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 1.45 by root, Sun Dec 14 05:48:39 2014 UTC vs.
Revision 1.66 by root, Sun Nov 29 21:32:01 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_BYTES = 1,
109 AS_TEXT = 2,
110 AS_FLOAT16 = 3,
111 AS_FLOAT32 = 4,
112 AS_FLOAT64 = 5,
113 // possibly future enhancements: float, integer
114};
115
100#define F_SHRINK 0x00000001UL 116#define F_SHRINK 0x00000001UL
101#define F_ALLOW_UNKNOWN 0x00000002UL 117#define F_ALLOW_UNKNOWN 0x00000002UL
102#define F_ALLOW_SHARING 0x00000004UL 118#define F_ALLOW_SHARING 0x00000004UL
103#define F_ALLOW_CYCLES 0x00000008UL 119#define F_ALLOW_CYCLES 0x00000008UL
120#define F_FORBID_OBJECTS 0x00000010UL
104#define F_PACK_STRINGS 0x00000010UL 121#define F_PACK_STRINGS 0x00000020UL
122#define F_TEXT_KEYS 0x00000040UL
123#define F_TEXT_STRINGS 0x00000080UL
105#define F_VALIDATE_UTF8 0x00000020UL 124#define F_VALIDATE_UTF8 0x00000100UL
106 125
107#define INIT_SIZE 32 // initial scalar size to be allocated 126#define INIT_SIZE 32 // initial scalar size to be allocated
108 127
109#define SB do { 128#define SB do {
110#define SE } while (0) 129#define SE } while (0)
… …
183 202
184// minimum length of a string to be registered for stringref 203// minimum length of a string to be registered for stringref
185ecb_inline int 204ecb_inline int
186minimum_string_length (UV idx) 205minimum_string_length (UV idx)
187{ 206{
188 return idx > 23 207 return idx <= 23 ? 3
189 ? idx > 0xffU 208 : idx <= 0xffU ? 4
190 ? idx > 0xffffU 209 : idx <= 0xffffU ? 5
191 ? idx > 0xffffffffU 210 : idx <= 0xffffffffU ? 7
192 ? 11 211 : 11;
193 : 7
194 : 5
195 : 4
196 : 3;
197} 212}
198 213
199///////////////////////////////////////////////////////////////////////////// 214/////////////////////////////////////////////////////////////////////////////
200// encoder 215// encoder
201 216
… …
214} enc_t; 229} enc_t;
215 230
216ecb_inline void 231ecb_inline void
217need (enc_t *enc, STRLEN len) 232need (enc_t *enc, STRLEN len)
218{ 233{
219 if (ecb_expect_false (enc->cur + len >= enc->end)) 234 if (ecb_expect_false ((uintptr_t)(enc->end - enc->cur) < len))
220 { 235 {
221 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv); 236 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
222 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1); 237 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
223 enc->cur = SvPVX (enc->sv) + cur; 238 enc->cur = SvPVX (enc->sv) + cur;
224 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; 239 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
… …
230{ 245{
231 need (enc, 1); 246 need (enc, 1);
232 *enc->cur++ = ch; 247 *enc->cur++ = ch;
233} 248}
234 249
250// used for tags, intregers, element counts and so on
235static void 251static void
236encode_uint (enc_t *enc, int major, UV len) 252encode_uint (enc_t *enc, int major, UV len)
237{ 253{
238 need (enc, 9); 254 need (enc, 9);
239 255
… …
270 *enc->cur++ = len >> 8; 286 *enc->cur++ = len >> 8;
271 *enc->cur++ = len; 287 *enc->cur++ = len;
272 } 288 }
273} 289}
274 290
291// encodes a perl value into a CBOR integer
292ecb_inline
293encode_int (enc_t *enc, SV *sv)
294{
295 if (SvIsUV (sv))
296 encode_uint (enc, MAJOR_POS_INT, SvUVX (sv));
297 else if (SvIVX (sv) >= 0)
298 encode_uint (enc, MAJOR_POS_INT, SvIVX (sv));
299 else
300 encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1));
301}
302
275ecb_inline void 303ecb_inline void
276encode_tag (enc_t *enc, UV tag) 304encode_tag (enc_t *enc, UV tag)
277{ 305{
278 encode_uint (enc, MAJOR_TAG, tag); 306 encode_uint (enc, MAJOR_TAG, tag);
279} 307}
280 308
309// exceptional (hopefully) slow path for byte strings that need to be utf8-encoded
310ecb_noinline static void
311encode_str_utf8 (enc_t *enc, int utf8, char *str, STRLEN len)
312{
313 STRLEN ulen = len;
314 U8 *p, *pend = (U8 *)str + len;
315
316 for (p = (U8 *)str; p < pend; ++p)
317 ulen += *p >> 7; // count set high bits
318
319 encode_uint (enc, MAJOR_TEXT, ulen);
320
321 need (enc, ulen);
322 for (p = (U8 *)str; p < pend; ++p)
323 if (*p < 0x80)
324 *enc->cur++ = *p;
325 else
326 {
327 *enc->cur++ = 0xc0 + (*p >> 6);
328 *enc->cur++ = 0x80 + (*p & 63);
329 }
330}
331
281ecb_inline void 332ecb_inline void
282encode_str (enc_t *enc, int utf8, char *str, STRLEN len) 333encode_str (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len)
283{ 334{
335 if (ecb_expect_false (upgrade_utf8))
336 if (!utf8)
337 {
338 encode_str_utf8 (enc, utf8, str, len);
339 return;
340 }
341
284 encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len); 342 encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len);
285 need (enc, len); 343 need (enc, len);
286 memcpy (enc->cur, str, len); 344 memcpy (enc->cur, str, len);
287 enc->cur += len; 345 enc->cur += len;
288} 346}
289 347
290static void 348ecb_inline void
291encode_strref (enc_t *enc, int utf8, char *str, STRLEN len) 349encode_strref (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len)
292{ 350{
293 if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS)) 351 if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS))
294 { 352 {
295 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1); 353 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1);
296 354
… …
307 sv_setuv (*svp, enc->stringref_idx); 365 sv_setuv (*svp, enc->stringref_idx);
308 ++enc->stringref_idx; 366 ++enc->stringref_idx;
309 } 367 }
310 } 368 }
311 369
312 encode_str (enc, utf8, str, len); 370 encode_str (enc, upgrade_utf8, utf8, str, len);
371}
372
373ecb_inline void
374encode_float16 (enc_t *enc, NV nv)
375{
376 need (enc, 1+2);
377
378 *enc->cur++ = MAJOR_MISC | MISC_FLOAT16;
379
380 uint16_t fp = ecb_float_to_binary16 (nv);
381
382 if (!ecb_big_endian ())
383 fp = ecb_bswap16 (fp);
384
385 memcpy (enc->cur, &fp, 2);
386 enc->cur += 2;
387}
388
389ecb_inline void
390encode_float32 (enc_t *enc, NV nv)
391{
392 need (enc, 1+4);
393
394 *enc->cur++ = MAJOR_MISC | MISC_FLOAT32;
395
396 uint32_t fp = ecb_float_to_binary32 (nv);
397
398 if (!ecb_big_endian ())
399 fp = ecb_bswap32 (fp);
400
401 memcpy (enc->cur, &fp, 4);
402 enc->cur += 4;
403}
404
405ecb_inline void
406encode_float64 (enc_t *enc, NV nv)
407{
408 need (enc, 1+8);
409
410 *enc->cur++ = MAJOR_MISC | MISC_FLOAT64;
411
412 uint64_t fp = ecb_double_to_binary64 (nv);
413
414 if (!ecb_big_endian ())
415 fp = ecb_bswap64 (fp);
416
417 memcpy (enc->cur, &fp, 8);
418 enc->cur += 8;
419}
420
421ecb_inline void
422encode_forced (enc_t *enc, UV type, SV *sv)
423{
424 switch (type)
425 {
426 case AS_CBOR:
427 {
428 STRLEN len;
429 char *str = SvPVbyte (sv, len);
430
431 need (enc, len);
432 memcpy (enc->cur, str, len);
433 enc->cur += len;
434 }
435 break;
436
437 case AS_BYTES:
438 {
439 STRLEN len;
440 char *str = SvPVbyte (sv, len);
441 encode_strref (enc, 0, 0, str, len);
442 }
443 break;
444
445 case AS_TEXT:
446 {
447 STRLEN len;
448 char *str = SvPVutf8 (sv, len);
449 encode_strref (enc, 1, 1, str, len);
450 }
451 break;
452
453 case AS_FLOAT16: encode_float16 (enc, SvNV (sv)); break;
454 case AS_FLOAT32: encode_float32 (enc, SvNV (sv)); break;
455 case AS_FLOAT64: encode_float64 (enc, SvNV (sv)); break;
456
457 default:
458 croak ("encountered malformed CBOR::XS::Tagged object");
459 }
313} 460}
314 461
315static void encode_sv (enc_t *enc, SV *sv); 462static void encode_sv (enc_t *enc, SV *sv);
316 463
317static void 464static void
… …
324 471
325 ++enc->depth; 472 ++enc->depth;
326 473
327 encode_uint (enc, MAJOR_ARRAY, len + 1); 474 encode_uint (enc, MAJOR_ARRAY, len + 1);
328 475
329 if (SvMAGICAL (av)) 476 if (ecb_expect_false (SvMAGICAL (av)))
330 for (i = 0; i <= len; ++i) 477 for (i = 0; i <= len; ++i)
331 { 478 {
332 SV **svp = av_fetch (av, i, 0); 479 SV **svp = av_fetch (av, i, 0);
333 encode_sv (enc, svp ? *svp : &PL_sv_undef); 480 encode_sv (enc, svp ? *svp : &PL_sv_undef);
334 } 481 }
335 else 482 else
336 for (i = 0; i <= len; ++i) 483 for (i = 0; i <= len; ++i)
337 encode_sv (enc, AvARRAY (av)[i]); 484 {
485 SV *sv = AvARRAY (av)[i];
486 encode_sv (enc, sv ? sv : &PL_sv_undef);
487 }
338 488
339 --enc->depth; 489 --enc->depth;
340} 490}
341 491
342static void 492static void
… …
350 ++enc->depth; 500 ++enc->depth;
351 501
352 int pairs = hv_iterinit (hv); 502 int pairs = hv_iterinit (hv);
353 int mg = SvMAGICAL (hv); 503 int mg = SvMAGICAL (hv);
354 504
355 if (mg) 505 if (ecb_expect_false (mg))
356 encode_ch (enc, MAJOR_MAP | MINOR_INDEF); 506 encode_ch (enc, MAJOR_MAP | MINOR_INDEF);
357 else 507 else
358 encode_uint (enc, MAJOR_MAP, pairs); 508 encode_uint (enc, MAJOR_MAP, pairs);
359 509
360 while ((he = hv_iternext (hv))) 510 while ((he = hv_iternext (hv)))
361 { 511 {
362 if (HeKLEN (he) == HEf_SVKEY) 512 if (HeKLEN (he) == HEf_SVKEY)
363 encode_sv (enc, HeSVKEY (he)); 513 encode_sv (enc, HeSVKEY (he));
364 else 514 else
365 encode_strref (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he)); 515 encode_strref (enc, enc->cbor.flags & (F_TEXT_KEYS | F_TEXT_STRINGS), HeKUTF8 (he), HeKEY (he), HeKLEN (he));
366 516
367 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he)); 517 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he));
368 } 518 }
369 519
370 if (mg) 520 if (ecb_expect_false (mg))
371 encode_ch (enc, MAJOR_MISC | MINOR_INDEF); 521 encode_ch (enc, MAJOR_MISC | MINOR_INDEF);
372 522
373 --enc->depth; 523 --enc->depth;
374} 524}
375 525
… …
408 else if (stash == tagged_stash) 558 else if (stash == tagged_stash)
409 { 559 {
410 if (svt != SVt_PVAV) 560 if (svt != SVt_PVAV)
411 croak ("encountered CBOR::XS::Tagged object that isn't an array"); 561 croak ("encountered CBOR::XS::Tagged object that isn't an array");
412 562
563 switch (av_len ((AV *)sv))
564 {
565 case 2-1:
566 // actually a tagged value
413 encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1))); 567 encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1)));
414 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1)); 568 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1));
569 break;
570
571 case 3-1:
572 // a forced type [value, type, undef]
573 encode_forced (enc, SvUV (*av_fetch ((AV *)sv, 1, 1)), *av_fetch ((AV *)sv, 0, 1));
574 break;
575
576 default:
577 croak ("encountered malformed CBOR::XS::Tagged object");
578 }
415 579
416 return; 580 return;
417 } 581 }
418 } 582 }
419 583
420 if (ecb_expect_false (SvREFCNT (sv) > 1) 584 if (ecb_expect_false (SvREFCNT (sv) > 1)
421 && ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING)) 585 && ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING))
422 { 586 {
423 if (!enc->shareable) 587 if (ecb_expect_false (!enc->shareable))
424 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ()); 588 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
425 589
426 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1); 590 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
427 591
428 if (SvOK (*svp)) 592 if (SvOK (*svp))
… …
442 if (ecb_expect_false (SvOBJECT (sv))) 606 if (ecb_expect_false (SvOBJECT (sv)))
443 { 607 {
444 HV *stash = SvSTASH (sv); 608 HV *stash = SvSTASH (sv);
445 GV *method; 609 GV *method;
446 610
611 if (enc->cbor.flags & F_FORBID_OBJECTS)
612 croak ("encountered object '%s', but forbid_objects is enabled",
613 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
447 if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0))) 614 else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
448 { 615 {
449 dSP; 616 dSP;
450 617
451 ENTER; SAVETMPS; PUSHMARK (SP); 618 ENTER; SAVETMPS;
619 PUSHMARK (SP);
452 // we re-bless the reference to get overload and other niceties right 620 // we re-bless the reference to get overload and other niceties right
453 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); 621 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
454 622
455 PUTBACK; 623 PUTBACK;
456 // G_SCALAR ensures that return value is 1 624 // G_SCALAR ensures that return value is 1
… …
469 } 637 }
470 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0) 638 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)
471 { 639 {
472 dSP; 640 dSP;
473 641
474 ENTER; SAVETMPS; PUSHMARK (SP); 642 ENTER; SAVETMPS;
643 PUSHMARK (SP);
475 EXTEND (SP, 2); 644 EXTEND (SP, 2);
476 // we re-bless the reference to get overload and other niceties right 645 // we re-bless the reference to get overload and other niceties right
477 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); 646 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
478 PUSHs (sv_cbor); 647 PUSHs (sv_cbor);
479 648
… …
485 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv) 654 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)
486 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash)); 655 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));
487 656
488 encode_tag (enc, CBOR_TAG_PERL_OBJECT); 657 encode_tag (enc, CBOR_TAG_PERL_OBJECT);
489 encode_uint (enc, MAJOR_ARRAY, count + 1); 658 encode_uint (enc, MAJOR_ARRAY, count + 1);
490 encode_strref (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash)); 659 encode_strref (enc, 0, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
491 660
492 while (count) 661 {
662 int i;
663
664 for (i = 0; i < count; ++i)
493 encode_sv (enc, SP[1 - count--]); 665 encode_sv (enc, SP[i + 1 - count]);
666
667 SP -= count;
668 }
494 669
495 PUTBACK; 670 PUTBACK;
496 671
497 FREETMPS; LEAVE; 672 FREETMPS; LEAVE;
498 } 673 }
… …
520 695
521 if (ecb_expect_false (nv == (NV)(U32)nv)) 696 if (ecb_expect_false (nv == (NV)(U32)nv))
522 encode_uint (enc, MAJOR_POS_INT, (U32)nv); 697 encode_uint (enc, MAJOR_POS_INT, (U32)nv);
523 //TODO: maybe I32? 698 //TODO: maybe I32?
524 else if (ecb_expect_false (nv == (float)nv)) 699 else if (ecb_expect_false (nv == (float)nv))
525 { 700 encode_float32 (enc, nv);
526 uint32_t fp = ecb_float_to_binary32 (nv);
527
528 *enc->cur++ = MAJOR_MISC | MISC_FLOAT32;
529
530 if (!ecb_big_endian ())
531 fp = ecb_bswap32 (fp);
532
533 memcpy (enc->cur, &fp, 4);
534 enc->cur += 4;
535 }
536 else 701 else
537 { 702 encode_float64 (enc, nv);
538 uint64_t fp = ecb_double_to_binary64 (nv);
539
540 *enc->cur++ = MAJOR_MISC | MISC_FLOAT64;
541
542 if (!ecb_big_endian ())
543 fp = ecb_bswap64 (fp);
544
545 memcpy (enc->cur, &fp, 8);
546 enc->cur += 8;
547 }
548} 703}
549 704
550static void 705static void
551encode_sv (enc_t *enc, SV *sv) 706encode_sv (enc_t *enc, SV *sv)
552{ 707{
… …
554 709
555 if (SvPOKp (sv)) 710 if (SvPOKp (sv))
556 { 711 {
557 STRLEN len; 712 STRLEN len;
558 char *str = SvPV (sv, len); 713 char *str = SvPV (sv, len);
559 encode_strref (enc, SvUTF8 (sv), str, len); 714 encode_strref (enc, enc->cbor.flags & F_TEXT_STRINGS, SvUTF8 (sv), str, len);
560 } 715 }
561 else if (SvNOKp (sv)) 716 else if (SvNOKp (sv))
562 encode_nv (enc, sv); 717 encode_nv (enc, sv);
563 else if (SvIOKp (sv)) 718 else if (SvIOKp (sv))
564 { 719 encode_int (e,v sv);
565 if (SvIsUV (sv))
566 encode_uint (enc, MAJOR_POS_INT, SvUVX (sv));
567 else if (SvIVX (sv) >= 0)
568 encode_uint (enc, MAJOR_POS_INT, SvIVX (sv));
569 else
570 encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1));
571 }
572 else if (SvROK (sv)) 720 else if (SvROK (sv))
573 encode_rv (enc, SvRV (sv)); 721 encode_rv (enc, SvRV (sv));
574 else if (!SvOK (sv)) 722 else if (!SvOK (sv))
575 encode_ch (enc, MAJOR_MISC | SIMPLE_NULL); 723 encode_ch (enc, MAJOR_MISC | SIMPLE_NULL);
576 else if (enc->cbor.flags & F_ALLOW_UNKNOWN) 724 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
… …
581} 729}
582 730
583static SV * 731static SV *
584encode_cbor (SV *scalar, CBOR *cbor) 732encode_cbor (SV *scalar, CBOR *cbor)
585{ 733{
586 enc_t enc = { }; 734 enc_t enc = { 0 };
587 735
588 enc.cbor = *cbor; 736 enc.cbor = *cbor;
589 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 737 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
590 enc.cur = SvPVX (enc.sv); 738 enc.cur = SvPVX (enc.sv);
591 enc.end = SvEND (enc.sv); 739 enc.end = SvEND (enc.sv);
592 740
593 SvPOK_only (enc.sv); 741 SvPOK_only (enc.sv);
594 742
595 if (cbor->flags & F_PACK_STRINGS) 743 if (cbor->flags & F_PACK_STRINGS)
596 { 744 {
… …
623 U32 depth; // recursion depth 771 U32 depth; // recursion depth
624 U32 maxdepth; // recursion depth limit 772 U32 maxdepth; // recursion depth limit
625 AV *shareable; 773 AV *shareable;
626 AV *stringref; 774 AV *stringref;
627 SV *decode_tagged; 775 SV *decode_tagged;
776 SV *err_sv; // optional sv for error, needs to be freed
628} dec_t; 777} dec_t;
629 778
630#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE 779// set dec->err to ERRSV
780ecb_cold static void
781err_errsv (dec_t *dec)
782{
783 if (!dec->err)
784 {
785 dec->err_sv = newSVsv (ERRSV);
631 786
632#define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data") 787 // chop off the trailing \n
788 SvCUR_set (dec->err_sv, SvCUR (dec->err_sv) - 1);
789 *SvEND (dec->err_sv) = 0;
633 790
791 dec->err = SvPVutf8_nolen (dec->err_sv);
792 }
793}
794
795// the following functions are used to reduce code size and help the compiler to optimise
796ecb_cold static void
797err_set (dec_t *dec, const char *reason)
798{
799 if (!dec->err)
800 dec->err = reason;
801}
802
803ecb_cold static void
804err_unexpected_end (dec_t *dec)
805{
806 err_set (dec, "unexpected end of CBOR data");
807}
808
809#define ERR_DO(do) SB do; goto fail; SE
810#define ERR(reason) ERR_DO (err_set (dec, reason))
811#define ERR_ERRSV ERR_DO (err_errsv (dec))
812
813#define WANT(len) if (ecb_expect_false ((uintptr_t)(dec->end - dec->cur) < (STRLEN)len)) ERR_DO (err_unexpected_end (dec))
814
634#define DEC_INC_DEPTH if (++dec->depth > dec->cbor.max_depth) ERR (ERR_NESTING_EXCEEDED) 815#define DEC_INC_DEPTH if (ecb_expect_false (++dec->depth > dec->cbor.max_depth)) ERR (ERR_NESTING_EXCEEDED)
635#define DEC_DEC_DEPTH --dec->depth 816#define DEC_DEC_DEPTH --dec->depth
636 817
637static UV 818static UV
638decode_uint (dec_t *dec) 819decode_uint (dec_t *dec)
639{ 820{
… …
705 886
706 for (;;) 887 for (;;)
707 { 888 {
708 WANT (1); 889 WANT (1);
709 890
710 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF)) 891 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF) || dec->err)
711 { 892 {
712 ++dec->cur; 893 ++dec->cur;
713 break; 894 break;
714 } 895 }
715 896
716 av_push (av, decode_sv (dec)); 897 av_push (av, decode_sv (dec));
717 } 898 }
718 } 899 }
719 else 900 else
720 { 901 {
721 int i, len = decode_uint (dec); 902 UV i, len = decode_uint (dec);
722 903
723 WANT (len); // complexity check for av_fill - need at least one byte per value, do not allow supersize arrays 904 WANT (len); // complexity check for av_fill - need at least one byte per value, do not allow supersize arrays
724 av_fill (av, len - 1); 905 av_fill (av, len - 1);
725 906
726 for (i = 0; i < len; ++i) 907 for (i = 0; i < len; ++i)
… …
729 910
730 DEC_DEC_DEPTH; 911 DEC_DEC_DEPTH;
731 return newRV_noinc ((SV *)av); 912 return newRV_noinc ((SV *)av);
732 913
733fail: 914fail:
734 SvREFCNT_dec (av); 915 SvREFCNT_dec_NN (av);
735 DEC_DEC_DEPTH; 916 DEC_DEC_DEPTH;
736 return &PL_sv_undef; 917 return &PL_sv_undef;
737} 918}
738 919
739static void 920static void
… …
743 // byte or utf-8 strings as keys, but only when !stringref 924 // byte or utf-8 strings as keys, but only when !stringref
744 925
745 if (ecb_expect_true (!dec->stringref)) 926 if (ecb_expect_true (!dec->stringref))
746 if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8)) 927 if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8))
747 { 928 {
748 I32 len = decode_uint (dec); 929 STRLEN len = decode_uint (dec);
749 char *key = (char *)dec->cur; 930 char *key = (char *)dec->cur;
750 931
932 WANT (len);
751 dec->cur += len; 933 dec->cur += len;
752 934
753 hv_store (hv, key, len, decode_sv (dec), 0); 935 hv_store (hv, key, len, decode_sv (dec), 0);
754 936
755 return; 937 return;
756 } 938 }
757 else if (ecb_expect_true ((U8)(*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8)) 939 else if (ecb_expect_true ((U8)(*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8))
758 { 940 {
759 I32 len = decode_uint (dec); 941 STRLEN len = decode_uint (dec);
760 char *key = (char *)dec->cur; 942 char *key = (char *)dec->cur;
761 943
944 WANT (len);
762 dec->cur += len; 945 dec->cur += len;
763 946
764 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8)) 947 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
765 if (!is_utf8_string (key, len)) 948 if (!is_utf8_string (key, len))
766 ERR ("corrupted CBOR data (invalid UTF-8 in map key)"); 949 ERR ("corrupted CBOR data (invalid UTF-8 in map key)");
… …
771 } 954 }
772 955
773 SV *k = decode_sv (dec); 956 SV *k = decode_sv (dec);
774 SV *v = decode_sv (dec); 957 SV *v = decode_sv (dec);
775 958
959 // we leak memory if uncaught exceptions are thrown by random magical
960 // methods, and this is hopefully the only place where it can happen,
961 // so if there is a chance of an exception, take the very slow path.
962 // since catching exceptions is "undocumented/internal/forbidden" by
963 // the new p5p powers, we need to call out to a perl function :/
964 if (ecb_expect_false (SvAMAGIC (k)))
965 {
966 dSP;
967
968 ENTER; SAVETMPS;
969 PUSHMARK (SP);
970 EXTEND (SP, 3);
971 PUSHs (sv_2mortal (newRV_inc ((SV *)hv)));
972 PUSHs (sv_2mortal (k));
973 PUSHs (sv_2mortal (v));
974
975 PUTBACK;
976 call_pv ("CBOR::XS::_hv_store", G_VOID | G_DISCARD | G_EVAL);
977 SPAGAIN;
978
979 FREETMPS; LEAVE;
980
981 if (SvTRUE (ERRSV))
982 ERR_ERRSV;
983
984 return;
985 }
986
776 hv_store_ent (hv, k, v, 0); 987 hv_store_ent (hv, k, v, 0);
777 SvREFCNT_dec (k); 988 SvREFCNT_dec_NN (k);
778 989
779fail: 990fail:
780 ; 991 ;
781} 992}
782 993
… …
793 1004
794 for (;;) 1005 for (;;)
795 { 1006 {
796 WANT (1); 1007 WANT (1);
797 1008
798 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF)) 1009 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF) || dec->err)
799 { 1010 {
800 ++dec->cur; 1011 ++dec->cur;
801 break; 1012 break;
802 } 1013 }
803 1014
804 decode_he (dec, hv); 1015 decode_he (dec, hv);
805 } 1016 }
806 } 1017 }
807 else 1018 else
808 { 1019 {
809 int pairs = decode_uint (dec); 1020 UV pairs = decode_uint (dec);
1021
1022 WANT (pairs); // complexity check - need at least one byte per value, do not allow supersize hashes
810 1023
811 while (pairs--) 1024 while (pairs--)
812 decode_he (dec, hv); 1025 decode_he (dec, hv);
813 } 1026 }
814 1027
815 DEC_DEC_DEPTH; 1028 DEC_DEC_DEPTH;
816 return newRV_noinc ((SV *)hv); 1029 return newRV_noinc ((SV *)hv);
817 1030
818fail: 1031fail:
819 SvREFCNT_dec (hv); 1032 SvREFCNT_dec_NN (hv);
820 DEC_DEC_DEPTH; 1033 DEC_DEC_DEPTH;
821 return &PL_sv_undef; 1034 return &PL_sv_undef;
822} 1035}
823 1036
824static SV * 1037static SV *
825decode_str (dec_t *dec, int utf8) 1038decode_str (dec_t *dec, int utf8)
826{ 1039{
827 SV *sv = 0; 1040 SV *sv = 0;
828 1041
829 if ((*dec->cur & MINOR_MASK) == MINOR_INDEF) 1042 if (ecb_expect_false ((*dec->cur & MINOR_MASK) == MINOR_INDEF))
830 { 1043 {
831 // indefinite length strings 1044 // indefinite length strings
832 ++dec->cur; 1045 ++dec->cur;
833 1046
834 U8 major = *dec->cur & MAJOR_MISC; 1047 U8 major = *dec->cur & MAJOR_MISC;
… …
902 sv = newRV_noinc (decode_sv (dec)); 1115 sv = newRV_noinc (decode_sv (dec));
903 break; 1116 break;
904 1117
905 case CBOR_TAG_STRINGREF_NAMESPACE: 1118 case CBOR_TAG_STRINGREF_NAMESPACE:
906 { 1119 {
1120 // do not use SAVETMPS/FREETMPS, as these will
1121 // erase mortalised caches, e.g. "shareable"
907 ENTER; SAVETMPS; 1122 ENTER;
908 1123
909 SAVESPTR (dec->stringref); 1124 SAVESPTR (dec->stringref);
910 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ()); 1125 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ());
911 1126
912 sv = decode_sv (dec); 1127 sv = decode_sv (dec);
913 1128
914 FREETMPS; LEAVE; 1129 LEAVE;
915 } 1130 }
916 break; 1131 break;
917 1132
918 case CBOR_TAG_STRINGREF: 1133 case CBOR_TAG_STRINGREF:
919 { 1134 {
920 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT)) 1135 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
921 ERR ("corrupted CBOR data (stringref index not an unsigned integer)"); 1136 ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
922 1137
923 UV idx = decode_uint (dec); 1138 UV idx = decode_uint (dec);
924 1139
925 if (!dec->stringref || (int)idx > AvFILLp (dec->stringref)) 1140 if (!dec->stringref || idx >= (UV)(1 + AvFILLp (dec->stringref)))
926 ERR ("corrupted CBOR data (stringref index out of bounds or outside namespace)"); 1141 ERR ("corrupted CBOR data (stringref index out of bounds or outside namespace)");
927 1142
928 sv = newSVsv (AvARRAY (dec->stringref)[idx]); 1143 sv = newSVsv (AvARRAY (dec->stringref)[idx]);
929 } 1144 }
930 break; 1145 break;
… …
958 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT)) 1173 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
959 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)"); 1174 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
960 1175
961 UV idx = decode_uint (dec); 1176 UV idx = decode_uint (dec);
962 1177
963 if (!dec->shareable || (int)idx > AvFILLp (dec->shareable)) 1178 if (!dec->shareable || idx >= (UV)(1 + AvFILLp (dec->shareable)))
964 ERR ("corrupted CBOR data (sharedref index out of bounds)"); 1179 ERR ("corrupted CBOR data (sharedref index out of bounds)");
965 1180
966 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]); 1181 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]);
967 1182
968 if (sv == &PL_sv_undef) 1183 if (sv == &PL_sv_undef)
… …
970 } 1185 }
971 break; 1186 break;
972 1187
973 case CBOR_TAG_PERL_OBJECT: 1188 case CBOR_TAG_PERL_OBJECT:
974 { 1189 {
1190 if (dec->cbor.flags & F_FORBID_OBJECTS)
1191 goto filter;
1192
975 sv = decode_sv (dec); 1193 sv = decode_sv (dec);
976 1194
977 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV) 1195 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
978 ERR ("corrupted CBOR data (non-array perl object)"); 1196 ERR ("corrupted CBOR data (non-array perl object)");
979 1197
… …
989 if (!method) 1207 if (!method)
990 ERR ("cannot decode perl-object (package does not have a THAW method)"); 1208 ERR ("cannot decode perl-object (package does not have a THAW method)");
991 1209
992 dSP; 1210 dSP;
993 1211
994 ENTER; SAVETMPS; PUSHMARK (SP); 1212 ENTER; SAVETMPS;
1213 PUSHMARK (SP);
995 EXTEND (SP, len + 1); 1214 EXTEND (SP, len + 1);
996 // we re-bless the reference to get overload and other niceties right 1215 // we re-bless the reference to get overload and other niceties right
997 PUSHs (*av_fetch (av, 0, 1)); 1216 PUSHs (*av_fetch (av, 0, 1));
998 PUSHs (sv_cbor); 1217 PUSHs (sv_cbor);
999 1218
… …
1007 SPAGAIN; 1226 SPAGAIN;
1008 1227
1009 if (SvTRUE (ERRSV)) 1228 if (SvTRUE (ERRSV))
1010 { 1229 {
1011 FREETMPS; LEAVE; 1230 FREETMPS; LEAVE;
1012 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV)))); 1231 ERR_ERRSV;
1013 } 1232 }
1014 1233
1015 SvREFCNT_dec (sv); 1234 SvREFCNT_dec_NN (sv);
1016 sv = SvREFCNT_inc (POPs); 1235 sv = SvREFCNT_inc (POPs);
1017 1236
1018 PUTBACK; 1237 PUTBACK;
1019 1238
1020 FREETMPS; LEAVE; 1239 FREETMPS; LEAVE;
1021 } 1240 }
1022 break; 1241 break;
1023 1242
1024 default: 1243 default:
1244 filter:
1025 { 1245 {
1246 SV *tag_sv = newSVuv (tag);
1247
1026 sv = decode_sv (dec); 1248 sv = decode_sv (dec);
1027 1249
1028 dSP; 1250 dSP;
1029 ENTER; SAVETMPS; PUSHMARK (SP); 1251 ENTER; SAVETMPS;
1252 PUSHMARK (SP);
1030 EXTEND (SP, 2); 1253 EXTEND (SP, 2);
1031 PUSHs (newSVuv (tag)); 1254 PUSHs (tag_sv);
1032 PUSHs (sv); 1255 PUSHs (sv);
1033 1256
1034 PUTBACK; 1257 PUTBACK;
1035 int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL); 1258 int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL);
1036 SPAGAIN; 1259 SPAGAIN;
1037 1260
1038 if (SvTRUE (ERRSV)) 1261 if (SvTRUE (ERRSV))
1039 { 1262 {
1263 SvREFCNT_dec_NN (tag_sv);
1040 FREETMPS; LEAVE; 1264 FREETMPS; LEAVE;
1041 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV)))); 1265 ERR_ERRSV;
1042 } 1266 }
1043 1267
1044 if (count) 1268 if (count)
1045 { 1269 {
1270 SvREFCNT_dec_NN (tag_sv);
1046 SvREFCNT_dec (sv); 1271 SvREFCNT_dec_NN (sv);
1047 sv = SvREFCNT_inc (POPs); 1272 sv = SvREFCNT_inc_NN (TOPs);
1273 SP -= count;
1048 } 1274 }
1049 else 1275 else
1050 { 1276 {
1051 AV *av = newAV (); 1277 AV *av = newAV ();
1052 av_push (av, newSVuv (tag)); 1278 av_push (av, tag_sv);
1053 av_push (av, sv); 1279 av_push (av, sv);
1054 1280
1055 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 1281 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
1056 ? cbor_tagged_stash 1282 ? cbor_tagged_stash
1057 : gv_stashpv ("CBOR::XS::Tagged" , 1); 1283 : gv_stashpv ("CBOR::XS::Tagged" , 1);
… …
1160} 1386}
1161 1387
1162static SV * 1388static SV *
1163decode_cbor (SV *string, CBOR *cbor, char **offset_return) 1389decode_cbor (SV *string, CBOR *cbor, char **offset_return)
1164{ 1390{
1165 dec_t dec = { }; 1391 dec_t dec = { 0 };
1166 SV *sv; 1392 SV *sv;
1167 STRLEN len; 1393 STRLEN len;
1168 char *data = SvPVbyte (string, len); 1394 char *data = SvPVbyte (string, len);
1169 1395
1170 if (len > cbor->max_size && cbor->max_size) 1396 if (len > cbor->max_size && cbor->max_size)
… …
1186 1412
1187 if (dec.err) 1413 if (dec.err)
1188 { 1414 {
1189 if (dec.shareable) 1415 if (dec.shareable)
1190 { 1416 {
1191 // need to break cyclic links, which whould all be in shareable 1417 // need to break cyclic links, which would all be in shareable
1192 int i; 1418 int i;
1193 SV **svp; 1419 SV **svp;
1194 1420
1195 for (i = av_len (dec.shareable) + 1; i--; ) 1421 for (i = av_len (dec.shareable) + 1; i--; )
1196 if ((svp = av_fetch (dec.shareable, i, 0))) 1422 if ((svp = av_fetch (dec.shareable, i, 0)))
1197 sv_setsv (*svp, &PL_sv_undef); 1423 sv_setsv (*svp, &PL_sv_undef);
1198 } 1424 }
1199 1425
1200 SvREFCNT_dec (sv); 1426 SvREFCNT_dec_NN (sv);
1427
1428 if (dec.err_sv)
1429 sv_2mortal (dec.err_sv);
1430
1201 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur); 1431 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur);
1202 } 1432 }
1203 1433
1204 sv = sv_2mortal (sv); 1434 sv = sv_2mortal (sv);
1205 1435
… …
1261 1491
1262 int major = *p >> MAJOR_SHIFT; 1492 int major = *p >> MAJOR_SHIFT;
1263 1493
1264 switch (major) 1494 switch (major)
1265 { 1495 {
1496 case MAJOR_TAG >> MAJOR_SHIFT:
1497 ++count; // tags merely prefix another value
1498 break;
1499
1266 case MAJOR_BYTES >> MAJOR_SHIFT: 1500 case MAJOR_BYTES >> MAJOR_SHIFT:
1267 case MAJOR_TEXT >> MAJOR_SHIFT: 1501 case MAJOR_TEXT >> MAJOR_SHIFT:
1268 case MAJOR_ARRAY >> MAJOR_SHIFT: 1502 case MAJOR_ARRAY >> MAJOR_SHIFT:
1269 case MAJOR_MAP >> MAJOR_SHIFT: 1503 case MAJOR_MAP >> MAJOR_SHIFT:
1270 { 1504 {
… …
1345 1579
1346 default_filter = newSVpv ("CBOR::XS::default_filter", 0); 1580 default_filter = newSVpv ("CBOR::XS::default_filter", 0);
1347 1581
1348 sv_cbor = newSVpv ("CBOR", 0); 1582 sv_cbor = newSVpv ("CBOR", 0);
1349 SvREADONLY_on (sv_cbor); 1583 SvREADONLY_on (sv_cbor);
1584
1585 assert (("STRLEN must be an unsigned type", 0 <= (STRLEN)-1));
1350} 1586}
1351 1587
1352PROTOTYPES: DISABLE 1588PROTOTYPES: DISABLE
1353 1589
1354void CLONE (...) 1590void CLONE (...)
… …
1374 ALIAS: 1610 ALIAS:
1375 shrink = F_SHRINK 1611 shrink = F_SHRINK
1376 allow_unknown = F_ALLOW_UNKNOWN 1612 allow_unknown = F_ALLOW_UNKNOWN
1377 allow_sharing = F_ALLOW_SHARING 1613 allow_sharing = F_ALLOW_SHARING
1378 allow_cycles = F_ALLOW_CYCLES 1614 allow_cycles = F_ALLOW_CYCLES
1615 forbid_objects = F_FORBID_OBJECTS
1379 pack_strings = F_PACK_STRINGS 1616 pack_strings = F_PACK_STRINGS
1617 text_keys = F_TEXT_KEYS
1618 text_strings = F_TEXT_STRINGS
1380 validate_utf8 = F_VALIDATE_UTF8 1619 validate_utf8 = F_VALIDATE_UTF8
1381 PPCODE: 1620 PPCODE:
1382{ 1621{
1383 if (enable) 1622 if (enable)
1384 self->flags |= ix; 1623 self->flags |= ix;
… …
1392 ALIAS: 1631 ALIAS:
1393 get_shrink = F_SHRINK 1632 get_shrink = F_SHRINK
1394 get_allow_unknown = F_ALLOW_UNKNOWN 1633 get_allow_unknown = F_ALLOW_UNKNOWN
1395 get_allow_sharing = F_ALLOW_SHARING 1634 get_allow_sharing = F_ALLOW_SHARING
1396 get_allow_cycles = F_ALLOW_CYCLES 1635 get_allow_cycles = F_ALLOW_CYCLES
1636 get_forbid_objects = F_FORBID_OBJECTS
1397 get_pack_strings = F_PACK_STRINGS 1637 get_pack_strings = F_PACK_STRINGS
1638 get_text_keys = F_TEXT_KEYS
1639 get_text_strings = F_TEXT_STRINGS
1398 get_validate_utf8 = F_VALIDATE_UTF8 1640 get_validate_utf8 = F_VALIDATE_UTF8
1399 PPCODE: 1641 PPCODE:
1400 XPUSHs (boolSV (self->flags & ix)); 1642 XPUSHs (boolSV (self->flags & ix));
1401 1643
1402void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) 1644void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
… …
1532 cbor_init (&cbor); 1774 cbor_init (&cbor);
1533 PUTBACK; cborstr = decode_cbor (cborstr, &cbor, 0); SPAGAIN; 1775 PUTBACK; cborstr = decode_cbor (cborstr, &cbor, 0); SPAGAIN;
1534 XPUSHs (cborstr); 1776 XPUSHs (cborstr);
1535} 1777}
1536 1778
1779#ifdef __AFL_COMPILER
1780
1781void
1782afl_init ()
1783 CODE:
1784 __AFL_INIT ();
1785
1786int
1787afl_loop (unsigned int count = 10000)
1788 CODE:
1789 RETVAL = __AFL_LOOP (count);
1790 OUTPUT:
1791 RETVAL
1792
1793#endif
1794

Diff Legend

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