#!/usr/bin/perl

# Author: Pete Whitelock
# Start Date: 15.5.94

# simple KWIC
$usage = q!

Usage: kwic [-ikb] [-c int|-l int -r int] pattern (filename)

prints pattern in $opt_c characters of context on either side
or L,R characters of context on left or right respectively (defaults to 50)

-i means case-insensitive
-k means print a tab-separated initial key in output as well
-b means print &nbsp instead of spaces for html to browser

set -c to 0 and pipe to sus (sort|uniq -c|sort -nr) to count instances of string
in file

!;

use Getopt::Std;
getopts('hikbc:l:r:');

die "$usage" if $opt_h;

$printing_key = $opt_k;
$case_insensitive = $opt_i;
$html = $opt_b;
$left_context = $opt_l || $opt_c || 50;
$right_context = $opt_r || $opt_c || 50;

$pattern = shift(@ARGV);

if ($pattern eq '') {print "$usage\n"; exit;}

# protect pattern
$pattern =~ s#/#\\/#g;

if (defined $ARGV[0]) {
  open(INPUT,$ARGV[0]) || die "Couldn't open file $ARGV[0]\n";
}
else {open (INPUT,"-");}

while(<>) {
  if ($printing_key) {
    ($key) = /^([^\t]*\t)/;
    $key =~ s#//P/Corpora and Linguistic Tools/BNCTagged/./../##;
    s/^[^\t]*\t//;
  }
  else {
    $key = '';
  }
  $key =~ s/\t/     /g;
  $key_length = length($key);

  # updated this 9.12.99 to display tabs correctly
  s/\t/     /g;
  $left = '';
  $match = '';
  if ($case_insensitive) {
    while (/($pattern)/ig) {
      &do;
    }
  }
  else {
    while (/($pattern)/g) {
      &do
    }
  }
}

sub do {

    ($left,$match,$right) = ($`,$&,$');
    $left =~ /.{0,$left_context}$/;
    $tleft = $&;
    $pad = ' ' x ($left_context - length($tleft));
    $right =~ /^.{0,$right_context}/;

    printf "%-20s", $key if $key;
    if ($html) {
      $pad =~ s/ /&nbsp;/g;
    }
    print $pad,$tleft,$match,$&,"\n";
  }

  exit;