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 file was separated from mbfilter.c
26 * by moriyoshi koizumi <moriyoshi@php.net> on 20 dec 2002.
27 *
28 */
29
30 #include "mbfilter.h"
31 #include "mbfilter_utf32.h"
32
33 static int mbfl_filt_conv_utf32_wchar_flush(mbfl_convert_filter *filter);
34 static size_t mb_utf32_to_wchar(unsigned char **in, size_t *in_len, uint32_t *buf, size_t bufsize, unsigned int *state);
35 static size_t mb_utf32be_to_wchar(unsigned char **in, size_t *in_len, uint32_t *buf, size_t bufsize, unsigned int *state);
36 static void mb_wchar_to_utf32be(uint32_t *in, size_t len, mb_convert_buf *buf, bool end);
37 static size_t mb_utf32le_to_wchar(unsigned char **in, size_t *in_len, uint32_t *buf, size_t bufsize, unsigned int *state);
38 static void mb_wchar_to_utf32le(uint32_t *in, size_t len, mb_convert_buf *buf, bool end);
39
40 static const char *mbfl_encoding_utf32_aliases[] = {"utf32", NULL};
41
42 const mbfl_encoding mbfl_encoding_utf32 = {
43 mbfl_no_encoding_utf32,
44 "UTF-32",
45 "UTF-32",
46 mbfl_encoding_utf32_aliases,
47 NULL,
48 MBFL_ENCTYPE_WCS4,
49 &vtbl_utf32_wchar,
50 &vtbl_wchar_utf32,
51 mb_utf32_to_wchar,
52 mb_wchar_to_utf32be,
53 NULL,
54 NULL,
55 };
56
57 const mbfl_encoding mbfl_encoding_utf32be = {
58 mbfl_no_encoding_utf32be,
59 "UTF-32BE",
60 "UTF-32BE",
61 NULL,
62 NULL,
63 MBFL_ENCTYPE_WCS4,
64 &vtbl_utf32be_wchar,
65 &vtbl_wchar_utf32be,
66 mb_utf32be_to_wchar,
67 mb_wchar_to_utf32be,
68 NULL,
69 NULL,
70 };
71
72 const mbfl_encoding mbfl_encoding_utf32le = {
73 mbfl_no_encoding_utf32le,
74 "UTF-32LE",
75 "UTF-32LE",
76 NULL,
77 NULL,
78 MBFL_ENCTYPE_WCS4,
79 &vtbl_utf32le_wchar,
80 &vtbl_wchar_utf32le,
81 mb_utf32le_to_wchar,
82 mb_wchar_to_utf32le,
83 NULL,
84 NULL,
85 };
86
87 const struct mbfl_convert_vtbl vtbl_utf32_wchar = {
88 mbfl_no_encoding_utf32,
89 mbfl_no_encoding_wchar,
90 mbfl_filt_conv_common_ctor,
91 NULL,
92 mbfl_filt_conv_utf32_wchar,
93 mbfl_filt_conv_utf32_wchar_flush,
94 NULL,
95 };
96
97 const struct mbfl_convert_vtbl vtbl_wchar_utf32 = {
98 mbfl_no_encoding_wchar,
99 mbfl_no_encoding_utf32,
100 mbfl_filt_conv_common_ctor,
101 NULL,
102 mbfl_filt_conv_wchar_utf32be,
103 mbfl_filt_conv_common_flush,
104 NULL,
105 };
106
107 const struct mbfl_convert_vtbl vtbl_utf32be_wchar = {
108 mbfl_no_encoding_utf32be,
109 mbfl_no_encoding_wchar,
110 mbfl_filt_conv_common_ctor,
111 NULL,
112 mbfl_filt_conv_utf32be_wchar,
113 mbfl_filt_conv_utf32_wchar_flush,
114 NULL,
115 };
116
117 const struct mbfl_convert_vtbl vtbl_wchar_utf32be = {
118 mbfl_no_encoding_wchar,
119 mbfl_no_encoding_utf32be,
120 mbfl_filt_conv_common_ctor,
121 NULL,
122 mbfl_filt_conv_wchar_utf32be,
123 mbfl_filt_conv_common_flush,
124 NULL,
125 };
126
127 const struct mbfl_convert_vtbl vtbl_utf32le_wchar = {
128 mbfl_no_encoding_utf32le,
129 mbfl_no_encoding_wchar,
130 mbfl_filt_conv_common_ctor,
131 NULL,
132 mbfl_filt_conv_utf32le_wchar,
133 mbfl_filt_conv_utf32_wchar_flush,
134 NULL,
135 };
136
137 const struct mbfl_convert_vtbl vtbl_wchar_utf32le = {
138 mbfl_no_encoding_wchar,
139 mbfl_no_encoding_utf32le,
140 mbfl_filt_conv_common_ctor,
141 NULL,
142 mbfl_filt_conv_wchar_utf32le,
143 mbfl_filt_conv_common_flush,
144 NULL,
145 };
146
147 #define CK(statement) do { if ((statement) < 0) return (-1); } while (0)
148
emit_char_if_valid(int n,mbfl_convert_filter * filter)149 static int emit_char_if_valid(int n, mbfl_convert_filter *filter)
150 {
151 if (n >= 0 && n < MBFL_WCSPLANE_UTF32MAX && (n < 0xD800 || n > 0xDFFF)) {
152 CK((*filter->output_function)(n, filter->data));
153 } else {
154 CK((*filter->output_function)(MBFL_BAD_INPUT, filter->data));
155 }
156 return 0;
157 }
158
mbfl_filt_conv_utf32_wchar(int c,mbfl_convert_filter * filter)159 int mbfl_filt_conv_utf32_wchar(int c, mbfl_convert_filter *filter)
160 {
161 if (filter->status < 3) {
162 filter->cache = (filter->cache << 8) | (c & 0xFF);
163 filter->status++;
164 } else {
165 int n = ((unsigned int)filter->cache << 8) | (c & 0xFF);
166 filter->cache = filter->status = 0;
167
168 if (n == 0xFFFE0000) {
169 /* Found a little-endian byte order mark */
170 filter->filter_function = mbfl_filt_conv_utf32le_wchar;
171 } else {
172 filter->filter_function = mbfl_filt_conv_utf32be_wchar;
173 if (n != 0xFEFF) {
174 CK(emit_char_if_valid(n, filter));
175 }
176 }
177 }
178
179 return 0;
180 }
181
mbfl_filt_conv_utf32be_wchar(int c,mbfl_convert_filter * filter)182 int mbfl_filt_conv_utf32be_wchar(int c, mbfl_convert_filter *filter)
183 {
184 if (filter->status < 3) {
185 filter->cache = (filter->cache << 8) | (c & 0xFF);
186 filter->status++;
187 } else {
188 int n = ((unsigned int)filter->cache << 8) | (c & 0xFF);
189 filter->cache = filter->status = 0;
190 CK(emit_char_if_valid(n, filter));
191 }
192 return 0;
193 }
194
mbfl_filt_conv_wchar_utf32be(int c,mbfl_convert_filter * filter)195 int mbfl_filt_conv_wchar_utf32be(int c, mbfl_convert_filter *filter)
196 {
197 if (c >= 0 && c < MBFL_WCSPLANE_UTF32MAX) {
198 CK((*filter->output_function)((c >> 24) & 0xff, filter->data));
199 CK((*filter->output_function)((c >> 16) & 0xff, filter->data));
200 CK((*filter->output_function)((c >> 8) & 0xff, filter->data));
201 CK((*filter->output_function)(c & 0xff, filter->data));
202 } else {
203 CK(mbfl_filt_conv_illegal_output(c, filter));
204 }
205
206 return 0;
207 }
208
mbfl_filt_conv_utf32le_wchar(int c,mbfl_convert_filter * filter)209 int mbfl_filt_conv_utf32le_wchar(int c, mbfl_convert_filter *filter)
210 {
211 if (filter->status < 3) {
212 filter->cache |= ((c & 0xFFU) << (8 * filter->status));
213 filter->status++;
214 } else {
215 int n = ((c & 0xFFU) << 24) | filter->cache;
216 filter->cache = filter->status = 0;
217 CK(emit_char_if_valid(n, filter));
218 }
219 return 0;
220 }
221
mbfl_filt_conv_wchar_utf32le(int c,mbfl_convert_filter * filter)222 int mbfl_filt_conv_wchar_utf32le(int c, mbfl_convert_filter *filter)
223 {
224 if (c >= 0 && c < MBFL_WCSPLANE_UTF32MAX) {
225 CK((*filter->output_function)(c & 0xff, filter->data));
226 CK((*filter->output_function)((c >> 8) & 0xff, filter->data));
227 CK((*filter->output_function)((c >> 16) & 0xff, filter->data));
228 CK((*filter->output_function)((c >> 24) & 0xff, filter->data));
229 } else {
230 CK(mbfl_filt_conv_illegal_output(c, filter));
231 }
232
233 return 0;
234 }
235
mbfl_filt_conv_utf32_wchar_flush(mbfl_convert_filter * filter)236 static int mbfl_filt_conv_utf32_wchar_flush(mbfl_convert_filter *filter)
237 {
238 if (filter->status) {
239 /* Input string was truncated */
240 CK((*filter->output_function)(MBFL_BAD_INPUT, filter->data));
241 }
242 filter->cache = filter->status = 0;
243
244 if (filter->flush_function) {
245 (*filter->flush_function)(filter->data);
246 }
247
248 return 0;
249 }
250
251 #define DETECTED_BE 1
252 #define DETECTED_LE 2
253
mb_utf32_to_wchar(unsigned char ** in,size_t * in_len,uint32_t * buf,size_t bufsize,unsigned int * state)254 static size_t mb_utf32_to_wchar(unsigned char **in, size_t *in_len, uint32_t *buf, size_t bufsize, unsigned int *state)
255 {
256 if (*state == DETECTED_BE) {
257 return mb_utf32be_to_wchar(in, in_len, buf, bufsize, NULL);
258 } else if (*state == DETECTED_LE) {
259 return mb_utf32le_to_wchar(in, in_len, buf, bufsize, NULL);
260 } else if (*in_len >= 4) {
261 unsigned char *p = *in;
262 uint32_t c1 = *p++;
263 uint32_t c2 = *p++;
264 uint32_t c3 = *p++;
265 uint32_t c4 = *p++;
266 uint32_t w = (c1 << 24) | (c2 << 16) | (c3 << 8) | c4;
267
268 if (w == 0xFFFE0000) {
269 /* Little-endian BOM */
270 *in = p;
271 *in_len -= 4;
272 *state = DETECTED_LE;
273 return mb_utf32le_to_wchar(in, in_len, buf, bufsize, NULL);
274 } else if (w == 0xFEFF) {
275 /* Big-endian BOM; don't send it to output */
276 *in = p;
277 *in_len -= 4;
278 }
279 }
280
281 *state = DETECTED_BE;
282 return mb_utf32be_to_wchar(in, in_len, buf, bufsize, NULL);
283 }
284
mb_utf32be_to_wchar(unsigned char ** in,size_t * in_len,uint32_t * buf,size_t bufsize,unsigned int * state)285 static size_t mb_utf32be_to_wchar(unsigned char **in, size_t *in_len, uint32_t *buf, size_t bufsize, unsigned int *state)
286 {
287 unsigned char *p = *in, *e = p + (*in_len & ~3);
288 uint32_t *out = buf, *limit = buf + bufsize;
289
290 while (p < e && out < limit) {
291 uint32_t c1 = *p++;
292 uint32_t c2 = *p++;
293 uint32_t c3 = *p++;
294 uint32_t c4 = *p++;
295 uint32_t w = (c1 << 24) | (c2 << 16) | (c3 << 8) | c4;
296
297 if (w < MBFL_WCSPLANE_UTF32MAX && (w < 0xD800 || w > 0xDFFF)) {
298 *out++ = w;
299 } else {
300 *out++ = MBFL_BAD_INPUT;
301 }
302 }
303
304 if (p == e && (*in_len & 0x3) && out < limit) {
305 /* There are 1-3 trailing bytes, which shouldn't be there */
306 *out++ = MBFL_BAD_INPUT;
307 p = *in + *in_len;
308 }
309
310 *in_len -= (p - *in);
311 *in = p;
312 return out - buf;
313 }
314
mb_wchar_to_utf32be(uint32_t * in,size_t len,mb_convert_buf * buf,bool end)315 static void mb_wchar_to_utf32be(uint32_t *in, size_t len, mb_convert_buf *buf, bool end)
316 {
317 unsigned char *out, *limit;
318 MB_CONVERT_BUF_LOAD(buf, out, limit);
319 MB_CONVERT_BUF_ENSURE(buf, out, limit, len * 4);
320
321 while (len--) {
322 uint32_t w = *in++;
323 if (w < MBFL_WCSPLANE_UTF32MAX) {
324 out = mb_convert_buf_add4(out, (w >> 24) & 0xFF, (w >> 16) & 0xFF, (w >> 8) & 0xFF, w & 0xFF);
325 } else {
326 MB_CONVERT_ERROR(buf, out, limit, w, mb_wchar_to_utf32be);
327 MB_CONVERT_BUF_ENSURE(buf, out, limit, len * 4);
328 }
329 }
330
331 MB_CONVERT_BUF_STORE(buf, out, limit);
332 }
333
mb_utf32le_to_wchar(unsigned char ** in,size_t * in_len,uint32_t * buf,size_t bufsize,unsigned int * state)334 static size_t mb_utf32le_to_wchar(unsigned char **in, size_t *in_len, uint32_t *buf, size_t bufsize, unsigned int *state)
335 {
336 unsigned char *p = *in, *e = p + (*in_len & ~3);
337 uint32_t *out = buf, *limit = buf + bufsize;
338
339 while (p < e && out < limit) {
340 uint32_t c1 = *p++;
341 uint32_t c2 = *p++;
342 uint32_t c3 = *p++;
343 uint32_t c4 = *p++;
344 uint32_t w = (c4 << 24) | (c3 << 16) | (c2 << 8) | c1;
345
346 if (w < MBFL_WCSPLANE_UTF32MAX && (w < 0xD800 || w > 0xDFFF)) {
347 *out++ = w;
348 } else {
349 *out++ = MBFL_BAD_INPUT;
350 }
351 }
352
353 if (p == e && (*in_len & 0x3) && out < limit) {
354 /* There are 1-3 trailing bytes, which shouldn't be there */
355 *out++ = MBFL_BAD_INPUT;
356 p = *in + *in_len;
357 }
358
359 *in_len -= (p - *in);
360 *in = p;
361 return out - buf;
362 }
363
mb_wchar_to_utf32le(uint32_t * in,size_t len,mb_convert_buf * buf,bool end)364 static void mb_wchar_to_utf32le(uint32_t *in, size_t len, mb_convert_buf *buf, bool end)
365 {
366 unsigned char *out, *limit;
367 MB_CONVERT_BUF_LOAD(buf, out, limit);
368 MB_CONVERT_BUF_ENSURE(buf, out, limit, len * 4);
369
370 while (len--) {
371 uint32_t w = *in++;
372 if (w < MBFL_WCSPLANE_UTF32MAX) {
373 out = mb_convert_buf_add4(out, w & 0xFF, (w >> 8) & 0xFF, (w >> 16) & 0xFF, (w >> 24) & 0xFF);
374 } else {
375 MB_CONVERT_ERROR(buf, out, limit, w, mb_wchar_to_utf32le);
376 MB_CONVERT_BUF_ENSURE(buf, out, limit, len * 4);
377 }
378 }
379
380 MB_CONVERT_BUF_STORE(buf, out, limit);
381 }
382
