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 1.36 by root, Sat Nov 30 17:37:45 2013 UTC vs.
Revision 1.66 by root, Sun Nov 29 21:32:01 2020 UTC

… …
6#include <string.h> 6#include <string.h>
7#include <stdlib.h> 7#include <stdlib.h>
8#include <stdio.h> 8#include <stdio.h>
9#include <limits.h> 9#include <limits.h>
10#include <float.h> 10#include <float.h>
11#include <inttypes.h>
11 12
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_FLOAT16 = 3,
111 AS_FLOAT32 = 4,
112 AS_FLOAT64 = 5,
113 // possibly future enhancements: float, integer
114};
115
99#define F_SHRINK 0x00000001UL 116#define F_SHRINK 0x00000001UL
100#define F_ALLOW_UNKNOWN 0x00000002UL 117#define F_ALLOW_UNKNOWN 0x00000002UL
101#define F_ALLOW_SHARING 0x00000004UL 118#define F_ALLOW_SHARING 0x00000004UL
119#define F_ALLOW_CYCLES 0x00000008UL
120#define F_FORBID_OBJECTS 0x00000010UL
102#define F_PACK_STRINGS 0x00000008UL 121#define F_PACK_STRINGS 0x00000020UL
122#define F_TEXT_KEYS 0x00000040UL
123#define F_TEXT_STRINGS 0x00000080UL
124#define F_VALIDATE_UTF8 0x00000100UL
103 125
104#define INIT_SIZE 32 // initial scalar size to be allocated 126#define INIT_SIZE 32 // initial scalar size to be allocated
105 127
106#define SB do { 128#define SB do {
107#define SE } while (0) 129#define SE } while (0)
… …
126typedef struct { 148typedef struct {
127 U32 flags; 149 U32 flags;
128 U32 max_depth; 150 U32 max_depth;
129 STRLEN max_size; 151 STRLEN max_size;
130 SV *filter; 152 SV *filter;
153
154 // for the incremental parser
155 STRLEN incr_pos; // the current offset into the text
156 STRLEN incr_need; // minimum bytes needed to decode
157 AV *incr_count; // for every nesting level, the number of outstanding values, or -1 for indef.
131} CBOR; 158} CBOR;
132 159
133ecb_inline void 160ecb_inline void
134cbor_init (CBOR *cbor) 161cbor_init (CBOR *cbor)
135{ 162{
… …
139 166
140ecb_inline void 167ecb_inline void
141cbor_free (CBOR *cbor) 168cbor_free (CBOR *cbor)
142{ 169{
143 SvREFCNT_dec (cbor->filter); 170 SvREFCNT_dec (cbor->filter);
171 SvREFCNT_dec (cbor->incr_count);
144} 172}
145 173
146///////////////////////////////////////////////////////////////////////////// 174/////////////////////////////////////////////////////////////////////////////
147// utility functions 175// utility functions
148 176
… …
174 202
175// minimum length of a string to be registered for stringref 203// minimum length of a string to be registered for stringref
176ecb_inline int 204ecb_inline int
177minimum_string_length (UV idx) 205minimum_string_length (UV idx)
178{ 206{
179 return idx > 23 207 return idx <= 23 ? 3
180 ? idx > 0xffU 208 : idx <= 0xffU ? 4
181 ? idx > 0xffffU 209 : idx <= 0xffffU ? 5
182 ? idx > 0xffffffffU 210 : idx <= 0xffffffffU ? 7
183 ? 11 211 : 11;
184 : 7
185 : 5
186 : 4
187 : 3;
188} 212}
189 213
190///////////////////////////////////////////////////////////////////////////// 214/////////////////////////////////////////////////////////////////////////////
191// encoder 215// encoder
192 216
… …
205} enc_t; 229} enc_t;
206 230
207ecb_inline void 231ecb_inline void
208need (enc_t *enc, STRLEN len) 232need (enc_t *enc, STRLEN len)
209{ 233{
210 if (ecb_expect_false (enc->cur + len >= enc->end)) 234 if (ecb_expect_false ((uintptr_t)(enc->end - enc->cur) < len))
211 { 235 {
212 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv); 236 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
213 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1); 237 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
214 enc->cur = SvPVX (enc->sv) + cur; 238 enc->cur = SvPVX (enc->sv) + cur;
215 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; 239 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
… …
221{ 245{
222 need (enc, 1); 246 need (enc, 1);
223 *enc->cur++ = ch; 247 *enc->cur++ = ch;
224} 248}
225 249
250// used for tags, intregers, element counts and so on
226static void 251static void
227encode_uint (enc_t *enc, int major, UV len) 252encode_uint (enc_t *enc, int major, UV len)
228{ 253{
229 need (enc, 9); 254 need (enc, 9);
230 255
… …
261 *enc->cur++ = len >> 8; 286 *enc->cur++ = len >> 8;
262 *enc->cur++ = len; 287 *enc->cur++ = len;
263 } 288 }
264} 289}
265 290
291// encodes a perl value into a CBOR integer
292ecb_inline
293encode_int (enc_t *enc, SV *sv)
294{
295 if (SvIsUV (sv))
296 encode_uint (enc, MAJOR_POS_INT, SvUVX (sv));
297 else if (SvIVX (sv) >= 0)
298 encode_uint (enc, MAJOR_POS_INT, SvIVX (sv));
299 else
300 encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1));
301}
302
266ecb_inline void 303ecb_inline void
267encode_tag (enc_t *enc, UV tag) 304encode_tag (enc_t *enc, UV tag)
268{ 305{
269 encode_uint (enc, MAJOR_TAG, tag); 306 encode_uint (enc, MAJOR_TAG, tag);
270} 307}
271 308
309// exceptional (hopefully) slow path for byte strings that need to be utf8-encoded
310ecb_noinline static void
311encode_str_utf8 (enc_t *enc, int utf8, char *str, STRLEN len)
312{
313 STRLEN ulen = len;
314 U8 *p, *pend = (U8 *)str + len;
315
316 for (p = (U8 *)str; p < pend; ++p)
317 ulen += *p >> 7; // count set high bits
318
319 encode_uint (enc, MAJOR_TEXT, ulen);
320
321 need (enc, ulen);
322 for (p = (U8 *)str; p < pend; ++p)
323 if (*p < 0x80)
324 *enc->cur++ = *p;
325 else
326 {
327 *enc->cur++ = 0xc0 + (*p >> 6);
328 *enc->cur++ = 0x80 + (*p & 63);
329 }
330}
331
272ecb_inline void 332ecb_inline void
273encode_str (enc_t *enc, int utf8, char *str, STRLEN len) 333encode_str (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len)
274{ 334{
335 if (ecb_expect_false (upgrade_utf8))
336 if (!utf8)
337 {
338 encode_str_utf8 (enc, utf8, str, len);
339 return;
340 }
341
275 encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len); 342 encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len);
276 need (enc, len); 343 need (enc, len);
277 memcpy (enc->cur, str, len); 344 memcpy (enc->cur, str, len);
278 enc->cur += len; 345 enc->cur += len;
279} 346}
280 347
281static void 348ecb_inline void
282encode_strref (enc_t *enc, int utf8, char *str, STRLEN len) 349encode_strref (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len)
283{ 350{
284 if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS)) 351 if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS))
285 { 352 {
286 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1); 353 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1);
287 354
… …
298 sv_setuv (*svp, enc->stringref_idx); 365 sv_setuv (*svp, enc->stringref_idx);
299 ++enc->stringref_idx; 366 ++enc->stringref_idx;
300 } 367 }
301 } 368 }
302 369
303 encode_str (enc, utf8, str, len); 370 encode_str (enc, upgrade_utf8, utf8, str, len);
371}
372
373ecb_inline void
374encode_float16 (enc_t *enc, NV nv)
375{
376 need (enc, 1+2);
377
378 *enc->cur++ = MAJOR_MISC | MISC_FLOAT16;
379
380 uint16_t fp = ecb_float_to_binary16 (nv);
381
382 if (!ecb_big_endian ())
383 fp = ecb_bswap16 (fp);
384
385 memcpy (enc->cur, &fp, 2);
386 enc->cur += 2;
387}
388
389ecb_inline void
390encode_float32 (enc_t *enc, NV nv)
391{
392 need (enc, 1+4);
393
394 *enc->cur++ = MAJOR_MISC | MISC_FLOAT32;
395
396 uint32_t fp = ecb_float_to_binary32 (nv);
397
398 if (!ecb_big_endian ())
399 fp = ecb_bswap32 (fp);
400
401 memcpy (enc->cur, &fp, 4);
402 enc->cur += 4;
403}
404
405ecb_inline void
406encode_float64 (enc_t *enc, NV nv)
407{
408 need (enc, 1+8);
409
410 *enc->cur++ = MAJOR_MISC | MISC_FLOAT64;
411
412 uint64_t fp = ecb_double_to_binary64 (nv);
413
414 if (!ecb_big_endian ())
415 fp = ecb_bswap64 (fp);
416
417 memcpy (enc->cur, &fp, 8);
418 enc->cur += 8;
419}
420
421ecb_inline void
422encode_forced (enc_t *enc, UV type, SV *sv)
423{
424 switch (type)
425 {
426 case AS_CBOR:
427 {
428 STRLEN len;
429 char *str = SvPVbyte (sv, len);
430
431 need (enc, len);
432 memcpy (enc->cur, str, len);
433 enc->cur += len;
434 }
435 break;
436
437 case AS_BYTES:
438 {
439 STRLEN len;
440 char *str = SvPVbyte (sv, len);
441 encode_strref (enc, 0, 0, str, len);
442 }
443 break;
444
445 case AS_TEXT:
446 {
447 STRLEN len;
448 char *str = SvPVutf8 (sv, len);
449 encode_strref (enc, 1, 1, str, len);
450 }
451 break;
452
453 case AS_FLOAT16: encode_float16 (enc, SvNV (sv)); break;
454 case AS_FLOAT32: encode_float32 (enc, SvNV (sv)); break;
455 case AS_FLOAT64: encode_float64 (enc, SvNV (sv)); break;
456
457 default:
458 croak ("encountered malformed CBOR::XS::Tagged object");
459 }
304} 460}
305 461
306static void encode_sv (enc_t *enc, SV *sv); 462static void encode_sv (enc_t *enc, SV *sv);
307 463
308static void 464static void
… …
315 471
316 ++enc->depth; 472 ++enc->depth;
317 473
318 encode_uint (enc, MAJOR_ARRAY, len + 1); 474 encode_uint (enc, MAJOR_ARRAY, len + 1);
319 475
476 if (ecb_expect_false (SvMAGICAL (av)))
320 for (i = 0; i <= len; ++i) 477 for (i = 0; i <= len; ++i)
321 { 478 {
322 SV **svp = av_fetch (av, i, 0); 479 SV **svp = av_fetch (av, i, 0);
323 encode_sv (enc, svp ? *svp : &PL_sv_undef); 480 encode_sv (enc, svp ? *svp : &PL_sv_undef);
324 } 481 }
482 else
483 for (i = 0; i <= len; ++i)
484 {
485 SV *sv = AvARRAY (av)[i];
486 encode_sv (enc, sv ? sv : &PL_sv_undef);
487 }
325 488
326 --enc->depth; 489 --enc->depth;
327} 490}
328 491
329static void 492static void
… …
337 ++enc->depth; 500 ++enc->depth;
338 501
339 int pairs = hv_iterinit (hv); 502 int pairs = hv_iterinit (hv);
340 int mg = SvMAGICAL (hv); 503 int mg = SvMAGICAL (hv);
341 504
342 if (mg) 505 if (ecb_expect_false (mg))
343 encode_ch (enc, MAJOR_MAP | MINOR_INDEF); 506 encode_ch (enc, MAJOR_MAP | MINOR_INDEF);
344 else 507 else
345 encode_uint (enc, MAJOR_MAP, pairs); 508 encode_uint (enc, MAJOR_MAP, pairs);
346 509
347 while ((he = hv_iternext (hv))) 510 while ((he = hv_iternext (hv)))
348 { 511 {
349 if (HeKLEN (he) == HEf_SVKEY) 512 if (HeKLEN (he) == HEf_SVKEY)
350 encode_sv (enc, HeSVKEY (he)); 513 encode_sv (enc, HeSVKEY (he));
351 else 514 else
352 encode_strref (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he)); 515 encode_strref (enc, enc->cbor.flags & (F_TEXT_KEYS | F_TEXT_STRINGS), HeKUTF8 (he), HeKEY (he), HeKLEN (he));
353 516
354 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he)); 517 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he));
355 } 518 }
356 519
357 if (mg) 520 if (ecb_expect_false (mg))
358 encode_ch (enc, MAJOR_MISC | MINOR_INDEF); 521 encode_ch (enc, MAJOR_MISC | MINOR_INDEF);
359 522
360 --enc->depth; 523 --enc->depth;
361} 524}
362 525
… …
395 else if (stash == tagged_stash) 558 else if (stash == tagged_stash)
396 { 559 {
397 if (svt != SVt_PVAV) 560 if (svt != SVt_PVAV)
398 croak ("encountered CBOR::XS::Tagged object that isn't an array"); 561 croak ("encountered CBOR::XS::Tagged object that isn't an array");
399 562
563 switch (av_len ((AV *)sv))
564 {
565 case 2-1:
566 // actually a tagged value
400 encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1))); 567 encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1)));
401 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1)); 568 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1));
569 break;
570
571 case 3-1:
572 // a forced type [value, type, undef]
573 encode_forced (enc, SvUV (*av_fetch ((AV *)sv, 1, 1)), *av_fetch ((AV *)sv, 0, 1));
574 break;
575
576 default:
577 croak ("encountered malformed CBOR::XS::Tagged object");
578 }
402 579
403 return; 580 return;
404 } 581 }
405 } 582 }
406 583
407 if (ecb_expect_false (SvREFCNT (sv) > 1) 584 if (ecb_expect_false (SvREFCNT (sv) > 1)
408 && ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING)) 585 && ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING))
409 { 586 {
410 if (!enc->shareable) 587 if (ecb_expect_false (!enc->shareable))
411 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ()); 588 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
412 589
413 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1); 590 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
414 591
415 if (SvOK (*svp)) 592 if (SvOK (*svp))
… …
429 if (ecb_expect_false (SvOBJECT (sv))) 606 if (ecb_expect_false (SvOBJECT (sv)))
430 { 607 {
431 HV *stash = SvSTASH (sv); 608 HV *stash = SvSTASH (sv);
432 GV *method; 609 GV *method;
433 610
611 if (enc->cbor.flags & F_FORBID_OBJECTS)
612 croak ("encountered object '%s', but forbid_objects is enabled",
613 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
434 if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0))) 614 else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
435 { 615 {
436 dSP; 616 dSP;
437 617
438 ENTER; SAVETMPS; PUSHMARK (SP); 618 ENTER; SAVETMPS;
619 PUSHMARK (SP);
439 // we re-bless the reference to get overload and other niceties right 620 // we re-bless the reference to get overload and other niceties right
440 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); 621 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
441 622
442 PUTBACK; 623 PUTBACK;
443 // G_SCALAR ensures that return value is 1 624 // G_SCALAR ensures that return value is 1
… …
456 } 637 }
457 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0) 638 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)
458 { 639 {
459 dSP; 640 dSP;
460 641
461 ENTER; SAVETMPS; PUSHMARK (SP); 642 ENTER; SAVETMPS;
643 PUSHMARK (SP);
462 EXTEND (SP, 2); 644 EXTEND (SP, 2);
463 // we re-bless the reference to get overload and other niceties right 645 // we re-bless the reference to get overload and other niceties right
464 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); 646 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
465 PUSHs (sv_cbor); 647 PUSHs (sv_cbor);
466 648
… …
472 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv) 654 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)
473 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash)); 655 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));
474 656
475 encode_tag (enc, CBOR_TAG_PERL_OBJECT); 657 encode_tag (enc, CBOR_TAG_PERL_OBJECT);
476 encode_uint (enc, MAJOR_ARRAY, count + 1); 658 encode_uint (enc, MAJOR_ARRAY, count + 1);
477 encode_strref (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash)); 659 encode_strref (enc, 0, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
478 660
479 while (count) 661 {
662 int i;
663
664 for (i = 0; i < count; ++i)
480 encode_sv (enc, SP[1 - count--]); 665 encode_sv (enc, SP[i + 1 - count]);
666
667 SP -= count;
668 }
481 669
482 PUTBACK; 670 PUTBACK;
483 671
484 FREETMPS; LEAVE; 672 FREETMPS; LEAVE;
485 } 673 }
… …
507 695
508 if (ecb_expect_false (nv == (NV)(U32)nv)) 696 if (ecb_expect_false (nv == (NV)(U32)nv))
509 encode_uint (enc, MAJOR_POS_INT, (U32)nv); 697 encode_uint (enc, MAJOR_POS_INT, (U32)nv);
510 //TODO: maybe I32? 698 //TODO: maybe I32?
511 else if (ecb_expect_false (nv == (float)nv)) 699 else if (ecb_expect_false (nv == (float)nv))
512 { 700 encode_float32 (enc, nv);
513 uint32_t fp = ecb_float_to_binary32 (nv);
514
515 *enc->cur++ = MAJOR_MISC | MISC_FLOAT32;
516
517 if (!ecb_big_endian ())
518 fp = ecb_bswap32 (fp);
519
520 memcpy (enc->cur, &fp, 4);
521 enc->cur += 4;
522 }
523 else 701 else
524 { 702 encode_float64 (enc, nv);
525 uint64_t fp = ecb_double_to_binary64 (nv);
526
527 *enc->cur++ = MAJOR_MISC | MISC_FLOAT64;
528
529 if (!ecb_big_endian ())
530 fp = ecb_bswap64 (fp);
531
532 memcpy (enc->cur, &fp, 8);
533 enc->cur += 8;
534 }
535} 703}
536 704
537static void 705static void
538encode_sv (enc_t *enc, SV *sv) 706encode_sv (enc_t *enc, SV *sv)
539{ 707{
… …
541 709
542 if (SvPOKp (sv)) 710 if (SvPOKp (sv))
543 { 711 {
544 STRLEN len; 712 STRLEN len;
545 char *str = SvPV (sv, len); 713 char *str = SvPV (sv, len);
546 encode_strref (enc, SvUTF8 (sv), str, len); 714 encode_strref (enc, enc->cbor.flags & F_TEXT_STRINGS, SvUTF8 (sv), str, len);
547 } 715 }
548 else if (SvNOKp (sv)) 716 else if (SvNOKp (sv))
549 encode_nv (enc, sv); 717 encode_nv (enc, sv);
550 else if (SvIOKp (sv)) 718 else if (SvIOKp (sv))
551 { 719 encode_int (e,v sv);
552 if (SvIsUV (sv))
553 encode_uint (enc, MAJOR_POS_INT, SvUVX (sv));
554 else if (SvIVX (sv) >= 0)
555 encode_uint (enc, MAJOR_POS_INT, SvIVX (sv));
556 else
557 encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1));
558 }
559 else if (SvROK (sv)) 720 else if (SvROK (sv))
560 encode_rv (enc, SvRV (sv)); 721 encode_rv (enc, SvRV (sv));
561 else if (!SvOK (sv)) 722 else if (!SvOK (sv))
562 encode_ch (enc, MAJOR_MISC | SIMPLE_NULL); 723 encode_ch (enc, MAJOR_MISC | SIMPLE_NULL);
563 else if (enc->cbor.flags & F_ALLOW_UNKNOWN) 724 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
… …
568} 729}
569 730
570static SV * 731static SV *
571encode_cbor (SV *scalar, CBOR *cbor) 732encode_cbor (SV *scalar, CBOR *cbor)
572{ 733{
573 enc_t enc = { }; 734 enc_t enc = { 0 };
574 735
575 enc.cbor = *cbor; 736 enc.cbor = *cbor;
576 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 737 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
577 enc.cur = SvPVX (enc.sv); 738 enc.cur = SvPVX (enc.sv);
578 enc.end = SvEND (enc.sv); 739 enc.end = SvEND (enc.sv);
579 740
580 SvPOK_only (enc.sv); 741 SvPOK_only (enc.sv);
581 742
582 if (cbor->flags & F_PACK_STRINGS) 743 if (cbor->flags & F_PACK_STRINGS)
583 { 744 {
… …
610 U32 depth; // recursion depth 771 U32 depth; // recursion depth
611 U32 maxdepth; // recursion depth limit 772 U32 maxdepth; // recursion depth limit
612 AV *shareable; 773 AV *shareable;
613 AV *stringref; 774 AV *stringref;
614 SV *decode_tagged; 775 SV *decode_tagged;
776 SV *err_sv; // optional sv for error, needs to be freed
615} dec_t; 777} dec_t;
616 778
617#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE 779// set dec->err to ERRSV
780ecb_cold static void
781err_errsv (dec_t *dec)
782{
783 if (!dec->err)
784 {
785 dec->err_sv = newSVsv (ERRSV);
618 786
619#define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data") 787 // chop off the trailing \n
788 SvCUR_set (dec->err_sv, SvCUR (dec->err_sv) - 1);
789 *SvEND (dec->err_sv) = 0;
620 790
791 dec->err = SvPVutf8_nolen (dec->err_sv);
792 }
793}
794
795// the following functions are used to reduce code size and help the compiler to optimise
796ecb_cold static void
797err_set (dec_t *dec, const char *reason)
798{
799 if (!dec->err)
800 dec->err = reason;
801}
802
803ecb_cold static void
804err_unexpected_end (dec_t *dec)
805{
806 err_set (dec, "unexpected end of CBOR data");
807}
808
809#define ERR_DO(do) SB do; goto fail; SE
810#define ERR(reason) ERR_DO (err_set (dec, reason))
811#define ERR_ERRSV ERR_DO (err_errsv (dec))
812
813#define WANT(len) if (ecb_expect_false ((uintptr_t)(dec->end - dec->cur) < (STRLEN)len)) ERR_DO (err_unexpected_end (dec))
814
621#define DEC_INC_DEPTH if (++dec->depth > dec->cbor.max_depth) ERR (ERR_NESTING_EXCEEDED) 815#define DEC_INC_DEPTH if (ecb_expect_false (++dec->depth > dec->cbor.max_depth)) ERR (ERR_NESTING_EXCEEDED)
622#define DEC_DEC_DEPTH --dec->depth 816#define DEC_DEC_DEPTH --dec->depth
623 817
624static UV 818static UV
625decode_uint (dec_t *dec) 819decode_uint (dec_t *dec)
626{ 820{
… …
692 886
693 for (;;) 887 for (;;)
694 { 888 {
695 WANT (1); 889 WANT (1);
696 890
697 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF)) 891 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF) || dec->err)
698 { 892 {
699 ++dec->cur; 893 ++dec->cur;
700 break; 894 break;
701 } 895 }
702 896
703 av_push (av, decode_sv (dec)); 897 av_push (av, decode_sv (dec));
704 } 898 }
705 } 899 }
706 else 900 else
707 { 901 {
708 int i, len = decode_uint (dec); 902 UV i, len = decode_uint (dec);
709 903
710 WANT (len); // complexity check for av_fill - need at least one byte per value, do not allow supersize arrays 904 WANT (len); // complexity check for av_fill - need at least one byte per value, do not allow supersize arrays
711 av_fill (av, len - 1); 905 av_fill (av, len - 1);
712 906
713 for (i = 0; i < len; ++i) 907 for (i = 0; i < len; ++i)
… …
716 910
717 DEC_DEC_DEPTH; 911 DEC_DEC_DEPTH;
718 return newRV_noinc ((SV *)av); 912 return newRV_noinc ((SV *)av);
719 913
720fail: 914fail:
721 SvREFCNT_dec (av); 915 SvREFCNT_dec_NN (av);
722 DEC_DEC_DEPTH; 916 DEC_DEC_DEPTH;
723 return &PL_sv_undef; 917 return &PL_sv_undef;
724} 918}
725 919
726static void 920static void
… …
728{ 922{
729 // for speed reasons, we specialcase single-string 923 // for speed reasons, we specialcase single-string
730 // byte or utf-8 strings as keys, but only when !stringref 924 // byte or utf-8 strings as keys, but only when !stringref
731 925
732 if (ecb_expect_true (!dec->stringref)) 926 if (ecb_expect_true (!dec->stringref))
733 if (ecb_expect_true ((*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8)) 927 if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8))
734 { 928 {
735 I32 len = decode_uint (dec); 929 STRLEN len = decode_uint (dec);
736 char *key = (char *)dec->cur; 930 char *key = (char *)dec->cur;
737 931
932 WANT (len);
738 dec->cur += len; 933 dec->cur += len;
739 934
740 if (ecb_expect_false (dec->stringref))
741 av_push (dec->stringref, newSVpvn (key, len));
742
743 hv_store (hv, key, len, decode_sv (dec), 0); 935 hv_store (hv, key, len, decode_sv (dec), 0);
744 936
745 return; 937 return;
746 } 938 }
747 else if (ecb_expect_true ((*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8)) 939 else if (ecb_expect_true ((U8)(*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8))
748 { 940 {
749 I32 len = decode_uint (dec); 941 STRLEN len = decode_uint (dec);
750 char *key = (char *)dec->cur; 942 char *key = (char *)dec->cur;
751 943
944 WANT (len);
752 dec->cur += len; 945 dec->cur += len;
753 946
754 if (ecb_expect_false (dec->stringref)) 947 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
755 av_push (dec->stringref, newSVpvn_utf8 (key, len, 1)); 948 if (!is_utf8_string (key, len))
949 ERR ("corrupted CBOR data (invalid UTF-8 in map key)");
756 950
757 hv_store (hv, key, -len, decode_sv (dec), 0); 951 hv_store (hv, key, -len, decode_sv (dec), 0);
758 952
759 return; 953 return;
760 } 954 }
761 955
762 SV *k = decode_sv (dec); 956 SV *k = decode_sv (dec);
763 SV *v = decode_sv (dec); 957 SV *v = decode_sv (dec);
764 958
959 // we leak memory if uncaught exceptions are thrown by random magical
960 // methods, and this is hopefully the only place where it can happen,
961 // so if there is a chance of an exception, take the very slow path.
962 // since catching exceptions is "undocumented/internal/forbidden" by
963 // the new p5p powers, we need to call out to a perl function :/
964 if (ecb_expect_false (SvAMAGIC (k)))
965 {
966 dSP;
967
968 ENTER; SAVETMPS;
969 PUSHMARK (SP);
970 EXTEND (SP, 3);
971 PUSHs (sv_2mortal (newRV_inc ((SV *)hv)));
972 PUSHs (sv_2mortal (k));
973 PUSHs (sv_2mortal (v));
974
975 PUTBACK;
976 call_pv ("CBOR::XS::_hv_store", G_VOID | G_DISCARD | G_EVAL);
977 SPAGAIN;
978
979 FREETMPS; LEAVE;
980
981 if (SvTRUE (ERRSV))
982 ERR_ERRSV;
983
984 return;
985 }
986
765 hv_store_ent (hv, k, v, 0); 987 hv_store_ent (hv, k, v, 0);
766 SvREFCNT_dec (k); 988 SvREFCNT_dec_NN (k);
989
990fail:
991 ;
767} 992}
768 993
769static SV * 994static SV *
770decode_hv (dec_t *dec) 995decode_hv (dec_t *dec)
771{ 996{
… …
779 1004
780 for (;;) 1005 for (;;)
781 { 1006 {
782 WANT (1); 1007 WANT (1);
783 1008
784 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF)) 1009 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF) || dec->err)
785 { 1010 {
786 ++dec->cur; 1011 ++dec->cur;
787 break; 1012 break;
788 } 1013 }
789 1014
790 decode_he (dec, hv); 1015 decode_he (dec, hv);
791 } 1016 }
792 } 1017 }
793 else 1018 else
794 { 1019 {
795 int pairs = decode_uint (dec); 1020 UV pairs = decode_uint (dec);
1021
1022 WANT (pairs); // complexity check - need at least one byte per value, do not allow supersize hashes
796 1023
797 while (pairs--) 1024 while (pairs--)
798 decode_he (dec, hv); 1025 decode_he (dec, hv);
799 } 1026 }
800 1027
801 DEC_DEC_DEPTH; 1028 DEC_DEC_DEPTH;
802 return newRV_noinc ((SV *)hv); 1029 return newRV_noinc ((SV *)hv);
803 1030
804fail: 1031fail:
805 SvREFCNT_dec (hv); 1032 SvREFCNT_dec_NN (hv);
806 DEC_DEC_DEPTH; 1033 DEC_DEC_DEPTH;
807 return &PL_sv_undef; 1034 return &PL_sv_undef;
808} 1035}
809 1036
810static SV * 1037static SV *
811decode_str (dec_t *dec, int utf8) 1038decode_str (dec_t *dec, int utf8)
812{ 1039{
813 SV *sv = 0; 1040 SV *sv = 0;
814 1041
815 if ((*dec->cur & MINOR_MASK) == MINOR_INDEF) 1042 if (ecb_expect_false ((*dec->cur & MINOR_MASK) == MINOR_INDEF))
816 { 1043 {
817 // indefinite length strings 1044 // indefinite length strings
818 ++dec->cur; 1045 ++dec->cur;
819 1046
820 U8 major = *dec->cur & MAJOR_MISC; 1047 U8 major = *dec->cur & MAJOR_MISC;
… …
853 && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1)) 1080 && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1))
854 av_push (dec->stringref, SvREFCNT_inc_NN (sv)); 1081 av_push (dec->stringref, SvREFCNT_inc_NN (sv));
855 } 1082 }
856 1083
857 if (utf8) 1084 if (utf8)
1085 {
1086 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
1087 if (!is_utf8_string (SvPVX (sv), SvCUR (sv)))
1088 ERR ("corrupted CBOR data (invalid UTF-8 in text string)");
1089
858 SvUTF8_on (sv); 1090 SvUTF8_on (sv);
1091 }
859 1092
860 return sv; 1093 return sv;
861 1094
862fail: 1095fail:
863 SvREFCNT_dec (sv); 1096 SvREFCNT_dec (sv);
… …
882 sv = newRV_noinc (decode_sv (dec)); 1115 sv = newRV_noinc (decode_sv (dec));
883 break; 1116 break;
884 1117
885 case CBOR_TAG_STRINGREF_NAMESPACE: 1118 case CBOR_TAG_STRINGREF_NAMESPACE:
886 { 1119 {
1120 // do not use SAVETMPS/FREETMPS, as these will
1121 // erase mortalised caches, e.g. "shareable"
887 ENTER; SAVETMPS; 1122 ENTER;
888 1123
889 SAVESPTR (dec->stringref); 1124 SAVESPTR (dec->stringref);
890 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ()); 1125 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ());
891 1126
892 sv = decode_sv (dec); 1127 sv = decode_sv (dec);
893 1128
894 FREETMPS; LEAVE; 1129 LEAVE;
895 } 1130 }
896 break; 1131 break;
897 1132
898 case CBOR_TAG_STRINGREF: 1133 case CBOR_TAG_STRINGREF:
899 { 1134 {
900 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT)) 1135 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
901 ERR ("corrupted CBOR data (stringref index not an unsigned integer)"); 1136 ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
902 1137
903 UV idx = decode_uint (dec); 1138 UV idx = decode_uint (dec);
904 1139
905 if (!dec->stringref || (int)idx > AvFILLp (dec->stringref)) 1140 if (!dec->stringref || idx >= (UV)(1 + AvFILLp (dec->stringref)))
906 ERR ("corrupted CBOR data (stringref index out of bounds or outside namespace)"); 1141 ERR ("corrupted CBOR data (stringref index out of bounds or outside namespace)");
907 1142
908 sv = newSVsv (AvARRAY (dec->stringref)[idx]); 1143 sv = newSVsv (AvARRAY (dec->stringref)[idx]);
909 } 1144 }
910 break; 1145 break;
… …
912 case CBOR_TAG_VALUE_SHAREABLE: 1147 case CBOR_TAG_VALUE_SHAREABLE:
913 { 1148 {
914 if (ecb_expect_false (!dec->shareable)) 1149 if (ecb_expect_false (!dec->shareable))
915 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ()); 1150 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ());
916 1151
1152 if (dec->cbor.flags & F_ALLOW_CYCLES)
1153 {
917 sv = newSV (0); 1154 sv = newSV (0);
918 av_push (dec->shareable, SvREFCNT_inc_NN (sv)); 1155 av_push (dec->shareable, SvREFCNT_inc_NN (sv));
919 1156
920 SV *osv = decode_sv (dec); 1157 SV *osv = decode_sv (dec);
921 sv_setsv (sv, osv); 1158 sv_setsv (sv, osv);
922 SvREFCNT_dec_NN (osv); 1159 SvREFCNT_dec_NN (osv);
1160 }
1161 else
1162 {
1163 av_push (dec->shareable, &PL_sv_undef);
1164 int idx = AvFILLp (dec->shareable);
1165 sv = decode_sv (dec);
1166 av_store (dec->shareable, idx, SvREFCNT_inc_NN (sv));
1167 }
923 } 1168 }
924 break; 1169 break;
925 1170
926 case CBOR_TAG_VALUE_SHAREDREF: 1171 case CBOR_TAG_VALUE_SHAREDREF:
927 { 1172 {
928 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT)) 1173 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
929 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)"); 1174 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
930 1175
931 UV idx = decode_uint (dec); 1176 UV idx = decode_uint (dec);
932 1177
933 if (!dec->shareable || (int)idx > AvFILLp (dec->shareable)) 1178 if (!dec->shareable || idx >= (UV)(1 + AvFILLp (dec->shareable)))
934 ERR ("corrupted CBOR data (sharedref index out of bounds)"); 1179 ERR ("corrupted CBOR data (sharedref index out of bounds)");
935 1180
936 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]); 1181 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]);
1182
1183 if (sv == &PL_sv_undef)
1184 ERR ("cyclic CBOR data structure found, but allow_cycles is not enabled");
937 } 1185 }
938 break; 1186 break;
939 1187
940 case CBOR_TAG_PERL_OBJECT: 1188 case CBOR_TAG_PERL_OBJECT:
941 { 1189 {
1190 if (dec->cbor.flags & F_FORBID_OBJECTS)
1191 goto filter;
1192
942 sv = decode_sv (dec); 1193 sv = decode_sv (dec);
943 1194
944 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV) 1195 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
945 ERR ("corrupted CBOR data (non-array perl object)"); 1196 ERR ("corrupted CBOR data (non-array perl object)");
946 1197
… …
956 if (!method) 1207 if (!method)
957 ERR ("cannot decode perl-object (package does not have a THAW method)"); 1208 ERR ("cannot decode perl-object (package does not have a THAW method)");
958 1209
959 dSP; 1210 dSP;
960 1211
961 ENTER; SAVETMPS; PUSHMARK (SP); 1212 ENTER; SAVETMPS;
1213 PUSHMARK (SP);
962 EXTEND (SP, len + 1); 1214 EXTEND (SP, len + 1);
963 // we re-bless the reference to get overload and other niceties right 1215 // we re-bless the reference to get overload and other niceties right
964 PUSHs (*av_fetch (av, 0, 1)); 1216 PUSHs (*av_fetch (av, 0, 1));
965 PUSHs (sv_cbor); 1217 PUSHs (sv_cbor);
966 1218
… …
974 SPAGAIN; 1226 SPAGAIN;
975 1227
976 if (SvTRUE (ERRSV)) 1228 if (SvTRUE (ERRSV))
977 { 1229 {
978 FREETMPS; LEAVE; 1230 FREETMPS; LEAVE;
979 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV)))); 1231 ERR_ERRSV;
980 } 1232 }
981 1233
982 SvREFCNT_dec (sv); 1234 SvREFCNT_dec_NN (sv);
983 sv = SvREFCNT_inc (POPs); 1235 sv = SvREFCNT_inc (POPs);
984 1236
985 PUTBACK; 1237 PUTBACK;
986 1238
987 FREETMPS; LEAVE; 1239 FREETMPS; LEAVE;
988 } 1240 }
989 break; 1241 break;
990 1242
991 default: 1243 default:
1244 filter:
992 { 1245 {
1246 SV *tag_sv = newSVuv (tag);
1247
993 sv = decode_sv (dec); 1248 sv = decode_sv (dec);
994 1249
995 dSP; 1250 dSP;
996 ENTER; SAVETMPS; PUSHMARK (SP); 1251 ENTER; SAVETMPS;
1252 PUSHMARK (SP);
997 EXTEND (SP, 2); 1253 EXTEND (SP, 2);
998 PUSHs (newSVuv (tag)); 1254 PUSHs (tag_sv);
999 PUSHs (sv); 1255 PUSHs (sv);
1000 1256
1001 PUTBACK; 1257 PUTBACK;
1002 int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL); 1258 int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL);
1003 SPAGAIN; 1259 SPAGAIN;
1004 1260
1005 if (SvTRUE (ERRSV)) 1261 if (SvTRUE (ERRSV))
1006 { 1262 {
1263 SvREFCNT_dec_NN (tag_sv);
1007 FREETMPS; LEAVE; 1264 FREETMPS; LEAVE;
1008 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV)))); 1265 ERR_ERRSV;
1009 } 1266 }
1010 1267
1011 if (count) 1268 if (count)
1012 { 1269 {
1270 SvREFCNT_dec_NN (tag_sv);
1013 SvREFCNT_dec (sv); 1271 SvREFCNT_dec_NN (sv);
1014 sv = SvREFCNT_inc (POPs); 1272 sv = SvREFCNT_inc_NN (TOPs);
1273 SP -= count;
1015 } 1274 }
1016 else 1275 else
1017 { 1276 {
1018 AV *av = newAV (); 1277 AV *av = newAV ();
1019 av_push (av, newSVuv (tag)); 1278 av_push (av, tag_sv);
1020 av_push (av, sv); 1279 av_push (av, sv);
1021 1280
1022 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 1281 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
1023 ? cbor_tagged_stash 1282 ? cbor_tagged_stash
1024 : gv_stashpv ("CBOR::XS::Tagged" , 1); 1283 : gv_stashpv ("CBOR::XS::Tagged" , 1);
… …
1110 1369
1111 return newSVnv (ecb_binary64_to_double (fp)); 1370 return newSVnv (ecb_binary64_to_double (fp));
1112 } 1371 }
1113 1372
1114 // 0..19 unassigned simple 1373 // 0..19 unassigned simple
1115 // 24 reserved + unassigned (reserved values are not encodable) 1374 // 24 reserved + unassigned simple (reserved values are not encodable)
1375 // 28-30 unassigned misc
1376 // 31 break code
1116 default: 1377 default:
1117 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)"); 1378 ERR ("corrupted CBOR data (reserved/unassigned/unexpected major 7 value)");
1118 } 1379 }
1119 1380
1120 break; 1381 break;
1121 } 1382 }
1122 1383
… …
1125} 1386}
1126 1387
1127static SV * 1388static SV *
1128decode_cbor (SV *string, CBOR *cbor, char **offset_return) 1389decode_cbor (SV *string, CBOR *cbor, char **offset_return)
1129{ 1390{
1130 dec_t dec = { }; 1391 dec_t dec = { 0 };
1131 SV *sv; 1392 SV *sv;
1132 STRLEN len; 1393 STRLEN len;
1133 char *data = SvPVbyte (string, len); 1394 char *data = SvPVbyte (string, len);
1134 1395
1135 if (len > cbor->max_size && cbor->max_size) 1396 if (len > cbor->max_size && cbor->max_size)
… …
1149 if (dec.cur != dec.end && !dec.err) 1410 if (dec.cur != dec.end && !dec.err)
1150 dec.err = "garbage after CBOR object"; 1411 dec.err = "garbage after CBOR object";
1151 1412
1152 if (dec.err) 1413 if (dec.err)
1153 { 1414 {
1415 if (dec.shareable)
1416 {
1417 // need to break cyclic links, which would all be in shareable
1418 int i;
1419 SV **svp;
1420
1421 for (i = av_len (dec.shareable) + 1; i--; )
1422 if ((svp = av_fetch (dec.shareable, i, 0)))
1423 sv_setsv (*svp, &PL_sv_undef);
1424 }
1425
1154 SvREFCNT_dec (sv); 1426 SvREFCNT_dec_NN (sv);
1427
1428 if (dec.err_sv)
1429 sv_2mortal (dec.err_sv);
1430
1155 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur); 1431 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur);
1156 } 1432 }
1157 1433
1158 sv = sv_2mortal (sv); 1434 sv = sv_2mortal (sv);
1159 1435
1160 return sv; 1436 return sv;
1161} 1437}
1162 1438
1439/////////////////////////////////////////////////////////////////////////////
1440// incremental parser
1441
1442#define INCR_DONE(cbor) (AvFILLp (cbor->incr_count) < 0)
1443
1444// returns 0 for notyet, 1 for success or error
1445static int
1446incr_parse (CBOR *self, SV *cborstr)
1447{
1448 STRLEN cur;
1449 SvPV (cborstr, cur);
1450
1451 while (ecb_expect_true (self->incr_need <= cur))
1452 {
1453 // table of integer count bytes
1454 static I8 incr_len[MINOR_MASK + 1] = {
1455 0, 0, 0, 0, 0, 0, 0, 0,
1456 0, 0, 0, 0, 0, 0, 0, 0,
1457 0, 0, 0, 0, 0, 0, 0, 0,
1458 1, 2, 4, 8,-1,-1,-1,-2
1459 };
1460
1461 const U8 *p = SvPVX (cborstr) + self->incr_pos;
1462 U8 m = *p & MINOR_MASK;
1463 IV count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1464 I8 ilen = incr_len[m];
1465
1466 self->incr_need = self->incr_pos + 1;
1467
1468 if (ecb_expect_false (ilen < 0))
1469 {
1470 if (m != MINOR_INDEF)
1471 return 1; // error
1472
1473 if (*p == (MAJOR_MISC | MINOR_INDEF))
1474 {
1475 if (count >= 0)
1476 return 1; // error
1477
1478 count = 1;
1479 }
1480 else
1481 {
1482 av_push (self->incr_count, newSViv (-1)); //TODO: nest
1483 count = -1;
1484 }
1485 }
1486 else
1487 {
1488 self->incr_need += ilen;
1489 if (ecb_expect_false (self->incr_need > cur))
1490 return 0;
1491
1492 int major = *p >> MAJOR_SHIFT;
1493
1494 switch (major)
1495 {
1496 case MAJOR_TAG >> MAJOR_SHIFT:
1497 ++count; // tags merely prefix another value
1498 break;
1499
1500 case MAJOR_BYTES >> MAJOR_SHIFT:
1501 case MAJOR_TEXT >> MAJOR_SHIFT:
1502 case MAJOR_ARRAY >> MAJOR_SHIFT:
1503 case MAJOR_MAP >> MAJOR_SHIFT:
1504 {
1505 UV len;
1506
1507 if (ecb_expect_false (ilen))
1508 {
1509 len = 0;
1510
1511 do {
1512 len = (len << 8) | *++p;
1513 } while (--ilen);
1514 }
1515 else
1516 len = m;
1517
1518 switch (major)
1519 {
1520 case MAJOR_BYTES >> MAJOR_SHIFT:
1521 case MAJOR_TEXT >> MAJOR_SHIFT:
1522 self->incr_need += len;
1523 if (ecb_expect_false (self->incr_need > cur))
1524 return 0;
1525
1526 break;
1527
1528 case MAJOR_MAP >> MAJOR_SHIFT:
1529 len <<= 1;
1530 case MAJOR_ARRAY >> MAJOR_SHIFT:
1531 if (len)
1532 {
1533 av_push (self->incr_count, newSViv (len + 1)); //TODO: nest
1534 count = len + 1;
1535 }
1536 break;
1537 }
1538 }
1539 }
1540 }
1541
1542 self->incr_pos = self->incr_need;
1543
1544 if (count > 0)
1545 {
1546 while (!--count)
1547 {
1548 if (!AvFILLp (self->incr_count))
1549 return 1; // done
1550
1551 SvREFCNT_dec_NN (av_pop (self->incr_count));
1552 count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1553 }
1554
1555 SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]) = count;
1556 }
1557 }
1558
1559 return 0;
1560}
1561
1562
1163///////////////////////////////////////////////////////////////////////////// 1563/////////////////////////////////////////////////////////////////////////////
1164// XS interface functions 1564// XS interface functions
1165 1565
1166MODULE = CBOR::XS PACKAGE = CBOR::XS 1566MODULE = CBOR::XS PACKAGE = CBOR::XS
1167 1567
… …
1179 1579
1180 default_filter = newSVpv ("CBOR::XS::default_filter", 0); 1580 default_filter = newSVpv ("CBOR::XS::default_filter", 0);
1181 1581
1182 sv_cbor = newSVpv ("CBOR", 0); 1582 sv_cbor = newSVpv ("CBOR", 0);
1183 SvREADONLY_on (sv_cbor); 1583 SvREADONLY_on (sv_cbor);
1584
1585 assert (("STRLEN must be an unsigned type", 0 <= (STRLEN)-1));
1184} 1586}
1185 1587
1186PROTOTYPES: DISABLE 1588PROTOTYPES: DISABLE
1187 1589
1188void CLONE (...) 1590void CLONE (...)
… …
1207void shrink (CBOR *self, int enable = 1) 1609void shrink (CBOR *self, int enable = 1)
1208 ALIAS: 1610 ALIAS:
1209 shrink = F_SHRINK 1611 shrink = F_SHRINK
1210 allow_unknown = F_ALLOW_UNKNOWN 1612 allow_unknown = F_ALLOW_UNKNOWN
1211 allow_sharing = F_ALLOW_SHARING 1613 allow_sharing = F_ALLOW_SHARING
1614 allow_cycles = F_ALLOW_CYCLES
1615 forbid_objects = F_FORBID_OBJECTS
1212 pack_strings = F_PACK_STRINGS 1616 pack_strings = F_PACK_STRINGS
1617 text_keys = F_TEXT_KEYS
1618 text_strings = F_TEXT_STRINGS
1619 validate_utf8 = F_VALIDATE_UTF8
1213 PPCODE: 1620 PPCODE:
1214{ 1621{
1215 if (enable) 1622 if (enable)
1216 self->flags |= ix; 1623 self->flags |= ix;
1217 else 1624 else
… …
1223void get_shrink (CBOR *self) 1630void get_shrink (CBOR *self)
1224 ALIAS: 1631 ALIAS:
1225 get_shrink = F_SHRINK 1632 get_shrink = F_SHRINK
1226 get_allow_unknown = F_ALLOW_UNKNOWN 1633 get_allow_unknown = F_ALLOW_UNKNOWN
1227 get_allow_sharing = F_ALLOW_SHARING 1634 get_allow_sharing = F_ALLOW_SHARING
1635 get_allow_cycles = F_ALLOW_CYCLES
1636 get_forbid_objects = F_FORBID_OBJECTS
1228 get_pack_strings = F_PACK_STRINGS 1637 get_pack_strings = F_PACK_STRINGS
1638 get_text_keys = F_TEXT_KEYS
1639 get_text_strings = F_TEXT_STRINGS
1640 get_validate_utf8 = F_VALIDATE_UTF8
1229 PPCODE: 1641 PPCODE:
1230 XPUSHs (boolSV (self->flags & ix)); 1642 XPUSHs (boolSV (self->flags & ix));
1231 1643
1232void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) 1644void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
1233 PPCODE: 1645 PPCODE:
… …
1282 EXTEND (SP, 2); 1694 EXTEND (SP, 2);
1283 PUSHs (sv); 1695 PUSHs (sv);
1284 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); 1696 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
1285} 1697}
1286 1698
1699void incr_parse (CBOR *self, SV *cborstr)
1700 ALIAS:
1701 incr_parse_multiple = 1
1702 PPCODE:
1703{
1704 if (SvUTF8 (cborstr))
1705 sv_utf8_downgrade (cborstr, 0);
1706
1707 if (!self->incr_count)
1708 {
1709 self->incr_count = newAV ();
1710 self->incr_pos = 0;
1711 self->incr_need = 1;
1712
1713 av_push (self->incr_count, newSViv (1));
1714 }
1715
1716 do
1717 {
1718 if (!incr_parse (self, cborstr))
1719 {
1720 if (self->incr_need > self->max_size && self->max_size)
1721 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
1722 (unsigned long)self->incr_need, (unsigned long)self->max_size);
1723
1724 break;
1725 }
1726
1727 SV *sv;
1728 char *offset;
1729
1730 PUTBACK; sv = decode_cbor (cborstr, self, &offset); SPAGAIN;
1731 XPUSHs (sv);
1732
1733 sv_chop (cborstr, offset);
1734
1735 av_clear (self->incr_count);
1736 av_push (self->incr_count, newSViv (1));
1737
1738 self->incr_pos = 0;
1739 self->incr_need = self->incr_pos + 1;
1740 }
1741 while (ix);
1742}
1743
1744void incr_reset (CBOR *self)
1745 CODE:
1746{
1747 SvREFCNT_dec (self->incr_count);
1748 self->incr_count = 0;
1749}
1750
1287void DESTROY (CBOR *self) 1751void DESTROY (CBOR *self)
1288 PPCODE: 1752 PPCODE:
1289 cbor_free (self); 1753 cbor_free (self);
1290 1754
1291PROTOTYPES: ENABLE 1755PROTOTYPES: ENABLE
… …
1310 cbor_init (&cbor); 1774 cbor_init (&cbor);
1311 PUTBACK; cborstr = decode_cbor (cborstr, &cbor, 0); SPAGAIN; 1775 PUTBACK; cborstr = decode_cbor (cborstr, &cbor, 0); SPAGAIN;
1312 XPUSHs (cborstr); 1776 XPUSHs (cborstr);
1313} 1777}
1314 1778
1779#ifdef __AFL_COMPILER
1780
1781void
1782afl_init ()
1783 CODE:
1784 __AFL_INIT ();
1785
1786int
1787afl_loop (unsigned int count = 10000)
1788 CODE:
1789 RETVAL = __AFL_LOOP (count);
1790 OUTPUT:
1791 RETVAL
1792
1793#endif
1794

Diff Legend

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