* src/misc/charset.c: New function vlc_fix_readdir_charset. It is used

under Mac OS X to transform UTF-8-MAC returned by readdir() into
   UTF-8. Darwin always uses UTF-8 internally, so we vlc_current_charset
   should always return UTF-8. Also made vlc_current_charset()
   thread-safe and added a few missing const's.

include/charset.h

@@ -1,7 +1,7 @@
 /*****************************************************************************
  * charset.h: Determine a canonical name for the current locale's character encoding.
  *****************************************************************************
- * Copyright (C) 2003 the VideoLAN team
+ * Copyright (C) 2003-2005 the VideoLAN team
  * $Id$
  *
  * Author: Derk-Jan Hartman <thedj at users.sourceforge.net>
@@ -30,6 +30,7 @@ VLC_EXPORT( void, LocaleFree, ( const char * ) );
 VLC_EXPORT( char *, FromLocale, ( const char * ) );
 VLC_EXPORT( char *, ToLocale, ( const char * ) );
 VLC_EXPORT( char *, EnsureUTF8, ( char * ) );
+VLC_EXPORT( char *, vlc_fix_readdir_charset, ( vlc_object_t *, const char * ) );
 
 # ifdef __cplusplus
 }

include/main.h

@@ -69,6 +69,8 @@ struct libvlc_t
     char *                 psz_vlcpath;
 #elif defined( SYS_DARWIN )
     char *                 psz_vlcpath;
+    vlc_iconv_t            iconv_macosx; /* for HFS+ file names */
+    vlc_mutex_t            iconv_lock;
 #elif defined( WIN32 ) && !defined( UNDER_CE )
     SIGNALOBJECTANDWAIT    SignalObjectAndWait;
     vlc_bool_t             b_fast_mutex;

include/vlc_symbols.h

@@ -387,6 +387,7 @@ struct module_symbols_t
     void (*LocaleFree_inner) (const char *);
     char * (*ToLocale_inner) (const char *);
     char * (*EnsureUTF8_inner) (char *);
+    char * (*vlc_fix_readdir_charset_inner) (vlc_object_t *, const char *);
 };
 # if defined (__PLUGIN__)
 #  define aout_FiltersCreatePipeline (p_symbols)->aout_FiltersCreatePipeline_inner
@@ -759,6 +760,7 @@ struct module_symbols_t
 #  define LocaleFree (p_symbols)->LocaleFree_inner
 #  define ToLocale (p_symbols)->ToLocale_inner
 #  define EnsureUTF8 (p_symbols)->EnsureUTF8_inner
+#  define vlc_fix_readdir_charset (p_symbols)->vlc_fix_readdir_charset_inner
 # elif defined (HAVE_DYNAMIC_PLUGINS) && !defined (__BUILTIN__)
 /******************************************************************
  * STORE_SYMBOLS: store VLC APIs into p_symbols for plugin access.
@@ -1134,6 +1136,7 @@ struct module_symbols_t
     ((p_symbols)->LocaleFree_inner) = LocaleFree; \
     ((p_symbols)->ToLocale_inner) = ToLocale; \
     ((p_symbols)->EnsureUTF8_inner) = EnsureUTF8; \
+    ((p_symbols)->vlc_fix_readdir_charset_inner) = vlc_fix_readdir_charset; \
     (p_symbols)->net_ConvertIPv4_deprecated = NULL; \
 
 # endif /* __PLUGIN__ */

src/misc/charset.c

@@ -43,6 +43,11 @@
 # include <windows.h>
 #endif
 
+#ifdef SYS_DARWIN
+#   include <errno.h>
+#   include <string.h>
+#endif
+
 #include "charset.h"
 
 typedef struct VLCCharsetAlias
@@ -198,14 +203,14 @@ static const char* vlc_charset_aliases( const char *psz_name )
 
 /* Returns charset from "language_COUNTRY.charset@modifier" string */
 #if defined WIN32 || defined OS2 || !HAVE_LANGINFO_CODESET
-static char *vlc_encoding_from_locale( char *psz_locale )
+static const char *vlc_encoding_from_locale( char *psz_locale )
 {
     char *psz_dot = strchr( psz_locale, '.' );
 
     if( psz_dot != NULL )
     {
         const char *psz_modifier;
-        static char buf[2 + 10 + 1];
+        char buf[2 + 10 + 1];
 
         psz_dot++;
 
@@ -223,7 +228,7 @@ static char *vlc_encoding_from_locale( char *psz_locale )
         }
     }
     /* try language mapping */
-    return (char *)vlc_encoding_from_language( psz_locale );
+    return vlc_encoding_from_language( psz_locale );
 }
 #endif
 
@@ -231,7 +236,7 @@ vlc_bool_t vlc_current_charset( char **psz_charset )
 {
     const char *psz_codeset;
 
-#if !(defined WIN32 || defined OS2)
+#if !(defined WIN32 || defined OS2 || defined SYS_DARWIN)
 
 # if HAVE_LANGINFO_CODESET
     /* Most systems support nl_langinfo( CODESET ) nowadays.  */
@@ -263,9 +268,14 @@ vlc_bool_t vlc_current_charset( char **psz_charset )
     psz_codeset = vlc_encoding_from_locale( (char *)psz_locale );
 # endif /* HAVE_LANGINFO_CODESET */
 
+#elif defined SYS_DARWIN
+
+    /* Darwin is always using UTF-8 internally. */
+    psz_codeset = "UTF-8";
+
 #elif defined WIN32
 
-    static char buf[2 + 10 + 1];
+    char buf[2 + 10 + 1];
 
     /* Woe32 has a function returning the locale's codepage as a number.  */
     sprintf( buf, "CP%u", GetACP() );
@@ -274,7 +284,7 @@ vlc_bool_t vlc_current_charset( char **psz_charset )
 #elif defined OS2
 
     const char *psz_locale;
-    static char buf[2 + 10 + 1];
+    char buf[2 + 10 + 1];
     ULONG cp[3];
     ULONG cplen;
 
@@ -318,7 +328,7 @@ vlc_bool_t vlc_current_charset( char **psz_charset )
     }
 
     if( psz_charset )
-        *psz_charset = strdup((char *)psz_codeset);
+        *psz_charset = strdup(psz_codeset);
 
     if( !strcasecmp(psz_codeset, "UTF8") || !strcasecmp(psz_codeset, "UTF-8") )
         return VLC_TRUE;
@@ -326,3 +336,34 @@ vlc_bool_t vlc_current_charset( char **psz_charset )
     return VLC_FALSE;
 }
 
+char *vlc_fix_readdir_charset( vlc_object_t *p_this, const char *psz_string )
+{
+#ifdef SYS_DARWIN
+    if ( p_this->p_libvlc->iconv_macosx != (vlc_iconv_t)-1 )
+    {
+        const char *psz_in = psz_string;
+        size_t i_in = strlen(psz_in);
+        size_t i_out = i_in * 2;
+        char *psz_utf8 = malloc(i_out + 1);
+        char *psz_out = psz_utf8;
+
+        vlc_mutex_lock( &p_this->p_libvlc->iconv_lock );
+        size_t i_ret = vlc_iconv( p_this->p_libvlc->iconv_macosx,
+                                  &psz_in, &i_in, &psz_out, &i_out );
+        vlc_mutex_unlock( &p_this->p_libvlc->iconv_lock );
+        if( i_ret == (size_t)-1 || i_in )
+        {
+            msg_Warn( p_this,
+                      "failed to convert \"%s\" from HFS+ charset (%s)",
+                      psz_string, strerror(errno) );
+            free( psz_utf8 );
+            return strdup( psz_string );
+        }
+
+        *psz_out = '\0';
+        return psz_utf8;
+    }
+#endif
+
+    return strdup( psz_string );
+}

src/misc/darwin_specific.m

@@ -27,6 +27,7 @@
 #include <vlc/vlc.h>
 
 #include <Cocoa/Cocoa.h>
+#include <CoreFoundation/CFString.h>
 
 #ifdef HAVE_LOCALE_H
 #   include <locale.h>
@@ -114,6 +115,9 @@ void system_Init( vlc_t *p_this, int *pi_argc, char *ppsz_argv[] )
 
         [o_pool release];
     }
+
+    vlc_mutex_init( p_this, &p_this->p_libvlc->iconv_lock );
+    p_this->p_libvlc->iconv_macosx = vlc_iconv_open( "UTF-8", "UTF-8-MAC" );
 }
 
 /*****************************************************************************
@@ -130,5 +134,9 @@ void system_Configure( vlc_t *p_this, int *pi_argc, char *ppsz_argv[] )
 void system_End( vlc_t *p_this )
 {
     free( p_this->p_libvlc->psz_vlcpath );
+
+    if ( p_this->p_libvlc->iconv_macosx != (vlc_iconv_t)-1 )
+        vlc_iconv_close( p_this->p_libvlc->iconv_macosx );
+    vlc_mutex_destroy( &p_this->p_libvlc->iconv_lock );
 }