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