maintenance/dumpInterwiki.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  * @addtogroup Maintenance
   8  */
   9
  10 /** */
  11
  12 /**
  13  * @todo document
  14  * @addtogroup Maintenance
  15  */
  16 class Site {
  17         var $suffix, $lateral, $url;
  18
  19         function __construct( $s, $l, $u ) {
  20                 $this->suffix = $s;
  21                 $this->lateral = $l;
  22                 $this->url = $u;
  23         }
  24
  25         function getURL( $lang ) {
  26                 $xlang = str_replace( '_', '-', $lang );
  27                 return "http://$xlang.{$this->url}/wiki/\$1";
  28         }
  29 }
  30
  31 function getRebuildInterwikiDump() {
  32         global $langlist, $languageAliases, $prefixRewrites;
  33
  34         # Multi-language sites
  35         # db suffix => db suffix, iw prefix, hostname
  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                 'wikisource' => new Site( 'wikisource', 's', 'wikisource.org' ),
  43                 'wikimedia' => new Site( 'wikimedia', 'chapter', 'wikimedia.org' ),
  44                 'wikiversity' => new Site( 'wikiversity', 'v', 'wikiversity.org' ),
  45         );
  46
  47         # List of language prefixes likely to be found in multi-language sites
  48         $langlist = array_map( "trim", file( "/home/wikipedia/common/langlist" ) );
  49
  50         # List of all database names
  51         $dblist = array_map( "trim", file( "/home/wikipedia/common/all.dblist" ) );
  52
  53         # Special-case databases
  54         $specials = array_flip(
  55                 array_map( "trim",
  56                         file( "/home/wikipedia/common/special.dblist" ) ) );
  57
  58         # Extra interwiki links that can't be in the intermap for some reason
  59         $extraLinks = array(
  60                 array( 'm', 'http://meta.wikimedia.org/wiki/$1', 1 ),
  61                 array( 'meta', 'http://meta.wikimedia.org/wiki/$1', 1 ),
  62                 array( 'sep11', 'http://sep11.wikipedia.org/wiki/$1', 1 ),
  63         );
  64
  65         # Language aliases, usually configured as redirects to the real wiki in apache
  66         # Interlanguage links are made directly to the real wiki
  67         # Something horrible happens if you forget to list an alias here, I can't
  68         #   remember what
  69         $languageAliases = array(
  70                 'zh-cn' => 'zh',
  71                 'zh-tw' => 'zh',
  72                 'dk' => 'da',
  73                 'nb' => 'no',
  74         );
  75
  76         # Special case prefix rewrites, for the benefit of Swedish which uses s:t
  77         # as an abbreviation for saint
  78         $prefixRewrites = array(
  79                 'svwiki' => array ( 's' => 'src'),
  80         );
  81
  82         # Construct a list of reserved prefixes
  83         $reserved = array();
  84         foreach ( $langlist as $lang ) {
  85                 $reserved[$lang] = 1;
  86         }
  87         foreach ( $languageAliases as $alias => $lang ) {
  88                 $reserved[$alias] = 1;
  89         }
  90         foreach( $sites as $site ) {
  91                 $reserved[$site->lateral] = 1;
  92         }
  93
  94         # Extract the intermap from meta
  95         $intermap = Http::get( 'http://meta.wikimedia.org/w/index.php?title=Interwiki_map&action=raw', 30 );
  96         $lines = array_map( 'trim', explode( "\n", trim( $intermap ) ) );
  97
  98         if ( !$lines || count( $lines ) < 2 ) {
  99                 wfDie( "m:Interwiki_map not found" );
 100         }
 101
 102         # Global iterwiki map
 103         foreach ( $lines as $line ) {
 104                 if ( preg_match( '/^\|\s*(.*?)\s*\|\|\s*(.*?)\s*$/', $line, $matches ) ) {
 105                         $prefix = strtolower( $matches[1] );
 106                         $url = $matches[2];
 107                         if ( preg_match( '/(wikipedia|wiktionary|wikisource|wikiquote|wikibooks|wikimedia)\.org/', $url ) ) {
 108                                 $local = 1;
 109                         } else {
 110                                 $local = 0;
 111                         }
 112
 113                         if ( empty( $reserved[$prefix] ) ) {
 114                             $imap  = array( "iw_prefix" => $prefix, "iw_url" => $url, "iw_local" => $local );
 115                             makeLink ($imap, "__global");
 116                         }
 117                 }
 118         }
 119
 120         # Exclude Wikipedia for Wikipedia
 121         makeLink ( array ('iw_prefix' => 'wikipedia', 'is_url' => null ), "_wiki" );
 122
 123         #Multilanguage sites
 124         foreach ($sites as $site)
 125             makeLanguageLinks ( $site, "_".$site->suffix );
 126
 127
 128         foreach ( $dblist as $db ) {
 129                 if ( isset( $specials[$db] ) ) {
 130                         # Special wiki
 131                         # Has interwiki links and interlanguage links to wikipedia
 132
 133                         makeLink( array( 'iw_prefix' => $db, 'iw_url' => "wiki"), "__sites" );
 134                         # Links to multilanguage sites
 135                         foreach ( $sites as $targetSite ) {
 136                                 makeLink( array( 'iw_prefix' => $targetSite->lateral,
 137                                                     'iw_url' =>$targetSite->getURL( 'en' ),
 138                                                     'iw_local' => 1 ), $db );
 139                         }
 140
 141                 } else {
 142                         # Find out which site this DB belongs to
 143                         $site = false;
 144                         foreach( $sites as $candidateSite ) {
 145                                 $suffix = $candidateSite->suffix;
 146                                 if ( preg_match( "/(.*)$suffix$/", $db, $matches ) ) {
 147                                 $site = $candidateSite;
 148                                         break;
 149                                 }
 150                         }
 151                         makeLink( array( 'iw_prefix' => $db, 'iw_url' => $site->suffix), "__sites" );
 152                         if ( !$site ) {
 153                                 print "Invalid database $db\n";
 154                                 continue;
 155                         }
 156                         $lang = $matches[1];
 157
 158                         # Lateral links
 159                         foreach ( $sites as $targetSite ) {
 160                                 if ( $targetSite->suffix != $site->suffix ) {
 161                                         makeLink( array( 'iw_prefix' => $targetSite->lateral,
 162                                                             'iw_url' => $targetSite->getURL( $lang ),
 163                                                             'iw_local' => 1 ), $db );
 164                                 }
 165                         }
 166
 167                         if ( $site->suffix == "wiki" ) {
 168                                 makeLink( array('iw_prefix' => 'w',
 169                                                 'iw_url' => "http://en.wikipedia.org/wiki/$1",
 170                                                 'iw_local' => 1), $db );
 171                         }
 172
 173                 }
 174         }
 175         foreach ( $extraLinks as $link )
 176                 makeLink( $link, "__global" );
 177 }
 178
 179 # ------------------------------------------------------------------------------------------
 180
 181 # Executes part of an INSERT statement, corresponding to all interlanguage links to a particular site
 182 function makeLanguageLinks( &$site, $source ) {
 183         global $langlist, $languageAliases;
 184         # Actual languages with their own databases
 185         foreach ( $langlist as $targetLang ) {
 186                 makeLink( array( $targetLang, $site->getURL( $targetLang ), 1 ), $source );
 187         }
 188
 189         # Language aliases
 190         foreach ( $languageAliases as $alias => $lang ) {
 191                 makeLink( array( $alias, $site->getURL( $lang ), 1 ), $source );
 192         }
 193 }
 194
 195 function makeLink( $entry, $source ) {
 196         global $prefixRewrites, $dbFile;
 197         if ( isset( $prefixRewrites[$source] ) && isset( $prefixRewrites[$source][$entry[0]] ) )
 198                 $entry[0] = $prefixRewrites[$source][$entry[0]];
 199         if (!array_key_exists("iw_prefix",$entry))
 200             $entry = array("iw_prefix" => $entry[0], "iw_url" => $entry[1], "iw_local" => $entry[2]);
 201         if ( array_key_exists($source,$prefixRewrites) &&
 202                 array_key_exists($entry['iw_prefix'],$prefixRewrites[$source]))
 203                     $entry['iw_prefix'] = $prefixRewrites[$source][$entry['iw_prefix']];
 204         if ($dbFile)
 205             dba_insert("{$source}:{$entry['iw_prefix']}", trim("{$entry['iw_local']} {$entry['iw_url']}"),$dbFile);
 206         else
 207             print "{$source}:{$entry['iw_prefix']} {$entry['iw_url']} {$entry['iw_local']}\n";
 208
 209     }
 210
 211 ?>