ViewVC Help
View File | Revision Log | Show Annotations | Download File
/cvs/CBOR-XS/XS.xs
(Generate patch)

Comparing CBOR-XS/XS.xs (file contents):
Revision 1.39 by root, Sun Dec 1 14:45:03 2013 UTC vs.
Revision 1.70 by root, Fri Dec 4 02:57:14 2020 UTC

… …
6#include <string.h> 6#include <string.h>
7#include <stdlib.h> 7#include <stdlib.h>
8#include <stdio.h> 8#include <stdio.h>
9#include <limits.h> 9#include <limits.h>
10#include <float.h> 10#include <float.h>
11#include <inttypes.h>
11 12
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
104#define F_VALIDATE_UTF8 0x00000020UL 126#define F_VALIDATE_UTF8 0x00000100UL
105 127
106#define INIT_SIZE 32 // initial scalar size to be allocated 128#define INIT_SIZE 32 // initial scalar size to be allocated
107 129
108#define SB do { 130#define SB do {
109#define SE } while (0) 131#define SE } while (0)
… …
128typedef struct { 150typedef struct {
129 U32 flags; 151 U32 flags;
130 U32 max_depth; 152 U32 max_depth;
131 STRLEN max_size; 153 STRLEN max_size;
132 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.
133} CBOR; 160} CBOR;
134 161
135ecb_inline void 162ecb_inline void
136cbor_init (CBOR *cbor) 163cbor_init (CBOR *cbor)
137{ 164{
… …
141 168
142ecb_inline void 169ecb_inline void
143cbor_free (CBOR *cbor) 170cbor_free (CBOR *cbor)
144{ 171{
145 SvREFCNT_dec (cbor->filter); 172 SvREFCNT_dec (cbor->filter);
173 SvREFCNT_dec (cbor->incr_count);
146} 174}
147 175
148///////////////////////////////////////////////////////////////////////////// 176/////////////////////////////////////////////////////////////////////////////
149// utility functions 177// utility functions
150 178
… …
176 204
177// minimum length of a string to be registered for stringref 205// minimum length of a string to be registered for stringref
178ecb_inline int 206ecb_inline int
179minimum_string_length (UV idx) 207minimum_string_length (UV idx)
180{ 208{
181 return idx > 23 209 return idx <= 23 ? 3
182 ? idx > 0xffU 210 : idx <= 0xffU ? 4
183 ? idx > 0xffffU 211 : idx <= 0xffffU ? 5
184 ? idx > 0xffffffffU 212 : idx <= 0xffffffffU ? 7
185 ? 11 213 : 11;
186 : 7
187 : 5
188 : 4
189 : 3;
190} 214}
191 215
192///////////////////////////////////////////////////////////////////////////// 216/////////////////////////////////////////////////////////////////////////////
193// encoder 217// encoder
194 218
… …
207} enc_t; 231} enc_t;
208 232
209ecb_inline void 233ecb_inline void
210need (enc_t *enc, STRLEN len) 234need (enc_t *enc, STRLEN len)
211{ 235{
212 if (ecb_expect_false (enc->cur + len >= enc->end)) 236 if (ecb_expect_false ((uintptr_t)(enc->end - enc->cur) < len))
213 { 237 {
214 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv); 238 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
215 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1); 239 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
216 enc->cur = SvPVX (enc->sv) + cur; 240 enc->cur = SvPVX (enc->sv) + cur;
217 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; 241 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
218 } 242 }
219} 243}
220 244
245static void encode_sv (enc_t *enc, SV *sv);
246
221ecb_inline void 247ecb_inline void
222encode_ch (enc_t *enc, char ch) 248encode_ch (enc_t *enc, char ch)
223{ 249{
224 need (enc, 1); 250 need (enc, 1);
225 *enc->cur++ = ch; 251 *enc->cur++ = ch;
226} 252}
227 253
254// used for tags, intregers, element counts and so on
228static void 255static void
229encode_uint (enc_t *enc, int major, UV len) 256encode_uint (enc_t *enc, int major, UV len)
230{ 257{
231 need (enc, 9); 258 need (enc, 9);
232 259
… …
263 *enc->cur++ = len >> 8; 290 *enc->cur++ = len >> 8;
264 *enc->cur++ = len; 291 *enc->cur++ = len;
265 } 292 }
266} 293}
267 294
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
268ecb_inline void 307ecb_inline void
269encode_tag (enc_t *enc, UV tag) 308encode_tag (enc_t *enc, UV tag)
270{ 309{
271 encode_uint (enc, MAJOR_TAG, tag); 310 encode_uint (enc, MAJOR_TAG, tag);
272} 311}
273 312
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
274ecb_inline void 336ecb_inline void
275encode_str (enc_t *enc, int utf8, char *str, STRLEN len) 337encode_str (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len)
276{ 338{
339 if (ecb_expect_false (upgrade_utf8))
340 if (!utf8)
341 {
342 encode_str_utf8 (enc, utf8, str, len);
343 return;
344 }
345
277 encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len); 346 encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len);
278 need (enc, len); 347 need (enc, len);
279 memcpy (enc->cur, str, len); 348 memcpy (enc->cur, str, len);
280 enc->cur += len; 349 enc->cur += len;
281} 350}
282 351
283static void 352ecb_inline void
284encode_strref (enc_t *enc, int utf8, char *str, STRLEN len) 353encode_strref (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len)
285{ 354{
286 if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS)) 355 if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS))
287 { 356 {
288 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1); 357 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1);
289 358
… …
300 sv_setuv (*svp, enc->stringref_idx); 369 sv_setuv (*svp, enc->stringref_idx);
301 ++enc->stringref_idx; 370 ++enc->stringref_idx;
302 } 371 }
303 } 372 }
304 373
305 encode_str (enc, utf8, str, len); 374 encode_str (enc, upgrade_utf8, utf8, str, len);
306} 375}
307 376
308static 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}
309 507
310static void 508static void
311encode_av (enc_t *enc, AV *av) 509encode_av (enc_t *enc, AV *av)
312{ 510{
313 int i, len = av_len (av); 511 int i, len = av_len (av);
… …
317 515
318 ++enc->depth; 516 ++enc->depth;
319 517
320 encode_uint (enc, MAJOR_ARRAY, len + 1); 518 encode_uint (enc, MAJOR_ARRAY, len + 1);
321 519
520 if (ecb_expect_false (SvMAGICAL (av)))
322 for (i = 0; i <= len; ++i) 521 for (i = 0; i <= len; ++i)
323 { 522 {
324 SV **svp = av_fetch (av, i, 0); 523 SV **svp = av_fetch (av, i, 0);
325 encode_sv (enc, svp ? *svp : &PL_sv_undef); 524 encode_sv (enc, svp ? *svp : &PL_sv_undef);
326 } 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 }
327 532
328 --enc->depth; 533 --enc->depth;
329} 534}
330 535
331static void 536static void
… …
339 ++enc->depth; 544 ++enc->depth;
340 545
341 int pairs = hv_iterinit (hv); 546 int pairs = hv_iterinit (hv);
342 int mg = SvMAGICAL (hv); 547 int mg = SvMAGICAL (hv);
343 548
344 if (mg) 549 if (ecb_expect_false (mg))
345 encode_ch (enc, MAJOR_MAP | MINOR_INDEF); 550 encode_ch (enc, MAJOR_MAP | MINOR_INDEF);
346 else 551 else
347 encode_uint (enc, MAJOR_MAP, pairs); 552 encode_uint (enc, MAJOR_MAP, pairs);
348 553
349 while ((he = hv_iternext (hv))) 554 while ((he = hv_iternext (hv)))
350 { 555 {
351 if (HeKLEN (he) == HEf_SVKEY) 556 if (HeKLEN (he) == HEf_SVKEY)
352 encode_sv (enc, HeSVKEY (he)); 557 encode_sv (enc, HeSVKEY (he));
353 else 558 else
354 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));
355 560
356 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));
357 } 562 }
358 563
359 if (mg) 564 if (ecb_expect_false (mg))
360 encode_ch (enc, MAJOR_MISC | MINOR_INDEF); 565 encode_ch (enc, MAJOR_MISC | MINOR_INDEF);
361 566
362 --enc->depth; 567 --enc->depth;
363} 568}
364 569
… …
384 589
385 HV *stash = SvSTASH (sv); 590 HV *stash = SvSTASH (sv);
386 591
387 if (stash == boolean_stash) 592 if (stash == boolean_stash)
388 { 593 {
389 encode_ch (enc, SvIV (sv) ? MAJOR_MISC | SIMPLE_TRUE : MAJOR_MISC | SIMPLE_FALSE); 594 encode_bool (enc, SvIV (sv));
390 return; 595 return;
391 } 596 }
392 else if (stash == error_stash) 597 else if (stash == error_stash)
393 { 598 {
394 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF); 599 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
… …
397 else if (stash == tagged_stash) 602 else if (stash == tagged_stash)
398 { 603 {
399 if (svt != SVt_PVAV) 604 if (svt != SVt_PVAV)
400 croak ("encountered CBOR::XS::Tagged object that isn't an array"); 605 croak ("encountered CBOR::XS::Tagged object that isn't an array");
401 606
607 switch (av_len ((AV *)sv))
608 {
609 case 2-1:
610 // actually a tagged value
402 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)));
403 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 }
404 623
405 return; 624 return;
406 } 625 }
407 } 626 }
408 627
409 if (ecb_expect_false (SvREFCNT (sv) > 1) 628 if (ecb_expect_false (SvREFCNT (sv) > 1)
410 && ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING)) 629 && ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING))
411 { 630 {
412 if (!enc->shareable) 631 if (ecb_expect_false (!enc->shareable))
413 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ()); 632 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
414 633
415 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1); 634 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
416 635
417 if (SvOK (*svp)) 636 if (SvOK (*svp))
… …
431 if (ecb_expect_false (SvOBJECT (sv))) 650 if (ecb_expect_false (SvOBJECT (sv)))
432 { 651 {
433 HV *stash = SvSTASH (sv); 652 HV *stash = SvSTASH (sv);
434 GV *method; 653 GV *method;
435 654
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))));
436 if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0))) 658 else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
437 { 659 {
438 dSP; 660 dSP;
439 661
440 ENTER; SAVETMPS; PUSHMARK (SP); 662 ENTER; SAVETMPS;
663 PUSHMARK (SP);
441 // 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
442 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); 665 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
443 666
444 PUTBACK; 667 PUTBACK;
445 // G_SCALAR ensures that return value is 1 668 // G_SCALAR ensures that return value is 1
… …
458 } 681 }
459 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0) 682 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)
460 { 683 {
461 dSP; 684 dSP;
462 685
463 ENTER; SAVETMPS; PUSHMARK (SP); 686 ENTER; SAVETMPS;
687 PUSHMARK (SP);
464 EXTEND (SP, 2); 688 EXTEND (SP, 2);
465 // 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
466 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); 690 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
467 PUSHs (sv_cbor); 691 PUSHs (sv_cbor);
468 692
… …
474 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv) 698 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)
475 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash)); 699 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));
476 700
477 encode_tag (enc, CBOR_TAG_PERL_OBJECT); 701 encode_tag (enc, CBOR_TAG_PERL_OBJECT);
478 encode_uint (enc, MAJOR_ARRAY, count + 1); 702 encode_uint (enc, MAJOR_ARRAY, count + 1);
479 encode_strref (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash)); 703 encode_strref (enc, 0, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
480 704
481 while (count) 705 {
706 int i;
707
708 for (i = 0; i < count; ++i)
482 encode_sv (enc, SP[1 - count--]); 709 encode_sv (enc, SP[i + 1 - count]);
710
711 SP -= count;
712 }
483 713
484 PUTBACK; 714 PUTBACK;
485 715
486 FREETMPS; LEAVE; 716 FREETMPS; LEAVE;
487 } 717 }
… …
509 739
510 if (ecb_expect_false (nv == (NV)(U32)nv)) 740 if (ecb_expect_false (nv == (NV)(U32)nv))
511 encode_uint (enc, MAJOR_POS_INT, (U32)nv); 741 encode_uint (enc, MAJOR_POS_INT, (U32)nv);
512 //TODO: maybe I32? 742 //TODO: maybe I32?
513 else if (ecb_expect_false (nv == (float)nv)) 743 else if (ecb_expect_false (nv == (float)nv))
514 { 744 encode_float32 (enc, nv);
515 uint32_t fp = ecb_float_to_binary32 (nv);
516
517 *enc->cur++ = MAJOR_MISC | MISC_FLOAT32;
518
519 if (!ecb_big_endian ())
520 fp = ecb_bswap32 (fp);
521
522 memcpy (enc->cur, &fp, 4);
523 enc->cur += 4;
524 }
525 else 745 else
526 { 746 encode_float64 (enc, nv);
527 uint64_t fp = ecb_double_to_binary64 (nv);
528
529 *enc->cur++ = MAJOR_MISC | MISC_FLOAT64;
530
531 if (!ecb_big_endian ())
532 fp = ecb_bswap64 (fp);
533
534 memcpy (enc->cur, &fp, 8);
535 enc->cur += 8;
536 }
537} 747}
538 748
539static void 749static void
540encode_sv (enc_t *enc, SV *sv) 750encode_sv (enc_t *enc, SV *sv)
541{ 751{
… …
543 753
544 if (SvPOKp (sv)) 754 if (SvPOKp (sv))
545 { 755 {
546 STRLEN len; 756 STRLEN len;
547 char *str = SvPV (sv, len); 757 char *str = SvPV (sv, len);
548 encode_strref (enc, SvUTF8 (sv), str, len); 758 encode_strref (enc, enc->cbor.flags & F_TEXT_STRINGS, SvUTF8 (sv), str, len);
549 } 759 }
550 else if (SvNOKp (sv)) 760 else if (SvNOKp (sv))
551 encode_nv (enc, sv); 761 encode_nv (enc, sv);
552 else if (SvIOKp (sv)) 762 else if (SvIOKp (sv))
553 { 763 encode_int (enc, sv);
554 if (SvIsUV (sv))
555 encode_uint (enc, MAJOR_POS_INT, SvUVX (sv));
556 else if (SvIVX (sv) >= 0)
557 encode_uint (enc, MAJOR_POS_INT, SvIVX (sv));
558 else
559 encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1));
560 }
561 else if (SvROK (sv)) 764 else if (SvROK (sv))
562 encode_rv (enc, SvRV (sv)); 765 encode_rv (enc, SvRV (sv));
563 else if (!SvOK (sv)) 766 else if (!SvOK (sv))
564 encode_ch (enc, MAJOR_MISC | SIMPLE_NULL); 767 encode_ch (enc, MAJOR_MISC | SIMPLE_NULL);
565 else if (enc->cbor.flags & F_ALLOW_UNKNOWN) 768 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
… …
570} 773}
571 774
572static SV * 775static SV *
573encode_cbor (SV *scalar, CBOR *cbor) 776encode_cbor (SV *scalar, CBOR *cbor)
574{ 777{
575 enc_t enc = { }; 778 enc_t enc = { 0 };
576 779
577 enc.cbor = *cbor; 780 enc.cbor = *cbor;
578 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 781 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
579 enc.cur = SvPVX (enc.sv); 782 enc.cur = SvPVX (enc.sv);
580 enc.end = SvEND (enc.sv); 783 enc.end = SvEND (enc.sv);
581 784
582 SvPOK_only (enc.sv); 785 SvPOK_only (enc.sv);
583 786
584 if (cbor->flags & F_PACK_STRINGS) 787 if (cbor->flags & F_PACK_STRINGS)
585 { 788 {
… …
612 U32 depth; // recursion depth 815 U32 depth; // recursion depth
613 U32 maxdepth; // recursion depth limit 816 U32 maxdepth; // recursion depth limit
614 AV *shareable; 817 AV *shareable;
615 AV *stringref; 818 AV *stringref;
616 SV *decode_tagged; 819 SV *decode_tagged;
820 SV *err_sv; // optional sv for error, needs to be freed
617} dec_t; 821} dec_t;
618 822
619#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);
620 830
621#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;
622 834
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
623#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)
624#define DEC_DEC_DEPTH --dec->depth 860#define DEC_DEC_DEPTH --dec->depth
625 861
626static UV 862static UV
627decode_uint (dec_t *dec) 863decode_uint (dec_t *dec)
628{ 864{
… …
694 930
695 for (;;) 931 for (;;)
696 { 932 {
697 WANT (1); 933 WANT (1);
698 934
699 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF)) 935 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF) || dec->err)
700 { 936 {
701 ++dec->cur; 937 ++dec->cur;
702 break; 938 break;
703 } 939 }
704 940
705 av_push (av, decode_sv (dec)); 941 av_push (av, decode_sv (dec));
706 } 942 }
707 } 943 }
708 else 944 else
709 { 945 {
710 int i, len = decode_uint (dec); 946 UV i, len = decode_uint (dec);
711 947
712 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
713 av_fill (av, len - 1); 949 av_fill (av, len - 1);
714 950
715 for (i = 0; i < len; ++i) 951 for (i = 0; i < len; ++i)
… …
718 954
719 DEC_DEC_DEPTH; 955 DEC_DEC_DEPTH;
720 return newRV_noinc ((SV *)av); 956 return newRV_noinc ((SV *)av);
721 957
722fail: 958fail:
723 SvREFCNT_dec (av); 959 SvREFCNT_dec_NN (av);
724 DEC_DEC_DEPTH; 960 DEC_DEC_DEPTH;
725 return &PL_sv_undef; 961 return &PL_sv_undef;
726} 962}
727 963
728static void 964static void
… …
730{ 966{
731 // for speed reasons, we specialcase single-string 967 // for speed reasons, we specialcase single-string
732 // byte or utf-8 strings as keys, but only when !stringref 968 // byte or utf-8 strings as keys, but only when !stringref
733 969
734 if (ecb_expect_true (!dec->stringref)) 970 if (ecb_expect_true (!dec->stringref))
735 if (ecb_expect_true ((*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8)) 971 if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8))
736 { 972 {
737 I32 len = decode_uint (dec); 973 STRLEN len = decode_uint (dec);
738 char *key = (char *)dec->cur; 974 char *key = (char *)dec->cur;
739 975
976 WANT (len);
740 dec->cur += len; 977 dec->cur += len;
741 978
742 hv_store (hv, key, len, decode_sv (dec), 0); 979 hv_store (hv, key, len, decode_sv (dec), 0);
743 980
744 return; 981 return;
745 } 982 }
746 else if (ecb_expect_true ((*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8)) 983 else if (ecb_expect_true ((U8)(*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8))
747 { 984 {
748 I32 len = decode_uint (dec); 985 STRLEN len = decode_uint (dec);
749 char *key = (char *)dec->cur; 986 char *key = (char *)dec->cur;
750 987
988 WANT (len);
751 dec->cur += len; 989 dec->cur += len;
752 990
753 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8)) 991 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
754 if (!is_utf8_string (key, len)) 992 if (!is_utf8_string (key, len))
755 ERR ("corrupted CBOR data (invalid UTF-8 in map key)"); 993 ERR ("corrupted CBOR data (invalid UTF-8 in map key)");
… …
760 } 998 }
761 999
762 SV *k = decode_sv (dec); 1000 SV *k = decode_sv (dec);
763 SV *v = decode_sv (dec); 1001 SV *v = decode_sv (dec);
764 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
765 hv_store_ent (hv, k, v, 0); 1031 hv_store_ent (hv, k, v, 0);
766 SvREFCNT_dec (k); 1032 SvREFCNT_dec_NN (k);
767 1033
768fail: 1034fail:
769 ; 1035 ;
770} 1036}
771 1037
… …
782 1048
783 for (;;) 1049 for (;;)
784 { 1050 {
785 WANT (1); 1051 WANT (1);
786 1052
787 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF)) 1053 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF) || dec->err)
788 { 1054 {
789 ++dec->cur; 1055 ++dec->cur;
790 break; 1056 break;
791 } 1057 }
792 1058
793 decode_he (dec, hv); 1059 decode_he (dec, hv);
794 } 1060 }
795 } 1061 }
796 else 1062 else
797 { 1063 {
798 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
799 1067
800 while (pairs--) 1068 while (pairs--)
801 decode_he (dec, hv); 1069 decode_he (dec, hv);
802 } 1070 }
803 1071
804 DEC_DEC_DEPTH; 1072 DEC_DEC_DEPTH;
805 return newRV_noinc ((SV *)hv); 1073 return newRV_noinc ((SV *)hv);
806 1074
807fail: 1075fail:
808 SvREFCNT_dec (hv); 1076 SvREFCNT_dec_NN (hv);
809 DEC_DEC_DEPTH; 1077 DEC_DEC_DEPTH;
810 return &PL_sv_undef; 1078 return &PL_sv_undef;
811} 1079}
812 1080
813static SV * 1081static SV *
814decode_str (dec_t *dec, int utf8) 1082decode_str (dec_t *dec, int utf8)
815{ 1083{
816 SV *sv = 0; 1084 SV *sv = 0;
817 1085
818 if ((*dec->cur & MINOR_MASK) == MINOR_INDEF) 1086 if (ecb_expect_false ((*dec->cur & MINOR_MASK) == MINOR_INDEF))
819 { 1087 {
820 // indefinite length strings 1088 // indefinite length strings
821 ++dec->cur; 1089 ++dec->cur;
822 1090
823 U8 major = *dec->cur & MAJOR_MISC; 1091 U8 major = *dec->cur & MAJOR_MISC;
… …
891 sv = newRV_noinc (decode_sv (dec)); 1159 sv = newRV_noinc (decode_sv (dec));
892 break; 1160 break;
893 1161
894 case CBOR_TAG_STRINGREF_NAMESPACE: 1162 case CBOR_TAG_STRINGREF_NAMESPACE:
895 { 1163 {
1164 // do not use SAVETMPS/FREETMPS, as these will
1165 // erase mortalised caches, e.g. "shareable"
896 ENTER; SAVETMPS; 1166 ENTER;
897 1167
898 SAVESPTR (dec->stringref); 1168 SAVESPTR (dec->stringref);
899 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ()); 1169 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ());
900 1170
901 sv = decode_sv (dec); 1171 sv = decode_sv (dec);
902 1172
903 FREETMPS; LEAVE; 1173 LEAVE;
904 } 1174 }
905 break; 1175 break;
906 1176
907 case CBOR_TAG_STRINGREF: 1177 case CBOR_TAG_STRINGREF:
908 { 1178 {
909 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT)) 1179 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
910 ERR ("corrupted CBOR data (stringref index not an unsigned integer)"); 1180 ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
911 1181
912 UV idx = decode_uint (dec); 1182 UV idx = decode_uint (dec);
913 1183
914 if (!dec->stringref || (int)idx > AvFILLp (dec->stringref)) 1184 if (!dec->stringref || idx >= (UV)(1 + AvFILLp (dec->stringref)))
915 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)");
916 1186
917 sv = newSVsv (AvARRAY (dec->stringref)[idx]); 1187 sv = newSVsv (AvARRAY (dec->stringref)[idx]);
918 } 1188 }
919 break; 1189 break;
… …
947 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT)) 1217 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
948 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)"); 1218 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
949 1219
950 UV idx = decode_uint (dec); 1220 UV idx = decode_uint (dec);
951 1221
952 if (!dec->shareable || (int)idx > AvFILLp (dec->shareable)) 1222 if (!dec->shareable || idx >= (UV)(1 + AvFILLp (dec->shareable)))
953 ERR ("corrupted CBOR data (sharedref index out of bounds)"); 1223 ERR ("corrupted CBOR data (sharedref index out of bounds)");
954 1224
955 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]); 1225 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]);
956 1226
957 if (sv == &PL_sv_undef) 1227 if (sv == &PL_sv_undef)
… …
959 } 1229 }
960 break; 1230 break;
961 1231
962 case CBOR_TAG_PERL_OBJECT: 1232 case CBOR_TAG_PERL_OBJECT:
963 { 1233 {
1234 if (dec->cbor.flags & F_FORBID_OBJECTS)
1235 goto filter;
1236
964 sv = decode_sv (dec); 1237 sv = decode_sv (dec);
965 1238
966 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV) 1239 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
967 ERR ("corrupted CBOR data (non-array perl object)"); 1240 ERR ("corrupted CBOR data (non-array perl object)");
968 1241
… …
978 if (!method) 1251 if (!method)
979 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)");
980 1253
981 dSP; 1254 dSP;
982 1255
983 ENTER; SAVETMPS; PUSHMARK (SP); 1256 ENTER; SAVETMPS;
1257 PUSHMARK (SP);
984 EXTEND (SP, len + 1); 1258 EXTEND (SP, len + 1);
985 // 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
986 PUSHs (*av_fetch (av, 0, 1)); 1260 PUSHs (*av_fetch (av, 0, 1));
987 PUSHs (sv_cbor); 1261 PUSHs (sv_cbor);
988 1262
… …
996 SPAGAIN; 1270 SPAGAIN;
997 1271
998 if (SvTRUE (ERRSV)) 1272 if (SvTRUE (ERRSV))
999 { 1273 {
1000 FREETMPS; LEAVE; 1274 FREETMPS; LEAVE;
1001 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV)))); 1275 ERR_ERRSV;
1002 } 1276 }
1003 1277
1004 SvREFCNT_dec (sv); 1278 SvREFCNT_dec_NN (sv);
1005 sv = SvREFCNT_inc (POPs); 1279 sv = SvREFCNT_inc (POPs);
1006 1280
1007 PUTBACK; 1281 PUTBACK;
1008 1282
1009 FREETMPS; LEAVE; 1283 FREETMPS; LEAVE;
1010 } 1284 }
1011 break; 1285 break;
1012 1286
1013 default: 1287 default:
1288 filter:
1014 { 1289 {
1290 SV *tag_sv = newSVuv (tag);
1291
1015 sv = decode_sv (dec); 1292 sv = decode_sv (dec);
1016 1293
1017 dSP; 1294 dSP;
1018 ENTER; SAVETMPS; PUSHMARK (SP); 1295 ENTER; SAVETMPS;
1296 PUSHMARK (SP);
1019 EXTEND (SP, 2); 1297 EXTEND (SP, 2);
1020 PUSHs (newSVuv (tag)); 1298 PUSHs (tag_sv);
1021 PUSHs (sv); 1299 PUSHs (sv);
1022 1300
1023 PUTBACK; 1301 PUTBACK;
1024 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);
1025 SPAGAIN; 1303 SPAGAIN;
1026 1304
1027 if (SvTRUE (ERRSV)) 1305 if (SvTRUE (ERRSV))
1028 { 1306 {
1307 SvREFCNT_dec_NN (tag_sv);
1029 FREETMPS; LEAVE; 1308 FREETMPS; LEAVE;
1030 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV)))); 1309 ERR_ERRSV;
1031 } 1310 }
1032 1311
1033 if (count) 1312 if (count)
1034 { 1313 {
1314 SvREFCNT_dec_NN (tag_sv);
1035 SvREFCNT_dec (sv); 1315 SvREFCNT_dec_NN (sv);
1036 sv = SvREFCNT_inc (POPs); 1316 sv = SvREFCNT_inc_NN (TOPs);
1317 SP -= count;
1037 } 1318 }
1038 else 1319 else
1039 { 1320 {
1040 AV *av = newAV (); 1321 AV *av = newAV ();
1041 av_push (av, newSVuv (tag)); 1322 av_push (av, tag_sv);
1042 av_push (av, sv); 1323 av_push (av, sv);
1043 1324
1044 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 1325 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
1045 ? cbor_tagged_stash 1326 ? cbor_tagged_stash
1046 : gv_stashpv ("CBOR::XS::Tagged" , 1); 1327 : gv_stashpv ("CBOR::XS::Tagged" , 1);
… …
1132 1413
1133 return newSVnv (ecb_binary64_to_double (fp)); 1414 return newSVnv (ecb_binary64_to_double (fp));
1134 } 1415 }
1135 1416
1136 // 0..19 unassigned simple 1417 // 0..19 unassigned simple
1137 // 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
1138 default: 1421 default:
1139 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)"); 1422 ERR ("corrupted CBOR data (reserved/unassigned/unexpected major 7 value)");
1140 } 1423 }
1141 1424
1142 break; 1425 break;
1143 } 1426 }
1144 1427
… …
1147} 1430}
1148 1431
1149static SV * 1432static SV *
1150decode_cbor (SV *string, CBOR *cbor, char **offset_return) 1433decode_cbor (SV *string, CBOR *cbor, char **offset_return)
1151{ 1434{
1152 dec_t dec = { }; 1435 dec_t dec = { 0 };
1153 SV *sv; 1436 SV *sv;
1154 STRLEN len; 1437 STRLEN len;
1155 char *data = SvPVbyte (string, len); 1438 char *data = SvPVbyte (string, len);
1156 1439
1157 if (len > cbor->max_size && cbor->max_size) 1440 if (len > cbor->max_size && cbor->max_size)
… …
1173 1456
1174 if (dec.err) 1457 if (dec.err)
1175 { 1458 {
1176 if (dec.shareable) 1459 if (dec.shareable)
1177 { 1460 {
1178 // need to break cyclic links, which whould all be in shareable 1461 // need to break cyclic links, which would all be in shareable
1179 int i; 1462 int i;
1180 SV **svp; 1463 SV **svp;
1181 1464
1182 for (i = av_len (dec.shareable) + 1; i--; ) 1465 for (i = av_len (dec.shareable) + 1; i--; )
1183 if ((svp = av_fetch (dec.shareable, i, 0))) 1466 if ((svp = av_fetch (dec.shareable, i, 0)))
1184 sv_setsv (*svp, &PL_sv_undef); 1467 sv_setsv (*svp, &PL_sv_undef);
1185 } 1468 }
1186 1469
1187 SvREFCNT_dec (sv); 1470 SvREFCNT_dec_NN (sv);
1471
1472 if (dec.err_sv)
1473 sv_2mortal (dec.err_sv);
1474
1188 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);
1189 } 1476 }
1190 1477
1191 sv = sv_2mortal (sv); 1478 sv = sv_2mortal (sv);
1192 1479
1193 return sv; 1480 return sv;
1194} 1481}
1195 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 case MAJOR_ARRAY >> MAJOR_SHIFT:
1575 if (len)
1576 {
1577 av_push (self->incr_count, newSViv (len + 1)); //TODO: nest
1578 count = len + 1;
1579 }
1580 break;
1581 }
1582 }
1583 }
1584 }
1585
1586 self->incr_pos = self->incr_need;
1587
1588 if (count > 0)
1589 {
1590 while (!--count)
1591 {
1592 if (!AvFILLp (self->incr_count))
1593 return 1; // done
1594
1595 SvREFCNT_dec_NN (av_pop (self->incr_count));
1596 count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1597 }
1598
1599 SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]) = count;
1600 }
1601 }
1602
1603 return 0;
1604}
1605
1606
1196///////////////////////////////////////////////////////////////////////////// 1607/////////////////////////////////////////////////////////////////////////////
1197// XS interface functions 1608// XS interface functions
1198 1609
1199MODULE = CBOR::XS PACKAGE = CBOR::XS 1610MODULE = CBOR::XS PACKAGE = CBOR::XS
1200 1611
… …
1212 1623
1213 default_filter = newSVpv ("CBOR::XS::default_filter", 0); 1624 default_filter = newSVpv ("CBOR::XS::default_filter", 0);
1214 1625
1215 sv_cbor = newSVpv ("CBOR", 0); 1626 sv_cbor = newSVpv ("CBOR", 0);
1216 SvREADONLY_on (sv_cbor); 1627 SvREADONLY_on (sv_cbor);
1628
1629 assert (("STRLEN must be an unsigned type", 0 <= (STRLEN)-1));
1217} 1630}
1218 1631
1219PROTOTYPES: DISABLE 1632PROTOTYPES: DISABLE
1220 1633
1221void CLONE (...) 1634void CLONE (...)
… …
1241 ALIAS: 1654 ALIAS:
1242 shrink = F_SHRINK 1655 shrink = F_SHRINK
1243 allow_unknown = F_ALLOW_UNKNOWN 1656 allow_unknown = F_ALLOW_UNKNOWN
1244 allow_sharing = F_ALLOW_SHARING 1657 allow_sharing = F_ALLOW_SHARING
1245 allow_cycles = F_ALLOW_CYCLES 1658 allow_cycles = F_ALLOW_CYCLES
1659 forbid_objects = F_FORBID_OBJECTS
1246 pack_strings = F_PACK_STRINGS 1660 pack_strings = F_PACK_STRINGS
1661 text_keys = F_TEXT_KEYS
1662 text_strings = F_TEXT_STRINGS
1247 validate_utf8 = F_VALIDATE_UTF8 1663 validate_utf8 = F_VALIDATE_UTF8
1248 PPCODE: 1664 PPCODE:
1249{ 1665{
1250 if (enable) 1666 if (enable)
1251 self->flags |= ix; 1667 self->flags |= ix;
… …
1259 ALIAS: 1675 ALIAS:
1260 get_shrink = F_SHRINK 1676 get_shrink = F_SHRINK
1261 get_allow_unknown = F_ALLOW_UNKNOWN 1677 get_allow_unknown = F_ALLOW_UNKNOWN
1262 get_allow_sharing = F_ALLOW_SHARING 1678 get_allow_sharing = F_ALLOW_SHARING
1263 get_allow_cycles = F_ALLOW_CYCLES 1679 get_allow_cycles = F_ALLOW_CYCLES
1680 get_forbid_objects = F_FORBID_OBJECTS
1264 get_pack_strings = F_PACK_STRINGS 1681 get_pack_strings = F_PACK_STRINGS
1682 get_text_keys = F_TEXT_KEYS
1683 get_text_strings = F_TEXT_STRINGS
1265 get_validate_utf8 = F_VALIDATE_UTF8 1684 get_validate_utf8 = F_VALIDATE_UTF8
1266 PPCODE: 1685 PPCODE:
1267 XPUSHs (boolSV (self->flags & ix)); 1686 XPUSHs (boolSV (self->flags & ix));
1268 1687
1269void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) 1688void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
… …
1319 EXTEND (SP, 2); 1738 EXTEND (SP, 2);
1320 PUSHs (sv); 1739 PUSHs (sv);
1321 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); 1740 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
1322} 1741}
1323 1742
1743void incr_parse (CBOR *self, SV *cborstr)
1744 ALIAS:
1745 incr_parse_multiple = 1
1746 PPCODE:
1747{
1748 if (SvUTF8 (cborstr))
1749 sv_utf8_downgrade (cborstr, 0);
1750
1751 if (!self->incr_count)
1752 {
1753 self->incr_count = newAV ();
1754 self->incr_pos = 0;
1755 self->incr_need = 1;
1756
1757 av_push (self->incr_count, newSViv (1));
1758 }
1759
1760 do
1761 {
1762 if (!incr_parse (self, cborstr))
1763 {
1764 if (self->incr_need > self->max_size && self->max_size)
1765 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
1766 (unsigned long)self->incr_need, (unsigned long)self->max_size);
1767
1768 break;
1769 }
1770
1771 SV *sv;
1772 char *offset;
1773
1774 PUTBACK; sv = decode_cbor (cborstr, self, &offset); SPAGAIN;
1775 XPUSHs (sv);
1776
1777 sv_chop (cborstr, offset);
1778
1779 av_clear (self->incr_count);
1780 av_push (self->incr_count, newSViv (1));
1781
1782 self->incr_pos = 0;
1783 self->incr_need = self->incr_pos + 1;
1784 }
1785 while (ix);
1786}
1787
1788void incr_reset (CBOR *self)
1789 CODE:
1790{
1791 SvREFCNT_dec (self->incr_count);
1792 self->incr_count = 0;
1793}
1794
1324void DESTROY (CBOR *self) 1795void DESTROY (CBOR *self)
1325 PPCODE: 1796 PPCODE:
1326 cbor_free (self); 1797 cbor_free (self);
1327 1798
1328PROTOTYPES: ENABLE 1799PROTOTYPES: ENABLE
… …
1347 cbor_init (&cbor); 1818 cbor_init (&cbor);
1348 PUTBACK; cborstr = decode_cbor (cborstr, &cbor, 0); SPAGAIN; 1819 PUTBACK; cborstr = decode_cbor (cborstr, &cbor, 0); SPAGAIN;
1349 XPUSHs (cborstr); 1820 XPUSHs (cborstr);
1350} 1821}
1351 1822
1823#ifdef __AFL_COMPILER
1824
1825void
1826afl_init ()
1827 CODE:
1828 __AFL_INIT ();
1829
1830int
1831afl_loop (unsigned int count = 10000)
1832 CODE:
1833 RETVAL = __AFL_LOOP (count);
1834 OUTPUT:
1835 RETVAL
1836
1837#endif
1838

Diff Legend

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