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: Moriyoshi Koizumi <koizumi@gree.co.jp>
22 *
23 */
24
25 #include "mbfilter.h"
26 #include "mbfilter_cp5022x.h"
27 #include "mbfilter_jis.h"
28 #include "mbfilter_tl_jisx0201_jisx0208.h"
29
30 #include "unicode_table_cp932_ext.h"
31 #include "unicode_table_jis.h"
32 #include "cp932_table.h"
33
34 static int mbfl_filt_conv_cp5022x_wchar_flush(mbfl_convert_filter *filter);
35 static int mbfl_filt_conv_wchar_cp50220_flush(mbfl_convert_filter *filter);
36 static int mbfl_filt_conv_wchar_cp50220(int c, mbfl_convert_filter *filter);
37
38 /* Previously, a dubious 'encoding' called 'cp50220raw' was supported
39 * This was just CP50220, but the implementation was less strict regarding
40 * invalid characters; it would silently pass some through
41 * This 'encoding' only existed in mbstring. In case some poor, lost soul is
42 * still using it, retain minimal support by aliasing it to CP50220
43 *
44 * Further, mbstring also had a made-up encoding called "JIS-ms"
45 * This was the same as CP5022{0,1,2}, but without their special ways of
46 * handling conversion of Unicode half-width katakana */
47 static const char *cp50220_aliases[] = {"cp50220raw", "cp50220-raw", "JIS-ms", NULL};
48
49 const mbfl_encoding mbfl_encoding_cp50220 = {
50 mbfl_no_encoding_cp50220,
51 "CP50220",
52 "ISO-2022-JP",
53 cp50220_aliases,
54 NULL,
55 MBFL_ENCTYPE_GL_UNSAFE,
56 &vtbl_cp50220_wchar,
57 &vtbl_wchar_cp50220,
58 NULL
59 };
60
61 const mbfl_encoding mbfl_encoding_cp50221 = {
62 mbfl_no_encoding_cp50221,
63 "CP50221",
64 "ISO-2022-JP",
65 NULL,
66 NULL,
67 MBFL_ENCTYPE_GL_UNSAFE,
68 &vtbl_cp50221_wchar,
69 &vtbl_wchar_cp50221,
70 NULL
71 };
72
73 const mbfl_encoding mbfl_encoding_cp50222 = {
74 mbfl_no_encoding_cp50222,
75 "CP50222",
76 "ISO-2022-JP",
77 NULL,
78 NULL,
79 MBFL_ENCTYPE_GL_UNSAFE,
80 &vtbl_cp50222_wchar,
81 &vtbl_wchar_cp50222,
82 NULL
83 };
84
85 const struct mbfl_convert_vtbl vtbl_cp50220_wchar = {
86 mbfl_no_encoding_cp50220,
87 mbfl_no_encoding_wchar,
88 mbfl_filt_conv_common_ctor,
89 NULL,
90 mbfl_filt_conv_cp5022x_wchar,
91 mbfl_filt_conv_cp5022x_wchar_flush,
92 NULL,
93 };
94
95 const struct mbfl_convert_vtbl vtbl_wchar_cp50220 = {
96 mbfl_no_encoding_wchar,
97 mbfl_no_encoding_cp50220,
98 mbfl_filt_conv_common_ctor,
99 NULL,
100 mbfl_filt_conv_wchar_cp50220,
101 mbfl_filt_conv_wchar_cp50220_flush,
102 NULL,
103 };
104
105 const struct mbfl_convert_vtbl vtbl_cp50221_wchar = {
106 mbfl_no_encoding_cp50221,
107 mbfl_no_encoding_wchar,
108 mbfl_filt_conv_common_ctor,
109 NULL,
110 mbfl_filt_conv_cp5022x_wchar,
111 mbfl_filt_conv_cp5022x_wchar_flush,
112 NULL,
113 };
114
115 const struct mbfl_convert_vtbl vtbl_wchar_cp50221 = {
116 mbfl_no_encoding_wchar,
117 mbfl_no_encoding_cp50221,
118 mbfl_filt_conv_common_ctor,
119 NULL,
120 mbfl_filt_conv_wchar_cp50221,
121 mbfl_filt_conv_any_jis_flush,
122 NULL,
123 };
124
125 const struct mbfl_convert_vtbl vtbl_cp50222_wchar = {
126 mbfl_no_encoding_cp50222,
127 mbfl_no_encoding_wchar,
128 mbfl_filt_conv_common_ctor,
129 NULL,
130 mbfl_filt_conv_cp5022x_wchar,
131 mbfl_filt_conv_cp5022x_wchar_flush,
132 NULL,
133 };
134
135 const struct mbfl_convert_vtbl vtbl_wchar_cp50222 = {
136 mbfl_no_encoding_wchar,
137 mbfl_no_encoding_cp50222,
138 mbfl_filt_conv_common_ctor,
139 NULL,
140 mbfl_filt_conv_wchar_cp50222,
141 mbfl_filt_conv_wchar_cp50222_flush,
142 NULL,
143 };
144
145 #define CK(statement) do { if ((statement) < 0) return (-1); } while (0)
146
mbfl_filt_conv_cp5022x_wchar(int c,mbfl_convert_filter * filter)147 int mbfl_filt_conv_cp5022x_wchar(int c, mbfl_convert_filter *filter)
148 {
149 int c1, s, w;
150
151 retry:
152 switch (filter->status & 0xf) {
153 /* case 0x00: ASCII */
154 /* case 0x10: X 0201 latin */
155 /* case 0x20: X 0201 kana */
156 /* case 0x80: X 0208 */
157 /* case 0x90: X 0212 */
158 case 0:
159 if (c == 0x1b) {
160 filter->status += 2;
161 } else if (c == 0x0e) { /* "kana in" */
162 filter->status = 0x20;
163 } else if (c == 0x0f) { /* "kana out" */
164 filter->status = 0;
165 } else if (filter->status == 0x10 && c == 0x5c) { /* YEN SIGN */
166 CK((*filter->output_function)(0xa5, filter->data));
167 } else if (filter->status == 0x10 && c == 0x7e) { /* OVER LINE */
168 CK((*filter->output_function)(0x203e, filter->data));
169 } else if (filter->status == 0x20 && c > 0x20 && c < 0x60) { /* kana */
170 CK((*filter->output_function)(0xff40 + c, filter->data));
171 } else if ((filter->status == 0x80 || filter->status == 0x90) && c > 0x20 && c <= 0x97) { /* kanji first char */
172 filter->cache = c;
173 filter->status += 1;
174 } else if (c >= 0 && c < 0x80) { /* latin, CTLs */
175 CK((*filter->output_function)(c, filter->data));
176 } else if (c > 0xa0 && c < 0xe0) { /* GR kana */
177 CK((*filter->output_function)(0xfec0 + c, filter->data));
178 } else {
179 CK((*filter->output_function)(MBFL_BAD_INPUT, filter->data));
180 }
181 break;
182
183 /* case 0x81: X 0208 second char */
184 /* case 0x91: X 0212 second char */
185 case 1:
186 filter->status &= ~0xf;
187 c1 = filter->cache;
188 if (c > 0x20 && c < 0x7f) {
189 s = (c1 - 0x21)*94 + c - 0x21;
190 if (filter->status == 0x80) {
191 if (s >= cp932ext1_ucs_table_min && s < cp932ext1_ucs_table_max) {
192 w = cp932ext1_ucs_table[s - cp932ext1_ucs_table_min];
193 } else if (s >= 0 && s < jisx0208_ucs_table_size) {
194 w = jisx0208_ucs_table[s];
195 } else if (s >= cp932ext2_ucs_table_min && s < cp932ext2_ucs_table_max) {
196 w = cp932ext2_ucs_table[s - cp932ext2_ucs_table_min];
197 } else if (s >= cp932ext3_ucs_table_min && s < cp932ext3_ucs_table_max) {
198 w = cp932ext3_ucs_table[s - cp932ext3_ucs_table_min];
199 } else if (s >= 94 * 94 && s < 114 * 94) {
200 /* user-defined => PUA (Microsoft extended) */
201 w = s - 94*94 + 0xe000;
202 } else {
203 w = 0;
204 }
205
206 if (w <= 0) {
207 w = MBFL_BAD_INPUT;
208 }
209 } else {
210 if (s >= 0 && s < jisx0212_ucs_table_size) {
211 w = jisx0212_ucs_table[s];
212 } else {
213 w = 0;
214 }
215
216 if (w <= 0) {
217 w = MBFL_BAD_INPUT;
218 }
219 }
220 CK((*filter->output_function)(w, filter->data));
221 } else {
222 CK((*filter->output_function)(MBFL_BAD_INPUT, filter->data));
223 }
224 break;
225
226 /* ESC */
227 /* case 0x02: */
228 /* case 0x12: */
229 /* case 0x22: */
230 /* case 0x82: */
231 /* case 0x92: */
232 case 2:
233 if (c == 0x24) { /* '$' */
234 filter->status++;
235 } else if (c == 0x28) { /* '(' */
236 filter->status += 3;
237 } else {
238 filter->status &= ~0xf;
239 CK((*filter->output_function)(MBFL_BAD_INPUT, filter->data));
240 goto retry;
241 }
242 break;
243
244 /* ESC $ */
245 /* case 0x03: */
246 /* case 0x13: */
247 /* case 0x23: */
248 /* case 0x83: */
249 /* case 0x93: */
250 case 3:
251 if (c == 0x40 || c == 0x42) { /* '@' or 'B' */
252 filter->status = 0x80;
253 } else if (c == 0x28) { /* '(' */
254 filter->status++;
255 } else {
256 filter->status &= ~0xf;
257 CK((*filter->output_function)(MBFL_BAD_INPUT, filter->data));
258 CK((*filter->output_function)(0x24, filter->data));
259 goto retry;
260 }
261 break;
262
263 /* ESC $ ( */
264 /* case 0x04: */
265 /* case 0x14: */
266 /* case 0x24: */
267 /* case 0x84: */
268 /* case 0x94: */
269 case 4:
270 if (c == 0x40 || c == 0x42) { /* '@' or 'B' */
271 filter->status = 0x80;
272 } else if (c == 0x44) { /* 'D' */
273 filter->status = 0x90;
274 } else {
275 filter->status &= ~0xf;
276 CK((*filter->output_function)(MBFL_BAD_INPUT, filter->data));
277 CK((*filter->output_function)(0x24, filter->data));
278 CK((*filter->output_function)(0x28, filter->data));
279 goto retry;
280 }
281 break;
282
283 /* ESC ( */
284 /* case 0x05: */
285 /* case 0x15: */
286 /* case 0x25: */
287 /* case 0x85: */
288 /* case 0x95: */
289 case 5:
290 if (c == 0x42 || c == 0x48) { /* 'B' or 'H' */
291 filter->status = 0;
292 } else if (c == 0x4a) { /* 'J' */
293 filter->status = 0x10;
294 } else if (c == 0x49) { /* 'I' */
295 filter->status = 0x20;
296 } else {
297 filter->status &= ~0xf;
298 CK((*filter->output_function)(MBFL_BAD_INPUT, filter->data));
299 CK((*filter->output_function)(0x28, filter->data));
300 goto retry;
301 }
302 break;
303
304 default:
305 filter->status = 0;
306 break;
307 }
308
309 return 0;
310 }
311
mbfl_filt_conv_cp5022x_wchar_flush(mbfl_convert_filter * filter)312 static int mbfl_filt_conv_cp5022x_wchar_flush(mbfl_convert_filter *filter)
313 {
314 if ((filter->status & 0xF) == 1) {
315 /* 2-byte (JIS X 0208 or 0212) character was truncated */
316 CK((*filter->output_function)(MBFL_BAD_INPUT, filter->data));
317 }
318 filter->status = 0;
319
320 if (filter->flush_function) {
321 (*filter->flush_function)(filter->data);
322 }
323
324 return 0;
325 }
326
mbfl_filt_conv_wchar_cp50220(int c,mbfl_convert_filter * filter)327 static int mbfl_filt_conv_wchar_cp50220(int c, mbfl_convert_filter *filter)
328 {
329 int mode = MBFL_FILT_TL_HAN2ZEN_KATAKANA | MBFL_FILT_TL_HAN2ZEN_GLUE, second = 0;
330 bool consumed = false;
331
332 if (filter->cache) {
333 int s = mbfl_convert_kana(filter->cache, c, &consumed, &second, mode);
334 filter->cache = consumed ? 0 : c;
335 mbfl_filt_conv_wchar_cp50221(s, filter);
336 if (second) {
337 mbfl_filt_conv_wchar_cp50221(second, filter);
338 }
339 } else if (c == 0) {
340 /* This case has to be handled separately, since `filter->cache == 0` means
341 * no codepoint is cached */
342 (*filter->output_function)(0, filter->data);
343 } else {
344 filter->cache = c;
345 }
346
347 return 0;
348 }
349
mbfl_filt_conv_wchar_cp50220_flush(mbfl_convert_filter * filter)350 static int mbfl_filt_conv_wchar_cp50220_flush(mbfl_convert_filter *filter)
351 {
352 int mode = MBFL_FILT_TL_HAN2ZEN_KATAKANA | MBFL_FILT_TL_HAN2ZEN_GLUE, second = 0;
353
354 if (filter->cache) {
355 int s = mbfl_convert_kana(filter->cache, 0, NULL, &second, mode);
356 mbfl_filt_conv_wchar_cp50221(s, filter);
357 if (second) {
358 mbfl_filt_conv_wchar_cp50221(s, filter);
359 }
360 filter->cache = 0;
361 }
362
363 return mbfl_filt_conv_any_jis_flush(filter);
364 }
365
mbfl_filt_conv_wchar_cp50221(int c,mbfl_convert_filter * filter)366 int mbfl_filt_conv_wchar_cp50221(int c, mbfl_convert_filter *filter)
367 {
368 int s = 0;
369
370 if (c >= ucs_a1_jis_table_min && c < ucs_a1_jis_table_max) {
371 s = ucs_a1_jis_table[c - ucs_a1_jis_table_min];
372 } else if (c == 0x203E) { /* OVERLINE */
373 s = 0x1007E; /* Convert to JISX 0201 OVERLINE */
374 } else if (c >= ucs_a2_jis_table_min && c < ucs_a2_jis_table_max) {
375 s = ucs_a2_jis_table[c - ucs_a2_jis_table_min];
376 } else if (c >= ucs_i_jis_table_min && c < ucs_i_jis_table_max) {
377 s = ucs_i_jis_table[c - ucs_i_jis_table_min];
378 } else if (c >= ucs_r_jis_table_min && c < ucs_r_jis_table_max) {
379 s = ucs_r_jis_table[c - ucs_r_jis_table_min];
380 } else if (c >= 0xE000 && c <= 0xE757) {
381 /* 'private'/'user' codepoints */
382 s = c - 0xE000;
383 s = ((s / 94) + 0x7F) << 8 | ((s % 94) + 0x21);
384 }
385
386 if (s <= 0) {
387 if (c == 0xa5) { /* YEN SIGN */
388 s = 0x1005c;
389 } else if (c == 0xff3c) { /* FULLWIDTH REVERSE SOLIDUS */
390 s = 0x2140;
391 } else if (c == 0x2225) { /* PARALLEL TO */
392 s = 0x2142;
393 } else if (c == 0xff0d) { /* FULLWIDTH HYPHEN-MINUS */
394 s = 0x215d;
395 } else if (c == 0xffe0) { /* FULLWIDTH CENT SIGN */
396 s = 0x2171;
397 } else if (c == 0xffe1) { /* FULLWIDTH POUND SIGN */
398 s = 0x2172;
399 } else if (c == 0xffe2) { /* FULLWIDTH NOT SIGN */
400 s = 0x224c;
401 }
402 }
403
404 /* Above, we do a series of lookups in `ucs_*_jis_table` to find a
405 * corresponding kuten code for this Unicode codepoint
406 * If we get zero, that means the codepoint is not in JIS X 0208
407 * On the other hand, if we get a result with the high bits set on both
408 * upper and lower bytes, that is not a code in JIS X 0208 but rather
409 * in JIS X 0213
410 * In either case, check if this codepoint is one of the extensions added
411 * to JIS X 0208 by MicroSoft (to make CP932) */
412 if (s == 0 || ((s & 0x8000) && (s & 0x80))) {
413 int i;
414 s = -1;
415
416 for (i = 0;
417 i < cp932ext1_ucs_table_max - cp932ext1_ucs_table_min;
418 i++) {
419 const int oh = cp932ext1_ucs_table_min / 94;
420
421 if (c == cp932ext1_ucs_table[i]) {
422 s = ((i / 94 + oh + 0x21) << 8) + (i % 94 + 0x21);
423 break;
424 }
425 }
426
427 if (s < 0) {
428 const int oh = cp932ext2_ucs_table_min / 94;
429 const int cp932ext2_ucs_table_size =
430 cp932ext2_ucs_table_max - cp932ext2_ucs_table_min;
431 for (i = 0; i < cp932ext2_ucs_table_size; i++) {
432 if (c == cp932ext2_ucs_table[i]) {
433 s = ((i / 94 + oh + 0x21) << 8) + (i % 94 + 0x21);
434 break;
435 }
436 }
437 }
438
439 if (s < 0) {
440 const int cp932ext3_ucs_table_size =
441 cp932ext3_ucs_table_max - cp932ext3_ucs_table_min;
442 const int limit = cp932ext3_ucs_table_size >
443 cp932ext3_eucjp_table_size ?
444 cp932ext3_eucjp_table_size:
445 cp932ext3_ucs_table_size;
446 for (i = 0; i < limit; i++) {
447 if (c == cp932ext3_ucs_table[i]) {
448 s = cp932ext3_eucjp_table[i];
449 break;
450 }
451 }
452 }
453
454 if (c == 0) {
455 s = 0;
456 } else if (s <= 0) {
457 s = -1;
458 }
459 }
460
461 if (s >= 0) {
462 if (s < 0x80) { /* ASCII */
463 if ((filter->status & 0xff00) != 0) {
464 CK((*filter->output_function)(0x1b, filter->data)); /* ESC */
465 CK((*filter->output_function)(0x28, filter->data)); /* '(' */
466 CK((*filter->output_function)(0x42, filter->data)); /* 'B' */
467 filter->status = 0;
468 }
469 CK((*filter->output_function)(s, filter->data));
470 } else if (s >= 0xa0 && s < 0xe0) { /* X 0201 kana */
471 if ((filter->status & 0xff00) != 0x500) {
472 CK((*filter->output_function)(0x1b, filter->data)); /* ESC */
473 CK((*filter->output_function)(0x28, filter->data)); /* '(' */
474 CK((*filter->output_function)(0x49, filter->data)); /* 'I' */
475 filter->status = 0x500;
476 }
477 CK((*filter->output_function)(s - 0x80, filter->data));
478 } else if (s <= 0x927E) { /* X 0208 + extensions */
479 if ((filter->status & 0xff00) != 0x200) {
480 CK((*filter->output_function)(0x1b, filter->data)); /* ESC */
481 CK((*filter->output_function)(0x24, filter->data)); /* '$' */
482 CK((*filter->output_function)(0x42, filter->data)); /* 'B' */
483 filter->status = 0x200;
484 }
485 CK((*filter->output_function)((s >> 8) & 0xff, filter->data));
486 CK((*filter->output_function)(s & 0xff, filter->data));
487 } else if (s < 0x10000) { /* X0212 */
488 CK(mbfl_filt_conv_illegal_output(c, filter));
489 } else { /* X 0201 latin */
490 if ((filter->status & 0xff00) != 0x400) {
491 CK((*filter->output_function)(0x1b, filter->data)); /* ESC */
492 CK((*filter->output_function)(0x28, filter->data)); /* '(' */
493 CK((*filter->output_function)(0x4a, filter->data)); /* 'J' */
494 }
495 filter->status = 0x400;
496 CK((*filter->output_function)(s & 0x7f, filter->data));
497 }
498 } else {
499 CK(mbfl_filt_conv_illegal_output(c, filter));
500 }
501
502 return 0;
503 }
504
505 /*
506 * wchar => CP50222
507 */
mbfl_filt_conv_wchar_cp50222(int c,mbfl_convert_filter * filter)508 int mbfl_filt_conv_wchar_cp50222(int c, mbfl_convert_filter *filter)
509 {
510 int s = 0;
511
512 if (c >= ucs_a1_jis_table_min && c < ucs_a1_jis_table_max) {
513 s = ucs_a1_jis_table[c - ucs_a1_jis_table_min];
514 } else if (c == 0x203E) { /* OVERLINE */
515 s = 0x1007E; /* Convert to JISX 0201 OVERLINE */
516 } else if (c >= ucs_a2_jis_table_min && c < ucs_a2_jis_table_max) {
517 s = ucs_a2_jis_table[c - ucs_a2_jis_table_min];
518 } else if (c >= ucs_i_jis_table_min && c < ucs_i_jis_table_max) {
519 s = ucs_i_jis_table[c - ucs_i_jis_table_min];
520 } else if (c >= ucs_r_jis_table_min && c < ucs_r_jis_table_max) {
521 s = ucs_r_jis_table[c - ucs_r_jis_table_min];
522 } else if (c >= 0xE000 && c <= 0xE757) {
523 /* 'private'/'user' codepoints */
524 s = c - 0xE000;
525 s = ((s / 94) + 0x7F) << 8 | ((s % 94) + 0x21);
526 }
527
528 if (s <= 0) {
529 if (c == 0xa5) { /* YEN SIGN */
530 s = 0x1005c;
531 } else if (c == 0xff3c) { /* FULLWIDTH REVERSE SOLIDUS */
532 s = 0x2140;
533 } else if (c == 0x2225) { /* PARALLEL TO */
534 s = 0x2142;
535 } else if (c == 0xff0d) { /* FULLWIDTH HYPHEN-MINUS */
536 s = 0x215d;
537 } else if (c == 0xffe0) { /* FULLWIDTH CENT SIGN */
538 s = 0x2171;
539 } else if (c == 0xffe1) { /* FULLWIDTH POUND SIGN */
540 s = 0x2172;
541 } else if (c == 0xffe2) { /* FULLWIDTH NOT SIGN */
542 s = 0x224c;
543 }
544 }
545 if (s == 0 || ((s & 0x8000) && (s & 0x80))) {
546 int i;
547 s = -1;
548
549 for (i = 0;
550 i < cp932ext1_ucs_table_max - cp932ext1_ucs_table_min; i++) {
551 const int oh = cp932ext1_ucs_table_min / 94;
552
553 if (c == cp932ext1_ucs_table[i]) {
554 s = ((i / 94 + oh + 0x21) << 8) + (i % 94 + 0x21);
555 break;
556 }
557 }
558
559 if (s <= 0) {
560 const int oh = cp932ext2_ucs_table_min / 94;
561 const int cp932ext2_ucs_table_size =
562 cp932ext2_ucs_table_max - cp932ext2_ucs_table_min;
563 for (i = 0; i < cp932ext2_ucs_table_size; i++) {
564 if (c == cp932ext2_ucs_table[i]) {
565 s = ((i / 94 + oh + 0x21) << 8) + (i % 94 + 0x21);
566 break;
567 }
568 }
569 }
570
571 if (s <= 0) {
572 const int cp932ext3_ucs_table_size =
573 cp932ext3_ucs_table_max - cp932ext3_ucs_table_min;
574 const int limit = cp932ext3_ucs_table_size >
575 cp932ext3_eucjp_table_size ?
576 cp932ext3_eucjp_table_size:
577 cp932ext3_ucs_table_size;
578 for (i = 0; i < limit; i++) {
579 if (c == cp932ext3_ucs_table[i]) {
580 s = cp932ext3_eucjp_table[i];
581 break;
582 }
583 }
584 }
585
586 if (c == 0) {
587 s = 0;
588 } else if (s <= 0) {
589 s = -1;
590 }
591 }
592
593 if (s >= 0) {
594 if (s < 0x80) { /* ASCII */
595 if ((filter->status & 0xff00) == 0x500) {
596 CK((*filter->output_function)(0x0f, filter->data)); /* SO */
597 filter->status = 0;
598 } else if ((filter->status & 0xff00) != 0) {
599 CK((*filter->output_function)(0x1b, filter->data)); /* ESC */
600 CK((*filter->output_function)(0x28, filter->data)); /* '(' */
601 CK((*filter->output_function)(0x42, filter->data)); /* 'B' */
602 filter->status = 0;
603 }
604 CK((*filter->output_function)(s, filter->data));
605 } else if (s >= 0xa0 && s < 0xe0) { /* X 0201 kana */
606 if ((filter->status & 0xff00) != 0x500) {
607 CK((*filter->output_function)(0x0e, filter->data)); /* SI */
608 filter->status = 0x500;
609 }
610 CK((*filter->output_function)(s - 0x80, filter->data));
611 } else if (s <= 0x927E) { /* X 0208 */
612 if ((filter->status & 0xff00) == 0x500) {
613 CK((*filter->output_function)(0x0f, filter->data)); /* SO */
614 filter->status = 0;
615 }
616 if ((filter->status & 0xff00) != 0x200) {
617 CK((*filter->output_function)(0x1b, filter->data)); /* ESC */
618 CK((*filter->output_function)(0x24, filter->data)); /* '$' */
619 CK((*filter->output_function)(0x42, filter->data)); /* 'B' */
620 filter->status = 0x200;
621 }
622 CK((*filter->output_function)((s >> 8) & 0xff, filter->data));
623 CK((*filter->output_function)(s & 0xff, filter->data));
624 } else if (s < 0x10000) { /* X0212 */
625 CK(mbfl_filt_conv_illegal_output(c, filter));
626 } else { /* X 0201 latin */
627 if ((filter->status & 0xff00) == 0x500) {
628 CK((*filter->output_function)(0x0f, filter->data)); /* SO */
629 filter->status = 0;
630 }
631 if ((filter->status & 0xff00) != 0x400) {
632 CK((*filter->output_function)(0x1b, filter->data)); /* ESC */
633 CK((*filter->output_function)(0x28, filter->data)); /* '(' */
634 CK((*filter->output_function)(0x4a, filter->data)); /* 'J' */
635 }
636 filter->status = 0x400;
637 CK((*filter->output_function)(s & 0x7f, filter->data));
638 }
639 } else {
640 CK(mbfl_filt_conv_illegal_output(c, filter));
641 }
642
643 return 0;
644 }
645
646 int
mbfl_filt_conv_wchar_cp50222_flush(mbfl_convert_filter * filter)647 mbfl_filt_conv_wchar_cp50222_flush(mbfl_convert_filter *filter)
648 {
649 /* back to latin */
650 if ((filter->status & 0xff00) == 0x500) {
651 CK((*filter->output_function)(0x0f, filter->data)); /* SO */
652 } else if ((filter->status & 0xff00) != 0) {
653 CK((*filter->output_function)(0x1b, filter->data)); /* ESC */
654 CK((*filter->output_function)(0x28, filter->data)); /* '(' */
655 CK((*filter->output_function)(0x42, filter->data)); /* 'B' */
656 }
657 filter->status = 0;
658
659 if (filter->flush_function) {
660 (*filter->flush_function)(filter->data);
661 }
662
663 return 0;
664 }
665