auto-delete of corrupted word files during word-migration

git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@1047 6c8d7289-2bf4-0310-a012-ef5d649a1542

source/de/anomic/plasma/plasmaWordIndexCache.java

@@ -574,6 +574,7 @@ public final class plasmaWordIndexCache implements plasmaWordIndexInterface {
                     return -container.size();
                 } else {
                     // the combined container will fit, read the container
+                    try {
                         Iterator entries = entity.elements(true);
                         plasmaWordIndexEntry entry;
                         while (entries.hasNext()) {
@@ -586,6 +587,12 @@ public final class plasmaWordIndexCache implements plasmaWordIndexInterface {
                         // integrate the container into the assortments; this will work
                         assortmentCluster.storeTry(wordhash, container);
                         return size;
+                    } catch (kelondroException e) {
+                        // database corrupted, we simply give up the database and delete it
+                        try {entity.close();} catch (Exception ee) {} entity = null;
+                        try {db.delete();} catch (Exception ee) {}
+                        return 0;                        
+                    }
                 }
             }
         } finally {

source/yacy.java

@@ -663,7 +663,7 @@ public final class yacy {
                 wordhash = wordfile.getName().substring(0, 12);
                 migration = wordIndexCache.migrateWords2Assortment(wordhash);
                 if (migration == 0)
-                    log.logInfo("SKIPPED  " + wordhash + ": too big");
+                    log.logInfo("SKIPPED  " + wordhash + ": " + ((wordfile.exists()) ? "too big" : "database corrupted; deleted"));
                 else if (migration > 0)
                     log.logInfo("MIGRATED " + wordhash + ": " + migration + " entries");
                 else