maintenance/benchmarks/bench_utf8_title_check.php

   1 <?php
   2 /**
   3  * @file
   4  * @ingroup Benchmark
   5  */
   6
   7 require_once( dirname( __FILE__ ) . '/Benchmarker.php' );
   8
   9 /**
  10  * This little benchmark executes the regexp used in Language->checkTitleEncoding() and compares its execution time
  11  * against that of mb_check_encoding, if available.
  12  */
  13 class bench_utf8_title_check extends Benchmarker {
  14
  15         private $canRun;
  16
  17         private $data;
  18
  19         public function __construct() {
  20                 parent::__construct();
  21
  22                 $this->data = array (
  23                         "",
  24                         "United States of America", // 7bit ASCII
  25                         "S%C3%A9rie%20t%C3%A9l%C3%A9vis%C3%A9e",
  26                         "Acteur%7CAlbert%20Robbins%7CAnglais%7CAnn%20Donahue%7CAnthony%20E.%20Zuiker%7CCarol%20Mendelsohn",
  27                         // This comes from bug 36839
  28                         "Acteur%7CAlbert%20Robbins%7CAnglais%7CAnn%20Donahue%7CAnthony%20E.%20Zuiker%7CCarol%20Mendelsohn%7C"
  29                         . "Catherine%20Willows%7CDavid%20Hodges%7CDavid%20Phillips%7CGil%20Grissom%7CGreg%20Sanders%7CHodges%7C"
  30                         . "Internet%20Movie%20Database%7CJim%20Brass%7CLady%20Heather%7C"
  31                         . "Les%20Experts%20(s%C3%A9rie%20t%C3%A9l%C3%A9vis%C3%A9e)%7CLes%20Experts%20:%20Manhattan%7C"
  32                         . "Les%20Experts%20:%20Miami%7CListe%20des%20personnages%20des%20Experts%7C"
  33                         . "Liste%20des%20%C3%A9pisodes%20des%20Experts%7CMod%C3%A8le%20discussion:Palette%20Les%20Experts%7C"
  34                         . "Nick%20Stokes%7CPersonnage%20de%20fiction%7CPersonnage%20fictif%7CPersonnage%20de%20fiction%7C"
  35                         . "Personnages%20r%C3%A9currents%20dans%20Les%20Experts%7CRaymond%20Langston%7CRiley%20Adams%7C"
  36                         . "Saison%201%20des%20Experts%7CSaison%2010%20des%20Experts%7CSaison%2011%20des%20Experts%7C"
  37                         . "Saison%2012%20des%20Experts%7CSaison%202%20des%20Experts%7CSaison%203%20des%20Experts%7C"
  38                         . "Saison%204%20des%20Experts%7CSaison%205%20des%20Experts%7CSaison%206%20des%20Experts%7C"
  39                         . "Saison%207%20des%20Experts%7CSaison%208%20des%20Experts%7CSaison%209%20des%20Experts%7C"
  40                         . "Sara%20Sidle%7CSofia%20Curtis%7CS%C3%A9rie%20t%C3%A9l%C3%A9vis%C3%A9e%7CWallace%20Langham%7C"
  41                         . "Warrick%20Brown%7CWendy%20Simms%7C%C3%89tats-Unis"
  42                 );
  43
  44                 $this->canRun = function_exists ( 'mb_check_encoding' );
  45
  46                 if ( $this->canRun ) {
  47                         $this->mDescription = "Benchmark for using a regexp vs. mb_check_encoding to check for UTF-8 encoding.";
  48                         mb_internal_encoding( 'UTF-8' );
  49                 } else {
  50                         $this->mDescription = "CANNOT RUN benchmark using mb_check_encoding: function not available.";
  51                 }
  52         }
  53
  54         public function execute() {
  55                 if ( !$this->canRun ) {
  56                         return;
  57                 }
  58                 $benchmarks = array();
  59                 foreach ($this->data as $val) {
  60                         $benchmarks[] = array(
  61                                 'function' => array( $this, 'use_regexp' ),
  62                                 'args' => array( rawurldecode ( $val ) )
  63                         );
  64                         $benchmarks[] = array(
  65                                 'function' => array( $this, 'use_regexp_non_capturing' ),
  66                                 'args' => array( rawurldecode ( $val ) )
  67                         );
  68                         $benchmarks[] = array(
  69                                 'function' => array( $this, 'use_regexp_once_only' ),
  70                                 'args' => array( rawurldecode ( $val ) )
  71                         );
  72                         $benchmarks[] = array(
  73                                 'function' => array( $this, 'use_mb_check_encoding' ),
  74                                 'args' => array( rawurldecode ( $val ) )
  75                         );
  76                 }
  77                 $this->bench( $benchmarks );
  78                 print $this->getFormattedResults();
  79         }
  80
  81         private $isutf8;
  82
  83         function use_regexp( $s ) {
  84                 $this->isutf8 = preg_match( '/^([\x00-\x7f]|[\xc0-\xdf][\x80-\xbf]|' .
  85                                 '[\xe0-\xef][\x80-\xbf]{2}|[\xf0-\xf7][\x80-\xbf]{3})+$/', $s );
  86         }
  87
  88         function use_regexp_non_capturing( $s ) {
  89                 // Same as above with a non-capturing subgroup.
  90                 $this->isutf8 = preg_match( '/^(?:[\x00-\x7f]|[\xc0-\xdf][\x80-\xbf]|' .
  91                                 '[\xe0-\xef][\x80-\xbf]{2}|[\xf0-\xf7][\x80-\xbf]{3})+$/', $s );
  92         }
  93
  94         function use_regexp_once_only( $s ) {
  95                 // Same as above with a once-only subgroup.
  96                 $this->isutf8 = preg_match( '/^(?>[\x00-\x7f]|[\xc0-\xdf][\x80-\xbf]|' .
  97                                 '[\xe0-\xef][\x80-\xbf]{2}|[\xf0-\xf7][\x80-\xbf]{3})+$/', $s );
  98         }
  99
 100         function use_mb_check_encoding( $s ) {
 101                 $this->isutf8 = mb_check_encoding( $s, 'UTF-8' );
 102         }
 103
 104 }
 105
 106 $maintClass = 'bench_utf8_title_check';
 107 require_once( RUN_MAINTENANCE_IF_MAIN );