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 1.35 by root, Sat Nov 30 17:19:34 2013 UTC vs.
Revision 1.67 by root, Sun Nov 29 21:35:06 2020 UTC

6#include <string.h> 6#include <string.h>
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#include <inttypes.h>
11 12
13#define ECB_NO_THREADS 1
12#include "ecb.h" 14#include "ecb.h"
13 15
14// compatibility with perl <5.18 16// compatibility with perl <5.18
15#ifndef HvNAMELEN_get 17#ifndef HvNAMELEN_get
16# define HvNAMELEN_get(hv) strlen (HvNAME (hv)) 18# define HvNAMELEN_get(hv) strlen (HvNAME (hv))
19# define HvNAMELEN(hv) HvNAMELEN_get (hv) 21# define HvNAMELEN(hv) HvNAMELEN_get (hv)
20#endif 22#endif
21#ifndef HvNAMEUTF8 23#ifndef HvNAMEUTF8
22# define HvNAMEUTF8(hv) 0 24# define HvNAMEUTF8(hv) 0
23#endif 25#endif
26#ifndef SvREFCNT_inc_NN
27# define SvREFCNT_inc_NN(sv) SvREFCNT_inc (sv)
28#endif
24#ifndef SvREFCNT_dec_NN 29#ifndef SvREFCNT_dec_NN
25# define SvREFCNT_dec_NN(sv) SvREFCNT_dec (sv) 30# define SvREFCNT_dec_NN(sv) SvREFCNT_dec (sv)
26#endif 31#endif
27 32
28// known major and minor types 33// known major and minor types
94 CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8 99 CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8
95 100
96 CBOR_TAG_MAGIC = 55799, // self-describe cbor 101 CBOR_TAG_MAGIC = 55799, // self-describe cbor
97}; 102};
98 103
104// known forced types, also hardcoded in CBOR.pm
105enum
106{
107 AS_CBOR = 0,
108 AS_BYTES = 1,
109 AS_TEXT = 2,
110 AS_INT = 3,
111 AS_FLOAT16 = 4,
112 AS_FLOAT32 = 5,
113 AS_FLOAT64 = 6,
114 // possibly future enhancements: (generic) float, (generic) string
115};
116
99#define F_SHRINK 0x00000001UL 117#define F_SHRINK 0x00000001UL
100#define F_ALLOW_UNKNOWN 0x00000002UL 118#define F_ALLOW_UNKNOWN 0x00000002UL
101#define F_ALLOW_SHARING 0x00000004UL 119#define F_ALLOW_SHARING 0x00000004UL
120#define F_ALLOW_CYCLES 0x00000008UL
121#define F_FORBID_OBJECTS 0x00000010UL
102#define F_PACK_STRINGS 0x00000008UL 122#define F_PACK_STRINGS 0x00000020UL
123#define F_TEXT_KEYS 0x00000040UL
124#define F_TEXT_STRINGS 0x00000080UL
125#define F_VALIDATE_UTF8 0x00000100UL
103 126
104#define INIT_SIZE 32 // initial scalar size to be allocated 127#define INIT_SIZE 32 // initial scalar size to be allocated
105 128
106#define SB do { 129#define SB do {
107#define SE } while (0) 130#define SE } while (0)
126typedef struct { 149typedef struct {
127 U32 flags; 150 U32 flags;
128 U32 max_depth; 151 U32 max_depth;
129 STRLEN max_size; 152 STRLEN max_size;
130 SV *filter; 153 SV *filter;
154
155 // for the incremental parser
156 STRLEN incr_pos; // the current offset into the text
157 STRLEN incr_need; // minimum bytes needed to decode
158 AV *incr_count; // for every nesting level, the number of outstanding values, or -1 for indef.
131} CBOR; 159} CBOR;
132 160
133ecb_inline void 161ecb_inline void
134cbor_init (CBOR *cbor) 162cbor_init (CBOR *cbor)
135{ 163{
139 167
140ecb_inline void 168ecb_inline void
141cbor_free (CBOR *cbor) 169cbor_free (CBOR *cbor)
142{ 170{
143 SvREFCNT_dec (cbor->filter); 171 SvREFCNT_dec (cbor->filter);
172 SvREFCNT_dec (cbor->incr_count);
144} 173}
145 174
146///////////////////////////////////////////////////////////////////////////// 175/////////////////////////////////////////////////////////////////////////////
147// utility functions 176// utility functions
148 177
174 203
175// minimum length of a string to be registered for stringref 204// minimum length of a string to be registered for stringref
176ecb_inline int 205ecb_inline int
177minimum_string_length (UV idx) 206minimum_string_length (UV idx)
178{ 207{
179 return idx > 23 208 return idx <= 23 ? 3
180 ? idx > 0xffU 209 : idx <= 0xffU ? 4
181 ? idx > 0xffffU 210 : idx <= 0xffffU ? 5
182 ? idx > 0xffffffffU 211 : idx <= 0xffffffffU ? 7
183 ? 11 212 : 11;
184 : 7
185 : 5
186 : 4
187 : 3;
188} 213}
189 214
190///////////////////////////////////////////////////////////////////////////// 215/////////////////////////////////////////////////////////////////////////////
191// encoder 216// encoder
192 217
205} enc_t; 230} enc_t;
206 231
207ecb_inline void 232ecb_inline void
208need (enc_t *enc, STRLEN len) 233need (enc_t *enc, STRLEN len)
209{ 234{
210 if (ecb_expect_false (enc->cur + len >= enc->end)) 235 if (ecb_expect_false ((uintptr_t)(enc->end - enc->cur) < len))
211 { 236 {
212 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv); 237 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
213 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1); 238 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
214 enc->cur = SvPVX (enc->sv) + cur; 239 enc->cur = SvPVX (enc->sv) + cur;
215 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; 240 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
221{ 246{
222 need (enc, 1); 247 need (enc, 1);
223 *enc->cur++ = ch; 248 *enc->cur++ = ch;
224} 249}
225 250
251// used for tags, intregers, element counts and so on
226static void 252static void
227encode_uint (enc_t *enc, int major, UV len) 253encode_uint (enc_t *enc, int major, UV len)
228{ 254{
229 need (enc, 9); 255 need (enc, 9);
230 256
231 if (ecb_expect_true (len < LENGTH_EXT1)) 257 if (ecb_expect_true (len < LENGTH_EXT1))
232 *enc->cur++ = major | len; 258 *enc->cur++ = major | len;
233 else if (ecb_expect_true (len <= 0xff)) 259 else if (ecb_expect_true (len <= 0xffU))
234 { 260 {
235 *enc->cur++ = major | LENGTH_EXT1; 261 *enc->cur++ = major | LENGTH_EXT1;
236 *enc->cur++ = len; 262 *enc->cur++ = len;
237 } 263 }
238 else if (len <= 0xffff) 264 else if (len <= 0xffffU)
239 { 265 {
240 *enc->cur++ = major | LENGTH_EXT2; 266 *enc->cur++ = major | LENGTH_EXT2;
241 *enc->cur++ = len >> 8; 267 *enc->cur++ = len >> 8;
242 *enc->cur++ = len; 268 *enc->cur++ = len;
243 } 269 }
244 else if (len <= 0xffffffff) 270 else if (len <= 0xffffffffU)
245 { 271 {
246 *enc->cur++ = major | LENGTH_EXT4; 272 *enc->cur++ = major | LENGTH_EXT4;
247 *enc->cur++ = len >> 24; 273 *enc->cur++ = len >> 24;
248 *enc->cur++ = len >> 16; 274 *enc->cur++ = len >> 16;
249 *enc->cur++ = len >> 8; 275 *enc->cur++ = len >> 8;
261 *enc->cur++ = len >> 8; 287 *enc->cur++ = len >> 8;
262 *enc->cur++ = len; 288 *enc->cur++ = len;
263 } 289 }
264} 290}
265 291
292// encodes a perl value into a CBOR integer
293ecb_inline void
294encode_int (enc_t *enc, SV *sv)
295{
296 if (SvIsUV (sv))
297 encode_uint (enc, MAJOR_POS_INT, SvUVX (sv));
298 else if (SvIVX (sv) >= 0)
299 encode_uint (enc, MAJOR_POS_INT, SvIVX (sv));
300 else
301 encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1));
302}
303
266ecb_inline void 304ecb_inline void
267encode_tag (enc_t *enc, UV tag) 305encode_tag (enc_t *enc, UV tag)
268{ 306{
269 encode_uint (enc, MAJOR_TAG, tag); 307 encode_uint (enc, MAJOR_TAG, tag);
270} 308}
271 309
310// exceptional (hopefully) slow path for byte strings that need to be utf8-encoded
311ecb_noinline static void
312encode_str_utf8 (enc_t *enc, int utf8, char *str, STRLEN len)
313{
314 STRLEN ulen = len;
315 U8 *p, *pend = (U8 *)str + len;
316
317 for (p = (U8 *)str; p < pend; ++p)
318 ulen += *p >> 7; // count set high bits
319
320 encode_uint (enc, MAJOR_TEXT, ulen);
321
322 need (enc, ulen);
323 for (p = (U8 *)str; p < pend; ++p)
324 if (*p < 0x80)
325 *enc->cur++ = *p;
326 else
327 {
328 *enc->cur++ = 0xc0 + (*p >> 6);
329 *enc->cur++ = 0x80 + (*p & 63);
330 }
331}
332
272ecb_inline void 333ecb_inline void
273encode_str (enc_t *enc, int utf8, char *str, STRLEN len) 334encode_str (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len)
274{ 335{
336 if (ecb_expect_false (upgrade_utf8))
337 if (!utf8)
338 {
339 encode_str_utf8 (enc, utf8, str, len);
340 return;
341 }
342
275 encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len); 343 encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len);
276 need (enc, len); 344 need (enc, len);
277 memcpy (enc->cur, str, len); 345 memcpy (enc->cur, str, len);
278 enc->cur += len; 346 enc->cur += len;
279} 347}
280 348
281static void 349ecb_inline void
282encode_strref (enc_t *enc, int utf8, char *str, STRLEN len) 350encode_strref (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len)
283{ 351{
284 if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS)) 352 if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS))
285 { 353 {
286 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1); 354 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1);
287 355
298 sv_setuv (*svp, enc->stringref_idx); 366 sv_setuv (*svp, enc->stringref_idx);
299 ++enc->stringref_idx; 367 ++enc->stringref_idx;
300 } 368 }
301 } 369 }
302 370
303 encode_str (enc, utf8, str, len); 371 encode_str (enc, upgrade_utf8, utf8, str, len);
372}
373
374ecb_inline void
375encode_float16 (enc_t *enc, NV nv)
376{
377 need (enc, 1+2);
378
379 *enc->cur++ = MAJOR_MISC | MISC_FLOAT16;
380
381 uint16_t fp = ecb_float_to_binary16 (nv);
382
383 if (!ecb_big_endian ())
384 fp = ecb_bswap16 (fp);
385
386 memcpy (enc->cur, &fp, 2);
387 enc->cur += 2;
388}
389
390ecb_inline void
391encode_float32 (enc_t *enc, NV nv)
392{
393 need (enc, 1+4);
394
395 *enc->cur++ = MAJOR_MISC | MISC_FLOAT32;
396
397 uint32_t fp = ecb_float_to_binary32 (nv);
398
399 if (!ecb_big_endian ())
400 fp = ecb_bswap32 (fp);
401
402 memcpy (enc->cur, &fp, 4);
403 enc->cur += 4;
404}
405
406ecb_inline void
407encode_float64 (enc_t *enc, NV nv)
408{
409 need (enc, 1+8);
410
411 *enc->cur++ = MAJOR_MISC | MISC_FLOAT64;
412
413 uint64_t fp = ecb_double_to_binary64 (nv);
414
415 if (!ecb_big_endian ())
416 fp = ecb_bswap64 (fp);
417
418 memcpy (enc->cur, &fp, 8);
419 enc->cur += 8;
420}
421
422ecb_inline void
423encode_forced (enc_t *enc, UV type, SV *sv)
424{
425 switch (type)
426 {
427 case AS_CBOR:
428 {
429 STRLEN len;
430 char *str = SvPVbyte (sv, len);
431
432 need (enc, len);
433 memcpy (enc->cur, str, len);
434 enc->cur += len;
435 }
436 break;
437
438 case AS_BYTES:
439 {
440 STRLEN len;
441 char *str = SvPVbyte (sv, len);
442 encode_strref (enc, 0, 0, str, len);
443 }
444 break;
445
446 case AS_TEXT:
447 {
448 STRLEN len;
449 char *str = SvPVutf8 (sv, len);
450 encode_strref (enc, 1, 1, str, len);
451 }
452 break;
453
454 case AS_INT: encode_int (enc, sv); break;
455 case AS_FLOAT16: encode_float16 (enc, SvNV (sv)); break;
456 case AS_FLOAT32: encode_float32 (enc, SvNV (sv)); break;
457 case AS_FLOAT64: encode_float64 (enc, SvNV (sv)); break;
458
459 default:
460 croak ("encountered malformed CBOR::XS::Tagged object");
461 }
304} 462}
305 463
306static void encode_sv (enc_t *enc, SV *sv); 464static void encode_sv (enc_t *enc, SV *sv);
307 465
308static void 466static void
315 473
316 ++enc->depth; 474 ++enc->depth;
317 475
318 encode_uint (enc, MAJOR_ARRAY, len + 1); 476 encode_uint (enc, MAJOR_ARRAY, len + 1);
319 477
478 if (ecb_expect_false (SvMAGICAL (av)))
320 for (i = 0; i <= len; ++i) 479 for (i = 0; i <= len; ++i)
321 { 480 {
322 SV **svp = av_fetch (av, i, 0); 481 SV **svp = av_fetch (av, i, 0);
323 encode_sv (enc, svp ? *svp : &PL_sv_undef); 482 encode_sv (enc, svp ? *svp : &PL_sv_undef);
324 } 483 }
484 else
485 for (i = 0; i <= len; ++i)
486 {
487 SV *sv = AvARRAY (av)[i];
488 encode_sv (enc, sv ? sv : &PL_sv_undef);
489 }
325 490
326 --enc->depth; 491 --enc->depth;
327} 492}
328 493
329static void 494static void
337 ++enc->depth; 502 ++enc->depth;
338 503
339 int pairs = hv_iterinit (hv); 504 int pairs = hv_iterinit (hv);
340 int mg = SvMAGICAL (hv); 505 int mg = SvMAGICAL (hv);
341 506
342 if (mg) 507 if (ecb_expect_false (mg))
343 encode_ch (enc, MAJOR_MAP | MINOR_INDEF); 508 encode_ch (enc, MAJOR_MAP | MINOR_INDEF);
344 else 509 else
345 encode_uint (enc, MAJOR_MAP, pairs); 510 encode_uint (enc, MAJOR_MAP, pairs);
346 511
347 while ((he = hv_iternext (hv))) 512 while ((he = hv_iternext (hv)))
348 { 513 {
349 if (HeKLEN (he) == HEf_SVKEY) 514 if (HeKLEN (he) == HEf_SVKEY)
350 encode_sv (enc, HeSVKEY (he)); 515 encode_sv (enc, HeSVKEY (he));
351 else 516 else
352 encode_strref (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he)); 517 encode_strref (enc, enc->cbor.flags & (F_TEXT_KEYS | F_TEXT_STRINGS), HeKUTF8 (he), HeKEY (he), HeKLEN (he));
353 518
354 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he)); 519 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he));
355 } 520 }
356 521
357 if (mg) 522 if (ecb_expect_false (mg))
358 encode_ch (enc, MAJOR_MISC | MINOR_INDEF); 523 encode_ch (enc, MAJOR_MISC | MINOR_INDEF);
359 524
360 --enc->depth; 525 --enc->depth;
361} 526}
362 527
395 else if (stash == tagged_stash) 560 else if (stash == tagged_stash)
396 { 561 {
397 if (svt != SVt_PVAV) 562 if (svt != SVt_PVAV)
398 croak ("encountered CBOR::XS::Tagged object that isn't an array"); 563 croak ("encountered CBOR::XS::Tagged object that isn't an array");
399 564
565 switch (av_len ((AV *)sv))
566 {
567 case 2-1:
568 // actually a tagged value
400 encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1))); 569 encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1)));
401 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1)); 570 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1));
571 break;
572
573 case 3-1:
574 // a forced type [value, type, undef]
575 encode_forced (enc, SvUV (*av_fetch ((AV *)sv, 1, 1)), *av_fetch ((AV *)sv, 0, 1));
576 break;
577
578 default:
579 croak ("encountered malformed CBOR::XS::Tagged object");
580 }
402 581
403 return; 582 return;
404 } 583 }
405 } 584 }
406 585
407 if (ecb_expect_false (SvREFCNT (sv) > 1) 586 if (ecb_expect_false (SvREFCNT (sv) > 1)
408 && ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING)) 587 && ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING))
409 { 588 {
410 if (!enc->shareable) 589 if (ecb_expect_false (!enc->shareable))
411 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ()); 590 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
412 591
413 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1); 592 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
414 593
415 if (SvOK (*svp)) 594 if (SvOK (*svp))
429 if (ecb_expect_false (SvOBJECT (sv))) 608 if (ecb_expect_false (SvOBJECT (sv)))
430 { 609 {
431 HV *stash = SvSTASH (sv); 610 HV *stash = SvSTASH (sv);
432 GV *method; 611 GV *method;
433 612
613 if (enc->cbor.flags & F_FORBID_OBJECTS)
614 croak ("encountered object '%s', but forbid_objects is enabled",
615 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
434 if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0))) 616 else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
435 { 617 {
436 dSP; 618 dSP;
437 619
438 ENTER; SAVETMPS; PUSHMARK (SP); 620 ENTER; SAVETMPS;
621 PUSHMARK (SP);
439 // we re-bless the reference to get overload and other niceties right 622 // we re-bless the reference to get overload and other niceties right
440 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); 623 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
441 624
442 PUTBACK; 625 PUTBACK;
443 // G_SCALAR ensures that return value is 1 626 // G_SCALAR ensures that return value is 1
456 } 639 }
457 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0) 640 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)
458 { 641 {
459 dSP; 642 dSP;
460 643
461 ENTER; SAVETMPS; PUSHMARK (SP); 644 ENTER; SAVETMPS;
645 PUSHMARK (SP);
462 EXTEND (SP, 2); 646 EXTEND (SP, 2);
463 // we re-bless the reference to get overload and other niceties right 647 // we re-bless the reference to get overload and other niceties right
464 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); 648 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
465 PUSHs (sv_cbor); 649 PUSHs (sv_cbor);
466 650
472 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv) 656 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)
473 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash)); 657 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));
474 658
475 encode_tag (enc, CBOR_TAG_PERL_OBJECT); 659 encode_tag (enc, CBOR_TAG_PERL_OBJECT);
476 encode_uint (enc, MAJOR_ARRAY, count + 1); 660 encode_uint (enc, MAJOR_ARRAY, count + 1);
477 encode_strref (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash)); 661 encode_strref (enc, 0, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
478 662
479 while (count) 663 {
664 int i;
665
666 for (i = 0; i < count; ++i)
480 encode_sv (enc, SP[1 - count--]); 667 encode_sv (enc, SP[i + 1 - count]);
668
669 SP -= count;
670 }
481 671
482 PUTBACK; 672 PUTBACK;
483 673
484 FREETMPS; LEAVE; 674 FREETMPS; LEAVE;
485 } 675 }
507 697
508 if (ecb_expect_false (nv == (NV)(U32)nv)) 698 if (ecb_expect_false (nv == (NV)(U32)nv))
509 encode_uint (enc, MAJOR_POS_INT, (U32)nv); 699 encode_uint (enc, MAJOR_POS_INT, (U32)nv);
510 //TODO: maybe I32? 700 //TODO: maybe I32?
511 else if (ecb_expect_false (nv == (float)nv)) 701 else if (ecb_expect_false (nv == (float)nv))
512 { 702 encode_float32 (enc, nv);
513 uint32_t fp = ecb_float_to_binary32 (nv);
514
515 *enc->cur++ = MAJOR_MISC | MISC_FLOAT32;
516
517 if (!ecb_big_endian ())
518 fp = ecb_bswap32 (fp);
519
520 memcpy (enc->cur, &fp, 4);
521 enc->cur += 4;
522 }
523 else 703 else
524 { 704 encode_float64 (enc, nv);
525 uint64_t fp = ecb_double_to_binary64 (nv);
526
527 *enc->cur++ = MAJOR_MISC | MISC_FLOAT64;
528
529 if (!ecb_big_endian ())
530 fp = ecb_bswap64 (fp);
531
532 memcpy (enc->cur, &fp, 8);
533 enc->cur += 8;
534 }
535} 705}
536 706
537static void 707static void
538encode_sv (enc_t *enc, SV *sv) 708encode_sv (enc_t *enc, SV *sv)
539{ 709{
541 711
542 if (SvPOKp (sv)) 712 if (SvPOKp (sv))
543 { 713 {
544 STRLEN len; 714 STRLEN len;
545 char *str = SvPV (sv, len); 715 char *str = SvPV (sv, len);
546 encode_strref (enc, SvUTF8 (sv), str, len); 716 encode_strref (enc, enc->cbor.flags & F_TEXT_STRINGS, SvUTF8 (sv), str, len);
547 } 717 }
548 else if (SvNOKp (sv)) 718 else if (SvNOKp (sv))
549 encode_nv (enc, sv); 719 encode_nv (enc, sv);
550 else if (SvIOKp (sv)) 720 else if (SvIOKp (sv))
551 { 721 encode_int (enc, sv);
552 if (SvIsUV (sv))
553 encode_uint (enc, MAJOR_POS_INT, SvUVX (sv));
554 else if (SvIVX (sv) >= 0)
555 encode_uint (enc, MAJOR_POS_INT, SvIVX (sv));
556 else
557 encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1));
558 }
559 else if (SvROK (sv)) 722 else if (SvROK (sv))
560 encode_rv (enc, SvRV (sv)); 723 encode_rv (enc, SvRV (sv));
561 else if (!SvOK (sv)) 724 else if (!SvOK (sv))
562 encode_ch (enc, MAJOR_MISC | SIMPLE_NULL); 725 encode_ch (enc, MAJOR_MISC | SIMPLE_NULL);
563 else if (enc->cbor.flags & F_ALLOW_UNKNOWN) 726 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
568} 731}
569 732
570static SV * 733static SV *
571encode_cbor (SV *scalar, CBOR *cbor) 734encode_cbor (SV *scalar, CBOR *cbor)
572{ 735{
573 enc_t enc = { }; 736 enc_t enc = { 0 };
574 737
575 enc.cbor = *cbor; 738 enc.cbor = *cbor;
576 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 739 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
577 enc.cur = SvPVX (enc.sv); 740 enc.cur = SvPVX (enc.sv);
578 enc.end = SvEND (enc.sv); 741 enc.end = SvEND (enc.sv);
579 742
580 SvPOK_only (enc.sv); 743 SvPOK_only (enc.sv);
581 744
582 if (cbor->flags & F_PACK_STRINGS) 745 if (cbor->flags & F_PACK_STRINGS)
583 { 746 {
610 U32 depth; // recursion depth 773 U32 depth; // recursion depth
611 U32 maxdepth; // recursion depth limit 774 U32 maxdepth; // recursion depth limit
612 AV *shareable; 775 AV *shareable;
613 AV *stringref; 776 AV *stringref;
614 SV *decode_tagged; 777 SV *decode_tagged;
778 SV *err_sv; // optional sv for error, needs to be freed
615} dec_t; 779} dec_t;
616 780
617#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE 781// set dec->err to ERRSV
782ecb_cold static void
783err_errsv (dec_t *dec)
784{
785 if (!dec->err)
786 {
787 dec->err_sv = newSVsv (ERRSV);
618 788
619#define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data") 789 // chop off the trailing \n
790 SvCUR_set (dec->err_sv, SvCUR (dec->err_sv) - 1);
791 *SvEND (dec->err_sv) = 0;
620 792
793 dec->err = SvPVutf8_nolen (dec->err_sv);
794 }
795}
796
797// the following functions are used to reduce code size and help the compiler to optimise
798ecb_cold static void
799err_set (dec_t *dec, const char *reason)
800{
801 if (!dec->err)
802 dec->err = reason;
803}
804
805ecb_cold static void
806err_unexpected_end (dec_t *dec)
807{
808 err_set (dec, "unexpected end of CBOR data");
809}
810
811#define ERR_DO(do) SB do; goto fail; SE
812#define ERR(reason) ERR_DO (err_set (dec, reason))
813#define ERR_ERRSV ERR_DO (err_errsv (dec))
814
815#define WANT(len) if (ecb_expect_false ((uintptr_t)(dec->end - dec->cur) < (STRLEN)len)) ERR_DO (err_unexpected_end (dec))
816
621#define DEC_INC_DEPTH if (++dec->depth > dec->cbor.max_depth) ERR (ERR_NESTING_EXCEEDED) 817#define DEC_INC_DEPTH if (ecb_expect_false (++dec->depth > dec->cbor.max_depth)) ERR (ERR_NESTING_EXCEEDED)
622#define DEC_DEC_DEPTH --dec->depth 818#define DEC_DEC_DEPTH --dec->depth
623 819
624static UV 820static UV
625decode_uint (dec_t *dec) 821decode_uint (dec_t *dec)
626{ 822{
627 U8 m = *dec->cur & MINOR_MASK; 823 U8 m = *dec->cur & MINOR_MASK;
628 ++dec->cur; 824 ++dec->cur;
629 825
630 if (ecb_expect_true (m < LENGTH_EXT1)) 826 if (ecb_expect_true (m < LENGTH_EXT1))
631 return m; 827 return m;
632 828 else if (ecb_expect_true (m == LENGTH_EXT1))
633 switch (m)
634 { 829 {
635 case LENGTH_EXT1:
636 WANT (1); 830 WANT (1);
637 dec->cur += 1; 831 dec->cur += 1;
638 return dec->cur[-1]; 832 return dec->cur[-1];
639 833 }
640 case LENGTH_EXT2: 834 else if (ecb_expect_true (m == LENGTH_EXT2))
835 {
641 WANT (2); 836 WANT (2);
642 dec->cur += 2; 837 dec->cur += 2;
643 return (((UV)dec->cur[-2]) << 8) 838 return (((UV)dec->cur[-2]) << 8)
644 | ((UV)dec->cur[-1]); 839 | ((UV)dec->cur[-1]);
645 840 }
646 case LENGTH_EXT4: 841 else if (ecb_expect_true (m == LENGTH_EXT4))
842 {
647 WANT (4); 843 WANT (4);
648 dec->cur += 4; 844 dec->cur += 4;
649 return (((UV)dec->cur[-4]) << 24) 845 return (((UV)dec->cur[-4]) << 24)
650 | (((UV)dec->cur[-3]) << 16) 846 | (((UV)dec->cur[-3]) << 16)
651 | (((UV)dec->cur[-2]) << 8) 847 | (((UV)dec->cur[-2]) << 8)
652 | ((UV)dec->cur[-1]); 848 | ((UV)dec->cur[-1]);
653 849 }
654 case LENGTH_EXT8: 850 else if (ecb_expect_true (m == LENGTH_EXT8))
851 {
655 WANT (8); 852 WANT (8);
656 dec->cur += 8; 853 dec->cur += 8;
657 854
658 return 855 return
659#if UVSIZE < 8 856#if UVSIZE < 8
660 0 857 0
661#else 858#else
662 (((UV)dec->cur[-8]) << 56) 859 (((UV)dec->cur[-8]) << 56)
663 | (((UV)dec->cur[-7]) << 48) 860 | (((UV)dec->cur[-7]) << 48)
664 | (((UV)dec->cur[-6]) << 40) 861 | (((UV)dec->cur[-6]) << 40)
665 | (((UV)dec->cur[-5]) << 32) 862 | (((UV)dec->cur[-5]) << 32)
666#endif 863#endif
667 | (((UV)dec->cur[-4]) << 24) 864 | (((UV)dec->cur[-4]) << 24)
668 | (((UV)dec->cur[-3]) << 16) 865 | (((UV)dec->cur[-3]) << 16)
669 | (((UV)dec->cur[-2]) << 8) 866 | (((UV)dec->cur[-2]) << 8)
670 | ((UV)dec->cur[-1]); 867 | ((UV)dec->cur[-1]);
671 868 }
672 default: 869 else
673 ERR ("corrupted CBOR data (unsupported integer minor encoding)"); 870 ERR ("corrupted CBOR data (unsupported integer minor encoding)");
674 }
675 871
676fail: 872fail:
677 return 0; 873 return 0;
678} 874}
679 875
692 888
693 for (;;) 889 for (;;)
694 { 890 {
695 WANT (1); 891 WANT (1);
696 892
697 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF)) 893 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF) || dec->err)
698 { 894 {
699 ++dec->cur; 895 ++dec->cur;
700 break; 896 break;
701 } 897 }
702 898
703 av_push (av, decode_sv (dec)); 899 av_push (av, decode_sv (dec));
704 } 900 }
705 } 901 }
706 else 902 else
707 { 903 {
708 int i, len = decode_uint (dec); 904 UV i, len = decode_uint (dec);
709 905
906 WANT (len); // complexity check for av_fill - need at least one byte per value, do not allow supersize arrays
710 av_fill (av, len - 1); 907 av_fill (av, len - 1);
711 908
712 for (i = 0; i < len; ++i) 909 for (i = 0; i < len; ++i)
713 AvARRAY (av)[i] = decode_sv (dec); 910 AvARRAY (av)[i] = decode_sv (dec);
714 } 911 }
715 912
716 DEC_DEC_DEPTH; 913 DEC_DEC_DEPTH;
717 return newRV_noinc ((SV *)av); 914 return newRV_noinc ((SV *)av);
718 915
719fail: 916fail:
720 SvREFCNT_dec (av); 917 SvREFCNT_dec_NN (av);
721 DEC_DEC_DEPTH; 918 DEC_DEC_DEPTH;
722 return &PL_sv_undef; 919 return &PL_sv_undef;
723} 920}
724 921
725static void 922static void
727{ 924{
728 // for speed reasons, we specialcase single-string 925 // for speed reasons, we specialcase single-string
729 // byte or utf-8 strings as keys, but only when !stringref 926 // byte or utf-8 strings as keys, but only when !stringref
730 927
731 if (ecb_expect_true (!dec->stringref)) 928 if (ecb_expect_true (!dec->stringref))
732 if ((*dec->cur - MAJOR_BYTES) <= 27) 929 if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8))
733 { 930 {
734 I32 len = decode_uint (dec); 931 STRLEN len = decode_uint (dec);
735 char *key = (char *)dec->cur; 932 char *key = (char *)dec->cur;
736 933
934 WANT (len);
737 dec->cur += len; 935 dec->cur += len;
738 936
739 if (ecb_expect_false (dec->stringref))
740 av_push (dec->stringref, newSVpvn (key, len));
741
742 hv_store (hv, key, len, decode_sv (dec), 0); 937 hv_store (hv, key, len, decode_sv (dec), 0);
743 938
744 return; 939 return;
745 } 940 }
746 else if ((*dec->cur - MAJOR_TEXT) <= 27) 941 else if (ecb_expect_true ((U8)(*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8))
747 { 942 {
748 I32 len = decode_uint (dec); 943 STRLEN len = decode_uint (dec);
749 char *key = (char *)dec->cur; 944 char *key = (char *)dec->cur;
750 945
946 WANT (len);
751 dec->cur += len; 947 dec->cur += len;
752 948
753 if (ecb_expect_false (dec->stringref)) 949 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
754 av_push (dec->stringref, newSVpvn_utf8 (key, len, 1)); 950 if (!is_utf8_string (key, len))
951 ERR ("corrupted CBOR data (invalid UTF-8 in map key)");
755 952
756 hv_store (hv, key, -len, decode_sv (dec), 0); 953 hv_store (hv, key, -len, decode_sv (dec), 0);
757 954
758 return; 955 return;
759 } 956 }
760 957
761 SV *k = decode_sv (dec); 958 SV *k = decode_sv (dec);
762 SV *v = decode_sv (dec); 959 SV *v = decode_sv (dec);
763 960
961 // we leak memory if uncaught exceptions are thrown by random magical
962 // methods, and this is hopefully the only place where it can happen,
963 // so if there is a chance of an exception, take the very slow path.
964 // since catching exceptions is "undocumented/internal/forbidden" by
965 // the new p5p powers, we need to call out to a perl function :/
966 if (ecb_expect_false (SvAMAGIC (k)))
967 {
968 dSP;
969
970 ENTER; SAVETMPS;
971 PUSHMARK (SP);
972 EXTEND (SP, 3);
973 PUSHs (sv_2mortal (newRV_inc ((SV *)hv)));
974 PUSHs (sv_2mortal (k));
975 PUSHs (sv_2mortal (v));
976
977 PUTBACK;
978 call_pv ("CBOR::XS::_hv_store", G_VOID | G_DISCARD | G_EVAL);
979 SPAGAIN;
980
981 FREETMPS; LEAVE;
982
983 if (SvTRUE (ERRSV))
984 ERR_ERRSV;
985
986 return;
987 }
988
764 hv_store_ent (hv, k, v, 0); 989 hv_store_ent (hv, k, v, 0);
765 SvREFCNT_dec (k); 990 SvREFCNT_dec_NN (k);
991
992fail:
993 ;
766} 994}
767 995
768static SV * 996static SV *
769decode_hv (dec_t *dec) 997decode_hv (dec_t *dec)
770{ 998{
778 1006
779 for (;;) 1007 for (;;)
780 { 1008 {
781 WANT (1); 1009 WANT (1);
782 1010
783 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF)) 1011 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF) || dec->err)
784 { 1012 {
785 ++dec->cur; 1013 ++dec->cur;
786 break; 1014 break;
787 } 1015 }
788 1016
789 decode_he (dec, hv); 1017 decode_he (dec, hv);
790 } 1018 }
791 } 1019 }
792 else 1020 else
793 { 1021 {
794 int pairs = decode_uint (dec); 1022 UV pairs = decode_uint (dec);
1023
1024 WANT (pairs); // complexity check - need at least one byte per value, do not allow supersize hashes
795 1025
796 while (pairs--) 1026 while (pairs--)
797 decode_he (dec, hv); 1027 decode_he (dec, hv);
798 } 1028 }
799 1029
800 DEC_DEC_DEPTH; 1030 DEC_DEC_DEPTH;
801 return newRV_noinc ((SV *)hv); 1031 return newRV_noinc ((SV *)hv);
802 1032
803fail: 1033fail:
804 SvREFCNT_dec (hv); 1034 SvREFCNT_dec_NN (hv);
805 DEC_DEC_DEPTH; 1035 DEC_DEC_DEPTH;
806 return &PL_sv_undef; 1036 return &PL_sv_undef;
807} 1037}
808 1038
809static SV * 1039static SV *
810decode_str (dec_t *dec, int utf8) 1040decode_str (dec_t *dec, int utf8)
811{ 1041{
812 SV *sv = 0; 1042 SV *sv = 0;
813 1043
814 if ((*dec->cur & MINOR_MASK) == MINOR_INDEF) 1044 if (ecb_expect_false ((*dec->cur & MINOR_MASK) == MINOR_INDEF))
815 { 1045 {
816 // indefinite length strings 1046 // indefinite length strings
817 ++dec->cur; 1047 ++dec->cur;
818 1048
819 U8 major = *dec->cur & MAJOR_MISC; 1049 U8 major = *dec->cur & MAJOR_MISC;
852 && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1)) 1082 && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1))
853 av_push (dec->stringref, SvREFCNT_inc_NN (sv)); 1083 av_push (dec->stringref, SvREFCNT_inc_NN (sv));
854 } 1084 }
855 1085
856 if (utf8) 1086 if (utf8)
1087 {
1088 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
1089 if (!is_utf8_string (SvPVX (sv), SvCUR (sv)))
1090 ERR ("corrupted CBOR data (invalid UTF-8 in text string)");
1091
857 SvUTF8_on (sv); 1092 SvUTF8_on (sv);
1093 }
858 1094
859 return sv; 1095 return sv;
860 1096
861fail: 1097fail:
862 SvREFCNT_dec (sv); 1098 SvREFCNT_dec (sv);
881 sv = newRV_noinc (decode_sv (dec)); 1117 sv = newRV_noinc (decode_sv (dec));
882 break; 1118 break;
883 1119
884 case CBOR_TAG_STRINGREF_NAMESPACE: 1120 case CBOR_TAG_STRINGREF_NAMESPACE:
885 { 1121 {
1122 // do not use SAVETMPS/FREETMPS, as these will
1123 // erase mortalised caches, e.g. "shareable"
886 ENTER; SAVETMPS; 1124 ENTER;
887 1125
888 SAVESPTR (dec->stringref); 1126 SAVESPTR (dec->stringref);
889 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ()); 1127 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ());
890 1128
891 sv = decode_sv (dec); 1129 sv = decode_sv (dec);
892 1130
893 FREETMPS; LEAVE; 1131 LEAVE;
894 } 1132 }
895 break; 1133 break;
896 1134
897 case CBOR_TAG_STRINGREF: 1135 case CBOR_TAG_STRINGREF:
898 { 1136 {
899 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT)) 1137 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
900 ERR ("corrupted CBOR data (stringref index not an unsigned integer)"); 1138 ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
901 1139
902 UV idx = decode_uint (dec); 1140 UV idx = decode_uint (dec);
903 1141
904 if (!dec->stringref || (int)idx > AvFILLp (dec->stringref)) 1142 if (!dec->stringref || idx >= (UV)(1 + AvFILLp (dec->stringref)))
905 ERR ("corrupted CBOR data (stringref index out of bounds or outside namespace)"); 1143 ERR ("corrupted CBOR data (stringref index out of bounds or outside namespace)");
906 1144
907 sv = newSVsv (AvARRAY (dec->stringref)[idx]); 1145 sv = newSVsv (AvARRAY (dec->stringref)[idx]);
908 } 1146 }
909 break; 1147 break;
911 case CBOR_TAG_VALUE_SHAREABLE: 1149 case CBOR_TAG_VALUE_SHAREABLE:
912 { 1150 {
913 if (ecb_expect_false (!dec->shareable)) 1151 if (ecb_expect_false (!dec->shareable))
914 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ()); 1152 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ());
915 1153
1154 if (dec->cbor.flags & F_ALLOW_CYCLES)
1155 {
916 sv = newSV (0); 1156 sv = newSV (0);
917 av_push (dec->shareable, SvREFCNT_inc_NN (sv)); 1157 av_push (dec->shareable, SvREFCNT_inc_NN (sv));
918 1158
919 SV *osv = decode_sv (dec); 1159 SV *osv = decode_sv (dec);
920 sv_setsv (sv, osv); 1160 sv_setsv (sv, osv);
921 SvREFCNT_dec_NN (osv); 1161 SvREFCNT_dec_NN (osv);
1162 }
1163 else
1164 {
1165 av_push (dec->shareable, &PL_sv_undef);
1166 int idx = AvFILLp (dec->shareable);
1167 sv = decode_sv (dec);
1168 av_store (dec->shareable, idx, SvREFCNT_inc_NN (sv));
1169 }
922 } 1170 }
923 break; 1171 break;
924 1172
925 case CBOR_TAG_VALUE_SHAREDREF: 1173 case CBOR_TAG_VALUE_SHAREDREF:
926 { 1174 {
927 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT)) 1175 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
928 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)"); 1176 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
929 1177
930 UV idx = decode_uint (dec); 1178 UV idx = decode_uint (dec);
931 1179
932 if (!dec->shareable || (int)idx > AvFILLp (dec->shareable)) 1180 if (!dec->shareable || idx >= (UV)(1 + AvFILLp (dec->shareable)))
933 ERR ("corrupted CBOR data (sharedref index out of bounds)"); 1181 ERR ("corrupted CBOR data (sharedref index out of bounds)");
934 1182
935 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]); 1183 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]);
1184
1185 if (sv == &PL_sv_undef)
1186 ERR ("cyclic CBOR data structure found, but allow_cycles is not enabled");
936 } 1187 }
937 break; 1188 break;
938 1189
939 case CBOR_TAG_PERL_OBJECT: 1190 case CBOR_TAG_PERL_OBJECT:
940 { 1191 {
1192 if (dec->cbor.flags & F_FORBID_OBJECTS)
1193 goto filter;
1194
941 sv = decode_sv (dec); 1195 sv = decode_sv (dec);
942 1196
943 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV) 1197 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
944 ERR ("corrupted CBOR data (non-array perl object)"); 1198 ERR ("corrupted CBOR data (non-array perl object)");
945 1199
955 if (!method) 1209 if (!method)
956 ERR ("cannot decode perl-object (package does not have a THAW method)"); 1210 ERR ("cannot decode perl-object (package does not have a THAW method)");
957 1211
958 dSP; 1212 dSP;
959 1213
960 ENTER; SAVETMPS; PUSHMARK (SP); 1214 ENTER; SAVETMPS;
1215 PUSHMARK (SP);
961 EXTEND (SP, len + 1); 1216 EXTEND (SP, len + 1);
962 // we re-bless the reference to get overload and other niceties right 1217 // we re-bless the reference to get overload and other niceties right
963 PUSHs (*av_fetch (av, 0, 1)); 1218 PUSHs (*av_fetch (av, 0, 1));
964 PUSHs (sv_cbor); 1219 PUSHs (sv_cbor);
965 1220
973 SPAGAIN; 1228 SPAGAIN;
974 1229
975 if (SvTRUE (ERRSV)) 1230 if (SvTRUE (ERRSV))
976 { 1231 {
977 FREETMPS; LEAVE; 1232 FREETMPS; LEAVE;
978 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV)))); 1233 ERR_ERRSV;
979 } 1234 }
980 1235
981 SvREFCNT_dec (sv); 1236 SvREFCNT_dec_NN (sv);
982 sv = SvREFCNT_inc (POPs); 1237 sv = SvREFCNT_inc (POPs);
983 1238
984 PUTBACK; 1239 PUTBACK;
985 1240
986 FREETMPS; LEAVE; 1241 FREETMPS; LEAVE;
987 } 1242 }
988 break; 1243 break;
989 1244
990 default: 1245 default:
1246 filter:
991 { 1247 {
1248 SV *tag_sv = newSVuv (tag);
1249
992 sv = decode_sv (dec); 1250 sv = decode_sv (dec);
993 1251
994 dSP; 1252 dSP;
995 ENTER; SAVETMPS; PUSHMARK (SP); 1253 ENTER; SAVETMPS;
1254 PUSHMARK (SP);
996 EXTEND (SP, 2); 1255 EXTEND (SP, 2);
997 PUSHs (newSVuv (tag)); 1256 PUSHs (tag_sv);
998 PUSHs (sv); 1257 PUSHs (sv);
999 1258
1000 PUTBACK; 1259 PUTBACK;
1001 int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL); 1260 int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL);
1002 SPAGAIN; 1261 SPAGAIN;
1003 1262
1004 if (SvTRUE (ERRSV)) 1263 if (SvTRUE (ERRSV))
1005 { 1264 {
1265 SvREFCNT_dec_NN (tag_sv);
1006 FREETMPS; LEAVE; 1266 FREETMPS; LEAVE;
1007 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV)))); 1267 ERR_ERRSV;
1008 } 1268 }
1009 1269
1010 if (count) 1270 if (count)
1011 { 1271 {
1272 SvREFCNT_dec_NN (tag_sv);
1012 SvREFCNT_dec (sv); 1273 SvREFCNT_dec_NN (sv);
1013 sv = SvREFCNT_inc (POPs); 1274 sv = SvREFCNT_inc_NN (TOPs);
1275 SP -= count;
1014 } 1276 }
1015 else 1277 else
1016 { 1278 {
1017 AV *av = newAV (); 1279 AV *av = newAV ();
1018 av_push (av, newSVuv (tag)); 1280 av_push (av, tag_sv);
1019 av_push (av, sv); 1281 av_push (av, sv);
1020 1282
1021 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 1283 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
1022 ? cbor_tagged_stash 1284 ? cbor_tagged_stash
1023 : gv_stashpv ("CBOR::XS::Tagged" , 1); 1285 : gv_stashpv ("CBOR::XS::Tagged" , 1);
1109 1371
1110 return newSVnv (ecb_binary64_to_double (fp)); 1372 return newSVnv (ecb_binary64_to_double (fp));
1111 } 1373 }
1112 1374
1113 // 0..19 unassigned simple 1375 // 0..19 unassigned simple
1114 // 24 reserved + unassigned (reserved values are not encodable) 1376 // 24 reserved + unassigned simple (reserved values are not encodable)
1377 // 28-30 unassigned misc
1378 // 31 break code
1115 default: 1379 default:
1116 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)"); 1380 ERR ("corrupted CBOR data (reserved/unassigned/unexpected major 7 value)");
1117 } 1381 }
1118 1382
1119 break; 1383 break;
1120 } 1384 }
1121 1385
1124} 1388}
1125 1389
1126static SV * 1390static SV *
1127decode_cbor (SV *string, CBOR *cbor, char **offset_return) 1391decode_cbor (SV *string, CBOR *cbor, char **offset_return)
1128{ 1392{
1129 dec_t dec = { }; 1393 dec_t dec = { 0 };
1130 SV *sv; 1394 SV *sv;
1131 STRLEN len; 1395 STRLEN len;
1132 char *data = SvPVbyte (string, len); 1396 char *data = SvPVbyte (string, len);
1133 1397
1134 if (len > cbor->max_size && cbor->max_size) 1398 if (len > cbor->max_size && cbor->max_size)
1148 if (dec.cur != dec.end && !dec.err) 1412 if (dec.cur != dec.end && !dec.err)
1149 dec.err = "garbage after CBOR object"; 1413 dec.err = "garbage after CBOR object";
1150 1414
1151 if (dec.err) 1415 if (dec.err)
1152 { 1416 {
1417 if (dec.shareable)
1418 {
1419 // need to break cyclic links, which would all be in shareable
1420 int i;
1421 SV **svp;
1422
1423 for (i = av_len (dec.shareable) + 1; i--; )
1424 if ((svp = av_fetch (dec.shareable, i, 0)))
1425 sv_setsv (*svp, &PL_sv_undef);
1426 }
1427
1153 SvREFCNT_dec (sv); 1428 SvREFCNT_dec_NN (sv);
1429
1430 if (dec.err_sv)
1431 sv_2mortal (dec.err_sv);
1432
1154 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur); 1433 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur);
1155 } 1434 }
1156 1435
1157 sv = sv_2mortal (sv); 1436 sv = sv_2mortal (sv);
1158 1437
1159 return sv; 1438 return sv;
1160} 1439}
1161 1440
1441/////////////////////////////////////////////////////////////////////////////
1442// incremental parser
1443
1444#define INCR_DONE(cbor) (AvFILLp (cbor->incr_count) < 0)
1445
1446// returns 0 for notyet, 1 for success or error
1447static int
1448incr_parse (CBOR *self, SV *cborstr)
1449{
1450 STRLEN cur;
1451 SvPV (cborstr, cur);
1452
1453 while (ecb_expect_true (self->incr_need <= cur))
1454 {
1455 // table of integer count bytes
1456 static I8 incr_len[MINOR_MASK + 1] = {
1457 0, 0, 0, 0, 0, 0, 0, 0,
1458 0, 0, 0, 0, 0, 0, 0, 0,
1459 0, 0, 0, 0, 0, 0, 0, 0,
1460 1, 2, 4, 8,-1,-1,-1,-2
1461 };
1462
1463 const U8 *p = SvPVX (cborstr) + self->incr_pos;
1464 U8 m = *p & MINOR_MASK;
1465 IV count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1466 I8 ilen = incr_len[m];
1467
1468 self->incr_need = self->incr_pos + 1;
1469
1470 if (ecb_expect_false (ilen < 0))
1471 {
1472 if (m != MINOR_INDEF)
1473 return 1; // error
1474
1475 if (*p == (MAJOR_MISC | MINOR_INDEF))
1476 {
1477 if (count >= 0)
1478 return 1; // error
1479
1480 count = 1;
1481 }
1482 else
1483 {
1484 av_push (self->incr_count, newSViv (-1)); //TODO: nest
1485 count = -1;
1486 }
1487 }
1488 else
1489 {
1490 self->incr_need += ilen;
1491 if (ecb_expect_false (self->incr_need > cur))
1492 return 0;
1493
1494 int major = *p >> MAJOR_SHIFT;
1495
1496 switch (major)
1497 {
1498 case MAJOR_TAG >> MAJOR_SHIFT:
1499 ++count; // tags merely prefix another value
1500 break;
1501
1502 case MAJOR_BYTES >> MAJOR_SHIFT:
1503 case MAJOR_TEXT >> MAJOR_SHIFT:
1504 case MAJOR_ARRAY >> MAJOR_SHIFT:
1505 case MAJOR_MAP >> MAJOR_SHIFT:
1506 {
1507 UV len;
1508
1509 if (ecb_expect_false (ilen))
1510 {
1511 len = 0;
1512
1513 do {
1514 len = (len << 8) | *++p;
1515 } while (--ilen);
1516 }
1517 else
1518 len = m;
1519
1520 switch (major)
1521 {
1522 case MAJOR_BYTES >> MAJOR_SHIFT:
1523 case MAJOR_TEXT >> MAJOR_SHIFT:
1524 self->incr_need += len;
1525 if (ecb_expect_false (self->incr_need > cur))
1526 return 0;
1527
1528 break;
1529
1530 case MAJOR_MAP >> MAJOR_SHIFT:
1531 len <<= 1;
1532 case MAJOR_ARRAY >> MAJOR_SHIFT:
1533 if (len)
1534 {
1535 av_push (self->incr_count, newSViv (len + 1)); //TODO: nest
1536 count = len + 1;
1537 }
1538 break;
1539 }
1540 }
1541 }
1542 }
1543
1544 self->incr_pos = self->incr_need;
1545
1546 if (count > 0)
1547 {
1548 while (!--count)
1549 {
1550 if (!AvFILLp (self->incr_count))
1551 return 1; // done
1552
1553 SvREFCNT_dec_NN (av_pop (self->incr_count));
1554 count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1555 }
1556
1557 SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]) = count;
1558 }
1559 }
1560
1561 return 0;
1562}
1563
1564
1162///////////////////////////////////////////////////////////////////////////// 1565/////////////////////////////////////////////////////////////////////////////
1163// XS interface functions 1566// XS interface functions
1164 1567
1165MODULE = CBOR::XS PACKAGE = CBOR::XS 1568MODULE = CBOR::XS PACKAGE = CBOR::XS
1166 1569
1178 1581
1179 default_filter = newSVpv ("CBOR::XS::default_filter", 0); 1582 default_filter = newSVpv ("CBOR::XS::default_filter", 0);
1180 1583
1181 sv_cbor = newSVpv ("CBOR", 0); 1584 sv_cbor = newSVpv ("CBOR", 0);
1182 SvREADONLY_on (sv_cbor); 1585 SvREADONLY_on (sv_cbor);
1586
1587 assert (("STRLEN must be an unsigned type", 0 <= (STRLEN)-1));
1183} 1588}
1184 1589
1185PROTOTYPES: DISABLE 1590PROTOTYPES: DISABLE
1186 1591
1187void CLONE (...) 1592void CLONE (...)
1206void shrink (CBOR *self, int enable = 1) 1611void shrink (CBOR *self, int enable = 1)
1207 ALIAS: 1612 ALIAS:
1208 shrink = F_SHRINK 1613 shrink = F_SHRINK
1209 allow_unknown = F_ALLOW_UNKNOWN 1614 allow_unknown = F_ALLOW_UNKNOWN
1210 allow_sharing = F_ALLOW_SHARING 1615 allow_sharing = F_ALLOW_SHARING
1616 allow_cycles = F_ALLOW_CYCLES
1617 forbid_objects = F_FORBID_OBJECTS
1211 pack_strings = F_PACK_STRINGS 1618 pack_strings = F_PACK_STRINGS
1619 text_keys = F_TEXT_KEYS
1620 text_strings = F_TEXT_STRINGS
1621 validate_utf8 = F_VALIDATE_UTF8
1212 PPCODE: 1622 PPCODE:
1213{ 1623{
1214 if (enable) 1624 if (enable)
1215 self->flags |= ix; 1625 self->flags |= ix;
1216 else 1626 else
1222void get_shrink (CBOR *self) 1632void get_shrink (CBOR *self)
1223 ALIAS: 1633 ALIAS:
1224 get_shrink = F_SHRINK 1634 get_shrink = F_SHRINK
1225 get_allow_unknown = F_ALLOW_UNKNOWN 1635 get_allow_unknown = F_ALLOW_UNKNOWN
1226 get_allow_sharing = F_ALLOW_SHARING 1636 get_allow_sharing = F_ALLOW_SHARING
1637 get_allow_cycles = F_ALLOW_CYCLES
1638 get_forbid_objects = F_FORBID_OBJECTS
1227 get_pack_strings = F_PACK_STRINGS 1639 get_pack_strings = F_PACK_STRINGS
1640 get_text_keys = F_TEXT_KEYS
1641 get_text_strings = F_TEXT_STRINGS
1642 get_validate_utf8 = F_VALIDATE_UTF8
1228 PPCODE: 1643 PPCODE:
1229 XPUSHs (boolSV (self->flags & ix)); 1644 XPUSHs (boolSV (self->flags & ix));
1230 1645
1231void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) 1646void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
1232 PPCODE: 1647 PPCODE:
1281 EXTEND (SP, 2); 1696 EXTEND (SP, 2);
1282 PUSHs (sv); 1697 PUSHs (sv);
1283 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); 1698 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
1284} 1699}
1285 1700
1701void incr_parse (CBOR *self, SV *cborstr)
1702 ALIAS:
1703 incr_parse_multiple = 1
1704 PPCODE:
1705{
1706 if (SvUTF8 (cborstr))
1707 sv_utf8_downgrade (cborstr, 0);
1708
1709 if (!self->incr_count)
1710 {
1711 self->incr_count = newAV ();
1712 self->incr_pos = 0;
1713 self->incr_need = 1;
1714
1715 av_push (self->incr_count, newSViv (1));
1716 }
1717
1718 do
1719 {
1720 if (!incr_parse (self, cborstr))
1721 {
1722 if (self->incr_need > self->max_size && self->max_size)
1723 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
1724 (unsigned long)self->incr_need, (unsigned long)self->max_size);
1725
1726 break;
1727 }
1728
1729 SV *sv;
1730 char *offset;
1731
1732 PUTBACK; sv = decode_cbor (cborstr, self, &offset); SPAGAIN;
1733 XPUSHs (sv);
1734
1735 sv_chop (cborstr, offset);
1736
1737 av_clear (self->incr_count);
1738 av_push (self->incr_count, newSViv (1));
1739
1740 self->incr_pos = 0;
1741 self->incr_need = self->incr_pos + 1;
1742 }
1743 while (ix);
1744}
1745
1746void incr_reset (CBOR *self)
1747 CODE:
1748{
1749 SvREFCNT_dec (self->incr_count);
1750 self->incr_count = 0;
1751}
1752
1286void DESTROY (CBOR *self) 1753void DESTROY (CBOR *self)
1287 PPCODE: 1754 PPCODE:
1288 cbor_free (self); 1755 cbor_free (self);
1289 1756
1290PROTOTYPES: ENABLE 1757PROTOTYPES: ENABLE
1291 1758
1292void encode_cbor (SV *scalar) 1759void encode_cbor (SV *scalar)
1760 ALIAS:
1761 encode_cbor = 0
1762 encode_cbor_sharing = F_ALLOW_SHARING
1293 PPCODE: 1763 PPCODE:
1294{ 1764{
1295 CBOR cbor; 1765 CBOR cbor;
1296 cbor_init (&cbor); 1766 cbor_init (&cbor);
1767 cbor.flags |= ix;
1297 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN; 1768 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;
1298 XPUSHs (scalar); 1769 XPUSHs (scalar);
1299} 1770}
1300 1771
1301void decode_cbor (SV *cborstr) 1772void decode_cbor (SV *cborstr)
1305 cbor_init (&cbor); 1776 cbor_init (&cbor);
1306 PUTBACK; cborstr = decode_cbor (cborstr, &cbor, 0); SPAGAIN; 1777 PUTBACK; cborstr = decode_cbor (cborstr, &cbor, 0); SPAGAIN;
1307 XPUSHs (cborstr); 1778 XPUSHs (cborstr);
1308} 1779}
1309 1780
1781#ifdef __AFL_COMPILER
1782
1783void
1784afl_init ()
1785 CODE:
1786 __AFL_INIT ();
1787
1788int
1789afl_loop (unsigned int count = 10000)
1790 CODE:
1791 RETVAL = __AFL_LOOP (count);
1792 OUTPUT:
1793 RETVAL
1794
1795#endif
1796

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines