maintenance/rebuildInterwiki.inc

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