source: trunk/gsdl3/src/java/org/greenstone/gsdl3/gs3build/doctypes/HTMLRecogniser.java@ 5944

package org.greenstone.gsdl3.gs3build.doctypes;

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

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

public class HTMLRecogniser implements RecogniserInterface
{
  DocumentList listRepository;

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

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

    if (url.toString().startsWith("file://")) {
      fileName = url.toString().substring(7);
    }
    else if (url.toString().startsWith("file:/")) {
      fileName = url.toString().substring(5);
    }

    if (fileName != null) {
      if (fileName.endsWith(".htm") ||
      fileName.endsWith(".html"))
      { System.out.println("Posting HTML Document " + fileName);

        HTMLDocument doc = new HTMLDocument(url);
    this.listRepository.addDocument(doc);
    return true;
      }
    }
    else {
      // Get Mime type remotely, and then proceed if required
      String mimeType = HTTPTools.getMIMEType(url);

      if (mimeType == "text/html")
      { System.out.println("Posting HTML Document " + url.toString());
            
        HTMLDocument doc = new HTMLDocument(url);
    this.listRepository.addDocument(doc);
    return true;
      }
    }
    return false;
  }
}