1 /*****************************************************************************
2 * unicode.c: Unicode <-> locale functions
3 *****************************************************************************
4 * Copyright (C) 2005-2006 VLC authors and VideoLAN
5 * Copyright © 2005-2010 Rémi Denis-Courmont
7 * Authors: Rémi Denis-Courmont <rem # videolan.org>
9 * This program is free software; you can redistribute it and/or modify it
10 * under the terms of the GNU Lesser General Public License as published by
11 * the Free Software Foundation; either version 2.1 of the License, or
12 * (at your option) any later version.
14 * This program is distributed in the hope that it will be useful,
15 * but WITHOUT ANY WARRANTY; without even the implied warranty of
16 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
17 * GNU Lesser General Public License for more details.
19 * You should have received a copy of the GNU Lesser General Public License
20 * along with this program; if not, write to the Free Software Foundation,
21 * Inc., 51 Franklin Street, Fifth Floor, Boston MA 02110-1301, USA.
22 *****************************************************************************/
24 /*****************************************************************************
26 *****************************************************************************/
31 #include <vlc_common.h>
34 #include <vlc_charset.h>
41 #include <sys/types.h>
49 * Formats an UTF-8 string as vfprintf(), then print it, with
50 * appropriate conversion to local encoding.
52 int utf8_vfprintf( FILE *stream, const char *fmt, va_list ap )
55 return vfprintf (stream, fmt, ap);
58 int res = vasprintf (&str, fmt, ap);
59 if (unlikely(res == -1))
63 /* Writing to the console is a lot of fun on Microsoft Windows.
64 * If you use the standard I/O functions, you must use the OEM code page,
65 * which is different from the usual ANSI code page. Or maybe not, if the
66 * user called "chcp". Anyway, we prefer Unicode. */
67 int fd = _fileno (stream);
68 if (likely(fd != -1) && _isatty (fd))
70 wchar_t *wide = ToWide (str);
71 if (likely(wide != NULL))
73 HANDLE h = (HANDLE)((uintptr_t)_get_osfhandle (fd));
75 /* XXX: It is not clear whether WriteConsole() wants the number of
76 * Unicode characters or the size of the wchar_t array. */
77 BOOL ok = WriteConsoleW (h, wide, wcslen (wide), &out, NULL);
84 wchar_t *wide = ToWide(str);
85 if (likely(wide != NULL))
87 res = fputws(wide, stream);
99 * Formats an UTF-8 string as fprintf(), then print it, with
100 * appropriate conversion to local encoding.
102 int utf8_fprintf( FILE *stream, const char *fmt, ... )
108 res = utf8_vfprintf( stream, fmt, ap );
115 * Converts the first character from a UTF-8 sequence into a code point.
117 * @param str an UTF-8 bytes sequence
118 * @return 0 if str points to an empty string, i.e. the first character is NUL;
119 * number of bytes that the first character occupies (from 1 to 4) otherwise;
120 * -1 if the byte sequence was not a valid UTF-8 sequence.
122 size_t vlc_towc (const char *str, uint32_t *restrict pwc)
124 uint8_t *ptr = (uint8_t *)str, c;
127 assert (str != NULL);
130 if (unlikely(c > 0xF4))
133 int charlen = clz8 (c ^ 0xFF);
136 case 0: // 7-bit ASCII character -> short cut
140 case 1: // continuation byte -> error
144 if (unlikely(c < 0xC2)) // ASCII overlong
146 cp = (c & 0x1F) << 6;
150 cp = (c & 0x0F) << 12;
154 cp = (c & 0x07) << 16;
161 /* Unrolled continuation bytes decoding */
166 if (unlikely((c >> 6) != 2)) // not a continuation byte
168 cp |= (c & 0x3f) << 12;
170 if (unlikely(cp >= 0x110000)) // beyond Unicode range
175 if (unlikely((c >> 6) != 2)) // not a continuation byte
177 cp |= (c & 0x3f) << 6;
179 if (unlikely(cp >= 0xD800 && cp < 0xE000)) // UTF-16 surrogate
181 if (unlikely(cp < (1u << (5 * charlen - 4)))) // non-ASCII overlong
186 if (unlikely((c >> 6) != 2)) // not a continuation byte
197 * Look for an UTF-8 string within another one in a case-insensitive fashion.
198 * Beware that this is quite slow. Contrary to strcasestr(), this function
199 * works regardless of the system character encoding, and handles multibyte
200 * code points correctly.
202 * @param haystack string to look into
203 * @param needle string to look for
204 * @return a pointer to the first occurence of the needle within the haystack,
205 * or NULL if no occurence were found.
207 char *vlc_strcasestr (const char *haystack, const char *needle)
213 const char *h = haystack, *n = needle;
219 s = vlc_towc (n, &cpn);
221 return (char *)haystack;
226 s = vlc_towc (h, &cph);
227 if (s <= 0 || towlower (cph) != towlower (cpn))
232 s = vlc_towc (haystack, &(uint32_t) { 0 });
241 * Replaces invalid/overlong UTF-8 sequences with question marks.
242 * Note that it is not possible to convert from Latin-1 to UTF-8 on the fly,
243 * so we don't try that, even though it would be less disruptive.
245 * @return str if it was valid UTF-8, NULL if not.
247 char *EnsureUTF8( char *str )
253 while ((n = vlc_towc (str, &cp)) != 0)
254 if (likely(n != (size_t)-1))
266 * Checks whether a string is a valid UTF-8 byte sequence.
268 * @param str nul-terminated string to be checked
270 * @return str if it was valid UTF-8, NULL if not.
272 const char *IsUTF8( const char *str )
277 while ((n = vlc_towc (str, &cp)) != 0)
278 if (likely(n != (size_t)-1))
286 * Converts a string from the given character encoding to utf-8.
288 * @return a nul-terminated utf-8 string, or null in case of error.
289 * The result must be freed using free().
291 char *FromCharset(const char *charset, const void *data, size_t data_size)
293 vlc_iconv_t handle = vlc_iconv_open ("UTF-8", charset);
294 if (handle == (vlc_iconv_t)(-1))
298 for(unsigned mul = 4; mul < 8; mul++ )
300 size_t in_size = data_size;
301 const char *in = data;
302 size_t out_max = mul * data_size;
303 char *tmp = out = malloc (1 + out_max);
307 if (vlc_iconv (handle, &in, &in_size, &tmp, &out_max) != (size_t)(-1)) {
317 vlc_iconv_close(handle);
322 * Converts a nul-terminated UTF-8 string to a given character encoding.
323 * @param charset iconv name of the character set
324 * @param in nul-terminated UTF-8 string
325 * @param outsize pointer to hold the byte size of result
327 * @return A pointer to the result, which must be released using free().
328 * The UTF-8 nul terminator is included in the conversion if the target
329 * character encoding supports it. However it is not included in the returned
331 * In case of error, NULL is returned and the byte size is undefined.
333 void *ToCharset(const char *charset, const char *in, size_t *outsize)
335 vlc_iconv_t hd = vlc_iconv_open (charset, "UTF-8");
336 if (hd == (vlc_iconv_t)(-1))
339 const size_t inlen = strlen (in);
342 for (unsigned mul = 4; mul < 16; mul++)
344 size_t outlen = mul * (inlen + 1);
345 res = malloc (outlen);
346 if (unlikely(res == NULL))
349 const char *inp = in;
352 size_t outb = outlen - mul;
354 if (vlc_iconv (hd, &inp, &inb, &outp, &outb) != (size_t)(-1))
356 *outsize = outlen - mul - outb;
358 inb = 1; /* append nul terminator if possible */
359 if (vlc_iconv (hd, &inp, &inb, &outp, &outb) != (size_t)(-1))
361 if (errno == EILSEQ) /* cannot translate nul terminator!? */
367 if (errno != E2BIG) /* conversion failure */
370 vlc_iconv_close (hd);