1 /*
2  * "streamable kanji code filter and converter"
3  * Copyright (c) 1998-2002 HappySize, Inc. All rights reserved.
4  *
5  * LICENSE NOTICES
6  *
7  * This file is part of "streamable kanji code filter and converter",
8  * which is distributed under the terms of GNU Lesser General Public
9  * License (version 2) as published by the Free Software Foundation.
10  *
11  * This software is distributed in the hope that it will be useful,
12  * but WITHOUT ANY WARRANTY; without even the implied warranty of
13  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
14  * GNU Lesser General Public License for more details.
15  *
16  * You should have received a copy of the GNU Lesser General Public
17  * License along with "streamable kanji code filter and converter";
18  * if not, write to the Free Software Foundation, Inc., 59 Temple Place,
19  * Suite 330, Boston, MA  02111-1307  USA
20  *
21  * The author of this file:
22  *
23  */
24 /*
25  * The source code included in this files was separated from mbfilter_ja.c
26  * by moriyoshi koizumi <moriyoshi@php.net> on 4 dec 2002.
27  *
28  */
29 
30 #ifdef HAVE_CONFIG_H
31 #include "config.h"
32 #endif
33 
34 #include "mbfilter.h"
35 #include "mbfilter_jis.h"
36 
37 #include "unicode_table_cp932_ext.h"
38 #include "unicode_table_jis.h"
39 
40 static int mbfl_filt_ident_jis(int c, mbfl_identify_filter *filter);
41 static int mbfl_filt_ident_2022jp(int c, mbfl_identify_filter *filter);
42 
43 const mbfl_encoding mbfl_encoding_jis = {
44 	mbfl_no_encoding_jis,
45 	"JIS",
46 	"ISO-2022-JP",
47 	NULL,
48 	NULL,
49 	MBFL_ENCTYPE_MBCS | MBFL_ENCTYPE_SHFTCODE | MBFL_ENCTYPE_GL_UNSAFE,
50 	&vtbl_jis_wchar,
51 	&vtbl_wchar_jis
52 };
53 
54 const mbfl_encoding mbfl_encoding_2022jp = {
55 	mbfl_no_encoding_2022jp,
56 	"ISO-2022-JP",
57 	"ISO-2022-JP",
58 	NULL,
59 	NULL,
60 	MBFL_ENCTYPE_MBCS | MBFL_ENCTYPE_SHFTCODE | MBFL_ENCTYPE_GL_UNSAFE,
61 	&vtbl_2022jp_wchar,
62 	&vtbl_wchar_2022jp
63 };
64 
65 const struct mbfl_identify_vtbl vtbl_identify_jis = {
66 	mbfl_no_encoding_jis,
67 	mbfl_filt_ident_common_ctor,
68 	mbfl_filt_ident_common_dtor,
69 	mbfl_filt_ident_jis
70 };
71 
72 const struct mbfl_identify_vtbl vtbl_identify_2022jp = {
73 	mbfl_no_encoding_2022jp,
74 	mbfl_filt_ident_common_ctor,
75 	mbfl_filt_ident_common_dtor,
76 	mbfl_filt_ident_2022jp
77 };
78 
79 const struct mbfl_convert_vtbl vtbl_jis_wchar = {
80 	mbfl_no_encoding_jis,
81 	mbfl_no_encoding_wchar,
82 	mbfl_filt_conv_common_ctor,
83 	mbfl_filt_conv_common_dtor,
84 	mbfl_filt_conv_jis_wchar,
85 	mbfl_filt_conv_common_flush
86 };
87 
88 const struct mbfl_convert_vtbl vtbl_wchar_jis = {
89 	mbfl_no_encoding_wchar,
90 	mbfl_no_encoding_jis,
91 	mbfl_filt_conv_common_ctor,
92 	mbfl_filt_conv_common_dtor,
93 	mbfl_filt_conv_wchar_jis,
94 	mbfl_filt_conv_any_jis_flush
95 };
96 
97 const struct mbfl_convert_vtbl vtbl_2022jp_wchar = {
98 	mbfl_no_encoding_2022jp,
99 	mbfl_no_encoding_wchar,
100 	mbfl_filt_conv_common_ctor,
101 	mbfl_filt_conv_common_dtor,
102 	mbfl_filt_conv_jis_wchar,
103 	mbfl_filt_conv_common_flush
104 };
105 
106 const struct mbfl_convert_vtbl vtbl_wchar_2022jp = {
107 	mbfl_no_encoding_wchar,
108 	mbfl_no_encoding_2022jp,
109 	mbfl_filt_conv_common_ctor,
110 	mbfl_filt_conv_common_dtor,
111 	mbfl_filt_conv_wchar_2022jp,
112 	mbfl_filt_conv_any_jis_flush
113 };
114 
115 #define CK(statement)	do { if ((statement) < 0) return (-1); } while (0)
116 
117 /*
118  * JIS => wchar
119  */
120 int
mbfl_filt_conv_jis_wchar(int c,mbfl_convert_filter * filter)121 mbfl_filt_conv_jis_wchar(int c, mbfl_convert_filter *filter)
122 {
123 	int c1, s, w;
124 
125 retry:
126 	switch (filter->status & 0xf) {
127 /*	case 0x00:	 ASCII */
128 /*	case 0x10:	 X 0201 latin */
129 /*	case 0x20:	 X 0201 kana */
130 /*	case 0x80:	 X 0208 */
131 /*	case 0x90:	 X 0212 */
132 	case 0:
133 		if (c == 0x1b) {
134 			filter->status += 2;
135 		} else if (c == 0x0e) {		/* "kana in" */
136 			filter->status = 0x20;
137 		} else if (c == 0x0f) {		/* "kana out" */
138 			filter->status = 0;
139 		} else if (filter->status == 0x10 && c == 0x5c) {	/* YEN SIGN */
140 			CK((*filter->output_function)(0xa5, filter->data));
141 		} else if (filter->status == 0x10 && c == 0x7e) {	/* OVER LINE */
142 			CK((*filter->output_function)(0x203e, filter->data));
143 		} else if (filter->status == 0x20 && c > 0x20 && c < 0x60) {		/* kana */
144 			CK((*filter->output_function)(0xff40 + c, filter->data));
145 		} else if ((filter->status == 0x80 || filter->status == 0x90) && c > 0x20 && c < 0x7f) {		/* kanji first char */
146 			filter->cache = c;
147 			filter->status += 1;
148 		} else if (c >= 0 && c < 0x80) {		/* latin, CTLs */
149 			CK((*filter->output_function)(c, filter->data));
150 		} else if (c > 0xa0 && c < 0xe0) {	/* GR kana */
151 			CK((*filter->output_function)(0xfec0 + c, filter->data));
152 		} else {
153 			w = c & MBFL_WCSGROUP_MASK;
154 			w |= MBFL_WCSGROUP_THROUGH;
155 			CK((*filter->output_function)(w, filter->data));
156 		}
157 		break;
158 
159 /*	case 0x81:	 X 0208 second char */
160 /*	case 0x91:	 X 0212 second char */
161 	case 1:
162 		filter->status &= ~0xf;
163 		c1 = filter->cache;
164 		if (c > 0x20 && c < 0x7f) {
165 			s = (c1 - 0x21)*94 + c - 0x21;
166 			if (filter->status == 0x80) {
167 				if (s >= 0 && s < jisx0208_ucs_table_size) {
168 					w = jisx0208_ucs_table[s];
169 				} else {
170 					w = 0;
171 				}
172 				if (w <= 0) {
173 					w = (c1 << 8) | c;
174 					w &= MBFL_WCSPLANE_MASK;
175 					w |= MBFL_WCSPLANE_JIS0208;
176 				}
177 			} else {
178 				if (s >= 0 && s < jisx0212_ucs_table_size) {
179 					w = jisx0212_ucs_table[s];
180 				} else {
181 					w = 0;
182 				}
183 				if (w <= 0) {
184 					w = (c1 << 8) | c;
185 					w &= MBFL_WCSPLANE_MASK;
186 					w |= MBFL_WCSPLANE_JIS0212;
187 				}
188 			}
189 			CK((*filter->output_function)(w, filter->data));
190 		} else if (c == 0x1b) {
191 			filter->status += 2;
192 		} else if ((c >= 0 && c < 0x21) || c == 0x7f) {		/* CTLs */
193 			CK((*filter->output_function)(c, filter->data));
194 		} else {
195 			w = (c1 << 8) | c;
196 			w &= MBFL_WCSGROUP_MASK;
197 			w |= MBFL_WCSGROUP_THROUGH;
198 			CK((*filter->output_function)(w, filter->data));
199 		}
200 		break;
201 
202 	/* ESC */
203 /*	case 0x02:	*/
204 /*	case 0x12:	*/
205 /*	case 0x22:	*/
206 /*	case 0x82:	*/
207 /*	case 0x92:	*/
208 	case 2:
209 		if (c == 0x24) {		/* '$' */
210 			filter->status++;
211 		} else if (c == 0x28) {		/* '(' */
212 			filter->status += 3;
213 		} else {
214 			filter->status &= ~0xf;
215 			CK((*filter->output_function)(0x1b, filter->data));
216 			goto retry;
217 		}
218 		break;
219 
220 	/* ESC $ */
221 /*	case 0x03:	*/
222 /*	case 0x13:	*/
223 /*	case 0x23:	*/
224 /*	case 0x83:	*/
225 /*	case 0x93:	*/
226 	case 3:
227 		if (c == 0x40 || c == 0x42) {	/* '@' or 'B' */
228 			filter->status = 0x80;
229 		} else if (c == 0x28) {			/* '(' */
230 			filter->status++;
231 		} else {
232 			filter->status &= ~0xf;
233 			CK((*filter->output_function)(0x1b, filter->data));
234 			CK((*filter->output_function)(0x24, filter->data));
235 			goto retry;
236 		}
237 		break;
238 
239 	/* ESC $ ( */
240 /*	case 0x04:	*/
241 /*	case 0x14:	*/
242 /*	case 0x24:	*/
243 /*	case 0x84:	*/
244 /*	case 0x94:	*/
245 	case 4:
246 		if (c == 0x40 || c == 0x42) {	/* '@' or 'B' */
247 			filter->status = 0x80;
248 		} else if (c == 0x44) {			/* 'D' */
249 			filter->status = 0x90;
250 		} else {
251 			filter->status &= ~0xf;
252 			CK((*filter->output_function)(0x1b, filter->data));
253 			CK((*filter->output_function)(0x24, filter->data));
254 			CK((*filter->output_function)(0x28, filter->data));
255 			goto retry;
256 		}
257 		break;
258 
259 	/* ESC ( */
260 /*	case 0x05:	*/
261 /*	case 0x15:	*/
262 /*	case 0x25:	*/
263 /*	case 0x85:	*/
264 /*	case 0x95:	*/
265 	case 5:
266 		if (c == 0x42 || c == 0x48) {		/* 'B' or 'H' */
267 			filter->status = 0;
268 		} else if (c == 0x4a) {		/* 'J' */
269 			filter->status = 0x10;
270 		} else if (c == 0x49) {		/* 'I' */
271 			filter->status = 0x20;
272 		} else {
273 			filter->status &= ~0xf;
274 			CK((*filter->output_function)(0x1b, filter->data));
275 			CK((*filter->output_function)(0x28, filter->data));
276 			goto retry;
277 		}
278 		break;
279 
280 	default:
281 		filter->status = 0;
282 		break;
283 	}
284 
285 	return c;
286 }
287 
288 /*
289  * wchar => JIS
290  */
291 int
mbfl_filt_conv_wchar_jis(int c,mbfl_convert_filter * filter)292 mbfl_filt_conv_wchar_jis(int c, mbfl_convert_filter *filter)
293 {
294 	int c1, s;
295 
296 	s = 0;
297 	if (c >= ucs_a1_jis_table_min && c < ucs_a1_jis_table_max) {
298 		s = ucs_a1_jis_table[c - ucs_a1_jis_table_min];
299 	} else if (c >= ucs_a2_jis_table_min && c < ucs_a2_jis_table_max) {
300 		s = ucs_a2_jis_table[c - ucs_a2_jis_table_min];
301 	} else if (c >= ucs_i_jis_table_min && c < ucs_i_jis_table_max) {
302 		s = ucs_i_jis_table[c - ucs_i_jis_table_min];
303 	} else if (c >= ucs_r_jis_table_min && c < ucs_r_jis_table_max) {
304 		s = ucs_r_jis_table[c - ucs_r_jis_table_min];
305 	}
306 	if (s <= 0) {
307 		c1 = c & ~MBFL_WCSPLANE_MASK;
308 		if (c1 == MBFL_WCSPLANE_JIS0208) {
309 			s = c & MBFL_WCSPLANE_MASK;
310 		} else if (c1 == MBFL_WCSPLANE_JIS0212) {
311 			s = c & MBFL_WCSPLANE_MASK;
312 			s |= 0x8080;
313 		} else if (c == 0xa5) {		/* YEN SIGN */
314 			s = 0x1005c;
315 		} else if (c == 0x203e) {	/* OVER LINE */
316 			s = 0x1007e;
317 		} else if (c == 0xff3c) {	/* FULLWIDTH REVERSE SOLIDUS */
318 			s = 0x2140;
319 		} else if (c == 0xff5e) {	/* FULLWIDTH TILDE */
320 			s = 0x2141;
321 		} else if (c == 0x2225) {	/* PARALLEL TO */
322 			s = 0x2142;
323 		} else if (c == 0xff0d) {	/* FULLWIDTH HYPHEN-MINUS */
324 			s = 0x215d;
325 		} else if (c == 0xffe0) {	/* FULLWIDTH CENT SIGN */
326 			s = 0x2171;
327 		} else if (c == 0xffe1) {	/* FULLWIDTH POUND SIGN */
328 			s = 0x2172;
329 		} else if (c == 0xffe2) {	/* FULLWIDTH NOT SIGN */
330 			s = 0x224c;
331 		}
332 		if (c == 0) {
333 			s = 0;
334 		} else if (s <= 0) {
335 			s = -1;
336 		}
337 	}
338 	if (s >= 0) {
339 		if (s < 0x80) { /* ASCII */
340 			if ((filter->status & 0xff00) != 0) {
341 				CK((*filter->output_function)(0x1b, filter->data));		/* ESC */
342 				CK((*filter->output_function)(0x28, filter->data));		/* '(' */
343 				CK((*filter->output_function)(0x42, filter->data));		/* 'B' */
344 			}
345 			filter->status = 0;
346 			CK((*filter->output_function)(s, filter->data));
347 		} else if (s < 0x100) { /* kana */
348 			if ((filter->status & 0xff00) != 0x100) {
349 				CK((*filter->output_function)(0x1b, filter->data));		/* ESC */
350 				CK((*filter->output_function)(0x28, filter->data));		/* '(' */
351 				CK((*filter->output_function)(0x49, filter->data));		/* 'I' */
352 			}
353 			filter->status = 0x100;
354 			CK((*filter->output_function)(s & 0x7f, filter->data));
355 		} else if (s < 0x8080) { /* X 0208 */
356 			if ((filter->status & 0xff00) != 0x200) {
357 				CK((*filter->output_function)(0x1b, filter->data));		/* ESC */
358 				CK((*filter->output_function)(0x24, filter->data));		/* '$' */
359 				CK((*filter->output_function)(0x42, filter->data));		/* 'B' */
360 			}
361 			filter->status = 0x200;
362 			CK((*filter->output_function)((s >> 8) & 0x7f, filter->data));
363 			CK((*filter->output_function)(s & 0x7f, filter->data));
364 		} else if (s < 0x10000) { /* X 0212 */
365 			if ((filter->status & 0xff00) != 0x300) {
366 				CK((*filter->output_function)(0x1b, filter->data));		/* ESC */
367 				CK((*filter->output_function)(0x24, filter->data));		/* '$' */
368 				CK((*filter->output_function)(0x28, filter->data));		/* '(' */
369 				CK((*filter->output_function)(0x44, filter->data));		/* 'D' */
370 			}
371 			filter->status = 0x300;
372 			CK((*filter->output_function)((s >> 8) & 0x7f, filter->data));
373 			CK((*filter->output_function)(s & 0x7f, filter->data));
374 		} else { /* X 0201 latin */
375 			if ((filter->status & 0xff00) != 0x400) {
376 				CK((*filter->output_function)(0x1b, filter->data));		/* ESC */
377 				CK((*filter->output_function)(0x28, filter->data));		/* '(' */
378 				CK((*filter->output_function)(0x4a, filter->data));		/* 'J' */
379 			}
380 			filter->status = 0x400;
381 			CK((*filter->output_function)(s & 0x7f, filter->data));
382 		}
383 	} else {
384 		CK(mbfl_filt_conv_illegal_output(c, filter));
385 	}
386 
387 	return c;
388 }
389 
390 
391 /*
392  * wchar => ISO-2022-JP
393  */
394 int
mbfl_filt_conv_wchar_2022jp(int c,mbfl_convert_filter * filter)395 mbfl_filt_conv_wchar_2022jp(int c, mbfl_convert_filter *filter)
396 {
397 	int s;
398 
399 	s = 0;
400 	if (c >= ucs_a1_jis_table_min && c < ucs_a1_jis_table_max) {
401 		s = ucs_a1_jis_table[c - ucs_a1_jis_table_min];
402 	} else if (c >= ucs_a2_jis_table_min && c < ucs_a2_jis_table_max) {
403 		s = ucs_a2_jis_table[c - ucs_a2_jis_table_min];
404 	} else if (c >= ucs_i_jis_table_min && c < ucs_i_jis_table_max) {
405 		s = ucs_i_jis_table[c - ucs_i_jis_table_min];
406 	} else if (c >= ucs_r_jis_table_min && c < ucs_r_jis_table_max) {
407 		s = ucs_r_jis_table[c - ucs_r_jis_table_min];
408 	}
409 	if (s <= 0) {
410 		if (c == 0xa5) {			/* YEN SIGN */
411 			s = 0x1005c;
412 		} else if (c == 0x203e) {	/* OVER LINE */
413 			s = 0x1007e;
414 		} else if (c == 0xff3c) {	/* FULLWIDTH REVERSE SOLIDUS */
415 			s = 0x2140;
416 		} else if (c == 0xff5e) {	/* FULLWIDTH TILDE */
417 			s = 0x2141;
418 		} else if (c == 0x2225) {	/* PARALLEL TO */
419 			s = 0x2142;
420 		} else if (c == 0xff0d) {	/* FULLWIDTH HYPHEN-MINUS */
421 			s = 0x215d;
422 		} else if (c == 0xffe0) {	/* FULLWIDTH CENT SIGN */
423 			s = 0x2171;
424 		} else if (c == 0xffe1) {	/* FULLWIDTH POUND SIGN */
425 			s = 0x2172;
426 		} else if (c == 0xffe2) {	/* FULLWIDTH NOT SIGN */
427 			s = 0x224c;
428 		}
429 		if (c == 0) {
430 			s = 0;
431 		} else if (s <= 0) {
432 			s = -1;
433 		}
434 	} else if ((s >= 0x80 && s < 0x2121) || (s > 0x8080)) {
435 		s = -1;
436 	}
437 	if (s >= 0) {
438 		if (s < 0x80) { /* ASCII */
439 			if ((filter->status & 0xff00) != 0) {
440 				CK((*filter->output_function)(0x1b, filter->data));		/* ESC */
441 				CK((*filter->output_function)(0x28, filter->data));		/* '(' */
442 				CK((*filter->output_function)(0x42, filter->data));		/* 'B' */
443 			}
444 			filter->status = 0;
445 			CK((*filter->output_function)(s, filter->data));
446 		} else if (s < 0x10000) { /* X 0208 */
447 			if ((filter->status & 0xff00) != 0x200) {
448 				CK((*filter->output_function)(0x1b, filter->data));		/* ESC */
449 				CK((*filter->output_function)(0x24, filter->data));		/* '$' */
450 				CK((*filter->output_function)(0x42, filter->data));		/* 'B' */
451 			}
452 			filter->status = 0x200;
453 			CK((*filter->output_function)((s >> 8) & 0x7f, filter->data));
454 			CK((*filter->output_function)(s & 0x7f, filter->data));
455 		} else { /* X 0201 latin */
456 			if ((filter->status & 0xff00) != 0x400) {
457 				CK((*filter->output_function)(0x1b, filter->data));		/* ESC */
458 				CK((*filter->output_function)(0x28, filter->data));		/* '(' */
459 				CK((*filter->output_function)(0x4a, filter->data));		/* 'J' */
460 			}
461 			filter->status = 0x400;
462 			CK((*filter->output_function)(s & 0x7f, filter->data));
463 		}
464 	} else {
465 		CK(mbfl_filt_conv_illegal_output(c, filter));
466 	}
467 
468 	return c;
469 }
470 
471 int
mbfl_filt_conv_any_jis_flush(mbfl_convert_filter * filter)472 mbfl_filt_conv_any_jis_flush(mbfl_convert_filter *filter)
473 {
474 	/* back to latin */
475 	if ((filter->status & 0xff00) != 0) {
476 		CK((*filter->output_function)(0x1b, filter->data));		/* ESC */
477 		CK((*filter->output_function)(0x28, filter->data));		/* '(' */
478 		CK((*filter->output_function)(0x42, filter->data));		/* 'B' */
479 	}
480 	filter->status &= 0xff;
481 
482 	if (filter->flush_function != NULL) {
483 		return (*filter->flush_function)(filter->data);
484 	}
485 
486 	return 0;
487 }
488 
mbfl_filt_ident_jis(int c,mbfl_identify_filter * filter)489 static int mbfl_filt_ident_jis(int c, mbfl_identify_filter *filter)
490 {
491 retry:
492 	switch (filter->status & 0xf) {
493 /*	case 0x00:	 ASCII */
494 /*	case 0x10:	 X 0201 latin */
495 /*	case 0x20:	 X 0201 kana */
496 /*	case 0x80:	 X 0208 */
497 /*	case 0x90:	 X 0212 */
498 	case 0:
499 		if (c == 0x1b) {
500 			filter->status += 2;
501 		} else if (c == 0x0e) {			/* "kana in" */
502 			filter->status = 0x20;
503 		} else if (c == 0x0f) {			/* "kana out" */
504 			filter->status = 0;
505 		} else if ((filter->status == 0x80 || filter->status == 0x90) && c > 0x20 && c < 0x7f) {		/* kanji first char */
506 			filter->status += 1;
507 		} else if (c >= 0 && c < 0x80) {		/* latin, CTLs */
508 			;
509 		} else {
510 			filter->flag = 1;	/* bad */
511 		}
512 		break;
513 
514 /*	case 0x81:	 X 0208 second char */
515 /*	case 0x91:	 X 0212 second char */
516 	case 1:
517 		filter->status &= ~0xf;
518 		if (c == 0x1b) {
519 			goto retry;
520 		} else if (c < 0x21 || c > 0x7e) {		/* bad */
521 			filter->flag = 1;
522 		}
523 		break;
524 
525 	/* ESC */
526 	case 2:
527 		if (c == 0x24) {		/* '$' */
528 			filter->status++;
529 		} else if (c == 0x28) {		/* '(' */
530 			filter->status += 3;
531 		} else {
532 			filter->flag = 1;	/* bad */
533 			filter->status &= ~0xf;
534 			goto retry;
535 		}
536 		break;
537 
538 	/* ESC $ */
539 	case 3:
540 		if (c == 0x40 || c == 0x42) {		/* '@' or 'B' */
541 			filter->status = 0x80;
542 		} else if (c == 0x28) {		/* '(' */
543 			filter->status++;
544 		} else {
545 			filter->flag = 1;	/* bad */
546 			filter->status &= ~0xf;
547 			goto retry;
548 		}
549 		break;
550 
551 	/* ESC $ ( */
552 	case 4:
553 		if (c == 0x40 || c == 0x42) {		/* '@' or 'B' */
554 			filter->status = 0x80;
555 		} else if (c == 0x44) {		/* 'D' */
556 			filter->status = 0x90;
557 		} else {
558 			filter->flag = 1;	/* bad */
559 			filter->status &= ~0xf;
560 			goto retry;
561 		}
562 		break;
563 
564 	/* ESC ( */
565 	case 5:
566 		if (c == 0x42 || c == 0x48) {		/* 'B' or 'H' */
567 			filter->status = 0;
568 		} else if (c == 0x4a) {		/* 'J' */
569 			filter->status = 0x10;
570 		} else if (c == 0x49) {		/* 'I' */
571 			filter->status = 0x20;
572 		} else {
573 			filter->flag = 1;	/* bad */
574 			filter->status &= ~0xf;
575 			goto retry;
576 		}
577 		break;
578 
579 	default:
580 		filter->status = 0;
581 		break;
582 	}
583 
584 	return c;
585 }
586 
mbfl_filt_ident_2022jp(int c,mbfl_identify_filter * filter)587 static int mbfl_filt_ident_2022jp(int c, mbfl_identify_filter *filter)
588 {
589 retry:
590 	switch (filter->status & 0xf) {
591 /*	case 0x00:	 ASCII */
592 /*	case 0x10:	 X 0201 latin */
593 /*	case 0x80:	 X 0208 */
594 	case 0:
595 		if (c == 0x1b) {
596 			filter->status += 2;
597 		} else if (filter->status == 0x80 && c > 0x20 && c < 0x7f) {		/* kanji first char */
598 			filter->status += 1;
599 		} else if (c >= 0 && c < 0x80) {		/* latin, CTLs */
600 			;
601 		} else {
602 			filter->flag = 1;	/* bad */
603 		}
604 		break;
605 
606 /*	case 0x81:	 X 0208 second char */
607 	case 1:
608 		if (c == 0x1b) {
609 			filter->status++;
610 		} else {
611 			filter->status &= ~0xf;
612 			if (c < 0x21 || c > 0x7e) {		/* bad */
613 				filter->flag = 1;
614 			}
615 		}
616 		break;
617 
618 	/* ESC */
619 	case 2:
620 		if (c == 0x24) {		/* '$' */
621 			filter->status++;
622 		} else if (c == 0x28) {		/* '(' */
623 			filter->status += 3;
624 		} else {
625 			filter->flag = 1;	/* bad */
626 			filter->status &= ~0xf;
627 			goto retry;
628 		}
629 		break;
630 
631 	/* ESC $ */
632 	case 3:
633 		if (c == 0x40 || c == 0x42) {		/* '@' or 'B' */
634 			filter->status = 0x80;
635 		} else {
636 			filter->flag = 1;	/* bad */
637 			filter->status &= ~0xf;
638 			goto retry;
639 		}
640 		break;
641 
642 	/* ESC ( */
643 	case 5:
644 		if (c == 0x42) {		/* 'B' */
645 			filter->status = 0;
646 		} else if (c == 0x4a) {		/* 'J' */
647 			filter->status = 0x10;
648 		} else {
649 			filter->flag = 1;	/* bad */
650 			filter->status &= ~0xf;
651 			goto retry;
652 		}
653 		break;
654 
655 	default:
656 		filter->status = 0;
657 		break;
658 	}
659 
660 	return c;
661 }
662