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 1.70 by root, Fri Dec 4 02:57:14 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 AS_MAP = 7,
115 // possibly future enhancements: (generic) float, (generic) string
116};
117
100#define F_SHRINK 0x00000001UL 118#define F_SHRINK 0x00000001UL
101#define F_ALLOW_UNKNOWN 0x00000002UL 119#define F_ALLOW_UNKNOWN 0x00000002UL
102#define F_ALLOW_SHARING 0x00000004UL 120#define F_ALLOW_SHARING 0x00000004UL
103#define F_ALLOW_CYCLES 0x00000008UL 121#define F_ALLOW_CYCLES 0x00000008UL
122#define F_FORBID_OBJECTS 0x00000010UL
104#define F_PACK_STRINGS 0x00000010UL 123#define F_PACK_STRINGS 0x00000020UL
124#define F_TEXT_KEYS 0x00000040UL
105#define F_UTF8_STRINGS 0x00000020UL 125#define F_TEXT_STRINGS 0x00000080UL
106#define F_VALIDATE_UTF8 0x00000040UL 126#define F_VALIDATE_UTF8 0x00000100UL
107 127
108#define INIT_SIZE 32 // initial scalar size to be allocated 128#define INIT_SIZE 32 // initial scalar size to be allocated
109 129
110#define SB do { 130#define SB do {
111#define SE } while (0) 131#define SE } while (0)
… …
184 204
185// minimum length of a string to be registered for stringref 205// minimum length of a string to be registered for stringref
186ecb_inline int 206ecb_inline int
187minimum_string_length (UV idx) 207minimum_string_length (UV idx)
188{ 208{
189 return idx > 23 209 return idx <= 23 ? 3
190 ? idx > 0xffU 210 : idx <= 0xffU ? 4
191 ? idx > 0xffffU 211 : idx <= 0xffffU ? 5
192 ? idx > 0xffffffffU 212 : idx <= 0xffffffffU ? 7
193 ? 11 213 : 11;
194 : 7
195 : 5
196 : 4
197 : 3;
198} 214}
199 215
200///////////////////////////////////////////////////////////////////////////// 216/////////////////////////////////////////////////////////////////////////////
201// encoder 217// encoder
202 218
… …
215} enc_t; 231} enc_t;
216 232
217ecb_inline void 233ecb_inline void
218need (enc_t *enc, STRLEN len) 234need (enc_t *enc, STRLEN len)
219{ 235{
220 if (ecb_expect_false (enc->cur + len >= enc->end)) 236 if (ecb_expect_false ((uintptr_t)(enc->end - enc->cur) < len))
221 { 237 {
222 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv); 238 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
223 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1); 239 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
224 enc->cur = SvPVX (enc->sv) + cur; 240 enc->cur = SvPVX (enc->sv) + cur;
225 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; 241 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
226 } 242 }
227} 243}
228 244
245static void encode_sv (enc_t *enc, SV *sv);
246
229ecb_inline void 247ecb_inline void
230encode_ch (enc_t *enc, char ch) 248encode_ch (enc_t *enc, char ch)
231{ 249{
232 need (enc, 1); 250 need (enc, 1);
233 *enc->cur++ = ch; 251 *enc->cur++ = ch;
234} 252}
235 253
254// used for tags, intregers, element counts and so on
236static void 255static void
237encode_uint (enc_t *enc, int major, UV len) 256encode_uint (enc_t *enc, int major, UV len)
238{ 257{
239 need (enc, 9); 258 need (enc, 9);
240 259
… …
271 *enc->cur++ = len >> 8; 290 *enc->cur++ = len >> 8;
272 *enc->cur++ = len; 291 *enc->cur++ = len;
273 } 292 }
274} 293}
275 294
295// encodes a perl value into a CBOR integer
296ecb_inline void
297encode_int (enc_t *enc, SV *sv)
298{
299 if (SvIsUV (sv))
300 encode_uint (enc, MAJOR_POS_INT, SvUVX (sv));
301 else if (SvIVX (sv) >= 0)
302 encode_uint (enc, MAJOR_POS_INT, SvIVX (sv));
303 else
304 encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1));
305}
306
276ecb_inline void 307ecb_inline void
277encode_tag (enc_t *enc, UV tag) 308encode_tag (enc_t *enc, UV tag)
278{ 309{
279 encode_uint (enc, MAJOR_TAG, tag); 310 encode_uint (enc, MAJOR_TAG, tag);
280} 311}
281 312
313// exceptional (hopefully) slow path for byte strings that need to be utf8-encoded
314ecb_noinline static void
315encode_str_utf8 (enc_t *enc, int utf8, char *str, STRLEN len)
316{
317 STRLEN ulen = len;
318 U8 *p, *pend = (U8 *)str + len;
319
320 for (p = (U8 *)str; p < pend; ++p)
321 ulen += *p >> 7; // count set high bits
322
323 encode_uint (enc, MAJOR_TEXT, ulen);
324
325 need (enc, ulen);
326 for (p = (U8 *)str; p < pend; ++p)
327 if (*p < 0x80)
328 *enc->cur++ = *p;
329 else
330 {
331 *enc->cur++ = 0xc0 + (*p >> 6);
332 *enc->cur++ = 0x80 + (*p & 63);
333 }
334}
335
282ecb_inline void 336ecb_inline void
283encode_str (enc_t *enc, int utf8, char *str, STRLEN len) 337encode_str (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len)
284{ 338{
285 if (ecb_expect_false (enc->cbor.flags & F_UTF8_STRINGS)) 339 if (ecb_expect_false (upgrade_utf8))
286 if (!utf8) 340 if (!utf8)
287 { 341 {
288 // exceptional path for bytze strings that need to be utf8-encoded 342 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; 343 return;
308 } 344 }
309 345
310 encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len); 346 encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len);
311 need (enc, len); 347 need (enc, len);
312 memcpy (enc->cur, str, len); 348 memcpy (enc->cur, str, len);
313 enc->cur += len; 349 enc->cur += len;
314} 350}
315 351
316static void 352ecb_inline void
317encode_strref (enc_t *enc, int utf8, char *str, STRLEN len) 353encode_strref (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len)
318{ 354{
319 if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS)) 355 if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS))
320 { 356 {
321 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1); 357 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1);
322 358
… …
333 sv_setuv (*svp, enc->stringref_idx); 369 sv_setuv (*svp, enc->stringref_idx);
334 ++enc->stringref_idx; 370 ++enc->stringref_idx;
335 } 371 }
336 } 372 }
337 373
338 encode_str (enc, utf8, str, len); 374 encode_str (enc, upgrade_utf8, utf8, str, len);
339} 375}
340 376
341static void encode_sv (enc_t *enc, SV *sv); 377ecb_inline void
378encode_float16 (enc_t *enc, NV nv)
379{
380 need (enc, 1+2);
381
382 *enc->cur++ = MAJOR_MISC | MISC_FLOAT16;
383
384 uint16_t fp = ecb_float_to_binary16 (nv);
385
386 if (!ecb_big_endian ())
387 fp = ecb_bswap16 (fp);
388
389 memcpy (enc->cur, &fp, 2);
390 enc->cur += 2;
391}
392
393ecb_inline void
394encode_float32 (enc_t *enc, NV nv)
395{
396 need (enc, 1+4);
397
398 *enc->cur++ = MAJOR_MISC | MISC_FLOAT32;
399
400 uint32_t fp = ecb_float_to_binary32 (nv);
401
402 if (!ecb_big_endian ())
403 fp = ecb_bswap32 (fp);
404
405 memcpy (enc->cur, &fp, 4);
406 enc->cur += 4;
407}
408
409ecb_inline void
410encode_float64 (enc_t *enc, NV nv)
411{
412 need (enc, 1+8);
413
414 *enc->cur++ = MAJOR_MISC | MISC_FLOAT64;
415
416 uint64_t fp = ecb_double_to_binary64 (nv);
417
418 if (!ecb_big_endian ())
419 fp = ecb_bswap64 (fp);
420
421 memcpy (enc->cur, &fp, 8);
422 enc->cur += 8;
423}
424
425ecb_inline void
426encode_bool (enc_t *enc, int istrue)
427{
428 encode_ch (enc, istrue ? MAJOR_MISC | SIMPLE_TRUE : MAJOR_MISC | SIMPLE_FALSE);
429}
430
431// encodes an arrayref containing key-value pairs as CBOR map
432ecb_inline void
433encode_array_as_map (enc_t *enc, SV *sv)
434{
435 if (enc->depth >= enc->cbor.max_depth)
436 croak (ERR_NESTING_EXCEEDED);
437
438 ++enc->depth;
439
440 // as_map does error checking for us, but we re-check in case
441 // things have changed.
442
443 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
444 croak ("CBOR::XS::as_map requires an array reference (did you change the array after calling as_map?)");
445
446 AV *av = (AV *)SvRV (sv);
447 int i, len = av_len (av);
448
449 if (!(len & 1))
450 croak ("CBOR::XS::as_map requires an even number of elements (did you change the array after calling as_map?)");
451
452 encode_uint (enc, MAJOR_MAP, (len + 1) >> 1);
453
454 for (i = 0; i <= len; ++i)
455 {
456 SV **svp = av_fetch (av, i, 0);
457 encode_sv (enc, svp ? *svp : &PL_sv_undef);
458 }
459
460 --enc->depth;
461}
462
463ecb_inline void
464encode_forced (enc_t *enc, UV type, SV *sv)
465{
466 switch (type)
467 {
468 case AS_CBOR:
469 {
470 STRLEN len;
471 char *str = SvPVbyte (sv, len);
472
473 need (enc, len);
474 memcpy (enc->cur, str, len);
475 enc->cur += len;
476 }
477 break;
478
479 case AS_BYTES:
480 {
481 STRLEN len;
482 char *str = SvPVbyte (sv, len);
483 encode_strref (enc, 0, 0, str, len);
484 }
485 break;
486
487 case AS_TEXT:
488 {
489 STRLEN len;
490 char *str = SvPVutf8 (sv, len);
491 encode_strref (enc, 1, 1, str, len);
492 }
493 break;
494
495 case AS_INT: encode_int (enc, sv); break;
496
497 case AS_FLOAT16: encode_float16 (enc, SvNV (sv)); break;
498 case AS_FLOAT32: encode_float32 (enc, SvNV (sv)); break;
499 case AS_FLOAT64: encode_float64 (enc, SvNV (sv)); break;
500
501 case AS_MAP: encode_array_as_map (enc, sv); break;
502
503 default:
504 croak ("encountered malformed CBOR::XS::Tagged object");
505 }
506}
342 507
343static void 508static void
344encode_av (enc_t *enc, AV *av) 509encode_av (enc_t *enc, AV *av)
345{ 510{
346 int i, len = av_len (av); 511 int i, len = av_len (av);
… …
350 515
351 ++enc->depth; 516 ++enc->depth;
352 517
353 encode_uint (enc, MAJOR_ARRAY, len + 1); 518 encode_uint (enc, MAJOR_ARRAY, len + 1);
354 519
355 if (SvMAGICAL (av)) 520 if (ecb_expect_false (SvMAGICAL (av)))
356 for (i = 0; i <= len; ++i) 521 for (i = 0; i <= len; ++i)
357 { 522 {
358 SV **svp = av_fetch (av, i, 0); 523 SV **svp = av_fetch (av, i, 0);
359 encode_sv (enc, svp ? *svp : &PL_sv_undef); 524 encode_sv (enc, svp ? *svp : &PL_sv_undef);
360 } 525 }
… …
379 ++enc->depth; 544 ++enc->depth;
380 545
381 int pairs = hv_iterinit (hv); 546 int pairs = hv_iterinit (hv);
382 int mg = SvMAGICAL (hv); 547 int mg = SvMAGICAL (hv);
383 548
384 if (mg) 549 if (ecb_expect_false (mg))
385 encode_ch (enc, MAJOR_MAP | MINOR_INDEF); 550 encode_ch (enc, MAJOR_MAP | MINOR_INDEF);
386 else 551 else
387 encode_uint (enc, MAJOR_MAP, pairs); 552 encode_uint (enc, MAJOR_MAP, pairs);
388 553
389 while ((he = hv_iternext (hv))) 554 while ((he = hv_iternext (hv)))
390 { 555 {
391 if (HeKLEN (he) == HEf_SVKEY) 556 if (HeKLEN (he) == HEf_SVKEY)
392 encode_sv (enc, HeSVKEY (he)); 557 encode_sv (enc, HeSVKEY (he));
393 else 558 else
394 encode_strref (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he)); 559 encode_strref (enc, enc->cbor.flags & (F_TEXT_KEYS | F_TEXT_STRINGS), HeKUTF8 (he), HeKEY (he), HeKLEN (he));
395 560
396 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he)); 561 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he));
397 } 562 }
398 563
399 if (mg) 564 if (ecb_expect_false (mg))
400 encode_ch (enc, MAJOR_MISC | MINOR_INDEF); 565 encode_ch (enc, MAJOR_MISC | MINOR_INDEF);
401 566
402 --enc->depth; 567 --enc->depth;
403} 568}
404 569
… …
424 589
425 HV *stash = SvSTASH (sv); 590 HV *stash = SvSTASH (sv);
426 591
427 if (stash == boolean_stash) 592 if (stash == boolean_stash)
428 { 593 {
429 encode_ch (enc, SvIV (sv) ? MAJOR_MISC | SIMPLE_TRUE : MAJOR_MISC | SIMPLE_FALSE); 594 encode_bool (enc, SvIV (sv));
430 return; 595 return;
431 } 596 }
432 else if (stash == error_stash) 597 else if (stash == error_stash)
433 { 598 {
434 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF); 599 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
… …
437 else if (stash == tagged_stash) 602 else if (stash == tagged_stash)
438 { 603 {
439 if (svt != SVt_PVAV) 604 if (svt != SVt_PVAV)
440 croak ("encountered CBOR::XS::Tagged object that isn't an array"); 605 croak ("encountered CBOR::XS::Tagged object that isn't an array");
441 606
607 switch (av_len ((AV *)sv))
608 {
609 case 2-1:
610 // actually a tagged value
442 encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1))); 611 encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1)));
443 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1)); 612 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1));
613 break;
614
615 case 3-1:
616 // a forced type [value, type, undef]
617 encode_forced (enc, SvUV (*av_fetch ((AV *)sv, 1, 1)), *av_fetch ((AV *)sv, 0, 1));
618 break;
619
620 default:
621 croak ("encountered malformed CBOR::XS::Tagged object");
622 }
444 623
445 return; 624 return;
446 } 625 }
447 } 626 }
448 627
449 if (ecb_expect_false (SvREFCNT (sv) > 1) 628 if (ecb_expect_false (SvREFCNT (sv) > 1)
450 && ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING)) 629 && ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING))
451 { 630 {
452 if (!enc->shareable) 631 if (ecb_expect_false (!enc->shareable))
453 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ()); 632 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
454 633
455 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1); 634 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
456 635
457 if (SvOK (*svp)) 636 if (SvOK (*svp))
… …
471 if (ecb_expect_false (SvOBJECT (sv))) 650 if (ecb_expect_false (SvOBJECT (sv)))
472 { 651 {
473 HV *stash = SvSTASH (sv); 652 HV *stash = SvSTASH (sv);
474 GV *method; 653 GV *method;
475 654
655 if (enc->cbor.flags & F_FORBID_OBJECTS)
656 croak ("encountered object '%s', but forbid_objects is enabled",
657 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
476 if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0))) 658 else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
477 { 659 {
478 dSP; 660 dSP;
479 661
480 ENTER; SAVETMPS; 662 ENTER; SAVETMPS;
481 PUSHMARK (SP); 663 PUSHMARK (SP);
… …
500 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0) 682 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)
501 { 683 {
502 dSP; 684 dSP;
503 685
504 ENTER; SAVETMPS; 686 ENTER; SAVETMPS;
505 SAVESTACK_POS ();
506 PUSHMARK (SP); 687 PUSHMARK (SP);
507 EXTEND (SP, 2); 688 EXTEND (SP, 2);
508 // we re-bless the reference to get overload and other niceties right 689 // we re-bless the reference to get overload and other niceties right
509 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); 690 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
510 PUSHs (sv_cbor); 691 PUSHs (sv_cbor);
… …
517 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv) 698 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)); 699 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));
519 700
520 encode_tag (enc, CBOR_TAG_PERL_OBJECT); 701 encode_tag (enc, CBOR_TAG_PERL_OBJECT);
521 encode_uint (enc, MAJOR_ARRAY, count + 1); 702 encode_uint (enc, MAJOR_ARRAY, count + 1);
522 encode_strref (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash)); 703 encode_strref (enc, 0, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
523 704
524 while (count) 705 {
706 int i;
707
708 for (i = 0; i < count; ++i)
525 encode_sv (enc, SP[1 - count--]); 709 encode_sv (enc, SP[i + 1 - count]);
710
711 SP -= count;
712 }
526 713
527 PUTBACK; 714 PUTBACK;
528 715
529 FREETMPS; LEAVE; 716 FREETMPS; LEAVE;
530 } 717 }
… …
552 739
553 if (ecb_expect_false (nv == (NV)(U32)nv)) 740 if (ecb_expect_false (nv == (NV)(U32)nv))
554 encode_uint (enc, MAJOR_POS_INT, (U32)nv); 741 encode_uint (enc, MAJOR_POS_INT, (U32)nv);
555 //TODO: maybe I32? 742 //TODO: maybe I32?
556 else if (ecb_expect_false (nv == (float)nv)) 743 else if (ecb_expect_false (nv == (float)nv))
557 { 744 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 745 else
569 { 746 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} 747}
581 748
582static void 749static void
583encode_sv (enc_t *enc, SV *sv) 750encode_sv (enc_t *enc, SV *sv)
584{ 751{
… …
586 753
587 if (SvPOKp (sv)) 754 if (SvPOKp (sv))
588 { 755 {
589 STRLEN len; 756 STRLEN len;
590 char *str = SvPV (sv, len); 757 char *str = SvPV (sv, len);
591 encode_strref (enc, SvUTF8 (sv), str, len); 758 encode_strref (enc, enc->cbor.flags & F_TEXT_STRINGS, SvUTF8 (sv), str, len);
592 } 759 }
593 else if (SvNOKp (sv)) 760 else if (SvNOKp (sv))
594 encode_nv (enc, sv); 761 encode_nv (enc, sv);
595 else if (SvIOKp (sv)) 762 else if (SvIOKp (sv))
596 { 763 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)) 764 else if (SvROK (sv))
605 encode_rv (enc, SvRV (sv)); 765 encode_rv (enc, SvRV (sv));
606 else if (!SvOK (sv)) 766 else if (!SvOK (sv))
607 encode_ch (enc, MAJOR_MISC | SIMPLE_NULL); 767 encode_ch (enc, MAJOR_MISC | SIMPLE_NULL);
608 else if (enc->cbor.flags & F_ALLOW_UNKNOWN) 768 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
… …
615static SV * 775static SV *
616encode_cbor (SV *scalar, CBOR *cbor) 776encode_cbor (SV *scalar, CBOR *cbor)
617{ 777{
618 enc_t enc = { 0 }; 778 enc_t enc = { 0 };
619 779
620 enc.cbor = *cbor; 780 enc.cbor = *cbor;
621 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 781 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
622 enc.cur = SvPVX (enc.sv); 782 enc.cur = SvPVX (enc.sv);
623 enc.end = SvEND (enc.sv); 783 enc.end = SvEND (enc.sv);
624 784
625 SvPOK_only (enc.sv); 785 SvPOK_only (enc.sv);
626 786
627 if (cbor->flags & F_PACK_STRINGS) 787 if (cbor->flags & F_PACK_STRINGS)
628 { 788 {
… …
655 U32 depth; // recursion depth 815 U32 depth; // recursion depth
656 U32 maxdepth; // recursion depth limit 816 U32 maxdepth; // recursion depth limit
657 AV *shareable; 817 AV *shareable;
658 AV *stringref; 818 AV *stringref;
659 SV *decode_tagged; 819 SV *decode_tagged;
820 SV *err_sv; // optional sv for error, needs to be freed
660} dec_t; 821} dec_t;
661 822
662#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE 823// set dec->err to ERRSV
824ecb_cold static void
825err_errsv (dec_t *dec)
826{
827 if (!dec->err)
828 {
829 dec->err_sv = newSVsv (ERRSV);
663 830
664#define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data") 831 // chop off the trailing \n
832 SvCUR_set (dec->err_sv, SvCUR (dec->err_sv) - 1);
833 *SvEND (dec->err_sv) = 0;
665 834
835 dec->err = SvPVutf8_nolen (dec->err_sv);
836 }
837}
838
839// the following functions are used to reduce code size and help the compiler to optimise
840ecb_cold static void
841err_set (dec_t *dec, const char *reason)
842{
843 if (!dec->err)
844 dec->err = reason;
845}
846
847ecb_cold static void
848err_unexpected_end (dec_t *dec)
849{
850 err_set (dec, "unexpected end of CBOR data");
851}
852
853#define ERR_DO(do) SB do; goto fail; SE
854#define ERR(reason) ERR_DO (err_set (dec, reason))
855#define ERR_ERRSV ERR_DO (err_errsv (dec))
856
857#define WANT(len) if (ecb_expect_false ((uintptr_t)(dec->end - dec->cur) < (STRLEN)len)) ERR_DO (err_unexpected_end (dec))
858
666#define DEC_INC_DEPTH if (++dec->depth > dec->cbor.max_depth) ERR (ERR_NESTING_EXCEEDED) 859#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 860#define DEC_DEC_DEPTH --dec->depth
668 861
669static UV 862static UV
670decode_uint (dec_t *dec) 863decode_uint (dec_t *dec)
671{ 864{
… …
737 930
738 for (;;) 931 for (;;)
739 { 932 {
740 WANT (1); 933 WANT (1);
741 934
742 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF)) 935 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF) || dec->err)
743 { 936 {
744 ++dec->cur; 937 ++dec->cur;
745 break; 938 break;
746 } 939 }
747 940
748 av_push (av, decode_sv (dec)); 941 av_push (av, decode_sv (dec));
749 } 942 }
750 } 943 }
751 else 944 else
752 { 945 {
753 int i, len = decode_uint (dec); 946 UV i, len = decode_uint (dec);
754 947
755 WANT (len); // complexity check for av_fill - need at least one byte per value, do not allow supersize arrays 948 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); 949 av_fill (av, len - 1);
757 950
758 for (i = 0; i < len; ++i) 951 for (i = 0; i < len; ++i)
… …
761 954
762 DEC_DEC_DEPTH; 955 DEC_DEC_DEPTH;
763 return newRV_noinc ((SV *)av); 956 return newRV_noinc ((SV *)av);
764 957
765fail: 958fail:
766 SvREFCNT_dec (av); 959 SvREFCNT_dec_NN (av);
767 DEC_DEC_DEPTH; 960 DEC_DEC_DEPTH;
768 return &PL_sv_undef; 961 return &PL_sv_undef;
769} 962}
770 963
771static void 964static void
… …
775 // byte or utf-8 strings as keys, but only when !stringref 968 // byte or utf-8 strings as keys, but only when !stringref
776 969
777 if (ecb_expect_true (!dec->stringref)) 970 if (ecb_expect_true (!dec->stringref))
778 if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8)) 971 if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8))
779 { 972 {
780 I32 len = decode_uint (dec); 973 STRLEN len = decode_uint (dec);
781 char *key = (char *)dec->cur; 974 char *key = (char *)dec->cur;
782 975
783 WANT (len); 976 WANT (len);
784 dec->cur += len; 977 dec->cur += len;
785 978
… …
787 980
788 return; 981 return;
789 } 982 }
790 else if (ecb_expect_true ((U8)(*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8)) 983 else if (ecb_expect_true ((U8)(*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8))
791 { 984 {
792 I32 len = decode_uint (dec); 985 STRLEN len = decode_uint (dec);
793 char *key = (char *)dec->cur; 986 char *key = (char *)dec->cur;
794 987
795 WANT (len); 988 WANT (len);
796 dec->cur += len; 989 dec->cur += len;
797 990
… …
805 } 998 }
806 999
807 SV *k = decode_sv (dec); 1000 SV *k = decode_sv (dec);
808 SV *v = decode_sv (dec); 1001 SV *v = decode_sv (dec);
809 1002
1003 // we leak memory if uncaught exceptions are thrown by random magical
1004 // methods, and this is hopefully the only place where it can happen,
1005 // so if there is a chance of an exception, take the very slow path.
1006 // since catching exceptions is "undocumented/internal/forbidden" by
1007 // the new p5p powers, we need to call out to a perl function :/
1008 if (ecb_expect_false (SvAMAGIC (k)))
1009 {
1010 dSP;
1011
1012 ENTER; SAVETMPS;
1013 PUSHMARK (SP);
1014 EXTEND (SP, 3);
1015 PUSHs (sv_2mortal (newRV_inc ((SV *)hv)));
1016 PUSHs (sv_2mortal (k));
1017 PUSHs (sv_2mortal (v));
1018
1019 PUTBACK;
1020 call_pv ("CBOR::XS::_hv_store", G_VOID | G_DISCARD | G_EVAL);
1021 SPAGAIN;
1022
1023 FREETMPS; LEAVE;
1024
1025 if (SvTRUE (ERRSV))
1026 ERR_ERRSV;
1027
1028 return;
1029 }
1030
810 hv_store_ent (hv, k, v, 0); 1031 hv_store_ent (hv, k, v, 0);
811 SvREFCNT_dec (k); 1032 SvREFCNT_dec_NN (k);
812 1033
813fail: 1034fail:
814 ; 1035 ;
815} 1036}
816 1037
… …
827 1048
828 for (;;) 1049 for (;;)
829 { 1050 {
830 WANT (1); 1051 WANT (1);
831 1052
832 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF)) 1053 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF) || dec->err)
833 { 1054 {
834 ++dec->cur; 1055 ++dec->cur;
835 break; 1056 break;
836 } 1057 }
837 1058
838 decode_he (dec, hv); 1059 decode_he (dec, hv);
839 } 1060 }
840 } 1061 }
841 else 1062 else
842 { 1063 {
843 int pairs = decode_uint (dec); 1064 UV pairs = decode_uint (dec);
1065
1066 WANT (pairs); // complexity check - need at least one byte per value, do not allow supersize hashes
844 1067
845 while (pairs--) 1068 while (pairs--)
846 decode_he (dec, hv); 1069 decode_he (dec, hv);
847 } 1070 }
848 1071
849 DEC_DEC_DEPTH; 1072 DEC_DEC_DEPTH;
850 return newRV_noinc ((SV *)hv); 1073 return newRV_noinc ((SV *)hv);
851 1074
852fail: 1075fail:
853 SvREFCNT_dec (hv); 1076 SvREFCNT_dec_NN (hv);
854 DEC_DEC_DEPTH; 1077 DEC_DEC_DEPTH;
855 return &PL_sv_undef; 1078 return &PL_sv_undef;
856} 1079}
857 1080
858static SV * 1081static SV *
859decode_str (dec_t *dec, int utf8) 1082decode_str (dec_t *dec, int utf8)
860{ 1083{
861 SV *sv = 0; 1084 SV *sv = 0;
862 1085
863 if ((*dec->cur & MINOR_MASK) == MINOR_INDEF) 1086 if (ecb_expect_false ((*dec->cur & MINOR_MASK) == MINOR_INDEF))
864 { 1087 {
865 // indefinite length strings 1088 // indefinite length strings
866 ++dec->cur; 1089 ++dec->cur;
867 1090
868 U8 major = *dec->cur & MAJOR_MISC; 1091 U8 major = *dec->cur & MAJOR_MISC;
… …
936 sv = newRV_noinc (decode_sv (dec)); 1159 sv = newRV_noinc (decode_sv (dec));
937 break; 1160 break;
938 1161
939 case CBOR_TAG_STRINGREF_NAMESPACE: 1162 case CBOR_TAG_STRINGREF_NAMESPACE:
940 { 1163 {
1164 // do not use SAVETMPS/FREETMPS, as these will
1165 // erase mortalised caches, e.g. "shareable"
941 ENTER; SAVETMPS; 1166 ENTER;
942 1167
943 SAVESPTR (dec->stringref); 1168 SAVESPTR (dec->stringref);
944 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ()); 1169 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ());
945 1170
946 sv = decode_sv (dec); 1171 sv = decode_sv (dec);
947 1172
948 FREETMPS; LEAVE; 1173 LEAVE;
949 } 1174 }
950 break; 1175 break;
951 1176
952 case CBOR_TAG_STRINGREF: 1177 case CBOR_TAG_STRINGREF:
953 { 1178 {
954 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT)) 1179 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
955 ERR ("corrupted CBOR data (stringref index not an unsigned integer)"); 1180 ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
956 1181
957 UV idx = decode_uint (dec); 1182 UV idx = decode_uint (dec);
958 1183
959 if (!dec->stringref || (int)idx > AvFILLp (dec->stringref)) 1184 if (!dec->stringref || idx >= (UV)(1 + AvFILLp (dec->stringref)))
960 ERR ("corrupted CBOR data (stringref index out of bounds or outside namespace)"); 1185 ERR ("corrupted CBOR data (stringref index out of bounds or outside namespace)");
961 1186
962 sv = newSVsv (AvARRAY (dec->stringref)[idx]); 1187 sv = newSVsv (AvARRAY (dec->stringref)[idx]);
963 } 1188 }
964 break; 1189 break;
… …
992 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT)) 1217 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
993 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)"); 1218 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
994 1219
995 UV idx = decode_uint (dec); 1220 UV idx = decode_uint (dec);
996 1221
997 if (!dec->shareable || (int)idx > AvFILLp (dec->shareable)) 1222 if (!dec->shareable || idx >= (UV)(1 + AvFILLp (dec->shareable)))
998 ERR ("corrupted CBOR data (sharedref index out of bounds)"); 1223 ERR ("corrupted CBOR data (sharedref index out of bounds)");
999 1224
1000 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]); 1225 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]);
1001 1226
1002 if (sv == &PL_sv_undef) 1227 if (sv == &PL_sv_undef)
… …
1004 } 1229 }
1005 break; 1230 break;
1006 1231
1007 case CBOR_TAG_PERL_OBJECT: 1232 case CBOR_TAG_PERL_OBJECT:
1008 { 1233 {
1234 if (dec->cbor.flags & F_FORBID_OBJECTS)
1235 goto filter;
1236
1009 sv = decode_sv (dec); 1237 sv = decode_sv (dec);
1010 1238
1011 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV) 1239 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
1012 ERR ("corrupted CBOR data (non-array perl object)"); 1240 ERR ("corrupted CBOR data (non-array perl object)");
1013 1241
… …
1042 SPAGAIN; 1270 SPAGAIN;
1043 1271
1044 if (SvTRUE (ERRSV)) 1272 if (SvTRUE (ERRSV))
1045 { 1273 {
1046 FREETMPS; LEAVE; 1274 FREETMPS; LEAVE;
1047 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV)))); 1275 ERR_ERRSV;
1048 } 1276 }
1049 1277
1050 SvREFCNT_dec (sv); 1278 SvREFCNT_dec_NN (sv);
1051 sv = SvREFCNT_inc (POPs); 1279 sv = SvREFCNT_inc (POPs);
1052 1280
1053 PUTBACK; 1281 PUTBACK;
1054 1282
1055 FREETMPS; LEAVE; 1283 FREETMPS; LEAVE;
1056 } 1284 }
1057 break; 1285 break;
1058 1286
1059 default: 1287 default:
1288 filter:
1060 { 1289 {
1290 SV *tag_sv = newSVuv (tag);
1291
1061 sv = decode_sv (dec); 1292 sv = decode_sv (dec);
1062 1293
1063 dSP; 1294 dSP;
1064 ENTER; SAVETMPS; 1295 ENTER; SAVETMPS;
1065 SAVESTACK_POS ();
1066 PUSHMARK (SP); 1296 PUSHMARK (SP);
1067 EXTEND (SP, 2); 1297 EXTEND (SP, 2);
1068 PUSHs (newSVuv (tag)); 1298 PUSHs (tag_sv);
1069 PUSHs (sv); 1299 PUSHs (sv);
1070 1300
1071 PUTBACK; 1301 PUTBACK;
1072 int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL); 1302 int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL);
1073 SPAGAIN; 1303 SPAGAIN;
1074 1304
1075 if (SvTRUE (ERRSV)) 1305 if (SvTRUE (ERRSV))
1076 { 1306 {
1307 SvREFCNT_dec_NN (tag_sv);
1077 FREETMPS; LEAVE; 1308 FREETMPS; LEAVE;
1078 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV)))); 1309 ERR_ERRSV;
1079 } 1310 }
1080 1311
1081 if (count) 1312 if (count)
1082 { 1313 {
1314 SvREFCNT_dec_NN (tag_sv);
1083 SvREFCNT_dec (sv); 1315 SvREFCNT_dec_NN (sv);
1084 sv = SvREFCNT_inc (POPs); 1316 sv = SvREFCNT_inc_NN (TOPs);
1317 SP -= count;
1085 } 1318 }
1086 else 1319 else
1087 { 1320 {
1088 AV *av = newAV (); 1321 AV *av = newAV ();
1089 av_push (av, newSVuv (tag)); 1322 av_push (av, tag_sv);
1090 av_push (av, sv); 1323 av_push (av, sv);
1091 1324
1092 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 1325 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
1093 ? cbor_tagged_stash 1326 ? cbor_tagged_stash
1094 : gv_stashpv ("CBOR::XS::Tagged" , 1); 1327 : gv_stashpv ("CBOR::XS::Tagged" , 1);
… …
1223 1456
1224 if (dec.err) 1457 if (dec.err)
1225 { 1458 {
1226 if (dec.shareable) 1459 if (dec.shareable)
1227 { 1460 {
1228 // need to break cyclic links, which whould all be in shareable 1461 // need to break cyclic links, which would all be in shareable
1229 int i; 1462 int i;
1230 SV **svp; 1463 SV **svp;
1231 1464
1232 for (i = av_len (dec.shareable) + 1; i--; ) 1465 for (i = av_len (dec.shareable) + 1; i--; )
1233 if ((svp = av_fetch (dec.shareable, i, 0))) 1466 if ((svp = av_fetch (dec.shareable, i, 0)))
1234 sv_setsv (*svp, &PL_sv_undef); 1467 sv_setsv (*svp, &PL_sv_undef);
1235 } 1468 }
1236 1469
1237 SvREFCNT_dec (sv); 1470 SvREFCNT_dec_NN (sv);
1471
1472 if (dec.err_sv)
1473 sv_2mortal (dec.err_sv);
1474
1238 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur); 1475 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur);
1239 } 1476 }
1240 1477
1241 sv = sv_2mortal (sv); 1478 sv = sv_2mortal (sv);
1242 1479
… …
1386 1623
1387 default_filter = newSVpv ("CBOR::XS::default_filter", 0); 1624 default_filter = newSVpv ("CBOR::XS::default_filter", 0);
1388 1625
1389 sv_cbor = newSVpv ("CBOR", 0); 1626 sv_cbor = newSVpv ("CBOR", 0);
1390 SvREADONLY_on (sv_cbor); 1627 SvREADONLY_on (sv_cbor);
1628
1629 assert (("STRLEN must be an unsigned type", 0 <= (STRLEN)-1));
1391} 1630}
1392 1631
1393PROTOTYPES: DISABLE 1632PROTOTYPES: DISABLE
1394 1633
1395void CLONE (...) 1634void CLONE (...)
… …
1415 ALIAS: 1654 ALIAS:
1416 shrink = F_SHRINK 1655 shrink = F_SHRINK
1417 allow_unknown = F_ALLOW_UNKNOWN 1656 allow_unknown = F_ALLOW_UNKNOWN
1418 allow_sharing = F_ALLOW_SHARING 1657 allow_sharing = F_ALLOW_SHARING
1419 allow_cycles = F_ALLOW_CYCLES 1658 allow_cycles = F_ALLOW_CYCLES
1659 forbid_objects = F_FORBID_OBJECTS
1420 pack_strings = F_PACK_STRINGS 1660 pack_strings = F_PACK_STRINGS
1661 text_keys = F_TEXT_KEYS
1421 utf8_strings = F_UTF8_STRINGS 1662 text_strings = F_TEXT_STRINGS
1422 validate_utf8 = F_VALIDATE_UTF8 1663 validate_utf8 = F_VALIDATE_UTF8
1423 PPCODE: 1664 PPCODE:
1424{ 1665{
1425 if (enable) 1666 if (enable)
1426 self->flags |= ix; 1667 self->flags |= ix;
… …
1434 ALIAS: 1675 ALIAS:
1435 get_shrink = F_SHRINK 1676 get_shrink = F_SHRINK
1436 get_allow_unknown = F_ALLOW_UNKNOWN 1677 get_allow_unknown = F_ALLOW_UNKNOWN
1437 get_allow_sharing = F_ALLOW_SHARING 1678 get_allow_sharing = F_ALLOW_SHARING
1438 get_allow_cycles = F_ALLOW_CYCLES 1679 get_allow_cycles = F_ALLOW_CYCLES
1680 get_forbid_objects = F_FORBID_OBJECTS
1439 get_pack_strings = F_PACK_STRINGS 1681 get_pack_strings = F_PACK_STRINGS
1682 get_text_keys = F_TEXT_KEYS
1683 get_text_strings = F_TEXT_STRINGS
1440 get_validate_utf8 = F_VALIDATE_UTF8 1684 get_validate_utf8 = F_VALIDATE_UTF8
1441 PPCODE: 1685 PPCODE:
1442 XPUSHs (boolSV (self->flags & ix)); 1686 XPUSHs (boolSV (self->flags & ix));
1443 1687
1444void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) 1688void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
… …
1574 cbor_init (&cbor); 1818 cbor_init (&cbor);
1575 PUTBACK; cborstr = decode_cbor (cborstr, &cbor, 0); SPAGAIN; 1819 PUTBACK; cborstr = decode_cbor (cborstr, &cbor, 0); SPAGAIN;
1576 XPUSHs (cborstr); 1820 XPUSHs (cborstr);
1577} 1821}
1578 1822
1823#ifdef __AFL_COMPILER
1824
1825void
1826afl_init ()
1827 CODE:
1828 __AFL_INIT ();
1829
1830int
1831afl_loop (unsigned int count = 10000)
1832 CODE:
1833 RETVAL = __AFL_LOOP (count);
1834 OUTPUT:
1835 RETVAL
1836
1837#endif
1838

Diff Legend

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