#!/usr/bin/perl

#--------------------------------------------------------------------#
# Concordance Generator
#       Date Written:   13-Aug-2001 04:02:11 PM
#       Last Modified:  14-Aug-2001 04:14:00 PM
#       Author:         Kurt Kincaid
#
#         This is free software and may be distributed under the
#         same terms as Perl itself.
#
#  A simple concordance generator, particularly useful for linguistic
#  analysis.
#--------------------------------------------------------------------#

use strict;
use vars qw($opt_h $opt_s);
use Getopt::Std;

my @theseWords;
my @theseLines;
my @found;
my %Count;
my %Line;
my ( $line, $word, $count, $LineNum );
my $VERSION = "1.0";

getopts( "hs:" );

if ( $opt_h ) {
    Usage();
}

my $file = shift || Usage();

open ( IN, $file ) || die "$file not found\n";
@theseLines = <IN>;
close (IN);
chomp @theseLines;

if ( $opt_s ) {
    Word($opt_s);
}

foreach $line ( @theseLines ) {
    $count++;
    $line = lc $line;
    $line =~ s/[.,:;?!]//g;
    while ( $line =~ /\b\w+\b/g ) {
        $word = $&;
        if ( $word =~ /\s/ || $word eq "" ) { next }
        $Count{$word}++;
        if ( defined $Line{$word} ) {
            $Line{$word} =~ m/(\d*?)$/;
            if ( $1 == $count ) {
                next;
            } else {
                $Line{$word} .= ", $count";
            }
        } else {
            $Line{$word} = $count;
        }
#        push @{$Line{$word}}, $count unless exists $Line{$word} && $L
+ine{$word}[-1] == $count;
    }
}

@theseWords = keys %Count;
@theseWords = sort @theseWords;
foreach $word ( @theseWords ) {
#    print ( "$word ($Count{$word}): ", join ', ', @{$Line{$word}}, "\
+n\n" );
    print ("$word ($Count{$word}): $Line{$word}\n\n");
}

sub Word {
    my $word = shift;
    foreach $line ( @theseLines ) {
        $LineNum++;
        $Line{$line} = $LineNum;
    }

    @found = grep { /$word/i } @theseLines;

    foreach $line ( @found ) {
        print ("$Line{$line}: $line\n");
    }
    exit;
}

sub Usage {
    print <<END;
Concordance Generator v$VERSION
    $0 [-h] [-s word] filename
    -h  Print this screen.
    -s  Perform a search for a specific word with immediate context.

END
    exit;
}
[download]