<?php
/**
 * Pluralize English words.
 * 
 * @file plural.php
 * 
 * Based on the algorithm by Damian Conway. See
 * http://www.csse.monash.edu.au/~damian/papers/HTML/Plurals.html
 * for more information on the algorithm used.
 */

include_once 'FileReader.php';


class Pluralizer {
    var $categories;
    var $classical = false;
    
    function Pluralizer($classical = false) {
        $this->categories = $this->getCategories();
        $this->classical = $classical;
    }
    
    function classical($classical = true) {
        $this->classical = $classical;
    }
    
    function is_classical() {
        return $this->classical;
    }
    
    function suffix($word, $suffix) {
        $suffix_len = strlen($suffix);
        if(preg_match('/'.$suffix.'$/',$word))
            return true;
        return false;
    }
    
    function stem($word, $suffix) {
        if(!$this->suffix($word, $suffix))
            return false;
        return substr($word, 0, -strlen($suffix));
    }
    
    function inflection($word, $singular_suffix, $plural_suffix) {
        $head = $this->stem($word, $singular_suffix);
        if($head == false)
            return false;
        return $head.$plural_suffix;
    }
    
    function category($word, $singular_suffix = null, $plural_suffix = null) {
        if($singular_suffix === null && $plural_suffix === null) {
            $key = 'default';
        } else {
            $key = $singular_suffix.','.$plural_suffix;
        }
        
        if(isset($this->categories[$key]['r'][$word])) {
            if($this->classical && $this->categories[$key]['r'][$word]['c'] !== null)
                    return $this->categories[$key]['r'][$word]['c'];
            
            if($this->categories[$key]['r'][$word]['p'] !== null)
                    return $this->categories[$key]['r'][$word]['p'];
        }
        return false;
    }
    
    function plural($word, $dict = null) {
        //special cases
        if($dict != null) {
            if(isset($dict[$word]))
                return $dict[$word];
        }
        
        if($this->suffix($word, 'fish') || $this->suffix($word, 'ois') || $this->suffix($word, 'sheep') ||
            $this->suffix($word, 'deer') || $this->suffix($word, 'pox') || $this->category($word, '', '') ||
            $this->suffix($word, '[A-Z].*ese')) {
                return $word;
        }
        
        if($this->category($word)) {
            return $this->category($word);
        }
        
        if($this->suffix($word, 'man'))
            return $this->inflection($word, 'man', 'men');
        if($this->suffix($word, '[lm]ouse'))
            return $this->inflection($word, 'ouse', 'ice');
        if($this->suffix($word,'tooth'))
            return $this->inflection($word, 'tooth','teeth');
        if($this->suffix($word,'goose'))
            return $this->inflection($word, 'goose','geese');
        if($this->suffix($word,'foot'))
            return $this->inflection($word, 'foot','feet');
        if($this->suffix($word,'zoon'))
            return $this->inflection($word, 'zoon','zoa');
        if($this->suffix($word,'[csx]is'))
            return $this->inflection($word, 'is','es');
            
        if($this->category($word, 'us', 'uses'))
            return $this->category($word, 'us', 'uses');
        if($this->category($word, 'ex', 'ices'))
            return $this->inflection($word, 'ex', 'ices');
        if($this->category($word, 'um', 'a'))
            return $this->inflection($word, 'um', 'a');
        if($this->category($word, 'on', 'a'))
            return $this->inflection($word, 'on', 'a');
        if($this->category($word, 'a', 'ae'))
            return $this->inflection($word, 'a', 'ae');
            
        if($this->classical) {
            if($this->suffix($word, 'trix'))
                return $this->inflection('trix','trices');
            if($this->suffix($word, 'eau'))
                return $this->inflection('eau','eaux');
            if($this->suffix($word, 'ieu'))
                return $this->inflection('ieu','ieux');
            if($this->suffix($word, '[iay]nx'))
                return $this->inflection('nx','nges');
            if($this->category($word,'en','ina'))
                return $this->category($word, 'en', 'ina');
            if($this->category($word, 'a', 'ata'))
                return $this->category($word, 'a', 'ata');
            if($this->category($word, 'is', 'ides'))
                return $this->category($word, 'is', 'ides');
            if($this->category($word, 'us', 'i'))
                return $this->category($word, 'us', 'i');
            if($this->category($word, 'us', 'us'))
                return $this->category($word, 'us', 'us');
            if($this->category($word, '', 'i'))
                return $this->category($word, '', 'i');
            if($this->category($word, '', 'im'))
                return $this->category($word, '','im');
        }
            
        if($this->suffix($word,'[cs]h'))
            return $this->inflection($word, 'h', 'hes');
        if($this->suffix($word,'ss'))
            return $this->inflection($word, 'ss', 'sses');
        if($this->suffix($word,'us'))
            return $this->inflection($word, 'us', 'uses');
            
        //barves??
        if($this->suffix($word,'[aeo]lf') || $this->suffix($word,'[^d]eaf') 
            || $this->suffix($word, '[^b]arf'))
            return $this->inflection($word, 'f', 'ves');
        
        if($this->suffix($word,'[nlw]ife'))
            return $this->inflection($word, 'fe', 'ves');
            
        if($this->suffix($word, '[aeiou]y'))
            return $this->inflection($word, 'y','ys');
        if($this->suffix($word, '[A-Z].*y'))
            return $this->inflection($word, 'y', 'ys');
        if($this->suffix($word, 'y'))
            return $this->inflection($word, 'y', 'ies');
        
        if($this->suffix($word, '[aeiou]o'))
            return $this->inflection($word, 'o', 'os');
        if($this->suffix($word, 'o'))
            return $this->inflection($word, 'o', 'oes');
            
        if($this->suffix($word, '[aeiou]x'))
            return $this->inflection($word, 'x', 'xes');
        
        //add an s
        return $word.'s';
    }
    
    function getCategories($filename = 'plural.txt') {
        $f = new FileReader($filename);
        $categories = array( 'default' => array( 'r' => array()));
        $current_cat = null;
        while(($l = $f->getLine()) !== false) {
            $line = trim($l);
            
            //comment
            if($line[0] == ';')
                continue;
                
            //processing instruction
            if($line[0] == ':') {
                preg_match('/(\:)([a-zA-Z_]+)((\(([\w \t]+(\,[\w \t]+)*)?\))?)/', $line, $matches);
                
                $instruction = $matches[2];
                if(isset($matches[5])) {
                    $params = explode(',',$matches[5]);
                    foreach($params as $i => $param) {
                        $params[$i] = trim($params[$i]);
                    }
                }
                
                if($instruction == 'cat') {
                    $cat_name = $params[0].','.$params[1];
                    $categories[$cat_name] = array('s' => $params[0], 'p' => $params[1], 'r' => array());
                    $current_cat = &$categories[$cat_name];
                } else if($instruction == 'end_cat') {
                    $current_cat = &$categories['default'];
                }
                continue;
            }
            
            if(strlen($line) > 1) {
                $parts = preg_split("/[\s]+/", $line);
                if(count($parts) == 1) {
                    $rule = array('s' => $parts[0], 'p' => $parts[0], 'c' => null);
                } else if(count($parts) == 2) {
                    $rule = array('s' => $parts[0], 'p' => $parts[1], 'c' => null);
                } else if(count($parts) == 3) {
                    $rule = array('s' => $parts[0], 'p' => null, 'c' => $parts[2]);
                } else if(count($parts) == 4) {
                    $rule = array('s' => $parts[0], 'p' => $parts[1], 'c' => $parts[3]);
                }
                
                if($current_cat != null)
                    $current_cat['r'][$rule['s']] = $rule;
                else
                    $categories['default']['r'][$rule['s']] = $rule;
            }
        }
        $f->close();
    
        return $categories;
    }
}
?>