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 #include "mbfilter.h"
31 #include "mbfilter_iso2022_kr.h"
32 #include "unicode_table_uhc.h"
33
34 static int mbfl_filt_ident_2022kr(int c, mbfl_identify_filter *filter);
35
36 const mbfl_encoding mbfl_encoding_2022kr = {
37 mbfl_no_encoding_2022kr,
38 "ISO-2022-KR",
39 "ISO-2022-KR",
40 NULL,
41 NULL,
42 MBFL_ENCTYPE_MBCS | MBFL_ENCTYPE_GL_UNSAFE,
43 &vtbl_2022kr_wchar,
44 &vtbl_wchar_2022kr
45 };
46
47 const struct mbfl_identify_vtbl vtbl_identify_2022kr = {
48 mbfl_no_encoding_2022kr,
49 mbfl_filt_ident_common_ctor,
50 mbfl_filt_ident_2022kr
51 };
52
53 const struct mbfl_convert_vtbl vtbl_wchar_2022kr = {
54 mbfl_no_encoding_wchar,
55 mbfl_no_encoding_2022kr,
56 mbfl_filt_conv_common_ctor,
57 NULL,
58 mbfl_filt_conv_wchar_2022kr,
59 mbfl_filt_conv_any_2022kr_flush,
60 NULL,
61 };
62
63 const struct mbfl_convert_vtbl vtbl_2022kr_wchar = {
64 mbfl_no_encoding_2022kr,
65 mbfl_no_encoding_wchar,
66 mbfl_filt_conv_common_ctor,
67 NULL,
68 mbfl_filt_conv_2022kr_wchar,
69 mbfl_filt_conv_common_flush,
70 NULL,
71 };
72
73 #define CK(statement) do { if ((statement) < 0) return (-1); } while (0)
74
75 /*
76 * ISO-2022-KR => wchar
77 */
78 int
mbfl_filt_conv_2022kr_wchar(int c,mbfl_convert_filter * filter)79 mbfl_filt_conv_2022kr_wchar(int c, mbfl_convert_filter *filter)
80 {
81 int c1, w, flag;
82
83 retry:
84 switch (filter->status & 0xf) {
85 /* case 0x00: ASCII */
86 /* case 0x10: KSC5601 */
87 case 0:
88 if (c == 0x1b) { /* ESC */
89 filter->status += 2;
90 } else if (c == 0x0f) { /* SI (ASCII) */
91 filter->status &= ~0xff;
92 } else if (c == 0x0e) { /* SO (KSC5601) */
93 filter->status |= 0x10;
94 } else if ((filter->status & 0x10) != 0 && c > 0x20 && c < 0x7f) {
95 /* KSC5601 lead byte */
96 filter->cache = c;
97 filter->status += 1;
98 } else if ((filter->status & 0x10) == 0 && c >= 0 && c < 0x80) {
99 /* latin, CTLs */
100 CK((*filter->output_function)(c, filter->data));
101 } else {
102 w = c & MBFL_WCSGROUP_MASK;
103 w |= MBFL_WCSGROUP_THROUGH;
104 CK((*filter->output_function)(w, filter->data));
105 }
106 break;
107
108 case 1: /* dbcs second byte */
109 filter->status &= ~0xf;
110 c1 = filter->cache;
111 flag = 0;
112 if (c1 > 0x20 && c1 < 0x47) {
113 flag = 1;
114 } else if (c1 >= 0x47 && c1 <= 0x7e && c1 != 0x49) {
115 flag = 2;
116 }
117 if (flag > 0 && c > 0x20 && c < 0x7f) {
118 if (flag == 1){
119 w = (c1 - 0x21)*190 + (c - 0x41) + 0x80;
120 if (w >= 0 && w < uhc2_ucs_table_size) {
121 w = uhc2_ucs_table[w];
122 } else {
123 w = 0;
124 }
125 } else {
126 w = (c1 - 0x47)*94 + (c - 0x21);
127 if (w >= 0 && w < uhc3_ucs_table_size) {
128 w = uhc3_ucs_table[w];
129 } else {
130 w = 0;
131 }
132 }
133
134 if (w <= 0) {
135 w = (c1 << 8) | c;
136 w &= MBFL_WCSPLANE_MASK;
137 w |= MBFL_WCSPLANE_KSC5601;
138 }
139 CK((*filter->output_function)(w, filter->data));
140 } else if (c == 0x1b) { /* ESC */
141 filter->status++;
142 } else if ((c >= 0 && c < 0x21) || c == 0x7f) { /* CTLs */
143 CK((*filter->output_function)(c, filter->data));
144 } else {
145 w = (c1 << 8) | c;
146 w &= MBFL_WCSGROUP_MASK;
147 w |= MBFL_WCSGROUP_THROUGH;
148 CK((*filter->output_function)(w, filter->data));
149 }
150 break;
151
152 case 2: /* ESC */
153 if (c == 0x24) { /* '$' */
154 filter->status++;
155 } else {
156 filter->status &= ~0xf;
157 CK((*filter->output_function)(0x1b, filter->data));
158 goto retry;
159 }
160 break;
161 case 3: /* ESC $ */
162 if (c == 0x29) { /* ')' */
163 filter->status++;
164 } else {
165 filter->status &= ~0xf;
166 CK((*filter->output_function)(0x1b, filter->data));
167 CK((*filter->output_function)(0x24, filter->data));
168 goto retry;
169 }
170 break;
171 case 4: /* ESC $ ) */
172 if (c == 0x43) { /* 'C' */
173 filter->status &= ~0xf;
174 filter->status |= 0x100;
175 } else {
176 filter->status &= ~0xf;
177 CK((*filter->output_function)(0x1b, filter->data));
178 CK((*filter->output_function)(0x24, filter->data));
179 CK((*filter->output_function)(0x29, filter->data));
180 goto retry;
181 }
182 break;
183 default:
184 filter->status = 0;
185 break;
186 }
187
188 return c;
189 }
190
191 /*
192 * wchar => ISO-2022-KR
193 */
194 int
mbfl_filt_conv_wchar_2022kr(int c,mbfl_convert_filter * filter)195 mbfl_filt_conv_wchar_2022kr(int c, mbfl_convert_filter *filter)
196 {
197 int c1, c2, s;
198
199 s = 0;
200
201 if (c >= ucs_a1_uhc_table_min && c < ucs_a1_uhc_table_max) {
202 s = ucs_a1_uhc_table[c - ucs_a1_uhc_table_min];
203 } else if (c >= ucs_a2_uhc_table_min && c < ucs_a2_uhc_table_max) {
204 s = ucs_a2_uhc_table[c - ucs_a2_uhc_table_min];
205 } else if (c >= ucs_a3_uhc_table_min && c < ucs_a3_uhc_table_max) {
206 s = ucs_a3_uhc_table[c - ucs_a3_uhc_table_min];
207 } else if (c >= ucs_i_uhc_table_min && c < ucs_i_uhc_table_max) {
208 s = ucs_i_uhc_table[c - ucs_i_uhc_table_min];
209 } else if (c >= ucs_s_uhc_table_min && c < ucs_s_uhc_table_max) {
210 s = ucs_s_uhc_table[c - ucs_s_uhc_table_min];
211 } else if (c >= ucs_r1_uhc_table_min && c < ucs_r1_uhc_table_max) {
212 s = ucs_r1_uhc_table[c - ucs_r1_uhc_table_min];
213 } else if (c >= ucs_r2_uhc_table_min && c < ucs_r2_uhc_table_max) {
214 s = ucs_r2_uhc_table[c - ucs_r2_uhc_table_min];
215 }
216
217 c1 = (s >> 8) & 0xff;
218 c2 = s & 0xff;
219 /* exclude UHC extension area */
220 if (c1 < 0xa1 || c2 < 0xa1){
221 s = c;
222 }
223 if (s & 0x8000) {
224 s -= 0x8080;
225 }
226
227 if (s <= 0) {
228 c1 = c & ~MBFL_WCSPLANE_MASK;
229 if (c1 == MBFL_WCSPLANE_KSC5601) {
230 s = c & MBFL_WCSPLANE_MASK;
231 }
232 if (c == 0) {
233 s = 0;
234 } else if (s <= 0) {
235 s = -1;
236 }
237 } else if ((s >= 0x80 && s < 0x2121) || (s > 0x8080)) {
238 s = -1;
239 }
240 if (s >= 0) {
241 if (s < 0x80 && s > 0) { /* ASCII */
242 if ((filter->status & 0x10) != 0) {
243 CK((*filter->output_function)(0x0f, filter->data)); /* SI */
244 filter->status &= ~0x10;
245 }
246 CK((*filter->output_function)(s, filter->data));
247 } else {
248 if ( (filter->status & 0x100) == 0) {
249 CK((*filter->output_function)(0x1b, filter->data)); /* ESC */
250 CK((*filter->output_function)(0x24, filter->data)); /* '$' */
251 CK((*filter->output_function)(0x29, filter->data)); /* ')' */
252 CK((*filter->output_function)(0x43, filter->data)); /* 'C' */
253 filter->status |= 0x100;
254 }
255 if ((filter->status & 0x10) == 0) {
256 CK((*filter->output_function)(0x0e, filter->data)); /* SO */
257 filter->status |= 0x10;
258 }
259 CK((*filter->output_function)((s >> 8) & 0xff, filter->data));
260 CK((*filter->output_function)(s & 0xff, filter->data));
261 }
262 } else {
263 CK(mbfl_filt_conv_illegal_output(c, filter));
264 }
265
266 return c;
267 }
268
269 int
mbfl_filt_conv_any_2022kr_flush(mbfl_convert_filter * filter)270 mbfl_filt_conv_any_2022kr_flush(mbfl_convert_filter *filter)
271 {
272 /* back to ascii */
273 if ((filter->status & 0xff00) != 0) {
274 CK((*filter->output_function)(0x0f, filter->data)); /* SI */
275 }
276
277 filter->status &= 0xff;
278
279 if (filter->flush_function != NULL) {
280 return (*filter->flush_function)(filter->data);
281 }
282
283 return 0;
284 }
285
mbfl_filt_ident_2022kr(int c,mbfl_identify_filter * filter)286 static int mbfl_filt_ident_2022kr(int c, mbfl_identify_filter *filter)
287 {
288 retry:
289 switch (filter->status & 0xf) {
290 /* case 0x00: ASCII */
291 /* case 0x10: KSC5601 mode */
292 /* case 0x20: KSC5601 DBCS */
293 /* case 0x40: KSC5601 SBCS */
294 case 0:
295 if (!(filter->status & 0x10)) {
296 if (c == 0x1b)
297 filter->status += 2;
298 } else if (filter->status == 0x20 && c > 0x20 && c < 0x7f) { /* kanji first char */
299 filter->status += 1;
300 } else if (c >= 0 && c < 0x80) { /* latin, CTLs */
301 ;
302 } else {
303 filter->flag = 1; /* bad */
304 }
305 break;
306
307 /* case 0x21: KSC5601 second char */
308 case 1:
309 filter->status &= ~0xf;
310 if (c < 0x21 || c > 0x7e) { /* bad */
311 filter->flag = 1;
312 }
313 break;
314
315 /* ESC */
316 case 2:
317 if (c == 0x24) { /* '$' */
318 filter->status++;
319 } else {
320 filter->flag = 1; /* bad */
321 filter->status &= ~0xf;
322 goto retry;
323 }
324 break;
325
326 /* ESC $ */
327 case 3:
328 if (c == 0x29) { /* ')' */
329 filter->status++;
330 } else {
331 filter->flag = 1; /* bad */
332 filter->status &= ~0xf;
333 goto retry;
334 }
335 break;
336
337 /* ESC $) */
338 case 5:
339 if (c == 0x43) { /* 'C' */
340 filter->status = 0x10;
341 } else {
342 filter->flag = 1; /* bad */
343 filter->status &= ~0xf;
344 goto retry;
345 }
346 break;
347
348 default:
349 filter->status = 0;
350 break;
351 }
352
353 return c;
354 }
355