#!/usr/bin/perl
# Copyright 22nd October 2008, Cotman University 
# http://www.cotmanuniversity.com
# Created by Dr. Peter Kjeldsen
# Department for Computational Linguistics
# Conversion of Trados TM format to  CSV format
# File name: tracsv.pl
#
# Instructions
# Change name of input file, source and target language
# as required.
# Run from command prompt perl tracsv.pl  
#


# Opening input file
open (TXT, "<lingualizerWFtradosTM.txt") or die ("Cannot open file!");
@text = <TXT>;
close (TXT);


# Looping through input file content line by line
foreach $lines(@text)
{

    # Splitting line content
    @line = split (/\>/, $lines);

    if ($line[0] eq '<Seg L=EN-US')
      {
        # Opening/creating output file
        open (SRC, ">>src.txt") or die ("Cannot open file!");
        
        $rawsource = chomp ($line[1]);
        $source = $line[1];
                
         print SRC '"' . $source . '"' . "\n";
      }; 
}
 
# Closing output file
close (SRC); 
 
# Looping through input file content line by line
foreach $lines(@text)
{

    # Splitting line content
    @line = split (/\>/, $lines);
               
    if ($line[0] eq '<Seg L=DA-01') 
      {
        # Opening/creating output file
        open (TGT, ">>tgt.txt") or die ("Cannot open file!");
      
        $rawtarget = chomp ($line[1]);
        $target = $line[1];
        
         print TGT '"' . $target . '"' . "\n";
      };      
              
}
# Closing output file
close (TGT);

# Opening source input file
open (SRC, "<src.txt") or die ("Cannot open file!");

# Opening source input file
open (TGT, "<tgt.txt") or die ("Cannot open file!");

$sep = ",";

# Opening/creating output file
open (TM, ">>tracsv.csv") or die ("Cannot open file!");

# Creating header for output file
$header = '"Created by Lingualizer tmXl8tr","Created by Lingualizer tmXl8tr"' . "\n";

# Writing header in the output file
print TM $header;

# Looping through input file content line by line
while ($line = <SRC>)
{
$line =~ s/\n//;

print TM $line . $sep . <TGT>
 
} 
close (SRC);
close (TGT);
close (TM);