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 };
51 
52 const mbfl_encoding mbfl_encoding_2022jp = {
53 	mbfl_no_encoding_2022jp,
54 	"ISO-2022-JP",
55 	"ISO-2022-JP",
56 	NULL,
57 	NULL,
58 	MBFL_ENCTYPE_MBCS | MBFL_ENCTYPE_SHFTCODE | MBFL_ENCTYPE_GL_UNSAFE
59 };
60 
61 const struct mbfl_identify_vtbl vtbl_identify_jis = {
62 	mbfl_no_encoding_jis,
63 	mbfl_filt_ident_common_ctor,
64 	mbfl_filt_ident_common_dtor,
65 	mbfl_filt_ident_jis
66 };
67 
68 const struct mbfl_identify_vtbl vtbl_identify_2022jp = {
69 	mbfl_no_encoding_2022jp,
70 	mbfl_filt_ident_common_ctor,
71 	mbfl_filt_ident_common_dtor,
72 	mbfl_filt_ident_2022jp
73 };
74 
75 const struct mbfl_convert_vtbl vtbl_jis_wchar = {
76 	mbfl_no_encoding_jis,
77 	mbfl_no_encoding_wchar,
78 	mbfl_filt_conv_common_ctor,
79 	mbfl_filt_conv_common_dtor,
80 	mbfl_filt_conv_jis_wchar,
81 	mbfl_filt_conv_common_flush
82 };
83 
84 const struct mbfl_convert_vtbl vtbl_wchar_jis = {
85 	mbfl_no_encoding_wchar,
86 	mbfl_no_encoding_jis,
87 	mbfl_filt_conv_common_ctor,
88 	mbfl_filt_conv_common_dtor,
89 	mbfl_filt_conv_wchar_jis,
90 	mbfl_filt_conv_any_jis_flush
91 };
92 
93 const struct mbfl_convert_vtbl vtbl_2022jp_wchar = {
94 	mbfl_no_encoding_2022jp,
95 	mbfl_no_encoding_wchar,
96 	mbfl_filt_conv_common_ctor,
97 	mbfl_filt_conv_common_dtor,
98 	mbfl_filt_conv_jis_wchar,
99 	mbfl_filt_conv_common_flush
100 };
101 
102 const struct mbfl_convert_vtbl vtbl_wchar_2022jp = {
103 	mbfl_no_encoding_wchar,
104 	mbfl_no_encoding_2022jp,
105 	mbfl_filt_conv_common_ctor,
106 	mbfl_filt_conv_common_dtor,
107 	mbfl_filt_conv_wchar_2022jp,
108 	mbfl_filt_conv_any_jis_flush
109 };
110 
111 #define CK(statement)	do { if ((statement) < 0) return (-1); } while (0)
112 
113 /*
114  * JIS => wchar
115  */
116 int
mbfl_filt_conv_jis_wchar(int c,mbfl_convert_filter * filter)117 mbfl_filt_conv_jis_wchar(int c, mbfl_convert_filter *filter)
118 {
119 	int c1, s, w;
120 
121 retry:
122 	switch (filter->status & 0xf) {
123 /*	case 0x00:	 ASCII */
124 /*	case 0x10:	 X 0201 latin */
125 /*	case 0x20:	 X 0201 kana */
126 /*	case 0x80:	 X 0208 */
127 /*	case 0x90:	 X 0212 */
128 	case 0:
129 		if (c == 0x1b) {
130 			filter->status += 2;
131 		} else if (c == 0x0e) {		/* "kana in" */
132 			filter->status = 0x20;
133 		} else if (c == 0x0f) {		/* "kana out" */
134 			filter->status = 0;
135 		} else if (filter->status == 0x10 && c == 0x5c) {	/* YEN SIGN */
136 			CK((*filter->output_function)(0xa5, filter->data));
137 		} else if (filter->status == 0x10 && c == 0x7e) {	/* OVER LINE */
138 			CK((*filter->output_function)(0x203e, filter->data));
139 		} else if (filter->status == 0x20 && c > 0x20 && c < 0x60) {		/* kana */
140 			CK((*filter->output_function)(0xff40 + c, filter->data));
141 		} else if ((filter->status == 0x80 || filter->status == 0x90) && c > 0x20 && c < 0x7f) {		/* kanji first char */
142 			filter->cache = c;
143 			filter->status += 1;
144 		} else if (c >= 0 && c < 0x80) {		/* latin, CTLs */
145 			CK((*filter->output_function)(c, filter->data));
146 		} else if (c > 0xa0 && c < 0xe0) {	/* GR kana */
147 			CK((*filter->output_function)(0xfec0 + c, filter->data));
148 		} else {
149 			w = c & MBFL_WCSGROUP_MASK;
150 			w |= MBFL_WCSGROUP_THROUGH;
151 			CK((*filter->output_function)(w, filter->data));
152 		}
153 		break;
154 
155 /*	case 0x81:	 X 0208 second char */
156 /*	case 0x91:	 X 0212 second char */
157 	case 1:
158 		filter->status &= ~0xf;
159 		c1 = filter->cache;
160 		if (c > 0x20 && c < 0x7f) {
161 			s = (c1 - 0x21)*94 + c - 0x21;
162 			if (filter->status == 0x80) {
163 				if (s >= 0 && s < jisx0208_ucs_table_size) {
164 					w = jisx0208_ucs_table[s];
165 				} else {
166 					w = 0;
167 				}
168 				if (w <= 0) {
169 					w = (c1 << 8) | c;
170 					w &= MBFL_WCSPLANE_MASK;
171 					w |= MBFL_WCSPLANE_JIS0208;
172 				}
173 			} else {
174 				if (s >= 0 && s < jisx0212_ucs_table_size) {
175 					w = jisx0212_ucs_table[s];
176 				} else {
177 					w = 0;
178 				}
179 				if (w <= 0) {
180 					w = (c1 << 8) | c;
181 					w &= MBFL_WCSPLANE_MASK;
182 					w |= MBFL_WCSPLANE_JIS0212;
183 				}
184 			}
185 			CK((*filter->output_function)(w, filter->data));
186 		} else if (c == 0x1b) {
187 			filter->status += 2;
188 		} else if ((c >= 0 && c < 0x21) || c == 0x7f) {		/* CTLs */
189 			CK((*filter->output_function)(c, filter->data));
190 		} else {
191 			w = (c1 << 8) | c;
192 			w &= MBFL_WCSGROUP_MASK;
193 			w |= MBFL_WCSGROUP_THROUGH;
194 			CK((*filter->output_function)(w, filter->data));
195 		}
196 		break;
197 
198 	/* ESC */
199 /*	case 0x02:	*/
200 /*	case 0x12:	*/
201 /*	case 0x22:	*/
202 /*	case 0x82:	*/
203 /*	case 0x92:	*/
204 	case 2:
205 		if (c == 0x24) {		/* '$' */
206 			filter->status++;
207 		} else if (c == 0x28) {		/* '(' */
208 			filter->status += 3;
209 		} else {
210 			filter->status &= ~0xf;
211 			CK((*filter->output_function)(0x1b, filter->data));
212 			goto retry;
213 		}
214 		break;
215 
216 	/* ESC $ */
217 /*	case 0x03:	*/
218 /*	case 0x13:	*/
219 /*	case 0x23:	*/
220 /*	case 0x83:	*/
221 /*	case 0x93:	*/
222 	case 3:
223 		if (c == 0x40 || c == 0x42) {	/* '@' or 'B' */
224 			filter->status = 0x80;
225 		} else if (c == 0x28) {			/* '(' */
226 			filter->status++;
227 		} else {
228 			filter->status &= ~0xf;
229 			CK((*filter->output_function)(0x1b, filter->data));
230 			CK((*filter->output_function)(0x24, filter->data));
231 			goto retry;
232 		}
233 		break;
234 
235 	/* ESC $ ( */
236 /*	case 0x04:	*/
237 /*	case 0x14:	*/
238 /*	case 0x24:	*/
239 /*	case 0x84:	*/
240 /*	case 0x94:	*/
241 	case 4:
242 		if (c == 0x40 || c == 0x42) {	/* '@' or 'B' */
243 			filter->status = 0x80;
244 		} else if (c == 0x44) {			/* 'D' */
245 			filter->status = 0x90;
246 		} else {
247 			filter->status &= ~0xf;
248 			CK((*filter->output_function)(0x1b, filter->data));
249 			CK((*filter->output_function)(0x24, filter->data));
250 			CK((*filter->output_function)(0x28, filter->data));
251 			goto retry;
252 		}
253 		break;
254 
255 	/* ESC ( */
256 /*	case 0x05:	*/
257 /*	case 0x15:	*/
258 /*	case 0x25:	*/
259 /*	case 0x85:	*/
260 /*	case 0x95:	*/
261 	case 5:
262 		if (c == 0x42 || c == 0x48) {		/* 'B' or 'H' */
263 			filter->status = 0;
264 		} else if (c == 0x4a) {		/* 'J' */
265 			filter->status = 0x10;
266 		} else if (c == 0x49) {		/* 'I' */
267 			filter->status = 0x20;
268 		} else {
269 			filter->status &= ~0xf;
270 			CK((*filter->output_function)(0x1b, filter->data));
271 			CK((*filter->output_function)(0x28, filter->data));
272 			goto retry;
273 		}
274 		break;
275 
276 	default:
277 		filter->status = 0;
278 		break;
279 	}
280 
281 	return c;
282 }
283 
284 /*
285  * wchar => JIS
286  */
287 int
mbfl_filt_conv_wchar_jis(int c,mbfl_convert_filter * filter)288 mbfl_filt_conv_wchar_jis(int c, mbfl_convert_filter *filter)
289 {
290 	int c1, s;
291 
292 	s = 0;
293 	if (c >= ucs_a1_jis_table_min && c < ucs_a1_jis_table_max) {
294 		s = ucs_a1_jis_table[c - ucs_a1_jis_table_min];
295 	} else if (c >= ucs_a2_jis_table_min && c < ucs_a2_jis_table_max) {
296 		s = ucs_a2_jis_table[c - ucs_a2_jis_table_min];
297 	} else if (c >= ucs_i_jis_table_min && c < ucs_i_jis_table_max) {
298 		s = ucs_i_jis_table[c - ucs_i_jis_table_min];
299 	} else if (c >= ucs_r_jis_table_min && c < ucs_r_jis_table_max) {
300 		s = ucs_r_jis_table[c - ucs_r_jis_table_min];
301 	}
302 	if (s <= 0) {
303 		c1 = c & ~MBFL_WCSPLANE_MASK;
304 		if (c1 == MBFL_WCSPLANE_JIS0208) {
305 			s = c & MBFL_WCSPLANE_MASK;
306 		} else if (c1 == MBFL_WCSPLANE_JIS0212) {
307 			s = c & MBFL_WCSPLANE_MASK;
308 			s |= 0x8080;
309 		} else if (c == 0xa5) {		/* YEN SIGN */
310 			s = 0x1005c;
311 		} else if (c == 0x203e) {	/* OVER LINE */
312 			s = 0x1007e;
313 		} else if (c == 0xff3c) {	/* FULLWIDTH REVERSE SOLIDUS */
314 			s = 0x2140;
315 		} else if (c == 0xff5e) {	/* FULLWIDTH TILDE */
316 			s = 0x2141;
317 		} else if (c == 0x2225) {	/* PARALLEL TO */
318 			s = 0x2142;
319 		} else if (c == 0xff0d) {	/* FULLWIDTH HYPHEN-MINUS */
320 			s = 0x215d;
321 		} else if (c == 0xffe0) {	/* FULLWIDTH CENT SIGN */
322 			s = 0x2171;
323 		} else if (c == 0xffe1) {	/* FULLWIDTH POUND SIGN */
324 			s = 0x2172;
325 		} else if (c == 0xffe2) {	/* FULLWIDTH NOT SIGN */
326 			s = 0x224c;
327 		}
328 		if (c == 0) {
329 			s = 0;
330 		} else if (s <= 0) {
331 			s = -1;
332 		}
333 	}
334 	if (s >= 0) {
335 		if (s < 0x80) { /* ASCII */
336 			if ((filter->status & 0xff00) != 0) {
337 				CK((*filter->output_function)(0x1b, filter->data));		/* ESC */
338 				CK((*filter->output_function)(0x28, filter->data));		/* '(' */
339 				CK((*filter->output_function)(0x42, filter->data));		/* 'B' */
340 			}
341 			filter->status = 0;
342 			CK((*filter->output_function)(s, filter->data));
343 		} else if (s < 0x100) { /* kana */
344 			if ((filter->status & 0xff00) != 0x100) {
345 				CK((*filter->output_function)(0x1b, filter->data));		/* ESC */
346 				CK((*filter->output_function)(0x28, filter->data));		/* '(' */
347 				CK((*filter->output_function)(0x49, filter->data));		/* 'I' */
348 			}
349 			filter->status = 0x100;
350 			CK((*filter->output_function)(s & 0x7f, filter->data));
351 		} else if (s < 0x8080) { /* X 0208 */
352 			if ((filter->status & 0xff00) != 0x200) {
353 				CK((*filter->output_function)(0x1b, filter->data));		/* ESC */
354 				CK((*filter->output_function)(0x24, filter->data));		/* '$' */
355 				CK((*filter->output_function)(0x42, filter->data));		/* 'B' */
356 			}
357 			filter->status = 0x200;
358 			CK((*filter->output_function)((s >> 8) & 0x7f, filter->data));
359 			CK((*filter->output_function)(s & 0x7f, filter->data));
360 		} else if (s < 0x10000) { /* X 0212 */
361 			if ((filter->status & 0xff00) != 0x300) {
362 				CK((*filter->output_function)(0x1b, filter->data));		/* ESC */
363 				CK((*filter->output_function)(0x24, filter->data));		/* '$' */
364 				CK((*filter->output_function)(0x28, filter->data));		/* '(' */
365 				CK((*filter->output_function)(0x44, filter->data));		/* 'D' */
366 			}
367 			filter->status = 0x300;
368 			CK((*filter->output_function)((s >> 8) & 0x7f, filter->data));
369 			CK((*filter->output_function)(s & 0x7f, filter->data));
370 		} else { /* X 0201 latin */
371 			if ((filter->status & 0xff00) != 0x400) {
372 				CK((*filter->output_function)(0x1b, filter->data));		/* ESC */
373 				CK((*filter->output_function)(0x28, filter->data));		/* '(' */
374 				CK((*filter->output_function)(0x4a, filter->data));		/* 'J' */
375 			}
376 			filter->status = 0x400;
377 			CK((*filter->output_function)(s & 0x7f, filter->data));
378 		}
379 	} else {
380 		if (filter->illegal_mode != MBFL_OUTPUTFILTER_ILLEGAL_MODE_NONE) {
381 			CK(mbfl_filt_conv_illegal_output(c, filter));
382 		}
383 	}
384 
385 	return c;
386 }
387 
388 
389 /*
390  * wchar => ISO-2022-JP
391  */
392 int
mbfl_filt_conv_wchar_2022jp(int c,mbfl_convert_filter * filter)393 mbfl_filt_conv_wchar_2022jp(int c, mbfl_convert_filter *filter)
394 {
395 	int s;
396 
397 	s = 0;
398 	if (c >= ucs_a1_jis_table_min && c < ucs_a1_jis_table_max) {
399 		s = ucs_a1_jis_table[c - ucs_a1_jis_table_min];
400 	} else if (c >= ucs_a2_jis_table_min && c < ucs_a2_jis_table_max) {
401 		s = ucs_a2_jis_table[c - ucs_a2_jis_table_min];
402 	} else if (c >= ucs_i_jis_table_min && c < ucs_i_jis_table_max) {
403 		s = ucs_i_jis_table[c - ucs_i_jis_table_min];
404 	} else if (c >= ucs_r_jis_table_min && c < ucs_r_jis_table_max) {
405 		s = ucs_r_jis_table[c - ucs_r_jis_table_min];
406 	}
407 	if (s <= 0) {
408 		if (c == 0xa5) {			/* YEN SIGN */
409 			s = 0x1005c;
410 		} else if (c == 0x203e) {	/* OVER LINE */
411 			s = 0x1007e;
412 		} else if (c == 0xff3c) {	/* FULLWIDTH REVERSE SOLIDUS */
413 			s = 0x2140;
414 		} else if (c == 0xff5e) {	/* FULLWIDTH TILDE */
415 			s = 0x2141;
416 		} else if (c == 0x2225) {	/* PARALLEL TO */
417 			s = 0x2142;
418 		} else if (c == 0xff0d) {	/* FULLWIDTH HYPHEN-MINUS */
419 			s = 0x215d;
420 		} else if (c == 0xffe0) {	/* FULLWIDTH CENT SIGN */
421 			s = 0x2171;
422 		} else if (c == 0xffe1) {	/* FULLWIDTH POUND SIGN */
423 			s = 0x2172;
424 		} else if (c == 0xffe2) {	/* FULLWIDTH NOT SIGN */
425 			s = 0x224c;
426 		}
427 		if (c == 0) {
428 			s = 0;
429 		} else if (s <= 0) {
430 			s = -1;
431 		}
432 	} else if ((s >= 0x80 && s < 0x2121) || (s > 0x8080)) {
433 		s = -1;
434 	}
435 	if (s >= 0) {
436 		if (s < 0x80) { /* ASCII */
437 			if ((filter->status & 0xff00) != 0) {
438 				CK((*filter->output_function)(0x1b, filter->data));		/* ESC */
439 				CK((*filter->output_function)(0x28, filter->data));		/* '(' */
440 				CK((*filter->output_function)(0x42, filter->data));		/* 'B' */
441 			}
442 			filter->status = 0;
443 			CK((*filter->output_function)(s, filter->data));
444 		} else if (s < 0x10000) { /* X 0208 */
445 			if ((filter->status & 0xff00) != 0x200) {
446 				CK((*filter->output_function)(0x1b, filter->data));		/* ESC */
447 				CK((*filter->output_function)(0x24, filter->data));		/* '$' */
448 				CK((*filter->output_function)(0x42, filter->data));		/* 'B' */
449 			}
450 			filter->status = 0x200;
451 			CK((*filter->output_function)((s >> 8) & 0x7f, filter->data));
452 			CK((*filter->output_function)(s & 0x7f, filter->data));
453 		} else { /* X 0201 latin */
454 			if ((filter->status & 0xff00) != 0x400) {
455 				CK((*filter->output_function)(0x1b, filter->data));		/* ESC */
456 				CK((*filter->output_function)(0x28, filter->data));		/* '(' */
457 				CK((*filter->output_function)(0x4a, filter->data));		/* 'J' */
458 			}
459 			filter->status = 0x400;
460 			CK((*filter->output_function)(s & 0x7f, filter->data));
461 		}
462 	} else {
463 		if (filter->illegal_mode != MBFL_OUTPUTFILTER_ILLEGAL_MODE_NONE) {
464 			CK(mbfl_filt_conv_illegal_output(c, filter));
465 		}
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