MediaWiki
REL1_22
|
00001 <?php 00029 require_once __DIR__ . '/Maintenance.php'; 00030 00036 abstract class DumpIterator extends Maintenance { 00037 00038 private $count = 0; 00039 private $startTime; 00040 00041 public function __construct() { 00042 parent::__construct(); 00043 $this->mDescription = "Does something with a dump"; 00044 $this->addOption( 'file', 'File with text to run.', false, true ); 00045 $this->addOption( 'dump', 'XML dump to execute all revisions.', false, true ); 00046 $this->addOption( 'from', 'Article from XML dump to start from.', false, true ); 00047 } 00048 00049 public function execute() { 00050 if ( !( $this->hasOption( 'file' ) ^ $this->hasOption( 'dump' ) ) ) { 00051 $this->error( "You must provide a file or dump", true ); 00052 } 00053 00054 $this->checkOptions(); 00055 00056 if ( $this->hasOption( 'file' ) ) { 00057 $revision = new WikiRevision; 00058 00059 $revision->setText( file_get_contents( $this->getOption( 'file' ) ) ); 00060 $revision->setTitle( Title::newFromText( rawurldecode( basename( $this->getOption( 'file' ), '.txt' ) ) ) ); 00061 $this->handleRevision( $revision ); 00062 return; 00063 } 00064 00065 $this->startTime = microtime( true ); 00066 00067 if ( $this->getOption( 'dump' ) == '-' ) { 00068 $source = new ImportStreamSource( $this->getStdin() ); 00069 } else { 00070 $this->error( "Sorry, I don't support dump filenames yet. Use - and provide it on stdin on the meantime.", true ); 00071 } 00072 $importer = new WikiImporter( $source ); 00073 00074 $importer->setRevisionCallback( 00075 array( &$this, 'handleRevision' ) ); 00076 00077 $this->from = $this->getOption( 'from', null ); 00078 $this->count = 0; 00079 $importer->doImport(); 00080 00081 $this->conclusions(); 00082 00083 $delta = microtime( true ) - $this->startTime; 00084 $this->error( "Done {$this->count} revisions in " . round( $delta, 2 ) . " seconds " ); 00085 if ( $delta > 0 ) { 00086 $this->error( round( $this->count / $delta, 2 ) . " pages/sec" ); 00087 } 00088 00089 # Perform the memory_get_peak_usage() when all the other data has been output so there's no damage if it dies. 00090 # It is only available since 5.2.0 (since 5.2.1 if you haven't compiled with --enable-memory-limit) 00091 $this->error( "Memory peak usage of " . memory_get_peak_usage() . " bytes\n" ); 00092 } 00093 00094 public function finalSetup() { 00095 parent::finalSetup(); 00096 00097 if ( $this->getDbType() == Maintenance::DB_NONE ) { 00098 global $wgUseDatabaseMessages, $wgLocalisationCacheConf, $wgHooks; 00099 $wgUseDatabaseMessages = false; 00100 $wgLocalisationCacheConf['storeClass'] = 'LCStore_Null'; 00101 $wgHooks['InterwikiLoadPrefix'][] = 'DumpIterator::disableInterwikis'; 00102 } 00103 } 00104 00105 static function disableInterwikis( $prefix, &$data ) { 00106 # Title::newFromText will check on each namespaced article if it's an interwiki. 00107 # We always answer that it is not. 00108 00109 return false; 00110 } 00111 00117 public function handleRevision( $rev ) { 00118 $title = $rev->getTitle(); 00119 if ( !$title ) { 00120 $this->error( "Got bogus revision with null title!" ); 00121 return; 00122 } 00123 00124 $this->count++; 00125 if ( isset( $this->from ) ) { 00126 if ( $this->from != $title ) { 00127 return; 00128 } 00129 $this->output( "Skipped " . ( $this->count - 1 ) . " pages\n" ); 00130 00131 $this->count = 1; 00132 $this->from = null; 00133 } 00134 00135 $this->processRevision( $rev ); 00136 } 00137 00138 /* Stub function for processing additional options */ 00139 public function checkOptions() { 00140 return; 00141 } 00142 00143 /* Stub function for giving data about what was computed */ 00144 public function conclusions() { 00145 return; 00146 } 00147 00148 /* Core function which does whatever the maintenance script is designed to do */ 00149 abstract public function processRevision( $rev ); 00150 } 00151 00157 class SearchDump extends DumpIterator { 00158 00159 public function __construct() { 00160 parent::__construct(); 00161 $this->mDescription = "Runs a regex in the revisions from a dump"; 00162 $this->addOption( 'regex', 'Searching regex', true, true ); 00163 } 00164 00165 public function getDbType() { 00166 return Maintenance::DB_NONE; 00167 } 00168 00172 public function processRevision( $rev ) { 00173 if ( preg_match( $this->getOption( 'regex' ), $rev->getContent()->getTextForSearchIndex() ) ) { 00174 $this->output( $rev->getTitle() . " matches at edit from " . $rev->getTimestamp() . "\n" ); 00175 } 00176 } 00177 } 00178 00179 $maintClass = "SearchDump"; 00180 require_once RUN_MAINTENANCE_IF_MAIN;