… | |
… | |
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 |
|
|
19 | |
|
|
20 | #define DEC_DEPTH(flags) (1UL << ((flags & F_MAXDEPTH) >> S_MAXDEPTH)) |
|
|
21 | |
|
|
22 | // F_SELFCONVERT? <=> to_json/toJson |
|
|
23 | // F_BLESSED? <=> { $__class__$ => } |
17 | |
24 | |
18 | #define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER |
25 | #define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER |
19 | #define F_DEFAULT 0 |
26 | #define F_DEFAULT (13UL << S_MAXDEPTH) |
20 | |
27 | |
21 | #define INIT_SIZE 32 // initial scalar size to be allocated |
28 | #define INIT_SIZE 32 // initial scalar size to be allocated |
22 | #define INDENT_STEP 3 // spaces per indentation level |
29 | #define INDENT_STEP 3 // spaces per indentation level |
23 | |
30 | |
24 | #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 |
… | |
… | |
81 | typedef struct |
88 | typedef struct |
82 | { |
89 | { |
83 | char *cur; // SvPVX (sv) + current output position |
90 | char *cur; // SvPVX (sv) + current output position |
84 | char *end; // SvEND (sv) |
91 | char *end; // SvEND (sv) |
85 | SV *sv; // result scalar |
92 | SV *sv; // result scalar |
86 | UV flags; // F_* |
93 | U32 flags; // F_* |
87 | int indent; // indentation level |
94 | U32 indent; // indentation level |
88 | int max_depth; // max. recursion level |
95 | U32 maxdepth; // max. indentation/recursion level |
89 | } enc_t; |
96 | } enc_t; |
90 | |
97 | |
91 | static void |
98 | static void |
92 | need (enc_t *enc, STRLEN len) |
99 | need (enc_t *enc, STRLEN len) |
93 | { |
100 | { |
… | |
… | |
444 | } |
451 | } |
445 | else if (SvROK (sv)) |
452 | else if (SvROK (sv)) |
446 | { |
453 | { |
447 | SV *rv = SvRV (sv); |
454 | SV *rv = SvRV (sv); |
448 | |
455 | |
449 | if (enc->indent >= enc->max_depth) |
456 | if (enc->indent >= enc->maxdepth) |
450 | croak ("data structure too deep (hit recursion limit)"); |
457 | croak ("data structure too deep (hit recursion limit)"); |
451 | |
458 | |
452 | switch (SvTYPE (rv)) |
459 | switch (SvTYPE (rv)) |
453 | { |
460 | { |
454 | case SVt_PVAV: encode_av (enc, (AV *)rv); break; |
461 | case SVt_PVAV: encode_av (enc, (AV *)rv); break; |
… | |
… | |
465 | 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", |
466 | SvPV_nolen (sv), SvFLAGS (sv)); |
473 | SvPV_nolen (sv), SvFLAGS (sv)); |
467 | } |
474 | } |
468 | |
475 | |
469 | static SV * |
476 | static SV * |
470 | encode_json (SV *scalar, UV flags) |
477 | encode_json (SV *scalar, U32 flags) |
471 | { |
478 | { |
472 | if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar)) |
479 | if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar)) |
473 | 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)"); |
474 | |
481 | |
475 | enc_t enc; |
482 | enc_t enc; |
476 | enc.flags = flags; |
483 | enc.flags = flags; |
477 | enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); |
484 | enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); |
478 | enc.cur = SvPVX (enc.sv); |
485 | enc.cur = SvPVX (enc.sv); |
479 | enc.end = SvEND (enc.sv); |
486 | enc.end = SvEND (enc.sv); |
480 | enc.indent = 0; |
487 | enc.indent = 0; |
481 | enc.max_depth = 0x7fffffffUL; |
488 | enc.maxdepth = DEC_DEPTH (flags); |
482 | |
489 | |
483 | SvPOK_only (enc.sv); |
490 | SvPOK_only (enc.sv); |
484 | encode_sv (&enc, scalar); |
491 | encode_sv (&enc, scalar); |
485 | |
492 | |
486 | if (!(flags & (F_ASCII | F_UTF8))) |
493 | if (!(flags & (F_ASCII | F_UTF8))) |
… | |
… | |
501 | typedef struct |
508 | typedef struct |
502 | { |
509 | { |
503 | char *cur; // current parser pointer |
510 | char *cur; // current parser pointer |
504 | char *end; // end of input string |
511 | char *end; // end of input string |
505 | const char *err; // parse error, if != 0 |
512 | const char *err; // parse error, if != 0 |
506 | UV flags; // F_* |
513 | U32 flags; // F_* |
|
|
514 | U32 depth; // recursion depth |
|
|
515 | U32 maxdepth; // recursion depth limit |
507 | } dec_t; |
516 | } dec_t; |
508 | |
517 | |
509 | static void |
518 | static void |
510 | decode_ws (dec_t *dec) |
519 | decode_ws (dec_t *dec) |
511 | { |
520 | { |
… | |
… | |
520 | ++dec->cur; |
529 | ++dec->cur; |
521 | } |
530 | } |
522 | } |
531 | } |
523 | |
532 | |
524 | #define ERR(reason) SB dec->err = reason; goto fail; SE |
533 | #define ERR(reason) SB dec->err = reason; goto fail; SE |
|
|
534 | |
525 | #define EXPECT_CH(ch) SB \ |
535 | #define EXPECT_CH(ch) SB \ |
526 | if (*dec->cur != ch) \ |
536 | if (*dec->cur != ch) \ |
527 | ERR (# ch " expected"); \ |
537 | ERR (# ch " expected"); \ |
528 | ++dec->cur; \ |
538 | ++dec->cur; \ |
529 | 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 |
530 | |
543 | |
531 | static SV *decode_sv (dec_t *dec); |
544 | static SV *decode_sv (dec_t *dec); |
532 | |
545 | |
533 | static signed char decode_hexdigit[256]; |
546 | static signed char decode_hexdigit[256]; |
534 | |
547 | |
… | |
… | |
778 | static SV * |
791 | static SV * |
779 | decode_av (dec_t *dec) |
792 | decode_av (dec_t *dec) |
780 | { |
793 | { |
781 | AV *av = newAV (); |
794 | AV *av = newAV (); |
782 | |
795 | |
|
|
796 | DEC_INC_DEPTH; |
783 | decode_ws (dec); |
797 | decode_ws (dec); |
|
|
798 | |
784 | if (*dec->cur == ']') |
799 | if (*dec->cur == ']') |
785 | ++dec->cur; |
800 | ++dec->cur; |
786 | else |
801 | else |
787 | for (;;) |
802 | for (;;) |
788 | { |
803 | { |
… | |
… | |
806 | ERR (", or ] expected while parsing array"); |
821 | ERR (", or ] expected while parsing array"); |
807 | |
822 | |
808 | ++dec->cur; |
823 | ++dec->cur; |
809 | } |
824 | } |
810 | |
825 | |
|
|
826 | DEC_DEC_DEPTH; |
811 | return newRV_noinc ((SV *)av); |
827 | return newRV_noinc ((SV *)av); |
812 | |
828 | |
813 | fail: |
829 | fail: |
814 | SvREFCNT_dec (av); |
830 | SvREFCNT_dec (av); |
|
|
831 | DEC_DEC_DEPTH; |
815 | return 0; |
832 | return 0; |
816 | } |
833 | } |
817 | |
834 | |
818 | static SV * |
835 | static SV * |
819 | decode_hv (dec_t *dec) |
836 | decode_hv (dec_t *dec) |
820 | { |
837 | { |
821 | HV *hv = newHV (); |
838 | HV *hv = newHV (); |
822 | |
839 | |
|
|
840 | DEC_INC_DEPTH; |
823 | decode_ws (dec); |
841 | decode_ws (dec); |
|
|
842 | |
824 | if (*dec->cur == '}') |
843 | if (*dec->cur == '}') |
825 | ++dec->cur; |
844 | ++dec->cur; |
826 | else |
845 | else |
827 | for (;;) |
846 | for (;;) |
828 | { |
847 | { |
… | |
… | |
841 | { |
860 | { |
842 | SvREFCNT_dec (key); |
861 | SvREFCNT_dec (key); |
843 | goto fail; |
862 | goto fail; |
844 | } |
863 | } |
845 | |
864 | |
846 | //TODO: optimise |
|
|
847 | hv_store_ent (hv, key, value, 0); |
865 | hv_store_ent (hv, key, value, 0); |
|
|
866 | SvREFCNT_dec (key); |
848 | |
867 | |
849 | decode_ws (dec); |
868 | decode_ws (dec); |
850 | |
869 | |
851 | if (*dec->cur == '}') |
870 | if (*dec->cur == '}') |
852 | { |
871 | { |
… | |
… | |
858 | ERR (", or } expected while parsing object/hash"); |
877 | ERR (", or } expected while parsing object/hash"); |
859 | |
878 | |
860 | ++dec->cur; |
879 | ++dec->cur; |
861 | } |
880 | } |
862 | |
881 | |
|
|
882 | DEC_DEC_DEPTH; |
863 | return newRV_noinc ((SV *)hv); |
883 | return newRV_noinc ((SV *)hv); |
864 | |
884 | |
865 | fail: |
885 | fail: |
866 | SvREFCNT_dec (hv); |
886 | SvREFCNT_dec (hv); |
|
|
887 | DEC_DEC_DEPTH; |
867 | return 0; |
888 | return 0; |
868 | } |
889 | } |
869 | |
890 | |
870 | static SV * |
891 | static SV * |
871 | decode_sv (dec_t *dec) |
892 | decode_sv (dec_t *dec) |
… | |
… | |
923 | fail: |
944 | fail: |
924 | return 0; |
945 | return 0; |
925 | } |
946 | } |
926 | |
947 | |
927 | static SV * |
948 | static SV * |
928 | decode_json (SV *string, UV flags) |
949 | decode_json (SV *string, U32 flags) |
929 | { |
950 | { |
930 | SV *sv; |
951 | SV *sv; |
931 | |
952 | |
932 | if (flags & F_UTF8) |
953 | if (flags & F_UTF8) |
933 | sv_utf8_downgrade (string, 0); |
954 | sv_utf8_downgrade (string, 0); |
… | |
… | |
935 | sv_utf8_upgrade (string); |
956 | sv_utf8_upgrade (string); |
936 | |
957 | |
937 | SvGROW (string, SvCUR (string) + 1); // should basically be a NOP |
958 | SvGROW (string, SvCUR (string) + 1); // should basically be a NOP |
938 | |
959 | |
939 | dec_t dec; |
960 | dec_t dec; |
940 | dec.flags = flags; |
961 | dec.flags = flags; |
941 | dec.cur = SvPVX (string); |
962 | dec.cur = SvPVX (string); |
942 | dec.end = SvEND (string); |
963 | dec.end = SvEND (string); |
943 | dec.err = 0; |
964 | dec.err = 0; |
|
|
965 | dec.depth = 0; |
|
|
966 | dec.maxdepth = DEC_DEPTH (dec.flags); |
944 | |
967 | |
|
|
968 | *dec.end = 0; // this should basically be a nop, too, but make sure its there |
945 | sv = decode_sv (&dec); |
969 | sv = decode_sv (&dec); |
946 | |
970 | |
947 | if (!sv) |
971 | if (!sv) |
948 | { |
972 | { |
949 | IV offset = dec.flags & F_UTF8 |
973 | IV offset = dec.flags & F_UTF8 |
… | |
… | |
982 | BOOT: |
1006 | BOOT: |
983 | { |
1007 | { |
984 | int i; |
1008 | int i; |
985 | |
1009 | |
986 | memset (decode_hexdigit, 0xff, 256); |
1010 | memset (decode_hexdigit, 0xff, 256); |
|
|
1011 | |
987 | for (i = 10; i--; ) |
1012 | for (i = 0; i < 256; ++i) |
988 | decode_hexdigit ['0' + i] = i; |
1013 | decode_hexdigit [i] = |
989 | |
1014 | i >= '0' && i <= '9' ? i - '0' |
990 | for (i = 7; i--; ) |
1015 | : i >= 'a' && i <= 'f' ? i - 'a' + 10 |
991 | { |
1016 | : i >= 'A' && i <= 'F' ? i - 'A' + 10 |
992 | decode_hexdigit ['a' + i] = 10 + i; |
1017 | : -1; |
993 | decode_hexdigit ['A' + i] = 10 + i; |
|
|
994 | } |
|
|
995 | |
1018 | |
996 | json_stash = gv_stashpv ("JSON::XS", 1); |
1019 | json_stash = gv_stashpv ("JSON::XS", 1); |
997 | } |
1020 | } |
998 | |
1021 | |
999 | PROTOTYPES: DISABLE |
1022 | PROTOTYPES: DISABLE |
… | |
… | |
1026 | RETVAL = newSVsv (self); |
1049 | RETVAL = newSVsv (self); |
1027 | } |
1050 | } |
1028 | OUTPUT: |
1051 | OUTPUT: |
1029 | RETVAL |
1052 | RETVAL |
1030 | |
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 | |
1031 | void encode (SV *self, SV *scalar) |
1072 | void encode (SV *self, SV *scalar) |
1032 | PPCODE: |
1073 | PPCODE: |
1033 | XPUSHs (encode_json (scalar, *SvJSON (self))); |
1074 | XPUSHs (encode_json (scalar, *SvJSON (self))); |
1034 | |
1075 | |
1035 | void decode (SV *self, SV *jsonstr) |
1076 | void decode (SV *self, SV *jsonstr) |
… | |
… | |
1040 | |
1081 | |
1041 | void to_json (SV *scalar) |
1082 | void to_json (SV *scalar) |
1042 | ALIAS: |
1083 | ALIAS: |
1043 | objToJson = 0 |
1084 | objToJson = 0 |
1044 | PPCODE: |
1085 | PPCODE: |
1045 | XPUSHs (encode_json (scalar, F_UTF8)); |
1086 | XPUSHs (encode_json (scalar, F_DEFAULT | F_UTF8)); |
1046 | |
1087 | |
1047 | void from_json (SV *jsonstr) |
1088 | void from_json (SV *jsonstr) |
1048 | ALIAS: |
1089 | ALIAS: |
1049 | jsonToObj = 0 |
1090 | jsonToObj = 0 |
1050 | PPCODE: |
1091 | PPCODE: |
1051 | XPUSHs (decode_json (jsonstr, F_UTF8)); |
1092 | XPUSHs (decode_json (jsonstr, F_DEFAULT | F_UTF8)); |
1052 | |
1093 | |