From cecfa244e00c6de2652a311cbf2427feabf2b224 Mon Sep 17 00:00:00 2001
From: David Huynh <dfhuynh@gmail.com>
Date: Wed, 6 Apr 2011 21:09:21 +0000
Subject: [PATCH] Changed to UTF-8 encoding

git-svn-id: http://google-refine.googlecode.com/svn/trunk@2039 7d457c2a-affb-35e4-300a-418c747d4874
---
 .../refine/clustering/binning/Metaphone3.java | 72 +++++++++----------
 1 file changed, 36 insertions(+), 36 deletions(-)

diff --git a/main/src/com/google/refine/clustering/binning/Metaphone3.java b/main/src/com/google/refine/clustering/binning/Metaphone3.java
index 56f4cd6fc..366e8863e 100644
--- a/main/src/com/google/refine/clustering/binning/Metaphone3.java
+++ b/main/src/com/google/refine/clustering/binning/Metaphone3.java
@@ -496,34 +496,34 @@ public class Metaphone3 {
 			|| (inChar == 'O') 
 			|| (inChar == 'U') 
 			|| (inChar == 'Y') 
-			|| (inChar == '�')
-			|| (inChar == '�')
-			|| (inChar == '�')
-			|| (inChar == '�')
-			|| (inChar == '�')
-			|| (inChar == '�')
-			|| (inChar == '�')
-			|| (inChar == '�')
-			|| (inChar == '�')
-	 		|| (inChar == '�')
-			|| (inChar == '�')
-			|| (inChar == '�')
-			|| (inChar == '�')
-			|| (inChar == '�')
-	 		|| (inChar == '�')
-			|| (inChar == '�')
-			|| (inChar == '�')
-			|| (inChar == '�')
-			|| (inChar == '�')
-			|| (inChar == '�')
-			|| (inChar == '�')
-			|| (inChar == '�')
-			|| (inChar == '�')
-			|| (inChar == '�')
-			|| (inChar == '�')
-			|| (inChar == '�')
-			|| (inChar == '�')
-			|| (inChar == '�'))
+			|| (inChar == 'À')
+			|| (inChar == 'Á')
+			|| (inChar == 'Â')
+			|| (inChar == 'Ã')
+			|| (inChar == 'Ä')
+			|| (inChar == 'Å')
+			|| (inChar == 'Æ')
+			|| (inChar == 'È')
+			|| (inChar == 'É')
+	 		|| (inChar == 'Ê')
+			|| (inChar == 'Ë')
+			|| (inChar == 'Ì')
+			|| (inChar == 'Í')
+			|| (inChar == 'Î')
+	 		|| (inChar == 'Ï')
+			|| (inChar == 'Ò')
+			|| (inChar == 'Ó')
+			|| (inChar == 'Ô')
+			|| (inChar == 'Õ')
+			|| (inChar == 'Ö')
+			|| (inChar == 'Œ')
+			|| (inChar == 'Ø')
+			|| (inChar == 'Ù')
+			|| (inChar == 'Ú')
+			|| (inChar == 'Û')
+			|| (inChar == 'Ü')
+			|| (inChar == 'Ý')
+			|| (inChar == 'Ÿ'))
 		{
 	        return true;
 		}
@@ -780,8 +780,8 @@ public class Metaphone3 {
 					Encode_B();
 	                break;
 	                        
-	            case '�':
-				case '�':
+	            case 'ß':
+				case 'Ç':
 
 	                MetaphAdd("S");
 	                m_current++;
@@ -837,7 +837,7 @@ public class Metaphone3 {
 					Encode_N();
 	                break;
 
-	            case '�':
+	            case 'Ñ':
 	                
 	                MetaphAdd("N");
 	                m_current++;
@@ -868,8 +868,8 @@ public class Metaphone3 {
 					Encode_T();
 	                break;
 
-	            case '�': // eth
-				case '�': // thorn
+	            case 'Ð': // eth
+				case 'Þ': // thorn
 
 	                MetaphAdd("0");
 	                m_current++;
@@ -890,13 +890,13 @@ public class Metaphone3 {
 					Encode_X();
 	                break;
 
-	            case '�':
+	            case 'Š':
 
 	                MetaphAdd("X");
 	                m_current++;
 	                break;
 
-				case '�':
+				case 'Ž':
 
 	                MetaphAdd("S");
 	                m_current++;
@@ -6137,7 +6137,7 @@ public class Metaphone3 {
 				return true;
 			}
 
-			// old '�cole fran�aise d'Extr�me-Orient' chinese pinyin where 'ts-' => 'X'
+			// old 'École française d'Extrême-Orient' chinese pinyin where 'ts-' => 'X'
 			if (((m_length == 3) && StringAt((m_current + 1), 2, "SO", "SA", "SU", ""))
 				||	((m_length == 4) && StringAt((m_current + 1), 3, "SAO", "SAI", ""))
 				||	((m_length == 5) && StringAt((m_current + 1), 4, "SING", "SANG", "")))