[Archive-access-cvs] SF.net SVN: archive-access: [2357] trunk/archive-access/projects/wayback/ wayback-core/src/main/java/org/archive/wayback/accesscontrol/robotstxt

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

Revision: 2357
          http://archive-access.svn.sourceforge.net/archive-access/?rev=2357&view=rev
Author:   bradtofel
Date:     2008-07-01 16:25:55 -0700 (Tue, 01 Jul 2008)

Log Message:
-----------
REFACTOR: SearchResult => (Url|Capture)SearchResult

Modified Paths:
--------------
    trunk/archive-access/projects/wayback/wayback-core/src/main/java/org/archive/wayback/accesscontrol/robotstxt/RobotExclusionFilter.java
    trunk/archive-access/projects/wayback/wayback-core/src/main/java/org/archive/wayback/accesscontrol/robotstxt/RobotExclusionFilterFactory.java

Modified: trunk/archive-access/projects/wayback/wayback-core/src/main/java/org/archive/wayback/accesscontrol/robotstxt/RobotExclusionFilter.java
===================================================================
--- trunk/archive-access/projects/wayback/wayback-core/src/main/java/org/archive/wayback/accesscontrol/robotstxt/RobotExclusionFilter.java	2008-07-01 23:25:27 UTC (rev 2356)
+++ trunk/archive-access/projects/wayback/wayback-core/src/main/java/org/archive/wayback/accesscontrol/robotstxt/RobotExclusionFilter.java	2008-07-01 23:25:55 UTC (rev 2357)
@@ -35,15 +35,14 @@
 import java.util.regex.Pattern;
 
 import org.archive.util.ArchiveUtils;
-import org.archive.wayback.WaybackConstants;
 import org.archive.wayback.core.Resource;
-import org.archive.wayback.core.SearchResult;
+import org.archive.wayback.core.CaptureSearchResult;
 import org.archive.wayback.exception.LiveDocumentNotAvailableException;
 import org.archive.wayback.liveweb.LiveWebCache;
 import org.archive.wayback.util.ObjectFilter;
 
 /**
- * SearchResultFilter that uses a LiveWebCache to retrieve robots.txt documents
+ * CaptureSearchResult Filter that uses a LiveWebCache to retrieve robots.txt documents
  * from the live web, and filters SearchResults based on the rules therein.
  * 
  * This class caches parsed RobotRules that are retrieved, so using the same 
@@ -56,7 +55,7 @@
  * @author brad
  * @version $Date$, $Revision$
  */
-public class RobotExclusionFilter implements ObjectFilter<SearchResult> {
+public class RobotExclusionFilter implements ObjectFilter<CaptureSearchResult> {
 
 	private final static String HTTP_PREFIX = "http://";
 	private final static String ROBOT_SUFFIX = "/robots.txt";
@@ -127,10 +126,10 @@
 		return list;
 	}
 	
-	private RobotRules getRules(SearchResult result) {
+	private RobotRules getRules(CaptureSearchResult result) {
 		RobotRules rules = null;
 		RobotRules tmpRules = null;
-		String host = result.get(WaybackConstants.RESULT_ORIG_HOST);
+		String host = result.getOriginalHost();
 		List<String> urlStrings = searchResultToRobotUrlStrings(host);
 		Iterator<String> itr = urlStrings.iterator();
 		String firstUrlString = null;
@@ -174,12 +173,12 @@
 	/* (non-Javadoc)
 	 * @see org.archive.wayback.resourceindex.SearchResultFilter#filterSearchResult(org.archive.wayback.core.SearchResult)
 	 */
-	public int filterObject(SearchResult r) {
+	public int filterObject(CaptureSearchResult r) {
 
 		int filterResult = ObjectFilter.FILTER_EXCLUDE; 
 		RobotRules rules = getRules(r);
 		if(rules != null) {
-			String resultURL = r.get(WaybackConstants.RESULT_URL);
+			String resultURL = r.getOriginalUrl();
 			URL url;
 			try {
 				url = new URL(ArchiveUtils.addImpliedHttpIfNecessary(resultURL));

Modified: trunk/archive-access/projects/wayback/wayback-core/src/main/java/org/archive/wayback/accesscontrol/robotstxt/RobotExclusionFilterFactory.java
===================================================================
--- trunk/archive-access/projects/wayback/wayback-core/src/main/java/org/archive/wayback/accesscontrol/robotstxt/RobotExclusionFilterFactory.java	2008-07-01 23:25:27 UTC (rev 2356)
+++ trunk/archive-access/projects/wayback/wayback-core/src/main/java/org/archive/wayback/accesscontrol/robotstxt/RobotExclusionFilterFactory.java	2008-07-01 23:25:55 UTC (rev 2357)
@@ -25,7 +25,7 @@
 package org.archive.wayback.accesscontrol.robotstxt;
 
 import org.archive.wayback.accesscontrol.ExclusionFilterFactory;
-import org.archive.wayback.core.SearchResult;
+import org.archive.wayback.core.CaptureSearchResult;
 import org.archive.wayback.liveweb.LiveWebCache;
 import org.archive.wayback.util.ObjectFilter;
 
@@ -44,7 +44,7 @@
 	/* (non-Javadoc)
 	 * @see org.archive.wayback.resourceindex.ExclusionFilterFactory#get()
 	 */
-	public ObjectFilter<SearchResult> get() {
+	public ObjectFilter<CaptureSearchResult> get() {
 		return new RobotExclusionFilter(webCache,userAgent,maxCacheMS);
 	}
 


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