maintenance/rebuildInterwiki.php

   1 <?
   2
   3 # Rebuild interwiki table using the file on meta and the language list
   4 # Wikimedia specific!
   5 $oldCwd = getcwd();
   6
   7 $optionsWithArgs = array( "o" );
   8 include_once( "commandLine.inc" );
   9
  10 class Site {
  11         var $suffix, $lateral, $url;
  12
  13         function Site( $s, $l, $u ) {
  14                 $this->suffix = $s;
  15                 $this->lateral = $l;
  16                 $this->url = $u;
  17         }
  18
  19         function getURL( $lang ) {
  20                 return "http://$lang.{$this->url}/wiki/\$1";
  21         }
  22 }
  23
  24 # Initialise lists of wikis
  25 $sites = array(
  26         'wiki' => new Site( 'wiki', 'w', 'wikipedia.org' ),
  27         'wiktionary' => new Site( 'wiktionary', 'wikt', 'wiktionary.org' )
  28 );
  29 $langlist = array_map( "trim", file( "/home/wikipedia/common/langlist" ) );
  30
  31 $specials = array(
  32         'sourceswiki' => 'sources.wikipedia.org',
  33         'quotewiki' => 'wikiquote.org',
  34         'textbookwiki' => 'wikibooks.org',
  35         'sep11wiki' => 'sep11.wikipedia.org',
  36         'metawiki' => 'meta.wikipedia.org',
  37 );
  38
  39 $extraLinks = array(
  40         array( 'm', 'http://meta.wikipedia.org/wiki/$1', 1 ),
  41         array( 'meta', 'http://meta.wikipedia.org/wiki/$1', 1 ),
  42         array( 'sep11', 'http://sep11.wikipedia.org/wiki/$1', 1 ),
  43 );
  44
  45 $languageAliases = array(
  46         'zh-cn' => 'zh',
  47         'zh-tw' => 'zh',
  48 );
  49
  50 # Extract the intermap from meta
  51
  52 $row = wfGetArray( "metawiki.cur", array( "cur_text" ), array( "cur_namespace" => 0, "cur_title" => "Interwiki_map" ) );
  53
  54 if ( !$row ) {
  55         die( "m:Interwiki_map not found" );
  56 }
  57
  58 $lines = explode( "\n", $row->cur_text );
  59 $iwArray = array();
  60
  61 foreach ( $lines as $line ) {
  62         if ( preg_match( '/^\|\s*(.*?)\s*\|\|\s*(.*?)\s*$/', $line, $matches ) ) {
  63                 $prefix = $matches[1];
  64                 $url = $matches[2];
  65                 if ( preg_match( '/(wikipedia|wiktionary|wikisource|wikiquote|wikibooks)\.org/', $url ) ) {
  66                         $local = 1;
  67                 } else {
  68                         $local = 0;
  69                 }
  70
  71                 $iwArray[] = array( "iw_prefix" => $prefix, "iw_url" => $url, "iw_local" => $local );
  72         }
  73 }
  74
  75
  76 # Insert links into special wikis
  77 # These have intermap links and interlanguage links pointing to wikipedia
  78
  79 $sql = "-- Generated by rebuildInterwiki.php";
  80
  81 foreach ( $specials as $db => $host ) {
  82         $sql .= "\nUSE $db;\n" .
  83                         "TRUNCATE TABLE interwiki;\n" .
  84                         "INSERT INTO interwiki (iw_prefix, iw_url, iw_local) VALUES \n";
  85         $first = true;
  86
  87         # Intermap links
  88         foreach ( $iwArray as $iwEntry ) {
  89                 # Suppress links to self
  90                 if ( strpos( $iwEntry['iw_url'], $host ) === false ) {
  91                         $sql .= makeLink( $iwEntry, $first );
  92                 }
  93         }
  94         # w link
  95         $sql .= makeLink( array("w", "http://en.wikipedia.org/wiki/$1", 1 ), $first );
  96
  97         # Interlanguage links to wikipedia
  98         $sql .= makeLanguageLinks( $sites['wiki'], $first );
  99
 100         # Extra links
 101         foreach ( $extraLinks as $link ) {
 102                         $sql .= makeLink( $link, $first );
 103         }
 104
 105         $sql .= ";\n";
 106 }
 107 $sql .= "\n";
 108
 109 # Insert links into multilanguage sites
 110
 111 foreach ( $sites as $site ) {
 112         $sql .= <<<EOS
 113
 114 ---
 115 --- {$site->suffix}
 116 ---
 117
 118 EOS;
 119         foreach ( $langlist as $lang ) {
 120                 $db = $lang . $site->suffix;
 121                 $db = str_replace( "-", "_", $db );
 122
 123                 $sql .= "USE $db;\n" .
 124                                 "TRUNCATE TABLE interwiki;\n" .
 125                                 "INSERT INTO interwiki (iw_prefix,iw_url,iw_local) VALUES\n";
 126                 $first = true;
 127
 128                 # Intermap links
 129                 foreach ( $iwArray as $iwEntry ) {
 130                         # Suppress links to self
 131                         if ( strpos( $iwEntry['iw_url'], $site->url ) === false ||
 132                           strpos( $iwEntry['iw_url'], 'meta.wikipedia.org' ) !== false ) {
 133                                 $sql .= makeLink( $iwEntry, $first );
 134                         }
 135                 }
 136
 137                 # Lateral links
 138                 foreach ( $sites as $targetSite ) {
 139                         # Suppress link to self
 140                         if ( $targetSite->suffix != $site->suffix ) {
 141                                 $sql .= makeLink( array( $targetSite->lateral, $targetSite->getURL( $lang ), 1 ), $first );
 142                         }
 143                 }
 144
 145                 # Interlanguage links
 146                 $sql .= makeLanguageLinks( $site, $first );
 147
 148                 # w link within wikipedias
 149                 # Other sites already have it as a lateral link
 150                 if ( $site->suffix == "wiki" ) {
 151                         $sql .= makeLink( array("w", "http://en.wikipedia.org/wiki/$1", 1), $first );
 152                 }
 153
 154                 # Extra links
 155                 foreach ( $extraLinks as $link ){
 156                                 $sql .= makeLink( $link, $first );
 157                 }
 158                 $sql .= ";\n\n";
 159         }
 160 }
 161
 162 # Output
 163 if ( isset( $options['o'] ) ) {
 164         # To file specified with -o
 165         chdir( $oldCwd );
 166         $file = fopen( $options['o'], "w" );
 167         fwrite( $file, $sql );
 168         fclose( $file );
 169 } else {
 170         # To stdout
 171         print $sql;
 172 }
 173
 174 # ------------------------------------------------------------------------------------------
 175
 176 # Returns part of an INSERT statement, corresponding to all interlanguage links to a particular site
 177 function makeLanguageLinks( &$site, &$first ) {
 178         global $langlist, $languageAliases;
 179
 180         $sql = "";
 181
 182         # Actual languages with their own databases
 183         foreach ( $langlist as $targetLang ) {
 184                 $sql .= makeLink( array( $targetLang, $site->getURL( $targetLang ), 1 ), $first );
 185         }
 186
 187         # Language aliases
 188         foreach ( $languageAliases as $alias => $lang ) {
 189                 $sql .= makeLink( array( $alias, $site->getURL( $lang ), 1 ), $first );
 190         }
 191         return $sql;
 192 }
 193
 194 # Make SQL for a single link from an array
 195 function makeLink( $entry, &$first ) {
 196         $sql = "";
 197         # Add comma
 198         if ( $first ) {
 199                 $first = false;
 200         } else {
 201                 $sql .= ",\n";
 202         }
 203         $sql .= "(" . Database::makeList( $entry ) . ")";
 204         return $sql;
 205 }
 206
 207 ?>