]>
Commit | Line | Data |
---|---|---|
7eb75bcc DM |
1 | /*\r |
2 | * _codecs_jp.c: Codecs collection for Japanese encodings\r | |
3 | *\r | |
4 | * Written by Hye-Shik Chang <perky@FreeBSD.org>\r | |
5 | */\r | |
6 | \r | |
7 | #define USING_BINARY_PAIR_SEARCH\r | |
8 | #define EMPBASE 0x20000\r | |
9 | \r | |
10 | #include "cjkcodecs.h"\r | |
11 | #include "mappings_jp.h"\r | |
12 | #include "mappings_jisx0213_pair.h"\r | |
13 | #include "alg_jisx0201.h"\r | |
14 | #include "emu_jisx0213_2000.h"\r | |
15 | \r | |
16 | /*\r | |
17 | * CP932 codec\r | |
18 | */\r | |
19 | \r | |
20 | ENCODER(cp932)\r | |
21 | {\r | |
22 | while (inleft > 0) {\r | |
23 | Py_UNICODE c = IN1;\r | |
24 | DBCHAR code;\r | |
25 | unsigned char c1, c2;\r | |
26 | \r | |
27 | if (c <= 0x80) {\r | |
28 | WRITE1((unsigned char)c)\r | |
29 | NEXT(1, 1)\r | |
30 | continue;\r | |
31 | }\r | |
32 | else if (c >= 0xff61 && c <= 0xff9f) {\r | |
33 | WRITE1(c - 0xfec0)\r | |
34 | NEXT(1, 1)\r | |
35 | continue;\r | |
36 | }\r | |
37 | else if (c >= 0xf8f0 && c <= 0xf8f3) {\r | |
38 | /* Windows compatibility */\r | |
39 | REQUIRE_OUTBUF(1)\r | |
40 | if (c == 0xf8f0)\r | |
41 | OUT1(0xa0)\r | |
42 | else\r | |
43 | OUT1(c - 0xfef1 + 0xfd)\r | |
44 | NEXT(1, 1)\r | |
45 | continue;\r | |
46 | }\r | |
47 | \r | |
48 | UCS4INVALID(c)\r | |
49 | REQUIRE_OUTBUF(2)\r | |
50 | \r | |
51 | TRYMAP_ENC(cp932ext, code, c) {\r | |
52 | OUT1(code >> 8)\r | |
53 | OUT2(code & 0xff)\r | |
54 | }\r | |
55 | else TRYMAP_ENC(jisxcommon, code, c) {\r | |
56 | if (code & 0x8000) /* MSB set: JIS X 0212 */\r | |
57 | return 1;\r | |
58 | \r | |
59 | /* JIS X 0208 */\r | |
60 | c1 = code >> 8;\r | |
61 | c2 = code & 0xff;\r | |
62 | c2 = (((c1 - 0x21) & 1) ? 0x5e : 0) + (c2 - 0x21);\r | |
63 | c1 = (c1 - 0x21) >> 1;\r | |
64 | OUT1(c1 < 0x1f ? c1 + 0x81 : c1 + 0xc1)\r | |
65 | OUT2(c2 < 0x3f ? c2 + 0x40 : c2 + 0x41)\r | |
66 | }\r | |
67 | else if (c >= 0xe000 && c < 0xe758) {\r | |
68 | /* User-defined area */\r | |
69 | c1 = (Py_UNICODE)(c - 0xe000) / 188;\r | |
70 | c2 = (Py_UNICODE)(c - 0xe000) % 188;\r | |
71 | OUT1(c1 + 0xf0)\r | |
72 | OUT2(c2 < 0x3f ? c2 + 0x40 : c2 + 0x41)\r | |
73 | }\r | |
74 | else\r | |
75 | return 1;\r | |
76 | \r | |
77 | NEXT(1, 2)\r | |
78 | }\r | |
79 | \r | |
80 | return 0;\r | |
81 | }\r | |
82 | \r | |
83 | DECODER(cp932)\r | |
84 | {\r | |
85 | while (inleft > 0) {\r | |
86 | unsigned char c = IN1, c2;\r | |
87 | \r | |
88 | REQUIRE_OUTBUF(1)\r | |
89 | if (c <= 0x80) {\r | |
90 | OUT1(c)\r | |
91 | NEXT(1, 1)\r | |
92 | continue;\r | |
93 | }\r | |
94 | else if (c >= 0xa0 && c <= 0xdf) {\r | |
95 | if (c == 0xa0)\r | |
96 | OUT1(0xf8f0) /* half-width katakana */\r | |
97 | else\r | |
98 | OUT1(0xfec0 + c)\r | |
99 | NEXT(1, 1)\r | |
100 | continue;\r | |
101 | }\r | |
102 | else if (c >= 0xfd/* && c <= 0xff*/) {\r | |
103 | /* Windows compatibility */\r | |
104 | OUT1(0xf8f1 - 0xfd + c)\r | |
105 | NEXT(1, 1)\r | |
106 | continue;\r | |
107 | }\r | |
108 | \r | |
109 | REQUIRE_INBUF(2)\r | |
110 | c2 = IN2;\r | |
111 | \r | |
112 | TRYMAP_DEC(cp932ext, **outbuf, c, c2);\r | |
113 | else if ((c >= 0x81 && c <= 0x9f) || (c >= 0xe0 && c <= 0xea)){\r | |
114 | if (c2 < 0x40 || (c2 > 0x7e && c2 < 0x80) || c2 > 0xfc)\r | |
115 | return 2;\r | |
116 | \r | |
117 | c = (c < 0xe0 ? c - 0x81 : c - 0xc1);\r | |
118 | c2 = (c2 < 0x80 ? c2 - 0x40 : c2 - 0x41);\r | |
119 | c = (2 * c + (c2 < 0x5e ? 0 : 1) + 0x21);\r | |
120 | c2 = (c2 < 0x5e ? c2 : c2 - 0x5e) + 0x21;\r | |
121 | \r | |
122 | TRYMAP_DEC(jisx0208, **outbuf, c, c2);\r | |
123 | else return 2;\r | |
124 | }\r | |
125 | else if (c >= 0xf0 && c <= 0xf9) {\r | |
126 | if ((c2 >= 0x40 && c2 <= 0x7e) ||\r | |
127 | (c2 >= 0x80 && c2 <= 0xfc))\r | |
128 | OUT1(0xe000 + 188 * (c - 0xf0) +\r | |
129 | (c2 < 0x80 ? c2 - 0x40 : c2 - 0x41))\r | |
130 | else\r | |
131 | return 2;\r | |
132 | }\r | |
133 | else\r | |
134 | return 2;\r | |
135 | \r | |
136 | NEXT(2, 1)\r | |
137 | }\r | |
138 | \r | |
139 | return 0;\r | |
140 | }\r | |
141 | \r | |
142 | \r | |
143 | /*\r | |
144 | * EUC-JIS-2004 codec\r | |
145 | */\r | |
146 | \r | |
147 | ENCODER(euc_jis_2004)\r | |
148 | {\r | |
149 | while (inleft > 0) {\r | |
150 | ucs4_t c = IN1;\r | |
151 | DBCHAR code;\r | |
152 | Py_ssize_t insize;\r | |
153 | \r | |
154 | if (c < 0x80) {\r | |
155 | WRITE1(c)\r | |
156 | NEXT(1, 1)\r | |
157 | continue;\r | |
158 | }\r | |
159 | \r | |
160 | DECODE_SURROGATE(c)\r | |
161 | insize = GET_INSIZE(c);\r | |
162 | \r | |
163 | if (c <= 0xFFFF) {\r | |
164 | EMULATE_JISX0213_2000_ENCODE_BMP(code, c)\r | |
165 | else TRYMAP_ENC(jisx0213_bmp, code, c) {\r | |
166 | if (code == MULTIC) {\r | |
167 | if (inleft < 2) {\r | |
168 | if (flags & MBENC_FLUSH) {\r | |
169 | code = find_pairencmap(\r | |
170 | (ucs2_t)c, 0,\r | |
171 | jisx0213_pair_encmap,\r | |
172 | JISX0213_ENCPAIRS);\r | |
173 | if (code == DBCINV)\r | |
174 | return 1;\r | |
175 | }\r | |
176 | else\r | |
177 | return MBERR_TOOFEW;\r | |
178 | }\r | |
179 | else {\r | |
180 | code = find_pairencmap(\r | |
181 | (ucs2_t)c, (*inbuf)[1],\r | |
182 | jisx0213_pair_encmap,\r | |
183 | JISX0213_ENCPAIRS);\r | |
184 | if (code == DBCINV) {\r | |
185 | code = find_pairencmap(\r | |
186 | (ucs2_t)c, 0,\r | |
187 | jisx0213_pair_encmap,\r | |
188 | JISX0213_ENCPAIRS);\r | |
189 | if (code == DBCINV)\r | |
190 | return 1;\r | |
191 | } else\r | |
192 | insize = 2;\r | |
193 | }\r | |
194 | }\r | |
195 | }\r | |
196 | else TRYMAP_ENC(jisxcommon, code, c);\r | |
197 | else if (c >= 0xff61 && c <= 0xff9f) {\r | |
198 | /* JIS X 0201 half-width katakana */\r | |
199 | WRITE2(0x8e, c - 0xfec0)\r | |
200 | NEXT(1, 2)\r | |
201 | continue;\r | |
202 | }\r | |
203 | else if (c == 0xff3c)\r | |
204 | /* F/W REVERSE SOLIDUS (see NOTES) */\r | |
205 | code = 0x2140;\r | |
206 | else if (c == 0xff5e)\r | |
207 | /* F/W TILDE (see NOTES) */\r | |
208 | code = 0x2232;\r | |
209 | else\r | |
210 | return 1;\r | |
211 | }\r | |
212 | else if (c >> 16 == EMPBASE >> 16) {\r | |
213 | EMULATE_JISX0213_2000_ENCODE_EMP(code, c)\r | |
214 | else TRYMAP_ENC(jisx0213_emp, code, c & 0xffff);\r | |
215 | else return insize;\r | |
216 | }\r | |
217 | else\r | |
218 | return insize;\r | |
219 | \r | |
220 | if (code & 0x8000) {\r | |
221 | /* Codeset 2 */\r | |
222 | WRITE3(0x8f, code >> 8, (code & 0xFF) | 0x80)\r | |
223 | NEXT(insize, 3)\r | |
224 | } else {\r | |
225 | /* Codeset 1 */\r | |
226 | WRITE2((code >> 8) | 0x80, (code & 0xFF) | 0x80)\r | |
227 | NEXT(insize, 2)\r | |
228 | }\r | |
229 | }\r | |
230 | \r | |
231 | return 0;\r | |
232 | }\r | |
233 | \r | |
234 | DECODER(euc_jis_2004)\r | |
235 | {\r | |
236 | while (inleft > 0) {\r | |
237 | unsigned char c = IN1;\r | |
238 | ucs4_t code;\r | |
239 | \r | |
240 | REQUIRE_OUTBUF(1)\r | |
241 | \r | |
242 | if (c < 0x80) {\r | |
243 | OUT1(c)\r | |
244 | NEXT(1, 1)\r | |
245 | continue;\r | |
246 | }\r | |
247 | \r | |
248 | if (c == 0x8e) {\r | |
249 | /* JIS X 0201 half-width katakana */\r | |
250 | unsigned char c2;\r | |
251 | \r | |
252 | REQUIRE_INBUF(2)\r | |
253 | c2 = IN2;\r | |
254 | if (c2 >= 0xa1 && c2 <= 0xdf) {\r | |
255 | OUT1(0xfec0 + c2)\r | |
256 | NEXT(2, 1)\r | |
257 | }\r | |
258 | else\r | |
259 | return 2;\r | |
260 | }\r | |
261 | else if (c == 0x8f) {\r | |
262 | unsigned char c2, c3;\r | |
263 | \r | |
264 | REQUIRE_INBUF(3)\r | |
265 | c2 = IN2 ^ 0x80;\r | |
266 | c3 = IN3 ^ 0x80;\r | |
267 | \r | |
268 | /* JIS X 0213 Plane 2 or JIS X 0212 (see NOTES) */\r | |
269 | EMULATE_JISX0213_2000_DECODE_PLANE2(**outbuf, c2, c3)\r | |
270 | else TRYMAP_DEC(jisx0213_2_bmp, **outbuf, c2, c3) ;\r | |
271 | else TRYMAP_DEC(jisx0213_2_emp, code, c2, c3) {\r | |
272 | WRITEUCS4(EMPBASE | code)\r | |
273 | NEXT_IN(3)\r | |
274 | continue;\r | |
275 | }\r | |
276 | else TRYMAP_DEC(jisx0212, **outbuf, c2, c3) ;\r | |
277 | else return 3;\r | |
278 | NEXT(3, 1)\r | |
279 | }\r | |
280 | else {\r | |
281 | unsigned char c2;\r | |
282 | \r | |
283 | REQUIRE_INBUF(2)\r | |
284 | c ^= 0x80;\r | |
285 | c2 = IN2 ^ 0x80;\r | |
286 | \r | |
287 | /* JIS X 0213 Plane 1 */\r | |
288 | EMULATE_JISX0213_2000_DECODE_PLANE1(**outbuf, c, c2)\r | |
289 | else if (c == 0x21 && c2 == 0x40) **outbuf = 0xff3c;\r | |
290 | else if (c == 0x22 && c2 == 0x32) **outbuf = 0xff5e;\r | |
291 | else TRYMAP_DEC(jisx0208, **outbuf, c, c2);\r | |
292 | else TRYMAP_DEC(jisx0213_1_bmp, **outbuf, c, c2);\r | |
293 | else TRYMAP_DEC(jisx0213_1_emp, code, c, c2) {\r | |
294 | WRITEUCS4(EMPBASE | code)\r | |
295 | NEXT_IN(2)\r | |
296 | continue;\r | |
297 | }\r | |
298 | else TRYMAP_DEC(jisx0213_pair, code, c, c2) {\r | |
299 | WRITE2(code >> 16, code & 0xffff)\r | |
300 | NEXT(2, 2)\r | |
301 | continue;\r | |
302 | }\r | |
303 | else return 2;\r | |
304 | NEXT(2, 1)\r | |
305 | }\r | |
306 | }\r | |
307 | \r | |
308 | return 0;\r | |
309 | }\r | |
310 | \r | |
311 | \r | |
312 | /*\r | |
313 | * EUC-JP codec\r | |
314 | */\r | |
315 | \r | |
316 | ENCODER(euc_jp)\r | |
317 | {\r | |
318 | while (inleft > 0) {\r | |
319 | Py_UNICODE c = IN1;\r | |
320 | DBCHAR code;\r | |
321 | \r | |
322 | if (c < 0x80) {\r | |
323 | WRITE1((unsigned char)c)\r | |
324 | NEXT(1, 1)\r | |
325 | continue;\r | |
326 | }\r | |
327 | \r | |
328 | UCS4INVALID(c)\r | |
329 | \r | |
330 | TRYMAP_ENC(jisxcommon, code, c);\r | |
331 | else if (c >= 0xff61 && c <= 0xff9f) {\r | |
332 | /* JIS X 0201 half-width katakana */\r | |
333 | WRITE2(0x8e, c - 0xfec0)\r | |
334 | NEXT(1, 2)\r | |
335 | continue;\r | |
336 | }\r | |
337 | #ifndef STRICT_BUILD\r | |
338 | else if (c == 0xff3c) /* FULL-WIDTH REVERSE SOLIDUS */\r | |
339 | code = 0x2140;\r | |
340 | else if (c == 0xa5) { /* YEN SIGN */\r | |
341 | WRITE1(0x5c);\r | |
342 | NEXT(1, 1)\r | |
343 | continue;\r | |
344 | } else if (c == 0x203e) { /* OVERLINE */\r | |
345 | WRITE1(0x7e);\r | |
346 | NEXT(1, 1)\r | |
347 | continue;\r | |
348 | }\r | |
349 | #endif\r | |
350 | else\r | |
351 | return 1;\r | |
352 | \r | |
353 | if (code & 0x8000) {\r | |
354 | /* JIS X 0212 */\r | |
355 | WRITE3(0x8f, code >> 8, (code & 0xFF) | 0x80)\r | |
356 | NEXT(1, 3)\r | |
357 | } else {\r | |
358 | /* JIS X 0208 */\r | |
359 | WRITE2((code >> 8) | 0x80, (code & 0xFF) | 0x80)\r | |
360 | NEXT(1, 2)\r | |
361 | }\r | |
362 | }\r | |
363 | \r | |
364 | return 0;\r | |
365 | }\r | |
366 | \r | |
367 | DECODER(euc_jp)\r | |
368 | {\r | |
369 | while (inleft > 0) {\r | |
370 | unsigned char c = IN1;\r | |
371 | \r | |
372 | REQUIRE_OUTBUF(1)\r | |
373 | \r | |
374 | if (c < 0x80) {\r | |
375 | OUT1(c)\r | |
376 | NEXT(1, 1)\r | |
377 | continue;\r | |
378 | }\r | |
379 | \r | |
380 | if (c == 0x8e) {\r | |
381 | /* JIS X 0201 half-width katakana */\r | |
382 | unsigned char c2;\r | |
383 | \r | |
384 | REQUIRE_INBUF(2)\r | |
385 | c2 = IN2;\r | |
386 | if (c2 >= 0xa1 && c2 <= 0xdf) {\r | |
387 | OUT1(0xfec0 + c2)\r | |
388 | NEXT(2, 1)\r | |
389 | }\r | |
390 | else\r | |
391 | return 2;\r | |
392 | }\r | |
393 | else if (c == 0x8f) {\r | |
394 | unsigned char c2, c3;\r | |
395 | \r | |
396 | REQUIRE_INBUF(3)\r | |
397 | c2 = IN2;\r | |
398 | c3 = IN3;\r | |
399 | /* JIS X 0212 */\r | |
400 | TRYMAP_DEC(jisx0212, **outbuf, c2 ^ 0x80, c3 ^ 0x80) {\r | |
401 | NEXT(3, 1)\r | |
402 | }\r | |
403 | else\r | |
404 | return 3;\r | |
405 | }\r | |
406 | else {\r | |
407 | unsigned char c2;\r | |
408 | \r | |
409 | REQUIRE_INBUF(2)\r | |
410 | c2 = IN2;\r | |
411 | /* JIS X 0208 */\r | |
412 | #ifndef STRICT_BUILD\r | |
413 | if (c == 0xa1 && c2 == 0xc0)\r | |
414 | /* FULL-WIDTH REVERSE SOLIDUS */\r | |
415 | **outbuf = 0xff3c;\r | |
416 | else\r | |
417 | #endif\r | |
418 | TRYMAP_DEC(jisx0208, **outbuf,\r | |
419 | c ^ 0x80, c2 ^ 0x80) ;\r | |
420 | else return 2;\r | |
421 | NEXT(2, 1)\r | |
422 | }\r | |
423 | }\r | |
424 | \r | |
425 | return 0;\r | |
426 | }\r | |
427 | \r | |
428 | \r | |
429 | /*\r | |
430 | * SHIFT_JIS codec\r | |
431 | */\r | |
432 | \r | |
433 | ENCODER(shift_jis)\r | |
434 | {\r | |
435 | while (inleft > 0) {\r | |
436 | Py_UNICODE c = IN1;\r | |
437 | DBCHAR code;\r | |
438 | unsigned char c1, c2;\r | |
439 | \r | |
440 | #ifdef STRICT_BUILD\r | |
441 | JISX0201_R_ENCODE(c, code)\r | |
442 | #else\r | |
443 | if (c < 0x80) code = c;\r | |
444 | else if (c == 0x00a5) code = 0x5c; /* YEN SIGN */\r | |
445 | else if (c == 0x203e) code = 0x7e; /* OVERLINE */\r | |
446 | #endif\r | |
447 | else JISX0201_K_ENCODE(c, code)\r | |
448 | else UCS4INVALID(c)\r | |
449 | else code = NOCHAR;\r | |
450 | \r | |
451 | if (code < 0x80 || (code >= 0xa1 && code <= 0xdf)) {\r | |
452 | REQUIRE_OUTBUF(1)\r | |
453 | \r | |
454 | OUT1((unsigned char)code)\r | |
455 | NEXT(1, 1)\r | |
456 | continue;\r | |
457 | }\r | |
458 | \r | |
459 | REQUIRE_OUTBUF(2)\r | |
460 | \r | |
461 | if (code == NOCHAR) {\r | |
462 | TRYMAP_ENC(jisxcommon, code, c);\r | |
463 | #ifndef STRICT_BUILD\r | |
464 | else if (c == 0xff3c)\r | |
465 | code = 0x2140; /* FULL-WIDTH REVERSE SOLIDUS */\r | |
466 | #endif\r | |
467 | else\r | |
468 | return 1;\r | |
469 | \r | |
470 | if (code & 0x8000) /* MSB set: JIS X 0212 */\r | |
471 | return 1;\r | |
472 | }\r | |
473 | \r | |
474 | c1 = code >> 8;\r | |
475 | c2 = code & 0xff;\r | |
476 | c2 = (((c1 - 0x21) & 1) ? 0x5e : 0) + (c2 - 0x21);\r | |
477 | c1 = (c1 - 0x21) >> 1;\r | |
478 | OUT1(c1 < 0x1f ? c1 + 0x81 : c1 + 0xc1)\r | |
479 | OUT2(c2 < 0x3f ? c2 + 0x40 : c2 + 0x41)\r | |
480 | NEXT(1, 2)\r | |
481 | }\r | |
482 | \r | |
483 | return 0;\r | |
484 | }\r | |
485 | \r | |
486 | DECODER(shift_jis)\r | |
487 | {\r | |
488 | while (inleft > 0) {\r | |
489 | unsigned char c = IN1;\r | |
490 | \r | |
491 | REQUIRE_OUTBUF(1)\r | |
492 | \r | |
493 | #ifdef STRICT_BUILD\r | |
494 | JISX0201_R_DECODE(c, **outbuf)\r | |
495 | #else\r | |
496 | if (c < 0x80) **outbuf = c;\r | |
497 | #endif\r | |
498 | else JISX0201_K_DECODE(c, **outbuf)\r | |
499 | else if ((c >= 0x81 && c <= 0x9f) || (c >= 0xe0 && c <= 0xea)){\r | |
500 | unsigned char c1, c2;\r | |
501 | \r | |
502 | REQUIRE_INBUF(2)\r | |
503 | c2 = IN2;\r | |
504 | if (c2 < 0x40 || (c2 > 0x7e && c2 < 0x80) || c2 > 0xfc)\r | |
505 | return 2;\r | |
506 | \r | |
507 | c1 = (c < 0xe0 ? c - 0x81 : c - 0xc1);\r | |
508 | c2 = (c2 < 0x80 ? c2 - 0x40 : c2 - 0x41);\r | |
509 | c1 = (2 * c1 + (c2 < 0x5e ? 0 : 1) + 0x21);\r | |
510 | c2 = (c2 < 0x5e ? c2 : c2 - 0x5e) + 0x21;\r | |
511 | \r | |
512 | #ifndef STRICT_BUILD\r | |
513 | if (c1 == 0x21 && c2 == 0x40) {\r | |
514 | /* FULL-WIDTH REVERSE SOLIDUS */\r | |
515 | OUT1(0xff3c)\r | |
516 | NEXT(2, 1)\r | |
517 | continue;\r | |
518 | }\r | |
519 | #endif\r | |
520 | TRYMAP_DEC(jisx0208, **outbuf, c1, c2) {\r | |
521 | NEXT(2, 1)\r | |
522 | continue;\r | |
523 | }\r | |
524 | else\r | |
525 | return 2;\r | |
526 | }\r | |
527 | else\r | |
528 | return 2;\r | |
529 | \r | |
530 | NEXT(1, 1) /* JIS X 0201 */\r | |
531 | }\r | |
532 | \r | |
533 | return 0;\r | |
534 | }\r | |
535 | \r | |
536 | \r | |
537 | /*\r | |
538 | * SHIFT_JIS-2004 codec\r | |
539 | */\r | |
540 | \r | |
541 | ENCODER(shift_jis_2004)\r | |
542 | {\r | |
543 | while (inleft > 0) {\r | |
544 | ucs4_t c = IN1;\r | |
545 | DBCHAR code = NOCHAR;\r | |
546 | int c1, c2;\r | |
547 | Py_ssize_t insize;\r | |
548 | \r | |
549 | JISX0201_ENCODE(c, code)\r | |
550 | else DECODE_SURROGATE(c)\r | |
551 | \r | |
552 | if (code < 0x80 || (code >= 0xa1 && code <= 0xdf)) {\r | |
553 | WRITE1((unsigned char)code)\r | |
554 | NEXT(1, 1)\r | |
555 | continue;\r | |
556 | }\r | |
557 | \r | |
558 | REQUIRE_OUTBUF(2)\r | |
559 | insize = GET_INSIZE(c);\r | |
560 | \r | |
561 | if (code == NOCHAR) {\r | |
562 | if (c <= 0xffff) {\r | |
563 | EMULATE_JISX0213_2000_ENCODE_BMP(code, c)\r | |
564 | else TRYMAP_ENC(jisx0213_bmp, code, c) {\r | |
565 | if (code == MULTIC) {\r | |
566 | if (inleft < 2) {\r | |
567 | if (flags & MBENC_FLUSH) {\r | |
568 | code = find_pairencmap\r | |
569 | ((ucs2_t)c, 0,\r | |
570 | jisx0213_pair_encmap,\r | |
571 | JISX0213_ENCPAIRS);\r | |
572 | if (code == DBCINV)\r | |
573 | return 1;\r | |
574 | }\r | |
575 | else\r | |
576 | return MBERR_TOOFEW;\r | |
577 | }\r | |
578 | else {\r | |
579 | code = find_pairencmap(\r | |
580 | (ucs2_t)c, IN2,\r | |
581 | jisx0213_pair_encmap,\r | |
582 | JISX0213_ENCPAIRS);\r | |
583 | if (code == DBCINV) {\r | |
584 | code = find_pairencmap(\r | |
585 | (ucs2_t)c, 0,\r | |
586 | jisx0213_pair_encmap,\r | |
587 | JISX0213_ENCPAIRS);\r | |
588 | if (code == DBCINV)\r | |
589 | return 1;\r | |
590 | }\r | |
591 | else\r | |
592 | insize = 2;\r | |
593 | }\r | |
594 | }\r | |
595 | }\r | |
596 | else TRYMAP_ENC(jisxcommon, code, c) {\r | |
597 | /* abandon JIS X 0212 codes */\r | |
598 | if (code & 0x8000)\r | |
599 | return 1;\r | |
600 | }\r | |
601 | else return 1;\r | |
602 | }\r | |
603 | else if (c >> 16 == EMPBASE >> 16) {\r | |
604 | EMULATE_JISX0213_2000_ENCODE_EMP(code, c)\r | |
605 | else TRYMAP_ENC(jisx0213_emp, code, c&0xffff);\r | |
606 | else return insize;\r | |
607 | }\r | |
608 | else\r | |
609 | return insize;\r | |
610 | }\r | |
611 | \r | |
612 | c1 = code >> 8;\r | |
613 | c2 = (code & 0xff) - 0x21;\r | |
614 | \r | |
615 | if (c1 & 0x80) { /* Plane 2 */\r | |
616 | if (c1 >= 0xee) c1 -= 0x87;\r | |
617 | else if (c1 >= 0xac || c1 == 0xa8) c1 -= 0x49;\r | |
618 | else c1 -= 0x43;\r | |
619 | }\r | |
620 | else /* Plane 1 */\r | |
621 | c1 -= 0x21;\r | |
622 | \r | |
623 | if (c1 & 1) c2 += 0x5e;\r | |
624 | c1 >>= 1;\r | |
625 | OUT1(c1 + (c1 < 0x1f ? 0x81 : 0xc1))\r | |
626 | OUT2(c2 + (c2 < 0x3f ? 0x40 : 0x41))\r | |
627 | \r | |
628 | NEXT(insize, 2)\r | |
629 | }\r | |
630 | \r | |
631 | return 0;\r | |
632 | }\r | |
633 | \r | |
634 | DECODER(shift_jis_2004)\r | |
635 | {\r | |
636 | while (inleft > 0) {\r | |
637 | unsigned char c = IN1;\r | |
638 | \r | |
639 | REQUIRE_OUTBUF(1)\r | |
640 | JISX0201_DECODE(c, **outbuf)\r | |
641 | else if ((c >= 0x81 && c <= 0x9f) || (c >= 0xe0 && c <= 0xfc)){\r | |
642 | unsigned char c1, c2;\r | |
643 | ucs4_t code;\r | |
644 | \r | |
645 | REQUIRE_INBUF(2)\r | |
646 | c2 = IN2;\r | |
647 | if (c2 < 0x40 || (c2 > 0x7e && c2 < 0x80) || c2 > 0xfc)\r | |
648 | return 2;\r | |
649 | \r | |
650 | c1 = (c < 0xe0 ? c - 0x81 : c - 0xc1);\r | |
651 | c2 = (c2 < 0x80 ? c2 - 0x40 : c2 - 0x41);\r | |
652 | c1 = (2 * c1 + (c2 < 0x5e ? 0 : 1));\r | |
653 | c2 = (c2 < 0x5e ? c2 : c2 - 0x5e) + 0x21;\r | |
654 | \r | |
655 | if (c1 < 0x5e) { /* Plane 1 */\r | |
656 | c1 += 0x21;\r | |
657 | EMULATE_JISX0213_2000_DECODE_PLANE1(**outbuf,\r | |
658 | c1, c2)\r | |
659 | else TRYMAP_DEC(jisx0208, **outbuf, c1, c2) {\r | |
660 | NEXT_OUT(1)\r | |
661 | }\r | |
662 | else TRYMAP_DEC(jisx0213_1_bmp, **outbuf,\r | |
663 | c1, c2) {\r | |
664 | NEXT_OUT(1)\r | |
665 | }\r | |
666 | else TRYMAP_DEC(jisx0213_1_emp, code, c1, c2) {\r | |
667 | WRITEUCS4(EMPBASE | code)\r | |
668 | }\r | |
669 | else TRYMAP_DEC(jisx0213_pair, code, c1, c2) {\r | |
670 | WRITE2(code >> 16, code & 0xffff)\r | |
671 | NEXT_OUT(2)\r | |
672 | }\r | |
673 | else\r | |
674 | return 2;\r | |
675 | NEXT_IN(2)\r | |
676 | }\r | |
677 | else { /* Plane 2 */\r | |
678 | if (c1 >= 0x67) c1 += 0x07;\r | |
679 | else if (c1 >= 0x63 || c1 == 0x5f) c1 -= 0x37;\r | |
680 | else c1 -= 0x3d;\r | |
681 | \r | |
682 | EMULATE_JISX0213_2000_DECODE_PLANE2(**outbuf,\r | |
683 | c1, c2)\r | |
684 | else TRYMAP_DEC(jisx0213_2_bmp, **outbuf,\r | |
685 | c1, c2) ;\r | |
686 | else TRYMAP_DEC(jisx0213_2_emp, code, c1, c2) {\r | |
687 | WRITEUCS4(EMPBASE | code)\r | |
688 | NEXT_IN(2)\r | |
689 | continue;\r | |
690 | }\r | |
691 | else\r | |
692 | return 2;\r | |
693 | NEXT(2, 1)\r | |
694 | }\r | |
695 | continue;\r | |
696 | }\r | |
697 | else\r | |
698 | return 2;\r | |
699 | \r | |
700 | NEXT(1, 1) /* JIS X 0201 */\r | |
701 | }\r | |
702 | \r | |
703 | return 0;\r | |
704 | }\r | |
705 | \r | |
706 | \r | |
707 | BEGIN_MAPPINGS_LIST\r | |
708 | MAPPING_DECONLY(jisx0208)\r | |
709 | MAPPING_DECONLY(jisx0212)\r | |
710 | MAPPING_ENCONLY(jisxcommon)\r | |
711 | MAPPING_DECONLY(jisx0213_1_bmp)\r | |
712 | MAPPING_DECONLY(jisx0213_2_bmp)\r | |
713 | MAPPING_ENCONLY(jisx0213_bmp)\r | |
714 | MAPPING_DECONLY(jisx0213_1_emp)\r | |
715 | MAPPING_DECONLY(jisx0213_2_emp)\r | |
716 | MAPPING_ENCONLY(jisx0213_emp)\r | |
717 | MAPPING_ENCDEC(jisx0213_pair)\r | |
718 | MAPPING_ENCDEC(cp932ext)\r | |
719 | END_MAPPINGS_LIST\r | |
720 | \r | |
721 | BEGIN_CODECS_LIST\r | |
722 | CODEC_STATELESS(shift_jis)\r | |
723 | CODEC_STATELESS(cp932)\r | |
724 | CODEC_STATELESS(euc_jp)\r | |
725 | CODEC_STATELESS(shift_jis_2004)\r | |
726 | CODEC_STATELESS(euc_jis_2004)\r | |
727 | { "euc_jisx0213", (void *)2000, NULL, _STATELESS_METHODS(euc_jis_2004) },\r | |
728 | { "shift_jisx0213", (void *)2000, NULL, _STATELESS_METHODS(shift_jis_2004) },\r | |
729 | END_CODECS_LIST\r | |
730 | \r | |
731 | I_AM_A_MODULE_FOR(jp)\r |