maintenance/rebuildInterwiki.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  * @package MediaWiki
   8  * @subpackage Maintenance
   9  */
  10
  11 /** */
  12
  13 /**
  14  * @todo document
  15  * @package MediaWiki
  16  * @subpackage Maintenance
  17  */
  18 class Site {
  19         var $suffix, $lateral, $url;
  20
  21         function Site( $s, $l, $u ) {
  22                 $this->suffix = $s;
  23                 $this->lateral = $l;
  24                 $this->url = $u;
  25         }
  26
  27         function getURL( $lang ) {
  28                 return "http://$lang.{$this->url}/wiki/\$1";
  29         }
  30 }
  31
  32 function getRebuildInterwikiSQL() {
  33         global $langlist, $languageAliases, $prefixRewrites, $wgDBname;
  34
  35         # Multi-language sites
  36         # db suffix => db suffix, iw prefix, hostname
  37         $sites = array(
  38                 'wiki' => new Site( 'wiki', 'w', 'wikipedia.org' ),
  39                 'wiktionary' => new Site( 'wiktionary', 'wikt', 'wiktionary.org' ),
  40                 'wikiquote' => new Site( 'wikiquote', 'q', 'wikiquote.org' ),
  41                 'wikibooks' => new Site( 'wikibooks', 'b', 'wikibooks.org' ),
  42                 'wikinews' => new Site( 'wikinews', 'n', 'wikinews.org' ),
  43                 'wikisource' => new Site( 'wikisource', 's', 'wikisource.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 hostnames
  53         $specials = array(
  54                 'sourceswiki' => 'sources.wikipedia.org',
  55                 'quotewiki' => 'wikiquote.org',
  56                 'textbookwiki' => 'wikibooks.org',
  57                 'sep11wiki' => 'sep11.wikipedia.org',
  58                 'metawiki' => 'meta.wikimedia.org',
  59                 'commonswiki' => 'commons.wikimedia.org',
  60         );
  61
  62         # Extra interwiki links that can't be in the intermap for some reason
  63         $extraLinks = array(
  64                 array( 'm', 'http://meta.wikimedia.org/wiki/$1', 1 ),
  65                 array( 'meta', 'http://meta.wikimedia.org/wiki/$1', 1 ),
  66                 array( 'sep11', 'http://sep11.wikipedia.org/wiki/$1', 1 ),
  67         );
  68
  69         # Language aliases, usually configured as redirects to the real wiki in apache
  70         # Interlanguage links are made directly to the real wiki
  71         # Something horrible happens if you forget to list an alias here, I can't
  72         #   remember what
  73         $languageAliases = array(
  74                 'zh-cn' => 'zh',
  75                 'zh-tw' => 'zh',
  76                 'dk' => 'da',
  77                 'nb' => 'no',
  78         );
  79
  80         # Special case prefix rewrites, for the benefit of Swedish which uses s:t
  81         # as an abbreviation for saint
  82         $prefixRewrites = array(
  83                 'svwiki' => array( 's' => 'src' ),
  84         );
  85
  86         # Construct a list of reserved prefixes
  87         $reserved = array();
  88         foreach ( $langlist as $lang ) {
  89                 $reserved[$lang] = 1;
  90         }
  91         foreach ( $languageAliases as $alias => $lang ) {
  92                 $reserved[$alias] = 1;
  93         }
  94         foreach( $sites as $site ) {
  95                 $reserved[$site->lateral] = 1;
  96         }
  97
  98         # Extract the intermap from meta
  99         $intermap = wfGetHTTP( 'http://meta.wikimedia.org/w/index.php?title=Interwiki_map&action=raw', 30 );
 100         $lines = array_map( 'trim', explode( "\n", trim( $intermap ) ) );
 101
 102         if ( !$lines || count( $lines ) < 2 ) {
 103                 die( "m:Interwiki_map not found" );
 104         }
 105
 106         $iwArray = array();
 107
 108         foreach ( $lines as $line ) {
 109                 if ( preg_match( '/^\|\s*(.*?)\s*\|\|\s*(.*?)\s*$/', $line, $matches ) ) {
 110                         $prefix = strtolower( $matches[1] );
 111                         $url = $matches[2];
 112                         if ( preg_match( '/(wikipedia|wiktionary|wikisource|wikiquote|wikibooks)\.org/', $url ) ) {
 113                                 $local = 1;
 114                         } else {
 115                                 $local = 0;
 116                         }
 117
 118                         if ( empty( $reserved[$prefix] ) ) {
 119                                 $iwArray[$prefix] = array( "iw_prefix" => $prefix, "iw_url" => $url, "iw_local" => $local );
 120                         }
 121                 }
 122         }
 123
 124         $sql = "-- Generated by rebuildInterwiki.php";
 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                         $host = $specials[$db];
 133                         $sql .= "\n--$host\n\n";
 134                         $sql .= "USE $db;\n" .
 135                                         "TRUNCATE TABLE interwiki;\n" .
 136                                         "INSERT INTO interwiki (iw_prefix, iw_url, iw_local) VALUES \n";
 137                         $first = true;
 138
 139                         # Intermap links
 140                         foreach ( $iwArray as $iwEntry ) {
 141                                 $sql .= makeLink( $iwEntry, $first, $db );
 142                         }
 143
 144                         # Links to multilanguage sites
 145                         foreach ( $sites as $targetSite ) {
 146                                 $sql .= makeLink( array( $targetSite->lateral, $targetSite->getURL( 'en' ), 1 ), $first, $db );
 147                         }
 148
 149                         # Interlanguage links to wikipedia
 150                         $sql .= makeLanguageLinks( $sites['wiki'], $first, $db );
 151
 152                         # Extra links
 153                         foreach ( $extraLinks as $link ) {
 154                                 $sql .= makeLink( $link, $first, $db );
 155                         }
 156
 157                         $sql .= ";\n";
 158                 } else {
 159                         # Find out which site this DB belongs to
 160                         $site = false;
 161                         foreach( $sites as $candidateSite ) {
 162                                 $suffix = $candidateSite->suffix;
 163                                 if ( preg_match( "/(.*)$suffix$/", $db, $matches ) ) {
 164                                         $site = $candidateSite;
 165                                         break;
 166                                 }
 167                         }
 168                         if ( !$site ) {
 169                                 print "Invalid database $db\n";
 170                                 continue;
 171                         }
 172                         $lang = $matches[1];
 173                         $host = "$lang." . $site->url;
 174                         $sql .= "\n--$host\n\n";
 175
 176                         $sql .= "USE $db;\n" .
 177                                         "TRUNCATE TABLE interwiki;\n" .
 178                                         "INSERT INTO interwiki (iw_prefix,iw_url,iw_local) VALUES\n";
 179                         $first = true;
 180
 181                         # Intermap links
 182                         foreach ( $iwArray as $iwEntry ) {
 183                                 # Suppress links with the same name as the site
 184                                 if ( ( $suffix == 'wiki' && $iwEntry['iw_prefix'] != 'wikipedia' ) ||
 185                                   ( $suffix != 'wiki' && $suffix != $iwEntry['iw_prefix'] ) )
 186                                 {
 187                                         $sql .= makeLink( $iwEntry, $first, $db );
 188                                 }
 189                         }
 190
 191                         # Lateral links
 192                         foreach ( $sites as $targetSite ) {
 193                                 # Suppress link to self
 194                                 if ( $targetSite->suffix != $site->suffix ) {
 195                                         $sql .= makeLink( array( $targetSite->lateral, $targetSite->getURL( $lang ), 1 ), $first, $db );
 196                                 }
 197                         }
 198
 199                         # Interlanguage links
 200                         $sql .= makeLanguageLinks( $site, $first, $db );
 201
 202                         # w link within wikipedias
 203                         # Other sites already have it as a lateral link
 204                         if ( $site->suffix == "wiki" ) {
 205                                 $sql .= makeLink( array("w", "http://en.wikipedia.org/wiki/$1", 1), $first, $db );
 206                         }
 207
 208                         # Extra links
 209                         foreach ( $extraLinks as $link ){
 210                                         $sql .= makeLink( $link, $first, $db );
 211                         }
 212                         $sql .= ";\n\n";
 213                 }
 214         }
 215         return $sql;
 216 }
 217
 218 # ------------------------------------------------------------------------------------------
 219
 220 # Returns part of an INSERT statement, corresponding to all interlanguage links to a particular site
 221 function makeLanguageLinks( &$site, &$first, $source ) {
 222         global $langlist, $languageAliases;
 223
 224         $sql = "";
 225
 226         # Actual languages with their own databases
 227         foreach ( $langlist as $targetLang ) {
 228                 $sql .= makeLink( array( $targetLang, $site->getURL( $targetLang ), 1 ), $first, $source );
 229         }
 230
 231         # Language aliases
 232         foreach ( $languageAliases as $alias => $lang ) {
 233                 $sql .= makeLink( array( $alias, $site->getURL( $lang ), 1 ), $first, $source );
 234         }
 235         return $sql;
 236 }
 237
 238 # Make SQL for a single link from an array
 239 function makeLink( $entry, &$first, $source ) {
 240         global $prefixRewrites;
 241
 242         if ( isset( $prefixRewrites[$source] ) && isset( $prefixRewrites[$source][$entry[0]] ) ) {
 243                 $entry[0] = $prefixRewrites[$source][$entry[0]];
 244         }
 245
 246         $sql = "";
 247         # Add comma
 248         if ( $first ) {
 249                 $first = false;
 250         } else {
 251                 $sql .= ",\n";
 252         }
 253         $dbr =& wfGetDB( DB_SLAVE );
 254         $sql .= "(" . $dbr->makeList( $entry ) . ")";
 255         return $sql;
 256 }
 257
 258 ?>