MediaWiki  master
benchmarkParse.php
Go to the documentation of this file.
1 <?php
25 require_once __DIR__ . '/../Maintenance.php';
26 
31 
38 class BenchmarkParse extends Maintenance {
40  private $templateTimestamp = null;
41 
42  private $clearLinkCache = false;
43 
47  private $linkCache;
48 
50  private $idCache = [];
51 
52  public function __construct() {
53  parent::__construct();
54  $this->addDescription( 'Benchmark parse operation' );
55  $this->addArg( 'title', 'The name of the page to parse' );
56  $this->addOption( 'warmup', 'Repeat the parse operation this number of times to warm the cache',
57  false, true );
58  $this->addOption( 'loops', 'Number of times to repeat parse operation post-warmup',
59  false, true );
60  $this->addOption( 'page-time',
61  'Use the version of the page which was current at the given time',
62  false, true );
63  $this->addOption( 'tpl-time',
64  'Use templates which were current at the given time (except that moves and ' .
65  'deletes are not handled properly)',
66  false, true );
67  $this->addOption( 'reset-linkcache', 'Reset the LinkCache after every parse.',
68  false, false );
69  }
70 
71  public function execute() {
72  if ( $this->hasOption( 'tpl-time' ) ) {
73  $this->templateTimestamp = wfTimestamp( TS_MW, strtotime( $this->getOption( 'tpl-time' ) ) );
74  Hooks::register( 'BeforeParserFetchTemplateRevisionRecord', [ $this, 'onFetchTemplate' ] );
75  }
76 
77  $this->clearLinkCache = $this->hasOption( 'reset-linkcache' );
78  // Set as a member variable to avoid function calls when we're timing the parse
79  $this->linkCache = MediaWikiServices::getInstance()->getLinkCache();
80 
81  $title = Title::newFromText( $this->getArg( 0 ) );
82  if ( !$title ) {
83  $this->fatalError( "Invalid title" );
84  }
85 
86  $revLookup = MediaWikiServices::getInstance()->getRevisionLookup();
87  if ( $this->hasOption( 'page-time' ) ) {
88  $pageTimestamp = wfTimestamp( TS_MW, strtotime( $this->getOption( 'page-time' ) ) );
89  $id = $this->getRevIdForTime( $title, $pageTimestamp );
90  if ( !$id ) {
91  $this->fatalError( "The page did not exist at that time" );
92  }
93 
94  $revision = $revLookup->getRevisionById( $id );
95  } else {
96  $revision = $revLookup->getRevisionByTitle( $title );
97  }
98 
99  if ( !$revision ) {
100  $this->fatalError( "Unable to load revision, incorrect title?" );
101  }
102 
103  $warmup = $this->getOption( 'warmup', 1 );
104  for ( $i = 0; $i < $warmup; $i++ ) {
105  $this->runParser( $revision );
106  }
107 
108  $loops = $this->getOption( 'loops', 1 );
109  if ( $loops < 1 ) {
110  $this->fatalError( 'Invalid number of loops specified' );
111  }
112  $startUsage = getrusage();
113  $startTime = microtime( true );
114  for ( $i = 0; $i < $loops; $i++ ) {
115  $this->runParser( $revision );
116  }
117  $endUsage = getrusage();
118  $endTime = microtime( true );
119 
120  printf( "CPU time = %.3f s, wall clock time = %.3f s\n",
121  // CPU time
122  ( $endUsage['ru_utime.tv_sec'] + $endUsage['ru_utime.tv_usec'] * 1e-6
123  - $startUsage['ru_utime.tv_sec'] - $startUsage['ru_utime.tv_usec'] * 1e-6 ) / $loops,
124  // Wall clock time
125  ( $endTime - $startTime ) / $loops
126  );
127  }
128 
136  private function getRevIdForTime( Title $title, $timestamp ) {
137  $dbr = $this->getDB( DB_REPLICA );
138 
139  $id = $dbr->selectField(
140  [ 'revision', 'page' ],
141  'rev_id',
142  [
143  'page_namespace' => $title->getNamespace(),
144  'page_title' => $title->getDBkey(),
145  'rev_timestamp <= ' . $dbr->addQuotes( $timestamp )
146  ],
147  __METHOD__,
148  [ 'ORDER BY' => 'rev_timestamp DESC' ],
149  [ 'revision' => [ 'JOIN', 'rev_page=page_id' ] ]
150  );
151 
152  return $id;
153  }
154 
160  private function runParser( RevisionRecord $revision ) {
161  $content = $revision->getContent( SlotRecord::MAIN );
163 
164  $content->getParserOutput( $title, $revision->getId() );
165  if ( $this->clearLinkCache ) {
166  $this->linkCache->clear();
167  }
168  }
169 
180  private function onFetchTemplate(
181  ?LinkTarget $contextTitle,
182  LinkTarget $titleTarget,
183  bool &$skip,
184  ?RevisionRecord &$revRecord
185  ): bool {
186  $title = Title::castFromLinkTarget( $titleTarget );
187 
188  $pdbk = $title->getPrefixedDBkey();
189  if ( !isset( $this->idCache[$pdbk] ) ) {
190  $proposedId = $this->getRevIdForTime( $title, $this->templateTimestamp );
191  $this->idCache[$pdbk] = $proposedId;
192  }
193  if ( $this->idCache[$pdbk] !== false ) {
194  $revLookup = MediaWikiServices::getInstance()->getRevisionLookup();
195  $revRecord = $revLookup->getRevisionById( $this->idCache[$pdbk] );
196  }
197 
198  return true;
199  }
200 }
201 
202 $maintClass = BenchmarkParse::class;
203 require_once RUN_MAINTENANCE_IF_MAIN;
LinkCache
Cache for article titles (prefixed DB keys) and ids linked from one source.
Definition: LinkCache.php:40
MediaWiki\Revision\RevisionRecord\getContent
getContent( $role, $audience=self::FOR_PUBLIC, Authority $performer=null)
Returns the Content of the given slot of this revision.
Definition: RevisionRecord.php:156
Title\newFromText
static newFromText( $text, $defaultNamespace=NS_MAIN)
Create a new Title from text, such as what one would find in a link.
Definition: Title.php:383
MediaWiki\Revision\RevisionRecord
Page revision base class.
Definition: RevisionRecord.php:47
MediaWiki\MediaWikiServices
MediaWikiServices is the service locator for the application scope of MediaWiki.
Definition: MediaWikiServices.php:193
Maintenance\fatalError
fatalError( $msg, $exitCode=1)
Output a message and terminate the current script.
Definition: Maintenance.php:489
Maintenance\addDescription
addDescription( $text)
Set the description text.
Definition: Maintenance.php:329
wfTimestamp
wfTimestamp( $outputtype=TS_UNIX, $ts=0)
Get a timestamp string in one of various formats.
Definition: GlobalFunctions.php:1668
BenchmarkParse\onFetchTemplate
onFetchTemplate(?LinkTarget $contextTitle, LinkTarget $titleTarget, bool &$skip, ?RevisionRecord &$revRecord)
Hook into the parser's revision ID fetcher.
Definition: benchmarkParse.php:180
Maintenance
Abstract maintenance class for quickly writing and churning out maintenance scripts with minimal effo...
Definition: Maintenance.php:59
$dbr
$dbr
Definition: testCompression.php:54
BenchmarkParse
Maintenance script to benchmark how long it takes to parse a given title at an optionally specified t...
Definition: benchmarkParse.php:38
$maintClass
$maintClass
Definition: benchmarkParse.php:202
Maintenance\addOption
addOption( $name, $description, $required=false, $withArg=false, $shortName=false, $multiOccurrence=false)
Add a parameter to the script.
Definition: Maintenance.php:249
BenchmarkParse\$templateTimestamp
string $templateTimestamp
MediaWiki concatenated string timestamp (YYYYMMDDHHMMSS)
Definition: benchmarkParse.php:40
$title
$title
Definition: testCompression.php:38
DB_REPLICA
const DB_REPLICA
Definition: defines.php:25
BenchmarkParse\$linkCache
LinkCache $linkCache
Definition: benchmarkParse.php:47
$content
$content
Definition: router.php:76
Hooks\register
static register( $name, $callback)
Attach an event handler to a given hook in both legacy and non-legacy hook systems.
Definition: Hooks.php:48
MediaWiki\Revision\RevisionRecord\getPageAsLinkTarget
getPageAsLinkTarget()
Returns the title of the page this revision is associated with as a LinkTarget object.
Definition: RevisionRecord.php:355
Maintenance\getDB
getDB( $db, $groups=[], $dbDomain=false)
Returns a database to be used by current maintenance script.
Definition: Maintenance.php:1362
Title\newFromLinkTarget
static newFromLinkTarget(LinkTarget $linkTarget, $forceClone='')
Returns a Title given a LinkTarget.
Definition: Title.php:295
BenchmarkParse\execute
execute()
Do the actual work.
Definition: benchmarkParse.php:71
Title
Represents a title within MediaWiki.
Definition: Title.php:48
Maintenance\getOption
getOption( $name, $default=null)
Get an option, or return the default.
Definition: Maintenance.php:286
MediaWiki\Revision\RevisionRecord\getId
getId( $wikiId=self::LOCAL)
Get revision ID.
Definition: RevisionRecord.php:279
Maintenance\addArg
addArg( $arg, $description, $required=true)
Add some args that are needed.
Definition: Maintenance.php:300
BenchmarkParse\$clearLinkCache
$clearLinkCache
Definition: benchmarkParse.php:42
BenchmarkParse\__construct
__construct()
Default constructor.
Definition: benchmarkParse.php:52
Title\castFromLinkTarget
static castFromLinkTarget( $linkTarget)
Same as newFromLinkTarget, but if passed null, returns null.
Definition: Title.php:319
MediaWiki\Linker\LinkTarget
Definition: LinkTarget.php:26
Maintenance\hasOption
hasOption( $name)
Checks to see if a particular option was set.
Definition: Maintenance.php:271
Maintenance\getArg
getArg( $argId=0, $default=null)
Get an argument.
Definition: Maintenance.php:353
BenchmarkParse\runParser
runParser(RevisionRecord $revision)
Parse the text from a given RevisionRecord.
Definition: benchmarkParse.php:160
BenchmarkParse\$idCache
array $idCache
Cache that maps a Title DB key to revision ID for the requested timestamp.
Definition: benchmarkParse.php:50
MediaWiki\Revision\SlotRecord
Value object representing a content slot associated with a page revision.
Definition: SlotRecord.php:40
BenchmarkParse\getRevIdForTime
getRevIdForTime(Title $title, $timestamp)
Fetch the ID of the revision of a Title that occurred.
Definition: benchmarkParse.php:136