Commit | Line | Data |
---|---|---|
8690e634 | 1 | /* Convert multibyte character to wide character. |
5e8754f9 | 2 | Copyright (C) 1999-2002, 2005-2016 Free Software Foundation, Inc. |
8690e634 JK |
3 | Written by Bruno Haible <bruno@clisp.org>, 2008. |
4 | ||
5 | This program is free software: you can redistribute it and/or modify | |
6 | it under the terms of the GNU General Public License as published by | |
7 | the Free Software Foundation; either version 3 of the License, or | |
8 | (at your option) any later version. | |
9 | ||
10 | This program is distributed in the hope that it will be useful, | |
11 | but WITHOUT ANY WARRANTY; without even the implied warranty of | |
12 | MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the | |
13 | GNU General Public License for more details. | |
14 | ||
15 | You should have received a copy of the GNU General Public License | |
5e8754f9 | 16 | along with this program. If not, see <http://www.gnu.org/licenses/>. */ |
8690e634 JK |
17 | |
18 | #include <config.h> | |
19 | ||
20 | /* Specification. */ | |
21 | #include <wchar.h> | |
22 | ||
49e4877c PA |
23 | #if C_LOCALE_MAYBE_EILSEQ |
24 | # include "hard-locale.h" | |
25 | # include <locale.h> | |
26 | #endif | |
27 | ||
8690e634 JK |
28 | #if GNULIB_defined_mbstate_t |
29 | /* Implement mbrtowc() on top of mbtowc(). */ | |
30 | ||
31 | # include <errno.h> | |
32 | # include <stdlib.h> | |
33 | ||
34 | # include "localcharset.h" | |
35 | # include "streq.h" | |
36 | # include "verify.h" | |
37 | ||
38 | ||
39 | verify (sizeof (mbstate_t) >= 4); | |
40 | ||
41 | static char internal_state[4]; | |
42 | ||
43 | size_t | |
44 | mbrtowc (wchar_t *pwc, const char *s, size_t n, mbstate_t *ps) | |
45 | { | |
46 | char *pstate = (char *)ps; | |
47 | ||
48 | if (s == NULL) | |
49 | { | |
50 | pwc = NULL; | |
51 | s = ""; | |
52 | n = 1; | |
53 | } | |
54 | ||
55 | if (n == 0) | |
56 | return (size_t)(-2); | |
57 | ||
58 | /* Here n > 0. */ | |
59 | ||
60 | if (pstate == NULL) | |
61 | pstate = internal_state; | |
62 | ||
63 | { | |
64 | size_t nstate = pstate[0]; | |
65 | char buf[4]; | |
66 | const char *p; | |
67 | size_t m; | |
68 | ||
69 | switch (nstate) | |
70 | { | |
71 | case 0: | |
72 | p = s; | |
73 | m = n; | |
74 | break; | |
75 | case 3: | |
76 | buf[2] = pstate[3]; | |
5e8754f9 | 77 | /*FALLTHROUGH*/ |
8690e634 JK |
78 | case 2: |
79 | buf[1] = pstate[2]; | |
5e8754f9 | 80 | /*FALLTHROUGH*/ |
8690e634 JK |
81 | case 1: |
82 | buf[0] = pstate[1]; | |
83 | p = buf; | |
84 | m = nstate; | |
85 | buf[m++] = s[0]; | |
86 | if (n >= 2 && m < 4) | |
87 | { | |
88 | buf[m++] = s[1]; | |
89 | if (n >= 3 && m < 4) | |
90 | buf[m++] = s[2]; | |
91 | } | |
92 | break; | |
93 | default: | |
94 | errno = EINVAL; | |
95 | return (size_t)(-1); | |
96 | } | |
97 | ||
98 | /* Here m > 0. */ | |
99 | ||
100 | # if __GLIBC__ || defined __UCLIBC__ | |
5e8754f9 | 101 | /* Work around bug <http://sourceware.org/bugzilla/show_bug.cgi?id=9674> */ |
8690e634 JK |
102 | mbtowc (NULL, NULL, 0); |
103 | # endif | |
104 | { | |
105 | int res = mbtowc (pwc, p, m); | |
106 | ||
107 | if (res >= 0) | |
108 | { | |
109 | if (pwc != NULL && ((*pwc == 0) != (res == 0))) | |
110 | abort (); | |
111 | if (nstate >= (res > 0 ? res : 1)) | |
112 | abort (); | |
113 | res -= nstate; | |
114 | pstate[0] = 0; | |
115 | return res; | |
116 | } | |
117 | ||
118 | /* mbtowc does not distinguish between invalid and incomplete multibyte | |
119 | sequences. But mbrtowc needs to make this distinction. | |
120 | There are two possible approaches: | |
121 | - Use iconv() and its return value. | |
122 | - Use built-in knowledge about the possible encodings. | |
123 | Given the low quality of implementation of iconv() on the systems that | |
124 | lack mbrtowc(), we use the second approach. | |
125 | The possible encodings are: | |
126 | - 8-bit encodings, | |
127 | - EUC-JP, EUC-KR, GB2312, EUC-TW, BIG5, GB18030, SJIS, | |
128 | - UTF-8. | |
129 | Use specialized code for each. */ | |
130 | if (m >= 4 || m >= MB_CUR_MAX) | |
131 | goto invalid; | |
132 | /* Here MB_CUR_MAX > 1 and 0 < m < 4. */ | |
5e8754f9 SDJ |
133 | { |
134 | const char *encoding = locale_charset (); | |
135 | ||
136 | if (STREQ_OPT (encoding, "UTF-8", 'U', 'T', 'F', '-', '8', 0, 0, 0, 0)) | |
8690e634 JK |
137 | { |
138 | /* Cf. unistr/u8-mblen.c. */ | |
139 | unsigned char c = (unsigned char) p[0]; | |
140 | ||
141 | if (c >= 0xc2) | |
142 | { | |
143 | if (c < 0xe0) | |
144 | { | |
145 | if (m == 1) | |
146 | goto incomplete; | |
147 | } | |
148 | else if (c < 0xf0) | |
149 | { | |
150 | if (m == 1) | |
151 | goto incomplete; | |
152 | if (m == 2) | |
153 | { | |
154 | unsigned char c2 = (unsigned char) p[1]; | |
155 | ||
156 | if ((c2 ^ 0x80) < 0x40 | |
157 | && (c >= 0xe1 || c2 >= 0xa0) | |
158 | && (c != 0xed || c2 < 0xa0)) | |
159 | goto incomplete; | |
160 | } | |
161 | } | |
162 | else if (c <= 0xf4) | |
163 | { | |
164 | if (m == 1) | |
165 | goto incomplete; | |
166 | else /* m == 2 || m == 3 */ | |
167 | { | |
168 | unsigned char c2 = (unsigned char) p[1]; | |
169 | ||
170 | if ((c2 ^ 0x80) < 0x40 | |
171 | && (c >= 0xf1 || c2 >= 0x90) | |
172 | && (c < 0xf4 || (c == 0xf4 && c2 < 0x90))) | |
173 | { | |
174 | if (m == 2) | |
175 | goto incomplete; | |
176 | else /* m == 3 */ | |
177 | { | |
178 | unsigned char c3 = (unsigned char) p[2]; | |
179 | ||
180 | if ((c3 ^ 0x80) < 0x40) | |
181 | goto incomplete; | |
182 | } | |
183 | } | |
184 | } | |
185 | } | |
186 | } | |
187 | goto invalid; | |
188 | } | |
189 | ||
190 | /* As a reference for this code, you can use the GNU libiconv | |
191 | implementation. Look for uses of the RET_TOOFEW macro. */ | |
192 | ||
5e8754f9 SDJ |
193 | if (STREQ_OPT (encoding, |
194 | "EUC-JP", 'E', 'U', 'C', '-', 'J', 'P', 0, 0, 0)) | |
8690e634 JK |
195 | { |
196 | if (m == 1) | |
197 | { | |
198 | unsigned char c = (unsigned char) p[0]; | |
199 | ||
200 | if ((c >= 0xa1 && c < 0xff) || c == 0x8e || c == 0x8f) | |
201 | goto incomplete; | |
202 | } | |
203 | if (m == 2) | |
204 | { | |
205 | unsigned char c = (unsigned char) p[0]; | |
206 | ||
207 | if (c == 0x8f) | |
208 | { | |
209 | unsigned char c2 = (unsigned char) p[1]; | |
210 | ||
211 | if (c2 >= 0xa1 && c2 < 0xff) | |
212 | goto incomplete; | |
213 | } | |
214 | } | |
215 | goto invalid; | |
216 | } | |
5e8754f9 SDJ |
217 | if (STREQ_OPT (encoding, |
218 | "EUC-KR", 'E', 'U', 'C', '-', 'K', 'R', 0, 0, 0) | |
219 | || STREQ_OPT (encoding, | |
220 | "GB2312", 'G', 'B', '2', '3', '1', '2', 0, 0, 0) | |
221 | || STREQ_OPT (encoding, | |
222 | "BIG5", 'B', 'I', 'G', '5', 0, 0, 0, 0, 0)) | |
8690e634 JK |
223 | { |
224 | if (m == 1) | |
225 | { | |
226 | unsigned char c = (unsigned char) p[0]; | |
227 | ||
228 | if (c >= 0xa1 && c < 0xff) | |
229 | goto incomplete; | |
230 | } | |
231 | goto invalid; | |
232 | } | |
5e8754f9 SDJ |
233 | if (STREQ_OPT (encoding, |
234 | "EUC-TW", 'E', 'U', 'C', '-', 'T', 'W', 0, 0, 0)) | |
8690e634 JK |
235 | { |
236 | if (m == 1) | |
237 | { | |
238 | unsigned char c = (unsigned char) p[0]; | |
239 | ||
240 | if ((c >= 0xa1 && c < 0xff) || c == 0x8e) | |
241 | goto incomplete; | |
242 | } | |
243 | else /* m == 2 || m == 3 */ | |
244 | { | |
245 | unsigned char c = (unsigned char) p[0]; | |
246 | ||
247 | if (c == 0x8e) | |
248 | goto incomplete; | |
249 | } | |
250 | goto invalid; | |
251 | } | |
5e8754f9 SDJ |
252 | if (STREQ_OPT (encoding, |
253 | "GB18030", 'G', 'B', '1', '8', '0', '3', '0', 0, 0)) | |
8690e634 JK |
254 | { |
255 | if (m == 1) | |
256 | { | |
257 | unsigned char c = (unsigned char) p[0]; | |
258 | ||
259 | if ((c >= 0x90 && c <= 0xe3) || (c >= 0xf8 && c <= 0xfe)) | |
260 | goto incomplete; | |
261 | } | |
262 | else /* m == 2 || m == 3 */ | |
263 | { | |
264 | unsigned char c = (unsigned char) p[0]; | |
265 | ||
266 | if (c >= 0x90 && c <= 0xe3) | |
267 | { | |
268 | unsigned char c2 = (unsigned char) p[1]; | |
269 | ||
270 | if (c2 >= 0x30 && c2 <= 0x39) | |
271 | { | |
272 | if (m == 2) | |
273 | goto incomplete; | |
274 | else /* m == 3 */ | |
275 | { | |
276 | unsigned char c3 = (unsigned char) p[2]; | |
277 | ||
278 | if (c3 >= 0x81 && c3 <= 0xfe) | |
279 | goto incomplete; | |
280 | } | |
281 | } | |
282 | } | |
283 | } | |
284 | goto invalid; | |
285 | } | |
5e8754f9 | 286 | if (STREQ_OPT (encoding, "SJIS", 'S', 'J', 'I', 'S', 0, 0, 0, 0, 0)) |
8690e634 JK |
287 | { |
288 | if (m == 1) | |
289 | { | |
290 | unsigned char c = (unsigned char) p[0]; | |
291 | ||
292 | if ((c >= 0x81 && c <= 0x9f) || (c >= 0xe0 && c <= 0xea) | |
293 | || (c >= 0xf0 && c <= 0xf9)) | |
294 | goto incomplete; | |
295 | } | |
296 | goto invalid; | |
297 | } | |
298 | ||
5e8754f9 SDJ |
299 | /* An unknown multibyte encoding. */ |
300 | goto incomplete; | |
301 | } | |
8690e634 JK |
302 | |
303 | incomplete: | |
304 | { | |
305 | size_t k = nstate; | |
306 | /* Here 0 <= k < m < 4. */ | |
307 | pstate[++k] = s[0]; | |
308 | if (k < m) | |
309 | { | |
310 | pstate[++k] = s[1]; | |
311 | if (k < m) | |
312 | pstate[++k] = s[2]; | |
313 | } | |
314 | if (k != m) | |
315 | abort (); | |
316 | } | |
317 | pstate[0] = m; | |
318 | return (size_t)(-2); | |
319 | ||
320 | invalid: | |
321 | errno = EILSEQ; | |
322 | /* The conversion state is undefined, says POSIX. */ | |
323 | return (size_t)(-1); | |
324 | } | |
325 | } | |
326 | } | |
327 | ||
328 | #else | |
329 | /* Override the system's mbrtowc() function. */ | |
330 | ||
331 | # undef mbrtowc | |
332 | ||
333 | size_t | |
334 | rpl_mbrtowc (wchar_t *pwc, const char *s, size_t n, mbstate_t *ps) | |
335 | { | |
49e4877c PA |
336 | size_t ret; |
337 | wchar_t wc; | |
338 | ||
4a626d0a | 339 | # if MBRTOWC_NULL_ARG2_BUG || MBRTOWC_RETVAL_BUG || MBRTOWC_EMPTY_INPUT_BUG |
8690e634 JK |
340 | if (s == NULL) |
341 | { | |
342 | pwc = NULL; | |
343 | s = ""; | |
344 | n = 1; | |
345 | } | |
346 | # endif | |
347 | ||
4a626d0a PA |
348 | # if MBRTOWC_EMPTY_INPUT_BUG |
349 | if (n == 0) | |
350 | return (size_t) -2; | |
351 | # endif | |
352 | ||
49e4877c PA |
353 | if (! pwc) |
354 | pwc = &wc; | |
355 | ||
8690e634 JK |
356 | # if MBRTOWC_RETVAL_BUG |
357 | { | |
358 | static mbstate_t internal_state; | |
359 | ||
360 | /* Override mbrtowc's internal state. We cannot call mbsinit() on the | |
361 | hidden internal state, but we can call it on our variable. */ | |
362 | if (ps == NULL) | |
363 | ps = &internal_state; | |
364 | ||
365 | if (!mbsinit (ps)) | |
366 | { | |
367 | /* Parse the rest of the multibyte character byte for byte. */ | |
368 | size_t count = 0; | |
369 | for (; n > 0; s++, n--) | |
370 | { | |
49e4877c | 371 | ret = mbrtowc (&wc, s, 1, ps); |
8690e634 JK |
372 | |
373 | if (ret == (size_t)(-1)) | |
374 | return (size_t)(-1); | |
375 | count++; | |
376 | if (ret != (size_t)(-2)) | |
377 | { | |
378 | /* The multibyte character has been completed. */ | |
49e4877c | 379 | *pwc = wc; |
8690e634 JK |
380 | return (wc == 0 ? 0 : count); |
381 | } | |
382 | } | |
383 | return (size_t)(-2); | |
384 | } | |
385 | } | |
386 | # endif | |
387 | ||
49e4877c | 388 | ret = mbrtowc (pwc, s, n, ps); |
8690e634 | 389 | |
49e4877c PA |
390 | # if MBRTOWC_NUL_RETVAL_BUG |
391 | if (ret < (size_t) -2 && !*pwc) | |
392 | return 0; | |
393 | # endif | |
8690e634 | 394 | |
49e4877c PA |
395 | # if C_LOCALE_MAYBE_EILSEQ |
396 | if ((size_t) -2 <= ret && n != 0 && ! hard_locale (LC_CTYPE)) | |
397 | { | |
398 | unsigned char uc = *s; | |
399 | *pwc = uc; | |
400 | return 1; | |
401 | } | |
8690e634 | 402 | # endif |
49e4877c PA |
403 | |
404 | return ret; | |
8690e634 JK |
405 | } |
406 | ||
407 | #endif |