[Archive-access-cvs] SF.net SVN: archive-access:[3622] trunk/archive-access/projects/ archive-commons/src/main/java/org/archive/url/GoogleURLCanonicalizer.java

[<nl...@us...>]

Revision: 3622
          http://archive-access.svn.sourceforge.net/archive-access/?rev=3622&view=rev
Author:   nlevitt
Date:     2012-02-28 22:58:11 +0000 (Tue, 28 Feb 2012)
Log Message:
-----------
BUGFIX: avoid NPE on urls with null host, e.g. dns:archive.org

Modified Paths:
--------------
    trunk/archive-access/projects/archive-commons/src/main/java/org/archive/url/GoogleURLCanonicalizer.java

Modified: trunk/archive-access/projects/archive-commons/src/main/java/org/archive/url/GoogleURLCanonicalizer.java
===================================================================
--- trunk/archive-access/projects/archive-commons/src/main/java/org/archive/url/GoogleURLCanonicalizer.java	2012-02-28 04:09:45 UTC (rev 3621)
+++ trunk/archive-access/projects/archive-commons/src/main/java/org/archive/url/GoogleURLCanonicalizer.java	2012-02-28 22:58:11 UTC (rev 3622)
@@ -23,19 +23,22 @@
 		url.setQuery(minimalEscape(url.getQuery()));
 		String hostE = unescapeRepeatedly(url.getHost());
 		String host = null;
-		try {
-			host = IDN.toASCII(hostE);
-		} catch(IllegalArgumentException e) {
-			if(!e.getMessage().contains("A prohibited code point was found")) {
-				// TODO: What to do???
-//				throw e;
+		if (hostE != null) {
+			try {
+				host = IDN.toASCII(hostE);
+			} catch(IllegalArgumentException e) {
+				if(!e.getMessage().contains("A prohibited code point was found")) {
+					// TODO: What to do???
+					//				throw e;
+				}
+				host = hostE;
+
 			}
-			host = hostE;
-			
-		}
-		host = host.replaceAll("^\\.+", "").
+			host = host.replaceAll("^\\.+", "").
 					replaceAll("\\.\\.+", ".").
 					replaceAll("\\.$", "");
+		}
+		
 		String ip = null;
 //			try {
 				ip = attemptIPFormats(host);
@@ -44,7 +47,7 @@
 //			}
 		if(ip != null) {
 			host = ip;
-		} else {
+		} else if (host != null) {
 			host = escapeOnce(host.toLowerCase());
 		}
 		url.setHost(host);

This was sent by the SourceForge.net collaborative development platform, the world's largest Open Source development site.