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 1.71 by root, Fri Mar 19 17:30:27 2021 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 AS_MAP = 7,
115 // possibly future enhancements: (generic) float, (generic) string
116};
117
99#define F_SHRINK 0x00000001UL 118#define F_SHRINK 0x00000001UL
100#define F_ALLOW_UNKNOWN 0x00000002UL 119#define F_ALLOW_UNKNOWN 0x00000002UL
101#define F_ALLOW_SHARING 0x00000004UL 120#define F_ALLOW_SHARING 0x00000004UL
102#define F_ALLOW_CYCLES 0x00000008UL 121#define F_ALLOW_CYCLES 0x00000008UL
122#define F_FORBID_OBJECTS 0x00000010UL
103#define F_PACK_STRINGS 0x00000010UL 123#define F_PACK_STRINGS 0x00000020UL
124#define F_TEXT_KEYS 0x00000040UL
125#define F_TEXT_STRINGS 0x00000080UL
126#define F_VALIDATE_UTF8 0x00000100UL
104 127
105#define INIT_SIZE 32 // initial scalar size to be allocated 128#define INIT_SIZE 32 // initial scalar size to be allocated
106 129
107#define SB do { 130#define SB do {
108#define SE } while (0) 131#define SE } while (0)
… …
127typedef struct { 150typedef struct {
128 U32 flags; 151 U32 flags;
129 U32 max_depth; 152 U32 max_depth;
130 STRLEN max_size; 153 STRLEN max_size;
131 SV *filter; 154 SV *filter;
155
156 // for the incremental parser
157 STRLEN incr_pos; // the current offset into the text
158 STRLEN incr_need; // minimum bytes needed to decode
159 AV *incr_count; // for every nesting level, the number of outstanding values, or -1 for indef.
132} CBOR; 160} CBOR;
133 161
134ecb_inline void 162ecb_inline void
135cbor_init (CBOR *cbor) 163cbor_init (CBOR *cbor)
136{ 164{
… …
140 168
141ecb_inline void 169ecb_inline void
142cbor_free (CBOR *cbor) 170cbor_free (CBOR *cbor)
143{ 171{
144 SvREFCNT_dec (cbor->filter); 172 SvREFCNT_dec (cbor->filter);
173 SvREFCNT_dec (cbor->incr_count);
145} 174}
146 175
147///////////////////////////////////////////////////////////////////////////// 176/////////////////////////////////////////////////////////////////////////////
148// utility functions 177// utility functions
149 178
… …
172#endif 201#endif
173 } 202 }
174} 203}
175 204
176// minimum length of a string to be registered for stringref 205// minimum length of a string to be registered for stringref
177ecb_inline int 206ecb_inline STRLEN
178minimum_string_length (UV idx) 207minimum_string_length (UV idx)
179{ 208{
180 return idx > 23 209 return idx <= 23 ? 3
181 ? idx > 0xffU 210 : idx <= 0xffU ? 4
182 ? idx > 0xffffU 211 : idx <= 0xffffU ? 5
183 ? idx > 0xffffffffU 212 : idx <= 0xffffffffU ? 7
184 ? 11 213 : 11;
185 : 7
186 : 5
187 : 4
188 : 3;
189} 214}
190 215
191///////////////////////////////////////////////////////////////////////////// 216/////////////////////////////////////////////////////////////////////////////
192// encoder 217// encoder
193 218
… …
206} enc_t; 231} enc_t;
207 232
208ecb_inline void 233ecb_inline void
209need (enc_t *enc, STRLEN len) 234need (enc_t *enc, STRLEN len)
210{ 235{
211 if (ecb_expect_false (enc->cur + len >= enc->end)) 236 if (ecb_expect_false ((uintptr_t)(enc->end - enc->cur) < len))
212 { 237 {
213 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv); 238 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
214 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1); 239 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
215 enc->cur = SvPVX (enc->sv) + cur; 240 enc->cur = SvPVX (enc->sv) + cur;
216 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; 241 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
217 } 242 }
218} 243}
219 244
245static void encode_sv (enc_t *enc, SV *sv);
246
220ecb_inline void 247ecb_inline void
221encode_ch (enc_t *enc, char ch) 248encode_ch (enc_t *enc, char ch)
222{ 249{
223 need (enc, 1); 250 need (enc, 1);
224 *enc->cur++ = ch; 251 *enc->cur++ = ch;
225} 252}
226 253
254// used for tags, intregers, element counts and so on
227static void 255static void
228encode_uint (enc_t *enc, int major, UV len) 256encode_uint (enc_t *enc, int major, UV len)
229{ 257{
230 need (enc, 9); 258 need (enc, 9);
231 259
… …
262 *enc->cur++ = len >> 8; 290 *enc->cur++ = len >> 8;
263 *enc->cur++ = len; 291 *enc->cur++ = len;
264 } 292 }
265} 293}
266 294
295// encodes a perl value into a CBOR integer
296ecb_inline void
297encode_int (enc_t *enc, SV *sv)
298{
299 if (SvIsUV (sv))
300 encode_uint (enc, MAJOR_POS_INT, SvUVX (sv));
301 else if (SvIVX (sv) >= 0)
302 encode_uint (enc, MAJOR_POS_INT, SvIVX (sv));
303 else
304 encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1));
305}
306
267ecb_inline void 307ecb_inline void
268encode_tag (enc_t *enc, UV tag) 308encode_tag (enc_t *enc, UV tag)
269{ 309{
270 encode_uint (enc, MAJOR_TAG, tag); 310 encode_uint (enc, MAJOR_TAG, tag);
271} 311}
272 312
313// exceptional (hopefully) slow path for byte strings that need to be utf8-encoded
314ecb_noinline static void
315encode_str_utf8 (enc_t *enc, int utf8, char *str, STRLEN len)
316{
317 STRLEN ulen = len;
318 U8 *p, *pend = (U8 *)str + len;
319
320 for (p = (U8 *)str; p < pend; ++p)
321 ulen += *p >> 7; // count set high bits
322
323 encode_uint (enc, MAJOR_TEXT, ulen);
324
325 need (enc, ulen);
326 for (p = (U8 *)str; p < pend; ++p)
327 if (*p < 0x80)
328 *enc->cur++ = *p;
329 else
330 {
331 *enc->cur++ = 0xc0 + (*p >> 6);
332 *enc->cur++ = 0x80 + (*p & 63);
333 }
334}
335
273ecb_inline void 336ecb_inline void
274encode_str (enc_t *enc, int utf8, char *str, STRLEN len) 337encode_str (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len)
275{ 338{
339 if (ecb_expect_false (upgrade_utf8))
340 if (!utf8)
341 {
342 encode_str_utf8 (enc, utf8, str, len);
343 return;
344 }
345
276 encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len); 346 encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len);
277 need (enc, len); 347 need (enc, len);
278 memcpy (enc->cur, str, len); 348 memcpy (enc->cur, str, len);
279 enc->cur += len; 349 enc->cur += len;
280} 350}
281 351
282static void 352ecb_inline void
283encode_strref (enc_t *enc, int utf8, char *str, STRLEN len) 353encode_strref (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len)
284{ 354{
285 if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS)) 355 if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS))
286 { 356 {
287 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1); 357 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1);
288 358
… …
299 sv_setuv (*svp, enc->stringref_idx); 369 sv_setuv (*svp, enc->stringref_idx);
300 ++enc->stringref_idx; 370 ++enc->stringref_idx;
301 } 371 }
302 } 372 }
303 373
304 encode_str (enc, utf8, str, len); 374 encode_str (enc, upgrade_utf8, utf8, str, len);
305} 375}
306 376
307static void encode_sv (enc_t *enc, SV *sv); 377ecb_inline void
378encode_float16 (enc_t *enc, NV nv)
379{
380 need (enc, 1+2);
381
382 *enc->cur++ = MAJOR_MISC | MISC_FLOAT16;
383
384 uint16_t fp = ecb_float_to_binary16 (nv);
385
386 if (!ecb_big_endian ())
387 fp = ecb_bswap16 (fp);
388
389 memcpy (enc->cur, &fp, 2);
390 enc->cur += 2;
391}
392
393ecb_inline void
394encode_float32 (enc_t *enc, NV nv)
395{
396 need (enc, 1+4);
397
398 *enc->cur++ = MAJOR_MISC | MISC_FLOAT32;
399
400 uint32_t fp = ecb_float_to_binary32 (nv);
401
402 if (!ecb_big_endian ())
403 fp = ecb_bswap32 (fp);
404
405 memcpy (enc->cur, &fp, 4);
406 enc->cur += 4;
407}
408
409ecb_inline void
410encode_float64 (enc_t *enc, NV nv)
411{
412 need (enc, 1+8);
413
414 *enc->cur++ = MAJOR_MISC | MISC_FLOAT64;
415
416 uint64_t fp = ecb_double_to_binary64 (nv);
417
418 if (!ecb_big_endian ())
419 fp = ecb_bswap64 (fp);
420
421 memcpy (enc->cur, &fp, 8);
422 enc->cur += 8;
423}
424
425ecb_inline void
426encode_bool (enc_t *enc, int istrue)
427{
428 encode_ch (enc, istrue ? MAJOR_MISC | SIMPLE_TRUE : MAJOR_MISC | SIMPLE_FALSE);
429}
430
431// encodes an arrayref containing key-value pairs as CBOR map
432ecb_inline void
433encode_array_as_map (enc_t *enc, SV *sv)
434{
435 if (enc->depth >= enc->cbor.max_depth)
436 croak (ERR_NESTING_EXCEEDED);
437
438 ++enc->depth;
439
440 // as_map does error checking for us, but we re-check in case
441 // things have changed.
442
443 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
444 croak ("CBOR::XS::as_map requires an array reference (did you change the array after calling as_map?)");
445
446 AV *av = (AV *)SvRV (sv);
447 int i, len = av_len (av);
448
449 if (!(len & 1))
450 croak ("CBOR::XS::as_map requires an even number of elements (did you change the array after calling as_map?)");
451
452 encode_uint (enc, MAJOR_MAP, (len + 1) >> 1);
453
454 for (i = 0; i <= len; ++i)
455 {
456 SV **svp = av_fetch (av, i, 0);
457 encode_sv (enc, svp ? *svp : &PL_sv_undef);
458 }
459
460 --enc->depth;
461}
462
463ecb_inline void
464encode_forced (enc_t *enc, UV type, SV *sv)
465{
466 switch (type)
467 {
468 case AS_CBOR:
469 {
470 STRLEN len;
471 char *str = SvPVbyte (sv, len);
472
473 need (enc, len);
474 memcpy (enc->cur, str, len);
475 enc->cur += len;
476 }
477 break;
478
479 case AS_BYTES:
480 {
481 STRLEN len;
482 char *str = SvPVbyte (sv, len);
483 encode_strref (enc, 0, 0, str, len);
484 }
485 break;
486
487 case AS_TEXT:
488 {
489 STRLEN len;
490 char *str = SvPVutf8 (sv, len);
491 encode_strref (enc, 1, 1, str, len);
492 }
493 break;
494
495 case AS_INT: encode_int (enc, sv); break;
496
497 case AS_FLOAT16: encode_float16 (enc, SvNV (sv)); break;
498 case AS_FLOAT32: encode_float32 (enc, SvNV (sv)); break;
499 case AS_FLOAT64: encode_float64 (enc, SvNV (sv)); break;
500
501 case AS_MAP: encode_array_as_map (enc, sv); break;
502
503 default:
504 croak ("encountered malformed CBOR::XS::Tagged object");
505 }
506}
308 507
309static void 508static void
310encode_av (enc_t *enc, AV *av) 509encode_av (enc_t *enc, AV *av)
311{ 510{
312 int i, len = av_len (av); 511 int i, len = av_len (av);
… …
316 515
317 ++enc->depth; 516 ++enc->depth;
318 517
319 encode_uint (enc, MAJOR_ARRAY, len + 1); 518 encode_uint (enc, MAJOR_ARRAY, len + 1);
320 519
520 if (ecb_expect_false (SvMAGICAL (av)))
321 for (i = 0; i <= len; ++i) 521 for (i = 0; i <= len; ++i)
322 { 522 {
323 SV **svp = av_fetch (av, i, 0); 523 SV **svp = av_fetch (av, i, 0);
324 encode_sv (enc, svp ? *svp : &PL_sv_undef); 524 encode_sv (enc, svp ? *svp : &PL_sv_undef);
325 } 525 }
526 else
527 for (i = 0; i <= len; ++i)
528 {
529 SV *sv = AvARRAY (av)[i];
530 encode_sv (enc, sv ? sv : &PL_sv_undef);
531 }
326 532
327 --enc->depth; 533 --enc->depth;
328} 534}
329 535
330static void 536static void
… …
338 ++enc->depth; 544 ++enc->depth;
339 545
340 int pairs = hv_iterinit (hv); 546 int pairs = hv_iterinit (hv);
341 int mg = SvMAGICAL (hv); 547 int mg = SvMAGICAL (hv);
342 548
343 if (mg) 549 if (ecb_expect_false (mg))
344 encode_ch (enc, MAJOR_MAP | MINOR_INDEF); 550 encode_ch (enc, MAJOR_MAP | MINOR_INDEF);
345 else 551 else
346 encode_uint (enc, MAJOR_MAP, pairs); 552 encode_uint (enc, MAJOR_MAP, pairs);
347 553
348 while ((he = hv_iternext (hv))) 554 while ((he = hv_iternext (hv)))
349 { 555 {
350 if (HeKLEN (he) == HEf_SVKEY) 556 if (HeKLEN (he) == HEf_SVKEY)
351 encode_sv (enc, HeSVKEY (he)); 557 encode_sv (enc, HeSVKEY (he));
352 else 558 else
353 encode_strref (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he)); 559 encode_strref (enc, enc->cbor.flags & (F_TEXT_KEYS | F_TEXT_STRINGS), HeKUTF8 (he), HeKEY (he), HeKLEN (he));
354 560
355 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he)); 561 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he));
356 } 562 }
357 563
358 if (mg) 564 if (ecb_expect_false (mg))
359 encode_ch (enc, MAJOR_MISC | MINOR_INDEF); 565 encode_ch (enc, MAJOR_MISC | MINOR_INDEF);
360 566
361 --enc->depth; 567 --enc->depth;
362} 568}
363 569
… …
383 589
384 HV *stash = SvSTASH (sv); 590 HV *stash = SvSTASH (sv);
385 591
386 if (stash == boolean_stash) 592 if (stash == boolean_stash)
387 { 593 {
388 encode_ch (enc, SvIV (sv) ? MAJOR_MISC | SIMPLE_TRUE : MAJOR_MISC | SIMPLE_FALSE); 594 encode_bool (enc, SvIV (sv));
389 return; 595 return;
390 } 596 }
391 else if (stash == error_stash) 597 else if (stash == error_stash)
392 { 598 {
393 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF); 599 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
… …
396 else if (stash == tagged_stash) 602 else if (stash == tagged_stash)
397 { 603 {
398 if (svt != SVt_PVAV) 604 if (svt != SVt_PVAV)
399 croak ("encountered CBOR::XS::Tagged object that isn't an array"); 605 croak ("encountered CBOR::XS::Tagged object that isn't an array");
400 606
607 switch (av_len ((AV *)sv))
608 {
609 case 2-1:
610 // actually a tagged value
401 encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1))); 611 encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1)));
402 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1)); 612 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1));
613 break;
614
615 case 3-1:
616 // a forced type [value, type, undef]
617 encode_forced (enc, SvUV (*av_fetch ((AV *)sv, 1, 1)), *av_fetch ((AV *)sv, 0, 1));
618 break;
619
620 default:
621 croak ("encountered malformed CBOR::XS::Tagged object");
622 }
403 623
404 return; 624 return;
405 } 625 }
406 } 626 }
407 627
408 if (ecb_expect_false (SvREFCNT (sv) > 1) 628 if (ecb_expect_false (SvREFCNT (sv) > 1)
409 && ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING)) 629 && ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING))
410 { 630 {
411 if (!enc->shareable) 631 if (ecb_expect_false (!enc->shareable))
412 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ()); 632 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
413 633
414 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1); 634 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
415 635
416 if (SvOK (*svp)) 636 if (SvOK (*svp))
… …
430 if (ecb_expect_false (SvOBJECT (sv))) 650 if (ecb_expect_false (SvOBJECT (sv)))
431 { 651 {
432 HV *stash = SvSTASH (sv); 652 HV *stash = SvSTASH (sv);
433 GV *method; 653 GV *method;
434 654
655 if (enc->cbor.flags & F_FORBID_OBJECTS)
656 croak ("encountered object '%s', but forbid_objects is enabled",
657 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
435 if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0))) 658 else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
436 { 659 {
437 dSP; 660 dSP;
438 661
439 ENTER; SAVETMPS; PUSHMARK (SP); 662 ENTER; SAVETMPS;
663 PUSHMARK (SP);
440 // we re-bless the reference to get overload and other niceties right 664 // we re-bless the reference to get overload and other niceties right
441 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); 665 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
442 666
443 PUTBACK; 667 PUTBACK;
444 // G_SCALAR ensures that return value is 1 668 // G_SCALAR ensures that return value is 1
… …
457 } 681 }
458 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0) 682 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)
459 { 683 {
460 dSP; 684 dSP;
461 685
462 ENTER; SAVETMPS; PUSHMARK (SP); 686 ENTER; SAVETMPS;
687 PUSHMARK (SP);
463 EXTEND (SP, 2); 688 EXTEND (SP, 2);
464 // we re-bless the reference to get overload and other niceties right 689 // we re-bless the reference to get overload and other niceties right
465 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); 690 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
466 PUSHs (sv_cbor); 691 PUSHs (sv_cbor);
467 692
… …
473 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv) 698 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)); 699 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));
475 700
476 encode_tag (enc, CBOR_TAG_PERL_OBJECT); 701 encode_tag (enc, CBOR_TAG_PERL_OBJECT);
477 encode_uint (enc, MAJOR_ARRAY, count + 1); 702 encode_uint (enc, MAJOR_ARRAY, count + 1);
478 encode_strref (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash)); 703 encode_strref (enc, 0, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
479 704
480 while (count) 705 {
706 int i;
707
708 for (i = 0; i < count; ++i)
481 encode_sv (enc, SP[1 - count--]); 709 encode_sv (enc, SP[i + 1 - count]);
710
711 SP -= count;
712 }
482 713
483 PUTBACK; 714 PUTBACK;
484 715
485 FREETMPS; LEAVE; 716 FREETMPS; LEAVE;
486 } 717 }
… …
508 739
509 if (ecb_expect_false (nv == (NV)(U32)nv)) 740 if (ecb_expect_false (nv == (NV)(U32)nv))
510 encode_uint (enc, MAJOR_POS_INT, (U32)nv); 741 encode_uint (enc, MAJOR_POS_INT, (U32)nv);
511 //TODO: maybe I32? 742 //TODO: maybe I32?
512 else if (ecb_expect_false (nv == (float)nv)) 743 else if (ecb_expect_false (nv == (float)nv))
513 { 744 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 745 else
525 { 746 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} 747}
537 748
538static void 749static void
539encode_sv (enc_t *enc, SV *sv) 750encode_sv (enc_t *enc, SV *sv)
540{ 751{
… …
542 753
543 if (SvPOKp (sv)) 754 if (SvPOKp (sv))
544 { 755 {
545 STRLEN len; 756 STRLEN len;
546 char *str = SvPV (sv, len); 757 char *str = SvPV (sv, len);
547 encode_strref (enc, SvUTF8 (sv), str, len); 758 encode_strref (enc, enc->cbor.flags & F_TEXT_STRINGS, SvUTF8 (sv), str, len);
548 } 759 }
549 else if (SvNOKp (sv)) 760 else if (SvNOKp (sv))
550 encode_nv (enc, sv); 761 encode_nv (enc, sv);
551 else if (SvIOKp (sv)) 762 else if (SvIOKp (sv))
552 { 763 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)) 764 else if (SvROK (sv))
561 encode_rv (enc, SvRV (sv)); 765 encode_rv (enc, SvRV (sv));
562 else if (!SvOK (sv)) 766 else if (!SvOK (sv))
563 encode_ch (enc, MAJOR_MISC | SIMPLE_NULL); 767 encode_ch (enc, MAJOR_MISC | SIMPLE_NULL);
564 else if (enc->cbor.flags & F_ALLOW_UNKNOWN) 768 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
… …
569} 773}
570 774
571static SV * 775static SV *
572encode_cbor (SV *scalar, CBOR *cbor) 776encode_cbor (SV *scalar, CBOR *cbor)
573{ 777{
574 enc_t enc = { }; 778 enc_t enc = { 0 };
575 779
576 enc.cbor = *cbor; 780 enc.cbor = *cbor;
577 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 781 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
578 enc.cur = SvPVX (enc.sv); 782 enc.cur = SvPVX (enc.sv);
579 enc.end = SvEND (enc.sv); 783 enc.end = SvEND (enc.sv);
580 784
581 SvPOK_only (enc.sv); 785 SvPOK_only (enc.sv);
582 786
583 if (cbor->flags & F_PACK_STRINGS) 787 if (cbor->flags & F_PACK_STRINGS)
584 { 788 {
… …
611 U32 depth; // recursion depth 815 U32 depth; // recursion depth
612 U32 maxdepth; // recursion depth limit 816 U32 maxdepth; // recursion depth limit
613 AV *shareable; 817 AV *shareable;
614 AV *stringref; 818 AV *stringref;
615 SV *decode_tagged; 819 SV *decode_tagged;
820 SV *err_sv; // optional sv for error, needs to be freed
616} dec_t; 821} dec_t;
617 822
618#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE 823// set dec->err to ERRSV
824ecb_cold static void
825err_errsv (dec_t *dec)
826{
827 if (!dec->err)
828 {
829 dec->err_sv = newSVsv (ERRSV);
619 830
620#define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data") 831 // chop off the trailing \n
832 SvCUR_set (dec->err_sv, SvCUR (dec->err_sv) - 1);
833 *SvEND (dec->err_sv) = 0;
621 834
835 dec->err = SvPVutf8_nolen (dec->err_sv);
836 }
837}
838
839// the following functions are used to reduce code size and help the compiler to optimise
840ecb_cold static void
841err_set (dec_t *dec, const char *reason)
842{
843 if (!dec->err)
844 dec->err = reason;
845}
846
847ecb_cold static void
848err_unexpected_end (dec_t *dec)
849{
850 err_set (dec, "unexpected end of CBOR data");
851}
852
853#define ERR_DO(do) SB do; goto fail; SE
854#define ERR(reason) ERR_DO (err_set (dec, reason))
855#define ERR_ERRSV ERR_DO (err_errsv (dec))
856
857#define WANT(len) if (ecb_expect_false ((uintptr_t)(dec->end - dec->cur) < (STRLEN)len)) ERR_DO (err_unexpected_end (dec))
858
622#define DEC_INC_DEPTH if (++dec->depth > dec->cbor.max_depth) ERR (ERR_NESTING_EXCEEDED) 859#define DEC_INC_DEPTH if (ecb_expect_false (++dec->depth > dec->cbor.max_depth)) ERR (ERR_NESTING_EXCEEDED)
623#define DEC_DEC_DEPTH --dec->depth 860#define DEC_DEC_DEPTH --dec->depth
624 861
625static UV 862static UV
626decode_uint (dec_t *dec) 863decode_uint (dec_t *dec)
627{ 864{
… …
693 930
694 for (;;) 931 for (;;)
695 { 932 {
696 WANT (1); 933 WANT (1);
697 934
698 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF)) 935 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF) || dec->err)
699 { 936 {
700 ++dec->cur; 937 ++dec->cur;
701 break; 938 break;
702 } 939 }
703 940
704 av_push (av, decode_sv (dec)); 941 av_push (av, decode_sv (dec));
705 } 942 }
706 } 943 }
707 else 944 else
708 { 945 {
709 int i, len = decode_uint (dec); 946 UV i, len = decode_uint (dec);
710 947
711 WANT (len); // complexity check for av_fill - need at least one byte per value, do not allow supersize arrays 948 WANT (len); // complexity check for av_fill - need at least one byte per value, do not allow supersize arrays
712 av_fill (av, len - 1); 949 av_fill (av, len - 1);
713 950
714 for (i = 0; i < len; ++i) 951 for (i = 0; i < len; ++i)
… …
717 954
718 DEC_DEC_DEPTH; 955 DEC_DEC_DEPTH;
719 return newRV_noinc ((SV *)av); 956 return newRV_noinc ((SV *)av);
720 957
721fail: 958fail:
722 SvREFCNT_dec (av); 959 SvREFCNT_dec_NN (av);
723 DEC_DEC_DEPTH; 960 DEC_DEC_DEPTH;
724 return &PL_sv_undef; 961 return &PL_sv_undef;
725} 962}
726 963
727static void 964static void
… …
729{ 966{
730 // for speed reasons, we specialcase single-string 967 // for speed reasons, we specialcase single-string
731 // byte or utf-8 strings as keys, but only when !stringref 968 // byte or utf-8 strings as keys, but only when !stringref
732 969
733 if (ecb_expect_true (!dec->stringref)) 970 if (ecb_expect_true (!dec->stringref))
734 if (ecb_expect_true ((*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8)) 971 if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8))
735 { 972 {
736 I32 len = decode_uint (dec); 973 STRLEN len = decode_uint (dec);
737 char *key = (char *)dec->cur; 974 char *key = (char *)dec->cur;
738 975
976 WANT (len);
739 dec->cur += len; 977 dec->cur += len;
740 978
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); 979 hv_store (hv, key, len, decode_sv (dec), 0);
745 980
746 return; 981 return;
747 } 982 }
748 else if (ecb_expect_true ((*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8)) 983 else if (ecb_expect_true ((U8)(*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8))
749 { 984 {
750 I32 len = decode_uint (dec); 985 STRLEN len = decode_uint (dec);
751 char *key = (char *)dec->cur; 986 char *key = (char *)dec->cur;
752 987
988 WANT (len);
753 dec->cur += len; 989 dec->cur += len;
754 990
755 if (ecb_expect_false (dec->stringref)) 991 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
756 av_push (dec->stringref, newSVpvn_utf8 (key, len, 1)); 992 if (!is_utf8_string ((U8 *)key, len))
993 ERR ("corrupted CBOR data (invalid UTF-8 in map key)");
757 994
758 hv_store (hv, key, -len, decode_sv (dec), 0); 995 hv_store (hv, key, -len, decode_sv (dec), 0);
759 996
760 return; 997 return;
761 } 998 }
762 999
763 SV *k = decode_sv (dec); 1000 SV *k = decode_sv (dec);
764 SV *v = decode_sv (dec); 1001 SV *v = decode_sv (dec);
765 1002
1003 // we leak memory if uncaught exceptions are thrown by random magical
1004 // methods, and this is hopefully the only place where it can happen,
1005 // so if there is a chance of an exception, take the very slow path.
1006 // since catching exceptions is "undocumented/internal/forbidden" by
1007 // the new p5p powers, we need to call out to a perl function :/
1008 if (ecb_expect_false (SvAMAGIC (k)))
1009 {
1010 dSP;
1011
1012 ENTER; SAVETMPS;
1013 PUSHMARK (SP);
1014 EXTEND (SP, 3);
1015 PUSHs (sv_2mortal (newRV_inc ((SV *)hv)));
1016 PUSHs (sv_2mortal (k));
1017 PUSHs (sv_2mortal (v));
1018
1019 PUTBACK;
1020 call_pv ("CBOR::XS::_hv_store", G_VOID | G_DISCARD | G_EVAL);
1021 SPAGAIN;
1022
1023 FREETMPS; LEAVE;
1024
1025 if (SvTRUE (ERRSV))
1026 ERR_ERRSV;
1027
1028 return;
1029 }
1030
766 hv_store_ent (hv, k, v, 0); 1031 hv_store_ent (hv, k, v, 0);
767 SvREFCNT_dec (k); 1032 SvREFCNT_dec_NN (k);
1033
1034fail:
1035 ;
768} 1036}
769 1037
770static SV * 1038static SV *
771decode_hv (dec_t *dec) 1039decode_hv (dec_t *dec)
772{ 1040{
… …
780 1048
781 for (;;) 1049 for (;;)
782 { 1050 {
783 WANT (1); 1051 WANT (1);
784 1052
785 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF)) 1053 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF) || dec->err)
786 { 1054 {
787 ++dec->cur; 1055 ++dec->cur;
788 break; 1056 break;
789 } 1057 }
790 1058
791 decode_he (dec, hv); 1059 decode_he (dec, hv);
792 } 1060 }
793 } 1061 }
794 else 1062 else
795 { 1063 {
796 int pairs = decode_uint (dec); 1064 UV pairs = decode_uint (dec);
1065
1066 WANT (pairs); // complexity check - need at least one byte per value, do not allow supersize hashes
797 1067
798 while (pairs--) 1068 while (pairs--)
799 decode_he (dec, hv); 1069 decode_he (dec, hv);
800 } 1070 }
801 1071
802 DEC_DEC_DEPTH; 1072 DEC_DEC_DEPTH;
803 return newRV_noinc ((SV *)hv); 1073 return newRV_noinc ((SV *)hv);
804 1074
805fail: 1075fail:
806 SvREFCNT_dec (hv); 1076 SvREFCNT_dec_NN (hv);
807 DEC_DEC_DEPTH; 1077 DEC_DEC_DEPTH;
808 return &PL_sv_undef; 1078 return &PL_sv_undef;
809} 1079}
810 1080
811static SV * 1081static SV *
812decode_str (dec_t *dec, int utf8) 1082decode_str (dec_t *dec, int utf8)
813{ 1083{
814 SV *sv = 0; 1084 SV *sv = 0;
815 1085
816 if ((*dec->cur & MINOR_MASK) == MINOR_INDEF) 1086 if (ecb_expect_false ((*dec->cur & MINOR_MASK) == MINOR_INDEF))
817 { 1087 {
818 // indefinite length strings 1088 // indefinite length strings
819 ++dec->cur; 1089 ++dec->cur;
820 1090
821 U8 major = *dec->cur & MAJOR_MISC; 1091 U8 major = *dec->cur & MAJOR_MISC;
… …
854 && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1)) 1124 && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1))
855 av_push (dec->stringref, SvREFCNT_inc_NN (sv)); 1125 av_push (dec->stringref, SvREFCNT_inc_NN (sv));
856 } 1126 }
857 1127
858 if (utf8) 1128 if (utf8)
1129 {
1130 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
1131 if (!is_utf8_string (SvPVX (sv), SvCUR (sv)))
1132 ERR ("corrupted CBOR data (invalid UTF-8 in text string)");
1133
859 SvUTF8_on (sv); 1134 SvUTF8_on (sv);
1135 }
860 1136
861 return sv; 1137 return sv;
862 1138
863fail: 1139fail:
864 SvREFCNT_dec (sv); 1140 SvREFCNT_dec (sv);
… …
883 sv = newRV_noinc (decode_sv (dec)); 1159 sv = newRV_noinc (decode_sv (dec));
884 break; 1160 break;
885 1161
886 case CBOR_TAG_STRINGREF_NAMESPACE: 1162 case CBOR_TAG_STRINGREF_NAMESPACE:
887 { 1163 {
1164 // do not use SAVETMPS/FREETMPS, as these will
1165 // erase mortalised caches, e.g. "shareable"
888 ENTER; SAVETMPS; 1166 ENTER;
889 1167
890 SAVESPTR (dec->stringref); 1168 SAVESPTR (dec->stringref);
891 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ()); 1169 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ());
892 1170
893 sv = decode_sv (dec); 1171 sv = decode_sv (dec);
894 1172
895 FREETMPS; LEAVE; 1173 LEAVE;
896 } 1174 }
897 break; 1175 break;
898 1176
899 case CBOR_TAG_STRINGREF: 1177 case CBOR_TAG_STRINGREF:
900 { 1178 {
901 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT)) 1179 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
902 ERR ("corrupted CBOR data (stringref index not an unsigned integer)"); 1180 ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
903 1181
904 UV idx = decode_uint (dec); 1182 UV idx = decode_uint (dec);
905 1183
906 if (!dec->stringref || (int)idx > AvFILLp (dec->stringref)) 1184 if (!dec->stringref || idx >= (UV)(1 + AvFILLp (dec->stringref)))
907 ERR ("corrupted CBOR data (stringref index out of bounds or outside namespace)"); 1185 ERR ("corrupted CBOR data (stringref index out of bounds or outside namespace)");
908 1186
909 sv = newSVsv (AvARRAY (dec->stringref)[idx]); 1187 sv = newSVsv (AvARRAY (dec->stringref)[idx]);
910 } 1188 }
911 break; 1189 break;
… …
939 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT)) 1217 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
940 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)"); 1218 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
941 1219
942 UV idx = decode_uint (dec); 1220 UV idx = decode_uint (dec);
943 1221
944 if (!dec->shareable || (int)idx > AvFILLp (dec->shareable)) 1222 if (!dec->shareable || idx >= (UV)(1 + AvFILLp (dec->shareable)))
945 ERR ("corrupted CBOR data (sharedref index out of bounds)"); 1223 ERR ("corrupted CBOR data (sharedref index out of bounds)");
946 1224
947 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]); 1225 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]);
948 1226
949 if (sv == &PL_sv_undef) 1227 if (sv == &PL_sv_undef)
… …
951 } 1229 }
952 break; 1230 break;
953 1231
954 case CBOR_TAG_PERL_OBJECT: 1232 case CBOR_TAG_PERL_OBJECT:
955 { 1233 {
1234 if (dec->cbor.flags & F_FORBID_OBJECTS)
1235 goto filter;
1236
956 sv = decode_sv (dec); 1237 sv = decode_sv (dec);
957 1238
958 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV) 1239 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
959 ERR ("corrupted CBOR data (non-array perl object)"); 1240 ERR ("corrupted CBOR data (non-array perl object)");
960 1241
… …
970 if (!method) 1251 if (!method)
971 ERR ("cannot decode perl-object (package does not have a THAW method)"); 1252 ERR ("cannot decode perl-object (package does not have a THAW method)");
972 1253
973 dSP; 1254 dSP;
974 1255
975 ENTER; SAVETMPS; PUSHMARK (SP); 1256 ENTER; SAVETMPS;
1257 PUSHMARK (SP);
976 EXTEND (SP, len + 1); 1258 EXTEND (SP, len + 1);
977 // we re-bless the reference to get overload and other niceties right 1259 // we re-bless the reference to get overload and other niceties right
978 PUSHs (*av_fetch (av, 0, 1)); 1260 PUSHs (*av_fetch (av, 0, 1));
979 PUSHs (sv_cbor); 1261 PUSHs (sv_cbor);
980 1262
… …
988 SPAGAIN; 1270 SPAGAIN;
989 1271
990 if (SvTRUE (ERRSV)) 1272 if (SvTRUE (ERRSV))
991 { 1273 {
992 FREETMPS; LEAVE; 1274 FREETMPS; LEAVE;
993 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV)))); 1275 ERR_ERRSV;
994 } 1276 }
995 1277
996 SvREFCNT_dec (sv); 1278 SvREFCNT_dec_NN (sv);
997 sv = SvREFCNT_inc (POPs); 1279 sv = SvREFCNT_inc (POPs);
998 1280
999 PUTBACK; 1281 PUTBACK;
1000 1282
1001 FREETMPS; LEAVE; 1283 FREETMPS; LEAVE;
1002 } 1284 }
1003 break; 1285 break;
1004 1286
1005 default: 1287 default:
1288 filter:
1006 { 1289 {
1290 SV *tag_sv = newSVuv (tag);
1291
1007 sv = decode_sv (dec); 1292 sv = decode_sv (dec);
1008 1293
1009 dSP; 1294 dSP;
1010 ENTER; SAVETMPS; PUSHMARK (SP); 1295 ENTER; SAVETMPS;
1296 PUSHMARK (SP);
1011 EXTEND (SP, 2); 1297 EXTEND (SP, 2);
1012 PUSHs (newSVuv (tag)); 1298 PUSHs (tag_sv);
1013 PUSHs (sv); 1299 PUSHs (sv);
1014 1300
1015 PUTBACK; 1301 PUTBACK;
1016 int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL); 1302 int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL);
1017 SPAGAIN; 1303 SPAGAIN;
1018 1304
1019 if (SvTRUE (ERRSV)) 1305 if (SvTRUE (ERRSV))
1020 { 1306 {
1307 SvREFCNT_dec_NN (tag_sv);
1021 FREETMPS; LEAVE; 1308 FREETMPS; LEAVE;
1022 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV)))); 1309 ERR_ERRSV;
1023 } 1310 }
1024 1311
1025 if (count) 1312 if (count)
1026 { 1313 {
1314 SvREFCNT_dec_NN (tag_sv);
1027 SvREFCNT_dec (sv); 1315 SvREFCNT_dec_NN (sv);
1028 sv = SvREFCNT_inc (POPs); 1316 sv = SvREFCNT_inc_NN (TOPs);
1317 SP -= count;
1029 } 1318 }
1030 else 1319 else
1031 { 1320 {
1032 AV *av = newAV (); 1321 AV *av = newAV ();
1033 av_push (av, newSVuv (tag)); 1322 av_push (av, tag_sv);
1034 av_push (av, sv); 1323 av_push (av, sv);
1035 1324
1036 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 1325 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
1037 ? cbor_tagged_stash 1326 ? cbor_tagged_stash
1038 : gv_stashpv ("CBOR::XS::Tagged" , 1); 1327 : gv_stashpv ("CBOR::XS::Tagged" , 1);
… …
1124 1413
1125 return newSVnv (ecb_binary64_to_double (fp)); 1414 return newSVnv (ecb_binary64_to_double (fp));
1126 } 1415 }
1127 1416
1128 // 0..19 unassigned simple 1417 // 0..19 unassigned simple
1129 // 24 reserved + unassigned (reserved values are not encodable) 1418 // 24 reserved + unassigned simple (reserved values are not encodable)
1419 // 28-30 unassigned misc
1420 // 31 break code
1130 default: 1421 default:
1131 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)"); 1422 ERR ("corrupted CBOR data (reserved/unassigned/unexpected major 7 value)");
1132 } 1423 }
1133 1424
1134 break; 1425 break;
1135 } 1426 }
1136 1427
… …
1139} 1430}
1140 1431
1141static SV * 1432static SV *
1142decode_cbor (SV *string, CBOR *cbor, char **offset_return) 1433decode_cbor (SV *string, CBOR *cbor, char **offset_return)
1143{ 1434{
1144 dec_t dec = { }; 1435 dec_t dec = { 0 };
1145 SV *sv; 1436 SV *sv;
1146 STRLEN len; 1437 STRLEN len;
1147 char *data = SvPVbyte (string, len); 1438 char *data = SvPVbyte (string, len);
1148 1439
1149 if (len > cbor->max_size && cbor->max_size) 1440 if (len > cbor->max_size && cbor->max_size)
… …
1163 if (dec.cur != dec.end && !dec.err) 1454 if (dec.cur != dec.end && !dec.err)
1164 dec.err = "garbage after CBOR object"; 1455 dec.err = "garbage after CBOR object";
1165 1456
1166 if (dec.err) 1457 if (dec.err)
1167 { 1458 {
1459 if (dec.shareable)
1460 {
1461 // need to break cyclic links, which would all be in shareable
1462 int i;
1463 SV **svp;
1464
1465 for (i = av_len (dec.shareable) + 1; i--; )
1466 if ((svp = av_fetch (dec.shareable, i, 0)))
1467 sv_setsv (*svp, &PL_sv_undef);
1468 }
1469
1168 SvREFCNT_dec (sv); 1470 SvREFCNT_dec_NN (sv);
1471
1472 if (dec.err_sv)
1473 sv_2mortal (dec.err_sv);
1474
1169 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur); 1475 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur);
1170 } 1476 }
1171 1477
1172 sv = sv_2mortal (sv); 1478 sv = sv_2mortal (sv);
1173 1479
1174 return sv; 1480 return sv;
1175} 1481}
1176 1482
1483/////////////////////////////////////////////////////////////////////////////
1484// incremental parser
1485
1486#define INCR_DONE(cbor) (AvFILLp (cbor->incr_count) < 0)
1487
1488// returns 0 for notyet, 1 for success or error
1489static int
1490incr_parse (CBOR *self, SV *cborstr)
1491{
1492 STRLEN cur;
1493 SvPV (cborstr, cur);
1494
1495 while (ecb_expect_true (self->incr_need <= cur))
1496 {
1497 // table of integer count bytes
1498 static I8 incr_len[MINOR_MASK + 1] = {
1499 0, 0, 0, 0, 0, 0, 0, 0,
1500 0, 0, 0, 0, 0, 0, 0, 0,
1501 0, 0, 0, 0, 0, 0, 0, 0,
1502 1, 2, 4, 8,-1,-1,-1,-2
1503 };
1504
1505 const U8 *p = SvPVX (cborstr) + self->incr_pos;
1506 U8 m = *p & MINOR_MASK;
1507 IV count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1508 I8 ilen = incr_len[m];
1509
1510 self->incr_need = self->incr_pos + 1;
1511
1512 if (ecb_expect_false (ilen < 0))
1513 {
1514 if (m != MINOR_INDEF)
1515 return 1; // error
1516
1517 if (*p == (MAJOR_MISC | MINOR_INDEF))
1518 {
1519 if (count >= 0)
1520 return 1; // error
1521
1522 count = 1;
1523 }
1524 else
1525 {
1526 av_push (self->incr_count, newSViv (-1)); //TODO: nest
1527 count = -1;
1528 }
1529 }
1530 else
1531 {
1532 self->incr_need += ilen;
1533 if (ecb_expect_false (self->incr_need > cur))
1534 return 0;
1535
1536 int major = *p >> MAJOR_SHIFT;
1537
1538 switch (major)
1539 {
1540 case MAJOR_TAG >> MAJOR_SHIFT:
1541 ++count; // tags merely prefix another value
1542 break;
1543
1544 case MAJOR_BYTES >> MAJOR_SHIFT:
1545 case MAJOR_TEXT >> MAJOR_SHIFT:
1546 case MAJOR_ARRAY >> MAJOR_SHIFT:
1547 case MAJOR_MAP >> MAJOR_SHIFT:
1548 {
1549 UV len;
1550
1551 if (ecb_expect_false (ilen))
1552 {
1553 len = 0;
1554
1555 do {
1556 len = (len << 8) | *++p;
1557 } while (--ilen);
1558 }
1559 else
1560 len = m;
1561
1562 switch (major)
1563 {
1564 case MAJOR_BYTES >> MAJOR_SHIFT:
1565 case MAJOR_TEXT >> MAJOR_SHIFT:
1566 self->incr_need += len;
1567 if (ecb_expect_false (self->incr_need > cur))
1568 return 0;
1569
1570 break;
1571
1572 case MAJOR_MAP >> MAJOR_SHIFT:
1573 len <<= 1;
1574 /* FALLTHROUGH */
1575 case MAJOR_ARRAY >> MAJOR_SHIFT:
1576 if (len)
1577 {
1578 av_push (self->incr_count, newSViv (len + 1)); //TODO: nest
1579 count = len + 1;
1580 }
1581 break;
1582 }
1583 }
1584 }
1585 }
1586
1587 self->incr_pos = self->incr_need;
1588
1589 if (count > 0)
1590 {
1591 while (!--count)
1592 {
1593 if (!AvFILLp (self->incr_count))
1594 return 1; // done
1595
1596 SvREFCNT_dec_NN (av_pop (self->incr_count));
1597 count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1598 }
1599
1600 SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]) = count;
1601 }
1602 }
1603
1604 return 0;
1605}
1606
1607
1177///////////////////////////////////////////////////////////////////////////// 1608/////////////////////////////////////////////////////////////////////////////
1178// XS interface functions 1609// XS interface functions
1179 1610
1180MODULE = CBOR::XS PACKAGE = CBOR::XS 1611MODULE = CBOR::XS PACKAGE = CBOR::XS
1181 1612
… …
1193 1624
1194 default_filter = newSVpv ("CBOR::XS::default_filter", 0); 1625 default_filter = newSVpv ("CBOR::XS::default_filter", 0);
1195 1626
1196 sv_cbor = newSVpv ("CBOR", 0); 1627 sv_cbor = newSVpv ("CBOR", 0);
1197 SvREADONLY_on (sv_cbor); 1628 SvREADONLY_on (sv_cbor);
1629
1630 assert (("STRLEN must be an unsigned type", 0 <= (STRLEN)-1));
1198} 1631}
1199 1632
1200PROTOTYPES: DISABLE 1633PROTOTYPES: DISABLE
1201 1634
1202void CLONE (...) 1635void CLONE (...)
… …
1222 ALIAS: 1655 ALIAS:
1223 shrink = F_SHRINK 1656 shrink = F_SHRINK
1224 allow_unknown = F_ALLOW_UNKNOWN 1657 allow_unknown = F_ALLOW_UNKNOWN
1225 allow_sharing = F_ALLOW_SHARING 1658 allow_sharing = F_ALLOW_SHARING
1226 allow_cycles = F_ALLOW_CYCLES 1659 allow_cycles = F_ALLOW_CYCLES
1660 forbid_objects = F_FORBID_OBJECTS
1227 pack_strings = F_PACK_STRINGS 1661 pack_strings = F_PACK_STRINGS
1662 text_keys = F_TEXT_KEYS
1663 text_strings = F_TEXT_STRINGS
1664 validate_utf8 = F_VALIDATE_UTF8
1228 PPCODE: 1665 PPCODE:
1229{ 1666{
1230 if (enable) 1667 if (enable)
1231 self->flags |= ix; 1668 self->flags |= ix;
1232 else 1669 else
… …
1239 ALIAS: 1676 ALIAS:
1240 get_shrink = F_SHRINK 1677 get_shrink = F_SHRINK
1241 get_allow_unknown = F_ALLOW_UNKNOWN 1678 get_allow_unknown = F_ALLOW_UNKNOWN
1242 get_allow_sharing = F_ALLOW_SHARING 1679 get_allow_sharing = F_ALLOW_SHARING
1243 get_allow_cycles = F_ALLOW_CYCLES 1680 get_allow_cycles = F_ALLOW_CYCLES
1681 get_forbid_objects = F_FORBID_OBJECTS
1244 get_pack_strings = F_PACK_STRINGS 1682 get_pack_strings = F_PACK_STRINGS
1683 get_text_keys = F_TEXT_KEYS
1684 get_text_strings = F_TEXT_STRINGS
1685 get_validate_utf8 = F_VALIDATE_UTF8
1245 PPCODE: 1686 PPCODE:
1246 XPUSHs (boolSV (self->flags & ix)); 1687 XPUSHs (boolSV (self->flags & ix));
1247 1688
1248void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) 1689void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
1249 PPCODE: 1690 PPCODE:
… …
1298 EXTEND (SP, 2); 1739 EXTEND (SP, 2);
1299 PUSHs (sv); 1740 PUSHs (sv);
1300 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); 1741 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
1301} 1742}
1302 1743
1744void incr_parse (CBOR *self, SV *cborstr)
1745 ALIAS:
1746 incr_parse_multiple = 1
1747 PPCODE:
1748{
1749 if (SvUTF8 (cborstr))
1750 sv_utf8_downgrade (cborstr, 0);
1751
1752 if (!self->incr_count)
1753 {
1754 self->incr_count = newAV ();
1755 self->incr_pos = 0;
1756 self->incr_need = 1;
1757
1758 av_push (self->incr_count, newSViv (1));
1759 }
1760
1761 do
1762 {
1763 if (!incr_parse (self, cborstr))
1764 {
1765 if (self->incr_need > self->max_size && self->max_size)
1766 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
1767 (unsigned long)self->incr_need, (unsigned long)self->max_size);
1768
1769 break;
1770 }
1771
1772 SV *sv;
1773 char *offset;
1774
1775 PUTBACK; sv = decode_cbor (cborstr, self, &offset); SPAGAIN;
1776 XPUSHs (sv);
1777
1778 sv_chop (cborstr, offset);
1779
1780 av_clear (self->incr_count);
1781 av_push (self->incr_count, newSViv (1));
1782
1783 self->incr_pos = 0;
1784 self->incr_need = self->incr_pos + 1;
1785 }
1786 while (ix);
1787}
1788
1789void incr_reset (CBOR *self)
1790 CODE:
1791{
1792 SvREFCNT_dec (self->incr_count);
1793 self->incr_count = 0;
1794}
1795
1303void DESTROY (CBOR *self) 1796void DESTROY (CBOR *self)
1304 PPCODE: 1797 PPCODE:
1305 cbor_free (self); 1798 cbor_free (self);
1306 1799
1307PROTOTYPES: ENABLE 1800PROTOTYPES: ENABLE
… …
1326 cbor_init (&cbor); 1819 cbor_init (&cbor);
1327 PUTBACK; cborstr = decode_cbor (cborstr, &cbor, 0); SPAGAIN; 1820 PUTBACK; cborstr = decode_cbor (cborstr, &cbor, 0); SPAGAIN;
1328 XPUSHs (cborstr); 1821 XPUSHs (cborstr);
1329} 1822}
1330 1823
1824#ifdef __AFL_COMPILER
1825
1826void
1827afl_init ()
1828 CODE:
1829 __AFL_INIT ();
1830
1831int
1832afl_loop (unsigned int count = 10000)
1833 CODE:
1834 RETVAL = __AFL_LOOP (count);
1835 OUTPUT:
1836 RETVAL
1837
1838#endif
1839

Diff Legend

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