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  * Checks ASCII validity.
  78  *
  79  * Checks whether a null-terminated string is a valid ASCII bytes sequence
  80  * (non-printable ASCII characters 1-31 are permitted).
  81  *
  82  * \param str string to check
  83  *
  84  * \retval str the string is a valid null-terminated ASCII sequence
  85  * \retval NULL the string is not an ASCII sequence
  86  */
  87 VLC_USED static inline const char *IsASCII(const char *str)
  88 {
  89     unsigned char c;
  90
  91     for (const char *p = str; (c = *p) != '\0'; p++)
  92         if (c >= 0x80)
  93             return NULL;
  94     return str;
  95 }
  96
  97 /**
  98  * Removes non-UTF-8 sequences.
  99  *
 100  * Replaces invalid or <i>over-long</i> UTF-8 bytes sequences within a
 101  * null-terminated string with question marks. This is so that the string can
 102  * be printed at least partially.
 103  *
 104  * \warning Do not use this were correctness is critical. use IsUTF8() and
 105  * handle the error case instead. This function is mainly for display or debug.
 106  *
 107  * \note Converting from Latin-1 to UTF-8 in place is not possible (the string
 108  * size would be increased). So it is not attempted even if it would otherwise
 109  * be less disruptive.
 110  *
 111  * \retval str the string is a valid null-terminated UTF-8 sequence
 112  *             (i.e. no changes were made)
 113  * \retval NULL the string is not an UTF-8 sequence
 114  */
 115 static inline char *EnsureUTF8(char *str)
 116 {
 117     char *ret = str;
 118     size_t n;
 119     uint32_t cp;
 120
 121     while ((n = vlc_towc(str, &cp)) != 0)
 122         if (likely(n != (size_t)-1))
 123             str += n;
 124         else
 125         {
 126             *str++ = '?';
 127             ret = NULL;
 128         }
 129     return ret;
 130 }
 131
 132 /* iconv wrappers (defined in src/extras/libc.c) */
 133 #define VLC_ICONV_ERR ((size_t) -1)
 134 typedef void *vlc_iconv_t;
 135 VLC_API vlc_iconv_t vlc_iconv_open( const char *, const char * ) VLC_USED;
 136 VLC_API size_t vlc_iconv( vlc_iconv_t, const char **, size_t *, char **, size_t * ) VLC_USED;
 137 VLC_API int vlc_iconv_close( vlc_iconv_t );
 138
 139 #include <stdarg.h>
 140
 141 VLC_API int utf8_vfprintf( FILE *stream, const char *fmt, va_list ap );
 142 VLC_API int utf8_fprintf( FILE *, const char *, ... ) VLC_FORMAT( 2, 3 );
 143 VLC_API char * vlc_strcasestr(const char *, const char *) VLC_USED;
 144
 145 VLC_API char * FromCharset( const char *charset, const void *data, size_t data_size ) VLC_USED;
 146 VLC_API void * ToCharset( const char *charset, const char *in, size_t *outsize ) VLC_USED;
 147
 148 #ifdef _WIN32
 149 VLC_USED
 150 static inline char *FromWide (const wchar_t *wide)
 151 {
 152     size_t len = WideCharToMultiByte (CP_UTF8, 0, wide, -1, NULL, 0, NULL, NULL);
 153     if (len == 0)
 154         return NULL;
 155
 156     char *out = (char *)malloc (len);
 157
 158     if (likely(out))
 159         WideCharToMultiByte (CP_UTF8, 0, wide, -1, out, len, NULL, NULL);
 160     return out;
 161 }
 162
 163 VLC_USED
 164 static inline wchar_t *ToWide (const char *utf8)
 165 {
 166     int len = MultiByteToWideChar (CP_UTF8, 0, utf8, -1, NULL, 0);
 167     if (len == 0)
 168         return NULL;
 169
 170     wchar_t *out = (wchar_t *)malloc (len * sizeof (wchar_t));
 171
 172     if (likely(out))
 173         MultiByteToWideChar (CP_UTF8, 0, utf8, -1, out, len);
 174     return out;
 175 }
 176
 177 VLC_USED VLC_MALLOC
 178 static inline char *ToCodePage (unsigned cp, const char *utf8)
 179 {
 180     wchar_t *wide = ToWide (utf8);
 181     if (wide == NULL)
 182         return NULL;
 183
 184     size_t len = WideCharToMultiByte (cp, 0, wide, -1, NULL, 0, NULL, NULL);
 185     if (len == 0) {
 186         free(wide);
 187         return NULL;
 188     }
 189
 190     char *out = (char *)malloc (len);
 191     if (likely(out != NULL))
 192         WideCharToMultiByte (cp, 0, wide, -1, out, len, NULL, NULL);
 193     free (wide);
 194     return out;
 195 }
 196
 197 VLC_USED VLC_MALLOC
 198 static inline char *FromCodePage (unsigned cp, const char *mb)
 199 {
 200     int len = MultiByteToWideChar (cp, 0, mb, -1, NULL, 0);
 201     if (len == 0)
 202         return NULL;
 203
 204     wchar_t *wide = (wchar_t *)malloc (len * sizeof (wchar_t));
 205     if (unlikely(wide == NULL))
 206         return NULL;
 207     MultiByteToWideChar (cp, 0, mb, -1, wide, len);
 208
 209     char *utf8 = FromWide (wide);
 210     free (wide);
 211     return utf8;
 212 }
 213
 214 VLC_USED VLC_MALLOC
 215 static inline char *FromANSI (const char *ansi)
 216 {
 217     return FromCodePage (GetACP (), ansi);
 218 }
 219
 220 VLC_USED VLC_MALLOC
 221 static inline char *ToANSI (const char *utf8)
 222 {
 223     return ToCodePage (GetACP (), utf8);
 224 }
 225
 226 # ifdef UNICODE
 227 #  define FromT FromWide
 228 #  define ToT   ToWide
 229 # else
 230 #  define FromT FromANSI
 231 #  define ToT   ToANSI
 232 # endif
 233 # define FromLocale    FromANSI
 234 # define ToLocale      ToANSI
 235 # define LocaleFree(s) free((char *)(s))
 236 # define FromLocaleDup FromANSI
 237 # define ToLocaleDup   ToANSI
 238
 239 #elif defined(__OS2__)
 240
 241 VLC_USED static inline char *FromLocale (const char *locale)
 242 {
 243     return locale ? FromCharset ((char *)"", locale, strlen(locale)) : NULL;
 244 }
 245
 246 VLC_USED static inline char *ToLocale (const char *utf8)
 247 {
 248     size_t outsize;
 249     return utf8 ? (char *)ToCharset ("", utf8, &outsize) : NULL;
 250 }
 251
 252 VLC_USED static inline void LocaleFree (const char *str)
 253 {
 254     free ((char *)str);
 255 }
 256
 257 VLC_USED static inline char *FromLocaleDup (const char *locale)
 258 {
 259     return FromCharset ("", locale, strlen(locale));
 260 }
 261
 262 VLC_USED static inline char *ToLocaleDup (const char *utf8)
 263 {
 264     size_t outsize;
 265     return (char *)ToCharset ("", utf8, &outsize);
 266 }
 267
 268 #else
 269
 270 # define FromLocale(l) (l)
 271 # define ToLocale(u)   (u)
 272 # define LocaleFree(s) ((void)(s))
 273 # define FromLocaleDup strdup
 274 # define ToLocaleDup   strdup
 275 #endif
 276
 277 /**
 278  * Converts a nul-terminated string from ISO-8859-1 to UTF-8.
 279  */
 280 static inline char *FromLatin1 (const char *latin)
 281 {
 282     char *str = (char *)malloc (2 * strlen (latin) + 1), *utf8 = str;
 283     unsigned char c;
 284
 285     if (str == NULL)
 286         return NULL;
 287
 288     while ((c = *(latin++)) != '\0')
 289     {
 290          if (c >= 0x80)
 291          {
 292              *(utf8++) = 0xC0 | (c >> 6);
 293              *(utf8++) = 0x80 | (c & 0x3F);
 294          }
 295          else
 296              *(utf8++) = c;
 297     }
 298     *(utf8++) = '\0';
 299
 300     utf8 = (char *)realloc (str, utf8 - str);
 301     return utf8 ? utf8 : str;
 302 }
 303
 304 /** @} */
 305
 306 VLC_API double us_strtod( const char *, char ** ) VLC_USED;
 307 VLC_API float us_strtof( const char *, char ** ) VLC_USED;
 308 VLC_API double us_atof( const char * ) VLC_USED;
 309 VLC_API int us_vasprintf( char **, const char *, va_list );
 310 VLC_API int us_asprintf( char **, const char *, ... ) VLC_USED;
 311
 312 #endif