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 1.76 by root, Thu Sep 7 23:57:58 2023 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
32
33// perl's is_utf8_string interprets len=0 as "calculate len", but we want it to mean 0
34#define cbor_is_utf8_string(str,len) (!(len) || is_utf8_string ((str), (len)))
27 35
28// known major and minor types 36// known major and minor types
29enum cbor_type 37enum cbor_type
30{ 38{
31 MAJOR_SHIFT = 5, 39 MAJOR_SHIFT = 5,
… …
94 CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8 102 CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8
95 103
96 CBOR_TAG_MAGIC = 55799, // self-describe cbor 104 CBOR_TAG_MAGIC = 55799, // self-describe cbor
97}; 105};
98 106
107// known forced types, also hardcoded in CBOR.pm
108enum
109{
110 AS_CBOR = 0,
111 AS_INT = 1,
112 AS_BYTES = 2,
113 AS_TEXT = 3,
114 AS_FLOAT16 = 4,
115 AS_FLOAT32 = 5,
116 AS_FLOAT64 = 6,
117 AS_MAP = 7,
118 // possibly future enhancements: (generic) float, (generic) string
119};
120
99#define F_SHRINK 0x00000001UL 121#define F_SHRINK 0x00000001UL
100#define F_ALLOW_UNKNOWN 0x00000002UL 122#define F_ALLOW_UNKNOWN 0x00000002UL
101#define F_ALLOW_SHARING 0x00000004UL 123#define F_ALLOW_SHARING 0x00000004UL
102#define F_ALLOW_CYCLES 0x00000008UL 124#define F_ALLOW_CYCLES 0x00000008UL
125#define F_FORBID_OBJECTS 0x00000010UL
103#define F_PACK_STRINGS 0x00000010UL 126#define F_PACK_STRINGS 0x00000020UL
127#define F_TEXT_KEYS 0x00000040UL
128#define F_TEXT_STRINGS 0x00000080UL
129#define F_VALIDATE_UTF8 0x00000100UL
104 130
105#define INIT_SIZE 32 // initial scalar size to be allocated 131#define INIT_SIZE 32 // initial scalar size to be allocated
106 132
107#define SB do { 133#define SB do {
108#define SE } while (0) 134#define SE } while (0)
… …
127typedef struct { 153typedef struct {
128 U32 flags; 154 U32 flags;
129 U32 max_depth; 155 U32 max_depth;
130 STRLEN max_size; 156 STRLEN max_size;
131 SV *filter; 157 SV *filter;
158
159 // for the incremental parser
160 STRLEN incr_pos; // the current offset into the text
161 STRLEN incr_need; // minimum bytes needed to decode
162 AV *incr_count; // for every nesting level, the number of outstanding values, or -1 for indef.
132} CBOR; 163} CBOR;
133 164
134ecb_inline void 165ecb_inline void
135cbor_init (CBOR *cbor) 166cbor_init (CBOR *cbor)
136{ 167{
… …
140 171
141ecb_inline void 172ecb_inline void
142cbor_free (CBOR *cbor) 173cbor_free (CBOR *cbor)
143{ 174{
144 SvREFCNT_dec (cbor->filter); 175 SvREFCNT_dec (cbor->filter);
176 SvREFCNT_dec (cbor->incr_count);
145} 177}
146 178
147///////////////////////////////////////////////////////////////////////////// 179/////////////////////////////////////////////////////////////////////////////
148// utility functions 180// utility functions
149 181
… …
172#endif 204#endif
173 } 205 }
174} 206}
175 207
176// minimum length of a string to be registered for stringref 208// minimum length of a string to be registered for stringref
177ecb_inline int 209ecb_inline STRLEN
178minimum_string_length (UV idx) 210minimum_string_length (UV idx)
179{ 211{
180 return idx > 23 212 return idx <= 23 ? 3
181 ? idx > 0xffU 213 : idx <= 0xffU ? 4
182 ? idx > 0xffffU 214 : idx <= 0xffffU ? 5
183 ? idx > 0xffffffffU 215 : idx <= 0xffffffffU ? 7
184 ? 11 216 : 11;
185 : 7
186 : 5
187 : 4
188 : 3;
189} 217}
190 218
191///////////////////////////////////////////////////////////////////////////// 219/////////////////////////////////////////////////////////////////////////////
192// encoder 220// encoder
193 221
… …
206} enc_t; 234} enc_t;
207 235
208ecb_inline void 236ecb_inline void
209need (enc_t *enc, STRLEN len) 237need (enc_t *enc, STRLEN len)
210{ 238{
211 if (ecb_expect_false (enc->cur + len >= enc->end)) 239 if (ecb_expect_false ((uintptr_t)(enc->end - enc->cur) < len))
212 { 240 {
213 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv); 241 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
214 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1); 242 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
215 enc->cur = SvPVX (enc->sv) + cur; 243 enc->cur = SvPVX (enc->sv) + cur;
216 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; 244 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
217 } 245 }
218} 246}
219 247
248static void encode_sv (enc_t *enc, SV *sv);
249
220ecb_inline void 250ecb_inline void
221encode_ch (enc_t *enc, char ch) 251encode_ch (enc_t *enc, char ch)
222{ 252{
223 need (enc, 1); 253 need (enc, 1);
224 *enc->cur++ = ch; 254 *enc->cur++ = ch;
225} 255}
226 256
257// used for tags, intregers, element counts and so on
227static void 258static void
228encode_uint (enc_t *enc, int major, UV len) 259encode_uint (enc_t *enc, int major, UV len)
229{ 260{
230 need (enc, 9); 261 need (enc, 9);
231 262
… …
262 *enc->cur++ = len >> 8; 293 *enc->cur++ = len >> 8;
263 *enc->cur++ = len; 294 *enc->cur++ = len;
264 } 295 }
265} 296}
266 297
298// encodes a perl value into a CBOR integer
299ecb_inline void
300encode_int (enc_t *enc, SV *sv)
301{
302 if (SvIsUV (sv))
303 encode_uint (enc, MAJOR_POS_INT, SvUVX (sv));
304 else if (SvIVX (sv) >= 0)
305 encode_uint (enc, MAJOR_POS_INT, SvIVX (sv));
306 else
307 encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1));
308}
309
267ecb_inline void 310ecb_inline void
268encode_tag (enc_t *enc, UV tag) 311encode_tag (enc_t *enc, UV tag)
269{ 312{
270 encode_uint (enc, MAJOR_TAG, tag); 313 encode_uint (enc, MAJOR_TAG, tag);
271} 314}
272 315
316// exceptional (hopefully) slow path for byte strings that need to be utf8-encoded
317ecb_noinline static void
318encode_str_utf8 (enc_t *enc, int utf8, char *str, STRLEN len)
319{
320 STRLEN ulen = len;
321 U8 *p, *pend = (U8 *)str + len;
322
323 for (p = (U8 *)str; p < pend; ++p)
324 ulen += *p >> 7; // count set high bits
325
326 encode_uint (enc, MAJOR_TEXT, ulen);
327
328 need (enc, ulen);
329 for (p = (U8 *)str; p < pend; ++p)
330 if (*p < 0x80)
331 *enc->cur++ = *p;
332 else
333 {
334 *enc->cur++ = 0xc0 + (*p >> 6);
335 *enc->cur++ = 0x80 + (*p & 63);
336 }
337}
338
273ecb_inline void 339ecb_inline void
274encode_str (enc_t *enc, int utf8, char *str, STRLEN len) 340encode_str (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len)
275{ 341{
342 if (ecb_expect_false (upgrade_utf8))
343 if (!utf8)
344 {
345 encode_str_utf8 (enc, utf8, str, len);
346 return;
347 }
348
276 encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len); 349 encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len);
277 need (enc, len); 350 need (enc, len);
278 memcpy (enc->cur, str, len); 351 memcpy (enc->cur, str, len);
279 enc->cur += len; 352 enc->cur += len;
280} 353}
281 354
282static void 355ecb_inline void
283encode_strref (enc_t *enc, int utf8, char *str, STRLEN len) 356encode_strref (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len)
284{ 357{
285 if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS)) 358 if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS))
286 { 359 {
287 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1); 360 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1);
288 361
… …
299 sv_setuv (*svp, enc->stringref_idx); 372 sv_setuv (*svp, enc->stringref_idx);
300 ++enc->stringref_idx; 373 ++enc->stringref_idx;
301 } 374 }
302 } 375 }
303 376
304 encode_str (enc, utf8, str, len); 377 encode_str (enc, upgrade_utf8, utf8, str, len);
305} 378}
306 379
307static void encode_sv (enc_t *enc, SV *sv); 380ecb_inline void
381encode_float16 (enc_t *enc, NV nv)
382{
383 need (enc, 1+2);
384
385 *enc->cur++ = MAJOR_MISC | MISC_FLOAT16;
386
387 uint16_t fp = ecb_float_to_binary16 (nv);
388
389 if (!ecb_big_endian ())
390 fp = ecb_bswap16 (fp);
391
392 memcpy (enc->cur, &fp, 2);
393 enc->cur += 2;
394}
395
396ecb_inline void
397encode_float32 (enc_t *enc, NV nv)
398{
399 need (enc, 1+4);
400
401 *enc->cur++ = MAJOR_MISC | MISC_FLOAT32;
402
403 uint32_t fp = ecb_float_to_binary32 (nv);
404
405 if (!ecb_big_endian ())
406 fp = ecb_bswap32 (fp);
407
408 memcpy (enc->cur, &fp, 4);
409 enc->cur += 4;
410}
411
412ecb_inline void
413encode_float64 (enc_t *enc, NV nv)
414{
415 need (enc, 1+8);
416
417 *enc->cur++ = MAJOR_MISC | MISC_FLOAT64;
418
419 uint64_t fp = ecb_double_to_binary64 (nv);
420
421 if (!ecb_big_endian ())
422 fp = ecb_bswap64 (fp);
423
424 memcpy (enc->cur, &fp, 8);
425 enc->cur += 8;
426}
427
428ecb_inline void
429encode_bool (enc_t *enc, int istrue)
430{
431 encode_ch (enc, istrue ? MAJOR_MISC | SIMPLE_TRUE : MAJOR_MISC | SIMPLE_FALSE);
432}
433
434// encodes an arrayref containing key-value pairs as CBOR map
435ecb_inline void
436encode_array_as_map (enc_t *enc, SV *sv)
437{
438 if (enc->depth >= enc->cbor.max_depth)
439 croak (ERR_NESTING_EXCEEDED);
440
441 ++enc->depth;
442
443 // as_map does error checking for us, but we re-check in case
444 // things have changed.
445
446 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
447 croak ("CBOR::XS::as_map requires an array reference (did you change the array after calling as_map?)");
448
449 AV *av = (AV *)SvRV (sv);
450 int i, len = av_len (av);
451
452 if (!(len & 1))
453 croak ("CBOR::XS::as_map requires an even number of elements (did you change the array after calling as_map?)");
454
455 encode_uint (enc, MAJOR_MAP, (len + 1) >> 1);
456
457 for (i = 0; i <= len; ++i)
458 {
459 SV **svp = av_fetch (av, i, 0);
460 encode_sv (enc, svp ? *svp : &PL_sv_undef);
461 }
462
463 --enc->depth;
464}
465
466ecb_inline void
467encode_forced (enc_t *enc, UV type, SV *sv)
468{
469 switch (type)
470 {
471 case AS_CBOR:
472 {
473 STRLEN len;
474 char *str = SvPVbyte (sv, len);
475
476 need (enc, len);
477 memcpy (enc->cur, str, len);
478 enc->cur += len;
479 }
480 break;
481
482 case AS_BYTES:
483 {
484 STRLEN len;
485 char *str = SvPVbyte (sv, len);
486 encode_strref (enc, 0, 0, str, len);
487 }
488 break;
489
490 case AS_TEXT:
491 {
492 STRLEN len;
493 char *str = SvPVutf8 (sv, len);
494 encode_strref (enc, 1, 1, str, len);
495 }
496 break;
497
498 case AS_INT: encode_int (enc, sv); break;
499
500 case AS_FLOAT16: encode_float16 (enc, SvNV (sv)); break;
501 case AS_FLOAT32: encode_float32 (enc, SvNV (sv)); break;
502 case AS_FLOAT64: encode_float64 (enc, SvNV (sv)); break;
503
504 case AS_MAP: encode_array_as_map (enc, sv); break;
505
506 default:
507 croak ("encountered malformed CBOR::XS::Tagged object");
508 }
509}
308 510
309static void 511static void
310encode_av (enc_t *enc, AV *av) 512encode_av (enc_t *enc, AV *av)
311{ 513{
312 int i, len = av_len (av); 514 int i, len = av_len (av);
… …
316 518
317 ++enc->depth; 519 ++enc->depth;
318 520
319 encode_uint (enc, MAJOR_ARRAY, len + 1); 521 encode_uint (enc, MAJOR_ARRAY, len + 1);
320 522
523 if (ecb_expect_false (SvMAGICAL (av)))
321 for (i = 0; i <= len; ++i) 524 for (i = 0; i <= len; ++i)
322 { 525 {
323 SV **svp = av_fetch (av, i, 0); 526 SV **svp = av_fetch (av, i, 0);
324 encode_sv (enc, svp ? *svp : &PL_sv_undef); 527 encode_sv (enc, svp ? *svp : &PL_sv_undef);
325 } 528 }
529 else
530 for (i = 0; i <= len; ++i)
531 {
532 SV *sv = AvARRAY (av)[i];
533 encode_sv (enc, sv ? sv : &PL_sv_undef);
534 }
326 535
327 --enc->depth; 536 --enc->depth;
328} 537}
329 538
330static void 539static void
… …
338 ++enc->depth; 547 ++enc->depth;
339 548
340 int pairs = hv_iterinit (hv); 549 int pairs = hv_iterinit (hv);
341 int mg = SvMAGICAL (hv); 550 int mg = SvMAGICAL (hv);
342 551
343 if (mg) 552 if (ecb_expect_false (mg))
344 encode_ch (enc, MAJOR_MAP | MINOR_INDEF); 553 encode_ch (enc, MAJOR_MAP | MINOR_INDEF);
345 else 554 else
346 encode_uint (enc, MAJOR_MAP, pairs); 555 encode_uint (enc, MAJOR_MAP, pairs);
347 556
348 while ((he = hv_iternext (hv))) 557 while ((he = hv_iternext (hv)))
349 { 558 {
350 if (HeKLEN (he) == HEf_SVKEY) 559 if (HeKLEN (he) == HEf_SVKEY)
351 encode_sv (enc, HeSVKEY (he)); 560 encode_sv (enc, HeSVKEY (he));
352 else 561 else
353 encode_strref (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he)); 562 encode_strref (enc, enc->cbor.flags & (F_TEXT_KEYS | F_TEXT_STRINGS), HeKUTF8 (he), HeKEY (he), HeKLEN (he));
354 563
355 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he)); 564 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he));
356 } 565 }
357 566
358 if (mg) 567 if (ecb_expect_false (mg))
359 encode_ch (enc, MAJOR_MISC | MINOR_INDEF); 568 encode_ch (enc, MAJOR_MISC | MINOR_INDEF);
360 569
361 --enc->depth; 570 --enc->depth;
362} 571}
363 572
… …
383 592
384 HV *stash = SvSTASH (sv); 593 HV *stash = SvSTASH (sv);
385 594
386 if (stash == boolean_stash) 595 if (stash == boolean_stash)
387 { 596 {
388 encode_ch (enc, SvIV (sv) ? MAJOR_MISC | SIMPLE_TRUE : MAJOR_MISC | SIMPLE_FALSE); 597 encode_bool (enc, SvIV (sv));
389 return; 598 return;
390 } 599 }
391 else if (stash == error_stash) 600 else if (stash == error_stash)
392 { 601 {
393 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF); 602 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
… …
396 else if (stash == tagged_stash) 605 else if (stash == tagged_stash)
397 { 606 {
398 if (svt != SVt_PVAV) 607 if (svt != SVt_PVAV)
399 croak ("encountered CBOR::XS::Tagged object that isn't an array"); 608 croak ("encountered CBOR::XS::Tagged object that isn't an array");
400 609
610 switch (av_len ((AV *)sv))
611 {
612 case 2-1:
613 // actually a tagged value
401 encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1))); 614 encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1)));
402 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1)); 615 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1));
616 break;
617
618 case 3-1:
619 // a forced type [value, type, undef]
620 encode_forced (enc, SvUV (*av_fetch ((AV *)sv, 1, 1)), *av_fetch ((AV *)sv, 0, 1));
621 break;
622
623 default:
624 croak ("encountered malformed CBOR::XS::Tagged object");
625 }
403 626
404 return; 627 return;
405 } 628 }
406 } 629 }
407 630
408 if (ecb_expect_false (SvREFCNT (sv) > 1) 631 if (ecb_expect_false (SvREFCNT (sv) > 1)
409 && ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING)) 632 && ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING))
410 { 633 {
411 if (!enc->shareable) 634 if (ecb_expect_false (!enc->shareable))
412 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ()); 635 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
413 636
414 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1); 637 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
415 638
416 if (SvOK (*svp)) 639 if (SvOK (*svp))
… …
430 if (ecb_expect_false (SvOBJECT (sv))) 653 if (ecb_expect_false (SvOBJECT (sv)))
431 { 654 {
432 HV *stash = SvSTASH (sv); 655 HV *stash = SvSTASH (sv);
433 GV *method; 656 GV *method;
434 657
658 if (enc->cbor.flags & F_FORBID_OBJECTS)
659 croak ("encountered object '%s', but forbid_objects is enabled",
660 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
435 if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0))) 661 else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
436 { 662 {
437 dSP; 663 dSP;
438 664
439 ENTER; SAVETMPS; PUSHMARK (SP); 665 ENTER; SAVETMPS;
666 PUSHMARK (SP);
440 // we re-bless the reference to get overload and other niceties right 667 // we re-bless the reference to get overload and other niceties right
441 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); 668 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
442 669
443 PUTBACK; 670 PUTBACK;
444 // G_SCALAR ensures that return value is 1 671 // G_SCALAR ensures that return value is 1
… …
457 } 684 }
458 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0) 685 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)
459 { 686 {
460 dSP; 687 dSP;
461 688
462 ENTER; SAVETMPS; PUSHMARK (SP); 689 ENTER; SAVETMPS;
690 PUSHMARK (SP);
463 EXTEND (SP, 2); 691 EXTEND (SP, 2);
464 // we re-bless the reference to get overload and other niceties right 692 // we re-bless the reference to get overload and other niceties right
465 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); 693 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
466 PUSHs (sv_cbor); 694 PUSHs (sv_cbor);
467 695
… …
473 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv) 701 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)); 702 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));
475 703
476 encode_tag (enc, CBOR_TAG_PERL_OBJECT); 704 encode_tag (enc, CBOR_TAG_PERL_OBJECT);
477 encode_uint (enc, MAJOR_ARRAY, count + 1); 705 encode_uint (enc, MAJOR_ARRAY, count + 1);
478 encode_strref (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash)); 706 encode_strref (enc, 0, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
479 707
480 while (count) 708 {
709 int i;
710
711 for (i = 0; i < count; ++i)
481 encode_sv (enc, SP[1 - count--]); 712 encode_sv (enc, SP[i + 1 - count]);
713
714 SP -= count;
715 }
482 716
483 PUTBACK; 717 PUTBACK;
484 718
485 FREETMPS; LEAVE; 719 FREETMPS; LEAVE;
486 } 720 }
… …
508 742
509 if (ecb_expect_false (nv == (NV)(U32)nv)) 743 if (ecb_expect_false (nv == (NV)(U32)nv))
510 encode_uint (enc, MAJOR_POS_INT, (U32)nv); 744 encode_uint (enc, MAJOR_POS_INT, (U32)nv);
511 //TODO: maybe I32? 745 //TODO: maybe I32?
512 else if (ecb_expect_false (nv == (float)nv)) 746 else if (ecb_expect_false (nv == (float)nv))
513 { 747 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 748 else
525 { 749 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} 750}
537 751
538static void 752static void
539encode_sv (enc_t *enc, SV *sv) 753encode_sv (enc_t *enc, SV *sv)
540{ 754{
… …
542 756
543 if (SvPOKp (sv)) 757 if (SvPOKp (sv))
544 { 758 {
545 STRLEN len; 759 STRLEN len;
546 char *str = SvPV (sv, len); 760 char *str = SvPV (sv, len);
547 encode_strref (enc, SvUTF8 (sv), str, len); 761 encode_strref (enc, enc->cbor.flags & F_TEXT_STRINGS, SvUTF8 (sv), str, len);
548 } 762 }
549 else if (SvNOKp (sv)) 763 else if (SvNOKp (sv))
550 encode_nv (enc, sv); 764 encode_nv (enc, sv);
551 else if (SvIOKp (sv)) 765 else if (SvIOKp (sv))
552 { 766 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)) 767 else if (SvROK (sv))
561 encode_rv (enc, SvRV (sv)); 768 encode_rv (enc, SvRV (sv));
562 else if (!SvOK (sv)) 769 else if (!SvOK (sv))
563 encode_ch (enc, MAJOR_MISC | SIMPLE_NULL); 770 encode_ch (enc, MAJOR_MISC | SIMPLE_NULL);
564 else if (enc->cbor.flags & F_ALLOW_UNKNOWN) 771 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
… …
569} 776}
570 777
571static SV * 778static SV *
572encode_cbor (SV *scalar, CBOR *cbor) 779encode_cbor (SV *scalar, CBOR *cbor)
573{ 780{
574 enc_t enc = { }; 781 enc_t enc = { 0 };
575 782
576 enc.cbor = *cbor; 783 enc.cbor = *cbor;
577 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 784 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
578 enc.cur = SvPVX (enc.sv); 785 enc.cur = SvPVX (enc.sv);
579 enc.end = SvEND (enc.sv); 786 enc.end = SvEND (enc.sv);
580 787
581 SvPOK_only (enc.sv); 788 SvPOK_only (enc.sv);
582 789
583 if (cbor->flags & F_PACK_STRINGS) 790 if (cbor->flags & F_PACK_STRINGS)
584 { 791 {
… …
611 U32 depth; // recursion depth 818 U32 depth; // recursion depth
612 U32 maxdepth; // recursion depth limit 819 U32 maxdepth; // recursion depth limit
613 AV *shareable; 820 AV *shareable;
614 AV *stringref; 821 AV *stringref;
615 SV *decode_tagged; 822 SV *decode_tagged;
823 SV *err_sv; // optional sv for error, needs to be freed
616} dec_t; 824} dec_t;
617 825
618#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE 826// set dec->err to ERRSV
827ecb_cold static void
828err_errsv (dec_t *dec)
829{
830 if (!dec->err)
831 {
832 dec->err_sv = newSVsv (ERRSV);
619 833
620#define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data") 834 // chop off the trailing \n
835 SvCUR_set (dec->err_sv, SvCUR (dec->err_sv) - 1);
836 *SvEND (dec->err_sv) = 0;
621 837
838 dec->err = SvPVutf8_nolen (dec->err_sv);
839 }
840}
841
842// the following functions are used to reduce code size and help the compiler to optimise
843ecb_cold static void
844err_set (dec_t *dec, const char *reason)
845{
846 if (!dec->err)
847 dec->err = reason;
848}
849
850ecb_cold static void
851err_unexpected_end (dec_t *dec)
852{
853 err_set (dec, "unexpected end of CBOR data");
854}
855
856#define ERR_DO(do) SB do; goto fail; SE
857#define ERR(reason) ERR_DO (err_set (dec, reason))
858#define ERR_ERRSV ERR_DO (err_errsv (dec))
859
860#define WANT(len) if (ecb_expect_false ((uintptr_t)(dec->end - dec->cur) < (STRLEN)len)) ERR_DO (err_unexpected_end (dec))
861
622#define DEC_INC_DEPTH if (++dec->depth > dec->cbor.max_depth) ERR (ERR_NESTING_EXCEEDED) 862#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 863#define DEC_DEC_DEPTH --dec->depth
624 864
625static UV 865static UV
626decode_uint (dec_t *dec) 866decode_uint (dec_t *dec)
627{ 867{
… …
693 933
694 for (;;) 934 for (;;)
695 { 935 {
696 WANT (1); 936 WANT (1);
697 937
698 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF)) 938 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF) || dec->err)
699 { 939 {
700 ++dec->cur; 940 ++dec->cur;
701 break; 941 break;
702 } 942 }
703 943
704 av_push (av, decode_sv (dec)); 944 av_push (av, decode_sv (dec));
705 } 945 }
706 } 946 }
707 else 947 else
708 { 948 {
709 int i, len = decode_uint (dec); 949 UV i, len = decode_uint (dec);
710 950
711 WANT (len); // complexity check for av_fill - need at least one byte per value, do not allow supersize arrays 951 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); 952 av_fill (av, len - 1);
713 953
714 for (i = 0; i < len; ++i) 954 for (i = 0; i < len; ++i)
… …
717 957
718 DEC_DEC_DEPTH; 958 DEC_DEC_DEPTH;
719 return newRV_noinc ((SV *)av); 959 return newRV_noinc ((SV *)av);
720 960
721fail: 961fail:
722 SvREFCNT_dec (av); 962 SvREFCNT_dec_NN (av);
723 DEC_DEC_DEPTH; 963 DEC_DEC_DEPTH;
724 return &PL_sv_undef; 964 return &PL_sv_undef;
725} 965}
726 966
727static void 967static void
… …
729{ 969{
730 // for speed reasons, we specialcase single-string 970 // for speed reasons, we specialcase single-string
731 // byte or utf-8 strings as keys, but only when !stringref 971 // byte or utf-8 strings as keys, but only when !stringref
732 972
733 if (ecb_expect_true (!dec->stringref)) 973 if (ecb_expect_true (!dec->stringref))
734 if (ecb_expect_true ((*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8)) 974 if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8))
735 { 975 {
736 I32 len = decode_uint (dec); 976 STRLEN len = decode_uint (dec);
737 char *key = (char *)dec->cur; 977 char *key = (char *)dec->cur;
738 978
979 WANT (len);
739 dec->cur += len; 980 dec->cur += len;
740 981
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); 982 hv_store (hv, key, len, decode_sv (dec), 0);
745 983
746 return; 984 return;
747 } 985 }
748 else if (ecb_expect_true ((*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8)) 986 else if (ecb_expect_true ((U8)(*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8))
749 { 987 {
750 I32 len = decode_uint (dec); 988 STRLEN len = decode_uint (dec);
751 char *key = (char *)dec->cur; 989 char *key = (char *)dec->cur;
752 990
991 WANT (len);
753 dec->cur += len; 992 dec->cur += len;
754 993
755 if (ecb_expect_false (dec->stringref)) 994 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
756 av_push (dec->stringref, newSVpvn_utf8 (key, len, 1)); 995 if (!cbor_is_utf8_string ((U8 *)key, len))
996 ERR ("corrupted CBOR data (invalid UTF-8 in map key)");
757 997
758 hv_store (hv, key, -len, decode_sv (dec), 0); 998 hv_store (hv, key, -len, decode_sv (dec), 0);
759 999
760 return; 1000 return;
761 } 1001 }
762 1002
763 SV *k = decode_sv (dec); 1003 SV *k = decode_sv (dec);
764 SV *v = decode_sv (dec); 1004 SV *v = decode_sv (dec);
765 1005
1006 // we leak memory if uncaught exceptions are thrown by random magical
1007 // methods, and this is hopefully the only place where it can happen,
1008 // so if there is a chance of an exception, take the very slow path.
1009 // since catching exceptions is "undocumented/internal/forbidden" by
1010 // the new p5p powers, we need to call out to a perl function :/
1011 if (ecb_expect_false (SvAMAGIC (k)))
1012 {
1013 dSP;
1014
1015 ENTER; SAVETMPS;
1016 PUSHMARK (SP);
1017 EXTEND (SP, 3);
1018 PUSHs (sv_2mortal (newRV_inc ((SV *)hv)));
1019 PUSHs (sv_2mortal (k));
1020 PUSHs (sv_2mortal (v));
1021
1022 PUTBACK;
1023 call_pv ("CBOR::XS::_hv_store", G_VOID | G_DISCARD | G_EVAL);
1024 SPAGAIN;
1025
1026 FREETMPS; LEAVE;
1027
1028 if (SvTRUE (ERRSV))
1029 ERR_ERRSV;
1030
1031 return;
1032 }
1033
766 hv_store_ent (hv, k, v, 0); 1034 hv_store_ent (hv, k, v, 0);
767 SvREFCNT_dec (k); 1035 SvREFCNT_dec_NN (k);
1036
1037fail:
1038 ;
768} 1039}
769 1040
770static SV * 1041static SV *
771decode_hv (dec_t *dec) 1042decode_hv (dec_t *dec)
772{ 1043{
… …
780 1051
781 for (;;) 1052 for (;;)
782 { 1053 {
783 WANT (1); 1054 WANT (1);
784 1055
785 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF)) 1056 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF) || dec->err)
786 { 1057 {
787 ++dec->cur; 1058 ++dec->cur;
788 break; 1059 break;
789 } 1060 }
790 1061
791 decode_he (dec, hv); 1062 decode_he (dec, hv);
792 } 1063 }
793 } 1064 }
794 else 1065 else
795 { 1066 {
796 int pairs = decode_uint (dec); 1067 UV pairs = decode_uint (dec);
1068
1069 WANT (pairs); // complexity check - need at least one byte per value, do not allow supersize hashes
797 1070
798 while (pairs--) 1071 while (pairs--)
799 decode_he (dec, hv); 1072 decode_he (dec, hv);
800 } 1073 }
801 1074
802 DEC_DEC_DEPTH; 1075 DEC_DEC_DEPTH;
803 return newRV_noinc ((SV *)hv); 1076 return newRV_noinc ((SV *)hv);
804 1077
805fail: 1078fail:
806 SvREFCNT_dec (hv); 1079 SvREFCNT_dec_NN (hv);
807 DEC_DEC_DEPTH; 1080 DEC_DEC_DEPTH;
808 return &PL_sv_undef; 1081 return &PL_sv_undef;
809} 1082}
810 1083
811static SV * 1084static SV *
812decode_str (dec_t *dec, int utf8) 1085decode_str (dec_t *dec, int utf8)
813{ 1086{
814 SV *sv = 0; 1087 SV *sv = 0;
815 1088
816 if ((*dec->cur & MINOR_MASK) == MINOR_INDEF) 1089 if (ecb_expect_false ((*dec->cur & MINOR_MASK) == MINOR_INDEF))
817 { 1090 {
818 // indefinite length strings 1091 // indefinite length strings
819 ++dec->cur; 1092 ++dec->cur;
820 1093
821 U8 major = *dec->cur & MAJOR_MISC; 1094 U8 major = *dec->cur & MAJOR_MISC;
… …
854 && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1)) 1127 && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1))
855 av_push (dec->stringref, SvREFCNT_inc_NN (sv)); 1128 av_push (dec->stringref, SvREFCNT_inc_NN (sv));
856 } 1129 }
857 1130
858 if (utf8) 1131 if (utf8)
1132 {
1133 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
1134 if (!cbor_is_utf8_string (SvPVX (sv), SvCUR (sv)))
1135 ERR ("corrupted CBOR data (invalid UTF-8 in text string)");
1136
859 SvUTF8_on (sv); 1137 SvUTF8_on (sv);
1138 }
860 1139
861 return sv; 1140 return sv;
862 1141
863fail: 1142fail:
864 SvREFCNT_dec (sv); 1143 SvREFCNT_dec (sv);
… …
883 sv = newRV_noinc (decode_sv (dec)); 1162 sv = newRV_noinc (decode_sv (dec));
884 break; 1163 break;
885 1164
886 case CBOR_TAG_STRINGREF_NAMESPACE: 1165 case CBOR_TAG_STRINGREF_NAMESPACE:
887 { 1166 {
1167 // do not use SAVETMPS/FREETMPS, as these will
1168 // erase mortalised caches, e.g. "shareable"
888 ENTER; SAVETMPS; 1169 ENTER;
889 1170
890 SAVESPTR (dec->stringref); 1171 SAVESPTR (dec->stringref);
891 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ()); 1172 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ());
892 1173
893 sv = decode_sv (dec); 1174 sv = decode_sv (dec);
894 1175
895 FREETMPS; LEAVE; 1176 LEAVE;
896 } 1177 }
897 break; 1178 break;
898 1179
899 case CBOR_TAG_STRINGREF: 1180 case CBOR_TAG_STRINGREF:
900 { 1181 {
901 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT)) 1182 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
902 ERR ("corrupted CBOR data (stringref index not an unsigned integer)"); 1183 ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
903 1184
904 UV idx = decode_uint (dec); 1185 UV idx = decode_uint (dec);
905 1186
906 if (!dec->stringref || (int)idx > AvFILLp (dec->stringref)) 1187 if (!dec->stringref || idx >= (UV)(1 + AvFILLp (dec->stringref)))
907 ERR ("corrupted CBOR data (stringref index out of bounds or outside namespace)"); 1188 ERR ("corrupted CBOR data (stringref index out of bounds or outside namespace)");
908 1189
909 sv = newSVsv (AvARRAY (dec->stringref)[idx]); 1190 sv = newSVsv (AvARRAY (dec->stringref)[idx]);
910 } 1191 }
911 break; 1192 break;
… …
939 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT)) 1220 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
940 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)"); 1221 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
941 1222
942 UV idx = decode_uint (dec); 1223 UV idx = decode_uint (dec);
943 1224
944 if (!dec->shareable || (int)idx > AvFILLp (dec->shareable)) 1225 if (!dec->shareable || idx >= (UV)(1 + AvFILLp (dec->shareable)))
945 ERR ("corrupted CBOR data (sharedref index out of bounds)"); 1226 ERR ("corrupted CBOR data (sharedref index out of bounds)");
946 1227
947 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]); 1228 sv = newSVsv (AvARRAY (dec->shareable)[idx]);
948 1229
949 if (sv == &PL_sv_undef) 1230 if (sv == &PL_sv_undef)
950 ERR ("cyclic CBOR data structure found, but allow_cycles is not enabled"); 1231 ERR ("cyclic CBOR data structure found, but allow_cycles is not enabled");
951 } 1232 }
952 break; 1233 break;
953 1234
954 case CBOR_TAG_PERL_OBJECT: 1235 case CBOR_TAG_PERL_OBJECT:
955 { 1236 {
1237 if (dec->cbor.flags & F_FORBID_OBJECTS)
1238 goto filter;
1239
956 sv = decode_sv (dec); 1240 sv = decode_sv (dec);
957 1241
958 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV) 1242 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
959 ERR ("corrupted CBOR data (non-array perl object)"); 1243 ERR ("corrupted CBOR data (non-array perl object)");
960 1244
… …
970 if (!method) 1254 if (!method)
971 ERR ("cannot decode perl-object (package does not have a THAW method)"); 1255 ERR ("cannot decode perl-object (package does not have a THAW method)");
972 1256
973 dSP; 1257 dSP;
974 1258
975 ENTER; SAVETMPS; PUSHMARK (SP); 1259 ENTER; SAVETMPS;
1260 PUSHMARK (SP);
976 EXTEND (SP, len + 1); 1261 EXTEND (SP, len + 1);
977 // we re-bless the reference to get overload and other niceties right 1262 // we re-bless the reference to get overload and other niceties right
978 PUSHs (*av_fetch (av, 0, 1)); 1263 PUSHs (*av_fetch (av, 0, 1));
979 PUSHs (sv_cbor); 1264 PUSHs (sv_cbor);
980 1265
… …
988 SPAGAIN; 1273 SPAGAIN;
989 1274
990 if (SvTRUE (ERRSV)) 1275 if (SvTRUE (ERRSV))
991 { 1276 {
992 FREETMPS; LEAVE; 1277 FREETMPS; LEAVE;
993 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV)))); 1278 ERR_ERRSV;
994 } 1279 }
995 1280
996 SvREFCNT_dec (sv); 1281 SvREFCNT_dec_NN (sv);
997 sv = SvREFCNT_inc (POPs); 1282 sv = SvREFCNT_inc (POPs);
998 1283
999 PUTBACK; 1284 PUTBACK;
1000 1285
1001 FREETMPS; LEAVE; 1286 FREETMPS; LEAVE;
1002 } 1287 }
1003 break; 1288 break;
1004 1289
1005 default: 1290 default:
1291 filter:
1006 { 1292 {
1293 SV *tag_sv = newSVuv (tag);
1294
1007 sv = decode_sv (dec); 1295 sv = decode_sv (dec);
1008 1296
1009 dSP; 1297 dSP;
1010 ENTER; SAVETMPS; PUSHMARK (SP); 1298 ENTER; SAVETMPS;
1299 PUSHMARK (SP);
1011 EXTEND (SP, 2); 1300 EXTEND (SP, 2);
1012 PUSHs (newSVuv (tag)); 1301 PUSHs (tag_sv);
1013 PUSHs (sv); 1302 PUSHs (sv);
1014 1303
1015 PUTBACK; 1304 PUTBACK;
1016 int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL); 1305 int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL);
1017 SPAGAIN; 1306 SPAGAIN;
1018 1307
1019 if (SvTRUE (ERRSV)) 1308 if (SvTRUE (ERRSV))
1020 { 1309 {
1310 SvREFCNT_dec_NN (tag_sv);
1021 FREETMPS; LEAVE; 1311 FREETMPS; LEAVE;
1022 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV)))); 1312 ERR_ERRSV;
1023 } 1313 }
1024 1314
1025 if (count) 1315 if (count)
1026 { 1316 {
1317 SvREFCNT_dec_NN (tag_sv);
1027 SvREFCNT_dec (sv); 1318 SvREFCNT_dec_NN (sv);
1028 sv = SvREFCNT_inc (POPs); 1319 sv = SvREFCNT_inc_NN (TOPs);
1320 SP -= count;
1029 } 1321 }
1030 else 1322 else
1031 { 1323 {
1032 AV *av = newAV (); 1324 AV *av = newAV ();
1033 av_push (av, newSVuv (tag)); 1325 av_push (av, tag_sv);
1034 av_push (av, sv); 1326 av_push (av, sv);
1035 1327
1036 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 1328 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
1037 ? cbor_tagged_stash 1329 ? cbor_tagged_stash
1038 : gv_stashpv ("CBOR::XS::Tagged" , 1); 1330 : gv_stashpv ("CBOR::XS::Tagged" , 1);
… …
1124 1416
1125 return newSVnv (ecb_binary64_to_double (fp)); 1417 return newSVnv (ecb_binary64_to_double (fp));
1126 } 1418 }
1127 1419
1128 // 0..19 unassigned simple 1420 // 0..19 unassigned simple
1129 // 24 reserved + unassigned (reserved values are not encodable) 1421 // 24 reserved + unassigned simple (reserved values are not encodable)
1422 // 28-30 unassigned misc
1423 // 31 break code
1130 default: 1424 default:
1131 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)"); 1425 ERR ("corrupted CBOR data (reserved/unassigned/unexpected major 7 value)");
1132 } 1426 }
1133 1427
1134 break; 1428 break;
1135 } 1429 }
1136 1430
… …
1139} 1433}
1140 1434
1141static SV * 1435static SV *
1142decode_cbor (SV *string, CBOR *cbor, char **offset_return) 1436decode_cbor (SV *string, CBOR *cbor, char **offset_return)
1143{ 1437{
1144 dec_t dec = { }; 1438 dec_t dec = { 0 };
1145 SV *sv; 1439 SV *sv;
1146 STRLEN len; 1440 STRLEN len;
1147 char *data = SvPVbyte (string, len); 1441 char *data = SvPVbyte (string, len);
1148 1442
1149 if (len > cbor->max_size && cbor->max_size) 1443 if (len > cbor->max_size && cbor->max_size)
… …
1163 if (dec.cur != dec.end && !dec.err) 1457 if (dec.cur != dec.end && !dec.err)
1164 dec.err = "garbage after CBOR object"; 1458 dec.err = "garbage after CBOR object";
1165 1459
1166 if (dec.err) 1460 if (dec.err)
1167 { 1461 {
1462 if (dec.shareable)
1463 {
1464 // need to break cyclic links, which would all be in shareable
1465 int i;
1466 SV **svp;
1467
1468 for (i = av_len (dec.shareable) + 1; i--; )
1469 if ((svp = av_fetch (dec.shareable, i, 0)))
1470 sv_setsv (*svp, &PL_sv_undef);
1471 }
1472
1168 SvREFCNT_dec (sv); 1473 SvREFCNT_dec_NN (sv);
1474
1475 if (dec.err_sv)
1476 sv_2mortal (dec.err_sv);
1477
1169 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur); 1478 croak ("%s, at offset %ld (octet 0x%02x)", dec.err, (long)(dec.cur - (U8 *)data), (int)(uint8_t)*dec.cur);
1170 } 1479 }
1171 1480
1172 sv = sv_2mortal (sv); 1481 sv = sv_2mortal (sv);
1173 1482
1174 return sv; 1483 return sv;
1175} 1484}
1176 1485
1486/////////////////////////////////////////////////////////////////////////////
1487// incremental parser
1488
1489#define INCR_DONE(cbor) (AvFILLp (cbor->incr_count) < 0)
1490
1491// returns 0 for notyet, 1 for success or error
1492static int
1493incr_parse (CBOR *self, SV *cborstr)
1494{
1495 STRLEN cur;
1496 SvPV (cborstr, cur);
1497
1498 while (ecb_expect_true (self->incr_need <= cur))
1499 {
1500 // table of integer count bytes
1501 static I8 incr_len[MINOR_MASK + 1] = {
1502 0, 0, 0, 0, 0, 0, 0, 0,
1503 0, 0, 0, 0, 0, 0, 0, 0,
1504 0, 0, 0, 0, 0, 0, 0, 0,
1505 1, 2, 4, 8,-1,-1,-1,-2
1506 };
1507
1508 const U8 *p = SvPVX (cborstr) + self->incr_pos;
1509 U8 m = *p & MINOR_MASK;
1510 IV count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1511 I8 ilen = incr_len[m];
1512
1513 self->incr_need = self->incr_pos + 1;
1514
1515 if (ecb_expect_false (ilen < 0))
1516 {
1517 if (m != MINOR_INDEF)
1518 return 1; // error
1519
1520 if (*p == (MAJOR_MISC | MINOR_INDEF))
1521 {
1522 if (count >= 0)
1523 return 1; // error
1524
1525 count = 1;
1526 }
1527 else
1528 {
1529 av_push (self->incr_count, newSViv (-1)); //TODO: nest
1530 count = -1;
1531 }
1532 }
1533 else
1534 {
1535 self->incr_need += ilen;
1536 if (ecb_expect_false (self->incr_need > cur))
1537 return 0;
1538
1539 int major = *p >> MAJOR_SHIFT;
1540
1541 switch (major)
1542 {
1543 case MAJOR_TAG >> MAJOR_SHIFT:
1544 ++count; // tags merely prefix another value
1545 break;
1546
1547 case MAJOR_BYTES >> MAJOR_SHIFT:
1548 case MAJOR_TEXT >> MAJOR_SHIFT:
1549 case MAJOR_ARRAY >> MAJOR_SHIFT:
1550 case MAJOR_MAP >> MAJOR_SHIFT:
1551 {
1552 UV len;
1553
1554 if (ecb_expect_false (ilen))
1555 {
1556 len = 0;
1557
1558 do {
1559 len = (len << 8) | *++p;
1560 } while (--ilen);
1561 }
1562 else
1563 len = m;
1564
1565 switch (major)
1566 {
1567 case MAJOR_BYTES >> MAJOR_SHIFT:
1568 case MAJOR_TEXT >> MAJOR_SHIFT:
1569 self->incr_need += len;
1570 if (ecb_expect_false (self->incr_need > cur))
1571 return 0;
1572
1573 break;
1574
1575 case MAJOR_MAP >> MAJOR_SHIFT:
1576 len <<= 1;
1577 /* FALLTHROUGH */
1578 case MAJOR_ARRAY >> MAJOR_SHIFT:
1579 if (len)
1580 {
1581 av_push (self->incr_count, newSViv (len + 1)); //TODO: nest
1582 count = len + 1;
1583 }
1584 break;
1585 }
1586 }
1587 }
1588 }
1589
1590 self->incr_pos = self->incr_need;
1591
1592 if (count > 0)
1593 {
1594 while (!--count)
1595 {
1596 if (!AvFILLp (self->incr_count))
1597 return 1; // done
1598
1599 SvREFCNT_dec_NN (av_pop (self->incr_count));
1600 count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1601 }
1602
1603 SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]) = count;
1604 }
1605 }
1606
1607 return 0;
1608}
1609
1610
1177///////////////////////////////////////////////////////////////////////////// 1611/////////////////////////////////////////////////////////////////////////////
1178// XS interface functions 1612// XS interface functions
1179 1613
1180MODULE = CBOR::XS PACKAGE = CBOR::XS 1614MODULE = CBOR::XS PACKAGE = CBOR::XS
1181 1615
… …
1193 1627
1194 default_filter = newSVpv ("CBOR::XS::default_filter", 0); 1628 default_filter = newSVpv ("CBOR::XS::default_filter", 0);
1195 1629
1196 sv_cbor = newSVpv ("CBOR", 0); 1630 sv_cbor = newSVpv ("CBOR", 0);
1197 SvREADONLY_on (sv_cbor); 1631 SvREADONLY_on (sv_cbor);
1632
1633 assert (("STRLEN must be an unsigned type", 0 <= (STRLEN)-1));
1198} 1634}
1199 1635
1200PROTOTYPES: DISABLE 1636PROTOTYPES: DISABLE
1201 1637
1202void CLONE (...) 1638void CLONE (...)
… …
1222 ALIAS: 1658 ALIAS:
1223 shrink = F_SHRINK 1659 shrink = F_SHRINK
1224 allow_unknown = F_ALLOW_UNKNOWN 1660 allow_unknown = F_ALLOW_UNKNOWN
1225 allow_sharing = F_ALLOW_SHARING 1661 allow_sharing = F_ALLOW_SHARING
1226 allow_cycles = F_ALLOW_CYCLES 1662 allow_cycles = F_ALLOW_CYCLES
1663 forbid_objects = F_FORBID_OBJECTS
1227 pack_strings = F_PACK_STRINGS 1664 pack_strings = F_PACK_STRINGS
1665 text_keys = F_TEXT_KEYS
1666 text_strings = F_TEXT_STRINGS
1667 validate_utf8 = F_VALIDATE_UTF8
1228 PPCODE: 1668 PPCODE:
1229{ 1669{
1230 if (enable) 1670 if (enable)
1231 self->flags |= ix; 1671 self->flags |= ix;
1232 else 1672 else
… …
1239 ALIAS: 1679 ALIAS:
1240 get_shrink = F_SHRINK 1680 get_shrink = F_SHRINK
1241 get_allow_unknown = F_ALLOW_UNKNOWN 1681 get_allow_unknown = F_ALLOW_UNKNOWN
1242 get_allow_sharing = F_ALLOW_SHARING 1682 get_allow_sharing = F_ALLOW_SHARING
1243 get_allow_cycles = F_ALLOW_CYCLES 1683 get_allow_cycles = F_ALLOW_CYCLES
1684 get_forbid_objects = F_FORBID_OBJECTS
1244 get_pack_strings = F_PACK_STRINGS 1685 get_pack_strings = F_PACK_STRINGS
1686 get_text_keys = F_TEXT_KEYS
1687 get_text_strings = F_TEXT_STRINGS
1688 get_validate_utf8 = F_VALIDATE_UTF8
1245 PPCODE: 1689 PPCODE:
1246 XPUSHs (boolSV (self->flags & ix)); 1690 XPUSHs (boolSV (self->flags & ix));
1247 1691
1248void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) 1692void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
1249 PPCODE: 1693 PPCODE:
… …
1298 EXTEND (SP, 2); 1742 EXTEND (SP, 2);
1299 PUSHs (sv); 1743 PUSHs (sv);
1300 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); 1744 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
1301} 1745}
1302 1746
1747void incr_parse (CBOR *self, SV *cborstr)
1748 ALIAS:
1749 incr_parse_multiple = 1
1750 PPCODE:
1751{
1752 if (SvUTF8 (cborstr))
1753 sv_utf8_downgrade (cborstr, 0);
1754
1755 if (!self->incr_count)
1756 {
1757 self->incr_count = newAV ();
1758 self->incr_pos = 0;
1759 self->incr_need = 1;
1760
1761 av_push (self->incr_count, newSViv (1));
1762 }
1763
1764 do
1765 {
1766 if (!incr_parse (self, cborstr))
1767 {
1768 if (self->incr_need > self->max_size && self->max_size)
1769 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
1770 (unsigned long)self->incr_need, (unsigned long)self->max_size);
1771
1772 break;
1773 }
1774
1775 SV *sv;
1776 char *offset;
1777
1778 PUTBACK; sv = decode_cbor (cborstr, self, &offset); SPAGAIN;
1779 XPUSHs (sv);
1780
1781 sv_chop (cborstr, offset);
1782
1783 av_clear (self->incr_count);
1784 av_push (self->incr_count, newSViv (1));
1785
1786 self->incr_pos = 0;
1787 self->incr_need = self->incr_pos + 1;
1788 }
1789 while (ix);
1790}
1791
1792void incr_reset (CBOR *self)
1793 CODE:
1794{
1795 SvREFCNT_dec (self->incr_count);
1796 self->incr_count = 0;
1797}
1798
1303void DESTROY (CBOR *self) 1799void DESTROY (CBOR *self)
1304 PPCODE: 1800 PPCODE:
1305 cbor_free (self); 1801 cbor_free (self);
1306 1802
1307PROTOTYPES: ENABLE 1803PROTOTYPES: ENABLE
… …
1326 cbor_init (&cbor); 1822 cbor_init (&cbor);
1327 PUTBACK; cborstr = decode_cbor (cborstr, &cbor, 0); SPAGAIN; 1823 PUTBACK; cborstr = decode_cbor (cborstr, &cbor, 0); SPAGAIN;
1328 XPUSHs (cborstr); 1824 XPUSHs (cborstr);
1329} 1825}
1330 1826
1827#ifdef __AFL_COMPILER
1828
1829void
1830afl_init ()
1831 CODE:
1832 __AFL_INIT ();
1833
1834int
1835afl_loop (unsigned int count = 10000)
1836 CODE:
1837 RETVAL = __AFL_LOOP (count);
1838 OUTPUT:
1839 RETVAL
1840
1841#endif
1842

Diff Legend

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