apply blacklist on rwis during dht receive

very experimental!

git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@1865 6c8d7289-2bf4-0310-a012-ef5d649a1542

htroot/yacy/transferRWI.java

@@ -126,19 +126,18 @@ public final class transferRWI {
                     wordhashes[received] = wordHash;
                     entry = new plasmaWordIndexEntry(estring.substring(p));
                     sb.wordIndex.addEntry(wordHash, entry, System.currentTimeMillis(), true);
-                    //sb.wordIndex.addEntries(plasmaWordIndexEntryContainer.instantContainer(wordHash, System.currentTimeMillis(), entry), System.currentTimeMillis(), true);
                     serverCore.checkInterruption();
                     
                     urlHash = entry.getUrlHash();
                     try {
-                        if (
-                                (!(unknownURL.contains(urlHash))) &&
-                                (!(sb.urlPool.loadedURL.exists(urlHash)))
-                        ) {
+                        if ((!(unknownURL.contains(urlHash))) &&
+                            (!(sb.urlPool.loadedURL.exists(urlHash)))) {
                             unknownURL.add(urlHash);
                         }
                     } catch (Exception ex) {
-                        sb.getLog().logWarning("transferRWI: DB-Error while trying to determine if URL with hash '" + urlHash + "' is known.",ex);
+                        sb.getLog().logWarning(
+                          "transferRWI: DB-Error while trying to determine if URL with hash '" +
+                          urlHash + "' is known.", ex);
                         unknownURL.add(urlHash);
                     }
                     received++;

htroot/yacy/transferURL.java

@@ -95,15 +95,16 @@ public final class transferURL {
                 } else {
                     lEntry = sb.urlPool.loadedURL.newEntry(urls, true);
                     if ((lEntry != null) && (lEntry.url() != null)) {
-                        if (
-                                (blockBlacklist) &&
-                                (plasmaSwitchboard.urlBlacklist.isListed(lEntry.url().getHost().toLowerCase(), lEntry.url().getPath()))
-                        ){
-                            yacyCore.log.logFine("transferURL: blocked blacklisted URL '" + lEntry.url() + "' from peer " + otherPeerName);
+                        if ((blockBlacklist) &&
+                            (plasmaSwitchboard.urlBlacklist.isListed( lEntry.url().getHost().toLowerCase(), lEntry.url().getPath()))) {
+                            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;
                         } else {
                             sb.urlPool.loadedURL.addEntry(lEntry, iam, iam, 3);
-                            yacyCore.log.logFine("transferURL: received URL '" + lEntry.url() + "' from peer " + otherPeerName);
+                            yacyCore.log.logFine("transferURL: received URL '"
+                                    + lEntry.url() + "' from peer "
+                                    + otherPeerName);
                             received++;
                         }
                     } else {

source/de/anomic/plasma/plasmaWordIndex.java

@@ -358,6 +358,14 @@ public final class plasmaWordIndex {
         return removed;
     }
     
+    public synchronized int tryRemoveURLs(String urlHash) {
+        // this tries to delete an index from the cache that has this
+        // urlHash assigned. This can only work if the entry is really fresh
+        // and can be found in the RAM cache
+        // this returns the number of deletion that had been possible
+        return ramCache.tryRemoveURLs(urlHash);
+    }
+    
     public static final int RL_RAMCACHE    = 0;
     public static final int RL_FILECACHE   = 1;
     public static final int RL_ASSORTMENTS = 2;

source/de/anomic/plasma/plasmaWordIndexCache.java

@@ -61,7 +61,8 @@ public final class plasmaWordIndexCache implements plasmaWordIndexInterface {
     // environment constants
     private static final String indexArrayFileName = "indexDump1.array";
     public static final int  ramCacheReferenceLimit = 50;
-    public static final long ramCacheAgeLimit       = 60 * 60 * 2 * 1000; // milliseconds; 2 Hours
+    public static final long ramCacheMaxAge         = 1000 * 60 * 60 * 2; // milliseconds; 2 Hours
+    public static final long ramCacheMinAge         = 1000 * 60 * 2; // milliseconds; 2 Minutes (Karenz for DHT Receive)
     
     // class variables
     private final File databaseRoot;
@@ -257,12 +258,13 @@ public final class plasmaWordIndexCache implements plasmaWordIndexInterface {
                 String hash = null;
                 int count = hashScore.getMaxScore();
                 if ((count > ramCacheReferenceLimit) &&
-                    ((hash = (String) hashScore.getMaxObject()) != null)) {
-                    // flush high-score entries
+                    ((hash = (String) hashScore.getMaxObject()) != null) &&
+                    (System.currentTimeMillis() - longEmit(hashDate.getScore(hash)) > ramCacheMinAge)) {
+                    // flush high-score entries, but not if they are too 'young'
                     return hash;
                 }
                 long oldestTime = longEmit(hashDate.getMinScore());
-                if (((System.currentTimeMillis() - oldestTime) > ramCacheAgeLimit) &&
+                if (((System.currentTimeMillis() - oldestTime) > ramCacheMaxAge) &&
                     ((hash = (String) hashDate.getMinObject()) != null)) {
                     // flush out-dated entries
                     return hash;
@@ -271,6 +273,10 @@ public final class plasmaWordIndexCache implements plasmaWordIndexInterface {
                 if (Runtime.getRuntime().freeMemory() < 10000000) {
                     // low-memory case
                     hash = (String) hashScore.getMaxObject(); // flush high-score entries (saves RAM)
+                    if (System.currentTimeMillis() - longEmit(hashDate.getScore(hash)) < ramCacheMinAge) {
+                        // to young, take it from the oldest entries
+                        hash = (String) hashDate.getMinObject();
+                    }
                 } else {
                     // not-efficient-so-far case
                     hash = (String) hashDate.getMinObject(); // flush oldest entries (makes indexing faster)
@@ -335,6 +341,30 @@ public final class plasmaWordIndexCache implements plasmaWordIndexInterface {
         return count;
     }
 
+    public synchronized int tryRemoveURLs(String urlHash) {
+        // this tries to delete an index from the cache that has this
+        // urlHash assigned. This can only work if the entry is really fresh
+        // Such entries must be searched in the latest entries
+        Iterator i = hashDate.scores(false);
+        String wordHash;
+        long t;
+        plasmaWordIndexEntryContainer c;
+        int delCount = 0;
+        while (i.hasNext()) {
+            wordHash = (String) i.next();
+            // check time
+            t = longEmit(hashDate.getScore(wordHash));
+            if (System.currentTimeMillis() - t > ramCacheMinAge) return delCount;
+            // get container
+            c = (plasmaWordIndexEntryContainer) cache.get(wordHash);
+            if (c.remove(urlHash) != null) {
+                cache.put(wordHash, c);
+                delCount++;
+            }
+        }
+        return delCount;
+    }
+    
     public int addEntries(plasmaWordIndexEntryContainer container, long updateTime, boolean highPriority) {
         // this puts the entries into the cache, not into the assortment directly