package gate.creole.vu.util;

import java.io.*;

/**
 * Created by IntelliJ IDEA.
 * User: marta
 * Date: Jan 17, 2005
 * Time: 12:23:49 PM
 * Utility to transform minipar output to a gate output.
 */
public class MiniparToGate {

    public static void main (String args[]){
        MiniparToGate mtg = new MiniparToGate();

        mtg.run();
    }

   public void run(){
       //this is the directory containing the files processed by MInipar
      File directory = new File("D:\\Test\\in\\TestCorpus\\");
      File currentFile;
      File[] allFiles = directory.listFiles();

       for (int i = 0; i < allFiles.length; i++){
           currentFile = allFiles[i];
           System.out.println("Current File " + currentFile.getName());
           processFile(currentFile);
       }
   }

    private void processFile(File file){
          String line, out, rel, token, pos, head, tag;
          //this is the directory where the trasnformed files will be written.
          File newFile = new File("D:\\Test\\out\\TestCorpus\\" + file.getName().replaceAll(".txt", ".xml"));
          int govStart, sentenceNr = 0;
          boolean firstSentence = true;


          try{
              BufferedReader fr = new BufferedReader(new FileReader(file));
              BufferedWriter fw = new BufferedWriter(new FileWriter(newFile));

              fw.write("<?xml version=\"1.0\" encoding=\"windows-1252\" ?>");
              fw.newLine();
              fw.write("<doc>");

              line = fr.readLine();

              while (line != null){

                  if (firstSentence){
                      firstSentence = false;
                      fw.write("<sentence nr=\"1\">");
                      sentenceNr = 1;
                  } else {
                        //at new sentence println.
                        if ( line.startsWith(">")){
                            fw.write("</sentence>");
                            fw.newLine();
                            sentenceNr = sentenceNr + 1;
                            fw.write("<sentence nr=\"" + sentenceNr + "\">");
                        }
                  }
                  //processes only lines that do not start with tehse symbols
                 if (!(line.startsWith(">") |
                       line.startsWith(")")|
                       line.startsWith("E"))) {

                     //eliminate "gov"
                     govStart = line.indexOf("(gov");
                     if (govStart > -1) {
                        line = line.substring(0, govStart);
                     }

                     line = line.trim();
                     line = line.replaceAll("\t", " ");

                    out = "<word ";


                     pos =line.substring(0, line.indexOf(" ")); pos = pos.trim();
                     out = out + "pos=\"" + pos + "\" ";
                     line = line.substring(line.indexOf(" "), line.length());
                     line = line.trim();


                     //1-to eliminate "("
                     token = line.substring(1, line.indexOf(" ")); token = token.trim();
                     if (token.startsWith("\"")){
                         token = token.substring(1, token.length() - 1);
                     }
                     out = out + "token=\"" + token + "\" ";
                     line = line.substring(line.indexOf(" "), line.length());
                     line = line.trim();

                     rel = line.substring(line.lastIndexOf(" "), line.length());
                     rel = rel.trim();
                     if (rel.endsWith(")")){
                         rel = rel.substring(0, rel.length() - 1);
                     }

                     out = out + "rel=\"" + rel + "\" ";
                     line = line.substring(0, line.lastIndexOf(" "));
                     line = line.trim();

                     head = line.substring(line.lastIndexOf(" "), line.length()); head = head.trim();
                     out = out + "head=\"" + head + "\" ";
                     line = line.substring(0, line.lastIndexOf(" "));
                     line = line.trim();

                     tag = line.substring(line.lastIndexOf(" "), line.length()); tag = tag.trim();
                     out = out + "tag=\"" + tag  + "\" ";
                     line = line.substring(0, line.lastIndexOf(" "));
                     line = line.trim();

                     if (line.startsWith("\"")){
                         line = line.substring(1, line.length() - 1);
                     }

                     if (line.startsWith("~")) {
                         line = token;
                     }
                     out = out + "lemma=\"" + line + "\" ";
                     out = out + ">" + token + "</word>";
                     fw.write(out);

                     }
                  line = fr.readLine();
              }

              fw.write("</sentence>");
              fw.write("</doc>");
              fr.close();
              fw.flush(); fw.close();

          } catch (Exception exc){
              exc.printStackTrace();
          }



    }

}