maintenance/dumpCategoriesAsRdf.php

   1 <?php
   2 /**
   3  * This program is free software; you can redistribute it and/or modify
   4  * it under the terms of the GNU General Public License as published by
   5  * the Free Software Foundation; either version 2 of the License, or
   6  * (at your option) any later version.
   7  *
   8  * This program is distributed in the hope that it will be useful,
   9  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  10  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
  11  * GNU General Public License for more details.
  12  *
  13  * You should have received a copy of the GNU General Public License along
  14  * with this program; if not, write to the Free Software Foundation, Inc.,
  15  * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
  16  * http://www.gnu.org/copyleft/gpl.html
  17  *
  18  */
  19 use Wikimedia\Purtle\RdfWriter;
  20 use Wikimedia\Purtle\RdfWriterFactory;
  21 use Wikimedia\Rdbms\IDatabase;
  22
  23 require_once __DIR__ . '/Maintenance.php';
  24
  25 /**
  26  * Maintenance script to provide RDF representation of the category tree.
  27  *
  28  * @ingroup Maintenance
  29  * @since 1.30
  30  */
  31 class DumpCategoriesAsRdf extends Maintenance {
  32         /**
  33          * @var RdfWriter
  34          */
  35         private $rdfWriter;
  36         /**
  37          * Categories RDF helper.
  38          * @var CategoriesRdf
  39          */
  40         private $categoriesRdf;
  41
  42         public function __construct() {
  43                 parent::__construct();
  44
  45                 $this->addDescription( "Generate RDF dump of categories in a wiki." );
  46
  47                 $this->setBatchSize( 200 );
  48                 $this->addOption( 'output', "Output file (default is stdout). Will be overwritten.",
  49                         false, true );
  50                 $this->addOption( 'format', "Set the dump format.", false, true );
  51         }
  52
  53         /**
  54          * Produce row iterator for categories.
  55          * @param IDatabase $dbr Database connection
  56          * @return RecursiveIterator
  57          */
  58         public function getCategoryIterator( IDatabase $dbr ) {
  59                 $it = new BatchRowIterator(
  60                         $dbr,
  61                         'page',
  62                         [ 'page_title' ],
  63                         $this->mBatchSize
  64                 );
  65                 $it->addConditions( [
  66                         'page_namespace' => NS_CATEGORY,
  67                 ] );
  68                 $it->setFetchColumns( [ 'page_title', 'page_id' ] );
  69                 return $it;
  70         }
  71
  72         /**
  73          * Get iterator for links for categories.
  74          * @param IDatabase $dbr
  75          * @param array $ids List of page IDs
  76          * @return Traversable
  77          */
  78         public function getCategoryLinksIterator( IDatabase $dbr, array $ids ) {
  79                 $it = new BatchRowIterator(
  80                         $dbr,
  81                         'categorylinks',
  82                         [ 'cl_from', 'cl_to' ],
  83                         $this->mBatchSize
  84                 );
  85                 $it->addConditions( [
  86                         'cl_type' => 'subcat',
  87                         'cl_from' => $ids
  88                 ] );
  89                 $it->setFetchColumns( [ 'cl_from', 'cl_to' ] );
  90                 return new RecursiveIteratorIterator( $it );
  91         }
  92
  93         public function addDumpHeader( $timestamp ) {
  94                 global $wgRightsUrl;
  95                 $licenseUrl = $wgRightsUrl;
  96                 if ( substr( $licenseUrl, 0, 2 ) == '//' ) {
  97                         $licenseUrl = 'https:' . $licenseUrl;
  98                 }
  99                 $this->rdfWriter->about( wfExpandUrl( '/categoriesDump', PROTO_CANONICAL ) )
 100                         ->a( 'schema', 'Dataset' )
 101                         ->a( 'owl', 'Ontology' )
 102                         ->say( 'cc', 'license' )->is( $licenseUrl )
 103                         ->say( 'schema', 'softwareVersion' )->value( CategoriesRdf::FORMAT_VERSION )
 104                         ->say( 'schema', 'dateModified' )
 105                                 ->value( wfTimestamp( TS_ISO_8601, $timestamp ), 'xsd', 'dateTime' )
 106                         ->say( 'schema', 'isPartOf' )->is( wfExpandUrl( '/', PROTO_CANONICAL ) )
 107                         ->say( 'owl', 'imports' )->is( CategoriesRdf::OWL_URL );
 108         }
 109
 110         public function execute() {
 111                 $outFile = $this->getOption( 'output', 'php://stdout' );
 112
 113                 if ( $outFile === '-' ) {
 114                         $outFile = 'php://stdout';
 115                 }
 116
 117                 $output = fopen( $outFile, 'w' );
 118                 $this->rdfWriter = $this->createRdfWriter( $this->getOption( 'format', 'ttl' ) );
 119                 $this->categoriesRdf = new CategoriesRdf( $this->rdfWriter );
 120
 121                 $this->categoriesRdf->setupPrefixes();
 122                 $this->rdfWriter->start();
 123
 124                 $this->addDumpHeader( time() );
 125                 fwrite( $output, $this->rdfWriter->drain() );
 126
 127                 $dbr = $this->getDB( DB_REPLICA, [ 'vslow' ] );
 128
 129                 foreach ( $this->getCategoryIterator( $dbr ) as $batch ) {
 130                         $pages = [];
 131                         foreach ( $batch as $row ) {
 132                                 $this->categoriesRdf->writeCategoryData( $row->page_title );
 133                                 $pages[$row->page_id] = $row->page_title;
 134                         }
 135
 136                         foreach ( $this->getCategoryLinksIterator( $dbr, array_keys( $pages ) ) as $row ) {
 137                                 $this->categoriesRdf->writeCategoryLinkData( $pages[$row->cl_from], $row->cl_to );
 138                         }
 139                         fwrite( $output, $this->rdfWriter->drain() );
 140                 }
 141                 fflush( $output );
 142                 if ( $outFile !== '-' ) {
 143                         fclose( $output );
 144                 }
 145         }
 146
 147         /**
 148          * @param string $format Writer format
 149          * @return RdfWriter
 150          */
 151         private function createRdfWriter( $format ) {
 152                 $factory = new RdfWriterFactory();
 153                 return $factory->getWriter( $factory->getFormatName( $format ) );
 154         }
 155 }
 156
 157 $maintClass = "DumpCategoriesAsRdf";
 158 require_once RUN_MAINTENANCE_IF_MAIN;