… | |
… | |
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 |
|
|
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))) |
28 | |
35 | |
29 | // known major and minor types |
36 | // known major and minor types |
30 | enum cbor_type |
37 | enum cbor_type |
31 | { |
38 | { |
32 | MAJOR_SHIFT = 5, |
39 | MAJOR_SHIFT = 5, |
… | |
… | |
93 | CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8 |
100 | CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8 |
94 | CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8 |
101 | CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8 |
95 | CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8 |
102 | CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8 |
96 | |
103 | |
97 | CBOR_TAG_MAGIC = 55799, // self-describe cbor |
104 | CBOR_TAG_MAGIC = 55799, // self-describe cbor |
|
|
105 | }; |
|
|
106 | |
|
|
107 | // known forced types, also hardcoded in CBOR.pm |
|
|
108 | enum |
|
|
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 |
98 | }; |
119 | }; |
99 | |
120 | |
100 | #define F_SHRINK 0x00000001UL |
121 | #define F_SHRINK 0x00000001UL |
101 | #define F_ALLOW_UNKNOWN 0x00000002UL |
122 | #define F_ALLOW_UNKNOWN 0x00000002UL |
102 | #define F_ALLOW_SHARING 0x00000004UL |
123 | #define F_ALLOW_SHARING 0x00000004UL |
… | |
… | |
183 | #endif |
204 | #endif |
184 | } |
205 | } |
185 | } |
206 | } |
186 | |
207 | |
187 | // minimum length of a string to be registered for stringref |
208 | // minimum length of a string to be registered for stringref |
188 | ecb_inline int |
209 | ecb_inline STRLEN |
189 | minimum_string_length (UV idx) |
210 | minimum_string_length (UV idx) |
190 | { |
211 | { |
191 | return idx > 23 |
212 | return idx <= 23 ? 3 |
192 | ? idx > 0xffU |
213 | : idx <= 0xffU ? 4 |
193 | ? idx > 0xffffU |
214 | : idx <= 0xffffU ? 5 |
194 | ? idx > 0xffffffffU |
215 | : idx <= 0xffffffffU ? 7 |
195 | ? 11 |
216 | : 11; |
196 | : 7 |
|
|
197 | : 5 |
|
|
198 | : 4 |
|
|
199 | : 3; |
|
|
200 | } |
217 | } |
201 | |
218 | |
202 | ///////////////////////////////////////////////////////////////////////////// |
219 | ///////////////////////////////////////////////////////////////////////////// |
203 | // encoder |
220 | // encoder |
204 | |
221 | |
… | |
… | |
217 | } enc_t; |
234 | } enc_t; |
218 | |
235 | |
219 | ecb_inline void |
236 | ecb_inline void |
220 | need (enc_t *enc, STRLEN len) |
237 | need (enc_t *enc, STRLEN len) |
221 | { |
238 | { |
222 | if (ecb_expect_false (enc->cur + len >= enc->end)) |
239 | if (ecb_expect_false ((uintptr_t)(enc->end - enc->cur) < len)) |
223 | { |
240 | { |
224 | STRLEN cur = enc->cur - (char *)SvPVX (enc->sv); |
241 | STRLEN cur = enc->cur - (char *)SvPVX (enc->sv); |
225 | SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1); |
242 | SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1); |
226 | enc->cur = SvPVX (enc->sv) + cur; |
243 | enc->cur = SvPVX (enc->sv) + cur; |
227 | enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; |
244 | enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; |
228 | } |
245 | } |
229 | } |
246 | } |
230 | |
247 | |
|
|
248 | static void encode_sv (enc_t *enc, SV *sv); |
|
|
249 | |
231 | ecb_inline void |
250 | ecb_inline void |
232 | encode_ch (enc_t *enc, char ch) |
251 | encode_ch (enc_t *enc, char ch) |
233 | { |
252 | { |
234 | need (enc, 1); |
253 | need (enc, 1); |
235 | *enc->cur++ = ch; |
254 | *enc->cur++ = ch; |
236 | } |
255 | } |
237 | |
256 | |
|
|
257 | // used for tags, intregers, element counts and so on |
238 | static void |
258 | static void |
239 | encode_uint (enc_t *enc, int major, UV len) |
259 | encode_uint (enc_t *enc, int major, UV len) |
240 | { |
260 | { |
241 | need (enc, 9); |
261 | need (enc, 9); |
242 | |
262 | |
… | |
… | |
273 | *enc->cur++ = len >> 8; |
293 | *enc->cur++ = len >> 8; |
274 | *enc->cur++ = len; |
294 | *enc->cur++ = len; |
275 | } |
295 | } |
276 | } |
296 | } |
277 | |
297 | |
|
|
298 | // encodes a perl value into a CBOR integer |
|
|
299 | ecb_inline void |
|
|
300 | encode_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 | |
278 | ecb_inline void |
310 | ecb_inline void |
279 | encode_tag (enc_t *enc, UV tag) |
311 | encode_tag (enc_t *enc, UV tag) |
280 | { |
312 | { |
281 | encode_uint (enc, MAJOR_TAG, tag); |
313 | encode_uint (enc, MAJOR_TAG, tag); |
282 | } |
314 | } |
… | |
… | |
343 | } |
375 | } |
344 | |
376 | |
345 | encode_str (enc, upgrade_utf8, utf8, str, len); |
377 | encode_str (enc, upgrade_utf8, utf8, str, len); |
346 | } |
378 | } |
347 | |
379 | |
348 | static void encode_sv (enc_t *enc, SV *sv); |
380 | ecb_inline void |
|
|
381 | encode_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 | |
|
|
396 | ecb_inline void |
|
|
397 | encode_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 | |
|
|
412 | ecb_inline void |
|
|
413 | encode_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 | |
|
|
428 | ecb_inline void |
|
|
429 | encode_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 |
|
|
435 | ecb_inline void |
|
|
436 | encode_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 | |
|
|
466 | ecb_inline void |
|
|
467 | encode_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 | } |
349 | |
510 | |
350 | static void |
511 | static void |
351 | encode_av (enc_t *enc, AV *av) |
512 | encode_av (enc_t *enc, AV *av) |
352 | { |
513 | { |
353 | int i, len = av_len (av); |
514 | int i, len = av_len (av); |
… | |
… | |
357 | |
518 | |
358 | ++enc->depth; |
519 | ++enc->depth; |
359 | |
520 | |
360 | encode_uint (enc, MAJOR_ARRAY, len + 1); |
521 | encode_uint (enc, MAJOR_ARRAY, len + 1); |
361 | |
522 | |
362 | if (SvMAGICAL (av)) |
523 | if (ecb_expect_false (SvMAGICAL (av))) |
363 | for (i = 0; i <= len; ++i) |
524 | for (i = 0; i <= len; ++i) |
364 | { |
525 | { |
365 | SV **svp = av_fetch (av, i, 0); |
526 | SV **svp = av_fetch (av, i, 0); |
366 | encode_sv (enc, svp ? *svp : &PL_sv_undef); |
527 | encode_sv (enc, svp ? *svp : &PL_sv_undef); |
367 | } |
528 | } |
… | |
… | |
386 | ++enc->depth; |
547 | ++enc->depth; |
387 | |
548 | |
388 | int pairs = hv_iterinit (hv); |
549 | int pairs = hv_iterinit (hv); |
389 | int mg = SvMAGICAL (hv); |
550 | int mg = SvMAGICAL (hv); |
390 | |
551 | |
391 | if (mg) |
552 | if (ecb_expect_false (mg)) |
392 | encode_ch (enc, MAJOR_MAP | MINOR_INDEF); |
553 | encode_ch (enc, MAJOR_MAP | MINOR_INDEF); |
393 | else |
554 | else |
394 | encode_uint (enc, MAJOR_MAP, pairs); |
555 | encode_uint (enc, MAJOR_MAP, pairs); |
395 | |
556 | |
396 | while ((he = hv_iternext (hv))) |
557 | while ((he = hv_iternext (hv))) |
… | |
… | |
401 | encode_strref (enc, enc->cbor.flags & (F_TEXT_KEYS | F_TEXT_STRINGS), 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)); |
402 | |
563 | |
403 | 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)); |
404 | } |
565 | } |
405 | |
566 | |
406 | if (mg) |
567 | if (ecb_expect_false (mg)) |
407 | encode_ch (enc, MAJOR_MISC | MINOR_INDEF); |
568 | encode_ch (enc, MAJOR_MISC | MINOR_INDEF); |
408 | |
569 | |
409 | --enc->depth; |
570 | --enc->depth; |
410 | } |
571 | } |
411 | |
572 | |
… | |
… | |
431 | |
592 | |
432 | HV *stash = SvSTASH (sv); |
593 | HV *stash = SvSTASH (sv); |
433 | |
594 | |
434 | if (stash == boolean_stash) |
595 | if (stash == boolean_stash) |
435 | { |
596 | { |
436 | encode_ch (enc, SvIV (sv) ? MAJOR_MISC | SIMPLE_TRUE : MAJOR_MISC | SIMPLE_FALSE); |
597 | encode_bool (enc, SvIV (sv)); |
437 | return; |
598 | return; |
438 | } |
599 | } |
439 | else if (stash == error_stash) |
600 | else if (stash == error_stash) |
440 | { |
601 | { |
441 | encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF); |
602 | encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF); |
… | |
… | |
444 | else if (stash == tagged_stash) |
605 | else if (stash == tagged_stash) |
445 | { |
606 | { |
446 | if (svt != SVt_PVAV) |
607 | if (svt != SVt_PVAV) |
447 | croak ("encountered CBOR::XS::Tagged object that isn't an array"); |
608 | croak ("encountered CBOR::XS::Tagged object that isn't an array"); |
448 | |
609 | |
|
|
610 | switch (av_len ((AV *)sv)) |
|
|
611 | { |
|
|
612 | case 2-1: |
|
|
613 | // actually a tagged value |
449 | 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))); |
450 | 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 | } |
451 | |
626 | |
452 | return; |
627 | return; |
453 | } |
628 | } |
454 | } |
629 | } |
455 | |
630 | |
456 | if (ecb_expect_false (SvREFCNT (sv) > 1) |
631 | if (ecb_expect_false (SvREFCNT (sv) > 1) |
457 | && ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING)) |
632 | && ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING)) |
458 | { |
633 | { |
459 | if (!enc->shareable) |
634 | if (ecb_expect_false (!enc->shareable)) |
460 | enc->shareable = (HV *)sv_2mortal ((SV *)newHV ()); |
635 | enc->shareable = (HV *)sv_2mortal ((SV *)newHV ()); |
461 | |
636 | |
462 | SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1); |
637 | SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1); |
463 | |
638 | |
464 | if (SvOK (*svp)) |
639 | if (SvOK (*svp)) |
… | |
… | |
510 | else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0) |
685 | else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0) |
511 | { |
686 | { |
512 | dSP; |
687 | dSP; |
513 | |
688 | |
514 | ENTER; SAVETMPS; |
689 | ENTER; SAVETMPS; |
515 | SAVESTACK_POS (); |
|
|
516 | PUSHMARK (SP); |
690 | PUSHMARK (SP); |
517 | EXTEND (SP, 2); |
691 | EXTEND (SP, 2); |
518 | // 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 |
519 | PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); |
693 | PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); |
520 | PUSHs (sv_cbor); |
694 | PUSHs (sv_cbor); |
… | |
… | |
529 | |
703 | |
530 | encode_tag (enc, CBOR_TAG_PERL_OBJECT); |
704 | encode_tag (enc, CBOR_TAG_PERL_OBJECT); |
531 | encode_uint (enc, MAJOR_ARRAY, count + 1); |
705 | encode_uint (enc, MAJOR_ARRAY, count + 1); |
532 | encode_strref (enc, 0, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash)); |
706 | encode_strref (enc, 0, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash)); |
533 | |
707 | |
534 | while (count) |
708 | { |
|
|
709 | int i; |
|
|
710 | |
|
|
711 | for (i = 0; i < count; ++i) |
535 | encode_sv (enc, SP[1 - count--]); |
712 | encode_sv (enc, SP[i + 1 - count]); |
|
|
713 | |
|
|
714 | SP -= count; |
|
|
715 | } |
536 | |
716 | |
537 | PUTBACK; |
717 | PUTBACK; |
538 | |
718 | |
539 | FREETMPS; LEAVE; |
719 | FREETMPS; LEAVE; |
540 | } |
720 | } |
… | |
… | |
562 | |
742 | |
563 | if (ecb_expect_false (nv == (NV)(U32)nv)) |
743 | if (ecb_expect_false (nv == (NV)(U32)nv)) |
564 | encode_uint (enc, MAJOR_POS_INT, (U32)nv); |
744 | encode_uint (enc, MAJOR_POS_INT, (U32)nv); |
565 | //TODO: maybe I32? |
745 | //TODO: maybe I32? |
566 | else if (ecb_expect_false (nv == (float)nv)) |
746 | else if (ecb_expect_false (nv == (float)nv)) |
567 | { |
747 | encode_float32 (enc, nv); |
568 | uint32_t fp = ecb_float_to_binary32 (nv); |
|
|
569 | |
|
|
570 | *enc->cur++ = MAJOR_MISC | MISC_FLOAT32; |
|
|
571 | |
|
|
572 | if (!ecb_big_endian ()) |
|
|
573 | fp = ecb_bswap32 (fp); |
|
|
574 | |
|
|
575 | memcpy (enc->cur, &fp, 4); |
|
|
576 | enc->cur += 4; |
|
|
577 | } |
|
|
578 | else |
748 | else |
579 | { |
749 | encode_float64 (enc, nv); |
580 | uint64_t fp = ecb_double_to_binary64 (nv); |
|
|
581 | |
|
|
582 | *enc->cur++ = MAJOR_MISC | MISC_FLOAT64; |
|
|
583 | |
|
|
584 | if (!ecb_big_endian ()) |
|
|
585 | fp = ecb_bswap64 (fp); |
|
|
586 | |
|
|
587 | memcpy (enc->cur, &fp, 8); |
|
|
588 | enc->cur += 8; |
|
|
589 | } |
|
|
590 | } |
750 | } |
591 | |
751 | |
592 | static void |
752 | static void |
593 | encode_sv (enc_t *enc, SV *sv) |
753 | encode_sv (enc_t *enc, SV *sv) |
594 | { |
754 | { |
… | |
… | |
601 | encode_strref (enc, enc->cbor.flags & F_TEXT_STRINGS, SvUTF8 (sv), str, len); |
761 | encode_strref (enc, enc->cbor.flags & F_TEXT_STRINGS, SvUTF8 (sv), str, len); |
602 | } |
762 | } |
603 | else if (SvNOKp (sv)) |
763 | else if (SvNOKp (sv)) |
604 | encode_nv (enc, sv); |
764 | encode_nv (enc, sv); |
605 | else if (SvIOKp (sv)) |
765 | else if (SvIOKp (sv)) |
606 | { |
766 | encode_int (enc, sv); |
607 | if (SvIsUV (sv)) |
|
|
608 | encode_uint (enc, MAJOR_POS_INT, SvUVX (sv)); |
|
|
609 | else if (SvIVX (sv) >= 0) |
|
|
610 | encode_uint (enc, MAJOR_POS_INT, SvIVX (sv)); |
|
|
611 | else |
|
|
612 | encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1)); |
|
|
613 | } |
|
|
614 | else if (SvROK (sv)) |
767 | else if (SvROK (sv)) |
615 | encode_rv (enc, SvRV (sv)); |
768 | encode_rv (enc, SvRV (sv)); |
616 | else if (!SvOK (sv)) |
769 | else if (!SvOK (sv)) |
617 | encode_ch (enc, MAJOR_MISC | SIMPLE_NULL); |
770 | encode_ch (enc, MAJOR_MISC | SIMPLE_NULL); |
618 | else if (enc->cbor.flags & F_ALLOW_UNKNOWN) |
771 | else if (enc->cbor.flags & F_ALLOW_UNKNOWN) |
… | |
… | |
625 | static SV * |
778 | static SV * |
626 | encode_cbor (SV *scalar, CBOR *cbor) |
779 | encode_cbor (SV *scalar, CBOR *cbor) |
627 | { |
780 | { |
628 | enc_t enc = { 0 }; |
781 | enc_t enc = { 0 }; |
629 | |
782 | |
630 | enc.cbor = *cbor; |
783 | enc.cbor = *cbor; |
631 | enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); |
784 | enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); |
632 | enc.cur = SvPVX (enc.sv); |
785 | enc.cur = SvPVX (enc.sv); |
633 | enc.end = SvEND (enc.sv); |
786 | enc.end = SvEND (enc.sv); |
634 | |
787 | |
635 | SvPOK_only (enc.sv); |
788 | SvPOK_only (enc.sv); |
636 | |
789 | |
637 | if (cbor->flags & F_PACK_STRINGS) |
790 | if (cbor->flags & F_PACK_STRINGS) |
638 | { |
791 | { |
… | |
… | |
698 | err_unexpected_end (dec_t *dec) |
851 | err_unexpected_end (dec_t *dec) |
699 | { |
852 | { |
700 | err_set (dec, "unexpected end of CBOR data"); |
853 | err_set (dec, "unexpected end of CBOR data"); |
701 | } |
854 | } |
702 | |
855 | |
703 | ecb_cold static void |
|
|
704 | err_nesting_exceeded (dec_t *dec) |
|
|
705 | { |
|
|
706 | err_set (dec, ERR_NESTING_EXCEEDED); |
|
|
707 | } |
|
|
708 | |
|
|
709 | #define ERR_DO(do) SB do; goto fail; SE |
856 | #define ERR_DO(do) SB do; goto fail; SE |
710 | #define ERR(reason) ERR_DO (err_set (dec, reason)) |
857 | #define ERR(reason) ERR_DO (err_set (dec, reason)) |
711 | #define ERR_ERRSV ERR_DO (err_errsv (dec)) |
858 | #define ERR_ERRSV ERR_DO (err_errsv (dec)) |
712 | |
859 | |
713 | #define WANT(len) if (ecb_expect_false ((UV)(dec->end - dec->cur) < (UV)len)) ERR_DO (err_unexpected_end (dec)) |
860 | #define WANT(len) if (ecb_expect_false ((uintptr_t)(dec->end - dec->cur) < (STRLEN)len)) ERR_DO (err_unexpected_end (dec)) |
714 | |
861 | |
715 | #define DEC_INC_DEPTH if (ecb_expect_false (++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) |
716 | #define DEC_DEC_DEPTH --dec->depth |
863 | #define DEC_DEC_DEPTH --dec->depth |
717 | |
864 | |
718 | static UV |
865 | static UV |
… | |
… | |
786 | |
933 | |
787 | for (;;) |
934 | for (;;) |
788 | { |
935 | { |
789 | WANT (1); |
936 | WANT (1); |
790 | |
937 | |
791 | if (*dec->cur == (MAJOR_MISC | MINOR_INDEF)) |
938 | if (*dec->cur == (MAJOR_MISC | MINOR_INDEF) || dec->err) |
792 | { |
939 | { |
793 | ++dec->cur; |
940 | ++dec->cur; |
794 | break; |
941 | break; |
795 | } |
942 | } |
796 | |
943 | |
… | |
… | |
810 | |
957 | |
811 | DEC_DEC_DEPTH; |
958 | DEC_DEC_DEPTH; |
812 | return newRV_noinc ((SV *)av); |
959 | return newRV_noinc ((SV *)av); |
813 | |
960 | |
814 | fail: |
961 | fail: |
815 | SvREFCNT_dec (av); |
962 | SvREFCNT_dec_NN (av); |
816 | DEC_DEC_DEPTH; |
963 | DEC_DEC_DEPTH; |
817 | return &PL_sv_undef; |
964 | return &PL_sv_undef; |
818 | } |
965 | } |
819 | |
966 | |
820 | static void |
967 | static void |
… | |
… | |
843 | |
990 | |
844 | WANT (len); |
991 | WANT (len); |
845 | dec->cur += len; |
992 | dec->cur += len; |
846 | |
993 | |
847 | if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8)) |
994 | if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8)) |
848 | if (!is_utf8_string (key, len)) |
995 | if (!cbor_is_utf8_string ((U8 *)key, len)) |
849 | ERR ("corrupted CBOR data (invalid UTF-8 in map key)"); |
996 | ERR ("corrupted CBOR data (invalid UTF-8 in map key)"); |
850 | |
997 | |
851 | hv_store (hv, key, -len, decode_sv (dec), 0); |
998 | hv_store (hv, key, -len, decode_sv (dec), 0); |
852 | |
999 | |
853 | return; |
1000 | return; |
… | |
… | |
883 | |
1030 | |
884 | return; |
1031 | return; |
885 | } |
1032 | } |
886 | |
1033 | |
887 | hv_store_ent (hv, k, v, 0); |
1034 | hv_store_ent (hv, k, v, 0); |
888 | SvREFCNT_dec (k); |
1035 | SvREFCNT_dec_NN (k); |
889 | |
1036 | |
890 | fail: |
1037 | fail: |
891 | ; |
1038 | ; |
892 | } |
1039 | } |
893 | |
1040 | |
… | |
… | |
904 | |
1051 | |
905 | for (;;) |
1052 | for (;;) |
906 | { |
1053 | { |
907 | WANT (1); |
1054 | WANT (1); |
908 | |
1055 | |
909 | if (*dec->cur == (MAJOR_MISC | MINOR_INDEF)) |
1056 | if (*dec->cur == (MAJOR_MISC | MINOR_INDEF) || dec->err) |
910 | { |
1057 | { |
911 | ++dec->cur; |
1058 | ++dec->cur; |
912 | break; |
1059 | break; |
913 | } |
1060 | } |
914 | |
1061 | |
… | |
… | |
927 | |
1074 | |
928 | DEC_DEC_DEPTH; |
1075 | DEC_DEC_DEPTH; |
929 | return newRV_noinc ((SV *)hv); |
1076 | return newRV_noinc ((SV *)hv); |
930 | |
1077 | |
931 | fail: |
1078 | fail: |
932 | SvREFCNT_dec (hv); |
1079 | SvREFCNT_dec_NN (hv); |
933 | DEC_DEC_DEPTH; |
1080 | DEC_DEC_DEPTH; |
934 | return &PL_sv_undef; |
1081 | return &PL_sv_undef; |
935 | } |
1082 | } |
936 | |
1083 | |
937 | static SV * |
1084 | static SV * |
938 | decode_str (dec_t *dec, int utf8) |
1085 | decode_str (dec_t *dec, int utf8) |
939 | { |
1086 | { |
940 | SV *sv = 0; |
1087 | SV *sv = 0; |
941 | |
1088 | |
942 | if ((*dec->cur & MINOR_MASK) == MINOR_INDEF) |
1089 | if (ecb_expect_false ((*dec->cur & MINOR_MASK) == MINOR_INDEF)) |
943 | { |
1090 | { |
944 | // indefinite length strings |
1091 | // indefinite length strings |
945 | ++dec->cur; |
1092 | ++dec->cur; |
946 | |
1093 | |
947 | U8 major = *dec->cur & MAJOR_MISC; |
1094 | U8 major = *dec->cur & MAJOR_MISC; |
… | |
… | |
982 | } |
1129 | } |
983 | |
1130 | |
984 | if (utf8) |
1131 | if (utf8) |
985 | { |
1132 | { |
986 | if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8)) |
1133 | if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8)) |
987 | if (!is_utf8_string (SvPVX (sv), SvCUR (sv))) |
1134 | if (!cbor_is_utf8_string (SvPVX (sv), SvCUR (sv))) |
988 | ERR ("corrupted CBOR data (invalid UTF-8 in text string)"); |
1135 | ERR ("corrupted CBOR data (invalid UTF-8 in text string)"); |
989 | |
1136 | |
990 | SvUTF8_on (sv); |
1137 | SvUTF8_on (sv); |
991 | } |
1138 | } |
992 | |
1139 | |
… | |
… | |
1015 | sv = newRV_noinc (decode_sv (dec)); |
1162 | sv = newRV_noinc (decode_sv (dec)); |
1016 | break; |
1163 | break; |
1017 | |
1164 | |
1018 | case CBOR_TAG_STRINGREF_NAMESPACE: |
1165 | case CBOR_TAG_STRINGREF_NAMESPACE: |
1019 | { |
1166 | { |
1020 | // do nmot use SAVETMPS/FREETMPS, as these will |
1167 | // do not use SAVETMPS/FREETMPS, as these will |
1021 | // erase mortalised caches, e.g. "shareable" |
1168 | // erase mortalised caches, e.g. "shareable" |
1022 | ENTER; |
1169 | ENTER; |
1023 | |
1170 | |
1024 | SAVESPTR (dec->stringref); |
1171 | SAVESPTR (dec->stringref); |
1025 | dec->stringref = (AV *)sv_2mortal ((SV *)newAV ()); |
1172 | dec->stringref = (AV *)sv_2mortal ((SV *)newAV ()); |
… | |
… | |
1035 | if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT)) |
1182 | if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT)) |
1036 | ERR ("corrupted CBOR data (stringref index not an unsigned integer)"); |
1183 | ERR ("corrupted CBOR data (stringref index not an unsigned integer)"); |
1037 | |
1184 | |
1038 | UV idx = decode_uint (dec); |
1185 | UV idx = decode_uint (dec); |
1039 | |
1186 | |
1040 | if (!dec->stringref || (int)idx > AvFILLp (dec->stringref)) |
1187 | if (!dec->stringref || idx >= (UV)(1 + AvFILLp (dec->stringref))) |
1041 | 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)"); |
1042 | |
1189 | |
1043 | sv = newSVsv (AvARRAY (dec->stringref)[idx]); |
1190 | sv = newSVsv (AvARRAY (dec->stringref)[idx]); |
1044 | } |
1191 | } |
1045 | break; |
1192 | break; |
… | |
… | |
1073 | if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT)) |
1220 | if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT)) |
1074 | ERR ("corrupted CBOR data (sharedref index not an unsigned integer)"); |
1221 | ERR ("corrupted CBOR data (sharedref index not an unsigned integer)"); |
1075 | |
1222 | |
1076 | UV idx = decode_uint (dec); |
1223 | UV idx = decode_uint (dec); |
1077 | |
1224 | |
1078 | if (!dec->shareable || (int)idx > AvFILLp (dec->shareable)) |
1225 | if (!dec->shareable || idx >= (UV)(1 + AvFILLp (dec->shareable))) |
1079 | ERR ("corrupted CBOR data (sharedref index out of bounds)"); |
1226 | ERR ("corrupted CBOR data (sharedref index out of bounds)"); |
1080 | |
1227 | |
1081 | sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]); |
1228 | sv = newSVsv (AvARRAY (dec->shareable)[idx]); |
1082 | |
1229 | |
1083 | if (sv == &PL_sv_undef) |
1230 | if (sv == &PL_sv_undef) |
1084 | ERR ("cyclic CBOR data structure found, but allow_cycles is not enabled"); |
1231 | ERR ("cyclic CBOR data structure found, but allow_cycles is not enabled"); |
1085 | } |
1232 | } |
1086 | break; |
1233 | break; |
… | |
… | |
1129 | { |
1276 | { |
1130 | FREETMPS; LEAVE; |
1277 | FREETMPS; LEAVE; |
1131 | ERR_ERRSV; |
1278 | ERR_ERRSV; |
1132 | } |
1279 | } |
1133 | |
1280 | |
1134 | SvREFCNT_dec (sv); |
1281 | SvREFCNT_dec_NN (sv); |
1135 | sv = SvREFCNT_inc (POPs); |
1282 | sv = SvREFCNT_inc (POPs); |
1136 | |
1283 | |
1137 | PUTBACK; |
1284 | PUTBACK; |
1138 | |
1285 | |
1139 | FREETMPS; LEAVE; |
1286 | FREETMPS; LEAVE; |
… | |
… | |
1147 | |
1294 | |
1148 | sv = decode_sv (dec); |
1295 | sv = decode_sv (dec); |
1149 | |
1296 | |
1150 | dSP; |
1297 | dSP; |
1151 | ENTER; SAVETMPS; |
1298 | ENTER; SAVETMPS; |
1152 | SAVESTACK_POS (); |
|
|
1153 | PUSHMARK (SP); |
1299 | PUSHMARK (SP); |
1154 | EXTEND (SP, 2); |
1300 | EXTEND (SP, 2); |
1155 | PUSHs (tag_sv); |
1301 | PUSHs (tag_sv); |
1156 | PUSHs (sv); |
1302 | PUSHs (sv); |
1157 | |
1303 | |
… | |
… | |
1159 | int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL); |
1305 | int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL); |
1160 | SPAGAIN; |
1306 | SPAGAIN; |
1161 | |
1307 | |
1162 | if (SvTRUE (ERRSV)) |
1308 | if (SvTRUE (ERRSV)) |
1163 | { |
1309 | { |
1164 | SvREFCNT_dec (tag_sv); |
1310 | SvREFCNT_dec_NN (tag_sv); |
1165 | FREETMPS; LEAVE; |
1311 | FREETMPS; LEAVE; |
1166 | ERR_ERRSV; |
1312 | ERR_ERRSV; |
1167 | } |
1313 | } |
1168 | |
1314 | |
1169 | if (count) |
1315 | if (count) |
1170 | { |
1316 | { |
1171 | SvREFCNT_dec (tag_sv); |
1317 | SvREFCNT_dec_NN (tag_sv); |
1172 | SvREFCNT_dec (sv); |
1318 | SvREFCNT_dec_NN (sv); |
1173 | sv = SvREFCNT_inc (POPs); |
1319 | sv = SvREFCNT_inc_NN (TOPs); |
|
|
1320 | SP -= count; |
1174 | } |
1321 | } |
1175 | else |
1322 | else |
1176 | { |
1323 | { |
1177 | AV *av = newAV (); |
1324 | AV *av = newAV (); |
1178 | av_push (av, tag_sv); |
1325 | av_push (av, tag_sv); |
… | |
… | |
1321 | for (i = av_len (dec.shareable) + 1; i--; ) |
1468 | for (i = av_len (dec.shareable) + 1; i--; ) |
1322 | if ((svp = av_fetch (dec.shareable, i, 0))) |
1469 | if ((svp = av_fetch (dec.shareable, i, 0))) |
1323 | sv_setsv (*svp, &PL_sv_undef); |
1470 | sv_setsv (*svp, &PL_sv_undef); |
1324 | } |
1471 | } |
1325 | |
1472 | |
1326 | SvREFCNT_dec (sv); |
1473 | SvREFCNT_dec_NN (sv); |
1327 | |
1474 | |
1328 | if (dec.err_sv) |
1475 | if (dec.err_sv) |
1329 | sv_2mortal (dec.err_sv); |
1476 | sv_2mortal (dec.err_sv); |
1330 | |
1477 | |
1331 | croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur); |
1478 | croak ("%s, at offset %ld (octet 0x%02x)", dec.err, (long)(dec.cur - (U8 *)data), (int)(uint8_t)*dec.cur); |
1332 | } |
1479 | } |
1333 | |
1480 | |
1334 | sv = sv_2mortal (sv); |
1481 | sv = sv_2mortal (sv); |
1335 | |
1482 | |
1336 | return sv; |
1483 | return sv; |
… | |
… | |
1425 | |
1572 | |
1426 | break; |
1573 | break; |
1427 | |
1574 | |
1428 | case MAJOR_MAP >> MAJOR_SHIFT: |
1575 | case MAJOR_MAP >> MAJOR_SHIFT: |
1429 | len <<= 1; |
1576 | len <<= 1; |
|
|
1577 | /* FALLTHROUGH */ |
1430 | case MAJOR_ARRAY >> MAJOR_SHIFT: |
1578 | case MAJOR_ARRAY >> MAJOR_SHIFT: |
1431 | if (len) |
1579 | if (len) |
1432 | { |
1580 | { |
1433 | av_push (self->incr_count, newSViv (len + 1)); //TODO: nest |
1581 | av_push (self->incr_count, newSViv (len + 1)); //TODO: nest |
1434 | count = len + 1; |
1582 | count = len + 1; |
… | |
… | |
1674 | cbor_init (&cbor); |
1822 | cbor_init (&cbor); |
1675 | PUTBACK; cborstr = decode_cbor (cborstr, &cbor, 0); SPAGAIN; |
1823 | PUTBACK; cborstr = decode_cbor (cborstr, &cbor, 0); SPAGAIN; |
1676 | XPUSHs (cborstr); |
1824 | XPUSHs (cborstr); |
1677 | } |
1825 | } |
1678 | |
1826 | |
|
|
1827 | #ifdef __AFL_COMPILER |
|
|
1828 | |
|
|
1829 | void |
|
|
1830 | afl_init () |
|
|
1831 | CODE: |
|
|
1832 | __AFL_INIT (); |
|
|
1833 | |
|
|
1834 | int |
|
|
1835 | afl_loop (unsigned int count = 10000) |
|
|
1836 | CODE: |
|
|
1837 | RETVAL = __AFL_LOOP (count); |
|
|
1838 | OUTPUT: |
|
|
1839 | RETVAL |
|
|
1840 | |
|
|
1841 | #endif |
|
|
1842 | |