maintenance/rebuildInterwiki.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  * @ingroup Wikimedia
  10  */
  11
  12 /**
  13  * @todo document
  14  * @ingroup 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 makeInterwikiSQL( $destDir ) {
  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 hostnames
  54         $specials = array(
  55                 'sourceswiki' => 'sources.wikipedia.org',
  56                 'quotewiki' => 'wikiquote.org',
  57                 'textbookwiki' => 'wikibooks.org',
  58                 'sep11wiki' => 'sep11.wikipedia.org',
  59                 'metawiki' => 'meta.wikimedia.org',
  60                 'commonswiki' => 'commons.wikimedia.org',
  61                 'specieswiki' => 'species.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 = Http::get( '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                 wfDie( "m:Interwiki_map not found" );
 106         }
 107
 108         $iwArray = array();
 109
 110         foreach ( $lines as $line ) {
 111                 $matches = array();
 112                 if ( preg_match( '/^\|\s*(.*?)\s*\|\|\s*(https?:\/\/.*?)\s*$/', $line, $matches ) ) {
 113                         $prefix = strtolower( $matches[1] );
 114                         $url = $matches[2];
 115                         if ( preg_match( '/(wikipedia|wiktionary|wikisource|wikiquote|wikibooks|wikimedia)\.org/', $url ) ) {
 116                                 $local = 1;
 117                         } else {
 118                                 $local = 0;
 119                         }
 120
 121                         if ( empty( $reserved[$prefix] ) ) {
 122                                 $iwArray[$prefix] = array( "iw_prefix" => $prefix, "iw_url" => $url, "iw_local" => $local );
 123                         }
 124                 }
 125         }
 126
 127
 128
 129         foreach ( $dblist as $db ) {
 130                 $sql = "-- Generated by rebuildInterwiki.php";
 131                 if ( isset( $specials[$db] ) ) {
 132                         # Special wiki
 133                         # Has interwiki links and interlanguage links to wikipedia
 134
 135                         $host = $specials[$db];
 136                         $sql .= "\n--$host\n\n";
 137                         $sql .= "USE $db;\n" .
 138                                         "TRUNCATE TABLE interwiki;\n" .
 139                                         "INSERT INTO interwiki (iw_prefix, iw_url, iw_local) VALUES \n";
 140                         $first = true;
 141
 142                         # Intermap links
 143                         foreach ( $iwArray as $iwEntry ) {
 144                                 $sql .= makeLink( $iwEntry, $first, $db );
 145                         }
 146
 147                         # Links to multilanguage sites
 148                         foreach ( $sites as $targetSite ) {
 149                                 $sql .= makeLink( array( $targetSite->lateral, $targetSite->getURL( 'en' ), 1 ), $first, $db );
 150                         }
 151
 152                         # Interlanguage links to wikipedia
 153                         $sql .= makeLanguageLinks( $sites['wiki'], $first, $db );
 154
 155                         # Extra links
 156                         foreach ( $extraLinks as $link ) {
 157                                 $sql .= makeLink( $link, $first, $db );
 158                         }
 159
 160                         $sql .= ";\n";
 161                 } else {
 162                         # Find out which site this DB belongs to
 163                         $site = false;
 164                         foreach ( $sites as $candidateSite ) {
 165                                 $suffix = $candidateSite->suffix;
 166                                 if ( preg_match( "/(.*)$suffix$/", $db, $matches ) ) {
 167                                         $site = $candidateSite;
 168                                         break;
 169                                 }
 170                         }
 171                         if ( !$site ) {
 172                                 print "Invalid database $db\n";
 173                                 continue;
 174                         }
 175                         $lang = $matches[1];
 176                         $host = "$lang." . $site->url;
 177                         $sql .= "\n--$host\n\n";
 178
 179                         $sql .= "USE $db;\n" .
 180                                         "TRUNCATE TABLE interwiki;\n" .
 181                                         "INSERT INTO interwiki (iw_prefix,iw_url,iw_local) VALUES\n";
 182                         $first = true;
 183
 184                         # Intermap links
 185                         foreach ( $iwArray as $iwEntry ) {
 186                                 # Suppress links with the same name as the site
 187                                 if ( ( $suffix == 'wiki' && $iwEntry['iw_prefix'] != 'wikipedia' ) ||
 188                                   ( $suffix != 'wiki' && $suffix != $iwEntry['iw_prefix'] ) )
 189                                 {
 190                                         $sql .= makeLink( $iwEntry, $first, $db );
 191                                 }
 192                         }
 193
 194                         # Lateral links
 195                         foreach ( $sites as $targetSite ) {
 196                                 # Suppress link to self
 197                                 if ( $targetSite->suffix != $site->suffix ) {
 198                                         $sql .= makeLink( array( $targetSite->lateral, $targetSite->getURL( $lang ), 1 ), $first, $db );
 199                                 }
 200                         }
 201
 202                         # Interlanguage links
 203                         $sql .= makeLanguageLinks( $site, $first, $db );
 204
 205                         # w link within wikipedias
 206                         # Other sites already have it as a lateral link
 207                         if ( $site->suffix == "wiki" ) {
 208                                 $sql .= makeLink( array( "w", "http://en.wikipedia.org/wiki/$1", 1 ), $first, $db );
 209                         }
 210
 211                         # Extra links
 212                         foreach ( $extraLinks as $link ) {
 213                                         $sql .= makeLink( $link, $first, $db );
 214                         }
 215                         $sql .= ";\n";
 216                 }
 217                 file_put_contents( "$destDir/$db.sql", $sql );
 218         }
 219 }
 220
 221 # ------------------------------------------------------------------------------------------
 222
 223 # Returns part of an INSERT statement, corresponding to all interlanguage links to a particular site
 224 function makeLanguageLinks( &$site, &$first, $source ) {
 225         global $langlist, $languageAliases;
 226
 227         $sql = "";
 228
 229         # Actual languages with their own databases
 230         foreach ( $langlist as $targetLang ) {
 231                 $sql .= makeLink( array( $targetLang, $site->getURL( $targetLang ), 1 ), $first, $source );
 232         }
 233
 234         # Language aliases
 235         foreach ( $languageAliases as $alias => $lang ) {
 236                 $sql .= makeLink( array( $alias, $site->getURL( $lang ), 1 ), $first, $source );
 237         }
 238         return $sql;
 239 }
 240
 241 # Make SQL for a single link from an array
 242 function makeLink( $entry, &$first, $source ) {
 243         global $prefixRewrites;
 244
 245         if ( isset( $prefixRewrites[$source] ) && isset( $prefixRewrites[$source][$entry[0]] ) ) {
 246                 $entry[0] = $prefixRewrites[$source][$entry[0]];
 247         }
 248
 249         $sql = "";
 250         # Add comma
 251         if ( $first ) {
 252                 $first = false;
 253         } else {
 254                 $sql .= ",\n";
 255         }
 256         $dbr = wfGetDB( DB_SLAVE );
 257         $sql .= "(" . $dbr->makeList( $entry ) . ")";
 258         return $sql;
 259 }