From 02f80130137279b6888285bef731961198fc6ce8 Mon Sep 17 00:00:00 2001
From: orbiter <orbiter@6c8d7289-2bf4-0310-a012-ef5d649a1542>
Date: Mon, 7 Nov 2005 14:57:37 +0000
Subject: [PATCH] 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
---
 .../anomic/plasma/plasmaWordIndexCache.java   | 29 ++++++++++++-------
 source/yacy.java                              |  2 +-
 2 files changed, 19 insertions(+), 12 deletions(-)

diff --git a/source/de/anomic/plasma/plasmaWordIndexCache.java b/source/de/anomic/plasma/plasmaWordIndexCache.java
index 8da039703..268afc3f9 100644
--- a/source/de/anomic/plasma/plasmaWordIndexCache.java
+++ b/source/de/anomic/plasma/plasmaWordIndexCache.java
@@ -574,18 +574,25 @@ public final class plasmaWordIndexCache implements plasmaWordIndexInterface {
                     return -container.size();
                 } else {
                     // the combined container will fit, read the container
-                    Iterator entries = entity.elements(true);
-                    plasmaWordIndexEntry entry;
-                    while (entries.hasNext()) {
-                        entry = (plasmaWordIndexEntry) entries.next();
-                        container.add(new plasmaWordIndexEntry[]{entry}, System.currentTimeMillis());
+                    try {
+                        Iterator entries = entity.elements(true);
+                        plasmaWordIndexEntry entry;
+                        while (entries.hasNext()) {
+                            entry = (plasmaWordIndexEntry) entries.next();
+                            container.add(new plasmaWordIndexEntry[]{entry}, System.currentTimeMillis());
+                        }
+                        // we have read all elements, now delete the entity
+                        entity.deleteComplete();
+                        entity.close(); entity = null;
+                        // 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;                        
                     }
-                    // we have read all elements, now delete the entity
-                    entity.deleteComplete();
-                    entity.close(); entity = null;
-                    // integrate the container into the assortments; this will work
-                    assortmentCluster.storeTry(wordhash, container);
-                    return size;
                 }
             }
         } finally {
diff --git a/source/yacy.java b/source/yacy.java
index 62fe195df..c9e3cc81f 100644
--- a/source/yacy.java
+++ b/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