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 1.69 by root, Mon Nov 30 20:38:25 2020 UTC

6#include <string.h> 6#include <string.h>
7#include <stdlib.h> 7#include <stdlib.h>
8#include <stdio.h> 8#include <stdio.h>
9#include <limits.h> 9#include <limits.h>
10#include <float.h> 10#include <float.h>
11#include <inttypes.h>
11 12
13#define ECB_NO_THREADS 1
12#include "ecb.h" 14#include "ecb.h"
13 15
14// compatibility with perl <5.18 16// compatibility with perl <5.18
15#ifndef HvNAMELEN_get 17#ifndef HvNAMELEN_get
16# define HvNAMELEN_get(hv) strlen (HvNAME (hv)) 18# define HvNAMELEN_get(hv) strlen (HvNAME (hv))
19# define HvNAMELEN(hv) HvNAMELEN_get (hv) 21# define HvNAMELEN(hv) HvNAMELEN_get (hv)
20#endif 22#endif
21#ifndef HvNAMEUTF8 23#ifndef HvNAMEUTF8
22# define HvNAMEUTF8(hv) 0 24# define HvNAMEUTF8(hv) 0
23#endif 25#endif
26#ifndef SvREFCNT_inc_NN
27# define SvREFCNT_inc_NN(sv) SvREFCNT_inc (sv)
28#endif
24#ifndef SvREFCNT_dec_NN 29#ifndef SvREFCNT_dec_NN
25# define SvREFCNT_dec_NN(sv) SvREFCNT_dec (sv) 30# define SvREFCNT_dec_NN(sv) SvREFCNT_dec (sv)
26#endif 31#endif
27 32
28// known major and minor types 33// known major and minor types
94 CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8 99 CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8
95 100
96 CBOR_TAG_MAGIC = 55799, // self-describe cbor 101 CBOR_TAG_MAGIC = 55799, // self-describe cbor
97}; 102};
98 103
104// known forced types, also hardcoded in CBOR.pm
105enum
106{
107 AS_CBOR = 0,
108 AS_INT = 1,
109 AS_BYTES = 2,
110 AS_TEXT = 3,
111 AS_FLOAT16 = 4,
112 AS_FLOAT32 = 5,
113 AS_FLOAT64 = 6,
114 // possibly future enhancements: (generic) float, (generic) string
115};
116
99#define F_SHRINK 0x00000001UL 117#define F_SHRINK 0x00000001UL
100#define F_ALLOW_UNKNOWN 0x00000002UL 118#define F_ALLOW_UNKNOWN 0x00000002UL
101#define F_ALLOW_SHARING 0x00000004UL 119#define F_ALLOW_SHARING 0x00000004UL
120#define F_ALLOW_CYCLES 0x00000008UL
121#define F_FORBID_OBJECTS 0x00000010UL
102#define F_PACK_STRINGS 0x00000008UL 122#define F_PACK_STRINGS 0x00000020UL
123#define F_TEXT_KEYS 0x00000040UL
124#define F_TEXT_STRINGS 0x00000080UL
125#define F_VALIDATE_UTF8 0x00000100UL
103 126
104#define INIT_SIZE 32 // initial scalar size to be allocated 127#define INIT_SIZE 32 // initial scalar size to be allocated
105 128
106#define SB do { 129#define SB do {
107#define SE } while (0) 130#define SE } while (0)
126typedef struct { 149typedef struct {
127 U32 flags; 150 U32 flags;
128 U32 max_depth; 151 U32 max_depth;
129 STRLEN max_size; 152 STRLEN max_size;
130 SV *filter; 153 SV *filter;
154
155 // for the incremental parser
156 STRLEN incr_pos; // the current offset into the text
157 STRLEN incr_need; // minimum bytes needed to decode
158 AV *incr_count; // for every nesting level, the number of outstanding values, or -1 for indef.
131} CBOR; 159} CBOR;
132 160
133ecb_inline void 161ecb_inline void
134cbor_init (CBOR *cbor) 162cbor_init (CBOR *cbor)
135{ 163{
139 167
140ecb_inline void 168ecb_inline void
141cbor_free (CBOR *cbor) 169cbor_free (CBOR *cbor)
142{ 170{
143 SvREFCNT_dec (cbor->filter); 171 SvREFCNT_dec (cbor->filter);
172 SvREFCNT_dec (cbor->incr_count);
144} 173}
145 174
146///////////////////////////////////////////////////////////////////////////// 175/////////////////////////////////////////////////////////////////////////////
147// utility functions 176// utility functions
148 177
174 203
175// minimum length of a string to be registered for stringref 204// minimum length of a string to be registered for stringref
176ecb_inline int 205ecb_inline int
177minimum_string_length (UV idx) 206minimum_string_length (UV idx)
178{ 207{
179 return idx > 23 208 return idx <= 23 ? 3
180 ? idx > 0xffU 209 : idx <= 0xffU ? 4
181 ? idx > 0xffffU 210 : idx <= 0xffffU ? 5
182 ? idx > 0xffffffffU 211 : idx <= 0xffffffffU ? 7
183 ? 11 212 : 11;
184 : 7
185 : 5
186 : 4
187 : 3;
188} 213}
189 214
190///////////////////////////////////////////////////////////////////////////// 215/////////////////////////////////////////////////////////////////////////////
191// encoder 216// encoder
192 217
205} enc_t; 230} enc_t;
206 231
207ecb_inline void 232ecb_inline void
208need (enc_t *enc, STRLEN len) 233need (enc_t *enc, STRLEN len)
209{ 234{
210 if (ecb_expect_false (enc->cur + len >= enc->end)) 235 if (ecb_expect_false ((uintptr_t)(enc->end - enc->cur) < len))
211 { 236 {
212 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv); 237 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
213 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1); 238 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
214 enc->cur = SvPVX (enc->sv) + cur; 239 enc->cur = SvPVX (enc->sv) + cur;
215 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; 240 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
221{ 246{
222 need (enc, 1); 247 need (enc, 1);
223 *enc->cur++ = ch; 248 *enc->cur++ = ch;
224} 249}
225 250
251// used for tags, intregers, element counts and so on
226static void 252static void
227encode_uint (enc_t *enc, int major, UV len) 253encode_uint (enc_t *enc, int major, UV len)
228{ 254{
229 need (enc, 9); 255 need (enc, 9);
230 256
231 if (ecb_expect_true (len < LENGTH_EXT1)) 257 if (ecb_expect_true (len < LENGTH_EXT1))
232 *enc->cur++ = major | len; 258 *enc->cur++ = major | len;
233 else if (ecb_expect_true (len <= 0xff)) 259 else if (ecb_expect_true (len <= 0xffU))
234 { 260 {
235 *enc->cur++ = major | LENGTH_EXT1; 261 *enc->cur++ = major | LENGTH_EXT1;
236 *enc->cur++ = len; 262 *enc->cur++ = len;
237 } 263 }
238 else if (len <= 0xffff) 264 else if (len <= 0xffffU)
239 { 265 {
240 *enc->cur++ = major | LENGTH_EXT2; 266 *enc->cur++ = major | LENGTH_EXT2;
241 *enc->cur++ = len >> 8; 267 *enc->cur++ = len >> 8;
242 *enc->cur++ = len; 268 *enc->cur++ = len;
243 } 269 }
244 else if (len <= 0xffffffff) 270 else if (len <= 0xffffffffU)
245 { 271 {
246 *enc->cur++ = major | LENGTH_EXT4; 272 *enc->cur++ = major | LENGTH_EXT4;
247 *enc->cur++ = len >> 24; 273 *enc->cur++ = len >> 24;
248 *enc->cur++ = len >> 16; 274 *enc->cur++ = len >> 16;
249 *enc->cur++ = len >> 8; 275 *enc->cur++ = len >> 8;
261 *enc->cur++ = len >> 8; 287 *enc->cur++ = len >> 8;
262 *enc->cur++ = len; 288 *enc->cur++ = len;
263 } 289 }
264} 290}
265 291
292// encodes a perl value into a CBOR integer
293ecb_inline void
294encode_int (enc_t *enc, SV *sv)
295{
296 if (SvIsUV (sv))
297 encode_uint (enc, MAJOR_POS_INT, SvUVX (sv));
298 else if (SvIVX (sv) >= 0)
299 encode_uint (enc, MAJOR_POS_INT, SvIVX (sv));
300 else
301 encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1));
302}
303
266ecb_inline void 304ecb_inline void
267encode_tag (enc_t *enc, UV tag) 305encode_tag (enc_t *enc, UV tag)
268{ 306{
269 encode_uint (enc, MAJOR_TAG, tag); 307 encode_uint (enc, MAJOR_TAG, tag);
270} 308}
271 309
310// exceptional (hopefully) slow path for byte strings that need to be utf8-encoded
311ecb_noinline static void
312encode_str_utf8 (enc_t *enc, int utf8, char *str, STRLEN len)
313{
314 STRLEN ulen = len;
315 U8 *p, *pend = (U8 *)str + len;
316
317 for (p = (U8 *)str; p < pend; ++p)
318 ulen += *p >> 7; // count set high bits
319
320 encode_uint (enc, MAJOR_TEXT, ulen);
321
322 need (enc, ulen);
323 for (p = (U8 *)str; p < pend; ++p)
324 if (*p < 0x80)
325 *enc->cur++ = *p;
326 else
327 {
328 *enc->cur++ = 0xc0 + (*p >> 6);
329 *enc->cur++ = 0x80 + (*p & 63);
330 }
331}
332
272ecb_inline void 333ecb_inline void
273encode_str (enc_t *enc, int utf8, char *str, STRLEN len) 334encode_str (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len)
274{ 335{
336 if (ecb_expect_false (upgrade_utf8))
337 if (!utf8)
338 {
339 encode_str_utf8 (enc, utf8, str, len);
340 return;
341 }
342
275 encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len); 343 encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len);
276 need (enc, len); 344 need (enc, len);
277 memcpy (enc->cur, str, len); 345 memcpy (enc->cur, str, len);
278 enc->cur += len; 346 enc->cur += len;
279} 347}
280 348
281static void 349ecb_inline void
282encode_strref (enc_t *enc, int utf8, char *str, STRLEN len) 350encode_strref (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len)
283{ 351{
284 if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS)) 352 if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS))
285 { 353 {
286 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1); 354 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1);
287 355
298 sv_setuv (*svp, enc->stringref_idx); 366 sv_setuv (*svp, enc->stringref_idx);
299 ++enc->stringref_idx; 367 ++enc->stringref_idx;
300 } 368 }
301 } 369 }
302 370
303 encode_str (enc, utf8, str, len); 371 encode_str (enc, upgrade_utf8, utf8, str, len);
372}
373
374ecb_inline void
375encode_float16 (enc_t *enc, NV nv)
376{
377 need (enc, 1+2);
378
379 *enc->cur++ = MAJOR_MISC | MISC_FLOAT16;
380
381 uint16_t fp = ecb_float_to_binary16 (nv);
382
383 if (!ecb_big_endian ())
384 fp = ecb_bswap16 (fp);
385
386 memcpy (enc->cur, &fp, 2);
387 enc->cur += 2;
388}
389
390ecb_inline void
391encode_float32 (enc_t *enc, NV nv)
392{
393 need (enc, 1+4);
394
395 *enc->cur++ = MAJOR_MISC | MISC_FLOAT32;
396
397 uint32_t fp = ecb_float_to_binary32 (nv);
398
399 if (!ecb_big_endian ())
400 fp = ecb_bswap32 (fp);
401
402 memcpy (enc->cur, &fp, 4);
403 enc->cur += 4;
404}
405
406ecb_inline void
407encode_float64 (enc_t *enc, NV nv)
408{
409 need (enc, 1+8);
410
411 *enc->cur++ = MAJOR_MISC | MISC_FLOAT64;
412
413 uint64_t fp = ecb_double_to_binary64 (nv);
414
415 if (!ecb_big_endian ())
416 fp = ecb_bswap64 (fp);
417
418 memcpy (enc->cur, &fp, 8);
419 enc->cur += 8;
420}
421
422ecb_inline void
423encode_bool (enc_t *enc, int istrue)
424{
425 encode_ch (enc, istrue ? MAJOR_MISC | SIMPLE_TRUE : MAJOR_MISC | SIMPLE_FALSE);
426}
427
428ecb_inline void
429encode_forced (enc_t *enc, UV type, SV *sv)
430{
431 switch (type)
432 {
433 case AS_CBOR:
434 {
435 STRLEN len;
436 char *str = SvPVbyte (sv, len);
437
438 need (enc, len);
439 memcpy (enc->cur, str, len);
440 enc->cur += len;
441 }
442 break;
443
444 case AS_BYTES:
445 {
446 STRLEN len;
447 char *str = SvPVbyte (sv, len);
448 encode_strref (enc, 0, 0, str, len);
449 }
450 break;
451
452 case AS_TEXT:
453 {
454 STRLEN len;
455 char *str = SvPVutf8 (sv, len);
456 encode_strref (enc, 1, 1, str, len);
457 }
458 break;
459
460 case AS_INT: encode_int (enc, sv); break;
461
462 case AS_FLOAT16: encode_float16 (enc, SvNV (sv)); break;
463 case AS_FLOAT32: encode_float32 (enc, SvNV (sv)); break;
464 case AS_FLOAT64: encode_float64 (enc, SvNV (sv)); break;
465
466 default:
467 croak ("encountered malformed CBOR::XS::Tagged object");
468 }
304} 469}
305 470
306static void encode_sv (enc_t *enc, SV *sv); 471static void encode_sv (enc_t *enc, SV *sv);
307 472
308static void 473static void
315 480
316 ++enc->depth; 481 ++enc->depth;
317 482
318 encode_uint (enc, MAJOR_ARRAY, len + 1); 483 encode_uint (enc, MAJOR_ARRAY, len + 1);
319 484
485 if (ecb_expect_false (SvMAGICAL (av)))
320 for (i = 0; i <= len; ++i) 486 for (i = 0; i <= len; ++i)
321 { 487 {
322 SV **svp = av_fetch (av, i, 0); 488 SV **svp = av_fetch (av, i, 0);
323 encode_sv (enc, svp ? *svp : &PL_sv_undef); 489 encode_sv (enc, svp ? *svp : &PL_sv_undef);
324 } 490 }
491 else
492 for (i = 0; i <= len; ++i)
493 {
494 SV *sv = AvARRAY (av)[i];
495 encode_sv (enc, sv ? sv : &PL_sv_undef);
496 }
325 497
326 --enc->depth; 498 --enc->depth;
327} 499}
328 500
329static void 501static void
337 ++enc->depth; 509 ++enc->depth;
338 510
339 int pairs = hv_iterinit (hv); 511 int pairs = hv_iterinit (hv);
340 int mg = SvMAGICAL (hv); 512 int mg = SvMAGICAL (hv);
341 513
342 if (mg) 514 if (ecb_expect_false (mg))
343 encode_ch (enc, MAJOR_MAP | MINOR_INDEF); 515 encode_ch (enc, MAJOR_MAP | MINOR_INDEF);
344 else 516 else
345 encode_uint (enc, MAJOR_MAP, pairs); 517 encode_uint (enc, MAJOR_MAP, pairs);
346 518
347 while ((he = hv_iternext (hv))) 519 while ((he = hv_iternext (hv)))
348 { 520 {
349 if (HeKLEN (he) == HEf_SVKEY) 521 if (HeKLEN (he) == HEf_SVKEY)
350 encode_sv (enc, HeSVKEY (he)); 522 encode_sv (enc, HeSVKEY (he));
351 else 523 else
352 encode_strref (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he)); 524 encode_strref (enc, enc->cbor.flags & (F_TEXT_KEYS | F_TEXT_STRINGS), HeKUTF8 (he), HeKEY (he), HeKLEN (he));
353 525
354 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he)); 526 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he));
355 } 527 }
356 528
357 if (mg) 529 if (ecb_expect_false (mg))
358 encode_ch (enc, MAJOR_MISC | MINOR_INDEF); 530 encode_ch (enc, MAJOR_MISC | MINOR_INDEF);
359 531
360 --enc->depth; 532 --enc->depth;
361} 533}
362 534
382 554
383 HV *stash = SvSTASH (sv); 555 HV *stash = SvSTASH (sv);
384 556
385 if (stash == boolean_stash) 557 if (stash == boolean_stash)
386 { 558 {
387 encode_ch (enc, SvIV (sv) ? MAJOR_MISC | SIMPLE_TRUE : MAJOR_MISC | SIMPLE_FALSE); 559 encode_bool (enc, SvIV (sv));
388 return; 560 return;
389 } 561 }
390 else if (stash == error_stash) 562 else if (stash == error_stash)
391 { 563 {
392 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF); 564 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
395 else if (stash == tagged_stash) 567 else if (stash == tagged_stash)
396 { 568 {
397 if (svt != SVt_PVAV) 569 if (svt != SVt_PVAV)
398 croak ("encountered CBOR::XS::Tagged object that isn't an array"); 570 croak ("encountered CBOR::XS::Tagged object that isn't an array");
399 571
572 switch (av_len ((AV *)sv))
573 {
574 case 2-1:
575 // actually a tagged value
400 encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1))); 576 encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1)));
401 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1)); 577 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1));
578 break;
579
580 case 3-1:
581 // a forced type [value, type, undef]
582 encode_forced (enc, SvUV (*av_fetch ((AV *)sv, 1, 1)), *av_fetch ((AV *)sv, 0, 1));
583 break;
584
585 default:
586 croak ("encountered malformed CBOR::XS::Tagged object");
587 }
402 588
403 return; 589 return;
404 } 590 }
405 } 591 }
406 592
407 if (ecb_expect_false (SvREFCNT (sv) > 1) 593 if (ecb_expect_false (SvREFCNT (sv) > 1)
408 && ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING)) 594 && ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING))
409 { 595 {
410 if (!enc->shareable) 596 if (ecb_expect_false (!enc->shareable))
411 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ()); 597 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
412 598
413 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1); 599 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
414 600
415 if (SvOK (*svp)) 601 if (SvOK (*svp))
429 if (ecb_expect_false (SvOBJECT (sv))) 615 if (ecb_expect_false (SvOBJECT (sv)))
430 { 616 {
431 HV *stash = SvSTASH (sv); 617 HV *stash = SvSTASH (sv);
432 GV *method; 618 GV *method;
433 619
620 if (enc->cbor.flags & F_FORBID_OBJECTS)
621 croak ("encountered object '%s', but forbid_objects is enabled",
622 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
434 if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0))) 623 else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
435 { 624 {
436 dSP; 625 dSP;
437 626
438 ENTER; SAVETMPS; PUSHMARK (SP); 627 ENTER; SAVETMPS;
628 PUSHMARK (SP);
439 // we re-bless the reference to get overload and other niceties right 629 // we re-bless the reference to get overload and other niceties right
440 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); 630 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
441 631
442 PUTBACK; 632 PUTBACK;
443 // G_SCALAR ensures that return value is 1 633 // G_SCALAR ensures that return value is 1
456 } 646 }
457 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0) 647 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)
458 { 648 {
459 dSP; 649 dSP;
460 650
461 ENTER; SAVETMPS; PUSHMARK (SP); 651 ENTER; SAVETMPS;
652 PUSHMARK (SP);
462 EXTEND (SP, 2); 653 EXTEND (SP, 2);
463 // we re-bless the reference to get overload and other niceties right 654 // we re-bless the reference to get overload and other niceties right
464 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); 655 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
465 PUSHs (sv_cbor); 656 PUSHs (sv_cbor);
466 657
472 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv) 663 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)); 664 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));
474 665
475 encode_tag (enc, CBOR_TAG_PERL_OBJECT); 666 encode_tag (enc, CBOR_TAG_PERL_OBJECT);
476 encode_uint (enc, MAJOR_ARRAY, count + 1); 667 encode_uint (enc, MAJOR_ARRAY, count + 1);
477 encode_strref (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash)); 668 encode_strref (enc, 0, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
478 669
479 while (count) 670 {
671 int i;
672
673 for (i = 0; i < count; ++i)
480 encode_sv (enc, SP[1 - count--]); 674 encode_sv (enc, SP[i + 1 - count]);
675
676 SP -= count;
677 }
481 678
482 PUTBACK; 679 PUTBACK;
483 680
484 FREETMPS; LEAVE; 681 FREETMPS; LEAVE;
485 } 682 }
507 704
508 if (ecb_expect_false (nv == (NV)(U32)nv)) 705 if (ecb_expect_false (nv == (NV)(U32)nv))
509 encode_uint (enc, MAJOR_POS_INT, (U32)nv); 706 encode_uint (enc, MAJOR_POS_INT, (U32)nv);
510 //TODO: maybe I32? 707 //TODO: maybe I32?
511 else if (ecb_expect_false (nv == (float)nv)) 708 else if (ecb_expect_false (nv == (float)nv))
512 { 709 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 710 else
524 { 711 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} 712}
536 713
537static void 714static void
538encode_sv (enc_t *enc, SV *sv) 715encode_sv (enc_t *enc, SV *sv)
539{ 716{
541 718
542 if (SvPOKp (sv)) 719 if (SvPOKp (sv))
543 { 720 {
544 STRLEN len; 721 STRLEN len;
545 char *str = SvPV (sv, len); 722 char *str = SvPV (sv, len);
546 encode_strref (enc, SvUTF8 (sv), str, len); 723 encode_strref (enc, enc->cbor.flags & F_TEXT_STRINGS, SvUTF8 (sv), str, len);
547 } 724 }
548 else if (SvNOKp (sv)) 725 else if (SvNOKp (sv))
549 encode_nv (enc, sv); 726 encode_nv (enc, sv);
550 else if (SvIOKp (sv)) 727 else if (SvIOKp (sv))
551 { 728 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)) 729 else if (SvROK (sv))
560 encode_rv (enc, SvRV (sv)); 730 encode_rv (enc, SvRV (sv));
561 else if (!SvOK (sv)) 731 else if (!SvOK (sv))
562 encode_ch (enc, MAJOR_MISC | SIMPLE_NULL); 732 encode_ch (enc, MAJOR_MISC | SIMPLE_NULL);
563 else if (enc->cbor.flags & F_ALLOW_UNKNOWN) 733 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
568} 738}
569 739
570static SV * 740static SV *
571encode_cbor (SV *scalar, CBOR *cbor) 741encode_cbor (SV *scalar, CBOR *cbor)
572{ 742{
573 enc_t enc = { }; 743 enc_t enc = { 0 };
574 744
575 enc.cbor = *cbor; 745 enc.cbor = *cbor;
576 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 746 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
577 enc.cur = SvPVX (enc.sv); 747 enc.cur = SvPVX (enc.sv);
578 enc.end = SvEND (enc.sv); 748 enc.end = SvEND (enc.sv);
579 749
580 SvPOK_only (enc.sv); 750 SvPOK_only (enc.sv);
581 751
582 if (cbor->flags & F_PACK_STRINGS) 752 if (cbor->flags & F_PACK_STRINGS)
583 { 753 {
610 U32 depth; // recursion depth 780 U32 depth; // recursion depth
611 U32 maxdepth; // recursion depth limit 781 U32 maxdepth; // recursion depth limit
612 AV *shareable; 782 AV *shareable;
613 AV *stringref; 783 AV *stringref;
614 SV *decode_tagged; 784 SV *decode_tagged;
785 SV *err_sv; // optional sv for error, needs to be freed
615} dec_t; 786} dec_t;
616 787
617#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE 788// set dec->err to ERRSV
789ecb_cold static void
790err_errsv (dec_t *dec)
791{
792 if (!dec->err)
793 {
794 dec->err_sv = newSVsv (ERRSV);
618 795
619#define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data") 796 // chop off the trailing \n
797 SvCUR_set (dec->err_sv, SvCUR (dec->err_sv) - 1);
798 *SvEND (dec->err_sv) = 0;
620 799
800 dec->err = SvPVutf8_nolen (dec->err_sv);
801 }
802}
803
804// the following functions are used to reduce code size and help the compiler to optimise
805ecb_cold static void
806err_set (dec_t *dec, const char *reason)
807{
808 if (!dec->err)
809 dec->err = reason;
810}
811
812ecb_cold static void
813err_unexpected_end (dec_t *dec)
814{
815 err_set (dec, "unexpected end of CBOR data");
816}
817
818#define ERR_DO(do) SB do; goto fail; SE
819#define ERR(reason) ERR_DO (err_set (dec, reason))
820#define ERR_ERRSV ERR_DO (err_errsv (dec))
821
822#define WANT(len) if (ecb_expect_false ((uintptr_t)(dec->end - dec->cur) < (STRLEN)len)) ERR_DO (err_unexpected_end (dec))
823
621#define DEC_INC_DEPTH if (++dec->depth > dec->cbor.max_depth) ERR (ERR_NESTING_EXCEEDED) 824#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 825#define DEC_DEC_DEPTH --dec->depth
623 826
624static UV 827static UV
625decode_uint (dec_t *dec) 828decode_uint (dec_t *dec)
626{ 829{
627 U8 m = *dec->cur & MINOR_MASK; 830 U8 m = *dec->cur & MINOR_MASK;
628 ++dec->cur; 831 ++dec->cur;
629 832
630 if (ecb_expect_true (m < LENGTH_EXT1)) 833 if (ecb_expect_true (m < LENGTH_EXT1))
631 return m; 834 return m;
632 835 else if (ecb_expect_true (m == LENGTH_EXT1))
633 switch (m)
634 { 836 {
635 case LENGTH_EXT1:
636 WANT (1); 837 WANT (1);
637 dec->cur += 1; 838 dec->cur += 1;
638 return dec->cur[-1]; 839 return dec->cur[-1];
639 840 }
640 case LENGTH_EXT2: 841 else if (ecb_expect_true (m == LENGTH_EXT2))
842 {
641 WANT (2); 843 WANT (2);
642 dec->cur += 2; 844 dec->cur += 2;
643 return (((UV)dec->cur[-2]) << 8) 845 return (((UV)dec->cur[-2]) << 8)
644 | ((UV)dec->cur[-1]); 846 | ((UV)dec->cur[-1]);
645 847 }
646 case LENGTH_EXT4: 848 else if (ecb_expect_true (m == LENGTH_EXT4))
849 {
647 WANT (4); 850 WANT (4);
648 dec->cur += 4; 851 dec->cur += 4;
649 return (((UV)dec->cur[-4]) << 24) 852 return (((UV)dec->cur[-4]) << 24)
650 | (((UV)dec->cur[-3]) << 16) 853 | (((UV)dec->cur[-3]) << 16)
651 | (((UV)dec->cur[-2]) << 8) 854 | (((UV)dec->cur[-2]) << 8)
652 | ((UV)dec->cur[-1]); 855 | ((UV)dec->cur[-1]);
653 856 }
654 case LENGTH_EXT8: 857 else if (ecb_expect_true (m == LENGTH_EXT8))
858 {
655 WANT (8); 859 WANT (8);
656 dec->cur += 8; 860 dec->cur += 8;
657 861
658 return 862 return
659#if UVSIZE < 8 863#if UVSIZE < 8
660 0 864 0
661#else 865#else
662 (((UV)dec->cur[-8]) << 56) 866 (((UV)dec->cur[-8]) << 56)
663 | (((UV)dec->cur[-7]) << 48) 867 | (((UV)dec->cur[-7]) << 48)
664 | (((UV)dec->cur[-6]) << 40) 868 | (((UV)dec->cur[-6]) << 40)
665 | (((UV)dec->cur[-5]) << 32) 869 | (((UV)dec->cur[-5]) << 32)
666#endif 870#endif
667 | (((UV)dec->cur[-4]) << 24) 871 | (((UV)dec->cur[-4]) << 24)
668 | (((UV)dec->cur[-3]) << 16) 872 | (((UV)dec->cur[-3]) << 16)
669 | (((UV)dec->cur[-2]) << 8) 873 | (((UV)dec->cur[-2]) << 8)
670 | ((UV)dec->cur[-1]); 874 | ((UV)dec->cur[-1]);
671 875 }
672 default: 876 else
673 ERR ("corrupted CBOR data (unsupported integer minor encoding)"); 877 ERR ("corrupted CBOR data (unsupported integer minor encoding)");
674 }
675 878
676fail: 879fail:
677 return 0; 880 return 0;
678} 881}
679 882
692 895
693 for (;;) 896 for (;;)
694 { 897 {
695 WANT (1); 898 WANT (1);
696 899
697 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF)) 900 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF) || dec->err)
698 { 901 {
699 ++dec->cur; 902 ++dec->cur;
700 break; 903 break;
701 } 904 }
702 905
703 av_push (av, decode_sv (dec)); 906 av_push (av, decode_sv (dec));
704 } 907 }
705 } 908 }
706 else 909 else
707 { 910 {
708 int i, len = decode_uint (dec); 911 UV i, len = decode_uint (dec);
709 912
913 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); 914 av_fill (av, len - 1);
711 915
712 for (i = 0; i < len; ++i) 916 for (i = 0; i < len; ++i)
713 AvARRAY (av)[i] = decode_sv (dec); 917 AvARRAY (av)[i] = decode_sv (dec);
714 } 918 }
715 919
716 DEC_DEC_DEPTH; 920 DEC_DEC_DEPTH;
717 return newRV_noinc ((SV *)av); 921 return newRV_noinc ((SV *)av);
718 922
719fail: 923fail:
720 SvREFCNT_dec (av); 924 SvREFCNT_dec_NN (av);
721 DEC_DEC_DEPTH; 925 DEC_DEC_DEPTH;
722 return &PL_sv_undef; 926 return &PL_sv_undef;
723} 927}
724 928
725static void 929static void
727{ 931{
728 // for speed reasons, we specialcase single-string 932 // for speed reasons, we specialcase single-string
729 // byte or utf-8 strings as keys, but only when !stringref 933 // byte or utf-8 strings as keys, but only when !stringref
730 934
731 if (ecb_expect_true (!dec->stringref)) 935 if (ecb_expect_true (!dec->stringref))
732 if ((*dec->cur - MAJOR_BYTES) <= 27) 936 if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8))
733 { 937 {
734 I32 len = decode_uint (dec); 938 STRLEN len = decode_uint (dec);
735 char *key = (char *)dec->cur; 939 char *key = (char *)dec->cur;
736 940
941 WANT (len);
737 dec->cur += len; 942 dec->cur += len;
738 943
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); 944 hv_store (hv, key, len, decode_sv (dec), 0);
743 945
744 return; 946 return;
745 } 947 }
746 else if ((*dec->cur - MAJOR_TEXT) <= 27) 948 else if (ecb_expect_true ((U8)(*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8))
747 { 949 {
748 I32 len = decode_uint (dec); 950 STRLEN len = decode_uint (dec);
749 char *key = (char *)dec->cur; 951 char *key = (char *)dec->cur;
750 952
953 WANT (len);
751 dec->cur += len; 954 dec->cur += len;
752 955
753 if (ecb_expect_false (dec->stringref)) 956 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
754 av_push (dec->stringref, newSVpvn_utf8 (key, len, 1)); 957 if (!is_utf8_string (key, len))
958 ERR ("corrupted CBOR data (invalid UTF-8 in map key)");
755 959
756 hv_store (hv, key, -len, decode_sv (dec), 0); 960 hv_store (hv, key, -len, decode_sv (dec), 0);
757 961
758 return; 962 return;
759 } 963 }
760 964
761 SV *k = decode_sv (dec); 965 SV *k = decode_sv (dec);
762 SV *v = decode_sv (dec); 966 SV *v = decode_sv (dec);
763 967
968 // we leak memory if uncaught exceptions are thrown by random magical
969 // methods, and this is hopefully the only place where it can happen,
970 // so if there is a chance of an exception, take the very slow path.
971 // since catching exceptions is "undocumented/internal/forbidden" by
972 // the new p5p powers, we need to call out to a perl function :/
973 if (ecb_expect_false (SvAMAGIC (k)))
974 {
975 dSP;
976
977 ENTER; SAVETMPS;
978 PUSHMARK (SP);
979 EXTEND (SP, 3);
980 PUSHs (sv_2mortal (newRV_inc ((SV *)hv)));
981 PUSHs (sv_2mortal (k));
982 PUSHs (sv_2mortal (v));
983
984 PUTBACK;
985 call_pv ("CBOR::XS::_hv_store", G_VOID | G_DISCARD | G_EVAL);
986 SPAGAIN;
987
988 FREETMPS; LEAVE;
989
990 if (SvTRUE (ERRSV))
991 ERR_ERRSV;
992
993 return;
994 }
995
764 hv_store_ent (hv, k, v, 0); 996 hv_store_ent (hv, k, v, 0);
765 SvREFCNT_dec (k); 997 SvREFCNT_dec_NN (k);
998
999fail:
1000 ;
766} 1001}
767 1002
768static SV * 1003static SV *
769decode_hv (dec_t *dec) 1004decode_hv (dec_t *dec)
770{ 1005{
778 1013
779 for (;;) 1014 for (;;)
780 { 1015 {
781 WANT (1); 1016 WANT (1);
782 1017
783 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF)) 1018 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF) || dec->err)
784 { 1019 {
785 ++dec->cur; 1020 ++dec->cur;
786 break; 1021 break;
787 } 1022 }
788 1023
789 decode_he (dec, hv); 1024 decode_he (dec, hv);
790 } 1025 }
791 } 1026 }
792 else 1027 else
793 { 1028 {
794 int pairs = decode_uint (dec); 1029 UV pairs = decode_uint (dec);
1030
1031 WANT (pairs); // complexity check - need at least one byte per value, do not allow supersize hashes
795 1032
796 while (pairs--) 1033 while (pairs--)
797 decode_he (dec, hv); 1034 decode_he (dec, hv);
798 } 1035 }
799 1036
800 DEC_DEC_DEPTH; 1037 DEC_DEC_DEPTH;
801 return newRV_noinc ((SV *)hv); 1038 return newRV_noinc ((SV *)hv);
802 1039
803fail: 1040fail:
804 SvREFCNT_dec (hv); 1041 SvREFCNT_dec_NN (hv);
805 DEC_DEC_DEPTH; 1042 DEC_DEC_DEPTH;
806 return &PL_sv_undef; 1043 return &PL_sv_undef;
807} 1044}
808 1045
809static SV * 1046static SV *
810decode_str (dec_t *dec, int utf8) 1047decode_str (dec_t *dec, int utf8)
811{ 1048{
812 SV *sv = 0; 1049 SV *sv = 0;
813 1050
814 if ((*dec->cur & MINOR_MASK) == MINOR_INDEF) 1051 if (ecb_expect_false ((*dec->cur & MINOR_MASK) == MINOR_INDEF))
815 { 1052 {
816 // indefinite length strings 1053 // indefinite length strings
817 ++dec->cur; 1054 ++dec->cur;
818 1055
819 U8 major = *dec->cur & MAJOR_MISC; 1056 U8 major = *dec->cur & MAJOR_MISC;
852 && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1)) 1089 && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1))
853 av_push (dec->stringref, SvREFCNT_inc_NN (sv)); 1090 av_push (dec->stringref, SvREFCNT_inc_NN (sv));
854 } 1091 }
855 1092
856 if (utf8) 1093 if (utf8)
1094 {
1095 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
1096 if (!is_utf8_string (SvPVX (sv), SvCUR (sv)))
1097 ERR ("corrupted CBOR data (invalid UTF-8 in text string)");
1098
857 SvUTF8_on (sv); 1099 SvUTF8_on (sv);
1100 }
858 1101
859 return sv; 1102 return sv;
860 1103
861fail: 1104fail:
862 SvREFCNT_dec (sv); 1105 SvREFCNT_dec (sv);
881 sv = newRV_noinc (decode_sv (dec)); 1124 sv = newRV_noinc (decode_sv (dec));
882 break; 1125 break;
883 1126
884 case CBOR_TAG_STRINGREF_NAMESPACE: 1127 case CBOR_TAG_STRINGREF_NAMESPACE:
885 { 1128 {
1129 // do not use SAVETMPS/FREETMPS, as these will
1130 // erase mortalised caches, e.g. "shareable"
886 ENTER; SAVETMPS; 1131 ENTER;
887 1132
888 SAVESPTR (dec->stringref); 1133 SAVESPTR (dec->stringref);
889 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ()); 1134 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ());
890 1135
891 sv = decode_sv (dec); 1136 sv = decode_sv (dec);
892 1137
893 FREETMPS; LEAVE; 1138 LEAVE;
894 } 1139 }
895 break; 1140 break;
896 1141
897 case CBOR_TAG_STRINGREF: 1142 case CBOR_TAG_STRINGREF:
898 { 1143 {
899 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT)) 1144 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
900 ERR ("corrupted CBOR data (stringref index not an unsigned integer)"); 1145 ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
901 1146
902 UV idx = decode_uint (dec); 1147 UV idx = decode_uint (dec);
903 1148
904 if (!dec->stringref || (int)idx > AvFILLp (dec->stringref)) 1149 if (!dec->stringref || idx >= (UV)(1 + AvFILLp (dec->stringref)))
905 ERR ("corrupted CBOR data (stringref index out of bounds or outside namespace)"); 1150 ERR ("corrupted CBOR data (stringref index out of bounds or outside namespace)");
906 1151
907 sv = newSVsv (AvARRAY (dec->stringref)[idx]); 1152 sv = newSVsv (AvARRAY (dec->stringref)[idx]);
908 } 1153 }
909 break; 1154 break;
911 case CBOR_TAG_VALUE_SHAREABLE: 1156 case CBOR_TAG_VALUE_SHAREABLE:
912 { 1157 {
913 if (ecb_expect_false (!dec->shareable)) 1158 if (ecb_expect_false (!dec->shareable))
914 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ()); 1159 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ());
915 1160
1161 if (dec->cbor.flags & F_ALLOW_CYCLES)
1162 {
916 sv = newSV (0); 1163 sv = newSV (0);
917 av_push (dec->shareable, SvREFCNT_inc_NN (sv)); 1164 av_push (dec->shareable, SvREFCNT_inc_NN (sv));
918 1165
919 SV *osv = decode_sv (dec); 1166 SV *osv = decode_sv (dec);
920 sv_setsv (sv, osv); 1167 sv_setsv (sv, osv);
921 SvREFCNT_dec_NN (osv); 1168 SvREFCNT_dec_NN (osv);
1169 }
1170 else
1171 {
1172 av_push (dec->shareable, &PL_sv_undef);
1173 int idx = AvFILLp (dec->shareable);
1174 sv = decode_sv (dec);
1175 av_store (dec->shareable, idx, SvREFCNT_inc_NN (sv));
1176 }
922 } 1177 }
923 break; 1178 break;
924 1179
925 case CBOR_TAG_VALUE_SHAREDREF: 1180 case CBOR_TAG_VALUE_SHAREDREF:
926 { 1181 {
927 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT)) 1182 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
928 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)"); 1183 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
929 1184
930 UV idx = decode_uint (dec); 1185 UV idx = decode_uint (dec);
931 1186
932 if (!dec->shareable || (int)idx > AvFILLp (dec->shareable)) 1187 if (!dec->shareable || idx >= (UV)(1 + AvFILLp (dec->shareable)))
933 ERR ("corrupted CBOR data (sharedref index out of bounds)"); 1188 ERR ("corrupted CBOR data (sharedref index out of bounds)");
934 1189
935 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]); 1190 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]);
1191
1192 if (sv == &PL_sv_undef)
1193 ERR ("cyclic CBOR data structure found, but allow_cycles is not enabled");
936 } 1194 }
937 break; 1195 break;
938 1196
939 case CBOR_TAG_PERL_OBJECT: 1197 case CBOR_TAG_PERL_OBJECT:
940 { 1198 {
1199 if (dec->cbor.flags & F_FORBID_OBJECTS)
1200 goto filter;
1201
941 sv = decode_sv (dec); 1202 sv = decode_sv (dec);
942 1203
943 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV) 1204 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
944 ERR ("corrupted CBOR data (non-array perl object)"); 1205 ERR ("corrupted CBOR data (non-array perl object)");
945 1206
955 if (!method) 1216 if (!method)
956 ERR ("cannot decode perl-object (package does not have a THAW method)"); 1217 ERR ("cannot decode perl-object (package does not have a THAW method)");
957 1218
958 dSP; 1219 dSP;
959 1220
960 ENTER; SAVETMPS; PUSHMARK (SP); 1221 ENTER; SAVETMPS;
1222 PUSHMARK (SP);
961 EXTEND (SP, len + 1); 1223 EXTEND (SP, len + 1);
962 // we re-bless the reference to get overload and other niceties right 1224 // we re-bless the reference to get overload and other niceties right
963 PUSHs (*av_fetch (av, 0, 1)); 1225 PUSHs (*av_fetch (av, 0, 1));
964 PUSHs (sv_cbor); 1226 PUSHs (sv_cbor);
965 1227
973 SPAGAIN; 1235 SPAGAIN;
974 1236
975 if (SvTRUE (ERRSV)) 1237 if (SvTRUE (ERRSV))
976 { 1238 {
977 FREETMPS; LEAVE; 1239 FREETMPS; LEAVE;
978 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV)))); 1240 ERR_ERRSV;
979 } 1241 }
980 1242
981 SvREFCNT_dec (sv); 1243 SvREFCNT_dec_NN (sv);
982 sv = SvREFCNT_inc (POPs); 1244 sv = SvREFCNT_inc (POPs);
983 1245
984 PUTBACK; 1246 PUTBACK;
985 1247
986 FREETMPS; LEAVE; 1248 FREETMPS; LEAVE;
987 } 1249 }
988 break; 1250 break;
989 1251
990 default: 1252 default:
1253 filter:
991 { 1254 {
1255 SV *tag_sv = newSVuv (tag);
1256
992 sv = decode_sv (dec); 1257 sv = decode_sv (dec);
993 1258
994 dSP; 1259 dSP;
995 ENTER; SAVETMPS; PUSHMARK (SP); 1260 ENTER; SAVETMPS;
1261 PUSHMARK (SP);
996 EXTEND (SP, 2); 1262 EXTEND (SP, 2);
997 PUSHs (newSVuv (tag)); 1263 PUSHs (tag_sv);
998 PUSHs (sv); 1264 PUSHs (sv);
999 1265
1000 PUTBACK; 1266 PUTBACK;
1001 int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL); 1267 int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL);
1002 SPAGAIN; 1268 SPAGAIN;
1003 1269
1004 if (SvTRUE (ERRSV)) 1270 if (SvTRUE (ERRSV))
1005 { 1271 {
1272 SvREFCNT_dec_NN (tag_sv);
1006 FREETMPS; LEAVE; 1273 FREETMPS; LEAVE;
1007 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV)))); 1274 ERR_ERRSV;
1008 } 1275 }
1009 1276
1010 if (count) 1277 if (count)
1011 { 1278 {
1279 SvREFCNT_dec_NN (tag_sv);
1012 SvREFCNT_dec (sv); 1280 SvREFCNT_dec_NN (sv);
1013 sv = SvREFCNT_inc (POPs); 1281 sv = SvREFCNT_inc_NN (TOPs);
1282 SP -= count;
1014 } 1283 }
1015 else 1284 else
1016 { 1285 {
1017 AV *av = newAV (); 1286 AV *av = newAV ();
1018 av_push (av, newSVuv (tag)); 1287 av_push (av, tag_sv);
1019 av_push (av, sv); 1288 av_push (av, sv);
1020 1289
1021 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 1290 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
1022 ? cbor_tagged_stash 1291 ? cbor_tagged_stash
1023 : gv_stashpv ("CBOR::XS::Tagged" , 1); 1292 : gv_stashpv ("CBOR::XS::Tagged" , 1);
1109 1378
1110 return newSVnv (ecb_binary64_to_double (fp)); 1379 return newSVnv (ecb_binary64_to_double (fp));
1111 } 1380 }
1112 1381
1113 // 0..19 unassigned simple 1382 // 0..19 unassigned simple
1114 // 24 reserved + unassigned (reserved values are not encodable) 1383 // 24 reserved + unassigned simple (reserved values are not encodable)
1384 // 28-30 unassigned misc
1385 // 31 break code
1115 default: 1386 default:
1116 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)"); 1387 ERR ("corrupted CBOR data (reserved/unassigned/unexpected major 7 value)");
1117 } 1388 }
1118 1389
1119 break; 1390 break;
1120 } 1391 }
1121 1392
1124} 1395}
1125 1396
1126static SV * 1397static SV *
1127decode_cbor (SV *string, CBOR *cbor, char **offset_return) 1398decode_cbor (SV *string, CBOR *cbor, char **offset_return)
1128{ 1399{
1129 dec_t dec = { }; 1400 dec_t dec = { 0 };
1130 SV *sv; 1401 SV *sv;
1131 STRLEN len; 1402 STRLEN len;
1132 char *data = SvPVbyte (string, len); 1403 char *data = SvPVbyte (string, len);
1133 1404
1134 if (len > cbor->max_size && cbor->max_size) 1405 if (len > cbor->max_size && cbor->max_size)
1148 if (dec.cur != dec.end && !dec.err) 1419 if (dec.cur != dec.end && !dec.err)
1149 dec.err = "garbage after CBOR object"; 1420 dec.err = "garbage after CBOR object";
1150 1421
1151 if (dec.err) 1422 if (dec.err)
1152 { 1423 {
1424 if (dec.shareable)
1425 {
1426 // need to break cyclic links, which would all be in shareable
1427 int i;
1428 SV **svp;
1429
1430 for (i = av_len (dec.shareable) + 1; i--; )
1431 if ((svp = av_fetch (dec.shareable, i, 0)))
1432 sv_setsv (*svp, &PL_sv_undef);
1433 }
1434
1153 SvREFCNT_dec (sv); 1435 SvREFCNT_dec_NN (sv);
1436
1437 if (dec.err_sv)
1438 sv_2mortal (dec.err_sv);
1439
1154 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur); 1440 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur);
1155 } 1441 }
1156 1442
1157 sv = sv_2mortal (sv); 1443 sv = sv_2mortal (sv);
1158 1444
1159 return sv; 1445 return sv;
1160} 1446}
1161 1447
1448/////////////////////////////////////////////////////////////////////////////
1449// incremental parser
1450
1451#define INCR_DONE(cbor) (AvFILLp (cbor->incr_count) < 0)
1452
1453// returns 0 for notyet, 1 for success or error
1454static int
1455incr_parse (CBOR *self, SV *cborstr)
1456{
1457 STRLEN cur;
1458 SvPV (cborstr, cur);
1459
1460 while (ecb_expect_true (self->incr_need <= cur))
1461 {
1462 // table of integer count bytes
1463 static I8 incr_len[MINOR_MASK + 1] = {
1464 0, 0, 0, 0, 0, 0, 0, 0,
1465 0, 0, 0, 0, 0, 0, 0, 0,
1466 0, 0, 0, 0, 0, 0, 0, 0,
1467 1, 2, 4, 8,-1,-1,-1,-2
1468 };
1469
1470 const U8 *p = SvPVX (cborstr) + self->incr_pos;
1471 U8 m = *p & MINOR_MASK;
1472 IV count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1473 I8 ilen = incr_len[m];
1474
1475 self->incr_need = self->incr_pos + 1;
1476
1477 if (ecb_expect_false (ilen < 0))
1478 {
1479 if (m != MINOR_INDEF)
1480 return 1; // error
1481
1482 if (*p == (MAJOR_MISC | MINOR_INDEF))
1483 {
1484 if (count >= 0)
1485 return 1; // error
1486
1487 count = 1;
1488 }
1489 else
1490 {
1491 av_push (self->incr_count, newSViv (-1)); //TODO: nest
1492 count = -1;
1493 }
1494 }
1495 else
1496 {
1497 self->incr_need += ilen;
1498 if (ecb_expect_false (self->incr_need > cur))
1499 return 0;
1500
1501 int major = *p >> MAJOR_SHIFT;
1502
1503 switch (major)
1504 {
1505 case MAJOR_TAG >> MAJOR_SHIFT:
1506 ++count; // tags merely prefix another value
1507 break;
1508
1509 case MAJOR_BYTES >> MAJOR_SHIFT:
1510 case MAJOR_TEXT >> MAJOR_SHIFT:
1511 case MAJOR_ARRAY >> MAJOR_SHIFT:
1512 case MAJOR_MAP >> MAJOR_SHIFT:
1513 {
1514 UV len;
1515
1516 if (ecb_expect_false (ilen))
1517 {
1518 len = 0;
1519
1520 do {
1521 len = (len << 8) | *++p;
1522 } while (--ilen);
1523 }
1524 else
1525 len = m;
1526
1527 switch (major)
1528 {
1529 case MAJOR_BYTES >> MAJOR_SHIFT:
1530 case MAJOR_TEXT >> MAJOR_SHIFT:
1531 self->incr_need += len;
1532 if (ecb_expect_false (self->incr_need > cur))
1533 return 0;
1534
1535 break;
1536
1537 case MAJOR_MAP >> MAJOR_SHIFT:
1538 len <<= 1;
1539 case MAJOR_ARRAY >> MAJOR_SHIFT:
1540 if (len)
1541 {
1542 av_push (self->incr_count, newSViv (len + 1)); //TODO: nest
1543 count = len + 1;
1544 }
1545 break;
1546 }
1547 }
1548 }
1549 }
1550
1551 self->incr_pos = self->incr_need;
1552
1553 if (count > 0)
1554 {
1555 while (!--count)
1556 {
1557 if (!AvFILLp (self->incr_count))
1558 return 1; // done
1559
1560 SvREFCNT_dec_NN (av_pop (self->incr_count));
1561 count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1562 }
1563
1564 SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]) = count;
1565 }
1566 }
1567
1568 return 0;
1569}
1570
1571
1162///////////////////////////////////////////////////////////////////////////// 1572/////////////////////////////////////////////////////////////////////////////
1163// XS interface functions 1573// XS interface functions
1164 1574
1165MODULE = CBOR::XS PACKAGE = CBOR::XS 1575MODULE = CBOR::XS PACKAGE = CBOR::XS
1166 1576
1178 1588
1179 default_filter = newSVpv ("CBOR::XS::default_filter", 0); 1589 default_filter = newSVpv ("CBOR::XS::default_filter", 0);
1180 1590
1181 sv_cbor = newSVpv ("CBOR", 0); 1591 sv_cbor = newSVpv ("CBOR", 0);
1182 SvREADONLY_on (sv_cbor); 1592 SvREADONLY_on (sv_cbor);
1593
1594 assert (("STRLEN must be an unsigned type", 0 <= (STRLEN)-1));
1183} 1595}
1184 1596
1185PROTOTYPES: DISABLE 1597PROTOTYPES: DISABLE
1186 1598
1187void CLONE (...) 1599void CLONE (...)
1206void shrink (CBOR *self, int enable = 1) 1618void shrink (CBOR *self, int enable = 1)
1207 ALIAS: 1619 ALIAS:
1208 shrink = F_SHRINK 1620 shrink = F_SHRINK
1209 allow_unknown = F_ALLOW_UNKNOWN 1621 allow_unknown = F_ALLOW_UNKNOWN
1210 allow_sharing = F_ALLOW_SHARING 1622 allow_sharing = F_ALLOW_SHARING
1623 allow_cycles = F_ALLOW_CYCLES
1624 forbid_objects = F_FORBID_OBJECTS
1211 pack_strings = F_PACK_STRINGS 1625 pack_strings = F_PACK_STRINGS
1626 text_keys = F_TEXT_KEYS
1627 text_strings = F_TEXT_STRINGS
1628 validate_utf8 = F_VALIDATE_UTF8
1212 PPCODE: 1629 PPCODE:
1213{ 1630{
1214 if (enable) 1631 if (enable)
1215 self->flags |= ix; 1632 self->flags |= ix;
1216 else 1633 else
1222void get_shrink (CBOR *self) 1639void get_shrink (CBOR *self)
1223 ALIAS: 1640 ALIAS:
1224 get_shrink = F_SHRINK 1641 get_shrink = F_SHRINK
1225 get_allow_unknown = F_ALLOW_UNKNOWN 1642 get_allow_unknown = F_ALLOW_UNKNOWN
1226 get_allow_sharing = F_ALLOW_SHARING 1643 get_allow_sharing = F_ALLOW_SHARING
1644 get_allow_cycles = F_ALLOW_CYCLES
1645 get_forbid_objects = F_FORBID_OBJECTS
1227 get_pack_strings = F_PACK_STRINGS 1646 get_pack_strings = F_PACK_STRINGS
1647 get_text_keys = F_TEXT_KEYS
1648 get_text_strings = F_TEXT_STRINGS
1649 get_validate_utf8 = F_VALIDATE_UTF8
1228 PPCODE: 1650 PPCODE:
1229 XPUSHs (boolSV (self->flags & ix)); 1651 XPUSHs (boolSV (self->flags & ix));
1230 1652
1231void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) 1653void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
1232 PPCODE: 1654 PPCODE:
1281 EXTEND (SP, 2); 1703 EXTEND (SP, 2);
1282 PUSHs (sv); 1704 PUSHs (sv);
1283 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); 1705 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
1284} 1706}
1285 1707
1708void incr_parse (CBOR *self, SV *cborstr)
1709 ALIAS:
1710 incr_parse_multiple = 1
1711 PPCODE:
1712{
1713 if (SvUTF8 (cborstr))
1714 sv_utf8_downgrade (cborstr, 0);
1715
1716 if (!self->incr_count)
1717 {
1718 self->incr_count = newAV ();
1719 self->incr_pos = 0;
1720 self->incr_need = 1;
1721
1722 av_push (self->incr_count, newSViv (1));
1723 }
1724
1725 do
1726 {
1727 if (!incr_parse (self, cborstr))
1728 {
1729 if (self->incr_need > self->max_size && self->max_size)
1730 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
1731 (unsigned long)self->incr_need, (unsigned long)self->max_size);
1732
1733 break;
1734 }
1735
1736 SV *sv;
1737 char *offset;
1738
1739 PUTBACK; sv = decode_cbor (cborstr, self, &offset); SPAGAIN;
1740 XPUSHs (sv);
1741
1742 sv_chop (cborstr, offset);
1743
1744 av_clear (self->incr_count);
1745 av_push (self->incr_count, newSViv (1));
1746
1747 self->incr_pos = 0;
1748 self->incr_need = self->incr_pos + 1;
1749 }
1750 while (ix);
1751}
1752
1753void incr_reset (CBOR *self)
1754 CODE:
1755{
1756 SvREFCNT_dec (self->incr_count);
1757 self->incr_count = 0;
1758}
1759
1286void DESTROY (CBOR *self) 1760void DESTROY (CBOR *self)
1287 PPCODE: 1761 PPCODE:
1288 cbor_free (self); 1762 cbor_free (self);
1289 1763
1290PROTOTYPES: ENABLE 1764PROTOTYPES: ENABLE
1291 1765
1292void encode_cbor (SV *scalar) 1766void encode_cbor (SV *scalar)
1767 ALIAS:
1768 encode_cbor = 0
1769 encode_cbor_sharing = F_ALLOW_SHARING
1293 PPCODE: 1770 PPCODE:
1294{ 1771{
1295 CBOR cbor; 1772 CBOR cbor;
1296 cbor_init (&cbor); 1773 cbor_init (&cbor);
1774 cbor.flags |= ix;
1297 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN; 1775 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;
1298 XPUSHs (scalar); 1776 XPUSHs (scalar);
1299} 1777}
1300 1778
1301void decode_cbor (SV *cborstr) 1779void decode_cbor (SV *cborstr)
1305 cbor_init (&cbor); 1783 cbor_init (&cbor);
1306 PUTBACK; cborstr = decode_cbor (cborstr, &cbor, 0); SPAGAIN; 1784 PUTBACK; cborstr = decode_cbor (cborstr, &cbor, 0); SPAGAIN;
1307 XPUSHs (cborstr); 1785 XPUSHs (cborstr);
1308} 1786}
1309 1787
1788#ifdef __AFL_COMPILER
1789
1790void
1791afl_init ()
1792 CODE:
1793 __AFL_INIT ();
1794
1795int
1796afl_loop (unsigned int count = 10000)
1797 CODE:
1798 RETVAL = __AFL_LOOP (count);
1799 OUTPUT:
1800 RETVAL
1801
1802#endif
1803

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines