From d443ec5620269be16df27f97c0119745c5ccf28e Mon Sep 17 00:00:00 2001
From: Konstantin Isakov <ikm@users.berlios.de>
Date: Fri, 8 May 2009 10:21:03 +0000
Subject: [PATCH] + Support UTF8 for the .dsl files which begin with UTF8 BOM.

---
 src/dsl_details.cc | 15 +++++++++++++++
 src/dsl_details.hh |  3 ++-
 2 files changed, 17 insertions(+), 1 deletion(-)

diff --git a/src/dsl_details.cc b/src/dsl_details.cc
index 5c63f107..66a8cc67 100644
--- a/src/dsl_details.cc
+++ b/src/dsl_details.cc
@@ -471,6 +471,19 @@ DslScanner::DslScanner( string const & fileName ) throw( Ex, Iconv::Ex ):
   if ( firstBytes[ 0 ] == 0xFE && firstBytes[ 1 ] == 0xFF )
     encoding = Utf16BE;
   else
+  if ( firstBytes[ 0 ] == 0xEF && firstBytes[ 1 ] == 0xBB )
+  {
+    // Looks like Utf8, read one more byte
+    if ( gzread( f, firstBytes, 1 ) != 1 || firstBytes[ 0 ] != 0xBF )
+    {
+      // Either the file's too short, or the BOM is weird
+      gzclose( f );
+      throw exMalformedDslFile( fileName );
+    }
+    
+    encoding = Utf8;
+  }
+  else
   {
     if ( firstBytes[ 0 ] && !firstBytes[ 1 ] )
       encoding = Utf16LE;
@@ -720,6 +733,8 @@ char const * DslIconv::getEncodingNameFor( DslEncoding e )
       return "WINDOWS-1252";
     case Windows1251:
       return "WINDOWS-1251";
+    case Details::Utf8:
+      return "UTF-8";
     case Windows1250:
     default:
       return "WINDOWS-1250";
diff --git a/src/dsl_details.hh b/src/dsl_details.hh
index daf90f88..ad7d7113 100644
--- a/src/dsl_details.hh
+++ b/src/dsl_details.hh
@@ -28,7 +28,8 @@ enum DslEncoding
   Utf16BE,
   Windows1252,
   Windows1251,
-  Windows1250
+  Windows1250,
+  Utf8 // This is an extension. Detected solely by the UTF8 BOM.
 };