benchmarks/Lexer.php

   1 <?php
   2
   3 // emulates inserting a dir called HTMLPurifier into your class dir
   4 set_include_path(get_include_path() . PATH_SEPARATOR . '../library/');
   5
   6 require_once 'HTMLPurifier/ConfigDef.php';
   7 require_once 'HTMLPurifier/Config.php';
   8 require_once 'HTMLPurifier/Lexer/DirectLex.php';
   9 require_once 'HTMLPurifier/Lexer/PEARSax3.php';
  10
  11 $LEXERS = array(
  12     'DirectLex' => new HTMLPurifier_Lexer_DirectLex(),
  13     'PEARSax3'  => new HTMLPurifier_Lexer_PEARSax3()
  14 );
  15
  16 if (version_compare(PHP_VERSION, '5', '>=')) {
  17     require_once 'HTMLPurifier/Lexer/DOMLex.php';
  18     $LEXERS['DOMLex'] = new HTMLPurifier_Lexer_DOMLex();
  19 }
  20
  21 // PEAR
  22 require_once 'Benchmark/Timer.php'; // to do the timing
  23 require_once 'Text/Password.php'; // for generating random input
  24
  25 // custom class to aid unit testing
  26 class RowTimer extends Benchmark_Timer
  27 {
  28
  29     var $name;
  30
  31     function RowTimer($name, $auto = false) {
  32         $this->name = htmlentities($name);
  33         $this->Benchmark_Timer($auto);
  34     }
  35
  36     function getOutput() {
  37
  38         $total  = $this->TimeElapsed();
  39         $result = $this->getProfiling();
  40         $dashes = '';
  41
  42         $out = '<tr>';
  43
  44         $out .= "<td>{$this->name}</td>";
  45
  46         $standard = false;
  47
  48         foreach ($result as $k => $v) {
  49             if ($v['name'] == 'Start' || $v['name'] == 'Stop') continue;
  50
  51             //$perc = (($v['diff'] * 100) / $total);
  52             //$tperc = (($v['total'] * 100) / $total);
  53
  54             //$out .= '<td align="right">' . $v['diff'] . '</td>';
  55
  56             if ($standard == false) $standard = $v['diff'];
  57
  58             $perc = $v['diff'] * 100 / $standard;
  59
  60             $out .= '<td align="right">' . number_format($perc, 2, '.', '') .
  61                    '%</td>';
  62
  63         }
  64
  65         $out .= '</tr>';
  66
  67         return $out;
  68     }
  69 }
  70
  71 function print_lexers() {
  72     global $LEXERS;
  73     $first = true;
  74     foreach ($LEXERS as $key => $value) {
  75         if (!$first) echo ' / ';
  76         echo htmlspecialchars($key);
  77         $first = false;
  78     }
  79 }
  80
  81 function do_benchmark($name, $document) {
  82     global $LEXERS;
  83
  84     $timer = new RowTimer($name);
  85     $timer->start();
  86
  87     foreach($LEXERS as $key => $lexer) {
  88         $tokens = $lexer->tokenizeHTML($document);
  89         $timer->setMarker($key);
  90     }
  91
  92     $timer->stop();
  93     $timer->display();
  94 }
  95
  96 ?>
  97 <html>
  98 <head>
  99 <title>Benchmark: <?php print_lexers(); ?></title>
 100 </head>
 101 <body>
 102 <h1>Benchmark: <?php print_lexers(); ?></h1>
 103 <table border="1">
 104 <tr><th>Case</th><?php
 105 foreach ($LEXERS as $key => $value) {
 106     echo '<th>' . htmlspecialchars($key) . '</th>';
 107 }
 108 ?></tr>
 109 <?php
 110
 111 // ************************************************************************** //
 112
 113 // sample of html pages
 114
 115 $dir = 'samples/Lexer';
 116 $dh  = opendir($dir);
 117 while (false !== ($filename = readdir($dh))) {
 118
 119     if (strpos($filename, '.html') !== strlen($filename) - 5) continue;
 120     $document = file_get_contents($dir . '/' . $filename);
 121     do_benchmark("File: $filename", $document);
 122
 123 }
 124
 125 // crashers, caused infinite loops before
 126
 127 $snippets = array();
 128 $snippets[] = '<a href="foo>';
 129 $snippets[] = '<a "=>';
 130
 131 foreach ($snippets as $snippet) {
 132     do_benchmark($snippet, $snippet);
 133 }
 134
 135 // random input
 136
 137 $random = Text_Password::create(80, 'unpronounceable', 'qwerty <>="\'');
 138
 139 do_benchmark('Random input', $random);
 140
 141 ?></table>
 142
 143 <?php
 144
 145 echo '<div>Random input was: ' .
 146   '<span colspan="4" style="font-family:monospace;">' .
 147   htmlspecialchars($random) . '</span></div>';
 148
 149 ?>
 150
 151
 152 </body></html>