#!/usr/bin/perl
# Copyright 13th. October 2008, Cotman University 
# http://www.cotmanuniversity.com
# Created by Dr. Peter Kjeldsen
# Department for Computational Linguistics
# Conversion of CVS format to  Trados Text TM format
# File name: csvtratext.pl
#
# Instructions
# Change name of input file, source and target language
# as required.
# Run from command prompt perl csvtratext.pl  
#

# Setting up formated variable for handling date and time
($sec, $min, $hour, $mday, $mon, $year) = localtime(time);

$mon++;
$year+= 1900;
$hour = sprintf("%02d", $hour);
$min = sprintf("%02d", $min);
$sec = sprintf("%02d", $sec);

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

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

# Creating header for output file
$hoved = '<TWBExportFile version="7.0" generator="Lingualizer tmXl8r" build="1.0">' . "\n";

print TM $hoved;
print TM "<RTF Preamble>\n";
print TM "<FontTable>\n";
print TM "</RTF Preamble>\n";

# Looping through input file content line by line
foreach $lines(@text)
{
# Splitting line content and creating translation units

    @line = split (/","/, $lines);
        
        # Chopping leading " from source string by reversing, chopping and reversing
        $words = reverse($line[0]);
        $char = chop($words);
        $rev2str = reverse($words);
        $CleanSource = $rev2str;
        
        # Creating the source element of the translation unit
        $SourceFull = '<Seg L=EN-US>' . $CleanSource;

                
        # Chopping \n from the end of target string
        $char = chop($line[1]);
        $TargetChop = $line[1];
        # Chopping " from the end of the target string
        $char = chop($TargetChop);
        $CleanTarget = $TargetChop;   

        # Creating the target element of the translation unit
        $TargetFull = '<Seg L=DA>' . $CleanTarget;
   
    print TM "<TrU>\n";
    print TM "<CrD> $mday" . "$mon" . "$year, $hour:$min:$sec" . "\n";
    print TM "<CrU>Lingualizer\n";
    print TM "$SourceFull\n";
    print TM "$TargetFull\n";
    print TM "</TrU>\n";
}
# Closing output file
close (TM);