fix for bugs in IODispatcher

git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@5921 6c8d7289-2bf4-0310-a012-ef5d649a1542

source/de/anomic/kelondro/index/RowSet.java

@@ -33,6 +33,7 @@ import java.util.Random;
 import de.anomic.kelondro.order.Base64Order;
 import de.anomic.kelondro.order.CloneableIterator;
 import de.anomic.kelondro.order.NaturalOrder;
+import de.anomic.kelondro.util.Log;
 
 public class RowSet extends RowCollection implements ObjectIndex, Iterable<Row.Entry> {
 
@@ -410,8 +411,20 @@ public class RowSet extends RowCollection implements ObjectIndex, Iterable<Row.E
     protected static RowSet mergeEnum(RowCollection c0, RowCollection c1) {
         assert c0.rowdef == c1.rowdef : c0.rowdef.toString() + " != " + c1.rowdef.toString();
         RowSet r = new RowSet(c0.rowdef, c0.size() + c1.size());
-        c0.sort();
-        c1.sort();
+        try {
+        	c0.sort();
+        } catch (Exception e) {
+        	Log.logSevere("RowSet", "collection corrupted. cleaned. " + e.getMessage());
+        	e.printStackTrace();
+        	c0.clear();
+        }
+        try {
+        	c1.sort();
+        } catch (Exception e) {
+        	Log.logSevere("RowSet", "collection corrupted. cleaned. " + e.getMessage());
+        	e.printStackTrace();
+        	c1.clear();
+        }
         int c0i = 0, c1i = 0;
         int c0p, c1p;
         int o;

source/de/anomic/kelondro/text/IODispatcher.java

@@ -134,24 +134,36 @@ public class IODispatcher <ReferenceType extends Reference> extends Thread {
                 
                 // prefer dump actions to flush memory to disc
                 if (dumpQueue.size() > 0) {
+                	File f = null;
                     try {
                         dumpJob = dumpQueue.take();
+                        f = dumpJob.file;
                         dumpJob.dump();
                     } catch (InterruptedException e) {
-                        e.printStackTrace();
                         Log.logSevere("IODispatcher", "main run job was interrupted (1)", e);
+                        e.printStackTrace();
+                    } catch (Exception e) {
+                        Log.logSevere("IODispatcher", "main run job had errors (1), dump to " + f + " failed.", e);
+                    	e.printStackTrace();
                     }
                     continue loop;
                 }
                 
                 // otherwise do a merge operation
                 if (mergeQueue.size() > 0) {
+                	File f = null, f1 = null, f2 = null;
                     try {
                         mergeJob = mergeQueue.take();
+                        f = mergeJob.newFile;
+                        f1 = mergeJob.f1;
+                        f2 = mergeJob.f2;
                         mergeJob.merge();
                     } catch (InterruptedException e) {
-                        e.printStackTrace();
                         Log.logSevere("IODispatcher", "main run job was interrupted (2)", e);
+                        e.printStackTrace();
+                    } catch (Exception e) {
+                        Log.logSevere("IODispatcher", "main run job had errors (2), dump to " + f + " failed. Input files are " + f1 + " and " + f2, e);
+                    	e.printStackTrace();
                     }
                     continue loop;
                 }
@@ -167,11 +179,11 @@ public class IODispatcher <ReferenceType extends Reference> extends Thread {
             }
             Log.logInfo("IODispatcher", "loop terminated");
         } catch (InterruptedException e) {
-            e.printStackTrace();
             Log.logSevere("IODispatcher", "main run job was interrupted (3)", e);
-        } catch (Exception e) {
             e.printStackTrace();
+        } catch (Exception e) {
             Log.logSevere("IODispatcher", "main run job failed (4)", e);
+            e.printStackTrace();
         } finally {
             Log.logInfo("IODispatcher", "terminating run job");
             controlQueue = null;

source/de/anomic/plasma/plasmaWordIndex.java

@@ -429,12 +429,6 @@ public final class plasmaWordIndex {
         final int urlLength = url.toNormalform(true, true).length();
         final int urlComps = htmlFilterContentScraper.urlComps(url.toString()).length;
         
-        // check if merger is running
-        if (this.merger != null && !this.merger.isAlive()) {
-        	log.logSevere("re-starting IODispatcher");
-        	this.merger.start();
-        }
-        
         // iterate over all words of context text
         final Iterator<Map.Entry<String, Word>> i = condenser.words().entrySet().iterator();
         Map.Entry<String, Word> wentry;