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_kr.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_kr.h"
36 #include "unicode_table_uhc.h"
37
38 static int mbfl_filt_ident_2022kr(int c, mbfl_identify_filter *filter);
39
40 const mbfl_encoding mbfl_encoding_2022kr = {
41 mbfl_no_encoding_2022kr,
42 "ISO-2022-KR",
43 "ISO-2022-KR",
44 NULL,
45 NULL,
46 MBFL_ENCTYPE_MBCS | MBFL_ENCTYPE_SHFTCODE | MBFL_ENCTYPE_GL_UNSAFE
47 };
48
49 const struct mbfl_identify_vtbl vtbl_identify_2022kr = {
50 mbfl_no_encoding_2022kr,
51 mbfl_filt_ident_common_ctor,
52 mbfl_filt_ident_common_dtor,
53 mbfl_filt_ident_2022kr
54 };
55
56 const struct mbfl_convert_vtbl vtbl_wchar_2022kr = {
57 mbfl_no_encoding_wchar,
58 mbfl_no_encoding_2022kr,
59 mbfl_filt_conv_common_ctor,
60 mbfl_filt_conv_common_dtor,
61 mbfl_filt_conv_wchar_2022kr,
62 mbfl_filt_conv_any_2022kr_flush
63 };
64
65 const struct mbfl_convert_vtbl vtbl_2022kr_wchar = {
66 mbfl_no_encoding_2022kr,
67 mbfl_no_encoding_wchar,
68 mbfl_filt_conv_common_ctor,
69 mbfl_filt_conv_common_dtor,
70 mbfl_filt_conv_2022kr_wchar,
71 mbfl_filt_conv_common_flush
72 };
73
74 #define CK(statement) do { if ((statement) < 0) return (-1); } while (0)
75
76 /*
77 * ISO-2022-KR => wchar
78 */
79 int
mbfl_filt_conv_2022kr_wchar(int c,mbfl_convert_filter * filter)80 mbfl_filt_conv_2022kr_wchar(int c, mbfl_convert_filter *filter)
81 {
82 int c1, w, flag;
83
84 retry:
85 switch (filter->status & 0xf) {
86 /* case 0x00: ASCII */
87 /* case 0x10: KSC5601 */
88 case 0:
89 if (c == 0x1b) { /* ESC */
90 filter->status += 2;
91 } else if (c == 0x0f) { /* SI (ASCII) */
92 filter->status &= ~0xff;
93 } else if (c == 0x0e) { /* SO (KSC5601) */
94 filter->status |= 0x10;
95 } else if ((filter->status & 0x10) != 0 && c > 0x20 && c < 0x7f) {
96 /* KSC5601 lead byte */
97 filter->cache = c;
98 filter->status += 1;
99 } else if ((filter->status & 0x10) == 0 && c >= 0 && c < 0x80) {
100 /* latin, CTLs */
101 CK((*filter->output_function)(c, filter->data));
102 } else {
103 w = c & MBFL_WCSGROUP_MASK;
104 w |= MBFL_WCSGROUP_THROUGH;
105 CK((*filter->output_function)(w, filter->data));
106 }
107 break;
108
109 case 1: /* dbcs second byte */
110 filter->status &= ~0xf;
111 c1 = filter->cache;
112 flag = 0;
113 if (c1 > 0x20 && c1 < 0x47) {
114 flag = 1;
115 } else if (c1 >= 0x47 && c1 <= 0x7e && c1 != 0x49) {
116 flag = 2;
117 }
118 if (flag > 0 && c > 0x20 && c < 0x7f) {
119 if (flag == 1){
120 w = (c1 - 0x21)*190 + (c - 0x41) + 0x80;
121 if (w >= 0 && w < uhc2_ucs_table_size) {
122 w = uhc2_ucs_table[w];
123 } else {
124 w = 0;
125 }
126 } else {
127 w = (c1 - 0x47)*94 + (c - 0x21);
128 if (w >= 0 && w < uhc3_ucs_table_size) {
129 w = uhc3_ucs_table[w];
130 } else {
131 w = 0;
132 }
133 }
134
135 if (w <= 0) {
136 w = (c1 << 8) | c;
137 w &= MBFL_WCSPLANE_MASK;
138 w |= MBFL_WCSPLANE_KSC5601;
139 }
140 CK((*filter->output_function)(w, filter->data));
141 } else if (c == 0x1b) { /* ESC */
142 filter->status++;
143 } else if ((c >= 0 && c < 0x21) || c == 0x7f) { /* CTLs */
144 CK((*filter->output_function)(c, filter->data));
145 } else {
146 w = (c1 << 8) | c;
147 w &= MBFL_WCSGROUP_MASK;
148 w |= MBFL_WCSGROUP_THROUGH;
149 CK((*filter->output_function)(w, filter->data));
150 }
151 break;
152
153 case 2: /* ESC */
154 if (c == 0x24) { /* '$' */
155 filter->status++;
156 } else {
157 filter->status &= ~0xf;
158 CK((*filter->output_function)(0x1b, filter->data));
159 goto retry;
160 }
161 break;
162 case 3: /* ESC $ */
163 if (c == 0x29) { /* ')' */
164 filter->status++;
165 } else {
166 filter->status &= ~0xf;
167 CK((*filter->output_function)(0x1b, filter->data));
168 CK((*filter->output_function)(0x24, filter->data));
169 goto retry;
170 }
171 break;
172 case 4: /* ESC $ ) */
173 if (c == 0x43) { /* 'C' */
174 filter->status &= ~0xf;
175 filter->status |= 0x100;
176 } else {
177 filter->status &= ~0xf;
178 CK((*filter->output_function)(0x1b, filter->data));
179 CK((*filter->output_function)(0x24, filter->data));
180 CK((*filter->output_function)(0x29, filter->data));
181 goto retry;
182 }
183 break;
184 default:
185 filter->status = 0;
186 break;
187 }
188
189 return c;
190 }
191
192 /*
193 * wchar => ISO-2022-KR
194 */
195 int
mbfl_filt_conv_wchar_2022kr(int c,mbfl_convert_filter * filter)196 mbfl_filt_conv_wchar_2022kr(int c, mbfl_convert_filter *filter)
197 {
198 int c1, c2, s;
199
200 s = 0;
201
202 if (c >= ucs_a1_uhc_table_min && c < ucs_a1_uhc_table_max) {
203 s = ucs_a1_uhc_table[c - ucs_a1_uhc_table_min];
204 } else if (c >= ucs_a2_uhc_table_min && c < ucs_a2_uhc_table_max) {
205 s = ucs_a2_uhc_table[c - ucs_a2_uhc_table_min];
206 } else if (c >= ucs_a3_uhc_table_min && c < ucs_a3_uhc_table_max) {
207 s = ucs_a3_uhc_table[c - ucs_a3_uhc_table_min];
208 } else if (c >= ucs_i_uhc_table_min && c < ucs_i_uhc_table_max) {
209 s = ucs_i_uhc_table[c - ucs_i_uhc_table_min];
210 } else if (c >= ucs_s_uhc_table_min && c < ucs_s_uhc_table_max) {
211 s = ucs_s_uhc_table[c - ucs_s_uhc_table_min];
212 } else if (c >= ucs_r1_uhc_table_min && c < ucs_r1_uhc_table_max) {
213 s = ucs_r1_uhc_table[c - ucs_r1_uhc_table_min];
214 } else if (c >= ucs_r2_uhc_table_min && c < ucs_r2_uhc_table_max) {
215 s = ucs_r2_uhc_table[c - ucs_r2_uhc_table_min];
216 }
217
218 c1 = (s >> 8) & 0xff;
219 c2 = s & 0xff;
220 /* exclude UHC extension area */
221 if (c1 < 0xa1 || c2 < 0xa1){
222 s = c;
223 }
224 if (s & 0x8000) {
225 s -= 0x8080;
226 }
227
228 if (s <= 0) {
229 c1 = c & ~MBFL_WCSPLANE_MASK;
230 if (c1 == MBFL_WCSPLANE_KSC5601) {
231 s = c & MBFL_WCSPLANE_MASK;
232 }
233 if (c == 0) {
234 s = 0;
235 } else if (s <= 0) {
236 s = -1;
237 }
238 } else if ((s >= 0x80 && s < 0x2121) || (s > 0x8080)) {
239 s = -1;
240 }
241 if (s >= 0) {
242 if (s < 0x80 && s > 0) { /* ASCII */
243 if ((filter->status & 0x10) != 0) {
244 CK((*filter->output_function)(0x0f, filter->data)); /* SI */
245 filter->status &= ~0x10;
246 }
247 CK((*filter->output_function)(s, filter->data));
248 } else {
249 if ( (filter->status & 0x100) == 0) {
250 CK((*filter->output_function)(0x1b, filter->data)); /* ESC */
251 CK((*filter->output_function)(0x24, filter->data)); /* '$' */
252 CK((*filter->output_function)(0x29, filter->data)); /* ')' */
253 CK((*filter->output_function)(0x43, filter->data)); /* 'C' */
254 filter->status |= 0x100;
255 }
256 if ((filter->status & 0x10) == 0) {
257 CK((*filter->output_function)(0x0e, filter->data)); /* SO */
258 filter->status |= 0x10;
259 }
260 CK((*filter->output_function)((s >> 8) & 0xff, filter->data));
261 CK((*filter->output_function)(s & 0xff, filter->data));
262 }
263 } else {
264 if (filter->illegal_mode != MBFL_OUTPUTFILTER_ILLEGAL_MODE_NONE) {
265 CK(mbfl_filt_conv_illegal_output(c, filter));
266 }
267 }
268
269 return c;
270 }
271
272 int
mbfl_filt_conv_any_2022kr_flush(mbfl_convert_filter * filter)273 mbfl_filt_conv_any_2022kr_flush(mbfl_convert_filter *filter)
274 {
275 /* back to ascii */
276 if ((filter->status & 0xff00) != 0) {
277 CK((*filter->output_function)(0x0f, filter->data)); /* SI */
278 }
279
280 filter->status &= 0xff;
281
282 if (filter->flush_function != NULL) {
283 return (*filter->flush_function)(filter->data);
284 }
285
286 return 0;
287 }
288
mbfl_filt_ident_2022kr(int c,mbfl_identify_filter * filter)289 static int mbfl_filt_ident_2022kr(int c, mbfl_identify_filter *filter)
290 {
291 retry:
292 switch (filter->status & 0xf) {
293 /* case 0x00: ASCII */
294 /* case 0x10: KSC5601 mode */
295 /* case 0x20: KSC5601 DBCS */
296 /* case 0x40: KSC5601 SBCS */
297 case 0:
298 if (!(filter->status & 0x10)) {
299 if (c == 0x1b)
300 filter->status += 2;
301 } else if (filter->status == 0x20 && c > 0x20 && c < 0x7f) { /* kanji first char */
302 filter->status += 1;
303 } else if (c >= 0 && c < 0x80) { /* latin, CTLs */
304 ;
305 } else {
306 filter->flag = 1; /* bad */
307 }
308 break;
309
310 /* case 0x21: KSC5601 second char */
311 case 1:
312 filter->status &= ~0xf;
313 if (c < 0x21 || c > 0x7e) { /* bad */
314 filter->flag = 1;
315 }
316 break;
317
318 /* ESC */
319 case 2:
320 if (c == 0x24) { /* '$' */
321 filter->status++;
322 } else {
323 filter->flag = 1; /* bad */
324 filter->status &= ~0xf;
325 goto retry;
326 }
327 break;
328
329 /* ESC $ */
330 case 3:
331 if (c == 0x29) { /* ')' */
332 filter->status++;
333 } else {
334 filter->flag = 1; /* bad */
335 filter->status &= ~0xf;
336 goto retry;
337 }
338 break;
339
340 /* ESC $) */
341 case 5:
342 if (c == 0x43) { /* 'C' */
343 filter->status = 0x10;
344 } else {
345 filter->flag = 1; /* bad */
346 filter->status &= ~0xf;
347 goto retry;
348 }
349 break;
350
351 default:
352 filter->status = 0;
353 break;
354 }
355
356 return c;
357 }
358
359
360
