include/vlc_charset.h

   1 /*****************************************************************************
   2  * vlc_charset.h: Unicode UTF-8 wrappers function
   3  *****************************************************************************
   4  * Copyright (C) 2003-2005 VLC authors and VideoLAN
   5  * Copyright © 2005-2010 Rémi Denis-Courmont
   6  * $Id$
   7  *
   8  * Author: Rémi Denis-Courmont <rem # videolan,org>
   9  *
  10  * This program is free software; you can redistribute it and/or modify it
  11  * under the terms of the GNU Lesser General Public License as published by
  12  * the Free Software Foundation; either version 2.1 of the License, or
  13  * (at your option) any later version.
  14  *
  15  * This program is distributed in the hope that it will be useful,
  16  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  17  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
  18  * GNU Lesser General Public License for more details.
  19  *
  20  * You should have received a copy of the GNU Lesser General Public License
  21  * along with this program; if not, write to the Free Software Foundation,
  22  * Inc., 51 Franklin Street, Fifth Floor, Boston MA 02110-1301, USA.
  23  *****************************************************************************/
  24
  25 #ifndef VLC_CHARSET_H
  26 #define VLC_CHARSET_H 1
  27
  28 /**
  29  * \file
  30  * Characters sets handling
  31  *
  32  * \ingroup strings
  33  * @{
  34  */
  35
  36 /**
  37  * Decodes a code point from UTF-8.
  38  *
  39  * Converts the first character in a UTF-8 sequence into a Unicode code point.
  40  *
  41  * \param str an UTF-8 bytes sequence [IN]
  42  * \param pwc address of a location to store the code point [OUT]
  43  *
  44  * \return the number of bytes occupied by the decoded code point
  45  *
  46  * \retval (size_t)-1 not a valid UTF-8 sequence
  47  * \retval 0 null character (i.e. str points to an empty string)
  48  * \retval 1 (non-null) ASCII character
  49  * \retval 2-4 non-ASCII character
  50  */
  51 VLC_API size_t vlc_towc(const char *str, uint32_t *restrict pwc);
  52
  53 /**
  54  * Checks UTF-8 validity.
  55  *
  56  * Checks whether a null-terminated string is a valid UTF-8 bytes sequence.
  57  *
  58  * \param str string to check
  59  *
  60  * \retval str the string is a valid null-terminated UTF-8 sequence
  61  * \retval NULL the string is not an UTF-8 sequence
  62  */
  63 VLC_USED static inline const char *IsUTF8(const char *str)
  64 {
  65     size_t n;
  66     uint32_t cp;
  67
  68     while ((n = vlc_towc(str, &cp)) != 0)
  69         if (likely(n != (size_t)-1))
  70             str += n;
  71         else
  72             return NULL;
  73     return str;
  74 }
  75
  76 /**
  77  * Removes non-UTF-8 sequences.
  78  *
  79  * Replaces invalid or <i>over-long</i> UTF-8 bytes sequences within a
  80  * null-terminated string with question marks. This is so that the string can
  81  * be printed at least partially.
  82  *
  83  * \warning Do not use this were correctness is critical. use IsUTF8() and
  84  * handle the error case instead. This function is mainly for display or debug.
  85  *
  86  * \note Converting from Latin-1 to UTF-8 in place is not possible (the string
  87  * size would be increased). So it is not attempted even if it would otherwise
  88  * be less disruptive.
  89  *
  90  * \retval str the string is a valid null-terminated UTF-8 sequence
  91  *             (i.e. no changes were made)
  92  * \retval NULL the string is not an UTF-8 sequence
  93  */
  94 static inline char *EnsureUTF8(char *str)
  95 {
  96     char *ret = str;
  97     size_t n;
  98     uint32_t cp;
  99
 100     while ((n = vlc_towc(str, &cp)) != 0)
 101         if (likely(n != (size_t)-1))
 102             str += n;
 103         else
 104         {
 105             *str++ = '?';
 106             ret = NULL;
 107         }
 108     return ret;
 109 }
 110
 111 /* iconv wrappers (defined in src/extras/libc.c) */
 112 #define VLC_ICONV_ERR ((size_t) -1)
 113 typedef void *vlc_iconv_t;
 114 VLC_API vlc_iconv_t vlc_iconv_open( const char *, const char * ) VLC_USED;
 115 VLC_API size_t vlc_iconv( vlc_iconv_t, const char **, size_t *, char **, size_t * ) VLC_USED;
 116 VLC_API int vlc_iconv_close( vlc_iconv_t );
 117
 118 #include <stdarg.h>
 119
 120 VLC_API int utf8_vfprintf( FILE *stream, const char *fmt, va_list ap );
 121 VLC_API int utf8_fprintf( FILE *, const char *, ... ) VLC_FORMAT( 2, 3 );
 122 VLC_API char * vlc_strcasestr(const char *, const char *) VLC_USED;
 123
 124 VLC_API char * FromCharset( const char *charset, const void *data, size_t data_size ) VLC_USED;
 125 VLC_API void * ToCharset( const char *charset, const char *in, size_t *outsize ) VLC_USED;
 126
 127 #ifdef _WIN32
 128 VLC_USED
 129 static inline char *FromWide (const wchar_t *wide)
 130 {
 131     size_t len = WideCharToMultiByte (CP_UTF8, 0, wide, -1, NULL, 0, NULL, NULL);
 132     if (len == 0)
 133         return NULL;
 134
 135     char *out = (char *)malloc (len);
 136
 137     if (likely(out))
 138         WideCharToMultiByte (CP_UTF8, 0, wide, -1, out, len, NULL, NULL);
 139     return out;
 140 }
 141
 142 VLC_USED
 143 static inline wchar_t *ToWide (const char *utf8)
 144 {
 145     int len = MultiByteToWideChar (CP_UTF8, 0, utf8, -1, NULL, 0);
 146     if (len == 0)
 147         return NULL;
 148
 149     wchar_t *out = (wchar_t *)malloc (len * sizeof (wchar_t));
 150
 151     if (likely(out))
 152         MultiByteToWideChar (CP_UTF8, 0, utf8, -1, out, len);
 153     return out;
 154 }
 155
 156 VLC_USED VLC_MALLOC
 157 static inline char *ToCodePage (unsigned cp, const char *utf8)
 158 {
 159     wchar_t *wide = ToWide (utf8);
 160     if (wide == NULL)
 161         return NULL;
 162
 163     size_t len = WideCharToMultiByte (cp, 0, wide, -1, NULL, 0, NULL, NULL);
 164     if (len == 0) {
 165         free(wide);
 166         return NULL;
 167     }
 168
 169     char *out = (char *)malloc (len);
 170     if (likely(out != NULL))
 171         WideCharToMultiByte (cp, 0, wide, -1, out, len, NULL, NULL);
 172     free (wide);
 173     return out;
 174 }
 175
 176 VLC_USED VLC_MALLOC
 177 static inline char *FromCodePage (unsigned cp, const char *mb)
 178 {
 179     int len = MultiByteToWideChar (cp, 0, mb, -1, NULL, 0);
 180     if (len == 0)
 181         return NULL;
 182
 183     wchar_t *wide = (wchar_t *)malloc (len * sizeof (wchar_t));
 184     if (unlikely(wide == NULL))
 185         return NULL;
 186     MultiByteToWideChar (cp, 0, mb, -1, wide, len);
 187
 188     char *utf8 = FromWide (wide);
 189     free (wide);
 190     return utf8;
 191 }
 192
 193 VLC_USED VLC_MALLOC
 194 static inline char *FromANSI (const char *ansi)
 195 {
 196     return FromCodePage (GetACP (), ansi);
 197 }
 198
 199 VLC_USED VLC_MALLOC
 200 static inline char *ToANSI (const char *utf8)
 201 {
 202     return ToCodePage (GetACP (), utf8);
 203 }
 204
 205 # ifdef UNICODE
 206 #  define FromT FromWide
 207 #  define ToT   ToWide
 208 # else
 209 #  define FromT FromANSI
 210 #  define ToT   ToANSI
 211 # endif
 212 # define FromLocale    FromANSI
 213 # define ToLocale      ToANSI
 214 # define LocaleFree(s) free((char *)(s))
 215 # define FromLocaleDup FromANSI
 216 # define ToLocaleDup   ToANSI
 217
 218 #elif defined(__OS2__)
 219
 220 VLC_USED static inline char *FromLocale (const char *locale)
 221 {
 222     return locale ? FromCharset ((char *)"", locale, strlen(locale)) : NULL;
 223 }
 224
 225 VLC_USED static inline char *ToLocale (const char *utf8)
 226 {
 227     size_t outsize;
 228     return utf8 ? (char *)ToCharset ("", utf8, &outsize) : NULL;
 229 }
 230
 231 VLC_USED static inline void LocaleFree (const char *str)
 232 {
 233     free ((char *)str);
 234 }
 235
 236 VLC_USED static inline char *FromLocaleDup (const char *locale)
 237 {
 238     return FromCharset ("", locale, strlen(locale));
 239 }
 240
 241 VLC_USED static inline char *ToLocaleDup (const char *utf8)
 242 {
 243     size_t outsize;
 244     return (char *)ToCharset ("", utf8, &outsize);
 245 }
 246
 247 #else
 248
 249 # define FromLocale(l) (l)
 250 # define ToLocale(u)   (u)
 251 # define LocaleFree(s) ((void)(s))
 252 # define FromLocaleDup strdup
 253 # define ToLocaleDup   strdup
 254 #endif
 255
 256 /**
 257  * Converts a nul-terminated string from ISO-8859-1 to UTF-8.
 258  */
 259 static inline char *FromLatin1 (const char *latin)
 260 {
 261     char *str = (char *)malloc (2 * strlen (latin) + 1), *utf8 = str;
 262     unsigned char c;
 263
 264     if (str == NULL)
 265         return NULL;
 266
 267     while ((c = *(latin++)) != '\0')
 268     {
 269          if (c >= 0x80)
 270          {
 271              *(utf8++) = 0xC0 | (c >> 6);
 272              *(utf8++) = 0x80 | (c & 0x3F);
 273          }
 274          else
 275              *(utf8++) = c;
 276     }
 277     *(utf8++) = '\0';
 278
 279     utf8 = (char *)realloc (str, utf8 - str);
 280     return utf8 ? utf8 : str;
 281 }
 282
 283 /** @} */
 284
 285 VLC_API double us_strtod( const char *, char ** ) VLC_USED;
 286 VLC_API float us_strtof( const char *, char ** ) VLC_USED;
 287 VLC_API double us_atof( const char * ) VLC_USED;
 288 VLC_API int us_vasprintf( char **, const char *, va_list );
 289 VLC_API int us_asprintf( char **, const char *, ... ) VLC_USED;
 290
 291 #endif