00001 <?php
00016 class Site {
00017 var $suffix, $lateral, $url;
00018
00019 function __construct( $s, $l, $u ) {
00020 $this->suffix = $s;
00021 $this->lateral = $l;
00022 $this->url = $u;
00023 }
00024
00025 function getURL( $lang ) {
00026 $xlang = str_replace( '_', '-', $lang );
00027 return "http://$xlang.{$this->url}/wiki/\$1";
00028 }
00029 }
00030
00031 function getRebuildInterwikiDump() {
00032 global $langlist, $languageAliases, $prefixRewrites;
00033
00034 # Multi-language sites
00035 # db suffix => db suffix, iw prefix, hostname
00036 $sites = array(
00037 'wiki' => new Site( 'wiki', 'w', 'wikipedia.org' ),
00038 'wiktionary' => new Site( 'wiktionary', 'wikt', 'wiktionary.org' ),
00039 'wikiquote' => new Site( 'wikiquote', 'q', 'wikiquote.org' ),
00040 'wikibooks' => new Site( 'wikibooks', 'b', 'wikibooks.org' ),
00041 'wikinews' => new Site( 'wikinews', 'n', 'wikinews.org' ),
00042 'wikisource' => new Site( 'wikisource', 's', 'wikisource.org' ),
00043 'wikimedia' => new Site( 'wikimedia', 'chapter', 'wikimedia.org' ),
00044 'wikiversity' => new Site( 'wikiversity', 'v', 'wikiversity.org' ),
00045 );
00046
00047 # List of language prefixes likely to be found in multi-language sites
00048 $langlist = array_map( "trim", file( "/home/wikipedia/common/langlist" ) );
00049
00050 # List of all database names
00051 $dblist = array_map( "trim", file( "/home/wikipedia/common/all.dblist" ) );
00052
00053 # Special-case databases
00054 $specials = array_flip(
00055 array_map( "trim",
00056 file( "/home/wikipedia/common/special.dblist" ) ) );
00057
00058 # Extra interwiki links that can't be in the intermap for some reason
00059 $extraLinks = array(
00060 array( 'm', 'http://meta.wikimedia.org/wiki/$1', 1 ),
00061 array( 'meta', 'http://meta.wikimedia.org/wiki/$1', 1 ),
00062 array( 'sep11', 'http://sep11.wikipedia.org/wiki/$1', 1 ),
00063 );
00064
00065 # Language aliases, usually configured as redirects to the real wiki in apache
00066 # Interlanguage links are made directly to the real wiki
00067 # Something horrible happens if you forget to list an alias here, I can't
00068 # remember what
00069 $languageAliases = array(
00070 'zh-cn' => 'zh',
00071 'zh-tw' => 'zh',
00072 'dk' => 'da',
00073 'nb' => 'no',
00074 );
00075
00076 # Special case prefix rewrites, for the benefit of Swedish which uses s:t
00077 # as an abbreviation for saint
00078 $prefixRewrites = array(
00079 'svwiki' => array ( 's' => 'src'),
00080 );
00081
00082 # Construct a list of reserved prefixes
00083 $reserved = array();
00084 foreach ( $langlist as $lang ) {
00085 $reserved[$lang] = 1;
00086 }
00087 foreach ( $languageAliases as $alias => $lang ) {
00088 $reserved[$alias] = 1;
00089 }
00090 foreach( $sites as $site ) {
00091 $reserved[$site->lateral] = 1;
00092 }
00093
00094 # Extract the intermap from meta
00095 $intermap = Http::get( 'http://meta.wikimedia.org/w/index.php?title=Interwiki_map&action=raw', 30 );
00096 $lines = array_map( 'trim', explode( "\n", trim( $intermap ) ) );
00097
00098 if ( !$lines || count( $lines ) < 2 ) {
00099 wfDie( "m:Interwiki_map not found" );
00100 }
00101
00102 # Global iterwiki map
00103 foreach ( $lines as $line ) {
00104 if ( preg_match( '/^\|\s*(.*?)\s*\|\|\s*(.*?)\s*$/', $line, $matches ) ) {
00105 $prefix = strtolower( $matches[1] );
00106 $url = $matches[2];
00107 if ( preg_match( '/(wikipedia|wiktionary|wikisource|wikiquote|wikibooks|wikimedia)\.org/', $url ) ) {
00108 $local = 1;
00109 } else {
00110 $local = 0;
00111 }
00112
00113 if ( empty( $reserved[$prefix] ) ) {
00114 $imap = array( "iw_prefix" => $prefix, "iw_url" => $url, "iw_local" => $local );
00115 makeLink ($imap, "__global");
00116 }
00117 }
00118 }
00119
00120 # Exclude Wikipedia for Wikipedia
00121 makeLink ( array ('iw_prefix' => 'wikipedia', 'is_url' => null ), "_wiki" );
00122
00123 #Multilanguage sites
00124 foreach ($sites as $site)
00125 makeLanguageLinks ( $site, "_".$site->suffix );
00126
00127
00128 foreach ( $dblist as $db ) {
00129 if ( isset( $specials[$db] ) ) {
00130 # Special wiki
00131 # Has interwiki links and interlanguage links to wikipedia
00132
00133 makeLink( array( 'iw_prefix' => $db, 'iw_url' => "wiki"), "__sites" );
00134 # Links to multilanguage sites
00135 foreach ( $sites as $targetSite ) {
00136 makeLink( array( 'iw_prefix' => $targetSite->lateral,
00137 'iw_url' =>$targetSite->getURL( 'en' ),
00138 'iw_local' => 1 ), $db );
00139 }
00140
00141 } else {
00142 # Find out which site this DB belongs to
00143 $site = false;
00144 foreach( $sites as $candidateSite ) {
00145 $suffix = $candidateSite->suffix;
00146 if ( preg_match( "/(.*)$suffix$/", $db, $matches ) ) {
00147 $site = $candidateSite;
00148 break;
00149 }
00150 }
00151 makeLink( array( 'iw_prefix' => $db, 'iw_url' => $site->suffix), "__sites" );
00152 if ( !$site ) {
00153 print "Invalid database $db\n";
00154 continue;
00155 }
00156 $lang = $matches[1];
00157
00158 # Lateral links
00159 foreach ( $sites as $targetSite ) {
00160 if ( $targetSite->suffix != $site->suffix ) {
00161 makeLink( array( 'iw_prefix' => $targetSite->lateral,
00162 'iw_url' => $targetSite->getURL( $lang ),
00163 'iw_local' => 1 ), $db );
00164 }
00165 }
00166
00167 if ( $site->suffix == "wiki" ) {
00168 makeLink( array('iw_prefix' => 'w',
00169 'iw_url' => "http://en.wikipedia.org/wiki/$1",
00170 'iw_local' => 1), $db );
00171 }
00172
00173 }
00174 }
00175 foreach ( $extraLinks as $link )
00176 makeLink( $link, "__global" );
00177 }
00178
00179 # ------------------------------------------------------------------------------------------
00180
00181 # Executes part of an INSERT statement, corresponding to all interlanguage links to a particular site
00182 function makeLanguageLinks( &$site, $source ) {
00183 global $langlist, $languageAliases;
00184 # Actual languages with their own databases
00185 foreach ( $langlist as $targetLang ) {
00186 makeLink( array( $targetLang, $site->getURL( $targetLang ), 1 ), $source );
00187 }
00188
00189 # Language aliases
00190 foreach ( $languageAliases as $alias => $lang ) {
00191 makeLink( array( $alias, $site->getURL( $lang ), 1 ), $source );
00192 }
00193 }
00194
00195 function makeLink( $entry, $source ) {
00196 global $prefixRewrites, $dbFile;
00197 if ( isset( $prefixRewrites[$source] ) && isset( $prefixRewrites[$source][$entry[0]] ) )
00198 $entry[0] = $prefixRewrites[$source][$entry[0]];
00199 if (!array_key_exists("iw_prefix",$entry))
00200 $entry = array("iw_prefix" => $entry[0], "iw_url" => $entry[1], "iw_local" => $entry[2]);
00201 if ( array_key_exists($source,$prefixRewrites) &&
00202 array_key_exists($entry['iw_prefix'],$prefixRewrites[$source]))
00203 $entry['iw_prefix'] = $prefixRewrites[$source][$entry['iw_prefix']];
00204 if ($dbFile)
00205 $dbFile->set( "{$source}:{$entry['iw_prefix']}", trim("{$entry['iw_local']} {$entry['iw_url']}") );
00206 else
00207 print "{$source}:{$entry['iw_prefix']} {$entry['iw_url']} {$entry['iw_local']}\n";
00208
00209 }