enc->cur++ = major | 26; 258 *enc->cur++ = major | LENGTH_EXT4;
210 *enc->cur++ = len >> 24; 259 *enc->cur++ = len >> 24;
211 *enc->cur++ = len >> 16; 260 *enc->cur++ = len >> 16;
212 *enc->cur++ = len >> 8; 261 *enc->cur++ = len >> 8;
213 *enc->cur++ = len; 262 *enc->cur++ = len;
214 } 263 }
215 else 264 else
216 { 265 {
217 *enc->cur++ = major | 27; 266 *enc->cur++ = major | LENGTH_EXT8;
218 *enc->cur++ = len >> 56; 267 *enc->cur++ = len >> 56;
219 *enc->cur++ = len >> 48; 268 *enc->cur++ = len >> 48;
220 *enc->cur++ = len >> 40; 269 *enc->cur++ = len >> 40;
221 *enc->cur++ = len >> 32; 270 *enc->cur++ = len >> 32;
222 *enc->cur++ = len >> 24; 271 *enc->cur++ = len >> 24;
227} 276}
228 277
229ecb_inline void 278ecb_inline void
230encode_tag (enc_t *enc, UV tag) 279encode_tag (enc_t *enc, UV tag)
231{ 280{
232 encode_uint (enc, 0xc0, tag); 281 encode_uint (enc, MAJOR_TAG, tag);
233} 282}
234 283
235static void 284// exceptional (hopefully) slow path for byte strings that need to be utf8-encoded
285ecb_noinline static void
236encode_str (enc_t *enc, int utf8, char *str, STRLEN len) 286encode_str_utf8 (enc_t *enc, int utf8, char *str, STRLEN len)
237{ 287{
288 STRLEN ulen = len;
289 U8 *p, *pend = (U8 *)str + len;
290
291 for (p = (U8 *)str; p < pend; ++p)
292 ulen += *p >> 7; // count set high bits
293
294 encode_uint (enc, MAJOR_TEXT, ulen);
295
296 need (enc, ulen);
297 for (p = (U8 *)str; p < pend; ++p)
298 if (*p < 0x80)
299 *enc->cur++ = *p;
300 else
301 {
302 *enc->cur++ = 0xc0 + (*p >> 6);
303 *enc->cur++ = 0x80 + (*p & 63);
304 }
305}
306
307ecb_inline void
308encode_str (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len)
309{
310 if (ecb_expect_false (upgrade_utf8))
311 if (!utf8)
312 {
313 encode_str_utf8 (enc, utf8, str, len);
314 return;
315 }
316
317 encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len);
318 need (enc, len);
319 memcpy (enc->cur, str, len);
320 enc->cur += len;
321}
322
323ecb_inline void
324encode_strref (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len)
325{
238 if (ecb_expect_false (enc->cbor.flags & F_ALLOW_STRINGREF)) 326 if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS))
239 { 327 {
240 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1); 328 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1);
241 329
242 if (SvOK (*svp)) 330 if (SvOK (*svp))
243 { 331 {
244 // already registered, use stringref 332 // already registered, use stringref
245 encode_tag (enc, CBOR_TAG_STRINGREF); 333 encode_tag (enc, CBOR_TAG_STRINGREF);
246 encode_uint (enc, 0x00, SvUV (*svp)); 334 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
247 return; 335 return;
248 } 336 }
249 else if (len >= minimum_string_length (enc->stringref_idx)) 337 else if (len >= minimum_string_length (enc->stringref_idx))
250 { 338 {
251 // register only 339 // register only
252 sv_setuv (*svp, enc->stringref_idx); 340 sv_setuv (*svp, enc->stringref_idx);
253 ++enc->stringref_idx; 341 ++enc->stringref_idx;
254 } 342 }
255 } 343 }
256 344
257 encode_uint (enc, utf8 ? 0x60 : 0x40, len); 345 encode_str (enc, upgrade_utf8, utf8, str, len);
258 need (enc, len);
259 memcpy (enc->cur, str, len);
260 enc->cur += len;
261} 346}
262 347
263static void encode_sv (enc_t *enc, SV *sv); 348static void encode_sv (enc_t *enc, SV *sv);
264 349
265static void 350static void
270 if (enc->depth >= enc->cbor.max_depth) 355 if (enc->depth >= enc->cbor.max_depth)
271 croak (ERR_NESTING_EXCEEDED); 356 croak (ERR_NESTING_EXCEEDED);
272 357
273 ++enc->depth; 358 ++enc->depth;
274 359
275 encode_uint (enc, 0x80, len + 1); 360 encode_uint (enc, MAJOR_ARRAY, len + 1);
276 361
362 if (SvMAGICAL (av))
277 for (i = 0; i <= len; ++i) 363 for (i = 0; i <= len; ++i)
278 { 364 {
279 SV **svp = av_fetch (av, i, 0); 365 SV **svp = av_fetch (av, i, 0);
280 encode_sv (enc, svp ? *svp : &PL_sv_undef); 366 encode_sv (enc, svp ? *svp : &PL_sv_undef);
281 } 367 }
368 else
369 for (i = 0; i <= len; ++i)
370 {
371 SV *sv = AvARRAY (av)[i];
372 encode_sv (enc, sv ? sv : &PL_sv_undef);
373 }
282 374
283 --enc->depth; 375 --enc->depth;
284} 376}
285 377
286static void 378static void
295 387
296 int pairs = hv_iterinit (hv); 388 int pairs = hv_iterinit (hv);
297 int mg = SvMAGICAL (hv); 389 int mg = SvMAGICAL (hv);
298 390
299 if (mg) 391 if (mg)
300 encode_ch (enc, 0xa0 | 31); 392 encode_ch (enc, MAJOR_MAP | MINOR_INDEF);
301 else 393 else
302 encode_uint (enc, 0xa0, pairs); 394 encode_uint (enc, MAJOR_MAP, pairs);
303 395
304 while ((he = hv_iternext (hv))) 396 while ((he = hv_iternext (hv)))
305 { 397 {
306 if (HeKLEN (he) == HEf_SVKEY) 398 if (HeKLEN (he) == HEf_SVKEY)
307 encode_sv (enc, HeSVKEY (he)); 399 encode_sv (enc, HeSVKEY (he));
308 else 400 else
309 encode_str (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he)); 401 encode_strref (enc, enc->cbor.flags & (F_TEXT_KEYS | F_TEXT_STRINGS), HeKUTF8 (he), HeKEY (he), HeKLEN (he));
310 402
311 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he)); 403 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he));
312 } 404 }
313 405
314 if (mg) 406 if (mg)
315 encode_ch (enc, 0xe0 | 31); 407 encode_ch (enc, MAJOR_MISC | MINOR_INDEF);
316 408
317 --enc->depth; 409 --enc->depth;
318} 410}
319 411
320// encode objects, arrays and special \0=false and \1=true values. 412// encode objects, arrays and special \0=false and \1=true values.
321static void 413static void
322encode_rv (enc_t *enc, SV *sv) 414encode_rv (enc_t *enc, SV *sv)
323{ 415{
324 SvGETMAGIC (sv); 416 SvGETMAGIC (sv);
325
326 if (ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING)
327 && ecb_expect_false (SvREFCNT (sv) > 1))
328 {
329 if (!enc->shareable)
330 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
331
332 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
333
334 if (SvOK (*svp))
335 {
336 encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF);
337 encode_uint (enc, 0x00, SvUV (*svp));
338 return;
339 }
340 else
341 {
342 sv_setuv (*svp, enc->shareable_idx);
343 ++enc->shareable_idx;
344 encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE);
345 }
346 }
347 417
348 svtype svt = SvTYPE (sv); 418 svtype svt = SvTYPE (sv);
349 419
350 if (ecb_expect_false (SvOBJECT (sv))) 420 if (ecb_expect_false (SvOBJECT (sv)))
351 { 421 {
358 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 428 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
359 ? cbor_tagged_stash 429 ? cbor_tagged_stash
360 : gv_stashpv ("CBOR::XS::Tagged" , 1); 430 : gv_stashpv ("CBOR::XS::Tagged" , 1);
361 431
362 HV *stash = SvSTASH (sv); 432 HV *stash = SvSTASH (sv);
363 GV *method;
364 433
365 if (stash == boolean_stash) 434 if (stash == boolean_stash)
366 encode_ch (enc, SvIV (sv) ? 0xe0 | 21 : 0xe0 | 20); 435 {
436 encode_ch (enc, SvIV (sv) ? MAJOR_MISC | SIMPLE_TRUE : MAJOR_MISC | SIMPLE_FALSE);
437 return;
438 }
367 else if (stash == error_stash) 439 else if (stash == error_stash)
368 encode_ch (enc, 0xe0 | 23); 440 {
441 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
442 return;
443 }
369 else if (stash == tagged_stash) 444 else if (stash == tagged_stash)
370 { 445 {
371 if (svt != SVt_PVAV) 446 if (svt != SVt_PVAV)
372 croak ("encountered CBOR::XS::Tagged object that isn't an array"); 447 croak ("encountered CBOR::XS::Tagged object that isn't an array");
373 448
374 encode_uint (enc, 0xc0, SvUV (*av_fetch ((AV *)sv, 0, 1))); 449 encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1)));
375 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1)); 450 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1));
451
452 return;
453 }
454 }
455
456 if (ecb_expect_false (SvREFCNT (sv) > 1)
457 && ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING))
458 {
459 if (!enc->shareable)
460 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
461
462 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
463
464 if (SvOK (*svp))
376 } 465 {
466 encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF);
467 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
468 return;
469 }
470 else
471 {
472 sv_setuv (*svp, enc->shareable_idx);
473 ++enc->shareable_idx;
474 encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE);
475 }
476 }
477
478 if (ecb_expect_false (SvOBJECT (sv)))
479 {
480 HV *stash = SvSTASH (sv);
481 GV *method;
482
483 if (enc->cbor.flags & F_FORBID_OBJECTS)
484 croak ("encountered object '%s', but forbid_objects is enabled",
485 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
377 else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0))) 486 else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
378 { 487 {
379 dSP; 488 dSP;
380 489
381 ENTER; SAVETMPS; PUSHMARK (SP); 490 ENTER; SAVETMPS;
491 PUSHMARK (SP);
382 // we re-bless the reference to get overload and other niceties right 492 // we re-bless the reference to get overload and other niceties right
383 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); 493 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
384 494
385 PUTBACK; 495 PUTBACK;
386 // G_SCALAR ensures that return value is 1 496 // G_SCALAR ensures that return value is 1
399 } 509 }
400 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0) 510 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)
401 { 511 {
402 dSP; 512 dSP;
403 513
404 ENTER; SAVETMPS; PUSHMARK (SP); 514 ENTER; SAVETMPS;
515 SAVESTACK_POS ();
516 PUSHMARK (SP);
405 EXTEND (SP, 2); 517 EXTEND (SP, 2);
406 // we re-bless the reference to get overload and other niceties right 518 // we re-bless the reference to get overload and other niceties right
407 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); 519 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
408 PUSHs (sv_cbor); 520 PUSHs (sv_cbor);
409 521
414 // catch this surprisingly common error 526 // catch this surprisingly common error
415 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv) 527 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)
416 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash)); 528 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));
417 529
418 encode_tag (enc, CBOR_TAG_PERL_OBJECT); 530 encode_tag (enc, CBOR_TAG_PERL_OBJECT);
419 encode_uint (enc, 0x80, count + 1); 531 encode_uint (enc, MAJOR_ARRAY, count + 1);
420 encode_str (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash)); 532 encode_strref (enc, 0, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
421 533
422 while (count) 534 while (count)
423 encode_sv (enc, SP[1 - count--]); 535 encode_sv (enc, SP[1 - count--]);
424 536
425 PUTBACK; 537 PUTBACK;
446{ 558{
447 double nv = SvNVX (sv); 559 double nv = SvNVX (sv);
448 560
449 need (enc, 9); 561 need (enc, 9);
450 562
451 if (ecb_expect_false (nv == (U32)nv)) 563 if (ecb_expect_false (nv == (NV)(U32)nv))
452 encode_uint (enc, 0x00, (U32)nv); 564 encode_uint (enc, MAJOR_POS_INT, (U32)nv);
453 //TODO: maybe I32? 565 //TODO: maybe I32?
454 else if (ecb_expect_false (nv == (float)nv)) 566 else if (ecb_expect_false (nv == (float)nv))
455 { 567 {
456 uint32_t fp = ecb_float_to_binary32 (nv); 568 uint32_t fp = ecb_float_to_binary32 (nv);
457 569
458 *enc->cur++ = 0xe0 | 26; 570 *enc->cur++ = MAJOR_MISC | MISC_FLOAT32;
459 571
460 if (!ecb_big_endian ()) 572 if (!ecb_big_endian ())
461 fp = ecb_bswap32 (fp); 573 fp = ecb_bswap32 (fp);
462 574
463 memcpy (enc->cur, &fp, 4); 575 memcpy (enc->cur, &fp, 4);
465 } 577 }
466 else 578 else
467 { 579 {
468 uint64_t fp = ecb_double_to_binary64 (nv); 580 uint64_t fp = ecb_double_to_binary64 (nv);
469 581
470 *enc->cur++ = 0xe0 | 27; 582 *enc->cur++ = MAJOR_MISC | MISC_FLOAT64;
471 583
472 if (!ecb_big_endian ()) 584 if (!ecb_big_endian ())
473 fp = ecb_bswap64 (fp); 585 fp = ecb_bswap64 (fp);
474 586
475 memcpy (enc->cur, &fp, 8); 587 memcpy (enc->cur, &fp, 8);
484 596
485 if (SvPOKp (sv)) 597 if (SvPOKp (sv))
486 { 598 {
487 STRLEN len; 599 STRLEN len;
488 char *str = SvPV (sv, len); 600 char *str = SvPV (sv, len);
489 encode_str (enc, SvUTF8 (sv), str, len); 601 encode_strref (enc, enc->cbor.flags & F_TEXT_STRINGS, SvUTF8 (sv), str, len);
490 } 602 }
491 else if (SvNOKp (sv)) 603 else if (SvNOKp (sv))
492 encode_nv (enc, sv); 604 encode_nv (enc, sv);
493 else if (SvIOKp (sv)) 605 else if (SvIOKp (sv))
494 { 606 {
495 if (SvIsUV (sv)) 607 if (SvIsUV (sv))
496 encode_uint (enc, 0x00, SvUVX (sv)); 608 encode_uint (enc, MAJOR_POS_INT, SvUVX (sv));
497 else if (SvIVX (sv) >= 0) 609 else if (SvIVX (sv) >= 0)
498 encode_uint (enc, 0x00, SvIVX (sv)); 610 encode_uint (enc, MAJOR_POS_INT, SvIVX (sv));
499 else 611 else
500 encode_uint (enc, 0x20, -(SvIVX (sv) + 1)); 612 encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1));
501 } 613 }
502 else if (SvROK (sv)) 614 else if (SvROK (sv))
503 encode_rv (enc, SvRV (sv)); 615 encode_rv (enc, SvRV (sv));
504 else if (!SvOK (sv)) 616 else if (!SvOK (sv))
505 encode_ch (enc, 0xe0 | 22); 617 encode_ch (enc, MAJOR_MISC | SIMPLE_NULL);
506 else if (enc->cbor.flags & F_ALLOW_UNKNOWN) 618 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
507 encode_ch (enc, 0xe0 | 23); 619 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
508 else 620 else
509 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data", 621 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data",
510 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv)); 622 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
511} 623}
512 624
513static SV * 625static SV *
514encode_cbor (SV *scalar, CBOR *cbor) 626encode_cbor (SV *scalar, CBOR *cbor)
515{ 627{
516 enc_t enc = { }; 628 enc_t enc = { 0 };
517 629
518 enc.cbor = *cbor; 630 enc.cbor = *cbor;
519 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 631 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
520 enc.cur = SvPVX (enc.sv); 632 enc.cur = SvPVX (enc.sv);
521 enc.end = SvEND (enc.sv); 633 enc.end = SvEND (enc.sv);
522 634
523 SvPOK_only (enc.sv); 635 SvPOK_only (enc.sv);
524 636
525 if (cbor->flags & F_ALLOW_STRINGREF) 637 if (cbor->flags & F_PACK_STRINGS)
526 { 638 {
527 encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE); 639 encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE);
528 enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ()); 640 enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ());
529 enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ()); 641 enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ());
530 } 642 }
553 U32 depth; // recursion depth 665 U32 depth; // recursion depth
554 U32 maxdepth; // recursion depth limit 666 U32 maxdepth; // recursion depth limit
555 AV *shareable; 667 AV *shareable;
556 AV *stringref; 668 AV *stringref;
557 SV *decode_tagged; 669 SV *decode_tagged;
670 SV *err_sv; // optional sv for error, needs to be freed
558} dec_t; 671} dec_t;
559 672
560#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE 673// set dec->err to ERRSV
674ecb_cold static void
675err_errsv (dec_t *dec)
676{
677 if (!dec->err)
678 {
679 dec->err_sv = newSVsv (ERRSV);
561 680
562#define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data") 681 // chop off the trailing \n
682 SvCUR_set (dec->err_sv, SvCUR (dec->err_sv) - 1);
683 *SvEND (dec->err_sv) = 0;
563 684
685 dec->err = SvPVutf8_nolen (dec->err_sv);
686 }
687}
688
689// the following functions are used to reduce code size and help the compiler to optimise
690ecb_cold static void
691err_set (dec_t *dec, const char *reason)
692{
693 if (!dec->err)
694 dec->err = reason;
695}
696
697ecb_cold static void
698err_unexpected_end (dec_t *dec)
699{
700 err_set (dec, "unexpected end of CBOR data");
701}
702
703ecb_cold static void
704err_nesting_exceeded (dec_t *dec)
705{
706 err_set (dec, ERR_NESTING_EXCEEDED);
707}
708
709#define ERR_DO(do) SB do; goto fail; SE
710#define ERR(reason) ERR_DO (err_set (dec, reason))
711#define ERR_ERRSV ERR_DO (err_errsv (dec))
712
713#define WANT(len) if (ecb_expect_false ((UV)(dec->end - dec->cur) < (UV)len)) ERR_DO (err_unexpected_end (dec))
714
564#define DEC_INC_DEPTH if (++dec->depth > dec->cbor.max_depth) ERR (ERR_NESTING_EXCEEDED) 715#define DEC_INC_DEPTH if (ecb_expect_false (++dec->depth > dec->cbor.max_depth)) ERR (ERR_NESTING_EXCEEDED)
565#define DEC_DEC_DEPTH --dec->depth 716#define DEC_DEC_DEPTH --dec->depth
566 717
567static UV 718static UV
568decode_uint (dec_t *dec) 719decode_uint (dec_t *dec)
569{ 720{
570 switch (*dec->cur & 31) 721 U8 m = *dec->cur & MINOR_MASK;
571 { 722 ++dec->cur;
572 case 0: case 1: case 2: case 3: case 4: case 5: case 6: case 7:
573 case 8: case 9: case 10: case 11: case 12: case 13: case 14: case 15:
574 case 16: case 17: case 18: case 19: case 20: case 21: case 22: case 23:
575 return *dec->cur++ & 31;
576 723
577 case 24: 724 if (ecb_expect_true (m < LENGTH_EXT1))
725 return m;
726 else if (ecb_expect_true (m == LENGTH_EXT1))
727 {
578 WANT (2); 728 WANT (1);
579 dec->cur += 2; 729 dec->cur += 1;
580 return dec->cur[-1]; 730 return dec->cur[-1];
581 731 }
582 case 25: 732 else if (ecb_expect_true (m == LENGTH_EXT2))
733 {
583 WANT (3); 734 WANT (2);
584 dec->cur += 3; 735 dec->cur += 2;
585 return (((UV)dec->cur[-2]) << 8) 736 return (((UV)dec->cur[-2]) << 8)
586 | ((UV)dec->cur[-1]); 737 | ((UV)dec->cur[-1]);
587 738 }
588 case 26: 739 else if (ecb_expect_true (m == LENGTH_EXT4))
740 {
589 WANT (5); 741 WANT (4);
590 dec->cur += 5; 742 dec->cur += 4;
591 return (((UV)dec->cur[-4]) << 24) 743 return (((UV)dec->cur[-4]) << 24)
592 | (((UV)dec->cur[-3]) << 16) 744 | (((UV)dec->cur[-3]) << 16)
593 | (((UV)dec->cur[-2]) << 8) 745 | (((UV)dec->cur[-2]) << 8)
594 | ((UV)dec->cur[-1]); 746 | ((UV)dec->cur[-1]);
595 747 }
596 case 27: 748 else if (ecb_expect_true (m == LENGTH_EXT8))
749 {
597 WANT (9); 750 WANT (8);
598 dec->cur += 9; 751 dec->cur += 8;
752
753 return
754#if UVSIZE < 8
755 0
756#else
599 return (((UV)dec->cur[-8]) << 56) 757 (((UV)dec->cur[-8]) << 56)
600 | (((UV)dec->cur[-7]) << 48) 758 | (((UV)dec->cur[-7]) << 48)
601 | (((UV)dec->cur[-6]) << 40) 759 | (((UV)dec->cur[-6]) << 40)
602 | (((UV)dec->cur[-5]) << 32) 760 | (((UV)dec->cur[-5]) << 32)
761#endif
603 | (((UV)dec->cur[-4]) << 24) 762 | (((UV)dec->cur[-4]) << 24)
604 | (((UV)dec->cur[-3]) << 16) 763 | (((UV)dec->cur[-3]) << 16)
605 | (((UV)dec->cur[-2]) << 8) 764 | (((UV)dec->cur[-2]) << 8)
606 | ((UV)dec->cur[-1]); 765 | ((UV)dec->cur[-1]);
607 766 }
608 default: 767 else
609 ERR ("corrupted CBOR data (unsupported integer minor encoding)"); 768 ERR ("corrupted CBOR data (unsupported integer minor encoding)");
610 }
611 769
612fail: 770fail:
613 return 0; 771 return 0;
614} 772}
615 773
620{ 778{
621 AV *av = newAV (); 779 AV *av = newAV ();
622 780
623 DEC_INC_DEPTH; 781 DEC_INC_DEPTH;
624 782
625 if ((*dec->cur & 31) == 31) 783 if (*dec->cur == (MAJOR_ARRAY | MINOR_INDEF))
626 { 784 {
627 ++dec->cur; 785 ++dec->cur;
628 786
629 for (;;) 787 for (;;)
630 { 788 {
631 WANT (1); 789 WANT (1);
632 790
633 if (*dec->cur == (0xe0 | 31)) 791 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
634 { 792 {
635 ++dec->cur; 793 ++dec->cur;
636 break; 794 break;
637 } 795 }
638 796
639 av_push (av, decode_sv (dec)); 797 av_push (av, decode_sv (dec));
640 } 798 }
641 } 799 }
642 else 800 else
643 { 801 {
644 int i, len = decode_uint (dec); 802 UV i, len = decode_uint (dec);
645 803
804 WANT (len); // complexity check for av_fill - need at least one byte per value, do not allow supersize arrays
646 av_fill (av, len - 1); 805 av_fill (av, len - 1);
647 806
648 for (i = 0; i < len; ++i) 807 for (i = 0; i < len; ++i)
649 AvARRAY (av)[i] = decode_sv (dec); 808 AvARRAY (av)[i] = decode_sv (dec);
650 } 809 }
663{ 822{
664 // for speed reasons, we specialcase single-string 823 // for speed reasons, we specialcase single-string
665 // byte or utf-8 strings as keys, but only when !stringref 824 // byte or utf-8 strings as keys, but only when !stringref
666 825
667 if (ecb_expect_true (!dec->stringref)) 826 if (ecb_expect_true (!dec->stringref))
668 if (*dec->cur >= 0x40 && *dec->cur <= 0x40 + 27) 827 if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8))
669 { 828 {
670 I32 len = decode_uint (dec); 829 STRLEN len = decode_uint (dec);
671 char *key = (char *)dec->cur; 830 char *key = (char *)dec->cur;
672 831
832 WANT (len);
673 dec->cur += len; 833 dec->cur += len;
674 834
675 if (ecb_expect_false (dec->stringref))
676 av_push (dec->stringref, newSVpvn (key, len));
677
678 hv_store (hv, key, len, decode_sv (dec), 0); 835 hv_store (hv, key, len, decode_sv (dec), 0);
679 836
680 return; 837 return;
681 } 838 }
682 else if (*dec->cur >= 0x60 && *dec->cur <= 0x60 + 27) 839 else if (ecb_expect_true ((U8)(*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8))
683 { 840 {
684 I32 len = decode_uint (dec); 841 STRLEN len = decode_uint (dec);
685 char *key = (char *)dec->cur; 842 char *key = (char *)dec->cur;
686 843
844 WANT (len);
687 dec->cur += len; 845 dec->cur += len;
688 846
689 if (ecb_expect_false (dec->stringref)) 847 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
690 av_push (dec->stringref, newSVpvn_utf8 (key, len, 1)); 848 if (!is_utf8_string (key, len))
849 ERR ("corrupted CBOR data (invalid UTF-8 in map key)");
691 850
692 hv_store (hv, key, -len, decode_sv (dec), 0); 851 hv_store (hv, key, -len, decode_sv (dec), 0);
693 852
694 return; 853 return;
695 } 854 }
696 855
697 SV *k = decode_sv (dec); 856 SV *k = decode_sv (dec);
698 SV *v = decode_sv (dec); 857 SV *v = decode_sv (dec);
699 858
859 // we leak memory if uncaught exceptions are thrown by random magical
860 // methods, and this is hopefully the only place where it can happen,
861 // so if there is a chance of an exception, take the very slow path.
862 // since catching exceptions is "undocumented/internal/forbidden" by
863 // the new p5p powers, we need to call out to a perl function :/
864 if (ecb_expect_false (SvAMAGIC (k)))
865 {
866 dSP;
867
868 ENTER; SAVETMPS;
869 PUSHMARK (SP);
870 EXTEND (SP, 3);
871 PUSHs (sv_2mortal (newRV_inc ((SV *)hv)));
872 PUSHs (sv_2mortal (k));
873 PUSHs (sv_2mortal (v));
874
875 PUTBACK;
876 call_pv ("CBOR::XS::_hv_store", G_VOID | G_DISCARD | G_EVAL);
877 SPAGAIN;
878
879 FREETMPS; LEAVE;
880
881 if (SvTRUE (ERRSV))
882 ERR_ERRSV;
883
884 return;
885 }
886
700 hv_store_ent (hv, k, v, 0); 887 hv_store_ent (hv, k, v, 0);
701 SvREFCNT_dec (k); 888 SvREFCNT_dec (k);
889
890fail:
891 ;
702} 892}
703 893
704static SV * 894static SV *
705decode_hv (dec_t *dec) 895decode_hv (dec_t *dec)
706{ 896{
707 HV *hv = newHV (); 897 HV *hv = newHV ();
708 898
709 DEC_INC_DEPTH; 899 DEC_INC_DEPTH;
710 900
711 if ((*dec->cur & 31) == 31) 901 if (*dec->cur == (MAJOR_MAP | MINOR_INDEF))
712 { 902 {
713 ++dec->cur; 903 ++dec->cur;
714 904
715 for (;;) 905 for (;;)
716 { 906 {
717 WANT (1); 907 WANT (1);
718 908
719 if (*dec->cur == (0xe0 | 31)) 909 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
720 { 910 {
721 ++dec->cur; 911 ++dec->cur;
722 break; 912 break;
723 } 913 }
724 914
725 decode_he (dec, hv); 915 decode_he (dec, hv);
726 } 916 }
727 } 917 }
728 else 918 else
729 { 919 {
730 int pairs = decode_uint (dec); 920 UV pairs = decode_uint (dec);
921
922 WANT (pairs); // complexity check - need at least one byte per value, do not allow supersize hashes
731 923
732 while (pairs--) 924 while (pairs--)
733 decode_he (dec, hv); 925 decode_he (dec, hv);
734 } 926 }
735 927
745static SV * 937static SV *
746decode_str (dec_t *dec, int utf8) 938decode_str (dec_t *dec, int utf8)
747{ 939{
748 SV *sv = 0; 940 SV *sv = 0;
749 941
750 if ((*dec->cur & 31) == 31) 942 if ((*dec->cur & MINOR_MASK) == MINOR_INDEF)
751 { 943 {
944 // indefinite length strings
752 ++dec->cur; 945 ++dec->cur;
753 946
947 U8 major = *dec->cur & MAJOR_MISC;
948
754 sv = newSVpvn ("", 0); 949 sv = newSVpvn ("", 0);
755 950
756 // not very fast, and certainly not robust against illegal input
757 for (;;) 951 for (;;)
758 { 952 {
759 WANT (1); 953 WANT (1);
760 954
761 if (*dec->cur == (0xe0 | 31)) 955 if ((*dec->cur - major) > LENGTH_EXT8)
956 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
762 { 957 {
763 ++dec->cur; 958 ++dec->cur;
764 break; 959 break;
765 } 960 }
961 else
962 ERR ("corrupted CBOR data (invalid chunks in indefinite length string)");
766 963
767 sv_catsv (sv, decode_sv (dec)); 964 STRLEN len = decode_uint (dec);
965
966 WANT (len);
967 sv_catpvn (sv, dec->cur, len);
968 dec->cur += len;
768 } 969 }
769 } 970 }
770 else 971 else
771 { 972 {
772 STRLEN len = decode_uint (dec); 973 STRLEN len = decode_uint (dec);
779 && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1)) 980 && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1))
780 av_push (dec->stringref, SvREFCNT_inc_NN (sv)); 981 av_push (dec->stringref, SvREFCNT_inc_NN (sv));
781 } 982 }
782 983
783 if (utf8) 984 if (utf8)
985 {
986 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
987 if (!is_utf8_string (SvPVX (sv), SvCUR (sv)))
988 ERR ("corrupted CBOR data (invalid UTF-8 in text string)");
989
784 SvUTF8_on (sv); 990 SvUTF8_on (sv);
991 }
785 992
786 return sv; 993 return sv;
787 994
788fail: 995fail:
789 SvREFCNT_dec (sv); 996 SvREFCNT_dec (sv);
808 sv = newRV_noinc (decode_sv (dec)); 1015 sv = newRV_noinc (decode_sv (dec));
809 break; 1016 break;
810 1017
811 case CBOR_TAG_STRINGREF_NAMESPACE: 1018 case CBOR_TAG_STRINGREF_NAMESPACE:
812 { 1019 {
1020 // do nmot use SAVETMPS/FREETMPS, as these will
1021 // erase mortalised caches, e.g. "shareable"
813 ENTER; SAVETMPS; 1022 ENTER;
814 1023
815 SAVESPTR (dec->stringref); 1024 SAVESPTR (dec->stringref);
816 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ()); 1025 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ());
817 1026
818 sv = decode_sv (dec); 1027 sv = decode_sv (dec);
819 1028
820 FREETMPS; LEAVE; 1029 LEAVE;
821 } 1030 }
822 break; 1031 break;
823 1032
824 case CBOR_TAG_STRINGREF: 1033 case CBOR_TAG_STRINGREF:
825 { 1034 {
826 if ((*dec->cur >> 5) != 0) 1035 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
827 ERR ("corrupted CBOR data (stringref index not an unsigned integer)"); 1036 ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
828 1037
829 UV idx = decode_uint (dec); 1038 UV idx = decode_uint (dec);
830 1039
831 if (!dec->stringref || (int)idx > AvFILLp (dec->stringref)) 1040 if (!dec->stringref || (int)idx > AvFILLp (dec->stringref))
838 case CBOR_TAG_VALUE_SHAREABLE: 1047 case CBOR_TAG_VALUE_SHAREABLE:
839 { 1048 {
840 if (ecb_expect_false (!dec->shareable)) 1049 if (ecb_expect_false (!dec->shareable))
841 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ()); 1050 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ());
842 1051
1052 if (dec->cbor.flags & F_ALLOW_CYCLES)
1053 {
843 sv = newSV (0); 1054 sv = newSV (0);
844 av_push (dec->shareable, SvREFCNT_inc_NN (sv)); 1055 av_push (dec->shareable, SvREFCNT_inc_NN (sv));
845 1056
846 SV *osv = decode_sv (dec); 1057 SV *osv = decode_sv (dec);
847 sv_setsv (sv, osv); 1058 sv_setsv (sv, osv);
848 SvREFCNT_dec_NN (osv); 1059 SvREFCNT_dec_NN (osv);
1060 }
1061 else
1062 {
1063 av_push (dec->shareable, &PL_sv_undef);
1064 int idx = AvFILLp (dec->shareable);
1065 sv = decode_sv (dec);
1066 av_store (dec->shareable, idx, SvREFCNT_inc_NN (sv));
1067 }
849 } 1068 }
850 break; 1069 break;
851 1070
852 case CBOR_TAG_VALUE_SHAREDREF: 1071 case CBOR_TAG_VALUE_SHAREDREF:
853 { 1072 {
854 if ((*dec->cur >> 5) != 0) 1073 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
855 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)"); 1074 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
856 1075
857 UV idx = decode_uint (dec); 1076 UV idx = decode_uint (dec);
858 1077
859 if (!dec->shareable || (int)idx > AvFILLp (dec->shareable)) 1078 if (!dec->shareable || (int)idx > AvFILLp (dec->shareable))
860 ERR ("corrupted CBOR data (sharedref index out of bounds)"); 1079 ERR ("corrupted CBOR data (sharedref index out of bounds)");
861 1080
862 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]); 1081 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]);
1082
1083 if (sv == &PL_sv_undef)
1084 ERR ("cyclic CBOR data structure found, but allow_cycles is not enabled");
863 } 1085 }
864 break; 1086 break;
865 1087
866 case CBOR_TAG_PERL_OBJECT: 1088 case CBOR_TAG_PERL_OBJECT:
867 { 1089 {
1090 if (dec->cbor.flags & F_FORBID_OBJECTS)
1091 goto filter;
1092
868 sv = decode_sv (dec); 1093 sv = decode_sv (dec);
869 1094
870 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV) 1095 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
871 ERR ("corrupted CBOR data (non-array perl object)"); 1096 ERR ("corrupted CBOR data (non-array perl object)");
872 1097
882 if (!method) 1107 if (!method)
883 ERR ("cannot decode perl-object (package does not have a THAW method)"); 1108 ERR ("cannot decode perl-object (package does not have a THAW method)");
884 1109
885 dSP; 1110 dSP;
886 1111
887 ENTER; SAVETMPS; PUSHMARK (SP); 1112 ENTER; SAVETMPS;
1113 PUSHMARK (SP);
888 EXTEND (SP, len + 1); 1114 EXTEND (SP, len + 1);
889 // we re-bless the reference to get overload and other niceties right 1115 // we re-bless the reference to get overload and other niceties right
890 PUSHs (*av_fetch (av, 0, 1)); 1116 PUSHs (*av_fetch (av, 0, 1));
891 PUSHs (sv_cbor); 1117 PUSHs (sv_cbor);
892 1118
900 SPAGAIN; 1126 SPAGAIN;
901 1127
902 if (SvTRUE (ERRSV)) 1128 if (SvTRUE (ERRSV))
903 { 1129 {
904 FREETMPS; LEAVE; 1130 FREETMPS; LEAVE;
905 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV)))); 1131 ERR_ERRSV;
906 } 1132 }
907 1133
908 SvREFCNT_dec (sv); 1134 SvREFCNT_dec (sv);
909 sv = SvREFCNT_inc (POPs); 1135 sv = SvREFCNT_inc (POPs);
910 1136
913 FREETMPS; LEAVE; 1139 FREETMPS; LEAVE;
914 } 1140 }
915 break; 1141 break;
916 1142
917 default: 1143 default:
1144 filter:
918 { 1145 {
1146 SV *tag_sv = newSVuv (tag);
1147
919 sv = decode_sv (dec); 1148 sv = decode_sv (dec);
920 1149
921 dSP; 1150 dSP;
922 ENTER; SAVETMPS; PUSHMARK (SP); 1151 ENTER; SAVETMPS;
1152 SAVESTACK_POS ();
1153 PUSHMARK (SP);
923 EXTEND (SP, 2); 1154 EXTEND (SP, 2);
924 PUSHs (newSVuv (tag)); 1155 PUSHs (tag_sv);
925 PUSHs (sv); 1156 PUSHs (sv);
926 1157
927 PUTBACK; 1158 PUTBACK;
928 int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL); 1159 int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL);
929 SPAGAIN; 1160 SPAGAIN;
930 1161
931 if (SvTRUE (ERRSV)) 1162 if (SvTRUE (ERRSV))
932 { 1163 {
1164 SvREFCNT_dec (tag_sv);
933 FREETMPS; LEAVE; 1165 FREETMPS; LEAVE;
934 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV)))); 1166 ERR_ERRSV;
935 } 1167 }
936 1168
937 if (count) 1169 if (count)
938 { 1170 {
1171 SvREFCNT_dec (tag_sv);
939 SvREFCNT_dec (sv); 1172 SvREFCNT_dec (sv);
940 sv = SvREFCNT_inc (POPs); 1173 sv = SvREFCNT_inc (POPs);
941 } 1174 }
942 else 1175 else
943 { 1176 {
944 AV *av = newAV (); 1177 AV *av = newAV ();
945 av_push (av, newSVuv (tag)); 1178 av_push (av, tag_sv);
946 av_push (av, sv); 1179 av_push (av, sv);
947 1180
948 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 1181 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
949 ? cbor_tagged_stash 1182 ? cbor_tagged_stash
950 : gv_stashpv ("CBOR::XS::Tagged" , 1); 1183 : gv_stashpv ("CBOR::XS::Tagged" , 1);
968static SV * 1201static SV *
969decode_sv (dec_t *dec) 1202decode_sv (dec_t *dec)
970{ 1203{
971 WANT (1); 1204 WANT (1);
972 1205
973 switch (*dec->cur >> 5) 1206 switch (*dec->cur >> MAJOR_SHIFT)
974 { 1207 {
975 case 0: // unsigned int 1208 case MAJOR_POS_INT >> MAJOR_SHIFT: return newSVuv (decode_uint (dec));
976 return newSVuv (decode_uint (dec)); 1209 case MAJOR_NEG_INT >> MAJOR_SHIFT: return newSViv (-1 - (IV)decode_uint (dec));
977 case 1: // negative int 1210 case MAJOR_BYTES >> MAJOR_SHIFT: return decode_str (dec, 0);
978 return newSViv (-1 - (IV)decode_uint (dec)); 1211 case MAJOR_TEXT >> MAJOR_SHIFT: return decode_str (dec, 1);
979 case 2: // octet string 1212 case MAJOR_ARRAY >> MAJOR_SHIFT: return decode_av (dec);
980 return decode_str (dec, 0); 1213 case MAJOR_MAP >> MAJOR_SHIFT: return decode_hv (dec);
981 case 3: // utf-8 string 1214 case MAJOR_TAG >> MAJOR_SHIFT: return decode_tagged (dec);
982 return decode_str (dec, 1); 1215
983 case 4: // array 1216 case MAJOR_MISC >> MAJOR_SHIFT:
984 return decode_av (dec);
985 case 5: // map
986 return decode_hv (dec);
987 case 6: // tag
988 return decode_tagged (dec);
989 case 7: // misc
990 switch (*dec->cur++ & 31) 1217 switch (*dec->cur++ & MINOR_MASK)
991 { 1218 {
992 case 20: 1219 case SIMPLE_FALSE:
993#if CBOR_SLOW 1220#if CBOR_SLOW
994 types_false = get_bool ("Types::Serialiser::false"); 1221 types_false = get_bool ("Types::Serialiser::false");
995#endif 1222#endif
996 return newSVsv (types_false); 1223 return newSVsv (types_false);
997 case 21: 1224 case SIMPLE_TRUE:
998#if CBOR_SLOW 1225#if CBOR_SLOW
999 types_true = get_bool ("Types::Serialiser::true"); 1226 types_true = get_bool ("Types::Serialiser::true");
1000#endif 1227#endif
1001 return newSVsv (types_true); 1228 return newSVsv (types_true);
1002 case 22: 1229 case SIMPLE_NULL:
1003 return newSVsv (&PL_sv_undef); 1230 return newSVsv (&PL_sv_undef);
1004 case 23: 1231 case SIMPLE_UNDEF:
1005#if CBOR_SLOW 1232#if CBOR_SLOW
1006 types_error = get_bool ("Types::Serialiser::error"); 1233 types_error = get_bool ("Types::Serialiser::error");
1007#endif 1234#endif
1008 return newSVsv (types_error); 1235 return newSVsv (types_error);
1009 1236
1010 case 25: 1237 case MISC_FLOAT16:
1011 { 1238 {
1012 WANT (2); 1239 WANT (2);
1013 1240
1014 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1]; 1241 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1];
1015 dec->cur += 2; 1242 dec->cur += 2;
1016 1243
1017 return newSVnv (ecb_binary16_to_float (fp)); 1244 return newSVnv (ecb_binary16_to_float (fp));
1018 } 1245 }
1019 1246
1020 case 26: 1247 case MISC_FLOAT32:
1021 { 1248 {
1022 uint32_t fp; 1249 uint32_t fp;
1023 WANT (4); 1250 WANT (4);
1024 memcpy (&fp, dec->cur, 4); 1251 memcpy (&fp, dec->cur, 4);
1025 dec->cur += 4; 1252 dec->cur += 4;
1028 fp = ecb_bswap32 (fp); 1255 fp = ecb_bswap32 (fp);
1029 1256
1030 return newSVnv (ecb_binary32_to_float (fp)); 1257 return newSVnv (ecb_binary32_to_float (fp));
1031 } 1258 }
1032 1259
1033 case 27: 1260 case MISC_FLOAT64:
1034 { 1261 {
1035 uint64_t fp; 1262 uint64_t fp;
1036 WANT (8); 1263 WANT (8);
1037 memcpy (&fp, dec->cur, 8); 1264 memcpy (&fp, dec->cur, 8);
1038 dec->cur += 8; 1265 dec->cur += 8;
1041 fp = ecb_bswap64 (fp); 1268 fp = ecb_bswap64 (fp);
1042 1269
1043 return newSVnv (ecb_binary64_to_double (fp)); 1270 return newSVnv (ecb_binary64_to_double (fp));
1044 } 1271 }
1045 1272
1046 // 0..19 unassigned 1273 // 0..19 unassigned simple
1047 // 24 reserved + unassigned (reserved values are not encodable) 1274 // 24 reserved + unassigned simple (reserved values are not encodable)
1275 // 28-30 unassigned misc
1276 // 31 break code
1048 default: 1277 default:
1049 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)"); 1278 ERR ("corrupted CBOR data (reserved/unassigned/unexpected major 7 value)");
1050 } 1279 }
1051 1280
1052 break; 1281 break;
1053 } 1282 }
1054 1283
1057} 1286}
1058 1287
1059static SV * 1288static SV *
1060decode_cbor (SV *string, CBOR *cbor, char **offset_return) 1289decode_cbor (SV *string, CBOR *cbor, char **offset_return)
1061{ 1290{
1062 dec_t dec = { }; 1291 dec_t dec = { 0 };
1063 SV *sv; 1292 SV *sv;
1064 STRLEN len; 1293 STRLEN len;
1065 char *data = SvPVbyte (string, len); 1294 char *data = SvPVbyte (string, len);
1066 1295
1067 if (len > cbor->max_size && cbor->max_size) 1296 if (len > cbor->max_size && cbor->max_size)
1081 if (dec.cur != dec.end && !dec.err) 1310 if (dec.cur != dec.end && !dec.err)
1082 dec.err = "garbage after CBOR object"; 1311 dec.err = "garbage after CBOR object";
1083 1312
1084 if (dec.err) 1313 if (dec.err)
1085 { 1314 {
1315 if (dec.shareable)
1316 {
1317 // need to break cyclic links, which would all be in shareable
1318 int i;
1319 SV **svp;
1320
1321 for (i = av_len (dec.shareable) + 1; i--; )
1322 if ((svp = av_fetch (dec.shareable, i, 0)))
1323 sv_setsv (*svp, &PL_sv_undef);
1324 }
1325
1086 SvREFCNT_dec (sv); 1326 SvREFCNT_dec (sv);
1327
1328 if (dec.err_sv)
1329 sv_2mortal (dec.err_sv);
1330
1087 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur); 1331 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur);
1088 } 1332 }
1089 1333
1090 sv = sv_2mortal (sv); 1334 sv = sv_2mortal (sv);
1091 1335
1092 return sv; 1336 return sv;
1093} 1337}
1094 1338
1339/////////////////////////////////////////////////////////////////////////////
1340// incremental parser
1341
1342#define INCR_DONE(cbor) (AvFILLp (cbor->incr_count) < 0)
1343
1344// returns 0 for notyet, 1 for success or error
1345static int
1346incr_parse (CBOR *self, SV *cborstr)
1347{
1348 STRLEN cur;
1349 SvPV (cborstr, cur);
1350
1351 while (ecb_expect_true (self->incr_need <= cur))
1352 {
1353 // table of integer count bytes
1354 static I8 incr_len[MINOR_MASK + 1] = {
1355 0, 0, 0, 0, 0, 0, 0, 0,
1356 0, 0, 0, 0, 0, 0, 0, 0,
1357 0, 0, 0, 0, 0, 0, 0, 0,
1358 1, 2, 4, 8,-1,-1,-1,-2
1359 };
1360
1361 const U8 *p = SvPVX (cborstr) + self->incr_pos;
1362 U8 m = *p & MINOR_MASK;
1363 IV count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1364 I8 ilen = incr_len[m];
1365
1366 self->incr_need = self->incr_pos + 1;
1367
1368 if (ecb_expect_false (ilen < 0))
1369 {
1370 if (m != MINOR_INDEF)
1371 return 1; // error
1372
1373 if (*p == (MAJOR_MISC | MINOR_INDEF))
1374 {
1375 if (count >= 0)
1376 return 1; // error
1377
1378 count = 1;
1379 }
1380 else
1381 {
1382 av_push (self->incr_count, newSViv (-1)); //TODO: nest
1383 count = -1;
1384 }
1385 }
1386 else
1387 {
1388 self->incr_need += ilen;
1389 if (ecb_expect_false (self->incr_need > cur))
1390 return 0;
1391
1392 int major = *p >> MAJOR_SHIFT;
1393
1394 switch (major)
1395 {
1396 case MAJOR_TAG >> MAJOR_SHIFT:
1397 ++count; // tags merely prefix another value
1398 break;
1399
1400 case MAJOR_BYTES >> MAJOR_SHIFT:
1401 case MAJOR_TEXT >> MAJOR_SHIFT:
1402 case MAJOR_ARRAY >> MAJOR_SHIFT:
1403 case MAJOR_MAP >> MAJOR_SHIFT:
1404 {
1405 UV len;
1406
1407 if (ecb_expect_false (ilen))
1408 {
1409 len = 0;
1410
1411 do {
1412 len = (len << 8) | *++p;
1413 } while (--ilen);
1414 }
1415 else
1416 len = m;
1417
1418 switch (major)
1419 {
1420 case MAJOR_BYTES >> MAJOR_SHIFT:
1421 case MAJOR_TEXT >> MAJOR_SHIFT:
1422 self->incr_need += len;
1423 if (ecb_expect_false (self->incr_need > cur))
1424 return 0;
1425
1426 break;
1427
1428 case MAJOR_MAP >> MAJOR_SHIFT:
1429 len <<= 1;
1430 case MAJOR_ARRAY >> MAJOR_SHIFT:
1431 if (len)
1432 {
1433 av_push (self->incr_count, newSViv (len + 1)); //TODO: nest
1434 count = len + 1;
1435 }
1436 break;
1437 }
1438 }
1439 }
1440 }
1441
1442 self->incr_pos = self->incr_need;
1443
1444 if (count > 0)
1445 {
1446 while (!--count)
1447 {
1448 if (!AvFILLp (self->incr_count))
1449 return 1; // done
1450
1451 SvREFCNT_dec_NN (av_pop (self->incr_count));
1452 count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1453 }
1454
1455 SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]) = count;
1456 }
1457 }
1458
1459 return 0;
1460}
1461
1462
1095///////////////////////////////////////////////////////////////////////////// 1463/////////////////////////////////////////////////////////////////////////////
1096// XS interface functions 1464// XS interface functions
1097 1465
1098MODULE = CBOR::XS PACKAGE = CBOR::XS 1466MODULE = CBOR::XS PACKAGE = CBOR::XS
1099 1467
1111 1479
1112 default_filter = newSVpv ("CBOR::XS::default_filter", 0); 1480 default_filter = newSVpv ("CBOR::XS::default_filter", 0);
1113 1481
1114 sv_cbor = newSVpv ("CBOR", 0); 1482 sv_cbor = newSVpv ("CBOR", 0);
1115 SvREADONLY_on (sv_cbor); 1483 SvREADONLY_on (sv_cbor);
1484
1485 assert (("STRLEN must be an unsigned type", 0 <= (STRLEN)-1));
1116} 1486}
1117 1487
1118PROTOTYPES: DISABLE 1488PROTOTYPES: DISABLE
1119 1489
1120void CLONE (...) 1490void CLONE (...)
1139void shrink (CBOR *self, int enable = 1) 1509void shrink (CBOR *self, int enable = 1)
1140 ALIAS: 1510 ALIAS:
1141 shrink = F_SHRINK 1511 shrink = F_SHRINK
1142 allow_unknown = F_ALLOW_UNKNOWN 1512 allow_unknown = F_ALLOW_UNKNOWN
1143 allow_sharing = F_ALLOW_SHARING 1513 allow_sharing = F_ALLOW_SHARING
1144 allow_stringref = F_ALLOW_STRINGREF 1514 allow_cycles = F_ALLOW_CYCLES
1515 forbid_objects = F_FORBID_OBJECTS
1516 pack_strings = F_PACK_STRINGS
1517 text_keys = F_TEXT_KEYS
1518 text_strings = F_TEXT_STRINGS
1519 validate_utf8 = F_VALIDATE_UTF8
1145 PPCODE: 1520 PPCODE:
1146{ 1521{
1147 if (enable) 1522 if (enable)
1148 self->flags |= ix; 1523 self->flags |= ix;
1149 else 1524 else
1155void get_shrink (CBOR *self) 1530void get_shrink (CBOR *self)
1156 ALIAS: 1531 ALIAS:
1157 get_shrink = F_SHRINK 1532 get_shrink = F_SHRINK
1158 get_allow_unknown = F_ALLOW_UNKNOWN 1533 get_allow_unknown = F_ALLOW_UNKNOWN
1159 get_allow_sharing = F_ALLOW_SHARING 1534 get_allow_sharing = F_ALLOW_SHARING
1160 get_allow_stringref = F_ALLOW_STRINGREF 1535 get_allow_cycles = F_ALLOW_CYCLES
1536 get_forbid_objects = F_FORBID_OBJECTS
1537 get_pack_strings = F_PACK_STRINGS
1538 get_text_keys = F_TEXT_KEYS
1539 get_text_strings = F_TEXT_STRINGS
1540 get_validate_utf8 = F_VALIDATE_UTF8
1161 PPCODE: 1541 PPCODE:
1162 XPUSHs (boolSV (self->flags & ix)); 1542 XPUSHs (boolSV (self->flags & ix));
1163 1543
1164void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) 1544void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
1165 PPCODE: 1545 PPCODE:
1214 EXTEND (SP, 2); 1594 EXTEND (SP, 2);
1215 PUSHs (sv); 1595 PUSHs (sv);
1216 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); 1596 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
1217} 1597}
1218 1598
1599void incr_parse (CBOR *self, SV *cborstr)
1600 ALIAS:
1601 incr_parse_multiple = 1
1602 PPCODE:
1603{
1604 if (SvUTF8 (cborstr))
1605 sv_utf8_downgrade (cborstr, 0);
1606
1607 if (!self->incr_count)
1608 {
1609 self->incr_count = newAV ();
1610 self->incr_pos = 0;
1611 self->incr_need = 1;
1612
1613 av_push (self->incr_count, newSViv (1));
1614 }
1615
1616 do
1617 {
1618 if (!incr_parse (self, cborstr))
1619 {
1620 if (self->incr_need > self->max_size && self->max_size)
1621 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
1622 (unsigned long)self->incr_need, (unsigned long)self->max_size);
1623
1624 break;
1625 }
1626
1627 SV *sv;
1628 char *offset;
1629
1630 PUTBACK; sv = decode_cbor (cborstr, self, &offset); SPAGAIN;
1631 XPUSHs (sv);
1632
1633 sv_chop (cborstr, offset);
1634
1635 av_clear (self->incr_count);
1636 av_push (self->incr_count, newSViv (1));
1637
1638 self->incr_pos = 0;
1639 self->incr_need = self->incr_pos + 1;
1640 }
1641 while (ix);
1642}
1643
1644void incr_reset (CBOR *self)
1645 CODE:
1646{
1647 SvREFCNT_dec (self->incr_count);
1648 self->incr_count = 0;
1649}
1650
1219void DESTROY (CBOR *self) 1651void DESTROY (CBOR *self)
1220 PPCODE: 1652 PPCODE:
1221 cbor_free (self); 1653 cbor_free (self);
1222 1654
1223PROTOTYPES: ENABLE 1655PROTOTYPES: ENABLE
1224 1656
1225void encode_cbor (SV *scalar) 1657void encode_cbor (SV *scalar)
1658 ALIAS:
1659 encode_cbor = 0
1660 encode_cbor_sharing = F_ALLOW_SHARING
1226 PPCODE: 1661 PPCODE:
1227{ 1662{
1228 CBOR cbor; 1663 CBOR cbor;
1229 cbor_init (&cbor); 1664 cbor_init (&cbor);
1665 cbor.flags |= ix;
1230 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN; 1666 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;
1231 XPUSHs (scalar); 1667 XPUSHs (scalar);
1232} 1668}
1233 1669
1234void decode_cbor (SV *cborstr) 1670void decode_cbor (SV *cborstr)

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines