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 #include "mbfilter.h"
31 #include "mbfilter_iso2022jp_mobile.h"
32 #include "mbfilter_sjis_mobile.h"
33
34 #include "unicode_table_cp932_ext.h"
35 #include "unicode_table_jis.h"
36 #include "cp932_table.h"
37
38 extern int mbfl_filt_conv_any_jis_flush(mbfl_convert_filter *filter);
39 extern int mbfl_filt_ident_2022jpms(int c, mbfl_identify_filter *filter);
40
41 static const char *mbfl_encoding_2022jp_kddi_aliases[] = {"ISO-2022-JP-KDDI", NULL};
42
43 const mbfl_encoding mbfl_encoding_2022jp_kddi = {
44 mbfl_no_encoding_2022jp_kddi,
45 "ISO-2022-JP-MOBILE#KDDI",
46 "ISO-2022-JP",
47 (const char *(*)[])&mbfl_encoding_2022jp_kddi_aliases,
48 NULL,
49 MBFL_ENCTYPE_MBCS | MBFL_ENCTYPE_GL_UNSAFE,
50 &vtbl_2022jp_kddi_wchar,
51 &vtbl_wchar_2022jp_kddi
52 };
53
54 const struct mbfl_identify_vtbl vtbl_identify_2022jp_kddi = {
55 mbfl_no_encoding_2022jp_kddi,
56 mbfl_filt_ident_common_ctor,
57 mbfl_filt_ident_2022jpms
58 };
59
60 const struct mbfl_convert_vtbl vtbl_2022jp_kddi_wchar = {
61 mbfl_no_encoding_2022jp_kddi,
62 mbfl_no_encoding_wchar,
63 mbfl_filt_conv_common_ctor,
64 NULL,
65 mbfl_filt_conv_2022jp_mobile_wchar,
66 mbfl_filt_conv_common_flush,
67 NULL,
68 };
69
70 const struct mbfl_convert_vtbl vtbl_wchar_2022jp_kddi = {
71 mbfl_no_encoding_wchar,
72 mbfl_no_encoding_2022jp_kddi,
73 mbfl_filt_conv_common_ctor,
74 NULL,
75 mbfl_filt_conv_wchar_2022jp_mobile,
76 mbfl_filt_conv_any_jis_flush,
77 NULL,
78 };
79
80 #define CK(statement) do { if ((statement) < 0) return (-1); } while (0)
81
82 #define sjistoidx(c1, c2) \
83 (((c1) > 0x9f) \
84 ? (((c1) - 0xc1) * 188 + (c2) - (((c2) > 0x7e) ? 0x41 : 0x40)) \
85 : (((c1) - 0x81) * 188 + (c2) - (((c2) > 0x7e) ? 0x41 : 0x40)))
86 #define idxtojis1(c) (((c) / 94) + 0x21)
87 #define idxtojis2(c) (((c) % 94) + 0x21)
88
89 #define SJIS_ENCODE(c1,c2,s1,s2) \
90 do { \
91 s1 = c1; \
92 s1--; \
93 s1 >>= 1; \
94 if ((c1) < 0x5f) { \
95 s1 += 0x71; \
96 } else { \
97 s1 += 0xb1; \
98 } \
99 s2 = c2; \
100 if ((c1) & 1) { \
101 if ((c2) < 0x60) { \
102 s2--; \
103 } \
104 s2 += 0x20; \
105 } else { \
106 s2 += 0x7e; \
107 } \
108 } while (0)
109
110 #define SJIS_DECODE(c1,c2,s1,s2) \
111 do { \
112 s1 = c1; \
113 if (s1 < 0xa0) { \
114 s1 -= 0x81; \
115 } else { \
116 s1 -= 0xc1; \
117 } \
118 s1 <<= 1; \
119 s1 += 0x21; \
120 s2 = c2; \
121 if (s2 < 0x9f) { \
122 if (s2 < 0x7f) { \
123 s2++; \
124 } \
125 s2 -= 0x20; \
126 } else { \
127 s1++; \
128 s2 -= 0x7e; \
129 } \
130 } while (0)
131
132 #define CODE2JIS(c1,c2,s1,s2) \
133 c1 = (s1)/94+0x21; \
134 c2 = (s1)-94*((c1)-0x21)+0x21; \
135 s1 = ((c1) << 8) | (c2); \
136 s2 = 1
137
138 /*
139 * ISO-2022-JP-Mobile => wchar
140 */
141 int
mbfl_filt_conv_2022jp_mobile_wchar(int c,mbfl_convert_filter * filter)142 mbfl_filt_conv_2022jp_mobile_wchar(int c, mbfl_convert_filter *filter)
143 {
144 int c1, s, w, snd = 0;
145
146 retry:
147 switch (filter->status & 0xf) {
148 /* case 0x00: ASCII */
149 /* case 0x10: X 0201 latin */
150 /* case 0x20: X 0201 kana */
151 /* case 0x80: X 0208 */
152 case 0:
153 if (c == 0x1b) {
154 filter->status += 2;
155 } else if (filter->status == 0x20 && c > 0x20 && c < 0x60) { /* kana */
156 CK((*filter->output_function)(0xff40 + c, filter->data));
157 } else if (filter->status == 0x80 && c > 0x20 && c < 0x80) { /* kanji first char */
158 filter->cache = c;
159 filter->status += 1;
160 } else if (c >= 0 && c < 0x80) { /* latin, CTLs */
161 CK((*filter->output_function)(c, filter->data));
162 } else if (c > 0xa0 && c < 0xe0) { /* GR kana */
163 CK((*filter->output_function)(0xfec0 + c, filter->data));
164 } else {
165 w = c & MBFL_WCSGROUP_MASK;
166 w |= MBFL_WCSGROUP_THROUGH;
167 CK((*filter->output_function)(w, filter->data));
168 }
169 break;
170
171 /* case 0x81: X 0208 second char */
172 case 1:
173 w = 0;
174 filter->status &= ~0xf;
175 c1 = filter->cache;
176 if (c > 0x20 && c < 0x7f) {
177 s = (c1 - 0x21)*94 + c - 0x21;
178
179 if (s <= 137) {
180 if (s == 31) {
181 w = 0xff3c; /* FULLWIDTH REVERSE SOLIDUS */
182 } else if (s == 32) {
183 w = 0xff5e; /* FULLWIDTH TILDE */
184 } else if (s == 33) {
185 w = 0x2225; /* PARALLEL TO */
186 } else if (s == 60) {
187 w = 0xff0d; /* FULLWIDTH HYPHEN-MINUS */
188 } else if (s == 80) {
189 w = 0xffe0; /* FULLWIDTH CENT SIGN */
190 } else if (s == 81) {
191 w = 0xffe1; /* FULLWIDTH POUND SIGN */
192 } else if (s == 137) {
193 w = 0xffe2; /* FULLWIDTH NOT SIGN */
194 }
195 }
196
197 if (w == 0) {
198 if (s >= cp932ext1_ucs_table_min && s < cp932ext1_ucs_table_max) { /* vendor ext1 (13ku) */
199 w = cp932ext1_ucs_table[s - cp932ext1_ucs_table_min];
200 } else if (s >= 0 && s < jisx0208_ucs_table_size) {
201 w = jisx0208_ucs_table[s];
202 } else {
203 w = 0;
204 }
205 }
206
207 if (s >= (84*94) && s < 91*94) {
208 s += 22*94;
209 if (filter->from->no_encoding == mbfl_no_encoding_2022jp_kddi) {
210 w = mbfilter_sjis_emoji_kddi2unicode(s, &snd);
211 }
212 if (w > 0 && snd > 0) {
213 CK((*filter->output_function)(snd, filter->data));
214 }
215 }
216
217 if (w <= 0) {
218 w = (c1 << 8) | c;
219 w &= MBFL_WCSPLANE_MASK;
220 w |= MBFL_WCSPLANE_JIS0208;
221 }
222 CK((*filter->output_function)(w, filter->data));
223 } else if (c == 0x1b) {
224 filter->status += 2;
225 } else if ((c >= 0 && c < 0x21) || c == 0x7f) { /* CTLs */
226 CK((*filter->output_function)(c, filter->data));
227 } else {
228 w = (c1 << 8) | c;
229 w &= MBFL_WCSGROUP_MASK;
230 w |= MBFL_WCSGROUP_THROUGH;
231 CK((*filter->output_function)(w, filter->data));
232 }
233 break;
234
235 /* ESC */
236 /* case 0x02: */
237 /* case 0x12: */
238 /* case 0x22: */
239 /* case 0x82: */
240 case 2:
241 if (c == 0x24) { /* '$' */
242 filter->status++;
243 } else if (c == 0x28) { /* '(' */
244 filter->status += 3;
245 } else {
246 filter->status &= ~0xf;
247 CK((*filter->output_function)(0x1b, filter->data));
248 goto retry;
249 }
250 break;
251
252 /* ESC $ */
253 /* case 0x03: */
254 /* case 0x13: */
255 /* case 0x23: */
256 /* case 0x83: */
257 case 3:
258 if (c == 0x40 || c == 0x42) { /* '@' or 'B' */
259 filter->status = 0x80;
260 } else if (c == 0x28) { /* '(' */
261 filter->status++;
262 } else {
263 filter->status &= ~0xf;
264 CK((*filter->output_function)(0x1b, filter->data));
265 CK((*filter->output_function)(0x24, filter->data));
266 goto retry;
267 }
268 break;
269
270 /* ESC $ ( */
271 /* case 0x04: */
272 /* case 0x14: */
273 /* case 0x24: */
274 /* case 0x84: */
275 case 4:
276 if (c == 0x40 || c == 0x42) { /* '@' or 'B' */
277 filter->status = 0x80;
278 } else {
279 filter->status &= ~0xf;
280 CK((*filter->output_function)(0x1b, filter->data));
281 CK((*filter->output_function)(0x24, filter->data));
282 CK((*filter->output_function)(0x28, filter->data));
283 goto retry;
284 }
285 break;
286
287 /* ESC ( */
288 /* case 0x05: */
289 /* case 0x15: */
290 /* case 0x25: */
291 /* case 0x85: */
292 case 5:
293 if (c == 0x42) { /* 'B' */
294 filter->status = 0;
295 } else if (c == 0x4a) { /* 'J' */
296 filter->status = 0;
297 } else if (c == 0x49) { /* 'I' */
298 filter->status = 0x20;
299 } else {
300 filter->status &= ~0xf;
301 CK((*filter->output_function)(0x1b, filter->data));
302 CK((*filter->output_function)(0x28, filter->data));
303 goto retry;
304 }
305 break;
306
307 default:
308 filter->status = 0;
309 break;
310 }
311
312 return c;
313 }
314
315 /*
316 * wchar => ISO-2022-JP-Mobile
317 */
318 int
mbfl_filt_conv_wchar_2022jp_mobile(int c,mbfl_convert_filter * filter)319 mbfl_filt_conv_wchar_2022jp_mobile(int c, mbfl_convert_filter *filter)
320 {
321 int c1, c2, s1, s2;
322
323 s1 = 0;
324 s2 = 0;
325 if (c >= ucs_a1_jis_table_min && c < ucs_a1_jis_table_max) {
326 s1 = ucs_a1_jis_table[c - ucs_a1_jis_table_min];
327 } else if (c >= ucs_a2_jis_table_min && c < ucs_a2_jis_table_max) {
328 s1 = ucs_a2_jis_table[c - ucs_a2_jis_table_min];
329 } else if (c >= ucs_i_jis_table_min && c < ucs_i_jis_table_max) {
330 s1 = ucs_i_jis_table[c - ucs_i_jis_table_min];
331 } else if (c >= ucs_r_jis_table_min && c < ucs_r_jis_table_max) {
332 s1 = ucs_r_jis_table[c - ucs_r_jis_table_min];
333 } else if (c >= 0xe000 && c < (0xe000 + 20*94)) { /* user (95ku - 114ku) */
334 s1 = c - 0xe000;
335 c1 = s1/94 + 0x7f;
336 c2 = s1%94 + 0x21;
337 s1 = (c1 << 8) | c2;
338 }
339 if (s1 <= 0) {
340 c1 = c & ~MBFL_WCSPLANE_MASK;
341 if (c1 == MBFL_WCSPLANE_WINCP932) {
342 s1 = c & MBFL_WCSPLANE_MASK;
343 s2 = 1;
344 } else if (c1 == MBFL_WCSPLANE_JIS0208) {
345 s1 = c & MBFL_WCSPLANE_MASK;
346 } else if (c1 == MBFL_WCSPLANE_JIS0212) {
347 s1 = c & MBFL_WCSPLANE_MASK;
348 s1 |= 0x8080;
349 } else if (c == 0xa5) { /* YEN SIGN */
350 s1 = 0x216f; /* FULLWIDTH YEN SIGN */
351 } else if (c == 0x203e) { /* OVER LINE */
352 s1 = 0x2131; /* FULLWIDTH MACRON */
353 } else if (c == 0xff3c) { /* FULLWIDTH REVERSE SOLIDUS */
354 s1 = 0x2140;
355 } else if (c == 0xff5e) { /* FULLWIDTH TILDE */
356 s1 = 0x2141;
357 } else if (c == 0x2225) { /* PARALLEL TO */
358 s1 = 0x2142;
359 } else if (c == 0xff0d) { /* FULLWIDTH HYPHEN-MINUS */
360 s1 = 0x215d;
361 } else if (c == 0xffe0) { /* FULLWIDTH CENT SIGN */
362 s1 = 0x2171;
363 } else if (c == 0xffe1) { /* FULLWIDTH POUND SIGN */
364 s1 = 0x2172;
365 } else if (c == 0xffe2) { /* FULLWIDTH NOT SIGN */
366 s1 = 0x224c;
367 }
368 }
369
370 if ((s1 <= 0) || (s1 >= 0xa1a1 && s2 == 0)) { /* not found or X 0212 */
371 s1 = -1;
372 c1 = 0;
373 c2 = cp932ext1_ucs_table_max - cp932ext1_ucs_table_min;
374 while (c1 < c2) { /* CP932 vendor ext1 (13ku) */
375 if (c == cp932ext1_ucs_table[c1]) {
376 s1 = ((c1/94 + 0x2d) << 8) + (c1%94 + 0x21);
377 break;
378 }
379 c1++;
380 }
381 if (c == 0) {
382 s1 = 0;
383 } else if (s1 <= 0) {
384 s1 = -1;
385 }
386 }
387
388 if (filter->to->no_encoding == mbfl_no_encoding_2022jp_kddi &&
389 mbfilter_unicode2sjis_emoji_kddi(c, &s1, filter) > 0) {
390 CODE2JIS(c1,c2,s1,s2);
391 s1 -= 0x1600;
392 }
393
394 if (filter->status == 1 && filter->cache > 0) {
395 return c;
396 }
397
398 if (s1 >= 0) {
399 if (s1 < 0x80) { /* latin */
400 if ((filter->status & 0xff00) != 0) {
401 CK((*filter->output_function)(0x1b, filter->data)); /* ESC */
402 CK((*filter->output_function)(0x28, filter->data)); /* '(' */
403 CK((*filter->output_function)(0x42, filter->data)); /* 'B' */
404 }
405 CK((*filter->output_function)(s1, filter->data));
406 filter->status = 0;
407 } else if (s1 > 0xa0 && s1 < 0xe0) { /* kana */
408 if ((filter->status & 0xff00) != 0x100) {
409 CK((*filter->output_function)(0x1b, filter->data)); /* ESC */
410 CK((*filter->output_function)(0x28, filter->data)); /* '(' */
411 CK((*filter->output_function)(0x49, filter->data)); /* 'I' */
412 }
413 filter->status = 0x100;
414 CK((*filter->output_function)(s1 & 0x7f, filter->data));
415 } else if (s1 < 0x7e7f) { /* X 0208 */
416 if ((filter->status & 0xff00) != 0x200) {
417 CK((*filter->output_function)(0x1b, filter->data)); /* ESC */
418 CK((*filter->output_function)(0x24, filter->data)); /* '$' */
419 CK((*filter->output_function)(0x42, filter->data)); /* 'B' */
420 }
421 filter->status = 0x200;
422 CK((*filter->output_function)((s1 >> 8) & 0xff, filter->data));
423 CK((*filter->output_function)(s1 & 0x7f, filter->data));
424 }
425 } else {
426 CK(mbfl_filt_conv_illegal_output(c, filter));
427 }
428
429 return c;
430 }
431