php/frekvencni_analyza/analyze_language.php

   1 #!/usr/bin/php
   2 <?php
   3 //Frequency analysis 0.2
   4 //Harvie 2oo7
   5 /*
   6  * Use this to crack monoalphabetic ciphers (vigenere,...)
   7  * It needs good language file (ex: one for technical another for medical language, etc...)
   8  * For more see this: http://en.wikipedia.org/wiki/Frequency_analysis_(cryptanalysis)
   9  *
  10  * Known bugs: this can assign one letter to more crypted letters
  11  */
  12
  13
  14 function freq_analyze($infile, $cereg = '[a-z ]') { //Make frequency fingerprint
  15         $debug = 100;
  16         $total = 0;
  17         $in = fopen($infile, 'r');
  18         while(!feof($in)) {
  19                 $c = strtolower(fgetc($in));
  20                 if(eregi($cereg, $c)) {
  21                         if(!isset($data[$c])) $data[$c] = 0;
  22                         $data[$c]++;
  23                         $total++;
  24                 }
  25         }
  26         fclose($in);
  27
  28         //Compute percents
  29         foreach($data as $c => $n) {
  30                 $data[$c] = ($n/$total)*100;
  31                 $debug -= $data[$c];
  32         }
  33
  34         //echo("Debug: $debug\n"); //Debug
  35         return($data);
  36 }
  37
  38 function freq_compare($cipher, $language) { //Do analysis (from two fingerprints)
  39         foreach($cipher as $c => $n) {
  40                 $min = 1024;
  41                 foreach($language as $dc => $dn) {
  42                         $nmin = abs($n-$dn);
  43                         //echo $nmin."\n"; //Debug
  44                         if($nmin < $min) {
  45                                 $min = $nmin;
  46                                 $decrypted[$c] = $dc;
  47                         }
  48                 }
  49         }
  50
  51         return($decrypted);
  52 }
  53
  54 function freq_errors($decrypted) { //Test success (run with uncrypted cipher)
  55         $errors = 0;
  56         $total = 0;
  57         foreach ($decrypted as $a => $b) {
  58                 if($a != $b) $errors++;
  59                 $total++;
  60         }
  61         $hits = $total - $errors;
  62         $perc = round(($hits/$total)*100);
  63         echo("$hits hits and $errors errors / $total total - that means, you are $perc% successfull\n");
  64         return $errors;
  65 }
  66
  67 ///CODE//////////////////////////////////////////////////////////////////////////
  68 $czech = freq_analyze('czech.txt');
  69 //print_r($czech); //Debug
  70 $cipher = freq_analyze('cipher.txt');
  71 //print_r($cipher); //Debug
  72
  73 $decrypted = freq_compare($cipher, $czech);
  74 freq_errors($decrypted);
  75 print_r($decrypted);