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  *
   7  * Author: Rémi Denis-Courmont <rem # videolan,org>
   8  *
   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.
  13  *
  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.
  18  *
  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  *****************************************************************************/
  23
  24 #ifndef VLC_CHARSET_H
  25 #define VLC_CHARSET_H 1
  26
  27 /**
  28  * \file
  29  * Characters sets handling
  30  *
  31  * \ingroup strings
  32  * @{
  33  */
  34
  35 /**
  36  * Decodes a code point from UTF-8.
  37  *
  38  * Converts the first character in a UTF-8 sequence into a Unicode code point.
  39  *
  40  * \param str an UTF-8 bytes sequence [IN]
  41  * \param pwc address of a location to store the code point [OUT]
  42  *
  43  * \return the number of bytes occupied by the decoded code point
  44  *
  45  * \retval (size_t)-1 not a valid UTF-8 sequence
  46  * \retval 0 null character (i.e. str points to an empty string)
  47  * \retval 1 (non-null) ASCII character
  48  * \retval 2-4 non-ASCII character
  49  */
  50 VLC_API size_t vlc_towc(const char *str, uint32_t *restrict pwc);
  51
  52 /**
  53  * Checks UTF-8 validity.
  54  *
  55  * Checks whether a null-terminated string is a valid UTF-8 bytes sequence.
  56  *
  57  * \param str string to check
  58  *
  59  * \retval str the string is a valid null-terminated UTF-8 sequence
  60  * \retval NULL the string is not an UTF-8 sequence
  61  */
  62 VLC_USED static inline const char *IsUTF8(const char *str)
  63 {
  64     size_t n;
  65     uint32_t cp;
  66
  67     while ((n = vlc_towc(str, &cp)) != 0)
  68         if (likely(n != (size_t)-1))
  69             str += n;
  70         else
  71             return NULL;
  72     return str;
  73 }
  74
  75 /**
  76  * Checks ASCII validity.
  77  *
  78  * Checks whether a null-terminated string is a valid ASCII bytes sequence
  79  * (non-printable ASCII characters 1-31 are permitted).
  80  *
  81  * \param str string to check
  82  *
  83  * \retval str the string is a valid null-terminated ASCII sequence
  84  * \retval NULL the string is not an ASCII sequence
  85  */
  86 VLC_USED static inline const char *IsASCII(const char *str)
  87 {
  88     unsigned char c;
  89
  90     for (const char *p = str; (c = *p) != '\0'; p++)
  91         if (c >= 0x80)
  92             return NULL;
  93     return str;
  94 }
  95
  96 /**
  97  * Removes non-UTF-8 sequences.
  98  *
  99  * Replaces invalid or <i>over-long</i> UTF-8 bytes sequences within a
 100  * null-terminated string with question marks. This is so that the string can
 101  * be printed at least partially.
 102  *
 103  * \warning Do not use this were correctness is critical. use IsUTF8() and
 104  * handle the error case instead. This function is mainly for display or debug.
 105  *
 106  * \note Converting from Latin-1 to UTF-8 in place is not possible (the string
 107  * size would be increased). So it is not attempted even if it would otherwise
 108  * be less disruptive.
 109  *
 110  * \retval str the string is a valid null-terminated UTF-8 sequence
 111  *             (i.e. no changes were made)
 112  * \retval NULL the string is not an UTF-8 sequence
 113  */
 114 static inline char *EnsureUTF8(char *str)
 115 {
 116     char *ret = str;
 117     size_t n;
 118     uint32_t cp;
 119
 120     while ((n = vlc_towc(str, &cp)) != 0)
 121         if (likely(n != (size_t)-1))
 122             str += n;
 123         else
 124         {
 125             *str++ = '?';
 126             ret = NULL;
 127         }
 128     return ret;
 129 }
 130
 131 /* iconv wrappers (defined in src/extras/libc.c) */
 132 #define VLC_ICONV_ERR ((size_t) -1)
 133 typedef void *vlc_iconv_t;
 134 VLC_API vlc_iconv_t vlc_iconv_open( const char *, const char * ) VLC_USED;
 135 VLC_API size_t vlc_iconv( vlc_iconv_t, const char **, size_t *, char **, size_t * ) VLC_USED;
 136 VLC_API int vlc_iconv_close( vlc_iconv_t );
 137
 138 #include <stdarg.h>
 139
 140 VLC_API int utf8_vfprintf( FILE *stream, const char *fmt, va_list ap );
 141 VLC_API int utf8_fprintf( FILE *, const char *, ... ) VLC_FORMAT( 2, 3 );
 142 VLC_API char * vlc_strcasestr(const char *, const char *) VLC_USED;
 143
 144 VLC_API char * FromCharset( const char *charset, const void *data, size_t data_size ) VLC_USED;
 145 VLC_API void * ToCharset( const char *charset, const char *in, size_t *outsize ) VLC_USED;
 146
 147 #ifdef __APPLE__
 148 # include <CoreFoundation/CoreFoundation.h>
 149
 150 /* Obtains a copy of the contents of a CFString in specified encoding.
 151  * Returns char* (must be freed by caller) or NULL on failure.
 152  */
 153 VLC_USED static inline char *FromCFString(const CFStringRef cfString,
 154     const CFStringEncoding cfStringEncoding)
 155 {
 156     // Try the quick way to obtain the buffer
 157     const char *tmpBuffer = CFStringGetCStringPtr(cfString, cfStringEncoding);
 158
 159     if (tmpBuffer != NULL) {
 160        return strdup(tmpBuffer);
 161     }
 162
 163     // The quick way did not work, try the long way
 164     CFIndex length = CFStringGetLength(cfString);
 165     CFIndex maxSize =
 166         CFStringGetMaximumSizeForEncoding(length, cfStringEncoding);
 167
 168     // If result would exceed LONG_MAX, kCFNotFound is returned
 169     if (unlikely(maxSize == kCFNotFound)) {
 170         return NULL;
 171     }
 172
 173     // Account for the null terminator
 174     maxSize++;
 175
 176     char *buffer = (char *)malloc(maxSize);
 177
 178     if (unlikely(buffer == NULL)) {
 179         return NULL;
 180     }
 181
 182     // Copy CFString in requested encoding to buffer
 183     Boolean success = CFStringGetCString(cfString, buffer, maxSize, cfStringEncoding);
 184
 185     if (!success)
 186         FREENULL(buffer);
 187     return buffer;
 188 }
 189 #endif
 190
 191 #ifdef _WIN32
 192 VLC_USED
 193 static inline char *FromWide (const wchar_t *wide)
 194 {
 195     size_t len = WideCharToMultiByte (CP_UTF8, 0, wide, -1, NULL, 0, NULL, NULL);
 196     if (len == 0)
 197         return NULL;
 198
 199     char *out = (char *)malloc (len);
 200
 201     if (likely(out))
 202         WideCharToMultiByte (CP_UTF8, 0, wide, -1, out, len, NULL, NULL);
 203     return out;
 204 }
 205
 206 VLC_USED
 207 static inline wchar_t *ToWide (const char *utf8)
 208 {
 209     int len = MultiByteToWideChar (CP_UTF8, 0, utf8, -1, NULL, 0);
 210     if (len == 0)
 211         return NULL;
 212
 213     wchar_t *out = (wchar_t *)malloc (len * sizeof (wchar_t));
 214
 215     if (likely(out))
 216         MultiByteToWideChar (CP_UTF8, 0, utf8, -1, out, len);
 217     return out;
 218 }
 219
 220 VLC_USED VLC_MALLOC
 221 static inline char *ToCodePage (unsigned cp, const char *utf8)
 222 {
 223     wchar_t *wide = ToWide (utf8);
 224     if (wide == NULL)
 225         return NULL;
 226
 227     size_t len = WideCharToMultiByte (cp, 0, wide, -1, NULL, 0, NULL, NULL);
 228     if (len == 0) {
 229         free(wide);
 230         return NULL;
 231     }
 232
 233     char *out = (char *)malloc (len);
 234     if (likely(out != NULL))
 235         WideCharToMultiByte (cp, 0, wide, -1, out, len, NULL, NULL);
 236     free (wide);
 237     return out;
 238 }
 239
 240 VLC_USED VLC_MALLOC
 241 static inline char *FromCodePage (unsigned cp, const char *mb)
 242 {
 243     int len = MultiByteToWideChar (cp, 0, mb, -1, NULL, 0);
 244     if (len == 0)
 245         return NULL;
 246
 247     wchar_t *wide = (wchar_t *)malloc (len * sizeof (wchar_t));
 248     if (unlikely(wide == NULL))
 249         return NULL;
 250     MultiByteToWideChar (cp, 0, mb, -1, wide, len);
 251
 252     char *utf8 = FromWide (wide);
 253     free (wide);
 254     return utf8;
 255 }
 256
 257 VLC_USED VLC_MALLOC
 258 static inline char *FromANSI (const char *ansi)
 259 {
 260     return FromCodePage (GetACP (), ansi);
 261 }
 262
 263 VLC_USED VLC_MALLOC
 264 static inline char *ToANSI (const char *utf8)
 265 {
 266     return ToCodePage (GetACP (), utf8);
 267 }
 268
 269 # define FromLocale    FromANSI
 270 # define ToLocale      ToANSI
 271 # define LocaleFree(s) free((char *)(s))
 272 # define FromLocaleDup FromANSI
 273 # define ToLocaleDup   ToANSI
 274
 275 #elif defined(__OS2__)
 276
 277 VLC_USED static inline char *FromLocale (const char *locale)
 278 {
 279     return locale ? FromCharset ((char *)"", locale, strlen(locale)) : NULL;
 280 }
 281
 282 VLC_USED static inline char *ToLocale (const char *utf8)
 283 {
 284     size_t outsize;
 285     return utf8 ? (char *)ToCharset ("", utf8, &outsize) : NULL;
 286 }
 287
 288 VLC_USED static inline void LocaleFree (const char *str)
 289 {
 290     free ((char *)str);
 291 }
 292
 293 VLC_USED static inline char *FromLocaleDup (const char *locale)
 294 {
 295     return FromCharset ("", locale, strlen(locale));
 296 }
 297
 298 VLC_USED static inline char *ToLocaleDup (const char *utf8)
 299 {
 300     size_t outsize;
 301     return (char *)ToCharset ("", utf8, &outsize);
 302 }
 303
 304 #else
 305
 306 # define FromLocale(l) (l)
 307 # define ToLocale(u)   (u)
 308 # define LocaleFree(s) ((void)(s))
 309 # define FromLocaleDup strdup
 310 # define ToLocaleDup   strdup
 311 #endif
 312
 313 /**
 314  * Converts a nul-terminated string from ISO-8859-1 to UTF-8.
 315  */
 316 static inline char *FromLatin1 (const char *latin)
 317 {
 318     char *str = (char *)malloc (2 * strlen (latin) + 1), *utf8 = str;
 319     unsigned char c;
 320
 321     if (str == NULL)
 322         return NULL;
 323
 324     while ((c = *(latin++)) != '\0')
 325     {
 326          if (c >= 0x80)
 327          {
 328              *(utf8++) = 0xC0 | (c >> 6);
 329              *(utf8++) = 0x80 | (c & 0x3F);
 330          }
 331          else
 332              *(utf8++) = c;
 333     }
 334     *(utf8++) = '\0';
 335
 336     utf8 = (char *)realloc (str, utf8 - str);
 337     return utf8 ? utf8 : str;
 338 }
 339
 340 /** @} */
 341
 342 VLC_API double us_strtod( const char *, char ** ) VLC_USED;
 343 VLC_API float us_strtof( const char *, char ** ) VLC_USED;
 344 VLC_API double us_atof( const char * ) VLC_USED;
 345 VLC_API int us_vasprintf( char **, const char *, va_list );
 346 VLC_API int us_asprintf( char **, const char *, ... ) VLC_USED;
 347
 348 #endif