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                 $xlang = str_replace( '_', '-', $lang );
  29                 return "http://$xlang.{$this->url}/wiki/\$1";
  30         }
  31 }
  32
  33 function getRebuildInterwikiSQL() {
  34         global $langlist, $languageAliases, $prefixRewrites, $wgDBname;
  35
  36         # Multi-language sites
  37         # db suffix => db suffix, iw prefix, hostname
  38         $sites = array(
  39                 'wiki' => new Site( 'wiki', 'w', 'wikipedia.org' ),
  40                 'wiktionary' => new Site( 'wiktionary', 'wikt', 'wiktionary.org' ),
  41                 'wikiquote' => new Site( 'wikiquote', 'q', 'wikiquote.org' ),
  42                 'wikibooks' => new Site( 'wikibooks', 'b', 'wikibooks.org' ),
  43                 'wikinews' => new Site( 'wikinews', 'n', 'wikinews.org' ),
  44                 'wikisource' => new Site( 'wikisource', 's', 'wikisource.org' ),
  45                 'wikimedia' => new Site( 'wikimedia', 'chapter', 'wikimedia.org' ),
  46         );
  47
  48         # List of language prefixes likely to be found in multi-language sites
  49         $langlist = array_map( "trim", file( "/home/wikipedia/common/langlist" ) );
  50
  51         # List of all database names
  52         $dblist = array_map( "trim", file( "/home/wikipedia/common/all.dblist" ) );
  53
  54         # Special-case hostnames
  55         $specials = array(
  56                 'sourceswiki' => 'sources.wikipedia.org',
  57                 'quotewiki' => 'wikiquote.org',
  58                 'textbookwiki' => 'wikibooks.org',
  59                 'sep11wiki' => 'sep11.wikipedia.org',
  60                 'metawiki' => 'meta.wikimedia.org',
  61                 'commonswiki' => 'commons.wikimedia.org',
  62         );
  63
  64         # Extra interwiki links that can't be in the intermap for some reason
  65         $extraLinks = array(
  66                 array( 'm', 'http://meta.wikimedia.org/wiki/$1', 1 ),
  67                 array( 'meta', 'http://meta.wikimedia.org/wiki/$1', 1 ),
  68                 array( 'sep11', 'http://sep11.wikipedia.org/wiki/$1', 1 ),
  69         );
  70
  71         # Language aliases, usually configured as redirects to the real wiki in apache
  72         # Interlanguage links are made directly to the real wiki
  73         # Something horrible happens if you forget to list an alias here, I can't
  74         #   remember what
  75         $languageAliases = array(
  76                 'zh-cn' => 'zh',
  77                 'zh-tw' => 'zh',
  78                 'dk' => 'da',
  79                 'nb' => 'no',
  80         );
  81
  82         # Special case prefix rewrites, for the benefit of Swedish which uses s:t
  83         # as an abbreviation for saint
  84         $prefixRewrites = array(
  85                 'svwiki' => array( 's' => 'src' ),
  86         );
  87
  88         # Construct a list of reserved prefixes
  89         $reserved = array();
  90         foreach ( $langlist as $lang ) {
  91                 $reserved[$lang] = 1;
  92         }
  93         foreach ( $languageAliases as $alias => $lang ) {
  94                 $reserved[$alias] = 1;
  95         }
  96         foreach( $sites as $site ) {
  97                 $reserved[$site->lateral] = 1;
  98         }
  99
 100         # Extract the intermap from meta
 101         $intermap = wfGetHTTP( 'http://meta.wikimedia.org/w/index.php?title=Interwiki_map&action=raw', 30 );
 102         $lines = array_map( 'trim', explode( "\n", trim( $intermap ) ) );
 103
 104         if ( !$lines || count( $lines ) < 2 ) {
 105                 die( "m:Interwiki_map not found" );
 106         }
 107
 108         $iwArray = array();
 109
 110         foreach ( $lines as $line ) {
 111                 if ( preg_match( '/^\|\s*(.*?)\s*\|\|\s*(.*?)\s*$/', $line, $matches ) ) {
 112                         $prefix = strtolower( $matches[1] );
 113                         $url = $matches[2];
 114                         if ( preg_match( '/(wikipedia|wiktionary|wikisource|wikiquote|wikibooks)\.org/', $url ) ) {
 115                                 $local = 1;
 116                         } else {
 117                                 $local = 0;
 118                         }
 119
 120                         if ( empty( $reserved[$prefix] ) ) {
 121                                 $iwArray[$prefix] = array( "iw_prefix" => $prefix, "iw_url" => $url, "iw_local" => $local );
 122                         }
 123                 }
 124         }
 125
 126         $sql = "-- Generated by rebuildInterwiki.php";
 127
 128
 129         foreach ( $dblist as $db ) {
 130                 if ( isset( $specials[$db] ) ) {
 131                         # Special wiki
 132                         # Has interwiki links and interlanguage links to wikipedia
 133
 134                         $host = $specials[$db];
 135                         $sql .= "\n--$host\n\n";
 136                         $sql .= "USE $db;\n" .
 137                                         "TRUNCATE TABLE interwiki;\n" .
 138                                         "INSERT INTO interwiki (iw_prefix, iw_url, iw_local) VALUES \n";
 139                         $first = true;
 140
 141                         # Intermap links
 142                         foreach ( $iwArray as $iwEntry ) {
 143                                 $sql .= makeLink( $iwEntry, $first, $db );
 144                         }
 145
 146                         # Links to multilanguage sites
 147                         foreach ( $sites as $targetSite ) {
 148                                 $sql .= makeLink( array( $targetSite->lateral, $targetSite->getURL( 'en' ), 1 ), $first, $db );
 149                         }
 150
 151                         # Interlanguage links to wikipedia
 152                         $sql .= makeLanguageLinks( $sites['wiki'], $first, $db );
 153
 154                         # Extra links
 155                         foreach ( $extraLinks as $link ) {
 156                                 $sql .= makeLink( $link, $first, $db );
 157                         }
 158
 159                         $sql .= ";\n";
 160                 } else {
 161                         # Find out which site this DB belongs to
 162                         $site = false;
 163                         foreach( $sites as $candidateSite ) {
 164                                 $suffix = $candidateSite->suffix;
 165                                 if ( preg_match( "/(.*)$suffix$/", $db, $matches ) ) {
 166                                         $site = $candidateSite;
 167                                         break;
 168                                 }
 169                         }
 170                         if ( !$site ) {
 171                                 print "Invalid database $db\n";
 172                                 continue;
 173                         }
 174                         $lang = $matches[1];
 175                         $host = "$lang." . $site->url;
 176                         $sql .= "\n--$host\n\n";
 177
 178                         $sql .= "USE $db;\n" .
 179                                         "TRUNCATE TABLE interwiki;\n" .
 180                                         "INSERT INTO interwiki (iw_prefix,iw_url,iw_local) VALUES\n";
 181                         $first = true;
 182
 183                         # Intermap links
 184                         foreach ( $iwArray as $iwEntry ) {
 185                                 # Suppress links with the same name as the site
 186                                 if ( ( $suffix == 'wiki' && $iwEntry['iw_prefix'] != 'wikipedia' ) ||
 187                                   ( $suffix != 'wiki' && $suffix != $iwEntry['iw_prefix'] ) )
 188                                 {
 189                                         $sql .= makeLink( $iwEntry, $first, $db );
 190                                 }
 191                         }
 192
 193                         # Lateral links
 194                         foreach ( $sites as $targetSite ) {
 195                                 # Suppress link to self
 196                                 if ( $targetSite->suffix != $site->suffix ) {
 197                                         $sql .= makeLink( array( $targetSite->lateral, $targetSite->getURL( $lang ), 1 ), $first, $db );
 198                                 }
 199                         }
 200
 201                         # Interlanguage links
 202                         $sql .= makeLanguageLinks( $site, $first, $db );
 203
 204                         # w link within wikipedias
 205                         # Other sites already have it as a lateral link
 206                         if ( $site->suffix == "wiki" ) {
 207                                 $sql .= makeLink( array("w", "http://en.wikipedia.org/wiki/$1", 1), $first, $db );
 208                         }
 209
 210                         # Extra links
 211                         foreach ( $extraLinks as $link ){
 212                                         $sql .= makeLink( $link, $first, $db );
 213                         }
 214                         $sql .= ";\n\n";
 215                 }
 216         }
 217         return $sql;
 218 }
 219
 220 # ------------------------------------------------------------------------------------------
 221
 222 # Returns part of an INSERT statement, corresponding to all interlanguage links to a particular site
 223 function makeLanguageLinks( &$site, &$first, $source ) {
 224         global $langlist, $languageAliases;
 225
 226         $sql = "";
 227
 228         # Actual languages with their own databases
 229         foreach ( $langlist as $targetLang ) {
 230                 $sql .= makeLink( array( $targetLang, $site->getURL( $targetLang ), 1 ), $first, $source );
 231         }
 232
 233         # Language aliases
 234         foreach ( $languageAliases as $alias => $lang ) {
 235                 $sql .= makeLink( array( $alias, $site->getURL( $lang ), 1 ), $first, $source );
 236         }
 237         return $sql;
 238 }
 239
 240 # Make SQL for a single link from an array
 241 function makeLink( $entry, &$first, $source ) {
 242         global $prefixRewrites;
 243
 244         if ( isset( $prefixRewrites[$source] ) && isset( $prefixRewrites[$source][$entry[0]] ) ) {
 245                 $entry[0] = $prefixRewrites[$source][$entry[0]];
 246         }
 247
 248         $sql = "";
 249         # Add comma
 250         if ( $first ) {
 251                 $first = false;
 252         } else {
 253                 $sql .= ",\n";
 254         }
 255         $dbr =& wfGetDB( DB_SLAVE );
 256         $sql .= "(" . $dbr->makeList( $entry ) . ")";
 257         return $sql;
 258 }
 259
 260 ?>