MediaWiki  master
bench_utf8_title_check.php
Go to the documentation of this file.
1 <?php
24 require_once __DIR__ . '/Benchmarker.php';
25 
33  private $data;
34 
35  private $isutf8;
36 
37  public function __construct() {
38  parent::__construct();
39 
40  // phpcs:disable Generic.Files.LineLength
41  $this->data = [
42  "",
43  "United States of America", // 7bit ASCII
44  "S%C3%A9rie%20t%C3%A9l%C3%A9vis%C3%A9e",
45  "Acteur%7CAlbert%20Robbins%7CAnglais%7CAnn%20Donahue%7CAnthony%20E.%20Zuiker%7CCarol%20Mendelsohn",
46  // This comes from T38839
47  "Acteur%7CAlbert%20Robbins%7CAnglais%7CAnn%20Donahue%7CAnthony%20E.%20Zuiker%7CCarol%20Mendelsohn%7C"
48  . "Catherine%20Willows%7CDavid%20Hodges%7CDavid%20Phillips%7CGil%20Grissom%7CGreg%20Sanders%7CHodges%7C"
49  . "Internet%20Movie%20Database%7CJim%20Brass%7CLady%20Heather%7C"
50  . "Les%20Experts%20(s%C3%A9rie%20t%C3%A9l%C3%A9vis%C3%A9e)%7CLes%20Experts%20:%20Manhattan%7C"
51  . "Les%20Experts%20:%20Miami%7CListe%20des%20personnages%20des%20Experts%7C"
52  . "Liste%20des%20%C3%A9pisodes%20des%20Experts%7CMod%C3%A8le%20discussion:Palette%20Les%20Experts%7C"
53  . "Nick%20Stokes%7CPersonnage%20de%20fiction%7CPersonnage%20fictif%7CPersonnage%20de%20fiction%7C"
54  . "Personnages%20r%C3%A9currents%20dans%20Les%20Experts%7CRaymond%20Langston%7CRiley%20Adams%7C"
55  . "Saison%201%20des%20Experts%7CSaison%2010%20des%20Experts%7CSaison%2011%20des%20Experts%7C"
56  . "Saison%2012%20des%20Experts%7CSaison%202%20des%20Experts%7CSaison%203%20des%20Experts%7C"
57  . "Saison%204%20des%20Experts%7CSaison%205%20des%20Experts%7CSaison%206%20des%20Experts%7C"
58  . "Saison%207%20des%20Experts%7CSaison%208%20des%20Experts%7CSaison%209%20des%20Experts%7C"
59  . "Sara%20Sidle%7CSofia%20Curtis%7CS%C3%A9rie%20t%C3%A9l%C3%A9vis%C3%A9e%7CWallace%20Langham%7C"
60  . "Warrick%20Brown%7CWendy%20Simms%7C%C3%89tats-Unis"
61  ];
62  // phpcs:enable
63 
64  $this->addDescription( "Benchmark for using a regexp vs. mb_check_encoding " .
65  "to check for UTF-8 encoding." );
66  }
67 
68  public function execute() {
69  $benchmarks = [];
70  foreach ( $this->data as $val ) {
71  $benchmarks[] = [
72  'function' => [ $this, 'use_regexp' ],
73  'args' => [ rawurldecode( $val ) ]
74  ];
75  $benchmarks[] = [
76  'function' => [ $this, 'use_regexp_non_capturing' ],
77  'args' => [ rawurldecode( $val ) ]
78  ];
79  $benchmarks[] = [
80  'function' => [ $this, 'use_regexp_once_only' ],
81  'args' => [ rawurldecode( $val ) ]
82  ];
83  $benchmarks[] = [
84  'function' => [ $this, 'use_mb_check_encoding' ],
85  'args' => [ rawurldecode( $val ) ]
86  ];
87  }
88  $this->bench( $benchmarks );
89  }
90 
91  protected function use_regexp( $s ) {
92  $this->isutf8 = preg_match( '/^([\x00-\x7f]|[\xc0-\xdf][\x80-\xbf]|' .
93  '[\xe0-\xef][\x80-\xbf]{2}|[\xf0-\xf7][\x80-\xbf]{3})+$/', $s );
94  }
95 
96  protected function use_regexp_non_capturing( $s ) {
97  // Same as above with a non-capturing subgroup.
98  $this->isutf8 = preg_match( '/^(?:[\x00-\x7f]|[\xc0-\xdf][\x80-\xbf]|' .
99  '[\xe0-\xef][\x80-\xbf]{2}|[\xf0-\xf7][\x80-\xbf]{3})+$/', $s );
100  }
101 
102  protected function use_regexp_once_only( $s ) {
103  // Same as above with a once-only subgroup.
104  $this->isutf8 = preg_match( '/^(?>[\x00-\x7f]|[\xc0-\xdf][\x80-\xbf]|' .
105  '[\xe0-\xef][\x80-\xbf]{2}|[\xf0-\xf7][\x80-\xbf]{3})+$/', $s );
106  }
107 
108  protected function use_mb_check_encoding( $s ) {
109  $this->isutf8 = mb_check_encoding( $s, 'UTF-8' );
110  }
111 }
112 
113 $maintClass = BenchUtf8TitleCheck::class;
114 require_once RUN_MAINTENANCE_IF_MAIN;
const RUN_MAINTENANCE_IF_MAIN
Definition: Maintenance.php:39
bench(array $benchs)
Definition: Benchmarker.php:50
This little benchmark executes the regexp formerly used in Language->checkTitleEncoding() and compare...
addDescription( $text)
Set the description text.
Base class for benchmark scripts.
Definition: Benchmarker.php:40