… | |
… | |
66 | { "KOI8U", CS_KOI8_U }, |
66 | { "KOI8U", CS_KOI8_U }, |
67 | |
67 | |
68 | { "VISCII*", CS_VISCII }, |
68 | { "VISCII*", CS_VISCII }, |
69 | |
69 | |
70 | { "JISX0201*", CS_JIS0201_1976_0 }, |
70 | { "JISX0201*", CS_JIS0201_1976_0 }, |
|
|
71 | { "JISC6226*", CS_JIS0208_1990_0 }, // also wrongly matches -1987-0? (check Encode::JP) |
71 | { "JISX0208*", CS_JIS0208_1990_0 }, // also wrongly matches -1987-0? (check Encode::JP) |
72 | { "JISX0208*", CS_JIS0208_1990_0 }, // also wrongly matches -1987-0? (check Encode::JP) |
72 | { "JISX0212*", CS_JIS0212_1990_0 }, |
73 | { "JISX0212*", CS_JIS0212_1990_0 }, |
73 | { "JISX021320001", CS_JIS0213_1 }, |
74 | { "JISX021320001", CS_JIS0213_1 }, |
74 | { "JISX021320002", CS_JIS0213_2 }, |
75 | { "JISX021320002", CS_JIS0213_2 }, |
75 | { "JISX0221*", CS_UNICODE }, // _very_ close |
76 | { "JISX0221*", CS_UNICODE }, // _very_ close |
… | |
… | |
80 | |
81 | |
81 | { "BIG5P*", CS_BIG5_PLUS }, |
82 | { "BIG5P*", CS_BIG5_PLUS }, |
82 | { "BIG5ETEN*", CS_BIG5_EXT }, |
83 | { "BIG5ETEN*", CS_BIG5_EXT }, |
83 | { "BIG5*", CS_BIG5 }, |
84 | { "BIG5*", CS_BIG5 }, |
84 | { "GB2312*", CS_GB2312_1980_0 }, |
85 | { "GB2312*", CS_GB2312_1980_0 }, |
|
|
86 | { "GBK*", CS_GBK_0 }, |
85 | { "GB6345*", CS_GB2312_1980_0 }, // slightly different to gb2312?? |
87 | { "GB6345*", CS_GB2312_1980_0 }, // slightly different to gb2312?? |
86 | { "GB8565*", CS_GB2312_1980_0 }, // a superset of gb2312?? |
88 | { "GB8565*", CS_GB2312_1980_0 }, // a superset of gb2312?? |
87 | { "GB13000*", CS_UNICODE }, |
89 | { "GB13000*", CS_UNICODE }, |
88 | { "CNS1164319921", CS_CNS11643_1992_1 }, |
90 | { "CNS1164319921", CS_CNS11643_1992_1 }, |
89 | { "CNS1164319922", CS_CNS11643_1992_2 }, |
91 | { "CNS1164319922", CS_CNS11643_1992_2 }, |
… | |
… | |
136 | } while ((++i)->name); |
138 | } while ((++i)->name); |
137 | |
139 | |
138 | return CS_UNKNOWN; |
140 | return CS_UNKNOWN; |
139 | } |
141 | } |
140 | |
142 | |
141 | struct rxvt_codeset_conv_unknown : rxvt_codeset_conv { |
|
|
142 | unicode_t to_unicode (uint32_t enc) const { return NOCHAR; } |
143 | static unicode_t cs_unknown_to_unicode (uint32_t enc) { return NOCHAR; } |
143 | uint32_t from_unicode (unicode_t unicode) const { return NOCHAR; } |
144 | static uint32_t cs_unknown_from_unicode (unicode_t unicode) { return NOCHAR; } |
144 | } rxvt_codeset_conv_unknown; |
|
|
145 | |
145 | |
146 | struct rxvt_codeset_conv_us_ascii : rxvt_codeset_conv { |
146 | static unicode_t cs_unicode_to_unicode (uint32_t enc) { return enc; } |
|
|
147 | static uint32_t cs_unicode_from_unicode (unicode_t unicode) { return unicode; } |
|
|
148 | |
|
|
149 | #define cs_us_ascii_to_unicode cs_unicode_to_unicode |
147 | uint32_t from_unicode (unicode_t unicode) const { return unicode <= 127 ? unicode : NOCHAR; } |
150 | static uint32_t cs_us_ascii_from_unicode (unicode_t unicode) { return unicode <= 127 ? unicode : NOCHAR; } |
148 | } rxvt_codeset_conv_us_ascii; |
|
|
149 | |
151 | |
150 | struct rxvt_codeset_conv_unicode : rxvt_codeset_conv { |
152 | #define cs_us_ascii_to_unicode_16 cs_unicode_to_unicode |
151 | /* transparent */ |
|
|
152 | } rxvt_codeset_conv_unicode; |
|
|
153 | |
|
|
154 | struct rxvt_codeset_conv_unicode_16 : rxvt_codeset_conv { |
|
|
155 | unicode_t to_unicode (uint32_t enc) const { return enc; } |
|
|
156 | uint32_t from_unicode (unicode_t unicode) const { return unicode <= 65535 ? unicode : NOCHAR; } |
153 | static uint32_t cs_unicode_16_from_unicode (unicode_t unicode) { return unicode <= 65535 ? unicode : NOCHAR; } |
157 | } rxvt_codeset_conv_unicode_16; |
|
|
158 | |
154 | |
159 | #define ENCODING_DEFAULT |
155 | #define ENCODING_DEFAULT |
160 | |
156 | |
161 | #include "table/iso8859_1.h" |
157 | #include "table/iso8859_1.h" |
162 | #include "table/iso8859_15.h" |
158 | #include "table/iso8859_15.h" |
… | |
… | |
182 | |
178 | |
183 | //#define ENCODING_KR |
179 | //#define ENCODING_KR |
184 | |
180 | |
185 | #include "table/ksc5601_1987_0.h" |
181 | #include "table/ksc5601_1987_0.h" |
186 | |
182 | |
187 | //#define ENCODING_CN |
183 | //#define ENCODING_ZH |
188 | |
184 | |
|
|
185 | #include "table/big5.h" |
|
|
186 | #include "table/gbk_0.h" |
189 | #include "table/gb2312_1980_0.h" |
187 | #include "table/gb2312_1980_0.h" |
190 | #include "table/big5.h" |
|
|
191 | |
188 | |
192 | //#define ENCODING_CN_EXT |
189 | //#define ENCODING_ZH_EXT |
193 | |
190 | |
194 | #include "table/cns11643_1992_1.h" |
191 | #include "table/cns11643_1992_1.h" |
195 | #include "table/cns11643_1992_2.h" |
192 | #include "table/cns11643_1992_2.h" |
196 | #include "table/cns11643_1992_3.h" |
193 | #include "table/cns11643_1992_3.h" |
197 | #include "table/cns11643_1992_4.h" |
194 | #include "table/cns11643_1992_4.h" |
… | |
… | |
215 | //#define ENCODING_JP_EXT |
212 | //#define ENCODING_JP_EXT |
216 | |
213 | |
217 | #include "table/jis0213_1.h" |
214 | #include "table/jis0213_1.h" |
218 | #include "table/jis0213_2.h" |
215 | #include "table/jis0213_2.h" |
219 | |
216 | |
|
|
217 | #if ENCODING_TO_UNICODE |
|
|
218 | # define ENC(base) { cs_ ## base ## _from_unicode, cs_ ## base ## _to_unicode } |
|
|
219 | #else |
|
|
220 | # define ENC(base) { cs_ ## base ## _from_unicode } |
|
|
221 | #endif |
|
|
222 | |
|
|
223 | |
220 | // order must match table in encoding.h(!) |
224 | // order must match table in encoding.h(!) |
221 | const rxvt_codeset_conv *rxvt_codeset[NUM_CODESETS] = { |
225 | const rxvt_codeset_conv rxvt_codeset[NUM_CODESETS] = { |
222 | &rxvt_codeset_conv_unknown, |
226 | ENC (unknown), |
223 | |
227 | |
224 | &rxvt_codeset_conv_us_ascii, |
228 | ENC (us_ascii), |
225 | |
229 | |
226 | &rxvt_codeset_conv_iso8859_1, |
230 | ENC (iso8859_1), |
227 | &rxvt_codeset_conv_iso8859_2, |
231 | ENC (iso8859_2), |
228 | &rxvt_codeset_conv_iso8859_3, |
232 | ENC (iso8859_3), |
229 | &rxvt_codeset_conv_iso8859_4, |
233 | ENC (iso8859_4), |
230 | &rxvt_codeset_conv_iso8859_5, |
234 | ENC (iso8859_5), |
231 | &rxvt_codeset_conv_iso8859_6, |
235 | ENC (iso8859_6), |
232 | &rxvt_codeset_conv_iso8859_7, |
236 | ENC (iso8859_7), |
233 | &rxvt_codeset_conv_iso8859_8, |
237 | ENC (iso8859_8), |
234 | &rxvt_codeset_conv_iso8859_9, |
238 | ENC (iso8859_9), |
235 | &rxvt_codeset_conv_iso8859_10, |
239 | ENC (iso8859_10), |
236 | &rxvt_codeset_conv_iso8859_11, |
240 | ENC (iso8859_11), |
237 | &rxvt_codeset_conv_iso8859_13, |
241 | ENC (iso8859_13), |
238 | &rxvt_codeset_conv_iso8859_14, |
242 | ENC (iso8859_14), |
239 | &rxvt_codeset_conv_iso8859_15, |
243 | ENC (iso8859_15), |
240 | &rxvt_codeset_conv_iso8859_16, |
244 | ENC (iso8859_16), |
241 | |
245 | |
242 | &rxvt_codeset_conv_koi8_r, |
246 | ENC (koi8_r), |
243 | &rxvt_codeset_conv_koi8_u, |
247 | ENC (koi8_u), |
244 | |
248 | |
245 | &rxvt_codeset_conv_jis0201_1976_0, |
249 | ENC (jis0201_1976_0), |
246 | &rxvt_codeset_conv_jis0208_1990_0, |
250 | ENC (jis0208_1990_0), |
247 | &rxvt_codeset_conv_jis0212_1990_0, |
251 | ENC (jis0212_1990_0), |
248 | |
252 | |
249 | &rxvt_codeset_conv_jis0213_1, |
253 | ENC (jis0213_1), |
250 | &rxvt_codeset_conv_jis0213_2, |
254 | ENC (jis0213_2), |
251 | |
255 | |
252 | &rxvt_codeset_conv_ksc5601_1987_0, |
256 | ENC (ksc5601_1987_0), |
253 | |
257 | |
254 | &rxvt_codeset_conv_gb2312_1980_0, |
258 | ENC (gb2312_1980_0), |
|
|
259 | ENC (gbk_0), |
255 | |
260 | |
256 | &rxvt_codeset_conv_cns11643_1992_1, |
261 | ENC (cns11643_1992_1), |
257 | &rxvt_codeset_conv_cns11643_1992_2, |
262 | ENC (cns11643_1992_2), |
258 | &rxvt_codeset_conv_cns11643_1992_3, |
263 | ENC (cns11643_1992_3), |
259 | &rxvt_codeset_conv_cns11643_1992_4, |
264 | ENC (cns11643_1992_4), |
260 | &rxvt_codeset_conv_cns11643_1992_5, |
265 | ENC (cns11643_1992_5), |
261 | &rxvt_codeset_conv_cns11643_1992_6, |
266 | ENC (cns11643_1992_6), |
262 | &rxvt_codeset_conv_cns11643_1992_7, |
267 | ENC (cns11643_1992_7), |
263 | &rxvt_codeset_conv_cns11643_1992_f, |
268 | ENC (cns11643_1992_f), |
264 | &rxvt_codeset_conv_big5, |
269 | ENC (big5), |
265 | &rxvt_codeset_conv_big5_ext, |
270 | ENC (big5_ext), |
266 | &rxvt_codeset_conv_big5_plus, |
271 | ENC (big5_plus), |
267 | |
272 | |
268 | &rxvt_codeset_conv_viscii, |
273 | ENC (viscii), |
269 | |
274 | |
270 | &rxvt_codeset_conv_unicode_16, |
275 | ENC (unicode_16), |
271 | &rxvt_codeset_conv_unicode |
276 | ENC (unicode), |
272 | }; |
277 | }; |
273 | |
278 | |
274 | #if ENABLE_COMBINING |
279 | #if ENABLE_COMBINING |
275 | # define ENCODING_COMPOSE |
280 | # define ENCODING_COMPOSE |
276 | #endif |
281 | #endif |
… | |
… | |
302 | |
307 | |
303 | #include "table/category.h" |
308 | #include "table/category.h" |
304 | |
309 | |
305 | bool unicode::is_space (unicode_t c) |
310 | bool unicode::is_space (unicode_t c) |
306 | { |
311 | { |
307 | return IS_SPACE (c); |
312 | return IS_SPACE (c) |
|
|
313 | || c == 0x09; // exclude tabs, too, as we store them in the buffer |
308 | } |
314 | } |