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