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 1.37 by root, Sat Nov 30 18:13:53 2013 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
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_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
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
125#define F_VALIDATE_UTF8 0x00000100UL
104 126
105#define INIT_SIZE 32 // initial scalar size to be allocated 127#define INIT_SIZE 32 // initial scalar size to be allocated
106 128
107#define SB do { 129#define SB do {
108#define SE } while (0) 130#define SE } while (0)
… …
127typedef struct { 149typedef struct {
128 U32 flags; 150 U32 flags;
129 U32 max_depth; 151 U32 max_depth;
130 STRLEN max_size; 152 STRLEN max_size;
131 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.
132} CBOR; 159} CBOR;
133 160
134ecb_inline void 161ecb_inline void
135cbor_init (CBOR *cbor) 162cbor_init (CBOR *cbor)
136{ 163{
… …
140 167
141ecb_inline void 168ecb_inline void
142cbor_free (CBOR *cbor) 169cbor_free (CBOR *cbor)
143{ 170{
144 SvREFCNT_dec (cbor->filter); 171 SvREFCNT_dec (cbor->filter);
172 SvREFCNT_dec (cbor->incr_count);
145} 173}
146 174
147///////////////////////////////////////////////////////////////////////////// 175/////////////////////////////////////////////////////////////////////////////
148// utility functions 176// utility functions
149 177
… …
175 203
176// minimum length of a string to be registered for stringref 204// minimum length of a string to be registered for stringref
177ecb_inline int 205ecb_inline int
178minimum_string_length (UV idx) 206minimum_string_length (UV idx)
179{ 207{
180 return idx > 23 208 return idx <= 23 ? 3
181 ? idx > 0xffU 209 : idx <= 0xffU ? 4
182 ? idx > 0xffffU 210 : idx <= 0xffffU ? 5
183 ? idx > 0xffffffffU 211 : idx <= 0xffffffffU ? 7
184 ? 11 212 : 11;
185 : 7
186 : 5
187 : 4
188 : 3;
189} 213}
190 214
191///////////////////////////////////////////////////////////////////////////// 215/////////////////////////////////////////////////////////////////////////////
192// encoder 216// encoder
193 217
… …
206} enc_t; 230} enc_t;
207 231
208ecb_inline void 232ecb_inline void
209need (enc_t *enc, STRLEN len) 233need (enc_t *enc, STRLEN len)
210{ 234{
211 if (ecb_expect_false (enc->cur + len >= enc->end)) 235 if (ecb_expect_false ((uintptr_t)(enc->end - enc->cur) < len))
212 { 236 {
213 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv); 237 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
214 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1); 238 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
215 enc->cur = SvPVX (enc->sv) + cur; 239 enc->cur = SvPVX (enc->sv) + cur;
216 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; 240 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
… …
222{ 246{
223 need (enc, 1); 247 need (enc, 1);
224 *enc->cur++ = ch; 248 *enc->cur++ = ch;
225} 249}
226 250
251// used for tags, intregers, element counts and so on
227static void 252static void
228encode_uint (enc_t *enc, int major, UV len) 253encode_uint (enc_t *enc, int major, UV len)
229{ 254{
230 need (enc, 9); 255 need (enc, 9);
231 256
… …
262 *enc->cur++ = len >> 8; 287 *enc->cur++ = len >> 8;
263 *enc->cur++ = len; 288 *enc->cur++ = len;
264 } 289 }
265} 290}
266 291
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
267ecb_inline void 304ecb_inline void
268encode_tag (enc_t *enc, UV tag) 305encode_tag (enc_t *enc, UV tag)
269{ 306{
270 encode_uint (enc, MAJOR_TAG, tag); 307 encode_uint (enc, MAJOR_TAG, tag);
271} 308}
272 309
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
273ecb_inline void 333ecb_inline void
274encode_str (enc_t *enc, int utf8, char *str, STRLEN len) 334encode_str (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len)
275{ 335{
336 if (ecb_expect_false (upgrade_utf8))
337 if (!utf8)
338 {
339 encode_str_utf8 (enc, utf8, str, len);
340 return;
341 }
342
276 encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len); 343 encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len);
277 need (enc, len); 344 need (enc, len);
278 memcpy (enc->cur, str, len); 345 memcpy (enc->cur, str, len);
279 enc->cur += len; 346 enc->cur += len;
280} 347}
281 348
282static void 349ecb_inline void
283encode_strref (enc_t *enc, int utf8, char *str, STRLEN len) 350encode_strref (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len)
284{ 351{
285 if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS)) 352 if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS))
286 { 353 {
287 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1); 354 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1);
288 355
… …
299 sv_setuv (*svp, enc->stringref_idx); 366 sv_setuv (*svp, enc->stringref_idx);
300 ++enc->stringref_idx; 367 ++enc->stringref_idx;
301 } 368 }
302 } 369 }
303 370
304 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 }
305} 469}
306 470
307static void encode_sv (enc_t *enc, SV *sv); 471static void encode_sv (enc_t *enc, SV *sv);
308 472
309static void 473static void
… …
316 480
317 ++enc->depth; 481 ++enc->depth;
318 482
319 encode_uint (enc, MAJOR_ARRAY, len + 1); 483 encode_uint (enc, MAJOR_ARRAY, len + 1);
320 484
485 if (ecb_expect_false (SvMAGICAL (av)))
321 for (i = 0; i <= len; ++i) 486 for (i = 0; i <= len; ++i)
322 { 487 {
323 SV **svp = av_fetch (av, i, 0); 488 SV **svp = av_fetch (av, i, 0);
324 encode_sv (enc, svp ? *svp : &PL_sv_undef); 489 encode_sv (enc, svp ? *svp : &PL_sv_undef);
325 } 490 }
491 else
492 for (i = 0; i <= len; ++i)
493 {
494 SV *sv = AvARRAY (av)[i];
495 encode_sv (enc, sv ? sv : &PL_sv_undef);
496 }
326 497
327 --enc->depth; 498 --enc->depth;
328} 499}
329 500
330static void 501static void
… …
338 ++enc->depth; 509 ++enc->depth;
339 510
340 int pairs = hv_iterinit (hv); 511 int pairs = hv_iterinit (hv);
341 int mg = SvMAGICAL (hv); 512 int mg = SvMAGICAL (hv);
342 513
343 if (mg) 514 if (ecb_expect_false (mg))
344 encode_ch (enc, MAJOR_MAP | MINOR_INDEF); 515 encode_ch (enc, MAJOR_MAP | MINOR_INDEF);
345 else 516 else
346 encode_uint (enc, MAJOR_MAP, pairs); 517 encode_uint (enc, MAJOR_MAP, pairs);
347 518
348 while ((he = hv_iternext (hv))) 519 while ((he = hv_iternext (hv)))
349 { 520 {
350 if (HeKLEN (he) == HEf_SVKEY) 521 if (HeKLEN (he) == HEf_SVKEY)
351 encode_sv (enc, HeSVKEY (he)); 522 encode_sv (enc, HeSVKEY (he));
352 else 523 else
353 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));
354 525
355 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));
356 } 527 }
357 528
358 if (mg) 529 if (ecb_expect_false (mg))
359 encode_ch (enc, MAJOR_MISC | MINOR_INDEF); 530 encode_ch (enc, MAJOR_MISC | MINOR_INDEF);
360 531
361 --enc->depth; 532 --enc->depth;
362} 533}
363 534
… …
383 554
384 HV *stash = SvSTASH (sv); 555 HV *stash = SvSTASH (sv);
385 556
386 if (stash == boolean_stash) 557 if (stash == boolean_stash)
387 { 558 {
388 encode_ch (enc, SvIV (sv) ? MAJOR_MISC | SIMPLE_TRUE : MAJOR_MISC | SIMPLE_FALSE); 559 encode_bool (enc, SvIV (sv));
389 return; 560 return;
390 } 561 }
391 else if (stash == error_stash) 562 else if (stash == error_stash)
392 { 563 {
393 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF); 564 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
… …
396 else if (stash == tagged_stash) 567 else if (stash == tagged_stash)
397 { 568 {
398 if (svt != SVt_PVAV) 569 if (svt != SVt_PVAV)
399 croak ("encountered CBOR::XS::Tagged object that isn't an array"); 570 croak ("encountered CBOR::XS::Tagged object that isn't an array");
400 571
572 switch (av_len ((AV *)sv))
573 {
574 case 2-1:
575 // actually a tagged value
401 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)));
402 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 }
403 588
404 return; 589 return;
405 } 590 }
406 } 591 }
407 592
408 if (ecb_expect_false (SvREFCNT (sv) > 1) 593 if (ecb_expect_false (SvREFCNT (sv) > 1)
409 && ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING)) 594 && ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING))
410 { 595 {
411 if (!enc->shareable) 596 if (ecb_expect_false (!enc->shareable))
412 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ()); 597 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
413 598
414 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1); 599 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
415 600
416 if (SvOK (*svp)) 601 if (SvOK (*svp))
… …
430 if (ecb_expect_false (SvOBJECT (sv))) 615 if (ecb_expect_false (SvOBJECT (sv)))
431 { 616 {
432 HV *stash = SvSTASH (sv); 617 HV *stash = SvSTASH (sv);
433 GV *method; 618 GV *method;
434 619
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))));
435 if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0))) 623 else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
436 { 624 {
437 dSP; 625 dSP;
438 626
439 ENTER; SAVETMPS; PUSHMARK (SP); 627 ENTER; SAVETMPS;
628 PUSHMARK (SP);
440 // we re-bless the reference to get overload and other niceties right 629 // we re-bless the reference to get overload and other niceties right
441 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); 630 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
442 631
443 PUTBACK; 632 PUTBACK;
444 // G_SCALAR ensures that return value is 1 633 // G_SCALAR ensures that return value is 1
… …
457 } 646 }
458 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0) 647 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)
459 { 648 {
460 dSP; 649 dSP;
461 650
462 ENTER; SAVETMPS; PUSHMARK (SP); 651 ENTER; SAVETMPS;
652 PUSHMARK (SP);
463 EXTEND (SP, 2); 653 EXTEND (SP, 2);
464 // 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
465 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); 655 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
466 PUSHs (sv_cbor); 656 PUSHs (sv_cbor);
467 657
… …
473 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv) 663 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)
474 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));
475 665
476 encode_tag (enc, CBOR_TAG_PERL_OBJECT); 666 encode_tag (enc, CBOR_TAG_PERL_OBJECT);
477 encode_uint (enc, MAJOR_ARRAY, count + 1); 667 encode_uint (enc, MAJOR_ARRAY, count + 1);
478 encode_strref (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash)); 668 encode_strref (enc, 0, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
479 669
480 while (count) 670 {
671 int i;
672
673 for (i = 0; i < count; ++i)
481 encode_sv (enc, SP[1 - count--]); 674 encode_sv (enc, SP[i + 1 - count]);
675
676 SP -= count;
677 }
482 678
483 PUTBACK; 679 PUTBACK;
484 680
485 FREETMPS; LEAVE; 681 FREETMPS; LEAVE;
486 } 682 }
… …
508 704
509 if (ecb_expect_false (nv == (NV)(U32)nv)) 705 if (ecb_expect_false (nv == (NV)(U32)nv))
510 encode_uint (enc, MAJOR_POS_INT, (U32)nv); 706 encode_uint (enc, MAJOR_POS_INT, (U32)nv);
511 //TODO: maybe I32? 707 //TODO: maybe I32?
512 else if (ecb_expect_false (nv == (float)nv)) 708 else if (ecb_expect_false (nv == (float)nv))
513 { 709 encode_float32 (enc, nv);
514 uint32_t fp = ecb_float_to_binary32 (nv);
515
516 *enc->cur++ = MAJOR_MISC | MISC_FLOAT32;
517
518 if (!ecb_big_endian ())
519 fp = ecb_bswap32 (fp);
520
521 memcpy (enc->cur, &fp, 4);
522 enc->cur += 4;
523 }
524 else 710 else
525 { 711 encode_float64 (enc, nv);
526 uint64_t fp = ecb_double_to_binary64 (nv);
527
528 *enc->cur++ = MAJOR_MISC | MISC_FLOAT64;
529
530 if (!ecb_big_endian ())
531 fp = ecb_bswap64 (fp);
532
533 memcpy (enc->cur, &fp, 8);
534 enc->cur += 8;
535 }
536} 712}
537 713
538static void 714static void
539encode_sv (enc_t *enc, SV *sv) 715encode_sv (enc_t *enc, SV *sv)
540{ 716{
… …
542 718
543 if (SvPOKp (sv)) 719 if (SvPOKp (sv))
544 { 720 {
545 STRLEN len; 721 STRLEN len;
546 char *str = SvPV (sv, len); 722 char *str = SvPV (sv, len);
547 encode_strref (enc, SvUTF8 (sv), str, len); 723 encode_strref (enc, enc->cbor.flags & F_TEXT_STRINGS, SvUTF8 (sv), str, len);
548 } 724 }
549 else if (SvNOKp (sv)) 725 else if (SvNOKp (sv))
550 encode_nv (enc, sv); 726 encode_nv (enc, sv);
551 else if (SvIOKp (sv)) 727 else if (SvIOKp (sv))
552 { 728 encode_int (enc, sv);
553 if (SvIsUV (sv))
554 encode_uint (enc, MAJOR_POS_INT, SvUVX (sv));
555 else if (SvIVX (sv) >= 0)
556 encode_uint (enc, MAJOR_POS_INT, SvIVX (sv));
557 else
558 encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1));
559 }
560 else if (SvROK (sv)) 729 else if (SvROK (sv))
561 encode_rv (enc, SvRV (sv)); 730 encode_rv (enc, SvRV (sv));
562 else if (!SvOK (sv)) 731 else if (!SvOK (sv))
563 encode_ch (enc, MAJOR_MISC | SIMPLE_NULL); 732 encode_ch (enc, MAJOR_MISC | SIMPLE_NULL);
564 else if (enc->cbor.flags & F_ALLOW_UNKNOWN) 733 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
… …
569} 738}
570 739
571static SV * 740static SV *
572encode_cbor (SV *scalar, CBOR *cbor) 741encode_cbor (SV *scalar, CBOR *cbor)
573{ 742{
574 enc_t enc = { }; 743 enc_t enc = { 0 };
575 744
576 enc.cbor = *cbor; 745 enc.cbor = *cbor;
577 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 746 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
578 enc.cur = SvPVX (enc.sv); 747 enc.cur = SvPVX (enc.sv);
579 enc.end = SvEND (enc.sv); 748 enc.end = SvEND (enc.sv);
580 749
581 SvPOK_only (enc.sv); 750 SvPOK_only (enc.sv);
582 751
583 if (cbor->flags & F_PACK_STRINGS) 752 if (cbor->flags & F_PACK_STRINGS)
584 { 753 {
… …
611 U32 depth; // recursion depth 780 U32 depth; // recursion depth
612 U32 maxdepth; // recursion depth limit 781 U32 maxdepth; // recursion depth limit
613 AV *shareable; 782 AV *shareable;
614 AV *stringref; 783 AV *stringref;
615 SV *decode_tagged; 784 SV *decode_tagged;
785 SV *err_sv; // optional sv for error, needs to be freed
616} dec_t; 786} dec_t;
617 787
618#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);
619 795
620#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;
621 799
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
622#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)
623#define DEC_DEC_DEPTH --dec->depth 825#define DEC_DEC_DEPTH --dec->depth
624 826
625static UV 827static UV
626decode_uint (dec_t *dec) 828decode_uint (dec_t *dec)
627{ 829{
… …
693 895
694 for (;;) 896 for (;;)
695 { 897 {
696 WANT (1); 898 WANT (1);
697 899
698 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF)) 900 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF) || dec->err)
699 { 901 {
700 ++dec->cur; 902 ++dec->cur;
701 break; 903 break;
702 } 904 }
703 905
704 av_push (av, decode_sv (dec)); 906 av_push (av, decode_sv (dec));
705 } 907 }
706 } 908 }
707 else 909 else
708 { 910 {
709 int i, len = decode_uint (dec); 911 UV i, len = decode_uint (dec);
710 912
711 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
712 av_fill (av, len - 1); 914 av_fill (av, len - 1);
713 915
714 for (i = 0; i < len; ++i) 916 for (i = 0; i < len; ++i)
… …
717 919
718 DEC_DEC_DEPTH; 920 DEC_DEC_DEPTH;
719 return newRV_noinc ((SV *)av); 921 return newRV_noinc ((SV *)av);
720 922
721fail: 923fail:
722 SvREFCNT_dec (av); 924 SvREFCNT_dec_NN (av);
723 DEC_DEC_DEPTH; 925 DEC_DEC_DEPTH;
724 return &PL_sv_undef; 926 return &PL_sv_undef;
725} 927}
726 928
727static void 929static void
… …
729{ 931{
730 // for speed reasons, we specialcase single-string 932 // for speed reasons, we specialcase single-string
731 // byte or utf-8 strings as keys, but only when !stringref 933 // byte or utf-8 strings as keys, but only when !stringref
732 934
733 if (ecb_expect_true (!dec->stringref)) 935 if (ecb_expect_true (!dec->stringref))
734 if (ecb_expect_true ((*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8)) 936 if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8))
735 { 937 {
736 I32 len = decode_uint (dec); 938 STRLEN len = decode_uint (dec);
737 char *key = (char *)dec->cur; 939 char *key = (char *)dec->cur;
738 940
941 WANT (len);
739 dec->cur += len; 942 dec->cur += len;
740 943
741 if (ecb_expect_false (dec->stringref))
742 av_push (dec->stringref, newSVpvn (key, len));
743
744 hv_store (hv, key, len, decode_sv (dec), 0); 944 hv_store (hv, key, len, decode_sv (dec), 0);
745 945
746 return; 946 return;
747 } 947 }
748 else if (ecb_expect_true ((*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8)) 948 else if (ecb_expect_true ((U8)(*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8))
749 { 949 {
750 I32 len = decode_uint (dec); 950 STRLEN len = decode_uint (dec);
751 char *key = (char *)dec->cur; 951 char *key = (char *)dec->cur;
752 952
953 WANT (len);
753 dec->cur += len; 954 dec->cur += len;
754 955
755 if (ecb_expect_false (dec->stringref)) 956 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
756 av_push (dec->stringref, newSVpvn_utf8 (key, len, 1)); 957 if (!is_utf8_string (key, len))
958 ERR ("corrupted CBOR data (invalid UTF-8 in map key)");
757 959
758 hv_store (hv, key, -len, decode_sv (dec), 0); 960 hv_store (hv, key, -len, decode_sv (dec), 0);
759 961
760 return; 962 return;
761 } 963 }
762 964
763 SV *k = decode_sv (dec); 965 SV *k = decode_sv (dec);
764 SV *v = decode_sv (dec); 966 SV *v = decode_sv (dec);
765 967
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
766 hv_store_ent (hv, k, v, 0); 996 hv_store_ent (hv, k, v, 0);
767 SvREFCNT_dec (k); 997 SvREFCNT_dec_NN (k);
998
999fail:
1000 ;
768} 1001}
769 1002
770static SV * 1003static SV *
771decode_hv (dec_t *dec) 1004decode_hv (dec_t *dec)
772{ 1005{
… …
780 1013
781 for (;;) 1014 for (;;)
782 { 1015 {
783 WANT (1); 1016 WANT (1);
784 1017
785 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF)) 1018 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF) || dec->err)
786 { 1019 {
787 ++dec->cur; 1020 ++dec->cur;
788 break; 1021 break;
789 } 1022 }
790 1023
791 decode_he (dec, hv); 1024 decode_he (dec, hv);
792 } 1025 }
793 } 1026 }
794 else 1027 else
795 { 1028 {
796 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
797 1032
798 while (pairs--) 1033 while (pairs--)
799 decode_he (dec, hv); 1034 decode_he (dec, hv);
800 } 1035 }
801 1036
802 DEC_DEC_DEPTH; 1037 DEC_DEC_DEPTH;
803 return newRV_noinc ((SV *)hv); 1038 return newRV_noinc ((SV *)hv);
804 1039
805fail: 1040fail:
806 SvREFCNT_dec (hv); 1041 SvREFCNT_dec_NN (hv);
807 DEC_DEC_DEPTH; 1042 DEC_DEC_DEPTH;
808 return &PL_sv_undef; 1043 return &PL_sv_undef;
809} 1044}
810 1045
811static SV * 1046static SV *
812decode_str (dec_t *dec, int utf8) 1047decode_str (dec_t *dec, int utf8)
813{ 1048{
814 SV *sv = 0; 1049 SV *sv = 0;
815 1050
816 if ((*dec->cur & MINOR_MASK) == MINOR_INDEF) 1051 if (ecb_expect_false ((*dec->cur & MINOR_MASK) == MINOR_INDEF))
817 { 1052 {
818 // indefinite length strings 1053 // indefinite length strings
819 ++dec->cur; 1054 ++dec->cur;
820 1055
821 U8 major = *dec->cur & MAJOR_MISC; 1056 U8 major = *dec->cur & MAJOR_MISC;
… …
854 && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1)) 1089 && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1))
855 av_push (dec->stringref, SvREFCNT_inc_NN (sv)); 1090 av_push (dec->stringref, SvREFCNT_inc_NN (sv));
856 } 1091 }
857 1092
858 if (utf8) 1093 if (utf8)
1094 {
1095 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
1096 if (!is_utf8_string (SvPVX (sv), SvCUR (sv)))
1097 ERR ("corrupted CBOR data (invalid UTF-8 in text string)");
1098
859 SvUTF8_on (sv); 1099 SvUTF8_on (sv);
1100 }
860 1101
861 return sv; 1102 return sv;
862 1103
863fail: 1104fail:
864 SvREFCNT_dec (sv); 1105 SvREFCNT_dec (sv);
… …
883 sv = newRV_noinc (decode_sv (dec)); 1124 sv = newRV_noinc (decode_sv (dec));
884 break; 1125 break;
885 1126
886 case CBOR_TAG_STRINGREF_NAMESPACE: 1127 case CBOR_TAG_STRINGREF_NAMESPACE:
887 { 1128 {
1129 // do not use SAVETMPS/FREETMPS, as these will
1130 // erase mortalised caches, e.g. "shareable"
888 ENTER; SAVETMPS; 1131 ENTER;
889 1132
890 SAVESPTR (dec->stringref); 1133 SAVESPTR (dec->stringref);
891 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ()); 1134 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ());
892 1135
893 sv = decode_sv (dec); 1136 sv = decode_sv (dec);
894 1137
895 FREETMPS; LEAVE; 1138 LEAVE;
896 } 1139 }
897 break; 1140 break;
898 1141
899 case CBOR_TAG_STRINGREF: 1142 case CBOR_TAG_STRINGREF:
900 { 1143 {
901 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT)) 1144 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
902 ERR ("corrupted CBOR data (stringref index not an unsigned integer)"); 1145 ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
903 1146
904 UV idx = decode_uint (dec); 1147 UV idx = decode_uint (dec);
905 1148
906 if (!dec->stringref || (int)idx > AvFILLp (dec->stringref)) 1149 if (!dec->stringref || idx >= (UV)(1 + AvFILLp (dec->stringref)))
907 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)");
908 1151
909 sv = newSVsv (AvARRAY (dec->stringref)[idx]); 1152 sv = newSVsv (AvARRAY (dec->stringref)[idx]);
910 } 1153 }
911 break; 1154 break;
… …
939 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT)) 1182 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
940 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)"); 1183 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
941 1184
942 UV idx = decode_uint (dec); 1185 UV idx = decode_uint (dec);
943 1186
944 if (!dec->shareable || (int)idx > AvFILLp (dec->shareable)) 1187 if (!dec->shareable || idx >= (UV)(1 + AvFILLp (dec->shareable)))
945 ERR ("corrupted CBOR data (sharedref index out of bounds)"); 1188 ERR ("corrupted CBOR data (sharedref index out of bounds)");
946 1189
947 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]); 1190 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]);
948 1191
949 if (sv == &PL_sv_undef) 1192 if (sv == &PL_sv_undef)
… …
951 } 1194 }
952 break; 1195 break;
953 1196
954 case CBOR_TAG_PERL_OBJECT: 1197 case CBOR_TAG_PERL_OBJECT:
955 { 1198 {
1199 if (dec->cbor.flags & F_FORBID_OBJECTS)
1200 goto filter;
1201
956 sv = decode_sv (dec); 1202 sv = decode_sv (dec);
957 1203
958 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV) 1204 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
959 ERR ("corrupted CBOR data (non-array perl object)"); 1205 ERR ("corrupted CBOR data (non-array perl object)");
960 1206
… …
970 if (!method) 1216 if (!method)
971 ERR ("cannot decode perl-object (package does not have a THAW method)"); 1217 ERR ("cannot decode perl-object (package does not have a THAW method)");
972 1218
973 dSP; 1219 dSP;
974 1220
975 ENTER; SAVETMPS; PUSHMARK (SP); 1221 ENTER; SAVETMPS;
1222 PUSHMARK (SP);
976 EXTEND (SP, len + 1); 1223 EXTEND (SP, len + 1);
977 // we re-bless the reference to get overload and other niceties right 1224 // we re-bless the reference to get overload and other niceties right
978 PUSHs (*av_fetch (av, 0, 1)); 1225 PUSHs (*av_fetch (av, 0, 1));
979 PUSHs (sv_cbor); 1226 PUSHs (sv_cbor);
980 1227
… …
988 SPAGAIN; 1235 SPAGAIN;
989 1236
990 if (SvTRUE (ERRSV)) 1237 if (SvTRUE (ERRSV))
991 { 1238 {
992 FREETMPS; LEAVE; 1239 FREETMPS; LEAVE;
993 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV)))); 1240 ERR_ERRSV;
994 } 1241 }
995 1242
996 SvREFCNT_dec (sv); 1243 SvREFCNT_dec_NN (sv);
997 sv = SvREFCNT_inc (POPs); 1244 sv = SvREFCNT_inc (POPs);
998 1245
999 PUTBACK; 1246 PUTBACK;
1000 1247
1001 FREETMPS; LEAVE; 1248 FREETMPS; LEAVE;
1002 } 1249 }
1003 break; 1250 break;
1004 1251
1005 default: 1252 default:
1253 filter:
1006 { 1254 {
1255 SV *tag_sv = newSVuv (tag);
1256
1007 sv = decode_sv (dec); 1257 sv = decode_sv (dec);
1008 1258
1009 dSP; 1259 dSP;
1010 ENTER; SAVETMPS; PUSHMARK (SP); 1260 ENTER; SAVETMPS;
1261 PUSHMARK (SP);
1011 EXTEND (SP, 2); 1262 EXTEND (SP, 2);
1012 PUSHs (newSVuv (tag)); 1263 PUSHs (tag_sv);
1013 PUSHs (sv); 1264 PUSHs (sv);
1014 1265
1015 PUTBACK; 1266 PUTBACK;
1016 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);
1017 SPAGAIN; 1268 SPAGAIN;
1018 1269
1019 if (SvTRUE (ERRSV)) 1270 if (SvTRUE (ERRSV))
1020 { 1271 {
1272 SvREFCNT_dec_NN (tag_sv);
1021 FREETMPS; LEAVE; 1273 FREETMPS; LEAVE;
1022 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV)))); 1274 ERR_ERRSV;
1023 } 1275 }
1024 1276
1025 if (count) 1277 if (count)
1026 { 1278 {
1279 SvREFCNT_dec_NN (tag_sv);
1027 SvREFCNT_dec (sv); 1280 SvREFCNT_dec_NN (sv);
1028 sv = SvREFCNT_inc (POPs); 1281 sv = SvREFCNT_inc_NN (TOPs);
1282 SP -= count;
1029 } 1283 }
1030 else 1284 else
1031 { 1285 {
1032 AV *av = newAV (); 1286 AV *av = newAV ();
1033 av_push (av, newSVuv (tag)); 1287 av_push (av, tag_sv);
1034 av_push (av, sv); 1288 av_push (av, sv);
1035 1289
1036 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 1290 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
1037 ? cbor_tagged_stash 1291 ? cbor_tagged_stash
1038 : gv_stashpv ("CBOR::XS::Tagged" , 1); 1292 : gv_stashpv ("CBOR::XS::Tagged" , 1);
… …
1124 1378
1125 return newSVnv (ecb_binary64_to_double (fp)); 1379 return newSVnv (ecb_binary64_to_double (fp));
1126 } 1380 }
1127 1381
1128 // 0..19 unassigned simple 1382 // 0..19 unassigned simple
1129 // 24 reserved + unassigned (reserved values are not encodable) 1383 // 24 reserved + unassigned simple (reserved values are not encodable)
1384 // 28-30 unassigned misc
1385 // 31 break code
1130 default: 1386 default:
1131 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)"); 1387 ERR ("corrupted CBOR data (reserved/unassigned/unexpected major 7 value)");
1132 } 1388 }
1133 1389
1134 break; 1390 break;
1135 } 1391 }
1136 1392
… …
1139} 1395}
1140 1396
1141static SV * 1397static SV *
1142decode_cbor (SV *string, CBOR *cbor, char **offset_return) 1398decode_cbor (SV *string, CBOR *cbor, char **offset_return)
1143{ 1399{
1144 dec_t dec = { }; 1400 dec_t dec = { 0 };
1145 SV *sv; 1401 SV *sv;
1146 STRLEN len; 1402 STRLEN len;
1147 char *data = SvPVbyte (string, len); 1403 char *data = SvPVbyte (string, len);
1148 1404
1149 if (len > cbor->max_size && cbor->max_size) 1405 if (len > cbor->max_size && cbor->max_size)
… …
1163 if (dec.cur != dec.end && !dec.err) 1419 if (dec.cur != dec.end && !dec.err)
1164 dec.err = "garbage after CBOR object"; 1420 dec.err = "garbage after CBOR object";
1165 1421
1166 if (dec.err) 1422 if (dec.err)
1167 { 1423 {
1424 if (dec.shareable)
1425 {
1426 // need to break cyclic links, which would all be in shareable
1427 int i;
1428 SV **svp;
1429
1430 for (i = av_len (dec.shareable) + 1; i--; )
1431 if ((svp = av_fetch (dec.shareable, i, 0)))
1432 sv_setsv (*svp, &PL_sv_undef);
1433 }
1434
1168 SvREFCNT_dec (sv); 1435 SvREFCNT_dec_NN (sv);
1436
1437 if (dec.err_sv)
1438 sv_2mortal (dec.err_sv);
1439
1169 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);
1170 } 1441 }
1171 1442
1172 sv = sv_2mortal (sv); 1443 sv = sv_2mortal (sv);
1173 1444
1174 return sv; 1445 return sv;
1175} 1446}
1176 1447
1448/////////////////////////////////////////////////////////////////////////////
1449// incremental parser
1450
1451#define INCR_DONE(cbor) (AvFILLp (cbor->incr_count) < 0)
1452
1453// returns 0 for notyet, 1 for success or error
1454static int
1455incr_parse (CBOR *self, SV *cborstr)
1456{
1457 STRLEN cur;
1458 SvPV (cborstr, cur);
1459
1460 while (ecb_expect_true (self->incr_need <= cur))
1461 {
1462 // table of integer count bytes
1463 static I8 incr_len[MINOR_MASK + 1] = {
1464 0, 0, 0, 0, 0, 0, 0, 0,
1465 0, 0, 0, 0, 0, 0, 0, 0,
1466 0, 0, 0, 0, 0, 0, 0, 0,
1467 1, 2, 4, 8,-1,-1,-1,-2
1468 };
1469
1470 const U8 *p = SvPVX (cborstr) + self->incr_pos;
1471 U8 m = *p & MINOR_MASK;
1472 IV count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1473 I8 ilen = incr_len[m];
1474
1475 self->incr_need = self->incr_pos + 1;
1476
1477 if (ecb_expect_false (ilen < 0))
1478 {
1479 if (m != MINOR_INDEF)
1480 return 1; // error
1481
1482 if (*p == (MAJOR_MISC | MINOR_INDEF))
1483 {
1484 if (count >= 0)
1485 return 1; // error
1486
1487 count = 1;
1488 }
1489 else
1490 {
1491 av_push (self->incr_count, newSViv (-1)); //TODO: nest
1492 count = -1;
1493 }
1494 }
1495 else
1496 {
1497 self->incr_need += ilen;
1498 if (ecb_expect_false (self->incr_need > cur))
1499 return 0;
1500
1501 int major = *p >> MAJOR_SHIFT;
1502
1503 switch (major)
1504 {
1505 case MAJOR_TAG >> MAJOR_SHIFT:
1506 ++count; // tags merely prefix another value
1507 break;
1508
1509 case MAJOR_BYTES >> MAJOR_SHIFT:
1510 case MAJOR_TEXT >> MAJOR_SHIFT:
1511 case MAJOR_ARRAY >> MAJOR_SHIFT:
1512 case MAJOR_MAP >> MAJOR_SHIFT:
1513 {
1514 UV len;
1515
1516 if (ecb_expect_false (ilen))
1517 {
1518 len = 0;
1519
1520 do {
1521 len = (len << 8) | *++p;
1522 } while (--ilen);
1523 }
1524 else
1525 len = m;
1526
1527 switch (major)
1528 {
1529 case MAJOR_BYTES >> MAJOR_SHIFT:
1530 case MAJOR_TEXT >> MAJOR_SHIFT:
1531 self->incr_need += len;
1532 if (ecb_expect_false (self->incr_need > cur))
1533 return 0;
1534
1535 break;
1536
1537 case MAJOR_MAP >> MAJOR_SHIFT:
1538 len <<= 1;
1539 case MAJOR_ARRAY >> MAJOR_SHIFT:
1540 if (len)
1541 {
1542 av_push (self->incr_count, newSViv (len + 1)); //TODO: nest
1543 count = len + 1;
1544 }
1545 break;
1546 }
1547 }
1548 }
1549 }
1550
1551 self->incr_pos = self->incr_need;
1552
1553 if (count > 0)
1554 {
1555 while (!--count)
1556 {
1557 if (!AvFILLp (self->incr_count))
1558 return 1; // done
1559
1560 SvREFCNT_dec_NN (av_pop (self->incr_count));
1561 count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1562 }
1563
1564 SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]) = count;
1565 }
1566 }
1567
1568 return 0;
1569}
1570
1571
1177///////////////////////////////////////////////////////////////////////////// 1572/////////////////////////////////////////////////////////////////////////////
1178// XS interface functions 1573// XS interface functions
1179 1574
1180MODULE = CBOR::XS PACKAGE = CBOR::XS 1575MODULE = CBOR::XS PACKAGE = CBOR::XS
1181 1576
… …
1193 1588
1194 default_filter = newSVpv ("CBOR::XS::default_filter", 0); 1589 default_filter = newSVpv ("CBOR::XS::default_filter", 0);
1195 1590
1196 sv_cbor = newSVpv ("CBOR", 0); 1591 sv_cbor = newSVpv ("CBOR", 0);
1197 SvREADONLY_on (sv_cbor); 1592 SvREADONLY_on (sv_cbor);
1593
1594 assert (("STRLEN must be an unsigned type", 0 <= (STRLEN)-1));
1198} 1595}
1199 1596
1200PROTOTYPES: DISABLE 1597PROTOTYPES: DISABLE
1201 1598
1202void CLONE (...) 1599void CLONE (...)
… …
1222 ALIAS: 1619 ALIAS:
1223 shrink = F_SHRINK 1620 shrink = F_SHRINK
1224 allow_unknown = F_ALLOW_UNKNOWN 1621 allow_unknown = F_ALLOW_UNKNOWN
1225 allow_sharing = F_ALLOW_SHARING 1622 allow_sharing = F_ALLOW_SHARING
1226 allow_cycles = F_ALLOW_CYCLES 1623 allow_cycles = F_ALLOW_CYCLES
1624 forbid_objects = F_FORBID_OBJECTS
1227 pack_strings = F_PACK_STRINGS 1625 pack_strings = F_PACK_STRINGS
1626 text_keys = F_TEXT_KEYS
1627 text_strings = F_TEXT_STRINGS
1628 validate_utf8 = F_VALIDATE_UTF8
1228 PPCODE: 1629 PPCODE:
1229{ 1630{
1230 if (enable) 1631 if (enable)
1231 self->flags |= ix; 1632 self->flags |= ix;
1232 else 1633 else
… …
1239 ALIAS: 1640 ALIAS:
1240 get_shrink = F_SHRINK 1641 get_shrink = F_SHRINK
1241 get_allow_unknown = F_ALLOW_UNKNOWN 1642 get_allow_unknown = F_ALLOW_UNKNOWN
1242 get_allow_sharing = F_ALLOW_SHARING 1643 get_allow_sharing = F_ALLOW_SHARING
1243 get_allow_cycles = F_ALLOW_CYCLES 1644 get_allow_cycles = F_ALLOW_CYCLES
1645 get_forbid_objects = F_FORBID_OBJECTS
1244 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
1649 get_validate_utf8 = F_VALIDATE_UTF8
1245 PPCODE: 1650 PPCODE:
1246 XPUSHs (boolSV (self->flags & ix)); 1651 XPUSHs (boolSV (self->flags & ix));
1247 1652
1248void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) 1653void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
1249 PPCODE: 1654 PPCODE:
… …
1298 EXTEND (SP, 2); 1703 EXTEND (SP, 2);
1299 PUSHs (sv); 1704 PUSHs (sv);
1300 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); 1705 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
1301} 1706}
1302 1707
1708void incr_parse (CBOR *self, SV *cborstr)
1709 ALIAS:
1710 incr_parse_multiple = 1
1711 PPCODE:
1712{
1713 if (SvUTF8 (cborstr))
1714 sv_utf8_downgrade (cborstr, 0);
1715
1716 if (!self->incr_count)
1717 {
1718 self->incr_count = newAV ();
1719 self->incr_pos = 0;
1720 self->incr_need = 1;
1721
1722 av_push (self->incr_count, newSViv (1));
1723 }
1724
1725 do
1726 {
1727 if (!incr_parse (self, cborstr))
1728 {
1729 if (self->incr_need > self->max_size && self->max_size)
1730 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
1731 (unsigned long)self->incr_need, (unsigned long)self->max_size);
1732
1733 break;
1734 }
1735
1736 SV *sv;
1737 char *offset;
1738
1739 PUTBACK; sv = decode_cbor (cborstr, self, &offset); SPAGAIN;
1740 XPUSHs (sv);
1741
1742 sv_chop (cborstr, offset);
1743
1744 av_clear (self->incr_count);
1745 av_push (self->incr_count, newSViv (1));
1746
1747 self->incr_pos = 0;
1748 self->incr_need = self->incr_pos + 1;
1749 }
1750 while (ix);
1751}
1752
1753void incr_reset (CBOR *self)
1754 CODE:
1755{
1756 SvREFCNT_dec (self->incr_count);
1757 self->incr_count = 0;
1758}
1759
1303void DESTROY (CBOR *self) 1760void DESTROY (CBOR *self)
1304 PPCODE: 1761 PPCODE:
1305 cbor_free (self); 1762 cbor_free (self);
1306 1763
1307PROTOTYPES: ENABLE 1764PROTOTYPES: ENABLE
… …
1326 cbor_init (&cbor); 1783 cbor_init (&cbor);
1327 PUTBACK; cborstr = decode_cbor (cborstr, &cbor, 0); SPAGAIN; 1784 PUTBACK; cborstr = decode_cbor (cborstr, &cbor, 0); SPAGAIN;
1328 XPUSHs (cborstr); 1785 XPUSHs (cborstr);
1329} 1786}
1330 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