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 };
53
54 const struct mbfl_identify_vtbl vtbl_identify_2022jpms = {
55 mbfl_no_encoding_2022jpms,
56 mbfl_filt_ident_common_ctor,
57 mbfl_filt_ident_common_dtor,
58 mbfl_filt_ident_2022jpms
59 };
60
61 const struct mbfl_convert_vtbl vtbl_2022jpms_wchar = {
62 mbfl_no_encoding_2022jpms,
63 mbfl_no_encoding_wchar,
64 mbfl_filt_conv_common_ctor,
65 mbfl_filt_conv_common_dtor,
66 mbfl_filt_conv_2022jpms_wchar,
67 mbfl_filt_conv_common_flush
68 };
69
70 const struct mbfl_convert_vtbl vtbl_wchar_2022jpms = {
71 mbfl_no_encoding_wchar,
72 mbfl_no_encoding_2022jpms,
73 mbfl_filt_conv_common_ctor,
74 mbfl_filt_conv_common_dtor,
75 mbfl_filt_conv_wchar_2022jpms,
76 mbfl_filt_conv_any_2022jpms_flush
77 };
78
79 #define CK(statement) do { if ((statement) < 0) return (-1); } while (0)
80
81 #define sjistoidx(c1, c2) \
82 (((c1) > 0x9f) \
83 ? (((c1) - 0xc1) * 188 + (c2) - (((c2) > 0x7e) ? 0x41 : 0x40)) \
84 : (((c1) - 0x81) * 188 + (c2) - (((c2) > 0x7e) ? 0x41 : 0x40)))
85 #define idxtojis1(c) (((c) / 94) + 0x21)
86 #define idxtojis2(c) (((c) % 94) + 0x21)
87
88 /*
89 * ISO-2022-JP-MS => wchar
90 */
91 int
mbfl_filt_conv_2022jpms_wchar(int c,mbfl_convert_filter * filter)92 mbfl_filt_conv_2022jpms_wchar(int c, mbfl_convert_filter *filter)
93 {
94 int c1, s, w;
95
96 retry:
97 switch (filter->status & 0xf) {
98 /* case 0x00: ASCII */
99 /* case 0x10: X 0201 latin */
100 /* case 0x20: X 0201 kana */
101 /* case 0x80: X 0208 */
102 /* case 0xa0: UDC */
103 case 0:
104 if (c == 0x1b) {
105 filter->status += 2;
106 } else if (filter->status == 0x20 && c > 0x20 && c < 0x60) { /* kana */
107 CK((*filter->output_function)(0xff40 + c, filter->data));
108 } else if ((filter->status == 0x80 || filter->status == 0xa0) && c > 0x20 && c < 0x80) { /* kanji first char */
109 filter->cache = c;
110 filter->status += 1;
111 } else if (c >= 0 && c < 0x80) { /* latin, CTLs */
112 CK((*filter->output_function)(c, filter->data));
113 } else if (c > 0xa0 && c < 0xe0) { /* GR kana */
114 CK((*filter->output_function)(0xfec0 + c, filter->data));
115 } else {
116 w = c & MBFL_WCSGROUP_MASK;
117 w |= MBFL_WCSGROUP_THROUGH;
118 CK((*filter->output_function)(w, filter->data));
119 }
120 break;
121
122 /* case 0x81: X 0208 second char */
123 /* case 0xa1: UDC second char */
124 case 1:
125 w = 0;
126 filter->status &= ~0xf;
127 c1 = filter->cache;
128 if (c > 0x20 && c < 0x7f) {
129 s = (c1 - 0x21)*94 + c - 0x21;
130 if (filter->status == 0x80) {
131 if (s <= 137) {
132 if (s == 31) {
133 w = 0xff3c; /* FULLWIDTH REVERSE SOLIDUS */
134 } else if (s == 32) {
135 w = 0xff5e; /* FULLWIDTH TILDE */
136 } else if (s == 33) {
137 w = 0x2225; /* PARALLEL TO */
138 } else if (s == 60) {
139 w = 0xff0d; /* FULLWIDTH HYPHEN-MINUS */
140 } else if (s == 80) {
141 w = 0xffe0; /* FULLWIDTH CENT SIGN */
142 } else if (s == 81) {
143 w = 0xffe1; /* FULLWIDTH POUND SIGN */
144 } else if (s == 137) {
145 w = 0xffe2; /* FULLWIDTH NOT SIGN */
146 }
147 }
148 if (w == 0) {
149 if (s >= cp932ext1_ucs_table_min && s < cp932ext1_ucs_table_max) { /* vendor ext1 (13ku) */
150 w = cp932ext1_ucs_table[s - cp932ext1_ucs_table_min];
151 } else if (s >= 0 && s < jisx0208_ucs_table_size) {
152 w = jisx0208_ucs_table[s];
153 } else if (s >= cp932ext2_ucs_table_min && s < cp932ext2_ucs_table_max) { /* vendor ext2 (89ku - 92ku) */
154 w = cp932ext2_ucs_table[s - cp932ext2_ucs_table_min];
155 } else {
156 w = 0;
157 }
158 }
159 if (w <= 0) {
160 w = (c1 << 8) | c;
161 w &= MBFL_WCSPLANE_MASK;
162 w |= MBFL_WCSPLANE_JIS0208;
163 }
164 CK((*filter->output_function)(w, filter->data));
165 } else {
166 if (c1 > 0x20 && c1 < 0x35) {
167 w = 0xe000 + (c1 - 0x21)*94 + c - 0x21;
168 }
169 if (w <= 0) {
170 w = (((c1 - 0x21) + 0x7f) << 8) | c;
171 w &= MBFL_WCSPLANE_MASK;
172 w |= MBFL_WCSPLANE_JIS0208;
173 }
174 CK((*filter->output_function)(w, filter->data));
175 }
176 } else if (c == 0x1b) {
177 filter->status += 2;
178 } else if ((c >= 0 && c < 0x21) || c == 0x7f) { /* CTLs */
179 CK((*filter->output_function)(c, filter->data));
180 } else {
181 w = (c1 << 8) | c;
182 w &= MBFL_WCSGROUP_MASK;
183 w |= MBFL_WCSGROUP_THROUGH;
184 CK((*filter->output_function)(w, filter->data));
185 }
186 break;
187
188 /* ESC */
189 /* case 0x02: */
190 /* case 0x12: */
191 /* case 0x22: */
192 /* case 0x82: */
193 /* case 0xa2: */
194 case 2:
195 if (c == 0x24) { /* '$' */
196 filter->status++;
197 } else if (c == 0x28) { /* '(' */
198 filter->status += 3;
199 } else {
200 filter->status &= ~0xf;
201 CK((*filter->output_function)(0x1b, filter->data));
202 goto retry;
203 }
204 break;
205
206 /* ESC $ */
207 /* case 0x03: */
208 /* case 0x13: */
209 /* case 0x23: */
210 /* case 0x83: */
211 /* case 0xa3: */
212 case 3:
213 if (c == 0x40 || c == 0x42) { /* '@' or 'B' */
214 filter->status = 0x80;
215 } else if (c == 0x28) { /* '(' */
216 filter->status++;
217 } else {
218 filter->status &= ~0xf;
219 CK((*filter->output_function)(0x1b, filter->data));
220 CK((*filter->output_function)(0x24, filter->data));
221 goto retry;
222 }
223 break;
224
225 /* ESC $ ( */
226 /* case 0x04: */
227 /* case 0x14: */
228 /* case 0x24: */
229 /* case 0x84: */
230 /* case 0xa4: */
231 case 4:
232 if (c == 0x40 || c == 0x42) { /* '@' or 'B' */
233 filter->status = 0x80;
234 } else if (c == 0x3f) { /* '?' */
235 filter->status = 0xa0;
236 } else {
237 filter->status &= ~0xf;
238 CK((*filter->output_function)(0x1b, filter->data));
239 CK((*filter->output_function)(0x24, filter->data));
240 CK((*filter->output_function)(0x28, filter->data));
241 goto retry;
242 }
243 break;
244
245 /* ESC ( */
246 /* case 0x05: */
247 /* case 0x15: */
248 /* case 0x25: */
249 /* case 0x85: */
250 /* case 0xa5: */
251 case 5:
252 if (c == 0x42) { /* 'B' */
253 filter->status = 0;
254 } else if (c == 0x4a) { /* 'J' */
255 filter->status = 0;
256 } else if (c == 0x49) { /* 'I' */
257 filter->status = 0x20;
258 } else {
259 filter->status &= ~0xf;
260 CK((*filter->output_function)(0x1b, filter->data));
261 CK((*filter->output_function)(0x28, filter->data));
262 goto retry;
263 }
264 break;
265
266 default:
267 filter->status = 0;
268 break;
269 }
270
271 return c;
272 }
273
274 static int
cp932ext3_cp932ext2_jis(int c)275 cp932ext3_cp932ext2_jis(int c)
276 {
277 int idx;
278
279 idx = sjistoidx(0xfa, 0x40) + c;
280 if (idx >= sjistoidx(0xfa, 0x5c))
281 idx -= sjistoidx(0xfa, 0x5c) - sjistoidx(0xed, 0x40);
282 else if (idx >= sjistoidx(0xfa, 0x55))
283 idx -= sjistoidx(0xfa, 0x55) - sjistoidx(0xee, 0xfa);
284 else if (idx >= sjistoidx(0xfa, 0x40))
285 idx -= sjistoidx(0xfa, 0x40) - sjistoidx(0xee, 0xef);
286 return idxtojis1(idx) << 8 | idxtojis2(idx);
287 }
288
289 /*
290 * wchar => ISO-2022-JP-MS
291 */
292 int
mbfl_filt_conv_wchar_2022jpms(int c,mbfl_convert_filter * filter)293 mbfl_filt_conv_wchar_2022jpms(int c, mbfl_convert_filter *filter)
294 {
295 int c1, c2, s1, s2;
296
297 s1 = 0;
298 s2 = 0;
299 if (c >= ucs_a1_jis_table_min && c < ucs_a1_jis_table_max) {
300 s1 = ucs_a1_jis_table[c - ucs_a1_jis_table_min];
301 } else if (c >= ucs_a2_jis_table_min && c < ucs_a2_jis_table_max) {
302 s1 = ucs_a2_jis_table[c - ucs_a2_jis_table_min];
303 } else if (c >= ucs_i_jis_table_min && c < ucs_i_jis_table_max) {
304 s1 = ucs_i_jis_table[c - ucs_i_jis_table_min];
305 } else if (c >= ucs_r_jis_table_min && c < ucs_r_jis_table_max) {
306 s1 = ucs_r_jis_table[c - ucs_r_jis_table_min];
307 } else if (c >= 0xe000 && c < (0xe000 + 20*94)) { /* user (95ku - 114ku) */
308 s1 = c - 0xe000;
309 c1 = s1/94 + 0x7f;
310 c2 = s1%94 + 0x21;
311 s1 = (c1 << 8) | c2;
312 }
313 if (s1 <= 0) {
314 c1 = c & ~MBFL_WCSPLANE_MASK;
315 if (c1 == MBFL_WCSPLANE_WINCP932) {
316 s1 = c & MBFL_WCSPLANE_MASK;
317 s2 = 1;
318 } else if (c1 == MBFL_WCSPLANE_JIS0208) {
319 s1 = c & MBFL_WCSPLANE_MASK;
320 } else if (c1 == MBFL_WCSPLANE_JIS0212) {
321 s1 = c & MBFL_WCSPLANE_MASK;
322 s1 |= 0x8080;
323 } else if (c == 0xa5) { /* YEN SIGN */
324 s1 = 0x216f; /* FULLWIDTH YEN SIGN */
325 } else if (c == 0x203e) { /* OVER LINE */
326 s1 = 0x2131; /* FULLWIDTH MACRON */
327 } else if (c == 0xff3c) { /* FULLWIDTH REVERSE SOLIDUS */
328 s1 = 0x2140;
329 } else if (c == 0xff5e) { /* FULLWIDTH TILDE */
330 s1 = 0x2141;
331 } else if (c == 0x2225) { /* PARALLEL TO */
332 s1 = 0x2142;
333 } else if (c == 0xff0d) { /* FULLWIDTH HYPHEN-MINUS */
334 s1 = 0x215d;
335 } else if (c == 0xffe0) { /* FULLWIDTH CENT SIGN */
336 s1 = 0x2171;
337 } else if (c == 0xffe1) { /* FULLWIDTH POUND SIGN */
338 s1 = 0x2172;
339 } else if (c == 0xffe2) { /* FULLWIDTH NOT SIGN */
340 s1 = 0x224c;
341 }
342 }
343 if ((s1 <= 0) || (s1 >= 0xa1a1 && s2 == 0)) { /* not found or X 0212 */
344 s1 = -1;
345 c1 = 0;
346 c2 = cp932ext1_ucs_table_max - cp932ext1_ucs_table_min;
347 while (c1 < c2) { /* CP932 vendor ext1 (13ku) */
348 if (c == cp932ext1_ucs_table[c1]) {
349 s1 = ((c1/94 + 0x2d) << 8) + (c1%94 + 0x21);
350 break;
351 }
352 c1++;
353 }
354 if (s1 <= 0) {
355 c1 = 0;
356 c2 = cp932ext3_ucs_table_max - cp932ext3_ucs_table_min;
357 while (c1 < c2) { /* CP932 vendor ext3 (115ku - 119ku) */
358 if (c == cp932ext3_ucs_table[c1]) {
359 s1 = cp932ext3_cp932ext2_jis(c1);
360 break;
361 }
362 c1++;
363 }
364 }
365 if (c == 0) {
366 s1 = 0;
367 } else if (s1 <= 0) {
368 s1 = -1;
369 }
370 }
371 if (s1 >= 0) {
372 if (s1 < 0x80) { /* latin */
373 if ((filter->status & 0xff00) != 0) {
374 CK((*filter->output_function)(0x1b, filter->data)); /* ESC */
375 CK((*filter->output_function)(0x28, filter->data)); /* '(' */
376 CK((*filter->output_function)(0x42, filter->data)); /* 'B' */
377 }
378 CK((*filter->output_function)(s1, filter->data));
379 filter->status = 0;
380 } else if (s1 > 0xa0 && s1 < 0xe0) { /* kana */
381 if ((filter->status & 0xff00) != 0x100) {
382 CK((*filter->output_function)(0x1b, filter->data)); /* ESC */
383 CK((*filter->output_function)(0x28, filter->data)); /* '(' */
384 CK((*filter->output_function)(0x49, filter->data)); /* 'I' */
385 }
386 filter->status = 0x100;
387 CK((*filter->output_function)(s1 & 0x7f, filter->data));
388 } else if (s1 < 0x7e7f) { /* X 0208 */
389 if ((filter->status & 0xff00) != 0x200) {
390 CK((*filter->output_function)(0x1b, filter->data)); /* ESC */
391 CK((*filter->output_function)(0x24, filter->data)); /* '$' */
392 CK((*filter->output_function)(0x42, filter->data)); /* 'B' */
393 }
394 filter->status = 0x200;
395 CK((*filter->output_function)((s1 >> 8) & 0xff, filter->data));
396 CK((*filter->output_function)(s1 & 0x7f, filter->data));
397 } else if (s1 < 0x927f) { /* UDC */
398 if ((filter->status & 0xff00) != 0x800) {
399 CK((*filter->output_function)(0x1b, filter->data)); /* ESC */
400 CK((*filter->output_function)(0x24, filter->data)); /* '$' */
401 CK((*filter->output_function)(0x28, filter->data)); /* '(' */
402 CK((*filter->output_function)(0x3f, filter->data)); /* '?' */
403 }
404 filter->status = 0x800;
405 CK((*filter->output_function)(((s1 >> 8) - 0x5e) & 0x7f, filter->data));
406 CK((*filter->output_function)(s1 & 0x7f, filter->data));
407 }
408 } else {
409 if (filter->illegal_mode != MBFL_OUTPUTFILTER_ILLEGAL_MODE_NONE) {
410 CK(mbfl_filt_conv_illegal_output(c, filter));
411 }
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
