| Index: trunk/extensions/MobileFrontend/collectExtractStats.php |
| — | — | @@ -0,0 +1,73 @@ |
| | 2 | +<?php |
| | 3 | + |
| | 4 | +$IP = getenv( 'MW_INSTALL_PATH' ); |
| | 5 | +if ( $IP === false ) { |
| | 6 | + $IP = dirname( __FILE__ ) . '/../..'; |
| | 7 | +} |
| | 8 | +require_once( "$IP/maintenance/Maintenance.php" ); |
| | 9 | + |
| | 10 | +class CollectStats extends Maintenance { |
| | 11 | + public function __construct() { |
| | 12 | + parent::__construct(); |
| | 13 | + $this->mDescription = 'Developer script that calculates average full extract size'; |
| | 14 | + $this->addArg( 'rate', 'Check excerpt length for one page of this number', true ); |
| | 15 | + } |
| | 16 | + |
| | 17 | + public function execute() { |
| | 18 | + if ( !class_exists( 'ApiQueryExcerpt' ) ) { |
| | 19 | + $this->error( 'This script requires MobileFrontend to be properly installed', true ); |
| | 20 | + } |
| | 21 | + $rate = $this->getArg( 0 ); |
| | 22 | + $ns = array( NS_MAIN ); |
| | 23 | + $pageId = 0; |
| | 24 | + $dbr = $this->getDB( DB_SLAVE ); |
| | 25 | + $total = 0; |
| | 26 | + $calls = 0; |
| | 27 | + $html = 0; |
| | 28 | + $plain = 0; |
| | 29 | + do { |
| | 30 | + $res = $dbr->select( 'page', |
| | 31 | + array( 'page_id', 'page_namespace', 'page_title' ), |
| | 32 | + array( 'page_namespace' => $ns, 'page_is_redirect' => 0, "page_id > $pageId" ), |
| | 33 | + __METHOD__, |
| | 34 | + array( 'ORDER BY' => 'page_id', 'LIMIT' => 500 ) |
| | 35 | + ); |
| | 36 | + foreach ( $res as $row ) { |
| | 37 | + $pageId = $row->page_id; |
| | 38 | + if ( $total++ % $rate == 0 ) { |
| | 39 | + $title = Title::newFromRow( $row ); |
| | 40 | + $html += $this->getLength( $title, false ); |
| | 41 | + $plain += $this->getLength( $title, true ); |
| | 42 | + if ( ++$calls % 10 == 0 ) { |
| | 43 | + $this->output( "$calls\n" ); |
| | 44 | + } |
| | 45 | + } |
| | 46 | + } |
| | 47 | + } while( $res->numRows() > 0 ); |
| | 48 | + |
| | 49 | + $this->output( "Total pages processed: $calls\n" ); |
| | 50 | + if ( $calls > 0 ) { |
| | 51 | + $html /= $calls; |
| | 52 | + $plain /= $calls; |
| | 53 | + $this->output( " Average HTML length: $html\n Average plaintext length: $plain" ); |
| | 54 | + } |
| | 55 | + } |
| | 56 | + |
| | 57 | + private function getLength( Title $title, $plainText ) { |
| | 58 | + $params = array( |
| | 59 | + 'action' => 'query', |
| | 60 | + 'prop' => 'excerpt', |
| | 61 | + 'titles' => $title->getPrefixedText(), |
| | 62 | + ); |
| | 63 | + if ( $plainText ) { |
| | 64 | + $params['explaintext'] = 1; |
| | 65 | + } |
| | 66 | + $main = new ApiMain( new FauxRequest( $params ) ); |
| | 67 | + $main->execute(); |
| | 68 | + $data = $main->getResultData(); |
| | 69 | + return strlen( $data['query']['pages'][$title->getArticleID()]['excerpt'][0] ); |
| | 70 | + } |
| | 71 | +} |
| | 72 | + |
| | 73 | +$maintClass = 'CollectStats'; |
| | 74 | +require_once( DO_MAINTENANCE ); |
| \ No newline at end of file |
| Property changes on: trunk/extensions/MobileFrontend/collectExtractStats.php |
| ___________________________________________________________________ |
| Added: svn:eol-style |
| 1 | 75 | + native |