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 1.52 by root, Sun Apr 24 19:31:55 2016 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
105#define F_UTF8_STRINGS 0x00000020UL 123#define F_TEXT_STRINGS 0x00000080UL
106#define F_VALIDATE_UTF8 0x00000040UL 124#define F_VALIDATE_UTF8 0x00000100UL
107 125
108#define INIT_SIZE 32 // initial scalar size to be allocated 126#define INIT_SIZE 32 // initial scalar size to be allocated
109 127
110#define SB do { 128#define SB do {
111#define SE } while (0) 129#define SE } while (0)
… …
184 202
185// minimum length of a string to be registered for stringref 203// minimum length of a string to be registered for stringref
186ecb_inline int 204ecb_inline int
187minimum_string_length (UV idx) 205minimum_string_length (UV idx)
188{ 206{
189 return idx > 23 207 return idx <= 23 ? 3
190 ? idx > 0xffU 208 : idx <= 0xffU ? 4
191 ? idx > 0xffffU 209 : idx <= 0xffffU ? 5
192 ? idx > 0xffffffffU 210 : idx <= 0xffffffffU ? 7
193 ? 11 211 : 11;
194 : 7
195 : 5
196 : 4
197 : 3;
198} 212}
199 213
200///////////////////////////////////////////////////////////////////////////// 214/////////////////////////////////////////////////////////////////////////////
201// encoder 215// encoder
202 216
… …
215} enc_t; 229} enc_t;
216 230
217ecb_inline void 231ecb_inline void
218need (enc_t *enc, STRLEN len) 232need (enc_t *enc, STRLEN len)
219{ 233{
220 if (ecb_expect_false (enc->cur + len >= enc->end)) 234 if (ecb_expect_false ((uintptr_t)(enc->end - enc->cur) < len))
221 { 235 {
222 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv); 236 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
223 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1); 237 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
224 enc->cur = SvPVX (enc->sv) + cur; 238 enc->cur = SvPVX (enc->sv) + cur;
225 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; 239 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
… …
231{ 245{
232 need (enc, 1); 246 need (enc, 1);
233 *enc->cur++ = ch; 247 *enc->cur++ = ch;
234} 248}
235 249
250// used for tags, intregers, element counts and so on
236static void 251static void
237encode_uint (enc_t *enc, int major, UV len) 252encode_uint (enc_t *enc, int major, UV len)
238{ 253{
239 need (enc, 9); 254 need (enc, 9);
240 255
… …
271 *enc->cur++ = len >> 8; 286 *enc->cur++ = len >> 8;
272 *enc->cur++ = len; 287 *enc->cur++ = len;
273 } 288 }
274} 289}
275 290
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
276ecb_inline void 303ecb_inline void
277encode_tag (enc_t *enc, UV tag) 304encode_tag (enc_t *enc, UV tag)
278{ 305{
279 encode_uint (enc, MAJOR_TAG, tag); 306 encode_uint (enc, MAJOR_TAG, tag);
280} 307}
281 308
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
282ecb_inline void 332ecb_inline void
283encode_str (enc_t *enc, int utf8, char *str, STRLEN len) 333encode_str (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len)
284{ 334{
285 if (ecb_expect_false (enc->cbor.flags & F_UTF8_STRINGS)) 335 if (ecb_expect_false (upgrade_utf8))
286 if (!utf8) 336 if (!utf8)
287 { 337 {
288 // exceptional path for bytze strings that need to be utf8-encoded 338 encode_str_utf8 (enc, utf8, str, len);
289 STRLEN ulen = len;
290 U8 *p, *pend = (U8 *)str + len;
291
292 for (p = (U8 *)str; p < pend; ++p)
293 ulen += *p >> 7; // count set high bits
294
295 encode_uint (enc, MAJOR_TEXT, ulen);
296
297 need (enc, ulen);
298 for (p = (U8 *)str; p < pend; ++p)
299 if (*p < 0x80)
300 *enc->cur++ = *p;
301 else
302 {
303 *enc->cur++ = 0xc0 + (*p >> 6);
304 *enc->cur++ = 0x80 + (*p & 63);
305 }
306
307 return; 339 return;
308 } 340 }
309 341
310 encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len); 342 encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len);
311 need (enc, len); 343 need (enc, len);
312 memcpy (enc->cur, str, len); 344 memcpy (enc->cur, str, len);
313 enc->cur += len; 345 enc->cur += len;
314} 346}
315 347
316static void 348ecb_inline void
317encode_strref (enc_t *enc, int utf8, char *str, STRLEN len) 349encode_strref (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len)
318{ 350{
319 if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS)) 351 if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS))
320 { 352 {
321 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1); 353 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1);
322 354
… …
333 sv_setuv (*svp, enc->stringref_idx); 365 sv_setuv (*svp, enc->stringref_idx);
334 ++enc->stringref_idx; 366 ++enc->stringref_idx;
335 } 367 }
336 } 368 }
337 369
338 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 }
339} 460}
340 461
341static void encode_sv (enc_t *enc, SV *sv); 462static void encode_sv (enc_t *enc, SV *sv);
342 463
343static void 464static void
… …
350 471
351 ++enc->depth; 472 ++enc->depth;
352 473
353 encode_uint (enc, MAJOR_ARRAY, len + 1); 474 encode_uint (enc, MAJOR_ARRAY, len + 1);
354 475
355 if (SvMAGICAL (av)) 476 if (ecb_expect_false (SvMAGICAL (av)))
356 for (i = 0; i <= len; ++i) 477 for (i = 0; i <= len; ++i)
357 { 478 {
358 SV **svp = av_fetch (av, i, 0); 479 SV **svp = av_fetch (av, i, 0);
359 encode_sv (enc, svp ? *svp : &PL_sv_undef); 480 encode_sv (enc, svp ? *svp : &PL_sv_undef);
360 } 481 }
… …
379 ++enc->depth; 500 ++enc->depth;
380 501
381 int pairs = hv_iterinit (hv); 502 int pairs = hv_iterinit (hv);
382 int mg = SvMAGICAL (hv); 503 int mg = SvMAGICAL (hv);
383 504
384 if (mg) 505 if (ecb_expect_false (mg))
385 encode_ch (enc, MAJOR_MAP | MINOR_INDEF); 506 encode_ch (enc, MAJOR_MAP | MINOR_INDEF);
386 else 507 else
387 encode_uint (enc, MAJOR_MAP, pairs); 508 encode_uint (enc, MAJOR_MAP, pairs);
388 509
389 while ((he = hv_iternext (hv))) 510 while ((he = hv_iternext (hv)))
390 { 511 {
391 if (HeKLEN (he) == HEf_SVKEY) 512 if (HeKLEN (he) == HEf_SVKEY)
392 encode_sv (enc, HeSVKEY (he)); 513 encode_sv (enc, HeSVKEY (he));
393 else 514 else
394 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));
395 516
396 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));
397 } 518 }
398 519
399 if (mg) 520 if (ecb_expect_false (mg))
400 encode_ch (enc, MAJOR_MISC | MINOR_INDEF); 521 encode_ch (enc, MAJOR_MISC | MINOR_INDEF);
401 522
402 --enc->depth; 523 --enc->depth;
403} 524}
404 525
… …
437 else if (stash == tagged_stash) 558 else if (stash == tagged_stash)
438 { 559 {
439 if (svt != SVt_PVAV) 560 if (svt != SVt_PVAV)
440 croak ("encountered CBOR::XS::Tagged object that isn't an array"); 561 croak ("encountered CBOR::XS::Tagged object that isn't an array");
441 562
563 switch (av_len ((AV *)sv))
564 {
565 case 2-1:
566 // actually a tagged value
442 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)));
443 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 }
444 579
445 return; 580 return;
446 } 581 }
447 } 582 }
448 583
449 if (ecb_expect_false (SvREFCNT (sv) > 1) 584 if (ecb_expect_false (SvREFCNT (sv) > 1)
450 && ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING)) 585 && ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING))
451 { 586 {
452 if (!enc->shareable) 587 if (ecb_expect_false (!enc->shareable))
453 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ()); 588 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
454 589
455 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1); 590 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
456 591
457 if (SvOK (*svp)) 592 if (SvOK (*svp))
… …
471 if (ecb_expect_false (SvOBJECT (sv))) 606 if (ecb_expect_false (SvOBJECT (sv)))
472 { 607 {
473 HV *stash = SvSTASH (sv); 608 HV *stash = SvSTASH (sv);
474 GV *method; 609 GV *method;
475 610
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))));
476 if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0))) 614 else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
477 { 615 {
478 dSP; 616 dSP;
479 617
480 ENTER; SAVETMPS; 618 ENTER; SAVETMPS;
481 PUSHMARK (SP); 619 PUSHMARK (SP);
… …
500 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0) 638 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)
501 { 639 {
502 dSP; 640 dSP;
503 641
504 ENTER; SAVETMPS; 642 ENTER; SAVETMPS;
505 SAVESTACK_POS ();
506 PUSHMARK (SP); 643 PUSHMARK (SP);
507 EXTEND (SP, 2); 644 EXTEND (SP, 2);
508 // 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
509 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); 646 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
510 PUSHs (sv_cbor); 647 PUSHs (sv_cbor);
… …
517 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv) 654 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)
518 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));
519 656
520 encode_tag (enc, CBOR_TAG_PERL_OBJECT); 657 encode_tag (enc, CBOR_TAG_PERL_OBJECT);
521 encode_uint (enc, MAJOR_ARRAY, count + 1); 658 encode_uint (enc, MAJOR_ARRAY, count + 1);
522 encode_strref (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash)); 659 encode_strref (enc, 0, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
523 660
524 while (count) 661 {
662 int i;
663
664 for (i = 0; i < count; ++i)
525 encode_sv (enc, SP[1 - count--]); 665 encode_sv (enc, SP[i + 1 - count]);
666
667 SP -= count;
668 }
526 669
527 PUTBACK; 670 PUTBACK;
528 671
529 FREETMPS; LEAVE; 672 FREETMPS; LEAVE;
530 } 673 }
… …
552 695
553 if (ecb_expect_false (nv == (NV)(U32)nv)) 696 if (ecb_expect_false (nv == (NV)(U32)nv))
554 encode_uint (enc, MAJOR_POS_INT, (U32)nv); 697 encode_uint (enc, MAJOR_POS_INT, (U32)nv);
555 //TODO: maybe I32? 698 //TODO: maybe I32?
556 else if (ecb_expect_false (nv == (float)nv)) 699 else if (ecb_expect_false (nv == (float)nv))
557 { 700 encode_float32 (enc, nv);
558 uint32_t fp = ecb_float_to_binary32 (nv);
559
560 *enc->cur++ = MAJOR_MISC | MISC_FLOAT32;
561
562 if (!ecb_big_endian ())
563 fp = ecb_bswap32 (fp);
564
565 memcpy (enc->cur, &fp, 4);
566 enc->cur += 4;
567 }
568 else 701 else
569 { 702 encode_float64 (enc, nv);
570 uint64_t fp = ecb_double_to_binary64 (nv);
571
572 *enc->cur++ = MAJOR_MISC | MISC_FLOAT64;
573
574 if (!ecb_big_endian ())
575 fp = ecb_bswap64 (fp);
576
577 memcpy (enc->cur, &fp, 8);
578 enc->cur += 8;
579 }
580} 703}
581 704
582static void 705static void
583encode_sv (enc_t *enc, SV *sv) 706encode_sv (enc_t *enc, SV *sv)
584{ 707{
… …
586 709
587 if (SvPOKp (sv)) 710 if (SvPOKp (sv))
588 { 711 {
589 STRLEN len; 712 STRLEN len;
590 char *str = SvPV (sv, len); 713 char *str = SvPV (sv, len);
591 encode_strref (enc, SvUTF8 (sv), str, len); 714 encode_strref (enc, enc->cbor.flags & F_TEXT_STRINGS, SvUTF8 (sv), str, len);
592 } 715 }
593 else if (SvNOKp (sv)) 716 else if (SvNOKp (sv))
594 encode_nv (enc, sv); 717 encode_nv (enc, sv);
595 else if (SvIOKp (sv)) 718 else if (SvIOKp (sv))
596 { 719 encode_int (e,v sv);
597 if (SvIsUV (sv))
598 encode_uint (enc, MAJOR_POS_INT, SvUVX (sv));
599 else if (SvIVX (sv) >= 0)
600 encode_uint (enc, MAJOR_POS_INT, SvIVX (sv));
601 else
602 encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1));
603 }
604 else if (SvROK (sv)) 720 else if (SvROK (sv))
605 encode_rv (enc, SvRV (sv)); 721 encode_rv (enc, SvRV (sv));
606 else if (!SvOK (sv)) 722 else if (!SvOK (sv))
607 encode_ch (enc, MAJOR_MISC | SIMPLE_NULL); 723 encode_ch (enc, MAJOR_MISC | SIMPLE_NULL);
608 else if (enc->cbor.flags & F_ALLOW_UNKNOWN) 724 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
… …
615static SV * 731static SV *
616encode_cbor (SV *scalar, CBOR *cbor) 732encode_cbor (SV *scalar, CBOR *cbor)
617{ 733{
618 enc_t enc = { 0 }; 734 enc_t enc = { 0 };
619 735
620 enc.cbor = *cbor; 736 enc.cbor = *cbor;
621 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 737 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
622 enc.cur = SvPVX (enc.sv); 738 enc.cur = SvPVX (enc.sv);
623 enc.end = SvEND (enc.sv); 739 enc.end = SvEND (enc.sv);
624 740
625 SvPOK_only (enc.sv); 741 SvPOK_only (enc.sv);
626 742
627 if (cbor->flags & F_PACK_STRINGS) 743 if (cbor->flags & F_PACK_STRINGS)
628 { 744 {
… …
655 U32 depth; // recursion depth 771 U32 depth; // recursion depth
656 U32 maxdepth; // recursion depth limit 772 U32 maxdepth; // recursion depth limit
657 AV *shareable; 773 AV *shareable;
658 AV *stringref; 774 AV *stringref;
659 SV *decode_tagged; 775 SV *decode_tagged;
776 SV *err_sv; // optional sv for error, needs to be freed
660} dec_t; 777} dec_t;
661 778
662#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);
663 786
664#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;
665 790
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
666#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)
667#define DEC_DEC_DEPTH --dec->depth 816#define DEC_DEC_DEPTH --dec->depth
668 817
669static UV 818static UV
670decode_uint (dec_t *dec) 819decode_uint (dec_t *dec)
671{ 820{
… …
737 886
738 for (;;) 887 for (;;)
739 { 888 {
740 WANT (1); 889 WANT (1);
741 890
742 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF)) 891 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF) || dec->err)
743 { 892 {
744 ++dec->cur; 893 ++dec->cur;
745 break; 894 break;
746 } 895 }
747 896
748 av_push (av, decode_sv (dec)); 897 av_push (av, decode_sv (dec));
749 } 898 }
750 } 899 }
751 else 900 else
752 { 901 {
753 int i, len = decode_uint (dec); 902 UV i, len = decode_uint (dec);
754 903
755 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
756 av_fill (av, len - 1); 905 av_fill (av, len - 1);
757 906
758 for (i = 0; i < len; ++i) 907 for (i = 0; i < len; ++i)
… …
761 910
762 DEC_DEC_DEPTH; 911 DEC_DEC_DEPTH;
763 return newRV_noinc ((SV *)av); 912 return newRV_noinc ((SV *)av);
764 913
765fail: 914fail:
766 SvREFCNT_dec (av); 915 SvREFCNT_dec_NN (av);
767 DEC_DEC_DEPTH; 916 DEC_DEC_DEPTH;
768 return &PL_sv_undef; 917 return &PL_sv_undef;
769} 918}
770 919
771static void 920static void
… …
775 // byte or utf-8 strings as keys, but only when !stringref 924 // byte or utf-8 strings as keys, but only when !stringref
776 925
777 if (ecb_expect_true (!dec->stringref)) 926 if (ecb_expect_true (!dec->stringref))
778 if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8)) 927 if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8))
779 { 928 {
780 I32 len = decode_uint (dec); 929 STRLEN len = decode_uint (dec);
781 char *key = (char *)dec->cur; 930 char *key = (char *)dec->cur;
782 931
783 WANT (len); 932 WANT (len);
784 dec->cur += len; 933 dec->cur += len;
785 934
… …
787 936
788 return; 937 return;
789 } 938 }
790 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))
791 { 940 {
792 I32 len = decode_uint (dec); 941 STRLEN len = decode_uint (dec);
793 char *key = (char *)dec->cur; 942 char *key = (char *)dec->cur;
794 943
795 WANT (len); 944 WANT (len);
796 dec->cur += len; 945 dec->cur += len;
797 946
… …
805 } 954 }
806 955
807 SV *k = decode_sv (dec); 956 SV *k = decode_sv (dec);
808 SV *v = decode_sv (dec); 957 SV *v = decode_sv (dec);
809 958
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
810 hv_store_ent (hv, k, v, 0); 987 hv_store_ent (hv, k, v, 0);
811 SvREFCNT_dec (k); 988 SvREFCNT_dec_NN (k);
812 989
813fail: 990fail:
814 ; 991 ;
815} 992}
816 993
… …
827 1004
828 for (;;) 1005 for (;;)
829 { 1006 {
830 WANT (1); 1007 WANT (1);
831 1008
832 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF)) 1009 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF) || dec->err)
833 { 1010 {
834 ++dec->cur; 1011 ++dec->cur;
835 break; 1012 break;
836 } 1013 }
837 1014
838 decode_he (dec, hv); 1015 decode_he (dec, hv);
839 } 1016 }
840 } 1017 }
841 else 1018 else
842 { 1019 {
843 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
844 1023
845 while (pairs--) 1024 while (pairs--)
846 decode_he (dec, hv); 1025 decode_he (dec, hv);
847 } 1026 }
848 1027
849 DEC_DEC_DEPTH; 1028 DEC_DEC_DEPTH;
850 return newRV_noinc ((SV *)hv); 1029 return newRV_noinc ((SV *)hv);
851 1030
852fail: 1031fail:
853 SvREFCNT_dec (hv); 1032 SvREFCNT_dec_NN (hv);
854 DEC_DEC_DEPTH; 1033 DEC_DEC_DEPTH;
855 return &PL_sv_undef; 1034 return &PL_sv_undef;
856} 1035}
857 1036
858static SV * 1037static SV *
859decode_str (dec_t *dec, int utf8) 1038decode_str (dec_t *dec, int utf8)
860{ 1039{
861 SV *sv = 0; 1040 SV *sv = 0;
862 1041
863 if ((*dec->cur & MINOR_MASK) == MINOR_INDEF) 1042 if (ecb_expect_false ((*dec->cur & MINOR_MASK) == MINOR_INDEF))
864 { 1043 {
865 // indefinite length strings 1044 // indefinite length strings
866 ++dec->cur; 1045 ++dec->cur;
867 1046
868 U8 major = *dec->cur & MAJOR_MISC; 1047 U8 major = *dec->cur & MAJOR_MISC;
… …
936 sv = newRV_noinc (decode_sv (dec)); 1115 sv = newRV_noinc (decode_sv (dec));
937 break; 1116 break;
938 1117
939 case CBOR_TAG_STRINGREF_NAMESPACE: 1118 case CBOR_TAG_STRINGREF_NAMESPACE:
940 { 1119 {
1120 // do not use SAVETMPS/FREETMPS, as these will
1121 // erase mortalised caches, e.g. "shareable"
941 ENTER; SAVETMPS; 1122 ENTER;
942 1123
943 SAVESPTR (dec->stringref); 1124 SAVESPTR (dec->stringref);
944 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ()); 1125 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ());
945 1126
946 sv = decode_sv (dec); 1127 sv = decode_sv (dec);
947 1128
948 FREETMPS; LEAVE; 1129 LEAVE;
949 } 1130 }
950 break; 1131 break;
951 1132
952 case CBOR_TAG_STRINGREF: 1133 case CBOR_TAG_STRINGREF:
953 { 1134 {
954 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT)) 1135 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
955 ERR ("corrupted CBOR data (stringref index not an unsigned integer)"); 1136 ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
956 1137
957 UV idx = decode_uint (dec); 1138 UV idx = decode_uint (dec);
958 1139
959 if (!dec->stringref || (int)idx > AvFILLp (dec->stringref)) 1140 if (!dec->stringref || idx >= (UV)(1 + AvFILLp (dec->stringref)))
960 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)");
961 1142
962 sv = newSVsv (AvARRAY (dec->stringref)[idx]); 1143 sv = newSVsv (AvARRAY (dec->stringref)[idx]);
963 } 1144 }
964 break; 1145 break;
… …
992 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT)) 1173 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
993 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)"); 1174 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
994 1175
995 UV idx = decode_uint (dec); 1176 UV idx = decode_uint (dec);
996 1177
997 if (!dec->shareable || (int)idx > AvFILLp (dec->shareable)) 1178 if (!dec->shareable || idx >= (UV)(1 + AvFILLp (dec->shareable)))
998 ERR ("corrupted CBOR data (sharedref index out of bounds)"); 1179 ERR ("corrupted CBOR data (sharedref index out of bounds)");
999 1180
1000 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]); 1181 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]);
1001 1182
1002 if (sv == &PL_sv_undef) 1183 if (sv == &PL_sv_undef)
… …
1004 } 1185 }
1005 break; 1186 break;
1006 1187
1007 case CBOR_TAG_PERL_OBJECT: 1188 case CBOR_TAG_PERL_OBJECT:
1008 { 1189 {
1190 if (dec->cbor.flags & F_FORBID_OBJECTS)
1191 goto filter;
1192
1009 sv = decode_sv (dec); 1193 sv = decode_sv (dec);
1010 1194
1011 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV) 1195 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
1012 ERR ("corrupted CBOR data (non-array perl object)"); 1196 ERR ("corrupted CBOR data (non-array perl object)");
1013 1197
… …
1042 SPAGAIN; 1226 SPAGAIN;
1043 1227
1044 if (SvTRUE (ERRSV)) 1228 if (SvTRUE (ERRSV))
1045 { 1229 {
1046 FREETMPS; LEAVE; 1230 FREETMPS; LEAVE;
1047 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV)))); 1231 ERR_ERRSV;
1048 } 1232 }
1049 1233
1050 SvREFCNT_dec (sv); 1234 SvREFCNT_dec_NN (sv);
1051 sv = SvREFCNT_inc (POPs); 1235 sv = SvREFCNT_inc (POPs);
1052 1236
1053 PUTBACK; 1237 PUTBACK;
1054 1238
1055 FREETMPS; LEAVE; 1239 FREETMPS; LEAVE;
1056 } 1240 }
1057 break; 1241 break;
1058 1242
1059 default: 1243 default:
1244 filter:
1060 { 1245 {
1246 SV *tag_sv = newSVuv (tag);
1247
1061 sv = decode_sv (dec); 1248 sv = decode_sv (dec);
1062 1249
1063 dSP; 1250 dSP;
1064 ENTER; SAVETMPS; 1251 ENTER; SAVETMPS;
1065 SAVESTACK_POS ();
1066 PUSHMARK (SP); 1252 PUSHMARK (SP);
1067 EXTEND (SP, 2); 1253 EXTEND (SP, 2);
1068 PUSHs (newSVuv (tag)); 1254 PUSHs (tag_sv);
1069 PUSHs (sv); 1255 PUSHs (sv);
1070 1256
1071 PUTBACK; 1257 PUTBACK;
1072 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);
1073 SPAGAIN; 1259 SPAGAIN;
1074 1260
1075 if (SvTRUE (ERRSV)) 1261 if (SvTRUE (ERRSV))
1076 { 1262 {
1263 SvREFCNT_dec_NN (tag_sv);
1077 FREETMPS; LEAVE; 1264 FREETMPS; LEAVE;
1078 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV)))); 1265 ERR_ERRSV;
1079 } 1266 }
1080 1267
1081 if (count) 1268 if (count)
1082 { 1269 {
1270 SvREFCNT_dec_NN (tag_sv);
1083 SvREFCNT_dec (sv); 1271 SvREFCNT_dec_NN (sv);
1084 sv = SvREFCNT_inc (POPs); 1272 sv = SvREFCNT_inc_NN (TOPs);
1273 SP -= count;
1085 } 1274 }
1086 else 1275 else
1087 { 1276 {
1088 AV *av = newAV (); 1277 AV *av = newAV ();
1089 av_push (av, newSVuv (tag)); 1278 av_push (av, tag_sv);
1090 av_push (av, sv); 1279 av_push (av, sv);
1091 1280
1092 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 1281 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
1093 ? cbor_tagged_stash 1282 ? cbor_tagged_stash
1094 : gv_stashpv ("CBOR::XS::Tagged" , 1); 1283 : gv_stashpv ("CBOR::XS::Tagged" , 1);
… …
1223 1412
1224 if (dec.err) 1413 if (dec.err)
1225 { 1414 {
1226 if (dec.shareable) 1415 if (dec.shareable)
1227 { 1416 {
1228 // need to break cyclic links, which whould all be in shareable 1417 // need to break cyclic links, which would all be in shareable
1229 int i; 1418 int i;
1230 SV **svp; 1419 SV **svp;
1231 1420
1232 for (i = av_len (dec.shareable) + 1; i--; ) 1421 for (i = av_len (dec.shareable) + 1; i--; )
1233 if ((svp = av_fetch (dec.shareable, i, 0))) 1422 if ((svp = av_fetch (dec.shareable, i, 0)))
1234 sv_setsv (*svp, &PL_sv_undef); 1423 sv_setsv (*svp, &PL_sv_undef);
1235 } 1424 }
1236 1425
1237 SvREFCNT_dec (sv); 1426 SvREFCNT_dec_NN (sv);
1427
1428 if (dec.err_sv)
1429 sv_2mortal (dec.err_sv);
1430
1238 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);
1239 } 1432 }
1240 1433
1241 sv = sv_2mortal (sv); 1434 sv = sv_2mortal (sv);
1242 1435
… …
1386 1579
1387 default_filter = newSVpv ("CBOR::XS::default_filter", 0); 1580 default_filter = newSVpv ("CBOR::XS::default_filter", 0);
1388 1581
1389 sv_cbor = newSVpv ("CBOR", 0); 1582 sv_cbor = newSVpv ("CBOR", 0);
1390 SvREADONLY_on (sv_cbor); 1583 SvREADONLY_on (sv_cbor);
1584
1585 assert (("STRLEN must be an unsigned type", 0 <= (STRLEN)-1));
1391} 1586}
1392 1587
1393PROTOTYPES: DISABLE 1588PROTOTYPES: DISABLE
1394 1589
1395void CLONE (...) 1590void CLONE (...)
… …
1415 ALIAS: 1610 ALIAS:
1416 shrink = F_SHRINK 1611 shrink = F_SHRINK
1417 allow_unknown = F_ALLOW_UNKNOWN 1612 allow_unknown = F_ALLOW_UNKNOWN
1418 allow_sharing = F_ALLOW_SHARING 1613 allow_sharing = F_ALLOW_SHARING
1419 allow_cycles = F_ALLOW_CYCLES 1614 allow_cycles = F_ALLOW_CYCLES
1615 forbid_objects = F_FORBID_OBJECTS
1420 pack_strings = F_PACK_STRINGS 1616 pack_strings = F_PACK_STRINGS
1617 text_keys = F_TEXT_KEYS
1421 utf8_strings = F_UTF8_STRINGS 1618 text_strings = F_TEXT_STRINGS
1422 validate_utf8 = F_VALIDATE_UTF8 1619 validate_utf8 = F_VALIDATE_UTF8
1423 PPCODE: 1620 PPCODE:
1424{ 1621{
1425 if (enable) 1622 if (enable)
1426 self->flags |= ix; 1623 self->flags |= ix;
… …
1434 ALIAS: 1631 ALIAS:
1435 get_shrink = F_SHRINK 1632 get_shrink = F_SHRINK
1436 get_allow_unknown = F_ALLOW_UNKNOWN 1633 get_allow_unknown = F_ALLOW_UNKNOWN
1437 get_allow_sharing = F_ALLOW_SHARING 1634 get_allow_sharing = F_ALLOW_SHARING
1438 get_allow_cycles = F_ALLOW_CYCLES 1635 get_allow_cycles = F_ALLOW_CYCLES
1636 get_forbid_objects = F_FORBID_OBJECTS
1439 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
1440 get_validate_utf8 = F_VALIDATE_UTF8 1640 get_validate_utf8 = F_VALIDATE_UTF8
1441 PPCODE: 1641 PPCODE:
1442 XPUSHs (boolSV (self->flags & ix)); 1642 XPUSHs (boolSV (self->flags & ix));
1443 1643
1444void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) 1644void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
… …
1574 cbor_init (&cbor); 1774 cbor_init (&cbor);
1575 PUTBACK; cborstr = decode_cbor (cborstr, &cbor, 0); SPAGAIN; 1775 PUTBACK; cborstr = decode_cbor (cborstr, &cbor, 0); SPAGAIN;
1576 XPUSHs (cborstr); 1776 XPUSHs (cborstr);
1577} 1777}
1578 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