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_iso2022_jp_ms.c
26 * by Rui Hirokawa <hirokawa@php.net> on 25 July 2011.
27 *
28 */
29
30 #ifdef HAVE_CONFIG_H
31 #include "config.h"
32 #endif
33
34 #include "mbfilter.h"
35 #include "mbfilter_iso2022jp_mobile.h"
36 #include "mbfilter_sjis_mobile.h"
37
38 #include "unicode_table_cp932_ext.h"
39 #include "unicode_table_jis.h"
40 #include "cp932_table.h"
41
42 extern int mbfl_filt_conv_any_jis_flush(mbfl_convert_filter *filter);
43 extern int mbfl_filt_ident_2022jpms(int c, mbfl_identify_filter *filter);
44
45 static const char *mbfl_encoding_2022jp_kddi_aliases[] = {"ISO-2022-JP-KDDI", NULL};
46
47 const mbfl_encoding mbfl_encoding_2022jp_kddi = {
48 mbfl_no_encoding_2022jp_kddi,
49 "ISO-2022-JP-MOBILE#KDDI",
50 "ISO-2022-JP",
51 &mbfl_encoding_2022jp_kddi_aliases,
52 NULL,
53 MBFL_ENCTYPE_MBCS | MBFL_ENCTYPE_SHFTCODE | MBFL_ENCTYPE_GL_UNSAFE
54 };
55
56 const struct mbfl_identify_vtbl vtbl_identify_2022jp_kddi = {
57 mbfl_no_encoding_2022jp_kddi,
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_2022jp_kddi_wchar = {
64 mbfl_no_encoding_2022jp_kddi,
65 mbfl_no_encoding_wchar,
66 mbfl_filt_conv_common_ctor,
67 mbfl_filt_conv_common_dtor,
68 mbfl_filt_conv_2022jp_mobile_wchar,
69 mbfl_filt_conv_common_flush
70 };
71
72 const struct mbfl_convert_vtbl vtbl_wchar_2022jp_kddi = {
73 mbfl_no_encoding_wchar,
74 mbfl_no_encoding_2022jp_kddi,
75 mbfl_filt_conv_common_ctor,
76 mbfl_filt_conv_common_dtor,
77 mbfl_filt_conv_wchar_2022jp_mobile,
78 mbfl_filt_conv_any_jis_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 #define SJIS_ENCODE(c1,c2,s1,s2) \
91 do { \
92 s1 = c1; \
93 s1--; \
94 s1 >>= 1; \
95 if ((c1) < 0x5f) { \
96 s1 += 0x71; \
97 } else { \
98 s1 += 0xb1; \
99 } \
100 s2 = c2; \
101 if ((c1) & 1) { \
102 if ((c2) < 0x60) { \
103 s2--; \
104 } \
105 s2 += 0x20; \
106 } else { \
107 s2 += 0x7e; \
108 } \
109 } while (0)
110
111 #define SJIS_DECODE(c1,c2,s1,s2) \
112 do { \
113 s1 = c1; \
114 if (s1 < 0xa0) { \
115 s1 -= 0x81; \
116 } else { \
117 s1 -= 0xc1; \
118 } \
119 s1 <<= 1; \
120 s1 += 0x21; \
121 s2 = c2; \
122 if (s2 < 0x9f) { \
123 if (s2 < 0x7f) { \
124 s2++; \
125 } \
126 s2 -= 0x20; \
127 } else { \
128 s1++; \
129 s2 -= 0x7e; \
130 } \
131 } while (0)
132
133 #define CODE2JIS(c1,c2,s1,s2) \
134 c1 = (s1)/94+0x21; \
135 c2 = (s1)-94*((c1)-0x21)+0x21; \
136 s1 = ((c1) << 8) | (c2); \
137 s2 = 1
138
139 /*
140 * ISO-2022-JP-Mobile => wchar
141 */
142 int
143 mbfl_filt_conv_2022jp_mobile_wchar(int c, mbfl_convert_filter *filter)
144 {
145 int c1, s, w, snd = 0;
146
147 retry:
148 switch (filter->status & 0xf) {
149 /* case 0x00: ASCII */
150 /* case 0x10: X 0201 latin */
151 /* case 0x20: X 0201 kana */
152 /* case 0x80: X 0208 */
153 case 0:
154 if (c == 0x1b) {
155 filter->status += 2;
156 } else if (filter->status == 0x20 && c > 0x20 && c < 0x60) { /* kana */
157 CK((*filter->output_function)(0xff40 + c, filter->data));
158 } else if (filter->status == 0x80 && c > 0x20 && c < 0x80) { /* kanji first char */
159 filter->cache = c;
160 filter->status += 1;
161 } else if (c >= 0 && c < 0x80) { /* latin, CTLs */
162 CK((*filter->output_function)(c, filter->data));
163 } else if (c > 0xa0 && c < 0xe0) { /* GR kana */
164 CK((*filter->output_function)(0xfec0 + c, filter->data));
165 } else {
166 w = c & MBFL_WCSGROUP_MASK;
167 w |= MBFL_WCSGROUP_THROUGH;
168 CK((*filter->output_function)(w, filter->data));
169 }
170 break;
171
172 /* case 0x81: X 0208 second char */
173 case 1:
174 w = 0;
175 filter->status &= ~0xf;
176 c1 = filter->cache;
177 if (c > 0x20 && c < 0x7f) {
178 s = (c1 - 0x21)*94 + c - 0x21;
179
180 if (s <= 137) {
181 if (s == 31) {
182 w = 0xff3c; /* FULLWIDTH REVERSE SOLIDUS */
183 } else if (s == 32) {
184 w = 0xff5e; /* FULLWIDTH TILDE */
185 } else if (s == 33) {
186 w = 0x2225; /* PARALLEL TO */
187 } else if (s == 60) {
188 w = 0xff0d; /* FULLWIDTH HYPHEN-MINUS */
189 } else if (s == 80) {
190 w = 0xffe0; /* FULLWIDTH CENT SIGN */
191 } else if (s == 81) {
192 w = 0xffe1; /* FULLWIDTH POUND SIGN */
193 } else if (s == 137) {
194 w = 0xffe2; /* FULLWIDTH NOT SIGN */
195 }
196 }
197
198 if (w == 0) {
199 if (s >= cp932ext1_ucs_table_min && s < cp932ext1_ucs_table_max) { /* vendor ext1 (13ku) */
200 w = cp932ext1_ucs_table[s - cp932ext1_ucs_table_min];
201 } else if (s >= 0 && s < jisx0208_ucs_table_size) {
202 w = jisx0208_ucs_table[s];
203 } else {
204 w = 0;
205 }
206 }
207
208 if (s >= (84*94) && s < 91*94) {
209 s += 22*94;
210 if (filter->from->no_encoding == mbfl_no_encoding_2022jp_kddi) {
211 w = mbfilter_sjis_emoji_kddi2unicode(s, &snd);
212 }
213 if (w > 0 && snd > 0) {
214 CK((*filter->output_function)(snd, filter->data));
215 }
216 }
217
218 if (w <= 0) {
219 w = (c1 << 8) | c;
220 w &= MBFL_WCSPLANE_MASK;
221 w |= MBFL_WCSPLANE_JIS0208;
222 }
223 CK((*filter->output_function)(w, filter->data));
224 } else if (c == 0x1b) {
225 filter->status += 2;
226 } else if ((c >= 0 && c < 0x21) || c == 0x7f) { /* CTLs */
227 CK((*filter->output_function)(c, filter->data));
228 } else {
229 w = (c1 << 8) | c;
230 w &= MBFL_WCSGROUP_MASK;
231 w |= MBFL_WCSGROUP_THROUGH;
232 CK((*filter->output_function)(w, filter->data));
233 }
234 break;
235
236 /* ESC */
237 /* case 0x02: */
238 /* case 0x12: */
239 /* case 0x22: */
240 /* case 0x82: */
241 case 2:
242 if (c == 0x24) { /* '$' */
243 filter->status++;
244 } else if (c == 0x28) { /* '(' */
245 filter->status += 3;
246 } else {
247 filter->status &= ~0xf;
248 CK((*filter->output_function)(0x1b, filter->data));
249 goto retry;
250 }
251 break;
252
253 /* ESC $ */
254 /* case 0x03: */
255 /* case 0x13: */
256 /* case 0x23: */
257 /* case 0x83: */
258 case 3:
259 if (c == 0x40 || c == 0x42) { /* '@' or 'B' */
260 filter->status = 0x80;
261 } else if (c == 0x28) { /* '(' */
262 filter->status++;
263 } else {
264 filter->status &= ~0xf;
265 CK((*filter->output_function)(0x1b, filter->data));
266 CK((*filter->output_function)(0x24, filter->data));
267 goto retry;
268 }
269 break;
270
271 /* ESC $ ( */
272 /* case 0x04: */
273 /* case 0x14: */
274 /* case 0x24: */
275 /* case 0x84: */
276 case 4:
277 if (c == 0x40 || c == 0x42) { /* '@' or 'B' */
278 filter->status = 0x80;
279 } else {
280 filter->status &= ~0xf;
281 CK((*filter->output_function)(0x1b, filter->data));
282 CK((*filter->output_function)(0x24, filter->data));
283 CK((*filter->output_function)(0x28, filter->data));
284 goto retry;
285 }
286 break;
287
288 /* ESC ( */
289 /* case 0x05: */
290 /* case 0x15: */
291 /* case 0x25: */
292 /* case 0x85: */
293 case 5:
294 if (c == 0x42) { /* 'B' */
295 filter->status = 0;
296 } else if (c == 0x4a) { /* 'J' */
297 filter->status = 0;
298 } else if (c == 0x49) { /* 'I' */
299 filter->status = 0x20;
300 } else {
301 filter->status &= ~0xf;
302 CK((*filter->output_function)(0x1b, filter->data));
303 CK((*filter->output_function)(0x28, filter->data));
304 goto retry;
305 }
306 break;
307
308 default:
309 filter->status = 0;
310 break;
311 }
312
313 return c;
314 }
315
316 /*
317 * wchar => ISO-2022-JP-Mobile
318 */
319 int
320 mbfl_filt_conv_wchar_2022jp_mobile(int c, mbfl_convert_filter *filter)
321 {
322 int c1, c2, s1, s2;
323
324 s1 = 0;
325 s2 = 0;
326 if (c >= ucs_a1_jis_table_min && c < ucs_a1_jis_table_max) {
327 s1 = ucs_a1_jis_table[c - ucs_a1_jis_table_min];
328 } else if (c >= ucs_a2_jis_table_min && c < ucs_a2_jis_table_max) {
329 s1 = ucs_a2_jis_table[c - ucs_a2_jis_table_min];
330 } else if (c >= ucs_i_jis_table_min && c < ucs_i_jis_table_max) {
331 s1 = ucs_i_jis_table[c - ucs_i_jis_table_min];
332 } else if (c >= ucs_r_jis_table_min && c < ucs_r_jis_table_max) {
333 s1 = ucs_r_jis_table[c - ucs_r_jis_table_min];
334 } else if (c >= 0xe000 && c < (0xe000 + 20*94)) { /* user (95ku - 114ku) */
335 s1 = c - 0xe000;
336 c1 = s1/94 + 0x7f;
337 c2 = s1%94 + 0x21;
338 s1 = (c1 << 8) | c2;
339 }
340 if (s1 <= 0) {
341 c1 = c & ~MBFL_WCSPLANE_MASK;
342 if (c1 == MBFL_WCSPLANE_WINCP932) {
343 s1 = c & MBFL_WCSPLANE_MASK;
344 s2 = 1;
345 } else if (c1 == MBFL_WCSPLANE_JIS0208) {
346 s1 = c & MBFL_WCSPLANE_MASK;
347 } else if (c1 == MBFL_WCSPLANE_JIS0212) {
348 s1 = c & MBFL_WCSPLANE_MASK;
349 s1 |= 0x8080;
350 } else if (c == 0xa5) { /* YEN SIGN */
351 s1 = 0x216f; /* FULLWIDTH YEN SIGN */
352 } else if (c == 0x203e) { /* OVER LINE */
353 s1 = 0x2131; /* FULLWIDTH MACRON */
354 } else if (c == 0xff3c) { /* FULLWIDTH REVERSE SOLIDUS */
355 s1 = 0x2140;
356 } else if (c == 0xff5e) { /* FULLWIDTH TILDE */
357 s1 = 0x2141;
358 } else if (c == 0x2225) { /* PARALLEL TO */
359 s1 = 0x2142;
360 } else if (c == 0xff0d) { /* FULLWIDTH HYPHEN-MINUS */
361 s1 = 0x215d;
362 } else if (c == 0xffe0) { /* FULLWIDTH CENT SIGN */
363 s1 = 0x2171;
364 } else if (c == 0xffe1) { /* FULLWIDTH POUND SIGN */
365 s1 = 0x2172;
366 } else if (c == 0xffe2) { /* FULLWIDTH NOT SIGN */
367 s1 = 0x224c;
368 }
369 }
370
371 if ((s1 <= 0) || (s1 >= 0xa1a1 && s2 == 0)) { /* not found or X 0212 */
372 s1 = -1;
373 c1 = 0;
374 c2 = cp932ext1_ucs_table_max - cp932ext1_ucs_table_min;
375 while (c1 < c2) { /* CP932 vendor ext1 (13ku) */
376 if (c == cp932ext1_ucs_table[c1]) {
377 s1 = ((c1/94 + 0x2d) << 8) + (c1%94 + 0x21);
378 break;
379 }
380 c1++;
381 }
382 if (c == 0) {
383 s1 = 0;
384 } else if (s1 <= 0) {
385 s1 = -1;
386 }
387 }
388
389 if (filter->to->no_encoding == mbfl_no_encoding_2022jp_kddi &&
390 mbfilter_unicode2sjis_emoji_kddi(c, &s1, filter) > 0) {
391 CODE2JIS(c1,c2,s1,s2);
392 s1 -= 0x1600;
393 }
394
395 if (filter->status == 1 && filter->cache > 0) {
396 return c;
397 }
398
399 if (s1 >= 0) {
400 if (s1 < 0x80) { /* latin */
401 if ((filter->status & 0xff00) != 0) {
402 CK((*filter->output_function)(0x1b, filter->data)); /* ESC */
403 CK((*filter->output_function)(0x28, filter->data)); /* '(' */
404 CK((*filter->output_function)(0x42, filter->data)); /* 'B' */
405 }
406 CK((*filter->output_function)(s1, filter->data));
407 filter->status = 0;
408 } else if (s1 > 0xa0 && s1 < 0xe0) { /* kana */
409 if ((filter->status & 0xff00) != 0x100) {
410 CK((*filter->output_function)(0x1b, filter->data)); /* ESC */
411 CK((*filter->output_function)(0x28, filter->data)); /* '(' */
412 CK((*filter->output_function)(0x49, filter->data)); /* 'I' */
413 }
414 filter->status = 0x100;
415 CK((*filter->output_function)(s1 & 0x7f, filter->data));
416 } else if (s1 < 0x7e7f) { /* X 0208 */
417 if ((filter->status & 0xff00) != 0x200) {
418 CK((*filter->output_function)(0x1b, filter->data)); /* ESC */
419 CK((*filter->output_function)(0x24, filter->data)); /* '$' */
420 CK((*filter->output_function)(0x42, filter->data)); /* 'B' */
421 }
422 filter->status = 0x200;
423 CK((*filter->output_function)((s1 >> 8) & 0xff, filter->data));
424 CK((*filter->output_function)(s1 & 0x7f, filter->data));
425 }
426 } else {
427 if (filter->illegal_mode != MBFL_OUTPUTFILTER_ILLEGAL_MODE_NONE) {
428 CK(mbfl_filt_conv_illegal_output(c, filter));
429 }
430 }
431
432 return c;
433 }