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 1.64 by root, Tue Jun 27 02:03:24 2017 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
97}; 102};
98 103
99#define F_SHRINK 0x00000001UL 104#define F_SHRINK 0x00000001UL
100#define F_ALLOW_UNKNOWN 0x00000002UL 105#define F_ALLOW_UNKNOWN 0x00000002UL
101#define F_ALLOW_SHARING 0x00000004UL 106#define F_ALLOW_SHARING 0x00000004UL
107#define F_ALLOW_CYCLES 0x00000008UL
108#define F_FORBID_OBJECTS 0x00000010UL
102#define F_PACK_STRINGS 0x00000008UL 109#define F_PACK_STRINGS 0x00000020UL
110#define F_TEXT_KEYS 0x00000040UL
111#define F_TEXT_STRINGS 0x00000080UL
112#define F_VALIDATE_UTF8 0x00000100UL
103 113
104#define INIT_SIZE 32 // initial scalar size to be allocated 114#define INIT_SIZE 32 // initial scalar size to be allocated
105 115
106#define SB do { 116#define SB do {
107#define SE } while (0) 117#define SE } while (0)
126typedef struct { 136typedef struct {
127 U32 flags; 137 U32 flags;
128 U32 max_depth; 138 U32 max_depth;
129 STRLEN max_size; 139 STRLEN max_size;
130 SV *filter; 140 SV *filter;
141
142 // for the incremental parser
143 STRLEN incr_pos; // the current offset into the text
144 STRLEN incr_need; // minimum bytes needed to decode
145 AV *incr_count; // for every nesting level, the number of outstanding values, or -1 for indef.
131} CBOR; 146} CBOR;
132 147
133ecb_inline void 148ecb_inline void
134cbor_init (CBOR *cbor) 149cbor_init (CBOR *cbor)
135{ 150{
139 154
140ecb_inline void 155ecb_inline void
141cbor_free (CBOR *cbor) 156cbor_free (CBOR *cbor)
142{ 157{
143 SvREFCNT_dec (cbor->filter); 158 SvREFCNT_dec (cbor->filter);
159 SvREFCNT_dec (cbor->incr_count);
144} 160}
145 161
146///////////////////////////////////////////////////////////////////////////// 162/////////////////////////////////////////////////////////////////////////////
147// utility functions 163// utility functions
148 164
174 190
175// minimum length of a string to be registered for stringref 191// minimum length of a string to be registered for stringref
176ecb_inline int 192ecb_inline int
177minimum_string_length (UV idx) 193minimum_string_length (UV idx)
178{ 194{
179 return idx > 23 195 return idx <= 23 ? 3
180 ? idx > 0xffU 196 : idx <= 0xffU ? 4
181 ? idx > 0xffffU 197 : idx <= 0xffffU ? 5
182 ? idx > 0xffffffffU 198 : idx <= 0xffffffffU ? 7
183 ? 11 199 : 11;
184 : 7
185 : 5
186 : 4
187 : 3;
188} 200}
189 201
190///////////////////////////////////////////////////////////////////////////// 202/////////////////////////////////////////////////////////////////////////////
191// encoder 203// encoder
192 204
205} enc_t; 217} enc_t;
206 218
207ecb_inline void 219ecb_inline void
208need (enc_t *enc, STRLEN len) 220need (enc_t *enc, STRLEN len)
209{ 221{
210 if (ecb_expect_false (enc->cur + len >= enc->end)) 222 if (ecb_expect_false ((uintptr_t)(enc->end - enc->cur) < len))
211 { 223 {
212 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv); 224 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
213 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1); 225 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
214 enc->cur = SvPVX (enc->sv) + cur; 226 enc->cur = SvPVX (enc->sv) + cur;
215 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; 227 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
228{ 240{
229 need (enc, 9); 241 need (enc, 9);
230 242
231 if (ecb_expect_true (len < LENGTH_EXT1)) 243 if (ecb_expect_true (len < LENGTH_EXT1))
232 *enc->cur++ = major | len; 244 *enc->cur++ = major | len;
233 else if (ecb_expect_true (len <= 0xff)) 245 else if (ecb_expect_true (len <= 0xffU))
234 { 246 {
235 *enc->cur++ = major | LENGTH_EXT1; 247 *enc->cur++ = major | LENGTH_EXT1;
236 *enc->cur++ = len; 248 *enc->cur++ = len;
237 } 249 }
238 else if (len <= 0xffff) 250 else if (len <= 0xffffU)
239 { 251 {
240 *enc->cur++ = major | LENGTH_EXT2; 252 *enc->cur++ = major | LENGTH_EXT2;
241 *enc->cur++ = len >> 8; 253 *enc->cur++ = len >> 8;
242 *enc->cur++ = len; 254 *enc->cur++ = len;
243 } 255 }
244 else if (len <= 0xffffffff) 256 else if (len <= 0xffffffffU)
245 { 257 {
246 *enc->cur++ = major | LENGTH_EXT4; 258 *enc->cur++ = major | LENGTH_EXT4;
247 *enc->cur++ = len >> 24; 259 *enc->cur++ = len >> 24;
248 *enc->cur++ = len >> 16; 260 *enc->cur++ = len >> 16;
249 *enc->cur++ = len >> 8; 261 *enc->cur++ = len >> 8;
267encode_tag (enc_t *enc, UV tag) 279encode_tag (enc_t *enc, UV tag)
268{ 280{
269 encode_uint (enc, MAJOR_TAG, tag); 281 encode_uint (enc, MAJOR_TAG, tag);
270} 282}
271 283
284// exceptional (hopefully) slow path for byte strings that need to be utf8-encoded
285ecb_noinline static void
286encode_str_utf8 (enc_t *enc, int utf8, char *str, STRLEN len)
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
272ecb_inline void 307ecb_inline void
273encode_str (enc_t *enc, int utf8, char *str, STRLEN len) 308encode_str (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len)
274{ 309{
310 if (ecb_expect_false (upgrade_utf8))
311 if (!utf8)
312 {
313 encode_str_utf8 (enc, utf8, str, len);
314 return;
315 }
316
275 encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len); 317 encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len);
276 need (enc, len); 318 need (enc, len);
277 memcpy (enc->cur, str, len); 319 memcpy (enc->cur, str, len);
278 enc->cur += len; 320 enc->cur += len;
279} 321}
280 322
281static void 323ecb_inline void
282encode_strref (enc_t *enc, int utf8, char *str, STRLEN len) 324encode_strref (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len)
283{ 325{
284 if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS)) 326 if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS))
285 { 327 {
286 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1); 328 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1);
287 329
298 sv_setuv (*svp, enc->stringref_idx); 340 sv_setuv (*svp, enc->stringref_idx);
299 ++enc->stringref_idx; 341 ++enc->stringref_idx;
300 } 342 }
301 } 343 }
302 344
303 encode_str (enc, utf8, str, len); 345 encode_str (enc, upgrade_utf8, utf8, str, len);
304} 346}
305 347
306static void encode_sv (enc_t *enc, SV *sv); 348static void encode_sv (enc_t *enc, SV *sv);
307 349
308static void 350static void
315 357
316 ++enc->depth; 358 ++enc->depth;
317 359
318 encode_uint (enc, MAJOR_ARRAY, len + 1); 360 encode_uint (enc, MAJOR_ARRAY, len + 1);
319 361
362 if (ecb_expect_false (SvMAGICAL (av)))
320 for (i = 0; i <= len; ++i) 363 for (i = 0; i <= len; ++i)
321 { 364 {
322 SV **svp = av_fetch (av, i, 0); 365 SV **svp = av_fetch (av, i, 0);
323 encode_sv (enc, svp ? *svp : &PL_sv_undef); 366 encode_sv (enc, svp ? *svp : &PL_sv_undef);
324 } 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 }
325 374
326 --enc->depth; 375 --enc->depth;
327} 376}
328 377
329static void 378static void
337 ++enc->depth; 386 ++enc->depth;
338 387
339 int pairs = hv_iterinit (hv); 388 int pairs = hv_iterinit (hv);
340 int mg = SvMAGICAL (hv); 389 int mg = SvMAGICAL (hv);
341 390
342 if (mg) 391 if (ecb_expect_false (mg))
343 encode_ch (enc, MAJOR_MAP | MINOR_INDEF); 392 encode_ch (enc, MAJOR_MAP | MINOR_INDEF);
344 else 393 else
345 encode_uint (enc, MAJOR_MAP, pairs); 394 encode_uint (enc, MAJOR_MAP, pairs);
346 395
347 while ((he = hv_iternext (hv))) 396 while ((he = hv_iternext (hv)))
348 { 397 {
349 if (HeKLEN (he) == HEf_SVKEY) 398 if (HeKLEN (he) == HEf_SVKEY)
350 encode_sv (enc, HeSVKEY (he)); 399 encode_sv (enc, HeSVKEY (he));
351 else 400 else
352 encode_strref (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));
353 402
354 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));
355 } 404 }
356 405
357 if (mg) 406 if (ecb_expect_false (mg))
358 encode_ch (enc, MAJOR_MISC | MINOR_INDEF); 407 encode_ch (enc, MAJOR_MISC | MINOR_INDEF);
359 408
360 --enc->depth; 409 --enc->depth;
361} 410}
362 411
405 } 454 }
406 455
407 if (ecb_expect_false (SvREFCNT (sv) > 1) 456 if (ecb_expect_false (SvREFCNT (sv) > 1)
408 && ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING)) 457 && ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING))
409 { 458 {
410 if (!enc->shareable) 459 if (ecb_expect_false (!enc->shareable))
411 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ()); 460 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
412 461
413 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1); 462 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
414 463
415 if (SvOK (*svp)) 464 if (SvOK (*svp))
429 if (ecb_expect_false (SvOBJECT (sv))) 478 if (ecb_expect_false (SvOBJECT (sv)))
430 { 479 {
431 HV *stash = SvSTASH (sv); 480 HV *stash = SvSTASH (sv);
432 GV *method; 481 GV *method;
433 482
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))));
434 if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0))) 486 else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
435 { 487 {
436 dSP; 488 dSP;
437 489
438 ENTER; SAVETMPS; PUSHMARK (SP); 490 ENTER; SAVETMPS;
491 PUSHMARK (SP);
439 // 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
440 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); 493 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
441 494
442 PUTBACK; 495 PUTBACK;
443 // G_SCALAR ensures that return value is 1 496 // G_SCALAR ensures that return value is 1
456 } 509 }
457 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0) 510 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)
458 { 511 {
459 dSP; 512 dSP;
460 513
461 ENTER; SAVETMPS; PUSHMARK (SP); 514 ENTER; SAVETMPS;
515 SAVESTACK_POS ();
516 PUSHMARK (SP);
462 EXTEND (SP, 2); 517 EXTEND (SP, 2);
463 // 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
464 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); 519 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
465 PUSHs (sv_cbor); 520 PUSHs (sv_cbor);
466 521
472 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv) 527 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)); 528 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));
474 529
475 encode_tag (enc, CBOR_TAG_PERL_OBJECT); 530 encode_tag (enc, CBOR_TAG_PERL_OBJECT);
476 encode_uint (enc, MAJOR_ARRAY, count + 1); 531 encode_uint (enc, MAJOR_ARRAY, count + 1);
477 encode_strref (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash)); 532 encode_strref (enc, 0, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
478 533
479 while (count) 534 while (count)
480 encode_sv (enc, SP[1 - count--]); 535 encode_sv (enc, SP[1 - count--]);
481 536
482 PUTBACK; 537 PUTBACK;
508 if (ecb_expect_false (nv == (NV)(U32)nv)) 563 if (ecb_expect_false (nv == (NV)(U32)nv))
509 encode_uint (enc, MAJOR_POS_INT, (U32)nv); 564 encode_uint (enc, MAJOR_POS_INT, (U32)nv);
510 //TODO: maybe I32? 565 //TODO: maybe I32?
511 else if (ecb_expect_false (nv == (float)nv)) 566 else if (ecb_expect_false (nv == (float)nv))
512 { 567 {
568 *enc->cur++ = MAJOR_MISC | MISC_FLOAT32;
569
513 uint32_t fp = ecb_float_to_binary32 (nv); 570 uint32_t fp = ecb_float_to_binary32 (nv);
514
515 *enc->cur++ = MAJOR_MISC | MISC_FLOAT32;
516 571
517 if (!ecb_big_endian ()) 572 if (!ecb_big_endian ())
518 fp = ecb_bswap32 (fp); 573 fp = ecb_bswap32 (fp);
519 574
520 memcpy (enc->cur, &fp, 4); 575 memcpy (enc->cur, &fp, 4);
521 enc->cur += 4; 576 enc->cur += 4;
522 } 577 }
523 else 578 else
524 { 579 {
580 *enc->cur++ = MAJOR_MISC | MISC_FLOAT64;
581
525 uint64_t fp = ecb_double_to_binary64 (nv); 582 uint64_t fp = ecb_double_to_binary64 (nv);
526
527 *enc->cur++ = MAJOR_MISC | MISC_FLOAT64;
528 583
529 if (!ecb_big_endian ()) 584 if (!ecb_big_endian ())
530 fp = ecb_bswap64 (fp); 585 fp = ecb_bswap64 (fp);
531 586
532 memcpy (enc->cur, &fp, 8); 587 memcpy (enc->cur, &fp, 8);
541 596
542 if (SvPOKp (sv)) 597 if (SvPOKp (sv))
543 { 598 {
544 STRLEN len; 599 STRLEN len;
545 char *str = SvPV (sv, len); 600 char *str = SvPV (sv, len);
546 encode_strref (enc, SvUTF8 (sv), str, len); 601 encode_strref (enc, enc->cbor.flags & F_TEXT_STRINGS, SvUTF8 (sv), str, len);
547 } 602 }
548 else if (SvNOKp (sv)) 603 else if (SvNOKp (sv))
549 encode_nv (enc, sv); 604 encode_nv (enc, sv);
550 else if (SvIOKp (sv)) 605 else if (SvIOKp (sv))
551 { 606 {
568} 623}
569 624
570static SV * 625static SV *
571encode_cbor (SV *scalar, CBOR *cbor) 626encode_cbor (SV *scalar, CBOR *cbor)
572{ 627{
573 enc_t enc = { }; 628 enc_t enc = { 0 };
574 629
575 enc.cbor = *cbor; 630 enc.cbor = *cbor;
576 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 631 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
577 enc.cur = SvPVX (enc.sv); 632 enc.cur = SvPVX (enc.sv);
578 enc.end = SvEND (enc.sv); 633 enc.end = SvEND (enc.sv);
579 634
580 SvPOK_only (enc.sv); 635 SvPOK_only (enc.sv);
581 636
582 if (cbor->flags & F_PACK_STRINGS) 637 if (cbor->flags & F_PACK_STRINGS)
583 { 638 {
610 U32 depth; // recursion depth 665 U32 depth; // recursion depth
611 U32 maxdepth; // recursion depth limit 666 U32 maxdepth; // recursion depth limit
612 AV *shareable; 667 AV *shareable;
613 AV *stringref; 668 AV *stringref;
614 SV *decode_tagged; 669 SV *decode_tagged;
670 SV *err_sv; // optional sv for error, needs to be freed
615} dec_t; 671} dec_t;
616 672
617#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);
618 680
619#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;
620 684
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
703#define ERR_DO(do) SB do; goto fail; SE
704#define ERR(reason) ERR_DO (err_set (dec, reason))
705#define ERR_ERRSV ERR_DO (err_errsv (dec))
706
707#define WANT(len) if (ecb_expect_false ((uintptr_t)(dec->end - dec->cur) < (STRLEN)len)) ERR_DO (err_unexpected_end (dec))
708
621#define DEC_INC_DEPTH if (++dec->depth > dec->cbor.max_depth) ERR (ERR_NESTING_EXCEEDED) 709#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 710#define DEC_DEC_DEPTH --dec->depth
623 711
624static UV 712static UV
625decode_uint (dec_t *dec) 713decode_uint (dec_t *dec)
626{ 714{
627 U8 m = *dec->cur & MINOR_MASK; 715 U8 m = *dec->cur & MINOR_MASK;
628 ++dec->cur; 716 ++dec->cur;
629 717
630 if (ecb_expect_true (m < LENGTH_EXT1)) 718 if (ecb_expect_true (m < LENGTH_EXT1))
631 return m; 719 return m;
632 720 else if (ecb_expect_true (m == LENGTH_EXT1))
633 switch (m)
634 { 721 {
635 case LENGTH_EXT1:
636 WANT (1); 722 WANT (1);
637 dec->cur += 1; 723 dec->cur += 1;
638 return dec->cur[-1]; 724 return dec->cur[-1];
639 725 }
640 case LENGTH_EXT2: 726 else if (ecb_expect_true (m == LENGTH_EXT2))
727 {
641 WANT (2); 728 WANT (2);
642 dec->cur += 2; 729 dec->cur += 2;
643 return (((UV)dec->cur[-2]) << 8) 730 return (((UV)dec->cur[-2]) << 8)
644 | ((UV)dec->cur[-1]); 731 | ((UV)dec->cur[-1]);
645 732 }
646 case LENGTH_EXT4: 733 else if (ecb_expect_true (m == LENGTH_EXT4))
734 {
647 WANT (4); 735 WANT (4);
648 dec->cur += 4; 736 dec->cur += 4;
649 return (((UV)dec->cur[-4]) << 24) 737 return (((UV)dec->cur[-4]) << 24)
650 | (((UV)dec->cur[-3]) << 16) 738 | (((UV)dec->cur[-3]) << 16)
651 | (((UV)dec->cur[-2]) << 8) 739 | (((UV)dec->cur[-2]) << 8)
652 | ((UV)dec->cur[-1]); 740 | ((UV)dec->cur[-1]);
653 741 }
654 case LENGTH_EXT8: 742 else if (ecb_expect_true (m == LENGTH_EXT8))
743 {
655 WANT (8); 744 WANT (8);
656 dec->cur += 8; 745 dec->cur += 8;
657 746
658 return 747 return
659#if UVSIZE < 8 748#if UVSIZE < 8
660 0 749 0
661#else 750#else
662 (((UV)dec->cur[-8]) << 56) 751 (((UV)dec->cur[-8]) << 56)
663 | (((UV)dec->cur[-7]) << 48) 752 | (((UV)dec->cur[-7]) << 48)
664 | (((UV)dec->cur[-6]) << 40) 753 | (((UV)dec->cur[-6]) << 40)
665 | (((UV)dec->cur[-5]) << 32) 754 | (((UV)dec->cur[-5]) << 32)
666#endif 755#endif
667 | (((UV)dec->cur[-4]) << 24) 756 | (((UV)dec->cur[-4]) << 24)
668 | (((UV)dec->cur[-3]) << 16) 757 | (((UV)dec->cur[-3]) << 16)
669 | (((UV)dec->cur[-2]) << 8) 758 | (((UV)dec->cur[-2]) << 8)
670 | ((UV)dec->cur[-1]); 759 | ((UV)dec->cur[-1]);
671 760 }
672 default: 761 else
673 ERR ("corrupted CBOR data (unsupported integer minor encoding)"); 762 ERR ("corrupted CBOR data (unsupported integer minor encoding)");
674 }
675 763
676fail: 764fail:
677 return 0; 765 return 0;
678} 766}
679 767
692 780
693 for (;;) 781 for (;;)
694 { 782 {
695 WANT (1); 783 WANT (1);
696 784
697 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF)) 785 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF) || dec->err)
698 { 786 {
699 ++dec->cur; 787 ++dec->cur;
700 break; 788 break;
701 } 789 }
702 790
703 av_push (av, decode_sv (dec)); 791 av_push (av, decode_sv (dec));
704 } 792 }
705 } 793 }
706 else 794 else
707 { 795 {
708 int i, len = decode_uint (dec); 796 UV i, len = decode_uint (dec);
709 797
798 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); 799 av_fill (av, len - 1);
711 800
712 for (i = 0; i < len; ++i) 801 for (i = 0; i < len; ++i)
713 AvARRAY (av)[i] = decode_sv (dec); 802 AvARRAY (av)[i] = decode_sv (dec);
714 } 803 }
715 804
716 DEC_DEC_DEPTH; 805 DEC_DEC_DEPTH;
717 return newRV_noinc ((SV *)av); 806 return newRV_noinc ((SV *)av);
718 807
719fail: 808fail:
720 SvREFCNT_dec (av); 809 SvREFCNT_dec_NN (av);
721 DEC_DEC_DEPTH; 810 DEC_DEC_DEPTH;
722 return &PL_sv_undef; 811 return &PL_sv_undef;
723} 812}
724 813
725static void 814static void
727{ 816{
728 // for speed reasons, we specialcase single-string 817 // for speed reasons, we specialcase single-string
729 // byte or utf-8 strings as keys, but only when !stringref 818 // byte or utf-8 strings as keys, but only when !stringref
730 819
731 if (ecb_expect_true (!dec->stringref)) 820 if (ecb_expect_true (!dec->stringref))
732 if ((*dec->cur - MAJOR_BYTES) <= 27) 821 if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8))
733 { 822 {
734 I32 len = decode_uint (dec); 823 STRLEN len = decode_uint (dec);
735 char *key = (char *)dec->cur; 824 char *key = (char *)dec->cur;
736 825
826 WANT (len);
737 dec->cur += len; 827 dec->cur += len;
738 828
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); 829 hv_store (hv, key, len, decode_sv (dec), 0);
743 830
744 return; 831 return;
745 } 832 }
746 else if ((*dec->cur - MAJOR_TEXT) <= 27) 833 else if (ecb_expect_true ((U8)(*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8))
747 { 834 {
748 I32 len = decode_uint (dec); 835 STRLEN len = decode_uint (dec);
749 char *key = (char *)dec->cur; 836 char *key = (char *)dec->cur;
750 837
838 WANT (len);
751 dec->cur += len; 839 dec->cur += len;
752 840
753 if (ecb_expect_false (dec->stringref)) 841 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
754 av_push (dec->stringref, newSVpvn_utf8 (key, len, 1)); 842 if (!is_utf8_string (key, len))
843 ERR ("corrupted CBOR data (invalid UTF-8 in map key)");
755 844
756 hv_store (hv, key, -len, decode_sv (dec), 0); 845 hv_store (hv, key, -len, decode_sv (dec), 0);
757 846
758 return; 847 return;
759 } 848 }
760 849
761 SV *k = decode_sv (dec); 850 SV *k = decode_sv (dec);
762 SV *v = decode_sv (dec); 851 SV *v = decode_sv (dec);
763 852
853 // we leak memory if uncaught exceptions are thrown by random magical
854 // methods, and this is hopefully the only place where it can happen,
855 // so if there is a chance of an exception, take the very slow path.
856 // since catching exceptions is "undocumented/internal/forbidden" by
857 // the new p5p powers, we need to call out to a perl function :/
858 if (ecb_expect_false (SvAMAGIC (k)))
859 {
860 dSP;
861
862 ENTER; SAVETMPS;
863 PUSHMARK (SP);
864 EXTEND (SP, 3);
865 PUSHs (sv_2mortal (newRV_inc ((SV *)hv)));
866 PUSHs (sv_2mortal (k));
867 PUSHs (sv_2mortal (v));
868
869 PUTBACK;
870 call_pv ("CBOR::XS::_hv_store", G_VOID | G_DISCARD | G_EVAL);
871 SPAGAIN;
872
873 FREETMPS; LEAVE;
874
875 if (SvTRUE (ERRSV))
876 ERR_ERRSV;
877
878 return;
879 }
880
764 hv_store_ent (hv, k, v, 0); 881 hv_store_ent (hv, k, v, 0);
765 SvREFCNT_dec (k); 882 SvREFCNT_dec_NN (k);
883
884fail:
885 ;
766} 886}
767 887
768static SV * 888static SV *
769decode_hv (dec_t *dec) 889decode_hv (dec_t *dec)
770{ 890{
778 898
779 for (;;) 899 for (;;)
780 { 900 {
781 WANT (1); 901 WANT (1);
782 902
783 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF)) 903 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF) || dec->err)
784 { 904 {
785 ++dec->cur; 905 ++dec->cur;
786 break; 906 break;
787 } 907 }
788 908
789 decode_he (dec, hv); 909 decode_he (dec, hv);
790 } 910 }
791 } 911 }
792 else 912 else
793 { 913 {
794 int pairs = decode_uint (dec); 914 UV pairs = decode_uint (dec);
915
916 WANT (pairs); // complexity check - need at least one byte per value, do not allow supersize hashes
795 917
796 while (pairs--) 918 while (pairs--)
797 decode_he (dec, hv); 919 decode_he (dec, hv);
798 } 920 }
799 921
800 DEC_DEC_DEPTH; 922 DEC_DEC_DEPTH;
801 return newRV_noinc ((SV *)hv); 923 return newRV_noinc ((SV *)hv);
802 924
803fail: 925fail:
804 SvREFCNT_dec (hv); 926 SvREFCNT_dec_NN (hv);
805 DEC_DEC_DEPTH; 927 DEC_DEC_DEPTH;
806 return &PL_sv_undef; 928 return &PL_sv_undef;
807} 929}
808 930
809static SV * 931static SV *
810decode_str (dec_t *dec, int utf8) 932decode_str (dec_t *dec, int utf8)
811{ 933{
812 SV *sv = 0; 934 SV *sv = 0;
813 935
814 if ((*dec->cur & MINOR_MASK) == MINOR_INDEF) 936 if (ecb_expect_false ((*dec->cur & MINOR_MASK) == MINOR_INDEF))
815 { 937 {
816 // indefinite length strings 938 // indefinite length strings
817 ++dec->cur; 939 ++dec->cur;
818 940
819 U8 major = *dec->cur & MAJOR_MISC; 941 U8 major = *dec->cur & MAJOR_MISC;
852 && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1)) 974 && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1))
853 av_push (dec->stringref, SvREFCNT_inc_NN (sv)); 975 av_push (dec->stringref, SvREFCNT_inc_NN (sv));
854 } 976 }
855 977
856 if (utf8) 978 if (utf8)
979 {
980 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
981 if (!is_utf8_string (SvPVX (sv), SvCUR (sv)))
982 ERR ("corrupted CBOR data (invalid UTF-8 in text string)");
983
857 SvUTF8_on (sv); 984 SvUTF8_on (sv);
985 }
858 986
859 return sv; 987 return sv;
860 988
861fail: 989fail:
862 SvREFCNT_dec (sv); 990 SvREFCNT_dec (sv);
881 sv = newRV_noinc (decode_sv (dec)); 1009 sv = newRV_noinc (decode_sv (dec));
882 break; 1010 break;
883 1011
884 case CBOR_TAG_STRINGREF_NAMESPACE: 1012 case CBOR_TAG_STRINGREF_NAMESPACE:
885 { 1013 {
1014 // do not use SAVETMPS/FREETMPS, as these will
1015 // erase mortalised caches, e.g. "shareable"
886 ENTER; SAVETMPS; 1016 ENTER;
887 1017
888 SAVESPTR (dec->stringref); 1018 SAVESPTR (dec->stringref);
889 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ()); 1019 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ());
890 1020
891 sv = decode_sv (dec); 1021 sv = decode_sv (dec);
892 1022
893 FREETMPS; LEAVE; 1023 LEAVE;
894 } 1024 }
895 break; 1025 break;
896 1026
897 case CBOR_TAG_STRINGREF: 1027 case CBOR_TAG_STRINGREF:
898 { 1028 {
899 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT)) 1029 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
900 ERR ("corrupted CBOR data (stringref index not an unsigned integer)"); 1030 ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
901 1031
902 UV idx = decode_uint (dec); 1032 UV idx = decode_uint (dec);
903 1033
904 if (!dec->stringref || (int)idx > AvFILLp (dec->stringref)) 1034 if (!dec->stringref || idx >= (UV)(1 + AvFILLp (dec->stringref)))
905 ERR ("corrupted CBOR data (stringref index out of bounds or outside namespace)"); 1035 ERR ("corrupted CBOR data (stringref index out of bounds or outside namespace)");
906 1036
907 sv = newSVsv (AvARRAY (dec->stringref)[idx]); 1037 sv = newSVsv (AvARRAY (dec->stringref)[idx]);
908 } 1038 }
909 break; 1039 break;
911 case CBOR_TAG_VALUE_SHAREABLE: 1041 case CBOR_TAG_VALUE_SHAREABLE:
912 { 1042 {
913 if (ecb_expect_false (!dec->shareable)) 1043 if (ecb_expect_false (!dec->shareable))
914 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ()); 1044 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ());
915 1045
1046 if (dec->cbor.flags & F_ALLOW_CYCLES)
1047 {
916 sv = newSV (0); 1048 sv = newSV (0);
917 av_push (dec->shareable, SvREFCNT_inc_NN (sv)); 1049 av_push (dec->shareable, SvREFCNT_inc_NN (sv));
918 1050
919 SV *osv = decode_sv (dec); 1051 SV *osv = decode_sv (dec);
920 sv_setsv (sv, osv); 1052 sv_setsv (sv, osv);
921 SvREFCNT_dec_NN (osv); 1053 SvREFCNT_dec_NN (osv);
1054 }
1055 else
1056 {
1057 av_push (dec->shareable, &PL_sv_undef);
1058 int idx = AvFILLp (dec->shareable);
1059 sv = decode_sv (dec);
1060 av_store (dec->shareable, idx, SvREFCNT_inc_NN (sv));
1061 }
922 } 1062 }
923 break; 1063 break;
924 1064
925 case CBOR_TAG_VALUE_SHAREDREF: 1065 case CBOR_TAG_VALUE_SHAREDREF:
926 { 1066 {
927 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT)) 1067 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
928 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)"); 1068 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
929 1069
930 UV idx = decode_uint (dec); 1070 UV idx = decode_uint (dec);
931 1071
932 if (!dec->shareable || (int)idx > AvFILLp (dec->shareable)) 1072 if (!dec->shareable || idx >= (UV)(1 + AvFILLp (dec->shareable)))
933 ERR ("corrupted CBOR data (sharedref index out of bounds)"); 1073 ERR ("corrupted CBOR data (sharedref index out of bounds)");
934 1074
935 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]); 1075 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]);
1076
1077 if (sv == &PL_sv_undef)
1078 ERR ("cyclic CBOR data structure found, but allow_cycles is not enabled");
936 } 1079 }
937 break; 1080 break;
938 1081
939 case CBOR_TAG_PERL_OBJECT: 1082 case CBOR_TAG_PERL_OBJECT:
940 { 1083 {
1084 if (dec->cbor.flags & F_FORBID_OBJECTS)
1085 goto filter;
1086
941 sv = decode_sv (dec); 1087 sv = decode_sv (dec);
942 1088
943 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV) 1089 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
944 ERR ("corrupted CBOR data (non-array perl object)"); 1090 ERR ("corrupted CBOR data (non-array perl object)");
945 1091
955 if (!method) 1101 if (!method)
956 ERR ("cannot decode perl-object (package does not have a THAW method)"); 1102 ERR ("cannot decode perl-object (package does not have a THAW method)");
957 1103
958 dSP; 1104 dSP;
959 1105
960 ENTER; SAVETMPS; PUSHMARK (SP); 1106 ENTER; SAVETMPS;
1107 PUSHMARK (SP);
961 EXTEND (SP, len + 1); 1108 EXTEND (SP, len + 1);
962 // we re-bless the reference to get overload and other niceties right 1109 // we re-bless the reference to get overload and other niceties right
963 PUSHs (*av_fetch (av, 0, 1)); 1110 PUSHs (*av_fetch (av, 0, 1));
964 PUSHs (sv_cbor); 1111 PUSHs (sv_cbor);
965 1112
973 SPAGAIN; 1120 SPAGAIN;
974 1121
975 if (SvTRUE (ERRSV)) 1122 if (SvTRUE (ERRSV))
976 { 1123 {
977 FREETMPS; LEAVE; 1124 FREETMPS; LEAVE;
978 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV)))); 1125 ERR_ERRSV;
979 } 1126 }
980 1127
981 SvREFCNT_dec (sv); 1128 SvREFCNT_dec_NN (sv);
982 sv = SvREFCNT_inc (POPs); 1129 sv = SvREFCNT_inc (POPs);
983 1130
984 PUTBACK; 1131 PUTBACK;
985 1132
986 FREETMPS; LEAVE; 1133 FREETMPS; LEAVE;
987 } 1134 }
988 break; 1135 break;
989 1136
990 default: 1137 default:
1138 filter:
991 { 1139 {
1140 SV *tag_sv = newSVuv (tag);
1141
992 sv = decode_sv (dec); 1142 sv = decode_sv (dec);
993 1143
994 dSP; 1144 dSP;
995 ENTER; SAVETMPS; PUSHMARK (SP); 1145 ENTER; SAVETMPS;
1146 SAVESTACK_POS ();
1147 PUSHMARK (SP);
996 EXTEND (SP, 2); 1148 EXTEND (SP, 2);
997 PUSHs (newSVuv (tag)); 1149 PUSHs (tag_sv);
998 PUSHs (sv); 1150 PUSHs (sv);
999 1151
1000 PUTBACK; 1152 PUTBACK;
1001 int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL); 1153 int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL);
1002 SPAGAIN; 1154 SPAGAIN;
1003 1155
1004 if (SvTRUE (ERRSV)) 1156 if (SvTRUE (ERRSV))
1005 { 1157 {
1158 SvREFCNT_dec_NN (tag_sv);
1006 FREETMPS; LEAVE; 1159 FREETMPS; LEAVE;
1007 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV)))); 1160 ERR_ERRSV;
1008 } 1161 }
1009 1162
1010 if (count) 1163 if (count)
1011 { 1164 {
1165 SvREFCNT_dec_NN (tag_sv);
1012 SvREFCNT_dec (sv); 1166 SvREFCNT_dec_NN (sv);
1013 sv = SvREFCNT_inc (POPs); 1167 sv = SvREFCNT_inc_NN (POPs);
1014 } 1168 }
1015 else 1169 else
1016 { 1170 {
1017 AV *av = newAV (); 1171 AV *av = newAV ();
1018 av_push (av, newSVuv (tag)); 1172 av_push (av, tag_sv);
1019 av_push (av, sv); 1173 av_push (av, sv);
1020 1174
1021 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 1175 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
1022 ? cbor_tagged_stash 1176 ? cbor_tagged_stash
1023 : gv_stashpv ("CBOR::XS::Tagged" , 1); 1177 : gv_stashpv ("CBOR::XS::Tagged" , 1);
1109 1263
1110 return newSVnv (ecb_binary64_to_double (fp)); 1264 return newSVnv (ecb_binary64_to_double (fp));
1111 } 1265 }
1112 1266
1113 // 0..19 unassigned simple 1267 // 0..19 unassigned simple
1114 // 24 reserved + unassigned (reserved values are not encodable) 1268 // 24 reserved + unassigned simple (reserved values are not encodable)
1269 // 28-30 unassigned misc
1270 // 31 break code
1115 default: 1271 default:
1116 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)"); 1272 ERR ("corrupted CBOR data (reserved/unassigned/unexpected major 7 value)");
1117 } 1273 }
1118 1274
1119 break; 1275 break;
1120 } 1276 }
1121 1277
1124} 1280}
1125 1281
1126static SV * 1282static SV *
1127decode_cbor (SV *string, CBOR *cbor, char **offset_return) 1283decode_cbor (SV *string, CBOR *cbor, char **offset_return)
1128{ 1284{
1129 dec_t dec = { }; 1285 dec_t dec = { 0 };
1130 SV *sv; 1286 SV *sv;
1131 STRLEN len; 1287 STRLEN len;
1132 char *data = SvPVbyte (string, len); 1288 char *data = SvPVbyte (string, len);
1133 1289
1134 if (len > cbor->max_size && cbor->max_size) 1290 if (len > cbor->max_size && cbor->max_size)
1148 if (dec.cur != dec.end && !dec.err) 1304 if (dec.cur != dec.end && !dec.err)
1149 dec.err = "garbage after CBOR object"; 1305 dec.err = "garbage after CBOR object";
1150 1306
1151 if (dec.err) 1307 if (dec.err)
1152 { 1308 {
1309 if (dec.shareable)
1310 {
1311 // need to break cyclic links, which would all be in shareable
1312 int i;
1313 SV **svp;
1314
1315 for (i = av_len (dec.shareable) + 1; i--; )
1316 if ((svp = av_fetch (dec.shareable, i, 0)))
1317 sv_setsv (*svp, &PL_sv_undef);
1318 }
1319
1153 SvREFCNT_dec (sv); 1320 SvREFCNT_dec_NN (sv);
1321
1322 if (dec.err_sv)
1323 sv_2mortal (dec.err_sv);
1324
1154 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur); 1325 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur);
1155 } 1326 }
1156 1327
1157 sv = sv_2mortal (sv); 1328 sv = sv_2mortal (sv);
1158 1329
1159 return sv; 1330 return sv;
1160} 1331}
1161 1332
1333/////////////////////////////////////////////////////////////////////////////
1334// incremental parser
1335
1336#define INCR_DONE(cbor) (AvFILLp (cbor->incr_count) < 0)
1337
1338// returns 0 for notyet, 1 for success or error
1339static int
1340incr_parse (CBOR *self, SV *cborstr)
1341{
1342 STRLEN cur;
1343 SvPV (cborstr, cur);
1344
1345 while (ecb_expect_true (self->incr_need <= cur))
1346 {
1347 // table of integer count bytes
1348 static I8 incr_len[MINOR_MASK + 1] = {
1349 0, 0, 0, 0, 0, 0, 0, 0,
1350 0, 0, 0, 0, 0, 0, 0, 0,
1351 0, 0, 0, 0, 0, 0, 0, 0,
1352 1, 2, 4, 8,-1,-1,-1,-2
1353 };
1354
1355 const U8 *p = SvPVX (cborstr) + self->incr_pos;
1356 U8 m = *p & MINOR_MASK;
1357 IV count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1358 I8 ilen = incr_len[m];
1359
1360 self->incr_need = self->incr_pos + 1;
1361
1362 if (ecb_expect_false (ilen < 0))
1363 {
1364 if (m != MINOR_INDEF)
1365 return 1; // error
1366
1367 if (*p == (MAJOR_MISC | MINOR_INDEF))
1368 {
1369 if (count >= 0)
1370 return 1; // error
1371
1372 count = 1;
1373 }
1374 else
1375 {
1376 av_push (self->incr_count, newSViv (-1)); //TODO: nest
1377 count = -1;
1378 }
1379 }
1380 else
1381 {
1382 self->incr_need += ilen;
1383 if (ecb_expect_false (self->incr_need > cur))
1384 return 0;
1385
1386 int major = *p >> MAJOR_SHIFT;
1387
1388 switch (major)
1389 {
1390 case MAJOR_TAG >> MAJOR_SHIFT:
1391 ++count; // tags merely prefix another value
1392 break;
1393
1394 case MAJOR_BYTES >> MAJOR_SHIFT:
1395 case MAJOR_TEXT >> MAJOR_SHIFT:
1396 case MAJOR_ARRAY >> MAJOR_SHIFT:
1397 case MAJOR_MAP >> MAJOR_SHIFT:
1398 {
1399 UV len;
1400
1401 if (ecb_expect_false (ilen))
1402 {
1403 len = 0;
1404
1405 do {
1406 len = (len << 8) | *++p;
1407 } while (--ilen);
1408 }
1409 else
1410 len = m;
1411
1412 switch (major)
1413 {
1414 case MAJOR_BYTES >> MAJOR_SHIFT:
1415 case MAJOR_TEXT >> MAJOR_SHIFT:
1416 self->incr_need += len;
1417 if (ecb_expect_false (self->incr_need > cur))
1418 return 0;
1419
1420 break;
1421
1422 case MAJOR_MAP >> MAJOR_SHIFT:
1423 len <<= 1;
1424 case MAJOR_ARRAY >> MAJOR_SHIFT:
1425 if (len)
1426 {
1427 av_push (self->incr_count, newSViv (len + 1)); //TODO: nest
1428 count = len + 1;
1429 }
1430 break;
1431 }
1432 }
1433 }
1434 }
1435
1436 self->incr_pos = self->incr_need;
1437
1438 if (count > 0)
1439 {
1440 while (!--count)
1441 {
1442 if (!AvFILLp (self->incr_count))
1443 return 1; // done
1444
1445 SvREFCNT_dec_NN (av_pop (self->incr_count));
1446 count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1447 }
1448
1449 SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]) = count;
1450 }
1451 }
1452
1453 return 0;
1454}
1455
1456
1162///////////////////////////////////////////////////////////////////////////// 1457/////////////////////////////////////////////////////////////////////////////
1163// XS interface functions 1458// XS interface functions
1164 1459
1165MODULE = CBOR::XS PACKAGE = CBOR::XS 1460MODULE = CBOR::XS PACKAGE = CBOR::XS
1166 1461
1178 1473
1179 default_filter = newSVpv ("CBOR::XS::default_filter", 0); 1474 default_filter = newSVpv ("CBOR::XS::default_filter", 0);
1180 1475
1181 sv_cbor = newSVpv ("CBOR", 0); 1476 sv_cbor = newSVpv ("CBOR", 0);
1182 SvREADONLY_on (sv_cbor); 1477 SvREADONLY_on (sv_cbor);
1478
1479 assert (("STRLEN must be an unsigned type", 0 <= (STRLEN)-1));
1183} 1480}
1184 1481
1185PROTOTYPES: DISABLE 1482PROTOTYPES: DISABLE
1186 1483
1187void CLONE (...) 1484void CLONE (...)
1206void shrink (CBOR *self, int enable = 1) 1503void shrink (CBOR *self, int enable = 1)
1207 ALIAS: 1504 ALIAS:
1208 shrink = F_SHRINK 1505 shrink = F_SHRINK
1209 allow_unknown = F_ALLOW_UNKNOWN 1506 allow_unknown = F_ALLOW_UNKNOWN
1210 allow_sharing = F_ALLOW_SHARING 1507 allow_sharing = F_ALLOW_SHARING
1508 allow_cycles = F_ALLOW_CYCLES
1509 forbid_objects = F_FORBID_OBJECTS
1211 pack_strings = F_PACK_STRINGS 1510 pack_strings = F_PACK_STRINGS
1511 text_keys = F_TEXT_KEYS
1512 text_strings = F_TEXT_STRINGS
1513 validate_utf8 = F_VALIDATE_UTF8
1212 PPCODE: 1514 PPCODE:
1213{ 1515{
1214 if (enable) 1516 if (enable)
1215 self->flags |= ix; 1517 self->flags |= ix;
1216 else 1518 else
1222void get_shrink (CBOR *self) 1524void get_shrink (CBOR *self)
1223 ALIAS: 1525 ALIAS:
1224 get_shrink = F_SHRINK 1526 get_shrink = F_SHRINK
1225 get_allow_unknown = F_ALLOW_UNKNOWN 1527 get_allow_unknown = F_ALLOW_UNKNOWN
1226 get_allow_sharing = F_ALLOW_SHARING 1528 get_allow_sharing = F_ALLOW_SHARING
1529 get_allow_cycles = F_ALLOW_CYCLES
1530 get_forbid_objects = F_FORBID_OBJECTS
1227 get_pack_strings = F_PACK_STRINGS 1531 get_pack_strings = F_PACK_STRINGS
1532 get_text_keys = F_TEXT_KEYS
1533 get_text_strings = F_TEXT_STRINGS
1534 get_validate_utf8 = F_VALIDATE_UTF8
1228 PPCODE: 1535 PPCODE:
1229 XPUSHs (boolSV (self->flags & ix)); 1536 XPUSHs (boolSV (self->flags & ix));
1230 1537
1231void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) 1538void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
1232 PPCODE: 1539 PPCODE:
1281 EXTEND (SP, 2); 1588 EXTEND (SP, 2);
1282 PUSHs (sv); 1589 PUSHs (sv);
1283 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); 1590 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
1284} 1591}
1285 1592
1593void incr_parse (CBOR *self, SV *cborstr)
1594 ALIAS:
1595 incr_parse_multiple = 1
1596 PPCODE:
1597{
1598 if (SvUTF8 (cborstr))
1599 sv_utf8_downgrade (cborstr, 0);
1600
1601 if (!self->incr_count)
1602 {
1603 self->incr_count = newAV ();
1604 self->incr_pos = 0;
1605 self->incr_need = 1;
1606
1607 av_push (self->incr_count, newSViv (1));
1608 }
1609
1610 do
1611 {
1612 if (!incr_parse (self, cborstr))
1613 {
1614 if (self->incr_need > self->max_size && self->max_size)
1615 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
1616 (unsigned long)self->incr_need, (unsigned long)self->max_size);
1617
1618 break;
1619 }
1620
1621 SV *sv;
1622 char *offset;
1623
1624 PUTBACK; sv = decode_cbor (cborstr, self, &offset); SPAGAIN;
1625 XPUSHs (sv);
1626
1627 sv_chop (cborstr, offset);
1628
1629 av_clear (self->incr_count);
1630 av_push (self->incr_count, newSViv (1));
1631
1632 self->incr_pos = 0;
1633 self->incr_need = self->incr_pos + 1;
1634 }
1635 while (ix);
1636}
1637
1638void incr_reset (CBOR *self)
1639 CODE:
1640{
1641 SvREFCNT_dec (self->incr_count);
1642 self->incr_count = 0;
1643}
1644
1286void DESTROY (CBOR *self) 1645void DESTROY (CBOR *self)
1287 PPCODE: 1646 PPCODE:
1288 cbor_free (self); 1647 cbor_free (self);
1289 1648
1290PROTOTYPES: ENABLE 1649PROTOTYPES: ENABLE
1291 1650
1292void encode_cbor (SV *scalar) 1651void encode_cbor (SV *scalar)
1652 ALIAS:
1653 encode_cbor = 0
1654 encode_cbor_sharing = F_ALLOW_SHARING
1293 PPCODE: 1655 PPCODE:
1294{ 1656{
1295 CBOR cbor; 1657 CBOR cbor;
1296 cbor_init (&cbor); 1658 cbor_init (&cbor);
1659 cbor.flags |= ix;
1297 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN; 1660 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;
1298 XPUSHs (scalar); 1661 XPUSHs (scalar);
1299} 1662}
1300 1663
1301void decode_cbor (SV *cborstr) 1664void decode_cbor (SV *cborstr)
1305 cbor_init (&cbor); 1668 cbor_init (&cbor);
1306 PUTBACK; cborstr = decode_cbor (cborstr, &cbor, 0); SPAGAIN; 1669 PUTBACK; cborstr = decode_cbor (cborstr, &cbor, 0); SPAGAIN;
1307 XPUSHs (cborstr); 1670 XPUSHs (cborstr);
1308} 1671}
1309 1672
1673#ifdef __AFL_COMPILER
1674
1675void
1676afl_init ()
1677 CODE:
1678 __AFL_INIT ();
1679
1680int
1681afl_loop (unsigned int count = 10000)
1682 CODE:
1683 RETVAL = __AFL_LOOP (count);
1684 OUTPUT:
1685 RETVAL
1686
1687#endif
1688

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines