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 1.49 by root, Mon Feb 8 04:11:11 2016 UTC vs.
Revision 1.69 by root, Mon Nov 30 20:38:25 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
124#define F_TEXT_STRINGS 0x00000080UL
105#define F_VALIDATE_UTF8 0x00000020UL 125#define F_VALIDATE_UTF8 0x00000100UL
106 126
107#define INIT_SIZE 32 // initial scalar size to be allocated 127#define INIT_SIZE 32 // initial scalar size to be allocated
108 128
109#define SB do { 129#define SB do {
110#define SE } while (0) 130#define SE } while (0)
… …
183 203
184// minimum length of a string to be registered for stringref 204// minimum length of a string to be registered for stringref
185ecb_inline int 205ecb_inline int
186minimum_string_length (UV idx) 206minimum_string_length (UV idx)
187{ 207{
188 return idx > 23 208 return idx <= 23 ? 3
189 ? idx > 0xffU 209 : idx <= 0xffU ? 4
190 ? idx > 0xffffU 210 : idx <= 0xffffU ? 5
191 ? idx > 0xffffffffU 211 : idx <= 0xffffffffU ? 7
192 ? 11 212 : 11;
193 : 7
194 : 5
195 : 4
196 : 3;
197} 213}
198 214
199///////////////////////////////////////////////////////////////////////////// 215/////////////////////////////////////////////////////////////////////////////
200// encoder 216// encoder
201 217
… …
214} enc_t; 230} enc_t;
215 231
216ecb_inline void 232ecb_inline void
217need (enc_t *enc, STRLEN len) 233need (enc_t *enc, STRLEN len)
218{ 234{
219 if (ecb_expect_false (enc->cur + len >= enc->end)) 235 if (ecb_expect_false ((uintptr_t)(enc->end - enc->cur) < len))
220 { 236 {
221 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv); 237 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
222 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1); 238 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
223 enc->cur = SvPVX (enc->sv) + cur; 239 enc->cur = SvPVX (enc->sv) + cur;
224 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; 240 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
… …
230{ 246{
231 need (enc, 1); 247 need (enc, 1);
232 *enc->cur++ = ch; 248 *enc->cur++ = ch;
233} 249}
234 250
251// used for tags, intregers, element counts and so on
235static void 252static void
236encode_uint (enc_t *enc, int major, UV len) 253encode_uint (enc_t *enc, int major, UV len)
237{ 254{
238 need (enc, 9); 255 need (enc, 9);
239 256
… …
270 *enc->cur++ = len >> 8; 287 *enc->cur++ = len >> 8;
271 *enc->cur++ = len; 288 *enc->cur++ = len;
272 } 289 }
273} 290}
274 291
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
275ecb_inline void 304ecb_inline void
276encode_tag (enc_t *enc, UV tag) 305encode_tag (enc_t *enc, UV tag)
277{ 306{
278 encode_uint (enc, MAJOR_TAG, tag); 307 encode_uint (enc, MAJOR_TAG, tag);
279} 308}
280 309
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
281ecb_inline void 333ecb_inline void
282encode_str (enc_t *enc, int utf8, char *str, STRLEN len) 334encode_str (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len)
283{ 335{
336 if (ecb_expect_false (upgrade_utf8))
337 if (!utf8)
338 {
339 encode_str_utf8 (enc, utf8, str, len);
340 return;
341 }
342
284 encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len); 343 encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len);
285 need (enc, len); 344 need (enc, len);
286 memcpy (enc->cur, str, len); 345 memcpy (enc->cur, str, len);
287 enc->cur += len; 346 enc->cur += len;
288} 347}
289 348
290static void 349ecb_inline void
291encode_strref (enc_t *enc, int utf8, char *str, STRLEN len) 350encode_strref (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len)
292{ 351{
293 if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS)) 352 if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS))
294 { 353 {
295 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1); 354 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1);
296 355
… …
307 sv_setuv (*svp, enc->stringref_idx); 366 sv_setuv (*svp, enc->stringref_idx);
308 ++enc->stringref_idx; 367 ++enc->stringref_idx;
309 } 368 }
310 } 369 }
311 370
312 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_bool (enc_t *enc, int istrue)
424{
425 encode_ch (enc, istrue ? MAJOR_MISC | SIMPLE_TRUE : MAJOR_MISC | SIMPLE_FALSE);
426}
427
428ecb_inline void
429encode_forced (enc_t *enc, UV type, SV *sv)
430{
431 switch (type)
432 {
433 case AS_CBOR:
434 {
435 STRLEN len;
436 char *str = SvPVbyte (sv, len);
437
438 need (enc, len);
439 memcpy (enc->cur, str, len);
440 enc->cur += len;
441 }
442 break;
443
444 case AS_BYTES:
445 {
446 STRLEN len;
447 char *str = SvPVbyte (sv, len);
448 encode_strref (enc, 0, 0, str, len);
449 }
450 break;
451
452 case AS_TEXT:
453 {
454 STRLEN len;
455 char *str = SvPVutf8 (sv, len);
456 encode_strref (enc, 1, 1, str, len);
457 }
458 break;
459
460 case AS_INT: encode_int (enc, sv); break;
461
462 case AS_FLOAT16: encode_float16 (enc, SvNV (sv)); break;
463 case AS_FLOAT32: encode_float32 (enc, SvNV (sv)); break;
464 case AS_FLOAT64: encode_float64 (enc, SvNV (sv)); break;
465
466 default:
467 croak ("encountered malformed CBOR::XS::Tagged object");
468 }
313} 469}
314 470
315static void encode_sv (enc_t *enc, SV *sv); 471static void encode_sv (enc_t *enc, SV *sv);
316 472
317static void 473static void
… …
324 480
325 ++enc->depth; 481 ++enc->depth;
326 482
327 encode_uint (enc, MAJOR_ARRAY, len + 1); 483 encode_uint (enc, MAJOR_ARRAY, len + 1);
328 484
329 if (SvMAGICAL (av)) 485 if (ecb_expect_false (SvMAGICAL (av)))
330 for (i = 0; i <= len; ++i) 486 for (i = 0; i <= len; ++i)
331 { 487 {
332 SV **svp = av_fetch (av, i, 0); 488 SV **svp = av_fetch (av, i, 0);
333 encode_sv (enc, svp ? *svp : &PL_sv_undef); 489 encode_sv (enc, svp ? *svp : &PL_sv_undef);
334 } 490 }
… …
353 ++enc->depth; 509 ++enc->depth;
354 510
355 int pairs = hv_iterinit (hv); 511 int pairs = hv_iterinit (hv);
356 int mg = SvMAGICAL (hv); 512 int mg = SvMAGICAL (hv);
357 513
358 if (mg) 514 if (ecb_expect_false (mg))
359 encode_ch (enc, MAJOR_MAP | MINOR_INDEF); 515 encode_ch (enc, MAJOR_MAP | MINOR_INDEF);
360 else 516 else
361 encode_uint (enc, MAJOR_MAP, pairs); 517 encode_uint (enc, MAJOR_MAP, pairs);
362 518
363 while ((he = hv_iternext (hv))) 519 while ((he = hv_iternext (hv)))
364 { 520 {
365 if (HeKLEN (he) == HEf_SVKEY) 521 if (HeKLEN (he) == HEf_SVKEY)
366 encode_sv (enc, HeSVKEY (he)); 522 encode_sv (enc, HeSVKEY (he));
367 else 523 else
368 encode_strref (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he)); 524 encode_strref (enc, enc->cbor.flags & (F_TEXT_KEYS | F_TEXT_STRINGS), HeKUTF8 (he), HeKEY (he), HeKLEN (he));
369 525
370 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he)); 526 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he));
371 } 527 }
372 528
373 if (mg) 529 if (ecb_expect_false (mg))
374 encode_ch (enc, MAJOR_MISC | MINOR_INDEF); 530 encode_ch (enc, MAJOR_MISC | MINOR_INDEF);
375 531
376 --enc->depth; 532 --enc->depth;
377} 533}
378 534
… …
398 554
399 HV *stash = SvSTASH (sv); 555 HV *stash = SvSTASH (sv);
400 556
401 if (stash == boolean_stash) 557 if (stash == boolean_stash)
402 { 558 {
403 encode_ch (enc, SvIV (sv) ? MAJOR_MISC | SIMPLE_TRUE : MAJOR_MISC | SIMPLE_FALSE); 559 encode_bool (enc, SvIV (sv));
404 return; 560 return;
405 } 561 }
406 else if (stash == error_stash) 562 else if (stash == error_stash)
407 { 563 {
408 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF); 564 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
… …
411 else if (stash == tagged_stash) 567 else if (stash == tagged_stash)
412 { 568 {
413 if (svt != SVt_PVAV) 569 if (svt != SVt_PVAV)
414 croak ("encountered CBOR::XS::Tagged object that isn't an array"); 570 croak ("encountered CBOR::XS::Tagged object that isn't an array");
415 571
572 switch (av_len ((AV *)sv))
573 {
574 case 2-1:
575 // actually a tagged value
416 encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1))); 576 encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1)));
417 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1)); 577 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1));
578 break;
579
580 case 3-1:
581 // a forced type [value, type, undef]
582 encode_forced (enc, SvUV (*av_fetch ((AV *)sv, 1, 1)), *av_fetch ((AV *)sv, 0, 1));
583 break;
584
585 default:
586 croak ("encountered malformed CBOR::XS::Tagged object");
587 }
418 588
419 return; 589 return;
420 } 590 }
421 } 591 }
422 592
423 if (ecb_expect_false (SvREFCNT (sv) > 1) 593 if (ecb_expect_false (SvREFCNT (sv) > 1)
424 && ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING)) 594 && ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING))
425 { 595 {
426 if (!enc->shareable) 596 if (ecb_expect_false (!enc->shareable))
427 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ()); 597 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
428 598
429 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1); 599 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
430 600
431 if (SvOK (*svp)) 601 if (SvOK (*svp))
… …
445 if (ecb_expect_false (SvOBJECT (sv))) 615 if (ecb_expect_false (SvOBJECT (sv)))
446 { 616 {
447 HV *stash = SvSTASH (sv); 617 HV *stash = SvSTASH (sv);
448 GV *method; 618 GV *method;
449 619
620 if (enc->cbor.flags & F_FORBID_OBJECTS)
621 croak ("encountered object '%s', but forbid_objects is enabled",
622 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
450 if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0))) 623 else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
451 { 624 {
452 dSP; 625 dSP;
453 626
454 ENTER; SAVETMPS; PUSHMARK (SP); 627 ENTER; SAVETMPS;
628 PUSHMARK (SP);
455 // we re-bless the reference to get overload and other niceties right 629 // we re-bless the reference to get overload and other niceties right
456 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); 630 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
457 631
458 PUTBACK; 632 PUTBACK;
459 // G_SCALAR ensures that return value is 1 633 // G_SCALAR ensures that return value is 1
… …
472 } 646 }
473 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0) 647 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)
474 { 648 {
475 dSP; 649 dSP;
476 650
477 ENTER; SAVETMPS; PUSHMARK (SP); 651 ENTER; SAVETMPS;
652 PUSHMARK (SP);
478 EXTEND (SP, 2); 653 EXTEND (SP, 2);
479 // we re-bless the reference to get overload and other niceties right 654 // we re-bless the reference to get overload and other niceties right
480 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); 655 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
481 PUSHs (sv_cbor); 656 PUSHs (sv_cbor);
482 657
… …
488 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv) 663 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)
489 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash)); 664 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));
490 665
491 encode_tag (enc, CBOR_TAG_PERL_OBJECT); 666 encode_tag (enc, CBOR_TAG_PERL_OBJECT);
492 encode_uint (enc, MAJOR_ARRAY, count + 1); 667 encode_uint (enc, MAJOR_ARRAY, count + 1);
493 encode_strref (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash)); 668 encode_strref (enc, 0, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
494 669
495 while (count) 670 {
671 int i;
672
673 for (i = 0; i < count; ++i)
496 encode_sv (enc, SP[1 - count--]); 674 encode_sv (enc, SP[i + 1 - count]);
675
676 SP -= count;
677 }
497 678
498 PUTBACK; 679 PUTBACK;
499 680
500 FREETMPS; LEAVE; 681 FREETMPS; LEAVE;
501 } 682 }
… …
523 704
524 if (ecb_expect_false (nv == (NV)(U32)nv)) 705 if (ecb_expect_false (nv == (NV)(U32)nv))
525 encode_uint (enc, MAJOR_POS_INT, (U32)nv); 706 encode_uint (enc, MAJOR_POS_INT, (U32)nv);
526 //TODO: maybe I32? 707 //TODO: maybe I32?
527 else if (ecb_expect_false (nv == (float)nv)) 708 else if (ecb_expect_false (nv == (float)nv))
528 { 709 encode_float32 (enc, nv);
529 uint32_t fp = ecb_float_to_binary32 (nv);
530
531 *enc->cur++ = MAJOR_MISC | MISC_FLOAT32;
532
533 if (!ecb_big_endian ())
534 fp = ecb_bswap32 (fp);
535
536 memcpy (enc->cur, &fp, 4);
537 enc->cur += 4;
538 }
539 else 710 else
540 { 711 encode_float64 (enc, nv);
541 uint64_t fp = ecb_double_to_binary64 (nv);
542
543 *enc->cur++ = MAJOR_MISC | MISC_FLOAT64;
544
545 if (!ecb_big_endian ())
546 fp = ecb_bswap64 (fp);
547
548 memcpy (enc->cur, &fp, 8);
549 enc->cur += 8;
550 }
551} 712}
552 713
553static void 714static void
554encode_sv (enc_t *enc, SV *sv) 715encode_sv (enc_t *enc, SV *sv)
555{ 716{
… …
557 718
558 if (SvPOKp (sv)) 719 if (SvPOKp (sv))
559 { 720 {
560 STRLEN len; 721 STRLEN len;
561 char *str = SvPV (sv, len); 722 char *str = SvPV (sv, len);
562 encode_strref (enc, SvUTF8 (sv), str, len); 723 encode_strref (enc, enc->cbor.flags & F_TEXT_STRINGS, SvUTF8 (sv), str, len);
563 } 724 }
564 else if (SvNOKp (sv)) 725 else if (SvNOKp (sv))
565 encode_nv (enc, sv); 726 encode_nv (enc, sv);
566 else if (SvIOKp (sv)) 727 else if (SvIOKp (sv))
567 { 728 encode_int (enc, sv);
568 if (SvIsUV (sv))
569 encode_uint (enc, MAJOR_POS_INT, SvUVX (sv));
570 else if (SvIVX (sv) >= 0)
571 encode_uint (enc, MAJOR_POS_INT, SvIVX (sv));
572 else
573 encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1));
574 }
575 else if (SvROK (sv)) 729 else if (SvROK (sv))
576 encode_rv (enc, SvRV (sv)); 730 encode_rv (enc, SvRV (sv));
577 else if (!SvOK (sv)) 731 else if (!SvOK (sv))
578 encode_ch (enc, MAJOR_MISC | SIMPLE_NULL); 732 encode_ch (enc, MAJOR_MISC | SIMPLE_NULL);
579 else if (enc->cbor.flags & F_ALLOW_UNKNOWN) 733 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
… …
586static SV * 740static SV *
587encode_cbor (SV *scalar, CBOR *cbor) 741encode_cbor (SV *scalar, CBOR *cbor)
588{ 742{
589 enc_t enc = { 0 }; 743 enc_t enc = { 0 };
590 744
591 enc.cbor = *cbor; 745 enc.cbor = *cbor;
592 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 746 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
593 enc.cur = SvPVX (enc.sv); 747 enc.cur = SvPVX (enc.sv);
594 enc.end = SvEND (enc.sv); 748 enc.end = SvEND (enc.sv);
595 749
596 SvPOK_only (enc.sv); 750 SvPOK_only (enc.sv);
597 751
598 if (cbor->flags & F_PACK_STRINGS) 752 if (cbor->flags & F_PACK_STRINGS)
599 { 753 {
… …
626 U32 depth; // recursion depth 780 U32 depth; // recursion depth
627 U32 maxdepth; // recursion depth limit 781 U32 maxdepth; // recursion depth limit
628 AV *shareable; 782 AV *shareable;
629 AV *stringref; 783 AV *stringref;
630 SV *decode_tagged; 784 SV *decode_tagged;
785 SV *err_sv; // optional sv for error, needs to be freed
631} dec_t; 786} dec_t;
632 787
633#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE 788// set dec->err to ERRSV
789ecb_cold static void
790err_errsv (dec_t *dec)
791{
792 if (!dec->err)
793 {
794 dec->err_sv = newSVsv (ERRSV);
634 795
635#define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data") 796 // chop off the trailing \n
797 SvCUR_set (dec->err_sv, SvCUR (dec->err_sv) - 1);
798 *SvEND (dec->err_sv) = 0;
636 799
800 dec->err = SvPVutf8_nolen (dec->err_sv);
801 }
802}
803
804// the following functions are used to reduce code size and help the compiler to optimise
805ecb_cold static void
806err_set (dec_t *dec, const char *reason)
807{
808 if (!dec->err)
809 dec->err = reason;
810}
811
812ecb_cold static void
813err_unexpected_end (dec_t *dec)
814{
815 err_set (dec, "unexpected end of CBOR data");
816}
817
818#define ERR_DO(do) SB do; goto fail; SE
819#define ERR(reason) ERR_DO (err_set (dec, reason))
820#define ERR_ERRSV ERR_DO (err_errsv (dec))
821
822#define WANT(len) if (ecb_expect_false ((uintptr_t)(dec->end - dec->cur) < (STRLEN)len)) ERR_DO (err_unexpected_end (dec))
823
637#define DEC_INC_DEPTH if (++dec->depth > dec->cbor.max_depth) ERR (ERR_NESTING_EXCEEDED) 824#define DEC_INC_DEPTH if (ecb_expect_false (++dec->depth > dec->cbor.max_depth)) ERR (ERR_NESTING_EXCEEDED)
638#define DEC_DEC_DEPTH --dec->depth 825#define DEC_DEC_DEPTH --dec->depth
639 826
640static UV 827static UV
641decode_uint (dec_t *dec) 828decode_uint (dec_t *dec)
642{ 829{
… …
708 895
709 for (;;) 896 for (;;)
710 { 897 {
711 WANT (1); 898 WANT (1);
712 899
713 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF)) 900 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF) || dec->err)
714 { 901 {
715 ++dec->cur; 902 ++dec->cur;
716 break; 903 break;
717 } 904 }
718 905
719 av_push (av, decode_sv (dec)); 906 av_push (av, decode_sv (dec));
720 } 907 }
721 } 908 }
722 else 909 else
723 { 910 {
724 int i, len = decode_uint (dec); 911 UV i, len = decode_uint (dec);
725 912
726 WANT (len); // complexity check for av_fill - need at least one byte per value, do not allow supersize arrays 913 WANT (len); // complexity check for av_fill - need at least one byte per value, do not allow supersize arrays
727 av_fill (av, len - 1); 914 av_fill (av, len - 1);
728 915
729 for (i = 0; i < len; ++i) 916 for (i = 0; i < len; ++i)
… …
732 919
733 DEC_DEC_DEPTH; 920 DEC_DEC_DEPTH;
734 return newRV_noinc ((SV *)av); 921 return newRV_noinc ((SV *)av);
735 922
736fail: 923fail:
737 SvREFCNT_dec (av); 924 SvREFCNT_dec_NN (av);
738 DEC_DEC_DEPTH; 925 DEC_DEC_DEPTH;
739 return &PL_sv_undef; 926 return &PL_sv_undef;
740} 927}
741 928
742static void 929static void
… …
746 // byte or utf-8 strings as keys, but only when !stringref 933 // byte or utf-8 strings as keys, but only when !stringref
747 934
748 if (ecb_expect_true (!dec->stringref)) 935 if (ecb_expect_true (!dec->stringref))
749 if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8)) 936 if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8))
750 { 937 {
751 I32 len = decode_uint (dec); 938 STRLEN len = decode_uint (dec);
752 char *key = (char *)dec->cur; 939 char *key = (char *)dec->cur;
753 940
754 WANT (len); 941 WANT (len);
755 dec->cur += len; 942 dec->cur += len;
756 943
… …
758 945
759 return; 946 return;
760 } 947 }
761 else if (ecb_expect_true ((U8)(*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8)) 948 else if (ecb_expect_true ((U8)(*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8))
762 { 949 {
763 I32 len = decode_uint (dec); 950 STRLEN len = decode_uint (dec);
764 char *key = (char *)dec->cur; 951 char *key = (char *)dec->cur;
765 952
766 WANT (len); 953 WANT (len);
767 dec->cur += len; 954 dec->cur += len;
768 955
… …
776 } 963 }
777 964
778 SV *k = decode_sv (dec); 965 SV *k = decode_sv (dec);
779 SV *v = decode_sv (dec); 966 SV *v = decode_sv (dec);
780 967
968 // we leak memory if uncaught exceptions are thrown by random magical
969 // methods, and this is hopefully the only place where it can happen,
970 // so if there is a chance of an exception, take the very slow path.
971 // since catching exceptions is "undocumented/internal/forbidden" by
972 // the new p5p powers, we need to call out to a perl function :/
973 if (ecb_expect_false (SvAMAGIC (k)))
974 {
975 dSP;
976
977 ENTER; SAVETMPS;
978 PUSHMARK (SP);
979 EXTEND (SP, 3);
980 PUSHs (sv_2mortal (newRV_inc ((SV *)hv)));
981 PUSHs (sv_2mortal (k));
982 PUSHs (sv_2mortal (v));
983
984 PUTBACK;
985 call_pv ("CBOR::XS::_hv_store", G_VOID | G_DISCARD | G_EVAL);
986 SPAGAIN;
987
988 FREETMPS; LEAVE;
989
990 if (SvTRUE (ERRSV))
991 ERR_ERRSV;
992
993 return;
994 }
995
781 hv_store_ent (hv, k, v, 0); 996 hv_store_ent (hv, k, v, 0);
782 SvREFCNT_dec (k); 997 SvREFCNT_dec_NN (k);
783 998
784fail: 999fail:
785 ; 1000 ;
786} 1001}
787 1002
… …
798 1013
799 for (;;) 1014 for (;;)
800 { 1015 {
801 WANT (1); 1016 WANT (1);
802 1017
803 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF)) 1018 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF) || dec->err)
804 { 1019 {
805 ++dec->cur; 1020 ++dec->cur;
806 break; 1021 break;
807 } 1022 }
808 1023
809 decode_he (dec, hv); 1024 decode_he (dec, hv);
810 } 1025 }
811 } 1026 }
812 else 1027 else
813 { 1028 {
814 int pairs = decode_uint (dec); 1029 UV pairs = decode_uint (dec);
1030
1031 WANT (pairs); // complexity check - need at least one byte per value, do not allow supersize hashes
815 1032
816 while (pairs--) 1033 while (pairs--)
817 decode_he (dec, hv); 1034 decode_he (dec, hv);
818 } 1035 }
819 1036
820 DEC_DEC_DEPTH; 1037 DEC_DEC_DEPTH;
821 return newRV_noinc ((SV *)hv); 1038 return newRV_noinc ((SV *)hv);
822 1039
823fail: 1040fail:
824 SvREFCNT_dec (hv); 1041 SvREFCNT_dec_NN (hv);
825 DEC_DEC_DEPTH; 1042 DEC_DEC_DEPTH;
826 return &PL_sv_undef; 1043 return &PL_sv_undef;
827} 1044}
828 1045
829static SV * 1046static SV *
830decode_str (dec_t *dec, int utf8) 1047decode_str (dec_t *dec, int utf8)
831{ 1048{
832 SV *sv = 0; 1049 SV *sv = 0;
833 1050
834 if ((*dec->cur & MINOR_MASK) == MINOR_INDEF) 1051 if (ecb_expect_false ((*dec->cur & MINOR_MASK) == MINOR_INDEF))
835 { 1052 {
836 // indefinite length strings 1053 // indefinite length strings
837 ++dec->cur; 1054 ++dec->cur;
838 1055
839 U8 major = *dec->cur & MAJOR_MISC; 1056 U8 major = *dec->cur & MAJOR_MISC;
… …
907 sv = newRV_noinc (decode_sv (dec)); 1124 sv = newRV_noinc (decode_sv (dec));
908 break; 1125 break;
909 1126
910 case CBOR_TAG_STRINGREF_NAMESPACE: 1127 case CBOR_TAG_STRINGREF_NAMESPACE:
911 { 1128 {
1129 // do not use SAVETMPS/FREETMPS, as these will
1130 // erase mortalised caches, e.g. "shareable"
912 ENTER; SAVETMPS; 1131 ENTER;
913 1132
914 SAVESPTR (dec->stringref); 1133 SAVESPTR (dec->stringref);
915 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ()); 1134 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ());
916 1135
917 sv = decode_sv (dec); 1136 sv = decode_sv (dec);
918 1137
919 FREETMPS; LEAVE; 1138 LEAVE;
920 } 1139 }
921 break; 1140 break;
922 1141
923 case CBOR_TAG_STRINGREF: 1142 case CBOR_TAG_STRINGREF:
924 { 1143 {
925 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT)) 1144 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
926 ERR ("corrupted CBOR data (stringref index not an unsigned integer)"); 1145 ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
927 1146
928 UV idx = decode_uint (dec); 1147 UV idx = decode_uint (dec);
929 1148
930 if (!dec->stringref || (int)idx > AvFILLp (dec->stringref)) 1149 if (!dec->stringref || idx >= (UV)(1 + AvFILLp (dec->stringref)))
931 ERR ("corrupted CBOR data (stringref index out of bounds or outside namespace)"); 1150 ERR ("corrupted CBOR data (stringref index out of bounds or outside namespace)");
932 1151
933 sv = newSVsv (AvARRAY (dec->stringref)[idx]); 1152 sv = newSVsv (AvARRAY (dec->stringref)[idx]);
934 } 1153 }
935 break; 1154 break;
… …
963 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT)) 1182 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
964 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)"); 1183 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
965 1184
966 UV idx = decode_uint (dec); 1185 UV idx = decode_uint (dec);
967 1186
968 if (!dec->shareable || (int)idx > AvFILLp (dec->shareable)) 1187 if (!dec->shareable || idx >= (UV)(1 + AvFILLp (dec->shareable)))
969 ERR ("corrupted CBOR data (sharedref index out of bounds)"); 1188 ERR ("corrupted CBOR data (sharedref index out of bounds)");
970 1189
971 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]); 1190 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]);
972 1191
973 if (sv == &PL_sv_undef) 1192 if (sv == &PL_sv_undef)
… …
975 } 1194 }
976 break; 1195 break;
977 1196
978 case CBOR_TAG_PERL_OBJECT: 1197 case CBOR_TAG_PERL_OBJECT:
979 { 1198 {
1199 if (dec->cbor.flags & F_FORBID_OBJECTS)
1200 goto filter;
1201
980 sv = decode_sv (dec); 1202 sv = decode_sv (dec);
981 1203
982 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV) 1204 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
983 ERR ("corrupted CBOR data (non-array perl object)"); 1205 ERR ("corrupted CBOR data (non-array perl object)");
984 1206
… …
994 if (!method) 1216 if (!method)
995 ERR ("cannot decode perl-object (package does not have a THAW method)"); 1217 ERR ("cannot decode perl-object (package does not have a THAW method)");
996 1218
997 dSP; 1219 dSP;
998 1220
999 ENTER; SAVETMPS; PUSHMARK (SP); 1221 ENTER; SAVETMPS;
1222 PUSHMARK (SP);
1000 EXTEND (SP, len + 1); 1223 EXTEND (SP, len + 1);
1001 // we re-bless the reference to get overload and other niceties right 1224 // we re-bless the reference to get overload and other niceties right
1002 PUSHs (*av_fetch (av, 0, 1)); 1225 PUSHs (*av_fetch (av, 0, 1));
1003 PUSHs (sv_cbor); 1226 PUSHs (sv_cbor);
1004 1227
… …
1012 SPAGAIN; 1235 SPAGAIN;
1013 1236
1014 if (SvTRUE (ERRSV)) 1237 if (SvTRUE (ERRSV))
1015 { 1238 {
1016 FREETMPS; LEAVE; 1239 FREETMPS; LEAVE;
1017 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV)))); 1240 ERR_ERRSV;
1018 } 1241 }
1019 1242
1020 SvREFCNT_dec (sv); 1243 SvREFCNT_dec_NN (sv);
1021 sv = SvREFCNT_inc (POPs); 1244 sv = SvREFCNT_inc (POPs);
1022 1245
1023 PUTBACK; 1246 PUTBACK;
1024 1247
1025 FREETMPS; LEAVE; 1248 FREETMPS; LEAVE;
1026 } 1249 }
1027 break; 1250 break;
1028 1251
1029 default: 1252 default:
1253 filter:
1030 { 1254 {
1255 SV *tag_sv = newSVuv (tag);
1256
1031 sv = decode_sv (dec); 1257 sv = decode_sv (dec);
1032 1258
1033 dSP; 1259 dSP;
1034 ENTER; SAVETMPS; PUSHMARK (SP); 1260 ENTER; SAVETMPS;
1261 PUSHMARK (SP);
1035 EXTEND (SP, 2); 1262 EXTEND (SP, 2);
1036 PUSHs (newSVuv (tag)); 1263 PUSHs (tag_sv);
1037 PUSHs (sv); 1264 PUSHs (sv);
1038 1265
1039 PUTBACK; 1266 PUTBACK;
1040 int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL); 1267 int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL);
1041 SPAGAIN; 1268 SPAGAIN;
1042 1269
1043 if (SvTRUE (ERRSV)) 1270 if (SvTRUE (ERRSV))
1044 { 1271 {
1272 SvREFCNT_dec_NN (tag_sv);
1045 FREETMPS; LEAVE; 1273 FREETMPS; LEAVE;
1046 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV)))); 1274 ERR_ERRSV;
1047 } 1275 }
1048 1276
1049 if (count) 1277 if (count)
1050 { 1278 {
1279 SvREFCNT_dec_NN (tag_sv);
1051 SvREFCNT_dec (sv); 1280 SvREFCNT_dec_NN (sv);
1052 sv = SvREFCNT_inc (POPs); 1281 sv = SvREFCNT_inc_NN (TOPs);
1282 SP -= count;
1053 } 1283 }
1054 else 1284 else
1055 { 1285 {
1056 AV *av = newAV (); 1286 AV *av = newAV ();
1057 av_push (av, newSVuv (tag)); 1287 av_push (av, tag_sv);
1058 av_push (av, sv); 1288 av_push (av, sv);
1059 1289
1060 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 1290 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
1061 ? cbor_tagged_stash 1291 ? cbor_tagged_stash
1062 : gv_stashpv ("CBOR::XS::Tagged" , 1); 1292 : gv_stashpv ("CBOR::XS::Tagged" , 1);
… …
1191 1421
1192 if (dec.err) 1422 if (dec.err)
1193 { 1423 {
1194 if (dec.shareable) 1424 if (dec.shareable)
1195 { 1425 {
1196 // need to break cyclic links, which whould all be in shareable 1426 // need to break cyclic links, which would all be in shareable
1197 int i; 1427 int i;
1198 SV **svp; 1428 SV **svp;
1199 1429
1200 for (i = av_len (dec.shareable) + 1; i--; ) 1430 for (i = av_len (dec.shareable) + 1; i--; )
1201 if ((svp = av_fetch (dec.shareable, i, 0))) 1431 if ((svp = av_fetch (dec.shareable, i, 0)))
1202 sv_setsv (*svp, &PL_sv_undef); 1432 sv_setsv (*svp, &PL_sv_undef);
1203 } 1433 }
1204 1434
1205 SvREFCNT_dec (sv); 1435 SvREFCNT_dec_NN (sv);
1436
1437 if (dec.err_sv)
1438 sv_2mortal (dec.err_sv);
1439
1206 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur); 1440 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur);
1207 } 1441 }
1208 1442
1209 sv = sv_2mortal (sv); 1443 sv = sv_2mortal (sv);
1210 1444
… …
1354 1588
1355 default_filter = newSVpv ("CBOR::XS::default_filter", 0); 1589 default_filter = newSVpv ("CBOR::XS::default_filter", 0);
1356 1590
1357 sv_cbor = newSVpv ("CBOR", 0); 1591 sv_cbor = newSVpv ("CBOR", 0);
1358 SvREADONLY_on (sv_cbor); 1592 SvREADONLY_on (sv_cbor);
1593
1594 assert (("STRLEN must be an unsigned type", 0 <= (STRLEN)-1));
1359} 1595}
1360 1596
1361PROTOTYPES: DISABLE 1597PROTOTYPES: DISABLE
1362 1598
1363void CLONE (...) 1599void CLONE (...)
… …
1383 ALIAS: 1619 ALIAS:
1384 shrink = F_SHRINK 1620 shrink = F_SHRINK
1385 allow_unknown = F_ALLOW_UNKNOWN 1621 allow_unknown = F_ALLOW_UNKNOWN
1386 allow_sharing = F_ALLOW_SHARING 1622 allow_sharing = F_ALLOW_SHARING
1387 allow_cycles = F_ALLOW_CYCLES 1623 allow_cycles = F_ALLOW_CYCLES
1624 forbid_objects = F_FORBID_OBJECTS
1388 pack_strings = F_PACK_STRINGS 1625 pack_strings = F_PACK_STRINGS
1626 text_keys = F_TEXT_KEYS
1627 text_strings = F_TEXT_STRINGS
1389 validate_utf8 = F_VALIDATE_UTF8 1628 validate_utf8 = F_VALIDATE_UTF8
1390 PPCODE: 1629 PPCODE:
1391{ 1630{
1392 if (enable) 1631 if (enable)
1393 self->flags |= ix; 1632 self->flags |= ix;
… …
1401 ALIAS: 1640 ALIAS:
1402 get_shrink = F_SHRINK 1641 get_shrink = F_SHRINK
1403 get_allow_unknown = F_ALLOW_UNKNOWN 1642 get_allow_unknown = F_ALLOW_UNKNOWN
1404 get_allow_sharing = F_ALLOW_SHARING 1643 get_allow_sharing = F_ALLOW_SHARING
1405 get_allow_cycles = F_ALLOW_CYCLES 1644 get_allow_cycles = F_ALLOW_CYCLES
1645 get_forbid_objects = F_FORBID_OBJECTS
1406 get_pack_strings = F_PACK_STRINGS 1646 get_pack_strings = F_PACK_STRINGS
1647 get_text_keys = F_TEXT_KEYS
1648 get_text_strings = F_TEXT_STRINGS
1407 get_validate_utf8 = F_VALIDATE_UTF8 1649 get_validate_utf8 = F_VALIDATE_UTF8
1408 PPCODE: 1650 PPCODE:
1409 XPUSHs (boolSV (self->flags & ix)); 1651 XPUSHs (boolSV (self->flags & ix));
1410 1652
1411void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) 1653void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
… …
1541 cbor_init (&cbor); 1783 cbor_init (&cbor);
1542 PUTBACK; cborstr = decode_cbor (cborstr, &cbor, 0); SPAGAIN; 1784 PUTBACK; cborstr = decode_cbor (cborstr, &cbor, 0); SPAGAIN;
1543 XPUSHs (cborstr); 1785 XPUSHs (cborstr);
1544} 1786}
1545 1787
1788#ifdef __AFL_COMPILER
1789
1790void
1791afl_init ()
1792 CODE:
1793 __AFL_INIT ();
1794
1795int
1796afl_loop (unsigned int count = 10000)
1797 CODE:
1798 RETVAL = __AFL_LOOP (count);
1799 OUTPUT:
1800 RETVAL
1801
1802#endif
1803

Diff Legend

– Removed lines
+ Added lines
< Changed lines
> Changed lines