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

Comparing CBOR-XS/XS.xs (file contents):
Revision 1.36 by root, Sat Nov 30 17:37:45 2013 UTC vs.
Revision 1.76 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
124#define F_ALLOW_CYCLES 0x00000008UL
125#define F_FORBID_OBJECTS 0x00000010UL
102#define F_PACK_STRINGS 0x00000008UL 126#define F_PACK_STRINGS 0x00000020UL
127#define F_TEXT_KEYS 0x00000040UL
128#define F_TEXT_STRINGS 0x00000080UL
129#define F_VALIDATE_UTF8 0x00000100UL
103 130
104#define INIT_SIZE 32 // initial scalar size to be allocated 131#define INIT_SIZE 32 // initial scalar size to be allocated
105 132
106#define SB do { 133#define SB do {
107#define SE } while (0) 134#define SE } while (0)
… …
126typedef struct { 153typedef struct {
127 U32 flags; 154 U32 flags;
128 U32 max_depth; 155 U32 max_depth;
129 STRLEN max_size; 156 STRLEN max_size;
130 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.
131} CBOR; 163} CBOR;
132 164
133ecb_inline void 165ecb_inline void
134cbor_init (CBOR *cbor) 166cbor_init (CBOR *cbor)
135{ 167{
… …
139 171
140ecb_inline void 172ecb_inline void
141cbor_free (CBOR *cbor) 173cbor_free (CBOR *cbor)
142{ 174{
143 SvREFCNT_dec (cbor->filter); 175 SvREFCNT_dec (cbor->filter);
176 SvREFCNT_dec (cbor->incr_count);
144} 177}
145 178
146///////////////////////////////////////////////////////////////////////////// 179/////////////////////////////////////////////////////////////////////////////
147// utility functions 180// utility functions
148 181
… …
171#endif 204#endif
172 } 205 }
173} 206}
174 207
175// minimum length of a string to be registered for stringref 208// minimum length of a string to be registered for stringref
176ecb_inline int 209ecb_inline STRLEN
177minimum_string_length (UV idx) 210minimum_string_length (UV idx)
178{ 211{
179 return idx > 23 212 return idx <= 23 ? 3
180 ? idx > 0xffU 213 : idx <= 0xffU ? 4
181 ? idx > 0xffffU 214 : idx <= 0xffffU ? 5
182 ? idx > 0xffffffffU 215 : idx <= 0xffffffffU ? 7
183 ? 11 216 : 11;
184 : 7
185 : 5
186 : 4
187 : 3;
188} 217}
189 218
190///////////////////////////////////////////////////////////////////////////// 219/////////////////////////////////////////////////////////////////////////////
191// encoder 220// encoder
192 221
… …
205} enc_t; 234} enc_t;
206 235
207ecb_inline void 236ecb_inline void
208need (enc_t *enc, STRLEN len) 237need (enc_t *enc, STRLEN len)
209{ 238{
210 if (ecb_expect_false (enc->cur + len >= enc->end)) 239 if (ecb_expect_false ((uintptr_t)(enc->end - enc->cur) < len))
211 { 240 {
212 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv); 241 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
213 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1); 242 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
214 enc->cur = SvPVX (enc->sv) + cur; 243 enc->cur = SvPVX (enc->sv) + cur;
215 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; 244 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
216 } 245 }
217} 246}
218 247
248static void encode_sv (enc_t *enc, SV *sv);
249
219ecb_inline void 250ecb_inline void
220encode_ch (enc_t *enc, char ch) 251encode_ch (enc_t *enc, char ch)
221{ 252{
222 need (enc, 1); 253 need (enc, 1);
223 *enc->cur++ = ch; 254 *enc->cur++ = ch;
224} 255}
225 256
257// used for tags, intregers, element counts and so on
226static void 258static void
227encode_uint (enc_t *enc, int major, UV len) 259encode_uint (enc_t *enc, int major, UV len)
228{ 260{
229 need (enc, 9); 261 need (enc, 9);
230 262
… …
261 *enc->cur++ = len >> 8; 293 *enc->cur++ = len >> 8;
262 *enc->cur++ = len; 294 *enc->cur++ = len;
263 } 295 }
264} 296}
265 297
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
266ecb_inline void 310ecb_inline void
267encode_tag (enc_t *enc, UV tag) 311encode_tag (enc_t *enc, UV tag)
268{ 312{
269 encode_uint (enc, MAJOR_TAG, tag); 313 encode_uint (enc, MAJOR_TAG, tag);
270} 314}
271 315
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
272ecb_inline void 339ecb_inline void
273encode_str (enc_t *enc, int utf8, char *str, STRLEN len) 340encode_str (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len)
274{ 341{
342 if (ecb_expect_false (upgrade_utf8))
343 if (!utf8)
344 {
345 encode_str_utf8 (enc, utf8, str, len);
346 return;
347 }
348
275 encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len); 349 encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len);
276 need (enc, len); 350 need (enc, len);
277 memcpy (enc->cur, str, len); 351 memcpy (enc->cur, str, len);
278 enc->cur += len; 352 enc->cur += len;
279} 353}
280 354
281static void 355ecb_inline void
282encode_strref (enc_t *enc, int utf8, char *str, STRLEN len) 356encode_strref (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len)
283{ 357{
284 if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS)) 358 if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS))
285 { 359 {
286 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1); 360 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1);
287 361
… …
298 sv_setuv (*svp, enc->stringref_idx); 372 sv_setuv (*svp, enc->stringref_idx);
299 ++enc->stringref_idx; 373 ++enc->stringref_idx;
300 } 374 }
301 } 375 }
302 376
303 encode_str (enc, utf8, str, len); 377 encode_str (enc, upgrade_utf8, utf8, str, len);
304} 378}
305 379
306static 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}
307 510
308static void 511static void
309encode_av (enc_t *enc, AV *av) 512encode_av (enc_t *enc, AV *av)
310{ 513{
311 int i, len = av_len (av); 514 int i, len = av_len (av);
… …
315 518
316 ++enc->depth; 519 ++enc->depth;
317 520
318 encode_uint (enc, MAJOR_ARRAY, len + 1); 521 encode_uint (enc, MAJOR_ARRAY, len + 1);
319 522
523 if (ecb_expect_false (SvMAGICAL (av)))
320 for (i = 0; i <= len; ++i) 524 for (i = 0; i <= len; ++i)
321 { 525 {
322 SV **svp = av_fetch (av, i, 0); 526 SV **svp = av_fetch (av, i, 0);
323 encode_sv (enc, svp ? *svp : &PL_sv_undef); 527 encode_sv (enc, svp ? *svp : &PL_sv_undef);
324 } 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 }
325 535
326 --enc->depth; 536 --enc->depth;
327} 537}
328 538
329static void 539static void
… …
337 ++enc->depth; 547 ++enc->depth;
338 548
339 int pairs = hv_iterinit (hv); 549 int pairs = hv_iterinit (hv);
340 int mg = SvMAGICAL (hv); 550 int mg = SvMAGICAL (hv);
341 551
342 if (mg) 552 if (ecb_expect_false (mg))
343 encode_ch (enc, MAJOR_MAP | MINOR_INDEF); 553 encode_ch (enc, MAJOR_MAP | MINOR_INDEF);
344 else 554 else
345 encode_uint (enc, MAJOR_MAP, pairs); 555 encode_uint (enc, MAJOR_MAP, pairs);
346 556
347 while ((he = hv_iternext (hv))) 557 while ((he = hv_iternext (hv)))
348 { 558 {
349 if (HeKLEN (he) == HEf_SVKEY) 559 if (HeKLEN (he) == HEf_SVKEY)
350 encode_sv (enc, HeSVKEY (he)); 560 encode_sv (enc, HeSVKEY (he));
351 else 561 else
352 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));
353 563
354 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));
355 } 565 }
356 566
357 if (mg) 567 if (ecb_expect_false (mg))
358 encode_ch (enc, MAJOR_MISC | MINOR_INDEF); 568 encode_ch (enc, MAJOR_MISC | MINOR_INDEF);
359 569
360 --enc->depth; 570 --enc->depth;
361} 571}
362 572
… …
382 592
383 HV *stash = SvSTASH (sv); 593 HV *stash = SvSTASH (sv);
384 594
385 if (stash == boolean_stash) 595 if (stash == boolean_stash)
386 { 596 {
387 encode_ch (enc, SvIV (sv) ? MAJOR_MISC | SIMPLE_TRUE : MAJOR_MISC | SIMPLE_FALSE); 597 encode_bool (enc, SvIV (sv));
388 return; 598 return;
389 } 599 }
390 else if (stash == error_stash) 600 else if (stash == error_stash)
391 { 601 {
392 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF); 602 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
… …
395 else if (stash == tagged_stash) 605 else if (stash == tagged_stash)
396 { 606 {
397 if (svt != SVt_PVAV) 607 if (svt != SVt_PVAV)
398 croak ("encountered CBOR::XS::Tagged object that isn't an array"); 608 croak ("encountered CBOR::XS::Tagged object that isn't an array");
399 609
610 switch (av_len ((AV *)sv))
611 {
612 case 2-1:
613 // actually a tagged value
400 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)));
401 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 }
402 626
403 return; 627 return;
404 } 628 }
405 } 629 }
406 630
407 if (ecb_expect_false (SvREFCNT (sv) > 1) 631 if (ecb_expect_false (SvREFCNT (sv) > 1)
408 && ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING)) 632 && ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING))
409 { 633 {
410 if (!enc->shareable) 634 if (ecb_expect_false (!enc->shareable))
411 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ()); 635 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
412 636
413 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1); 637 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
414 638
415 if (SvOK (*svp)) 639 if (SvOK (*svp))
… …
429 if (ecb_expect_false (SvOBJECT (sv))) 653 if (ecb_expect_false (SvOBJECT (sv)))
430 { 654 {
431 HV *stash = SvSTASH (sv); 655 HV *stash = SvSTASH (sv);
432 GV *method; 656 GV *method;
433 657
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))));
434 if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0))) 661 else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
435 { 662 {
436 dSP; 663 dSP;
437 664
438 ENTER; SAVETMPS; PUSHMARK (SP); 665 ENTER; SAVETMPS;
666 PUSHMARK (SP);
439 // 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
440 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); 668 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
441 669
442 PUTBACK; 670 PUTBACK;
443 // G_SCALAR ensures that return value is 1 671 // G_SCALAR ensures that return value is 1
… …
456 } 684 }
457 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0) 685 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)
458 { 686 {
459 dSP; 687 dSP;
460 688
461 ENTER; SAVETMPS; PUSHMARK (SP); 689 ENTER; SAVETMPS;
690 PUSHMARK (SP);
462 EXTEND (SP, 2); 691 EXTEND (SP, 2);
463 // 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
464 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); 693 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
465 PUSHs (sv_cbor); 694 PUSHs (sv_cbor);
466 695
… …
472 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv) 701 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)
473 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash)); 702 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));
474 703
475 encode_tag (enc, CBOR_TAG_PERL_OBJECT); 704 encode_tag (enc, CBOR_TAG_PERL_OBJECT);
476 encode_uint (enc, MAJOR_ARRAY, count + 1); 705 encode_uint (enc, MAJOR_ARRAY, count + 1);
477 encode_strref (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash)); 706 encode_strref (enc, 0, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
478 707
479 while (count) 708 {
709 int i;
710
711 for (i = 0; i < count; ++i)
480 encode_sv (enc, SP[1 - count--]); 712 encode_sv (enc, SP[i + 1 - count]);
713
714 SP -= count;
715 }
481 716
482 PUTBACK; 717 PUTBACK;
483 718
484 FREETMPS; LEAVE; 719 FREETMPS; LEAVE;
485 } 720 }
… …
507 742
508 if (ecb_expect_false (nv == (NV)(U32)nv)) 743 if (ecb_expect_false (nv == (NV)(U32)nv))
509 encode_uint (enc, MAJOR_POS_INT, (U32)nv); 744 encode_uint (enc, MAJOR_POS_INT, (U32)nv);
510 //TODO: maybe I32? 745 //TODO: maybe I32?
511 else if (ecb_expect_false (nv == (float)nv)) 746 else if (ecb_expect_false (nv == (float)nv))
512 { 747 encode_float32 (enc, nv);
513 uint32_t fp = ecb_float_to_binary32 (nv);
514
515 *enc->cur++ = MAJOR_MISC | MISC_FLOAT32;
516
517 if (!ecb_big_endian ())
518 fp = ecb_bswap32 (fp);
519
520 memcpy (enc->cur, &fp, 4);
521 enc->cur += 4;
522 }
523 else 748 else
524 { 749 encode_float64 (enc, nv);
525 uint64_t fp = ecb_double_to_binary64 (nv);
526
527 *enc->cur++ = MAJOR_MISC | MISC_FLOAT64;
528
529 if (!ecb_big_endian ())
530 fp = ecb_bswap64 (fp);
531
532 memcpy (enc->cur, &fp, 8);
533 enc->cur += 8;
534 }
535} 750}
536 751
537static void 752static void
538encode_sv (enc_t *enc, SV *sv) 753encode_sv (enc_t *enc, SV *sv)
539{ 754{
… …
541 756
542 if (SvPOKp (sv)) 757 if (SvPOKp (sv))
543 { 758 {
544 STRLEN len; 759 STRLEN len;
545 char *str = SvPV (sv, len); 760 char *str = SvPV (sv, len);
546 encode_strref (enc, SvUTF8 (sv), str, len); 761 encode_strref (enc, enc->cbor.flags & F_TEXT_STRINGS, SvUTF8 (sv), str, len);
547 } 762 }
548 else if (SvNOKp (sv)) 763 else if (SvNOKp (sv))
549 encode_nv (enc, sv); 764 encode_nv (enc, sv);
550 else if (SvIOKp (sv)) 765 else if (SvIOKp (sv))
551 { 766 encode_int (enc, sv);
552 if (SvIsUV (sv))
553 encode_uint (enc, MAJOR_POS_INT, SvUVX (sv));
554 else if (SvIVX (sv) >= 0)
555 encode_uint (enc, MAJOR_POS_INT, SvIVX (sv));
556 else
557 encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1));
558 }
559 else if (SvROK (sv)) 767 else if (SvROK (sv))
560 encode_rv (enc, SvRV (sv)); 768 encode_rv (enc, SvRV (sv));
561 else if (!SvOK (sv)) 769 else if (!SvOK (sv))
562 encode_ch (enc, MAJOR_MISC | SIMPLE_NULL); 770 encode_ch (enc, MAJOR_MISC | SIMPLE_NULL);
563 else if (enc->cbor.flags & F_ALLOW_UNKNOWN) 771 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
… …
568} 776}
569 777
570static SV * 778static SV *
571encode_cbor (SV *scalar, CBOR *cbor) 779encode_cbor (SV *scalar, CBOR *cbor)
572{ 780{
573 enc_t enc = { }; 781 enc_t enc = { 0 };
574 782
575 enc.cbor = *cbor; 783 enc.cbor = *cbor;
576 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 784 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
577 enc.cur = SvPVX (enc.sv); 785 enc.cur = SvPVX (enc.sv);
578 enc.end = SvEND (enc.sv); 786 enc.end = SvEND (enc.sv);
579 787
580 SvPOK_only (enc.sv); 788 SvPOK_only (enc.sv);
581 789
582 if (cbor->flags & F_PACK_STRINGS) 790 if (cbor->flags & F_PACK_STRINGS)
583 { 791 {
… …
610 U32 depth; // recursion depth 818 U32 depth; // recursion depth
611 U32 maxdepth; // recursion depth limit 819 U32 maxdepth; // recursion depth limit
612 AV *shareable; 820 AV *shareable;
613 AV *stringref; 821 AV *stringref;
614 SV *decode_tagged; 822 SV *decode_tagged;
823 SV *err_sv; // optional sv for error, needs to be freed
615} dec_t; 824} dec_t;
616 825
617#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);
618 833
619#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;
620 837
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
621#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)
622#define DEC_DEC_DEPTH --dec->depth 863#define DEC_DEC_DEPTH --dec->depth
623 864
624static UV 865static UV
625decode_uint (dec_t *dec) 866decode_uint (dec_t *dec)
626{ 867{
… …
692 933
693 for (;;) 934 for (;;)
694 { 935 {
695 WANT (1); 936 WANT (1);
696 937
697 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF)) 938 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF) || dec->err)
698 { 939 {
699 ++dec->cur; 940 ++dec->cur;
700 break; 941 break;
701 } 942 }
702 943
703 av_push (av, decode_sv (dec)); 944 av_push (av, decode_sv (dec));
704 } 945 }
705 } 946 }
706 else 947 else
707 { 948 {
708 int i, len = decode_uint (dec); 949 UV i, len = decode_uint (dec);
709 950
710 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
711 av_fill (av, len - 1); 952 av_fill (av, len - 1);
712 953
713 for (i = 0; i < len; ++i) 954 for (i = 0; i < len; ++i)
… …
716 957
717 DEC_DEC_DEPTH; 958 DEC_DEC_DEPTH;
718 return newRV_noinc ((SV *)av); 959 return newRV_noinc ((SV *)av);
719 960
720fail: 961fail:
721 SvREFCNT_dec (av); 962 SvREFCNT_dec_NN (av);
722 DEC_DEC_DEPTH; 963 DEC_DEC_DEPTH;
723 return &PL_sv_undef; 964 return &PL_sv_undef;
724} 965}
725 966
726static void 967static void
… …
728{ 969{
729 // for speed reasons, we specialcase single-string 970 // for speed reasons, we specialcase single-string
730 // byte or utf-8 strings as keys, but only when !stringref 971 // byte or utf-8 strings as keys, but only when !stringref
731 972
732 if (ecb_expect_true (!dec->stringref)) 973 if (ecb_expect_true (!dec->stringref))
733 if (ecb_expect_true ((*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8)) 974 if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8))
734 { 975 {
735 I32 len = decode_uint (dec); 976 STRLEN len = decode_uint (dec);
736 char *key = (char *)dec->cur; 977 char *key = (char *)dec->cur;
737 978
979 WANT (len);
738 dec->cur += len; 980 dec->cur += len;
739 981
740 if (ecb_expect_false (dec->stringref))
741 av_push (dec->stringref, newSVpvn (key, len));
742
743 hv_store (hv, key, len, decode_sv (dec), 0); 982 hv_store (hv, key, len, decode_sv (dec), 0);
744 983
745 return; 984 return;
746 } 985 }
747 else if (ecb_expect_true ((*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8)) 986 else if (ecb_expect_true ((U8)(*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8))
748 { 987 {
749 I32 len = decode_uint (dec); 988 STRLEN len = decode_uint (dec);
750 char *key = (char *)dec->cur; 989 char *key = (char *)dec->cur;
751 990
991 WANT (len);
752 dec->cur += len; 992 dec->cur += len;
753 993
754 if (ecb_expect_false (dec->stringref)) 994 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
755 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)");
756 997
757 hv_store (hv, key, -len, decode_sv (dec), 0); 998 hv_store (hv, key, -len, decode_sv (dec), 0);
758 999
759 return; 1000 return;
760 } 1001 }
761 1002
762 SV *k = decode_sv (dec); 1003 SV *k = decode_sv (dec);
763 SV *v = decode_sv (dec); 1004 SV *v = decode_sv (dec);
764 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
765 hv_store_ent (hv, k, v, 0); 1034 hv_store_ent (hv, k, v, 0);
766 SvREFCNT_dec (k); 1035 SvREFCNT_dec_NN (k);
1036
1037fail:
1038 ;
767} 1039}
768 1040
769static SV * 1041static SV *
770decode_hv (dec_t *dec) 1042decode_hv (dec_t *dec)
771{ 1043{
… …
779 1051
780 for (;;) 1052 for (;;)
781 { 1053 {
782 WANT (1); 1054 WANT (1);
783 1055
784 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF)) 1056 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF) || dec->err)
785 { 1057 {
786 ++dec->cur; 1058 ++dec->cur;
787 break; 1059 break;
788 } 1060 }
789 1061
790 decode_he (dec, hv); 1062 decode_he (dec, hv);
791 } 1063 }
792 } 1064 }
793 else 1065 else
794 { 1066 {
795 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
796 1070
797 while (pairs--) 1071 while (pairs--)
798 decode_he (dec, hv); 1072 decode_he (dec, hv);
799 } 1073 }
800 1074
801 DEC_DEC_DEPTH; 1075 DEC_DEC_DEPTH;
802 return newRV_noinc ((SV *)hv); 1076 return newRV_noinc ((SV *)hv);
803 1077
804fail: 1078fail:
805 SvREFCNT_dec (hv); 1079 SvREFCNT_dec_NN (hv);
806 DEC_DEC_DEPTH; 1080 DEC_DEC_DEPTH;
807 return &PL_sv_undef; 1081 return &PL_sv_undef;
808} 1082}
809 1083
810static SV * 1084static SV *
811decode_str (dec_t *dec, int utf8) 1085decode_str (dec_t *dec, int utf8)
812{ 1086{
813 SV *sv = 0; 1087 SV *sv = 0;
814 1088
815 if ((*dec->cur & MINOR_MASK) == MINOR_INDEF) 1089 if (ecb_expect_false ((*dec->cur & MINOR_MASK) == MINOR_INDEF))
816 { 1090 {
817 // indefinite length strings 1091 // indefinite length strings
818 ++dec->cur; 1092 ++dec->cur;
819 1093
820 U8 major = *dec->cur & MAJOR_MISC; 1094 U8 major = *dec->cur & MAJOR_MISC;
… …
853 && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1)) 1127 && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1))
854 av_push (dec->stringref, SvREFCNT_inc_NN (sv)); 1128 av_push (dec->stringref, SvREFCNT_inc_NN (sv));
855 } 1129 }
856 1130
857 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
858 SvUTF8_on (sv); 1137 SvUTF8_on (sv);
1138 }
859 1139
860 return sv; 1140 return sv;
861 1141
862fail: 1142fail:
863 SvREFCNT_dec (sv); 1143 SvREFCNT_dec (sv);
… …
882 sv = newRV_noinc (decode_sv (dec)); 1162 sv = newRV_noinc (decode_sv (dec));
883 break; 1163 break;
884 1164
885 case CBOR_TAG_STRINGREF_NAMESPACE: 1165 case CBOR_TAG_STRINGREF_NAMESPACE:
886 { 1166 {
1167 // do not use SAVETMPS/FREETMPS, as these will
1168 // erase mortalised caches, e.g. "shareable"
887 ENTER; SAVETMPS; 1169 ENTER;
888 1170
889 SAVESPTR (dec->stringref); 1171 SAVESPTR (dec->stringref);
890 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ()); 1172 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ());
891 1173
892 sv = decode_sv (dec); 1174 sv = decode_sv (dec);
893 1175
894 FREETMPS; LEAVE; 1176 LEAVE;
895 } 1177 }
896 break; 1178 break;
897 1179
898 case CBOR_TAG_STRINGREF: 1180 case CBOR_TAG_STRINGREF:
899 { 1181 {
900 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT)) 1182 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
901 ERR ("corrupted CBOR data (stringref index not an unsigned integer)"); 1183 ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
902 1184
903 UV idx = decode_uint (dec); 1185 UV idx = decode_uint (dec);
904 1186
905 if (!dec->stringref || (int)idx > AvFILLp (dec->stringref)) 1187 if (!dec->stringref || idx >= (UV)(1 + AvFILLp (dec->stringref)))
906 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)");
907 1189
908 sv = newSVsv (AvARRAY (dec->stringref)[idx]); 1190 sv = newSVsv (AvARRAY (dec->stringref)[idx]);
909 } 1191 }
910 break; 1192 break;
… …
912 case CBOR_TAG_VALUE_SHAREABLE: 1194 case CBOR_TAG_VALUE_SHAREABLE:
913 { 1195 {
914 if (ecb_expect_false (!dec->shareable)) 1196 if (ecb_expect_false (!dec->shareable))
915 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ()); 1197 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ());
916 1198
1199 if (dec->cbor.flags & F_ALLOW_CYCLES)
1200 {
917 sv = newSV (0); 1201 sv = newSV (0);
918 av_push (dec->shareable, SvREFCNT_inc_NN (sv)); 1202 av_push (dec->shareable, SvREFCNT_inc_NN (sv));
919 1203
920 SV *osv = decode_sv (dec); 1204 SV *osv = decode_sv (dec);
921 sv_setsv (sv, osv); 1205 sv_setsv (sv, osv);
922 SvREFCNT_dec_NN (osv); 1206 SvREFCNT_dec_NN (osv);
1207 }
1208 else
1209 {
1210 av_push (dec->shareable, &PL_sv_undef);
1211 int idx = AvFILLp (dec->shareable);
1212 sv = decode_sv (dec);
1213 av_store (dec->shareable, idx, SvREFCNT_inc_NN (sv));
1214 }
923 } 1215 }
924 break; 1216 break;
925 1217
926 case CBOR_TAG_VALUE_SHAREDREF: 1218 case CBOR_TAG_VALUE_SHAREDREF:
927 { 1219 {
928 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT)) 1220 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
929 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)"); 1221 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
930 1222
931 UV idx = decode_uint (dec); 1223 UV idx = decode_uint (dec);
932 1224
933 if (!dec->shareable || (int)idx > AvFILLp (dec->shareable)) 1225 if (!dec->shareable || idx >= (UV)(1 + AvFILLp (dec->shareable)))
934 ERR ("corrupted CBOR data (sharedref index out of bounds)"); 1226 ERR ("corrupted CBOR data (sharedref index out of bounds)");
935 1227
936 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]); 1228 sv = newSVsv (AvARRAY (dec->shareable)[idx]);
1229
1230 if (sv == &PL_sv_undef)
1231 ERR ("cyclic CBOR data structure found, but allow_cycles is not enabled");
937 } 1232 }
938 break; 1233 break;
939 1234
940 case CBOR_TAG_PERL_OBJECT: 1235 case CBOR_TAG_PERL_OBJECT:
941 { 1236 {
1237 if (dec->cbor.flags & F_FORBID_OBJECTS)
1238 goto filter;
1239
942 sv = decode_sv (dec); 1240 sv = decode_sv (dec);
943 1241
944 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV) 1242 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
945 ERR ("corrupted CBOR data (non-array perl object)"); 1243 ERR ("corrupted CBOR data (non-array perl object)");
946 1244
… …
956 if (!method) 1254 if (!method)
957 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)");
958 1256
959 dSP; 1257 dSP;
960 1258
961 ENTER; SAVETMPS; PUSHMARK (SP); 1259 ENTER; SAVETMPS;
1260 PUSHMARK (SP);
962 EXTEND (SP, len + 1); 1261 EXTEND (SP, len + 1);
963 // 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
964 PUSHs (*av_fetch (av, 0, 1)); 1263 PUSHs (*av_fetch (av, 0, 1));
965 PUSHs (sv_cbor); 1264 PUSHs (sv_cbor);
966 1265
… …
974 SPAGAIN; 1273 SPAGAIN;
975 1274
976 if (SvTRUE (ERRSV)) 1275 if (SvTRUE (ERRSV))
977 { 1276 {
978 FREETMPS; LEAVE; 1277 FREETMPS; LEAVE;
979 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV)))); 1278 ERR_ERRSV;
980 } 1279 }
981 1280
982 SvREFCNT_dec (sv); 1281 SvREFCNT_dec_NN (sv);
983 sv = SvREFCNT_inc (POPs); 1282 sv = SvREFCNT_inc (POPs);
984 1283
985 PUTBACK; 1284 PUTBACK;
986 1285
987 FREETMPS; LEAVE; 1286 FREETMPS; LEAVE;
988 } 1287 }
989 break; 1288 break;
990 1289
991 default: 1290 default:
1291 filter:
992 { 1292 {
1293 SV *tag_sv = newSVuv (tag);
1294
993 sv = decode_sv (dec); 1295 sv = decode_sv (dec);
994 1296
995 dSP; 1297 dSP;
996 ENTER; SAVETMPS; PUSHMARK (SP); 1298 ENTER; SAVETMPS;
1299 PUSHMARK (SP);
997 EXTEND (SP, 2); 1300 EXTEND (SP, 2);
998 PUSHs (newSVuv (tag)); 1301 PUSHs (tag_sv);
999 PUSHs (sv); 1302 PUSHs (sv);
1000 1303
1001 PUTBACK; 1304 PUTBACK;
1002 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);
1003 SPAGAIN; 1306 SPAGAIN;
1004 1307
1005 if (SvTRUE (ERRSV)) 1308 if (SvTRUE (ERRSV))
1006 { 1309 {
1310 SvREFCNT_dec_NN (tag_sv);
1007 FREETMPS; LEAVE; 1311 FREETMPS; LEAVE;
1008 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV)))); 1312 ERR_ERRSV;
1009 } 1313 }
1010 1314
1011 if (count) 1315 if (count)
1012 { 1316 {
1317 SvREFCNT_dec_NN (tag_sv);
1013 SvREFCNT_dec (sv); 1318 SvREFCNT_dec_NN (sv);
1014 sv = SvREFCNT_inc (POPs); 1319 sv = SvREFCNT_inc_NN (TOPs);
1320 SP -= count;
1015 } 1321 }
1016 else 1322 else
1017 { 1323 {
1018 AV *av = newAV (); 1324 AV *av = newAV ();
1019 av_push (av, newSVuv (tag)); 1325 av_push (av, tag_sv);
1020 av_push (av, sv); 1326 av_push (av, sv);
1021 1327
1022 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 1328 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
1023 ? cbor_tagged_stash 1329 ? cbor_tagged_stash
1024 : gv_stashpv ("CBOR::XS::Tagged" , 1); 1330 : gv_stashpv ("CBOR::XS::Tagged" , 1);
… …
1110 1416
1111 return newSVnv (ecb_binary64_to_double (fp)); 1417 return newSVnv (ecb_binary64_to_double (fp));
1112 } 1418 }
1113 1419
1114 // 0..19 unassigned simple 1420 // 0..19 unassigned simple
1115 // 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
1116 default: 1424 default:
1117 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)"); 1425 ERR ("corrupted CBOR data (reserved/unassigned/unexpected major 7 value)");
1118 } 1426 }
1119 1427
1120 break; 1428 break;
1121 } 1429 }
1122 1430
… …
1125} 1433}
1126 1434
1127static SV * 1435static SV *
1128decode_cbor (SV *string, CBOR *cbor, char **offset_return) 1436decode_cbor (SV *string, CBOR *cbor, char **offset_return)
1129{ 1437{
1130 dec_t dec = { }; 1438 dec_t dec = { 0 };
1131 SV *sv; 1439 SV *sv;
1132 STRLEN len; 1440 STRLEN len;
1133 char *data = SvPVbyte (string, len); 1441 char *data = SvPVbyte (string, len);
1134 1442
1135 if (len > cbor->max_size && cbor->max_size) 1443 if (len > cbor->max_size && cbor->max_size)
… …
1149 if (dec.cur != dec.end && !dec.err) 1457 if (dec.cur != dec.end && !dec.err)
1150 dec.err = "garbage after CBOR object"; 1458 dec.err = "garbage after CBOR object";
1151 1459
1152 if (dec.err) 1460 if (dec.err)
1153 { 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
1154 SvREFCNT_dec (sv); 1473 SvREFCNT_dec_NN (sv);
1474
1475 if (dec.err_sv)
1476 sv_2mortal (dec.err_sv);
1477
1155 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);
1156 } 1479 }
1157 1480
1158 sv = sv_2mortal (sv); 1481 sv = sv_2mortal (sv);
1159 1482
1160 return sv; 1483 return sv;
1161} 1484}
1162 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
1163///////////////////////////////////////////////////////////////////////////// 1611/////////////////////////////////////////////////////////////////////////////
1164// XS interface functions 1612// XS interface functions
1165 1613
1166MODULE = CBOR::XS PACKAGE = CBOR::XS 1614MODULE = CBOR::XS PACKAGE = CBOR::XS
1167 1615
… …
1179 1627
1180 default_filter = newSVpv ("CBOR::XS::default_filter", 0); 1628 default_filter = newSVpv ("CBOR::XS::default_filter", 0);
1181 1629
1182 sv_cbor = newSVpv ("CBOR", 0); 1630 sv_cbor = newSVpv ("CBOR", 0);
1183 SvREADONLY_on (sv_cbor); 1631 SvREADONLY_on (sv_cbor);
1632
1633 assert (("STRLEN must be an unsigned type", 0 <= (STRLEN)-1));
1184} 1634}
1185 1635
1186PROTOTYPES: DISABLE 1636PROTOTYPES: DISABLE
1187 1637
1188void CLONE (...) 1638void CLONE (...)
… …
1207void shrink (CBOR *self, int enable = 1) 1657void shrink (CBOR *self, int enable = 1)
1208 ALIAS: 1658 ALIAS:
1209 shrink = F_SHRINK 1659 shrink = F_SHRINK
1210 allow_unknown = F_ALLOW_UNKNOWN 1660 allow_unknown = F_ALLOW_UNKNOWN
1211 allow_sharing = F_ALLOW_SHARING 1661 allow_sharing = F_ALLOW_SHARING
1662 allow_cycles = F_ALLOW_CYCLES
1663 forbid_objects = F_FORBID_OBJECTS
1212 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
1213 PPCODE: 1668 PPCODE:
1214{ 1669{
1215 if (enable) 1670 if (enable)
1216 self->flags |= ix; 1671 self->flags |= ix;
1217 else 1672 else
… …
1223void get_shrink (CBOR *self) 1678void get_shrink (CBOR *self)
1224 ALIAS: 1679 ALIAS:
1225 get_shrink = F_SHRINK 1680 get_shrink = F_SHRINK
1226 get_allow_unknown = F_ALLOW_UNKNOWN 1681 get_allow_unknown = F_ALLOW_UNKNOWN
1227 get_allow_sharing = F_ALLOW_SHARING 1682 get_allow_sharing = F_ALLOW_SHARING
1683 get_allow_cycles = F_ALLOW_CYCLES
1684 get_forbid_objects = F_FORBID_OBJECTS
1228 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
1229 PPCODE: 1689 PPCODE:
1230 XPUSHs (boolSV (self->flags & ix)); 1690 XPUSHs (boolSV (self->flags & ix));
1231 1691
1232void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) 1692void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
1233 PPCODE: 1693 PPCODE:
… …
1282 EXTEND (SP, 2); 1742 EXTEND (SP, 2);
1283 PUSHs (sv); 1743 PUSHs (sv);
1284 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); 1744 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
1285} 1745}
1286 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
1287void DESTROY (CBOR *self) 1799void DESTROY (CBOR *self)
1288 PPCODE: 1800 PPCODE:
1289 cbor_free (self); 1801 cbor_free (self);
1290 1802
1291PROTOTYPES: ENABLE 1803PROTOTYPES: ENABLE
… …
1310 cbor_init (&cbor); 1822 cbor_init (&cbor);
1311 PUTBACK; cborstr = decode_cbor (cborstr, &cbor, 0); SPAGAIN; 1823 PUTBACK; cborstr = decode_cbor (cborstr, &cbor, 0); SPAGAIN;
1312 XPUSHs (cborstr); 1824 XPUSHs (cborstr);
1313} 1825}
1314 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