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 require_once 'HTMLPurifier/Context.php';
  11
  12 $LEXERS = array();
  13 $RUNS = isset($GLOBALS['HTMLPurifierTest']['Runs'])
  14     ? $GLOBALS['HTMLPurifierTest']['Runs'] : 2;
  15
  16 require_once 'HTMLPurifier/Lexer/DirectLex.php';
  17 $LEXERS['DirectLex'] = new HTMLPurifier_Lexer_DirectLex();
  18
  19 if (!empty($GLOBALS['HTMLPurifierTest']['PEAR'])) {
  20     require_once 'HTMLPurifier/Lexer/PEARSax3.php';
  21     $LEXERS['PEARSax3'] = new HTMLPurifier_Lexer_PEARSax3();
  22 } else {
  23     exit('PEAR required to perform benchmark.');
  24 }
  25
  26 if (version_compare(PHP_VERSION, '5', '>=')) {
  27     require_once 'HTMLPurifier/Lexer/DOMLex.php';
  28     $LEXERS['DOMLex'] = new HTMLPurifier_Lexer_DOMLex();
  29 }
  30
  31 // PEAR
  32 require_once 'Benchmark/Timer.php'; // to do the timing
  33 require_once 'Text/Password.php'; // for generating random input
  34
  35 // custom class to aid unit testing
  36 class RowTimer extends Benchmark_Timer
  37 {
  38
  39     var $name;
  40
  41     function RowTimer($name, $auto = false) {
  42         $this->name = htmlentities($name);
  43         $this->Benchmark_Timer($auto);
  44     }
  45
  46     function getOutput() {
  47
  48         $total  = $this->TimeElapsed();
  49         $result = $this->getProfiling();
  50         $dashes = '';
  51
  52         $out = '<tr>';
  53
  54         $out .= "<td>{$this->name}</td>";
  55
  56         $standard = false;
  57
  58         foreach ($result as $k => $v) {
  59             if ($v['name'] == 'Start' || $v['name'] == 'Stop') continue;
  60
  61             //$perc = (($v['diff'] * 100) / $total);
  62             //$tperc = (($v['total'] * 100) / $total);
  63
  64             //$out .= '<td align="right">' . $v['diff'] . '</td>';
  65
  66             if ($standard == false) $standard = $v['diff'];
  67
  68             $perc = $v['diff'] * 100 / $standard;
  69             $bad_run = ($v['diff'] < 0);
  70
  71             $out .= '<td align="right"'.
  72                    ($bad_run ? ' style="color:#AAA;"' : '').
  73                    '>' . number_format($perc, 2, '.', '') .
  74                    '%</td><td>'.number_format($v['diff'],4,'.','').'</td>';
  75
  76         }
  77
  78         $out .= '</tr>';
  79
  80         return $out;
  81     }
  82 }
  83
  84 function print_lexers() {
  85     global $LEXERS;
  86     $first = true;
  87     foreach ($LEXERS as $key => $value) {
  88         if (!$first) echo ' / ';
  89         echo htmlspecialchars($key);
  90         $first = false;
  91     }
  92 }
  93
  94 function do_benchmark($name, $document) {
  95     global $LEXERS, $RUNS;
  96
  97     $config = HTMLPurifier_Config::createDefault();
  98     $context = new HTMLPurifier_Context();
  99
 100     $timer = new RowTimer($name);
 101     $timer->start();
 102
 103     foreach($LEXERS as $key => $lexer) {
 104         for ($i=0; $i<$RUNS; $i++) $tokens = $lexer->tokenizeHTML($document, $config, $context);
 105         $timer->setMarker($key);
 106     }
 107
 108     $timer->stop();
 109     $timer->display();
 110 }
 111
 112 ?>
 113 <html>
 114 <head>
 115 <title>Benchmark: <?php print_lexers(); ?></title>
 116 </head>
 117 <body>
 118 <h1>Benchmark: <?php print_lexers(); ?></h1>
 119 <table border="1">
 120 <tr><th>Case</th><?php
 121 foreach ($LEXERS as $key => $value) {
 122     echo '<th colspan="2">' . htmlspecialchars($key) . '</th>';
 123 }
 124 ?></tr>
 125 <?php
 126
 127 // ************************************************************************** //
 128
 129 // sample of html pages
 130
 131 $dir = 'samples/Lexer';
 132 $dh  = opendir($dir);
 133 while (false !== ($filename = readdir($dh))) {
 134
 135     if (strpos($filename, '.html') !== strlen($filename) - 5) continue;
 136     $document = file_get_contents($dir . '/' . $filename);
 137     do_benchmark("File: $filename", $document);
 138
 139 }
 140
 141 // crashers, caused infinite loops before
 142
 143 $snippets = array();
 144 $snippets[] = '<a href="foo>';
 145 $snippets[] = '<a "=>';
 146
 147 foreach ($snippets as $snippet) {
 148     do_benchmark($snippet, $snippet);
 149 }
 150
 151 // random input
 152
 153 $random = Text_Password::create(80, 'unpronounceable', 'qwerty <>="\'');
 154
 155 do_benchmark('Random input', $random);
 156
 157 ?></table>
 158
 159 <?php
 160
 161 echo '<div>Random input was: ' .
 162   '<span colspan="4" style="font-family:monospace;">' .
 163   htmlspecialchars($random) . '</span></div>';
 164
 165 ?>
 166
 167
 168 </body></html>