/*****************************************************************************
* unicode.c: UTF8 <-> locale functions
*****************************************************************************
- * Copyright (C) 2005 the VideoLAN team
+ * Copyright (C) 2005-2006 the VideoLAN team
* $Id$
*
* Authors: RĂ©mi Denis-Courmont <rem # videolan.org>
*****************************************************************************/
#include <vlc/vlc.h>
#include "charset.h"
+
+#include <assert.h>
+
#include <stdio.h>
+#include <stdarg.h>
+#include <errno.h>
+#include <sys/types.h>
+#ifdef HAVE_DIRENT_H
+# include <dirent.h>
+#endif
+#ifdef UNDER_CE
+# include <tchar.h>
+#endif
+#ifdef HAVE_SYS_STAT_H
+# include <sys/stat.h>
+#endif
+#ifndef HAVE_LSTAT
+# define lstat( a, b ) stat(a, b)
+#endif
+
+#ifdef __APPLE__
+/* Define this if the OS always use UTF-8 internally */
+# define ASSUME_UTF8 1
+#endif
+
+#if !(defined (WIN32) || defined (UNDER_CE) || defined (ASSUME_UTF8))
+# define USE_ICONV 1
+#endif
+
+#if defined (USE_ICONV) && !defined (HAVE_ICONV)
+# error No UTF8 charset conversion implemented on this platform!
+#endif
+
+
+
+#ifdef USE_ICONV
+static struct {
+ vlc_iconv_t hd;
+ vlc_mutex_t lock;
+} from_locale, to_locale;
+#endif
+
+void LocaleInit( vlc_object_t *p_this )
+{
+#ifdef USE_ICONV
+ char *psz_charset;
+
+ if( vlc_current_charset( &psz_charset ) )
+ /* UTF-8 */
+ from_locale.hd = to_locale.hd = (vlc_iconv_t)(-1);
+ else
+ {
+ /* not UTF-8 */
+ char psz_buf[strlen( psz_charset ) + sizeof( "//translist" )];
+ const char *psz_conv;
+
+ /*
+ * Still allow non-ASCII characters when the locale is not set.
+ * Western Europeans are being favored for historical reasons.
+ */
+ if( strcmp( psz_charset, "ASCII" ) )
+ {
+ sprintf( psz_buf, "%s//translit", psz_charset );
+ psz_conv = psz_buf;
+ }
+ else
+ psz_conv = "ISO-8859-1//translit";
+
+ vlc_mutex_init( p_this, &from_locale.lock );
+ vlc_mutex_init( p_this, &to_locale.lock );
+ from_locale.hd = vlc_iconv_open( "UTF-8", psz_conv );
+ to_locale.hd = vlc_iconv_open( psz_conv, "UTF-8" );
+ }
+
+ free( psz_charset );
+
+ assert( (from_locale.hd == (vlc_iconv_t)(-1))
+ == (to_locale.hd == (vlc_iconv_t)(-1)) );
+#else
+ (void)p_this;
+#endif
+}
+
+void LocaleDeinit( void )
+{
+#ifdef USE_ICONV
+ if( to_locale.hd != (vlc_iconv_t)(-1) )
+ {
+ vlc_iconv_close( to_locale.hd );
+ vlc_mutex_destroy( &to_locale.lock );
+ }
+
+ if( from_locale.hd != (vlc_iconv_t)(-1) )
+ {
+ vlc_iconv_close( from_locale.hd );
+ vlc_mutex_destroy( &from_locale.lock );
+ }
+#endif
+}
+
+#if defined (WIN32) || defined (UNDER_CE)
+static char *MB2MB( const char *string, UINT fromCP, UINT toCP )
+{
+ char *out;
+ wchar_t *wide;
+ int len;
+
+ len = MultiByteToWideChar( fromCP, 0, string, -1, NULL, 0 );
+ assert( len > 0 );
+ wide = (wchar_t *)malloc (len * sizeof (wchar_t));
+ if( wide == NULL )
+ return NULL;
+
+ MultiByteToWideChar( fromCP, 0, string, -1, wide, len );
+ len = WideCharToMultiByte( toCP, 0, wide, -1, NULL, 0, NULL, NULL );
+ assert( len > 0 );
+ out = malloc( len );
+
+ WideCharToMultiByte( toCP, 0, wide, -1, out, len, NULL, NULL );
+ free( wide );
+ return out;
+}
+#endif
+
+/*****************************************************************************
+ * FromLocale: converts a locale string to UTF-8
+ *****************************************************************************/
+char *FromLocale( const char *locale )
+{
+ if( locale == NULL )
+ return NULL;
+
+#if !(defined WIN32 || defined (UNDER_CE))
+# ifdef USE_ICONV
+ if( from_locale.hd != (vlc_iconv_t)(-1) )
+ {
+ char *iptr = (char *)locale, *output, *optr;
+ size_t inb, outb;
+
+ /*
+ * We are not allowed to modify the locale pointer, even if we cast it
+ * to non-const.
+ */
+ inb = strlen( locale );
+ /* FIXME: I'm not sure about the value for the multiplication
+ * (for western people, multiplication by 3 (Latin9) is needed).
+ * While UTF-8 could reach 6 bytes, no existing code point exceeds
+ * 4 bytes. */
+ outb = inb * 4 + 1;
+
+ optr = output = malloc( outb );
+
+ vlc_mutex_lock( &from_locale.lock );
+ vlc_iconv( from_locale.hd, NULL, NULL, NULL, NULL );
+
+ while( vlc_iconv( from_locale.hd, &iptr, &inb, &optr, &outb )
+ == (size_t)-1 )
+ {
+ *optr++ = '?';
+ outb--;
+ iptr++;
+ inb--;
+ vlc_iconv( from_locale.hd, NULL, NULL, NULL, NULL );
+ }
+ vlc_mutex_unlock( &from_locale.lock );
+ *optr = '\0';
+
+ assert (inb == 0);
+ assert (*iptr == '\0');
+ assert (*optr == '\0');
+ assert (strlen( output ) == (size_t)(optr - output));
+ return realloc( output, optr - output + 1 );
+ }
+# endif /* USE_ICONV */
+ return (char *)locale;
+#else /* WIN32 */
+ return MB2MB( locale, CP_ACP, CP_UTF8 );
+#endif
+}
+
+char *FromLocaleDup( const char *locale )
+{
+#if defined (ASSUME_UTF8)
+ return strdup( locale );
+#else
+# ifdef USE_ICONV
+ if (from_locale.hd == (vlc_iconv_t)(-1))
+ return strdup( locale );
+# endif
+ return FromLocale( locale );
+#endif
+}
+
+
+/*****************************************************************************
+ * ToLocale: converts an UTF-8 string to locale
+ *****************************************************************************/
+char *ToLocale( const char *utf8 )
+{
+ if( utf8 == NULL )
+ return NULL;
+
+#if !(defined (WIN32) || defined (UNDER_CE))
+# ifdef USE_ICONV
+ if( to_locale.hd != (vlc_iconv_t)(-1) )
+ {
+ char *iptr = (char *)utf8, *output, *optr;
+ size_t inb, outb;
+
+ /*
+ * We are not allowed to modify the locale pointer, even if we cast it
+ * to non-const.
+ */
+ inb = strlen( utf8 );
+ /* FIXME: I'm not sure about the value for the multiplication
+ * (for western people, multiplication is not needed) */
+ outb = inb * 2 + 1;
+
+ optr = output = malloc( outb );
+ vlc_mutex_lock( &to_locale.lock );
+ vlc_iconv( to_locale.hd, NULL, NULL, NULL, NULL );
+
+ while( vlc_iconv( to_locale.hd, &iptr, &inb, &optr, &outb )
+ == (size_t)-1 )
+ {
+ *optr++ = '?'; /* should not happen, and yes, it sucks */
+ outb--;
+ iptr++;
+ inb--;
+ vlc_iconv( to_locale.hd, NULL, NULL, NULL, NULL );
+ }
+ vlc_mutex_unlock( &to_locale.lock );
+ *optr = '\0';
+
+ assert (inb == 0);
+ assert (*iptr == '\0');
+ assert (*optr == '\0');
+ assert (strlen( output ) == (size_t)(optr - output));
+ return realloc( output, optr - output + 1 );
+ }
+# endif /* USE_ICONV */
+ return (char *)utf8;
+#else /* WIN32 */
+ return MB2MB( utf8, CP_UTF8, CP_ACP );
+#endif
+}
+
+char *ToLocaleDup( const char *utf8 )
+{
+#if defined (ASSUME_UTF8)
+ return strdup( utf8 );
+#else
+# ifdef USE_ICONV
+ if (to_locale.hd == (vlc_iconv_t)(-1))
+ return strdup( utf8 );
+# endif
+ return ToLocale( utf8 );
+#endif
+}
+
+void LocaleFree( const char *str )
+{
+#ifdef USE_ICONV
+ if( to_locale.hd == (vlc_iconv_t)(-1) )
+ return;
+#endif
+
+#ifndef ASSUME_UTF8
+ if( str != NULL )
+ free( (char *)str );
+#endif
+}
/*****************************************************************************
- * vlc_fopen: Calls fopen() after conversion of file name to OS locale
+ * utf8_fopen: Calls fopen() after conversion of file name to OS locale
*****************************************************************************/
-FILE *vlc_fopen( const char *filename, const char *mode )
+FILE *utf8_fopen( const char *filename, const char *mode )
{
+#if !(defined (WIN32) || defined (UNDER_CE))
const char *local_name = ToLocale( filename );
if( local_name != NULL )
LocaleFree( local_name );
return stream;
}
+ else
+ errno = ENOENT;
+ return NULL;
+#else
+ wchar_t wpath[MAX_PATH + 1];
+ size_t len = strlen( mode ) + 1;
+ wchar_t wmode[len];
+
+ if( !MultiByteToWideChar( CP_UTF8, 0, filename, -1, wpath, MAX_PATH )
+ || !MultiByteToWideChar( CP_ACP, 0, mode, len, wmode, len ) )
+ {
+ errno = ENOENT;
+ return NULL;
+ }
+ wpath[MAX_PATH] = L'\0';
+
+ return _wfopen( wpath, wmode );
+#endif
+}
+
+/*****************************************************************************
+ * utf8_mkdir: Calls mkdir() after conversion of file name to OS locale
+ *****************************************************************************/
+int utf8_mkdir( const char *dirname )
+{
+#if defined (UNDER_CE) || defined (WIN32)
+ wchar_t wname[MAX_PATH + 1];
+ char mod[MAX_PATH + 1];
+ int i;
+
+ /* Convert '/' into '\' */
+ for( i = 0; *dirname; i++ )
+ {
+ if( i == MAX_PATH )
+ return -1; /* overflow */
+
+ if( *dirname == '/' )
+ mod[i] = '\\';
+ else
+ mod[i] = *dirname;
+ dirname++;
+
+ }
+ mod[i] = 0;
+
+ if( MultiByteToWideChar( CP_UTF8, 0, mod, -1, wname, MAX_PATH ) == 0 )
+ {
+ errno = ENOENT;
+ return -1;
+ }
+ wname[MAX_PATH] = L'\0';
+
+ if( CreateDirectoryW( wname, NULL ) == 0 )
+ {
+ if( GetLastError( ) == ERROR_ALREADY_EXISTS )
+ errno = EEXIST;
+ errno = ENOENT;
+ return -1;
+ }
+ return 0;
+#else
+ char *locname = ToLocale( dirname );
+ int res;
+
+ if( locname == NULL )
+ {
+ errno = ENOENT;
+ return -1;
+ }
+ res = mkdir( locname, 0755 );
+
+ LocaleFree( locname );
+ return res;
+#endif
+}
+
+
+void *utf8_opendir( const char *dirname )
+{
+ const char *local_name = ToLocale( dirname );
+
+ if( local_name != NULL )
+ {
+ DIR *dir = opendir( local_name );
+ LocaleFree( local_name );
+ return dir;
+ }
+ else
+ errno = ENOENT;
return NULL;
}
+const char *utf8_readdir( void *dir )
+{
+ struct dirent *ent;
+
+ ent = readdir( (DIR *)dir );
+ if( ent == NULL )
+ return NULL;
+
+ return FromLocale( ent->d_name );
+}
+
+
+static int utf8_statEx( const char *filename, void *buf,
+ vlc_bool_t deref )
+{
+#if !(defined (WIN32) || defined (UNDER_CE))
+# ifdef HAVE_SYS_STAT_H
+ const char *local_name = ToLocale( filename );
+
+ if( local_name != NULL )
+ {
+ int res = deref ? stat( local_name, (struct stat *)buf )
+ : lstat( local_name, (struct stat *)buf );
+ LocaleFree( local_name );
+ return res;
+ }
+ errno = ENOENT;
+# endif
+ return -1;
+#else
+ wchar_t wpath[MAX_PATH + 1];
+
+ if( !MultiByteToWideChar( CP_UTF8, 0, filename, -1, wpath, MAX_PATH ) )
+ {
+ errno = ENOENT;
+ return -1;
+ }
+ wpath[MAX_PATH] = L'\0';
+
+ /* struct _stat is just a silly Microsoft alias for struct stat */
+ return _wstat( wpath, (struct _stat *)buf );
+#endif
+}
+
+
+int utf8_stat( const char *filename, void *buf)
+{
+ return utf8_statEx( filename, buf, VLC_TRUE );
+}
+
+int utf8_lstat( const char *filename, void *buf)
+{
+ return utf8_statEx( filename, buf, VLC_FALSE );
+}
+
+/*****************************************************************************
+ * utf8_*printf: *printf with conversion from UTF-8 to local encoding
+ *****************************************************************************/
+static int utf8_vasprintf( char **str, const char *fmt, va_list ap )
+{
+ char *utf8;
+ int res = vasprintf( &utf8, fmt, ap );
+ if( res == -1 )
+ return -1;
+
+ *str = ToLocaleDup( utf8 );
+ free( utf8 );
+ return res;
+}
+
+static int utf8_vfprintf( FILE *stream, const char *fmt, va_list ap )
+{
+ char *str;
+ int res = utf8_vasprintf( &str, fmt, ap );
+ if( res == -1 )
+ return -1;
+
+ fputs( str, stream );
+ free( str );
+ return res;
+}
+
+int utf8_fprintf( FILE *stream, const char *fmt, ... )
+{
+ va_list ap;
+ int res;
+
+ va_start( ap, fmt );
+ res = utf8_vfprintf( stream, fmt, ap );
+ va_end( ap );
+ return res;
+}
+
/*****************************************************************************
* EnsureUTF8: replaces invalid/overlong UTF-8 sequences with question marks
*****************************************************************************
* Not Todo : convert Latin1 to UTF-8 on the flu
* It is not possible given UTF-8 needs more space
+ * Returns str if it was valid UTF-8, NULL if not.
*****************************************************************************/
#define isutf8cont( c ) (((c) >= 0x80) && ((c) <= 0xBF))
char *EnsureUTF8( char *str )
if( isutf8cont( c ) )
ptr += 2; /* OK */
else
+ {
*ptr++ = '?'; /* invalid */
+ str = NULL;
+ }
}
else
/* 3 bytes */
if( isutf8cont( c ) )
ptr += 3; /* OK */
else
+ {
*ptr++ = '?';
+ str = NULL;
+ }
}
else
+ {
*ptr++ = '?';
+ str = NULL;
+ }
}
else
if( ( ( c >= 0xE1 ) && ( c <= 0xEC ) ) || ( c == 0xEC )
if( isutf8cont( c ) )
ptr += 3; /* OK */
else
+ {
*ptr++ = '?';
+ str = NULL;
+ }
}
else
+ {
*ptr++ = '?';
+ str = NULL;
+ }
}
else
if( c == 0xED )
if( isutf8cont( c ) )
ptr += 3; /* OK */
else
+ {
*ptr++ = '?';
+ str = NULL;
+ }
}
else
+ {
*ptr++ = '?';
+ str = NULL;
+ }
}
else
/* 4 bytes */
if( isutf8cont( c ) )
ptr += 4; /* OK */
else
+ {
*ptr++ = '?';
+ str = NULL;
+ }
}
else
+ {
*ptr++ = '?';
+ str = NULL;
+ }
}
else
+ {
*ptr++ = '?';
+ str = NULL;
+ }
}
else
if( ( c >= 0xF1 ) && ( c <= 0xF3 ) )
if( isutf8cont( c ) )
ptr += 4; /* OK */
else
+ {
*ptr++ = '?';
+ str = NULL;
+ }
}
else
+ {
*ptr++ = '?';
+ str = NULL;
+ }
}
else
+ {
*ptr++ = '?';
+ str = NULL;
+ }
}
else
if( c == 0xF4 )
if( isutf8cont( c ) )
ptr += 4; /* OK */
else
+ {
*ptr++ = '?';
+ str = NULL;
+ }
}
else
+ {
*ptr++ = '?';
+ str = NULL;
+ }
}
else
+ {
*ptr++ = '?';
+ str = NULL;
+ }
}
else
+ {
*ptr++ = '?';
+ str = NULL;
+ }
}
return str;