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 1.39 by root, Sun Dec 1 14:45:03 2013 UTC vs.
Revision 1.67 by root, Sun Nov 29 21:35:06 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
13#define ECB_NO_THREADS 1
12#include "ecb.h" 14#include "ecb.h"
13 15
14// compatibility with perl <5.18 16// compatibility with perl <5.18
15#ifndef HvNAMELEN_get 17#ifndef HvNAMELEN_get
16# define HvNAMELEN_get(hv) strlen (HvNAME (hv)) 18# define HvNAMELEN_get(hv) strlen (HvNAME (hv))
… …
19# define HvNAMELEN(hv) HvNAMELEN_get (hv) 21# define HvNAMELEN(hv) HvNAMELEN_get (hv)
20#endif 22#endif
21#ifndef HvNAMEUTF8 23#ifndef HvNAMEUTF8
22# define HvNAMEUTF8(hv) 0 24# define HvNAMEUTF8(hv) 0
23#endif 25#endif
26#ifndef SvREFCNT_inc_NN
27# define SvREFCNT_inc_NN(sv) SvREFCNT_inc (sv)
28#endif
24#ifndef SvREFCNT_dec_NN 29#ifndef SvREFCNT_dec_NN
25# define SvREFCNT_dec_NN(sv) SvREFCNT_dec (sv) 30# define SvREFCNT_dec_NN(sv) SvREFCNT_dec (sv)
26#endif 31#endif
27 32
28// known major and minor types 33// known major and minor types
… …
94 CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8 99 CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8
95 100
96 CBOR_TAG_MAGIC = 55799, // self-describe cbor 101 CBOR_TAG_MAGIC = 55799, // self-describe cbor
97}; 102};
98 103
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_INT = 3,
111 AS_FLOAT16 = 4,
112 AS_FLOAT32 = 5,
113 AS_FLOAT64 = 6,
114 // possibly future enhancements: (generic) float, (generic) string
115};
116
99#define F_SHRINK 0x00000001UL 117#define F_SHRINK 0x00000001UL
100#define F_ALLOW_UNKNOWN 0x00000002UL 118#define F_ALLOW_UNKNOWN 0x00000002UL
101#define F_ALLOW_SHARING 0x00000004UL 119#define F_ALLOW_SHARING 0x00000004UL
102#define F_ALLOW_CYCLES 0x00000008UL 120#define F_ALLOW_CYCLES 0x00000008UL
121#define F_FORBID_OBJECTS 0x00000010UL
103#define F_PACK_STRINGS 0x00000010UL 122#define F_PACK_STRINGS 0x00000020UL
123#define F_TEXT_KEYS 0x00000040UL
124#define F_TEXT_STRINGS 0x00000080UL
104#define F_VALIDATE_UTF8 0x00000020UL 125#define F_VALIDATE_UTF8 0x00000100UL
105 126
106#define INIT_SIZE 32 // initial scalar size to be allocated 127#define INIT_SIZE 32 // initial scalar size to be allocated
107 128
108#define SB do { 129#define SB do {
109#define SE } while (0) 130#define SE } while (0)
… …
128typedef struct { 149typedef struct {
129 U32 flags; 150 U32 flags;
130 U32 max_depth; 151 U32 max_depth;
131 STRLEN max_size; 152 STRLEN max_size;
132 SV *filter; 153 SV *filter;
154
155 // for the incremental parser
156 STRLEN incr_pos; // the current offset into the text
157 STRLEN incr_need; // minimum bytes needed to decode
158 AV *incr_count; // for every nesting level, the number of outstanding values, or -1 for indef.
133} CBOR; 159} CBOR;
134 160
135ecb_inline void 161ecb_inline void
136cbor_init (CBOR *cbor) 162cbor_init (CBOR *cbor)
137{ 163{
… …
141 167
142ecb_inline void 168ecb_inline void
143cbor_free (CBOR *cbor) 169cbor_free (CBOR *cbor)
144{ 170{
145 SvREFCNT_dec (cbor->filter); 171 SvREFCNT_dec (cbor->filter);
172 SvREFCNT_dec (cbor->incr_count);
146} 173}
147 174
148///////////////////////////////////////////////////////////////////////////// 175/////////////////////////////////////////////////////////////////////////////
149// utility functions 176// utility functions
150 177
… …
176 203
177// minimum length of a string to be registered for stringref 204// minimum length of a string to be registered for stringref
178ecb_inline int 205ecb_inline int
179minimum_string_length (UV idx) 206minimum_string_length (UV idx)
180{ 207{
181 return idx > 23 208 return idx <= 23 ? 3
182 ? idx > 0xffU 209 : idx <= 0xffU ? 4
183 ? idx > 0xffffU 210 : idx <= 0xffffU ? 5
184 ? idx > 0xffffffffU 211 : idx <= 0xffffffffU ? 7
185 ? 11 212 : 11;
186 : 7
187 : 5
188 : 4
189 : 3;
190} 213}
191 214
192///////////////////////////////////////////////////////////////////////////// 215/////////////////////////////////////////////////////////////////////////////
193// encoder 216// encoder
194 217
… …
207} enc_t; 230} enc_t;
208 231
209ecb_inline void 232ecb_inline void
210need (enc_t *enc, STRLEN len) 233need (enc_t *enc, STRLEN len)
211{ 234{
212 if (ecb_expect_false (enc->cur + len >= enc->end)) 235 if (ecb_expect_false ((uintptr_t)(enc->end - enc->cur) < len))
213 { 236 {
214 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv); 237 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
215 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1); 238 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
216 enc->cur = SvPVX (enc->sv) + cur; 239 enc->cur = SvPVX (enc->sv) + cur;
217 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; 240 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
… …
223{ 246{
224 need (enc, 1); 247 need (enc, 1);
225 *enc->cur++ = ch; 248 *enc->cur++ = ch;
226} 249}
227 250
251// used for tags, intregers, element counts and so on
228static void 252static void
229encode_uint (enc_t *enc, int major, UV len) 253encode_uint (enc_t *enc, int major, UV len)
230{ 254{
231 need (enc, 9); 255 need (enc, 9);
232 256
… …
263 *enc->cur++ = len >> 8; 287 *enc->cur++ = len >> 8;
264 *enc->cur++ = len; 288 *enc->cur++ = len;
265 } 289 }
266} 290}
267 291
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
268ecb_inline void 304ecb_inline void
269encode_tag (enc_t *enc, UV tag) 305encode_tag (enc_t *enc, UV tag)
270{ 306{
271 encode_uint (enc, MAJOR_TAG, tag); 307 encode_uint (enc, MAJOR_TAG, tag);
272} 308}
273 309
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
274ecb_inline void 333ecb_inline void
275encode_str (enc_t *enc, int utf8, char *str, STRLEN len) 334encode_str (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len)
276{ 335{
336 if (ecb_expect_false (upgrade_utf8))
337 if (!utf8)
338 {
339 encode_str_utf8 (enc, utf8, str, len);
340 return;
341 }
342
277 encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len); 343 encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len);
278 need (enc, len); 344 need (enc, len);
279 memcpy (enc->cur, str, len); 345 memcpy (enc->cur, str, len);
280 enc->cur += len; 346 enc->cur += len;
281} 347}
282 348
283static void 349ecb_inline void
284encode_strref (enc_t *enc, int utf8, char *str, STRLEN len) 350encode_strref (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len)
285{ 351{
286 if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS)) 352 if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS))
287 { 353 {
288 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1); 354 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1);
289 355
… …
300 sv_setuv (*svp, enc->stringref_idx); 366 sv_setuv (*svp, enc->stringref_idx);
301 ++enc->stringref_idx; 367 ++enc->stringref_idx;
302 } 368 }
303 } 369 }
304 370
305 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_forced (enc_t *enc, UV type, SV *sv)
424{
425 switch (type)
426 {
427 case AS_CBOR:
428 {
429 STRLEN len;
430 char *str = SvPVbyte (sv, len);
431
432 need (enc, len);
433 memcpy (enc->cur, str, len);
434 enc->cur += len;
435 }
436 break;
437
438 case AS_BYTES:
439 {
440 STRLEN len;
441 char *str = SvPVbyte (sv, len);
442 encode_strref (enc, 0, 0, str, len);
443 }
444 break;
445
446 case AS_TEXT:
447 {
448 STRLEN len;
449 char *str = SvPVutf8 (sv, len);
450 encode_strref (enc, 1, 1, str, len);
451 }
452 break;
453
454 case AS_INT: encode_int (enc, sv); break;
455 case AS_FLOAT16: encode_float16 (enc, SvNV (sv)); break;
456 case AS_FLOAT32: encode_float32 (enc, SvNV (sv)); break;
457 case AS_FLOAT64: encode_float64 (enc, SvNV (sv)); break;
458
459 default:
460 croak ("encountered malformed CBOR::XS::Tagged object");
461 }
306} 462}
307 463
308static void encode_sv (enc_t *enc, SV *sv); 464static void encode_sv (enc_t *enc, SV *sv);
309 465
310static void 466static void
… …
317 473
318 ++enc->depth; 474 ++enc->depth;
319 475
320 encode_uint (enc, MAJOR_ARRAY, len + 1); 476 encode_uint (enc, MAJOR_ARRAY, len + 1);
321 477
478 if (ecb_expect_false (SvMAGICAL (av)))
322 for (i = 0; i <= len; ++i) 479 for (i = 0; i <= len; ++i)
323 { 480 {
324 SV **svp = av_fetch (av, i, 0); 481 SV **svp = av_fetch (av, i, 0);
325 encode_sv (enc, svp ? *svp : &PL_sv_undef); 482 encode_sv (enc, svp ? *svp : &PL_sv_undef);
326 } 483 }
484 else
485 for (i = 0; i <= len; ++i)
486 {
487 SV *sv = AvARRAY (av)[i];
488 encode_sv (enc, sv ? sv : &PL_sv_undef);
489 }
327 490
328 --enc->depth; 491 --enc->depth;
329} 492}
330 493
331static void 494static void
… …
339 ++enc->depth; 502 ++enc->depth;
340 503
341 int pairs = hv_iterinit (hv); 504 int pairs = hv_iterinit (hv);
342 int mg = SvMAGICAL (hv); 505 int mg = SvMAGICAL (hv);
343 506
344 if (mg) 507 if (ecb_expect_false (mg))
345 encode_ch (enc, MAJOR_MAP | MINOR_INDEF); 508 encode_ch (enc, MAJOR_MAP | MINOR_INDEF);
346 else 509 else
347 encode_uint (enc, MAJOR_MAP, pairs); 510 encode_uint (enc, MAJOR_MAP, pairs);
348 511
349 while ((he = hv_iternext (hv))) 512 while ((he = hv_iternext (hv)))
350 { 513 {
351 if (HeKLEN (he) == HEf_SVKEY) 514 if (HeKLEN (he) == HEf_SVKEY)
352 encode_sv (enc, HeSVKEY (he)); 515 encode_sv (enc, HeSVKEY (he));
353 else 516 else
354 encode_strref (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he)); 517 encode_strref (enc, enc->cbor.flags & (F_TEXT_KEYS | F_TEXT_STRINGS), HeKUTF8 (he), HeKEY (he), HeKLEN (he));
355 518
356 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he)); 519 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he));
357 } 520 }
358 521
359 if (mg) 522 if (ecb_expect_false (mg))
360 encode_ch (enc, MAJOR_MISC | MINOR_INDEF); 523 encode_ch (enc, MAJOR_MISC | MINOR_INDEF);
361 524
362 --enc->depth; 525 --enc->depth;
363} 526}
364 527
… …
397 else if (stash == tagged_stash) 560 else if (stash == tagged_stash)
398 { 561 {
399 if (svt != SVt_PVAV) 562 if (svt != SVt_PVAV)
400 croak ("encountered CBOR::XS::Tagged object that isn't an array"); 563 croak ("encountered CBOR::XS::Tagged object that isn't an array");
401 564
565 switch (av_len ((AV *)sv))
566 {
567 case 2-1:
568 // actually a tagged value
402 encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1))); 569 encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1)));
403 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1)); 570 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1));
571 break;
572
573 case 3-1:
574 // a forced type [value, type, undef]
575 encode_forced (enc, SvUV (*av_fetch ((AV *)sv, 1, 1)), *av_fetch ((AV *)sv, 0, 1));
576 break;
577
578 default:
579 croak ("encountered malformed CBOR::XS::Tagged object");
580 }
404 581
405 return; 582 return;
406 } 583 }
407 } 584 }
408 585
409 if (ecb_expect_false (SvREFCNT (sv) > 1) 586 if (ecb_expect_false (SvREFCNT (sv) > 1)
410 && ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING)) 587 && ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING))
411 { 588 {
412 if (!enc->shareable) 589 if (ecb_expect_false (!enc->shareable))
413 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ()); 590 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
414 591
415 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1); 592 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
416 593
417 if (SvOK (*svp)) 594 if (SvOK (*svp))
… …
431 if (ecb_expect_false (SvOBJECT (sv))) 608 if (ecb_expect_false (SvOBJECT (sv)))
432 { 609 {
433 HV *stash = SvSTASH (sv); 610 HV *stash = SvSTASH (sv);
434 GV *method; 611 GV *method;
435 612
613 if (enc->cbor.flags & F_FORBID_OBJECTS)
614 croak ("encountered object '%s', but forbid_objects is enabled",
615 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
436 if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0))) 616 else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
437 { 617 {
438 dSP; 618 dSP;
439 619
440 ENTER; SAVETMPS; PUSHMARK (SP); 620 ENTER; SAVETMPS;
621 PUSHMARK (SP);
441 // we re-bless the reference to get overload and other niceties right 622 // we re-bless the reference to get overload and other niceties right
442 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); 623 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
443 624
444 PUTBACK; 625 PUTBACK;
445 // G_SCALAR ensures that return value is 1 626 // G_SCALAR ensures that return value is 1
… …
458 } 639 }
459 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0) 640 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)
460 { 641 {
461 dSP; 642 dSP;
462 643
463 ENTER; SAVETMPS; PUSHMARK (SP); 644 ENTER; SAVETMPS;
645 PUSHMARK (SP);
464 EXTEND (SP, 2); 646 EXTEND (SP, 2);
465 // we re-bless the reference to get overload and other niceties right 647 // we re-bless the reference to get overload and other niceties right
466 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); 648 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
467 PUSHs (sv_cbor); 649 PUSHs (sv_cbor);
468 650
… …
474 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv) 656 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)
475 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash)); 657 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));
476 658
477 encode_tag (enc, CBOR_TAG_PERL_OBJECT); 659 encode_tag (enc, CBOR_TAG_PERL_OBJECT);
478 encode_uint (enc, MAJOR_ARRAY, count + 1); 660 encode_uint (enc, MAJOR_ARRAY, count + 1);
479 encode_strref (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash)); 661 encode_strref (enc, 0, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
480 662
481 while (count) 663 {
664 int i;
665
666 for (i = 0; i < count; ++i)
482 encode_sv (enc, SP[1 - count--]); 667 encode_sv (enc, SP[i + 1 - count]);
668
669 SP -= count;
670 }
483 671
484 PUTBACK; 672 PUTBACK;
485 673
486 FREETMPS; LEAVE; 674 FREETMPS; LEAVE;
487 } 675 }
… …
509 697
510 if (ecb_expect_false (nv == (NV)(U32)nv)) 698 if (ecb_expect_false (nv == (NV)(U32)nv))
511 encode_uint (enc, MAJOR_POS_INT, (U32)nv); 699 encode_uint (enc, MAJOR_POS_INT, (U32)nv);
512 //TODO: maybe I32? 700 //TODO: maybe I32?
513 else if (ecb_expect_false (nv == (float)nv)) 701 else if (ecb_expect_false (nv == (float)nv))
514 { 702 encode_float32 (enc, nv);
515 uint32_t fp = ecb_float_to_binary32 (nv);
516
517 *enc->cur++ = MAJOR_MISC | MISC_FLOAT32;
518
519 if (!ecb_big_endian ())
520 fp = ecb_bswap32 (fp);
521
522 memcpy (enc->cur, &fp, 4);
523 enc->cur += 4;
524 }
525 else 703 else
526 { 704 encode_float64 (enc, nv);
527 uint64_t fp = ecb_double_to_binary64 (nv);
528
529 *enc->cur++ = MAJOR_MISC | MISC_FLOAT64;
530
531 if (!ecb_big_endian ())
532 fp = ecb_bswap64 (fp);
533
534 memcpy (enc->cur, &fp, 8);
535 enc->cur += 8;
536 }
537} 705}
538 706
539static void 707static void
540encode_sv (enc_t *enc, SV *sv) 708encode_sv (enc_t *enc, SV *sv)
541{ 709{
… …
543 711
544 if (SvPOKp (sv)) 712 if (SvPOKp (sv))
545 { 713 {
546 STRLEN len; 714 STRLEN len;
547 char *str = SvPV (sv, len); 715 char *str = SvPV (sv, len);
548 encode_strref (enc, SvUTF8 (sv), str, len); 716 encode_strref (enc, enc->cbor.flags & F_TEXT_STRINGS, SvUTF8 (sv), str, len);
549 } 717 }
550 else if (SvNOKp (sv)) 718 else if (SvNOKp (sv))
551 encode_nv (enc, sv); 719 encode_nv (enc, sv);
552 else if (SvIOKp (sv)) 720 else if (SvIOKp (sv))
553 { 721 encode_int (enc, sv);
554 if (SvIsUV (sv))
555 encode_uint (enc, MAJOR_POS_INT, SvUVX (sv));
556 else if (SvIVX (sv) >= 0)
557 encode_uint (enc, MAJOR_POS_INT, SvIVX (sv));
558 else
559 encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1));
560 }
561 else if (SvROK (sv)) 722 else if (SvROK (sv))
562 encode_rv (enc, SvRV (sv)); 723 encode_rv (enc, SvRV (sv));
563 else if (!SvOK (sv)) 724 else if (!SvOK (sv))
564 encode_ch (enc, MAJOR_MISC | SIMPLE_NULL); 725 encode_ch (enc, MAJOR_MISC | SIMPLE_NULL);
565 else if (enc->cbor.flags & F_ALLOW_UNKNOWN) 726 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
… …
570} 731}
571 732
572static SV * 733static SV *
573encode_cbor (SV *scalar, CBOR *cbor) 734encode_cbor (SV *scalar, CBOR *cbor)
574{ 735{
575 enc_t enc = { }; 736 enc_t enc = { 0 };
576 737
577 enc.cbor = *cbor; 738 enc.cbor = *cbor;
578 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 739 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
579 enc.cur = SvPVX (enc.sv); 740 enc.cur = SvPVX (enc.sv);
580 enc.end = SvEND (enc.sv); 741 enc.end = SvEND (enc.sv);
581 742
582 SvPOK_only (enc.sv); 743 SvPOK_only (enc.sv);
583 744
584 if (cbor->flags & F_PACK_STRINGS) 745 if (cbor->flags & F_PACK_STRINGS)
585 { 746 {
… …
612 U32 depth; // recursion depth 773 U32 depth; // recursion depth
613 U32 maxdepth; // recursion depth limit 774 U32 maxdepth; // recursion depth limit
614 AV *shareable; 775 AV *shareable;
615 AV *stringref; 776 AV *stringref;
616 SV *decode_tagged; 777 SV *decode_tagged;
778 SV *err_sv; // optional sv for error, needs to be freed
617} dec_t; 779} dec_t;
618 780
619#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE 781// set dec->err to ERRSV
782ecb_cold static void
783err_errsv (dec_t *dec)
784{
785 if (!dec->err)
786 {
787 dec->err_sv = newSVsv (ERRSV);
620 788
621#define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data") 789 // chop off the trailing \n
790 SvCUR_set (dec->err_sv, SvCUR (dec->err_sv) - 1);
791 *SvEND (dec->err_sv) = 0;
622 792
793 dec->err = SvPVutf8_nolen (dec->err_sv);
794 }
795}
796
797// the following functions are used to reduce code size and help the compiler to optimise
798ecb_cold static void
799err_set (dec_t *dec, const char *reason)
800{
801 if (!dec->err)
802 dec->err = reason;
803}
804
805ecb_cold static void
806err_unexpected_end (dec_t *dec)
807{
808 err_set (dec, "unexpected end of CBOR data");
809}
810
811#define ERR_DO(do) SB do; goto fail; SE
812#define ERR(reason) ERR_DO (err_set (dec, reason))
813#define ERR_ERRSV ERR_DO (err_errsv (dec))
814
815#define WANT(len) if (ecb_expect_false ((uintptr_t)(dec->end - dec->cur) < (STRLEN)len)) ERR_DO (err_unexpected_end (dec))
816
623#define DEC_INC_DEPTH if (++dec->depth > dec->cbor.max_depth) ERR (ERR_NESTING_EXCEEDED) 817#define DEC_INC_DEPTH if (ecb_expect_false (++dec->depth > dec->cbor.max_depth)) ERR (ERR_NESTING_EXCEEDED)
624#define DEC_DEC_DEPTH --dec->depth 818#define DEC_DEC_DEPTH --dec->depth
625 819
626static UV 820static UV
627decode_uint (dec_t *dec) 821decode_uint (dec_t *dec)
628{ 822{
… …
694 888
695 for (;;) 889 for (;;)
696 { 890 {
697 WANT (1); 891 WANT (1);
698 892
699 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF)) 893 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF) || dec->err)
700 { 894 {
701 ++dec->cur; 895 ++dec->cur;
702 break; 896 break;
703 } 897 }
704 898
705 av_push (av, decode_sv (dec)); 899 av_push (av, decode_sv (dec));
706 } 900 }
707 } 901 }
708 else 902 else
709 { 903 {
710 int i, len = decode_uint (dec); 904 UV i, len = decode_uint (dec);
711 905
712 WANT (len); // complexity check for av_fill - need at least one byte per value, do not allow supersize arrays 906 WANT (len); // complexity check for av_fill - need at least one byte per value, do not allow supersize arrays
713 av_fill (av, len - 1); 907 av_fill (av, len - 1);
714 908
715 for (i = 0; i < len; ++i) 909 for (i = 0; i < len; ++i)
… …
718 912
719 DEC_DEC_DEPTH; 913 DEC_DEC_DEPTH;
720 return newRV_noinc ((SV *)av); 914 return newRV_noinc ((SV *)av);
721 915
722fail: 916fail:
723 SvREFCNT_dec (av); 917 SvREFCNT_dec_NN (av);
724 DEC_DEC_DEPTH; 918 DEC_DEC_DEPTH;
725 return &PL_sv_undef; 919 return &PL_sv_undef;
726} 920}
727 921
728static void 922static void
… …
730{ 924{
731 // for speed reasons, we specialcase single-string 925 // for speed reasons, we specialcase single-string
732 // byte or utf-8 strings as keys, but only when !stringref 926 // byte or utf-8 strings as keys, but only when !stringref
733 927
734 if (ecb_expect_true (!dec->stringref)) 928 if (ecb_expect_true (!dec->stringref))
735 if (ecb_expect_true ((*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8)) 929 if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8))
736 { 930 {
737 I32 len = decode_uint (dec); 931 STRLEN len = decode_uint (dec);
738 char *key = (char *)dec->cur; 932 char *key = (char *)dec->cur;
739 933
934 WANT (len);
740 dec->cur += len; 935 dec->cur += len;
741 936
742 hv_store (hv, key, len, decode_sv (dec), 0); 937 hv_store (hv, key, len, decode_sv (dec), 0);
743 938
744 return; 939 return;
745 } 940 }
746 else if (ecb_expect_true ((*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8)) 941 else if (ecb_expect_true ((U8)(*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8))
747 { 942 {
748 I32 len = decode_uint (dec); 943 STRLEN len = decode_uint (dec);
749 char *key = (char *)dec->cur; 944 char *key = (char *)dec->cur;
750 945
946 WANT (len);
751 dec->cur += len; 947 dec->cur += len;
752 948
753 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8)) 949 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
754 if (!is_utf8_string (key, len)) 950 if (!is_utf8_string (key, len))
755 ERR ("corrupted CBOR data (invalid UTF-8 in map key)"); 951 ERR ("corrupted CBOR data (invalid UTF-8 in map key)");
… …
760 } 956 }
761 957
762 SV *k = decode_sv (dec); 958 SV *k = decode_sv (dec);
763 SV *v = decode_sv (dec); 959 SV *v = decode_sv (dec);
764 960
961 // we leak memory if uncaught exceptions are thrown by random magical
962 // methods, and this is hopefully the only place where it can happen,
963 // so if there is a chance of an exception, take the very slow path.
964 // since catching exceptions is "undocumented/internal/forbidden" by
965 // the new p5p powers, we need to call out to a perl function :/
966 if (ecb_expect_false (SvAMAGIC (k)))
967 {
968 dSP;
969
970 ENTER; SAVETMPS;
971 PUSHMARK (SP);
972 EXTEND (SP, 3);
973 PUSHs (sv_2mortal (newRV_inc ((SV *)hv)));
974 PUSHs (sv_2mortal (k));
975 PUSHs (sv_2mortal (v));
976
977 PUTBACK;
978 call_pv ("CBOR::XS::_hv_store", G_VOID | G_DISCARD | G_EVAL);
979 SPAGAIN;
980
981 FREETMPS; LEAVE;
982
983 if (SvTRUE (ERRSV))
984 ERR_ERRSV;
985
986 return;
987 }
988
765 hv_store_ent (hv, k, v, 0); 989 hv_store_ent (hv, k, v, 0);
766 SvREFCNT_dec (k); 990 SvREFCNT_dec_NN (k);
767 991
768fail: 992fail:
769 ; 993 ;
770} 994}
771 995
… …
782 1006
783 for (;;) 1007 for (;;)
784 { 1008 {
785 WANT (1); 1009 WANT (1);
786 1010
787 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF)) 1011 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF) || dec->err)
788 { 1012 {
789 ++dec->cur; 1013 ++dec->cur;
790 break; 1014 break;
791 } 1015 }
792 1016
793 decode_he (dec, hv); 1017 decode_he (dec, hv);
794 } 1018 }
795 } 1019 }
796 else 1020 else
797 { 1021 {
798 int pairs = decode_uint (dec); 1022 UV pairs = decode_uint (dec);
1023
1024 WANT (pairs); // complexity check - need at least one byte per value, do not allow supersize hashes
799 1025
800 while (pairs--) 1026 while (pairs--)
801 decode_he (dec, hv); 1027 decode_he (dec, hv);
802 } 1028 }
803 1029
804 DEC_DEC_DEPTH; 1030 DEC_DEC_DEPTH;
805 return newRV_noinc ((SV *)hv); 1031 return newRV_noinc ((SV *)hv);
806 1032
807fail: 1033fail:
808 SvREFCNT_dec (hv); 1034 SvREFCNT_dec_NN (hv);
809 DEC_DEC_DEPTH; 1035 DEC_DEC_DEPTH;
810 return &PL_sv_undef; 1036 return &PL_sv_undef;
811} 1037}
812 1038
813static SV * 1039static SV *
814decode_str (dec_t *dec, int utf8) 1040decode_str (dec_t *dec, int utf8)
815{ 1041{
816 SV *sv = 0; 1042 SV *sv = 0;
817 1043
818 if ((*dec->cur & MINOR_MASK) == MINOR_INDEF) 1044 if (ecb_expect_false ((*dec->cur & MINOR_MASK) == MINOR_INDEF))
819 { 1045 {
820 // indefinite length strings 1046 // indefinite length strings
821 ++dec->cur; 1047 ++dec->cur;
822 1048
823 U8 major = *dec->cur & MAJOR_MISC; 1049 U8 major = *dec->cur & MAJOR_MISC;
… …
891 sv = newRV_noinc (decode_sv (dec)); 1117 sv = newRV_noinc (decode_sv (dec));
892 break; 1118 break;
893 1119
894 case CBOR_TAG_STRINGREF_NAMESPACE: 1120 case CBOR_TAG_STRINGREF_NAMESPACE:
895 { 1121 {
1122 // do not use SAVETMPS/FREETMPS, as these will
1123 // erase mortalised caches, e.g. "shareable"
896 ENTER; SAVETMPS; 1124 ENTER;
897 1125
898 SAVESPTR (dec->stringref); 1126 SAVESPTR (dec->stringref);
899 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ()); 1127 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ());
900 1128
901 sv = decode_sv (dec); 1129 sv = decode_sv (dec);
902 1130
903 FREETMPS; LEAVE; 1131 LEAVE;
904 } 1132 }
905 break; 1133 break;
906 1134
907 case CBOR_TAG_STRINGREF: 1135 case CBOR_TAG_STRINGREF:
908 { 1136 {
909 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT)) 1137 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
910 ERR ("corrupted CBOR data (stringref index not an unsigned integer)"); 1138 ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
911 1139
912 UV idx = decode_uint (dec); 1140 UV idx = decode_uint (dec);
913 1141
914 if (!dec->stringref || (int)idx > AvFILLp (dec->stringref)) 1142 if (!dec->stringref || idx >= (UV)(1 + AvFILLp (dec->stringref)))
915 ERR ("corrupted CBOR data (stringref index out of bounds or outside namespace)"); 1143 ERR ("corrupted CBOR data (stringref index out of bounds or outside namespace)");
916 1144
917 sv = newSVsv (AvARRAY (dec->stringref)[idx]); 1145 sv = newSVsv (AvARRAY (dec->stringref)[idx]);
918 } 1146 }
919 break; 1147 break;
… …
947 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT)) 1175 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
948 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)"); 1176 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
949 1177
950 UV idx = decode_uint (dec); 1178 UV idx = decode_uint (dec);
951 1179
952 if (!dec->shareable || (int)idx > AvFILLp (dec->shareable)) 1180 if (!dec->shareable || idx >= (UV)(1 + AvFILLp (dec->shareable)))
953 ERR ("corrupted CBOR data (sharedref index out of bounds)"); 1181 ERR ("corrupted CBOR data (sharedref index out of bounds)");
954 1182
955 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]); 1183 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]);
956 1184
957 if (sv == &PL_sv_undef) 1185 if (sv == &PL_sv_undef)
… …
959 } 1187 }
960 break; 1188 break;
961 1189
962 case CBOR_TAG_PERL_OBJECT: 1190 case CBOR_TAG_PERL_OBJECT:
963 { 1191 {
1192 if (dec->cbor.flags & F_FORBID_OBJECTS)
1193 goto filter;
1194
964 sv = decode_sv (dec); 1195 sv = decode_sv (dec);
965 1196
966 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV) 1197 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
967 ERR ("corrupted CBOR data (non-array perl object)"); 1198 ERR ("corrupted CBOR data (non-array perl object)");
968 1199
… …
978 if (!method) 1209 if (!method)
979 ERR ("cannot decode perl-object (package does not have a THAW method)"); 1210 ERR ("cannot decode perl-object (package does not have a THAW method)");
980 1211
981 dSP; 1212 dSP;
982 1213
983 ENTER; SAVETMPS; PUSHMARK (SP); 1214 ENTER; SAVETMPS;
1215 PUSHMARK (SP);
984 EXTEND (SP, len + 1); 1216 EXTEND (SP, len + 1);
985 // we re-bless the reference to get overload and other niceties right 1217 // we re-bless the reference to get overload and other niceties right
986 PUSHs (*av_fetch (av, 0, 1)); 1218 PUSHs (*av_fetch (av, 0, 1));
987 PUSHs (sv_cbor); 1219 PUSHs (sv_cbor);
988 1220
… …
996 SPAGAIN; 1228 SPAGAIN;
997 1229
998 if (SvTRUE (ERRSV)) 1230 if (SvTRUE (ERRSV))
999 { 1231 {
1000 FREETMPS; LEAVE; 1232 FREETMPS; LEAVE;
1001 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV)))); 1233 ERR_ERRSV;
1002 } 1234 }
1003 1235
1004 SvREFCNT_dec (sv); 1236 SvREFCNT_dec_NN (sv);
1005 sv = SvREFCNT_inc (POPs); 1237 sv = SvREFCNT_inc (POPs);
1006 1238
1007 PUTBACK; 1239 PUTBACK;
1008 1240
1009 FREETMPS; LEAVE; 1241 FREETMPS; LEAVE;
1010 } 1242 }
1011 break; 1243 break;
1012 1244
1013 default: 1245 default:
1246 filter:
1014 { 1247 {
1248 SV *tag_sv = newSVuv (tag);
1249
1015 sv = decode_sv (dec); 1250 sv = decode_sv (dec);
1016 1251
1017 dSP; 1252 dSP;
1018 ENTER; SAVETMPS; PUSHMARK (SP); 1253 ENTER; SAVETMPS;
1254 PUSHMARK (SP);
1019 EXTEND (SP, 2); 1255 EXTEND (SP, 2);
1020 PUSHs (newSVuv (tag)); 1256 PUSHs (tag_sv);
1021 PUSHs (sv); 1257 PUSHs (sv);
1022 1258
1023 PUTBACK; 1259 PUTBACK;
1024 int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL); 1260 int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL);
1025 SPAGAIN; 1261 SPAGAIN;
1026 1262
1027 if (SvTRUE (ERRSV)) 1263 if (SvTRUE (ERRSV))
1028 { 1264 {
1265 SvREFCNT_dec_NN (tag_sv);
1029 FREETMPS; LEAVE; 1266 FREETMPS; LEAVE;
1030 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV)))); 1267 ERR_ERRSV;
1031 } 1268 }
1032 1269
1033 if (count) 1270 if (count)
1034 { 1271 {
1272 SvREFCNT_dec_NN (tag_sv);
1035 SvREFCNT_dec (sv); 1273 SvREFCNT_dec_NN (sv);
1036 sv = SvREFCNT_inc (POPs); 1274 sv = SvREFCNT_inc_NN (TOPs);
1275 SP -= count;
1037 } 1276 }
1038 else 1277 else
1039 { 1278 {
1040 AV *av = newAV (); 1279 AV *av = newAV ();
1041 av_push (av, newSVuv (tag)); 1280 av_push (av, tag_sv);
1042 av_push (av, sv); 1281 av_push (av, sv);
1043 1282
1044 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 1283 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
1045 ? cbor_tagged_stash 1284 ? cbor_tagged_stash
1046 : gv_stashpv ("CBOR::XS::Tagged" , 1); 1285 : gv_stashpv ("CBOR::XS::Tagged" , 1);
… …
1132 1371
1133 return newSVnv (ecb_binary64_to_double (fp)); 1372 return newSVnv (ecb_binary64_to_double (fp));
1134 } 1373 }
1135 1374
1136 // 0..19 unassigned simple 1375 // 0..19 unassigned simple
1137 // 24 reserved + unassigned (reserved values are not encodable) 1376 // 24 reserved + unassigned simple (reserved values are not encodable)
1377 // 28-30 unassigned misc
1378 // 31 break code
1138 default: 1379 default:
1139 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)"); 1380 ERR ("corrupted CBOR data (reserved/unassigned/unexpected major 7 value)");
1140 } 1381 }
1141 1382
1142 break; 1383 break;
1143 } 1384 }
1144 1385
… …
1147} 1388}
1148 1389
1149static SV * 1390static SV *
1150decode_cbor (SV *string, CBOR *cbor, char **offset_return) 1391decode_cbor (SV *string, CBOR *cbor, char **offset_return)
1151{ 1392{
1152 dec_t dec = { }; 1393 dec_t dec = { 0 };
1153 SV *sv; 1394 SV *sv;
1154 STRLEN len; 1395 STRLEN len;
1155 char *data = SvPVbyte (string, len); 1396 char *data = SvPVbyte (string, len);
1156 1397
1157 if (len > cbor->max_size && cbor->max_size) 1398 if (len > cbor->max_size && cbor->max_size)
… …
1173 1414
1174 if (dec.err) 1415 if (dec.err)
1175 { 1416 {
1176 if (dec.shareable) 1417 if (dec.shareable)
1177 { 1418 {
1178 // need to break cyclic links, which whould all be in shareable 1419 // need to break cyclic links, which would all be in shareable
1179 int i; 1420 int i;
1180 SV **svp; 1421 SV **svp;
1181 1422
1182 for (i = av_len (dec.shareable) + 1; i--; ) 1423 for (i = av_len (dec.shareable) + 1; i--; )
1183 if ((svp = av_fetch (dec.shareable, i, 0))) 1424 if ((svp = av_fetch (dec.shareable, i, 0)))
1184 sv_setsv (*svp, &PL_sv_undef); 1425 sv_setsv (*svp, &PL_sv_undef);
1185 } 1426 }
1186 1427
1187 SvREFCNT_dec (sv); 1428 SvREFCNT_dec_NN (sv);
1429
1430 if (dec.err_sv)
1431 sv_2mortal (dec.err_sv);
1432
1188 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur); 1433 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur);
1189 } 1434 }
1190 1435
1191 sv = sv_2mortal (sv); 1436 sv = sv_2mortal (sv);
1192 1437
1193 return sv; 1438 return sv;
1194} 1439}
1195 1440
1441/////////////////////////////////////////////////////////////////////////////
1442// incremental parser
1443
1444#define INCR_DONE(cbor) (AvFILLp (cbor->incr_count) < 0)
1445
1446// returns 0 for notyet, 1 for success or error
1447static int
1448incr_parse (CBOR *self, SV *cborstr)
1449{
1450 STRLEN cur;
1451 SvPV (cborstr, cur);
1452
1453 while (ecb_expect_true (self->incr_need <= cur))
1454 {
1455 // table of integer count bytes
1456 static I8 incr_len[MINOR_MASK + 1] = {
1457 0, 0, 0, 0, 0, 0, 0, 0,
1458 0, 0, 0, 0, 0, 0, 0, 0,
1459 0, 0, 0, 0, 0, 0, 0, 0,
1460 1, 2, 4, 8,-1,-1,-1,-2
1461 };
1462
1463 const U8 *p = SvPVX (cborstr) + self->incr_pos;
1464 U8 m = *p & MINOR_MASK;
1465 IV count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1466 I8 ilen = incr_len[m];
1467
1468 self->incr_need = self->incr_pos + 1;
1469
1470 if (ecb_expect_false (ilen < 0))
1471 {
1472 if (m != MINOR_INDEF)
1473 return 1; // error
1474
1475 if (*p == (MAJOR_MISC | MINOR_INDEF))
1476 {
1477 if (count >= 0)
1478 return 1; // error
1479
1480 count = 1;
1481 }
1482 else
1483 {
1484 av_push (self->incr_count, newSViv (-1)); //TODO: nest
1485 count = -1;
1486 }
1487 }
1488 else
1489 {
1490 self->incr_need += ilen;
1491 if (ecb_expect_false (self->incr_need > cur))
1492 return 0;
1493
1494 int major = *p >> MAJOR_SHIFT;
1495
1496 switch (major)
1497 {
1498 case MAJOR_TAG >> MAJOR_SHIFT:
1499 ++count; // tags merely prefix another value
1500 break;
1501
1502 case MAJOR_BYTES >> MAJOR_SHIFT:
1503 case MAJOR_TEXT >> MAJOR_SHIFT:
1504 case MAJOR_ARRAY >> MAJOR_SHIFT:
1505 case MAJOR_MAP >> MAJOR_SHIFT:
1506 {
1507 UV len;
1508
1509 if (ecb_expect_false (ilen))
1510 {
1511 len = 0;
1512
1513 do {
1514 len = (len << 8) | *++p;
1515 } while (--ilen);
1516 }
1517 else
1518 len = m;
1519
1520 switch (major)
1521 {
1522 case MAJOR_BYTES >> MAJOR_SHIFT:
1523 case MAJOR_TEXT >> MAJOR_SHIFT:
1524 self->incr_need += len;
1525 if (ecb_expect_false (self->incr_need > cur))
1526 return 0;
1527
1528 break;
1529
1530 case MAJOR_MAP >> MAJOR_SHIFT:
1531 len <<= 1;
1532 case MAJOR_ARRAY >> MAJOR_SHIFT:
1533 if (len)
1534 {
1535 av_push (self->incr_count, newSViv (len + 1)); //TODO: nest
1536 count = len + 1;
1537 }
1538 break;
1539 }
1540 }
1541 }
1542 }
1543
1544 self->incr_pos = self->incr_need;
1545
1546 if (count > 0)
1547 {
1548 while (!--count)
1549 {
1550 if (!AvFILLp (self->incr_count))
1551 return 1; // done
1552
1553 SvREFCNT_dec_NN (av_pop (self->incr_count));
1554 count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1555 }
1556
1557 SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]) = count;
1558 }
1559 }
1560
1561 return 0;
1562}
1563
1564
1196///////////////////////////////////////////////////////////////////////////// 1565/////////////////////////////////////////////////////////////////////////////
1197// XS interface functions 1566// XS interface functions
1198 1567
1199MODULE = CBOR::XS PACKAGE = CBOR::XS 1568MODULE = CBOR::XS PACKAGE = CBOR::XS
1200 1569
… …
1212 1581
1213 default_filter = newSVpv ("CBOR::XS::default_filter", 0); 1582 default_filter = newSVpv ("CBOR::XS::default_filter", 0);
1214 1583
1215 sv_cbor = newSVpv ("CBOR", 0); 1584 sv_cbor = newSVpv ("CBOR", 0);
1216 SvREADONLY_on (sv_cbor); 1585 SvREADONLY_on (sv_cbor);
1586
1587 assert (("STRLEN must be an unsigned type", 0 <= (STRLEN)-1));
1217} 1588}
1218 1589
1219PROTOTYPES: DISABLE 1590PROTOTYPES: DISABLE
1220 1591
1221void CLONE (...) 1592void CLONE (...)
… …
1241 ALIAS: 1612 ALIAS:
1242 shrink = F_SHRINK 1613 shrink = F_SHRINK
1243 allow_unknown = F_ALLOW_UNKNOWN 1614 allow_unknown = F_ALLOW_UNKNOWN
1244 allow_sharing = F_ALLOW_SHARING 1615 allow_sharing = F_ALLOW_SHARING
1245 allow_cycles = F_ALLOW_CYCLES 1616 allow_cycles = F_ALLOW_CYCLES
1617 forbid_objects = F_FORBID_OBJECTS
1246 pack_strings = F_PACK_STRINGS 1618 pack_strings = F_PACK_STRINGS
1619 text_keys = F_TEXT_KEYS
1620 text_strings = F_TEXT_STRINGS
1247 validate_utf8 = F_VALIDATE_UTF8 1621 validate_utf8 = F_VALIDATE_UTF8
1248 PPCODE: 1622 PPCODE:
1249{ 1623{
1250 if (enable) 1624 if (enable)
1251 self->flags |= ix; 1625 self->flags |= ix;
… …
1259 ALIAS: 1633 ALIAS:
1260 get_shrink = F_SHRINK 1634 get_shrink = F_SHRINK
1261 get_allow_unknown = F_ALLOW_UNKNOWN 1635 get_allow_unknown = F_ALLOW_UNKNOWN
1262 get_allow_sharing = F_ALLOW_SHARING 1636 get_allow_sharing = F_ALLOW_SHARING
1263 get_allow_cycles = F_ALLOW_CYCLES 1637 get_allow_cycles = F_ALLOW_CYCLES
1638 get_forbid_objects = F_FORBID_OBJECTS
1264 get_pack_strings = F_PACK_STRINGS 1639 get_pack_strings = F_PACK_STRINGS
1640 get_text_keys = F_TEXT_KEYS
1641 get_text_strings = F_TEXT_STRINGS
1265 get_validate_utf8 = F_VALIDATE_UTF8 1642 get_validate_utf8 = F_VALIDATE_UTF8
1266 PPCODE: 1643 PPCODE:
1267 XPUSHs (boolSV (self->flags & ix)); 1644 XPUSHs (boolSV (self->flags & ix));
1268 1645
1269void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) 1646void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
… …
1319 EXTEND (SP, 2); 1696 EXTEND (SP, 2);
1320 PUSHs (sv); 1697 PUSHs (sv);
1321 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); 1698 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
1322} 1699}
1323 1700
1701void incr_parse (CBOR *self, SV *cborstr)
1702 ALIAS:
1703 incr_parse_multiple = 1
1704 PPCODE:
1705{
1706 if (SvUTF8 (cborstr))
1707 sv_utf8_downgrade (cborstr, 0);
1708
1709 if (!self->incr_count)
1710 {
1711 self->incr_count = newAV ();
1712 self->incr_pos = 0;
1713 self->incr_need = 1;
1714
1715 av_push (self->incr_count, newSViv (1));
1716 }
1717
1718 do
1719 {
1720 if (!incr_parse (self, cborstr))
1721 {
1722 if (self->incr_need > self->max_size && self->max_size)
1723 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
1724 (unsigned long)self->incr_need, (unsigned long)self->max_size);
1725
1726 break;
1727 }
1728
1729 SV *sv;
1730 char *offset;
1731
1732 PUTBACK; sv = decode_cbor (cborstr, self, &offset); SPAGAIN;
1733 XPUSHs (sv);
1734
1735 sv_chop (cborstr, offset);
1736
1737 av_clear (self->incr_count);
1738 av_push (self->incr_count, newSViv (1));
1739
1740 self->incr_pos = 0;
1741 self->incr_need = self->incr_pos + 1;
1742 }
1743 while (ix);
1744}
1745
1746void incr_reset (CBOR *self)
1747 CODE:
1748{
1749 SvREFCNT_dec (self->incr_count);
1750 self->incr_count = 0;
1751}
1752
1324void DESTROY (CBOR *self) 1753void DESTROY (CBOR *self)
1325 PPCODE: 1754 PPCODE:
1326 cbor_free (self); 1755 cbor_free (self);
1327 1756
1328PROTOTYPES: ENABLE 1757PROTOTYPES: ENABLE
… …
1347 cbor_init (&cbor); 1776 cbor_init (&cbor);
1348 PUTBACK; cborstr = decode_cbor (cborstr, &cbor, 0); SPAGAIN; 1777 PUTBACK; cborstr = decode_cbor (cborstr, &cbor, 0); SPAGAIN;
1349 XPUSHs (cborstr); 1778 XPUSHs (cborstr);
1350} 1779}
1351 1780
1781#ifdef __AFL_COMPILER
1782
1783void
1784afl_init ()
1785 CODE:
1786 __AFL_INIT ();
1787
1788int
1789afl_loop (unsigned int count = 10000)
1790 CODE:
1791 RETVAL = __AFL_LOOP (count);
1792 OUTPUT:
1793 RETVAL
1794
1795#endif
1796

Diff Legend

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