#!/usr/bin/perl -w

use DBI;
use Getopt::Long;

# Synthesizes text from a Markov Chain.
# Uses a mysql database, see genmarkov for details.
# Usage: mkmarkov table db_pass_for_root -l [number of words to generate]

$errors = &GetOptions("length|l=i" => \$length);

if(!$errors) {
	exit 1;
}

# warn "Reading mkv and mkp files...\n";

# open WORDS, "$ARGV[0].mkv" or die "Couldn't open $ARGV[0].mkv: $!\n";
# open PROBS, "$ARGV[0].mkp" or die "Couldn't open $ARGV[0].mkp: $!\n";

warn "Reading database markov, table $ARGV[0]...\n";

$dbh ||= DBI->connect("DBI:mysql:markov", "root", $ARGV[1]);
die "Unable to connect to SQL Server" unless $dbh;

$c = $dbh->prepare("SELECT prev,new,prob FROM $ARGV[0]");
$c->execute();

while (($prev, $new, $prob) = $c->fetchrow) {
	$words = 1;
	while ($new =~ m/ /g) {
		$words++;
	}
	$probabilities[$words] = 1;
	$words{$prev}->{$new} = $prob;
	# warn "words{$prev}->$new has been set to $prob\n";
}

$dbh->disconnect();

$words = 0;
while($words < $length) {
	$numwords = int (rand() * $#probabilities) + 1;
	if($words > 0) {
		$numwords = $words if $numwords > $words;
		$prevseq = "";
		for($i = $#probabilities; $i >= $#probabilities - ($numwords - 1); $i--) {
			$prevseq = "$prev[$i] $prevseq";
		}
		chop($prevseq);
	} else {
		$numwords = 1;
		$prevseq = "";
	}
	$words++; # += $numwords;
	warn "Getting a sequence of $numwords words...\n";
	$rval = rand();
	#warn "Target is $rval\n";
	$cprob = 0;
	warn "Prevseq is $prevseq\n";
	GETSEQ: foreach $word (sort randsort keys %{$words{$prevseq}}) {
		warn "Is $word the one?\n";
		$cprob += $words{$prevseq}->{$word};
		$theword = $word if $cprob >= $rval;
		#warn "cprob is $cprob\n";
		warn "Yep.\n" if $cprob >= $rval;
		last GETSEQ if $cprob >= $rval;
	}
	DONE: $theword =~ s/.+ (.+)$/$1/;
	warn "We got $theword.\n";
	print $theword . " ";
	push @prev, $theword;
	shift @prev if $#prev > $#probabilities;
}

sub randsort {
	return 1 if rand() <= .5;
	return -1;
}