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 1.69 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
105#define F_UTF8_STRINGS 0x00000020UL 124#define F_TEXT_STRINGS 0x00000080UL
106#define F_VALIDATE_UTF8 0x00000040UL 125#define F_VALIDATE_UTF8 0x00000100UL
107 126
108#define INIT_SIZE 32 // initial scalar size to be allocated 127#define INIT_SIZE 32 // initial scalar size to be allocated
109 128
110#define SB do { 129#define SB do {
111#define SE } while (0) 130#define SE } while (0)
… …
184 203
185// minimum length of a string to be registered for stringref 204// minimum length of a string to be registered for stringref
186ecb_inline int 205ecb_inline int
187minimum_string_length (UV idx) 206minimum_string_length (UV idx)
188{ 207{
189 return idx > 23 208 return idx <= 23 ? 3
190 ? idx > 0xffU 209 : idx <= 0xffU ? 4
191 ? idx > 0xffffU 210 : idx <= 0xffffU ? 5
192 ? idx > 0xffffffffU 211 : idx <= 0xffffffffU ? 7
193 ? 11 212 : 11;
194 : 7
195 : 5
196 : 4
197 : 3;
198} 213}
199 214
200///////////////////////////////////////////////////////////////////////////// 215/////////////////////////////////////////////////////////////////////////////
201// encoder 216// encoder
202 217
… …
215} enc_t; 230} enc_t;
216 231
217ecb_inline void 232ecb_inline void
218need (enc_t *enc, STRLEN len) 233need (enc_t *enc, STRLEN len)
219{ 234{
220 if (ecb_expect_false (enc->cur + len >= enc->end)) 235 if (ecb_expect_false ((uintptr_t)(enc->end - enc->cur) < len))
221 { 236 {
222 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv); 237 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
223 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1); 238 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
224 enc->cur = SvPVX (enc->sv) + cur; 239 enc->cur = SvPVX (enc->sv) + cur;
225 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; 240 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
… …
231{ 246{
232 need (enc, 1); 247 need (enc, 1);
233 *enc->cur++ = ch; 248 *enc->cur++ = ch;
234} 249}
235 250
251// used for tags, intregers, element counts and so on
236static void 252static void
237encode_uint (enc_t *enc, int major, UV len) 253encode_uint (enc_t *enc, int major, UV len)
238{ 254{
239 need (enc, 9); 255 need (enc, 9);
240 256
… …
271 *enc->cur++ = len >> 8; 287 *enc->cur++ = len >> 8;
272 *enc->cur++ = len; 288 *enc->cur++ = len;
273 } 289 }
274} 290}
275 291
292// encodes a perl value into a CBOR integer
293ecb_inline void
294encode_int (enc_t *enc, SV *sv)
295{
296 if (SvIsUV (sv))
297 encode_uint (enc, MAJOR_POS_INT, SvUVX (sv));
298 else if (SvIVX (sv) >= 0)
299 encode_uint (enc, MAJOR_POS_INT, SvIVX (sv));
300 else
301 encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1));
302}
303
276ecb_inline void 304ecb_inline void
277encode_tag (enc_t *enc, UV tag) 305encode_tag (enc_t *enc, UV tag)
278{ 306{
279 encode_uint (enc, MAJOR_TAG, tag); 307 encode_uint (enc, MAJOR_TAG, tag);
280} 308}
281 309
310// exceptional (hopefully) slow path for byte strings that need to be utf8-encoded
311ecb_noinline static void
312encode_str_utf8 (enc_t *enc, int utf8, char *str, STRLEN len)
313{
314 STRLEN ulen = len;
315 U8 *p, *pend = (U8 *)str + len;
316
317 for (p = (U8 *)str; p < pend; ++p)
318 ulen += *p >> 7; // count set high bits
319
320 encode_uint (enc, MAJOR_TEXT, ulen);
321
322 need (enc, ulen);
323 for (p = (U8 *)str; p < pend; ++p)
324 if (*p < 0x80)
325 *enc->cur++ = *p;
326 else
327 {
328 *enc->cur++ = 0xc0 + (*p >> 6);
329 *enc->cur++ = 0x80 + (*p & 63);
330 }
331}
332
282ecb_inline void 333ecb_inline void
283encode_str (enc_t *enc, int utf8, char *str, STRLEN len) 334encode_str (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len)
284{ 335{
285 if (ecb_expect_false (enc->cbor.flags & F_UTF8_STRINGS)) 336 if (ecb_expect_false (upgrade_utf8))
286 if (!utf8) 337 if (!utf8)
287 { 338 {
288 // exceptional path for bytze strings that need to be utf8-encoded 339 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; 340 return;
308 } 341 }
309 342
310 encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len); 343 encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len);
311 need (enc, len); 344 need (enc, len);
312 memcpy (enc->cur, str, len); 345 memcpy (enc->cur, str, len);
313 enc->cur += len; 346 enc->cur += len;
314} 347}
315 348
316static void 349ecb_inline void
317encode_strref (enc_t *enc, int utf8, char *str, STRLEN len) 350encode_strref (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len)
318{ 351{
319 if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS)) 352 if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS))
320 { 353 {
321 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1); 354 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1);
322 355
… …
333 sv_setuv (*svp, enc->stringref_idx); 366 sv_setuv (*svp, enc->stringref_idx);
334 ++enc->stringref_idx; 367 ++enc->stringref_idx;
335 } 368 }
336 } 369 }
337 370
338 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 }
339} 469}
340 470
341static void encode_sv (enc_t *enc, SV *sv); 471static void encode_sv (enc_t *enc, SV *sv);
342 472
343static void 473static void
… …
350 480
351 ++enc->depth; 481 ++enc->depth;
352 482
353 encode_uint (enc, MAJOR_ARRAY, len + 1); 483 encode_uint (enc, MAJOR_ARRAY, len + 1);
354 484
355 if (SvMAGICAL (av)) 485 if (ecb_expect_false (SvMAGICAL (av)))
356 for (i = 0; i <= len; ++i) 486 for (i = 0; i <= len; ++i)
357 { 487 {
358 SV **svp = av_fetch (av, i, 0); 488 SV **svp = av_fetch (av, i, 0);
359 encode_sv (enc, svp ? *svp : &PL_sv_undef); 489 encode_sv (enc, svp ? *svp : &PL_sv_undef);
360 } 490 }
… …
379 ++enc->depth; 509 ++enc->depth;
380 510
381 int pairs = hv_iterinit (hv); 511 int pairs = hv_iterinit (hv);
382 int mg = SvMAGICAL (hv); 512 int mg = SvMAGICAL (hv);
383 513
384 if (mg) 514 if (ecb_expect_false (mg))
385 encode_ch (enc, MAJOR_MAP | MINOR_INDEF); 515 encode_ch (enc, MAJOR_MAP | MINOR_INDEF);
386 else 516 else
387 encode_uint (enc, MAJOR_MAP, pairs); 517 encode_uint (enc, MAJOR_MAP, pairs);
388 518
389 while ((he = hv_iternext (hv))) 519 while ((he = hv_iternext (hv)))
390 { 520 {
391 if (HeKLEN (he) == HEf_SVKEY) 521 if (HeKLEN (he) == HEf_SVKEY)
392 encode_sv (enc, HeSVKEY (he)); 522 encode_sv (enc, HeSVKEY (he));
393 else 523 else
394 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));
395 525
396 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));
397 } 527 }
398 528
399 if (mg) 529 if (ecb_expect_false (mg))
400 encode_ch (enc, MAJOR_MISC | MINOR_INDEF); 530 encode_ch (enc, MAJOR_MISC | MINOR_INDEF);
401 531
402 --enc->depth; 532 --enc->depth;
403} 533}
404 534
… …
424 554
425 HV *stash = SvSTASH (sv); 555 HV *stash = SvSTASH (sv);
426 556
427 if (stash == boolean_stash) 557 if (stash == boolean_stash)
428 { 558 {
429 encode_ch (enc, SvIV (sv) ? MAJOR_MISC | SIMPLE_TRUE : MAJOR_MISC | SIMPLE_FALSE); 559 encode_bool (enc, SvIV (sv));
430 return; 560 return;
431 } 561 }
432 else if (stash == error_stash) 562 else if (stash == error_stash)
433 { 563 {
434 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF); 564 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
… …
437 else if (stash == tagged_stash) 567 else if (stash == tagged_stash)
438 { 568 {
439 if (svt != SVt_PVAV) 569 if (svt != SVt_PVAV)
440 croak ("encountered CBOR::XS::Tagged object that isn't an array"); 570 croak ("encountered CBOR::XS::Tagged object that isn't an array");
441 571
572 switch (av_len ((AV *)sv))
573 {
574 case 2-1:
575 // actually a tagged value
442 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)));
443 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 }
444 588
445 return; 589 return;
446 } 590 }
447 } 591 }
448 592
449 if (ecb_expect_false (SvREFCNT (sv) > 1) 593 if (ecb_expect_false (SvREFCNT (sv) > 1)
450 && ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING)) 594 && ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING))
451 { 595 {
452 if (!enc->shareable) 596 if (ecb_expect_false (!enc->shareable))
453 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ()); 597 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
454 598
455 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1); 599 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
456 600
457 if (SvOK (*svp)) 601 if (SvOK (*svp))
… …
471 if (ecb_expect_false (SvOBJECT (sv))) 615 if (ecb_expect_false (SvOBJECT (sv)))
472 { 616 {
473 HV *stash = SvSTASH (sv); 617 HV *stash = SvSTASH (sv);
474 GV *method; 618 GV *method;
475 619
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))));
476 if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0))) 623 else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
477 { 624 {
478 dSP; 625 dSP;
479 626
480 ENTER; SAVETMPS; 627 ENTER; SAVETMPS;
481 PUSHMARK (SP); 628 PUSHMARK (SP);
… …
500 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0) 647 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)
501 { 648 {
502 dSP; 649 dSP;
503 650
504 ENTER; SAVETMPS; 651 ENTER; SAVETMPS;
505 SAVESTACK_POS ();
506 PUSHMARK (SP); 652 PUSHMARK (SP);
507 EXTEND (SP, 2); 653 EXTEND (SP, 2);
508 // 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
509 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); 655 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
510 PUSHs (sv_cbor); 656 PUSHs (sv_cbor);
… …
517 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv) 663 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)); 664 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));
519 665
520 encode_tag (enc, CBOR_TAG_PERL_OBJECT); 666 encode_tag (enc, CBOR_TAG_PERL_OBJECT);
521 encode_uint (enc, MAJOR_ARRAY, count + 1); 667 encode_uint (enc, MAJOR_ARRAY, count + 1);
522 encode_strref (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash)); 668 encode_strref (enc, 0, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
523 669
524 while (count) 670 {
671 int i;
672
673 for (i = 0; i < count; ++i)
525 encode_sv (enc, SP[1 - count--]); 674 encode_sv (enc, SP[i + 1 - count]);
675
676 SP -= count;
677 }
526 678
527 PUTBACK; 679 PUTBACK;
528 680
529 FREETMPS; LEAVE; 681 FREETMPS; LEAVE;
530 } 682 }
… …
552 704
553 if (ecb_expect_false (nv == (NV)(U32)nv)) 705 if (ecb_expect_false (nv == (NV)(U32)nv))
554 encode_uint (enc, MAJOR_POS_INT, (U32)nv); 706 encode_uint (enc, MAJOR_POS_INT, (U32)nv);
555 //TODO: maybe I32? 707 //TODO: maybe I32?
556 else if (ecb_expect_false (nv == (float)nv)) 708 else if (ecb_expect_false (nv == (float)nv))
557 { 709 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 710 else
569 { 711 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} 712}
581 713
582static void 714static void
583encode_sv (enc_t *enc, SV *sv) 715encode_sv (enc_t *enc, SV *sv)
584{ 716{
… …
586 718
587 if (SvPOKp (sv)) 719 if (SvPOKp (sv))
588 { 720 {
589 STRLEN len; 721 STRLEN len;
590 char *str = SvPV (sv, len); 722 char *str = SvPV (sv, len);
591 encode_strref (enc, SvUTF8 (sv), str, len); 723 encode_strref (enc, enc->cbor.flags & F_TEXT_STRINGS, SvUTF8 (sv), str, len);
592 } 724 }
593 else if (SvNOKp (sv)) 725 else if (SvNOKp (sv))
594 encode_nv (enc, sv); 726 encode_nv (enc, sv);
595 else if (SvIOKp (sv)) 727 else if (SvIOKp (sv))
596 { 728 encode_int (enc, 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)) 729 else if (SvROK (sv))
605 encode_rv (enc, SvRV (sv)); 730 encode_rv (enc, SvRV (sv));
606 else if (!SvOK (sv)) 731 else if (!SvOK (sv))
607 encode_ch (enc, MAJOR_MISC | SIMPLE_NULL); 732 encode_ch (enc, MAJOR_MISC | SIMPLE_NULL);
608 else if (enc->cbor.flags & F_ALLOW_UNKNOWN) 733 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
… …
615static SV * 740static SV *
616encode_cbor (SV *scalar, CBOR *cbor) 741encode_cbor (SV *scalar, CBOR *cbor)
617{ 742{
618 enc_t enc = { 0 }; 743 enc_t enc = { 0 };
619 744
620 enc.cbor = *cbor; 745 enc.cbor = *cbor;
621 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 746 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
622 enc.cur = SvPVX (enc.sv); 747 enc.cur = SvPVX (enc.sv);
623 enc.end = SvEND (enc.sv); 748 enc.end = SvEND (enc.sv);
624 749
625 SvPOK_only (enc.sv); 750 SvPOK_only (enc.sv);
626 751
627 if (cbor->flags & F_PACK_STRINGS) 752 if (cbor->flags & F_PACK_STRINGS)
628 { 753 {
… …
655 U32 depth; // recursion depth 780 U32 depth; // recursion depth
656 U32 maxdepth; // recursion depth limit 781 U32 maxdepth; // recursion depth limit
657 AV *shareable; 782 AV *shareable;
658 AV *stringref; 783 AV *stringref;
659 SV *decode_tagged; 784 SV *decode_tagged;
785 SV *err_sv; // optional sv for error, needs to be freed
660} dec_t; 786} dec_t;
661 787
662#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);
663 795
664#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;
665 799
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
666#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)
667#define DEC_DEC_DEPTH --dec->depth 825#define DEC_DEC_DEPTH --dec->depth
668 826
669static UV 827static UV
670decode_uint (dec_t *dec) 828decode_uint (dec_t *dec)
671{ 829{
… …
737 895
738 for (;;) 896 for (;;)
739 { 897 {
740 WANT (1); 898 WANT (1);
741 899
742 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF)) 900 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF) || dec->err)
743 { 901 {
744 ++dec->cur; 902 ++dec->cur;
745 break; 903 break;
746 } 904 }
747 905
748 av_push (av, decode_sv (dec)); 906 av_push (av, decode_sv (dec));
749 } 907 }
750 } 908 }
751 else 909 else
752 { 910 {
753 int i, len = decode_uint (dec); 911 UV i, len = decode_uint (dec);
754 912
755 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
756 av_fill (av, len - 1); 914 av_fill (av, len - 1);
757 915
758 for (i = 0; i < len; ++i) 916 for (i = 0; i < len; ++i)
… …
761 919
762 DEC_DEC_DEPTH; 920 DEC_DEC_DEPTH;
763 return newRV_noinc ((SV *)av); 921 return newRV_noinc ((SV *)av);
764 922
765fail: 923fail:
766 SvREFCNT_dec (av); 924 SvREFCNT_dec_NN (av);
767 DEC_DEC_DEPTH; 925 DEC_DEC_DEPTH;
768 return &PL_sv_undef; 926 return &PL_sv_undef;
769} 927}
770 928
771static void 929static void
… …
775 // byte or utf-8 strings as keys, but only when !stringref 933 // byte or utf-8 strings as keys, but only when !stringref
776 934
777 if (ecb_expect_true (!dec->stringref)) 935 if (ecb_expect_true (!dec->stringref))
778 if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8)) 936 if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8))
779 { 937 {
780 I32 len = decode_uint (dec); 938 STRLEN len = decode_uint (dec);
781 char *key = (char *)dec->cur; 939 char *key = (char *)dec->cur;
782 940
783 WANT (len); 941 WANT (len);
784 dec->cur += len; 942 dec->cur += len;
785 943
… …
787 945
788 return; 946 return;
789 } 947 }
790 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))
791 { 949 {
792 I32 len = decode_uint (dec); 950 STRLEN len = decode_uint (dec);
793 char *key = (char *)dec->cur; 951 char *key = (char *)dec->cur;
794 952
795 WANT (len); 953 WANT (len);
796 dec->cur += len; 954 dec->cur += len;
797 955
… …
805 } 963 }
806 964
807 SV *k = decode_sv (dec); 965 SV *k = decode_sv (dec);
808 SV *v = decode_sv (dec); 966 SV *v = decode_sv (dec);
809 967
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
810 hv_store_ent (hv, k, v, 0); 996 hv_store_ent (hv, k, v, 0);
811 SvREFCNT_dec (k); 997 SvREFCNT_dec_NN (k);
812 998
813fail: 999fail:
814 ; 1000 ;
815} 1001}
816 1002
… …
827 1013
828 for (;;) 1014 for (;;)
829 { 1015 {
830 WANT (1); 1016 WANT (1);
831 1017
832 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF)) 1018 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF) || dec->err)
833 { 1019 {
834 ++dec->cur; 1020 ++dec->cur;
835 break; 1021 break;
836 } 1022 }
837 1023
838 decode_he (dec, hv); 1024 decode_he (dec, hv);
839 } 1025 }
840 } 1026 }
841 else 1027 else
842 { 1028 {
843 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
844 1032
845 while (pairs--) 1033 while (pairs--)
846 decode_he (dec, hv); 1034 decode_he (dec, hv);
847 } 1035 }
848 1036
849 DEC_DEC_DEPTH; 1037 DEC_DEC_DEPTH;
850 return newRV_noinc ((SV *)hv); 1038 return newRV_noinc ((SV *)hv);
851 1039
852fail: 1040fail:
853 SvREFCNT_dec (hv); 1041 SvREFCNT_dec_NN (hv);
854 DEC_DEC_DEPTH; 1042 DEC_DEC_DEPTH;
855 return &PL_sv_undef; 1043 return &PL_sv_undef;
856} 1044}
857 1045
858static SV * 1046static SV *
859decode_str (dec_t *dec, int utf8) 1047decode_str (dec_t *dec, int utf8)
860{ 1048{
861 SV *sv = 0; 1049 SV *sv = 0;
862 1050
863 if ((*dec->cur & MINOR_MASK) == MINOR_INDEF) 1051 if (ecb_expect_false ((*dec->cur & MINOR_MASK) == MINOR_INDEF))
864 { 1052 {
865 // indefinite length strings 1053 // indefinite length strings
866 ++dec->cur; 1054 ++dec->cur;
867 1055
868 U8 major = *dec->cur & MAJOR_MISC; 1056 U8 major = *dec->cur & MAJOR_MISC;
… …
936 sv = newRV_noinc (decode_sv (dec)); 1124 sv = newRV_noinc (decode_sv (dec));
937 break; 1125 break;
938 1126
939 case CBOR_TAG_STRINGREF_NAMESPACE: 1127 case CBOR_TAG_STRINGREF_NAMESPACE:
940 { 1128 {
1129 // do not use SAVETMPS/FREETMPS, as these will
1130 // erase mortalised caches, e.g. "shareable"
941 ENTER; SAVETMPS; 1131 ENTER;
942 1132
943 SAVESPTR (dec->stringref); 1133 SAVESPTR (dec->stringref);
944 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ()); 1134 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ());
945 1135
946 sv = decode_sv (dec); 1136 sv = decode_sv (dec);
947 1137
948 FREETMPS; LEAVE; 1138 LEAVE;
949 } 1139 }
950 break; 1140 break;
951 1141
952 case CBOR_TAG_STRINGREF: 1142 case CBOR_TAG_STRINGREF:
953 { 1143 {
954 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT)) 1144 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
955 ERR ("corrupted CBOR data (stringref index not an unsigned integer)"); 1145 ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
956 1146
957 UV idx = decode_uint (dec); 1147 UV idx = decode_uint (dec);
958 1148
959 if (!dec->stringref || (int)idx > AvFILLp (dec->stringref)) 1149 if (!dec->stringref || idx >= (UV)(1 + AvFILLp (dec->stringref)))
960 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)");
961 1151
962 sv = newSVsv (AvARRAY (dec->stringref)[idx]); 1152 sv = newSVsv (AvARRAY (dec->stringref)[idx]);
963 } 1153 }
964 break; 1154 break;
… …
992 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT)) 1182 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
993 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)"); 1183 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
994 1184
995 UV idx = decode_uint (dec); 1185 UV idx = decode_uint (dec);
996 1186
997 if (!dec->shareable || (int)idx > AvFILLp (dec->shareable)) 1187 if (!dec->shareable || idx >= (UV)(1 + AvFILLp (dec->shareable)))
998 ERR ("corrupted CBOR data (sharedref index out of bounds)"); 1188 ERR ("corrupted CBOR data (sharedref index out of bounds)");
999 1189
1000 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]); 1190 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]);
1001 1191
1002 if (sv == &PL_sv_undef) 1192 if (sv == &PL_sv_undef)
… …
1004 } 1194 }
1005 break; 1195 break;
1006 1196
1007 case CBOR_TAG_PERL_OBJECT: 1197 case CBOR_TAG_PERL_OBJECT:
1008 { 1198 {
1199 if (dec->cbor.flags & F_FORBID_OBJECTS)
1200 goto filter;
1201
1009 sv = decode_sv (dec); 1202 sv = decode_sv (dec);
1010 1203
1011 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV) 1204 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
1012 ERR ("corrupted CBOR data (non-array perl object)"); 1205 ERR ("corrupted CBOR data (non-array perl object)");
1013 1206
… …
1042 SPAGAIN; 1235 SPAGAIN;
1043 1236
1044 if (SvTRUE (ERRSV)) 1237 if (SvTRUE (ERRSV))
1045 { 1238 {
1046 FREETMPS; LEAVE; 1239 FREETMPS; LEAVE;
1047 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV)))); 1240 ERR_ERRSV;
1048 } 1241 }
1049 1242
1050 SvREFCNT_dec (sv); 1243 SvREFCNT_dec_NN (sv);
1051 sv = SvREFCNT_inc (POPs); 1244 sv = SvREFCNT_inc (POPs);
1052 1245
1053 PUTBACK; 1246 PUTBACK;
1054 1247
1055 FREETMPS; LEAVE; 1248 FREETMPS; LEAVE;
1056 } 1249 }
1057 break; 1250 break;
1058 1251
1059 default: 1252 default:
1253 filter:
1060 { 1254 {
1255 SV *tag_sv = newSVuv (tag);
1256
1061 sv = decode_sv (dec); 1257 sv = decode_sv (dec);
1062 1258
1063 dSP; 1259 dSP;
1064 ENTER; SAVETMPS; 1260 ENTER; SAVETMPS;
1065 SAVESTACK_POS ();
1066 PUSHMARK (SP); 1261 PUSHMARK (SP);
1067 EXTEND (SP, 2); 1262 EXTEND (SP, 2);
1068 PUSHs (newSVuv (tag)); 1263 PUSHs (tag_sv);
1069 PUSHs (sv); 1264 PUSHs (sv);
1070 1265
1071 PUTBACK; 1266 PUTBACK;
1072 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);
1073 SPAGAIN; 1268 SPAGAIN;
1074 1269
1075 if (SvTRUE (ERRSV)) 1270 if (SvTRUE (ERRSV))
1076 { 1271 {
1272 SvREFCNT_dec_NN (tag_sv);
1077 FREETMPS; LEAVE; 1273 FREETMPS; LEAVE;
1078 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV)))); 1274 ERR_ERRSV;
1079 } 1275 }
1080 1276
1081 if (count) 1277 if (count)
1082 { 1278 {
1279 SvREFCNT_dec_NN (tag_sv);
1083 SvREFCNT_dec (sv); 1280 SvREFCNT_dec_NN (sv);
1084 sv = SvREFCNT_inc (POPs); 1281 sv = SvREFCNT_inc_NN (TOPs);
1282 SP -= count;
1085 } 1283 }
1086 else 1284 else
1087 { 1285 {
1088 AV *av = newAV (); 1286 AV *av = newAV ();
1089 av_push (av, newSVuv (tag)); 1287 av_push (av, tag_sv);
1090 av_push (av, sv); 1288 av_push (av, sv);
1091 1289
1092 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 1290 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
1093 ? cbor_tagged_stash 1291 ? cbor_tagged_stash
1094 : gv_stashpv ("CBOR::XS::Tagged" , 1); 1292 : gv_stashpv ("CBOR::XS::Tagged" , 1);
… …
1223 1421
1224 if (dec.err) 1422 if (dec.err)
1225 { 1423 {
1226 if (dec.shareable) 1424 if (dec.shareable)
1227 { 1425 {
1228 // need to break cyclic links, which whould all be in shareable 1426 // need to break cyclic links, which would all be in shareable
1229 int i; 1427 int i;
1230 SV **svp; 1428 SV **svp;
1231 1429
1232 for (i = av_len (dec.shareable) + 1; i--; ) 1430 for (i = av_len (dec.shareable) + 1; i--; )
1233 if ((svp = av_fetch (dec.shareable, i, 0))) 1431 if ((svp = av_fetch (dec.shareable, i, 0)))
1234 sv_setsv (*svp, &PL_sv_undef); 1432 sv_setsv (*svp, &PL_sv_undef);
1235 } 1433 }
1236 1434
1237 SvREFCNT_dec (sv); 1435 SvREFCNT_dec_NN (sv);
1436
1437 if (dec.err_sv)
1438 sv_2mortal (dec.err_sv);
1439
1238 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);
1239 } 1441 }
1240 1442
1241 sv = sv_2mortal (sv); 1443 sv = sv_2mortal (sv);
1242 1444
… …
1386 1588
1387 default_filter = newSVpv ("CBOR::XS::default_filter", 0); 1589 default_filter = newSVpv ("CBOR::XS::default_filter", 0);
1388 1590
1389 sv_cbor = newSVpv ("CBOR", 0); 1591 sv_cbor = newSVpv ("CBOR", 0);
1390 SvREADONLY_on (sv_cbor); 1592 SvREADONLY_on (sv_cbor);
1593
1594 assert (("STRLEN must be an unsigned type", 0 <= (STRLEN)-1));
1391} 1595}
1392 1596
1393PROTOTYPES: DISABLE 1597PROTOTYPES: DISABLE
1394 1598
1395void CLONE (...) 1599void CLONE (...)
… …
1415 ALIAS: 1619 ALIAS:
1416 shrink = F_SHRINK 1620 shrink = F_SHRINK
1417 allow_unknown = F_ALLOW_UNKNOWN 1621 allow_unknown = F_ALLOW_UNKNOWN
1418 allow_sharing = F_ALLOW_SHARING 1622 allow_sharing = F_ALLOW_SHARING
1419 allow_cycles = F_ALLOW_CYCLES 1623 allow_cycles = F_ALLOW_CYCLES
1624 forbid_objects = F_FORBID_OBJECTS
1420 pack_strings = F_PACK_STRINGS 1625 pack_strings = F_PACK_STRINGS
1626 text_keys = F_TEXT_KEYS
1421 utf8_strings = F_UTF8_STRINGS 1627 text_strings = F_TEXT_STRINGS
1422 validate_utf8 = F_VALIDATE_UTF8 1628 validate_utf8 = F_VALIDATE_UTF8
1423 PPCODE: 1629 PPCODE:
1424{ 1630{
1425 if (enable) 1631 if (enable)
1426 self->flags |= ix; 1632 self->flags |= ix;
… …
1434 ALIAS: 1640 ALIAS:
1435 get_shrink = F_SHRINK 1641 get_shrink = F_SHRINK
1436 get_allow_unknown = F_ALLOW_UNKNOWN 1642 get_allow_unknown = F_ALLOW_UNKNOWN
1437 get_allow_sharing = F_ALLOW_SHARING 1643 get_allow_sharing = F_ALLOW_SHARING
1438 get_allow_cycles = F_ALLOW_CYCLES 1644 get_allow_cycles = F_ALLOW_CYCLES
1645 get_forbid_objects = F_FORBID_OBJECTS
1439 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
1440 get_validate_utf8 = F_VALIDATE_UTF8 1649 get_validate_utf8 = F_VALIDATE_UTF8
1441 PPCODE: 1650 PPCODE:
1442 XPUSHs (boolSV (self->flags & ix)); 1651 XPUSHs (boolSV (self->flags & ix));
1443 1652
1444void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) 1653void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
… …
1574 cbor_init (&cbor); 1783 cbor_init (&cbor);
1575 PUTBACK; cborstr = decode_cbor (cborstr, &cbor, 0); SPAGAIN; 1784 PUTBACK; cborstr = decode_cbor (cborstr, &cbor, 0); SPAGAIN;
1576 XPUSHs (cborstr); 1785 XPUSHs (cborstr);
1577} 1786}
1578 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