… | |
… | |
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 | |
12 | #define ECB_NO_THREADS 1 |
13 | #define ECB_NO_THREADS 1 |
13 | #include "ecb.h" |
14 | #include "ecb.h" |
14 | |
15 | |
15 | // compatibility with perl <5.18 |
16 | // compatibility with perl <5.18 |
… | |
… | |
20 | # define HvNAMELEN(hv) HvNAMELEN_get (hv) |
21 | # define HvNAMELEN(hv) HvNAMELEN_get (hv) |
21 | #endif |
22 | #endif |
22 | #ifndef HvNAMEUTF8 |
23 | #ifndef HvNAMEUTF8 |
23 | # define HvNAMEUTF8(hv) 0 |
24 | # define HvNAMEUTF8(hv) 0 |
24 | #endif |
25 | #endif |
|
|
26 | #ifndef SvREFCNT_inc_NN |
|
|
27 | # define SvREFCNT_inc_NN(sv) SvREFCNT_inc (sv) |
|
|
28 | #endif |
25 | #ifndef SvREFCNT_dec_NN |
29 | #ifndef SvREFCNT_dec_NN |
26 | # define SvREFCNT_dec_NN(sv) SvREFCNT_dec (sv) |
30 | # define SvREFCNT_dec_NN(sv) SvREFCNT_dec (sv) |
27 | #endif |
31 | #endif |
28 | |
32 | |
29 | // known major and minor types |
33 | // known major and minor types |
… | |
… | |
95 | CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8 |
99 | CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8 |
96 | |
100 | |
97 | CBOR_TAG_MAGIC = 55799, // self-describe cbor |
101 | CBOR_TAG_MAGIC = 55799, // self-describe cbor |
98 | }; |
102 | }; |
99 | |
103 | |
|
|
104 | // known forced types, also hardcoded in CBOR.pm |
|
|
105 | enum |
|
|
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 | |
100 | #define F_SHRINK 0x00000001UL |
117 | #define F_SHRINK 0x00000001UL |
101 | #define F_ALLOW_UNKNOWN 0x00000002UL |
118 | #define F_ALLOW_UNKNOWN 0x00000002UL |
102 | #define F_ALLOW_SHARING 0x00000004UL |
119 | #define F_ALLOW_SHARING 0x00000004UL |
103 | #define F_ALLOW_CYCLES 0x00000008UL |
120 | #define F_ALLOW_CYCLES 0x00000008UL |
|
|
121 | #define F_FORBID_OBJECTS 0x00000010UL |
104 | #define F_PACK_STRINGS 0x00000010UL |
122 | #define F_PACK_STRINGS 0x00000020UL |
105 | #define F_TEXT_KEYS 0x00000020UL |
123 | #define F_TEXT_KEYS 0x00000040UL |
106 | #define F_TEXT_STRINGS 0x00000040UL |
124 | #define F_TEXT_STRINGS 0x00000080UL |
107 | #define F_VALIDATE_UTF8 0x00000080UL |
125 | #define F_VALIDATE_UTF8 0x00000100UL |
108 | |
126 | |
109 | #define INIT_SIZE 32 // initial scalar size to be allocated |
127 | #define INIT_SIZE 32 // initial scalar size to be allocated |
110 | |
128 | |
111 | #define SB do { |
129 | #define SB do { |
112 | #define SE } while (0) |
130 | #define SE } while (0) |
… | |
… | |
185 | |
203 | |
186 | // minimum length of a string to be registered for stringref |
204 | // minimum length of a string to be registered for stringref |
187 | ecb_inline int |
205 | ecb_inline int |
188 | minimum_string_length (UV idx) |
206 | minimum_string_length (UV idx) |
189 | { |
207 | { |
190 | return idx > 23 |
208 | return idx <= 23 ? 3 |
191 | ? idx > 0xffU |
209 | : idx <= 0xffU ? 4 |
192 | ? idx > 0xffffU |
210 | : idx <= 0xffffU ? 5 |
193 | ? idx > 0xffffffffU |
211 | : idx <= 0xffffffffU ? 7 |
194 | ? 11 |
212 | : 11; |
195 | : 7 |
|
|
196 | : 5 |
|
|
197 | : 4 |
|
|
198 | : 3; |
|
|
199 | } |
213 | } |
200 | |
214 | |
201 | ///////////////////////////////////////////////////////////////////////////// |
215 | ///////////////////////////////////////////////////////////////////////////// |
202 | // encoder |
216 | // encoder |
203 | |
217 | |
… | |
… | |
216 | } enc_t; |
230 | } enc_t; |
217 | |
231 | |
218 | ecb_inline void |
232 | ecb_inline void |
219 | need (enc_t *enc, STRLEN len) |
233 | need (enc_t *enc, STRLEN len) |
220 | { |
234 | { |
221 | if (ecb_expect_false (enc->cur + len >= enc->end)) |
235 | if (ecb_expect_false ((uintptr_t)(enc->end - enc->cur) < len)) |
222 | { |
236 | { |
223 | STRLEN cur = enc->cur - (char *)SvPVX (enc->sv); |
237 | STRLEN cur = enc->cur - (char *)SvPVX (enc->sv); |
224 | SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1); |
238 | SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1); |
225 | enc->cur = SvPVX (enc->sv) + cur; |
239 | enc->cur = SvPVX (enc->sv) + cur; |
226 | enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; |
240 | enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; |
… | |
… | |
232 | { |
246 | { |
233 | need (enc, 1); |
247 | need (enc, 1); |
234 | *enc->cur++ = ch; |
248 | *enc->cur++ = ch; |
235 | } |
249 | } |
236 | |
250 | |
|
|
251 | // used for tags, intregers, element counts and so on |
237 | static void |
252 | static void |
238 | encode_uint (enc_t *enc, int major, UV len) |
253 | encode_uint (enc_t *enc, int major, UV len) |
239 | { |
254 | { |
240 | need (enc, 9); |
255 | need (enc, 9); |
241 | |
256 | |
… | |
… | |
272 | *enc->cur++ = len >> 8; |
287 | *enc->cur++ = len >> 8; |
273 | *enc->cur++ = len; |
288 | *enc->cur++ = len; |
274 | } |
289 | } |
275 | } |
290 | } |
276 | |
291 | |
|
|
292 | // encodes a perl value into a CBOR integer |
|
|
293 | ecb_inline void |
|
|
294 | encode_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 | |
277 | ecb_inline void |
304 | ecb_inline void |
278 | encode_tag (enc_t *enc, UV tag) |
305 | encode_tag (enc_t *enc, UV tag) |
279 | { |
306 | { |
280 | encode_uint (enc, MAJOR_TAG, tag); |
307 | encode_uint (enc, MAJOR_TAG, tag); |
281 | } |
308 | } |
… | |
… | |
342 | } |
369 | } |
343 | |
370 | |
344 | encode_str (enc, upgrade_utf8, utf8, str, len); |
371 | encode_str (enc, upgrade_utf8, utf8, str, len); |
345 | } |
372 | } |
346 | |
373 | |
|
|
374 | ecb_inline void |
|
|
375 | encode_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 | |
|
|
390 | ecb_inline void |
|
|
391 | encode_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 | |
|
|
406 | ecb_inline void |
|
|
407 | encode_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 | |
|
|
422 | ecb_inline void |
|
|
423 | encode_bool (enc_t *enc, int istrue) |
|
|
424 | { |
|
|
425 | encode_ch (enc, istrue ? MAJOR_MISC | SIMPLE_TRUE : MAJOR_MISC | SIMPLE_FALSE); |
|
|
426 | } |
|
|
427 | |
|
|
428 | ecb_inline void |
|
|
429 | encode_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 | } |
|
|
469 | } |
|
|
470 | |
347 | static void encode_sv (enc_t *enc, SV *sv); |
471 | static void encode_sv (enc_t *enc, SV *sv); |
348 | |
472 | |
349 | static void |
473 | static void |
350 | encode_av (enc_t *enc, AV *av) |
474 | encode_av (enc_t *enc, AV *av) |
351 | { |
475 | { |
… | |
… | |
356 | |
480 | |
357 | ++enc->depth; |
481 | ++enc->depth; |
358 | |
482 | |
359 | encode_uint (enc, MAJOR_ARRAY, len + 1); |
483 | encode_uint (enc, MAJOR_ARRAY, len + 1); |
360 | |
484 | |
361 | if (SvMAGICAL (av)) |
485 | if (ecb_expect_false (SvMAGICAL (av))) |
362 | for (i = 0; i <= len; ++i) |
486 | for (i = 0; i <= len; ++i) |
363 | { |
487 | { |
364 | SV **svp = av_fetch (av, i, 0); |
488 | SV **svp = av_fetch (av, i, 0); |
365 | encode_sv (enc, svp ? *svp : &PL_sv_undef); |
489 | encode_sv (enc, svp ? *svp : &PL_sv_undef); |
366 | } |
490 | } |
… | |
… | |
385 | ++enc->depth; |
509 | ++enc->depth; |
386 | |
510 | |
387 | int pairs = hv_iterinit (hv); |
511 | int pairs = hv_iterinit (hv); |
388 | int mg = SvMAGICAL (hv); |
512 | int mg = SvMAGICAL (hv); |
389 | |
513 | |
390 | if (mg) |
514 | if (ecb_expect_false (mg)) |
391 | encode_ch (enc, MAJOR_MAP | MINOR_INDEF); |
515 | encode_ch (enc, MAJOR_MAP | MINOR_INDEF); |
392 | else |
516 | else |
393 | encode_uint (enc, MAJOR_MAP, pairs); |
517 | encode_uint (enc, MAJOR_MAP, pairs); |
394 | |
518 | |
395 | while ((he = hv_iternext (hv))) |
519 | while ((he = hv_iternext (hv))) |
… | |
… | |
400 | encode_strref (enc, enc->cbor.flags & (F_TEXT_KEYS | F_TEXT_STRINGS), 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)); |
401 | |
525 | |
402 | 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)); |
403 | } |
527 | } |
404 | |
528 | |
405 | if (mg) |
529 | if (ecb_expect_false (mg)) |
406 | encode_ch (enc, MAJOR_MISC | MINOR_INDEF); |
530 | encode_ch (enc, MAJOR_MISC | MINOR_INDEF); |
407 | |
531 | |
408 | --enc->depth; |
532 | --enc->depth; |
409 | } |
533 | } |
410 | |
534 | |
… | |
… | |
430 | |
554 | |
431 | HV *stash = SvSTASH (sv); |
555 | HV *stash = SvSTASH (sv); |
432 | |
556 | |
433 | if (stash == boolean_stash) |
557 | if (stash == boolean_stash) |
434 | { |
558 | { |
435 | encode_ch (enc, SvIV (sv) ? MAJOR_MISC | SIMPLE_TRUE : MAJOR_MISC | SIMPLE_FALSE); |
559 | encode_bool (enc, SvIV (sv)); |
436 | return; |
560 | return; |
437 | } |
561 | } |
438 | else if (stash == error_stash) |
562 | else if (stash == error_stash) |
439 | { |
563 | { |
440 | encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF); |
564 | encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF); |
… | |
… | |
443 | else if (stash == tagged_stash) |
567 | else if (stash == tagged_stash) |
444 | { |
568 | { |
445 | if (svt != SVt_PVAV) |
569 | if (svt != SVt_PVAV) |
446 | croak ("encountered CBOR::XS::Tagged object that isn't an array"); |
570 | croak ("encountered CBOR::XS::Tagged object that isn't an array"); |
447 | |
571 | |
|
|
572 | switch (av_len ((AV *)sv)) |
|
|
573 | { |
|
|
574 | case 2-1: |
|
|
575 | // actually a tagged value |
448 | 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))); |
449 | 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 | } |
450 | |
588 | |
451 | return; |
589 | return; |
452 | } |
590 | } |
453 | } |
591 | } |
454 | |
592 | |
455 | if (ecb_expect_false (SvREFCNT (sv) > 1) |
593 | if (ecb_expect_false (SvREFCNT (sv) > 1) |
456 | && ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING)) |
594 | && ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING)) |
457 | { |
595 | { |
458 | if (!enc->shareable) |
596 | if (ecb_expect_false (!enc->shareable)) |
459 | enc->shareable = (HV *)sv_2mortal ((SV *)newHV ()); |
597 | enc->shareable = (HV *)sv_2mortal ((SV *)newHV ()); |
460 | |
598 | |
461 | SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1); |
599 | SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1); |
462 | |
600 | |
463 | if (SvOK (*svp)) |
601 | if (SvOK (*svp)) |
… | |
… | |
477 | if (ecb_expect_false (SvOBJECT (sv))) |
615 | if (ecb_expect_false (SvOBJECT (sv))) |
478 | { |
616 | { |
479 | HV *stash = SvSTASH (sv); |
617 | HV *stash = SvSTASH (sv); |
480 | GV *method; |
618 | GV *method; |
481 | |
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)))); |
482 | if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0))) |
623 | else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0))) |
483 | { |
624 | { |
484 | dSP; |
625 | dSP; |
485 | |
626 | |
486 | ENTER; SAVETMPS; |
627 | ENTER; SAVETMPS; |
487 | PUSHMARK (SP); |
628 | PUSHMARK (SP); |
… | |
… | |
506 | else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0) |
647 | else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0) |
507 | { |
648 | { |
508 | dSP; |
649 | dSP; |
509 | |
650 | |
510 | ENTER; SAVETMPS; |
651 | ENTER; SAVETMPS; |
511 | SAVESTACK_POS (); |
|
|
512 | PUSHMARK (SP); |
652 | PUSHMARK (SP); |
513 | EXTEND (SP, 2); |
653 | EXTEND (SP, 2); |
514 | // 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 |
515 | PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); |
655 | PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); |
516 | PUSHs (sv_cbor); |
656 | PUSHs (sv_cbor); |
… | |
… | |
525 | |
665 | |
526 | encode_tag (enc, CBOR_TAG_PERL_OBJECT); |
666 | encode_tag (enc, CBOR_TAG_PERL_OBJECT); |
527 | encode_uint (enc, MAJOR_ARRAY, count + 1); |
667 | encode_uint (enc, MAJOR_ARRAY, count + 1); |
528 | encode_strref (enc, 0, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash)); |
668 | encode_strref (enc, 0, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash)); |
529 | |
669 | |
530 | while (count) |
670 | { |
|
|
671 | int i; |
|
|
672 | |
|
|
673 | for (i = 0; i < count; ++i) |
531 | encode_sv (enc, SP[1 - count--]); |
674 | encode_sv (enc, SP[i + 1 - count]); |
|
|
675 | |
|
|
676 | SP -= count; |
|
|
677 | } |
532 | |
678 | |
533 | PUTBACK; |
679 | PUTBACK; |
534 | |
680 | |
535 | FREETMPS; LEAVE; |
681 | FREETMPS; LEAVE; |
536 | } |
682 | } |
… | |
… | |
558 | |
704 | |
559 | if (ecb_expect_false (nv == (NV)(U32)nv)) |
705 | if (ecb_expect_false (nv == (NV)(U32)nv)) |
560 | encode_uint (enc, MAJOR_POS_INT, (U32)nv); |
706 | encode_uint (enc, MAJOR_POS_INT, (U32)nv); |
561 | //TODO: maybe I32? |
707 | //TODO: maybe I32? |
562 | else if (ecb_expect_false (nv == (float)nv)) |
708 | else if (ecb_expect_false (nv == (float)nv)) |
563 | { |
709 | encode_float32 (enc, nv); |
564 | uint32_t fp = ecb_float_to_binary32 (nv); |
|
|
565 | |
|
|
566 | *enc->cur++ = MAJOR_MISC | MISC_FLOAT32; |
|
|
567 | |
|
|
568 | if (!ecb_big_endian ()) |
|
|
569 | fp = ecb_bswap32 (fp); |
|
|
570 | |
|
|
571 | memcpy (enc->cur, &fp, 4); |
|
|
572 | enc->cur += 4; |
|
|
573 | } |
|
|
574 | else |
710 | else |
575 | { |
711 | encode_float64 (enc, nv); |
576 | uint64_t fp = ecb_double_to_binary64 (nv); |
|
|
577 | |
|
|
578 | *enc->cur++ = MAJOR_MISC | MISC_FLOAT64; |
|
|
579 | |
|
|
580 | if (!ecb_big_endian ()) |
|
|
581 | fp = ecb_bswap64 (fp); |
|
|
582 | |
|
|
583 | memcpy (enc->cur, &fp, 8); |
|
|
584 | enc->cur += 8; |
|
|
585 | } |
|
|
586 | } |
712 | } |
587 | |
713 | |
588 | static void |
714 | static void |
589 | encode_sv (enc_t *enc, SV *sv) |
715 | encode_sv (enc_t *enc, SV *sv) |
590 | { |
716 | { |
… | |
… | |
597 | encode_strref (enc, enc->cbor.flags & F_TEXT_STRINGS, SvUTF8 (sv), str, len); |
723 | encode_strref (enc, enc->cbor.flags & F_TEXT_STRINGS, SvUTF8 (sv), str, len); |
598 | } |
724 | } |
599 | else if (SvNOKp (sv)) |
725 | else if (SvNOKp (sv)) |
600 | encode_nv (enc, sv); |
726 | encode_nv (enc, sv); |
601 | else if (SvIOKp (sv)) |
727 | else if (SvIOKp (sv)) |
602 | { |
728 | encode_int (enc, sv); |
603 | if (SvIsUV (sv)) |
|
|
604 | encode_uint (enc, MAJOR_POS_INT, SvUVX (sv)); |
|
|
605 | else if (SvIVX (sv) >= 0) |
|
|
606 | encode_uint (enc, MAJOR_POS_INT, SvIVX (sv)); |
|
|
607 | else |
|
|
608 | encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1)); |
|
|
609 | } |
|
|
610 | else if (SvROK (sv)) |
729 | else if (SvROK (sv)) |
611 | encode_rv (enc, SvRV (sv)); |
730 | encode_rv (enc, SvRV (sv)); |
612 | else if (!SvOK (sv)) |
731 | else if (!SvOK (sv)) |
613 | encode_ch (enc, MAJOR_MISC | SIMPLE_NULL); |
732 | encode_ch (enc, MAJOR_MISC | SIMPLE_NULL); |
614 | else if (enc->cbor.flags & F_ALLOW_UNKNOWN) |
733 | else if (enc->cbor.flags & F_ALLOW_UNKNOWN) |
… | |
… | |
621 | static SV * |
740 | static SV * |
622 | encode_cbor (SV *scalar, CBOR *cbor) |
741 | encode_cbor (SV *scalar, CBOR *cbor) |
623 | { |
742 | { |
624 | enc_t enc = { 0 }; |
743 | enc_t enc = { 0 }; |
625 | |
744 | |
626 | enc.cbor = *cbor; |
745 | enc.cbor = *cbor; |
627 | enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); |
746 | enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); |
628 | enc.cur = SvPVX (enc.sv); |
747 | enc.cur = SvPVX (enc.sv); |
629 | enc.end = SvEND (enc.sv); |
748 | enc.end = SvEND (enc.sv); |
630 | |
749 | |
631 | SvPOK_only (enc.sv); |
750 | SvPOK_only (enc.sv); |
632 | |
751 | |
633 | if (cbor->flags & F_PACK_STRINGS) |
752 | if (cbor->flags & F_PACK_STRINGS) |
634 | { |
753 | { |
… | |
… | |
694 | err_unexpected_end (dec_t *dec) |
813 | err_unexpected_end (dec_t *dec) |
695 | { |
814 | { |
696 | err_set (dec, "unexpected end of CBOR data"); |
815 | err_set (dec, "unexpected end of CBOR data"); |
697 | } |
816 | } |
698 | |
817 | |
699 | ecb_cold static void |
|
|
700 | err_nesting_exceeded (dec_t *dec) |
|
|
701 | { |
|
|
702 | err_set (dec, ERR_NESTING_EXCEEDED); |
|
|
703 | } |
|
|
704 | |
|
|
705 | #define ERR_DO(do) SB do; goto fail; SE |
818 | #define ERR_DO(do) SB do; goto fail; SE |
706 | #define ERR(reason) ERR_DO (err_set (dec, reason)) |
819 | #define ERR(reason) ERR_DO (err_set (dec, reason)) |
707 | #define ERR_ERRSV ERR_DO (err_errsv (dec)) |
820 | #define ERR_ERRSV ERR_DO (err_errsv (dec)) |
708 | |
821 | |
709 | #define WANT(len) if (ecb_expect_false ((UV)(dec->end - dec->cur) < (UV)len)) ERR_DO (err_unexpected_end (dec)) |
822 | #define WANT(len) if (ecb_expect_false ((uintptr_t)(dec->end - dec->cur) < (STRLEN)len)) ERR_DO (err_unexpected_end (dec)) |
710 | |
823 | |
711 | #define DEC_INC_DEPTH if (ecb_expect_false (++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) |
712 | #define DEC_DEC_DEPTH --dec->depth |
825 | #define DEC_DEC_DEPTH --dec->depth |
713 | |
826 | |
714 | static UV |
827 | static UV |
… | |
… | |
782 | |
895 | |
783 | for (;;) |
896 | for (;;) |
784 | { |
897 | { |
785 | WANT (1); |
898 | WANT (1); |
786 | |
899 | |
787 | if (*dec->cur == (MAJOR_MISC | MINOR_INDEF)) |
900 | if (*dec->cur == (MAJOR_MISC | MINOR_INDEF) || dec->err) |
788 | { |
901 | { |
789 | ++dec->cur; |
902 | ++dec->cur; |
790 | break; |
903 | break; |
791 | } |
904 | } |
792 | |
905 | |
… | |
… | |
806 | |
919 | |
807 | DEC_DEC_DEPTH; |
920 | DEC_DEC_DEPTH; |
808 | return newRV_noinc ((SV *)av); |
921 | return newRV_noinc ((SV *)av); |
809 | |
922 | |
810 | fail: |
923 | fail: |
811 | SvREFCNT_dec (av); |
924 | SvREFCNT_dec_NN (av); |
812 | DEC_DEC_DEPTH; |
925 | DEC_DEC_DEPTH; |
813 | return &PL_sv_undef; |
926 | return &PL_sv_undef; |
814 | } |
927 | } |
815 | |
928 | |
816 | static void |
929 | static void |
… | |
… | |
879 | |
992 | |
880 | return; |
993 | return; |
881 | } |
994 | } |
882 | |
995 | |
883 | hv_store_ent (hv, k, v, 0); |
996 | hv_store_ent (hv, k, v, 0); |
884 | SvREFCNT_dec (k); |
997 | SvREFCNT_dec_NN (k); |
885 | |
998 | |
886 | fail: |
999 | fail: |
887 | ; |
1000 | ; |
888 | } |
1001 | } |
889 | |
1002 | |
… | |
… | |
900 | |
1013 | |
901 | for (;;) |
1014 | for (;;) |
902 | { |
1015 | { |
903 | WANT (1); |
1016 | WANT (1); |
904 | |
1017 | |
905 | if (*dec->cur == (MAJOR_MISC | MINOR_INDEF)) |
1018 | if (*dec->cur == (MAJOR_MISC | MINOR_INDEF) || dec->err) |
906 | { |
1019 | { |
907 | ++dec->cur; |
1020 | ++dec->cur; |
908 | break; |
1021 | break; |
909 | } |
1022 | } |
910 | |
1023 | |
… | |
… | |
923 | |
1036 | |
924 | DEC_DEC_DEPTH; |
1037 | DEC_DEC_DEPTH; |
925 | return newRV_noinc ((SV *)hv); |
1038 | return newRV_noinc ((SV *)hv); |
926 | |
1039 | |
927 | fail: |
1040 | fail: |
928 | SvREFCNT_dec (hv); |
1041 | SvREFCNT_dec_NN (hv); |
929 | DEC_DEC_DEPTH; |
1042 | DEC_DEC_DEPTH; |
930 | return &PL_sv_undef; |
1043 | return &PL_sv_undef; |
931 | } |
1044 | } |
932 | |
1045 | |
933 | static SV * |
1046 | static SV * |
934 | decode_str (dec_t *dec, int utf8) |
1047 | decode_str (dec_t *dec, int utf8) |
935 | { |
1048 | { |
936 | SV *sv = 0; |
1049 | SV *sv = 0; |
937 | |
1050 | |
938 | if ((*dec->cur & MINOR_MASK) == MINOR_INDEF) |
1051 | if (ecb_expect_false ((*dec->cur & MINOR_MASK) == MINOR_INDEF)) |
939 | { |
1052 | { |
940 | // indefinite length strings |
1053 | // indefinite length strings |
941 | ++dec->cur; |
1054 | ++dec->cur; |
942 | |
1055 | |
943 | U8 major = *dec->cur & MAJOR_MISC; |
1056 | U8 major = *dec->cur & MAJOR_MISC; |
… | |
… | |
1011 | sv = newRV_noinc (decode_sv (dec)); |
1124 | sv = newRV_noinc (decode_sv (dec)); |
1012 | break; |
1125 | break; |
1013 | |
1126 | |
1014 | case CBOR_TAG_STRINGREF_NAMESPACE: |
1127 | case CBOR_TAG_STRINGREF_NAMESPACE: |
1015 | { |
1128 | { |
1016 | // do nmot use SAVETMPS/FREETMPS, as these will |
1129 | // do not use SAVETMPS/FREETMPS, as these will |
1017 | // erase mortalised caches, e.g. "shareable" |
1130 | // erase mortalised caches, e.g. "shareable" |
1018 | ENTER; |
1131 | ENTER; |
1019 | |
1132 | |
1020 | SAVESPTR (dec->stringref); |
1133 | SAVESPTR (dec->stringref); |
1021 | dec->stringref = (AV *)sv_2mortal ((SV *)newAV ()); |
1134 | dec->stringref = (AV *)sv_2mortal ((SV *)newAV ()); |
… | |
… | |
1031 | if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT)) |
1144 | if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT)) |
1032 | ERR ("corrupted CBOR data (stringref index not an unsigned integer)"); |
1145 | ERR ("corrupted CBOR data (stringref index not an unsigned integer)"); |
1033 | |
1146 | |
1034 | UV idx = decode_uint (dec); |
1147 | UV idx = decode_uint (dec); |
1035 | |
1148 | |
1036 | if (!dec->stringref || (int)idx > AvFILLp (dec->stringref)) |
1149 | if (!dec->stringref || idx >= (UV)(1 + AvFILLp (dec->stringref))) |
1037 | 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)"); |
1038 | |
1151 | |
1039 | sv = newSVsv (AvARRAY (dec->stringref)[idx]); |
1152 | sv = newSVsv (AvARRAY (dec->stringref)[idx]); |
1040 | } |
1153 | } |
1041 | break; |
1154 | break; |
… | |
… | |
1069 | if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT)) |
1182 | if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT)) |
1070 | ERR ("corrupted CBOR data (sharedref index not an unsigned integer)"); |
1183 | ERR ("corrupted CBOR data (sharedref index not an unsigned integer)"); |
1071 | |
1184 | |
1072 | UV idx = decode_uint (dec); |
1185 | UV idx = decode_uint (dec); |
1073 | |
1186 | |
1074 | if (!dec->shareable || (int)idx > AvFILLp (dec->shareable)) |
1187 | if (!dec->shareable || idx >= (UV)(1 + AvFILLp (dec->shareable))) |
1075 | ERR ("corrupted CBOR data (sharedref index out of bounds)"); |
1188 | ERR ("corrupted CBOR data (sharedref index out of bounds)"); |
1076 | |
1189 | |
1077 | sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]); |
1190 | sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]); |
1078 | |
1191 | |
1079 | if (sv == &PL_sv_undef) |
1192 | if (sv == &PL_sv_undef) |
… | |
… | |
1081 | } |
1194 | } |
1082 | break; |
1195 | break; |
1083 | |
1196 | |
1084 | case CBOR_TAG_PERL_OBJECT: |
1197 | case CBOR_TAG_PERL_OBJECT: |
1085 | { |
1198 | { |
|
|
1199 | if (dec->cbor.flags & F_FORBID_OBJECTS) |
|
|
1200 | goto filter; |
|
|
1201 | |
1086 | sv = decode_sv (dec); |
1202 | sv = decode_sv (dec); |
1087 | |
1203 | |
1088 | if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV) |
1204 | if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV) |
1089 | ERR ("corrupted CBOR data (non-array perl object)"); |
1205 | ERR ("corrupted CBOR data (non-array perl object)"); |
1090 | |
1206 | |
… | |
… | |
1122 | { |
1238 | { |
1123 | FREETMPS; LEAVE; |
1239 | FREETMPS; LEAVE; |
1124 | ERR_ERRSV; |
1240 | ERR_ERRSV; |
1125 | } |
1241 | } |
1126 | |
1242 | |
1127 | SvREFCNT_dec (sv); |
1243 | SvREFCNT_dec_NN (sv); |
1128 | sv = SvREFCNT_inc (POPs); |
1244 | sv = SvREFCNT_inc (POPs); |
1129 | |
1245 | |
1130 | PUTBACK; |
1246 | PUTBACK; |
1131 | |
1247 | |
1132 | FREETMPS; LEAVE; |
1248 | FREETMPS; LEAVE; |
1133 | } |
1249 | } |
1134 | break; |
1250 | break; |
1135 | |
1251 | |
1136 | default: |
1252 | default: |
|
|
1253 | filter: |
1137 | { |
1254 | { |
1138 | SV *tag_sv = newSVuv (tag); |
1255 | SV *tag_sv = newSVuv (tag); |
1139 | |
1256 | |
1140 | sv = decode_sv (dec); |
1257 | sv = decode_sv (dec); |
1141 | |
1258 | |
1142 | dSP; |
1259 | dSP; |
1143 | ENTER; SAVETMPS; |
1260 | ENTER; SAVETMPS; |
1144 | SAVESTACK_POS (); |
|
|
1145 | PUSHMARK (SP); |
1261 | PUSHMARK (SP); |
1146 | EXTEND (SP, 2); |
1262 | EXTEND (SP, 2); |
1147 | PUSHs (tag_sv); |
1263 | PUSHs (tag_sv); |
1148 | PUSHs (sv); |
1264 | PUSHs (sv); |
1149 | |
1265 | |
… | |
… | |
1151 | 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); |
1152 | SPAGAIN; |
1268 | SPAGAIN; |
1153 | |
1269 | |
1154 | if (SvTRUE (ERRSV)) |
1270 | if (SvTRUE (ERRSV)) |
1155 | { |
1271 | { |
1156 | SvREFCNT_dec (tag_sv); |
1272 | SvREFCNT_dec_NN (tag_sv); |
1157 | FREETMPS; LEAVE; |
1273 | FREETMPS; LEAVE; |
1158 | ERR_ERRSV; |
1274 | ERR_ERRSV; |
1159 | } |
1275 | } |
1160 | |
1276 | |
1161 | if (count) |
1277 | if (count) |
1162 | { |
1278 | { |
1163 | SvREFCNT_dec (tag_sv); |
1279 | SvREFCNT_dec_NN (tag_sv); |
1164 | SvREFCNT_dec (sv); |
1280 | SvREFCNT_dec_NN (sv); |
1165 | sv = SvREFCNT_inc (POPs); |
1281 | sv = SvREFCNT_inc_NN (TOPs); |
|
|
1282 | SP -= count; |
1166 | } |
1283 | } |
1167 | else |
1284 | else |
1168 | { |
1285 | { |
1169 | AV *av = newAV (); |
1286 | AV *av = newAV (); |
1170 | av_push (av, tag_sv); |
1287 | av_push (av, tag_sv); |
… | |
… | |
1313 | for (i = av_len (dec.shareable) + 1; i--; ) |
1430 | for (i = av_len (dec.shareable) + 1; i--; ) |
1314 | if ((svp = av_fetch (dec.shareable, i, 0))) |
1431 | if ((svp = av_fetch (dec.shareable, i, 0))) |
1315 | sv_setsv (*svp, &PL_sv_undef); |
1432 | sv_setsv (*svp, &PL_sv_undef); |
1316 | } |
1433 | } |
1317 | |
1434 | |
1318 | SvREFCNT_dec (sv); |
1435 | SvREFCNT_dec_NN (sv); |
1319 | |
1436 | |
1320 | if (dec.err_sv) |
1437 | if (dec.err_sv) |
1321 | sv_2mortal (dec.err_sv); |
1438 | sv_2mortal (dec.err_sv); |
1322 | |
1439 | |
1323 | 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); |
… | |
… | |
1502 | ALIAS: |
1619 | ALIAS: |
1503 | shrink = F_SHRINK |
1620 | shrink = F_SHRINK |
1504 | allow_unknown = F_ALLOW_UNKNOWN |
1621 | allow_unknown = F_ALLOW_UNKNOWN |
1505 | allow_sharing = F_ALLOW_SHARING |
1622 | allow_sharing = F_ALLOW_SHARING |
1506 | allow_cycles = F_ALLOW_CYCLES |
1623 | allow_cycles = F_ALLOW_CYCLES |
|
|
1624 | forbid_objects = F_FORBID_OBJECTS |
1507 | pack_strings = F_PACK_STRINGS |
1625 | pack_strings = F_PACK_STRINGS |
1508 | text_keys = F_TEXT_KEYS |
1626 | text_keys = F_TEXT_KEYS |
1509 | text_strings = F_TEXT_STRINGS |
1627 | text_strings = F_TEXT_STRINGS |
1510 | validate_utf8 = F_VALIDATE_UTF8 |
1628 | validate_utf8 = F_VALIDATE_UTF8 |
1511 | PPCODE: |
1629 | PPCODE: |
… | |
… | |
1522 | ALIAS: |
1640 | ALIAS: |
1523 | get_shrink = F_SHRINK |
1641 | get_shrink = F_SHRINK |
1524 | get_allow_unknown = F_ALLOW_UNKNOWN |
1642 | get_allow_unknown = F_ALLOW_UNKNOWN |
1525 | get_allow_sharing = F_ALLOW_SHARING |
1643 | get_allow_sharing = F_ALLOW_SHARING |
1526 | get_allow_cycles = F_ALLOW_CYCLES |
1644 | get_allow_cycles = F_ALLOW_CYCLES |
|
|
1645 | get_forbid_objects = F_FORBID_OBJECTS |
1527 | get_pack_strings = F_PACK_STRINGS |
1646 | get_pack_strings = F_PACK_STRINGS |
1528 | get_text_keys = F_TEXT_KEYS |
1647 | get_text_keys = F_TEXT_KEYS |
1529 | get_text_strings = F_TEXT_STRINGS |
1648 | get_text_strings = F_TEXT_STRINGS |
1530 | get_validate_utf8 = F_VALIDATE_UTF8 |
1649 | get_validate_utf8 = F_VALIDATE_UTF8 |
1531 | PPCODE: |
1650 | PPCODE: |
… | |
… | |
1664 | cbor_init (&cbor); |
1783 | cbor_init (&cbor); |
1665 | PUTBACK; cborstr = decode_cbor (cborstr, &cbor, 0); SPAGAIN; |
1784 | PUTBACK; cborstr = decode_cbor (cborstr, &cbor, 0); SPAGAIN; |
1666 | XPUSHs (cborstr); |
1785 | XPUSHs (cborstr); |
1667 | } |
1786 | } |
1668 | |
1787 | |
|
|
1788 | #ifdef __AFL_COMPILER |
|
|
1789 | |
|
|
1790 | void |
|
|
1791 | afl_init () |
|
|
1792 | CODE: |
|
|
1793 | __AFL_INIT (); |
|
|
1794 | |
|
|
1795 | int |
|
|
1796 | afl_loop (unsigned int count = 10000) |
|
|
1797 | CODE: |
|
|
1798 | RETVAL = __AFL_LOOP (count); |
|
|
1799 | OUTPUT: |
|
|
1800 | RETVAL |
|
|
1801 | |
|
|
1802 | #endif |
|
|
1803 | |