Line | Count | Source (jump to first uncovered line) |
1 | | |
2 | | #include "wc.h" |
3 | | #include "sjis.h" |
4 | | #include "jis.h" |
5 | | #include "wtf.h" |
6 | | #include "ucs.h" |
7 | | |
8 | | #include "map/jisx02132_sjis.map" |
9 | | wc_uchar *wc_jisx0212_jisx02132_map = jisx02132_sjis_map; |
10 | | |
11 | | #define C0 WC_SJIS_MAP_C0 |
12 | | #define GL WC_SJIS_MAP_GL |
13 | 328k | #define LB WC_SJIS_MAP_LB |
14 | 42.0k | #define S80 WC_SJIS_MAP_80 |
15 | 50.5M | #define SK WC_SJIS_MAP_SK |
16 | 185k | #define SL WC_SJIS_MAP_SL |
17 | 51.4k | #define SH WC_SJIS_MAP_SH |
18 | 90.9k | #define SX WC_SJIS_MAP_SX |
19 | 182k | #define C1 WC_SJIS_MAP_C1 |
20 | 53.1k | #define SA0 WC_SJIS_MAP_A0 |
21 | | |
22 | | wc_uint8 WC_SJIS_MAP[ 0x100 ] = { |
23 | | C0, C0, C0, C0, C0, C0, C0, C0, C0, C0, C0, C0, C0, C0, C0, C0, |
24 | | C0, C0, C0, C0, C0, C0, C0, C0, C0, C0, C0, C0, C0, C0, C0, C0, |
25 | | GL, GL, GL, GL, GL, GL, GL, GL, GL, GL, GL, GL, GL, GL, GL, GL, |
26 | | GL, GL, GL, GL, GL, GL, GL, GL, GL, GL, GL, GL, GL, GL, GL, GL, |
27 | | LB, LB, LB, LB, LB, LB, LB, LB, LB, LB, LB, LB, LB, LB, LB, LB, |
28 | | LB, LB, LB, LB, LB, LB, LB, LB, LB, LB, LB, LB, LB, LB, LB, LB, |
29 | | LB, LB, LB, LB, LB, LB, LB, LB, LB, LB, LB, LB, LB, LB, LB, LB, |
30 | | LB, LB, LB, LB, LB, LB, LB, LB, LB, LB, LB, LB, LB, LB, LB, C0, |
31 | | |
32 | | S80,SL, SL, SL, SL, SL, SL, SL, SL, SL, SL, SL, SL, SL, SL, SL, |
33 | | SL, SL, SL, SL, SL, SL, SL, SL, SL, SL, SL, SL, SL, SL, SL, SL, |
34 | | SA0,SK, SK, SK, SK, SK, SK, SK, SK, SK, SK, SK, SK, SK, SK, SK, |
35 | | SK, SK, SK, SK, SK, SK, SK, SK, SK, SK, SK, SK, SK, SK, SK, SK, |
36 | | SK, SK, SK, SK, SK, SK, SK, SK, SK, SK, SK, SK, SK, SK, SK, SK, |
37 | | SK, SK, SK, SK, SK, SK, SK, SK, SK, SK, SK, SK, SK, SK, SK, SK, |
38 | | SH, SH, SH, SH, SH, SH, SH, SH, SH, SH, SH, SH, SH, SH, SH, SH, |
39 | | SX, SX, SX, SX, SX, SX, SX, SX, SX, SX, SX, SX, SX, C1, C1, C1, |
40 | | }; |
41 | | |
42 | 317k | #define sjis_to_jisx0208(ub, lb) \ |
43 | 317k | { \ |
44 | 317k | ub -= (ub < 0xa0) ? 0x81 : 0xc1; \ |
45 | 317k | ub = (ub << 1) + 0x21; \ |
46 | 317k | if (lb < 0x9f) { \ |
47 | 147k | lb -= (lb > 0x7e) ? 0x20 : 0x1f; \ |
48 | 170k | } else { \ |
49 | 170k | ub++; \ |
50 | 170k | lb -= 0x7e; \ |
51 | 170k | } \ |
52 | 317k | } |
53 | 8.83k | #define sjis_to_jisx02132(ub, lb) \ |
54 | 8.83k | { \ |
55 | 8.83k | if (lb < 0x9f) { \ |
56 | 2.61k | ub = sjis1_jisx02132_map[ub - 0xf0]; \ |
57 | 2.61k | lb -= (lb > 0x7e) ? 0x20 : 0x1f; \ |
58 | 6.22k | } else { \ |
59 | 6.22k | ub = sjis2_jisx02132_map[ub - 0xf0]; \ |
60 | 6.22k | lb -= 0x7e; \ |
61 | 6.22k | } \ |
62 | 8.83k | } |
63 | 7.72M | #define jisx0208_to_sjis(ub, lb) \ |
64 | 7.72M | { \ |
65 | 7.72M | lb += (ub & 1) ? 0x1f : 0x7d; \ |
66 | 7.72M | if (lb > 0x7e) \ |
67 | 7.72M | lb++; \ |
68 | 7.72M | ub = (ub - 0x21) >> 1; \ |
69 | 7.72M | ub += (ub < 0x1f) ? 0x81 : 0xc1; \ |
70 | 7.72M | } |
71 | 0 | #define jisx02132_to_sjis(ub, lb) \ |
72 | 0 | { \ |
73 | 0 | lb += (ub & 1) ? 0x1f : 0x7d; \ |
74 | 0 | if (lb > 0x7e) \ |
75 | 0 | lb++; \ |
76 | 0 | ub = jisx02132_sjis_map[ ub ]; \ |
77 | 0 | } |
78 | | |
79 | | wc_wchar_t |
80 | | wc_sjis_to_jis(wc_wchar_t cc) |
81 | 0 | { |
82 | 0 | wc_uchar ub, lb; |
83 | |
|
84 | 0 | ub = cc.code >> 8; |
85 | 0 | lb = cc.code & 0xff; |
86 | 0 | if (ub < 0xf0) { |
87 | 0 | sjis_to_jisx0208(ub, lb); |
88 | 0 | cc.ccs = WC_CCS_JIS_X_0208; |
89 | 0 | } else { |
90 | 0 | sjis_to_jisx02132(ub, lb); |
91 | 0 | cc.ccs = WC_CCS_JIS_X_0213_2; |
92 | 0 | } |
93 | 0 | cc.code = ((wc_uint32)ub << 8) | lb; |
94 | 0 | return cc; |
95 | 0 | } |
96 | | |
97 | | wc_wchar_t |
98 | | wc_jis_to_sjis(wc_wchar_t cc) |
99 | 0 | { |
100 | 0 | wc_uchar ub, lb; |
101 | |
|
102 | 0 | ub = (cc.code >> 8) & 0x7f; |
103 | 0 | lb = cc.code & 0x7f; |
104 | 0 | if (cc.ccs == WC_CCS_JIS_X_0213_2) { |
105 | 0 | jisx02132_to_sjis(ub, lb); |
106 | 0 | if (! ub) { |
107 | 0 | cc.ccs = WC_CCS_UNKNOWN_W; |
108 | 0 | return cc; |
109 | 0 | } |
110 | 0 | } else { |
111 | 0 | jisx0208_to_sjis(ub, lb); |
112 | 0 | } |
113 | 0 | cc.code = ((wc_uint32)ub << 8) | lb; |
114 | 0 | return cc; |
115 | 0 | } |
116 | | |
117 | | wc_wchar_t |
118 | | wc_sjis_ext_to_cs94w(wc_wchar_t cc) |
119 | 147k | { |
120 | 147k | wc_uchar ub, lb; |
121 | | |
122 | 147k | ub = cc.code >> 8; |
123 | 147k | lb = cc.code & 0xff; |
124 | 147k | sjis_to_jisx0208(ub, lb); |
125 | 147k | if (ub <= 0x7e) { |
126 | 43.2k | cc.ccs = WC_CCS_SJIS_EXT_1; |
127 | 104k | } else { |
128 | 104k | ub -= 0x5e; |
129 | 104k | cc.ccs = WC_CCS_SJIS_EXT_2; |
130 | 104k | } |
131 | 147k | cc.code = ((wc_uint32)ub << 8) | lb; |
132 | 147k | return cc; |
133 | 147k | } |
134 | | |
135 | | wc_wchar_t |
136 | | wc_cs94w_to_sjis_ext(wc_wchar_t cc) |
137 | 96.2k | { |
138 | 96.2k | wc_uchar ub, lb; |
139 | | |
140 | 96.2k | ub = (cc.code >> 8) & 0x7f; |
141 | 96.2k | lb = cc.code & 0x7f; |
142 | 96.2k | if (cc.ccs == WC_CCS_SJIS_EXT_2) |
143 | 75.0k | ub += 0x5e; |
144 | 96.2k | jisx0208_to_sjis(ub, lb); |
145 | 96.2k | cc.ccs = WC_CCS_SJIS_EXT; |
146 | 96.2k | cc.code = ((wc_uint32)ub << 8) | lb; |
147 | 96.2k | return cc; |
148 | 96.2k | } |
149 | | |
150 | | wc_uint32 |
151 | | wc_sjis_ext1_to_N(wc_uint32 c) |
152 | 22.0k | { |
153 | 22.0k | wc_uchar ub; |
154 | | |
155 | 22.0k | ub = (c >> 8) & 0x7f; |
156 | 22.0k | switch(ub) { |
157 | 2.40k | case 0x2D: /* 0x8740 - */ |
158 | 2.40k | ub = 0; |
159 | 2.40k | break; |
160 | 590 | case 0x79: /* 0xED40 - */ |
161 | 1.97k | case 0x7A: /* 0xED9F - */ |
162 | 2.74k | case 0x7B: /* 0xEE40 - */ |
163 | 3.50k | case 0x7C: /* 0xEE9F - */ |
164 | 3.50k | ub -= 0x78; |
165 | 3.50k | break; |
166 | 16.1k | default: |
167 | 16.1k | return WC_C_SJIS_ERROR; |
168 | 22.0k | } |
169 | 5.91k | return ub * 0x5e + (c & 0x7f) - 0x21; |
170 | 22.0k | } |
171 | | |
172 | | wc_uint32 |
173 | | wc_sjis_ext2_to_N(wc_uint32 c) |
174 | 29.4k | { |
175 | 29.4k | wc_uchar ub; |
176 | | |
177 | 29.4k | ub = (c >> 8) & 0x7f; |
178 | 29.4k | switch(ub) { |
179 | 439 | case 0x35: /* 0xFA40 - */ |
180 | 4.81k | case 0x36: /* 0xFA9F - */ |
181 | 5.34k | case 0x37: /* 0xFB40 - */ |
182 | 7.02k | case 0x38: /* 0xFB9F - */ |
183 | 7.51k | case 0x39: /* 0xFC40 - */ |
184 | 7.51k | ub -= 0x30; |
185 | 7.51k | break; |
186 | 21.9k | default: |
187 | 21.9k | return WC_C_SJIS_ERROR; |
188 | 29.4k | } |
189 | 7.51k | return ub * 0x5e + (c & 0x7f) - 0x21; |
190 | 29.4k | } |
191 | | |
192 | | Str |
193 | | wc_conv_from_sjis(Str is, wc_ces ces) |
194 | 1.58k | { |
195 | 1.58k | Str os; |
196 | 1.58k | wc_uchar *sp = (wc_uchar *)is->ptr; |
197 | 1.58k | wc_uchar *ep = sp + is->length; |
198 | 1.58k | wc_uchar *p; |
199 | 1.58k | wc_uchar jis[2]; |
200 | 1.58k | int state = WC_SJIS_NOSTATE; |
201 | 1.58k | wc_wchar_t cc; |
202 | | |
203 | 2.56k | for (p = sp; p < ep && *p < 0x80; p++) |
204 | 980 | ; |
205 | 1.58k | if (p == ep) |
206 | 123 | return is; |
207 | 1.46k | os = Strnew_size(is->length); |
208 | 1.46k | if (p > sp) |
209 | 261 | Strcat_charp_n(os, is->ptr, (int)(p - sp)); |
210 | | |
211 | 49.7M | for (; p < ep; p++) { |
212 | 49.7M | switch (state) { |
213 | 49.4M | case WC_SJIS_NOSTATE: |
214 | 49.4M | switch (WC_SJIS_MAP[*p]) { |
215 | 140k | case SL: |
216 | 140k | state = WC_SJIS_SHIFT_L; |
217 | 140k | break; |
218 | 41.5k | case SH: |
219 | 41.5k | state = WC_SJIS_SHIFT_H; |
220 | 41.5k | break; |
221 | 79.9k | case SX: |
222 | 79.9k | state = WC_SJIS_SHIFT_X; |
223 | 79.9k | break; |
224 | 46.5M | case SK: |
225 | 46.5M | wtf_push(os, WC_CCS_JIS_X_0201K, (wc_uint32)*p); |
226 | 46.5M | break; |
227 | 25.3k | case S80: |
228 | 29.7k | case SA0: |
229 | 109k | case C1: |
230 | 109k | wtf_push_unknown(os, p, 1); |
231 | 109k | break; |
232 | 2.52M | default: |
233 | 2.52M | Strcat_char(os, (char)*p); |
234 | 2.52M | break; |
235 | 49.4M | } |
236 | 49.4M | break; |
237 | 49.4M | case WC_SJIS_SHIFT_L: |
238 | 182k | case WC_SJIS_SHIFT_H: |
239 | 182k | if (WC_SJIS_MAP[*p] & LB) { |
240 | 138k | jis[0] = *(p-1); |
241 | 138k | jis[1] = *p; |
242 | 138k | sjis_to_jisx0208(jis[0], jis[1]); |
243 | 138k | cc.code = ((wc_uint32)jis[0] << 8) | jis[1]; |
244 | 138k | cc.ccs = wc_jisx0208_or_jisx02131(cc.code); |
245 | 138k | if (cc.ccs == WC_CCS_JIS_X_0208) |
246 | 119k | wtf_push(os, cc.ccs, cc.code); |
247 | 19.2k | else |
248 | 19.2k | wtf_push(os, WC_CCS_SJIS_EXT, ((wc_uint32)*(p-1) << 8) | *p); |
249 | 138k | } else |
250 | 43.3k | wtf_push_unknown(os, p-1, 2); |
251 | 182k | state = WC_SJIS_NOSTATE; |
252 | 182k | break; |
253 | 79.9k | case WC_SJIS_SHIFT_X: |
254 | 79.9k | if (WC_SJIS_MAP[*p] & LB) |
255 | 73.9k | wtf_push(os, WC_CCS_SJIS_EXT, ((wc_uint32)*(p-1) << 8) | *p); |
256 | 6.01k | else |
257 | 6.01k | wtf_push_unknown(os, p-1, 2); |
258 | 79.9k | state = WC_SJIS_NOSTATE; |
259 | 79.9k | break; |
260 | 49.7M | } |
261 | 49.7M | } |
262 | 1.46k | switch (state) { |
263 | 34 | case WC_SJIS_SHIFT_L: |
264 | 54 | case WC_SJIS_SHIFT_H: |
265 | 61 | case WC_SJIS_SHIFT_X: |
266 | 61 | wtf_push_unknown(os, p-1, 1); |
267 | 61 | break; |
268 | 1.46k | } |
269 | 1.46k | return os; |
270 | 1.46k | } |
271 | | |
272 | | Str |
273 | | wc_conv_from_sjisx0213(Str is, wc_ces ces) |
274 | 198 | { |
275 | 198 | Str os; |
276 | 198 | wc_uchar *sp = (wc_uchar *)is->ptr; |
277 | 198 | wc_uchar *ep = sp + is->length; |
278 | 198 | wc_uchar *p; |
279 | 198 | wc_uchar jis[2]; |
280 | 198 | int state = WC_SJIS_NOSTATE; |
281 | 198 | wc_wchar_t cc; |
282 | | |
283 | 588 | for (p = sp; p < ep && *p < 0x80; p++) |
284 | 390 | ; |
285 | 198 | if (p == ep) |
286 | 13 | return is; |
287 | 185 | os = Strnew_size(is->length); |
288 | 185 | if (p > sp) |
289 | 53 | Strcat_charp_n(os, is->ptr, (int)(p - sp)); |
290 | | |
291 | 4.31M | for (; p < ep; p++) { |
292 | 4.31M | switch (state) { |
293 | 4.24M | case WC_SJIS_NOSTATE: |
294 | 4.24M | switch (WC_SJIS_MAP[*p]) { |
295 | 44.8k | case SL: |
296 | 44.8k | state = WC_SJIS_SHIFT_L; |
297 | 44.8k | break; |
298 | 9.94k | case SH: |
299 | 9.94k | state = WC_SJIS_SHIFT_H; |
300 | 9.94k | break; |
301 | 11.0k | case SX: |
302 | 11.0k | state = WC_SJIS_SHIFT_X; |
303 | 11.0k | break; |
304 | 3.92M | case SK: |
305 | 3.92M | wtf_push(os, WC_CCS_JIS_X_0201K, (wc_uint32)*p); |
306 | 3.92M | break; |
307 | 16.7k | case S80: |
308 | 23.4k | case SA0: |
309 | 72.7k | case C1: |
310 | 72.7k | wtf_push_unknown(os, p, 1); |
311 | 72.7k | break; |
312 | 177k | default: |
313 | 177k | Strcat_char(os, (char)*p); |
314 | 177k | break; |
315 | 4.24M | } |
316 | 4.24M | break; |
317 | 4.24M | case WC_SJIS_SHIFT_L: |
318 | 54.8k | case WC_SJIS_SHIFT_H: |
319 | 54.8k | if (WC_SJIS_MAP[*p] & LB) { |
320 | 30.8k | jis[0] = *(p-1); |
321 | 30.8k | jis[1] = *p; |
322 | 30.8k | sjis_to_jisx0208(jis[0], jis[1]); |
323 | 30.8k | cc.code = ((wc_uint32)jis[0] << 8) | jis[1]; |
324 | 30.8k | cc.ccs = wc_jisx0208_or_jisx02131(cc.code); |
325 | 30.8k | wtf_push(os, cc.ccs, cc.code); |
326 | 30.8k | } else |
327 | 24.0k | wtf_push_unknown(os, p-1, 2); |
328 | 54.8k | state = WC_SJIS_NOSTATE; |
329 | 54.8k | break; |
330 | 11.0k | case WC_SJIS_SHIFT_X: |
331 | 11.0k | if (WC_SJIS_MAP[*p] & LB) { |
332 | 8.83k | jis[0] = *(p-1); |
333 | 8.83k | jis[1] = *p; |
334 | 8.83k | sjis_to_jisx02132(jis[0], jis[1]); |
335 | 8.83k | wtf_push(os, WC_CCS_JIS_X_0213_2, ((wc_uint32)jis[0] << 8) | jis[1]); |
336 | 8.83k | } else |
337 | 2.16k | wtf_push_unknown(os, p-1, 2); |
338 | 11.0k | state = WC_SJIS_NOSTATE; |
339 | 11.0k | break; |
340 | 4.31M | } |
341 | 4.31M | } |
342 | 185 | switch (state) { |
343 | 8 | case WC_SJIS_SHIFT_L: |
344 | 11 | case WC_SJIS_SHIFT_H: |
345 | 17 | case WC_SJIS_SHIFT_X: |
346 | 17 | wtf_push_unknown(os, p-1, 1); |
347 | 17 | break; |
348 | 185 | } |
349 | 185 | return os; |
350 | 185 | } |
351 | | |
352 | | void |
353 | | wc_push_to_sjis(Str os, wc_wchar_t cc, wc_status *st) |
354 | 3.59M | { |
355 | 3.59M | wc_uchar ub, lb; |
356 | | |
357 | 3.70M | while (1) { |
358 | 3.70M | switch (cc.ccs) { |
359 | 82.1k | case WC_CCS_US_ASCII: |
360 | 82.1k | Strcat_char(os, cc.code); |
361 | 82.1k | return; |
362 | 6.84k | case WC_CCS_JIS_X_0201K: |
363 | 6.84k | if (WcOption.use_jisx0201k) { |
364 | 0 | Strcat_char(os, cc.code | 0x80); |
365 | 0 | return; |
366 | 6.84k | } else if (WcOption.fix_width_conv) |
367 | 6.84k | cc.ccs = WC_CCS_UNKNOWN; |
368 | 0 | else |
369 | 0 | cc = wc_jisx0201k_to_jisx0208(cc); |
370 | 6.84k | continue; |
371 | 3.34M | case WC_CCS_JIS_X_0208: |
372 | 3.34M | ub = (cc.code >> 8) & 0x7f; |
373 | 3.34M | lb = cc.code & 0x7f; |
374 | 3.34M | jisx0208_to_sjis(ub, lb); |
375 | 3.34M | Strcat_char(os, ub); |
376 | 3.34M | Strcat_char(os, lb); |
377 | 3.34M | return; |
378 | 1.97k | case WC_CCS_SJIS_EXT_1: |
379 | 3.03k | case WC_CCS_SJIS_EXT_2: |
380 | 3.03k | cc = wc_cs94w_to_sjis_ext(cc); |
381 | 54.2k | case WC_CCS_SJIS_EXT: |
382 | 54.2k | Strcat_char(os, (char)(cc.code >> 8)); |
383 | 54.2k | Strcat_char(os, (char)(cc.code & 0xff)); |
384 | 54.2k | return; |
385 | 8.44k | case WC_CCS_UNKNOWN_W: |
386 | 8.44k | if (!WcOption.no_replace) |
387 | 8.44k | Strcat_charp(os, WC_REPLACE_W); |
388 | 8.44k | return; |
389 | 102k | case WC_CCS_UNKNOWN: |
390 | 102k | if (!WcOption.no_replace) |
391 | 102k | Strcat_charp(os, WC_REPLACE); |
392 | 102k | return; |
393 | 101k | default: |
394 | 101k | #ifdef USE_UNICODE |
395 | 101k | if (WcOption.ucs_conv) |
396 | 101k | cc = wc_any_to_any_ces(cc, st); |
397 | 0 | else |
398 | 0 | #endif |
399 | 0 | cc.ccs = WC_CCS_IS_WIDE(cc.ccs) ? WC_CCS_UNKNOWN_W : WC_CCS_UNKNOWN; |
400 | 101k | continue; |
401 | 3.70M | } |
402 | 3.70M | } |
403 | 3.59M | } |
404 | | |
405 | | void |
406 | | wc_push_to_sjisx0213(Str os, wc_wchar_t cc, wc_status *st) |
407 | 4.43M | { |
408 | 4.43M | wc_uchar ub, lb; |
409 | | |
410 | 4.55M | while (1) { |
411 | 4.55M | switch (cc.ccs) { |
412 | 9.87k | case WC_CCS_US_ASCII: |
413 | 9.87k | Strcat_char(os, cc.code); |
414 | 9.87k | return; |
415 | 30.7k | case WC_CCS_JIS_X_0201K: |
416 | 30.7k | if (WcOption.use_jisx0201k) { |
417 | 0 | Strcat_char(os, cc.code | 0x80); |
418 | 0 | return; |
419 | 30.7k | } else if (WcOption.fix_width_conv) |
420 | 30.7k | cc.ccs = WC_CCS_UNKNOWN; |
421 | 0 | else |
422 | 0 | cc = wc_jisx0201k_to_jisx0208(cc); |
423 | 30.7k | continue; |
424 | 30.7k | case WC_CCS_JIS_X_0213_1: |
425 | 818 | if (! WcOption.use_jisx0213) { |
426 | 818 | cc.ccs = WC_CCS_UNKNOWN_W; |
427 | 818 | continue; |
428 | 818 | } |
429 | 4.28M | case WC_CCS_JIS_X_0208: |
430 | 4.28M | ub = (cc.code >> 8) & 0x7f; |
431 | 4.28M | lb = cc.code & 0x7f; |
432 | 4.28M | jisx0208_to_sjis(ub, lb); |
433 | 4.28M | Strcat_char(os, ub); |
434 | 4.28M | Strcat_char(os, lb); |
435 | 4.28M | return; |
436 | 2.88k | case WC_CCS_JIS_X_0213_2: |
437 | 2.88k | if (! WcOption.use_jisx0213) { |
438 | 2.88k | cc.ccs = WC_CCS_UNKNOWN_W; |
439 | 2.88k | continue; |
440 | 2.88k | } |
441 | 0 | ub = (cc.code >> 8) & 0x7f; |
442 | 0 | lb = cc.code & 0x7f; |
443 | 0 | jisx02132_to_sjis(ub, lb); |
444 | 0 | if (ub) { |
445 | 0 | Strcat_char(os, ub); |
446 | 0 | Strcat_char(os, lb); |
447 | 0 | return; |
448 | 0 | } |
449 | 9.95k | case WC_CCS_UNKNOWN_W: |
450 | 9.95k | if (!WcOption.no_replace) |
451 | 9.95k | Strcat_charp(os, WC_REPLACE_W); |
452 | 9.95k | return; |
453 | 131k | case WC_CCS_UNKNOWN: |
454 | 131k | if (!WcOption.no_replace) |
455 | 131k | Strcat_charp(os, WC_REPLACE); |
456 | 131k | return; |
457 | 84.9k | default: |
458 | 84.9k | #ifdef USE_UNICODE |
459 | 84.9k | if (WcOption.ucs_conv) |
460 | 84.9k | cc = wc_any_to_any_ces(cc, st); |
461 | 0 | else |
462 | 0 | #endif |
463 | 0 | cc.ccs = WC_CCS_IS_WIDE(cc.ccs) ? WC_CCS_UNKNOWN_W : WC_CCS_UNKNOWN; |
464 | 84.9k | continue; |
465 | 4.55M | } |
466 | 4.55M | } |
467 | 4.43M | } |
468 | | |
469 | | Str |
470 | | wc_char_conv_from_sjis(wc_uchar c, wc_status *st) |
471 | 0 | { |
472 | 0 | static Str os; |
473 | 0 | static wc_uchar jis[2]; |
474 | 0 | wc_wchar_t cc; |
475 | |
|
476 | 0 | if (st->state == -1) { |
477 | 0 | st->state = WC_SJIS_NOSTATE; |
478 | 0 | os = Strnew_size(8); |
479 | 0 | } |
480 | |
|
481 | 0 | switch (st->state) { |
482 | 0 | case WC_SJIS_NOSTATE: |
483 | 0 | switch (WC_SJIS_MAP[c]) { |
484 | 0 | case SL: |
485 | 0 | jis[0] = c; |
486 | 0 | st->state = WC_SJIS_SHIFT_L; |
487 | 0 | return NULL; |
488 | 0 | case SH: |
489 | 0 | jis[0] = c; |
490 | 0 | st->state = WC_SJIS_SHIFT_H; |
491 | 0 | return NULL; |
492 | 0 | case SX: |
493 | 0 | jis[0] = c; |
494 | 0 | st->state = WC_SJIS_SHIFT_X; |
495 | 0 | return NULL; |
496 | 0 | case SK: |
497 | 0 | wtf_push(os, WC_CCS_JIS_X_0201K, (wc_uint32)c); |
498 | 0 | break; |
499 | 0 | case S80: |
500 | 0 | case SA0: |
501 | 0 | case C1: |
502 | 0 | break; |
503 | 0 | default: |
504 | 0 | Strcat_char(os, (char)c); |
505 | 0 | break; |
506 | 0 | } |
507 | 0 | break; |
508 | 0 | case WC_SJIS_SHIFT_L: |
509 | 0 | case WC_SJIS_SHIFT_H: |
510 | 0 | if (WC_SJIS_MAP[c] & LB) { |
511 | 0 | jis[1] = c; |
512 | 0 | sjis_to_jisx0208(jis[0], jis[1]); |
513 | 0 | cc.code = ((wc_uint32)jis[0] << 8) | jis[1]; |
514 | 0 | cc.ccs = wc_jisx0208_or_jisx02131(cc.code); |
515 | 0 | if (cc.ccs == WC_CCS_JIS_X_0208) |
516 | 0 | wtf_push(os, cc.ccs, cc.code); |
517 | 0 | else |
518 | 0 | wtf_push(os, WC_CCS_SJIS_EXT, ((wc_uint32)jis[0] << 8) | jis[1]); |
519 | 0 | } |
520 | 0 | st->state = WC_SJIS_NOSTATE; |
521 | 0 | break; |
522 | 0 | case WC_SJIS_SHIFT_X: |
523 | 0 | if (WC_SJIS_MAP[c] & LB) { |
524 | 0 | jis[1] = c; |
525 | 0 | wtf_push(os, WC_CCS_SJIS_EXT, ((wc_uint32)jis[0] << 8) | jis[1]); |
526 | 0 | } |
527 | 0 | st->state = WC_SJIS_NOSTATE; |
528 | 0 | break; |
529 | 0 | } |
530 | 0 | st->state = -1; |
531 | 0 | return os; |
532 | 0 | } |
533 | | |
534 | | Str |
535 | | wc_char_conv_from_sjisx0213(wc_uchar c, wc_status *st) |
536 | 0 | { |
537 | 0 | static Str os; |
538 | 0 | static wc_uchar jis[2]; |
539 | 0 | wc_wchar_t cc; |
540 | |
|
541 | 0 | if (st->state == -1) { |
542 | 0 | st->state = WC_SJIS_NOSTATE; |
543 | 0 | os = Strnew_size(8); |
544 | 0 | } |
545 | |
|
546 | 0 | switch (st->state) { |
547 | 0 | case WC_SJIS_NOSTATE: |
548 | 0 | switch (WC_SJIS_MAP[c]) { |
549 | 0 | case SL: |
550 | 0 | jis[0] = c; |
551 | 0 | st->state = WC_SJIS_SHIFT_L; |
552 | 0 | return NULL; |
553 | 0 | case SH: |
554 | 0 | jis[0] = c; |
555 | 0 | st->state = WC_SJIS_SHIFT_H; |
556 | 0 | return NULL; |
557 | 0 | case SX: |
558 | 0 | jis[0] = c; |
559 | 0 | st->state = WC_SJIS_SHIFT_X; |
560 | 0 | return NULL; |
561 | 0 | case SK: |
562 | 0 | wtf_push(os, WC_CCS_JIS_X_0201K, (wc_uint32)c); |
563 | 0 | break; |
564 | 0 | case S80: |
565 | 0 | case SA0: |
566 | 0 | case C1: |
567 | 0 | break; |
568 | 0 | default: |
569 | 0 | Strcat_char(os, (char)c); |
570 | 0 | break; |
571 | 0 | } |
572 | 0 | break; |
573 | 0 | case WC_SJIS_SHIFT_L: |
574 | 0 | case WC_SJIS_SHIFT_H: |
575 | 0 | if (WC_SJIS_MAP[c] & LB) { |
576 | 0 | jis[1] = c; |
577 | 0 | sjis_to_jisx0208(jis[0], jis[1]); |
578 | 0 | cc.code = ((wc_uint32)jis[0] << 8) | jis[1]; |
579 | 0 | cc.ccs = wc_jisx0208_or_jisx02131(cc.code); |
580 | 0 | wtf_push(os, cc.ccs, cc.code); |
581 | 0 | } |
582 | 0 | st->state = WC_SJIS_NOSTATE; |
583 | 0 | break; |
584 | 0 | case WC_SJIS_SHIFT_X: |
585 | 0 | if (WC_SJIS_MAP[c] & LB) { |
586 | 0 | jis[1] = c; |
587 | 0 | sjis_to_jisx02132(jis[0], jis[1]); |
588 | 0 | wtf_push(os, WC_CCS_JIS_X_0213_2, ((wc_uint32)jis[0] << 8) | jis[1]); |
589 | 0 | } |
590 | 0 | st->state = WC_SJIS_NOSTATE; |
591 | 0 | break; |
592 | 0 | } |
593 | 0 | st->state = -1; |
594 | 0 | return os; |
595 | 0 | } |