maintenance/generateSitemap.php

   1 <?php
   2 define( 'GS_MAIN', -2 );
   3 define( 'GS_TALK', -1 );
   4 /**
   5  * Creates a sitemap for the site
   6  *
   7  * Copyright © 2005, Ævar Arnfjörð Bjarmason, Jens Frank <jeluf@gmx.de> and
   8  * Brion Vibber <brion@pobox.com>
   9  *
  10  * This program is free software; you can redistribute it and/or modify
  11  * it under the terms of the GNU General Public License as published by
  12  * the Free Software Foundation; either version 2 of the License, or
  13  * (at your option) any later version.
  14  *
  15  * This program is distributed in the hope that it will be useful,
  16  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  17  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
  18  * GNU General Public License for more details.
  19  *
  20  * You should have received a copy of the GNU General Public License along
  21  * with this program; if not, write to the Free Software Foundation, Inc.,
  22  * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
  23  * http://www.gnu.org/copyleft/gpl.html
  24  *
  25  * @file
  26  * @ingroup Maintenance
  27  * @see http://www.sitemaps.org/
  28  * @see http://www.sitemaps.org/schemas/sitemap/0.9/sitemap.xsd
  29  */
  30
  31 require_once( dirname( __FILE__ ) . '/Maintenance.php' );
  32
  33 class GenerateSitemap extends Maintenance {
  34         /**
  35          * The maximum amount of urls in a sitemap file
  36          *
  37          * @link http://www.sitemaps.org/schemas/sitemap/0.9/sitemap.xsd
  38          *
  39          * @var int
  40          */
  41         var $url_limit;
  42
  43         /**
  44          * The maximum size of a sitemap file
  45          *
  46          * @link http://www.sitemaps.org/faq.php#faq_sitemap_size
  47          *
  48          * @var int
  49          */
  50         var $size_limit;
  51
  52         /**
  53          * The path to prepend to the filename
  54          *
  55          * @var string
  56          */
  57         var $fspath;
  58
  59         /**
  60          * The path to append to the domain name
  61          *
  62          * @var string
  63          */
  64         var $path;
  65
  66         /**
  67          * Whether or not to use compression
  68          *
  69          * @var bool
  70          */
  71         var $compress;
  72
  73         /**
  74          * The number of entries to save in each sitemap file
  75          *
  76          * @var array
  77          */
  78         var $limit = array();
  79
  80         /**
  81          * Key => value entries of namespaces and their priorities
  82          *
  83          * @var array
  84          */
  85         var $priorities = array();
  86
  87         /**
  88          * A one-dimensional array of namespaces in the wiki
  89          *
  90          * @var array
  91          */
  92         var $namespaces = array();
  93
  94         /**
  95          * When this sitemap batch was generated
  96          *
  97          * @var string
  98          */
  99         var $timestamp;
 100
 101         /**
 102          * A database slave object
 103          *
 104          * @var object
 105          */
 106         var $dbr;
 107
 108         /**
 109          * A resource pointing to the sitemap index file
 110          *
 111          * @var resource
 112          */
 113         var $findex;
 114
 115
 116         /**
 117          * A resource pointing to a sitemap file
 118          *
 119          * @var resource
 120          */
 121         var $file;
 122
 123         /**
 124          * Constructor
 125          */
 126         public function __construct() {
 127                 parent::__construct();
 128                 $this->mDescription = "Creates a sitemap for the site";
 129                 $this->addOption( 'fspath', 'The file system path to save to, e.g. /tmp/sitemap' .
 130                                                                         "\n\t\tdefaults to current directory", false, true );
 131                 $this->addOption( 'server', "The protocol and server name to use in URLs, e.g.\n" .
 132                                                                         "\t\thttp://en.wikipedia.org. This is sometimes necessary because\n" .
 133                                                                         "\t\tserver name detection may fail in command line scripts.", false, true );
 134                 $this->addOption( 'compress', 'Compress the sitemap files, can take value yes|no, default yes', false, true );
 135         }
 136
 137         /**
 138          * Execute
 139          */
 140         public function execute() {
 141                 $this->setNamespacePriorities();
 142                 $this->url_limit = 50000;
 143                 $this->size_limit = pow( 2, 20 ) * 10;
 144                 $this->fspath = self::init_path( $this->getOption( 'fspath', getcwd() ) );
 145                 $this->compress = $this->getOption( 'compress', 'yes' ) !== 'no';
 146                 $this->dbr = wfGetDB( DB_SLAVE );
 147                 $this->generateNamespaces();
 148                 $this->timestamp = wfTimestamp( TS_ISO_8601, wfTimestampNow() );
 149                 $this->findex = fopen( "{$this->fspath}sitemap-index-" . wfWikiID() . ".xml", 'wb' );
 150                 $this->main();
 151         }
 152
 153         private function setNamespacePriorities() {
 154                 // Custom main namespaces
 155                 $this->priorities[GS_MAIN] = '0.5';
 156                 // Custom talk namesspaces
 157                 $this->priorities[GS_TALK] = '0.1';
 158                 // MediaWiki standard namespaces
 159                 $this->priorities[NS_MAIN] = '1.0';
 160                 $this->priorities[NS_TALK] = '0.1';
 161                 $this->priorities[NS_USER] = '0.5';
 162                 $this->priorities[NS_USER_TALK] = '0.1';
 163                 $this->priorities[NS_PROJECT] = '0.5';
 164                 $this->priorities[NS_PROJECT_TALK] = '0.1';
 165                 $this->priorities[NS_FILE] = '0.5';
 166                 $this->priorities[NS_FILE_TALK] = '0.1';
 167                 $this->priorities[NS_MEDIAWIKI] = '0.0';
 168                 $this->priorities[NS_MEDIAWIKI_TALK] = '0.1';
 169                 $this->priorities[NS_TEMPLATE] = '0.0';
 170                 $this->priorities[NS_TEMPLATE_TALK] = '0.1';
 171                 $this->priorities[NS_HELP] = '0.5';
 172                 $this->priorities[NS_HELP_TALK] = '0.1';
 173                 $this->priorities[NS_CATEGORY] = '0.5';
 174                 $this->priorities[NS_CATEGORY_TALK] = '0.1';
 175         }
 176
 177         /**
 178          * Create directory if it does not exist and return pathname with a trailing slash
 179          */
 180         private static function init_path( $fspath ) {
 181                 if ( !isset( $fspath ) ) {
 182                         return null;
 183                 }
 184                 # Create directory if needed
 185                 if ( $fspath && !is_dir( $fspath ) ) {
 186                         wfMkdirParents( $fspath ) or die( "Can not create directory $fspath.\n" );
 187                 }
 188
 189                 return realpath( $fspath ) . DIRECTORY_SEPARATOR ;
 190         }
 191
 192         /**
 193          * Generate a one-dimensional array of existing namespaces
 194          */
 195         function generateNamespaces() {
 196                 // Only generate for specific namespaces if $wgSitemapNamespaces is an array.
 197                 global $wgSitemapNamespaces;
 198                 if ( is_array( $wgSitemapNamespaces ) ) {
 199                         $this->namespaces = $wgSitemapNamespaces;
 200                         return;
 201                 }
 202
 203                 $res = $this->dbr->select( 'page',
 204                         array( 'page_namespace' ),
 205                         array(),
 206                         __METHOD__,
 207                         array(
 208                                 'GROUP BY' => 'page_namespace',
 209                                 'ORDER BY' => 'page_namespace',
 210                         )
 211                 );
 212
 213                 foreach ( $res as $row )
 214                         $this->namespaces[] = $row->page_namespace;
 215         }
 216
 217         /**
 218          * Get the priority of a given namespace
 219          *
 220          * @param $namespace Integer: the namespace to get the priority for
 221          * @return String
 222          */
 223         function priority( $namespace ) {
 224                 return isset( $this->priorities[$namespace] ) ? $this->priorities[$namespace] : $this->guessPriority( $namespace );
 225         }
 226
 227         /**
 228          * If the namespace isn't listed on the priority list return the
 229          * default priority for the namespace, varies depending on whether it's
 230          * a talkpage or not.
 231          *
 232          * @param $namespace Integer: the namespace to get the priority for
 233          * @return String
 234          */
 235         function guessPriority( $namespace ) {
 236                 return MWNamespace::isMain( $namespace ) ? $this->priorities[GS_MAIN] : $this->priorities[GS_TALK];
 237         }
 238
 239         /**
 240          * Return a database resolution of all the pages in a given namespace
 241          *
 242          * @param $namespace Integer: limit the query to this namespace
 243          * @return Resource
 244          */
 245         function getPageRes( $namespace ) {
 246                 return $this->dbr->select( 'page',
 247                         array(
 248                                 'page_namespace',
 249                                 'page_title',
 250                                 'page_touched',
 251                         ),
 252                         array( 'page_namespace' => $namespace ),
 253                         __METHOD__
 254                 );
 255         }
 256
 257         /**
 258          * Main loop
 259          */
 260         public function main() {
 261                 global $wgContLang;
 262
 263                 fwrite( $this->findex, $this->openIndex() );
 264
 265                 foreach ( $this->namespaces as $namespace ) {
 266                         $res = $this->getPageRes( $namespace );
 267                         $this->file = false;
 268                         $this->generateLimit( $namespace );
 269                         $length = $this->limit[0];
 270                         $i = $smcount = 0;
 271
 272                         $fns = $wgContLang->getFormattedNsText( $namespace );
 273                         $this->output( "$namespace ($fns)" );
 274                         foreach ( $res as $row ) {
 275                                 if ( $i++ === 0 || $i === $this->url_limit + 1 || $length + $this->limit[1] + $this->limit[2] > $this->size_limit ) {
 276                                         if ( $this->file !== false ) {
 277                                                 $this->write( $this->file, $this->closeFile() );
 278                                                 $this->close( $this->file );
 279                                         }
 280                                         $filename = $this->sitemapFilename( $namespace, $smcount++ );
 281                                         $this->file = $this->open( $this->fspath . $filename, 'wb' );
 282                                         $this->write( $this->file, $this->openFile() );
 283                                         fwrite( $this->findex, $this->indexEntry( $filename ) );
 284                                         $this->output( "\t$this->fspath$filename\n" );
 285                                         $length = $this->limit[0];
 286                                         $i = 1;
 287                                 }
 288                                 $title = Title::makeTitle( $row->page_namespace, $row->page_title );
 289                                 $date = wfTimestamp( TS_ISO_8601, $row->page_touched );
 290                                 $entry = $this->fileEntry( $title->getFullURL(), $date, $this->priority( $namespace ) );
 291                                 $length += strlen( $entry );
 292                                 $this->write( $this->file, $entry );
 293                                 // generate pages for language variants
 294                                 if ( $wgContLang->hasVariants() ) {
 295                                         $variants = $wgContLang->getVariants();
 296                                         foreach ( $variants as $vCode ) {
 297                                                 if ( $vCode == $wgContLang->getCode() ) continue; // we don't want default variant
 298                                                 $entry = $this->fileEntry( $title->getFullURL( '', $vCode ), $date, $this->priority( $namespace ) );
 299                                                 $length += strlen( $entry );
 300                                                 $this->write( $this->file, $entry );
 301                                         }
 302                                 }
 303                         }
 304                         if ( $this->file ) {
 305                                 $this->write( $this->file, $this->closeFile() );
 306                                 $this->close( $this->file );
 307                         }
 308                 }
 309                 fwrite( $this->findex, $this->closeIndex() );
 310                 fclose( $this->findex );
 311         }
 312
 313         /**
 314          * gzopen() / fopen() wrapper
 315          *
 316          * @return Resource
 317          */
 318         function open( $file, $flags ) {
 319                 return $this->compress ? gzopen( $file, $flags ) : fopen( $file, $flags );
 320         }
 321
 322         /**
 323          * gzwrite() / fwrite() wrapper
 324          */
 325         function write( &$handle, $str ) {
 326                 if ( $this->compress )
 327                         gzwrite( $handle, $str );
 328                 else
 329                         fwrite( $handle, $str );
 330         }
 331
 332         /**
 333          * gzclose() / fclose() wrapper
 334          */
 335         function close( &$handle ) {
 336                 if ( $this->compress )
 337                         gzclose( $handle );
 338                 else
 339                         fclose( $handle );
 340         }
 341
 342         /**
 343          * Get a sitemap filename
 344          *
 345          * @param $namespace Integer: the namespace
 346          * @param $count Integer: the count
 347          * @return String
 348          */
 349         function sitemapFilename( $namespace, $count ) {
 350                 $ext = $this->compress ? '.gz' : '';
 351                 return "sitemap-" . wfWikiID() . "-NS_$namespace-$count.xml$ext";
 352         }
 353
 354         /**
 355          * Return the XML required to open an XML file
 356          *
 357          * @return string
 358          */
 359         function xmlHead() {
 360                 return '<?xml version="1.0" encoding="UTF-8"?>' . "\n";
 361         }
 362
 363         /**
 364          * Return the XML schema being used
 365          *
 366          * @return String
 367          */
 368         function xmlSchema() {
 369                 return 'http://www.sitemaps.org/schemas/sitemap/0.9';
 370         }
 371
 372         /**
 373          * Return the XML required to open a sitemap index file
 374          *
 375          * @return String
 376          */
 377         function openIndex() {
 378                 return $this->xmlHead() . '<sitemapindex xmlns="' . $this->xmlSchema() . '">' . "\n";
 379         }
 380
 381         /**
 382          * Return the XML for a single sitemap indexfile entry
 383          *
 384          * @param $filename String: the filename of the sitemap file
 385          * @return String
 386          */
 387         function indexEntry( $filename ) {
 388                 return
 389                         "\t<sitemap>\n" .
 390                         "\t\t<loc>$filename</loc>\n" .
 391                         "\t\t<lastmod>{$this->timestamp}</lastmod>\n" .
 392                         "\t</sitemap>\n";
 393         }
 394
 395         /**
 396          * Return the XML required to close a sitemap index file
 397          *
 398          * @return String
 399          */
 400         function closeIndex() {
 401                 return "</sitemapindex>\n";
 402         }
 403
 404         /**
 405          * Return the XML required to open a sitemap file
 406          *
 407          * @return String
 408          */
 409         function openFile() {
 410                 return $this->xmlHead() . '<urlset xmlns="' . $this->xmlSchema() . '">' . "\n";
 411         }
 412
 413         /**
 414          * Return the XML for a single sitemap entry
 415          *
 416          * @param $url String: an RFC 2396 compliant URL
 417          * @param $date String: a ISO 8601 date
 418          * @param $priority String: a priority indicator, 0.0 - 1.0 inclusive with a 0.1 stepsize
 419          * @return String
 420          */
 421         function fileEntry( $url, $date, $priority ) {
 422                 return
 423                         "\t<url>\n" .
 424                         "\t\t<loc>$url</loc>\n" .
 425                         "\t\t<lastmod>$date</lastmod>\n" .
 426                         "\t\t<priority>$priority</priority>\n" .
 427                         "\t</url>\n";
 428         }
 429
 430         /**
 431          * Return the XML required to close sitemap file
 432          *
 433          * @return String
 434          */
 435         function closeFile() {
 436                 return "</urlset>\n";
 437         }
 438
 439         /**
 440          * Populate $this->limit
 441          */
 442         function generateLimit( $namespace ) {
 443                 $title = Title::makeTitle( $namespace, str_repeat( "\xf0\xa8\xae\x81", 63 ) . "\xe5\x96\x83" );
 444
 445                 $this->limit = array(
 446                         strlen( $this->openFile() ),
 447                         strlen( $this->fileEntry( $title->getFullUrl(), wfTimestamp( TS_ISO_8601, wfTimestamp() ), $this->priority( $namespace ) ) ),
 448                         strlen( $this->closeFile() )
 449                 );
 450         }
 451 }
 452
 453 $maintClass = "GenerateSitemap";
 454 require_once( DO_MAINTENANCE );