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 @include_once '../test-settings.php';
   7
   8 require_once 'HTMLPurifier/ConfigSchema.php';
   9 require_once 'HTMLPurifier/Config.php';
  10
  11 $LEXERS = array();
  12 $RUNS = isset($GLOBALS['HTMLPurifierTest']['Runs'])
  13     ? $GLOBALS['HTMLPurifierTest']['Runs'] : 2;
  14
  15 require_once 'HTMLPurifier/Lexer/DirectLex.php';
  16 $LEXERS['DirectLex'] = new HTMLPurifier_Lexer_DirectLex();
  17
  18 if (!empty($GLOBALS['HTMLPurifierTest']['PEAR'])) {
  19     require_once 'HTMLPurifier/Lexer/PEARSax3.php';
  20     $LEXERS['PEARSax3'] = new HTMLPurifier_Lexer_PEARSax3();
  21 } else {
  22     exit('PEAR required to perform benchmark.');
  23 }
  24
  25 if (version_compare(PHP_VERSION, '5', '>=')) {
  26     require_once 'HTMLPurifier/Lexer/DOMLex.php';
  27     $LEXERS['DOMLex'] = new HTMLPurifier_Lexer_DOMLex();
  28 }
  29
  30 // PEAR
  31 require_once 'Benchmark/Timer.php'; // to do the timing
  32 require_once 'Text/Password.php'; // for generating random input
  33
  34 // custom class to aid unit testing
  35 class RowTimer extends Benchmark_Timer
  36 {
  37
  38     var $name;
  39
  40     function RowTimer($name, $auto = false) {
  41         $this->name = htmlentities($name);
  42         $this->Benchmark_Timer($auto);
  43     }
  44
  45     function getOutput() {
  46
  47         $total  = $this->TimeElapsed();
  48         $result = $this->getProfiling();
  49         $dashes = '';
  50
  51         $out = '<tr>';
  52
  53         $out .= "<td>{$this->name}</td>";
  54
  55         $standard = false;
  56
  57         foreach ($result as $k => $v) {
  58             if ($v['name'] == 'Start' || $v['name'] == 'Stop') continue;
  59
  60             //$perc = (($v['diff'] * 100) / $total);
  61             //$tperc = (($v['total'] * 100) / $total);
  62
  63             //$out .= '<td align="right">' . $v['diff'] . '</td>';
  64
  65             if ($standard == false) $standard = $v['diff'];
  66
  67             $perc = $v['diff'] * 100 / $standard;
  68
  69             $out .= '<td align="right">' . number_format($perc, 2, '.', '') .
  70                    '%</td><td>'.number_format($v['diff'],4,'.','').'</td>';
  71
  72         }
  73
  74         $out .= '</tr>';
  75
  76         return $out;
  77     }
  78 }
  79
  80 function print_lexers() {
  81     global $LEXERS;
  82     $first = true;
  83     foreach ($LEXERS as $key => $value) {
  84         if (!$first) echo ' / ';
  85         echo htmlspecialchars($key);
  86         $first = false;
  87     }
  88 }
  89
  90 function do_benchmark($name, $document) {
  91     global $LEXERS, $RUNS;
  92
  93     $timer = new RowTimer($name);
  94     $timer->start();
  95
  96     foreach($LEXERS as $key => $lexer) {
  97         for ($i=0; $i<$RUNS; $i++) $tokens = $lexer->tokenizeHTML($document);
  98         $timer->setMarker($key);
  99     }
 100
 101     $timer->stop();
 102     $timer->display();
 103 }
 104
 105 ?>
 106 <html>
 107 <head>
 108 <title>Benchmark: <?php print_lexers(); ?></title>
 109 </head>
 110 <body>
 111 <h1>Benchmark: <?php print_lexers(); ?></h1>
 112 <table border="1">
 113 <tr><th>Case</th><?php
 114 foreach ($LEXERS as $key => $value) {
 115     echo '<th colspan="2">' . htmlspecialchars($key) . '</th>';
 116 }
 117 ?></tr>
 118 <?php
 119
 120 // ************************************************************************** //
 121
 122 // sample of html pages
 123
 124 $dir = 'samples/Lexer';
 125 $dh  = opendir($dir);
 126 while (false !== ($filename = readdir($dh))) {
 127
 128     if (strpos($filename, '.html') !== strlen($filename) - 5) continue;
 129     $document = file_get_contents($dir . '/' . $filename);
 130     do_benchmark("File: $filename", $document);
 131
 132 }
 133
 134 // crashers, caused infinite loops before
 135
 136 $snippets = array();
 137 $snippets[] = '<a href="foo>';
 138 $snippets[] = '<a "=>';
 139
 140 foreach ($snippets as $snippet) {
 141     do_benchmark($snippet, $snippet);
 142 }
 143
 144 // random input
 145
 146 $random = Text_Password::create(80, 'unpronounceable', 'qwerty <>="\'');
 147
 148 do_benchmark('Random input', $random);
 149
 150 ?></table>
 151
 152 <?php
 153
 154 echo '<div>Random input was: ' .
 155   '<span colspan="4" style="font-family:monospace;">' .
 156   htmlspecialchars($random) . '</span></div>';
 157
 158 ?>
 159
 160
 161 </body></html>