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_iso2022_jp_ms.h"
36
37 #include "unicode_table_cp932_ext.h"
38 #include "unicode_table_jis.h"
39 #include "cp932_table.h"
40
41 int mbfl_filt_ident_2022jpms(int c, mbfl_identify_filter *filter);
42
43 static const char *mbfl_encoding_2022jpms_aliases[] = {"ISO2022JPMS", NULL};
44
45 const mbfl_encoding mbfl_encoding_2022jpms = {
46 mbfl_no_encoding_2022jpms,
47 "ISO-2022-JP-MS",
48 "ISO-2022-JP",
49 (const char *(*)[])&mbfl_encoding_2022jpms_aliases,
50 NULL,
51 MBFL_ENCTYPE_MBCS | MBFL_ENCTYPE_SHFTCODE | MBFL_ENCTYPE_GL_UNSAFE,
52 &vtbl_2022jpms_wchar,
53 &vtbl_wchar_2022jpms
54 };
55
56 const struct mbfl_identify_vtbl vtbl_identify_2022jpms = {
57 mbfl_no_encoding_2022jpms,
58 mbfl_filt_ident_common_ctor,
59 mbfl_filt_ident_common_dtor,
60 mbfl_filt_ident_2022jpms
61 };
62
63 const struct mbfl_convert_vtbl vtbl_2022jpms_wchar = {
64 mbfl_no_encoding_2022jpms,
65 mbfl_no_encoding_wchar,
66 mbfl_filt_conv_common_ctor,
67 mbfl_filt_conv_common_dtor,
68 mbfl_filt_conv_2022jpms_wchar,
69 mbfl_filt_conv_common_flush
70 };
71
72 const struct mbfl_convert_vtbl vtbl_wchar_2022jpms = {
73 mbfl_no_encoding_wchar,
74 mbfl_no_encoding_2022jpms,
75 mbfl_filt_conv_common_ctor,
76 mbfl_filt_conv_common_dtor,
77 mbfl_filt_conv_wchar_2022jpms,
78 mbfl_filt_conv_any_2022jpms_flush
79 };
80
81 #define CK(statement) do { if ((statement) < 0) return (-1); } while (0)
82
83 #define sjistoidx(c1, c2) \
84 (((c1) > 0x9f) \
85 ? (((c1) - 0xc1) * 188 + (c2) - (((c2) > 0x7e) ? 0x41 : 0x40)) \
86 : (((c1) - 0x81) * 188 + (c2) - (((c2) > 0x7e) ? 0x41 : 0x40)))
87 #define idxtojis1(c) (((c) / 94) + 0x21)
88 #define idxtojis2(c) (((c) % 94) + 0x21)
89
90 /*
91 * ISO-2022-JP-MS => wchar
92 */
93 int
mbfl_filt_conv_2022jpms_wchar(int c,mbfl_convert_filter * filter)94 mbfl_filt_conv_2022jpms_wchar(int c, mbfl_convert_filter *filter)
95 {
96 int c1, s, w;
97
98 retry:
99 switch (filter->status & 0xf) {
100 /* case 0x00: ASCII */
101 /* case 0x10: X 0201 latin */
102 /* case 0x20: X 0201 kana */
103 /* case 0x80: X 0208 */
104 /* case 0xa0: UDC */
105 case 0:
106 if (c == 0x1b) {
107 filter->status += 2;
108 } else if (filter->status == 0x20 && c > 0x20 && c < 0x60) { /* kana */
109 CK((*filter->output_function)(0xff40 + c, filter->data));
110 } else if ((filter->status == 0x80 || filter->status == 0xa0) && c > 0x20 && c < 0x80) { /* kanji first char */
111 filter->cache = c;
112 filter->status += 1;
113 } else if (c >= 0 && c < 0x80) { /* latin, CTLs */
114 CK((*filter->output_function)(c, filter->data));
115 } else if (c > 0xa0 && c < 0xe0) { /* GR kana */
116 CK((*filter->output_function)(0xfec0 + c, filter->data));
117 } else {
118 w = c & MBFL_WCSGROUP_MASK;
119 w |= MBFL_WCSGROUP_THROUGH;
120 CK((*filter->output_function)(w, filter->data));
121 }
122 break;
123
124 /* case 0x81: X 0208 second char */
125 /* case 0xa1: UDC second char */
126 case 1:
127 w = 0;
128 filter->status &= ~0xf;
129 c1 = filter->cache;
130 if (c > 0x20 && c < 0x7f) {
131 s = (c1 - 0x21)*94 + c - 0x21;
132 if (filter->status == 0x80) {
133 if (s <= 137) {
134 if (s == 31) {
135 w = 0xff3c; /* FULLWIDTH REVERSE SOLIDUS */
136 } else if (s == 32) {
137 w = 0xff5e; /* FULLWIDTH TILDE */
138 } else if (s == 33) {
139 w = 0x2225; /* PARALLEL TO */
140 } else if (s == 60) {
141 w = 0xff0d; /* FULLWIDTH HYPHEN-MINUS */
142 } else if (s == 80) {
143 w = 0xffe0; /* FULLWIDTH CENT SIGN */
144 } else if (s == 81) {
145 w = 0xffe1; /* FULLWIDTH POUND SIGN */
146 } else if (s == 137) {
147 w = 0xffe2; /* FULLWIDTH NOT SIGN */
148 }
149 }
150 if (w == 0) {
151 if (s >= cp932ext1_ucs_table_min && s < cp932ext1_ucs_table_max) { /* vendor ext1 (13ku) */
152 w = cp932ext1_ucs_table[s - cp932ext1_ucs_table_min];
153 } else if (s >= 0 && s < jisx0208_ucs_table_size) {
154 w = jisx0208_ucs_table[s];
155 } else if (s >= cp932ext2_ucs_table_min && s < cp932ext2_ucs_table_max) { /* vendor ext2 (89ku - 92ku) */
156 w = cp932ext2_ucs_table[s - cp932ext2_ucs_table_min];
157 } else {
158 w = 0;
159 }
160 }
161 if (w <= 0) {
162 w = (c1 << 8) | c;
163 w &= MBFL_WCSPLANE_MASK;
164 w |= MBFL_WCSPLANE_JIS0208;
165 }
166 CK((*filter->output_function)(w, filter->data));
167 } else {
168 if (c1 > 0x20 && c1 < 0x35) {
169 w = 0xe000 + (c1 - 0x21)*94 + c - 0x21;
170 }
171 if (w <= 0) {
172 w = (((c1 - 0x21) + 0x7f) << 8) | c;
173 w &= MBFL_WCSPLANE_MASK;
174 w |= MBFL_WCSPLANE_JIS0208;
175 }
176 CK((*filter->output_function)(w, filter->data));
177 }
178 } else if (c == 0x1b) {
179 filter->status += 2;
180 } else if ((c >= 0 && c < 0x21) || c == 0x7f) { /* CTLs */
181 CK((*filter->output_function)(c, filter->data));
182 } else {
183 w = (c1 << 8) | c;
184 w &= MBFL_WCSGROUP_MASK;
185 w |= MBFL_WCSGROUP_THROUGH;
186 CK((*filter->output_function)(w, filter->data));
187 }
188 break;
189
190 /* ESC */
191 /* case 0x02: */
192 /* case 0x12: */
193 /* case 0x22: */
194 /* case 0x82: */
195 /* case 0xa2: */
196 case 2:
197 if (c == 0x24) { /* '$' */
198 filter->status++;
199 } else if (c == 0x28) { /* '(' */
200 filter->status += 3;
201 } else {
202 filter->status &= ~0xf;
203 CK((*filter->output_function)(0x1b, filter->data));
204 goto retry;
205 }
206 break;
207
208 /* ESC $ */
209 /* case 0x03: */
210 /* case 0x13: */
211 /* case 0x23: */
212 /* case 0x83: */
213 /* case 0xa3: */
214 case 3:
215 if (c == 0x40 || c == 0x42) { /* '@' or 'B' */
216 filter->status = 0x80;
217 } else if (c == 0x28) { /* '(' */
218 filter->status++;
219 } else {
220 filter->status &= ~0xf;
221 CK((*filter->output_function)(0x1b, filter->data));
222 CK((*filter->output_function)(0x24, filter->data));
223 goto retry;
224 }
225 break;
226
227 /* ESC $ ( */
228 /* case 0x04: */
229 /* case 0x14: */
230 /* case 0x24: */
231 /* case 0x84: */
232 /* case 0xa4: */
233 case 4:
234 if (c == 0x40 || c == 0x42) { /* '@' or 'B' */
235 filter->status = 0x80;
236 } else if (c == 0x3f) { /* '?' */
237 filter->status = 0xa0;
238 } else {
239 filter->status &= ~0xf;
240 CK((*filter->output_function)(0x1b, filter->data));
241 CK((*filter->output_function)(0x24, filter->data));
242 CK((*filter->output_function)(0x28, filter->data));
243 goto retry;
244 }
245 break;
246
247 /* ESC ( */
248 /* case 0x05: */
249 /* case 0x15: */
250 /* case 0x25: */
251 /* case 0x85: */
252 /* case 0xa5: */
253 case 5:
254 if (c == 0x42) { /* 'B' */
255 filter->status = 0;
256 } else if (c == 0x4a) { /* 'J' */
257 filter->status = 0;
258 } else if (c == 0x49) { /* 'I' */
259 filter->status = 0x20;
260 } else {
261 filter->status &= ~0xf;
262 CK((*filter->output_function)(0x1b, filter->data));
263 CK((*filter->output_function)(0x28, filter->data));
264 goto retry;
265 }
266 break;
267
268 default:
269 filter->status = 0;
270 break;
271 }
272
273 return c;
274 }
275
276 static int
cp932ext3_cp932ext2_jis(int c)277 cp932ext3_cp932ext2_jis(int c)
278 {
279 int idx;
280
281 idx = sjistoidx(0xfa, 0x40) + c;
282 if (idx >= sjistoidx(0xfa, 0x5c))
283 idx -= sjistoidx(0xfa, 0x5c) - sjistoidx(0xed, 0x40);
284 else if (idx >= sjistoidx(0xfa, 0x55))
285 idx -= sjistoidx(0xfa, 0x55) - sjistoidx(0xee, 0xfa);
286 else if (idx >= sjistoidx(0xfa, 0x40))
287 idx -= sjistoidx(0xfa, 0x40) - sjistoidx(0xee, 0xef);
288 return idxtojis1(idx) << 8 | idxtojis2(idx);
289 }
290
291 /*
292 * wchar => ISO-2022-JP-MS
293 */
294 int
mbfl_filt_conv_wchar_2022jpms(int c,mbfl_convert_filter * filter)295 mbfl_filt_conv_wchar_2022jpms(int c, mbfl_convert_filter *filter)
296 {
297 int c1, c2, s1, s2;
298
299 s1 = 0;
300 s2 = 0;
301 if (c >= ucs_a1_jis_table_min && c < ucs_a1_jis_table_max) {
302 s1 = ucs_a1_jis_table[c - ucs_a1_jis_table_min];
303 } else if (c >= ucs_a2_jis_table_min && c < ucs_a2_jis_table_max) {
304 s1 = ucs_a2_jis_table[c - ucs_a2_jis_table_min];
305 } else if (c >= ucs_i_jis_table_min && c < ucs_i_jis_table_max) {
306 s1 = ucs_i_jis_table[c - ucs_i_jis_table_min];
307 } else if (c >= ucs_r_jis_table_min && c < ucs_r_jis_table_max) {
308 s1 = ucs_r_jis_table[c - ucs_r_jis_table_min];
309 } else if (c >= 0xe000 && c < (0xe000 + 20*94)) { /* user (95ku - 114ku) */
310 s1 = c - 0xe000;
311 c1 = s1/94 + 0x7f;
312 c2 = s1%94 + 0x21;
313 s1 = (c1 << 8) | c2;
314 }
315 if (s1 <= 0) {
316 c1 = c & ~MBFL_WCSPLANE_MASK;
317 if (c1 == MBFL_WCSPLANE_WINCP932) {
318 s1 = c & MBFL_WCSPLANE_MASK;
319 s2 = 1;
320 } else if (c1 == MBFL_WCSPLANE_JIS0208) {
321 s1 = c & MBFL_WCSPLANE_MASK;
322 } else if (c1 == MBFL_WCSPLANE_JIS0212) {
323 s1 = c & MBFL_WCSPLANE_MASK;
324 s1 |= 0x8080;
325 } else if (c == 0xa5) { /* YEN SIGN */
326 s1 = 0x216f; /* FULLWIDTH YEN SIGN */
327 } else if (c == 0x203e) { /* OVER LINE */
328 s1 = 0x2131; /* FULLWIDTH MACRON */
329 } else if (c == 0xff3c) { /* FULLWIDTH REVERSE SOLIDUS */
330 s1 = 0x2140;
331 } else if (c == 0xff5e) { /* FULLWIDTH TILDE */
332 s1 = 0x2141;
333 } else if (c == 0x2225) { /* PARALLEL TO */
334 s1 = 0x2142;
335 } else if (c == 0xff0d) { /* FULLWIDTH HYPHEN-MINUS */
336 s1 = 0x215d;
337 } else if (c == 0xffe0) { /* FULLWIDTH CENT SIGN */
338 s1 = 0x2171;
339 } else if (c == 0xffe1) { /* FULLWIDTH POUND SIGN */
340 s1 = 0x2172;
341 } else if (c == 0xffe2) { /* FULLWIDTH NOT SIGN */
342 s1 = 0x224c;
343 }
344 }
345 if ((s1 <= 0) || (s1 >= 0xa1a1 && s2 == 0)) { /* not found or X 0212 */
346 s1 = -1;
347 c1 = 0;
348 c2 = cp932ext1_ucs_table_max - cp932ext1_ucs_table_min;
349 while (c1 < c2) { /* CP932 vendor ext1 (13ku) */
350 if (c == cp932ext1_ucs_table[c1]) {
351 s1 = ((c1/94 + 0x2d) << 8) + (c1%94 + 0x21);
352 break;
353 }
354 c1++;
355 }
356 if (s1 <= 0) {
357 c1 = 0;
358 c2 = cp932ext3_ucs_table_max - cp932ext3_ucs_table_min;
359 while (c1 < c2) { /* CP932 vendor ext3 (115ku - 119ku) */
360 if (c == cp932ext3_ucs_table[c1]) {
361 s1 = cp932ext3_cp932ext2_jis(c1);
362 break;
363 }
364 c1++;
365 }
366 }
367 if (c == 0) {
368 s1 = 0;
369 } else if (s1 <= 0) {
370 s1 = -1;
371 }
372 }
373 if (s1 >= 0) {
374 if (s1 < 0x80) { /* latin */
375 if ((filter->status & 0xff00) != 0) {
376 CK((*filter->output_function)(0x1b, filter->data)); /* ESC */
377 CK((*filter->output_function)(0x28, filter->data)); /* '(' */
378 CK((*filter->output_function)(0x42, filter->data)); /* 'B' */
379 }
380 CK((*filter->output_function)(s1, filter->data));
381 filter->status = 0;
382 } else if (s1 > 0xa0 && s1 < 0xe0) { /* kana */
383 if ((filter->status & 0xff00) != 0x100) {
384 CK((*filter->output_function)(0x1b, filter->data)); /* ESC */
385 CK((*filter->output_function)(0x28, filter->data)); /* '(' */
386 CK((*filter->output_function)(0x49, filter->data)); /* 'I' */
387 }
388 filter->status = 0x100;
389 CK((*filter->output_function)(s1 & 0x7f, filter->data));
390 } else if (s1 < 0x7e7f) { /* X 0208 */
391 if ((filter->status & 0xff00) != 0x200) {
392 CK((*filter->output_function)(0x1b, filter->data)); /* ESC */
393 CK((*filter->output_function)(0x24, filter->data)); /* '$' */
394 CK((*filter->output_function)(0x42, filter->data)); /* 'B' */
395 }
396 filter->status = 0x200;
397 CK((*filter->output_function)((s1 >> 8) & 0xff, filter->data));
398 CK((*filter->output_function)(s1 & 0x7f, filter->data));
399 } else if (s1 < 0x927f) { /* UDC */
400 if ((filter->status & 0xff00) != 0x800) {
401 CK((*filter->output_function)(0x1b, filter->data)); /* ESC */
402 CK((*filter->output_function)(0x24, filter->data)); /* '$' */
403 CK((*filter->output_function)(0x28, filter->data)); /* '(' */
404 CK((*filter->output_function)(0x3f, filter->data)); /* '?' */
405 }
406 filter->status = 0x800;
407 CK((*filter->output_function)(((s1 >> 8) - 0x5e) & 0x7f, filter->data));
408 CK((*filter->output_function)(s1 & 0x7f, filter->data));
409 }
410 } else {
411 CK(mbfl_filt_conv_illegal_output(c, filter));
412 }
413
414 return c;
415 }
416
417 int
mbfl_filt_conv_any_2022jpms_flush(mbfl_convert_filter * filter)418 mbfl_filt_conv_any_2022jpms_flush(mbfl_convert_filter *filter)
419 {
420 /* back to latin */
421 if ((filter->status & 0xff00) != 0) {
422 CK((*filter->output_function)(0x1b, filter->data)); /* ESC */
423 CK((*filter->output_function)(0x28, filter->data)); /* '(' */
424 CK((*filter->output_function)(0x42, filter->data)); /* 'B' */
425 }
426
427 filter->status &= 0xff;
428
429 if (filter->flush_function != NULL) {
430 return (*filter->flush_function)(filter->data);
431 }
432
433 return 0;
434 }
435
mbfl_filt_ident_2022jpms(int c,mbfl_identify_filter * filter)436 int mbfl_filt_ident_2022jpms(int c, mbfl_identify_filter *filter)
437 {
438 retry:
439 switch (filter->status & 0xf) {
440 /* case 0x00: ASCII */
441 /* case 0x10: X 0201 latin */
442 /* case 0x20: X 0201 kana */
443 /* case 0x80: X 0208 */
444 /* case 0xa0: X UDC */
445 case 0:
446 if (c == 0x1b) {
447 filter->status += 2;
448 } else if ((filter->status == 0x80 || filter->status == 0xa0) && c > 0x20 && c < 0x80) { /* kanji first char */
449 filter->status += 1;
450 } else if (c >= 0 && c < 0x80) { /* latin, CTLs */
451 ;
452 } else {
453 filter->flag = 1; /* bad */
454 }
455 break;
456
457 /* case 0x81: X 0208 second char */
458 /* case 0xa1: UDC second char */
459 case 1:
460 filter->status &= ~0xf;
461 if (c == 0x1b) {
462 goto retry;
463 } else if (c < 0x21 || c > 0x7e) { /* bad */
464 filter->flag = 1;
465 }
466 break;
467
468 /* ESC */
469 case 2:
470 if (c == 0x24) { /* '$' */
471 filter->status++;
472 } else if (c == 0x28) { /* '(' */
473 filter->status += 3;
474 } else {
475 filter->flag = 1; /* bad */
476 filter->status &= ~0xf;
477 goto retry;
478 }
479 break;
480
481 /* ESC $ */
482 case 3:
483 if (c == 0x40 || c == 0x42) { /* '@' or 'B' */
484 filter->status = 0x80;
485 } else if (c == 0x28) { /* '(' */
486 filter->status++;
487 } else {
488 filter->flag = 1; /* bad */
489 filter->status &= ~0xf;
490 goto retry;
491 }
492 break;
493
494 /* ESC $ ( */
495 case 4:
496 if (c == 0x40 || c == 0x42) { /* '@' or 'B' */
497 filter->status = 0x80;
498 } else if (c == 0x3f) { /* '?' */
499 filter->status = 0xa0;
500 } else {
501 filter->flag = 1; /* bad */
502 filter->status &= ~0xf;
503 goto retry;
504 }
505 break;
506
507 /* ESC ( */
508 case 5:
509 if (c == 0x42) { /* 'B' */
510 filter->status = 0;
511 } else if (c == 0x4a) { /* 'J' */
512 filter->status = 0;
513 } else if (c == 0x49) { /* 'I' */
514 filter->status = 0x20;
515 } else {
516 filter->flag = 1; /* bad */
517 filter->status &= ~0xf;
518 goto retry;
519 }
520 break;
521
522 default:
523 filter->status = 0;
524 break;
525 }
526
527 return c;
528 }
529
