Add the Postgres slash search fix to existing installations.
[lhc/web/wiklou.git] / maintenance / generateSitemap.php
index cb43594..657437b 100644 (file)
@@ -4,15 +4,14 @@ define( 'GS_TALK', -1 );
 /**
  * Creates a Google sitemap for the site
  *
- * @package MediaWiki
- * @subpackage Maintenance
+ * @addtogroup Maintenance
  *
  * @copyright Copyright © 2005, Ævar Arnfjörð Bjarmason
  * @copyright Copyright © 2005, Jens Frank <jeluf@gmx.de>
  * @copyright Copyright © 2005, Brion Vibber <brion@pobox.com>
  *
- * @link http://www.google.com/webmasters/sitemaps/docs/en/about.html
- * @link http://www.google.com/schemas/sitemap/0.84/sitemap.xsd
+ * @see http://www.google.com/webmasters/sitemaps/docs/en/about.html
+ * @see http://www.google.com/schemas/sitemap/0.84/sitemap.xsd
  *
  * @license http://www.gnu.org/copyleft/gpl.html GNU General Public License 2.0 or later
  */
@@ -35,7 +34,7 @@ class GenerateSitemap {
         * @var int
         */
        var $size_limit;
-       
+
        /**
         * The path to prepend to the filename
         *
@@ -49,14 +48,14 @@ class GenerateSitemap {
         * @var string
         */
        var $path;
-       
+
        /**
         * Whether or not to use compression
         *
         * @var bool
         */
        var $compress;
-       
+
        /**
         * The number of entries to save in each sitemap file
         *
@@ -113,14 +112,14 @@ class GenerateSitemap {
         * @var object
         */
        var $dbr;
-       
+
        /**
         * A resource pointing to the sitemap index file
         *
         * @var resource
         */
        var $findex;
-       
+
 
        /**
         * A resource pointing to a sitemap file
@@ -144,20 +143,19 @@ class GenerateSitemap {
         * @param string $path The path to append to the domain name
         * @param bool $compress Whether to compress the sitemap files
         */
-       function GenerateSitemap( $fspath, $path, $compress ) {
-               global $wgDBname, $wgScriptPath;
-               
+       function GenerateSitemap( $fspath, $compress ) {
+               global $wgScriptPath;
+
                $this->url_limit = 50000;
                $this->size_limit = pow( 2, 20 ) * 10;
                $this->fspath = isset( $fspath ) ? $fspath : '';
-               $this->path = isset( $path ) ? $path : $wgScriptPath;
                $this->compress = $compress;
 
                $this->stderr = fopen( 'php://stderr', 'wt' );
-               $this->dbr =& wfGetDB( DB_SLAVE );
+               $this->dbr = wfGetDB( DB_SLAVE );
                $this->generateNamespaces();
                $this->timestamp = wfTimestamp( TS_ISO_8601, wfTimestampNow() );
-               $this->findex = fopen( "{$this->fspath}sitemap-index-$wgDBname.xml", 'wb' );
+               $this->findex = fopen( "{$this->fspath}sitemap-index-" . wfWikiID() . ".xml", 'wb' );
        }
 
        /**
@@ -165,7 +163,7 @@ class GenerateSitemap {
         */
        function generateNamespaces() {
                $fname = 'GenerateSitemap::generateNamespaces';
-               
+
                $res = $this->dbr->select( 'page',
                        array( 'page_namespace' ),
                        array(),
@@ -196,7 +194,7 @@ class GenerateSitemap {
         * If the namespace isn't listed on the priority list return the
         * default priority for the namespace, varies depending on whether it's
         * a talkpage or not.
-        * 
+        *
         * @param int $namespace The namespace to get the priority for
         *
         * @return string
@@ -216,10 +214,9 @@ class GenerateSitemap {
                $fname = 'GenerateSitemap::getPageRes';
 
                return $this->dbr->select( 'page',
-                       array( 
+                       array(
                                'page_namespace',
                                'page_title',
-                               'page_is_redirect',
                                'page_touched',
                        ),
                        array( 'page_namespace' => $namespace ),
@@ -233,17 +230,17 @@ class GenerateSitemap {
         * @access public
         */
        function main() {
-               global $wgDBname, $wgContLang;
+               global $wgContLang;
 
                fwrite( $this->findex, $this->openIndex() );
-               
+
                foreach ( $this->namespaces as $namespace ) {
                        $res = $this->getPageRes( $namespace );
                        $this->file = false;
                        $this->generateLimit( $namespace );
                        $length = $this->limit[0];
                        $i = $smcount = 0;
-                       
+
                        $fns = $wgContLang->getFormattedNsText( $namespace );
                        $this->debug( "$namespace ($fns)" );
                        while ( $row = $this->dbr->fetchObject( $res ) ) {
@@ -265,6 +262,16 @@ class GenerateSitemap {
                                $entry = $this->fileEntry( $title->getFullURL(), $date, $this->priority( $namespace ) );
                                $length += strlen( $entry );
                                $this->write( $this->file, $entry );
+                               // generate pages for language variants
+                               if($wgContLang->hasVariants()){
+                                       $variants = $wgContLang->getVariants();
+                                       foreach($variants as $vCode){
+                                               if($vCode==$wgContLang->getCode()) continue; // we don't want default variant
+                                               $entry = $this->fileEntry( $title->getFullURL('',$vCode), $date, $this->priority( $namespace ) );
+                                               $length += strlen( $entry );
+                                               $this->write( $this->file, $entry );
+                                       }
+                               }
                        }
                        if ( $this->file ) {
                                $this->write( $this->file, $this->closeFile() );
@@ -283,14 +290,14 @@ class GenerateSitemap {
        function open( $file, $flags ) {
                return $this->compress ? gzopen( $file, $flags ) : fopen( $file, $flags );
        }
-       
+
        /**
         * gzwrite() / fwrite() wrapper
         */
        function write( &$handle, $str ) {
                if ( $this->compress )
                        gzwrite( $handle, $str );
-               else 
+               else
                        fwrite( $handle, $str );
        }
 
@@ -315,11 +322,8 @@ class GenerateSitemap {
         * @return string
         */
        function sitemapFilename( $namespace, $count ) {
-               global $wgDBname;
-               
                $ext = $this->compress ? '.gz' : '';
-               
-               return "sitemap-$wgDBname-NS_$namespace-$count.xml$ext";
+               return "sitemap-".wfWikiID()."-NS_$namespace-$count.xml$ext";
        }
 
        /**
@@ -363,12 +367,10 @@ class GenerateSitemap {
         * @return string
         */
        function indexEntry( $filename ) {
-               global $wgServer;
-               
                return
                        "\t<sitemap>\n" .
-                       "\t\t<loc>$wgServer{$this->path}/$filename</log>\n" .
-                       "\t\t<lastmod>{$this->timestamp}</lastmod>\n" . 
+                       "\t\t<loc>$filename</loc>\n" .
+                       "\t\t<lastmod>{$this->timestamp}</lastmod>\n" .
                        "\t</sitemap>\n";
        }
 
@@ -434,7 +436,7 @@ class GenerateSitemap {
         */
        function generateLimit( $namespace ) {
                $title = Title::makeTitle( $namespace, str_repeat( "\xf0\xa8\xae\x81", 63 ) . "\xe5\x96\x83" );
-               
+
                $this->limit = array(
                        strlen( $this->openFile() ),
                        strlen( $this->fileEntry( $title->getFullUrl(), wfTimestamp( TS_ISO_8601, wfTimestamp() ), $this->priority( $namespace ) ) ),
@@ -443,23 +445,30 @@ class GenerateSitemap {
        }
 }
 
-if ( in_array( '--help', $argv ) )
-       die(
-               "Usage: php generateSitemap.php [host] [options]\n" .
-               "\thost = hostname\n" .
-               "\toptions:\n" .
-               "\t\t--help\tshow this message\n" .
-               "\t\t--fspath\tThe file system path to save to, e.g /tmp/sitemap/\n" .
-               "\t\t--path\tThe http path to use, e.g. /wiki\n" .
-               "\t\t--compress=[yes|no]\tcompress the sitemap files, default yes\n"
-       );
+if ( in_array( '--help', $argv ) ) {
+       echo <<<EOT
+Usage: php generateSitemap.php [options]
+       --help                  show this message
+
+       --fspath=<path>         The file system path to save to, e.g /tmp/sitemap/
 
-if ( isset( $argv[1] ) && strpos( $argv[1], '--' ) !== 0 )
-       $_SERVER['SERVER_NAME'] = $argv[1];
+       --server=<server>       The protocol and server name to use in URLs, e.g.
+               http://en.wikipedia.org. This is sometimes necessary because
+               server name detection may fail in command line scripts.
 
-$optionsWithArgs = array( 'fspath', 'path', 'compress' );
-require_once 'commandLine.inc';
+       --compress=[yes|no]     compress the sitemap files, default yes
 
-$gs = new GenerateSitemap( @$options['fspath'], @$options['path'], @$options['compress'] !== 'no' );
+EOT;
+       die( -1 );
+}
+
+$optionsWithArgs = array( 'fspath', 'server', 'compress' );
+require_once( dirname( __FILE__ ) . '/commandLine.inc' );
+
+if ( isset( $options['server'] ) ) {
+       $wgServer = $options['server'];
+}
+
+$gs = new GenerateSitemap( @$options['fspath'], @$options['compress'] !== 'no' );
 $gs->main();
-?>
+