[Archive-access-cvs] SF.net SVN: archive-access:[2774] branches/wayback-1_4_2/wayback-core/src/ test/java/org/archive/wayback/util/url/AggressiveUrlCanonicalizerTest.java

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

Revision: 2774
          http://archive-access.svn.sourceforge.net/archive-access/?rev=2774&view=rev
Author:   bradtofel
Date:     2009-07-18 00:24:00 +0000 (Sat, 18 Jul 2009)

Log Message:
-----------
TWEAK: added tests for newly supported schemes.

Modified Paths:
--------------
    branches/wayback-1_4_2/wayback-core/src/test/java/org/archive/wayback/util/url/AggressiveUrlCanonicalizerTest.java

Modified: branches/wayback-1_4_2/wayback-core/src/test/java/org/archive/wayback/util/url/AggressiveUrlCanonicalizerTest.java
===================================================================
--- branches/wayback-1_4_2/wayback-core/src/test/java/org/archive/wayback/util/url/AggressiveUrlCanonicalizerTest.java	2009-07-18 00:22:52 UTC (rev 2773)
+++ branches/wayback-1_4_2/wayback-core/src/test/java/org/archive/wayback/util/url/AggressiveUrlCanonicalizerTest.java	2009-07-18 00:24:00 UTC (rev 2774)
@@ -45,16 +45,15 @@
 		// simple strip of http://
 		checkCanonicalization("http://foo.com/","foo.com/");
 
-// would be nice to handle other protocols...
-//		// simple strip of https://
-//		checkCanonicalization("https://foo.com/","foo.com/");
-//
-//		// simple strip of ftp://
-//		checkCanonicalization("ftp://foo.com/","foo.com/");
-//
-//		// simple strip of rtsp://
-//		checkCanonicalization("rtsp://foo.com/","foo.com/");
+		// simple strip of https://
+		checkCanonicalization("https://foo.com/","foo.com/");
 
+		// simple strip of ftp://
+		checkCanonicalization("ftp://foo.com/","foo.com/");
+
+		// simple strip of rtsp://
+		checkCanonicalization("rtsp://foo.com/","foo.com/");
+
 		// strip leading 'www.'
 		checkCanonicalization("http://www.foo.com/","foo.com/");
 		
@@ -63,6 +62,9 @@
 		
 		// strip leading 'www##.'
 		checkCanonicalization("http://www12.foo.com/","foo.com/");
+
+		// strip leading 'www##.' with https
+		checkCanonicalization("https://www12.foo.com/","foo.com/");
 		
 		// strip leading 'www##.' with no protocol
 		checkCanonicalization("www12.foo.com/","foo.com/");
@@ -174,13 +176,53 @@
 		checkCanonicalization(
 				"http://legislature.mi.gov/(a(4hqa0555fwsecu455xqckv45)S(4hqa0555fwsecu455xqckv45)f(4hqa0555fwsecu455xqckv45))/mileg.aspx?page=sessionschedules",
 				"legislature.mi.gov/mileg.aspx?page=sessionschedules");
+
+
+
+
+		// default port stripping:
 		
+		// FIRST the easy-on-the-eyes
+
 		// strip port 80
 		checkCanonicalization("http://www.chub.org:80/foo","chub.org/foo");
 
 		// but not other ports...
 		checkCanonicalization("http://www.chub.org:8080/foo","chub.org:8080/foo");
+		
+		// but not other ports... with "www#." massage
+		checkCanonicalization("http://www232.chub.org:8080/foo","chub.org:8080/foo");
 
+		// default HTTP (:80) stripping without a scheme:
+		checkCanonicalization("www.chub.org:80/foo","chub.org/foo");
+		
+		// no strip https port (443) without scheme:
+		checkCanonicalization("www.chub.org:443/foo","chub.org:443/foo");
+
+		// yes strip https port (443) with scheme:
+		checkCanonicalization("https://www.chub.org:443/foo","chub.org/foo");
+		
+		// NEXT the exhaustive:
+		String origHost = "www.chub.org";
+		String massagedHost = "chub.org";
+		String path = "/foo";
+		for(String scheme : UrlOperations.ALL_SCHEMES) {
+
+			int defaultPort = UrlOperations.schemeToDefaultPort(scheme);
+			int nonDefaultPort = 19991;
+
+			String origDefault = scheme + origHost + ":" + defaultPort + path;
+			String canonDefault = massagedHost + path;
+
+			String origNonDefault = 
+				scheme + origHost + ":" + nonDefaultPort + path;
+			String canonNonDefault =
+				massagedHost + ":" + nonDefaultPort + path;
+
+			checkCanonicalization(origDefault,canonDefault);
+			checkCanonicalization(origNonDefault,canonNonDefault);
+		}
+
 	}
 	
 	private void checkCanonicalization(String orig, String want) {


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