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
  13 /**
  14  * @todo document
  15  * @package MediaWiki
  16  * @subpackage Maintenance
  17  */
  18 class Site {
  19         var $suffix, $lateral, $url;
  20
  21         function Site( $s, $l, $u ) {
  22                 $this->suffix = $s;
  23                 $this->lateral = $l;
  24                 $this->url = $u;
  25         }
  26
  27         function getURL( $lang ) {
  28                 return "http://$lang.{$this->url}/wiki/\$1";
  29         }
  30 }
  31
  32 function getRebuildInterwikiSQL() {
  33         global $langlist, $languageAliases, $wgDBname;
  34
  35         # Initialise lists of wikis
  36         $sites = array(
  37                 'wiki' => new Site( 'wiki', 'w', 'wikipedia.org' ),
  38                 'wiktionary' => new Site( 'wiktionary', 'wikt', 'wiktionary.org' ),
  39                 'wikiquote' => new Site( 'wikiquote', 'q', 'wikiquote.org' ),
  40                 'wikibooks' => new Site( 'wikibooks', 'b', 'wikibooks.org' ),
  41                 'wikinews' => new Site( 'wikinews', 'n', 'wikinews.org' ),
  42         );
  43         $langlist = array_map( "trim", file( "/home/wikipedia/common/langlist" ) );
  44         $dblist = array_map( "trim", file( "/home/wikipedia/common/all.dblist" ) );
  45
  46         $specials = array(
  47                 'sourceswiki' => 'sources.wikipedia.org',
  48                 'quotewiki' => 'wikiquote.org',
  49                 'textbookwiki' => 'wikibooks.org',
  50                 'sep11wiki' => 'sep11.wikipedia.org',
  51                 'metawiki' => 'meta.wikimedia.org',
  52                 'commonswiki' => 'commons.wikimedia.org',
  53         );
  54
  55         $extraLinks = array(
  56                 array( 'm', 'http://meta.wikimedia.org/wiki/$1', 1 ),
  57                 array( 'meta', 'http://meta.wikimedia.org/wiki/$1', 1 ),
  58                 array( 'sep11', 'http://sep11.wikipedia.org/wiki/$1', 1 ),
  59         );
  60
  61         $languageAliases = array(
  62                 'zh-cn' => 'zh',
  63                 'zh-tw' => 'zh',
  64                 'dk' => 'da',
  65                 'nb' => 'no',
  66         );
  67
  68         # Construct a list of reserved prefixes
  69         $reserved = array();
  70         foreach ( $langlist as $lang ) {
  71                 $reserved[$lang] = 1;
  72         }
  73         foreach ( $languageAliases as $alias => $lang ) {
  74                 $reserved[$alias] = 1;
  75         }
  76         foreach( $sites as $site ) {
  77                 $reserved[$site->lateral] = 1;
  78         }
  79
  80         # Extract the intermap from meta
  81         $intermap = wfGetHTTP( 'http://meta.wikimedia.org/w/index.php?title=Interwiki_map&action=raw' );
  82         $lines = array_map( 'trim', explode( "\n", trim( $intermap ) ) );
  83
  84         if ( !$lines ) {
  85                 die( "m:Interwiki_map not found" );
  86         }
  87
  88         $iwArray = array();
  89
  90         foreach ( $lines as $line ) {
  91                 if ( preg_match( '/^\|\s*(.*?)\s*\|\|\s*(.*?)\s*$/', $line, $matches ) ) {
  92                         $prefix = strtolower( $matches[1] );
  93                         $url = $matches[2];
  94                         if ( preg_match( '/(wikipedia|wiktionary|wikisource|wikiquote|wikibooks)\.org/', $url ) ) {
  95                                 $local = 1;
  96                         } else {
  97                                 $local = 0;
  98                         }
  99
 100                         if ( empty( $reserved[$prefix] ) ) {
 101                                 $iwArray[$prefix] = array( "iw_prefix" => $prefix, "iw_url" => $url, "iw_local" => $local );
 102                         }
 103                 }
 104         }
 105
 106         $sql = "-- Generated by rebuildInterwiki.php";
 107
 108
 109         foreach ( $dblist as $db ) {
 110                 if ( isset( $specials[$db] ) ) {
 111                         # Special wiki
 112                         # Has interwiki links and interlanguage links to wikipedia
 113
 114                         $host = $specials[$db];
 115                         $sql .= "\n--$host\n\n";
 116                         $sql .= "USE $db;\n" .
 117                                         "TRUNCATE TABLE interwiki;\n" .
 118                                         "INSERT INTO interwiki (iw_prefix, iw_url, iw_local) VALUES \n";
 119                         $first = true;
 120
 121                         # Intermap links
 122                         foreach ( $iwArray as $iwEntry ) {
 123                                 $sql .= makeLink( $iwEntry, $first );
 124                         }
 125
 126                         # Links to multilanguage sites
 127                         foreach ( $sites as $targetSite ) {
 128                                 $sql .= makeLink( array( $targetSite->lateral, $targetSite->getURL( 'en' ), 1 ), $first );
 129                         }
 130
 131                         # Interlanguage links to wikipedia
 132                         $sql .= makeLanguageLinks( $sites['wiki'], $first );
 133
 134                         # Extra links
 135                         foreach ( $extraLinks as $link ) {
 136                                 $sql .= makeLink( $link, $first );
 137                         }
 138
 139                         $sql .= ";\n";
 140                 } else {
 141                         # Find out which site this DB belongs to
 142                         $site = false;
 143                         foreach( $sites as $candidateSite ) {
 144                                 $suffix = $candidateSite->suffix;
 145                                 if ( preg_match( "/(.*)$suffix$/", $db, $matches ) ) {
 146                                         $site = $candidateSite;
 147                                         break;
 148                                 }
 149                         }
 150                         if ( !$site ) {
 151                                 print "Invalid database $db\n";
 152                                 continue;
 153                         }
 154                         $lang = $matches[1];
 155                         $host = "$lang." . $site->url;
 156                         $sql .= "\n--$host\n\n";
 157
 158                         $sql .= "USE $db;\n" .
 159                                         "TRUNCATE TABLE interwiki;\n" .
 160                                         "INSERT INTO interwiki (iw_prefix,iw_url,iw_local) VALUES\n";
 161                         $first = true;
 162
 163                         # Intermap links
 164                         foreach ( $iwArray as $iwEntry ) {
 165                                 # Suppress links with the same name as the site
 166                                 if ( ( $suffix == 'wiki' && $iwEntry['iw_prefix'] != 'wikipedia' ) ||
 167                                   ( $suffix != 'wiki' && $suffix != $iwEntry['iw_prefix'] ) )
 168                                 {
 169                                         $sql .= makeLink( $iwEntry, $first );
 170                                 }
 171                         }
 172
 173                         # Lateral links
 174                         foreach ( $sites as $targetSite ) {
 175                                 # Suppress link to self
 176                                 if ( $targetSite->suffix != $site->suffix ) {
 177                                         $sql .= makeLink( array( $targetSite->lateral, $targetSite->getURL( $lang ), 1 ), $first );
 178                                 }
 179                         }
 180
 181                         # Interlanguage links
 182                         $sql .= makeLanguageLinks( $site, $first );
 183
 184                         # w link within wikipedias
 185                         # Other sites already have it as a lateral link
 186                         if ( $site->suffix == "wiki" ) {
 187                                 $sql .= makeLink( array("w", "http://en.wikipedia.org/wiki/$1", 1), $first );
 188                         }
 189
 190                         # Extra links
 191                         foreach ( $extraLinks as $link ){
 192                                         $sql .= makeLink( $link, $first );
 193                         }
 194                         $sql .= ";\n\n";
 195                 }
 196         }
 197         return $sql;
 198 }
 199
 200 # ------------------------------------------------------------------------------------------
 201
 202 # Returns part of an INSERT statement, corresponding to all interlanguage links to a particular site
 203 function makeLanguageLinks( &$site, &$first ) {
 204         global $langlist, $languageAliases;
 205
 206         $sql = "";
 207
 208         # Actual languages with their own databases
 209         foreach ( $langlist as $targetLang ) {
 210                 $sql .= makeLink( array( $targetLang, $site->getURL( $targetLang ), 1 ), $first );
 211         }
 212
 213         # Language aliases
 214         foreach ( $languageAliases as $alias => $lang ) {
 215                 $sql .= makeLink( array( $alias, $site->getURL( $lang ), 1 ), $first );
 216         }
 217         return $sql;
 218 }
 219
 220 # Make SQL for a single link from an array
 221 function makeLink( $entry, &$first ) {
 222         $sql = "";
 223         # Add comma
 224         if ( $first ) {
 225                 $first = false;
 226         } else {
 227                 $sql .= ",\n";
 228         }
 229         $dbr =& wfGetDB( DB_SLAVE );
 230         $sql .= "(" . $dbr->makeList( $entry ) . ")";
 231         return $sql;
 232 }
 233
 234 ?>