maintenance/rebuildInterwiki.inc

   1 <?php
   2 /**
   3  * Rebuild interwiki table using the file on meta and the language list
   4  * Wikimedia specific!
   5  *
   6  * @todo document
   7  * @package MediaWiki
   8  * @subpackage Maintenance
   9  */
  10
  11 /** */
  12 $oldCwd = getcwd();
  13
  14 $optionsWithArgs = array( "o" );
  15 include_once( "commandLine.inc" );
  16
  17 /**
  18  * @todo document
  19  * @package MediaWiki
  20  * @subpackage Maintenance
  21  */
  22 class Site {
  23         var $suffix, $lateral, $url;
  24
  25         function Site( $s, $l, $u ) {
  26                 $this->suffix = $s;
  27                 $this->lateral = $l;
  28                 $this->url = $u;
  29         }
  30
  31         function getURL( $lang ) {
  32                 return "http://$lang.{$this->url}/wiki/\$1";
  33         }
  34 }
  35
  36 function getRebuildInterwikiSQL() {
  37         global $langlist, $languageAliases;
  38
  39         # Initialise lists of wikis
  40         $sites = array(
  41                 'wiki' => new Site( 'wiki', 'w', 'wikipedia.org' ),
  42                 'wiktionary' => new Site( 'wiktionary', 'wikt', 'wiktionary.org' ),
  43                 'wikiquote' => new Site( 'wikiquote', 'q', 'wikiquote.org' ),
  44                 'wikibooks' => new Site( 'wikibooks', 'b', 'wikibooks.org' )
  45         );
  46         $langlist = array_map( "trim", file( "/home/wikipedia/common/langlist" ) );
  47         $dblist = array_map( "trim", file( "/home/wikipedia/common/all.dblist" ) );
  48
  49         $specials = array(
  50                 'sourceswiki' => 'sources.wikipedia.org',
  51                 'quotewiki' => 'wikiquote.org',
  52                 'textbookwiki' => 'wikibooks.org',
  53                 'sep11wiki' => 'sep11.wikipedia.org',
  54                 'metawiki' => 'meta.wikimedia.org',
  55         );
  56
  57         $extraLinks = array(
  58                 array( 'm', 'http://meta.wikimedia.org/wiki/$1', 1 ),
  59                 array( 'meta', 'http://meta.wikimedia.org/wiki/$1', 1 ),
  60                 array( 'sep11', 'http://sep11.wikipedia.org/wiki/$1', 1 ),
  61         );
  62
  63         $languageAliases = array(
  64                 'zh-cn' => 'zh',
  65                 'zh-tw' => 'zh',
  66                 'dk' => 'da',
  67         );
  68
  69         # Construct a list of reserved prefixes
  70         $reserved = array();
  71         foreach ( $langlist as $lang ) {
  72                 $reserved[$lang] = 1;
  73         }
  74         foreach ( $languageAliases as $alias => $lang ) {
  75                 $reserved[$alias] = 1;
  76         }
  77         foreach( $sites as $site ) {
  78                 $reserved[$site->lateral] = 1;
  79         }
  80
  81         # Extract the intermap from meta
  82         $dbr =& wfGetDB( DB_WRITE );
  83         $row = $dbr->getArray( "metawiki.cur", array( "cur_text" ),
  84                 array( "cur_namespace" => 0, "cur_title" => "Interwiki_map" ) );
  85
  86         if ( !$row ) {
  87                 die( "m:Interwiki_map not found" );
  88         }
  89
  90         $lines = explode( "\n", $row->cur_text );
  91         $iwArray = array();
  92
  93         foreach ( $lines as $line ) {
  94                 if ( preg_match( '/^\|\s*(.*?)\s*\|\|\s*(.*?)\s*$/', $line, $matches ) ) {
  95                         $prefix = strtolower( $matches[1] );
  96                         $url = $matches[2];
  97                         if ( preg_match( '/(wikipedia|wiktionary|wikisource|wikiquote|wikibooks)\.org/', $url ) ) {
  98                                 $local = 1;
  99                         } else {
 100                                 $local = 0;
 101                         }
 102
 103                         if ( empty( $reserved[$prefix] ) ) {
 104                                 $iwArray[] = array( "iw_prefix" => $prefix, "iw_url" => $url, "iw_local" => $local );
 105                         }
 106                 }
 107         }
 108
 109         $sql = "-- Generated by rebuildInterwiki.php";
 110
 111
 112         foreach ( $dblist as $db ) {
 113                 if ( isset( $specials[$db] ) ) {
 114                         # Special wiki
 115                         # Has interwiki links and interlanguage links to wikipedia
 116
 117                         $host = $specials[$db];
 118                         $sql .= "\n--$host\n\n";
 119                         $sql .= "USE $db;\n" .
 120                                         "TRUNCATE TABLE interwiki;\n" .
 121                                         "INSERT INTO interwiki (iw_prefix, iw_url, iw_local) VALUES \n";
 122                         $first = true;
 123
 124                         # Intermap links
 125                         foreach ( $iwArray as $iwEntry ) {
 126                                 $sql .= makeLink( $iwEntry, $first );
 127                         }
 128
 129                         # Links to multilanguage sites
 130                         foreach ( $sites as $targetSite ) {
 131                                 $sql .= makeLink( array( $targetSite->lateral, $targetSite->getURL( 'en' ), 1 ), $first );
 132                         }
 133
 134                         # Interlanguage links to wikipedia
 135                         $sql .= makeLanguageLinks( $sites['wiki'], $first );
 136
 137                         # Extra links
 138                         foreach ( $extraLinks as $link ) {
 139                                 $sql .= makeLink( $link, $first );
 140                         }
 141
 142                         $sql .= ";\n";
 143                 } else {
 144                         # Find out which site this DB belongs to
 145                         $site = false;
 146                         foreach( $sites as $candidateSite ) {
 147                                 $suffix = $candidateSite->suffix;
 148                                 if ( preg_match( "/(.*)$suffix$/", $db, $matches ) ) {
 149                                         $site = $candidateSite;
 150                                         break;
 151                                 }
 152                         }
 153                         if ( !$site ) {
 154                                 print "Invalid database $db\n";
 155                                 continue;
 156                         }
 157                         $lang = $matches[1];
 158                         $host = "$lang." . $site->url;
 159                         $sql .= "\n--$host\n\n";
 160
 161                         $sql .= "USE $db;\n" .
 162                                         "TRUNCATE TABLE interwiki;\n" .
 163                                         "INSERT INTO interwiki (iw_prefix,iw_url,iw_local) VALUES\n";
 164                         $first = true;
 165
 166                         # Intermap links
 167                         foreach ( $iwArray as $iwEntry ) {
 168                                 # Suppress links with the same name as the site
 169                                 if ( ( $suffix == 'wiki' && $iwEntry['iw_prefix'] != 'wikipedia' ) ||
 170                                   ( $suffix != 'wiki' && $suffix != $iwEntry['iw_prefix'] ) )
 171                                 {
 172                                         $sql .= makeLink( $iwEntry, $first );
 173                                 }
 174                         }
 175
 176                         # Lateral links
 177                         foreach ( $sites as $targetSite ) {
 178                                 # Suppress link to self
 179                                 if ( $targetSite->suffix != $site->suffix ) {
 180                                         $sql .= makeLink( array( $targetSite->lateral, $targetSite->getURL( $lang ), 1 ), $first );
 181                                 }
 182                         }
 183
 184                         # Interlanguage links
 185                         $sql .= makeLanguageLinks( $site, $first );
 186
 187                         # w link within wikipedias
 188                         # Other sites already have it as a lateral link
 189                         if ( $site->suffix == "wiki" ) {
 190                                 $sql .= makeLink( array("w", "http://en.wikipedia.org/wiki/$1", 1), $first );
 191                         }
 192
 193                         # Extra links
 194                         foreach ( $extraLinks as $link ){
 195                                         $sql .= makeLink( $link, $first );
 196                         }
 197                         $sql .= ";\n\n";
 198                 }
 199         }
 200         return $sql;
 201 }
 202
 203 # ------------------------------------------------------------------------------------------
 204
 205 # Returns part of an INSERT statement, corresponding to all interlanguage links to a particular site
 206 function makeLanguageLinks( &$site, &$first ) {
 207         global $langlist, $languageAliases;
 208
 209         $sql = "";
 210
 211         # Actual languages with their own databases
 212         foreach ( $langlist as $targetLang ) {
 213                 $sql .= makeLink( array( $targetLang, $site->getURL( $targetLang ), 1 ), $first );
 214         }
 215
 216         # Language aliases
 217         foreach ( $languageAliases as $alias => $lang ) {
 218                 $sql .= makeLink( array( $alias, $site->getURL( $lang ), 1 ), $first );
 219         }
 220         return $sql;
 221 }
 222
 223 # Make SQL for a single link from an array
 224 function makeLink( $entry, &$first ) {
 225         $sql = "";
 226         # Add comma
 227         if ( $first ) {
 228                 $first = false;
 229         } else {
 230                 $sql .= ",\n";
 231         }
 232         $sql .= "(" . Database::makeList( $entry ) . ")";
 233         return $sql;
 234 }
 235
 236 ?>