From 8ba8e2b7d9c24e6de9df7c64a00452fa98958516 Mon Sep 17 00:00:00 2001
From: hydrox <hydrox@6c8d7289-2bf4-0310-a012-ef5d649a1542>
Date: Wed, 28 Jun 2006 08:51:34 +0000
Subject: [PATCH] *) added cache for blacklists urlhashs recieved by DHT. DHT
 does not request URLs listed in this cache.

git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@2251 6c8d7289-2bf4-0310-a012-ef5d649a1542
---
 htroot/yacy/transferRWI.java                  | 10 ++++++++-
 htroot/yacy/transferURL.java                  |  2 +-
 source/de/anomic/plasma/plasmaURLPattern.java | 21 +++++++++++++++++++
 3 files changed, 31 insertions(+), 2 deletions(-)

diff --git a/htroot/yacy/transferRWI.java b/htroot/yacy/transferRWI.java
index 52cbe032f..012c9724e 100644
--- a/htroot/yacy/transferRWI.java
+++ b/htroot/yacy/transferRWI.java
@@ -78,6 +78,7 @@ public final class transferRWI {
         final int entryc      = Integer.parseInt(post.get("entryc", "")); // number of entries in indexes
         byte[] indexes        = post.get("indexes", "").getBytes();       // the indexes, as list of word entries
         boolean granted       = sb.getConfig("allowReceiveIndex", "false").equals("true");
+        boolean blockBlacklist = sb.getConfig("indexReceiveBlockBlacklist", "false").equals("true");
         boolean checkLimit    = sb.getConfigBool("indexDistribution.dhtReceiptLimitEnabled", true);
         final long cachelimit = sb.getConfigLong("indexDistribution.dhtReceiptLimit", 1000);
         final yacySeed otherPeer = yacyCore.seedDB.get(iam);
@@ -152,7 +153,14 @@ public final class transferRWI {
                     try {
                         if ((!(unknownURL.contains(urlHash))) &&
                             (!(sb.urlPool.loadedURL.exists(urlHash)))) {
-                            unknownURL.add(urlHash);
+                            if ((blockBlacklist) && (plasmaSwitchboard.urlBlacklist.hashInBlacklistedCache(urlHash))) {
+                                int deleted = sb.wordIndex.tryRemoveURLs(urlHash);
+                                yacyCore.log.logFine("transferRWI: blocked blacklisted URLHash '" + urlHash + "' from peer " + otherPeerName + "; deleted " + deleted + " URL entries from RWIs");
+                                //TODO: set to logFine if it works.
+                            }
+                            else {
+                                unknownURL.add(urlHash);                                
+                            }
                         }
                     } catch (Exception ex) {
                         sb.getLog().logWarning(
diff --git a/htroot/yacy/transferURL.java b/htroot/yacy/transferURL.java
index b27261fc9..6ab05512a 100644
--- a/htroot/yacy/transferURL.java
+++ b/htroot/yacy/transferURL.java
@@ -96,7 +96,7 @@ public final class transferURL {
                     lEntry = sb.urlPool.loadedURL.newEntry(urls, true);
                     if ((lEntry != null) && (lEntry.url() != null)) {
                         if ((blockBlacklist) &&
-                            (plasmaSwitchboard.urlBlacklist.isListed(lEntry.url()))) {
+                            (plasmaSwitchboard.urlBlacklist.isListed(lEntry.hash(), lEntry.url()))) {
                             int deleted = sb.wordIndex.tryRemoveURLs(lEntry.hash());
                             yacyCore.log.logFine("transferURL: blocked blacklisted URL '" + lEntry.url() + "' from peer " + otherPeerName + "; deleted " + deleted + " URL entries from RWIs");
                             lEntry = null;
diff --git a/source/de/anomic/plasma/plasmaURLPattern.java b/source/de/anomic/plasma/plasmaURLPattern.java
index 515db7340..644081c71 100644
--- a/source/de/anomic/plasma/plasmaURLPattern.java
+++ b/source/de/anomic/plasma/plasmaURLPattern.java
@@ -43,11 +43,16 @@ package de.anomic.plasma;
 
 import java.io.File;
 import java.net.URL;
+import java.util.Collections;
 import java.util.HashMap;
+import java.util.HashSet;
+import java.util.Set;
+
 import de.anomic.kelondro.kelondroMSetTools;
 
 public class plasmaURLPattern {
 
+    private Set cachedUrlHashs = Collections.synchronizedSet(new HashSet());
     private File rootPath = null;
     private HashMap hostpaths = null; // key=host, value=path; mapped url is http://host/path; path does not start with '/' here
 
@@ -85,6 +90,22 @@ public class plasmaURLPattern {
         hostpaths.put(host.toLowerCase(), path);
     }
 
+    public boolean hashInBlacklistedCache(String urlHash) {
+        return cachedUrlHashs.contains(urlHash);
+    }
+
+    public boolean isListed(String urlHash, URL url) {
+        if (!cachedUrlHashs.contains(urlHash)) {
+            boolean temp = isListed(url.getHost().toLowerCase(), url.getFile());
+            if (temp) 
+                {
+                    cachedUrlHashs.add(urlHash);
+                }
+            return temp;   
+        }        
+        return true;  
+    }
+    
     public boolean isListed(URL url) {
         return isListed(url.getHost().toLowerCase(), url.getFile());
     }