From 96930f0d2b9ade4a02bfc329db267ede6f7f6b9a Mon Sep 17 00:00:00 2001
From: hydrox <hydrox@6c8d7289-2bf4-0310-a012-ef5d649a1542>
Date: Wed, 7 Dec 2005 11:10:08 +0000
Subject: [PATCH] *)added function to removed malformed URLs from urlHash.db

git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@1182 6c8d7289-2bf4-0310-a012-ef5d649a1542
---
 source/de/anomic/plasma/plasmaCrawlLURL.java | 14 ++++++-
 source/yacy.java                             | 41 ++++++++++++++++++++
 2 files changed, 53 insertions(+), 2 deletions(-)

diff --git a/source/de/anomic/plasma/plasmaCrawlLURL.java b/source/de/anomic/plasma/plasmaCrawlLURL.java
index fe153bbbc..b45aa8eca 100644
--- a/source/de/anomic/plasma/plasmaCrawlLURL.java
+++ b/source/de/anomic/plasma/plasmaCrawlLURL.java
@@ -57,14 +57,18 @@ import java.io.IOException;
 import java.net.MalformedURLException;
 import java.net.URL;
 import java.text.SimpleDateFormat;
+import java.util.Collections;
 import java.util.Date;
+import java.util.HashSet;
 import java.util.Iterator;
 import java.util.LinkedList;
 import java.util.Locale;
 import java.util.Properties;
+import java.util.Set;
+
 import de.anomic.http.httpc;
-import de.anomic.kelondro.kelondroTree;
 import de.anomic.kelondro.kelondroException;
+import de.anomic.kelondro.kelondroTree;
 import de.anomic.server.serverCodings;
 import de.anomic.server.serverObjects;
 import de.anomic.server.logging.serverLog;
@@ -85,6 +89,8 @@ public final class plasmaCrawlLURL extends plasmaURL {
     private final LinkedList lcrawlResultStack; // 5 - local index: result of local crawling
     private final LinkedList gcrawlResultStack; // 6 - local index: triggered external
 
+    public static Set damagedURLS = Collections.synchronizedSet(new HashSet());
+    
     public plasmaCrawlLURL(File cachePath, int bufferkb) throws IOException {
         super();
         int[] ce = {
@@ -478,6 +484,10 @@ public final class plasmaCrawlLURL extends plasmaURL {
                 this.snippet = null;
                 return;
             }
+        } catch (MalformedURLException e) {
+            plasmaCrawlLURL.damagedURLS.add(this.urlHash);
+            System.out.println("DEBUG: Marked damaged Entry for removal (malformedURL). UrlHash: " + this.urlHash);
+            //serverLog.logSevere("PLASMA", "INTERNAL ERROR in plasmaLURL.entry/1: " + e.toString(), e);
         } catch (Exception e) {
             serverLog.logSevere("PLASMA", "INTERNAL ERROR in plasmaLURL.entry/1: " + e.toString(), e);
         }
@@ -774,4 +784,4 @@ public final class plasmaCrawlLURL extends plasmaURL {
         }
     }
 
-}
\ No newline at end of file
+}
diff --git a/source/yacy.java b/source/yacy.java
index ec192b1ff..426eddc9a 100644
--- a/source/yacy.java
+++ b/source/yacy.java
@@ -1170,6 +1170,43 @@ public final class yacy {
         return newargs;
     }
     
+    /**
+     * Uses an Iteration over urlHash.db to detect malformed URL-Entries.
+     * Damaged URL-Entries will be marked in a HashSet and removed at the end of the function.
+     *
+     * @param homePath Root-Path where all information is to be found.
+     */
+    private static void urldbcleanup(String homePath) {
+        File root = new File(homePath);
+        File dbroot = new File(root, "DATA/PLASMADB");
+        try {
+            plasmaCrawlLURL currentUrlDB = new plasmaCrawlLURL(new File(dbroot, "urlHash.db"), 4194304);
+            Iterator eiter = currentUrlDB.entries(true, false);
+            int iteratorCount=0;
+            while (eiter.hasNext()) {
+                eiter.next();
+                iteratorCount++;
+            }
+            try { Thread.sleep(1000); } catch (InterruptedException e) {}
+            System.out.println("URLs vorher: " + currentUrlDB.size() + " Entries loaded during Iteratorloop: " + iteratorCount + " kaputte URLs: " + plasmaCrawlLURL.damagedURLS.size());
+            synchronized(plasmaCrawlLURL.damagedURLS)
+            {
+                Iterator eiter2 = plasmaCrawlLURL.damagedURLS.iterator();
+                String urlHash;
+                while (eiter2.hasNext()) {
+                    urlHash = (String) eiter2.next();
+                    currentUrlDB.remove(urlHash);
+                    System.out.println("Removed UrlDB-Entry for urlHash: " + urlHash);
+                }
+            }
+            plasmaCrawlLURL.damagedURLS.clear();
+            System.out.println("URLs nachher: " + currentUrlDB.size() + " kaputte URLs: " + plasmaCrawlLURL.damagedURLS.size());
+            currentUrlDB.close();
+        } catch (IOException e) {
+            e.printStackTrace();
+        }
+    }
+    
     /**
     * Main-method which is started by java. Checks for special arguments or
     * starts up the application.
@@ -1260,6 +1297,10 @@ public final class yacy {
             if (args.length == 2) applicationRoot= args[1];
             String outfile = "urllist_" + System.currentTimeMillis() + ((html) ? ".html" : ".txt");
             urllist(applicationRoot, html, outfile);
+        } else if ((args.length >= 1) && (args[0].equals("-urldbcleanup"))) {
+            // generate a url list and save it in a file
+            if (args.length == 2) applicationRoot= args[1];
+            urldbcleanup(applicationRoot);
         } else {
             if (args.length == 1) applicationRoot= args[0];
             startup(applicationRoot, startupMemFree, startupMemTotal);