ensure minimal mime support in wxHTML even without wxMimeTypesMananger

[wxWidgets.git] / src / common / intl.cpp

diff --git a/src/common/intl.cpp b/src/common/intl.cpp
index 5e03cbdc18233d250b4d75459618e59fb4761dfc..de62ef5bb6ef3903dcd816130e4034a1508a1057 100644 (file)
--- a/src/common/intl.cpp
+++ b/src/common/intl.cpp
@@ -60,6 +60,10 @@
     #include "wx/fontmap.h"         // for CharsetToEncoding()
 #endif
 
+#if defined(__WXMAC__)
+  #include  "wx/mac/private.h"  // includes mac headers
+#endif
+
 // ----------------------------------------------------------------------------
 // simple types
 // ----------------------------------------------------------------------------
@@ -615,7 +619,12 @@ bool wxLocale::Init(const wxChar *szName,
   if ( m_strShort.IsEmpty() ) {
     // FIXME I don't know how these 2 letter abbreviations are formed,
     //       this wild guess is surely wrong
-    m_strShort = tolower(szLocale[0]) + tolower(szLocale[1]);
+    if ( szLocale[0] )
+    {
+        m_strShort += (wxChar)wxTolower(szLocale[0]);
+        if ( szLocale[1] )
+            m_strShort += (wxChar)wxTolower(szLocale[1]);
+    }
   }
 
   // save the old locale to be able to restore it later
@@ -796,7 +805,7 @@ void wxLocale::AddCatalogLookupPathPrefix(const wxString& prefix)
         return wxLANGUAGE_ENGLISH;
     }
 
-    if ( langFull == _T("C") )
+    if ( langFull == _T("C") || langFull == _T("POSIX") )
     {
         // default C locale
         return wxLANGUAGE_ENGLISH;
@@ -804,18 +813,31 @@ void wxLocale::AddCatalogLookupPathPrefix(const wxString& prefix)
 
     // the language string has the following form
     //
-    //      lang[_LANG[.encoding]]
+    //      lang[_LANG][.encoding][@modifier]
+    //
+    // (see environ(5) in the Open Unix specification)
     //
-    // where lang is the primary language, LANG is a sublang
+    // where lang is the primary language, LANG is a sublang/territory,
+    // encoding is the charset to use and modifier "allows the user to select
+    // a specific instance of localization data within a single category"
     //
     // for example, the following strings are valid:
     //      fr
     //      fr_FR
     //      de_DE.iso88591
+    //      de_DE@euro
+    //      de_DE.iso88591@euro
 
     // for now we don't use the encoding, although we probably should (doing
     // translations of the msg catalogs on the fly as required) (TODO)
-    langFull = langFull.BeforeFirst(_T('.'));
+    //
+    // we don't use the modifiers neither but we probably should translate
+    // "euro" into iso885915
+    size_t posEndLang = langFull.find_first_of(_T("@."));
+    if ( posEndLang != wxString::npos )
+    {
+        langFull.Truncate(posEndLang);
+    }
 
     // in addition to the format above, we also can have full language names
     // in LANG env var - for example, SuSE is known to use LANG="german" - so
@@ -1222,7 +1244,7 @@ wxString wxLocale::GetSystemEncodingName()
 {
     wxString encname;
 
-#ifdef __WIN32__
+#if defined(__WIN32__) && !defined(__WXMICROWIN__)
     // FIXME: what is the error return value for GetACP()?
     UINT codepage = ::GetACP();
     encname.Printf(_T("windows-%u"), codepage);
@@ -1238,7 +1260,19 @@ wxString wxLocale::GetSystemEncodingName()
     free(oldLocale);
     if (alang)
     {
-        encname = wxConvLibc.cMB2WX(alang);
+#ifdef __SOLARIS__
+        // nl_langinfo() under Solaris returns 646 by default which stands for
+        // ISO-646, i.e. 7 bit ASCII and we should recognize it to avoid
+        // warnings about unrecognized encoding on each program startup
+        if ( strcmp(alang, "646") == 0 )
+        {
+            encname = _T("US-ASCII");
+        }
+        else
+#endif // __SOLARIS__
+        {
+            encname = wxConvLibc.cMB2WX(alang);
+        }
     }
     else
 #endif // HAVE_LANGINFO_H
@@ -1274,20 +1308,52 @@ wxString wxLocale::GetSystemEncodingName()
 /* static */
 wxFontEncoding wxLocale::GetSystemEncoding()
 {
-#ifdef __WIN32__
+#if defined(__WIN32__) && !defined(__WXMICROWIN__)
     UINT codepage = ::GetACP();
 
-    // wxWindows only knows about CP1250-1257
+    // wxWindows only knows about CP1250-1257, 932, 936, 949, 950
     if ( codepage >= 1250 && codepage <= 1257 )
     {
         return (wxFontEncoding)(wxFONTENCODING_CP1250 + codepage - 1250);
     }
+
+    if ( codepage == 932 )
+    {
+        return wxFONTENCODING_CP932;
+    }
+
+    if ( codepage == 936 )
+    {
+        return wxFONTENCODING_CP936;
+    }
+
+    if ( codepage == 949 )
+    {
+        return wxFONTENCODING_CP949;
+    }
+
+    if ( codepage == 950 )
+    {
+        return wxFONTENCODING_CP950;
+    }
 #elif defined(__UNIX_LIKE__) && wxUSE_FONTMAP
     wxString encname = GetSystemEncodingName();
     if ( !encname.empty() )
     {
-        return wxTheFontMapper->
+        wxFontEncoding enc = wxTheFontMapper->
             CharsetToEncoding(encname, FALSE /* not interactive */);
+
+        // this should probably be considered as a bug in CharsetToEncoding():
+        // it shouldn't return wxFONTENCODING_DEFAULT at all - but it does it
+        // for US-ASCII charset
+        //
+        // we, OTOH, definitely shouldn't return it as it doesn't make sense at
+        // all (which encoding is it?)
+        if ( enc != wxFONTENCODING_DEFAULT )
+        {
+            return enc;
+        }
+        //else: return wxFONTENCODING_SYSTEM below
     }
 #endif // Win32/Unix
 
@@ -1326,7 +1392,7 @@ const wxMB2WXbuf wxLocale::GetString(const wxChar *szOrigString,
                                      const wxChar *szDomain) const
 {
   if ( wxIsEmpty(szOrigString) )
-      return szDomain;
+    return _T("");
 
   const char *pszTrans = NULL;
 #if wxUSE_UNICODE
@@ -1372,12 +1438,13 @@ const wxMB2WXbuf wxLocale::GetString(const wxChar *szOrigString,
 
     return (wxMB2WXbuf)(szOrigString);
   }
-  else
-  {
-    return wxConvertMB2WX(pszTrans); // or preferably wxCSConv(charset).cMB2WX(pszTrans) or something,
-                                     // a macro similar to wxConvertMB2WX could be written for that
-  }
 
+  // or preferably wxCSConv(charset).cMB2WX(pszTrans) or something, a macro
+  // similar to wxConvertMB2WX could be written for that
+
+  return wxConvertMB2WX(pszTrans);
+
+  // undo the hack from the beginning of this function
   #undef szOrgString
 }