#!/usr/bin/php
<?php // version="5+"

   /*
    *   Guess codepage - adaptive version
    *   A tool which determines in which codepage is the input
    *
    *   (C) 2005, Rozsnyo Daniel <daniel@rozsnyo.com>
    *
    *   This modification reads 4KB blocks while the codepage could not be determined
    *
    */

    class codepageGuess {
    
        const   BLOCKSIZE = 4096;
    
        private $handle;
        private $counts = array();
        
        public static function fromFile( $filename ) {
            $guess = new codepageGuess( fopen($filename,'rb') );
            $cp    = $guess->doGuess();
            fclose($guess->handle);
            return $cp;
        }
        
        private function __construct( $handle ) {
            // initialize input
            $this->handle = $handle;
            // initialize histogram
            for($i=0; $i<=255; $i++) $this->counts[$i] = 0;
        }
        
        private function processInputBlock() {
            $toGo = self::BLOCKSIZE;
            while(($toGo-->0)&&($c=fgetc($this->handle))!==false) $this->counts[ord($c)]++;
            return $toGo<0;
        }
        
        private function doGuess() {
            while(true) {
                // read a bit..
                $more = $this->processInputBlock();
                // make indices
                $iso  = $this->counts[0xA9]+$this->counts[0xAE]+$this->counts[0xB9]+$this->counts[0xBE];
                $utf  = $this->counts[0xC3]+$this->counts[0xC4]+$this->counts[0xC5]+$this->counts[0xC6];
                $win  = $this->counts[0x8A]+$this->counts[0x8E]+$this->counts[0x9A]+$this->counts[0x9E];
                // guess
                if ( $utf > ($iso+$win) ) return 'UTF-8';
                if ( $iso > $win ) return 'ISO-8859-2';
                if ( $iso < $win ) return 'WINDOWS-1250';
                // ..or not
                if (!$more) return 'WINDOWS-1250'; // default guess due to our geographical location
            }
        }
        
    }
    
    echo codepageGuess::fromFile('php://stdin'), "\n";
    
?>