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 1.51 by root, Sun Apr 24 13:15:19 2016 UTC vs.
Revision 1.68 by root, Mon Nov 30 18:30:29 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
12#define ECB_NO_THREADS 1 13#define ECB_NO_THREADS 1
13#include "ecb.h" 14#include "ecb.h"
14 15
15// compatibility with perl <5.18 16// compatibility with perl <5.18
… …
20# define HvNAMELEN(hv) HvNAMELEN_get (hv) 21# define HvNAMELEN(hv) HvNAMELEN_get (hv)
21#endif 22#endif
22#ifndef HvNAMEUTF8 23#ifndef HvNAMEUTF8
23# define HvNAMEUTF8(hv) 0 24# define HvNAMEUTF8(hv) 0
24#endif 25#endif
26#ifndef SvREFCNT_inc_NN
27# define SvREFCNT_inc_NN(sv) SvREFCNT_inc (sv)
28#endif
25#ifndef SvREFCNT_dec_NN 29#ifndef SvREFCNT_dec_NN
26# define SvREFCNT_dec_NN(sv) SvREFCNT_dec (sv) 30# define SvREFCNT_dec_NN(sv) SvREFCNT_dec (sv)
27#endif 31#endif
28 32
29// known major and minor types 33// known major and minor types
… …
95 CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8 99 CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8
96 100
97 CBOR_TAG_MAGIC = 55799, // self-describe cbor 101 CBOR_TAG_MAGIC = 55799, // self-describe cbor
98}; 102};
99 103
104// known forced types, also hardcoded in CBOR.pm
105enum
106{
107 AS_CBOR = 0,
108 AS_INT = 1,
109 AS_BYTES = 2,
110 AS_TEXT = 3,
111 AS_FLOAT16 = 4,
112 AS_FLOAT32 = 5,
113 AS_FLOAT64 = 6,
114 // possibly future enhancements: (generic) float, (generic) string
115};
116
100#define F_SHRINK 0x00000001UL 117#define F_SHRINK 0x00000001UL
101#define F_ALLOW_UNKNOWN 0x00000002UL 118#define F_ALLOW_UNKNOWN 0x00000002UL
102#define F_ALLOW_SHARING 0x00000004UL 119#define F_ALLOW_SHARING 0x00000004UL
103#define F_ALLOW_CYCLES 0x00000008UL 120#define F_ALLOW_CYCLES 0x00000008UL
121#define F_FORBID_OBJECTS 0x00000010UL
104#define F_PACK_STRINGS 0x00000010UL 122#define F_PACK_STRINGS 0x00000020UL
123#define F_TEXT_KEYS 0x00000040UL
105#define F_UTF8_STRINGS 0x00000020UL 124#define F_TEXT_STRINGS 0x00000080UL
106#define F_VALIDATE_UTF8 0x00000040UL 125#define F_VALIDATE_UTF8 0x00000100UL
107 126
108#define INIT_SIZE 32 // initial scalar size to be allocated 127#define INIT_SIZE 32 // initial scalar size to be allocated
109 128
110#define SB do { 129#define SB do {
111#define SE } while (0) 130#define SE } while (0)
… …
184 203
185// minimum length of a string to be registered for stringref 204// minimum length of a string to be registered for stringref
186ecb_inline int 205ecb_inline int
187minimum_string_length (UV idx) 206minimum_string_length (UV idx)
188{ 207{
189 return idx > 23 208 return idx <= 23 ? 3
190 ? idx > 0xffU 209 : idx <= 0xffU ? 4
191 ? idx > 0xffffU 210 : idx <= 0xffffU ? 5
192 ? idx > 0xffffffffU 211 : idx <= 0xffffffffU ? 7
193 ? 11 212 : 11;
194 : 7
195 : 5
196 : 4
197 : 3;
198} 213}
199 214
200///////////////////////////////////////////////////////////////////////////// 215/////////////////////////////////////////////////////////////////////////////
201// encoder 216// encoder
202 217
… …
215} enc_t; 230} enc_t;
216 231
217ecb_inline void 232ecb_inline void
218need (enc_t *enc, STRLEN len) 233need (enc_t *enc, STRLEN len)
219{ 234{
220 if (ecb_expect_false (enc->cur + len >= enc->end)) 235 if (ecb_expect_false ((uintptr_t)(enc->end - enc->cur) < len))
221 { 236 {
222 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv); 237 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
223 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1); 238 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
224 enc->cur = SvPVX (enc->sv) + cur; 239 enc->cur = SvPVX (enc->sv) + cur;
225 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; 240 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
… …
231{ 246{
232 need (enc, 1); 247 need (enc, 1);
233 *enc->cur++ = ch; 248 *enc->cur++ = ch;
234} 249}
235 250
251// used for tags, intregers, element counts and so on
236static void 252static void
237encode_uint (enc_t *enc, int major, UV len) 253encode_uint (enc_t *enc, int major, UV len)
238{ 254{
239 need (enc, 9); 255 need (enc, 9);
240 256
… …
271 *enc->cur++ = len >> 8; 287 *enc->cur++ = len >> 8;
272 *enc->cur++ = len; 288 *enc->cur++ = len;
273 } 289 }
274} 290}
275 291
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
276ecb_inline void 304ecb_inline void
277encode_tag (enc_t *enc, UV tag) 305encode_tag (enc_t *enc, UV tag)
278{ 306{
279 encode_uint (enc, MAJOR_TAG, tag); 307 encode_uint (enc, MAJOR_TAG, tag);
280} 308}
281 309
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
282ecb_inline void 333ecb_inline void
283encode_str (enc_t *enc, int utf8, char *str, STRLEN len) 334encode_str (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len)
284{ 335{
285 if (ecb_expect_false (enc->cbor.flags & F_UTF8_STRINGS)) 336 if (ecb_expect_false (upgrade_utf8))
286 if (!utf8) 337 if (!utf8)
287 { 338 {
288 SV *sv = sv_newmortal ();
289 char *s; STRLEN l;
290
291 sv_setpvn (sv, str, len);
292
293 s = SvPVutf8 (sv, l);
294 encode_str (enc, 1, s, l); 339 encode_str_utf8 (enc, utf8, str, len);
295 return; 340 return;
296 } 341 }
297 342
298 encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len); 343 encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len);
299 need (enc, len); 344 need (enc, len);
300 memcpy (enc->cur, str, len); 345 memcpy (enc->cur, str, len);
301 enc->cur += len; 346 enc->cur += len;
302} 347}
303 348
304static void 349ecb_inline void
305encode_strref (enc_t *enc, int utf8, char *str, STRLEN len) 350encode_strref (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len)
306{ 351{
307 if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS)) 352 if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS))
308 { 353 {
309 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1); 354 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1);
310 355
… …
321 sv_setuv (*svp, enc->stringref_idx); 366 sv_setuv (*svp, enc->stringref_idx);
322 ++enc->stringref_idx; 367 ++enc->stringref_idx;
323 } 368 }
324 } 369 }
325 370
326 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 }
327} 462}
328 463
329static void encode_sv (enc_t *enc, SV *sv); 464static void encode_sv (enc_t *enc, SV *sv);
330 465
331static void 466static void
… …
338 473
339 ++enc->depth; 474 ++enc->depth;
340 475
341 encode_uint (enc, MAJOR_ARRAY, len + 1); 476 encode_uint (enc, MAJOR_ARRAY, len + 1);
342 477
343 if (SvMAGICAL (av)) 478 if (ecb_expect_false (SvMAGICAL (av)))
344 for (i = 0; i <= len; ++i) 479 for (i = 0; i <= len; ++i)
345 { 480 {
346 SV **svp = av_fetch (av, i, 0); 481 SV **svp = av_fetch (av, i, 0);
347 encode_sv (enc, svp ? *svp : &PL_sv_undef); 482 encode_sv (enc, svp ? *svp : &PL_sv_undef);
348 } 483 }
… …
367 ++enc->depth; 502 ++enc->depth;
368 503
369 int pairs = hv_iterinit (hv); 504 int pairs = hv_iterinit (hv);
370 int mg = SvMAGICAL (hv); 505 int mg = SvMAGICAL (hv);
371 506
372 if (mg) 507 if (ecb_expect_false (mg))
373 encode_ch (enc, MAJOR_MAP | MINOR_INDEF); 508 encode_ch (enc, MAJOR_MAP | MINOR_INDEF);
374 else 509 else
375 encode_uint (enc, MAJOR_MAP, pairs); 510 encode_uint (enc, MAJOR_MAP, pairs);
376 511
377 while ((he = hv_iternext (hv))) 512 while ((he = hv_iternext (hv)))
378 { 513 {
379 if (HeKLEN (he) == HEf_SVKEY) 514 if (HeKLEN (he) == HEf_SVKEY)
380 encode_sv (enc, HeSVKEY (he)); 515 encode_sv (enc, HeSVKEY (he));
381 else 516 else
382 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));
383 518
384 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));
385 } 520 }
386 521
387 if (mg) 522 if (ecb_expect_false (mg))
388 encode_ch (enc, MAJOR_MISC | MINOR_INDEF); 523 encode_ch (enc, MAJOR_MISC | MINOR_INDEF);
389 524
390 --enc->depth; 525 --enc->depth;
391} 526}
392 527
… …
425 else if (stash == tagged_stash) 560 else if (stash == tagged_stash)
426 { 561 {
427 if (svt != SVt_PVAV) 562 if (svt != SVt_PVAV)
428 croak ("encountered CBOR::XS::Tagged object that isn't an array"); 563 croak ("encountered CBOR::XS::Tagged object that isn't an array");
429 564
565 switch (av_len ((AV *)sv))
566 {
567 case 2-1:
568 // actually a tagged value
430 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)));
431 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 }
432 581
433 return; 582 return;
434 } 583 }
435 } 584 }
436 585
437 if (ecb_expect_false (SvREFCNT (sv) > 1) 586 if (ecb_expect_false (SvREFCNT (sv) > 1)
438 && ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING)) 587 && ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING))
439 { 588 {
440 if (!enc->shareable) 589 if (ecb_expect_false (!enc->shareable))
441 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ()); 590 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
442 591
443 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1); 592 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
444 593
445 if (SvOK (*svp)) 594 if (SvOK (*svp))
… …
459 if (ecb_expect_false (SvOBJECT (sv))) 608 if (ecb_expect_false (SvOBJECT (sv)))
460 { 609 {
461 HV *stash = SvSTASH (sv); 610 HV *stash = SvSTASH (sv);
462 GV *method; 611 GV *method;
463 612
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))));
464 if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0))) 616 else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
465 { 617 {
466 dSP; 618 dSP;
467 619
468 ENTER; SAVETMPS; 620 ENTER; SAVETMPS;
469 PUSHMARK (SP); 621 PUSHMARK (SP);
… …
488 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0) 640 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)
489 { 641 {
490 dSP; 642 dSP;
491 643
492 ENTER; SAVETMPS; 644 ENTER; SAVETMPS;
493 SAVESTACK_POS ();
494 PUSHMARK (SP); 645 PUSHMARK (SP);
495 EXTEND (SP, 2); 646 EXTEND (SP, 2);
496 // 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
497 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); 648 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
498 PUSHs (sv_cbor); 649 PUSHs (sv_cbor);
… …
505 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv) 656 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)
506 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));
507 658
508 encode_tag (enc, CBOR_TAG_PERL_OBJECT); 659 encode_tag (enc, CBOR_TAG_PERL_OBJECT);
509 encode_uint (enc, MAJOR_ARRAY, count + 1); 660 encode_uint (enc, MAJOR_ARRAY, count + 1);
510 encode_strref (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash)); 661 encode_strref (enc, 0, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
511 662
512 while (count) 663 {
664 int i;
665
666 for (i = 0; i < count; ++i)
513 encode_sv (enc, SP[1 - count--]); 667 encode_sv (enc, SP[i + 1 - count]);
668
669 SP -= count;
670 }
514 671
515 PUTBACK; 672 PUTBACK;
516 673
517 FREETMPS; LEAVE; 674 FREETMPS; LEAVE;
518 } 675 }
… …
540 697
541 if (ecb_expect_false (nv == (NV)(U32)nv)) 698 if (ecb_expect_false (nv == (NV)(U32)nv))
542 encode_uint (enc, MAJOR_POS_INT, (U32)nv); 699 encode_uint (enc, MAJOR_POS_INT, (U32)nv);
543 //TODO: maybe I32? 700 //TODO: maybe I32?
544 else if (ecb_expect_false (nv == (float)nv)) 701 else if (ecb_expect_false (nv == (float)nv))
545 { 702 encode_float32 (enc, nv);
546 uint32_t fp = ecb_float_to_binary32 (nv);
547
548 *enc->cur++ = MAJOR_MISC | MISC_FLOAT32;
549
550 if (!ecb_big_endian ())
551 fp = ecb_bswap32 (fp);
552
553 memcpy (enc->cur, &fp, 4);
554 enc->cur += 4;
555 }
556 else 703 else
557 { 704 encode_float64 (enc, nv);
558 uint64_t fp = ecb_double_to_binary64 (nv);
559
560 *enc->cur++ = MAJOR_MISC | MISC_FLOAT64;
561
562 if (!ecb_big_endian ())
563 fp = ecb_bswap64 (fp);
564
565 memcpy (enc->cur, &fp, 8);
566 enc->cur += 8;
567 }
568} 705}
569 706
570static void 707static void
571encode_sv (enc_t *enc, SV *sv) 708encode_sv (enc_t *enc, SV *sv)
572{ 709{
… …
574 711
575 if (SvPOKp (sv)) 712 if (SvPOKp (sv))
576 { 713 {
577 STRLEN len; 714 STRLEN len;
578 char *str = SvPV (sv, len); 715 char *str = SvPV (sv, len);
579 encode_strref (enc, SvUTF8 (sv), str, len); 716 encode_strref (enc, enc->cbor.flags & F_TEXT_STRINGS, SvUTF8 (sv), str, len);
580 } 717 }
581 else if (SvNOKp (sv)) 718 else if (SvNOKp (sv))
582 encode_nv (enc, sv); 719 encode_nv (enc, sv);
583 else if (SvIOKp (sv)) 720 else if (SvIOKp (sv))
584 { 721 encode_int (enc, sv);
585 if (SvIsUV (sv))
586 encode_uint (enc, MAJOR_POS_INT, SvUVX (sv));
587 else if (SvIVX (sv) >= 0)
588 encode_uint (enc, MAJOR_POS_INT, SvIVX (sv));
589 else
590 encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1));
591 }
592 else if (SvROK (sv)) 722 else if (SvROK (sv))
593 encode_rv (enc, SvRV (sv)); 723 encode_rv (enc, SvRV (sv));
594 else if (!SvOK (sv)) 724 else if (!SvOK (sv))
595 encode_ch (enc, MAJOR_MISC | SIMPLE_NULL); 725 encode_ch (enc, MAJOR_MISC | SIMPLE_NULL);
596 else if (enc->cbor.flags & F_ALLOW_UNKNOWN) 726 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
… …
603static SV * 733static SV *
604encode_cbor (SV *scalar, CBOR *cbor) 734encode_cbor (SV *scalar, CBOR *cbor)
605{ 735{
606 enc_t enc = { 0 }; 736 enc_t enc = { 0 };
607 737
608 enc.cbor = *cbor; 738 enc.cbor = *cbor;
609 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 739 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
610 enc.cur = SvPVX (enc.sv); 740 enc.cur = SvPVX (enc.sv);
611 enc.end = SvEND (enc.sv); 741 enc.end = SvEND (enc.sv);
612 742
613 SvPOK_only (enc.sv); 743 SvPOK_only (enc.sv);
614 744
615 if (cbor->flags & F_PACK_STRINGS) 745 if (cbor->flags & F_PACK_STRINGS)
616 { 746 {
… …
643 U32 depth; // recursion depth 773 U32 depth; // recursion depth
644 U32 maxdepth; // recursion depth limit 774 U32 maxdepth; // recursion depth limit
645 AV *shareable; 775 AV *shareable;
646 AV *stringref; 776 AV *stringref;
647 SV *decode_tagged; 777 SV *decode_tagged;
778 SV *err_sv; // optional sv for error, needs to be freed
648} dec_t; 779} dec_t;
649 780
650#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);
651 788
652#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;
653 792
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
654#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)
655#define DEC_DEC_DEPTH --dec->depth 818#define DEC_DEC_DEPTH --dec->depth
656 819
657static UV 820static UV
658decode_uint (dec_t *dec) 821decode_uint (dec_t *dec)
659{ 822{
… …
725 888
726 for (;;) 889 for (;;)
727 { 890 {
728 WANT (1); 891 WANT (1);
729 892
730 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF)) 893 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF) || dec->err)
731 { 894 {
732 ++dec->cur; 895 ++dec->cur;
733 break; 896 break;
734 } 897 }
735 898
736 av_push (av, decode_sv (dec)); 899 av_push (av, decode_sv (dec));
737 } 900 }
738 } 901 }
739 else 902 else
740 { 903 {
741 int i, len = decode_uint (dec); 904 UV i, len = decode_uint (dec);
742 905
743 WANT (len); // complexity check for av_fill - need at least one byte per value, do not allow supersize arrays 906 WANT (len); // complexity check for av_fill - need at least one byte per value, do not allow supersize arrays
744 av_fill (av, len - 1); 907 av_fill (av, len - 1);
745 908
746 for (i = 0; i < len; ++i) 909 for (i = 0; i < len; ++i)
… …
749 912
750 DEC_DEC_DEPTH; 913 DEC_DEC_DEPTH;
751 return newRV_noinc ((SV *)av); 914 return newRV_noinc ((SV *)av);
752 915
753fail: 916fail:
754 SvREFCNT_dec (av); 917 SvREFCNT_dec_NN (av);
755 DEC_DEC_DEPTH; 918 DEC_DEC_DEPTH;
756 return &PL_sv_undef; 919 return &PL_sv_undef;
757} 920}
758 921
759static void 922static void
… …
763 // byte or utf-8 strings as keys, but only when !stringref 926 // byte or utf-8 strings as keys, but only when !stringref
764 927
765 if (ecb_expect_true (!dec->stringref)) 928 if (ecb_expect_true (!dec->stringref))
766 if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8)) 929 if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8))
767 { 930 {
768 I32 len = decode_uint (dec); 931 STRLEN len = decode_uint (dec);
769 char *key = (char *)dec->cur; 932 char *key = (char *)dec->cur;
770 933
771 WANT (len); 934 WANT (len);
772 dec->cur += len; 935 dec->cur += len;
773 936
… …
775 938
776 return; 939 return;
777 } 940 }
778 else if (ecb_expect_true ((U8)(*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8)) 941 else if (ecb_expect_true ((U8)(*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8))
779 { 942 {
780 I32 len = decode_uint (dec); 943 STRLEN len = decode_uint (dec);
781 char *key = (char *)dec->cur; 944 char *key = (char *)dec->cur;
782 945
783 WANT (len); 946 WANT (len);
784 dec->cur += len; 947 dec->cur += len;
785 948
… …
793 } 956 }
794 957
795 SV *k = decode_sv (dec); 958 SV *k = decode_sv (dec);
796 SV *v = decode_sv (dec); 959 SV *v = decode_sv (dec);
797 960
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
798 hv_store_ent (hv, k, v, 0); 989 hv_store_ent (hv, k, v, 0);
799 SvREFCNT_dec (k); 990 SvREFCNT_dec_NN (k);
800 991
801fail: 992fail:
802 ; 993 ;
803} 994}
804 995
… …
815 1006
816 for (;;) 1007 for (;;)
817 { 1008 {
818 WANT (1); 1009 WANT (1);
819 1010
820 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF)) 1011 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF) || dec->err)
821 { 1012 {
822 ++dec->cur; 1013 ++dec->cur;
823 break; 1014 break;
824 } 1015 }
825 1016
826 decode_he (dec, hv); 1017 decode_he (dec, hv);
827 } 1018 }
828 } 1019 }
829 else 1020 else
830 { 1021 {
831 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
832 1025
833 while (pairs--) 1026 while (pairs--)
834 decode_he (dec, hv); 1027 decode_he (dec, hv);
835 } 1028 }
836 1029
837 DEC_DEC_DEPTH; 1030 DEC_DEC_DEPTH;
838 return newRV_noinc ((SV *)hv); 1031 return newRV_noinc ((SV *)hv);
839 1032
840fail: 1033fail:
841 SvREFCNT_dec (hv); 1034 SvREFCNT_dec_NN (hv);
842 DEC_DEC_DEPTH; 1035 DEC_DEC_DEPTH;
843 return &PL_sv_undef; 1036 return &PL_sv_undef;
844} 1037}
845 1038
846static SV * 1039static SV *
847decode_str (dec_t *dec, int utf8) 1040decode_str (dec_t *dec, int utf8)
848{ 1041{
849 SV *sv = 0; 1042 SV *sv = 0;
850 1043
851 if ((*dec->cur & MINOR_MASK) == MINOR_INDEF) 1044 if (ecb_expect_false ((*dec->cur & MINOR_MASK) == MINOR_INDEF))
852 { 1045 {
853 // indefinite length strings 1046 // indefinite length strings
854 ++dec->cur; 1047 ++dec->cur;
855 1048
856 U8 major = *dec->cur & MAJOR_MISC; 1049 U8 major = *dec->cur & MAJOR_MISC;
… …
924 sv = newRV_noinc (decode_sv (dec)); 1117 sv = newRV_noinc (decode_sv (dec));
925 break; 1118 break;
926 1119
927 case CBOR_TAG_STRINGREF_NAMESPACE: 1120 case CBOR_TAG_STRINGREF_NAMESPACE:
928 { 1121 {
1122 // do not use SAVETMPS/FREETMPS, as these will
1123 // erase mortalised caches, e.g. "shareable"
929 ENTER; SAVETMPS; 1124 ENTER;
930 1125
931 SAVESPTR (dec->stringref); 1126 SAVESPTR (dec->stringref);
932 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ()); 1127 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ());
933 1128
934 sv = decode_sv (dec); 1129 sv = decode_sv (dec);
935 1130
936 FREETMPS; LEAVE; 1131 LEAVE;
937 } 1132 }
938 break; 1133 break;
939 1134
940 case CBOR_TAG_STRINGREF: 1135 case CBOR_TAG_STRINGREF:
941 { 1136 {
942 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT)) 1137 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
943 ERR ("corrupted CBOR data (stringref index not an unsigned integer)"); 1138 ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
944 1139
945 UV idx = decode_uint (dec); 1140 UV idx = decode_uint (dec);
946 1141
947 if (!dec->stringref || (int)idx > AvFILLp (dec->stringref)) 1142 if (!dec->stringref || idx >= (UV)(1 + AvFILLp (dec->stringref)))
948 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)");
949 1144
950 sv = newSVsv (AvARRAY (dec->stringref)[idx]); 1145 sv = newSVsv (AvARRAY (dec->stringref)[idx]);
951 } 1146 }
952 break; 1147 break;
… …
980 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT)) 1175 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
981 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)"); 1176 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
982 1177
983 UV idx = decode_uint (dec); 1178 UV idx = decode_uint (dec);
984 1179
985 if (!dec->shareable || (int)idx > AvFILLp (dec->shareable)) 1180 if (!dec->shareable || idx >= (UV)(1 + AvFILLp (dec->shareable)))
986 ERR ("corrupted CBOR data (sharedref index out of bounds)"); 1181 ERR ("corrupted CBOR data (sharedref index out of bounds)");
987 1182
988 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]); 1183 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]);
989 1184
990 if (sv == &PL_sv_undef) 1185 if (sv == &PL_sv_undef)
… …
992 } 1187 }
993 break; 1188 break;
994 1189
995 case CBOR_TAG_PERL_OBJECT: 1190 case CBOR_TAG_PERL_OBJECT:
996 { 1191 {
1192 if (dec->cbor.flags & F_FORBID_OBJECTS)
1193 goto filter;
1194
997 sv = decode_sv (dec); 1195 sv = decode_sv (dec);
998 1196
999 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV) 1197 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
1000 ERR ("corrupted CBOR data (non-array perl object)"); 1198 ERR ("corrupted CBOR data (non-array perl object)");
1001 1199
… …
1030 SPAGAIN; 1228 SPAGAIN;
1031 1229
1032 if (SvTRUE (ERRSV)) 1230 if (SvTRUE (ERRSV))
1033 { 1231 {
1034 FREETMPS; LEAVE; 1232 FREETMPS; LEAVE;
1035 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV)))); 1233 ERR_ERRSV;
1036 } 1234 }
1037 1235
1038 SvREFCNT_dec (sv); 1236 SvREFCNT_dec_NN (sv);
1039 sv = SvREFCNT_inc (POPs); 1237 sv = SvREFCNT_inc (POPs);
1040 1238
1041 PUTBACK; 1239 PUTBACK;
1042 1240
1043 FREETMPS; LEAVE; 1241 FREETMPS; LEAVE;
1044 } 1242 }
1045 break; 1243 break;
1046 1244
1047 default: 1245 default:
1246 filter:
1048 { 1247 {
1248 SV *tag_sv = newSVuv (tag);
1249
1049 sv = decode_sv (dec); 1250 sv = decode_sv (dec);
1050 1251
1051 dSP; 1252 dSP;
1052 ENTER; SAVETMPS; 1253 ENTER; SAVETMPS;
1053 SAVESTACK_POS ();
1054 PUSHMARK (SP); 1254 PUSHMARK (SP);
1055 EXTEND (SP, 2); 1255 EXTEND (SP, 2);
1056 PUSHs (newSVuv (tag)); 1256 PUSHs (tag_sv);
1057 PUSHs (sv); 1257 PUSHs (sv);
1058 1258
1059 PUTBACK; 1259 PUTBACK;
1060 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);
1061 SPAGAIN; 1261 SPAGAIN;
1062 1262
1063 if (SvTRUE (ERRSV)) 1263 if (SvTRUE (ERRSV))
1064 { 1264 {
1265 SvREFCNT_dec_NN (tag_sv);
1065 FREETMPS; LEAVE; 1266 FREETMPS; LEAVE;
1066 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV)))); 1267 ERR_ERRSV;
1067 } 1268 }
1068 1269
1069 if (count) 1270 if (count)
1070 { 1271 {
1272 SvREFCNT_dec_NN (tag_sv);
1071 SvREFCNT_dec (sv); 1273 SvREFCNT_dec_NN (sv);
1072 sv = SvREFCNT_inc (POPs); 1274 sv = SvREFCNT_inc_NN (TOPs);
1275 SP -= count;
1073 } 1276 }
1074 else 1277 else
1075 { 1278 {
1076 AV *av = newAV (); 1279 AV *av = newAV ();
1077 av_push (av, newSVuv (tag)); 1280 av_push (av, tag_sv);
1078 av_push (av, sv); 1281 av_push (av, sv);
1079 1282
1080 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 1283 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
1081 ? cbor_tagged_stash 1284 ? cbor_tagged_stash
1082 : gv_stashpv ("CBOR::XS::Tagged" , 1); 1285 : gv_stashpv ("CBOR::XS::Tagged" , 1);
… …
1211 1414
1212 if (dec.err) 1415 if (dec.err)
1213 { 1416 {
1214 if (dec.shareable) 1417 if (dec.shareable)
1215 { 1418 {
1216 // need to break cyclic links, which whould all be in shareable 1419 // need to break cyclic links, which would all be in shareable
1217 int i; 1420 int i;
1218 SV **svp; 1421 SV **svp;
1219 1422
1220 for (i = av_len (dec.shareable) + 1; i--; ) 1423 for (i = av_len (dec.shareable) + 1; i--; )
1221 if ((svp = av_fetch (dec.shareable, i, 0))) 1424 if ((svp = av_fetch (dec.shareable, i, 0)))
1222 sv_setsv (*svp, &PL_sv_undef); 1425 sv_setsv (*svp, &PL_sv_undef);
1223 } 1426 }
1224 1427
1225 SvREFCNT_dec (sv); 1428 SvREFCNT_dec_NN (sv);
1429
1430 if (dec.err_sv)
1431 sv_2mortal (dec.err_sv);
1432
1226 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);
1227 } 1434 }
1228 1435
1229 sv = sv_2mortal (sv); 1436 sv = sv_2mortal (sv);
1230 1437
… …
1374 1581
1375 default_filter = newSVpv ("CBOR::XS::default_filter", 0); 1582 default_filter = newSVpv ("CBOR::XS::default_filter", 0);
1376 1583
1377 sv_cbor = newSVpv ("CBOR", 0); 1584 sv_cbor = newSVpv ("CBOR", 0);
1378 SvREADONLY_on (sv_cbor); 1585 SvREADONLY_on (sv_cbor);
1586
1587 assert (("STRLEN must be an unsigned type", 0 <= (STRLEN)-1));
1379} 1588}
1380 1589
1381PROTOTYPES: DISABLE 1590PROTOTYPES: DISABLE
1382 1591
1383void CLONE (...) 1592void CLONE (...)
… …
1403 ALIAS: 1612 ALIAS:
1404 shrink = F_SHRINK 1613 shrink = F_SHRINK
1405 allow_unknown = F_ALLOW_UNKNOWN 1614 allow_unknown = F_ALLOW_UNKNOWN
1406 allow_sharing = F_ALLOW_SHARING 1615 allow_sharing = F_ALLOW_SHARING
1407 allow_cycles = F_ALLOW_CYCLES 1616 allow_cycles = F_ALLOW_CYCLES
1617 forbid_objects = F_FORBID_OBJECTS
1408 pack_strings = F_PACK_STRINGS 1618 pack_strings = F_PACK_STRINGS
1619 text_keys = F_TEXT_KEYS
1409 utf8_strings = F_UTF8_STRINGS 1620 text_strings = F_TEXT_STRINGS
1410 validate_utf8 = F_VALIDATE_UTF8 1621 validate_utf8 = F_VALIDATE_UTF8
1411 PPCODE: 1622 PPCODE:
1412{ 1623{
1413 if (enable) 1624 if (enable)
1414 self->flags |= ix; 1625 self->flags |= ix;
… …
1422 ALIAS: 1633 ALIAS:
1423 get_shrink = F_SHRINK 1634 get_shrink = F_SHRINK
1424 get_allow_unknown = F_ALLOW_UNKNOWN 1635 get_allow_unknown = F_ALLOW_UNKNOWN
1425 get_allow_sharing = F_ALLOW_SHARING 1636 get_allow_sharing = F_ALLOW_SHARING
1426 get_allow_cycles = F_ALLOW_CYCLES 1637 get_allow_cycles = F_ALLOW_CYCLES
1638 get_forbid_objects = F_FORBID_OBJECTS
1427 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
1428 get_validate_utf8 = F_VALIDATE_UTF8 1642 get_validate_utf8 = F_VALIDATE_UTF8
1429 PPCODE: 1643 PPCODE:
1430 XPUSHs (boolSV (self->flags & ix)); 1644 XPUSHs (boolSV (self->flags & ix));
1431 1645
1432void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) 1646void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
… …
1562 cbor_init (&cbor); 1776 cbor_init (&cbor);
1563 PUTBACK; cborstr = decode_cbor (cborstr, &cbor, 0); SPAGAIN; 1777 PUTBACK; cborstr = decode_cbor (cborstr, &cbor, 0); SPAGAIN;
1564 XPUSHs (cborstr); 1778 XPUSHs (cborstr);
1565} 1779}
1566 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