*) Trying to be more tolerant against wrong charset names

git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@2760 6c8d7289-2bf4-0310-a012-ef5d649a1542
yacy · Oct 13, 2006 · decb09d · decb09d
1 parent e9afe39
commit decb09d
Showing 1 changed file with 12 additions and 0 deletions.
diff --git a/source/de/anomic/plasma/plasmaParser.java b/source/de/anomic/plasma/plasmaParser.java
@@ -352,6 +352,18 @@ public static String getRealCharsetEncoding(String encoding) {
     		else if ((c >= '0') && (c <= '9')) encoding = "windows-" + encoding.substring(7);
     	}
 
+    	if (encoding.toLowerCase().startsWith("iso") && encoding.length() > 3) {
+    		char c = encoding.charAt(3);
+    		if (c == '_') encoding = "ISO-" + encoding.substring(4);
+    		else if ((c >= '0') && (c <= '9')) encoding = "ISO-" + encoding.substring(3);    		
+    	}
+
+    	if (encoding.toLowerCase().startsWith("iso") && encoding.length() > 8) {
+    		char c = encoding.charAt(8);
+    		if (c == '_') encoding = encoding.substring(0,8) + "-" + encoding.substring(9);
+    		else if ((c >= '0') && (c <= '9')) encoding = encoding.substring(0,8) + "-" + encoding.substring(8);    		
+    	}    	
+
     	return encoding;
     }