maintenance/rebuildInterwiki.php

   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 require_once( dirname( __FILE__ ) . '/Maintenance.php' );
  32
  33 class RebuildInterwiki extends Maintenance {
  34         public function __construct() {
  35                 parent::__construct();
  36                 $this->mDescription = "Protect or unprotect an article from the command line.";
  37                 $this->addOption( 'langlist', 'File with one language code per line', false, true );
  38                 $this->addOption( 'dblist', 'File with one db per line', false, true );
  39                 $this->addOption( 'd', 'Output folder', false, true );
  40         }
  41
  42         function execute() {
  43                 # List of language prefixes likely to be found in multi-language sites
  44                 $this->langlist = array_map( "trim", file( $this->getOption( 'langlist', "/home/wikipedia/common/langlist" ) ) );
  45
  46                 # List of all database names
  47                 $this->dblist = array_map( "trim", file( $this->getOption( 'dblist', "/home/wikipedia/common/all.dblist" ) ) );
  48
  49                 $this->makeInterwikiSQL( $this->getOption( 'd', '/home/wikipedia/conf/interwiki/sql' ) );
  50         }
  51
  52         function makeInterwikiSQL( $destDir ) {
  53                 $this->output( "Making new interwiki SQL files in $destDir\n" );
  54
  55                 # Multi-language sites
  56                 # db suffix => db suffix, iw prefix, hostname
  57                 $sites = array(
  58                         'wiki' => new Site( 'wiki', 'w', 'wikipedia.org' ),
  59                         'wiktionary' => new Site( 'wiktionary', 'wikt', 'wiktionary.org' ),
  60                         'wikiquote' => new Site( 'wikiquote', 'q', 'wikiquote.org' ),
  61                         'wikibooks' => new Site( 'wikibooks', 'b', 'wikibooks.org' ),
  62                         'wikinews' => new Site( 'wikinews', 'n', 'wikinews.org' ),
  63                         'wikisource' => new Site( 'wikisource', 's', 'wikisource.org' ),
  64                         'wikimedia' => new Site( 'wikimedia', 'chapter', 'wikimedia.org' ),
  65                         'wikiversity' => new Site( 'wikiversity', 'v', 'wikiversity.org' ),
  66                 );
  67
  68                 # Special-case hostnames
  69                 $specials = array(
  70                         'sourceswiki' => 'sources.wikipedia.org',
  71                         'quotewiki' => 'wikiquote.org',
  72                         'textbookwiki' => 'wikibooks.org',
  73                         'sep11wiki' => 'sep11.wikipedia.org',
  74                         'metawiki' => 'meta.wikimedia.org',
  75                         'commonswiki' => 'commons.wikimedia.org',
  76                         'specieswiki' => 'species.wikimedia.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                         wfDie( "m:Interwiki_map not found" );
 121                 }
 122
 123                 $iwArray = array();
 124
 125                 foreach ( $lines as $line ) {
 126                         $matches = array();
 127                         if ( preg_match( '/^\|\s*(.*?)\s*\|\|\s*(https?:\/\/.*?)\s*$/', $line, $matches ) ) {
 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                                         $iwArray[$prefix] = array( "iw_prefix" => $prefix, "iw_url" => $url, "iw_local" => $local );
 138                                 }
 139                         }
 140                 }
 141
 142                 foreach ( $this->dblist as $db ) {
 143                         $sql = "-- Generated by rebuildInterwiki.php";
 144                         if ( isset( $specials[$db] ) ) {
 145                                 # Special wiki
 146                                 # Has interwiki links and interlanguage links to wikipedia
 147
 148                                 $host = $specials[$db];
 149                                 $sql .= "\n--$host\n\n";
 150                                 $sql .= "USE $db;\n" .
 151                                                 "TRUNCATE TABLE interwiki;\n" .
 152                                                 "INSERT INTO interwiki (iw_prefix, iw_url, iw_local) VALUES \n";
 153                                 $first = true;
 154
 155                                 # Intermap links
 156                                 foreach ( $iwArray as $iwEntry ) {
 157                                         $sql .= $this->makeLink( $iwEntry, $first, $db );
 158                                 }
 159
 160                                 # Links to multilanguage sites
 161                                 foreach ( $sites as $targetSite ) {
 162                                         $sql .= $this->makeLink( array( $targetSite->lateral, $targetSite->getURL( 'en' ), 1 ), $first, $db );
 163                                 }
 164
 165                                 # Interlanguage links to wikipedia
 166                                 $sql .= $this->makeLanguageLinks( $sites['wiki'], $first, $db );
 167
 168                                 # Extra links
 169                                 foreach ( $extraLinks as $link ) {
 170                                         $sql .= $this->makeLink( $link, $first, $db );
 171                                 }
 172
 173                                 $sql .= ";\n";
 174                         } else {
 175                                 # Find out which site this DB belongs to
 176                                 $site = false;
 177                                 foreach ( $sites as $candidateSite ) {
 178                                         $suffix = $candidateSite->suffix;
 179                                         if ( preg_match( "/(.*)$suffix$/", $db, $matches ) ) {
 180                                                 $site = $candidateSite;
 181                                                 break;
 182                                         }
 183                                 }
 184                                 if ( !$site ) {
 185                                         print "Invalid database $db\n";
 186                                         continue;
 187                                 }
 188                                 $lang = $matches[1];
 189                                 $host = "$lang." . $site->url;
 190                                 $sql .= "\n--$host\n\n";
 191
 192                                 $sql .= "USE $db;\n" .
 193                                                 "TRUNCATE TABLE interwiki;\n" .
 194                                                 "INSERT INTO interwiki (iw_prefix,iw_url,iw_local) VALUES\n";
 195                                 $first = true;
 196
 197                                 # Intermap links
 198                                 foreach ( $iwArray as $iwEntry ) {
 199                                         # Suppress links with the same name as the site
 200                                         if ( ( $suffix == 'wiki' && $iwEntry['iw_prefix'] != 'wikipedia' ) ||
 201                                                 ( $suffix != 'wiki' && $suffix != $iwEntry['iw_prefix'] ) )
 202                                         {
 203                                                 $sql .= $this->makeLink( $iwEntry, $first, $db );
 204                                         }
 205                                 }
 206
 207                                 # Lateral links
 208                                 foreach ( $sites as $targetSite ) {
 209                                         # Suppress link to self
 210                                         if ( $targetSite->suffix != $site->suffix ) {
 211                                                 $sql .= $this->makeLink( array( $targetSite->lateral, $targetSite->getURL( $lang ), 1 ), $first, $db );
 212                                         }
 213                                 }
 214
 215                                 # Interlanguage links
 216                                 $sql .= $this->makeLanguageLinks( $site, $first, $db );
 217
 218                                 # w link within wikipedias
 219                                 # Other sites already have it as a lateral link
 220                                 if ( $site->suffix == "wiki" ) {
 221                                         $sql .= $this->makeLink( array( "w", "http://en.wikipedia.org/wiki/$1", 1 ), $first, $db );
 222                                 }
 223
 224                                 # Extra links
 225                                 foreach ( $extraLinks as $link ) {
 226                                                 $sql .= $this->makeLink( $link, $first, $db );
 227                                 }
 228                                 $sql .= ";\n";
 229                         }
 230                         file_put_contents( "$destDir/$db.sql", $sql );
 231                 }
 232         }
 233
 234         # ------------------------------------------------------------------------------------------
 235
 236         # Returns part of an INSERT statement, corresponding to all interlanguage links to a particular site
 237         function makeLanguageLinks( &$site, &$first, $source ) {
 238                 $sql = "";
 239
 240                 # Actual languages with their own databases
 241                 foreach ( $this->langlist as $targetLang ) {
 242                         $sql .= $this->makeLink( array( $targetLang, $site->getURL( $targetLang ), 1 ), $first, $source );
 243                 }
 244
 245                 # Language aliases
 246                 foreach ( $this->languageAliases as $alias => $lang ) {
 247                         $sql .= $this->makeLink( array( $alias, $site->getURL( $lang ), 1 ), $first, $source );
 248                 }
 249                 return $sql;
 250         }
 251
 252         # Make SQL for a single link from an array
 253         function makeLink( $entry, &$first, $source ) {
 254
 255                 if ( isset( $this->prefixRewrites[$source] ) && isset($entry[0]) && isset( $this->prefixRewrites[$source][$entry[0]] ) ) {
 256                         $entry[0] = $this->prefixRewrites[$source][$entry[0]];
 257                 }
 258
 259                 $sql = "";
 260                 # Add comma
 261                 if ( $first ) {
 262                         $first = false;
 263                 } else {
 264                         $sql .= ",\n";
 265                 }
 266                 $dbr = wfGetDB( DB_SLAVE );
 267                 $sql .= "(" . $dbr->makeList( $entry ) . ")";
 268                 return $sql;
 269         }
 270 }
 271
 272 $maintClass = "RebuildInterwiki";
 273 require_once( DO_MAINTENANCE );
 274