#!/usr/bin/perl

$story_count = 0;
$sentence_counter = 0;
$corpus_file = "/home/james/scifi/corpus/scifi.corpus.data";

open(CORPUS, "$corpus_file") || die "Can't open file $corpus_file\n";

print "What term?\n";
$term = <>;
chop($term);

print "Output type? (s/f/db)\n";
$output = <>;
chop($output);

if ($output =~ /f/)
{
    print "What filename for output?\n";
    $unix_filename = <>;
    chop($unix_filename);
    open(UNIXFILE, ">$unix_filename") || die "Can't open outfile $output_filename\n";
}
if ($output =~ /db/)
{
    print "What filename for output? (.txt)\n";
    $db_filename = <>;
    chop($db_filename);
    open(DATABASE, ">$db_filename") || die "Can't open outfile $output_filename\n";
}
$/ = "<ENDRECORD>";

print STDERR "\nSearching Early Science Fiction Corpus\n";
while (<CORPUS>)
{
    if ($term =~ / /)
    {
	$_ =~ s/\n/ /g;
	$_ =~ s/[ ]+/ /g;
    }
    if ($_ =~ /$term/i)
    {
	$_ =~ m|<AUTH>(.*?)</AUTH>|;
	$auth = $1;
	
	$_ =~ m|<TITLE>(.*?)</TITLE>|;
	$title = $1;
	
	if ($_ =~ /<STORY>[^<]/)
	{
	    $_ =~ m|<STORY>(.*?)</STORY>|;
	    $story = $1;
	    $story_count = 1;
	}
	else
	{
	    $story = "";
	    $story_count = 0;
	}
	
	$_ =~ m|<YEAR>(.*?)</YEAR>|;
	$year = $1;
	
	$_ =~ m|<CHAP>(.*?)</CHAP>|;
	$chap = $1;
	
	$rec = $_;
	if ($rec =~ /\n/)
	{
	    $rec =~ s/\n/ /g;
	}
	$rec =~ s/[ ]+/ /g;
	while ($rec =~ s/(.*?[\.\?\!])//)
	{
	    $sentence = $1;
	    $sentence =~ s/^ //;
	    if ($sentence =~ m/$term/)
	    {
		$sentence_counter++;
		if ($sentence =~ /<\/CHAP>/)
		{
		    $sentence =~ s/.*<\/CHAP>//g;
		}
		if ($output =~ /f/)
		{
		    if ($sentence_counter == 10)
		    {
			print STDERR "\nMore than ten!\n\n";
		    }
		    if ($sentence_counter == 50)
		    {
			print STDERR "\nMore than fifty!\n\n";
		    }
		    if ($sentence_counter == 200)
		    {
			print STDERR "\nMore than two-hundred!\n\n";
		    }
		    if ($story_count == 0)
		    {
			print UNIXFILE "\n$year\, $auth\, $title\, $chap\n$sentence\n";
		    }
		    else
		    {
			print UNIXFILE "\n$year\, $auth\, \`$story', $title\, $chap\n$sentence\n";
		    }
		}
		elsif ($output =~ /db/)
		{
		    $sortkey = $term;
		    $sortkey =~ s/[ ]+//g;
		    $sortkey =~ s/[\-\']//g;
		    print DATABASE "\%Start\:\n";
		    print DATABASE "\%sortkey\:$sortkey\n";
		    print DATABASE "\%headword\:$term\n";
		    print DATABASE "\%pos\:\n";
		    print DATABASE "\%page\:\n";
		    print DATABASE "\%col\:\n";
		    print DATABASE "\%chap\:$chap\n";
		    if ($title =~ /Hector S/)
		    {
			$year =~ s/ \(1911\)//;
		    }
		    print DATABASE "\%year\:$year\n";
		    print DATABASE "\%author\:$auth\n";
		    print DATABASE "\%story\:$story\n";
		    print DATABASE "\%mag\/book\:$title\n";
		    if ($title =~ /Hector S/)
		    {
			print DATABASE "\%date\:\(1911\)\n";
		    }
		    else
		    {
			print DATABASE "\%date\:\n";
		    }
		    print DATABASE "\%place\:(Project Gutenberg)\n";
		    print DATABASE "\%ed\:\n";
		    print DATABASE "\%anthology\:\n";
		    print DATABASE "\%section\:\n";
		    print DATABASE "\%cit\:$sentence\n";
		    print DATABASE "\%www\:\n";
		    print DATABASE "\%def\:\n";
		    print DATABASE "\%notes\:\n";
		}
		else
		{
		    if ($story_count == 0)
		    {
			print "\n$year\, $auth\, $title\, $chap\n$sentence\n";
		    }
		    else
		    {
			print "\n$year\, $auth\, \`$story', $title\, $chap\n$sentence\n";
		    }
		}
	    }
	}
    }
}
if ($output =~ /db/)
{
    print DATABASE "\%End\:\n";
}

if ($sentence_counter == 0)
{
    print "\nThere were none\!\n\n";
}
close(CORPUS);
print STDERR "\cG\cG\cG\cG";