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 1.35 by root, Sat Nov 30 17:19:34 2013 UTC vs.
Revision 1.78 by root, Fri Sep 8 06:01:12 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
231 if (ecb_expect_true (len < LENGTH_EXT1)) 263 if (ecb_expect_true (len < LENGTH_EXT1))
232 *enc->cur++ = major | len; 264 *enc->cur++ = major | len;
233 else if (ecb_expect_true (len <= 0xff)) 265 else if (ecb_expect_true (len <= 0xffU))
234 { 266 {
235 *enc->cur++ = major | LENGTH_EXT1; 267 *enc->cur++ = major | LENGTH_EXT1;
236 *enc->cur++ = len; 268 *enc->cur++ = len;
237 } 269 }
238 else if (len <= 0xffff) 270 else if (len <= 0xffffU)
239 { 271 {
240 *enc->cur++ = major | LENGTH_EXT2; 272 *enc->cur++ = major | LENGTH_EXT2;
241 *enc->cur++ = len >> 8; 273 *enc->cur++ = len >> 8;
242 *enc->cur++ = len; 274 *enc->cur++ = len;
243 } 275 }
244 else if (len <= 0xffffffff) 276 else if (len <= 0xffffffffU)
245 { 277 {
246 *enc->cur++ = major | LENGTH_EXT4; 278 *enc->cur++ = major | LENGTH_EXT4;
247 *enc->cur++ = len >> 24; 279 *enc->cur++ = len >> 24;
248 *enc->cur++ = len >> 16; 280 *enc->cur++ = len >> 16;
249 *enc->cur++ = len >> 8; 281 *enc->cur++ = len >> 8;
… …
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{
627 U8 m = *dec->cur & MINOR_MASK; 868 U8 m = *dec->cur & MINOR_MASK;
628 ++dec->cur; 869 ++dec->cur;
629 870
630 if (ecb_expect_true (m < LENGTH_EXT1)) 871 if (ecb_expect_true (m < LENGTH_EXT1))
631 return m; 872 return m;
632 873 else if (ecb_expect_true (m == LENGTH_EXT1))
633 switch (m)
634 { 874 {
635 case LENGTH_EXT1:
636 WANT (1); 875 WANT (1);
637 dec->cur += 1; 876 dec->cur += 1;
638 return dec->cur[-1]; 877 return dec->cur[-1];
639 878 }
640 case LENGTH_EXT2: 879 else if (ecb_expect_true (m == LENGTH_EXT2))
880 {
641 WANT (2); 881 WANT (2);
642 dec->cur += 2; 882 dec->cur += 2;
643 return (((UV)dec->cur[-2]) << 8) 883 return (((UV)dec->cur[-2]) << 8)
644 | ((UV)dec->cur[-1]); 884 | ((UV)dec->cur[-1]);
645 885 }
646 case LENGTH_EXT4: 886 else if (ecb_expect_true (m == LENGTH_EXT4))
887 {
647 WANT (4); 888 WANT (4);
648 dec->cur += 4; 889 dec->cur += 4;
649 return (((UV)dec->cur[-4]) << 24) 890 return (((UV)dec->cur[-4]) << 24)
650 | (((UV)dec->cur[-3]) << 16) 891 | (((UV)dec->cur[-3]) << 16)
651 | (((UV)dec->cur[-2]) << 8) 892 | (((UV)dec->cur[-2]) << 8)
652 | ((UV)dec->cur[-1]); 893 | ((UV)dec->cur[-1]);
653 894 }
654 case LENGTH_EXT8: 895 else if (ecb_expect_true (m == LENGTH_EXT8))
896 {
655 WANT (8); 897 WANT (8);
656 dec->cur += 8; 898 dec->cur += 8;
657 899
658 return 900 return
659#if UVSIZE < 8 901#if UVSIZE < 8
660 0 902 0
661#else 903#else
662 (((UV)dec->cur[-8]) << 56) 904 (((UV)dec->cur[-8]) << 56)
663 | (((UV)dec->cur[-7]) << 48) 905 | (((UV)dec->cur[-7]) << 48)
664 | (((UV)dec->cur[-6]) << 40) 906 | (((UV)dec->cur[-6]) << 40)
665 | (((UV)dec->cur[-5]) << 32) 907 | (((UV)dec->cur[-5]) << 32)
666#endif 908#endif
667 | (((UV)dec->cur[-4]) << 24) 909 | (((UV)dec->cur[-4]) << 24)
668 | (((UV)dec->cur[-3]) << 16) 910 | (((UV)dec->cur[-3]) << 16)
669 | (((UV)dec->cur[-2]) << 8) 911 | (((UV)dec->cur[-2]) << 8)
670 | ((UV)dec->cur[-1]); 912 | ((UV)dec->cur[-1]);
671 913 }
672 default: 914 else
673 ERR ("corrupted CBOR data (unsupported integer minor encoding)"); 915 ERR ("corrupted CBOR data (unsupported integer minor encoding)");
674 }
675 916
676fail: 917fail:
677 return 0; 918 return 0;
678} 919}
679 920
… …
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
951 WANT (len); // complexity check for av_fill - need at least one byte per value, do not allow supersize arrays
710 av_fill (av, len - 1); 952 av_fill (av, len - 1);
711 953
712 for (i = 0; i < len; ++i) 954 for (i = 0; i < len; ++i)
713 AvARRAY (av)[i] = decode_sv (dec); 955 AvARRAY (av)[i] = decode_sv (dec);
714 } 956 }
715 957
716 DEC_DEC_DEPTH; 958 DEC_DEC_DEPTH;
717 return newRV_noinc ((SV *)av); 959 return newRV_noinc ((SV *)av);
718 960
719fail: 961fail:
720 SvREFCNT_dec (av); 962 SvREFCNT_dec_NN (av);
721 DEC_DEC_DEPTH; 963 DEC_DEC_DEPTH;
722 return &PL_sv_undef; 964 return &PL_sv_undef;
723} 965}
724 966
725static void 967static void
… …
727{ 969{
728 // for speed reasons, we specialcase single-string 970 // for speed reasons, we specialcase single-string
729 // byte or utf-8 strings as keys, but only when !stringref 971 // byte or utf-8 strings as keys, but only when !stringref
730 972
731 if (ecb_expect_true (!dec->stringref)) 973 if (ecb_expect_true (!dec->stringref))
732 if ((*dec->cur - MAJOR_BYTES) <= 27) 974 if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8))
733 { 975 {
734 I32 len = decode_uint (dec); 976 STRLEN len = decode_uint (dec);
735 char *key = (char *)dec->cur; 977 char *key = (char *)dec->cur;
736 978
979 WANT (len);
737 dec->cur += len; 980 dec->cur += len;
738 981
739 if (ecb_expect_false (dec->stringref))
740 av_push (dec->stringref, newSVpvn (key, len));
741
742 hv_store (hv, key, len, decode_sv (dec), 0); 982 hv_store (hv, key, len, decode_sv (dec), 0);
743 983
744 return; 984 return;
745 } 985 }
746 else if ((*dec->cur - MAJOR_TEXT) <= 27) 986 else if (ecb_expect_true ((U8)(*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8))
747 { 987 {
748 I32 len = decode_uint (dec); 988 STRLEN len = decode_uint (dec);
749 char *key = (char *)dec->cur; 989 char *key = (char *)dec->cur;
750 990
991 WANT (len);
751 dec->cur += len; 992 dec->cur += len;
752 993
753 if (ecb_expect_false (dec->stringref)) 994 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
754 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)");
755 997
756 hv_store (hv, key, -len, decode_sv (dec), 0); 998 hv_store (hv, key, -len, decode_sv (dec), 0);
757 999
758 return; 1000 return;
759 } 1001 }
760 1002
761 SV *k = decode_sv (dec); 1003 SV *k = decode_sv (dec);
762 SV *v = decode_sv (dec); 1004 SV *v = decode_sv (dec);
763 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
764 hv_store_ent (hv, k, v, 0); 1034 hv_store_ent (hv, k, v, 0);
765 SvREFCNT_dec (k); 1035 SvREFCNT_dec_NN (k);
1036
1037fail:
1038 ;
766} 1039}
767 1040
768static SV * 1041static SV *
769decode_hv (dec_t *dec) 1042decode_hv (dec_t *dec)
770{ 1043{
… …
778 1051
779 for (;;) 1052 for (;;)
780 { 1053 {
781 WANT (1); 1054 WANT (1);
782 1055
783 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF)) 1056 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF) || dec->err)
784 { 1057 {
785 ++dec->cur; 1058 ++dec->cur;
786 break; 1059 break;
787 } 1060 }
788 1061
789 decode_he (dec, hv); 1062 decode_he (dec, hv);
790 } 1063 }
791 } 1064 }
792 else 1065 else
793 { 1066 {
794 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
795 1070
796 while (pairs--) 1071 while (pairs--)
797 decode_he (dec, hv); 1072 decode_he (dec, hv);
798 } 1073 }
799 1074
800 DEC_DEC_DEPTH; 1075 DEC_DEC_DEPTH;
801 return newRV_noinc ((SV *)hv); 1076 return newRV_noinc ((SV *)hv);
802 1077
803fail: 1078fail:
804 SvREFCNT_dec (hv); 1079 SvREFCNT_dec_NN (hv);
805 DEC_DEC_DEPTH; 1080 DEC_DEC_DEPTH;
806 return &PL_sv_undef; 1081 return &PL_sv_undef;
807} 1082}
808 1083
809static SV * 1084static SV *
810decode_str (dec_t *dec, int utf8) 1085decode_str (dec_t *dec, int utf8)
811{ 1086{
812 SV *sv = 0; 1087 SV *sv = 0;
813 1088
814 if ((*dec->cur & MINOR_MASK) == MINOR_INDEF) 1089 if (ecb_expect_false ((*dec->cur & MINOR_MASK) == MINOR_INDEF))
815 { 1090 {
816 // indefinite length strings 1091 // indefinite length strings
817 ++dec->cur; 1092 ++dec->cur;
818 1093
819 U8 major = *dec->cur & MAJOR_MISC; 1094 U8 major = *dec->cur & MAJOR_MISC;
… …
852 && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1)) 1127 && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1))
853 av_push (dec->stringref, SvREFCNT_inc_NN (sv)); 1128 av_push (dec->stringref, SvREFCNT_inc_NN (sv));
854 } 1129 }
855 1130
856 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
857 SvUTF8_on (sv); 1137 SvUTF8_on (sv);
1138 }
858 1139
859 return sv; 1140 return sv;
860 1141
861fail: 1142fail:
862 SvREFCNT_dec (sv); 1143 SvREFCNT_dec (sv);
… …
881 sv = newRV_noinc (decode_sv (dec)); 1162 sv = newRV_noinc (decode_sv (dec));
882 break; 1163 break;
883 1164
884 case CBOR_TAG_STRINGREF_NAMESPACE: 1165 case CBOR_TAG_STRINGREF_NAMESPACE:
885 { 1166 {
1167 // do not use SAVETMPS/FREETMPS, as these will
1168 // erase mortalised caches, e.g. "shareable"
886 ENTER; SAVETMPS; 1169 ENTER;
887 1170
888 SAVESPTR (dec->stringref); 1171 SAVESPTR (dec->stringref);
889 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ()); 1172 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ());
890 1173
891 sv = decode_sv (dec); 1174 sv = decode_sv (dec);
892 1175
893 FREETMPS; LEAVE; 1176 LEAVE;
894 } 1177 }
895 break; 1178 break;
896 1179
897 case CBOR_TAG_STRINGREF: 1180 case CBOR_TAG_STRINGREF:
898 { 1181 {
899 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT)) 1182 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
900 ERR ("corrupted CBOR data (stringref index not an unsigned integer)"); 1183 ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
901 1184
902 UV idx = decode_uint (dec); 1185 UV idx = decode_uint (dec);
903 1186
904 if (!dec->stringref || (int)idx > AvFILLp (dec->stringref)) 1187 if (!dec->stringref || idx >= (UV)(1 + AvFILLp (dec->stringref)))
905 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)");
906 1189
907 sv = newSVsv (AvARRAY (dec->stringref)[idx]); 1190 sv = newSVsv (AvARRAY (dec->stringref)[idx]);
908 } 1191 }
909 break; 1192 break;
… …
911 case CBOR_TAG_VALUE_SHAREABLE: 1194 case CBOR_TAG_VALUE_SHAREABLE:
912 { 1195 {
913 if (ecb_expect_false (!dec->shareable)) 1196 if (ecb_expect_false (!dec->shareable))
914 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ()); 1197 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ());
915 1198
1199 if (dec->cbor.flags & F_ALLOW_CYCLES)
1200 {
1201 // if cycles are allowed, then we store an AV as value
1202 // while it is being decoded, and gather unresolved
1203 // references in it, to be re4solved after decoding.
1204 int idx, i;
916 sv = newSV (0); 1205 AV *av = newAV ();
917 av_push (dec->shareable, SvREFCNT_inc_NN (sv)); 1206 av_push (dec->shareable, (SV *)av);
1207 idx = AvFILLp (dec->shareable);
918 1208
919 SV *osv = decode_sv (dec); 1209 sv = decode_sv (dec);
920 sv_setsv (sv, osv); 1210
1211 // the AV now contains \undef for all unresolved references,
1212 // so we fix them up here.
1213 for (i = 0; i <= AvFILLp (av); ++i)
1214 SvRV_set (AvARRAY (av)[i], SvREFCNT_inc_NN (SvRV (sv)));
1215
1216 // now replace the AV by a reference to the completed value
921 SvREFCNT_dec_NN (osv); 1217 SvREFCNT_dec_NN ((SV *)av);
1218 AvARRAY (dec->shareable)[idx] = SvREFCNT_inc_NN (sv);
1219 }
1220 else
1221 {
1222 av_push (dec->shareable, &PL_sv_undef);
1223 int idx = AvFILLp (dec->shareable);
1224 sv = decode_sv (dec);
1225 AvARRAY (dec->shareable)[idx] = SvREFCNT_inc_NN (sv);
1226 }
922 } 1227 }
923 break; 1228 break;
924 1229
925 case CBOR_TAG_VALUE_SHAREDREF: 1230 case CBOR_TAG_VALUE_SHAREDREF:
926 { 1231 {
927 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT)) 1232 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
928 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)"); 1233 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
929 1234
930 UV idx = decode_uint (dec); 1235 UV idx = decode_uint (dec);
931 1236
932 if (!dec->shareable || (int)idx > AvFILLp (dec->shareable)) 1237 if (!dec->shareable || idx >= (UV)(1 + AvFILLp (dec->shareable)))
933 ERR ("corrupted CBOR data (sharedref index out of bounds)"); 1238 ERR ("corrupted CBOR data (sharedref index out of bounds)");
934 1239
935 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]); 1240 sv = AvARRAY (dec->shareable)[idx];
1241
1242 // reference to cycle, we create a new \undef and use that, and also
1243 // registerr it in the AV for later fixing
1244 if (SvTYPE (sv) == SVt_PVAV)
1245 {
1246 AV *av = (AV *)sv;
1247 sv = newRV_noinc (&PL_sv_undef);
1248 av_push (av, sv);
1249 }
1250 else if (sv == &PL_sv_undef) // not yet decoded, but cycles not allowed
1251 ERR ("cyclic CBOR data structure found, but allow_cycles is not enabled");
1252 else // we decoded the object earlier, no cycle
1253 sv = newSVsv (sv);
936 } 1254 }
937 break; 1255 break;
938 1256
939 case CBOR_TAG_PERL_OBJECT: 1257 case CBOR_TAG_PERL_OBJECT:
940 { 1258 {
1259 if (dec->cbor.flags & F_FORBID_OBJECTS)
1260 goto filter;
1261
941 sv = decode_sv (dec); 1262 sv = decode_sv (dec);
942 1263
943 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV) 1264 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
944 ERR ("corrupted CBOR data (non-array perl object)"); 1265 ERR ("corrupted CBOR data (non-array perl object)");
945 1266
… …
955 if (!method) 1276 if (!method)
956 ERR ("cannot decode perl-object (package does not have a THAW method)"); 1277 ERR ("cannot decode perl-object (package does not have a THAW method)");
957 1278
958 dSP; 1279 dSP;
959 1280
960 ENTER; SAVETMPS; PUSHMARK (SP); 1281 ENTER; SAVETMPS;
1282 PUSHMARK (SP);
961 EXTEND (SP, len + 1); 1283 EXTEND (SP, len + 1);
962 // we re-bless the reference to get overload and other niceties right 1284 // we re-bless the reference to get overload and other niceties right
963 PUSHs (*av_fetch (av, 0, 1)); 1285 PUSHs (*av_fetch (av, 0, 1));
964 PUSHs (sv_cbor); 1286 PUSHs (sv_cbor);
965 1287
… …
973 SPAGAIN; 1295 SPAGAIN;
974 1296
975 if (SvTRUE (ERRSV)) 1297 if (SvTRUE (ERRSV))
976 { 1298 {
977 FREETMPS; LEAVE; 1299 FREETMPS; LEAVE;
978 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV)))); 1300 ERR_ERRSV;
979 } 1301 }
980 1302
981 SvREFCNT_dec (sv); 1303 SvREFCNT_dec_NN (sv);
982 sv = SvREFCNT_inc (POPs); 1304 sv = SvREFCNT_inc (POPs);
983 1305
984 PUTBACK; 1306 PUTBACK;
985 1307
986 FREETMPS; LEAVE; 1308 FREETMPS; LEAVE;
987 } 1309 }
988 break; 1310 break;
989 1311
990 default: 1312 default:
1313 filter:
991 { 1314 {
1315 SV *tag_sv = newSVuv (tag);
1316
992 sv = decode_sv (dec); 1317 sv = decode_sv (dec);
993 1318
994 dSP; 1319 dSP;
995 ENTER; SAVETMPS; PUSHMARK (SP); 1320 ENTER; SAVETMPS;
1321 PUSHMARK (SP);
996 EXTEND (SP, 2); 1322 EXTEND (SP, 2);
997 PUSHs (newSVuv (tag)); 1323 PUSHs (tag_sv);
998 PUSHs (sv); 1324 PUSHs (sv);
999 1325
1000 PUTBACK; 1326 PUTBACK;
1001 int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL); 1327 int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL);
1002 SPAGAIN; 1328 SPAGAIN;
1003 1329
1004 if (SvTRUE (ERRSV)) 1330 if (SvTRUE (ERRSV))
1005 { 1331 {
1332 SvREFCNT_dec_NN (tag_sv);
1006 FREETMPS; LEAVE; 1333 FREETMPS; LEAVE;
1007 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV)))); 1334 ERR_ERRSV;
1008 } 1335 }
1009 1336
1010 if (count) 1337 if (count)
1011 { 1338 {
1339 SvREFCNT_dec_NN (tag_sv);
1012 SvREFCNT_dec (sv); 1340 SvREFCNT_dec_NN (sv);
1013 sv = SvREFCNT_inc (POPs); 1341 sv = SvREFCNT_inc_NN (TOPs);
1342 SP -= count;
1014 } 1343 }
1015 else 1344 else
1016 { 1345 {
1017 AV *av = newAV (); 1346 AV *av = newAV ();
1018 av_push (av, newSVuv (tag)); 1347 av_push (av, tag_sv);
1019 av_push (av, sv); 1348 av_push (av, sv);
1020 1349
1021 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 1350 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
1022 ? cbor_tagged_stash 1351 ? cbor_tagged_stash
1023 : gv_stashpv ("CBOR::XS::Tagged" , 1); 1352 : gv_stashpv ("CBOR::XS::Tagged" , 1);
… …
1109 1438
1110 return newSVnv (ecb_binary64_to_double (fp)); 1439 return newSVnv (ecb_binary64_to_double (fp));
1111 } 1440 }
1112 1441
1113 // 0..19 unassigned simple 1442 // 0..19 unassigned simple
1114 // 24 reserved + unassigned (reserved values are not encodable) 1443 // 24 reserved + unassigned simple (reserved values are not encodable)
1444 // 28-30 unassigned misc
1445 // 31 break code
1115 default: 1446 default:
1116 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)"); 1447 ERR ("corrupted CBOR data (reserved/unassigned/unexpected major 7 value)");
1117 } 1448 }
1118 1449
1119 break; 1450 break;
1120 } 1451 }
1121 1452
… …
1124} 1455}
1125 1456
1126static SV * 1457static SV *
1127decode_cbor (SV *string, CBOR *cbor, char **offset_return) 1458decode_cbor (SV *string, CBOR *cbor, char **offset_return)
1128{ 1459{
1129 dec_t dec = { }; 1460 dec_t dec = { 0 };
1130 SV *sv; 1461 SV *sv;
1131 STRLEN len; 1462 STRLEN len;
1132 char *data = SvPVbyte (string, len); 1463 char *data = SvPVbyte (string, len);
1133 1464
1134 if (len > cbor->max_size && cbor->max_size) 1465 if (len > cbor->max_size && cbor->max_size)
… …
1148 if (dec.cur != dec.end && !dec.err) 1479 if (dec.cur != dec.end && !dec.err)
1149 dec.err = "garbage after CBOR object"; 1480 dec.err = "garbage after CBOR object";
1150 1481
1151 if (dec.err) 1482 if (dec.err)
1152 { 1483 {
1484 if (dec.shareable)
1485 {
1486 // need to break cyclic links, which would all be in shareable
1487 int i;
1488 SV **svp;
1489
1490 for (i = av_len (dec.shareable) + 1; i--; )
1491 if ((svp = av_fetch (dec.shareable, i, 0)))
1492 sv_setsv (*svp, &PL_sv_undef);
1493 }
1494
1153 SvREFCNT_dec (sv); 1495 SvREFCNT_dec_NN (sv);
1496
1497 if (dec.err_sv)
1498 sv_2mortal (dec.err_sv);
1499
1154 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur); 1500 croak ("%s, at offset %ld (octet 0x%02x)", dec.err, (long)(dec.cur - (U8 *)data), (int)(uint8_t)*dec.cur);
1155 } 1501 }
1156 1502
1157 sv = sv_2mortal (sv); 1503 sv = sv_2mortal (sv);
1158 1504
1159 return sv; 1505 return sv;
1160} 1506}
1161 1507
1508/////////////////////////////////////////////////////////////////////////////
1509// incremental parser
1510
1511#define INCR_DONE(cbor) (AvFILLp (cbor->incr_count) < 0)
1512
1513// returns 0 for notyet, 1 for success or error
1514static int
1515incr_parse (CBOR *self, SV *cborstr)
1516{
1517 STRLEN cur;
1518 SvPV (cborstr, cur);
1519
1520 while (ecb_expect_true (self->incr_need <= cur))
1521 {
1522 // table of integer count bytes
1523 static I8 incr_len[MINOR_MASK + 1] = {
1524 0, 0, 0, 0, 0, 0, 0, 0,
1525 0, 0, 0, 0, 0, 0, 0, 0,
1526 0, 0, 0, 0, 0, 0, 0, 0,
1527 1, 2, 4, 8,-1,-1,-1,-2
1528 };
1529
1530 const U8 *p = SvPVX (cborstr) + self->incr_pos;
1531 U8 m = *p & MINOR_MASK;
1532 IV count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1533 I8 ilen = incr_len[m];
1534
1535 self->incr_need = self->incr_pos + 1;
1536
1537 if (ecb_expect_false (ilen < 0))
1538 {
1539 if (m != MINOR_INDEF)
1540 return 1; // error
1541
1542 if (*p == (MAJOR_MISC | MINOR_INDEF))
1543 {
1544 if (count >= 0)
1545 return 1; // error
1546
1547 count = 1;
1548 }
1549 else
1550 {
1551 av_push (self->incr_count, newSViv (-1)); //TODO: nest
1552 count = -1;
1553 }
1554 }
1555 else
1556 {
1557 self->incr_need += ilen;
1558 if (ecb_expect_false (self->incr_need > cur))
1559 return 0;
1560
1561 int major = *p >> MAJOR_SHIFT;
1562
1563 switch (major)
1564 {
1565 case MAJOR_TAG >> MAJOR_SHIFT:
1566 ++count; // tags merely prefix another value
1567 break;
1568
1569 case MAJOR_BYTES >> MAJOR_SHIFT:
1570 case MAJOR_TEXT >> MAJOR_SHIFT:
1571 case MAJOR_ARRAY >> MAJOR_SHIFT:
1572 case MAJOR_MAP >> MAJOR_SHIFT:
1573 {
1574 UV len;
1575
1576 if (ecb_expect_false (ilen))
1577 {
1578 len = 0;
1579
1580 do {
1581 len = (len << 8) | *++p;
1582 } while (--ilen);
1583 }
1584 else
1585 len = m;
1586
1587 switch (major)
1588 {
1589 case MAJOR_BYTES >> MAJOR_SHIFT:
1590 case MAJOR_TEXT >> MAJOR_SHIFT:
1591 self->incr_need += len;
1592 if (ecb_expect_false (self->incr_need > cur))
1593 return 0;
1594
1595 break;
1596
1597 case MAJOR_MAP >> MAJOR_SHIFT:
1598 len <<= 1;
1599 /* FALLTHROUGH */
1600 case MAJOR_ARRAY >> MAJOR_SHIFT:
1601 if (len)
1602 {
1603 av_push (self->incr_count, newSViv (len + 1)); //TODO: nest
1604 count = len + 1;
1605 }
1606 break;
1607 }
1608 }
1609 }
1610 }
1611
1612 self->incr_pos = self->incr_need;
1613
1614 if (count > 0)
1615 {
1616 while (!--count)
1617 {
1618 if (!AvFILLp (self->incr_count))
1619 return 1; // done
1620
1621 SvREFCNT_dec_NN (av_pop (self->incr_count));
1622 count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1623 }
1624
1625 SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]) = count;
1626 }
1627 }
1628
1629 return 0;
1630}
1631
1632
1162///////////////////////////////////////////////////////////////////////////// 1633/////////////////////////////////////////////////////////////////////////////
1163// XS interface functions 1634// XS interface functions
1164 1635
1165MODULE = CBOR::XS PACKAGE = CBOR::XS 1636MODULE = CBOR::XS PACKAGE = CBOR::XS
1166 1637
… …
1178 1649
1179 default_filter = newSVpv ("CBOR::XS::default_filter", 0); 1650 default_filter = newSVpv ("CBOR::XS::default_filter", 0);
1180 1651
1181 sv_cbor = newSVpv ("CBOR", 0); 1652 sv_cbor = newSVpv ("CBOR", 0);
1182 SvREADONLY_on (sv_cbor); 1653 SvREADONLY_on (sv_cbor);
1654
1655 assert (("STRLEN must be an unsigned type", 0 <= (STRLEN)-1));
1183} 1656}
1184 1657
1185PROTOTYPES: DISABLE 1658PROTOTYPES: DISABLE
1186 1659
1187void CLONE (...) 1660void CLONE (...)
… …
1206void shrink (CBOR *self, int enable = 1) 1679void shrink (CBOR *self, int enable = 1)
1207 ALIAS: 1680 ALIAS:
1208 shrink = F_SHRINK 1681 shrink = F_SHRINK
1209 allow_unknown = F_ALLOW_UNKNOWN 1682 allow_unknown = F_ALLOW_UNKNOWN
1210 allow_sharing = F_ALLOW_SHARING 1683 allow_sharing = F_ALLOW_SHARING
1684 allow_cycles = F_ALLOW_CYCLES
1685 forbid_objects = F_FORBID_OBJECTS
1211 pack_strings = F_PACK_STRINGS 1686 pack_strings = F_PACK_STRINGS
1687 text_keys = F_TEXT_KEYS
1688 text_strings = F_TEXT_STRINGS
1689 validate_utf8 = F_VALIDATE_UTF8
1212 PPCODE: 1690 PPCODE:
1213{ 1691{
1214 if (enable) 1692 if (enable)
1215 self->flags |= ix; 1693 self->flags |= ix;
1216 else 1694 else
… …
1222void get_shrink (CBOR *self) 1700void get_shrink (CBOR *self)
1223 ALIAS: 1701 ALIAS:
1224 get_shrink = F_SHRINK 1702 get_shrink = F_SHRINK
1225 get_allow_unknown = F_ALLOW_UNKNOWN 1703 get_allow_unknown = F_ALLOW_UNKNOWN
1226 get_allow_sharing = F_ALLOW_SHARING 1704 get_allow_sharing = F_ALLOW_SHARING
1705 get_allow_cycles = F_ALLOW_CYCLES
1706 get_forbid_objects = F_FORBID_OBJECTS
1227 get_pack_strings = F_PACK_STRINGS 1707 get_pack_strings = F_PACK_STRINGS
1708 get_text_keys = F_TEXT_KEYS
1709 get_text_strings = F_TEXT_STRINGS
1710 get_validate_utf8 = F_VALIDATE_UTF8
1228 PPCODE: 1711 PPCODE:
1229 XPUSHs (boolSV (self->flags & ix)); 1712 XPUSHs (boolSV (self->flags & ix));
1230 1713
1231void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) 1714void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
1232 PPCODE: 1715 PPCODE:
… …
1281 EXTEND (SP, 2); 1764 EXTEND (SP, 2);
1282 PUSHs (sv); 1765 PUSHs (sv);
1283 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); 1766 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
1284} 1767}
1285 1768
1769void incr_parse (CBOR *self, SV *cborstr)
1770 ALIAS:
1771 incr_parse_multiple = 1
1772 PPCODE:
1773{
1774 if (SvUTF8 (cborstr))
1775 sv_utf8_downgrade (cborstr, 0);
1776
1777 if (!self->incr_count)
1778 {
1779 self->incr_count = newAV ();
1780 self->incr_pos = 0;
1781 self->incr_need = 1;
1782
1783 av_push (self->incr_count, newSViv (1));
1784 }
1785
1786 do
1787 {
1788 if (!incr_parse (self, cborstr))
1789 {
1790 if (self->incr_need > self->max_size && self->max_size)
1791 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
1792 (unsigned long)self->incr_need, (unsigned long)self->max_size);
1793
1794 break;
1795 }
1796
1797 SV *sv;
1798 char *offset;
1799
1800 PUTBACK; sv = decode_cbor (cborstr, self, &offset); SPAGAIN;
1801 XPUSHs (sv);
1802
1803 sv_chop (cborstr, offset);
1804
1805 av_clear (self->incr_count);
1806 av_push (self->incr_count, newSViv (1));
1807
1808 self->incr_pos = 0;
1809 self->incr_need = self->incr_pos + 1;
1810 }
1811 while (ix);
1812}
1813
1814void incr_reset (CBOR *self)
1815 CODE:
1816{
1817 SvREFCNT_dec (self->incr_count);
1818 self->incr_count = 0;
1819}
1820
1286void DESTROY (CBOR *self) 1821void DESTROY (CBOR *self)
1287 PPCODE: 1822 PPCODE:
1288 cbor_free (self); 1823 cbor_free (self);
1289 1824
1290PROTOTYPES: ENABLE 1825PROTOTYPES: ENABLE
1291 1826
1292void encode_cbor (SV *scalar) 1827void encode_cbor (SV *scalar)
1828 ALIAS:
1829 encode_cbor = 0
1830 encode_cbor_sharing = F_ALLOW_SHARING
1293 PPCODE: 1831 PPCODE:
1294{ 1832{
1295 CBOR cbor; 1833 CBOR cbor;
1296 cbor_init (&cbor); 1834 cbor_init (&cbor);
1835 cbor.flags |= ix;
1297 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN; 1836 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;
1298 XPUSHs (scalar); 1837 XPUSHs (scalar);
1299} 1838}
1300 1839
1301void decode_cbor (SV *cborstr) 1840void decode_cbor (SV *cborstr)
… …
1305 cbor_init (&cbor); 1844 cbor_init (&cbor);
1306 PUTBACK; cborstr = decode_cbor (cborstr, &cbor, 0); SPAGAIN; 1845 PUTBACK; cborstr = decode_cbor (cborstr, &cbor, 0); SPAGAIN;
1307 XPUSHs (cborstr); 1846 XPUSHs (cborstr);
1308} 1847}
1309 1848
1849#ifdef __AFL_COMPILER
1850
1851void
1852afl_init ()
1853 CODE:
1854 __AFL_INIT ();
1855
1856int
1857afl_loop (unsigned int count = 10000)
1858 CODE:
1859 RETVAL = __AFL_LOOP (count);
1860 OUTPUT:
1861 RETVAL
1862
1863#endif
1864

Diff Legend

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