… | |
… | |
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 | |
11 | |
|
|
12 | #define ECB_NO_THREADS 1 |
12 | #include "ecb.h" |
13 | #include "ecb.h" |
13 | |
14 | |
14 | // compatibility with perl <5.18 |
15 | // compatibility with perl <5.18 |
15 | #ifndef HvNAMELEN_get |
16 | #ifndef HvNAMELEN_get |
16 | # define HvNAMELEN_get(hv) strlen (HvNAME (hv)) |
17 | # define HvNAMELEN_get(hv) strlen (HvNAME (hv)) |
… | |
… | |
97 | }; |
98 | }; |
98 | |
99 | |
99 | #define F_SHRINK 0x00000001UL |
100 | #define F_SHRINK 0x00000001UL |
100 | #define F_ALLOW_UNKNOWN 0x00000002UL |
101 | #define F_ALLOW_UNKNOWN 0x00000002UL |
101 | #define F_ALLOW_SHARING 0x00000004UL |
102 | #define F_ALLOW_SHARING 0x00000004UL |
|
|
103 | #define F_ALLOW_CYCLES 0x00000008UL |
102 | #define F_PACK_STRINGS 0x00000008UL |
104 | #define F_PACK_STRINGS 0x00000010UL |
|
|
105 | #define F_TEXT_KEYS 0x00000020UL |
|
|
106 | #define F_TEXT_STRINGS 0x00000040UL |
|
|
107 | #define F_VALIDATE_UTF8 0x00000080UL |
103 | |
108 | |
104 | #define INIT_SIZE 32 // initial scalar size to be allocated |
109 | #define INIT_SIZE 32 // initial scalar size to be allocated |
105 | |
110 | |
106 | #define SB do { |
111 | #define SB do { |
107 | #define SE } while (0) |
112 | #define SE } while (0) |
… | |
… | |
126 | typedef struct { |
131 | typedef struct { |
127 | U32 flags; |
132 | U32 flags; |
128 | U32 max_depth; |
133 | U32 max_depth; |
129 | STRLEN max_size; |
134 | STRLEN max_size; |
130 | SV *filter; |
135 | SV *filter; |
|
|
136 | |
|
|
137 | // for the incremental parser |
|
|
138 | STRLEN incr_pos; // the current offset into the text |
|
|
139 | STRLEN incr_need; // minimum bytes needed to decode |
|
|
140 | AV *incr_count; // for every nesting level, the number of outstanding values, or -1 for indef. |
131 | } CBOR; |
141 | } CBOR; |
132 | |
142 | |
133 | ecb_inline void |
143 | ecb_inline void |
134 | cbor_init (CBOR *cbor) |
144 | cbor_init (CBOR *cbor) |
135 | { |
145 | { |
… | |
… | |
139 | |
149 | |
140 | ecb_inline void |
150 | ecb_inline void |
141 | cbor_free (CBOR *cbor) |
151 | cbor_free (CBOR *cbor) |
142 | { |
152 | { |
143 | SvREFCNT_dec (cbor->filter); |
153 | SvREFCNT_dec (cbor->filter); |
|
|
154 | SvREFCNT_dec (cbor->incr_count); |
144 | } |
155 | } |
145 | |
156 | |
146 | ///////////////////////////////////////////////////////////////////////////// |
157 | ///////////////////////////////////////////////////////////////////////////// |
147 | // utility functions |
158 | // utility functions |
148 | |
159 | |
… | |
… | |
267 | encode_tag (enc_t *enc, UV tag) |
278 | encode_tag (enc_t *enc, UV tag) |
268 | { |
279 | { |
269 | encode_uint (enc, MAJOR_TAG, tag); |
280 | encode_uint (enc, MAJOR_TAG, tag); |
270 | } |
281 | } |
271 | |
282 | |
|
|
283 | // exceptional (hopefully) slow path for byte strings that need to be utf8-encoded |
|
|
284 | ecb_noinline static void |
|
|
285 | encode_str_utf8 (enc_t *enc, int utf8, char *str, STRLEN len) |
|
|
286 | { |
|
|
287 | STRLEN ulen = len; |
|
|
288 | U8 *p, *pend = (U8 *)str + len; |
|
|
289 | |
|
|
290 | for (p = (U8 *)str; p < pend; ++p) |
|
|
291 | ulen += *p >> 7; // count set high bits |
|
|
292 | |
|
|
293 | encode_uint (enc, MAJOR_TEXT, ulen); |
|
|
294 | |
|
|
295 | need (enc, ulen); |
|
|
296 | for (p = (U8 *)str; p < pend; ++p) |
|
|
297 | if (*p < 0x80) |
|
|
298 | *enc->cur++ = *p; |
|
|
299 | else |
|
|
300 | { |
|
|
301 | *enc->cur++ = 0xc0 + (*p >> 6); |
|
|
302 | *enc->cur++ = 0x80 + (*p & 63); |
|
|
303 | } |
|
|
304 | } |
|
|
305 | |
272 | ecb_inline void |
306 | ecb_inline void |
273 | encode_str (enc_t *enc, int utf8, char *str, STRLEN len) |
307 | encode_str (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len) |
274 | { |
308 | { |
|
|
309 | if (ecb_expect_false (upgrade_utf8)) |
|
|
310 | if (!utf8) |
|
|
311 | { |
|
|
312 | encode_str_utf8 (enc, utf8, str, len); |
|
|
313 | return; |
|
|
314 | } |
|
|
315 | |
275 | encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len); |
316 | encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len); |
276 | need (enc, len); |
317 | need (enc, len); |
277 | memcpy (enc->cur, str, len); |
318 | memcpy (enc->cur, str, len); |
278 | enc->cur += len; |
319 | enc->cur += len; |
279 | } |
320 | } |
280 | |
321 | |
281 | static void |
322 | ecb_inline void |
282 | encode_strref (enc_t *enc, int utf8, char *str, STRLEN len) |
323 | encode_strref (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len) |
283 | { |
324 | { |
284 | if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS)) |
325 | if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS)) |
285 | { |
326 | { |
286 | SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1); |
327 | SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1); |
287 | |
328 | |
… | |
… | |
298 | sv_setuv (*svp, enc->stringref_idx); |
339 | sv_setuv (*svp, enc->stringref_idx); |
299 | ++enc->stringref_idx; |
340 | ++enc->stringref_idx; |
300 | } |
341 | } |
301 | } |
342 | } |
302 | |
343 | |
303 | encode_str (enc, utf8, str, len); |
344 | encode_str (enc, upgrade_utf8, utf8, str, len); |
304 | } |
345 | } |
305 | |
346 | |
306 | static void encode_sv (enc_t *enc, SV *sv); |
347 | static void encode_sv (enc_t *enc, SV *sv); |
307 | |
348 | |
308 | static void |
349 | static void |
… | |
… | |
315 | |
356 | |
316 | ++enc->depth; |
357 | ++enc->depth; |
317 | |
358 | |
318 | encode_uint (enc, MAJOR_ARRAY, len + 1); |
359 | encode_uint (enc, MAJOR_ARRAY, len + 1); |
319 | |
360 | |
|
|
361 | if (SvMAGICAL (av)) |
320 | for (i = 0; i <= len; ++i) |
362 | for (i = 0; i <= len; ++i) |
321 | { |
363 | { |
322 | SV **svp = av_fetch (av, i, 0); |
364 | SV **svp = av_fetch (av, i, 0); |
323 | encode_sv (enc, svp ? *svp : &PL_sv_undef); |
365 | encode_sv (enc, svp ? *svp : &PL_sv_undef); |
324 | } |
366 | } |
|
|
367 | else |
|
|
368 | for (i = 0; i <= len; ++i) |
|
|
369 | { |
|
|
370 | SV *sv = AvARRAY (av)[i]; |
|
|
371 | encode_sv (enc, sv ? sv : &PL_sv_undef); |
|
|
372 | } |
325 | |
373 | |
326 | --enc->depth; |
374 | --enc->depth; |
327 | } |
375 | } |
328 | |
376 | |
329 | static void |
377 | static void |
… | |
… | |
347 | while ((he = hv_iternext (hv))) |
395 | while ((he = hv_iternext (hv))) |
348 | { |
396 | { |
349 | if (HeKLEN (he) == HEf_SVKEY) |
397 | if (HeKLEN (he) == HEf_SVKEY) |
350 | encode_sv (enc, HeSVKEY (he)); |
398 | encode_sv (enc, HeSVKEY (he)); |
351 | else |
399 | else |
352 | encode_strref (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he)); |
400 | encode_strref (enc, enc->cbor.flags & (F_TEXT_KEYS | F_TEXT_STRINGS), HeKUTF8 (he), HeKEY (he), HeKLEN (he)); |
353 | |
401 | |
354 | encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he)); |
402 | encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he)); |
355 | } |
403 | } |
356 | |
404 | |
357 | if (mg) |
405 | if (mg) |
… | |
… | |
433 | |
481 | |
434 | if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0))) |
482 | if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0))) |
435 | { |
483 | { |
436 | dSP; |
484 | dSP; |
437 | |
485 | |
438 | ENTER; SAVETMPS; PUSHMARK (SP); |
486 | ENTER; SAVETMPS; |
|
|
487 | PUSHMARK (SP); |
439 | // we re-bless the reference to get overload and other niceties right |
488 | // we re-bless the reference to get overload and other niceties right |
440 | XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); |
489 | XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); |
441 | |
490 | |
442 | PUTBACK; |
491 | PUTBACK; |
443 | // G_SCALAR ensures that return value is 1 |
492 | // G_SCALAR ensures that return value is 1 |
… | |
… | |
456 | } |
505 | } |
457 | else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0) |
506 | else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0) |
458 | { |
507 | { |
459 | dSP; |
508 | dSP; |
460 | |
509 | |
461 | ENTER; SAVETMPS; PUSHMARK (SP); |
510 | ENTER; SAVETMPS; |
|
|
511 | SAVESTACK_POS (); |
|
|
512 | PUSHMARK (SP); |
462 | EXTEND (SP, 2); |
513 | EXTEND (SP, 2); |
463 | // we re-bless the reference to get overload and other niceties right |
514 | // we re-bless the reference to get overload and other niceties right |
464 | PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); |
515 | PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); |
465 | PUSHs (sv_cbor); |
516 | PUSHs (sv_cbor); |
466 | |
517 | |
… | |
… | |
472 | if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv) |
523 | if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv) |
473 | croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash)); |
524 | croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash)); |
474 | |
525 | |
475 | encode_tag (enc, CBOR_TAG_PERL_OBJECT); |
526 | encode_tag (enc, CBOR_TAG_PERL_OBJECT); |
476 | encode_uint (enc, MAJOR_ARRAY, count + 1); |
527 | encode_uint (enc, MAJOR_ARRAY, count + 1); |
477 | encode_strref (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash)); |
528 | encode_strref (enc, 0, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash)); |
478 | |
529 | |
479 | while (count) |
530 | while (count) |
480 | encode_sv (enc, SP[1 - count--]); |
531 | encode_sv (enc, SP[1 - count--]); |
481 | |
532 | |
482 | PUTBACK; |
533 | PUTBACK; |
… | |
… | |
541 | |
592 | |
542 | if (SvPOKp (sv)) |
593 | if (SvPOKp (sv)) |
543 | { |
594 | { |
544 | STRLEN len; |
595 | STRLEN len; |
545 | char *str = SvPV (sv, len); |
596 | char *str = SvPV (sv, len); |
546 | encode_strref (enc, SvUTF8 (sv), str, len); |
597 | encode_strref (enc, enc->cbor.flags & F_TEXT_STRINGS, SvUTF8 (sv), str, len); |
547 | } |
598 | } |
548 | else if (SvNOKp (sv)) |
599 | else if (SvNOKp (sv)) |
549 | encode_nv (enc, sv); |
600 | encode_nv (enc, sv); |
550 | else if (SvIOKp (sv)) |
601 | else if (SvIOKp (sv)) |
551 | { |
602 | { |
… | |
… | |
568 | } |
619 | } |
569 | |
620 | |
570 | static SV * |
621 | static SV * |
571 | encode_cbor (SV *scalar, CBOR *cbor) |
622 | encode_cbor (SV *scalar, CBOR *cbor) |
572 | { |
623 | { |
573 | enc_t enc = { }; |
624 | enc_t enc = { 0 }; |
574 | |
625 | |
575 | enc.cbor = *cbor; |
626 | enc.cbor = *cbor; |
576 | enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); |
627 | enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); |
577 | enc.cur = SvPVX (enc.sv); |
628 | enc.cur = SvPVX (enc.sv); |
578 | enc.end = SvEND (enc.sv); |
629 | enc.end = SvEND (enc.sv); |
… | |
… | |
616 | |
667 | |
617 | #define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE |
668 | #define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE |
618 | |
669 | |
619 | #define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data") |
670 | #define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data") |
620 | |
671 | |
621 | #define DEC_INC_DEPTH if (++dec->depth > dec->cbor.max_depth) ERR (ERR_NESTING_EXCEEDED) |
672 | #define DEC_INC_DEPTH if (ecb_expect_false (++dec->depth > dec->cbor.max_depth)) ERR (ERR_NESTING_EXCEEDED) |
622 | #define DEC_DEC_DEPTH --dec->depth |
673 | #define DEC_DEC_DEPTH --dec->depth |
623 | |
674 | |
624 | static UV |
675 | static UV |
625 | decode_uint (dec_t *dec) |
676 | decode_uint (dec_t *dec) |
626 | { |
677 | { |
… | |
… | |
728 | { |
779 | { |
729 | // for speed reasons, we specialcase single-string |
780 | // for speed reasons, we specialcase single-string |
730 | // byte or utf-8 strings as keys, but only when !stringref |
781 | // byte or utf-8 strings as keys, but only when !stringref |
731 | |
782 | |
732 | if (ecb_expect_true (!dec->stringref)) |
783 | if (ecb_expect_true (!dec->stringref)) |
733 | if (ecb_expect_true ((*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8)) |
784 | if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8)) |
734 | { |
785 | { |
735 | I32 len = decode_uint (dec); |
786 | STRLEN len = decode_uint (dec); |
736 | char *key = (char *)dec->cur; |
787 | char *key = (char *)dec->cur; |
737 | |
788 | |
|
|
789 | WANT (len); |
738 | dec->cur += len; |
790 | dec->cur += len; |
739 | |
791 | |
740 | if (ecb_expect_false (dec->stringref)) |
|
|
741 | av_push (dec->stringref, newSVpvn (key, len)); |
|
|
742 | |
|
|
743 | hv_store (hv, key, len, decode_sv (dec), 0); |
792 | hv_store (hv, key, len, decode_sv (dec), 0); |
744 | |
793 | |
745 | return; |
794 | return; |
746 | } |
795 | } |
747 | else if (ecb_expect_true ((*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8)) |
796 | else if (ecb_expect_true ((U8)(*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8)) |
748 | { |
797 | { |
749 | I32 len = decode_uint (dec); |
798 | STRLEN len = decode_uint (dec); |
750 | char *key = (char *)dec->cur; |
799 | char *key = (char *)dec->cur; |
751 | |
800 | |
|
|
801 | WANT (len); |
752 | dec->cur += len; |
802 | dec->cur += len; |
753 | |
803 | |
754 | if (ecb_expect_false (dec->stringref)) |
804 | if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8)) |
755 | av_push (dec->stringref, newSVpvn_utf8 (key, len, 1)); |
805 | if (!is_utf8_string (key, len)) |
|
|
806 | ERR ("corrupted CBOR data (invalid UTF-8 in map key)"); |
756 | |
807 | |
757 | hv_store (hv, key, -len, decode_sv (dec), 0); |
808 | hv_store (hv, key, -len, decode_sv (dec), 0); |
758 | |
809 | |
759 | return; |
810 | return; |
760 | } |
811 | } |
… | |
… | |
762 | SV *k = decode_sv (dec); |
813 | SV *k = decode_sv (dec); |
763 | SV *v = decode_sv (dec); |
814 | SV *v = decode_sv (dec); |
764 | |
815 | |
765 | hv_store_ent (hv, k, v, 0); |
816 | hv_store_ent (hv, k, v, 0); |
766 | SvREFCNT_dec (k); |
817 | SvREFCNT_dec (k); |
|
|
818 | |
|
|
819 | fail: |
|
|
820 | ; |
767 | } |
821 | } |
768 | |
822 | |
769 | static SV * |
823 | static SV * |
770 | decode_hv (dec_t *dec) |
824 | decode_hv (dec_t *dec) |
771 | { |
825 | { |
… | |
… | |
853 | && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1)) |
907 | && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1)) |
854 | av_push (dec->stringref, SvREFCNT_inc_NN (sv)); |
908 | av_push (dec->stringref, SvREFCNT_inc_NN (sv)); |
855 | } |
909 | } |
856 | |
910 | |
857 | if (utf8) |
911 | if (utf8) |
|
|
912 | { |
|
|
913 | if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8)) |
|
|
914 | if (!is_utf8_string (SvPVX (sv), SvCUR (sv))) |
|
|
915 | ERR ("corrupted CBOR data (invalid UTF-8 in text string)"); |
|
|
916 | |
858 | SvUTF8_on (sv); |
917 | SvUTF8_on (sv); |
|
|
918 | } |
859 | |
919 | |
860 | return sv; |
920 | return sv; |
861 | |
921 | |
862 | fail: |
922 | fail: |
863 | SvREFCNT_dec (sv); |
923 | SvREFCNT_dec (sv); |
… | |
… | |
882 | sv = newRV_noinc (decode_sv (dec)); |
942 | sv = newRV_noinc (decode_sv (dec)); |
883 | break; |
943 | break; |
884 | |
944 | |
885 | case CBOR_TAG_STRINGREF_NAMESPACE: |
945 | case CBOR_TAG_STRINGREF_NAMESPACE: |
886 | { |
946 | { |
|
|
947 | // do nmot use SAVETMPS/FREETMPS, as these will |
|
|
948 | // erase mortalised caches, e.g. "shareable" |
887 | ENTER; SAVETMPS; |
949 | ENTER; |
888 | |
950 | |
889 | SAVESPTR (dec->stringref); |
951 | SAVESPTR (dec->stringref); |
890 | dec->stringref = (AV *)sv_2mortal ((SV *)newAV ()); |
952 | dec->stringref = (AV *)sv_2mortal ((SV *)newAV ()); |
891 | |
953 | |
892 | sv = decode_sv (dec); |
954 | sv = decode_sv (dec); |
893 | |
955 | |
894 | FREETMPS; LEAVE; |
956 | LEAVE; |
895 | } |
957 | } |
896 | break; |
958 | break; |
897 | |
959 | |
898 | case CBOR_TAG_STRINGREF: |
960 | case CBOR_TAG_STRINGREF: |
899 | { |
961 | { |
… | |
… | |
912 | case CBOR_TAG_VALUE_SHAREABLE: |
974 | case CBOR_TAG_VALUE_SHAREABLE: |
913 | { |
975 | { |
914 | if (ecb_expect_false (!dec->shareable)) |
976 | if (ecb_expect_false (!dec->shareable)) |
915 | dec->shareable = (AV *)sv_2mortal ((SV *)newAV ()); |
977 | dec->shareable = (AV *)sv_2mortal ((SV *)newAV ()); |
916 | |
978 | |
|
|
979 | if (dec->cbor.flags & F_ALLOW_CYCLES) |
|
|
980 | { |
917 | sv = newSV (0); |
981 | sv = newSV (0); |
918 | av_push (dec->shareable, SvREFCNT_inc_NN (sv)); |
982 | av_push (dec->shareable, SvREFCNT_inc_NN (sv)); |
919 | |
983 | |
920 | SV *osv = decode_sv (dec); |
984 | SV *osv = decode_sv (dec); |
921 | sv_setsv (sv, osv); |
985 | sv_setsv (sv, osv); |
922 | SvREFCNT_dec_NN (osv); |
986 | SvREFCNT_dec_NN (osv); |
|
|
987 | } |
|
|
988 | else |
|
|
989 | { |
|
|
990 | av_push (dec->shareable, &PL_sv_undef); |
|
|
991 | int idx = AvFILLp (dec->shareable); |
|
|
992 | sv = decode_sv (dec); |
|
|
993 | av_store (dec->shareable, idx, SvREFCNT_inc_NN (sv)); |
|
|
994 | } |
923 | } |
995 | } |
924 | break; |
996 | break; |
925 | |
997 | |
926 | case CBOR_TAG_VALUE_SHAREDREF: |
998 | case CBOR_TAG_VALUE_SHAREDREF: |
927 | { |
999 | { |
… | |
… | |
932 | |
1004 | |
933 | if (!dec->shareable || (int)idx > AvFILLp (dec->shareable)) |
1005 | if (!dec->shareable || (int)idx > AvFILLp (dec->shareable)) |
934 | ERR ("corrupted CBOR data (sharedref index out of bounds)"); |
1006 | ERR ("corrupted CBOR data (sharedref index out of bounds)"); |
935 | |
1007 | |
936 | sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]); |
1008 | sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]); |
|
|
1009 | |
|
|
1010 | if (sv == &PL_sv_undef) |
|
|
1011 | ERR ("cyclic CBOR data structure found, but allow_cycles is not enabled"); |
937 | } |
1012 | } |
938 | break; |
1013 | break; |
939 | |
1014 | |
940 | case CBOR_TAG_PERL_OBJECT: |
1015 | case CBOR_TAG_PERL_OBJECT: |
941 | { |
1016 | { |
… | |
… | |
956 | if (!method) |
1031 | if (!method) |
957 | ERR ("cannot decode perl-object (package does not have a THAW method)"); |
1032 | ERR ("cannot decode perl-object (package does not have a THAW method)"); |
958 | |
1033 | |
959 | dSP; |
1034 | dSP; |
960 | |
1035 | |
961 | ENTER; SAVETMPS; PUSHMARK (SP); |
1036 | ENTER; SAVETMPS; |
|
|
1037 | PUSHMARK (SP); |
962 | EXTEND (SP, len + 1); |
1038 | EXTEND (SP, len + 1); |
963 | // we re-bless the reference to get overload and other niceties right |
1039 | // we re-bless the reference to get overload and other niceties right |
964 | PUSHs (*av_fetch (av, 0, 1)); |
1040 | PUSHs (*av_fetch (av, 0, 1)); |
965 | PUSHs (sv_cbor); |
1041 | PUSHs (sv_cbor); |
966 | |
1042 | |
… | |
… | |
991 | default: |
1067 | default: |
992 | { |
1068 | { |
993 | sv = decode_sv (dec); |
1069 | sv = decode_sv (dec); |
994 | |
1070 | |
995 | dSP; |
1071 | dSP; |
996 | ENTER; SAVETMPS; PUSHMARK (SP); |
1072 | ENTER; SAVETMPS; |
|
|
1073 | SAVESTACK_POS (); |
|
|
1074 | PUSHMARK (SP); |
997 | EXTEND (SP, 2); |
1075 | EXTEND (SP, 2); |
998 | PUSHs (newSVuv (tag)); |
1076 | PUSHs (newSVuv (tag)); |
999 | PUSHs (sv); |
1077 | PUSHs (sv); |
1000 | |
1078 | |
1001 | PUTBACK; |
1079 | PUTBACK; |
… | |
… | |
1110 | |
1188 | |
1111 | return newSVnv (ecb_binary64_to_double (fp)); |
1189 | return newSVnv (ecb_binary64_to_double (fp)); |
1112 | } |
1190 | } |
1113 | |
1191 | |
1114 | // 0..19 unassigned simple |
1192 | // 0..19 unassigned simple |
1115 | // 24 reserved + unassigned (reserved values are not encodable) |
1193 | // 24 reserved + unassigned simple (reserved values are not encodable) |
|
|
1194 | // 28-30 unassigned misc |
|
|
1195 | // 31 break code |
1116 | default: |
1196 | default: |
1117 | ERR ("corrupted CBOR data (reserved/unassigned major 7 value)"); |
1197 | ERR ("corrupted CBOR data (reserved/unassigned/unexpected major 7 value)"); |
1118 | } |
1198 | } |
1119 | |
1199 | |
1120 | break; |
1200 | break; |
1121 | } |
1201 | } |
1122 | |
1202 | |
… | |
… | |
1125 | } |
1205 | } |
1126 | |
1206 | |
1127 | static SV * |
1207 | static SV * |
1128 | decode_cbor (SV *string, CBOR *cbor, char **offset_return) |
1208 | decode_cbor (SV *string, CBOR *cbor, char **offset_return) |
1129 | { |
1209 | { |
1130 | dec_t dec = { }; |
1210 | dec_t dec = { 0 }; |
1131 | SV *sv; |
1211 | SV *sv; |
1132 | STRLEN len; |
1212 | STRLEN len; |
1133 | char *data = SvPVbyte (string, len); |
1213 | char *data = SvPVbyte (string, len); |
1134 | |
1214 | |
1135 | if (len > cbor->max_size && cbor->max_size) |
1215 | if (len > cbor->max_size && cbor->max_size) |
… | |
… | |
1149 | if (dec.cur != dec.end && !dec.err) |
1229 | if (dec.cur != dec.end && !dec.err) |
1150 | dec.err = "garbage after CBOR object"; |
1230 | dec.err = "garbage after CBOR object"; |
1151 | |
1231 | |
1152 | if (dec.err) |
1232 | if (dec.err) |
1153 | { |
1233 | { |
|
|
1234 | if (dec.shareable) |
|
|
1235 | { |
|
|
1236 | // need to break cyclic links, which whould all be in shareable |
|
|
1237 | int i; |
|
|
1238 | SV **svp; |
|
|
1239 | |
|
|
1240 | for (i = av_len (dec.shareable) + 1; i--; ) |
|
|
1241 | if ((svp = av_fetch (dec.shareable, i, 0))) |
|
|
1242 | sv_setsv (*svp, &PL_sv_undef); |
|
|
1243 | } |
|
|
1244 | |
1154 | SvREFCNT_dec (sv); |
1245 | SvREFCNT_dec (sv); |
1155 | croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur); |
1246 | croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur); |
1156 | } |
1247 | } |
1157 | |
1248 | |
1158 | sv = sv_2mortal (sv); |
1249 | sv = sv_2mortal (sv); |
1159 | |
1250 | |
1160 | return sv; |
1251 | return sv; |
1161 | } |
1252 | } |
1162 | |
1253 | |
|
|
1254 | ///////////////////////////////////////////////////////////////////////////// |
|
|
1255 | // incremental parser |
|
|
1256 | |
|
|
1257 | #define INCR_DONE(cbor) (AvFILLp (cbor->incr_count) < 0) |
|
|
1258 | |
|
|
1259 | // returns 0 for notyet, 1 for success or error |
|
|
1260 | static int |
|
|
1261 | incr_parse (CBOR *self, SV *cborstr) |
|
|
1262 | { |
|
|
1263 | STRLEN cur; |
|
|
1264 | SvPV (cborstr, cur); |
|
|
1265 | |
|
|
1266 | while (ecb_expect_true (self->incr_need <= cur)) |
|
|
1267 | { |
|
|
1268 | // table of integer count bytes |
|
|
1269 | static I8 incr_len[MINOR_MASK + 1] = { |
|
|
1270 | 0, 0, 0, 0, 0, 0, 0, 0, |
|
|
1271 | 0, 0, 0, 0, 0, 0, 0, 0, |
|
|
1272 | 0, 0, 0, 0, 0, 0, 0, 0, |
|
|
1273 | 1, 2, 4, 8,-1,-1,-1,-2 |
|
|
1274 | }; |
|
|
1275 | |
|
|
1276 | const U8 *p = SvPVX (cborstr) + self->incr_pos; |
|
|
1277 | U8 m = *p & MINOR_MASK; |
|
|
1278 | IV count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]); |
|
|
1279 | I8 ilen = incr_len[m]; |
|
|
1280 | |
|
|
1281 | self->incr_need = self->incr_pos + 1; |
|
|
1282 | |
|
|
1283 | if (ecb_expect_false (ilen < 0)) |
|
|
1284 | { |
|
|
1285 | if (m != MINOR_INDEF) |
|
|
1286 | return 1; // error |
|
|
1287 | |
|
|
1288 | if (*p == (MAJOR_MISC | MINOR_INDEF)) |
|
|
1289 | { |
|
|
1290 | if (count >= 0) |
|
|
1291 | return 1; // error |
|
|
1292 | |
|
|
1293 | count = 1; |
|
|
1294 | } |
|
|
1295 | else |
|
|
1296 | { |
|
|
1297 | av_push (self->incr_count, newSViv (-1)); //TODO: nest |
|
|
1298 | count = -1; |
|
|
1299 | } |
|
|
1300 | } |
|
|
1301 | else |
|
|
1302 | { |
|
|
1303 | self->incr_need += ilen; |
|
|
1304 | if (ecb_expect_false (self->incr_need > cur)) |
|
|
1305 | return 0; |
|
|
1306 | |
|
|
1307 | int major = *p >> MAJOR_SHIFT; |
|
|
1308 | |
|
|
1309 | switch (major) |
|
|
1310 | { |
|
|
1311 | case MAJOR_TAG >> MAJOR_SHIFT: |
|
|
1312 | ++count; // tags merely prefix another value |
|
|
1313 | break; |
|
|
1314 | |
|
|
1315 | case MAJOR_BYTES >> MAJOR_SHIFT: |
|
|
1316 | case MAJOR_TEXT >> MAJOR_SHIFT: |
|
|
1317 | case MAJOR_ARRAY >> MAJOR_SHIFT: |
|
|
1318 | case MAJOR_MAP >> MAJOR_SHIFT: |
|
|
1319 | { |
|
|
1320 | UV len; |
|
|
1321 | |
|
|
1322 | if (ecb_expect_false (ilen)) |
|
|
1323 | { |
|
|
1324 | len = 0; |
|
|
1325 | |
|
|
1326 | do { |
|
|
1327 | len = (len << 8) | *++p; |
|
|
1328 | } while (--ilen); |
|
|
1329 | } |
|
|
1330 | else |
|
|
1331 | len = m; |
|
|
1332 | |
|
|
1333 | switch (major) |
|
|
1334 | { |
|
|
1335 | case MAJOR_BYTES >> MAJOR_SHIFT: |
|
|
1336 | case MAJOR_TEXT >> MAJOR_SHIFT: |
|
|
1337 | self->incr_need += len; |
|
|
1338 | if (ecb_expect_false (self->incr_need > cur)) |
|
|
1339 | return 0; |
|
|
1340 | |
|
|
1341 | break; |
|
|
1342 | |
|
|
1343 | case MAJOR_MAP >> MAJOR_SHIFT: |
|
|
1344 | len <<= 1; |
|
|
1345 | case MAJOR_ARRAY >> MAJOR_SHIFT: |
|
|
1346 | if (len) |
|
|
1347 | { |
|
|
1348 | av_push (self->incr_count, newSViv (len + 1)); //TODO: nest |
|
|
1349 | count = len + 1; |
|
|
1350 | } |
|
|
1351 | break; |
|
|
1352 | } |
|
|
1353 | } |
|
|
1354 | } |
|
|
1355 | } |
|
|
1356 | |
|
|
1357 | self->incr_pos = self->incr_need; |
|
|
1358 | |
|
|
1359 | if (count > 0) |
|
|
1360 | { |
|
|
1361 | while (!--count) |
|
|
1362 | { |
|
|
1363 | if (!AvFILLp (self->incr_count)) |
|
|
1364 | return 1; // done |
|
|
1365 | |
|
|
1366 | SvREFCNT_dec_NN (av_pop (self->incr_count)); |
|
|
1367 | count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]); |
|
|
1368 | } |
|
|
1369 | |
|
|
1370 | SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]) = count; |
|
|
1371 | } |
|
|
1372 | } |
|
|
1373 | |
|
|
1374 | return 0; |
|
|
1375 | } |
|
|
1376 | |
|
|
1377 | |
1163 | ///////////////////////////////////////////////////////////////////////////// |
1378 | ///////////////////////////////////////////////////////////////////////////// |
1164 | // XS interface functions |
1379 | // XS interface functions |
1165 | |
1380 | |
1166 | MODULE = CBOR::XS PACKAGE = CBOR::XS |
1381 | MODULE = CBOR::XS PACKAGE = CBOR::XS |
1167 | |
1382 | |
… | |
… | |
1179 | |
1394 | |
1180 | default_filter = newSVpv ("CBOR::XS::default_filter", 0); |
1395 | default_filter = newSVpv ("CBOR::XS::default_filter", 0); |
1181 | |
1396 | |
1182 | sv_cbor = newSVpv ("CBOR", 0); |
1397 | sv_cbor = newSVpv ("CBOR", 0); |
1183 | SvREADONLY_on (sv_cbor); |
1398 | SvREADONLY_on (sv_cbor); |
|
|
1399 | |
|
|
1400 | assert (("STRLEN must be an unsigned type", 0 <= (STRLEN)-1)); |
1184 | } |
1401 | } |
1185 | |
1402 | |
1186 | PROTOTYPES: DISABLE |
1403 | PROTOTYPES: DISABLE |
1187 | |
1404 | |
1188 | void CLONE (...) |
1405 | void CLONE (...) |
… | |
… | |
1207 | void shrink (CBOR *self, int enable = 1) |
1424 | void shrink (CBOR *self, int enable = 1) |
1208 | ALIAS: |
1425 | ALIAS: |
1209 | shrink = F_SHRINK |
1426 | shrink = F_SHRINK |
1210 | allow_unknown = F_ALLOW_UNKNOWN |
1427 | allow_unknown = F_ALLOW_UNKNOWN |
1211 | allow_sharing = F_ALLOW_SHARING |
1428 | allow_sharing = F_ALLOW_SHARING |
|
|
1429 | allow_cycles = F_ALLOW_CYCLES |
1212 | pack_strings = F_PACK_STRINGS |
1430 | pack_strings = F_PACK_STRINGS |
|
|
1431 | text_keys = F_TEXT_KEYS |
|
|
1432 | text_strings = F_TEXT_STRINGS |
|
|
1433 | validate_utf8 = F_VALIDATE_UTF8 |
1213 | PPCODE: |
1434 | PPCODE: |
1214 | { |
1435 | { |
1215 | if (enable) |
1436 | if (enable) |
1216 | self->flags |= ix; |
1437 | self->flags |= ix; |
1217 | else |
1438 | else |
… | |
… | |
1223 | void get_shrink (CBOR *self) |
1444 | void get_shrink (CBOR *self) |
1224 | ALIAS: |
1445 | ALIAS: |
1225 | get_shrink = F_SHRINK |
1446 | get_shrink = F_SHRINK |
1226 | get_allow_unknown = F_ALLOW_UNKNOWN |
1447 | get_allow_unknown = F_ALLOW_UNKNOWN |
1227 | get_allow_sharing = F_ALLOW_SHARING |
1448 | get_allow_sharing = F_ALLOW_SHARING |
|
|
1449 | get_allow_cycles = F_ALLOW_CYCLES |
1228 | get_pack_strings = F_PACK_STRINGS |
1450 | get_pack_strings = F_PACK_STRINGS |
|
|
1451 | get_text_keys = F_TEXT_KEYS |
|
|
1452 | get_text_strings = F_TEXT_STRINGS |
|
|
1453 | get_validate_utf8 = F_VALIDATE_UTF8 |
1229 | PPCODE: |
1454 | PPCODE: |
1230 | XPUSHs (boolSV (self->flags & ix)); |
1455 | XPUSHs (boolSV (self->flags & ix)); |
1231 | |
1456 | |
1232 | void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) |
1457 | void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) |
1233 | PPCODE: |
1458 | PPCODE: |
… | |
… | |
1282 | EXTEND (SP, 2); |
1507 | EXTEND (SP, 2); |
1283 | PUSHs (sv); |
1508 | PUSHs (sv); |
1284 | PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); |
1509 | PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); |
1285 | } |
1510 | } |
1286 | |
1511 | |
|
|
1512 | void incr_parse (CBOR *self, SV *cborstr) |
|
|
1513 | ALIAS: |
|
|
1514 | incr_parse_multiple = 1 |
|
|
1515 | PPCODE: |
|
|
1516 | { |
|
|
1517 | if (SvUTF8 (cborstr)) |
|
|
1518 | sv_utf8_downgrade (cborstr, 0); |
|
|
1519 | |
|
|
1520 | if (!self->incr_count) |
|
|
1521 | { |
|
|
1522 | self->incr_count = newAV (); |
|
|
1523 | self->incr_pos = 0; |
|
|
1524 | self->incr_need = 1; |
|
|
1525 | |
|
|
1526 | av_push (self->incr_count, newSViv (1)); |
|
|
1527 | } |
|
|
1528 | |
|
|
1529 | do |
|
|
1530 | { |
|
|
1531 | if (!incr_parse (self, cborstr)) |
|
|
1532 | { |
|
|
1533 | if (self->incr_need > self->max_size && self->max_size) |
|
|
1534 | croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu", |
|
|
1535 | (unsigned long)self->incr_need, (unsigned long)self->max_size); |
|
|
1536 | |
|
|
1537 | break; |
|
|
1538 | } |
|
|
1539 | |
|
|
1540 | SV *sv; |
|
|
1541 | char *offset; |
|
|
1542 | |
|
|
1543 | PUTBACK; sv = decode_cbor (cborstr, self, &offset); SPAGAIN; |
|
|
1544 | XPUSHs (sv); |
|
|
1545 | |
|
|
1546 | sv_chop (cborstr, offset); |
|
|
1547 | |
|
|
1548 | av_clear (self->incr_count); |
|
|
1549 | av_push (self->incr_count, newSViv (1)); |
|
|
1550 | |
|
|
1551 | self->incr_pos = 0; |
|
|
1552 | self->incr_need = self->incr_pos + 1; |
|
|
1553 | } |
|
|
1554 | while (ix); |
|
|
1555 | } |
|
|
1556 | |
|
|
1557 | void incr_reset (CBOR *self) |
|
|
1558 | CODE: |
|
|
1559 | { |
|
|
1560 | SvREFCNT_dec (self->incr_count); |
|
|
1561 | self->incr_count = 0; |
|
|
1562 | } |
|
|
1563 | |
1287 | void DESTROY (CBOR *self) |
1564 | void DESTROY (CBOR *self) |
1288 | PPCODE: |
1565 | PPCODE: |
1289 | cbor_free (self); |
1566 | cbor_free (self); |
1290 | |
1567 | |
1291 | PROTOTYPES: ENABLE |
1568 | PROTOTYPES: ENABLE |