<?php

/**
 * This class use soundex algorithm to create a regular expressions 
 * based in a givem word this can perform a soundex comparisom on large texts
 */
class SoundRegex{
	
    private $_string = "";	
    private $_word = "";	
    private $_patterns = array();
    private $_pattern = "";	
    private $_matches = array();
    private $_tolerance = 50;	


    /** 
     * @param string $word
     * @param string $string 
     */
    public function __construct($word, $string){

            $this->_word = $word;
            $this->_string = $string;	

            $this->prepare();	
    }	

    /**
     * Set tolerance for word's matches
     * @param float $tolerance 
     */
    public function setTolerance($tolerance){
        $this->_tolerance = $tolerance;
    }

    /**
     * Load the patterns that will used in algorithm
     */
    private function prepare(){
        $this->_patterns[0] = array("B","P","F","V","b","p","f","v");
        $this->_patterns[1] = array("C","G","J","K","Q","S","X","Z","c","g","j","k","q","s","x","z","�","�");
        $this->_patterns[2] = array("D","T","d","t");
        $this->_patterns[3] = array("L","l");
        $this->_patterns[4] = array("M","N","m","n");
        $this->_patterns[5] = array("R","r");
        $this->_patterns[6] = array("A","Â","À","Á","Ã","a","â","à","á","ã","ä","Ä",
                                 "E","Ê","È","É","Ë","e","ê","ë","é","è",
                                 "I","Î","Ì","Í","Ï","i","î","ï","í","ì",
                                 "O","Ô","Ò","Ó","Ö","Õ","o","ô","ö","ó","ò","ö",
                                 "U","Û","Ù","Ú","Ü","u","û","û","ü","ú","ù",
                                 "Y","W","H","y","w","h");				
    }

    /**
     * Returns ER criated by algorithm
     * @return ER string
     */
    public function getPattern(){
        return $this->_pattern;
    }

    private function execute(&$matches = array()){

        if(!$this->_word || !$this->_string){
            return 0;
        }

        $this->_word = trim($this->_word);		

        $this->_pattern = "";

        for($i=0; $i < strlen($this->_word); $i++){
            $letter = $this->_word{$i};

            if($i == 0){
                $this->_pattern.= "/".$letter;
            }
            else{		
                if($letter != $this->_word[$i-1]){	

                    if(in_array($letter, $this->_patterns[0])){
                        $this->_pattern.= "[".implode("",$this->_patterns[0])."]{1,}";
                    }							
                    elseif(in_array($letter, $this->_patterns[1])){
                        $this->_pattern.= "[".implode("",$this->_patterns[1])."]{1,}";
                    }								
                    elseif(in_array($letter, $this->_patterns[2])){
                        $this->_pattern.= "[".implode("",$this->_patterns[2])."]{1,}";
                    }
                    elseif(in_array($letter, $this->_patterns[3])){
                        $this->_pattern.= "[".implode("",$this->_patterns[3])."]{1,}";
                    }				
                    elseif(in_array($letter, $this->_patterns[4])){
                        $this->_pattern.= "[".implode("",$this->_patterns[4])."]{1,}";
                    }				
                    elseif(in_array($letter, $this->_patterns[5])){
                        $this->_pattern.= "[".implode("",$this->_patterns[5])."]{1,}";
                    }							
                    elseif(in_array($letter, $this->_patterns[6])){
                        $this->_pattern.= "[".implode("",$this->_patterns[6])."]*";
                    }				
                    else{
                        $this->_pattern.= ".?";
                    }				
                }				
            }				
        }
        $this->_pattern.= "/i";
        return preg_match_all($this->_pattern, $this->_string, $matches);
    }	

    /**
     * Get the matches found by algorithm
     * @return mixed array of matches or false 
     */
    public function getMatches(){
        $matches = array();	
        if(empty($this->_matches)){			
            if($this->execute($matches)){						
                for($i=0;$i<count($matches);$i++){				
                    $matches[$i] = array_unique($matches[$i]);				
                }
                $this->_matches = $matches;	
                return $this->_matches;
            }
            else{
                return false;
            }
        }
        else{
            return $this->_matches;
        }		
    }	

    /**
     * The similarity percent is the index of array
     * @return array matches ordered for percent similarity
     */
    public function getMatchesSimilarity(){		
        $matches = $this->getMatches();	
        $orderby = array();		
        for($i=0;$i<count($matches);$i++){						
            foreach($matches[$i] as $key => $match){
                $percent = 0;
                similar_text($this->_word, $match, $percent);
                if($percent >= $this->_tolerance){
                    $orderby[$percent] = $match; 					
                }
            }
        }
        krsort($orderby);	
        return $orderby;	
    }

    /**
     * @return string the text analized by algorithm, but with the matches in bold tags
     */
    public function getString(){
        $string = $this->_string;
        $matches = $this->_matches;
        for($i=0;$i<count($matches);$i++){								
            foreach($matches[$i] as $match){
                $percent = 0;
                similar_text($this->_word, $match, $percent);
                if($percent >= $this->_tolerance){
                    $string = str_replace($match,"<b>".$match."</b>",$string);
                }
            }
        }	
        return $string;
    }
}

?>