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 moriyoshi koizumi <moriyoshi@php.net> on 4 dec 2002.
27 *
28 */
29
30 #include "mbfilter.h"
31 #include "mbfilter_ucs2.h"
32
33 static int mbfl_filt_conv_ucs2_wchar_flush(mbfl_convert_filter *filter);
34 static size_t mb_ucs2_to_wchar(unsigned char **in, size_t *in_len, uint32_t *buf, size_t bufsize, unsigned int *state);
35 static size_t mb_ucs2be_to_wchar(unsigned char **in, size_t *in_len, uint32_t *buf, size_t bufsize, unsigned int *state);
36 static void mb_wchar_to_ucs2be(uint32_t *in, size_t len, mb_convert_buf *buf, bool end);
37 static size_t mb_ucs2le_to_wchar(unsigned char **in, size_t *in_len, uint32_t *buf, size_t bufsize, unsigned int *state);
38 static void mb_wchar_to_ucs2le(uint32_t *in, size_t len, mb_convert_buf *buf, bool end);
39
40 static const char *mbfl_encoding_ucs2_aliases[] = {"ISO-10646-UCS-2", "UCS2" , "UNICODE", NULL};
41
42 /* This library historically had encodings called 'byte2be' and 'byte2le'
43 * which were almost identical to UCS-2, except that they would truncate
44 * Unicode codepoints higher than 0xFFFF quietly
45 * Maintain minimal support by aliasing to UCS-2 */
46 static const char *mbfl_encoding_ucs2be_aliases[] = {"byte2be", NULL};
47 static const char *mbfl_encoding_ucs2le_aliases[] = {"byte2le", NULL};
48
49 const mbfl_encoding mbfl_encoding_ucs2 = {
50 mbfl_no_encoding_ucs2,
51 "UCS-2",
52 "UCS-2",
53 mbfl_encoding_ucs2_aliases,
54 NULL,
55 MBFL_ENCTYPE_WCS2,
56 &vtbl_ucs2_wchar,
57 &vtbl_wchar_ucs2,
58 mb_ucs2_to_wchar,
59 mb_wchar_to_ucs2be,
60 NULL
61 };
62
63 const mbfl_encoding mbfl_encoding_ucs2be = {
64 mbfl_no_encoding_ucs2be,
65 "UCS-2BE",
66 "UCS-2BE",
67 mbfl_encoding_ucs2be_aliases,
68 NULL,
69 MBFL_ENCTYPE_WCS2,
70 &vtbl_ucs2be_wchar,
71 &vtbl_wchar_ucs2be,
72 mb_ucs2be_to_wchar,
73 mb_wchar_to_ucs2be,
74 NULL
75 };
76
77 const mbfl_encoding mbfl_encoding_ucs2le = {
78 mbfl_no_encoding_ucs2le,
79 "UCS-2LE",
80 "UCS-2LE",
81 mbfl_encoding_ucs2le_aliases,
82 NULL,
83 MBFL_ENCTYPE_WCS2,
84 &vtbl_ucs2le_wchar,
85 &vtbl_wchar_ucs2le,
86 mb_ucs2le_to_wchar,
87 mb_wchar_to_ucs2le,
88 NULL
89 };
90
91 const struct mbfl_convert_vtbl vtbl_ucs2_wchar = {
92 mbfl_no_encoding_ucs2,
93 mbfl_no_encoding_wchar,
94 mbfl_filt_conv_common_ctor,
95 NULL,
96 mbfl_filt_conv_ucs2_wchar,
97 mbfl_filt_conv_ucs2_wchar_flush,
98 NULL,
99 };
100
101 const struct mbfl_convert_vtbl vtbl_wchar_ucs2 = {
102 mbfl_no_encoding_wchar,
103 mbfl_no_encoding_ucs2,
104 mbfl_filt_conv_common_ctor,
105 NULL,
106 mbfl_filt_conv_wchar_ucs2be,
107 mbfl_filt_conv_common_flush,
108 NULL,
109 };
110
111 const struct mbfl_convert_vtbl vtbl_ucs2be_wchar = {
112 mbfl_no_encoding_ucs2be,
113 mbfl_no_encoding_wchar,
114 mbfl_filt_conv_common_ctor,
115 NULL,
116 mbfl_filt_conv_ucs2be_wchar,
117 mbfl_filt_conv_ucs2_wchar_flush,
118 NULL,
119 };
120
121 const struct mbfl_convert_vtbl vtbl_wchar_ucs2be = {
122 mbfl_no_encoding_wchar,
123 mbfl_no_encoding_ucs2be,
124 mbfl_filt_conv_common_ctor,
125 NULL,
126 mbfl_filt_conv_wchar_ucs2be,
127 mbfl_filt_conv_common_flush,
128 NULL,
129 };
130
131 const struct mbfl_convert_vtbl vtbl_ucs2le_wchar = {
132 mbfl_no_encoding_ucs2le,
133 mbfl_no_encoding_wchar,
134 mbfl_filt_conv_common_ctor,
135 NULL,
136 mbfl_filt_conv_ucs2le_wchar,
137 mbfl_filt_conv_ucs2_wchar_flush,
138 NULL,
139 };
140
141 const struct mbfl_convert_vtbl vtbl_wchar_ucs2le = {
142 mbfl_no_encoding_wchar,
143 mbfl_no_encoding_ucs2le,
144 mbfl_filt_conv_common_ctor,
145 NULL,
146 mbfl_filt_conv_wchar_ucs2le,
147 mbfl_filt_conv_common_flush,
148 NULL,
149 };
150
151 #define CK(statement) do { if ((statement) < 0) return (-1); } while (0)
152
mbfl_filt_conv_ucs2_wchar(int c,mbfl_convert_filter * filter)153 int mbfl_filt_conv_ucs2_wchar(int c, mbfl_convert_filter *filter)
154 {
155 if (filter->status == 0) {
156 filter->status = 1;
157 filter->cache = c & 0xFF;
158 } else {
159 filter->status = 0;
160 int n = (filter->cache << 8) | (c & 0xFF);
161 if (n == 0xFFFE) {
162 /* Found little-endian byte order mark */
163 filter->filter_function = mbfl_filt_conv_ucs2le_wchar;
164 } else {
165 filter->filter_function = mbfl_filt_conv_ucs2be_wchar;
166 if (n != 0xFEFF) {
167 CK((*filter->output_function)(n, filter->data));
168 }
169 }
170 }
171 return 0;
172 }
173
mbfl_filt_conv_ucs2be_wchar(int c,mbfl_convert_filter * filter)174 int mbfl_filt_conv_ucs2be_wchar(int c, mbfl_convert_filter *filter)
175 {
176 if (filter->status == 0) {
177 filter->status = 1;
178 filter->cache = (c & 0xFF) << 8;
179 } else {
180 filter->status = 0;
181 CK((*filter->output_function)((c & 0xFF) | filter->cache, filter->data));
182 }
183 return 0;
184 }
185
mbfl_filt_conv_wchar_ucs2be(int c,mbfl_convert_filter * filter)186 int mbfl_filt_conv_wchar_ucs2be(int c, mbfl_convert_filter *filter)
187 {
188 if (c >= 0 && c < MBFL_WCSPLANE_UCS2MAX) {
189 CK((*filter->output_function)((c >> 8) & 0xFF, filter->data));
190 CK((*filter->output_function)(c & 0xFF, filter->data));
191 } else {
192 CK(mbfl_filt_conv_illegal_output(c, filter));
193 }
194 return 0;
195 }
196
mbfl_filt_conv_ucs2le_wchar(int c,mbfl_convert_filter * filter)197 int mbfl_filt_conv_ucs2le_wchar(int c, mbfl_convert_filter *filter)
198 {
199 if (filter->status == 0) {
200 filter->status = 1;
201 filter->cache = c & 0xFF;
202 } else {
203 filter->status = 0;
204 CK((*filter->output_function)(((c & 0xFF) << 8) | filter->cache, filter->data));
205 }
206 return 0;
207 }
208
mbfl_filt_conv_wchar_ucs2le(int c,mbfl_convert_filter * filter)209 int mbfl_filt_conv_wchar_ucs2le(int c, mbfl_convert_filter *filter)
210 {
211 if (c >= 0 && c < MBFL_WCSPLANE_UCS2MAX) {
212 CK((*filter->output_function)(c & 0xFF, filter->data));
213 CK((*filter->output_function)((c >> 8) & 0xFF, filter->data));
214 } else {
215 CK(mbfl_filt_conv_illegal_output(c, filter));
216 }
217 return 0;
218 }
219
mbfl_filt_conv_ucs2_wchar_flush(mbfl_convert_filter * filter)220 static int mbfl_filt_conv_ucs2_wchar_flush(mbfl_convert_filter *filter)
221 {
222 if (filter->status) {
223 /* Input string was truncated */
224 filter->status = 0;
225 CK((*filter->output_function)(MBFL_BAD_INPUT, filter->data));
226 }
227
228 if (filter->flush_function) {
229 (*filter->flush_function)(filter->data);
230 }
231
232 return 0;
233 }
234
235 #define DETECTED_BE 1
236 #define DETECTED_LE 2
237
mb_ucs2_to_wchar(unsigned char ** in,size_t * in_len,uint32_t * buf,size_t bufsize,unsigned int * state)238 static size_t mb_ucs2_to_wchar(unsigned char **in, size_t *in_len, uint32_t *buf, size_t bufsize, unsigned int *state)
239 {
240 if (*state == DETECTED_BE) {
241 return mb_ucs2be_to_wchar(in, in_len, buf, bufsize, NULL);
242 } else if (*state == DETECTED_LE) {
243 return mb_ucs2le_to_wchar(in, in_len, buf, bufsize, NULL);
244 } else if (*in_len >= 2) {
245 unsigned char *p = *in;
246 unsigned char c1 = *p++;
247 unsigned char c2 = *p++;
248 uint32_t w = (c1 << 8) | c2;
249
250 if (w == 0xFFFE) {
251 /* Little-endian BOM */
252 *in = p;
253 *in_len -= 2;
254 *state = DETECTED_LE;
255 return mb_ucs2le_to_wchar(in, in_len, buf, bufsize, NULL);
256 } else if (w == 0xFEFF) {
257 /* Big-endian BOM; don't send it to output */
258 *in = p;
259 *in_len -= 2;
260 }
261 }
262
263 *state = DETECTED_BE;
264 return mb_ucs2be_to_wchar(in, in_len, buf, bufsize, NULL);
265 }
266
mb_ucs2be_to_wchar(unsigned char ** in,size_t * in_len,uint32_t * buf,size_t bufsize,unsigned int * state)267 static size_t mb_ucs2be_to_wchar(unsigned char **in, size_t *in_len, uint32_t *buf, size_t bufsize, unsigned int *state)
268 {
269 unsigned char *p = *in, *e = p + (*in_len & ~1);
270 uint32_t *out = buf, *limit = buf + bufsize;
271
272 while (p < e && out < limit) {
273 unsigned char c1 = *p++;
274 unsigned char c2 = *p++;
275 uint32_t w = (c1 << 8) | c2;
276 *out++ = w;
277 }
278
279 if (p == e && (*in_len & 0x1) && out < limit) {
280 /* There is 1 trailing byte, which shouldn't be there */
281 *out++ = MBFL_BAD_INPUT;
282 p++;
283 }
284
285 *in_len -= (p - *in);
286 *in = p;
287 return out - buf;
288 }
289
mb_wchar_to_ucs2be(uint32_t * in,size_t len,mb_convert_buf * buf,bool end)290 static void mb_wchar_to_ucs2be(uint32_t *in, size_t len, mb_convert_buf *buf, bool end)
291 {
292 unsigned char *out, *limit;
293 MB_CONVERT_BUF_LOAD(buf, out, limit);
294 MB_CONVERT_BUF_ENSURE(buf, out, limit, len * 2);
295
296 while (len--) {
297 uint32_t w = *in++;
298 if (w < MBFL_WCSPLANE_UCS2MAX) {
299 out = mb_convert_buf_add2(out, (w >> 8) & 0xFF, w & 0xFF);
300 } else {
301 MB_CONVERT_ERROR(buf, out, limit, w, mb_wchar_to_ucs2be);
302 MB_CONVERT_BUF_ENSURE(buf, out, limit, len * 2);
303 }
304 }
305
306 MB_CONVERT_BUF_STORE(buf, out, limit);
307 }
308
mb_ucs2le_to_wchar(unsigned char ** in,size_t * in_len,uint32_t * buf,size_t bufsize,unsigned int * state)309 static size_t mb_ucs2le_to_wchar(unsigned char **in, size_t *in_len, uint32_t *buf, size_t bufsize, unsigned int *state)
310 {
311 unsigned char *p = *in, *e = p + (*in_len & ~1);
312 uint32_t *out = buf, *limit = buf + bufsize;
313
314 while (p < e && out < limit) {
315 unsigned char c1 = *p++;
316 unsigned char c2 = *p++;
317 uint32_t w = (c2 << 8) | c1;
318 *out++ = w;
319 }
320
321 if (p == e && (*in_len & 0x1) && out < limit) {
322 /* There is 1 trailing byte, which shouldn't be there */
323 *out++ = MBFL_BAD_INPUT;
324 p++;
325 }
326
327 *in_len -= (p - *in);
328 *in = p;
329 return out - buf;
330 }
331
mb_wchar_to_ucs2le(uint32_t * in,size_t len,mb_convert_buf * buf,bool end)332 static void mb_wchar_to_ucs2le(uint32_t *in, size_t len, mb_convert_buf *buf, bool end)
333 {
334 unsigned char *out, *limit;
335 MB_CONVERT_BUF_LOAD(buf, out, limit);
336 MB_CONVERT_BUF_ENSURE(buf, out, limit, len * 2);
337
338 while (len--) {
339 uint32_t w = *in++;
340 if (w < MBFL_WCSPLANE_UCS2MAX) {
341 out = mb_convert_buf_add2(out, w & 0xFF, (w >> 8) & 0xFF);
342 } else {
343 MB_CONVERT_ERROR(buf, out, limit, w, mb_wchar_to_ucs2le);
344 MB_CONVERT_BUF_ENSURE(buf, out, limit, len * 2);
345 }
346 }
347
348 MB_CONVERT_BUF_STORE(buf, out, limit);
349 }
350
