other bugfix

git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@3048 6c8d7289-2bf4-0310-a012-ef5d649a1542

source/de/anomic/plasma/plasmaCrawlLURL.java

@@ -388,8 +388,9 @@ public final class plasmaCrawlLURL {
         }
 
         public Object next() throws RuntimeException {
-            kelondroRow.Entry e = (kelondroRow.Entry) i.next();
-            if (e == null) return null;
+            kelondroRow.Entry e = null;
+            if (i.hasNext()) { e = (kelondroRow.Entry) i.next(); }
+            if (e == null) { return null; }
             return new indexURLEntryNew(e, null);
         }
 
@@ -499,7 +500,7 @@ public final class plasmaCrawlLURL {
         public void run() {
             try {
                 serverLog.logInfo("URLDBCLEANER", "UrldbCleaner-Thread startet");
-                Iterator eiter = entries(true, false, null);
+                final Iterator eiter = entries(true, false, null);
                 while (eiter.hasNext() && run) {
                     synchronized (this) {
                         if (this.pause) {
@@ -512,19 +513,17 @@ public final class plasmaCrawlLURL {
                             }
                         }
                     }
-                    if (eiter.hasNext()) {
-                        indexURLEntry entry = (indexURLEntry) eiter.next();
+                    final indexURLEntry entry = (indexURLEntry) eiter.next();
                     if (entry == null) {
                         serverLog.logFine("URLDBCLEANER", "entry == null");
+                    } else if (entry.hash() == null) {
+                        serverLog.logFine("URLDBCLEANER", ++blacklistedUrls + " blacklisted (" + ((double) blacklistedUrls / totalSearchedUrls) * 100 + "%): " + "hash == null");
                     } else {
-                            indexURLEntry.Components comp = entry.comp();
+                        final indexURLEntry.Components comp = entry.comp();
                         totalSearchedUrls++;
-                            if (entry.hash() == null) {
-                                serverLog.logFine("URLDBCLEANER", ++blacklistedUrls + " blacklisted (" + ((double)blacklistedUrls/totalSearchedUrls)*100 + "%): " + " hash == null");
-                            } else if (comp.url() == null) {
+                        if (comp.url() == null) {
                             serverLog.logFine("URLDBCLEANER", ++blacklistedUrls + " blacklisted (" + ((double) blacklistedUrls / totalSearchedUrls) * 100 + "%): " + entry.hash() + "URL == null");
                             remove(entry.hash());
-                                lastHash = entry.hash();
                         } else if (plasmaSwitchboard.urlBlacklist.isListed(plasmaURLPattern.BLACKLIST_CRAWLER, comp.url()) ||
                                    plasmaSwitchboard.urlBlacklist.isListed(plasmaURLPattern.BLACKLIST_DHT, comp.url())) {
                             lastBlacklistedUrl = comp.url().toNormalform();
@@ -535,11 +534,8 @@ public final class plasmaCrawlLURL {
                                 serverLog.logInfo("URLDBCLEANER", "Deleted " + blacklistedUrls + " URLs until now. Last deleted URL-Hash: " + lastBlacklistedUrl);
                             }
                             lastUrl = comp.url().toNormalform();
-                                lastHash = entry.hash();
-                            }
                         }
-                    } else {
-                        serverLog.logFine("URLDBCLEANER", "Iterator == null");
+                        lastHash = entry.hash();
                     }
                 }
             } catch (RuntimeException e) {