Inline FromUTF16 and use Win32 API

(suppress one more ugly piece of code)

include/charset.h

@@ -2,9 +2,11 @@
  * charset.h: Determine a canonical name for the current locale's character encoding.
  *****************************************************************************
  * Copyright (C) 2003-2005 the VideoLAN team
+ * Copyright © 2005-2006 Rémi Denis-Courmont
  * $Id$
  *
- * Author: Derk-Jan Hartman <thedj at users.sourceforge.net>
+ * Authors: Rémi Denis-Courmont <rem # videolan,org>
+ *          Derk-Jan Hartman <thedj at users.sourceforge.net>
  *
  * This program is free software; you can redistribute it and/or modify
  * it under the terms of the GNU General Public License as published by
@@ -53,8 +55,7 @@ VLC_EXPORT( char *, EnsureUTF8, ( char * ) );
 VLC_EXPORT( const char *, IsUTF8, ( const char * ) );
 
 #ifdef WIN32
-VLC_EXPORT( char *, FromUTF16, ( const uint16_t * ) );
-#define FromWide FromUTF16
+VLC_EXPORT( char *, FromWide, ( const wchar_t * ) );
 #endif
 
 

include/vlc_symbols.h

@@ -483,7 +483,7 @@ struct module_symbols_t
     char * (*convert_xml_special_chars_inner) (const char *psz_content);
     char * (*unescape_URI_duplicate_inner) (const char *psz);
     void (*resolve_xml_special_chars_inner) (char *psz_value);
-    char * (*FromUTF16_inner) (const uint16_t *);
+    void *FromUTF16_deprecated;
     const char * (*IsUTF8_inner) (const char *);
     const char * (*GetFallbackEncoding_inner) (void);
     int (*utf8_scandir_inner) (const char *dirname, char ***namelist, int (*select)( const char * ), int (*compar)( const char **, const char ** ));
@@ -557,6 +557,7 @@ struct module_symbols_t
     int (*net_ListenSingle_inner) (vlc_object_t *p_this, const char *psz_host, int i_port, int family, int socktype, int protocol);
     char * (*str_format_time_inner) (char *);
     char * (*__str_format_meta_inner) (vlc_object_t *, char *);
+    char * (*FromWide_inner) (const wchar_t *);
 };
 # if defined (__PLUGIN__)
 #  define aout_FiltersCreatePipeline (p_symbols)->aout_FiltersCreatePipeline_inner
@@ -979,7 +980,6 @@ struct module_symbols_t
 #  define convert_xml_special_chars (p_symbols)->convert_xml_special_chars_inner
 #  define unescape_URI_duplicate (p_symbols)->unescape_URI_duplicate_inner
 #  define resolve_xml_special_chars (p_symbols)->resolve_xml_special_chars_inner
-#  define FromUTF16 (p_symbols)->FromUTF16_inner
 #  define IsUTF8 (p_symbols)->IsUTF8_inner
 #  define GetFallbackEncoding (p_symbols)->GetFallbackEncoding_inner
 #  define utf8_scandir (p_symbols)->utf8_scandir_inner
@@ -1035,6 +1035,7 @@ struct module_symbols_t
 #  define net_ListenSingle (p_symbols)->net_ListenSingle_inner
 #  define str_format_time (p_symbols)->str_format_time_inner
 #  define __str_format_meta (p_symbols)->__str_format_meta_inner
+#  define FromWide (p_symbols)->FromWide_inner
 # elif defined (HAVE_DYNAMIC_PLUGINS) && !defined (__BUILTIN__)
 /******************************************************************
  * STORE_SYMBOLS: store VLC APIs into p_symbols for plugin access.
@@ -1460,7 +1461,6 @@ struct module_symbols_t
     ((p_symbols)->convert_xml_special_chars_inner) = convert_xml_special_chars; \
     ((p_symbols)->unescape_URI_duplicate_inner) = unescape_URI_duplicate; \
     ((p_symbols)->resolve_xml_special_chars_inner) = resolve_xml_special_chars; \
-    ((p_symbols)->FromUTF16_inner) = FromUTF16; \
     ((p_symbols)->IsUTF8_inner) = IsUTF8; \
     ((p_symbols)->GetFallbackEncoding_inner) = GetFallbackEncoding; \
     ((p_symbols)->utf8_scandir_inner) = utf8_scandir; \
@@ -1516,6 +1516,7 @@ struct module_symbols_t
     ((p_symbols)->net_ListenSingle_inner) = net_ListenSingle; \
     ((p_symbols)->str_format_time_inner) = str_format_time; \
     ((p_symbols)->__str_format_meta_inner) = __str_format_meta; \
+    ((p_symbols)->FromWide_inner) = FromWide; \
     (p_symbols)->net_ConvertIPv4_deprecated = NULL; \
     (p_symbols)->__sout_CfgParse_deprecated = NULL; \
     (p_symbols)->sout_CfgCreate_deprecated = NULL; \
@@ -1560,6 +1561,7 @@ struct module_symbols_t
     (p_symbols)->vlc_HashInsert_deprecated = NULL; \
     (p_symbols)->vlc_HashLookup_deprecated = NULL; \
     (p_symbols)->vlc_HashRetrieve_deprecated = NULL; \
+    (p_symbols)->FromUTF16_deprecated = NULL; \
     (p_symbols)->playlist_ItemNewFromInput_deprecated = NULL; \
     (p_symbols)->playlist_PlaylistAdd_deprecated = NULL; \
     (p_symbols)->playlist_PlaylistAddExt_deprecated = NULL; \

src/misc/unicode.c

@@ -7,9 +7,6 @@
  *
  * Authors: Rémi Denis-Courmont <rem # videolan.org>
  *
- * UTF16toUTF8() adapted from Perl 5 (also GPL'd)
- * Copyright (C) 1998-2002, Larry Wall
- *
  * This program is free software; you can redistribute it and/or modify
  * it under the terms of the GNU General Public License as published by
  * the Free Software Foundation; either version 2 of the License, or
@@ -778,96 +775,3 @@ const char *IsUTF8( const char *str )
 {
     return CheckUTF8( (char *)str, 0 );
 }
-
-
-/**
- * UTF16toUTF8: converts UTF-16 (host byte order) to UTF-8
- *
- * @param src UTF-16 bytes sequence, aligned on a 16-bits boundary
- * @param len number of uint16_t to convert
- */
-static inline char *
-UTF16toUTF8( const uint16_t *in, size_t len, size_t *newlen )
-{
-    char *res, *out;
-
-    /* allocate memory */
-    out = res = (char *)malloc( 3 * len );
-    if( res == NULL )
-        return NULL;
-
-    while( len > 0 )
-    {
-        uint32_t uv = *in;
-
-        in++;
-        len--;
-
-        if( uv < 0x80 )
-        {
-            *out++ = uv;
-            continue;
-        }
-        if( uv < 0x800 )
-        {
-            *out++ = (( uv >>  6)         | 0xc0);
-            *out++ = (( uv        & 0x3f) | 0x80);
-            continue;
-        }
-        if( (uv >= 0xd800) && (uv < 0xdbff) )
-        {   /* surrogates */
-            uint16_t low = GetWBE( in );
-            in++;
-            len--;
-
-            if( (low < 0xdc00) || (low >= 0xdfff) )
-            {
-                *out++ = '?'; /* Malformed surrogate */
-                continue;
-            }
-            else
-                uv = ((uv - 0xd800) << 10) + (low - 0xdc00) + 0x10000;
-        }
-        if( uv < 0x10000 )
-        {
-            *out++ = (( uv >> 12)         | 0xe0);
-            *out++ = (((uv >>  6) & 0x3f) | 0x80);
-            *out++ = (( uv        & 0x3f) | 0x80);
-            continue;
-        }
-        else
-        {
-            *out++ = (( uv >> 18)         | 0xf0);
-            *out++ = (((uv >> 12) & 0x3f) | 0x80);
-            *out++ = (((uv >>  6) & 0x3f) | 0x80);
-            *out++ = (( uv        & 0x3f) | 0x80);
-            continue;
-        }
-    }
-    len = out - res;
-    res = realloc( res, len );
-    if( newlen != NULL )
-        *newlen = len;
-    return res;
-}
-
-
-/**
- * FromUTF16(): converts an UTF-16 string to UTF-8.
- *
- * @param src UTF-16 bytes sequence, aligned on a 16-bits boundary.
- *
- * @return the result of the conversion (must be free()'d),
- * or NULL in case of error.
- */
-char *FromUTF16( const uint16_t *src )
-{
-    const uint16_t *in;
-    size_t len;
-
-    /* determine the size of the string */
-    for( len = 1, in = src; *in; len++ )
-        in++;
-
-    return UTF16toUTF8( src, len, NULL );
-}