source: trunk/greenstone3-extensions/gs3build/src/org/greenstone/gsdl3/gs3build/doctypes/IndexRecogniser.java@ 12188

package org.greenstone.gsdl3.gs3build.doctypes;

import java.io.*;
import java.net.*;
import java.util.ArrayList;

import org.greenstone.gsdl3.gs3build.metadata.*;

public class IndexRecogniser extends AbstractRecogniser
{
    
    public IndexRecogniser(DocumentList listRepository)
    { 
    this.preferred_mime_type = "text/plain";
    this.filename_extensions = new ArrayList();
    this.filename_extensions.add("index.txt");
    this.document_type = IndexDocument.INDEX_DOCUMENT_TYPE;
    }
    

    public boolean parseDocument(URL url)
    { 
    String filename = null;

    if (url.getProtocol().equals("file")) {
        filename = url.getPath();
    }
    
    if (filename != null) {
        String leafName;
        int leafAt = filename.lastIndexOf(File.separator);
        if (leafAt >= 0) {
        leafName = filename.substring(leafAt+1);
        } else {
        leafName = filename;
        }

        if (leafName.equals("index.txt")) {
        
        System.out.println("Posting Index Document " + filename);
        IndexDocument doc = new IndexDocument(url);
        this.list_repository.addDocument(doc);
        // TODO: spawn knowledge of children too...
        //              System.out.println(doc.getDocumentText());
        return true;
        }
    } else {
        // TODO: get Mime type remotely, and then proceed if required
    }
    return false;
    }
}