Problem

How can I remove (update) anything in the index?

Solution

import org.egothor.dir.TankerImpl;
import org.egothor.parser.Tokenizer;
import org.egothor.data.*;
import org.egothor.parser.plain.Plain;
import org.egothor.parser.misc.*;

import java.io.StringReader;

public class Updater {
    public static void main(String[] args) throws Exception {
        TankerImpl index = new TankerImpl("index/",false,32,10);

        // generate 100 docs and append them to the index
        for (int i=0;i<100;i++) {
            FTField field = new MyTxtField("This is my document no. "+i);
            DocMetadata dm = new DocMetadata();
            dm.put("S","Some summary...");
            dm.put("L","Location of this doc is on my HDD...");
            dm.put("T","Title of the document "+i);
            dm.put("K",Integer.toString(i));
            Document doc = new Document(dm, field);
            index.append(doc);
        }
        index.close();

        // remove docs 0,2,4,6,8,10, etc.
        // see the "K" field

        // open index
        index = new TankerImpl("index/",false,32,10);
        // get enumeration of all barrels
        java.util.Enumeration e = index.elements();
        long sum = 0;
        while (e.hasMoreElements()) {
            Object o = e.nextElement();
            // if it is a barrel that is able to modify its data structure
            if (o instanceof BarrelShaker) {
                BarrelShaker bs = (BarrelShaker) o;

                System.out.println("Barrel size: "+bs.size()+
                    " (already removed "+bs.deleted()+
                    " docs)");
                // iterate over meta data
                IMetaReader imr = bs.openDocMeta();
                long d = 0;
                while (imr.hasMoreElements()) {
                    DocMetadata mt = (DocMetadata) imr.nextElement();
                    // uid of the document in the barrel bs
                    long uid = imr.getUid();

                    int id = Integer.parseInt(mt.get("K","1").toString());
                    if (id%2 == 0) {
                        bs.removeDoc(uid);
                        d++;
                    }
                }
                imr.close();

                bs.close();

                System.out.println("    removed: "+d+
                    " (removed docs in the barrel: "+bs.deleted()+")");
                sum += d;
            } else if (o instanceof Barrel) {
                // if it is a different sort of a barrel (i.e. meta searcher)
                Barrel b = (Barrel) o;
                // ...only close the object
                b.close();
            }
        }
        index.close();

        System.out.println("Removed (total): "+sum);
    }
}

class MyTxtField extends FTField {
    String text;

    MyTxtField(String text) { this.text = text; }

    public Tokenizer words() {
        return
            new LowerCase( new Plain( new StringReader(text) ) );
    }
}

After first run, you will see:

Barrel size: 64 (already removed 0 docs)
    removed: 32 (removed docs in the barrel: 32)
Barrel size: 36 (already removed 0 docs)
    removed: 18 (removed docs in the barrel: 18)
Removed (total): 50

After second run, the following lines are printed out:

Barrel size: 64 (already removed 32 docs)
    removed: 0 (removed docs in the barrel: 32)
Barrel size: 36 (already removed 18 docs)
    removed: 0 (removed docs in the barrel: 18)
Barrel size: 36 (already removed 0 docs)
    removed: 18 (removed docs in the barrel: 18)
Barrel size: 64 (already removed 0 docs)
    removed: 32 (removed docs in the barrel: 32)
Removed (total): 50

-- LeoGalambos - 26 May 2004