[Archive-access-cvs] SF.net SVN: archive-access:[3259] trunk/archive-access/projects/wayback/ wayback-core/src/main/java/org/archive/wayback/util/htmllex/ParseContext. java

[<bra...@us...>]

Revision: 3259
          http://archive-access.svn.sourceforge.net/archive-access/?rev=3259&view=rev
Author:   bradtofel
Date:     2010-09-28 21:22:39 +0000 (Tue, 28 Sep 2010)

Log Message:
-----------
Fixed URL canonicalization test case and underlying code

Modified Paths:
--------------
    trunk/archive-access/projects/wayback/wayback-core/src/main/java/org/archive/wayback/util/htmllex/ParseContext.java

Modified: trunk/archive-access/projects/wayback/wayback-core/src/main/java/org/archive/wayback/util/htmllex/ParseContext.java
===================================================================
--- trunk/archive-access/projects/wayback/wayback-core/src/main/java/org/archive/wayback/util/htmllex/ParseContext.java	2010-09-28 21:22:20 UTC (rev 3258)
+++ trunk/archive-access/projects/wayback/wayback-core/src/main/java/org/archive/wayback/util/htmllex/ParseContext.java	2010-09-28 21:22:39 UTC (rev 3259)
@@ -30,6 +30,7 @@
 import java.util.Map;
 
 import org.apache.commons.httpclient.URIException;
+import org.apache.commons.lang.StringEscapeUtils;
 import org.archive.net.UURI;
 import org.archive.net.UURIFactory;
 import org.htmlparser.util.Translate;
@@ -98,7 +99,10 @@
 	 * @throws URISyntaxException if the input URL is malformed
 	 */
 	public String resolve(String url) throws URISyntaxException {
-		url = Translate.decode(url);
+		// BUG in Translate.decode(): "foo?a=b&lang=en" acts as if it 
+		// was "⟨"
+//		url = Translate.decode(url);
+		url = StringEscapeUtils.unescapeHtml(url);
 		int hashIdx = url.indexOf('#');
 		String frag = "";
 		if(hashIdx != -1) {
@@ -111,7 +115,8 @@
 				System.err.println("No url to resolve!");
 				return url;
 			}
-			return baseUrl.resolve(url,false).toString() + frag;
+			return baseUrl.resolve(url,true).toString() + frag;
+//			return baseUrl.resolve(url,false).toString() + frag;
 //			return UURIFactory.getInstance(baseUrl, url).toString() + frag;
 		} catch (URIException e) {
 			e.printStackTrace();


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