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.c
26 * by rui hrokawa <hirokawa@php.net> on 8 aug 2011.
27 *
28 */
29
30 #ifdef HAVE_CONFIG_H
31 #include "config.h"
32 #endif
33
34 #include "mbfilter.h"
35
36 #include "mbfilter_utf8_mobile.h"
37 #include "mbfilter_sjis_mobile.h"
38
39 extern int mbfl_filt_ident_utf8(int c, mbfl_identify_filter *filter);
40 extern int mbfl_filt_conv_utf8_wchar_flush(mbfl_convert_filter *filter);
41
42 extern const unsigned char mblen_table_utf8[];
43
44 static const char *mbfl_encoding_utf8_docomo_aliases[] = {"UTF-8-DOCOMO", "UTF8-DOCOMO", NULL};
45 static const char *mbfl_encoding_utf8_kddi_a_aliases[] = {NULL};
46 static const char *mbfl_encoding_utf8_kddi_b_aliases[] = {"UTF-8-Mobile#KDDI", "UTF-8-KDDI", "UTF8-KDDI", NULL};
47 static const char *mbfl_encoding_utf8_sb_aliases[] = {"UTF-8-SOFTBANK", "UTF8-SOFTBANK", NULL};
48
49 const mbfl_encoding mbfl_encoding_utf8_docomo = {
50 mbfl_no_encoding_utf8_docomo,
51 "UTF-8-Mobile#DOCOMO",
52 "UTF-8",
53 (const char *(*)[])&mbfl_encoding_utf8_docomo_aliases,
54 mblen_table_utf8,
55 MBFL_ENCTYPE_MBCS
56 };
57
58 const mbfl_encoding mbfl_encoding_utf8_kddi_a = {
59 mbfl_no_encoding_utf8_kddi_a,
60 "UTF-8-Mobile#KDDI-A",
61 "UTF-8",
62 (const char *(*)[])&mbfl_encoding_utf8_kddi_a_aliases,
63 mblen_table_utf8,
64 MBFL_ENCTYPE_MBCS
65 };
66
67 const mbfl_encoding mbfl_encoding_utf8_kddi_b = {
68 mbfl_no_encoding_utf8_kddi_b,
69 "UTF-8-Mobile#KDDI-B",
70 "UTF-8",
71 (const char *(*)[])&mbfl_encoding_utf8_kddi_b_aliases,
72 mblen_table_utf8,
73 MBFL_ENCTYPE_MBCS
74 };
75
76 const mbfl_encoding mbfl_encoding_utf8_sb = {
77 mbfl_no_encoding_utf8_sb,
78 "UTF-8-Mobile#SOFTBANK",
79 "UTF-8",
80 (const char *(*)[])&mbfl_encoding_utf8_sb_aliases,
81 mblen_table_utf8,
82 MBFL_ENCTYPE_MBCS
83 };
84
85 const struct mbfl_identify_vtbl vtbl_identify_utf8_docomo = {
86 mbfl_no_encoding_utf8_docomo,
87 mbfl_filt_ident_common_ctor,
88 mbfl_filt_ident_common_dtor,
89 mbfl_filt_ident_utf8
90 };
91
92 const struct mbfl_identify_vtbl vtbl_identify_utf8_kddi_a = {
93 mbfl_no_encoding_utf8_kddi_a,
94 mbfl_filt_ident_common_ctor,
95 mbfl_filt_ident_common_dtor,
96 mbfl_filt_ident_utf8
97 };
98
99 const struct mbfl_identify_vtbl vtbl_identify_utf8_kddi_b = {
100 mbfl_no_encoding_utf8_kddi_b,
101 mbfl_filt_ident_common_ctor,
102 mbfl_filt_ident_common_dtor,
103 mbfl_filt_ident_utf8
104 };
105
106 const struct mbfl_identify_vtbl vtbl_identify_utf8_sb = {
107 mbfl_no_encoding_utf8_sb,
108 mbfl_filt_ident_common_ctor,
109 mbfl_filt_ident_common_dtor,
110 mbfl_filt_ident_utf8
111 };
112
113 const struct mbfl_convert_vtbl vtbl_utf8_docomo_wchar = {
114 mbfl_no_encoding_utf8_docomo,
115 mbfl_no_encoding_wchar,
116 mbfl_filt_conv_common_ctor,
117 mbfl_filt_conv_common_dtor,
118 mbfl_filt_conv_utf8_mobile_wchar,
119 mbfl_filt_conv_utf8_wchar_flush
120 };
121
122 const struct mbfl_convert_vtbl vtbl_wchar_utf8_docomo = {
123 mbfl_no_encoding_wchar,
124 mbfl_no_encoding_utf8_docomo,
125 mbfl_filt_conv_common_ctor,
126 mbfl_filt_conv_common_dtor,
127 mbfl_filt_conv_wchar_utf8_mobile,
128 mbfl_filt_conv_common_flush
129 };
130
131 const struct mbfl_convert_vtbl vtbl_utf8_kddi_a_wchar = {
132 mbfl_no_encoding_utf8_kddi_a,
133 mbfl_no_encoding_wchar,
134 mbfl_filt_conv_common_ctor,
135 mbfl_filt_conv_common_dtor,
136 mbfl_filt_conv_utf8_mobile_wchar,
137 mbfl_filt_conv_utf8_wchar_flush
138 };
139
140 const struct mbfl_convert_vtbl vtbl_wchar_utf8_kddi_a = {
141 mbfl_no_encoding_wchar,
142 mbfl_no_encoding_utf8_kddi_a,
143 mbfl_filt_conv_common_ctor,
144 mbfl_filt_conv_common_dtor,
145 mbfl_filt_conv_wchar_utf8_mobile,
146 mbfl_filt_conv_common_flush
147 };
148
149 const struct mbfl_convert_vtbl vtbl_utf8_kddi_b_wchar = {
150 mbfl_no_encoding_utf8_kddi_b,
151 mbfl_no_encoding_wchar,
152 mbfl_filt_conv_common_ctor,
153 mbfl_filt_conv_common_dtor,
154 mbfl_filt_conv_utf8_mobile_wchar,
155 mbfl_filt_conv_utf8_wchar_flush
156 };
157
158 const struct mbfl_convert_vtbl vtbl_wchar_utf8_kddi_b = {
159 mbfl_no_encoding_wchar,
160 mbfl_no_encoding_utf8_kddi_b,
161 mbfl_filt_conv_common_ctor,
162 mbfl_filt_conv_common_dtor,
163 mbfl_filt_conv_wchar_utf8_mobile,
164 mbfl_filt_conv_common_flush
165 };
166
167 const struct mbfl_convert_vtbl vtbl_utf8_sb_wchar = {
168 mbfl_no_encoding_utf8_sb,
169 mbfl_no_encoding_wchar,
170 mbfl_filt_conv_common_ctor,
171 mbfl_filt_conv_common_dtor,
172 mbfl_filt_conv_utf8_mobile_wchar,
173 mbfl_filt_conv_utf8_wchar_flush
174 };
175
176 const struct mbfl_convert_vtbl vtbl_wchar_utf8_sb = {
177 mbfl_no_encoding_wchar,
178 mbfl_no_encoding_utf8_sb,
179 mbfl_filt_conv_common_ctor,
180 mbfl_filt_conv_common_dtor,
181 mbfl_filt_conv_wchar_utf8_mobile,
182 mbfl_filt_conv_common_flush
183 };
184
185 #define CK(statement) do { if ((statement) < 0) return (-1); } while (0)
186 int mbfl_filt_put_invalid_char(int c, mbfl_convert_filter *filter);
187
188 /*
189 * UTF-8 => wchar
190 */
mbfl_filt_conv_utf8_mobile_wchar(int c,mbfl_convert_filter * filter)191 int mbfl_filt_conv_utf8_mobile_wchar(int c, mbfl_convert_filter *filter)
192 {
193 int s, s1 = 0, c1 = 0, snd = 0;
194
195 retry:
196 switch (filter->status & 0xff) {
197 case 0x00:
198 if (c < 0x80) {
199 CK((*filter->output_function)(c, filter->data));
200 } else if (c >= 0xc2 && c <= 0xdf) { /* 2byte code first char: 0xc2-0xdf */
201 filter->status = 0x10;
202 filter->cache = c & 0x1f;
203 } else if (c >= 0xe0 && c <= 0xef) { /* 3byte code first char: 0xe0-0xef */
204 filter->status = 0x20;
205 filter->cache = c & 0xf;
206 } else if (c >= 0xf0 && c <= 0xf4) { /* 3byte code first char: 0xf0-0xf4 */
207 filter->status = 0x30;
208 filter->cache = c & 0x7;
209 } else {
210 CK(mbfl_filt_put_invalid_char(c, filter));
211 }
212 break;
213 case 0x10: /* 2byte code 2nd char: 0x80-0xbf */
214 case 0x21: /* 3byte code 3rd char: 0x80-0xbf */
215 case 0x32: /* 4byte code 4th char: 0x80-0xbf */
216 filter->status = 0;
217 if (c >= 0x80 && c <= 0xbf) {
218 s = (filter->cache<<6) | (c & 0x3f);
219 filter->cache = 0;
220
221 if (filter->from->no_encoding == mbfl_no_encoding_utf8_docomo &&
222 mbfilter_conv_r_map_tbl(s, &s1, mbfl_docomo2uni_pua, 4) > 0) {
223 s = mbfilter_sjis_emoji_docomo2unicode(s1, &snd);
224 } else if (filter->from->no_encoding == mbfl_no_encoding_utf8_kddi_a &&
225 mbfilter_conv_r_map_tbl(s, &s1, mbfl_kddi2uni_pua, 7) > 0) {
226 s = mbfilter_sjis_emoji_kddi2unicode(s1, &snd);
227 } else if (filter->from->no_encoding == mbfl_no_encoding_utf8_kddi_b &&
228 mbfilter_conv_r_map_tbl(s, &s1, mbfl_kddi2uni_pua_b, 8) > 0) {
229 s = mbfilter_sjis_emoji_kddi2unicode(s1, &snd);
230 } else if (filter->from->no_encoding == mbfl_no_encoding_utf8_sb &&
231 mbfilter_conv_r_map_tbl(s, &s1, mbfl_sb2uni_pua, 6) > 0) {
232 s = mbfilter_sjis_emoji_sb2unicode(s1, &snd);
233 }
234
235 if (snd > 0) {
236 CK((*filter->output_function)(snd, filter->data));
237 }
238 CK((*filter->output_function)(s, filter->data));
239 } else {
240 CK(mbfl_filt_put_invalid_char(filter->cache, filter));
241 goto retry;
242 }
243 break;
244 case 0x20: /* 3byte code 2nd char: 0:0xa0-0xbf,D:0x80-9F,1-C,E-F:0x80-0x9f */
245 s = (filter->cache<<6) | (c & 0x3f);
246 c1 = filter->cache & 0xf;
247
248 if ((c >= 0x80 && c <= 0xbf) &&
249 ((c1 == 0x0 && c >= 0xa0) ||
250 (c1 == 0xd && c < 0xa0) ||
251 (c1 > 0x0 && c1 != 0xd))) {
252 filter->cache = s;
253 filter->status++;
254 } else {
255 CK(mbfl_filt_put_invalid_char(filter->cache, filter));
256 goto retry;
257 }
258 break;
259 case 0x30: /* 4byte code 2nd char: 0:0x90-0xbf,1-3:0x80-0xbf,4:0x80-0x8f */
260 s = (filter->cache<<6) | (c & 0x3f);
261 c1 = filter->cache & 0x7;
262
263 if ((c >= 0x80 && c <= 0xbf) &&
264 ((c1 == 0x0 && c >= 0x90) ||
265 (c1 == 0x4 && c < 0x90) ||
266 (c1 > 0x0 && c1 != 0x4))) {
267 filter->cache = s;
268 filter->status++;
269 } else {
270 CK(mbfl_filt_put_invalid_char(filter->cache, filter));
271 goto retry;
272 }
273 break;
274 case 0x31: /* 4byte code 3rd char: 0x80-0xbf */
275 if (c >= 0x80 && c <= 0xbf) {
276 filter->cache = (filter->cache<<6) | (c & 0x3f);
277 filter->status++;
278 } else {
279 CK(mbfl_filt_put_invalid_char(filter->cache, filter));
280 goto retry;
281 }
282 break;
283 default:
284 filter->status = 0;
285 break;
286 }
287
288 return c;
289 }
290
291 /*
292 * wchar => UTF-8
293 */
mbfl_filt_conv_wchar_utf8_mobile(int c,mbfl_convert_filter * filter)294 int mbfl_filt_conv_wchar_utf8_mobile(int c, mbfl_convert_filter *filter)
295 {
296 if (c >= 0 && c < 0x110000) {
297 int s1, c1;
298
299 if ((filter->to->no_encoding == mbfl_no_encoding_utf8_docomo &&
300 mbfilter_unicode2sjis_emoji_docomo(c, &s1, filter) > 0 &&
301 mbfilter_conv_map_tbl(s1, &c1, mbfl_docomo2uni_pua, 4) > 0) ||
302 (filter->to->no_encoding == mbfl_no_encoding_utf8_kddi_a &&
303 mbfilter_unicode2sjis_emoji_kddi(c, &s1, filter) > 0 &&
304 mbfilter_conv_map_tbl(s1, &c1, mbfl_kddi2uni_pua, 7) > 0) ||
305 (filter->to->no_encoding == mbfl_no_encoding_utf8_kddi_b &&
306 mbfilter_unicode2sjis_emoji_kddi(c, &s1, filter) > 0 &&
307 mbfilter_conv_map_tbl(s1, &c1, mbfl_kddi2uni_pua_b, 8) > 0) ||
308 (filter->to->no_encoding == mbfl_no_encoding_utf8_sb &&
309 mbfilter_unicode2sjis_emoji_sb(c, &s1, filter) > 0 &&
310 mbfilter_conv_map_tbl(s1, &c1, mbfl_sb2uni_pua, 6) > 0)) {
311 c = c1;
312 }
313
314 if (filter->status == 1 && filter->cache > 0) {
315 return c;
316 }
317
318 if (c < 0x80) {
319 CK((*filter->output_function)(c, filter->data));
320 } else if (c < 0x800) {
321 CK((*filter->output_function)(((c >> 6) & 0x1f) | 0xc0, filter->data));
322 CK((*filter->output_function)((c & 0x3f) | 0x80, filter->data));
323 } else if (c < 0x10000) {
324 CK((*filter->output_function)(((c >> 12) & 0x0f) | 0xe0, filter->data));
325 CK((*filter->output_function)(((c >> 6) & 0x3f) | 0x80, filter->data));
326 CK((*filter->output_function)((c & 0x3f) | 0x80, filter->data));
327 } else {
328 CK((*filter->output_function)(((c >> 18) & 0x07) | 0xf0, filter->data));
329 CK((*filter->output_function)(((c >> 12) & 0x3f) | 0x80, filter->data));
330 CK((*filter->output_function)(((c >> 6) & 0x3f) | 0x80, filter->data));
331 CK((*filter->output_function)((c & 0x3f) | 0x80, filter->data));
332 }
333 } else {
334 if (filter->illegal_mode != MBFL_OUTPUTFILTER_ILLEGAL_MODE_NONE) {
335 CK(mbfl_filt_conv_illegal_output(c, filter));
336 }
337 }
338
339 return c;
340 }
341