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     public $name;
  27
  28     public function __construct($name, $auto = false)
  29     {
  30         $this->name = htmlentities($name);
  31         $this->Benchmark_Timer($auto);
  32     }
  33
  34     public function getOutput()
  35     {
  36         $total  = $this->TimeElapsed();
  37         $result = $this->getProfiling();
  38         $dashes = '';
  39
  40         $out = '<tr>';
  41
  42         $out .= "<td>{$this->name}</td>";
  43
  44         $standard = false;
  45
  46         foreach ($result as $k => $v) {
  47             if ($v['name'] == 'Start' || $v['name'] == 'Stop') continue;
  48
  49             //$perc = (($v['diff'] * 100) / $total);
  50             //$tperc = (($v['total'] * 100) / $total);
  51
  52             //$out .= '<td align="right">' . $v['diff'] . '</td>';
  53
  54             if ($standard == false) $standard = $v['diff'];
  55
  56             $perc = $v['diff'] * 100 / $standard;
  57             $bad_run = ($v['diff'] < 0);
  58
  59             $out .= '<td align="right"'.
  60                    ($bad_run ? ' style="color:#AAA;"' : '').
  61                    '>' . number_format($perc, 2, '.', '') .
  62                    '%</td><td>'.number_format($v['diff'],4,'.','').'</td>';
  63
  64         }
  65
  66         $out .= '</tr>';
  67
  68         return $out;
  69     }
  70 }
  71
  72 function print_lexers()
  73 {
  74     global $LEXERS;
  75     $first = true;
  76     foreach ($LEXERS as $key => $value) {
  77         if (!$first) echo ' / ';
  78         echo htmlspecialchars($key);
  79         $first = false;
  80     }
  81 }
  82
  83 function do_benchmark($name, $document)
  84 {
  85     global $LEXERS, $RUNS;
  86
  87     $config = HTMLPurifier_Config::createDefault();
  88     $context = new HTMLPurifier_Context();
  89
  90     $timer = new RowTimer($name);
  91     $timer->start();
  92
  93     foreach($LEXERS as $key => $lexer) {
  94         for ($i=0; $i<$RUNS; $i++) $tokens = $lexer->tokenizeHTML($document, $config, $context);
  95         $timer->setMarker($key);
  96     }
  97
  98     $timer->stop();
  99     $timer->display();
 100 }
 101
 102 ?>
 103 <html>
 104 <head>
 105 <title>Benchmark: <?php print_lexers(); ?></title>
 106 </head>
 107 <body>
 108 <h1>Benchmark: <?php print_lexers(); ?></h1>
 109 <table border="1">
 110 <tr><th>Case</th><?php
 111 foreach ($LEXERS as $key => $value) {
 112     echo '<th colspan="2">' . htmlspecialchars($key) . '</th>';
 113 }
 114 ?></tr>
 115 <?php
 116
 117 // ************************************************************************** //
 118
 119 // sample of html pages
 120
 121 $dir = 'samples/Lexer';
 122 $dh  = opendir($dir);
 123 while (false !== ($filename = readdir($dh))) {
 124
 125     if (strpos($filename, '.html') !== strlen($filename) - 5) continue;
 126     $document = file_get_contents($dir . '/' . $filename);
 127     do_benchmark("File: $filename", $document);
 128
 129 }
 130
 131 // crashers, caused infinite loops before
 132
 133 $snippets = array();
 134 $snippets[] = '<a href="foo>';
 135 $snippets[] = '<a "=>';
 136
 137 foreach ($snippets as $snippet) {
 138     do_benchmark($snippet, $snippet);
 139 }
 140
 141 // random input
 142
 143 $random = Text_Password::create(80, 'unpronounceable', 'qwerty <>="\'');
 144
 145 do_benchmark('Random input', $random);
 146
 147 ?></table>
 148
 149 <?php
 150
 151 echo '<div>Random input was: ' .
 152   '<span colspan="4" style="font-family:monospace;">' .
 153   htmlspecialchars($random) . '</span></div>';
 154
 155 ?>
 156
 157
 158 </body></html>
 159 <?php
 160
 161 // vim: et sw=4 sts=4