… | |
… | |
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 |
… | |
… | |
99 | |
103 | |
100 | #define F_SHRINK 0x00000001UL |
104 | #define F_SHRINK 0x00000001UL |
101 | #define F_ALLOW_UNKNOWN 0x00000002UL |
105 | #define F_ALLOW_UNKNOWN 0x00000002UL |
102 | #define F_ALLOW_SHARING 0x00000004UL |
106 | #define F_ALLOW_SHARING 0x00000004UL |
103 | #define F_ALLOW_CYCLES 0x00000008UL |
107 | #define F_ALLOW_CYCLES 0x00000008UL |
|
|
108 | #define F_FORBID_OBJECTS 0x00000010UL |
104 | #define F_PACK_STRINGS 0x00000010UL |
109 | #define F_PACK_STRINGS 0x00000020UL |
|
|
110 | #define F_TEXT_KEYS 0x00000040UL |
105 | #define F_UTF8_STRINGS 0x00000020UL |
111 | #define F_TEXT_STRINGS 0x00000080UL |
106 | #define F_VALIDATE_UTF8 0x00000040UL |
112 | #define F_VALIDATE_UTF8 0x00000100UL |
107 | |
113 | |
108 | #define INIT_SIZE 32 // initial scalar size to be allocated |
114 | #define INIT_SIZE 32 // initial scalar size to be allocated |
109 | |
115 | |
110 | #define SB do { |
116 | #define SB do { |
111 | #define SE } while (0) |
117 | #define SE } while (0) |
… | |
… | |
184 | |
190 | |
185 | // minimum length of a string to be registered for stringref |
191 | // minimum length of a string to be registered for stringref |
186 | ecb_inline int |
192 | ecb_inline int |
187 | minimum_string_length (UV idx) |
193 | minimum_string_length (UV idx) |
188 | { |
194 | { |
189 | return idx > 23 |
195 | return idx <= 23 ? 3 |
190 | ? idx > 0xffU |
196 | : idx <= 0xffU ? 4 |
191 | ? idx > 0xffffU |
197 | : idx <= 0xffffU ? 5 |
192 | ? idx > 0xffffffffU |
198 | : idx <= 0xffffffffU ? 7 |
193 | ? 11 |
199 | : 11; |
194 | : 7 |
|
|
195 | : 5 |
|
|
196 | : 4 |
|
|
197 | : 3; |
|
|
198 | } |
200 | } |
199 | |
201 | |
200 | ///////////////////////////////////////////////////////////////////////////// |
202 | ///////////////////////////////////////////////////////////////////////////// |
201 | // encoder |
203 | // encoder |
202 | |
204 | |
… | |
… | |
215 | } enc_t; |
217 | } enc_t; |
216 | |
218 | |
217 | ecb_inline void |
219 | ecb_inline void |
218 | need (enc_t *enc, STRLEN len) |
220 | need (enc_t *enc, STRLEN len) |
219 | { |
221 | { |
220 | if (ecb_expect_false (enc->cur + len >= enc->end)) |
222 | if (ecb_expect_false ((uintptr_t)(enc->end - enc->cur) < len)) |
221 | { |
223 | { |
222 | STRLEN cur = enc->cur - (char *)SvPVX (enc->sv); |
224 | STRLEN cur = enc->cur - (char *)SvPVX (enc->sv); |
223 | SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1); |
225 | SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1); |
224 | enc->cur = SvPVX (enc->sv) + cur; |
226 | enc->cur = SvPVX (enc->sv) + cur; |
225 | enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; |
227 | enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; |
… | |
… | |
277 | encode_tag (enc_t *enc, UV tag) |
279 | encode_tag (enc_t *enc, UV tag) |
278 | { |
280 | { |
279 | encode_uint (enc, MAJOR_TAG, tag); |
281 | encode_uint (enc, MAJOR_TAG, tag); |
280 | } |
282 | } |
281 | |
283 | |
|
|
284 | // exceptional (hopefully) slow path for byte strings that need to be utf8-encoded |
|
|
285 | ecb_noinline static void |
|
|
286 | encode_str_utf8 (enc_t *enc, int utf8, char *str, STRLEN len) |
|
|
287 | { |
|
|
288 | STRLEN ulen = len; |
|
|
289 | U8 *p, *pend = (U8 *)str + len; |
|
|
290 | |
|
|
291 | for (p = (U8 *)str; p < pend; ++p) |
|
|
292 | ulen += *p >> 7; // count set high bits |
|
|
293 | |
|
|
294 | encode_uint (enc, MAJOR_TEXT, ulen); |
|
|
295 | |
|
|
296 | need (enc, ulen); |
|
|
297 | for (p = (U8 *)str; p < pend; ++p) |
|
|
298 | if (*p < 0x80) |
|
|
299 | *enc->cur++ = *p; |
|
|
300 | else |
|
|
301 | { |
|
|
302 | *enc->cur++ = 0xc0 + (*p >> 6); |
|
|
303 | *enc->cur++ = 0x80 + (*p & 63); |
|
|
304 | } |
|
|
305 | } |
|
|
306 | |
282 | ecb_inline void |
307 | ecb_inline void |
283 | encode_str (enc_t *enc, int utf8, char *str, STRLEN len) |
308 | encode_str (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len) |
284 | { |
309 | { |
285 | if (ecb_expect_false (enc->cbor.flags & F_UTF8_STRINGS)) |
310 | if (ecb_expect_false (upgrade_utf8)) |
286 | if (!utf8) |
311 | if (!utf8) |
287 | { |
312 | { |
288 | SV *sv = sv_newmortal (); |
|
|
289 | char *s; STRLEN l; |
|
|
290 | |
|
|
291 | sv_setpvn (sv, str, len); |
|
|
292 | |
|
|
293 | s = SvPVutf8 (sv, l); |
|
|
294 | encode_str (enc, 1, s, l); |
313 | encode_str_utf8 (enc, utf8, str, len); |
295 | return; |
314 | return; |
296 | } |
315 | } |
297 | |
316 | |
298 | encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len); |
317 | encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len); |
299 | need (enc, len); |
318 | need (enc, len); |
300 | memcpy (enc->cur, str, len); |
319 | memcpy (enc->cur, str, len); |
301 | enc->cur += len; |
320 | enc->cur += len; |
302 | } |
321 | } |
303 | |
322 | |
304 | static void |
323 | ecb_inline void |
305 | encode_strref (enc_t *enc, int utf8, char *str, STRLEN len) |
324 | encode_strref (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len) |
306 | { |
325 | { |
307 | if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS)) |
326 | if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS)) |
308 | { |
327 | { |
309 | SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1); |
328 | SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1); |
310 | |
329 | |
… | |
… | |
321 | sv_setuv (*svp, enc->stringref_idx); |
340 | sv_setuv (*svp, enc->stringref_idx); |
322 | ++enc->stringref_idx; |
341 | ++enc->stringref_idx; |
323 | } |
342 | } |
324 | } |
343 | } |
325 | |
344 | |
326 | encode_str (enc, utf8, str, len); |
345 | encode_str (enc, upgrade_utf8, utf8, str, len); |
327 | } |
346 | } |
328 | |
347 | |
329 | static void encode_sv (enc_t *enc, SV *sv); |
348 | static void encode_sv (enc_t *enc, SV *sv); |
330 | |
349 | |
331 | static void |
350 | static void |
… | |
… | |
338 | |
357 | |
339 | ++enc->depth; |
358 | ++enc->depth; |
340 | |
359 | |
341 | encode_uint (enc, MAJOR_ARRAY, len + 1); |
360 | encode_uint (enc, MAJOR_ARRAY, len + 1); |
342 | |
361 | |
343 | if (SvMAGICAL (av)) |
362 | if (ecb_expect_false (SvMAGICAL (av))) |
344 | for (i = 0; i <= len; ++i) |
363 | for (i = 0; i <= len; ++i) |
345 | { |
364 | { |
346 | SV **svp = av_fetch (av, i, 0); |
365 | SV **svp = av_fetch (av, i, 0); |
347 | encode_sv (enc, svp ? *svp : &PL_sv_undef); |
366 | encode_sv (enc, svp ? *svp : &PL_sv_undef); |
348 | } |
367 | } |
… | |
… | |
367 | ++enc->depth; |
386 | ++enc->depth; |
368 | |
387 | |
369 | int pairs = hv_iterinit (hv); |
388 | int pairs = hv_iterinit (hv); |
370 | int mg = SvMAGICAL (hv); |
389 | int mg = SvMAGICAL (hv); |
371 | |
390 | |
372 | if (mg) |
391 | if (ecb_expect_false (mg)) |
373 | encode_ch (enc, MAJOR_MAP | MINOR_INDEF); |
392 | encode_ch (enc, MAJOR_MAP | MINOR_INDEF); |
374 | else |
393 | else |
375 | encode_uint (enc, MAJOR_MAP, pairs); |
394 | encode_uint (enc, MAJOR_MAP, pairs); |
376 | |
395 | |
377 | while ((he = hv_iternext (hv))) |
396 | while ((he = hv_iternext (hv))) |
378 | { |
397 | { |
379 | if (HeKLEN (he) == HEf_SVKEY) |
398 | if (HeKLEN (he) == HEf_SVKEY) |
380 | encode_sv (enc, HeSVKEY (he)); |
399 | encode_sv (enc, HeSVKEY (he)); |
381 | else |
400 | else |
382 | encode_strref (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he)); |
401 | encode_strref (enc, enc->cbor.flags & (F_TEXT_KEYS | F_TEXT_STRINGS), HeKUTF8 (he), HeKEY (he), HeKLEN (he)); |
383 | |
402 | |
384 | encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he)); |
403 | encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he)); |
385 | } |
404 | } |
386 | |
405 | |
387 | if (mg) |
406 | if (ecb_expect_false (mg)) |
388 | encode_ch (enc, MAJOR_MISC | MINOR_INDEF); |
407 | encode_ch (enc, MAJOR_MISC | MINOR_INDEF); |
389 | |
408 | |
390 | --enc->depth; |
409 | --enc->depth; |
391 | } |
410 | } |
392 | |
411 | |
… | |
… | |
435 | } |
454 | } |
436 | |
455 | |
437 | if (ecb_expect_false (SvREFCNT (sv) > 1) |
456 | if (ecb_expect_false (SvREFCNT (sv) > 1) |
438 | && ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING)) |
457 | && ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING)) |
439 | { |
458 | { |
440 | if (!enc->shareable) |
459 | if (ecb_expect_false (!enc->shareable)) |
441 | enc->shareable = (HV *)sv_2mortal ((SV *)newHV ()); |
460 | enc->shareable = (HV *)sv_2mortal ((SV *)newHV ()); |
442 | |
461 | |
443 | SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1); |
462 | SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1); |
444 | |
463 | |
445 | if (SvOK (*svp)) |
464 | if (SvOK (*svp)) |
… | |
… | |
459 | if (ecb_expect_false (SvOBJECT (sv))) |
478 | if (ecb_expect_false (SvOBJECT (sv))) |
460 | { |
479 | { |
461 | HV *stash = SvSTASH (sv); |
480 | HV *stash = SvSTASH (sv); |
462 | GV *method; |
481 | GV *method; |
463 | |
482 | |
|
|
483 | if (enc->cbor.flags & F_FORBID_OBJECTS) |
|
|
484 | croak ("encountered object '%s', but forbid_objects is enabled", |
|
|
485 | SvPV_nolen (sv_2mortal (newRV_inc (sv)))); |
464 | if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0))) |
486 | else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0))) |
465 | { |
487 | { |
466 | dSP; |
488 | dSP; |
467 | |
489 | |
468 | ENTER; SAVETMPS; |
490 | ENTER; SAVETMPS; |
469 | PUSHMARK (SP); |
491 | PUSHMARK (SP); |
… | |
… | |
505 | if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv) |
527 | if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv) |
506 | croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash)); |
528 | croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash)); |
507 | |
529 | |
508 | encode_tag (enc, CBOR_TAG_PERL_OBJECT); |
530 | encode_tag (enc, CBOR_TAG_PERL_OBJECT); |
509 | encode_uint (enc, MAJOR_ARRAY, count + 1); |
531 | encode_uint (enc, MAJOR_ARRAY, count + 1); |
510 | encode_strref (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash)); |
532 | encode_strref (enc, 0, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash)); |
511 | |
533 | |
512 | while (count) |
534 | while (count) |
513 | encode_sv (enc, SP[1 - count--]); |
535 | encode_sv (enc, SP[1 - count--]); |
514 | |
536 | |
515 | PUTBACK; |
537 | PUTBACK; |
… | |
… | |
541 | if (ecb_expect_false (nv == (NV)(U32)nv)) |
563 | if (ecb_expect_false (nv == (NV)(U32)nv)) |
542 | encode_uint (enc, MAJOR_POS_INT, (U32)nv); |
564 | encode_uint (enc, MAJOR_POS_INT, (U32)nv); |
543 | //TODO: maybe I32? |
565 | //TODO: maybe I32? |
544 | else if (ecb_expect_false (nv == (float)nv)) |
566 | else if (ecb_expect_false (nv == (float)nv)) |
545 | { |
567 | { |
|
|
568 | *enc->cur++ = MAJOR_MISC | MISC_FLOAT32; |
|
|
569 | |
546 | uint32_t fp = ecb_float_to_binary32 (nv); |
570 | uint32_t fp = ecb_float_to_binary32 (nv); |
547 | |
|
|
548 | *enc->cur++ = MAJOR_MISC | MISC_FLOAT32; |
|
|
549 | |
571 | |
550 | if (!ecb_big_endian ()) |
572 | if (!ecb_big_endian ()) |
551 | fp = ecb_bswap32 (fp); |
573 | fp = ecb_bswap32 (fp); |
552 | |
574 | |
553 | memcpy (enc->cur, &fp, 4); |
575 | memcpy (enc->cur, &fp, 4); |
554 | enc->cur += 4; |
576 | enc->cur += 4; |
555 | } |
577 | } |
556 | else |
578 | else |
557 | { |
579 | { |
|
|
580 | *enc->cur++ = MAJOR_MISC | MISC_FLOAT64; |
|
|
581 | |
558 | uint64_t fp = ecb_double_to_binary64 (nv); |
582 | uint64_t fp = ecb_double_to_binary64 (nv); |
559 | |
|
|
560 | *enc->cur++ = MAJOR_MISC | MISC_FLOAT64; |
|
|
561 | |
583 | |
562 | if (!ecb_big_endian ()) |
584 | if (!ecb_big_endian ()) |
563 | fp = ecb_bswap64 (fp); |
585 | fp = ecb_bswap64 (fp); |
564 | |
586 | |
565 | memcpy (enc->cur, &fp, 8); |
587 | memcpy (enc->cur, &fp, 8); |
… | |
… | |
574 | |
596 | |
575 | if (SvPOKp (sv)) |
597 | if (SvPOKp (sv)) |
576 | { |
598 | { |
577 | STRLEN len; |
599 | STRLEN len; |
578 | char *str = SvPV (sv, len); |
600 | char *str = SvPV (sv, len); |
579 | encode_strref (enc, SvUTF8 (sv), str, len); |
601 | encode_strref (enc, enc->cbor.flags & F_TEXT_STRINGS, SvUTF8 (sv), str, len); |
580 | } |
602 | } |
581 | else if (SvNOKp (sv)) |
603 | else if (SvNOKp (sv)) |
582 | encode_nv (enc, sv); |
604 | encode_nv (enc, sv); |
583 | else if (SvIOKp (sv)) |
605 | else if (SvIOKp (sv)) |
584 | { |
606 | { |
… | |
… | |
603 | static SV * |
625 | static SV * |
604 | encode_cbor (SV *scalar, CBOR *cbor) |
626 | encode_cbor (SV *scalar, CBOR *cbor) |
605 | { |
627 | { |
606 | enc_t enc = { 0 }; |
628 | enc_t enc = { 0 }; |
607 | |
629 | |
608 | enc.cbor = *cbor; |
630 | enc.cbor = *cbor; |
609 | enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); |
631 | enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); |
610 | enc.cur = SvPVX (enc.sv); |
632 | enc.cur = SvPVX (enc.sv); |
611 | enc.end = SvEND (enc.sv); |
633 | enc.end = SvEND (enc.sv); |
612 | |
634 | |
613 | SvPOK_only (enc.sv); |
635 | SvPOK_only (enc.sv); |
614 | |
636 | |
615 | if (cbor->flags & F_PACK_STRINGS) |
637 | if (cbor->flags & F_PACK_STRINGS) |
616 | { |
638 | { |
… | |
… | |
643 | U32 depth; // recursion depth |
665 | U32 depth; // recursion depth |
644 | U32 maxdepth; // recursion depth limit |
666 | U32 maxdepth; // recursion depth limit |
645 | AV *shareable; |
667 | AV *shareable; |
646 | AV *stringref; |
668 | AV *stringref; |
647 | SV *decode_tagged; |
669 | SV *decode_tagged; |
|
|
670 | SV *err_sv; // optional sv for error, needs to be freed |
648 | } dec_t; |
671 | } dec_t; |
649 | |
672 | |
650 | #define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE |
673 | // set dec->err to ERRSV |
|
|
674 | ecb_cold static void |
|
|
675 | err_errsv (dec_t *dec) |
|
|
676 | { |
|
|
677 | if (!dec->err) |
|
|
678 | { |
|
|
679 | dec->err_sv = newSVsv (ERRSV); |
651 | |
680 | |
652 | #define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data") |
681 | // chop off the trailing \n |
|
|
682 | SvCUR_set (dec->err_sv, SvCUR (dec->err_sv) - 1); |
|
|
683 | *SvEND (dec->err_sv) = 0; |
653 | |
684 | |
|
|
685 | dec->err = SvPVutf8_nolen (dec->err_sv); |
|
|
686 | } |
|
|
687 | } |
|
|
688 | |
|
|
689 | // the following functions are used to reduce code size and help the compiler to optimise |
|
|
690 | ecb_cold static void |
|
|
691 | err_set (dec_t *dec, const char *reason) |
|
|
692 | { |
|
|
693 | if (!dec->err) |
|
|
694 | dec->err = reason; |
|
|
695 | } |
|
|
696 | |
|
|
697 | ecb_cold static void |
|
|
698 | err_unexpected_end (dec_t *dec) |
|
|
699 | { |
|
|
700 | err_set (dec, "unexpected end of CBOR data"); |
|
|
701 | } |
|
|
702 | |
|
|
703 | #define ERR_DO(do) SB do; goto fail; SE |
|
|
704 | #define ERR(reason) ERR_DO (err_set (dec, reason)) |
|
|
705 | #define ERR_ERRSV ERR_DO (err_errsv (dec)) |
|
|
706 | |
|
|
707 | #define WANT(len) if (ecb_expect_false ((uintptr_t)(dec->end - dec->cur) < (STRLEN)len)) ERR_DO (err_unexpected_end (dec)) |
|
|
708 | |
654 | #define DEC_INC_DEPTH if (++dec->depth > dec->cbor.max_depth) ERR (ERR_NESTING_EXCEEDED) |
709 | #define DEC_INC_DEPTH if (ecb_expect_false (++dec->depth > dec->cbor.max_depth)) ERR (ERR_NESTING_EXCEEDED) |
655 | #define DEC_DEC_DEPTH --dec->depth |
710 | #define DEC_DEC_DEPTH --dec->depth |
656 | |
711 | |
657 | static UV |
712 | static UV |
658 | decode_uint (dec_t *dec) |
713 | decode_uint (dec_t *dec) |
659 | { |
714 | { |
… | |
… | |
736 | av_push (av, decode_sv (dec)); |
791 | av_push (av, decode_sv (dec)); |
737 | } |
792 | } |
738 | } |
793 | } |
739 | else |
794 | else |
740 | { |
795 | { |
741 | int i, len = decode_uint (dec); |
796 | UV i, len = decode_uint (dec); |
742 | |
797 | |
743 | WANT (len); // complexity check for av_fill - need at least one byte per value, do not allow supersize arrays |
798 | WANT (len); // complexity check for av_fill - need at least one byte per value, do not allow supersize arrays |
744 | av_fill (av, len - 1); |
799 | av_fill (av, len - 1); |
745 | |
800 | |
746 | for (i = 0; i < len; ++i) |
801 | for (i = 0; i < len; ++i) |
… | |
… | |
749 | |
804 | |
750 | DEC_DEC_DEPTH; |
805 | DEC_DEC_DEPTH; |
751 | return newRV_noinc ((SV *)av); |
806 | return newRV_noinc ((SV *)av); |
752 | |
807 | |
753 | fail: |
808 | fail: |
754 | SvREFCNT_dec (av); |
809 | SvREFCNT_dec_NN (av); |
755 | DEC_DEC_DEPTH; |
810 | DEC_DEC_DEPTH; |
756 | return &PL_sv_undef; |
811 | return &PL_sv_undef; |
757 | } |
812 | } |
758 | |
813 | |
759 | static void |
814 | static void |
… | |
… | |
763 | // byte or utf-8 strings as keys, but only when !stringref |
818 | // byte or utf-8 strings as keys, but only when !stringref |
764 | |
819 | |
765 | if (ecb_expect_true (!dec->stringref)) |
820 | if (ecb_expect_true (!dec->stringref)) |
766 | if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8)) |
821 | if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8)) |
767 | { |
822 | { |
768 | I32 len = decode_uint (dec); |
823 | STRLEN len = decode_uint (dec); |
769 | char *key = (char *)dec->cur; |
824 | char *key = (char *)dec->cur; |
770 | |
825 | |
771 | WANT (len); |
826 | WANT (len); |
772 | dec->cur += len; |
827 | dec->cur += len; |
773 | |
828 | |
… | |
… | |
775 | |
830 | |
776 | return; |
831 | return; |
777 | } |
832 | } |
778 | else if (ecb_expect_true ((U8)(*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8)) |
833 | else if (ecb_expect_true ((U8)(*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8)) |
779 | { |
834 | { |
780 | I32 len = decode_uint (dec); |
835 | STRLEN len = decode_uint (dec); |
781 | char *key = (char *)dec->cur; |
836 | char *key = (char *)dec->cur; |
782 | |
837 | |
783 | WANT (len); |
838 | WANT (len); |
784 | dec->cur += len; |
839 | dec->cur += len; |
785 | |
840 | |
… | |
… | |
793 | } |
848 | } |
794 | |
849 | |
795 | SV *k = decode_sv (dec); |
850 | SV *k = decode_sv (dec); |
796 | SV *v = decode_sv (dec); |
851 | SV *v = decode_sv (dec); |
797 | |
852 | |
|
|
853 | // we leak memory if uncaught exceptions are thrown by random magical |
|
|
854 | // methods, and this is hopefully the only place where it can happen, |
|
|
855 | // so if there is a chance of an exception, take the very slow path. |
|
|
856 | // since catching exceptions is "undocumented/internal/forbidden" by |
|
|
857 | // the new p5p powers, we need to call out to a perl function :/ |
|
|
858 | if (ecb_expect_false (SvAMAGIC (k))) |
|
|
859 | { |
|
|
860 | dSP; |
|
|
861 | |
|
|
862 | ENTER; SAVETMPS; |
|
|
863 | PUSHMARK (SP); |
|
|
864 | EXTEND (SP, 3); |
|
|
865 | PUSHs (sv_2mortal (newRV_inc ((SV *)hv))); |
|
|
866 | PUSHs (sv_2mortal (k)); |
|
|
867 | PUSHs (sv_2mortal (v)); |
|
|
868 | |
|
|
869 | PUTBACK; |
|
|
870 | call_pv ("CBOR::XS::_hv_store", G_VOID | G_DISCARD | G_EVAL); |
|
|
871 | SPAGAIN; |
|
|
872 | |
|
|
873 | FREETMPS; LEAVE; |
|
|
874 | |
|
|
875 | if (SvTRUE (ERRSV)) |
|
|
876 | ERR_ERRSV; |
|
|
877 | |
|
|
878 | return; |
|
|
879 | } |
|
|
880 | |
798 | hv_store_ent (hv, k, v, 0); |
881 | hv_store_ent (hv, k, v, 0); |
799 | SvREFCNT_dec (k); |
882 | SvREFCNT_dec_NN (k); |
800 | |
883 | |
801 | fail: |
884 | fail: |
802 | ; |
885 | ; |
803 | } |
886 | } |
804 | |
887 | |
… | |
… | |
826 | decode_he (dec, hv); |
909 | decode_he (dec, hv); |
827 | } |
910 | } |
828 | } |
911 | } |
829 | else |
912 | else |
830 | { |
913 | { |
831 | int pairs = decode_uint (dec); |
914 | UV pairs = decode_uint (dec); |
|
|
915 | |
|
|
916 | WANT (pairs); // complexity check - need at least one byte per value, do not allow supersize hashes |
832 | |
917 | |
833 | while (pairs--) |
918 | while (pairs--) |
834 | decode_he (dec, hv); |
919 | decode_he (dec, hv); |
835 | } |
920 | } |
836 | |
921 | |
837 | DEC_DEC_DEPTH; |
922 | DEC_DEC_DEPTH; |
838 | return newRV_noinc ((SV *)hv); |
923 | return newRV_noinc ((SV *)hv); |
839 | |
924 | |
840 | fail: |
925 | fail: |
841 | SvREFCNT_dec (hv); |
926 | SvREFCNT_dec_NN (hv); |
842 | DEC_DEC_DEPTH; |
927 | DEC_DEC_DEPTH; |
843 | return &PL_sv_undef; |
928 | return &PL_sv_undef; |
844 | } |
929 | } |
845 | |
930 | |
846 | static SV * |
931 | static SV * |
847 | decode_str (dec_t *dec, int utf8) |
932 | decode_str (dec_t *dec, int utf8) |
848 | { |
933 | { |
849 | SV *sv = 0; |
934 | SV *sv = 0; |
850 | |
935 | |
851 | if ((*dec->cur & MINOR_MASK) == MINOR_INDEF) |
936 | if (ecb_expect_false ((*dec->cur & MINOR_MASK) == MINOR_INDEF)) |
852 | { |
937 | { |
853 | // indefinite length strings |
938 | // indefinite length strings |
854 | ++dec->cur; |
939 | ++dec->cur; |
855 | |
940 | |
856 | U8 major = *dec->cur & MAJOR_MISC; |
941 | U8 major = *dec->cur & MAJOR_MISC; |
… | |
… | |
924 | sv = newRV_noinc (decode_sv (dec)); |
1009 | sv = newRV_noinc (decode_sv (dec)); |
925 | break; |
1010 | break; |
926 | |
1011 | |
927 | case CBOR_TAG_STRINGREF_NAMESPACE: |
1012 | case CBOR_TAG_STRINGREF_NAMESPACE: |
928 | { |
1013 | { |
|
|
1014 | // do not use SAVETMPS/FREETMPS, as these will |
|
|
1015 | // erase mortalised caches, e.g. "shareable" |
929 | ENTER; SAVETMPS; |
1016 | ENTER; |
930 | |
1017 | |
931 | SAVESPTR (dec->stringref); |
1018 | SAVESPTR (dec->stringref); |
932 | dec->stringref = (AV *)sv_2mortal ((SV *)newAV ()); |
1019 | dec->stringref = (AV *)sv_2mortal ((SV *)newAV ()); |
933 | |
1020 | |
934 | sv = decode_sv (dec); |
1021 | sv = decode_sv (dec); |
935 | |
1022 | |
936 | FREETMPS; LEAVE; |
1023 | LEAVE; |
937 | } |
1024 | } |
938 | break; |
1025 | break; |
939 | |
1026 | |
940 | case CBOR_TAG_STRINGREF: |
1027 | case CBOR_TAG_STRINGREF: |
941 | { |
1028 | { |
… | |
… | |
992 | } |
1079 | } |
993 | break; |
1080 | break; |
994 | |
1081 | |
995 | case CBOR_TAG_PERL_OBJECT: |
1082 | case CBOR_TAG_PERL_OBJECT: |
996 | { |
1083 | { |
|
|
1084 | if (dec->cbor.flags & F_FORBID_OBJECTS) |
|
|
1085 | goto filter; |
|
|
1086 | |
997 | sv = decode_sv (dec); |
1087 | sv = decode_sv (dec); |
998 | |
1088 | |
999 | if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV) |
1089 | if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV) |
1000 | ERR ("corrupted CBOR data (non-array perl object)"); |
1090 | ERR ("corrupted CBOR data (non-array perl object)"); |
1001 | |
1091 | |
… | |
… | |
1030 | SPAGAIN; |
1120 | SPAGAIN; |
1031 | |
1121 | |
1032 | if (SvTRUE (ERRSV)) |
1122 | if (SvTRUE (ERRSV)) |
1033 | { |
1123 | { |
1034 | FREETMPS; LEAVE; |
1124 | FREETMPS; LEAVE; |
1035 | ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV)))); |
1125 | ERR_ERRSV; |
1036 | } |
1126 | } |
1037 | |
1127 | |
1038 | SvREFCNT_dec (sv); |
1128 | SvREFCNT_dec_NN (sv); |
1039 | sv = SvREFCNT_inc (POPs); |
1129 | sv = SvREFCNT_inc (POPs); |
1040 | |
1130 | |
1041 | PUTBACK; |
1131 | PUTBACK; |
1042 | |
1132 | |
1043 | FREETMPS; LEAVE; |
1133 | FREETMPS; LEAVE; |
1044 | } |
1134 | } |
1045 | break; |
1135 | break; |
1046 | |
1136 | |
1047 | default: |
1137 | default: |
|
|
1138 | filter: |
1048 | { |
1139 | { |
|
|
1140 | SV *tag_sv = newSVuv (tag); |
|
|
1141 | |
1049 | sv = decode_sv (dec); |
1142 | sv = decode_sv (dec); |
1050 | |
1143 | |
1051 | dSP; |
1144 | dSP; |
1052 | ENTER; SAVETMPS; |
1145 | ENTER; SAVETMPS; |
1053 | SAVESTACK_POS (); |
1146 | SAVESTACK_POS (); |
1054 | PUSHMARK (SP); |
1147 | PUSHMARK (SP); |
1055 | EXTEND (SP, 2); |
1148 | EXTEND (SP, 2); |
1056 | PUSHs (newSVuv (tag)); |
1149 | PUSHs (tag_sv); |
1057 | PUSHs (sv); |
1150 | PUSHs (sv); |
1058 | |
1151 | |
1059 | PUTBACK; |
1152 | PUTBACK; |
1060 | int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL); |
1153 | int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL); |
1061 | SPAGAIN; |
1154 | SPAGAIN; |
1062 | |
1155 | |
1063 | if (SvTRUE (ERRSV)) |
1156 | if (SvTRUE (ERRSV)) |
1064 | { |
1157 | { |
|
|
1158 | SvREFCNT_dec_NN (tag_sv); |
1065 | FREETMPS; LEAVE; |
1159 | FREETMPS; LEAVE; |
1066 | ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV)))); |
1160 | ERR_ERRSV; |
1067 | } |
1161 | } |
1068 | |
1162 | |
1069 | if (count) |
1163 | if (count) |
1070 | { |
1164 | { |
|
|
1165 | SvREFCNT_dec_NN (tag_sv); |
1071 | SvREFCNT_dec (sv); |
1166 | SvREFCNT_dec_NN (sv); |
1072 | sv = SvREFCNT_inc (POPs); |
1167 | sv = SvREFCNT_inc_NN (POPs); |
1073 | } |
1168 | } |
1074 | else |
1169 | else |
1075 | { |
1170 | { |
1076 | AV *av = newAV (); |
1171 | AV *av = newAV (); |
1077 | av_push (av, newSVuv (tag)); |
1172 | av_push (av, tag_sv); |
1078 | av_push (av, sv); |
1173 | av_push (av, sv); |
1079 | |
1174 | |
1080 | HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash |
1175 | HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash |
1081 | ? cbor_tagged_stash |
1176 | ? cbor_tagged_stash |
1082 | : gv_stashpv ("CBOR::XS::Tagged" , 1); |
1177 | : gv_stashpv ("CBOR::XS::Tagged" , 1); |
… | |
… | |
1211 | |
1306 | |
1212 | if (dec.err) |
1307 | if (dec.err) |
1213 | { |
1308 | { |
1214 | if (dec.shareable) |
1309 | if (dec.shareable) |
1215 | { |
1310 | { |
1216 | // need to break cyclic links, which whould all be in shareable |
1311 | // need to break cyclic links, which would all be in shareable |
1217 | int i; |
1312 | int i; |
1218 | SV **svp; |
1313 | SV **svp; |
1219 | |
1314 | |
1220 | for (i = av_len (dec.shareable) + 1; i--; ) |
1315 | for (i = av_len (dec.shareable) + 1; i--; ) |
1221 | if ((svp = av_fetch (dec.shareable, i, 0))) |
1316 | if ((svp = av_fetch (dec.shareable, i, 0))) |
1222 | sv_setsv (*svp, &PL_sv_undef); |
1317 | sv_setsv (*svp, &PL_sv_undef); |
1223 | } |
1318 | } |
1224 | |
1319 | |
1225 | SvREFCNT_dec (sv); |
1320 | SvREFCNT_dec_NN (sv); |
|
|
1321 | |
|
|
1322 | if (dec.err_sv) |
|
|
1323 | sv_2mortal (dec.err_sv); |
|
|
1324 | |
1226 | croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur); |
1325 | croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur); |
1227 | } |
1326 | } |
1228 | |
1327 | |
1229 | sv = sv_2mortal (sv); |
1328 | sv = sv_2mortal (sv); |
1230 | |
1329 | |
… | |
… | |
1374 | |
1473 | |
1375 | default_filter = newSVpv ("CBOR::XS::default_filter", 0); |
1474 | default_filter = newSVpv ("CBOR::XS::default_filter", 0); |
1376 | |
1475 | |
1377 | sv_cbor = newSVpv ("CBOR", 0); |
1476 | sv_cbor = newSVpv ("CBOR", 0); |
1378 | SvREADONLY_on (sv_cbor); |
1477 | SvREADONLY_on (sv_cbor); |
|
|
1478 | |
|
|
1479 | assert (("STRLEN must be an unsigned type", 0 <= (STRLEN)-1)); |
1379 | } |
1480 | } |
1380 | |
1481 | |
1381 | PROTOTYPES: DISABLE |
1482 | PROTOTYPES: DISABLE |
1382 | |
1483 | |
1383 | void CLONE (...) |
1484 | void CLONE (...) |
… | |
… | |
1403 | ALIAS: |
1504 | ALIAS: |
1404 | shrink = F_SHRINK |
1505 | shrink = F_SHRINK |
1405 | allow_unknown = F_ALLOW_UNKNOWN |
1506 | allow_unknown = F_ALLOW_UNKNOWN |
1406 | allow_sharing = F_ALLOW_SHARING |
1507 | allow_sharing = F_ALLOW_SHARING |
1407 | allow_cycles = F_ALLOW_CYCLES |
1508 | allow_cycles = F_ALLOW_CYCLES |
|
|
1509 | forbid_objects = F_FORBID_OBJECTS |
1408 | pack_strings = F_PACK_STRINGS |
1510 | pack_strings = F_PACK_STRINGS |
|
|
1511 | text_keys = F_TEXT_KEYS |
1409 | utf8_strings = F_UTF8_STRINGS |
1512 | text_strings = F_TEXT_STRINGS |
1410 | validate_utf8 = F_VALIDATE_UTF8 |
1513 | validate_utf8 = F_VALIDATE_UTF8 |
1411 | PPCODE: |
1514 | PPCODE: |
1412 | { |
1515 | { |
1413 | if (enable) |
1516 | if (enable) |
1414 | self->flags |= ix; |
1517 | self->flags |= ix; |
… | |
… | |
1422 | ALIAS: |
1525 | ALIAS: |
1423 | get_shrink = F_SHRINK |
1526 | get_shrink = F_SHRINK |
1424 | get_allow_unknown = F_ALLOW_UNKNOWN |
1527 | get_allow_unknown = F_ALLOW_UNKNOWN |
1425 | get_allow_sharing = F_ALLOW_SHARING |
1528 | get_allow_sharing = F_ALLOW_SHARING |
1426 | get_allow_cycles = F_ALLOW_CYCLES |
1529 | get_allow_cycles = F_ALLOW_CYCLES |
|
|
1530 | get_forbid_objects = F_FORBID_OBJECTS |
1427 | get_pack_strings = F_PACK_STRINGS |
1531 | get_pack_strings = F_PACK_STRINGS |
|
|
1532 | get_text_keys = F_TEXT_KEYS |
|
|
1533 | get_text_strings = F_TEXT_STRINGS |
1428 | get_validate_utf8 = F_VALIDATE_UTF8 |
1534 | get_validate_utf8 = F_VALIDATE_UTF8 |
1429 | PPCODE: |
1535 | PPCODE: |
1430 | XPUSHs (boolSV (self->flags & ix)); |
1536 | XPUSHs (boolSV (self->flags & ix)); |
1431 | |
1537 | |
1432 | void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) |
1538 | void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) |