… | |
… | |
3 | #include "XSUB.h" |
3 | #include "XSUB.h" |
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 | #include "stdio.h" |
|
|
9 | |
|
|
10 | #if defined(__BORLANDC__) || defined(_MSC_VER) |
|
|
11 | # define snprintf _snprintf // C compilers have this in stdio.h |
|
|
12 | #endif |
8 | |
13 | |
9 | #define F_ASCII 0x00000001UL |
14 | #define F_ASCII 0x00000001UL |
|
|
15 | #define F_LATIN1 0x00000002UL |
10 | #define F_UTF8 0x00000002UL |
16 | #define F_UTF8 0x00000004UL |
11 | #define F_INDENT 0x00000004UL |
17 | #define F_INDENT 0x00000008UL |
12 | #define F_CANONICAL 0x00000008UL |
18 | #define F_CANONICAL 0x00000010UL |
13 | #define F_SPACE_BEFORE 0x00000010UL |
19 | #define F_SPACE_BEFORE 0x00000020UL |
14 | #define F_SPACE_AFTER 0x00000020UL |
20 | #define F_SPACE_AFTER 0x00000040UL |
15 | #define F_ALLOW_NONREF 0x00000080UL |
21 | #define F_ALLOW_NONREF 0x00000100UL |
16 | #define F_SHRINK 0x00000100UL |
22 | #define F_SHRINK 0x00000200UL |
17 | #define F_MAXDEPTH 0xf8000000UL |
23 | #define F_MAXDEPTH 0xf8000000UL |
18 | #define S_MAXDEPTH 27 |
24 | #define S_MAXDEPTH 27 |
19 | |
25 | |
20 | #define DEC_DEPTH(flags) (1UL << ((flags & F_MAXDEPTH) >> S_MAXDEPTH)) |
26 | #define DEC_DEPTH(flags) (1UL << ((flags & F_MAXDEPTH) >> S_MAXDEPTH)) |
21 | |
27 | |
22 | // F_SELFCONVERT? <=> to_json/toJson |
28 | // F_SELFCONVERT? <=> to_json/toJson |
23 | // F_BLESSED? <=> { $__class__$ => } |
29 | // F_BLESSED? <=> { $__class__$ => } |
24 | |
30 | |
25 | #define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER |
31 | #define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER |
26 | #define F_DEFAULT (12UL << S_MAXDEPTH) |
32 | #define F_DEFAULT (9UL << S_MAXDEPTH) |
27 | |
33 | |
28 | #define INIT_SIZE 32 // initial scalar size to be allocated |
34 | #define INIT_SIZE 32 // initial scalar size to be allocated |
29 | #define INDENT_STEP 3 // spaces per indentation level |
35 | #define INDENT_STEP 3 // spaces per indentation level |
30 | |
36 | |
31 | #define UTF8_MAX_LEN 11 // for perls UTF-X: max. number of octets per character |
|
|
32 | #define SHORT_STRING_LEN 512 // special-case strings of up to this size |
37 | #define SHORT_STRING_LEN 512 // special-case strings of up to this size |
33 | |
38 | |
34 | #define SB do { |
39 | #define SB do { |
35 | #define SE } while (0) |
40 | #define SE } while (0) |
36 | |
41 | |
… | |
… | |
76 | { |
81 | { |
77 | *clen = 2; |
82 | *clen = 2; |
78 | return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f); |
83 | return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f); |
79 | } |
84 | } |
80 | else |
85 | else |
|
|
86 | { |
|
|
87 | *clen = (STRLEN)-1; |
81 | return (UV)-1; |
88 | return (UV)-1; |
|
|
89 | } |
82 | } |
90 | } |
83 | |
91 | |
84 | ///////////////////////////////////////////////////////////////////////////// |
92 | ///////////////////////////////////////////////////////////////////////////// |
85 | // encoder |
93 | // encoder |
86 | |
94 | |
… | |
… | |
101 | if (enc->cur + len >= enc->end) |
109 | if (enc->cur + len >= enc->end) |
102 | { |
110 | { |
103 | STRLEN cur = enc->cur - SvPVX (enc->sv); |
111 | STRLEN cur = enc->cur - SvPVX (enc->sv); |
104 | SvGROW (enc->sv, cur + len + 1); |
112 | SvGROW (enc->sv, cur + len + 1); |
105 | enc->cur = SvPVX (enc->sv) + cur; |
113 | enc->cur = SvPVX (enc->sv) + cur; |
106 | enc->end = SvPVX (enc->sv) + SvLEN (enc->sv); |
114 | enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; |
107 | } |
115 | } |
108 | } |
116 | } |
109 | |
117 | |
110 | static void |
118 | static void |
111 | encode_ch (enc_t *enc, char ch) |
119 | encode_ch (enc_t *enc, char ch) |
… | |
… | |
173 | } |
181 | } |
174 | |
182 | |
175 | if (uch > 0x10FFFFUL) |
183 | if (uch > 0x10FFFFUL) |
176 | croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch); |
184 | croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch); |
177 | |
185 | |
178 | if (uch < 0x80 || enc->flags & F_ASCII) |
186 | if (uch < 0x80 || enc->flags & F_ASCII || (enc->flags & F_LATIN1 && uch > 0xFF)) |
179 | { |
187 | { |
180 | if (uch > 0xFFFFUL) |
188 | if (uch > 0xFFFFUL) |
181 | { |
189 | { |
182 | need (enc, len += 11); |
190 | need (enc, len += 11); |
183 | sprintf (enc->cur, "\\u%04x\\u%04x", |
191 | sprintf (enc->cur, "\\u%04x\\u%04x", |
… | |
… | |
197 | *enc->cur++ = hexdigit [(uch >> 0) & 15]; |
205 | *enc->cur++ = hexdigit [(uch >> 0) & 15]; |
198 | } |
206 | } |
199 | |
207 | |
200 | str += clen; |
208 | str += clen; |
201 | } |
209 | } |
|
|
210 | else if (enc->flags & F_LATIN1) |
|
|
211 | { |
|
|
212 | *enc->cur++ = uch; |
|
|
213 | str += clen; |
|
|
214 | } |
202 | else if (is_utf8) |
215 | else if (is_utf8) |
203 | { |
216 | { |
204 | need (enc, len += clen); |
217 | need (enc, len += clen); |
205 | do |
218 | do |
206 | { |
219 | { |
… | |
… | |
208 | } |
221 | } |
209 | while (--clen); |
222 | while (--clen); |
210 | } |
223 | } |
211 | else |
224 | else |
212 | { |
225 | { |
213 | need (enc, len += UTF8_MAX_LEN - 1); // never more than 11 bytes needed |
226 | need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed |
214 | enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0); |
227 | enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0); |
215 | ++str; |
228 | ++str; |
216 | } |
229 | } |
217 | } |
230 | } |
218 | } |
231 | } |
… | |
… | |
405 | |
418 | |
406 | encode_nl (enc); |
419 | encode_nl (enc); |
407 | } |
420 | } |
408 | else |
421 | else |
409 | { |
422 | { |
410 | SV *sv; |
|
|
411 | HE *he = hv_iternext (hv); |
423 | HE *he = hv_iternext (hv); |
412 | |
424 | |
413 | for (;;) |
425 | for (;;) |
414 | { |
426 | { |
415 | encode_indent (enc); |
427 | encode_indent (enc); |
… | |
… | |
430 | |
442 | |
431 | // encode objects, arrays and special \0=false and \1=true values. |
443 | // encode objects, arrays and special \0=false and \1=true values. |
432 | static void |
444 | static void |
433 | encode_rv (enc_t *enc, SV *sv) |
445 | encode_rv (enc_t *enc, SV *sv) |
434 | { |
446 | { |
|
|
447 | svtype svt; |
|
|
448 | |
435 | SvGETMAGIC (sv); |
449 | SvGETMAGIC (sv); |
436 | |
|
|
437 | svtype svt = SvTYPE (sv); |
450 | svt = SvTYPE (sv); |
438 | |
451 | |
439 | if (svt == SVt_PVHV) |
452 | if (svt == SVt_PVHV) |
440 | encode_hv (enc, (HV *)sv); |
453 | encode_hv (enc, (HV *)sv); |
441 | else if (svt == SVt_PVAV) |
454 | else if (svt == SVt_PVAV) |
442 | encode_av (enc, (AV *)sv); |
455 | encode_av (enc, (AV *)sv); |
… | |
… | |
492 | } |
505 | } |
493 | |
506 | |
494 | static SV * |
507 | static SV * |
495 | encode_json (SV *scalar, U32 flags) |
508 | encode_json (SV *scalar, U32 flags) |
496 | { |
509 | { |
|
|
510 | enc_t enc; |
|
|
511 | |
497 | if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar)) |
512 | if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar)) |
498 | croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)"); |
513 | croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)"); |
499 | |
514 | |
500 | enc_t enc; |
|
|
501 | enc.flags = flags; |
515 | enc.flags = flags; |
502 | enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); |
516 | enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); |
503 | enc.cur = SvPVX (enc.sv); |
517 | enc.cur = SvPVX (enc.sv); |
504 | enc.end = SvEND (enc.sv); |
518 | enc.end = SvEND (enc.sv); |
505 | enc.indent = 0; |
519 | enc.indent = 0; |
506 | enc.maxdepth = DEC_DEPTH (flags); |
520 | enc.maxdepth = DEC_DEPTH (flags); |
507 | |
521 | |
508 | SvPOK_only (enc.sv); |
522 | SvPOK_only (enc.sv); |
509 | encode_sv (&enc, scalar); |
523 | encode_sv (&enc, scalar); |
510 | |
524 | |
|
|
525 | SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); |
|
|
526 | *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings |
|
|
527 | |
511 | if (!(flags & (F_ASCII | F_UTF8))) |
528 | if (!(flags & (F_ASCII | F_LATIN1 | F_UTF8))) |
512 | SvUTF8_on (enc.sv); |
529 | SvUTF8_on (enc.sv); |
513 | |
|
|
514 | SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); |
|
|
515 | |
530 | |
516 | if (enc.flags & F_SHRINK) |
531 | if (enc.flags & F_SHRINK) |
517 | shrink (enc.sv); |
532 | shrink (enc.sv); |
518 | |
533 | |
519 | return enc.sv; |
534 | return enc.sv; |
… | |
… | |
591 | SV *sv = 0; |
606 | SV *sv = 0; |
592 | int utf8 = 0; |
607 | int utf8 = 0; |
593 | |
608 | |
594 | do |
609 | do |
595 | { |
610 | { |
596 | char buf [SHORT_STRING_LEN + UTF8_MAX_LEN]; |
611 | char buf [SHORT_STRING_LEN + UTF8_MAXBYTES]; |
597 | char *cur = buf; |
612 | char *cur = buf; |
598 | |
613 | |
599 | do |
614 | do |
600 | { |
615 | { |
601 | unsigned char ch = *(unsigned char *)dec->cur++; |
616 | unsigned char ch = *(unsigned char *)dec->cur++; |
… | |
… | |
667 | } |
682 | } |
668 | else if (ch >= 0x20 && ch <= 0x7f) |
683 | else if (ch >= 0x20 && ch <= 0x7f) |
669 | *cur++ = ch; |
684 | *cur++ = ch; |
670 | else if (ch >= 0x80) |
685 | else if (ch >= 0x80) |
671 | { |
686 | { |
|
|
687 | STRLEN clen; |
|
|
688 | UV uch; |
|
|
689 | |
672 | --dec->cur; |
690 | --dec->cur; |
673 | |
691 | |
674 | STRLEN clen; |
|
|
675 | UV uch = decode_utf8 (dec->cur, dec->end - dec->cur, &clen); |
692 | uch = decode_utf8 (dec->cur, dec->end - dec->cur, &clen); |
676 | if (clen == (STRLEN)-1) |
693 | if (clen == (STRLEN)-1) |
677 | ERR ("malformed UTF-8 character in JSON string"); |
694 | ERR ("malformed UTF-8 character in JSON string"); |
678 | |
695 | |
679 | do |
696 | do |
680 | { |
|
|
681 | *cur++ = *dec->cur++; |
697 | *cur++ = *dec->cur++; |
682 | } |
|
|
683 | while (--clen); |
698 | while (--clen); |
684 | |
699 | |
685 | utf8 = 1; |
700 | utf8 = 1; |
686 | } |
701 | } |
687 | else if (!ch) |
|
|
688 | ERR ("unexpected end of string while parsing json string"); |
|
|
689 | else |
702 | else |
|
|
703 | { |
|
|
704 | --dec->cur; |
|
|
705 | |
|
|
706 | if (!ch) |
|
|
707 | ERR ("unexpected end of string while parsing JSON string"); |
|
|
708 | else |
690 | ERR ("invalid character encountered"); |
709 | ERR ("invalid character encountered while parsing JSON string"); |
691 | |
710 | } |
692 | } |
711 | } |
693 | while (cur < buf + SHORT_STRING_LEN); |
712 | while (cur < buf + SHORT_STRING_LEN); |
694 | |
713 | |
|
|
714 | { |
695 | STRLEN len = cur - buf; |
715 | STRLEN len = cur - buf; |
696 | |
716 | |
697 | if (sv) |
717 | if (sv) |
698 | { |
718 | { |
699 | SvGROW (sv, SvCUR (sv) + len + 1); |
719 | SvGROW (sv, SvCUR (sv) + len + 1); |
700 | memcpy (SvPVX (sv) + SvCUR (sv), buf, len); |
720 | memcpy (SvPVX (sv) + SvCUR (sv), buf, len); |
701 | SvCUR_set (sv, SvCUR (sv) + len); |
721 | SvCUR_set (sv, SvCUR (sv) + len); |
702 | } |
722 | } |
703 | else |
723 | else |
704 | sv = newSVpvn (buf, len); |
724 | sv = newSVpvn (buf, len); |
|
|
725 | } |
705 | } |
726 | } |
706 | while (*dec->cur != '"'); |
727 | while (*dec->cur != '"'); |
707 | |
728 | |
708 | ++dec->cur; |
729 | ++dec->cur; |
709 | |
730 | |
… | |
… | |
953 | ERR ("'null' expected"); |
974 | ERR ("'null' expected"); |
954 | |
975 | |
955 | break; |
976 | break; |
956 | |
977 | |
957 | default: |
978 | default: |
958 | ERR ("malformed json string, neither array, object, number, string or atom"); |
979 | ERR ("malformed JSON string, neither array, object, number, string or atom"); |
959 | break; |
980 | break; |
960 | } |
981 | } |
961 | |
982 | |
962 | fail: |
983 | fail: |
963 | return 0; |
984 | return 0; |
964 | } |
985 | } |
965 | |
986 | |
966 | static SV * |
987 | static SV * |
967 | decode_json (SV *string, U32 flags) |
988 | decode_json (SV *string, U32 flags, UV *offset_return) |
968 | { |
989 | { |
|
|
990 | dec_t dec; |
|
|
991 | UV offset; |
969 | SV *sv; |
992 | SV *sv; |
970 | |
993 | |
|
|
994 | SvGETMAGIC (string); |
971 | SvUPGRADE (string, SVt_PV); |
995 | SvUPGRADE (string, SVt_PV); |
972 | |
996 | |
973 | if (flags & F_UTF8) |
997 | if (flags & F_UTF8) |
974 | sv_utf8_downgrade (string, 0); |
998 | sv_utf8_downgrade (string, 0); |
975 | else |
999 | else |
976 | sv_utf8_upgrade (string); |
1000 | sv_utf8_upgrade (string); |
977 | |
1001 | |
978 | SvGROW (string, SvCUR (string) + 1); // should basically be a NOP |
1002 | SvGROW (string, SvCUR (string) + 1); // should basically be a NOP |
979 | |
1003 | |
980 | dec_t dec; |
|
|
981 | dec.flags = flags; |
1004 | dec.flags = flags; |
982 | dec.cur = SvPVX (string); |
1005 | dec.cur = SvPVX (string); |
983 | dec.end = SvEND (string); |
1006 | dec.end = SvEND (string); |
984 | dec.err = 0; |
1007 | dec.err = 0; |
985 | dec.depth = 0; |
1008 | dec.depth = 0; |
986 | dec.maxdepth = DEC_DEPTH (dec.flags); |
1009 | dec.maxdepth = DEC_DEPTH (dec.flags); |
987 | |
1010 | |
988 | *dec.end = 0; // this should basically be a nop, too, but make sure its there |
1011 | *dec.end = 0; // this should basically be a nop, too, but make sure it's there |
989 | sv = decode_sv (&dec); |
1012 | sv = decode_sv (&dec); |
990 | |
1013 | |
|
|
1014 | if (!(offset_return || !sv)) |
|
|
1015 | { |
|
|
1016 | // check for trailing garbage |
|
|
1017 | decode_ws (&dec); |
|
|
1018 | |
|
|
1019 | if (*dec.cur) |
|
|
1020 | { |
|
|
1021 | dec.err = "garbage after JSON object"; |
|
|
1022 | SvREFCNT_dec (sv); |
|
|
1023 | sv = 0; |
|
|
1024 | } |
|
|
1025 | } |
|
|
1026 | |
|
|
1027 | if (offset_return || !sv) |
|
|
1028 | { |
|
|
1029 | offset = dec.flags & F_UTF8 |
|
|
1030 | ? dec.cur - SvPVX (string) |
|
|
1031 | : utf8_distance (dec.cur, SvPVX (string)); |
|
|
1032 | |
|
|
1033 | if (offset_return) |
|
|
1034 | *offset_return = offset; |
|
|
1035 | } |
|
|
1036 | |
991 | if (!sv) |
1037 | if (!sv) |
992 | { |
1038 | { |
993 | IV offset = dec.flags & F_UTF8 |
|
|
994 | ? dec.cur - SvPVX (string) |
|
|
995 | : utf8_distance (dec.cur, SvPVX (string)); |
|
|
996 | SV *uni = sv_newmortal (); |
1039 | SV *uni = sv_newmortal (); |
997 | |
1040 | |
998 | // horrible hack to silence warning inside pv_uni_display |
1041 | // horrible hack to silence warning inside pv_uni_display |
999 | COP cop = *PL_curcop; |
1042 | COP cop = *PL_curcop; |
1000 | cop.cop_warnings = pWARN_NONE; |
1043 | cop.cop_warnings = pWARN_NONE; |
… | |
… | |
1002 | SAVEVPTR (PL_curcop); |
1045 | SAVEVPTR (PL_curcop); |
1003 | PL_curcop = &cop; |
1046 | PL_curcop = &cop; |
1004 | pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ); |
1047 | pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ); |
1005 | LEAVE; |
1048 | LEAVE; |
1006 | |
1049 | |
1007 | croak ("%s, at character offset %d (%s)", |
1050 | croak ("%s, at character offset %d [\"%s\"]", |
1008 | dec.err, |
1051 | dec.err, |
1009 | (int)offset, |
1052 | (int)offset, |
1010 | dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); |
1053 | dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); |
1011 | } |
1054 | } |
1012 | |
1055 | |
… | |
… | |
1048 | RETVAL |
1091 | RETVAL |
1049 | |
1092 | |
1050 | SV *ascii (SV *self, int enable = 1) |
1093 | SV *ascii (SV *self, int enable = 1) |
1051 | ALIAS: |
1094 | ALIAS: |
1052 | ascii = F_ASCII |
1095 | ascii = F_ASCII |
|
|
1096 | latin1 = F_LATIN1 |
1053 | utf8 = F_UTF8 |
1097 | utf8 = F_UTF8 |
1054 | indent = F_INDENT |
1098 | indent = F_INDENT |
1055 | canonical = F_CANONICAL |
1099 | canonical = F_CANONICAL |
1056 | space_before = F_SPACE_BEFORE |
1100 | space_before = F_SPACE_BEFORE |
1057 | space_after = F_SPACE_AFTER |
1101 | space_after = F_SPACE_AFTER |
… | |
… | |
1069 | RETVAL = newSVsv (self); |
1113 | RETVAL = newSVsv (self); |
1070 | } |
1114 | } |
1071 | OUTPUT: |
1115 | OUTPUT: |
1072 | RETVAL |
1116 | RETVAL |
1073 | |
1117 | |
1074 | SV *max_depth (SV *self, int max_depth = 0x80000000UL) |
1118 | SV *max_depth (SV *self, UV max_depth = 0x80000000UL) |
1075 | CODE: |
1119 | CODE: |
1076 | { |
1120 | { |
1077 | UV *uv = SvJSON (self); |
1121 | UV *uv = SvJSON (self); |
1078 | UV log2 = 0; |
1122 | UV log2 = 0; |
1079 | |
1123 | |
… | |
… | |
1093 | PPCODE: |
1137 | PPCODE: |
1094 | XPUSHs (encode_json (scalar, *SvJSON (self))); |
1138 | XPUSHs (encode_json (scalar, *SvJSON (self))); |
1095 | |
1139 | |
1096 | void decode (SV *self, SV *jsonstr) |
1140 | void decode (SV *self, SV *jsonstr) |
1097 | PPCODE: |
1141 | PPCODE: |
1098 | XPUSHs (decode_json (jsonstr, *SvJSON (self))); |
1142 | XPUSHs (decode_json (jsonstr, *SvJSON (self), 0)); |
|
|
1143 | |
|
|
1144 | void decode_prefix (SV *self, SV *jsonstr) |
|
|
1145 | PPCODE: |
|
|
1146 | { |
|
|
1147 | UV offset; |
|
|
1148 | EXTEND (SP, 2); |
|
|
1149 | PUSHs (decode_json (jsonstr, *SvJSON (self), &offset)); |
|
|
1150 | PUSHs (sv_2mortal (newSVuv (offset))); |
|
|
1151 | } |
1099 | |
1152 | |
1100 | PROTOTYPES: ENABLE |
1153 | PROTOTYPES: ENABLE |
1101 | |
1154 | |
1102 | void to_json (SV *scalar) |
1155 | void to_json (SV *scalar) |
1103 | ALIAS: |
1156 | ALIAS: |
… | |
… | |
1107 | |
1160 | |
1108 | void from_json (SV *jsonstr) |
1161 | void from_json (SV *jsonstr) |
1109 | ALIAS: |
1162 | ALIAS: |
1110 | jsonToObj = 0 |
1163 | jsonToObj = 0 |
1111 | PPCODE: |
1164 | PPCODE: |
1112 | XPUSHs (decode_json (jsonstr, F_DEFAULT | F_UTF8)); |
1165 | XPUSHs (decode_json (jsonstr, F_DEFAULT | F_UTF8, 0)); |
1113 | |
1166 | |