somewhatdirtyhack: build constant slightly compact database of interwiki prefixes

diff --git a/maintenance/dumpInterwiki.inc b/maintenance/dumpInterwiki.inc
new file mode 100644 (file)
index 0000000..5e54bea
--- /dev/null
+++ b/maintenance/dumpInterwiki.inc
@@ -0,0 +1,219 @@
+<?php
+/**
+ * Rebuild interwiki table using the file on meta and the language list
+ * Wikimedia specific!
+ *
+ * @todo document
+ * @package MediaWiki
+ * @subpackage Maintenance
+ */
+
+/** */
+
+/**
+ * @todo document
+ * @package MediaWiki
+ * @subpackage Maintenance
+ */
+class Site {
+       var $suffix, $lateral, $url;
+
+       function Site( $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";
+       }
+}
+
+function getRebuildInterwikiDump() {
+       global $langlist, $languageAliases, $prefixRewrites, $wgDBname;
+
+       # 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' ),
+       );
+
+       # 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',
+       );
+
+       # 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 = wfGetHTTP( '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();
+        # Global iterwiki map
+       foreach ( $lines as $line ) {
+               if ( preg_match( '/^\|\s*(.*?)\s*\|\|\s*(.*?)\s*$/', $line, $matches ) ) {
+                       $prefix = strtolower( $matches[1] );
+                       $url = $matches[2];
+                       if ( preg_match( '/(wikipedia|wiktionary|wikisource|wikiquote|wikibooks)\.org/', $url ) ) {
+                               $local = 1;
+                       } else {
+                               $local = 0;
+                       }
+
+                       if ( empty( $reserved[$prefix] ) ) {
+                           $imap  = array( "iw_prefix" => $prefix, "iw_url" => $url, "iw_local" => $local );
+                            makeLink ($imap, "__global");
+                       }
+               }
+       }
+
+        # Exclude Wikipedia for Wikipedia
+        makeLink ( array ('iw_prefix' => 'wikipedia', 'is_url' => null ), "_wiki" );
+        
+        #Multilanguage sites
+        foreach ($sites as $site) 
+            $sql .= makeLanguageLinks ( $site, "_".$site->suffix );
+
+
+        foreach ( $dblist as $db ) {
+               if ( isset( $specials[$db] ) ) {
+                       # Special wiki
+                       # Has interwiki links and interlanguage links to wikipedia
+
+                        makeLink( array( 'iw_prefix' => $db, 'iw_url' => "wiki"), "__sites" );
+                       # Links to multilanguage sites
+                       foreach ( $sites as $targetSite ) {
+                               makeLink( array( 'iw_prefix' => $targetSite->lateral, 
+                                                    'iw_url' =>$targetSite->getURL( 'en' ),
+                                                    'iw_local' => 1 ), $db );
+                       }
+
+               } 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;
+                               }
+                       }
+                        makeLink( array( 'iw_prefix' => $db, 'iw_url' => $site->suffix), "__sites" );
+                       if ( !$site ) {
+                               print "Invalid database $db\n";
+                               continue;
+                       }
+                       $lang = $matches[1];
+                       $host = "$lang." . $site->url;
+
+                       # Lateral links
+                       foreach ( $sites as $targetSite ) {
+                               if ( $targetSite->suffix != $site->suffix ) {
+                                       makeLink( array( 'iw_prefix' => $targetSite->lateral, 
+                                                            'iw_url' => $targetSite->getURL( $lang ), 
+                                                            'iw_local' => 1 ), $db );
+                               }
+                       }
+
+                       if ( $site->suffix == "wiki" ) {
+                               makeLink( array('iw_prefix' => 'w', 
+                                                'iw_url' => "http://en.wikipedia.org/wiki/$1", 
+                                                'iw_local' => 1), $db );
+                       }
+
+               }
+       }
+       foreach ( $extraLinks as $link )
+               makeLink( $link, "__global" );
+}
+
+# ------------------------------------------------------------------------------------------
+
+# Returns part of an INSERT statement, corresponding to all interlanguage links to a particular site
+function makeLanguageLinks( &$site, $source ) {
+       global $langlist, $languageAliases;
+       # Actual languages with their own databases
+       foreach ( $langlist as $targetLang ) {
+               makeLink( array( $targetLang, $site->getURL( $targetLang ), 1 ), $source );
+       }
+
+       # Language aliases
+       foreach ( $languageAliases as $alias => $lang ) {
+               makeLink( array( $alias, $site->getURL( $lang ), 1 ), $source );
+       }
+}
+
+function makeLink( $entry, $source ) {
+       global $prefixRewrites, $dbFile;
+       if ( isset( $prefixRewrites[$source] ) && isset( $prefixRewrites[$source][$entry[0]] ) )
+               $entry[0] = $prefixRewrites[$source][$entry[0]];
+        if (!array_key_exists("iw_prefix",$entry))
+            $entry = array("iw_prefix" => $entry[0], "iw_url" => $entry[1], "iw_local" => $entry[2]);
+        if ( array_key_exists($source,$prefixRewrites) && 
+                array_key_exists($entry['iw_prefix'],$prefixRewrites[$source]))
+                    $entry['iw_prefix'] = $prefixRewrites[$source][$entry['iw_prefix']];
+        if ($dbFile)
+            dba_insert("{$source}:{$entry['iw_prefix']}", trim("{$entry['iw_local']} {$entry['iw_url']}"),$dbFile);
+        else
+            print "{$source}:{$entry['iw_prefix']} {$entry['iw_url']} {$entry['iw_local']}\n";
+
+    }
+
+?>

diff --git a/maintenance/dumpInterwiki.php b/maintenance/dumpInterwiki.php
new file mode 100644 (file)
index 0000000..411260a
--- /dev/null
+++ b/maintenance/dumpInterwiki.php
@@ -0,0 +1,25 @@
+<?php
+/**
+ * Rebuild interwiki table using the file on meta and the language list
+ * Wikimedia specific!
+ * @todo document
+ * @package MediaWiki
+ * @subpackage Maintenance
+ */
+
+/** */
+$oldCwd = getcwd();
+
+$optionsWithArgs = array( "o" );
+include_once( "commandLine.inc" );
+include_once( "dumpInterwiki.inc" );
+chdir( $oldCwd );
+
+# Output
+if ( isset( $options['o'] ) ) {
+    # To database specified with -o
+    $dbFile = dba_open( $options['o'], "n", "cdb_make" );
+} 
+
+getRebuildInterwikiDump();
+?>