<?php

//first attempt at a markov chain, never did get it working perfectly on probablities but it's something i'll come back to.


$word_list = explode(" ",file_get_contents("test.txt"));
$word_list = clean_word_list($word_list);
$tree = create_wordtree($word_list);
$seed = array_rand($tree);
$words = 50;
$sentence = create_markov_sentence($seed,$words,$tree);
$sentence = ucfirst($seed)." ".$sentence;
echo $sentence.'<br/>';
echo "<pre>";
	print_r($tree);
echo "</pre>";

function clean_word_list($word_list){
	for($i=0;$i<count($word_list);$i++){
		$cur_word = preg_replace("/[^a-zA-Z 0-9 , . ']+/", "", $word_list[$i]);
		if(!empty($cur_word)){
			$word_list[$i] = $cur_word;
		}
	}
	return($word_list);
}

function create_wordtree($word_list){
	$tree = array();
	$word_size = count($word_list);
	for($i=0;$i<$word_size;){
		$root = $word_list[$i];
		$keys = array_keys($word_list,$root);
		foreach($keys as $key=>$num){
			$tree[$root][] = $word_list[$num+1];
		}
		$tree[$root] = array_unique($tree[$root]);
		$i++;
	}
	return $tree;
}

function create_markov_sentence($seed,$words,$tree){
	if($tree[$seed]){
		for($i = 0;$i<$words;$i++){
		$random_seed = array_rand($tree[$seed]);
			if($tree[$seed][$random_seed]){
				$results .= ' '.$tree[$seed][$random_seed];
				$seed = $tree[$seed][$random_seed];
			}
			else{
				$results .= "";
			}
		}
		return $results;
	}
	else{
		return false;
	}
}

?>