… | |
… | |
21 | # define HvNAMELEN(hv) HvNAMELEN_get (hv) |
21 | # define HvNAMELEN(hv) HvNAMELEN_get (hv) |
22 | #endif |
22 | #endif |
23 | #ifndef HvNAMEUTF8 |
23 | #ifndef HvNAMEUTF8 |
24 | # define HvNAMEUTF8(hv) 0 |
24 | # define HvNAMEUTF8(hv) 0 |
25 | #endif |
25 | #endif |
|
|
26 | #ifndef SvREFCNT_inc_NN |
|
|
27 | # define SvREFCNT_inc_NN(sv) SvREFCNT_inc (sv) |
|
|
28 | #endif |
26 | #ifndef SvREFCNT_dec_NN |
29 | #ifndef SvREFCNT_dec_NN |
27 | # define SvREFCNT_dec_NN(sv) SvREFCNT_dec (sv) |
30 | # define SvREFCNT_dec_NN(sv) SvREFCNT_dec (sv) |
28 | #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))) |
29 | |
35 | |
30 | // known major and minor types |
36 | // known major and minor types |
31 | enum cbor_type |
37 | enum cbor_type |
32 | { |
38 | { |
33 | MAJOR_SHIFT = 5, |
39 | MAJOR_SHIFT = 5, |
… | |
… | |
94 | CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8 |
100 | CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8 |
95 | CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8 |
101 | CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8 |
96 | CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8 |
102 | CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8 |
97 | |
103 | |
98 | 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 |
99 | }; |
119 | }; |
100 | |
120 | |
101 | #define F_SHRINK 0x00000001UL |
121 | #define F_SHRINK 0x00000001UL |
102 | #define F_ALLOW_UNKNOWN 0x00000002UL |
122 | #define F_ALLOW_UNKNOWN 0x00000002UL |
103 | #define F_ALLOW_SHARING 0x00000004UL |
123 | #define F_ALLOW_SHARING 0x00000004UL |
… | |
… | |
184 | #endif |
204 | #endif |
185 | } |
205 | } |
186 | } |
206 | } |
187 | |
207 | |
188 | // minimum length of a string to be registered for stringref |
208 | // minimum length of a string to be registered for stringref |
189 | ecb_inline int |
209 | ecb_inline STRLEN |
190 | minimum_string_length (UV idx) |
210 | minimum_string_length (UV idx) |
191 | { |
211 | { |
192 | |
|
|
193 | return idx <= 23 ? 3 |
212 | return idx <= 23 ? 3 |
194 | : idx <= 0xffU ? 4 |
213 | : idx <= 0xffU ? 4 |
195 | : idx <= 0xffffU ? 5 |
214 | : idx <= 0xffffU ? 5 |
196 | : idx <= 0xffffffffU ? 7 |
215 | : idx <= 0xffffffffU ? 7 |
197 | : 11; |
216 | : 11; |
… | |
… | |
224 | enc->cur = SvPVX (enc->sv) + cur; |
243 | enc->cur = SvPVX (enc->sv) + cur; |
225 | enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; |
244 | enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; |
226 | } |
245 | } |
227 | } |
246 | } |
228 | |
247 | |
|
|
248 | static void encode_sv (enc_t *enc, SV *sv); |
|
|
249 | |
229 | ecb_inline void |
250 | ecb_inline void |
230 | encode_ch (enc_t *enc, char ch) |
251 | encode_ch (enc_t *enc, char ch) |
231 | { |
252 | { |
232 | need (enc, 1); |
253 | need (enc, 1); |
233 | *enc->cur++ = ch; |
254 | *enc->cur++ = ch; |
234 | } |
255 | } |
235 | |
256 | |
|
|
257 | // used for tags, intregers, element counts and so on |
236 | static void |
258 | static void |
237 | encode_uint (enc_t *enc, int major, UV len) |
259 | encode_uint (enc_t *enc, int major, UV len) |
238 | { |
260 | { |
239 | need (enc, 9); |
261 | need (enc, 9); |
240 | |
262 | |
… | |
… | |
271 | *enc->cur++ = len >> 8; |
293 | *enc->cur++ = len >> 8; |
272 | *enc->cur++ = len; |
294 | *enc->cur++ = len; |
273 | } |
295 | } |
274 | } |
296 | } |
275 | |
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 | |
276 | ecb_inline void |
310 | ecb_inline void |
277 | encode_tag (enc_t *enc, UV tag) |
311 | encode_tag (enc_t *enc, UV tag) |
278 | { |
312 | { |
279 | encode_uint (enc, MAJOR_TAG, tag); |
313 | encode_uint (enc, MAJOR_TAG, tag); |
280 | } |
314 | } |
… | |
… | |
341 | } |
375 | } |
342 | |
376 | |
343 | encode_str (enc, upgrade_utf8, utf8, str, len); |
377 | encode_str (enc, upgrade_utf8, utf8, str, len); |
344 | } |
378 | } |
345 | |
379 | |
346 | 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 | } |
347 | |
510 | |
348 | static void |
511 | static void |
349 | encode_av (enc_t *enc, AV *av) |
512 | encode_av (enc_t *enc, AV *av) |
350 | { |
513 | { |
351 | int i, len = av_len (av); |
514 | int i, len = av_len (av); |
… | |
… | |
355 | |
518 | |
356 | ++enc->depth; |
519 | ++enc->depth; |
357 | |
520 | |
358 | encode_uint (enc, MAJOR_ARRAY, len + 1); |
521 | encode_uint (enc, MAJOR_ARRAY, len + 1); |
359 | |
522 | |
360 | if (SvMAGICAL (av)) |
523 | if (ecb_expect_false (SvMAGICAL (av))) |
361 | for (i = 0; i <= len; ++i) |
524 | for (i = 0; i <= len; ++i) |
362 | { |
525 | { |
363 | SV **svp = av_fetch (av, i, 0); |
526 | SV **svp = av_fetch (av, i, 0); |
364 | encode_sv (enc, svp ? *svp : &PL_sv_undef); |
527 | encode_sv (enc, svp ? *svp : &PL_sv_undef); |
365 | } |
528 | } |
… | |
… | |
384 | ++enc->depth; |
547 | ++enc->depth; |
385 | |
548 | |
386 | int pairs = hv_iterinit (hv); |
549 | int pairs = hv_iterinit (hv); |
387 | int mg = SvMAGICAL (hv); |
550 | int mg = SvMAGICAL (hv); |
388 | |
551 | |
389 | if (mg) |
552 | if (ecb_expect_false (mg)) |
390 | encode_ch (enc, MAJOR_MAP | MINOR_INDEF); |
553 | encode_ch (enc, MAJOR_MAP | MINOR_INDEF); |
391 | else |
554 | else |
392 | encode_uint (enc, MAJOR_MAP, pairs); |
555 | encode_uint (enc, MAJOR_MAP, pairs); |
393 | |
556 | |
394 | while ((he = hv_iternext (hv))) |
557 | while ((he = hv_iternext (hv))) |
… | |
… | |
399 | 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)); |
400 | |
563 | |
401 | 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)); |
402 | } |
565 | } |
403 | |
566 | |
404 | if (mg) |
567 | if (ecb_expect_false (mg)) |
405 | encode_ch (enc, MAJOR_MISC | MINOR_INDEF); |
568 | encode_ch (enc, MAJOR_MISC | MINOR_INDEF); |
406 | |
569 | |
407 | --enc->depth; |
570 | --enc->depth; |
408 | } |
571 | } |
409 | |
572 | |
… | |
… | |
429 | |
592 | |
430 | HV *stash = SvSTASH (sv); |
593 | HV *stash = SvSTASH (sv); |
431 | |
594 | |
432 | if (stash == boolean_stash) |
595 | if (stash == boolean_stash) |
433 | { |
596 | { |
434 | encode_ch (enc, SvIV (sv) ? MAJOR_MISC | SIMPLE_TRUE : MAJOR_MISC | SIMPLE_FALSE); |
597 | encode_bool (enc, SvIV (sv)); |
435 | return; |
598 | return; |
436 | } |
599 | } |
437 | else if (stash == error_stash) |
600 | else if (stash == error_stash) |
438 | { |
601 | { |
439 | encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF); |
602 | encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF); |
… | |
… | |
442 | else if (stash == tagged_stash) |
605 | else if (stash == tagged_stash) |
443 | { |
606 | { |
444 | if (svt != SVt_PVAV) |
607 | if (svt != SVt_PVAV) |
445 | croak ("encountered CBOR::XS::Tagged object that isn't an array"); |
608 | croak ("encountered CBOR::XS::Tagged object that isn't an array"); |
446 | |
609 | |
|
|
610 | switch (av_len ((AV *)sv)) |
|
|
611 | { |
|
|
612 | case 2-1: |
|
|
613 | // actually a tagged value |
447 | 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))); |
448 | 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 | } |
449 | |
626 | |
450 | return; |
627 | return; |
451 | } |
628 | } |
452 | } |
629 | } |
453 | |
630 | |
454 | if (ecb_expect_false (SvREFCNT (sv) > 1) |
631 | if (ecb_expect_false (SvREFCNT (sv) > 1) |
455 | && ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING)) |
632 | && ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING)) |
456 | { |
633 | { |
457 | if (!enc->shareable) |
634 | if (ecb_expect_false (!enc->shareable)) |
458 | enc->shareable = (HV *)sv_2mortal ((SV *)newHV ()); |
635 | enc->shareable = (HV *)sv_2mortal ((SV *)newHV ()); |
459 | |
636 | |
460 | SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1); |
637 | SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1); |
461 | |
638 | |
462 | if (SvOK (*svp)) |
639 | if (SvOK (*svp)) |
… | |
… | |
508 | else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0) |
685 | else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0) |
509 | { |
686 | { |
510 | dSP; |
687 | dSP; |
511 | |
688 | |
512 | ENTER; SAVETMPS; |
689 | ENTER; SAVETMPS; |
513 | SAVESTACK_POS (); |
|
|
514 | PUSHMARK (SP); |
690 | PUSHMARK (SP); |
515 | EXTEND (SP, 2); |
691 | EXTEND (SP, 2); |
516 | // 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 |
517 | PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); |
693 | PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); |
518 | PUSHs (sv_cbor); |
694 | PUSHs (sv_cbor); |
… | |
… | |
527 | |
703 | |
528 | encode_tag (enc, CBOR_TAG_PERL_OBJECT); |
704 | encode_tag (enc, CBOR_TAG_PERL_OBJECT); |
529 | encode_uint (enc, MAJOR_ARRAY, count + 1); |
705 | encode_uint (enc, MAJOR_ARRAY, count + 1); |
530 | encode_strref (enc, 0, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash)); |
706 | encode_strref (enc, 0, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash)); |
531 | |
707 | |
532 | while (count) |
708 | { |
|
|
709 | int i; |
|
|
710 | |
|
|
711 | for (i = 0; i < count; ++i) |
533 | encode_sv (enc, SP[1 - count--]); |
712 | encode_sv (enc, SP[i + 1 - count]); |
|
|
713 | |
|
|
714 | SP -= count; |
|
|
715 | } |
534 | |
716 | |
535 | PUTBACK; |
717 | PUTBACK; |
536 | |
718 | |
537 | FREETMPS; LEAVE; |
719 | FREETMPS; LEAVE; |
538 | } |
720 | } |
… | |
… | |
560 | |
742 | |
561 | if (ecb_expect_false (nv == (NV)(U32)nv)) |
743 | if (ecb_expect_false (nv == (NV)(U32)nv)) |
562 | encode_uint (enc, MAJOR_POS_INT, (U32)nv); |
744 | encode_uint (enc, MAJOR_POS_INT, (U32)nv); |
563 | //TODO: maybe I32? |
745 | //TODO: maybe I32? |
564 | else if (ecb_expect_false (nv == (float)nv)) |
746 | else if (ecb_expect_false (nv == (float)nv)) |
565 | { |
747 | encode_float32 (enc, nv); |
566 | uint32_t fp = ecb_float_to_binary32 (nv); |
|
|
567 | |
|
|
568 | *enc->cur++ = MAJOR_MISC | MISC_FLOAT32; |
|
|
569 | |
|
|
570 | if (!ecb_big_endian ()) |
|
|
571 | fp = ecb_bswap32 (fp); |
|
|
572 | |
|
|
573 | memcpy (enc->cur, &fp, 4); |
|
|
574 | enc->cur += 4; |
|
|
575 | } |
|
|
576 | else |
748 | else |
577 | { |
749 | encode_float64 (enc, nv); |
578 | uint64_t fp = ecb_double_to_binary64 (nv); |
|
|
579 | |
|
|
580 | *enc->cur++ = MAJOR_MISC | MISC_FLOAT64; |
|
|
581 | |
|
|
582 | if (!ecb_big_endian ()) |
|
|
583 | fp = ecb_bswap64 (fp); |
|
|
584 | |
|
|
585 | memcpy (enc->cur, &fp, 8); |
|
|
586 | enc->cur += 8; |
|
|
587 | } |
|
|
588 | } |
750 | } |
589 | |
751 | |
590 | static void |
752 | static void |
591 | encode_sv (enc_t *enc, SV *sv) |
753 | encode_sv (enc_t *enc, SV *sv) |
592 | { |
754 | { |
… | |
… | |
599 | 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); |
600 | } |
762 | } |
601 | else if (SvNOKp (sv)) |
763 | else if (SvNOKp (sv)) |
602 | encode_nv (enc, sv); |
764 | encode_nv (enc, sv); |
603 | else if (SvIOKp (sv)) |
765 | else if (SvIOKp (sv)) |
604 | { |
766 | encode_int (enc, sv); |
605 | if (SvIsUV (sv)) |
|
|
606 | encode_uint (enc, MAJOR_POS_INT, SvUVX (sv)); |
|
|
607 | else if (SvIVX (sv) >= 0) |
|
|
608 | encode_uint (enc, MAJOR_POS_INT, SvIVX (sv)); |
|
|
609 | else |
|
|
610 | encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1)); |
|
|
611 | } |
|
|
612 | else if (SvROK (sv)) |
767 | else if (SvROK (sv)) |
613 | encode_rv (enc, SvRV (sv)); |
768 | encode_rv (enc, SvRV (sv)); |
614 | else if (!SvOK (sv)) |
769 | else if (!SvOK (sv)) |
615 | encode_ch (enc, MAJOR_MISC | SIMPLE_NULL); |
770 | encode_ch (enc, MAJOR_MISC | SIMPLE_NULL); |
616 | else if (enc->cbor.flags & F_ALLOW_UNKNOWN) |
771 | else if (enc->cbor.flags & F_ALLOW_UNKNOWN) |
… | |
… | |
623 | static SV * |
778 | static SV * |
624 | encode_cbor (SV *scalar, CBOR *cbor) |
779 | encode_cbor (SV *scalar, CBOR *cbor) |
625 | { |
780 | { |
626 | enc_t enc = { 0 }; |
781 | enc_t enc = { 0 }; |
627 | |
782 | |
628 | enc.cbor = *cbor; |
783 | enc.cbor = *cbor; |
629 | enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); |
784 | enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); |
630 | enc.cur = SvPVX (enc.sv); |
785 | enc.cur = SvPVX (enc.sv); |
631 | enc.end = SvEND (enc.sv); |
786 | enc.end = SvEND (enc.sv); |
632 | |
787 | |
633 | SvPOK_only (enc.sv); |
788 | SvPOK_only (enc.sv); |
634 | |
789 | |
635 | if (cbor->flags & F_PACK_STRINGS) |
790 | if (cbor->flags & F_PACK_STRINGS) |
636 | { |
791 | { |
… | |
… | |
694 | |
849 | |
695 | ecb_cold static void |
850 | ecb_cold static void |
696 | err_unexpected_end (dec_t *dec) |
851 | err_unexpected_end (dec_t *dec) |
697 | { |
852 | { |
698 | err_set (dec, "unexpected end of CBOR data"); |
853 | err_set (dec, "unexpected end of CBOR data"); |
699 | } |
|
|
700 | |
|
|
701 | ecb_cold static void |
|
|
702 | err_nesting_exceeded (dec_t *dec) |
|
|
703 | { |
|
|
704 | err_set (dec, ERR_NESTING_EXCEEDED); |
|
|
705 | } |
854 | } |
706 | |
855 | |
707 | #define ERR_DO(do) SB do; goto fail; SE |
856 | #define ERR_DO(do) SB do; goto fail; SE |
708 | #define ERR(reason) ERR_DO (err_set (dec, reason)) |
857 | #define ERR(reason) ERR_DO (err_set (dec, reason)) |
709 | #define ERR_ERRSV ERR_DO (err_errsv (dec)) |
858 | #define ERR_ERRSV ERR_DO (err_errsv (dec)) |
… | |
… | |
784 | |
933 | |
785 | for (;;) |
934 | for (;;) |
786 | { |
935 | { |
787 | WANT (1); |
936 | WANT (1); |
788 | |
937 | |
789 | if (*dec->cur == (MAJOR_MISC | MINOR_INDEF)) |
938 | if (*dec->cur == (MAJOR_MISC | MINOR_INDEF) || dec->err) |
790 | { |
939 | { |
791 | ++dec->cur; |
940 | ++dec->cur; |
792 | break; |
941 | break; |
793 | } |
942 | } |
794 | |
943 | |
… | |
… | |
808 | |
957 | |
809 | DEC_DEC_DEPTH; |
958 | DEC_DEC_DEPTH; |
810 | return newRV_noinc ((SV *)av); |
959 | return newRV_noinc ((SV *)av); |
811 | |
960 | |
812 | fail: |
961 | fail: |
813 | SvREFCNT_dec (av); |
962 | SvREFCNT_dec_NN (av); |
814 | DEC_DEC_DEPTH; |
963 | DEC_DEC_DEPTH; |
815 | return &PL_sv_undef; |
964 | return &PL_sv_undef; |
816 | } |
965 | } |
817 | |
966 | |
818 | static void |
967 | static void |
… | |
… | |
841 | |
990 | |
842 | WANT (len); |
991 | WANT (len); |
843 | dec->cur += len; |
992 | dec->cur += len; |
844 | |
993 | |
845 | if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8)) |
994 | if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8)) |
846 | if (!is_utf8_string (key, len)) |
995 | if (!cbor_is_utf8_string ((U8 *)key, len)) |
847 | ERR ("corrupted CBOR data (invalid UTF-8 in map key)"); |
996 | ERR ("corrupted CBOR data (invalid UTF-8 in map key)"); |
848 | |
997 | |
849 | hv_store (hv, key, -len, decode_sv (dec), 0); |
998 | hv_store (hv, key, -len, decode_sv (dec), 0); |
850 | |
999 | |
851 | return; |
1000 | return; |
… | |
… | |
881 | |
1030 | |
882 | return; |
1031 | return; |
883 | } |
1032 | } |
884 | |
1033 | |
885 | hv_store_ent (hv, k, v, 0); |
1034 | hv_store_ent (hv, k, v, 0); |
886 | SvREFCNT_dec (k); |
1035 | SvREFCNT_dec_NN (k); |
887 | |
1036 | |
888 | fail: |
1037 | fail: |
889 | ; |
1038 | ; |
890 | } |
1039 | } |
891 | |
1040 | |
… | |
… | |
902 | |
1051 | |
903 | for (;;) |
1052 | for (;;) |
904 | { |
1053 | { |
905 | WANT (1); |
1054 | WANT (1); |
906 | |
1055 | |
907 | if (*dec->cur == (MAJOR_MISC | MINOR_INDEF)) |
1056 | if (*dec->cur == (MAJOR_MISC | MINOR_INDEF) || dec->err) |
908 | { |
1057 | { |
909 | ++dec->cur; |
1058 | ++dec->cur; |
910 | break; |
1059 | break; |
911 | } |
1060 | } |
912 | |
1061 | |
… | |
… | |
925 | |
1074 | |
926 | DEC_DEC_DEPTH; |
1075 | DEC_DEC_DEPTH; |
927 | return newRV_noinc ((SV *)hv); |
1076 | return newRV_noinc ((SV *)hv); |
928 | |
1077 | |
929 | fail: |
1078 | fail: |
930 | SvREFCNT_dec (hv); |
1079 | SvREFCNT_dec_NN (hv); |
931 | DEC_DEC_DEPTH; |
1080 | DEC_DEC_DEPTH; |
932 | return &PL_sv_undef; |
1081 | return &PL_sv_undef; |
933 | } |
1082 | } |
934 | |
1083 | |
935 | static SV * |
1084 | static SV * |
936 | decode_str (dec_t *dec, int utf8) |
1085 | decode_str (dec_t *dec, int utf8) |
937 | { |
1086 | { |
938 | SV *sv = 0; |
1087 | SV *sv = 0; |
939 | |
1088 | |
940 | if ((*dec->cur & MINOR_MASK) == MINOR_INDEF) |
1089 | if (ecb_expect_false ((*dec->cur & MINOR_MASK) == MINOR_INDEF)) |
941 | { |
1090 | { |
942 | // indefinite length strings |
1091 | // indefinite length strings |
943 | ++dec->cur; |
1092 | ++dec->cur; |
944 | |
1093 | |
945 | U8 major = *dec->cur & MAJOR_MISC; |
1094 | U8 major = *dec->cur & MAJOR_MISC; |
… | |
… | |
980 | } |
1129 | } |
981 | |
1130 | |
982 | if (utf8) |
1131 | if (utf8) |
983 | { |
1132 | { |
984 | if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8)) |
1133 | if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8)) |
985 | if (!is_utf8_string (SvPVX (sv), SvCUR (sv))) |
1134 | if (!cbor_is_utf8_string (SvPVX (sv), SvCUR (sv))) |
986 | ERR ("corrupted CBOR data (invalid UTF-8 in text string)"); |
1135 | ERR ("corrupted CBOR data (invalid UTF-8 in text string)"); |
987 | |
1136 | |
988 | SvUTF8_on (sv); |
1137 | SvUTF8_on (sv); |
989 | } |
1138 | } |
990 | |
1139 | |
… | |
… | |
1013 | sv = newRV_noinc (decode_sv (dec)); |
1162 | sv = newRV_noinc (decode_sv (dec)); |
1014 | break; |
1163 | break; |
1015 | |
1164 | |
1016 | case CBOR_TAG_STRINGREF_NAMESPACE: |
1165 | case CBOR_TAG_STRINGREF_NAMESPACE: |
1017 | { |
1166 | { |
1018 | // do nmot use SAVETMPS/FREETMPS, as these will |
1167 | // do not use SAVETMPS/FREETMPS, as these will |
1019 | // erase mortalised caches, e.g. "shareable" |
1168 | // erase mortalised caches, e.g. "shareable" |
1020 | ENTER; |
1169 | ENTER; |
1021 | |
1170 | |
1022 | SAVESPTR (dec->stringref); |
1171 | SAVESPTR (dec->stringref); |
1023 | dec->stringref = (AV *)sv_2mortal ((SV *)newAV ()); |
1172 | dec->stringref = (AV *)sv_2mortal ((SV *)newAV ()); |
… | |
… | |
1033 | if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT)) |
1182 | if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT)) |
1034 | ERR ("corrupted CBOR data (stringref index not an unsigned integer)"); |
1183 | ERR ("corrupted CBOR data (stringref index not an unsigned integer)"); |
1035 | |
1184 | |
1036 | UV idx = decode_uint (dec); |
1185 | UV idx = decode_uint (dec); |
1037 | |
1186 | |
1038 | if (!dec->stringref || (int)idx > AvFILLp (dec->stringref)) |
1187 | if (!dec->stringref || idx >= (UV)(1 + AvFILLp (dec->stringref))) |
1039 | 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)"); |
1040 | |
1189 | |
1041 | sv = newSVsv (AvARRAY (dec->stringref)[idx]); |
1190 | sv = newSVsv (AvARRAY (dec->stringref)[idx]); |
1042 | } |
1191 | } |
1043 | break; |
1192 | break; |
… | |
… | |
1047 | if (ecb_expect_false (!dec->shareable)) |
1196 | if (ecb_expect_false (!dec->shareable)) |
1048 | dec->shareable = (AV *)sv_2mortal ((SV *)newAV ()); |
1197 | dec->shareable = (AV *)sv_2mortal ((SV *)newAV ()); |
1049 | |
1198 | |
1050 | if (dec->cbor.flags & F_ALLOW_CYCLES) |
1199 | if (dec->cbor.flags & F_ALLOW_CYCLES) |
1051 | { |
1200 | { |
|
|
1201 | // if cycles are allowed, then we store an AV as value |
|
|
1202 | // while it is being decoded, and gather unresolved |
|
|
1203 | // references in it, to be re4solved after decoding. |
|
|
1204 | int idx, i; |
1052 | sv = newSV (0); |
1205 | AV *av = newAV (); |
1053 | av_push (dec->shareable, SvREFCNT_inc_NN (sv)); |
1206 | av_push (dec->shareable, SvREFCNT_inc_NN ((SV *)av)); |
|
|
1207 | idx = AvFILLp (dec->shareable); |
1054 | |
1208 | |
1055 | SV *osv = decode_sv (dec); |
1209 | sv = decode_sv (dec); |
1056 | sv_setsv (sv, osv); |
1210 | |
|
|
1211 | // the AV now contains \undef for all unresolved references, |
|
|
1212 | // so we fix them up here. |
|
|
1213 | for (i = 0; i <= AvFILLp (av); ++i) |
|
|
1214 | SvRV_set (AvARRAY (av)[i], SvREFCNT_inc_NN (SvRV (sv))); |
|
|
1215 | |
|
|
1216 | // now replace the AV by a reference to the completed value |
1057 | SvREFCNT_dec_NN (osv); |
1217 | SvREFCNT_dec_NN (av); |
|
|
1218 | AvARRAY (dec->shareable)[idx] = SvREFCNT_inc_NN (sv); |
1058 | } |
1219 | } |
1059 | else |
1220 | else |
1060 | { |
1221 | { |
1061 | av_push (dec->shareable, &PL_sv_undef); |
1222 | av_push (dec->shareable, &PL_sv_undef); |
1062 | int idx = AvFILLp (dec->shareable); |
1223 | int idx = AvFILLp (dec->shareable); |
… | |
… | |
1071 | if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT)) |
1232 | if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT)) |
1072 | ERR ("corrupted CBOR data (sharedref index not an unsigned integer)"); |
1233 | ERR ("corrupted CBOR data (sharedref index not an unsigned integer)"); |
1073 | |
1234 | |
1074 | UV idx = decode_uint (dec); |
1235 | UV idx = decode_uint (dec); |
1075 | |
1236 | |
1076 | if (!dec->shareable || (int)idx > AvFILLp (dec->shareable)) |
1237 | if (!dec->shareable || idx >= (UV)(1 + AvFILLp (dec->shareable))) |
1077 | ERR ("corrupted CBOR data (sharedref index out of bounds)"); |
1238 | ERR ("corrupted CBOR data (sharedref index out of bounds)"); |
1078 | |
1239 | |
1079 | sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]); |
1240 | sv = AvARRAY (dec->shareable)[idx]; |
1080 | |
1241 | |
1081 | if (sv == &PL_sv_undef) |
1242 | // reference to cycle, we create a new \undef and use that, and also |
|
|
1243 | // registerr it in the AV for later fixing |
|
|
1244 | if (SvTYPE (sv) == SVt_PVAV) |
|
|
1245 | { |
|
|
1246 | AV *av = (AV *)sv; |
|
|
1247 | sv = newRV_noinc (&PL_sv_undef); |
|
|
1248 | av_push (av, sv); |
|
|
1249 | } |
|
|
1250 | else if (sv == &PL_sv_undef) // not yet decoded, but cycles not allowed |
1082 | ERR ("cyclic CBOR data structure found, but allow_cycles is not enabled"); |
1251 | ERR ("cyclic CBOR data structure found, but allow_cycles is not enabled"); |
|
|
1252 | else // we decoded the object earlier, no cycle |
|
|
1253 | sv = newSVsv (sv); |
1083 | } |
1254 | } |
1084 | break; |
1255 | break; |
1085 | |
1256 | |
1086 | case CBOR_TAG_PERL_OBJECT: |
1257 | case CBOR_TAG_PERL_OBJECT: |
1087 | { |
1258 | { |
… | |
… | |
1127 | { |
1298 | { |
1128 | FREETMPS; LEAVE; |
1299 | FREETMPS; LEAVE; |
1129 | ERR_ERRSV; |
1300 | ERR_ERRSV; |
1130 | } |
1301 | } |
1131 | |
1302 | |
1132 | SvREFCNT_dec (sv); |
1303 | SvREFCNT_dec_NN (sv); |
1133 | sv = SvREFCNT_inc (POPs); |
1304 | sv = SvREFCNT_inc (POPs); |
1134 | |
1305 | |
1135 | PUTBACK; |
1306 | PUTBACK; |
1136 | |
1307 | |
1137 | FREETMPS; LEAVE; |
1308 | FREETMPS; LEAVE; |
… | |
… | |
1145 | |
1316 | |
1146 | sv = decode_sv (dec); |
1317 | sv = decode_sv (dec); |
1147 | |
1318 | |
1148 | dSP; |
1319 | dSP; |
1149 | ENTER; SAVETMPS; |
1320 | ENTER; SAVETMPS; |
1150 | SAVESTACK_POS (); |
|
|
1151 | PUSHMARK (SP); |
1321 | PUSHMARK (SP); |
1152 | EXTEND (SP, 2); |
1322 | EXTEND (SP, 2); |
1153 | PUSHs (tag_sv); |
1323 | PUSHs (tag_sv); |
1154 | PUSHs (sv); |
1324 | PUSHs (sv); |
1155 | |
1325 | |
… | |
… | |
1157 | int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL); |
1327 | int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL); |
1158 | SPAGAIN; |
1328 | SPAGAIN; |
1159 | |
1329 | |
1160 | if (SvTRUE (ERRSV)) |
1330 | if (SvTRUE (ERRSV)) |
1161 | { |
1331 | { |
1162 | SvREFCNT_dec (tag_sv); |
1332 | SvREFCNT_dec_NN (tag_sv); |
1163 | FREETMPS; LEAVE; |
1333 | FREETMPS; LEAVE; |
1164 | ERR_ERRSV; |
1334 | ERR_ERRSV; |
1165 | } |
1335 | } |
1166 | |
1336 | |
1167 | if (count) |
1337 | if (count) |
1168 | { |
1338 | { |
1169 | SvREFCNT_dec (tag_sv); |
1339 | SvREFCNT_dec_NN (tag_sv); |
1170 | SvREFCNT_dec (sv); |
1340 | SvREFCNT_dec_NN (sv); |
1171 | sv = SvREFCNT_inc (POPs); |
1341 | sv = SvREFCNT_inc_NN (TOPs); |
|
|
1342 | SP -= count; |
1172 | } |
1343 | } |
1173 | else |
1344 | else |
1174 | { |
1345 | { |
1175 | AV *av = newAV (); |
1346 | AV *av = newAV (); |
1176 | av_push (av, tag_sv); |
1347 | av_push (av, tag_sv); |
… | |
… | |
1319 | for (i = av_len (dec.shareable) + 1; i--; ) |
1490 | for (i = av_len (dec.shareable) + 1; i--; ) |
1320 | if ((svp = av_fetch (dec.shareable, i, 0))) |
1491 | if ((svp = av_fetch (dec.shareable, i, 0))) |
1321 | sv_setsv (*svp, &PL_sv_undef); |
1492 | sv_setsv (*svp, &PL_sv_undef); |
1322 | } |
1493 | } |
1323 | |
1494 | |
1324 | SvREFCNT_dec (sv); |
1495 | SvREFCNT_dec_NN (sv); |
1325 | |
1496 | |
1326 | if (dec.err_sv) |
1497 | if (dec.err_sv) |
1327 | sv_2mortal (dec.err_sv); |
1498 | sv_2mortal (dec.err_sv); |
1328 | |
1499 | |
1329 | croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur); |
1500 | croak ("%s, at offset %ld (octet 0x%02x)", dec.err, (long)(dec.cur - (U8 *)data), (int)(uint8_t)*dec.cur); |
1330 | } |
1501 | } |
1331 | |
1502 | |
1332 | sv = sv_2mortal (sv); |
1503 | sv = sv_2mortal (sv); |
1333 | |
1504 | |
1334 | return sv; |
1505 | return sv; |
… | |
… | |
1423 | |
1594 | |
1424 | break; |
1595 | break; |
1425 | |
1596 | |
1426 | case MAJOR_MAP >> MAJOR_SHIFT: |
1597 | case MAJOR_MAP >> MAJOR_SHIFT: |
1427 | len <<= 1; |
1598 | len <<= 1; |
|
|
1599 | /* FALLTHROUGH */ |
1428 | case MAJOR_ARRAY >> MAJOR_SHIFT: |
1600 | case MAJOR_ARRAY >> MAJOR_SHIFT: |
1429 | if (len) |
1601 | if (len) |
1430 | { |
1602 | { |
1431 | av_push (self->incr_count, newSViv (len + 1)); //TODO: nest |
1603 | av_push (self->incr_count, newSViv (len + 1)); //TODO: nest |
1432 | count = len + 1; |
1604 | count = len + 1; |
… | |
… | |
1672 | cbor_init (&cbor); |
1844 | cbor_init (&cbor); |
1673 | PUTBACK; cborstr = decode_cbor (cborstr, &cbor, 0); SPAGAIN; |
1845 | PUTBACK; cborstr = decode_cbor (cborstr, &cbor, 0); SPAGAIN; |
1674 | XPUSHs (cborstr); |
1846 | XPUSHs (cborstr); |
1675 | } |
1847 | } |
1676 | |
1848 | |
|
|
1849 | #ifdef __AFL_COMPILER |
|
|
1850 | |
|
|
1851 | void |
|
|
1852 | afl_init () |
|
|
1853 | CODE: |
|
|
1854 | __AFL_INIT (); |
|
|
1855 | |
|
|
1856 | int |
|
|
1857 | afl_loop (unsigned int count = 10000) |
|
|
1858 | CODE: |
|
|
1859 | RETVAL = __AFL_LOOP (count); |
|
|
1860 | OUTPUT: |
|
|
1861 | RETVAL |
|
|
1862 | |
|
|
1863 | #endif |
|
|
1864 | |