[Archive-access-cvs] SF.net SVN: archive-access:[3442] trunk/archive-access/projects/wayback/ wayback-core/src

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

Revision: 3442
          http://archive-access.svn.sourceforge.net/archive-access/?rev=3442&view=rev
Author:   bradtofel
Date:     2011-05-25 01:08:38 +0000 (Wed, 25 May 2011)

Log Message:
-----------
BUGFIX: urlToPath was not handling ports correctly

Modified Paths:
--------------
    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/test/java/org/archive/wayback/util/url/UrlOperationsTest.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	2011-05-25 01:04:11 UTC (rev 3441)
+++ trunk/archive-access/projects/wayback/wayback-core/src/main/java/org/archive/wayback/util/url/UrlOperations.java	2011-05-25 01:08:38 UTC (rev 3442)
@@ -220,25 +220,28 @@
 	 * @return the path component of the URL, or "" if it contains no path.
 	 */
 	public static String getURLPath(String url) {
-		int portIdx = url.indexOf(UrlOperations.PORT_SEPARATOR);
+		url = stripURLScheme(url);
 		int pathIdx = url.indexOf(UrlOperations.PATH_START);
-		if(portIdx == -1 && pathIdx == -1) {
-			return "";
-		}
-		if(portIdx == -1) {
-			return url.substring(pathIdx);
-		}
 		if(pathIdx == -1) {
-			return url.substring(portIdx);
+			return "/";
 		}
-		if(pathIdx > portIdx) {
-			return url.substring(portIdx);
-		} else {
-			return url.substring(pathIdx);
-		}
+		return url.substring(pathIdx);
 	}
-	
 	/**
+	 * Attempt to extract the path component of a url String argument.
+	 * @param url the URL which may contain a path, sans scheme.
+	 * @return the path component of the URL, or "" if it contains no path.
+	 */
+	public static String stripURLScheme(String url) {
+		String lcUrl = url.toLowerCase();
+		for(String scheme : ALL_SCHEMES) {
+			if(lcUrl.startsWith(scheme)) {
+				return url.substring(scheme.length());
+			}
+		}
+		return url;
+	}	
+	/**
 	 * Attempt to strip default ports out of URL strings.
 	 * @param url the original URL possibly including a port
 	 * @return the URL sans port, if the scheme was recognized and the default
@@ -279,6 +282,11 @@
 		return sb.toString();
 	}
 
+	/**
+	 * @param orig String containing a URL, possibly beginning with "http:/".
+	 * @return original string if orig begins with "http://", or a new String
+	 * with the extra slash, if orig only had one slash.
+	 */
 	public static String fixupHTTPUrlWithOneSlash(String orig) {
 		if(orig.startsWith("http:/") && ! orig.startsWith(HTTP_SCHEME)) {
 			// very likely the IE "you must have meant 1 slash, not 2 bug:

Modified: trunk/archive-access/projects/wayback/wayback-core/src/test/java/org/archive/wayback/util/url/UrlOperationsTest.java
===================================================================
--- trunk/archive-access/projects/wayback/wayback-core/src/test/java/org/archive/wayback/util/url/UrlOperationsTest.java	2011-05-25 01:04:11 UTC (rev 3441)
+++ trunk/archive-access/projects/wayback/wayback-core/src/test/java/org/archive/wayback/util/url/UrlOperationsTest.java	2011-05-25 01:08:38 UTC (rev 3442)
@@ -205,7 +205,29 @@
 		
 		
 	}
+	public void testUrlPath() {
+		assertEquals("/",UrlOperations.getURLPath("http://foo.com"));
+		assertEquals("/",UrlOperations.getURLPath("http://foo.com/"));
+		assertEquals("/",UrlOperations.getURLPath("http://foo.com:80/"));
+		assertEquals("/blue",UrlOperations.getURLPath("http://foo.com:80/blue"));
+		assertEquals("/blue/red",UrlOperations.getURLPath("http://foo.com:80/blue/red"));
+		assertEquals("/blue/red:colon",UrlOperations.getURLPath("http://foo.com:80/blue/red:colon"));
 
+		assertEquals("/",UrlOperations.getURLPath("foo.com"));
+		assertEquals("/",UrlOperations.getURLPath("foo.com:80"));
+		assertEquals("/",UrlOperations.getURLPath("foo.com:8080"));
+		assertEquals("/",UrlOperations.getURLPath("foo.com/"));
+		assertEquals("/",UrlOperations.getURLPath("foo.com:80/"));
+		assertEquals("/",UrlOperations.getURLPath("foo.com:8080/"));
+		assertEquals("/bar",UrlOperations.getURLPath("foo.com/bar"));
+		assertEquals("/bar",UrlOperations.getURLPath("foo.com:80/bar"));
+		assertEquals("/bar",UrlOperations.getURLPath("foo.com:8080/bar"));
+
+		assertEquals("/bar/baz",UrlOperations.getURLPath("foo.com/bar/baz"));
+		assertEquals("/bar/baz",UrlOperations.getURLPath("foo.com:80/bar/baz"));
+		assertEquals("/bar/baz",UrlOperations.getURLPath("foo.com:8080/bar/baz"));
+
+	}
 	public void testStripDefaultPort() {
 		assertSDP("http://foo.com/","http://foo.com/");
 		assertSDP("http://foo.com","http://foo.com");


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