1 |
pcg |
1.1 |
#include "../config.h" |
2 |
|
|
|
3 |
|
|
#include "encoding.h" |
4 |
|
|
|
5 |
|
|
#include <cstdlib> |
6 |
|
|
#include <cstring> |
7 |
|
|
|
8 |
|
|
const struct n2cs { |
9 |
|
|
const char *name; |
10 |
|
|
codeset cs; |
11 |
|
|
} n2cs[] = { |
12 |
|
|
/* first one found is the normalized one */ |
13 |
pcg |
1.3 |
{ "ISO88591", CS_ISO8859_1 }, |
14 |
|
|
{ "ISO8859PRIMARY", CS_ISO8859_1 }, // some stupid fonts use this (hi tigert) |
15 |
|
|
{ "ISO88592", CS_ISO8859_2 }, |
16 |
|
|
{ "ISO88593", CS_ISO8859_3 }, |
17 |
|
|
{ "ISO88594", CS_ISO8859_4 }, |
18 |
|
|
{ "ISO88595", CS_ISO8859_5 }, |
19 |
|
|
{ "ISO88596", CS_ISO8859_6 }, |
20 |
|
|
{ "ISO88597", CS_ISO8859_7 }, |
21 |
|
|
{ "ISO88598", CS_ISO8859_8 }, |
22 |
|
|
{ "ISO88599", CS_ISO8859_9 }, |
23 |
|
|
{ "ISO885910", CS_ISO8859_10 }, |
24 |
|
|
{ "ISO885911", CS_ISO8859_11 }, |
25 |
|
|
{ "ISO885913", CS_ISO8859_13 }, |
26 |
|
|
{ "ISO885914", CS_ISO8859_14 }, |
27 |
|
|
{ "ISO885915", CS_ISO8859_15 }, |
28 |
|
|
{ "FCD885915", CS_ISO8859_15 }, |
29 |
|
|
{ "ISO885916", CS_ISO8859_16 }, |
30 |
|
|
|
31 |
|
|
{ "ISO10646*", CS_UNICODE }, |
32 |
|
|
{ "UNICODE", CS_UNICODE }, |
33 |
|
|
{ "UTF8", CS_UNICODE }, |
34 |
|
|
|
35 |
|
|
{ "ASCII", CS_US_ASCII }, |
36 |
|
|
{ "USASCII", CS_US_ASCII }, |
37 |
|
|
{ "ANSIX341968", CS_US_ASCII }, |
38 |
|
|
|
39 |
|
|
{ "KOI8R", CS_KOI8_R }, |
40 |
|
|
{ "GOST1976874*", CS_KOI8_R }, |
41 |
|
|
{ "KOI8RU", CS_KOI8_U }, |
42 |
|
|
{ "KOI8U", CS_KOI8_U }, |
43 |
|
|
|
44 |
|
|
{ "VISCII*", CS_VISCII }, |
45 |
|
|
{ "TIS62025291", CS_VISCII }, // close enough |
46 |
|
|
|
47 |
|
|
{ "JISX0201*", CS_JIS0201_1976_0 }, |
48 |
|
|
{ "JISX0208*", CS_JIS0208_1983_0 }, // also wrongly matches -1990-0 (check Encode::JP) |
49 |
|
|
{ "JISX0212*", CS_JIS0212_1990_0 }, |
50 |
|
|
{ "JISX0221*", CS_UNICODE }, |
51 |
|
|
|
52 |
|
|
{ "KSC5601*", CS_KSC5601_1987_0 }, |
53 |
|
|
{ "KSX1001*", CS_KSC5601_1987_0 }, |
54 |
|
|
{ "KSC5700*", CS_UNICODE }, // unicode plus extensions |
55 |
|
|
|
56 |
|
|
{ "BIG5P*", CS_BIG5_PLUS }, |
57 |
|
|
{ "BIG5ETEN*", CS_BIG5_EXT }, |
58 |
|
|
{ "BIG5*", CS_BIG5 }, |
59 |
|
|
{ "GB2312*", CS_GB2312_1980_0 }, |
60 |
|
|
{ "GB6345*", CS_GB2312_1980_0 }, // slightly different to gb2312?? |
61 |
|
|
{ "GB8565*", CS_GB2312_1980_0 }, // a superset of gb2312?? |
62 |
|
|
{ "GB13000*", CS_UNICODE }, |
63 |
|
|
{ "CNS1164319921", CS_CNS11643_1992_1 }, |
64 |
|
|
{ "CNS1164319922", CS_CNS11643_1992_2 }, |
65 |
|
|
{ "CNS1164319923", CS_CNS11643_1992_3 }, |
66 |
|
|
{ "CNS1164319924", CS_CNS11643_1992_4 }, |
67 |
|
|
{ "CNS1164319925", CS_CNS11643_1992_5 }, |
68 |
|
|
{ "CNS1164319926", CS_CNS11643_1992_6 }, |
69 |
|
|
{ "CNS1164319927", CS_CNS11643_1992_7 }, |
70 |
|
|
{ "CNS116431992F", CS_CNS11643_1992_F }, |
71 |
pcg |
1.1 |
|
72 |
|
|
{ 0, CS_UNKNOWN } |
73 |
|
|
}; |
74 |
|
|
|
75 |
|
|
static const char * |
76 |
|
|
normalize_name (const char *name) |
77 |
|
|
{ |
78 |
|
|
static char res[16]; |
79 |
|
|
char *r; |
80 |
|
|
|
81 |
|
|
for (r = res; *name && r < res + 15; name++) |
82 |
|
|
if ((*name >= '0' && *name <= '9') |
83 |
|
|
|| (*name >= 'A' && *name <= 'Z')) |
84 |
|
|
*r++ = *name; |
85 |
|
|
else if (*name >= 'a' && *name <= 'z') |
86 |
|
|
*r++ = *name - ('a' - 'A'); |
87 |
|
|
|
88 |
|
|
*r = 0; |
89 |
|
|
|
90 |
|
|
return res; |
91 |
|
|
} |
92 |
|
|
|
93 |
|
|
codeset |
94 |
|
|
codeset_from_name (const char *name) |
95 |
|
|
{ |
96 |
|
|
if (!name) |
97 |
|
|
return CS_UNKNOWN; |
98 |
|
|
|
99 |
|
|
name = normalize_name (name); |
100 |
|
|
|
101 |
|
|
const struct n2cs *i = n2cs; |
102 |
|
|
|
103 |
|
|
do { |
104 |
pcg |
1.3 |
int len = strlen (i->name); |
105 |
|
|
|
106 |
|
|
if ((i->name[len - 1] == '*' |
107 |
|
|
&& !strncmp (name, i->name, len - 1)) |
108 |
|
|
|| !strcmp (name, i->name)) |
109 |
|
|
return i->cs; |
110 |
|
|
|
111 |
pcg |
1.1 |
} while ((++i)->name); |
112 |
|
|
|
113 |
|
|
return CS_UNKNOWN; |
114 |
|
|
} |
115 |
|
|
|
116 |
|
|
struct rxvt_codeset_conv_unknown : rxvt_codeset_conv { |
117 |
|
|
uint32_t to_unicode (uint32_t enc) const { return NOCHAR; } |
118 |
|
|
uint32_t from_unicode (uint32_t unicode) const { return NOCHAR; } |
119 |
|
|
} rxvt_codeset_conv_unknown; |
120 |
|
|
|
121 |
|
|
struct rxvt_codeset_conv_us_ascii : rxvt_codeset_conv { |
122 |
|
|
uint32_t from_unicode (uint32_t unicode) const { return unicode <= 127 ? unicode : NOCHAR; } |
123 |
|
|
} rxvt_codeset_conv_us_ascii; |
124 |
|
|
|
125 |
|
|
struct rxvt_codeset_conv_unicode : rxvt_codeset_conv { |
126 |
|
|
/* transparent */ |
127 |
|
|
} rxvt_codeset_conv_unicode; |
128 |
|
|
|
129 |
|
|
struct rxvt_codeset_conv_unicode_16 : rxvt_codeset_conv { |
130 |
|
|
uint32_t to_unicode (uint32_t enc) const { return enc; } |
131 |
|
|
uint32_t from_unicode (uint32_t unicode) const { return unicode <= 65535 ? unicode : NOCHAR; } |
132 |
|
|
} rxvt_codeset_conv_unicode_16; |
133 |
|
|
|
134 |
|
|
/* block character set, must conform to the dec special pseudofont in defaultfont.C */ |
135 |
|
|
struct rxvt_codeset_conv_special : rxvt_codeset_conv { |
136 |
|
|
uint32_t to_unicode (uint32_t enc) const { |
137 |
|
|
return enc; |
138 |
|
|
} |
139 |
|
|
|
140 |
|
|
uint32_t from_unicode (uint32_t unicode) const { |
141 |
|
|
return unicode; |
142 |
|
|
} |
143 |
|
|
} rxvt_codeset_conv_special; |
144 |
|
|
|
145 |
|
|
#define ENCODING_DEFAULT |
146 |
|
|
|
147 |
|
|
#include "table/iso8859_1.h" |
148 |
|
|
#include "table/iso8859_15.h" |
149 |
|
|
|
150 |
|
|
//#define ENCODING_EU |
151 |
|
|
|
152 |
|
|
#include "table/iso8859_2.h" |
153 |
|
|
#include "table/iso8859_3.h" |
154 |
|
|
#include "table/iso8859_4.h" |
155 |
|
|
#include "table/iso8859_5.h" |
156 |
|
|
#include "table/iso8859_6.h" |
157 |
|
|
#include "table/iso8859_7.h" |
158 |
|
|
#include "table/iso8859_8.h" |
159 |
|
|
#include "table/iso8859_9.h" |
160 |
|
|
#include "table/iso8859_10.h" |
161 |
|
|
#include "table/iso8859_11.h" |
162 |
|
|
#include "table/iso8859_13.h" |
163 |
|
|
#include "table/iso8859_14.h" |
164 |
|
|
#include "table/iso8859_16.h" |
165 |
|
|
|
166 |
|
|
#include "table/koi8_r.h" |
167 |
|
|
#include "table/koi8_u.h" |
168 |
|
|
|
169 |
|
|
//#define ENCODING_KR |
170 |
|
|
|
171 |
|
|
#include "table/ksc5601_1987_0.h" |
172 |
|
|
|
173 |
|
|
//#define ENCODING_CN |
174 |
|
|
|
175 |
|
|
#include "table/gb2312_1980_0.h" |
176 |
pcg |
1.3 |
#include "table/big5.h" |
177 |
pcg |
1.1 |
|
178 |
|
|
//#define ENCODING_CN_EXT |
179 |
|
|
|
180 |
|
|
#include "table/cns11643_1992_1.h" |
181 |
|
|
#include "table/cns11643_1992_2.h" |
182 |
|
|
#include "table/cns11643_1992_3.h" |
183 |
|
|
#include "table/cns11643_1992_4.h" |
184 |
|
|
#include "table/cns11643_1992_5.h" |
185 |
|
|
#include "table/cns11643_1992_6.h" |
186 |
|
|
#include "table/cns11643_1992_7.h" |
187 |
|
|
#include "table/cns11643_1992_f.h" |
188 |
|
|
#include "table/big5_ext.h" |
189 |
|
|
#include "table/big5_plus.h" |
190 |
|
|
|
191 |
|
|
//#define ENCODING_VN |
192 |
|
|
|
193 |
|
|
#include "table/viscii.h" |
194 |
|
|
|
195 |
|
|
//#define ENCODING_JP |
196 |
|
|
|
197 |
|
|
#include "table/jis0201_1976_0.h" |
198 |
|
|
#include "table/jis0208_1983_0.h" |
199 |
|
|
#include "table/jis0212_1990_0.h" |
200 |
|
|
|
201 |
|
|
//#define ENCODING_JP_EXT |
202 |
|
|
|
203 |
|
|
#include "table/jis0213_1.h" |
204 |
|
|
#include "table/jis0213_2.h" |
205 |
|
|
|
206 |
pcg |
1.3 |
// order must match table in encoding.h(!) |
207 |
pcg |
1.1 |
const rxvt_codeset_conv *rxvt_codeset[NUM_CODESETS] = { |
208 |
|
|
&rxvt_codeset_conv_unknown, |
209 |
|
|
&rxvt_codeset_conv_special, |
210 |
|
|
|
211 |
|
|
&rxvt_codeset_conv_us_ascii, |
212 |
|
|
|
213 |
|
|
&rxvt_codeset_conv_iso8859_1, |
214 |
|
|
&rxvt_codeset_conv_iso8859_2, |
215 |
|
|
&rxvt_codeset_conv_iso8859_3, |
216 |
|
|
&rxvt_codeset_conv_iso8859_4, |
217 |
|
|
&rxvt_codeset_conv_iso8859_5, |
218 |
|
|
&rxvt_codeset_conv_iso8859_6, |
219 |
|
|
&rxvt_codeset_conv_iso8859_7, |
220 |
|
|
&rxvt_codeset_conv_iso8859_8, |
221 |
|
|
&rxvt_codeset_conv_iso8859_9, |
222 |
|
|
&rxvt_codeset_conv_iso8859_10, |
223 |
|
|
&rxvt_codeset_conv_iso8859_11, |
224 |
|
|
&rxvt_codeset_conv_iso8859_13, |
225 |
|
|
&rxvt_codeset_conv_iso8859_14, |
226 |
|
|
&rxvt_codeset_conv_iso8859_15, |
227 |
|
|
&rxvt_codeset_conv_iso8859_16, |
228 |
|
|
|
229 |
|
|
&rxvt_codeset_conv_koi8_r, |
230 |
|
|
&rxvt_codeset_conv_koi8_u, |
231 |
|
|
|
232 |
|
|
&rxvt_codeset_conv_jis0201_1976_0, |
233 |
|
|
&rxvt_codeset_conv_jis0208_1983_0, |
234 |
|
|
&rxvt_codeset_conv_jis0212_1990_0, |
235 |
|
|
|
236 |
|
|
&rxvt_codeset_conv_jis0213_1, |
237 |
|
|
&rxvt_codeset_conv_jis0213_2, |
238 |
|
|
|
239 |
|
|
&rxvt_codeset_conv_ksc5601_1987_0, |
240 |
|
|
|
241 |
|
|
&rxvt_codeset_conv_gb2312_1980_0, |
242 |
|
|
|
243 |
|
|
&rxvt_codeset_conv_cns11643_1992_1, |
244 |
|
|
&rxvt_codeset_conv_cns11643_1992_2, |
245 |
|
|
&rxvt_codeset_conv_cns11643_1992_3, |
246 |
|
|
&rxvt_codeset_conv_cns11643_1992_4, |
247 |
|
|
&rxvt_codeset_conv_cns11643_1992_5, |
248 |
|
|
&rxvt_codeset_conv_cns11643_1992_6, |
249 |
|
|
&rxvt_codeset_conv_cns11643_1992_7, |
250 |
|
|
&rxvt_codeset_conv_cns11643_1992_f, |
251 |
pcg |
1.3 |
&rxvt_codeset_conv_big5, |
252 |
pcg |
1.1 |
&rxvt_codeset_conv_big5_ext, |
253 |
|
|
&rxvt_codeset_conv_big5_plus, |
254 |
|
|
|
255 |
|
|
&rxvt_codeset_conv_viscii, |
256 |
|
|
|
257 |
|
|
&rxvt_codeset_conv_unicode_16, |
258 |
|
|
&rxvt_codeset_conv_unicode |
259 |
|
|
}; |
260 |
|
|
|