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', 'wikiquote', 'wikiquote.org' ),
  32                 'wikibooks' => new Site( 'wikibooks', 'wikibooks', '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         );
  55
  56         # Construct a list of reserved prefixes
  57         $reserved = array();
  58         foreach ( $langlist as $lang ) {
  59                 $reserved[$lang] = 1;
  60         }
  61         foreach ( $languageAliases as $alias => $lang ) {
  62                 $reserved[$alias] = 1;
  63         }
  64         foreach( $sites as $site ) {
  65                 $reserved[$site->lateral] = 1;
  66         }
  67
  68         # Extract the intermap from meta
  69         $dbr =& wfGetDB( DB_WRITE );
  70         $row = $dbr->getArray( "metawiki.cur", array( "cur_text" ),
  71                 array( "cur_namespace" => 0, "cur_title" => "Interwiki_map" ) );
  72
  73         if ( !$row ) {
  74                 die( "m:Interwiki_map not found" );
  75         }
  76
  77         $lines = explode( "\n", $row->cur_text );
  78         $iwArray = array();
  79
  80         foreach ( $lines as $line ) {
  81                 if ( preg_match( '/^\|\s*(.*?)\s*\|\|\s*(.*?)\s*$/', $line, $matches ) ) {
  82                         $prefix = strtolower( $matches[1] );
  83                         $url = $matches[2];
  84                         if ( preg_match( '/(wikipedia|wiktionary|wikisource|wikiquote|wikibooks)\.org/', $url ) ) {
  85                                 $local = 1;
  86                         } else {
  87                                 $local = 0;
  88                         }
  89
  90                         if ( empty( $reserved[$prefix] ) ) {
  91                                 $iwArray[] = array( "iw_prefix" => $prefix, "iw_url" => $url, "iw_local" => $local );
  92                         }
  93                 }
  94         }
  95
  96         $sql = "-- Generated by rebuildInterwiki.php";
  97
  98
  99         foreach ( $dblist as $db ) {
 100                 if ( isset( $specials[$db] ) ) {
 101                         # Special wiki
 102                         # Has interwiki links and interlanguage links to wikipedia
 103
 104                         $host = $specials[$db];
 105                         $sql .= "\n--$host\n\n";
 106                         $sql .= "USE $db;\n" .
 107                                         "TRUNCATE TABLE interwiki;\n" .
 108                                         "INSERT INTO interwiki (iw_prefix, iw_url, iw_local) VALUES \n";
 109                         $first = true;
 110
 111                         # Intermap links
 112                         foreach ( $iwArray as $iwEntry ) {
 113                                 $sql .= makeLink( $iwEntry, $first );
 114                         }
 115
 116                         # Links to multilanguage sites
 117                         foreach ( $sites as $targetSite ) {
 118                                 $sql .= makeLink( array( $targetSite->lateral, $targetSite->getURL( 'en' ), 1 ), $first );
 119                         }
 120
 121                         # Interlanguage links to wikipedia
 122                         $sql .= makeLanguageLinks( $sites['wiki'], $first );
 123
 124                         # Extra links
 125                         foreach ( $extraLinks as $link ) {
 126                                         $sql .= makeLink( $link, $first );
 127                         }
 128
 129                         $sql .= ";\n";
 130                 } else {
 131                         # Find out which site this DB belongs to
 132                         $site = false;
 133                         foreach( $sites as $candidateSite ) {
 134                                 $suffix = $candidateSite->suffix;
 135                                 if ( preg_match( "/(.*)$suffix$/", $db, $matches ) ) {
 136                                         $site = $candidateSite;
 137                                         break;
 138                                 }
 139                         }
 140                         if ( !$site ) {
 141                                 print "Invalid database $db\n";
 142                                 continue;
 143                         }
 144                         $lang = $matches[1];
 145                         $host = "$lang." . $site->url;
 146                         $sql .= "\n--$host\n\n";
 147
 148                         $sql .= "USE $db;\n" .
 149                                         "TRUNCATE TABLE interwiki;\n" .
 150                                         "INSERT INTO interwiki (iw_prefix,iw_url,iw_local) VALUES\n";
 151                         $first = true;
 152
 153                         # Intermap links
 154                         foreach ( $iwArray as $iwEntry ) {
 155                                 $sql .= makeLink( $iwEntry, $first );
 156                         }
 157
 158                         # Lateral links
 159                         foreach ( $sites as $targetSite ) {
 160                                 # Suppress link to self
 161                                 if ( $targetSite->suffix != $site->suffix ) {
 162                                         $sql .= makeLink( array( $targetSite->lateral, $targetSite->getURL( $lang ), 1 ), $first );
 163                                 }
 164                         }
 165
 166                         # Interlanguage links
 167                         $sql .= makeLanguageLinks( $site, $first );
 168
 169                         # w link within wikipedias
 170                         # Other sites already have it as a lateral link
 171                         if ( $site->suffix == "wiki" ) {
 172                                 $sql .= makeLink( array("w", "http://en.wikipedia.org/wiki/$1", 1), $first );
 173                         }
 174
 175                         # Extra links
 176                         foreach ( $extraLinks as $link ){
 177                                         $sql .= makeLink( $link, $first );
 178                         }
 179                         $sql .= ";\n\n";
 180                 }
 181         }
 182         return $sql;
 183 }
 184
 185 # ------------------------------------------------------------------------------------------
 186
 187 # Returns part of an INSERT statement, corresponding to all interlanguage links to a particular site
 188 function makeLanguageLinks( &$site, &$first ) {
 189         global $langlist, $languageAliases;
 190
 191         $sql = "";
 192
 193         # Actual languages with their own databases
 194         foreach ( $langlist as $targetLang ) {
 195                 $sql .= makeLink( array( $targetLang, $site->getURL( $targetLang ), 1 ), $first );
 196         }
 197
 198         # Language aliases
 199         foreach ( $languageAliases as $alias => $lang ) {
 200                 $sql .= makeLink( array( $alias, $site->getURL( $lang ), 1 ), $first );
 201         }
 202         return $sql;
 203 }
 204
 205 # Make SQL for a single link from an array
 206 function makeLink( $entry, &$first ) {
 207         $sql = "";
 208         # Add comma
 209         if ( $first ) {
 210                 $first = false;
 211         } else {
 212                 $sql .= ",\n";
 213         }
 214         $sql .= "(" . Database::makeList( $entry ) . ")";
 215         return $sql;
 216 }
 217
 218 ?>