… | |
… | |
12 | #define F_CANONICAL 0x00000008 |
12 | #define F_CANONICAL 0x00000008 |
13 | #define F_SPACE_BEFORE 0x00000010 |
13 | #define F_SPACE_BEFORE 0x00000010 |
14 | #define F_SPACE_AFTER 0x00000020 |
14 | #define F_SPACE_AFTER 0x00000020 |
15 | #define F_JSON_RPC 0x00000040 |
15 | #define F_JSON_RPC 0x00000040 |
16 | #define F_ALLOW_NONREF 0x00000080 |
16 | #define F_ALLOW_NONREF 0x00000080 |
|
|
17 | #define F_SHRINK 0x00000100 |
17 | |
18 | |
18 | #define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER |
19 | #define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER |
19 | #define F_DEFAULT 0 |
20 | #define F_DEFAULT 0 |
20 | |
21 | |
21 | #define INIT_SIZE 32 // initial scalar size to be allocated |
22 | #define INIT_SIZE 32 // initial scalar size to be allocated |
… | |
… | |
85 | |
86 | |
86 | while (str < end) |
87 | while (str < end) |
87 | { |
88 | { |
88 | unsigned char ch = *(unsigned char *)str; |
89 | unsigned char ch = *(unsigned char *)str; |
89 | |
90 | |
90 | if (ch == '"') |
91 | if (ch >= 0x20 && ch < 0x80) // most common case |
91 | { |
92 | { |
|
|
93 | if (ch == '"') // but with slow exceptions |
|
|
94 | { |
92 | need (enc, len += 1); |
95 | need (enc, len += 1); |
93 | *enc->cur++ = '\\'; |
96 | *enc->cur++ = '\\'; |
94 | *enc->cur++ = '"'; |
97 | *enc->cur++ = '"'; |
95 | ++str; |
|
|
96 | } |
98 | } |
97 | else if (ch == '\\') |
99 | else if (ch == '\\') |
98 | { |
100 | { |
99 | need (enc, len += 1); |
101 | need (enc, len += 1); |
100 | *enc->cur++ = '\\'; |
102 | *enc->cur++ = '\\'; |
101 | *enc->cur++ = '\\'; |
103 | *enc->cur++ = '\\'; |
102 | ++str; |
|
|
103 | } |
104 | } |
104 | else if (ch >= 0x20 && ch < 0x80) // most common case |
105 | else |
105 | { |
|
|
106 | *enc->cur++ = ch; |
106 | *enc->cur++ = ch; |
107 | ++str; |
107 | |
108 | } |
|
|
109 | else if (ch == '\015') |
|
|
110 | { |
|
|
111 | need (enc, len += 1); |
|
|
112 | *enc->cur++ = '\\'; |
|
|
113 | *enc->cur++ = 'r'; |
|
|
114 | ++str; |
|
|
115 | } |
|
|
116 | else if (ch == '\012') |
|
|
117 | { |
|
|
118 | need (enc, len += 1); |
|
|
119 | *enc->cur++ = '\\'; |
|
|
120 | *enc->cur++ = 'n'; |
|
|
121 | ++str; |
108 | ++str; |
122 | } |
109 | } |
123 | else |
110 | else |
124 | { |
111 | { |
125 | STRLEN clen; |
112 | switch (ch) |
126 | UV uch; |
|
|
127 | |
|
|
128 | if (is_utf8) |
|
|
129 | { |
113 | { |
130 | uch = utf8n_to_uvuni (str, end - str, &clen, UTF8_CHECK_ONLY); |
114 | case '\010': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'b'; ++str; break; |
131 | if (clen == (STRLEN)-1) |
115 | case '\011': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 't'; ++str; break; |
132 | croak ("malformed UTF-8 character in string, cannot convert to JSON"); |
116 | case '\012': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'n'; ++str; break; |
133 | } |
117 | case '\014': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'f'; ++str; break; |
134 | else |
118 | case '\015': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'r'; ++str; break; |
135 | { |
|
|
136 | uch = ch; |
|
|
137 | clen = 1; |
|
|
138 | } |
|
|
139 | |
119 | |
140 | if (uch < 0x80 || enc->flags & F_ASCII) |
120 | default: |
141 | { |
|
|
142 | if (uch > 0xFFFFUL) |
|
|
143 | { |
121 | { |
|
|
122 | STRLEN clen; |
|
|
123 | UV uch; |
|
|
124 | |
|
|
125 | if (is_utf8) |
|
|
126 | { |
|
|
127 | uch = utf8n_to_uvuni (str, end - str, &clen, UTF8_CHECK_ONLY); |
|
|
128 | if (clen == (STRLEN)-1) |
|
|
129 | croak ("malformed UTF-8 character in string, cannot convert to JSON"); |
|
|
130 | } |
|
|
131 | else |
|
|
132 | { |
|
|
133 | uch = ch; |
|
|
134 | clen = 1; |
|
|
135 | } |
|
|
136 | |
|
|
137 | if (uch < 0x80 || enc->flags & F_ASCII) |
|
|
138 | { |
|
|
139 | if (uch > 0xFFFFUL) |
|
|
140 | { |
144 | need (enc, len += 11); |
141 | need (enc, len += 11); |
145 | sprintf (enc->cur, "\\u%04x\\u%04x", |
142 | sprintf (enc->cur, "\\u%04x\\u%04x", |
146 | (uch - 0x10000) / 0x400 + 0xD800, |
143 | (uch - 0x10000) / 0x400 + 0xD800, |
147 | (uch - 0x10000) % 0x400 + 0xDC00); |
144 | (uch - 0x10000) % 0x400 + 0xDC00); |
148 | enc->cur += 12; |
145 | enc->cur += 12; |
|
|
146 | } |
|
|
147 | else |
|
|
148 | { |
|
|
149 | static char hexdigit [16] = "0123456789abcdef"; |
|
|
150 | need (enc, len += 5); |
|
|
151 | *enc->cur++ = '\\'; |
|
|
152 | *enc->cur++ = 'u'; |
|
|
153 | *enc->cur++ = hexdigit [ uch >> 12 ]; |
|
|
154 | *enc->cur++ = hexdigit [(uch >> 8) & 15]; |
|
|
155 | *enc->cur++ = hexdigit [(uch >> 4) & 15]; |
|
|
156 | *enc->cur++ = hexdigit [(uch >> 0) & 15]; |
|
|
157 | } |
|
|
158 | |
|
|
159 | str += clen; |
|
|
160 | } |
|
|
161 | else if (is_utf8) |
|
|
162 | { |
|
|
163 | need (enc, len += clen); |
|
|
164 | do |
|
|
165 | { |
|
|
166 | *enc->cur++ = *str++; |
|
|
167 | } |
|
|
168 | while (--clen); |
|
|
169 | } |
|
|
170 | else |
|
|
171 | { |
|
|
172 | need (enc, 10); // never more than 11 bytes needed |
|
|
173 | enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0); |
|
|
174 | ++str; |
|
|
175 | } |
149 | } |
176 | } |
150 | else |
|
|
151 | { |
|
|
152 | static char hexdigit [16] = "0123456789abcdef"; |
|
|
153 | need (enc, len += 5); |
|
|
154 | *enc->cur++ = '\\'; |
|
|
155 | *enc->cur++ = 'u'; |
|
|
156 | *enc->cur++ = hexdigit [ uch >> 12 ]; |
|
|
157 | *enc->cur++ = hexdigit [(uch >> 8) & 15]; |
|
|
158 | *enc->cur++ = hexdigit [(uch >> 4) & 15]; |
|
|
159 | *enc->cur++ = hexdigit [(uch >> 0) & 15]; |
|
|
160 | } |
|
|
161 | |
|
|
162 | str += clen; |
|
|
163 | } |
|
|
164 | else if (is_utf8) |
|
|
165 | { |
|
|
166 | need (enc, len += clen); |
|
|
167 | do |
|
|
168 | { |
|
|
169 | *enc->cur++ = *str++; |
|
|
170 | } |
|
|
171 | while (--clen); |
|
|
172 | } |
|
|
173 | else |
|
|
174 | { |
|
|
175 | need (enc, 10); // never more than 11 bytes needed |
|
|
176 | enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0); |
|
|
177 | ++str; |
|
|
178 | } |
177 | } |
179 | } |
178 | } |
180 | |
179 | |
181 | --len; |
180 | --len; |
182 | } |
181 | } |
… | |
… | |
425 | |
424 | |
426 | if (!(flags & (F_ASCII | F_UTF8))) |
425 | if (!(flags & (F_ASCII | F_UTF8))) |
427 | SvUTF8_on (enc.sv); |
426 | SvUTF8_on (enc.sv); |
428 | |
427 | |
429 | SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); |
428 | SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); |
|
|
429 | |
|
|
430 | #ifdef SvPV_shrink_to_cur |
|
|
431 | if (enc.flags & F_SHRINK) |
|
|
432 | SvPV_shrink_to_cur (enc.sv); |
|
|
433 | #endif |
430 | return enc.sv; |
434 | return enc.sv; |
431 | } |
435 | } |
432 | |
436 | |
433 | ///////////////////////////////////////////////////////////////////////////// |
437 | ///////////////////////////////////////////////////////////////////////////// |
434 | |
438 | |
… | |
… | |
599 | *SvEND (sv) = 0; |
603 | *SvEND (sv) = 0; |
600 | |
604 | |
601 | if (utf8) |
605 | if (utf8) |
602 | SvUTF8_on (sv); |
606 | SvUTF8_on (sv); |
603 | |
607 | |
|
|
608 | #ifdef SvPV_shrink_to_cur |
|
|
609 | if (dec->flags & F_SHRINK) |
|
|
610 | SvPV_shrink_to_cur (sv); |
|
|
611 | #endif |
|
|
612 | |
604 | return sv; |
613 | return sv; |
605 | |
614 | |
606 | fail: |
615 | fail: |
607 | SvREFCNT_dec (sv); |
616 | SvREFCNT_dec (sv); |
608 | return 0; |
617 | return 0; |
… | |
… | |
910 | CODE: |
919 | CODE: |
911 | RETVAL = sv_bless (newRV_noinc (newSVuv (F_DEFAULT)), json_stash); |
920 | RETVAL = sv_bless (newRV_noinc (newSVuv (F_DEFAULT)), json_stash); |
912 | OUTPUT: |
921 | OUTPUT: |
913 | RETVAL |
922 | RETVAL |
914 | |
923 | |
915 | SV *ascii (SV *self, int enable) |
924 | SV *ascii (SV *self, int enable = 1) |
916 | ALIAS: |
925 | ALIAS: |
917 | ascii = F_ASCII |
926 | ascii = F_ASCII |
918 | utf8 = F_UTF8 |
927 | utf8 = F_UTF8 |
919 | indent = F_INDENT |
928 | indent = F_INDENT |
920 | canonical = F_CANONICAL |
929 | canonical = F_CANONICAL |
921 | space_before = F_SPACE_BEFORE |
930 | space_before = F_SPACE_BEFORE |
922 | space_after = F_SPACE_AFTER |
931 | space_after = F_SPACE_AFTER |
923 | json_rpc = F_JSON_RPC |
932 | json_rpc = F_JSON_RPC |
924 | pretty = F_PRETTY |
933 | pretty = F_PRETTY |
925 | allow_nonref = F_ALLOW_NONREF |
934 | allow_nonref = F_ALLOW_NONREF |
|
|
935 | shrink = F_SHRINK |
926 | CODE: |
936 | CODE: |
927 | { |
937 | { |
928 | UV *uv = SvJSON (self); |
938 | UV *uv = SvJSON (self); |
929 | if (enable) |
939 | if (enable) |
930 | *uv |= ix; |
940 | *uv |= ix; |