… | |
… | |
4 | |
4 | |
5 | #include "assert.h" |
5 | #include "assert.h" |
6 | #include "string.h" |
6 | #include "string.h" |
7 | #include "stdlib.h" |
7 | #include "stdlib.h" |
8 | |
8 | |
9 | #define F_ASCII 0x00000001 |
9 | #define F_ASCII 0x00000001UL |
10 | #define F_UTF8 0x00000002 |
10 | #define F_UTF8 0x00000002UL |
11 | #define F_INDENT 0x00000004 |
11 | #define F_INDENT 0x00000004UL |
12 | #define F_CANONICAL 0x00000008 |
12 | #define F_CANONICAL 0x00000008UL |
13 | #define F_SPACE_BEFORE 0x00000010 |
13 | #define F_SPACE_BEFORE 0x00000010UL |
14 | #define F_SPACE_AFTER 0x00000020 |
14 | #define F_SPACE_AFTER 0x00000020UL |
15 | #define F_ALLOW_NONREF 0x00000080 |
15 | #define F_ALLOW_NONREF 0x00000080UL |
16 | #define F_SHRINK 0x00000100 |
16 | #define F_SHRINK 0x00000100UL |
|
|
17 | #define F_MAXDEPTH 0xf8000000UL |
|
|
18 | #define S_MAXDEPTH 27 |
17 | |
19 | |
18 | // F_SKIPINVALID? |
20 | #define DEC_DEPTH(flags) (1UL << ((flags & F_MAXDEPTH) >> S_MAXDEPTH)) |
19 | // F_EXECCODEREF? |
21 | |
20 | // F_SELFCONVERT? <=> { &__class__ => } |
22 | // F_SELFCONVERT? <=> to_json/toJson |
|
|
23 | // F_BLESSED? <=> { $__class__$ => } |
21 | |
24 | |
22 | #define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER |
25 | #define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER |
23 | #define F_DEFAULT 0 |
26 | #define F_DEFAULT (13UL << S_MAXDEPTH) |
24 | |
27 | |
25 | #define INIT_SIZE 32 // initial scalar size to be allocated |
28 | #define INIT_SIZE 32 // initial scalar size to be allocated |
26 | #define INDENT_STEP 3 // spaces per indentation level |
29 | #define INDENT_STEP 3 // spaces per indentation level |
27 | |
30 | |
28 | #define UTF8_MAX_LEN 11 // for perls UTF-X: max. number of octets per character |
31 | #define UTF8_MAX_LEN 11 // for perls UTF-X: max. number of octets per character |
… | |
… | |
85 | typedef struct |
88 | typedef struct |
86 | { |
89 | { |
87 | char *cur; // SvPVX (sv) + current output position |
90 | char *cur; // SvPVX (sv) + current output position |
88 | char *end; // SvEND (sv) |
91 | char *end; // SvEND (sv) |
89 | SV *sv; // result scalar |
92 | SV *sv; // result scalar |
90 | UV flags; // F_* |
93 | U32 flags; // F_* |
91 | int indent; // indentation level |
94 | U32 indent; // indentation level |
92 | int max_depth; // max. recursion level |
95 | U32 maxdepth; // max. indentation/recursion level |
93 | } enc_t; |
96 | } enc_t; |
94 | |
97 | |
95 | static void |
98 | static void |
96 | need (enc_t *enc, STRLEN len) |
99 | need (enc_t *enc, STRLEN len) |
97 | { |
100 | { |
… | |
… | |
448 | } |
451 | } |
449 | else if (SvROK (sv)) |
452 | else if (SvROK (sv)) |
450 | { |
453 | { |
451 | SV *rv = SvRV (sv); |
454 | SV *rv = SvRV (sv); |
452 | |
455 | |
453 | if (enc->indent >= enc->max_depth) |
456 | if (enc->indent >= enc->maxdepth) |
454 | croak ("data structure too deep (hit recursion limit)"); |
457 | croak ("data structure too deep (hit recursion limit)"); |
455 | |
458 | |
456 | switch (SvTYPE (rv)) |
459 | switch (SvTYPE (rv)) |
457 | { |
460 | { |
458 | case SVt_PVAV: encode_av (enc, (AV *)rv); break; |
461 | case SVt_PVAV: encode_av (enc, (AV *)rv); break; |
… | |
… | |
469 | croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this", |
472 | croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this", |
470 | SvPV_nolen (sv), SvFLAGS (sv)); |
473 | SvPV_nolen (sv), SvFLAGS (sv)); |
471 | } |
474 | } |
472 | |
475 | |
473 | static SV * |
476 | static SV * |
474 | encode_json (SV *scalar, UV flags) |
477 | encode_json (SV *scalar, U32 flags) |
475 | { |
478 | { |
476 | if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar)) |
479 | if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar)) |
477 | croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)"); |
480 | croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)"); |
478 | |
481 | |
479 | enc_t enc; |
482 | enc_t enc; |
480 | enc.flags = flags; |
483 | enc.flags = flags; |
481 | enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); |
484 | enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); |
482 | enc.cur = SvPVX (enc.sv); |
485 | enc.cur = SvPVX (enc.sv); |
483 | enc.end = SvEND (enc.sv); |
486 | enc.end = SvEND (enc.sv); |
484 | enc.indent = 0; |
487 | enc.indent = 0; |
485 | enc.max_depth = 0x7fffffffUL; |
488 | enc.maxdepth = DEC_DEPTH (flags); |
486 | |
489 | |
487 | SvPOK_only (enc.sv); |
490 | SvPOK_only (enc.sv); |
488 | encode_sv (&enc, scalar); |
491 | encode_sv (&enc, scalar); |
489 | |
492 | |
490 | if (!(flags & (F_ASCII | F_UTF8))) |
493 | if (!(flags & (F_ASCII | F_UTF8))) |
… | |
… | |
505 | typedef struct |
508 | typedef struct |
506 | { |
509 | { |
507 | char *cur; // current parser pointer |
510 | char *cur; // current parser pointer |
508 | char *end; // end of input string |
511 | char *end; // end of input string |
509 | const char *err; // parse error, if != 0 |
512 | const char *err; // parse error, if != 0 |
510 | UV flags; // F_* |
513 | U32 flags; // F_* |
|
|
514 | U32 depth; // recursion depth |
|
|
515 | U32 maxdepth; // recursion depth limit |
511 | } dec_t; |
516 | } dec_t; |
512 | |
517 | |
513 | static void |
518 | static void |
514 | decode_ws (dec_t *dec) |
519 | decode_ws (dec_t *dec) |
515 | { |
520 | { |
… | |
… | |
524 | ++dec->cur; |
529 | ++dec->cur; |
525 | } |
530 | } |
526 | } |
531 | } |
527 | |
532 | |
528 | #define ERR(reason) SB dec->err = reason; goto fail; SE |
533 | #define ERR(reason) SB dec->err = reason; goto fail; SE |
|
|
534 | |
529 | #define EXPECT_CH(ch) SB \ |
535 | #define EXPECT_CH(ch) SB \ |
530 | if (*dec->cur != ch) \ |
536 | if (*dec->cur != ch) \ |
531 | ERR (# ch " expected"); \ |
537 | ERR (# ch " expected"); \ |
532 | ++dec->cur; \ |
538 | ++dec->cur; \ |
533 | SE |
539 | SE |
|
|
540 | |
|
|
541 | #define DEC_INC_DEPTH if (++dec->depth > dec->maxdepth) ERR ("json datastructure exceeds maximum nesting level (set a higher max_depth)") |
|
|
542 | #define DEC_DEC_DEPTH --dec->depth |
534 | |
543 | |
535 | static SV *decode_sv (dec_t *dec); |
544 | static SV *decode_sv (dec_t *dec); |
536 | |
545 | |
537 | static signed char decode_hexdigit[256]; |
546 | static signed char decode_hexdigit[256]; |
538 | |
547 | |
… | |
… | |
782 | static SV * |
791 | static SV * |
783 | decode_av (dec_t *dec) |
792 | decode_av (dec_t *dec) |
784 | { |
793 | { |
785 | AV *av = newAV (); |
794 | AV *av = newAV (); |
786 | |
795 | |
|
|
796 | DEC_INC_DEPTH; |
787 | decode_ws (dec); |
797 | decode_ws (dec); |
|
|
798 | |
788 | if (*dec->cur == ']') |
799 | if (*dec->cur == ']') |
789 | ++dec->cur; |
800 | ++dec->cur; |
790 | else |
801 | else |
791 | for (;;) |
802 | for (;;) |
792 | { |
803 | { |
… | |
… | |
810 | ERR (", or ] expected while parsing array"); |
821 | ERR (", or ] expected while parsing array"); |
811 | |
822 | |
812 | ++dec->cur; |
823 | ++dec->cur; |
813 | } |
824 | } |
814 | |
825 | |
|
|
826 | DEC_DEC_DEPTH; |
815 | return newRV_noinc ((SV *)av); |
827 | return newRV_noinc ((SV *)av); |
816 | |
828 | |
817 | fail: |
829 | fail: |
818 | SvREFCNT_dec (av); |
830 | SvREFCNT_dec (av); |
|
|
831 | DEC_DEC_DEPTH; |
819 | return 0; |
832 | return 0; |
820 | } |
833 | } |
821 | |
834 | |
822 | static SV * |
835 | static SV * |
823 | decode_hv (dec_t *dec) |
836 | decode_hv (dec_t *dec) |
824 | { |
837 | { |
825 | HV *hv = newHV (); |
838 | HV *hv = newHV (); |
826 | |
839 | |
|
|
840 | DEC_INC_DEPTH; |
827 | decode_ws (dec); |
841 | decode_ws (dec); |
|
|
842 | |
828 | if (*dec->cur == '}') |
843 | if (*dec->cur == '}') |
829 | ++dec->cur; |
844 | ++dec->cur; |
830 | else |
845 | else |
831 | for (;;) |
846 | for (;;) |
832 | { |
847 | { |
… | |
… | |
845 | { |
860 | { |
846 | SvREFCNT_dec (key); |
861 | SvREFCNT_dec (key); |
847 | goto fail; |
862 | goto fail; |
848 | } |
863 | } |
849 | |
864 | |
850 | //TODO: optimise |
|
|
851 | hv_store_ent (hv, key, value, 0); |
865 | hv_store_ent (hv, key, value, 0); |
|
|
866 | SvREFCNT_dec (key); |
852 | |
867 | |
853 | decode_ws (dec); |
868 | decode_ws (dec); |
854 | |
869 | |
855 | if (*dec->cur == '}') |
870 | if (*dec->cur == '}') |
856 | { |
871 | { |
… | |
… | |
862 | ERR (", or } expected while parsing object/hash"); |
877 | ERR (", or } expected while parsing object/hash"); |
863 | |
878 | |
864 | ++dec->cur; |
879 | ++dec->cur; |
865 | } |
880 | } |
866 | |
881 | |
|
|
882 | DEC_DEC_DEPTH; |
867 | return newRV_noinc ((SV *)hv); |
883 | return newRV_noinc ((SV *)hv); |
868 | |
884 | |
869 | fail: |
885 | fail: |
870 | SvREFCNT_dec (hv); |
886 | SvREFCNT_dec (hv); |
|
|
887 | DEC_DEC_DEPTH; |
871 | return 0; |
888 | return 0; |
872 | } |
889 | } |
873 | |
890 | |
874 | static SV * |
891 | static SV * |
875 | decode_sv (dec_t *dec) |
892 | decode_sv (dec_t *dec) |
… | |
… | |
927 | fail: |
944 | fail: |
928 | return 0; |
945 | return 0; |
929 | } |
946 | } |
930 | |
947 | |
931 | static SV * |
948 | static SV * |
932 | decode_json (SV *string, UV flags) |
949 | decode_json (SV *string, U32 flags) |
933 | { |
950 | { |
934 | SV *sv; |
951 | SV *sv; |
935 | |
952 | |
936 | if (flags & F_UTF8) |
953 | if (flags & F_UTF8) |
937 | sv_utf8_downgrade (string, 0); |
954 | sv_utf8_downgrade (string, 0); |
… | |
… | |
939 | sv_utf8_upgrade (string); |
956 | sv_utf8_upgrade (string); |
940 | |
957 | |
941 | SvGROW (string, SvCUR (string) + 1); // should basically be a NOP |
958 | SvGROW (string, SvCUR (string) + 1); // should basically be a NOP |
942 | |
959 | |
943 | dec_t dec; |
960 | dec_t dec; |
944 | dec.flags = flags; |
961 | dec.flags = flags; |
945 | dec.cur = SvPVX (string); |
962 | dec.cur = SvPVX (string); |
946 | dec.end = SvEND (string); |
963 | dec.end = SvEND (string); |
947 | dec.err = 0; |
964 | dec.err = 0; |
|
|
965 | dec.depth = 0; |
|
|
966 | dec.maxdepth = DEC_DEPTH (dec.flags); |
948 | |
967 | |
|
|
968 | *SvEND (sv) = 0; // this shou[ld basically be a nop, too |
949 | sv = decode_sv (&dec); |
969 | sv = decode_sv (&dec); |
950 | |
970 | |
951 | if (!sv) |
971 | if (!sv) |
952 | { |
972 | { |
953 | IV offset = dec.flags & F_UTF8 |
973 | IV offset = dec.flags & F_UTF8 |
… | |
… | |
986 | BOOT: |
1006 | BOOT: |
987 | { |
1007 | { |
988 | int i; |
1008 | int i; |
989 | |
1009 | |
990 | memset (decode_hexdigit, 0xff, 256); |
1010 | memset (decode_hexdigit, 0xff, 256); |
|
|
1011 | |
991 | for (i = 10; i--; ) |
1012 | for (i = 0; i < 256; ++i) |
992 | decode_hexdigit ['0' + i] = i; |
1013 | decode_hexdigit [i] = |
993 | |
1014 | i >= '0' && i <= '9' ? i - '0' |
994 | for (i = 7; i--; ) |
1015 | : i >= 'a' && i <= 'f' ? i - 'a' + 10 |
995 | { |
1016 | : i >= 'A' && i <= 'F' ? i - 'A' + 10 |
996 | decode_hexdigit ['a' + i] = 10 + i; |
1017 | : -1; |
997 | decode_hexdigit ['A' + i] = 10 + i; |
|
|
998 | } |
|
|
999 | |
1018 | |
1000 | json_stash = gv_stashpv ("JSON::XS", 1); |
1019 | json_stash = gv_stashpv ("JSON::XS", 1); |
1001 | } |
1020 | } |
1002 | |
1021 | |
1003 | PROTOTYPES: DISABLE |
1022 | PROTOTYPES: DISABLE |
… | |
… | |
1030 | RETVAL = newSVsv (self); |
1049 | RETVAL = newSVsv (self); |
1031 | } |
1050 | } |
1032 | OUTPUT: |
1051 | OUTPUT: |
1033 | RETVAL |
1052 | RETVAL |
1034 | |
1053 | |
|
|
1054 | SV *max_depth (SV *self, int max_depth = 0x80000000UL) |
|
|
1055 | CODE: |
|
|
1056 | { |
|
|
1057 | UV *uv = SvJSON (self); |
|
|
1058 | UV log2 = 0; |
|
|
1059 | |
|
|
1060 | if (max_depth > 0x80000000UL) max_depth = 0x80000000UL; |
|
|
1061 | |
|
|
1062 | while ((1UL << log2) < max_depth) |
|
|
1063 | ++log2; |
|
|
1064 | |
|
|
1065 | *uv = *uv & ~F_MAXDEPTH | (log2 << S_MAXDEPTH); |
|
|
1066 | |
|
|
1067 | RETVAL = newSVsv (self); |
|
|
1068 | } |
|
|
1069 | OUTPUT: |
|
|
1070 | RETVAL |
|
|
1071 | |
1035 | void encode (SV *self, SV *scalar) |
1072 | void encode (SV *self, SV *scalar) |
1036 | PPCODE: |
1073 | PPCODE: |
1037 | XPUSHs (encode_json (scalar, *SvJSON (self))); |
1074 | XPUSHs (encode_json (scalar, *SvJSON (self))); |
1038 | |
1075 | |
1039 | void decode (SV *self, SV *jsonstr) |
1076 | void decode (SV *self, SV *jsonstr) |
… | |
… | |
1044 | |
1081 | |
1045 | void to_json (SV *scalar) |
1082 | void to_json (SV *scalar) |
1046 | ALIAS: |
1083 | ALIAS: |
1047 | objToJson = 0 |
1084 | objToJson = 0 |
1048 | PPCODE: |
1085 | PPCODE: |
1049 | XPUSHs (encode_json (scalar, F_UTF8)); |
1086 | XPUSHs (encode_json (scalar, F_DEFAULT | F_UTF8)); |
1050 | |
1087 | |
1051 | void from_json (SV *jsonstr) |
1088 | void from_json (SV *jsonstr) |
1052 | ALIAS: |
1089 | ALIAS: |
1053 | jsonToObj = 0 |
1090 | jsonToObj = 0 |
1054 | PPCODE: |
1091 | PPCODE: |
1055 | XPUSHs (decode_json (jsonstr, F_UTF8)); |
1092 | XPUSHs (decode_json (jsonstr, F_DEFAULT | F_UTF8)); |
1056 | |
1093 | |