maintenance/dumpInterwiki.php

   1 <?php
   2 /**
   3  * Build constant slightly compact database of interwiki prefixes
   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 require_once( dirname( __FILE__ ) . '/Maintenance.php' );
  32
  33 class DumpInterwiki extends Maintenance {
  34
  35         public function __construct() {
  36                 parent::__construct();
  37                 $this->mDescription = "Build constant slightly compact database of interwiki prefixes.";
  38                 $this->addOption( 'langlist', 'File with one language code per line', false, true );
  39                 $this->addOption( 'dblist', 'File with one db per line', false, true );
  40                 $this->addOption( 'specialdbs', "File with one 'special' db per line", false, true );
  41                 $this->addOption( 'o', 'Cdb output file', false, true );
  42         }
  43
  44         function execute() {
  45                 # List of language prefixes likely to be found in multi-language sites
  46                 $this->langlist = array_map( "trim", file( $this->getOption( 'langlist', "/home/wikipedia/common/langlist" ) ) );
  47
  48                 # List of all database names
  49                 $this->dblist = array_map( "trim", file( $this->getOption( 'dblist', "/home/wikipedia/common/all.dblist" ) ) );
  50
  51                 # Special-case databases
  52                 $this->specials = array_flip(   array_map( "trim", file( $this->getOption( 'specialdbs', "/home/wikipedia/common/special.dblist" ) ) ) );
  53
  54                 if ( $this->hasOption( 'o' ) ) {
  55                         $this->dbFile = CdbWriter::open( $this->getOption( 'o' ) ) ;
  56                 } else {
  57                         $this->dbFile = false;
  58                 }
  59
  60                 $this->getRebuildInterwikiDump();
  61         }
  62
  63         function getRebuildInterwikiDump() {
  64                 global $wgContLang;
  65
  66                 # Multi-language sites
  67                 # db suffix => db suffix, iw prefix, hostname
  68                 $sites = array(
  69                         'wiki' => new Site( 'wiki', 'w', 'wikipedia.org' ),
  70                         'wiktionary' => new Site( 'wiktionary', 'wikt', 'wiktionary.org' ),
  71                         'wikiquote' => new Site( 'wikiquote', 'q', 'wikiquote.org' ),
  72                         'wikibooks' => new Site( 'wikibooks', 'b', 'wikibooks.org' ),
  73                         'wikinews' => new Site( 'wikinews', 'n', 'wikinews.org' ),
  74                         'wikisource' => new Site( 'wikisource', 's', 'wikisource.org' ),
  75                         'wikimedia' => new Site( 'wikimedia', 'chapter', 'wikimedia.org' ),
  76                         'wikiversity' => new Site( 'wikiversity', 'v', 'wikiversity.org' ),
  77                 );
  78
  79                 # Extra interwiki links that can't be in the intermap for some reason
  80                 $extraLinks = array(
  81                         array( 'm', 'http://meta.wikimedia.org/wiki/$1', 1 ),
  82                         array( 'meta', 'http://meta.wikimedia.org/wiki/$1', 1 ),
  83                         array( 'sep11', 'http://sep11.wikipedia.org/wiki/$1', 1 ),
  84                 );
  85
  86                 # Language aliases, usually configured as redirects to the real wiki in apache
  87                 # Interlanguage links are made directly to the real wiki
  88                 # Something horrible happens if you forget to list an alias here, I can't
  89                 #   remember what
  90                 $this->languageAliases = array(
  91                         'zh-cn' => 'zh',
  92                         'zh-tw' => 'zh',
  93                         'dk' => 'da',
  94                         'nb' => 'no',
  95                 );
  96
  97                 # Special case prefix rewrites, for the benefit of Swedish which uses s:t
  98                 # as an abbreviation for saint
  99                 $this->prefixRewrites = array(
 100                         'svwiki' => array( 's' => 'src' ),
 101                 );
 102
 103                 # Construct a list of reserved prefixes
 104                 $reserved = array();
 105                 foreach ( $this->langlist as $lang ) {
 106                         $reserved[$lang] = 1;
 107                 }
 108                 foreach ( $this->languageAliases as $alias => $lang ) {
 109                         $reserved[$alias] = 1;
 110                 }
 111                 foreach ( $sites as $site ) {
 112                         $reserved[$site->lateral] = 1;
 113                 }
 114
 115                 # Extract the intermap from meta
 116                 $intermap = Http::get( 'http://meta.wikimedia.org/w/index.php?title=Interwiki_map&action=raw', 30 );
 117                 $lines = array_map( 'trim', explode( "\n", trim( $intermap ) ) );
 118
 119                 if ( !$lines || count( $lines ) < 2 ) {
 120                         $this->error( "m:Interwiki_map not found", true );
 121                 }
 122
 123                 # Global iterwiki map
 124                 foreach ( $lines as $line ) {
 125                         if ( preg_match( '/^\|\s*(.*?)\s*\|\|\s*(.*?)\s*$/', $line, $matches ) ) {
 126                                 $prefix = $wgContLang->lc( $matches[1] );
 127                                 $prefix = str_replace( ' ', '_', $prefix );
 128                                 $prefix = strtolower( $matches[1] );
 129                                 $url = $matches[2];
 130                                 if ( preg_match( '/(wikipedia|wiktionary|wikisource|wikiquote|wikibooks|wikimedia)\.org/', $url ) ) {
 131                                         $local = 1;
 132                                 } else {
 133                                         $local = 0;
 134                                 }
 135
 136                                 if ( empty( $reserved[$prefix] ) ) {
 137                                         $imap  = array( "iw_prefix" => $prefix, "iw_url" => $url, "iw_local" => $local );
 138                                         $this->makeLink ( $imap, "__global" );
 139                                 }
 140                         }
 141                 }
 142
 143                 # Exclude Wikipedia for Wikipedia
 144                 $this->makeLink ( array ( 'iw_prefix' => 'wikipedia', 'is_url' => null ), "_wiki" );
 145
 146                 # Multilanguage sites
 147                 foreach ( $sites as $site ) {
 148                         $this->makeLanguageLinks ( $site, "_" . $site->suffix );
 149                 }
 150
 151
 152                 foreach ( $dblist as $db ) {
 153                         if ( isset( $this->specials[$db] ) ) {
 154                                 # Special wiki
 155                                 # Has interwiki links and interlanguage links to wikipedia
 156
 157                                 $this->makeLink( array( 'iw_prefix' => $db, 'iw_url' => "wiki" ), "__sites" );
 158                                 # Links to multilanguage sites
 159                                 foreach ( $sites as $targetSite ) {
 160                                         $this->makeLink( array( 'iw_prefix' => $targetSite->lateral,
 161                                                 'iw_url' => $targetSite->getURL( 'en' ),
 162                                                 'iw_local' => 1 ), $db );
 163                                 }
 164                         } else {
 165                                 # Find out which site this DB belongs to
 166                                 $site = false;
 167                                 foreach ( $sites as $candidateSite ) {
 168                                         $suffix = $candidateSite->suffix;
 169                                         if ( preg_match( "/(.*)$suffix$/", $db, $matches ) ) {
 170                                                 $site = $candidateSite;
 171                                                 break;
 172                                         }
 173                                 }
 174                                 $this->makeLink( array( 'iw_prefix' => $db, 'iw_url' => $site->suffix ), "__sites" );
 175                                 if ( !$site ) {
 176                                         $this->error( "Invalid database $db\n" );
 177                                         continue;
 178                                 }
 179                                 $lang = $matches[1];
 180
 181                                 # Lateral links
 182                                 foreach ( $sites as $targetSite ) {
 183                                         if ( $targetSite->suffix != $site->suffix ) {
 184                                                 $this->makeLink( array( 'iw_prefix' => $targetSite->lateral,
 185                                                         'iw_url' => $targetSite->getURL( $lang ),
 186                                                         'iw_local' => 1 ), $db );
 187                                         }
 188                                 }
 189
 190                                 if ( $site->suffix == "wiki" ) {
 191                                         $this->makeLink( array( 'iw_prefix' => 'w',
 192                                                 'iw_url' => "http://en.wikipedia.org/wiki/$1",
 193                                                 'iw_local' => 1 ), $db );
 194                                 }
 195
 196                         }
 197                 }
 198                 foreach ( $extraLinks as $link ) {
 199                         $this->makeLink( $link, "__global" );
 200                 }
 201         }
 202
 203         # ------------------------------------------------------------------------------------------
 204
 205         # Executes part of an INSERT statement, corresponding to all interlanguage links to a particular site
 206         function makeLanguageLinks( &$site, $source ) {
 207                 # Actual languages with their own databases
 208                 foreach ( $this->langlist as $targetLang ) {
 209                         $this->makeLink( array( $targetLang, $site->getURL( $targetLang ), 1 ), $source );
 210                 }
 211
 212                 # Language aliases
 213                 foreach ( $this->languageAliases as $alias => $lang ) {
 214                         $this->makeLink( array( $alias, $site->getURL( $lang ), 1 ), $source );
 215                 }
 216         }
 217
 218         function makeLink( $entry, $source ) {
 219                 if ( isset( $this->prefixRewrites[$source] ) && isset( $this->prefixRewrites[$source][$entry[0]] ) )
 220                         $entry[0] = $this->prefixRewrites[$source][$entry[0]];
 221
 222                 if ( !array_key_exists( "iw_prefix", $entry ) ) {
 223                         $entry = array( "iw_prefix" => $entry[0], "iw_url" => $entry[1], "iw_local" => $entry[2] );
 224                 }
 225                 if ( array_key_exists( $source, $this->prefixRewrites ) &&
 226                                 array_key_exists( $entry['iw_prefix'], $this->prefixRewrites[$source] ) ) {
 227                         $entry['iw_prefix'] = $this->prefixRewrites[$source][$entry['iw_prefix']];
 228                 }
 229
 230                 if ( $this->dbFile ) {
 231                         $this->dbFile->set( "{$source}:{$entry['iw_prefix']}", trim( "{$entry['iw_local']} {$entry['iw_url']}" ) );
 232                 } else {
 233                         $this->output( "{$source}:{$entry['iw_prefix']} {$entry['iw_url']} {$entry['iw_local']}\n" );
 234                 }
 235         }
 236 }
 237
 238 $maintClass = "DumpInterwiki";
 239 require_once( DO_MAINTENANCE );
 240