maintenance/dumpInterwiki.inc

   1 <?php
   2 /**
   3  * Rebuild interwiki table using the file on meta and the language list
   4  * Wikimedia specific!
   5  *
   6  * @file
   7  * @todo document
   8  * @ingroup Maintenance
   9  */
  10
  11 /**
  12  * @todo document
  13  * @ingroup Maintenance
  14  */
  15 class Site {
  16         var $suffix, $lateral, $url;
  17
  18         function __construct( $s, $l, $u ) {
  19                 $this->suffix = $s;
  20                 $this->lateral = $l;
  21                 $this->url = $u;
  22         }
  23
  24         function getURL( $lang ) {
  25                 $xlang = str_replace( '_', '-', $lang );
  26                 return "http://$xlang.{$this->url}/wiki/\$1";
  27         }
  28 }
  29
  30 function getRebuildInterwikiDump() {
  31         global $langlist, $languageAliases, $prefixRewrites;
  32
  33         # Multi-language sites
  34         # db suffix => db suffix, iw prefix, hostname
  35         $sites = array(
  36                 'wiki' => new Site( 'wiki', 'w', 'wikipedia.org' ),
  37                 'wiktionary' => new Site( 'wiktionary', 'wikt', 'wiktionary.org' ),
  38                 'wikiquote' => new Site( 'wikiquote', 'q', 'wikiquote.org' ),
  39                 'wikibooks' => new Site( 'wikibooks', 'b', 'wikibooks.org' ),
  40                 'wikinews' => new Site( 'wikinews', 'n', 'wikinews.org' ),
  41                 'wikisource' => new Site( 'wikisource', 's', 'wikisource.org' ),
  42                 'wikimedia' => new Site( 'wikimedia', 'chapter', 'wikimedia.org' ),
  43                 'wikiversity' => new Site( 'wikiversity', 'v', 'wikiversity.org' ),
  44         );
  45
  46         # List of language prefixes likely to be found in multi-language sites
  47         $langlist = array_map( "trim", file( "/home/wikipedia/common/langlist" ) );
  48
  49         # List of all database names
  50         $dblist = array_map( "trim", file( "/home/wikipedia/common/all.dblist" ) );
  51
  52         # Special-case databases
  53         $specials = array_flip(
  54                 array_map( "trim",
  55                         file( "/home/wikipedia/common/special.dblist" ) ) );
  56
  57         # Extra interwiki links that can't be in the intermap for some reason
  58         $extraLinks = array(
  59                 array( 'm', 'http://meta.wikimedia.org/wiki/$1', 1 ),
  60                 array( 'meta', 'http://meta.wikimedia.org/wiki/$1', 1 ),
  61                 array( 'sep11', 'http://sep11.wikipedia.org/wiki/$1', 1 ),
  62         );
  63
  64         # Language aliases, usually configured as redirects to the real wiki in apache
  65         # Interlanguage links are made directly to the real wiki
  66         # Something horrible happens if you forget to list an alias here, I can't
  67         #   remember what
  68         $languageAliases = array(
  69                 'zh-cn' => 'zh',
  70                 'zh-tw' => 'zh',
  71                 'dk' => 'da',
  72                 'nb' => 'no',
  73         );
  74
  75         # Special case prefix rewrites, for the benefit of Swedish which uses s:t
  76         # as an abbreviation for saint
  77         $prefixRewrites = array(
  78                 'svwiki' => array ( 's' => 'src'),
  79         );
  80
  81         # Construct a list of reserved prefixes
  82         $reserved = array();
  83         foreach ( $langlist as $lang ) {
  84                 $reserved[$lang] = 1;
  85         }
  86         foreach ( $languageAliases as $alias => $lang ) {
  87                 $reserved[$alias] = 1;
  88         }
  89         foreach( $sites as $site ) {
  90                 $reserved[$site->lateral] = 1;
  91         }
  92
  93         # Extract the intermap from meta
  94         $intermap = Http::get( 'http://meta.wikimedia.org/w/index.php?title=Interwiki_map&action=raw', 30 );
  95         $lines = array_map( 'trim', explode( "\n", trim( $intermap ) ) );
  96
  97         if ( !$lines || count( $lines ) < 2 ) {
  98                 wfDie( "m:Interwiki_map not found" );
  99         }
 100
 101         # Global iterwiki map
 102         foreach ( $lines as $line ) {
 103                 if ( preg_match( '/^\|\s*(.*?)\s*\|\|\s*(.*?)\s*$/', $line, $matches ) ) {
 104                         $prefix = strtolower( $matches[1] );
 105                         $url = $matches[2];
 106                         if ( preg_match( '/(wikipedia|wiktionary|wikisource|wikiquote|wikibooks|wikimedia)\.org/', $url ) ) {
 107                                 $local = 1;
 108                         } else {
 109                                 $local = 0;
 110                         }
 111
 112                         if ( empty( $reserved[$prefix] ) ) {
 113                             $imap  = array( "iw_prefix" => $prefix, "iw_url" => $url, "iw_local" => $local );
 114                             makeLink ($imap, "__global");
 115                         }
 116                 }
 117         }
 118
 119         # Exclude Wikipedia for Wikipedia
 120         makeLink ( array ('iw_prefix' => 'wikipedia', 'is_url' => null ), "_wiki" );
 121
 122         #Multilanguage sites
 123         foreach ($sites as $site)
 124             makeLanguageLinks ( $site, "_".$site->suffix );
 125
 126
 127         foreach ( $dblist as $db ) {
 128                 if ( isset( $specials[$db] ) ) {
 129                         # Special wiki
 130                         # Has interwiki links and interlanguage links to wikipedia
 131
 132                         makeLink( array( 'iw_prefix' => $db, 'iw_url' => "wiki"), "__sites" );
 133                         # Links to multilanguage sites
 134                         foreach ( $sites as $targetSite ) {
 135                                 makeLink( array( 'iw_prefix' => $targetSite->lateral,
 136                                                     'iw_url' =>$targetSite->getURL( 'en' ),
 137                                                     'iw_local' => 1 ), $db );
 138                         }
 139
 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                         makeLink( array( 'iw_prefix' => $db, 'iw_url' => $site->suffix), "__sites" );
 151                         if ( !$site ) {
 152                                 print "Invalid database $db\n";
 153                                 continue;
 154                         }
 155                         $lang = $matches[1];
 156
 157                         # Lateral links
 158                         foreach ( $sites as $targetSite ) {
 159                                 if ( $targetSite->suffix != $site->suffix ) {
 160                                         makeLink( array( 'iw_prefix' => $targetSite->lateral,
 161                                                             'iw_url' => $targetSite->getURL( $lang ),
 162                                                             'iw_local' => 1 ), $db );
 163                                 }
 164                         }
 165
 166                         if ( $site->suffix == "wiki" ) {
 167                                 makeLink( array('iw_prefix' => 'w',
 168                                                 'iw_url' => "http://en.wikipedia.org/wiki/$1",
 169                                                 'iw_local' => 1), $db );
 170                         }
 171
 172                 }
 173         }
 174         foreach ( $extraLinks as $link )
 175                 makeLink( $link, "__global" );
 176 }
 177
 178 # ------------------------------------------------------------------------------------------
 179
 180 # Executes part of an INSERT statement, corresponding to all interlanguage links to a particular site
 181 function makeLanguageLinks( &$site, $source ) {
 182         global $langlist, $languageAliases;
 183         # Actual languages with their own databases
 184         foreach ( $langlist as $targetLang ) {
 185                 makeLink( array( $targetLang, $site->getURL( $targetLang ), 1 ), $source );
 186         }
 187
 188         # Language aliases
 189         foreach ( $languageAliases as $alias => $lang ) {
 190                 makeLink( array( $alias, $site->getURL( $lang ), 1 ), $source );
 191         }
 192 }
 193
 194 function makeLink( $entry, $source ) {
 195         global $prefixRewrites, $dbFile;
 196         if ( isset( $prefixRewrites[$source] ) && isset( $prefixRewrites[$source][$entry[0]] ) )
 197                 $entry[0] = $prefixRewrites[$source][$entry[0]];
 198         if (!array_key_exists("iw_prefix",$entry))
 199             $entry = array("iw_prefix" => $entry[0], "iw_url" => $entry[1], "iw_local" => $entry[2]);
 200         if ( array_key_exists($source,$prefixRewrites) &&
 201                 array_key_exists($entry['iw_prefix'],$prefixRewrites[$source]))
 202                     $entry['iw_prefix'] = $prefixRewrites[$source][$entry['iw_prefix']];
 203         if ($dbFile)
 204             dba_insert("{$source}:{$entry['iw_prefix']}", trim("{$entry['iw_local']} {$entry['iw_url']}"),$dbFile);
 205         else
 206             print "{$source}:{$entry['iw_prefix']} {$entry['iw_url']} {$entry['iw_local']}\n";
 207
 208     }