removed wikisql user
[lhc/web/wiklou.git] / maintenance / dumpInterwiki.inc
1 <?php
2 /**
3 * Rebuild interwiki table using the file on meta and the language list
4 * Wikimedia specific!
5 *
6 * @todo document
7 * @package MediaWiki
8 * @subpackage Maintenance
9 */
10
11 /** */
12
13 /**
14 * @todo document
15 * @package MediaWiki
16 * @subpackage Maintenance
17 */
18 class Site {
19 var $suffix, $lateral, $url;
20
21 function Site( $s, $l, $u ) {
22 $this->suffix = $s;
23 $this->lateral = $l;
24 $this->url = $u;
25 }
26
27 function getURL( $lang ) {
28 $xlang = str_replace( '_', '-', $lang );
29 return "http://$xlang.{$this->url}/wiki/\$1";
30 }
31 }
32
33 function getRebuildInterwikiDump() {
34 global $langlist, $languageAliases, $prefixRewrites;
35
36 # Multi-language sites
37 # db suffix => db suffix, iw prefix, hostname
38 $sites = array(
39 'wiki' => new Site( 'wiki', 'w', 'wikipedia.org' ),
40 'wiktionary' => new Site( 'wiktionary', 'wikt', 'wiktionary.org' ),
41 'wikiquote' => new Site( 'wikiquote', 'q', 'wikiquote.org' ),
42 'wikibooks' => new Site( 'wikibooks', 'b', 'wikibooks.org' ),
43 'wikinews' => new Site( 'wikinews', 'n', 'wikinews.org' ),
44 'wikisource' => new Site( 'wikisource', 's', 'wikisource.org' ),
45 'wikimedia' => new Site( 'wikimedia', 'chapter', 'wikimedia.org' ),
46 'wikiversity' => new Site( 'wikiversity', 'v', 'wikiversity.org' ),
47 );
48
49 # List of language prefixes likely to be found in multi-language sites
50 $langlist = array_map( "trim", file( "/home/wikipedia/common/langlist" ) );
51
52 # List of all database names
53 $dblist = array_map( "trim", file( "/home/wikipedia/common/all.dblist" ) );
54
55 # Special-case hostnames
56 $specials = array(
57 'sourceswiki' => 'sources.wikipedia.org',
58 'quotewiki' => 'wikiquote.org',
59 'textbookwiki' => 'wikibooks.org',
60 'sep11wiki' => 'sep11.wikipedia.org',
61 'metawiki' => 'meta.wikimedia.org',
62 'commonswiki' => 'commons.wikimedia.org',
63 );
64
65 # Extra interwiki links that can't be in the intermap for some reason
66 $extraLinks = array(
67 array( 'm', 'http://meta.wikimedia.org/wiki/$1', 1 ),
68 array( 'meta', 'http://meta.wikimedia.org/wiki/$1', 1 ),
69 array( 'sep11', 'http://sep11.wikipedia.org/wiki/$1', 1 ),
70 );
71
72 # Language aliases, usually configured as redirects to the real wiki in apache
73 # Interlanguage links are made directly to the real wiki
74 # Something horrible happens if you forget to list an alias here, I can't
75 # remember what
76 $languageAliases = array(
77 'zh-cn' => 'zh',
78 'zh-tw' => 'zh',
79 'dk' => 'da',
80 'nb' => 'no',
81 );
82
83 # Special case prefix rewrites, for the benefit of Swedish which uses s:t
84 # as an abbreviation for saint
85 $prefixRewrites = array(
86 'svwiki' => array ( 's' => 'src'),
87 );
88
89 # Construct a list of reserved prefixes
90 $reserved = array();
91 foreach ( $langlist as $lang ) {
92 $reserved[$lang] = 1;
93 }
94 foreach ( $languageAliases as $alias => $lang ) {
95 $reserved[$alias] = 1;
96 }
97 foreach( $sites as $site ) {
98 $reserved[$site->lateral] = 1;
99 }
100
101 # Extract the intermap from meta
102 $intermap = wfGetHTTP( 'http://meta.wikimedia.org/w/index.php?title=Interwiki_map&action=raw', 30 );
103 $lines = array_map( 'trim', explode( "\n", trim( $intermap ) ) );
104
105 if ( !$lines || count( $lines ) < 2 ) {
106 wfDie( "m:Interwiki_map not found" );
107 }
108
109 $iwArray = array();
110 # Global iterwiki map
111 foreach ( $lines as $line ) {
112 if ( preg_match( '/^\|\s*(.*?)\s*\|\|\s*(.*?)\s*$/', $line, $matches ) ) {
113 $prefix = strtolower( $matches[1] );
114 $url = $matches[2];
115 if ( preg_match( '/(wikipedia|wiktionary|wikisource|wikiquote|wikibooks|wikimedia)\.org/', $url ) ) {
116 $local = 1;
117 } else {
118 $local = 0;
119 }
120
121 if ( empty( $reserved[$prefix] ) ) {
122 $imap = array( "iw_prefix" => $prefix, "iw_url" => $url, "iw_local" => $local );
123 makeLink ($imap, "__global");
124 }
125 }
126 }
127
128 # Exclude Wikipedia for Wikipedia
129 makeLink ( array ('iw_prefix' => 'wikipedia', 'is_url' => null ), "_wiki" );
130
131 #Multilanguage sites
132 foreach ($sites as $site)
133 $sql .= makeLanguageLinks ( $site, "_".$site->suffix );
134
135
136 foreach ( $dblist as $db ) {
137 if ( isset( $specials[$db] ) ) {
138 # Special wiki
139 # Has interwiki links and interlanguage links to wikipedia
140
141 makeLink( array( 'iw_prefix' => $db, 'iw_url' => "wiki"), "__sites" );
142 # Links to multilanguage sites
143 foreach ( $sites as $targetSite ) {
144 makeLink( array( 'iw_prefix' => $targetSite->lateral,
145 'iw_url' =>$targetSite->getURL( 'en' ),
146 'iw_local' => 1 ), $db );
147 }
148
149 } else {
150 # Find out which site this DB belongs to
151 $site = false;
152 foreach( $sites as $candidateSite ) {
153 $suffix = $candidateSite->suffix;
154 if ( preg_match( "/(.*)$suffix$/", $db, $matches ) ) {
155 $site = $candidateSite;
156 break;
157 }
158 }
159 makeLink( array( 'iw_prefix' => $db, 'iw_url' => $site->suffix), "__sites" );
160 if ( !$site ) {
161 print "Invalid database $db\n";
162 continue;
163 }
164 $lang = $matches[1];
165 $host = "$lang." . $site->url;
166
167 # Lateral links
168 foreach ( $sites as $targetSite ) {
169 if ( $targetSite->suffix != $site->suffix ) {
170 makeLink( array( 'iw_prefix' => $targetSite->lateral,
171 'iw_url' => $targetSite->getURL( $lang ),
172 'iw_local' => 1 ), $db );
173 }
174 }
175
176 if ( $site->suffix == "wiki" ) {
177 makeLink( array('iw_prefix' => 'w',
178 'iw_url' => "http://en.wikipedia.org/wiki/$1",
179 'iw_local' => 1), $db );
180 }
181
182 }
183 }
184 foreach ( $extraLinks as $link )
185 makeLink( $link, "__global" );
186 }
187
188 # ------------------------------------------------------------------------------------------
189
190 # Returns part of an INSERT statement, corresponding to all interlanguage links to a particular site
191 function makeLanguageLinks( &$site, $source ) {
192 global $langlist, $languageAliases;
193 # Actual languages with their own databases
194 foreach ( $langlist as $targetLang ) {
195 makeLink( array( $targetLang, $site->getURL( $targetLang ), 1 ), $source );
196 }
197
198 # Language aliases
199 foreach ( $languageAliases as $alias => $lang ) {
200 makeLink( array( $alias, $site->getURL( $lang ), 1 ), $source );
201 }
202 }
203
204 function makeLink( $entry, $source ) {
205 global $prefixRewrites, $dbFile;
206 if ( isset( $prefixRewrites[$source] ) && isset( $prefixRewrites[$source][$entry[0]] ) )
207 $entry[0] = $prefixRewrites[$source][$entry[0]];
208 if (!array_key_exists("iw_prefix",$entry))
209 $entry = array("iw_prefix" => $entry[0], "iw_url" => $entry[1], "iw_local" => $entry[2]);
210 if ( array_key_exists($source,$prefixRewrites) &&
211 array_key_exists($entry['iw_prefix'],$prefixRewrites[$source]))
212 $entry['iw_prefix'] = $prefixRewrites[$source][$entry['iw_prefix']];
213 if ($dbFile)
214 dba_insert("{$source}:{$entry['iw_prefix']}", trim("{$entry['iw_local']} {$entry['iw_url']}"),$dbFile);
215 else
216 print "{$source}:{$entry['iw_prefix']} {$entry['iw_url']} {$entry['iw_local']}\n";
217
218 }
219
220 ?>