#!/usr/bin/perl -T

use strict;
use CGI qw(:standard);
use CGI::Carp qw(fatalsToBrowser);
use DBI;
use Reviews;

$CGI::POST_MAX=1024 * 15;  # max 20K posts
$CGI::DISABLE_UPLOADS = 1;  # no uploads
$ENV{PATH} = "";

my $q = new CGI;
print $q->header(-expires=>'now');
print $q->start_html();

my $begin_time = time();

my $dbh = make_db_handle;

# ok, first...if we've seen this url before, just spit back what
# we returned last time.
# if we haven't seen it before, we need to cache our results at
# the end.

my $out;

my $url = $q->url(-relative=>1, -query=>1);
my $html = dbquery($dbh, qq/ 
	SELECT html FROM htmlcache WHERE url = "$url"
/);
if ($html) {
	#print "<h2>READING FROM CACHE...</h2>";
	print $html;
	exit;
}

#my $order = $q->param('sort') || "totDESC";
#$order =~ s/[^A-Za-z,]//g;
#$order = substr($order, 0, 30);
#$order =~ s/desc/DESC/g;
#$order =~ s/DESC/ DESC/g;

#my $limit = $q->param('limit') || 100; #only return 100 at a time
#$limit =~ s/[^\d]//g;
#$limit = substr($limit, 0, 4);
#my $offset = $q->param('offset') || 0;
#$offset =~ s/[^\d]//g;
#$offset = substr($offset, 0, 7);
# not interested in words that occur less than the "floor"
#my $floor = $q->param('floor') || 50; 
#$floor =~ s/[^\d]//g;
#$floor = substr($floor, 0, 4);

# get words from arguments (separated by space)
my $id = $q->param('id') || 1;
$id =~ s/[^\d ]//g;
my @word_ids = split(/ /, $id);
my @sentence_ids;
$out .= "<h3>words that coexist with:<br>";

my %seen_words;
# for each word, print words at the top, find sentences...
my $stuff_to_print;
my %word_id_to_word;
foreach my $word_id (@word_ids) {
	$word_id = substr($word_id, 0, 10);
	#print words at the top
	my $word = dbquery($dbh, qq/
		SELECT word FROM word WHERE id = $word_id
	/);
	$stuff_to_print .= "$word, ";
	$seen_words{$word} = 1;
	$word_id_to_word{$word_id} = $word;

	# find all sentences containing these words
	my $sentence_ids = dbquery($dbh, qq/
		SELECT sentence_id FROM word_sentence 
		WHERE word_id = $word_id
	/);
	push(@sentence_ids, @$sentence_ids);
}
$stuff_to_print =~ s/, $//;
$out .= "$stuff_to_print</h3>";

# next, find all other words in the sentences we collected
my %word_counts;
my %id_to_word;
my %seen_sentence;
my $seen_sentence_count;
foreach my $sentence_id (@sentence_ids) {
	next if $seen_sentence{$sentence_id};
	my $other_words = dbquery($dbh, qq/
		SELECT word_id,word FROM word_sentence,word
		WHERE sentence_id = $sentence_id
		AND word_id = word.id
	/);
	$seen_sentence{$sentence_id} = 1;
	$seen_sentence_count++;
	# if we only have one word, there is an error...this is the workaround
	if (! ref $$other_words[0]) { $other_words = [ $other_words]; }

	foreach my $other_word (@$other_words) {
		my $other_word_id = $$other_word[0];
		my $word_text = $$other_word[1];
		# don't count this word if it's the same as one of the 
		# words we already have
		next if ($seen_words{$word_text});
		$word_counts{$other_word_id}++;
		$id_to_word{$other_word_id} = $word_text;
	}
}

$out .= "<ol>\n";
# print 'em
my $count;
foreach my $other_word_id (sort { $word_counts{$b} <=> $word_counts{$a} } keys %word_counts) {
	# next if we have more than one word, but we only saw this once.
	#next if (($word_counts{$other_word_id} == 1) && ($#word_ids > 0));
	# next if we only saw this word once.
	#next if ($word_counts{$other_word_id} == 1);
	next if (length $id_to_word{$other_word_id} < 3);
	next if ($id_to_word{$other_word_id} eq "the");
	next if ($id_to_word{$other_word_id} eq "The");
	next if ($id_to_word{$other_word_id} eq "and");
	$count++;
	$out .= qq%<li><big><a href="word.pl?id=$other_word_id">$id_to_word{$other_word_id}</a>: $word_counts{$other_word_id}</big>%;
	# print a more advanced joint KWIC here, one for each word...
	$out .= "<small> [joint KWIC with: ";
	foreach my $word_id (@word_ids) {
		$out .= qq%<a href="concordance.pl?id=$word_id&id2=$other_word_id">$word_id_to_word{$word_id}</a> \n%;
	}
	$out .= "]</small>";
	last if $count == 500;
}
$out .= "</ol>";


my $end_time = time();
my $time_elapsed = $end_time - $begin_time;
$out .= "\n<p><b>query time: $time_elapsed seconds</b>\n";
my $total_sentences = $#sentence_ids + 1;
$out .= "<br>total sentences: $total_sentences, sentences after pruning: $seen_sentence_count\n";

print $out; 
print $q->end_html();

# now put this text in the database...
$url =~ s/"/\\"/g;
$out =~ s/"/\\"/g;
dbquery($dbh, qq/
	INSERT INTO htmlcache (url,html) VALUES ("$url","$out")
/);