source: trunk/greenstone3-extensions/gs3build/src/org/greenstone/gsdl3/gs3build/doctypes/GS2PerlRecogniser.java@ 12191

package org.greenstone.gsdl3.gs3build.doctypes;

import java.util.ArrayList;
import java.io.File;
import java.net.URL;
import org.w3c.dom.Element;

import org.greenstone.gsdl3.util.GSFile;
import org.greenstone.gsdl3.util.Processing;
import org.greenstone.gsdl3.gs3build.util.HTTPTools;

public class GS2PerlRecogniser extends AbstractRecogniser
{
    public String plugin_name;
    public GS2PerlRecogniser()
    { 
    //this.preferred_mime_type = "application/pdf";
    //this.preferred_mime_type = "text/plain";
    this.filename_extensions = new ArrayList();
    //this.filename_extensions.add(".pdf");
    //this.filename_extensions.add(".txt");
    this.document_type = METSDocument.METS_DOCUMENT_TYPE;
    }

    public boolean configure(Element config_elem) {
    String mime_type = config_elem.getAttribute("mimeType");
    if (mime_type.equals("")) {
        System.err.println("GS2PerlRecogniser configure Error: no mimetype specified");
        return false;
    }
    this.preferred_mime_type = mime_type;
    String plugin = config_elem.getAttribute("plugin");
    if (plugin.equals("")) {
        System.err.println("GS2PerlRecogniser configure Error: no plugin specified");
        return false;
    }
    this.plugin_name = plugin;
    
    String extensions = config_elem.getAttribute("filenameExtensions");
    if (extensions.equals("")) {
        System.err.println("GS2PerlRecogniser configure Error: no filename extensions specified");
        return false;
    }
    String [] exts = extensions.split("|");
    for (int i=0; i<exts.length; i++) {
        this.filename_extensions.add(exts[i]);
    }
    return true;
    
    }
    public boolean parseDocument(URL url)
    { 
    String filename = null;
    
    if (url.getProtocol().equals("file")) {
        filename = url.getPath();
    }
    
    if (filename != null) {
        if ( isAcceptedFilename(filename)) {
        
        System.out.println("Posting "+document_type+" Document " + filename);
        METSDocument doc = processFile(url, filename);
        
        if (doc != null) {
            this.list_repository.addDocument(doc);
            return true;
        }
        return false;
        }
    } else {
        // Get Mime type remotely, and then proceed if required
        String mimeType = HTTPTools.getMIMEType(url);
        
        if (mimeType == this.preferred_mime_type) {
        System.out.println("Posting "+document_type+" Document " + url.toString());
        METSDocument doc = processFile(url, null);
        if (doc != null) {
        
            this.list_repository.addDocument(doc);
            return true;
        }
        return false;
        }
    }
    return false;
    }


    protected METSDocument processFile(URL url, String filename) {

    // we can only do files at this stage
    if (filename == null) return null;
    File file = new File(filename);
    String local_dir = file.getParent();
    System.err.println("local dir = "+local_dir);
    
    //String command = "importone.pl -saveas METS -collectdir /research/kjdon/home/gsdl3/web/sites/localsite/collect -importdir "+local_dir+" -archivedir "+local_dir+" -- kath TEXTPlug "+ file.getName();
    String command = "importone.pl -saveas METS -collectdir "+GSFile.collectDir(this.coll_manager.siteHome)+" -importdir "+local_dir+" -archivedir "+local_dir+" -- "+this.coll_manager.collectionName +" " +this.plugin_name+" "+ file.getName();
    System.err.println("Command = "+command);
    
    int exit_value = Processing.runProcess(command);
    if (exit_value == 0) {
        System.out.println("converted success");
    } else {
        System.out.println("Unable to convert to mets");
        
        return null ;
    }
    
    //read in the archives.inf file to see the directory for the new mets
    File arcinfo = new File(local_dir, "archives.inf");
    if (!arcinfo.isFile()) {
        System.err.println("cant find archives.inf file ("+arcinfo.getPath()+", aborting");
        return null;
    }
    try {
    String content = new String(GSFile.readFile(arcinfo.getPath()));
    content = content.trim();
    System.err.println("content =  "+content);
    String items[] = content.split("\\s+");
    String filepath = items[1];
    // delete the archive file
    arcinfo.delete();
    File mets_file = new File(local_dir, filepath);
    System.err.println("mets file = "+mets_file.getPath());
    return new METSDocument(mets_file.toURL());
    } catch (Exception e) {
        System.err.println("processfile: "+e);
        return null;
    }
    }
}