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 __APPLE__
 149 # include <CoreFoundation/CoreFoundation.h>
 150
 151 /* Obtains a copy of the contents of a CFString in specified encoding.
 152  * Returns char* (must be freed by caller) or NULL on failure.
 153  */
 154 VLC_USED static inline char *FromCFString(const CFStringRef cfString,
 155     const CFStringEncoding cfStringEncoding)
 156 {
 157     // Try the quick way to obtain the buffer
 158     const char *tmpBuffer = CFStringGetCStringPtr(cfString, cfStringEncoding);
 159
 160     if (tmpBuffer != NULL) {
 161        return strdup(tmpBuffer);
 162     }
 163
 164     // The quick way did not work, try the long way
 165     CFIndex length = CFStringGetLength(cfString);
 166     CFIndex maxSize =
 167         CFStringGetMaximumSizeForEncoding(length, cfStringEncoding);
 168
 169     // If result would exceed LONG_MAX, kCFNotFound is returned
 170     if (unlikely(maxSize == kCFNotFound)) {
 171         return NULL;
 172     }
 173
 174     // Account for the null terminator
 175     maxSize++;
 176
 177     char *buffer = (char *)malloc(maxSize);
 178
 179     if (unlikely(buffer == NULL)) {
 180         return NULL;
 181     }
 182
 183     // Copy CFString in requested encoding to buffer
 184     Boolean success = CFStringGetCString(cfString, buffer, maxSize, cfStringEncoding);
 185
 186     if (!success)
 187         FREENULL(buffer);
 188     return buffer;
 189 }
 190 #endif
 191
 192 #ifdef _WIN32
 193 VLC_USED
 194 static inline char *FromWide (const wchar_t *wide)
 195 {
 196     size_t len = WideCharToMultiByte (CP_UTF8, 0, wide, -1, NULL, 0, NULL, NULL);
 197     if (len == 0)
 198         return NULL;
 199
 200     char *out = (char *)malloc (len);
 201
 202     if (likely(out))
 203         WideCharToMultiByte (CP_UTF8, 0, wide, -1, out, len, NULL, NULL);
 204     return out;
 205 }
 206
 207 VLC_USED
 208 static inline wchar_t *ToWide (const char *utf8)
 209 {
 210     int len = MultiByteToWideChar (CP_UTF8, 0, utf8, -1, NULL, 0);
 211     if (len == 0)
 212         return NULL;
 213
 214     wchar_t *out = (wchar_t *)malloc (len * sizeof (wchar_t));
 215
 216     if (likely(out))
 217         MultiByteToWideChar (CP_UTF8, 0, utf8, -1, out, len);
 218     return out;
 219 }
 220
 221 VLC_USED VLC_MALLOC
 222 static inline char *ToCodePage (unsigned cp, const char *utf8)
 223 {
 224     wchar_t *wide = ToWide (utf8);
 225     if (wide == NULL)
 226         return NULL;
 227
 228     size_t len = WideCharToMultiByte (cp, 0, wide, -1, NULL, 0, NULL, NULL);
 229     if (len == 0) {
 230         free(wide);
 231         return NULL;
 232     }
 233
 234     char *out = (char *)malloc (len);
 235     if (likely(out != NULL))
 236         WideCharToMultiByte (cp, 0, wide, -1, out, len, NULL, NULL);
 237     free (wide);
 238     return out;
 239 }
 240
 241 VLC_USED VLC_MALLOC
 242 static inline char *FromCodePage (unsigned cp, const char *mb)
 243 {
 244     int len = MultiByteToWideChar (cp, 0, mb, -1, NULL, 0);
 245     if (len == 0)
 246         return NULL;
 247
 248     wchar_t *wide = (wchar_t *)malloc (len * sizeof (wchar_t));
 249     if (unlikely(wide == NULL))
 250         return NULL;
 251     MultiByteToWideChar (cp, 0, mb, -1, wide, len);
 252
 253     char *utf8 = FromWide (wide);
 254     free (wide);
 255     return utf8;
 256 }
 257
 258 VLC_USED VLC_MALLOC
 259 static inline char *FromANSI (const char *ansi)
 260 {
 261     return FromCodePage (GetACP (), ansi);
 262 }
 263
 264 VLC_USED VLC_MALLOC
 265 static inline char *ToANSI (const char *utf8)
 266 {
 267     return ToCodePage (GetACP (), utf8);
 268 }
 269
 270 # ifdef UNICODE
 271 #  define FromT FromWide
 272 #  define ToT   ToWide
 273 # else
 274 #  define FromT FromANSI
 275 #  define ToT   ToANSI
 276 # endif
 277 # define FromLocale    FromANSI
 278 # define ToLocale      ToANSI
 279 # define LocaleFree(s) free((char *)(s))
 280 # define FromLocaleDup FromANSI
 281 # define ToLocaleDup   ToANSI
 282
 283 #elif defined(__OS2__)
 284
 285 VLC_USED static inline char *FromLocale (const char *locale)
 286 {
 287     return locale ? FromCharset ((char *)"", locale, strlen(locale)) : NULL;
 288 }
 289
 290 VLC_USED static inline char *ToLocale (const char *utf8)
 291 {
 292     size_t outsize;
 293     return utf8 ? (char *)ToCharset ("", utf8, &outsize) : NULL;
 294 }
 295
 296 VLC_USED static inline void LocaleFree (const char *str)
 297 {
 298     free ((char *)str);
 299 }
 300
 301 VLC_USED static inline char *FromLocaleDup (const char *locale)
 302 {
 303     return FromCharset ("", locale, strlen(locale));
 304 }
 305
 306 VLC_USED static inline char *ToLocaleDup (const char *utf8)
 307 {
 308     size_t outsize;
 309     return (char *)ToCharset ("", utf8, &outsize);
 310 }
 311
 312 #else
 313
 314 # define FromLocale(l) (l)
 315 # define ToLocale(u)   (u)
 316 # define LocaleFree(s) ((void)(s))
 317 # define FromLocaleDup strdup
 318 # define ToLocaleDup   strdup
 319 #endif
 320
 321 /**
 322  * Converts a nul-terminated string from ISO-8859-1 to UTF-8.
 323  */
 324 static inline char *FromLatin1 (const char *latin)
 325 {
 326     char *str = (char *)malloc (2 * strlen (latin) + 1), *utf8 = str;
 327     unsigned char c;
 328
 329     if (str == NULL)
 330         return NULL;
 331
 332     while ((c = *(latin++)) != '\0')
 333     {
 334          if (c >= 0x80)
 335          {
 336              *(utf8++) = 0xC0 | (c >> 6);
 337              *(utf8++) = 0x80 | (c & 0x3F);
 338          }
 339          else
 340              *(utf8++) = c;
 341     }
 342     *(utf8++) = '\0';
 343
 344     utf8 = (char *)realloc (str, utf8 - str);
 345     return utf8 ? utf8 : str;
 346 }
 347
 348 /** @} */
 349
 350 VLC_API double us_strtod( const char *, char ** ) VLC_USED;
 351 VLC_API float us_strtof( const char *, char ** ) VLC_USED;
 352 VLC_API double us_atof( const char * ) VLC_USED;
 353 VLC_API int us_vasprintf( char **, const char *, va_list );
 354 VLC_API int us_asprintf( char **, const char *, ... ) VLC_USED;
 355
 356 #endif