<?php

$aConf = parse_ini_file('conf.ini');

set_time_limit(0);
ini_set('memory_limit', '700M');

function microtime_float()
{
	list($usec, $sec) = explode(" ", microtime());
	return ((float) $usec + (float) $sec);
}

function convert($size)
{
	$unit = array('b', 'kb', 'mb', 'gb', 'tb', 'pb');
	return @round($size / pow(1024, ($i = floor(log($size, 1024)))), 2) . ' ' . $unit[$i];
}
$time_start = microtime_float();


class InvicoKeywords
{

	/**
	 *
	 * @var int Max word for suggest phrase
	 */
	private $_iMaxWords = 5;
	/**
	 *
	 * @var int min Word length for suggest word
	 */
	private $_iMinWordLength = 2;
	/**
	 *
	 * @var array List of phrases  (phrase => freq)
	 */
	private $_aPhrases = array();
	private $_aPhrasesTest = array();
	/**
	 *
	 * @var string uggest table name
	 */
	private $_tableName = 'keywords';
	/**
	 *
	 * @var type minimum freq for adding
	 */
	private $_iMinFreq = 2;

	/**
	 *
	 * @param string $name
	 */
	public function setTableName($name)
	{
		$this->_tableName = $name;
	}

	private function _addPhrase($i, $j, $aWordArray)
	{
		$sWord = '';
		$bAdd = true;
		for ($s = 0; $s <= $j; $s++) {
			$is = $i + $s;
			if (isset($aWordArray[$is])) {
				$sWord .= " " . $aWordArray[$is];
			}
			else {
				$bAdd = false;
			}
		}
		$sWord = trim($sWord);
		if ($bAdd) {
			if (isset($this->_aPhrases[$sWord])) {
				$this->_aPhrases[$sWord]++;
			}
			else {
				$this->_aPhrases[$sWord] = 1;
			}
		}
	}

	private function _allowParse($word)
	{
		if (is_numeric($word)) {
			return false;
		}
		//elseif (strlen($word) < $this->_iMinWordLength) {
		elseif (!isset($word{$this->_iMinWordLength})) {
			return false;
		}
		return true;
	}

	private function _parseArray($aWordArray)
	{
		$i = 0;
		while (isset($aWordArray[$i])) {
			if ($this->_allowParse($aWordArray[$i])) {
				$j = 0;
				while ($j < $this->_iMaxWords) {
					$this->_addPhrase($i, $j, $aWordArray);
					$j++;
				}
			}
			$i++;
		}
	}

	/**
	 * Parse givven text
	 *
	 * @param string $sText
	 */
	public function addText($sText)
	{
		$aTextList = strtolower(trim(strip_tags($sText)));

		preg_match_all("/[\w]+/", $aTextList, $matches);

		$this->_parseArray($matches[0]);
	}

	/**
	 *
	 * @return array
	 */
	public function getKeywords()
	{
		return $this->_aPhrases;
	}

	/**
	 *
	 * @return string
	 */
	function getSql()
	{
		$sSql = "
DROP TABLE IF EXISTS {$this->_tableName};

CREATE TABLE `{$this->_tableName}` (
  `keyword` varchar(255) NOT NULL,
  `freq` int(11) NOT NULL
) ENGINE=MyISAM DEFAULT CHARSET=utf8;
";

		$i = 0;
		foreach ($this->getKeywords() as $keyword => $freq) {
			if ($freq > $this->_iMinFreq) {
				if ($i != 0) {
					$sSql .= ",\n";
				}
				else {
					$sSql .= "\nINSERT INTO {$this->_tableName} VALUES ";
				}
				$sSql .= "('" . $keyword . "', $freq)";
				$i++;
				if ($i > 1000) {
					$sSql .= ";\n";
					$i = 0;
				}
			}
		}

		return $sSql;
	}

	/**
	 *  Sending data to php://stdout
	 */
	public function sendToStdout()
	{
		$out = fopen("php://stdout", "w+");

		fwrite($out, $this->getSql());

		fclose($out);
	}

}

mysql_connect($aConf['mysql.host'], $aConf['mysql.user'], $aConf['mysql.pass']);
mysql_select_db($aConf['mysql.dbname']);

$i = mysql_fetch_object(mysql_query($aConf['query.count']));
$chunks = ceil($i->number / $aConf['limit']);

for ($i=0; $i<$chunks; $i++) {
	$from = $i * $aConf['limit'];
	$oResult = mysql_query($aConf['query.data'] . " limit $from," . $aConf['limit']);
	$oKeywordParser = new InvicoKeywords();
	while ($obj = mysql_fetch_object($oResult)) {
		$oKeywordParser->addText($obj->mytext);
		unset($obj);
	}
}
mysql_free_result($oResult);

$oKeywordParser->sendToStdout();

$time_end = microtime_float();
$time = $time_end - $time_start;

//echo "Did nothing in $time seconds\n";
//echo "Peak: " . convert(memory_get_peak_usage(true)) . " Usage: " . convert(memory_get_usage(true)) . "\n";