]>
Commit | Line | Data |
---|---|---|
7eb75bcc DM |
1 | /*\r |
2 | * _codecs_kr.c: Codecs collection for Korean encodings\r | |
3 | *\r | |
4 | * Written by Hye-Shik Chang <perky@FreeBSD.org>\r | |
5 | */\r | |
6 | \r | |
7 | #include "cjkcodecs.h"\r | |
8 | #include "mappings_kr.h"\r | |
9 | \r | |
10 | /*\r | |
11 | * EUC-KR codec\r | |
12 | */\r | |
13 | \r | |
14 | #define EUCKR_JAMO_FIRSTBYTE 0xA4\r | |
15 | #define EUCKR_JAMO_FILLER 0xD4\r | |
16 | \r | |
17 | static const unsigned char u2cgk_choseong[19] = {\r | |
18 | 0xa1, 0xa2, 0xa4, 0xa7, 0xa8, 0xa9, 0xb1, 0xb2,\r | |
19 | 0xb3, 0xb5, 0xb6, 0xb7, 0xb8, 0xb9, 0xba, 0xbb,\r | |
20 | 0xbc, 0xbd, 0xbe\r | |
21 | };\r | |
22 | static const unsigned char u2cgk_jungseong[21] = {\r | |
23 | 0xbf, 0xc0, 0xc1, 0xc2, 0xc3, 0xc4, 0xc5, 0xc6,\r | |
24 | 0xc7, 0xc8, 0xc9, 0xca, 0xcb, 0xcc, 0xcd, 0xce,\r | |
25 | 0xcf, 0xd0, 0xd1, 0xd2, 0xd3\r | |
26 | };\r | |
27 | static const unsigned char u2cgk_jongseong[28] = {\r | |
28 | 0xd4, 0xa1, 0xa2, 0xa3, 0xa4, 0xa5, 0xa6, 0xa7,\r | |
29 | 0xa9, 0xaa, 0xab, 0xac, 0xad, 0xae, 0xaf, 0xb0,\r | |
30 | 0xb1, 0xb2, 0xb4, 0xb5, 0xb6, 0xb7, 0xb8, 0xba,\r | |
31 | 0xbb, 0xbc, 0xbd, 0xbe\r | |
32 | };\r | |
33 | \r | |
34 | ENCODER(euc_kr)\r | |
35 | {\r | |
36 | while (inleft > 0) {\r | |
37 | Py_UNICODE c = IN1;\r | |
38 | DBCHAR code;\r | |
39 | \r | |
40 | if (c < 0x80) {\r | |
41 | WRITE1((unsigned char)c)\r | |
42 | NEXT(1, 1)\r | |
43 | continue;\r | |
44 | }\r | |
45 | UCS4INVALID(c)\r | |
46 | \r | |
47 | REQUIRE_OUTBUF(2)\r | |
48 | TRYMAP_ENC(cp949, code, c);\r | |
49 | else return 1;\r | |
50 | \r | |
51 | if ((code & 0x8000) == 0) {\r | |
52 | /* KS X 1001 coded character */\r | |
53 | OUT1((code >> 8) | 0x80)\r | |
54 | OUT2((code & 0xFF) | 0x80)\r | |
55 | NEXT(1, 2)\r | |
56 | }\r | |
57 | else { /* Mapping is found in CP949 extension,\r | |
58 | * but we encode it in KS X 1001:1998 Annex 3,\r | |
59 | * make-up sequence for EUC-KR. */\r | |
60 | \r | |
61 | REQUIRE_OUTBUF(8)\r | |
62 | \r | |
63 | /* syllable composition precedence */\r | |
64 | OUT1(EUCKR_JAMO_FIRSTBYTE)\r | |
65 | OUT2(EUCKR_JAMO_FILLER)\r | |
66 | \r | |
67 | /* All code points in CP949 extension are in unicode\r | |
68 | * Hangul Syllable area. */\r | |
69 | assert(0xac00 <= c && c <= 0xd7a3);\r | |
70 | c -= 0xac00;\r | |
71 | \r | |
72 | OUT3(EUCKR_JAMO_FIRSTBYTE)\r | |
73 | OUT4(u2cgk_choseong[c / 588])\r | |
74 | NEXT_OUT(4)\r | |
75 | \r | |
76 | OUT1(EUCKR_JAMO_FIRSTBYTE)\r | |
77 | OUT2(u2cgk_jungseong[(c / 28) % 21])\r | |
78 | OUT3(EUCKR_JAMO_FIRSTBYTE)\r | |
79 | OUT4(u2cgk_jongseong[c % 28])\r | |
80 | NEXT(1, 4)\r | |
81 | }\r | |
82 | }\r | |
83 | \r | |
84 | return 0;\r | |
85 | }\r | |
86 | \r | |
87 | #define NONE 127\r | |
88 | \r | |
89 | static const unsigned char cgk2u_choseong[] = { /* [A1, BE] */\r | |
90 | 0, 1, NONE, 2, NONE, NONE, 3, 4,\r | |
91 | 5, NONE, NONE, NONE, NONE, NONE, NONE, NONE,\r | |
92 | 6, 7, 8, NONE, 9, 10, 11, 12,\r | |
93 | 13, 14, 15, 16, 17, 18\r | |
94 | };\r | |
95 | static const unsigned char cgk2u_jongseong[] = { /* [A1, BE] */\r | |
96 | 1, 2, 3, 4, 5, 6, 7, NONE,\r | |
97 | 8, 9, 10, 11, 12, 13, 14, 15,\r | |
98 | 16, 17, NONE, 18, 19, 20, 21, 22,\r | |
99 | NONE, 23, 24, 25, 26, 27\r | |
100 | };\r | |
101 | \r | |
102 | DECODER(euc_kr)\r | |
103 | {\r | |
104 | while (inleft > 0) {\r | |
105 | unsigned char c = IN1;\r | |
106 | \r | |
107 | REQUIRE_OUTBUF(1)\r | |
108 | \r | |
109 | if (c < 0x80) {\r | |
110 | OUT1(c)\r | |
111 | NEXT(1, 1)\r | |
112 | continue;\r | |
113 | }\r | |
114 | \r | |
115 | REQUIRE_INBUF(2)\r | |
116 | \r | |
117 | if (c == EUCKR_JAMO_FIRSTBYTE &&\r | |
118 | IN2 == EUCKR_JAMO_FILLER) {\r | |
119 | /* KS X 1001:1998 Annex 3 make-up sequence */\r | |
120 | DBCHAR cho, jung, jong;\r | |
121 | \r | |
122 | REQUIRE_INBUF(8)\r | |
123 | if ((*inbuf)[2] != EUCKR_JAMO_FIRSTBYTE ||\r | |
124 | (*inbuf)[4] != EUCKR_JAMO_FIRSTBYTE ||\r | |
125 | (*inbuf)[6] != EUCKR_JAMO_FIRSTBYTE)\r | |
126 | return 8;\r | |
127 | \r | |
128 | c = (*inbuf)[3];\r | |
129 | if (0xa1 <= c && c <= 0xbe)\r | |
130 | cho = cgk2u_choseong[c - 0xa1];\r | |
131 | else\r | |
132 | cho = NONE;\r | |
133 | \r | |
134 | c = (*inbuf)[5];\r | |
135 | jung = (0xbf <= c && c <= 0xd3) ? c - 0xbf : NONE;\r | |
136 | \r | |
137 | c = (*inbuf)[7];\r | |
138 | if (c == EUCKR_JAMO_FILLER)\r | |
139 | jong = 0;\r | |
140 | else if (0xa1 <= c && c <= 0xbe)\r | |
141 | jong = cgk2u_jongseong[c - 0xa1];\r | |
142 | else\r | |
143 | jong = NONE;\r | |
144 | \r | |
145 | if (cho == NONE || jung == NONE || jong == NONE)\r | |
146 | return 8;\r | |
147 | \r | |
148 | OUT1(0xac00 + cho*588 + jung*28 + jong);\r | |
149 | NEXT(8, 1)\r | |
150 | }\r | |
151 | else TRYMAP_DEC(ksx1001, **outbuf, c ^ 0x80, IN2 ^ 0x80) {\r | |
152 | NEXT(2, 1)\r | |
153 | }\r | |
154 | else\r | |
155 | return 2;\r | |
156 | }\r | |
157 | \r | |
158 | return 0;\r | |
159 | }\r | |
160 | #undef NONE\r | |
161 | \r | |
162 | \r | |
163 | /*\r | |
164 | * CP949 codec\r | |
165 | */\r | |
166 | \r | |
167 | ENCODER(cp949)\r | |
168 | {\r | |
169 | while (inleft > 0) {\r | |
170 | Py_UNICODE c = IN1;\r | |
171 | DBCHAR code;\r | |
172 | \r | |
173 | if (c < 0x80) {\r | |
174 | WRITE1((unsigned char)c)\r | |
175 | NEXT(1, 1)\r | |
176 | continue;\r | |
177 | }\r | |
178 | UCS4INVALID(c)\r | |
179 | \r | |
180 | REQUIRE_OUTBUF(2)\r | |
181 | TRYMAP_ENC(cp949, code, c);\r | |
182 | else return 1;\r | |
183 | \r | |
184 | OUT1((code >> 8) | 0x80)\r | |
185 | if (code & 0x8000)\r | |
186 | OUT2(code & 0xFF) /* MSB set: CP949 */\r | |
187 | else\r | |
188 | OUT2((code & 0xFF) | 0x80) /* MSB unset: ks x 1001 */\r | |
189 | NEXT(1, 2)\r | |
190 | }\r | |
191 | \r | |
192 | return 0;\r | |
193 | }\r | |
194 | \r | |
195 | DECODER(cp949)\r | |
196 | {\r | |
197 | while (inleft > 0) {\r | |
198 | unsigned char c = IN1;\r | |
199 | \r | |
200 | REQUIRE_OUTBUF(1)\r | |
201 | \r | |
202 | if (c < 0x80) {\r | |
203 | OUT1(c)\r | |
204 | NEXT(1, 1)\r | |
205 | continue;\r | |
206 | }\r | |
207 | \r | |
208 | REQUIRE_INBUF(2)\r | |
209 | TRYMAP_DEC(ksx1001, **outbuf, c ^ 0x80, IN2 ^ 0x80);\r | |
210 | else TRYMAP_DEC(cp949ext, **outbuf, c, IN2);\r | |
211 | else return 2;\r | |
212 | \r | |
213 | NEXT(2, 1)\r | |
214 | }\r | |
215 | \r | |
216 | return 0;\r | |
217 | }\r | |
218 | \r | |
219 | \r | |
220 | /*\r | |
221 | * JOHAB codec\r | |
222 | */\r | |
223 | \r | |
224 | static const unsigned char u2johabidx_choseong[32] = {\r | |
225 | 0x02, 0x03, 0x04, 0x05, 0x06, 0x07,\r | |
226 | 0x08, 0x09, 0x0a, 0x0b, 0x0c, 0x0d, 0x0e, 0x0f,\r | |
227 | 0x10, 0x11, 0x12, 0x13, 0x14,\r | |
228 | };\r | |
229 | static const unsigned char u2johabidx_jungseong[32] = {\r | |
230 | 0x03, 0x04, 0x05, 0x06, 0x07,\r | |
231 | 0x0a, 0x0b, 0x0c, 0x0d, 0x0e, 0x0f,\r | |
232 | 0x12, 0x13, 0x14, 0x15, 0x16, 0x17,\r | |
233 | 0x1a, 0x1b, 0x1c, 0x1d,\r | |
234 | };\r | |
235 | static const unsigned char u2johabidx_jongseong[32] = {\r | |
236 | 0x01, 0x02, 0x03, 0x04, 0x05, 0x06, 0x07,\r | |
237 | 0x08, 0x09, 0x0a, 0x0b, 0x0c, 0x0d, 0x0e, 0x0f,\r | |
238 | 0x10, 0x11, 0x13, 0x14, 0x15, 0x16, 0x17,\r | |
239 | 0x18, 0x19, 0x1a, 0x1b, 0x1c, 0x1d,\r | |
240 | };\r | |
241 | static const DBCHAR u2johabjamo[] = {\r | |
242 | 0x8841, 0x8c41, 0x8444, 0x9041, 0x8446, 0x8447, 0x9441,\r | |
243 | 0x9841, 0x9c41, 0x844a, 0x844b, 0x844c, 0x844d, 0x844e, 0x844f,\r | |
244 | 0x8450, 0xa041, 0xa441, 0xa841, 0x8454, 0xac41, 0xb041, 0xb441,\r | |
245 | 0xb841, 0xbc41, 0xc041, 0xc441, 0xc841, 0xcc41, 0xd041, 0x8461,\r | |
246 | 0x8481, 0x84a1, 0x84c1, 0x84e1, 0x8541, 0x8561, 0x8581, 0x85a1,\r | |
247 | 0x85c1, 0x85e1, 0x8641, 0x8661, 0x8681, 0x86a1, 0x86c1, 0x86e1,\r | |
248 | 0x8741, 0x8761, 0x8781, 0x87a1,\r | |
249 | };\r | |
250 | \r | |
251 | ENCODER(johab)\r | |
252 | {\r | |
253 | while (inleft > 0) {\r | |
254 | Py_UNICODE c = IN1;\r | |
255 | DBCHAR code;\r | |
256 | \r | |
257 | if (c < 0x80) {\r | |
258 | WRITE1((unsigned char)c)\r | |
259 | NEXT(1, 1)\r | |
260 | continue;\r | |
261 | }\r | |
262 | UCS4INVALID(c)\r | |
263 | \r | |
264 | REQUIRE_OUTBUF(2)\r | |
265 | \r | |
266 | if (c >= 0xac00 && c <= 0xd7a3) {\r | |
267 | c -= 0xac00;\r | |
268 | code = 0x8000 |\r | |
269 | (u2johabidx_choseong[c / 588] << 10) |\r | |
270 | (u2johabidx_jungseong[(c / 28) % 21] << 5) |\r | |
271 | u2johabidx_jongseong[c % 28];\r | |
272 | }\r | |
273 | else if (c >= 0x3131 && c <= 0x3163)\r | |
274 | code = u2johabjamo[c - 0x3131];\r | |
275 | else TRYMAP_ENC(cp949, code, c) {\r | |
276 | unsigned char c1, c2, t2;\r | |
277 | unsigned short t1;\r | |
278 | \r | |
279 | assert((code & 0x8000) == 0);\r | |
280 | c1 = code >> 8;\r | |
281 | c2 = code & 0xff;\r | |
282 | if (((c1 >= 0x21 && c1 <= 0x2c) ||\r | |
283 | (c1 >= 0x4a && c1 <= 0x7d)) &&\r | |
284 | (c2 >= 0x21 && c2 <= 0x7e)) {\r | |
285 | t1 = (c1 < 0x4a ? (c1 - 0x21 + 0x1b2) :\r | |
286 | (c1 - 0x21 + 0x197));\r | |
287 | t2 = ((t1 & 1) ? 0x5e : 0) + (c2 - 0x21);\r | |
288 | OUT1(t1 >> 1)\r | |
289 | OUT2(t2 < 0x4e ? t2 + 0x31 : t2 + 0x43)\r | |
290 | NEXT(1, 2)\r | |
291 | continue;\r | |
292 | }\r | |
293 | else\r | |
294 | return 1;\r | |
295 | }\r | |
296 | else\r | |
297 | return 1;\r | |
298 | \r | |
299 | OUT1(code >> 8)\r | |
300 | OUT2(code & 0xff)\r | |
301 | NEXT(1, 2)\r | |
302 | }\r | |
303 | \r | |
304 | return 0;\r | |
305 | }\r | |
306 | \r | |
307 | #define FILL 0xfd\r | |
308 | #define NONE 0xff\r | |
309 | \r | |
310 | static const unsigned char johabidx_choseong[32] = {\r | |
311 | NONE, FILL, 0x00, 0x01, 0x02, 0x03, 0x04, 0x05,\r | |
312 | 0x06, 0x07, 0x08, 0x09, 0x0a, 0x0b, 0x0c, 0x0d,\r | |
313 | 0x0e, 0x0f, 0x10, 0x11, 0x12, NONE, NONE, NONE,\r | |
314 | NONE, NONE, NONE, NONE, NONE, NONE, NONE, NONE,\r | |
315 | };\r | |
316 | static const unsigned char johabidx_jungseong[32] = {\r | |
317 | NONE, NONE, FILL, 0x00, 0x01, 0x02, 0x03, 0x04,\r | |
318 | NONE, NONE, 0x05, 0x06, 0x07, 0x08, 0x09, 0x0a,\r | |
319 | NONE, NONE, 0x0b, 0x0c, 0x0d, 0x0e, 0x0f, 0x10,\r | |
320 | NONE, NONE, 0x11, 0x12, 0x13, 0x14, NONE, NONE,\r | |
321 | };\r | |
322 | static const unsigned char johabidx_jongseong[32] = {\r | |
323 | NONE, FILL, 0x01, 0x02, 0x03, 0x04, 0x05, 0x06,\r | |
324 | 0x07, 0x08, 0x09, 0x0a, 0x0b, 0x0c, 0x0d, 0x0e,\r | |
325 | 0x0f, 0x10, NONE, 0x11, 0x12, 0x13, 0x14, 0x15,\r | |
326 | 0x16, 0x17, 0x18, 0x19, 0x1a, 0x1b, NONE, NONE,\r | |
327 | };\r | |
328 | \r | |
329 | static const unsigned char johabjamo_choseong[32] = {\r | |
330 | NONE, FILL, 0x31, 0x32, 0x34, 0x37, 0x38, 0x39,\r | |
331 | 0x41, 0x42, 0x43, 0x45, 0x46, 0x47, 0x48, 0x49,\r | |
332 | 0x4a, 0x4b, 0x4c, 0x4d, 0x4e, NONE, NONE, NONE,\r | |
333 | NONE, NONE, NONE, NONE, NONE, NONE, NONE, NONE,\r | |
334 | };\r | |
335 | static const unsigned char johabjamo_jungseong[32] = {\r | |
336 | NONE, NONE, FILL, 0x4f, 0x50, 0x51, 0x52, 0x53,\r | |
337 | NONE, NONE, 0x54, 0x55, 0x56, 0x57, 0x58, 0x59,\r | |
338 | NONE, NONE, 0x5a, 0x5b, 0x5c, 0x5d, 0x5e, 0x5f,\r | |
339 | NONE, NONE, 0x60, 0x61, 0x62, 0x63, NONE, NONE,\r | |
340 | };\r | |
341 | static const unsigned char johabjamo_jongseong[32] = {\r | |
342 | NONE, FILL, 0x31, 0x32, 0x33, 0x34, 0x35, 0x36,\r | |
343 | 0x37, 0x39, 0x3a, 0x3b, 0x3c, 0x3d, 0x3e, 0x3f,\r | |
344 | 0x40, 0x41, NONE, 0x42, 0x44, 0x45, 0x46, 0x47,\r | |
345 | 0x48, 0x4a, 0x4b, 0x4c, 0x4d, 0x4e, NONE, NONE,\r | |
346 | };\r | |
347 | \r | |
348 | DECODER(johab)\r | |
349 | {\r | |
350 | while (inleft > 0) {\r | |
351 | unsigned char c = IN1, c2;\r | |
352 | \r | |
353 | REQUIRE_OUTBUF(1)\r | |
354 | \r | |
355 | if (c < 0x80) {\r | |
356 | OUT1(c)\r | |
357 | NEXT(1, 1)\r | |
358 | continue;\r | |
359 | }\r | |
360 | \r | |
361 | REQUIRE_INBUF(2)\r | |
362 | c2 = IN2;\r | |
363 | \r | |
364 | if (c < 0xd8) {\r | |
365 | /* johab hangul */\r | |
366 | unsigned char c_cho, c_jung, c_jong;\r | |
367 | unsigned char i_cho, i_jung, i_jong;\r | |
368 | \r | |
369 | c_cho = (c >> 2) & 0x1f;\r | |
370 | c_jung = ((c << 3) | c2 >> 5) & 0x1f;\r | |
371 | c_jong = c2 & 0x1f;\r | |
372 | \r | |
373 | i_cho = johabidx_choseong[c_cho];\r | |
374 | i_jung = johabidx_jungseong[c_jung];\r | |
375 | i_jong = johabidx_jongseong[c_jong];\r | |
376 | \r | |
377 | if (i_cho == NONE || i_jung == NONE || i_jong == NONE)\r | |
378 | return 2;\r | |
379 | \r | |
380 | /* we don't use U+1100 hangul jamo yet. */\r | |
381 | if (i_cho == FILL) {\r | |
382 | if (i_jung == FILL) {\r | |
383 | if (i_jong == FILL)\r | |
384 | OUT1(0x3000)\r | |
385 | else\r | |
386 | OUT1(0x3100 |\r | |
387 | johabjamo_jongseong[c_jong])\r | |
388 | }\r | |
389 | else {\r | |
390 | if (i_jong == FILL)\r | |
391 | OUT1(0x3100 |\r | |
392 | johabjamo_jungseong[c_jung])\r | |
393 | else\r | |
394 | return 2;\r | |
395 | }\r | |
396 | } else {\r | |
397 | if (i_jung == FILL) {\r | |
398 | if (i_jong == FILL)\r | |
399 | OUT1(0x3100 |\r | |
400 | johabjamo_choseong[c_cho])\r | |
401 | else\r | |
402 | return 2;\r | |
403 | }\r | |
404 | else\r | |
405 | OUT1(0xac00 +\r | |
406 | i_cho * 588 +\r | |
407 | i_jung * 28 +\r | |
408 | (i_jong == FILL ? 0 : i_jong))\r | |
409 | }\r | |
410 | NEXT(2, 1)\r | |
411 | } else {\r | |
412 | /* KS X 1001 except hangul jamos and syllables */\r | |
413 | if (c == 0xdf || c > 0xf9 ||\r | |
414 | c2 < 0x31 || (c2 >= 0x80 && c2 < 0x91) ||\r | |
415 | (c2 & 0x7f) == 0x7f ||\r | |
416 | (c == 0xda && (c2 >= 0xa1 && c2 <= 0xd3)))\r | |
417 | return 2;\r | |
418 | else {\r | |
419 | unsigned char t1, t2;\r | |
420 | \r | |
421 | t1 = (c < 0xe0 ? 2 * (c - 0xd9) :\r | |
422 | 2 * c - 0x197);\r | |
423 | t2 = (c2 < 0x91 ? c2 - 0x31 : c2 - 0x43);\r | |
424 | t1 = t1 + (t2 < 0x5e ? 0 : 1) + 0x21;\r | |
425 | t2 = (t2 < 0x5e ? t2 : t2 - 0x5e) + 0x21;\r | |
426 | \r | |
427 | TRYMAP_DEC(ksx1001, **outbuf, t1, t2);\r | |
428 | else return 2;\r | |
429 | NEXT(2, 1)\r | |
430 | }\r | |
431 | }\r | |
432 | }\r | |
433 | \r | |
434 | return 0;\r | |
435 | }\r | |
436 | #undef NONE\r | |
437 | #undef FILL\r | |
438 | \r | |
439 | \r | |
440 | BEGIN_MAPPINGS_LIST\r | |
441 | MAPPING_DECONLY(ksx1001)\r | |
442 | MAPPING_ENCONLY(cp949)\r | |
443 | MAPPING_DECONLY(cp949ext)\r | |
444 | END_MAPPINGS_LIST\r | |
445 | \r | |
446 | BEGIN_CODECS_LIST\r | |
447 | CODEC_STATELESS(euc_kr)\r | |
448 | CODEC_STATELESS(cp949)\r | |
449 | CODEC_STATELESS(johab)\r | |
450 | END_CODECS_LIST\r | |
451 | \r | |
452 | I_AM_A_MODULE_FOR(kr)\r |