#!/usr/bin/perl
# Jussi Karlgren, april 2000
# Del av programpaketet PARA
# http://www.sics.se/jussi/Verktyg/Para
# omformar text till prologtermer mening för mening:
# item(Löpnummer,Etikett,ProloglistaAvOrd).
# använd växeln -l för att välja
# etikett till prologtermen. 

%abbr = {"ABl.", "abl"};

require "getopts.pl";
&Getopts('l:');
$i = 0;
print (":- multifile item/3.\n:- dynamic item/3.\n");
while(<>) {
    $opt_l = $ARGV unless $opt_l;
    tr/ÉČĀÂA-ZÅÄÖÜ/éčāâa-zåäöü/;
    s/([\)\(\]\[\/])/\ $1\ /g;	 # add some space
    s/n\'t/\ not\ /g;	 # n't -> not
    s/\'/\ \ /g;	 # ' -> space (for 'll or for l' d' etc)
    s/,/\ ,\ /g;	 # put some space around the comma
    s/\>/\ \>\ /g;	 # put some space around the bracket
    s/(\D)([\.;:!\?])(\D)/$1\ \.\ $3/g;	 # space around punctuation (sentence break) 
    s/(\d)\.(\s)/$1$2/g;	 # take out period after digit (ordinal in many lges)
    @words = split /\s+/; # split line into words 	
    $pw = "";
    $out = "";
    foreach $word (@words) {  
	next if $word eq '\\';
	$out = $out."'".$pw."'" unless $pw eq "";
	if ($out && $word eq ".") {
	    print "item(",++$i,",'$opt_l',[",$out,"]).\n";
	    $pw = "";
	    $out = "";
	} else {
	    $out = $out.", " unless $pw eq "";
	    $pw = $word;
	};
    };	
    $out = $out."'".$pw."'" unless $pw eq "";
    print "item(",++$i,",'$opt_l',[",$out,"]).\n" if $out;

};