From 180f85643cdbc6647f36951af1a4ef45dc5b39d3 Mon Sep 17 00:00:00 2001 From: Platonides Date: Sat, 4 Sep 2010 19:15:42 +0000 Subject: [PATCH] Port rebuildInterwiki to the new Maintenance system. --- maintenance/rebuildInterwiki.inc | 259 ----------------------------- maintenance/rebuildInterwiki.php | 275 +++++++++++++++++++++++++++++-- 2 files changed, 260 insertions(+), 274 deletions(-) delete mode 100644 maintenance/rebuildInterwiki.inc diff --git a/maintenance/rebuildInterwiki.inc b/maintenance/rebuildInterwiki.inc deleted file mode 100644 index 9ad9dbf522..0000000000 --- a/maintenance/rebuildInterwiki.inc +++ /dev/null @@ -1,259 +0,0 @@ -suffix = $s; - $this->lateral = $l; - $this->url = $u; - } - - function getURL( $lang ) { - $xlang = str_replace( '_', '-', $lang ); - return "http://$xlang.{$this->url}/wiki/\$1"; - } -} - -function makeInterwikiSQL( $destDir ) { - global $langlist, $languageAliases, $prefixRewrites; - - # Multi-language sites - # db suffix => db suffix, iw prefix, hostname - $sites = array( - 'wiki' => new Site( 'wiki', 'w', 'wikipedia.org' ), - 'wiktionary' => new Site( 'wiktionary', 'wikt', 'wiktionary.org' ), - 'wikiquote' => new Site( 'wikiquote', 'q', 'wikiquote.org' ), - 'wikibooks' => new Site( 'wikibooks', 'b', 'wikibooks.org' ), - 'wikinews' => new Site( 'wikinews', 'n', 'wikinews.org' ), - 'wikisource' => new Site( 'wikisource', 's', 'wikisource.org' ), - 'wikimedia' => new Site( 'wikimedia', 'chapter', 'wikimedia.org' ), - 'wikiversity' => new Site( 'wikiversity', 'v', 'wikiversity.org' ), - ); - - # List of language prefixes likely to be found in multi-language sites - $langlist = array_map( "trim", file( "/home/wikipedia/common/langlist" ) ); - - # List of all database names - $dblist = array_map( "trim", file( "/home/wikipedia/common/all.dblist" ) ); - - # Special-case hostnames - $specials = array( - 'sourceswiki' => 'sources.wikipedia.org', - 'quotewiki' => 'wikiquote.org', - 'textbookwiki' => 'wikibooks.org', - 'sep11wiki' => 'sep11.wikipedia.org', - 'metawiki' => 'meta.wikimedia.org', - 'commonswiki' => 'commons.wikimedia.org', - 'specieswiki' => 'species.wikimedia.org', - ); - - # Extra interwiki links that can't be in the intermap for some reason - $extraLinks = array( - array( 'm', 'http://meta.wikimedia.org/wiki/$1', 1 ), - array( 'meta', 'http://meta.wikimedia.org/wiki/$1', 1 ), - array( 'sep11', 'http://sep11.wikipedia.org/wiki/$1', 1 ), - ); - - # Language aliases, usually configured as redirects to the real wiki in apache - # Interlanguage links are made directly to the real wiki - # Something horrible happens if you forget to list an alias here, I can't - # remember what - $languageAliases = array( - 'zh-cn' => 'zh', - 'zh-tw' => 'zh', - 'dk' => 'da', - 'nb' => 'no', - ); - - # Special case prefix rewrites, for the benefit of Swedish which uses s:t - # as an abbreviation for saint - $prefixRewrites = array( - 'svwiki' => array( 's' => 'src' ), - ); - - # Construct a list of reserved prefixes - $reserved = array(); - foreach ( $langlist as $lang ) { - $reserved[$lang] = 1; - } - foreach ( $languageAliases as $alias => $lang ) { - $reserved[$alias] = 1; - } - foreach ( $sites as $site ) { - $reserved[$site->lateral] = 1; - } - - # Extract the intermap from meta - $intermap = Http::get( 'http://meta.wikimedia.org/w/index.php?title=Interwiki_map&action=raw', 30 ); - $lines = array_map( 'trim', explode( "\n", trim( $intermap ) ) ); - - if ( !$lines || count( $lines ) < 2 ) { - wfDie( "m:Interwiki_map not found" ); - } - - $iwArray = array(); - - foreach ( $lines as $line ) { - $matches = array(); - if ( preg_match( '/^\|\s*(.*?)\s*\|\|\s*(https?:\/\/.*?)\s*$/', $line, $matches ) ) { - $prefix = strtolower( $matches[1] ); - $url = $matches[2]; - if ( preg_match( '/(wikipedia|wiktionary|wikisource|wikiquote|wikibooks|wikimedia)\.org/', $url ) ) { - $local = 1; - } else { - $local = 0; - } - - if ( empty( $reserved[$prefix] ) ) { - $iwArray[$prefix] = array( "iw_prefix" => $prefix, "iw_url" => $url, "iw_local" => $local ); - } - } - } - - - - foreach ( $dblist as $db ) { - $sql = "-- Generated by rebuildInterwiki.php"; - if ( isset( $specials[$db] ) ) { - # Special wiki - # Has interwiki links and interlanguage links to wikipedia - - $host = $specials[$db]; - $sql .= "\n--$host\n\n"; - $sql .= "USE $db;\n" . - "TRUNCATE TABLE interwiki;\n" . - "INSERT INTO interwiki (iw_prefix, iw_url, iw_local) VALUES \n"; - $first = true; - - # Intermap links - foreach ( $iwArray as $iwEntry ) { - $sql .= makeLink( $iwEntry, $first, $db ); - } - - # Links to multilanguage sites - foreach ( $sites as $targetSite ) { - $sql .= makeLink( array( $targetSite->lateral, $targetSite->getURL( 'en' ), 1 ), $first, $db ); - } - - # Interlanguage links to wikipedia - $sql .= makeLanguageLinks( $sites['wiki'], $first, $db ); - - # Extra links - foreach ( $extraLinks as $link ) { - $sql .= makeLink( $link, $first, $db ); - } - - $sql .= ";\n"; - } else { - # Find out which site this DB belongs to - $site = false; - foreach ( $sites as $candidateSite ) { - $suffix = $candidateSite->suffix; - if ( preg_match( "/(.*)$suffix$/", $db, $matches ) ) { - $site = $candidateSite; - break; - } - } - if ( !$site ) { - print "Invalid database $db\n"; - continue; - } - $lang = $matches[1]; - $host = "$lang." . $site->url; - $sql .= "\n--$host\n\n"; - - $sql .= "USE $db;\n" . - "TRUNCATE TABLE interwiki;\n" . - "INSERT INTO interwiki (iw_prefix,iw_url,iw_local) VALUES\n"; - $first = true; - - # Intermap links - foreach ( $iwArray as $iwEntry ) { - # Suppress links with the same name as the site - if ( ( $suffix == 'wiki' && $iwEntry['iw_prefix'] != 'wikipedia' ) || - ( $suffix != 'wiki' && $suffix != $iwEntry['iw_prefix'] ) ) - { - $sql .= makeLink( $iwEntry, $first, $db ); - } - } - - # Lateral links - foreach ( $sites as $targetSite ) { - # Suppress link to self - if ( $targetSite->suffix != $site->suffix ) { - $sql .= makeLink( array( $targetSite->lateral, $targetSite->getURL( $lang ), 1 ), $first, $db ); - } - } - - # Interlanguage links - $sql .= makeLanguageLinks( $site, $first, $db ); - - # w link within wikipedias - # Other sites already have it as a lateral link - if ( $site->suffix == "wiki" ) { - $sql .= makeLink( array( "w", "http://en.wikipedia.org/wiki/$1", 1 ), $first, $db ); - } - - # Extra links - foreach ( $extraLinks as $link ) { - $sql .= makeLink( $link, $first, $db ); - } - $sql .= ";\n"; - } - file_put_contents( "$destDir/$db.sql", $sql ); - } -} - -# ------------------------------------------------------------------------------------------ - -# Returns part of an INSERT statement, corresponding to all interlanguage links to a particular site -function makeLanguageLinks( &$site, &$first, $source ) { - global $langlist, $languageAliases; - - $sql = ""; - - # Actual languages with their own databases - foreach ( $langlist as $targetLang ) { - $sql .= makeLink( array( $targetLang, $site->getURL( $targetLang ), 1 ), $first, $source ); - } - - # Language aliases - foreach ( $languageAliases as $alias => $lang ) { - $sql .= makeLink( array( $alias, $site->getURL( $lang ), 1 ), $first, $source ); - } - return $sql; -} - -# Make SQL for a single link from an array -function makeLink( $entry, &$first, $source ) { - global $prefixRewrites; - - if ( isset( $prefixRewrites[$source] ) && isset( $prefixRewrites[$source][$entry[0]] ) ) { - $entry[0] = $prefixRewrites[$source][$entry[0]]; - } - - $sql = ""; - # Add comma - if ( $first ) { - $first = false; - } else { - $sql .= ",\n"; - } - $dbr = wfGetDB( DB_SLAVE ); - $sql .= "(" . $dbr->makeList( $entry ) . ")"; - return $sql; -} diff --git a/maintenance/rebuildInterwiki.php b/maintenance/rebuildInterwiki.php index 01a74ea26c..c715597527 100644 --- a/maintenance/rebuildInterwiki.php +++ b/maintenance/rebuildInterwiki.php @@ -9,21 +9,266 @@ * @ingroup Wikimedia */ -/** */ -$oldCwd = getcwd(); - -$optionsWithArgs = array( "d" ); -require_once( dirname( __FILE__ ) . '/commandLine.inc' ); -require( "rebuildInterwiki.inc" ); -chdir( $oldCwd ); - -# Output -if ( isset( $options['d'] ) ) { - $destDir = $options['d']; -} else { - $destDir = '/home/wikipedia/conf/interwiki/sql'; +/** + * @todo document + * @ingroup Maintenance + */ +class Site { + var $suffix, $lateral, $url; + + function __construct( $s, $l, $u ) { + $this->suffix = $s; + $this->lateral = $l; + $this->url = $u; + } + + function getURL( $lang ) { + $xlang = str_replace( '_', '-', $lang ); + return "http://$xlang.{$this->url}/wiki/\$1"; + } +} + +require_once( dirname( __FILE__ ) . '/Maintenance.php' ); + +class RebuildInterwiki extends Maintenance { + public function __construct() { + parent::__construct(); + $this->mDescription = "Protect or unprotect an article from the command line."; + $this->addOption( 'langlist', 'File with one language code per line', false, true ); + $this->addOption( 'dblist', 'File with one db per line', false, true ); + $this->addOption( 'd', 'Output folder', false, true ); + } + + function execute() { + # List of language prefixes likely to be found in multi-language sites + $this->langlist = array_map( "trim", file( $this->getOption( 'langlist', "/home/wikipedia/common/langlist" ) ) ); + + # List of all database names + $this->dblist = array_map( "trim", file( $this->getOption( 'dblist', "/home/wikipedia/common/all.dblist" ) ) ); + + $this->makeInterwikiSQL( $this->getOption( 'd', '/home/wikipedia/conf/interwiki/sql' ) ); + } + + function makeInterwikiSQL( $destDir ) { + $this->output( "Making new interwiki SQL files in $destDir\n" ); + + # Multi-language sites + # db suffix => db suffix, iw prefix, hostname + $sites = array( + 'wiki' => new Site( 'wiki', 'w', 'wikipedia.org' ), + 'wiktionary' => new Site( 'wiktionary', 'wikt', 'wiktionary.org' ), + 'wikiquote' => new Site( 'wikiquote', 'q', 'wikiquote.org' ), + 'wikibooks' => new Site( 'wikibooks', 'b', 'wikibooks.org' ), + 'wikinews' => new Site( 'wikinews', 'n', 'wikinews.org' ), + 'wikisource' => new Site( 'wikisource', 's', 'wikisource.org' ), + 'wikimedia' => new Site( 'wikimedia', 'chapter', 'wikimedia.org' ), + 'wikiversity' => new Site( 'wikiversity', 'v', 'wikiversity.org' ), + ); + + # Special-case hostnames + $specials = array( + 'sourceswiki' => 'sources.wikipedia.org', + 'quotewiki' => 'wikiquote.org', + 'textbookwiki' => 'wikibooks.org', + 'sep11wiki' => 'sep11.wikipedia.org', + 'metawiki' => 'meta.wikimedia.org', + 'commonswiki' => 'commons.wikimedia.org', + 'specieswiki' => 'species.wikimedia.org', + ); + + # Extra interwiki links that can't be in the intermap for some reason + $extraLinks = array( + array( 'm', 'http://meta.wikimedia.org/wiki/$1', 1 ), + array( 'meta', 'http://meta.wikimedia.org/wiki/$1', 1 ), + array( 'sep11', 'http://sep11.wikipedia.org/wiki/$1', 1 ), + ); + + # Language aliases, usually configured as redirects to the real wiki in apache + # Interlanguage links are made directly to the real wiki + # Something horrible happens if you forget to list an alias here, I can't + # remember what + $this->languageAliases = array( + 'zh-cn' => 'zh', + 'zh-tw' => 'zh', + 'dk' => 'da', + 'nb' => 'no', + ); + + # Special case prefix rewrites, for the benefit of Swedish which uses s:t + # as an abbreviation for saint + $this->prefixRewrites = array( + 'svwiki' => array( 's' => 'src' ), + ); + + # Construct a list of reserved prefixes + $reserved = array(); + foreach ( $this->langlist as $lang ) { + $reserved[$lang] = 1; + } + foreach ( $this->languageAliases as $alias => $lang ) { + $reserved[$alias] = 1; + } + foreach ( $sites as $site ) { + $reserved[$site->lateral] = 1; + } + + # Extract the intermap from meta + $intermap = Http::get( 'http://meta.wikimedia.org/w/index.php?title=Interwiki_map&action=raw', 30 ); + $lines = array_map( 'trim', explode( "\n", trim( $intermap ) ) ); + + if ( !$lines || count( $lines ) < 2 ) { + wfDie( "m:Interwiki_map not found" ); + } + + $iwArray = array(); + + foreach ( $lines as $line ) { + $matches = array(); + if ( preg_match( '/^\|\s*(.*?)\s*\|\|\s*(https?:\/\/.*?)\s*$/', $line, $matches ) ) { + $prefix = strtolower( $matches[1] ); + $url = $matches[2]; + if ( preg_match( '/(wikipedia|wiktionary|wikisource|wikiquote|wikibooks|wikimedia)\.org/', $url ) ) { + $local = 1; + } else { + $local = 0; + } + + if ( empty( $reserved[$prefix] ) ) { + $iwArray[$prefix] = array( "iw_prefix" => $prefix, "iw_url" => $url, "iw_local" => $local ); + } + } + } + + foreach ( $this->dblist as $db ) { + $sql = "-- Generated by rebuildInterwiki.php"; + if ( isset( $specials[$db] ) ) { + # Special wiki + # Has interwiki links and interlanguage links to wikipedia + + $host = $specials[$db]; + $sql .= "\n--$host\n\n"; + $sql .= "USE $db;\n" . + "TRUNCATE TABLE interwiki;\n" . + "INSERT INTO interwiki (iw_prefix, iw_url, iw_local) VALUES \n"; + $first = true; + + # Intermap links + foreach ( $iwArray as $iwEntry ) { + $sql .= $this->makeLink( $iwEntry, $first, $db ); + } + + # Links to multilanguage sites + foreach ( $sites as $targetSite ) { + $sql .= $this->makeLink( array( $targetSite->lateral, $targetSite->getURL( 'en' ), 1 ), $first, $db ); + } + + # Interlanguage links to wikipedia + $sql .= $this->makeLanguageLinks( $sites['wiki'], $first, $db ); + + # Extra links + foreach ( $extraLinks as $link ) { + $sql .= $this->makeLink( $link, $first, $db ); + } + + $sql .= ";\n"; + } else { + # Find out which site this DB belongs to + $site = false; + foreach ( $sites as $candidateSite ) { + $suffix = $candidateSite->suffix; + if ( preg_match( "/(.*)$suffix$/", $db, $matches ) ) { + $site = $candidateSite; + break; + } + } + if ( !$site ) { + print "Invalid database $db\n"; + continue; + } + $lang = $matches[1]; + $host = "$lang." . $site->url; + $sql .= "\n--$host\n\n"; + + $sql .= "USE $db;\n" . + "TRUNCATE TABLE interwiki;\n" . + "INSERT INTO interwiki (iw_prefix,iw_url,iw_local) VALUES\n"; + $first = true; + + # Intermap links + foreach ( $iwArray as $iwEntry ) { + # Suppress links with the same name as the site + if ( ( $suffix == 'wiki' && $iwEntry['iw_prefix'] != 'wikipedia' ) || + ( $suffix != 'wiki' && $suffix != $iwEntry['iw_prefix'] ) ) + { + $sql .= $this->makeLink( $iwEntry, $first, $db ); + } + } + + # Lateral links + foreach ( $sites as $targetSite ) { + # Suppress link to self + if ( $targetSite->suffix != $site->suffix ) { + $sql .= $this->makeLink( array( $targetSite->lateral, $targetSite->getURL( $lang ), 1 ), $first, $db ); + } + } + + # Interlanguage links + $sql .= $this->makeLanguageLinks( $site, $first, $db ); + + # w link within wikipedias + # Other sites already have it as a lateral link + if ( $site->suffix == "wiki" ) { + $sql .= $this->makeLink( array( "w", "http://en.wikipedia.org/wiki/$1", 1 ), $first, $db ); + } + + # Extra links + foreach ( $extraLinks as $link ) { + $sql .= $this->makeLink( $link, $first, $db ); + } + $sql .= ";\n"; + } + file_put_contents( "$destDir/$db.sql", $sql ); + } + } + + # ------------------------------------------------------------------------------------------ + + # Returns part of an INSERT statement, corresponding to all interlanguage links to a particular site + function makeLanguageLinks( &$site, &$first, $source ) { + $sql = ""; + + # Actual languages with their own databases + foreach ( $this->langlist as $targetLang ) { + $sql .= $this->makeLink( array( $targetLang, $site->getURL( $targetLang ), 1 ), $first, $source ); + } + + # Language aliases + foreach ( $this->languageAliases as $alias => $lang ) { + $sql .= $this->makeLink( array( $alias, $site->getURL( $lang ), 1 ), $first, $source ); + } + return $sql; + } + + # Make SQL for a single link from an array + function makeLink( $entry, &$first, $source ) { + + if ( isset( $this->prefixRewrites[$source] ) && isset($entry[0]) && isset( $this->prefixRewrites[$source][$entry[0]] ) ) { + $entry[0] = $this->prefixRewrites[$source][$entry[0]]; + } + + $sql = ""; + # Add comma + if ( $first ) { + $first = false; + } else { + $sql .= ",\n"; + } + $dbr = wfGetDB( DB_SLAVE ); + $sql .= "(" . $dbr->makeList( $entry ) . ")"; + return $sql; + } } -echo "Making new interwiki SQL files in $destDir\n"; -makeInterwikiSQL( $destDir ); +$maintClass = "RebuildInterwiki"; +require_once( DO_MAINTENANCE ); -- 2.20.1