[Archive-access-cvs] SF.net SVN: archive-access:[2988] trunk/archive-access/projects/wayback/ wayback-core/src/main/java/org/archive/wayback/util/url/UrlOperations.java

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

Revision: 2988
          http://archive-access.svn.sourceforge.net/archive-access/?rev=2988&view=rev
Author:   bradtofel
Date:     2010-03-20 01:05:39 +0000 (Sat, 20 Mar 2010)

Log Message:
-----------
BUGFIX(unreported): replaced URL to host processing with REGEX, to better handle URLs with freakish illegal characters before the port/path start.

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

Modified: trunk/archive-access/projects/wayback/wayback-core/src/main/java/org/archive/wayback/util/url/UrlOperations.java
===================================================================
--- trunk/archive-access/projects/wayback/wayback-core/src/main/java/org/archive/wayback/util/url/UrlOperations.java	2010-03-20 01:02:49 UTC (rev 2987)
+++ trunk/archive-access/projects/wayback/wayback-core/src/main/java/org/archive/wayback/util/url/UrlOperations.java	2010-03-20 01:05:39 UTC (rev 2988)
@@ -92,6 +92,9 @@
         Pattern.compile("(([0-9a-z_.-]+)\\.(" + ALL_TLD_PATTERN + "))|" +
         		"(" + IP_PATTERN + ")");
 
+    private static final Pattern AUTHORITY_REGEX_SIMPLE =
+        Pattern.compile("([0-9a-z_.-]++)");
+
     /**
 	 * @param urlPart
 	 * @return boolean indicating whether urlPart might be an Authority.
@@ -186,22 +189,11 @@
 		for(String scheme : ALL_SCHEMES) {
 			if(url.startsWith(scheme)) {
 				int hostIdx = scheme.length();
-				int portIdx = url.indexOf(PORT_SEPARATOR, hostIdx + 1);
-				int pathIdx = url.indexOf(PATH_START, hostIdx + 1);
-				if(portIdx == -1 && pathIdx == -1) {
-					return url.substring(hostIdx);
+
+				Matcher m = AUTHORITY_REGEX_SIMPLE.matcher(url.substring(hostIdx));
+				if(m.find()) {
+					return m.group(0);
 				}
-				if(portIdx == -1) {
-					return url.substring(hostIdx,pathIdx);
-				}
-				if(pathIdx == -1) {
-					return url.substring(hostIdx,portIdx);
-				}
-				if(pathIdx > portIdx) {
-					return url.substring(hostIdx,portIdx);
-				} else {
-					return url.substring(hostIdx,pathIdx);
-				}
 			}
 		}
 		return url;


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