#!/usr/bin/perl
# reads a file who's name is given in the command line
# prints the name of each document (<DOCNO>) and its text (<TEXT>) on one line

use strict;

my ($line, $in_text, $name, $text, $all_text);

$in_text=0;

while($line =<>)
{
chomp $line;
if ($line =~ /\<DOCNO\> (.+) \<\/DOCNO\>/) {$name=$1;}
elsif ($line =~ /\<\/DOC\>/)
 {
    $all_text =~ s/ +$//go;
    $all_text =~ s/^ +//go;
    $all_text =~ s/ +/ /go;
 
    print $name, "***", $all_text, "\n";
    $all_text="";
 }
elsif ($line =~ /\<TEXT\>/) {$in_text=1; $text ="";}
elsif ($line =~ /\<\/TEXT\>/)
   {$in_text=0;
    $all_text .= $text." ";
   }
elsif ($in_text)
 {
  $text .= $line." ";
 }
#print $line;
}