MediaWiki master
renderDump.php
Go to the documentation of this file.
1<?php
17use MediaWiki\Import\ImportStreamSource;
18use MediaWiki\Import\WikiRevision;
19use MediaWiki\Maintenance\Maintenance;
20use MediaWiki\Parser\ParserOptions;
21use MediaWiki\Permissions\UltimateAuthority;
22use MediaWiki\Revision\MutableRevisionRecord;
23use MediaWiki\User\User;
24
25// @codeCoverageIgnoreStart
26require_once __DIR__ . '/Maintenance.php';
27// @codeCoverageIgnoreEnd
28
35class DumpRenderer extends Maintenance {
36
38 private $count = 0;
39 private string $outputDirectory;
40 private float $startTime;
42 private $prefix;
43
44 public function __construct() {
45 parent::__construct();
46 $this->addDescription(
47 'Take page text out of an XML dump file and render basic HTML out to files' );
48 $this->addOption( 'output-dir', 'The directory to output the HTML files to', true, true );
49 $this->addOption( 'prefix', 'Prefix for the rendered files (defaults to wiki)', false, true );
50 $this->addOption( 'parser', 'Use an alternative parser class', false, true );
51 }
52
53 public function execute() {
54 $this->outputDirectory = $this->getOption( 'output-dir' );
55 $this->prefix = $this->getOption( 'prefix', 'wiki' );
56 $this->startTime = microtime( true );
57
58 if ( $this->hasOption( 'parser' ) ) {
59 $this->prefix .= '-' . $this->getOption( 'parser' );
60 // T236809: We'll need to provide an alternate ParserFactory
61 // service to make this work.
62 $this->fatalError( 'Parser class configuration temporarily disabled.' );
63 }
64
65 $user = User::newSystemUser( User::MAINTENANCE_SCRIPT_USER, [ 'steal' => true ] );
66
67 $source = new ImportStreamSource( $this->getStdin() );
68 $importer = $this->getServiceContainer()
69 ->getWikiImporterFactory()
70 ->getWikiImporter( $source, new UltimateAuthority( $user ) );
71
72 $importer->setRevisionCallback(
73 $this->handleRevision( ... ) );
74 $importer->setNoticeCallback( static function ( $msg, $params ) {
75 echo wfMessage( $msg, $params )->text() . "\n";
76 } );
77
78 $importer->doImport();
79
80 $delta = microtime( true ) - $this->startTime;
81 $this->error( "Rendered {$this->count} pages in " . round( $delta, 2 ) . " seconds " );
82 if ( $delta > 0 ) {
83 $this->error( round( $this->count / $delta, 2 ) . " pages/sec" );
84 }
85 $this->error( "\n" );
86 }
87
91 public function handleRevision( WikiRevision $rev ) {
92 $title = $rev->getTitle();
93 if ( !$title ) {
94 $this->error( "Got bogus revision with null title!" );
95
96 return;
97 }
98 $display = $title->getPrefixedText();
99
100 $this->count++;
101
102 $sanitized = rawurlencode( $display );
103 $filename = sprintf( "%s/%s-%07d-%s.html",
104 $this->outputDirectory,
105 $this->prefix,
106 $this->count,
107 $sanitized );
108 $this->output( sprintf( "%s\t%s\n", $filename, $display ) );
109
110 $user = new User();
111 $options = ParserOptions::newFromUser( $user );
112
113 $content = $rev->getContent();
114 $contentRenderer = $this->getServiceContainer()->getContentRenderer();
115 // ContentRenderer expects a RevisionRecord, and all we have is a
116 // WikiRevision from the dump. Make a fake MutableRevisionRecord to
117 // satisfy it -- the only thing ::getParserOutput actually needs is
118 // the revision ID and revision timestamp.
119 $mutableRev = new MutableRevisionRecord( $rev->getTitle() );
120 $mutableRev->setId( $rev->getID() );
121 $mutableRev->setTimestamp( $rev->getTimestamp() );
122 $output = $contentRenderer->getParserOutput(
123 $content, $title, $mutableRev, $options
124 );
125
126 file_put_contents( $filename,
127 "<!DOCTYPE html>\n" .
128 "<html lang=\"en\" dir=\"ltr\">\n" .
129 "<head>\n" .
130 "<meta charset=\"UTF-8\" />\n" .
131 "<meta name=\"color-scheme\" content=\"light dark\">" .
132 "<title>" . htmlspecialchars( $display, ENT_COMPAT ) . "</title>\n" .
133 "</head>\n" .
134 "<body>\n" .
135 // TODO T371004 move runOutputPipeline out of $parserOutput
136 $output->runOutputPipeline( $options, [] )->getContentHolderText() .
137 "</body>\n" .
138 "</html>" );
139 }
140}
141
142// @codeCoverageIgnoreStart
143$maintClass = DumpRenderer::class;
144require_once RUN_MAINTENANCE_IF_MAIN;
145// @codeCoverageIgnoreEnd
wfMessage
wfMessage( $key,... $params)
This is the function for getting translated interface messages.
Definition GlobalFunctions.php:602
DumpRenderer
Maintenance script that takes page text out of an XML dump file and render basic HTML out to files.
Definition renderDump.php:35
DumpRenderer\handleRevision
handleRevision(WikiRevision $rev)
Callback function for each revision, turn into HTML and save.
Definition renderDump.php:91
DumpRenderer\__construct
__construct()
Default constructor.
Definition renderDump.php:44
DumpRenderer\execute
execute()
Do the actual work.
Definition renderDump.php:53
MediaWiki\Import\ImportStreamSource
Imports a XML dump from a file (either from file upload, files on disk, or HTTP)
Definition ImportStreamSource.php:23
MediaWiki\Import\WikiRevision
Represents a revision, log entry or upload during the import process.
Definition WikiRevision.php:35
MediaWiki\Import\WikiRevision\getContent
getContent( $role=SlotRecord::MAIN)
Definition WikiRevision.php:460
MediaWiki\Maintenance\Maintenance
Abstract maintenance class for quickly writing and churning out maintenance scripts with minimal effo...
Definition Maintenance.php:65
MediaWiki\Maintenance\Maintenance\output
output( $out, $channel=null)
Throw some output to the user.
Definition Maintenance.php:486
MediaWiki\Maintenance\Maintenance\fatalError
fatalError( $msg, $exitCode=1)
Output a message and terminate the current script.
Definition Maintenance.php:551
MediaWiki\Maintenance\Maintenance\addOption
addOption( $name, $description, $required=false, $withArg=false, $shortName=false, $multiOccurrence=false)
Add a parameter to the script.
Definition Maintenance.php:266
MediaWiki\Maintenance\Maintenance\hasOption
hasOption( $name)
Checks to see if a particular option was set.
Definition Maintenance.php:285
MediaWiki\Maintenance\Maintenance\getOption
getOption( $name, $default=null)
Get an option, or return the default.
Definition Maintenance.php:300
MediaWiki\Maintenance\Maintenance\error
error( $err, $die=0)
Throw an error to the user.
Definition Maintenance.php:513
MediaWiki\Maintenance\Maintenance\getServiceContainer
getServiceContainer()
Returns the main service container.
Definition Maintenance.php:680
MediaWiki\Maintenance\Maintenance\getStdin
getStdin( $len=null)
Return input from stdin.
Definition Maintenance.php:458
MediaWiki\Maintenance\Maintenance\addDescription
addDescription( $text)
Set the description text.
Definition Maintenance.php:336
MediaWiki\Parser\ParserOptions
Set options of the Parser.
Definition ParserOptions.php:70
MediaWiki\Permissions\UltimateAuthority
Represents an authority that has all permissions.
Definition UltimateAuthority.php:22
MediaWiki\User\User
User class for the MediaWiki software.
Definition User.php:130
$source
$source
Definition mwdoc-filter.php:36
$maintClass
$maintClass
Definition renderDump.php:143