source: trunk/gsdl3/src/java/org/greenstone/gsdl3/gs3build/doctypes/TextRecogniser.java@ 5800

package org.greenstone.gsdl3.gs3build.doctypes;

import java.io.*;
import java.net.*;

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

public class TextRecogniser implements RecogniserInterface
{
    DocumentList listRepository;

    public TextRecogniser(DocumentList listRepository)
    {   this.listRepository = listRepository;
    }

    public boolean parseDocument(METSFile file)
    {
        String MIMEType = file.getMIMEType();
        if (MIMEType == null || 
              MIMEType.equals("text/plain")) {
            URL location = file.getLocation();
            return this.parseDocument(location);
        }
        return false;
    }

    public boolean parseDocument(URL url)
    {   if (url.toString().startsWith("file://")) {
            String fileName = url.toString().substring(7);

            if (fileName.endsWith(".txt") ||
              fileName.endsWith(".text"))
            {   this.listRepository.addDocument(new TextDocument(url));
                // TODO: spawn knowledge of children too...
                System.out.println(">>> Posting text document " + fileName);
                return true;
            }
        }
        else
        {   // Check MIME type
        }
        
        return false;
    }
}