benchmarks/Lexer.php

   1 <?php
   2
   3 require_once '../library/HTMLPurifier.auto.php';
   4 @include_once '../test-settings.php';
   5
   6 // PEAR
   7 require_once 'Benchmark/Timer.php'; // to do the timing
   8 require_once 'Text/Password.php'; // for generating random input
   9
  10 $LEXERS = array();
  11 $RUNS = isset($GLOBALS['HTMLPurifierTest']['Runs'])
  12     ? $GLOBALS['HTMLPurifierTest']['Runs'] : 2;
  13
  14 require_once 'HTMLPurifier/Lexer/DirectLex.php';
  15 $LEXERS['DirectLex'] = new HTMLPurifier_Lexer_DirectLex();
  16
  17 if (version_compare(PHP_VERSION, '5', '>=')) {
  18     require_once 'HTMLPurifier/Lexer/DOMLex.php';
  19     $LEXERS['DOMLex'] = new HTMLPurifier_Lexer_DOMLex();
  20 }
  21
  22 // custom class to aid unit testing
  23 class RowTimer extends Benchmark_Timer
  24 {
  25
  26     var $name;
  27
  28     function RowTimer($name, $auto = false) {
  29         $this->name = htmlentities($name);
  30         $this->Benchmark_Timer($auto);
  31     }
  32
  33     function getOutput() {
  34
  35         $total  = $this->TimeElapsed();
  36         $result = $this->getProfiling();
  37         $dashes = '';
  38
  39         $out = '<tr>';
  40
  41         $out .= "<td>{$this->name}</td>";
  42
  43         $standard = false;
  44
  45         foreach ($result as $k => $v) {
  46             if ($v['name'] == 'Start' || $v['name'] == 'Stop') continue;
  47
  48             //$perc = (($v['diff'] * 100) / $total);
  49             //$tperc = (($v['total'] * 100) / $total);
  50
  51             //$out .= '<td align="right">' . $v['diff'] . '</td>';
  52
  53             if ($standard == false) $standard = $v['diff'];
  54
  55             $perc = $v['diff'] * 100 / $standard;
  56             $bad_run = ($v['diff'] < 0);
  57
  58             $out .= '<td align="right"'.
  59                    ($bad_run ? ' style="color:#AAA;"' : '').
  60                    '>' . number_format($perc, 2, '.', '') .
  61                    '%</td><td>'.number_format($v['diff'],4,'.','').'</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, $RUNS;
  83
  84     $config = HTMLPurifier_Config::createDefault();
  85     $context = new HTMLPurifier_Context();
  86
  87     $timer = new RowTimer($name);
  88     $timer->start();
  89
  90     foreach($LEXERS as $key => $lexer) {
  91         for ($i=0; $i<$RUNS; $i++) $tokens = $lexer->tokenizeHTML($document, $config, $context);
  92         $timer->setMarker($key);
  93     }
  94
  95     $timer->stop();
  96     $timer->display();
  97 }
  98
  99 ?>
 100 <html>
 101 <head>
 102 <title>Benchmark: <?php print_lexers(); ?></title>
 103 </head>
 104 <body>
 105 <h1>Benchmark: <?php print_lexers(); ?></h1>
 106 <table border="1">
 107 <tr><th>Case</th><?php
 108 foreach ($LEXERS as $key => $value) {
 109     echo '<th colspan="2">' . htmlspecialchars($key) . '</th>';
 110 }
 111 ?></tr>
 112 <?php
 113
 114 // ************************************************************************** //
 115
 116 // sample of html pages
 117
 118 $dir = 'samples/Lexer';
 119 $dh  = opendir($dir);
 120 while (false !== ($filename = readdir($dh))) {
 121
 122     if (strpos($filename, '.html') !== strlen($filename) - 5) continue;
 123     $document = file_get_contents($dir . '/' . $filename);
 124     do_benchmark("File: $filename", $document);
 125
 126 }
 127
 128 // crashers, caused infinite loops before
 129
 130 $snippets = array();
 131 $snippets[] = '<a href="foo>';
 132 $snippets[] = '<a "=>';
 133
 134 foreach ($snippets as $snippet) {
 135     do_benchmark($snippet, $snippet);
 136 }
 137
 138 // random input
 139
 140 $random = Text_Password::create(80, 'unpronounceable', 'qwerty <>="\'');
 141
 142 do_benchmark('Random input', $random);
 143
 144 ?></table>
 145
 146 <?php
 147
 148 echo '<div>Random input was: ' .
 149   '<span colspan="4" style="font-family:monospace;">' .
 150   htmlspecialchars($random) . '</span></div>';
 151
 152 ?>
 153
 154
 155 </body></html>
 156 <?php
 157
 158 // vim: et sw=4 sts=4