maintenance/generateSitemap.php

   1 <?php
   2 /**
   3  * Creates a sitemap for the site.
   4  *
   5  * Copyright © 2005, Ævar Arnfjörð Bjarmason, Jens Frank <jeluf@gmx.de> and
   6  * Brion Vibber <brion@pobox.com>
   7  *
   8  * This program is free software; you can redistribute it and/or modify
   9  * it under the terms of the GNU General Public License as published by
  10  * the Free Software Foundation; either version 2 of the License, or
  11  * (at your option) any later version.
  12  *
  13  * This program is distributed in the hope that it will be useful,
  14  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  15  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
  16  * GNU General Public License for more details.
  17  *
  18  * You should have received a copy of the GNU General Public License along
  19  * with this program; if not, write to the Free Software Foundation, Inc.,
  20  * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
  21  * http://www.gnu.org/copyleft/gpl.html
  22  *
  23  * @file
  24  * @ingroup Maintenance
  25  * @see http://www.sitemaps.org/
  26  * @see http://www.sitemaps.org/schemas/sitemap/0.9/sitemap.xsd
  27  */
  28
  29 require_once __DIR__ . '/Maintenance.php';
  30
  31 /**
  32  * Maintenance script that generates a sitemap for the site.
  33  *
  34  * @ingroup Maintenance
  35  */
  36 class GenerateSitemap extends Maintenance {
  37         const GS_MAIN = -2;
  38         const GS_TALK = -1;
  39
  40         /**
  41          * The maximum amount of urls in a sitemap file
  42          *
  43          * @link http://www.sitemaps.org/schemas/sitemap/0.9/sitemap.xsd
  44          *
  45          * @var int
  46          */
  47         public $url_limit;
  48
  49         /**
  50          * The maximum size of a sitemap file
  51          *
  52          * @link http://www.sitemaps.org/faq.php#faq_sitemap_size
  53          *
  54          * @var int
  55          */
  56         public $size_limit;
  57
  58         /**
  59          * The path to prepend to the filename
  60          *
  61          * @var string
  62          */
  63         public $fspath;
  64
  65         /**
  66          * The URL path to prepend to filenames in the index; should resolve to the same directory as $fspath
  67          *
  68          * @var string
  69          */
  70         public $urlpath;
  71
  72         /**
  73          * Whether or not to use compression
  74          *
  75          * @var bool
  76          */
  77         public $compress;
  78
  79         /**
  80          * Whether or not to include redirection pages
  81          *
  82          * @var bool
  83          */
  84         public $skipRedirects;
  85
  86         /**
  87          * The number of entries to save in each sitemap file
  88          *
  89          * @var array
  90          */
  91         public $limit = array();
  92
  93         /**
  94          * Key => value entries of namespaces and their priorities
  95          *
  96          * @var array
  97          */
  98         public $priorities = array();
  99
 100         /**
 101          * A one-dimensional array of namespaces in the wiki
 102          *
 103          * @var array
 104          */
 105         public $namespaces = array();
 106
 107         /**
 108          * When this sitemap batch was generated
 109          *
 110          * @var string
 111          */
 112         public $timestamp;
 113
 114         /**
 115          * A database slave object
 116          *
 117          * @var object
 118          */
 119         public $dbr;
 120
 121         /**
 122          * A resource pointing to the sitemap index file
 123          *
 124          * @var resource
 125          */
 126         public $findex;
 127
 128         /**
 129          * A resource pointing to a sitemap file
 130          *
 131          * @var resource
 132          */
 133         public $file;
 134
 135         /**
 136          * Identifier to use in filenames, default $wgDBname
 137          *
 138          * @var string
 139          */
 140         private $identifier;
 141
 142         /**
 143          * Constructor
 144          */
 145         public function __construct() {
 146                 parent::__construct();
 147                 $this->mDescription = "Creates a sitemap for the site";
 148                 $this->addOption( 'fspath', 'The file system path to save to, e.g. /tmp/sitemap; defaults to current directory', false, true );
 149                 $this->addOption( 'urlpath', 'The URL path corresponding to --fspath, prepended to filenames in the index; defaults to an empty string', false, true );
 150                 $this->addOption( 'compress', 'Compress the sitemap files, can take value yes|no, default yes', false, true );
 151                 $this->addOption( 'skip-redirects', 'Do not include redirecting articles in the sitemap' );
 152                 $this->addOption( 'identifier', 'What site identifier to use for the wiki, defaults to $wgDBname', false, true );
 153         }
 154
 155         /**
 156          * Execute
 157          */
 158         public function execute() {
 159                 $this->setNamespacePriorities();
 160                 $this->url_limit = 50000;
 161                 $this->size_limit = pow( 2, 20 ) * 10;
 162                 $this->fspath = self::init_path( $this->getOption( 'fspath', getcwd() ) );
 163                 $this->urlpath = $this->getOption( 'urlpath', "" );
 164                 if ( $this->urlpath !== "" && substr( $this->urlpath, -1 ) !== '/' ) {
 165                         $this->urlpath .= '/';
 166                 }
 167                 $this->identifier = $this->getOption( 'identifier', wfWikiID() );
 168                 $this->compress = $this->getOption( 'compress', 'yes' ) !== 'no';
 169                 $this->skipRedirects = $this->getOption( 'skip-redirects', false ) !== false;
 170                 $this->dbr = wfGetDB( DB_SLAVE );
 171                 $this->generateNamespaces();
 172                 $this->timestamp = wfTimestamp( TS_ISO_8601, wfTimestampNow() );
 173                 $this->findex = fopen( "{$this->fspath}sitemap-index-{$this->identifier}.xml", 'wb' );
 174                 $this->main();
 175         }
 176
 177         private function setNamespacePriorities() {
 178                 global $wgSitemapNamespacesPriorities;
 179
 180                 // Custom main namespaces
 181                 $this->priorities[self::GS_MAIN] = '0.5';
 182                 // Custom talk namesspaces
 183                 $this->priorities[self::GS_TALK] = '0.1';
 184                 // MediaWiki standard namespaces
 185                 $this->priorities[NS_MAIN] = '1.0';
 186                 $this->priorities[NS_TALK] = '0.1';
 187                 $this->priorities[NS_USER] = '0.5';
 188                 $this->priorities[NS_USER_TALK] = '0.1';
 189                 $this->priorities[NS_PROJECT] = '0.5';
 190                 $this->priorities[NS_PROJECT_TALK] = '0.1';
 191                 $this->priorities[NS_FILE] = '0.5';
 192                 $this->priorities[NS_FILE_TALK] = '0.1';
 193                 $this->priorities[NS_MEDIAWIKI] = '0.0';
 194                 $this->priorities[NS_MEDIAWIKI_TALK] = '0.1';
 195                 $this->priorities[NS_TEMPLATE] = '0.0';
 196                 $this->priorities[NS_TEMPLATE_TALK] = '0.1';
 197                 $this->priorities[NS_HELP] = '0.5';
 198                 $this->priorities[NS_HELP_TALK] = '0.1';
 199                 $this->priorities[NS_CATEGORY] = '0.5';
 200                 $this->priorities[NS_CATEGORY_TALK] = '0.1';
 201
 202                 // Custom priorities
 203                 if ( $wgSitemapNamespacesPriorities !== false ) {
 204                         /**
 205                          * @var $wgSitemapNamespacesPriorities array
 206                          */
 207                         foreach ( $wgSitemapNamespacesPriorities as $namespace => $priority ) {
 208                                 $float = floatval( $priority );
 209                                 if ( $float > 1.0 ) {
 210                                         $priority = '1.0';
 211                                 } elseif ( $float < 0.0 ) {
 212                                         $priority = '0.0';
 213                                 }
 214                                 $this->priorities[$namespace] = $priority;
 215                         }
 216                 }
 217         }
 218
 219         /**
 220          * Create directory if it does not exist and return pathname with a trailing slash
 221          * @param $fspath string
 222          * @return null|string
 223          */
 224         private static function init_path( $fspath ) {
 225                 if ( !isset( $fspath ) ) {
 226                         return null;
 227                 }
 228                 # Create directory if needed
 229                 if ( $fspath && !is_dir( $fspath ) ) {
 230                         wfMkdirParents( $fspath, null, __METHOD__ ) or die( "Can not create directory $fspath.\n" );
 231                 }
 232
 233                 return realpath( $fspath ) . DIRECTORY_SEPARATOR;
 234         }
 235
 236         /**
 237          * Generate a one-dimensional array of existing namespaces
 238          */
 239         function generateNamespaces() {
 240                 // Only generate for specific namespaces if $wgSitemapNamespaces is an array.
 241                 global $wgSitemapNamespaces;
 242                 if ( is_array( $wgSitemapNamespaces ) ) {
 243                         $this->namespaces = $wgSitemapNamespaces;
 244                         return;
 245                 }
 246
 247                 $res = $this->dbr->select( 'page',
 248                         array( 'page_namespace' ),
 249                         array(),
 250                         __METHOD__,
 251                         array(
 252                                 'GROUP BY' => 'page_namespace',
 253                                 'ORDER BY' => 'page_namespace',
 254                         )
 255                 );
 256
 257                 foreach ( $res as $row ) {
 258                         $this->namespaces[] = $row->page_namespace;
 259                 }
 260         }
 261
 262         /**
 263          * Get the priority of a given namespace
 264          *
 265          * @param $namespace Integer: the namespace to get the priority for
 266          * @return String
 267          */
 268         function priority( $namespace ) {
 269                 return isset( $this->priorities[$namespace] ) ? $this->priorities[$namespace] : $this->guessPriority( $namespace );
 270         }
 271
 272         /**
 273          * If the namespace isn't listed on the priority list return the
 274          * default priority for the namespace, varies depending on whether it's
 275          * a talkpage or not.
 276          *
 277          * @param $namespace Integer: the namespace to get the priority for
 278          * @return String
 279          */
 280         function guessPriority( $namespace ) {
 281                 return MWNamespace::isSubject( $namespace ) ? $this->priorities[self::GS_MAIN] : $this->priorities[self::GS_TALK];
 282         }
 283
 284         /**
 285          * Return a database resolution of all the pages in a given namespace
 286          *
 287          * @param $namespace Integer: limit the query to this namespace
 288          * @return Resource
 289          */
 290         function getPageRes( $namespace ) {
 291                 return $this->dbr->select( 'page',
 292                         array(
 293                                 'page_namespace',
 294                                 'page_title',
 295                                 'page_touched',
 296                                 'page_is_redirect'
 297                         ),
 298                         array( 'page_namespace' => $namespace ),
 299                         __METHOD__
 300                 );
 301         }
 302
 303         /**
 304          * Main loop
 305          */
 306         public function main() {
 307                 global $wgContLang;
 308
 309                 fwrite( $this->findex, $this->openIndex() );
 310
 311                 foreach ( $this->namespaces as $namespace ) {
 312                         $res = $this->getPageRes( $namespace );
 313                         $this->file = false;
 314                         $this->generateLimit( $namespace );
 315                         $length = $this->limit[0];
 316                         $i = $smcount = 0;
 317
 318                         $fns = $wgContLang->getFormattedNsText( $namespace );
 319                         $this->output( "$namespace ($fns)\n" );
 320                         $skippedRedirects = 0;  // Number of redirects skipped for that namespace
 321                         foreach ( $res as $row ) {
 322                                 if ( $this->skipRedirects && $row->page_is_redirect ) {
 323                                         $skippedRedirects++;
 324                                         continue;
 325                                 }
 326
 327                                 if ( $i++ === 0 || $i === $this->url_limit + 1 || $length + $this->limit[1] + $this->limit[2] > $this->size_limit ) {
 328                                         if ( $this->file !== false ) {
 329                                                 $this->write( $this->file, $this->closeFile() );
 330                                                 $this->close( $this->file );
 331                                         }
 332                                         $filename = $this->sitemapFilename( $namespace, $smcount++ );
 333                                         $this->file = $this->open( $this->fspath . $filename, 'wb' );
 334                                         $this->write( $this->file, $this->openFile() );
 335                                         fwrite( $this->findex, $this->indexEntry( $filename ) );
 336                                         $this->output( "\t$this->fspath$filename\n" );
 337                                         $length = $this->limit[0];
 338                                         $i = 1;
 339                                 }
 340                                 $title = Title::makeTitle( $row->page_namespace, $row->page_title );
 341                                 $date = wfTimestamp( TS_ISO_8601, $row->page_touched );
 342                                 $entry = $this->fileEntry( $title->getCanonicalURL(), $date, $this->priority( $namespace ) );
 343                                 $length += strlen( $entry );
 344                                 $this->write( $this->file, $entry );
 345                                 // generate pages for language variants
 346                                 if ( $wgContLang->hasVariants() ) {
 347                                         $variants = $wgContLang->getVariants();
 348                                         foreach ( $variants as $vCode ) {
 349                                                 if ( $vCode == $wgContLang->getCode() ) {
 350                                                         continue; // we don't want default variant
 351                                                 }
 352                                                 $entry = $this->fileEntry( $title->getCanonicalURL( '', $vCode ), $date, $this->priority( $namespace ) );
 353                                                 $length += strlen( $entry );
 354                                                 $this->write( $this->file, $entry );
 355                                         }
 356                                 }
 357                         }
 358
 359                         if ( $this->skipRedirects && $skippedRedirects > 0 ) {
 360                                 $this->output( "  skipped $skippedRedirects redirect(s)\n" );
 361                         }
 362
 363                         if ( $this->file ) {
 364                                 $this->write( $this->file, $this->closeFile() );
 365                                 $this->close( $this->file );
 366                         }
 367                 }
 368                 fwrite( $this->findex, $this->closeIndex() );
 369                 fclose( $this->findex );
 370         }
 371
 372         /**
 373          * gzopen() / fopen() wrapper
 374          *
 375          * @return Resource
 376          */
 377         function open( $file, $flags ) {
 378                 $resource = $this->compress ? gzopen( $file, $flags ) : fopen( $file, $flags );
 379                 if ( $resource === false ) {
 380                         throw new MWException( __METHOD__ . " error opening file $file with flags $flags. Check permissions?" );
 381                 }
 382                 return $resource;
 383         }
 384
 385         /**
 386          * gzwrite() / fwrite() wrapper
 387          */
 388         function write( &$handle, $str ) {
 389                 if ( $handle === true || $handle === false ) {
 390                         throw new MWException( __METHOD__ . " was passed a boolean as a file handle.\n" );
 391                 }
 392                 if ( $this->compress ) {
 393                         gzwrite( $handle, $str );
 394                 } else {
 395                         fwrite( $handle, $str );
 396                 }
 397         }
 398
 399         /**
 400          * gzclose() / fclose() wrapper
 401          */
 402         function close( &$handle ) {
 403                 if ( $this->compress ) {
 404                         gzclose( $handle );
 405                 } else {
 406                         fclose( $handle );
 407                 }
 408         }
 409
 410         /**
 411          * Get a sitemap filename
 412          *
 413          * @param $namespace Integer: the namespace
 414          * @param $count Integer: the count
 415          * @return String
 416          */
 417         function sitemapFilename( $namespace, $count ) {
 418                 $ext = $this->compress ? '.gz' : '';
 419                 return "sitemap-{$this->identifier}-NS_$namespace-$count.xml$ext";
 420         }
 421
 422         /**
 423          * Return the XML required to open an XML file
 424          *
 425          * @return string
 426          */
 427         function xmlHead() {
 428                 return '<?xml version="1.0" encoding="UTF-8"?>' . "\n";
 429         }
 430
 431         /**
 432          * Return the XML schema being used
 433          *
 434          * @return String
 435          */
 436         function xmlSchema() {
 437                 return 'http://www.sitemaps.org/schemas/sitemap/0.9';
 438         }
 439
 440         /**
 441          * Return the XML required to open a sitemap index file
 442          *
 443          * @return String
 444          */
 445         function openIndex() {
 446                 return $this->xmlHead() . '<sitemapindex xmlns="' . $this->xmlSchema() . '">' . "\n";
 447         }
 448
 449         /**
 450          * Return the XML for a single sitemap indexfile entry
 451          *
 452          * @param $filename String: the filename of the sitemap file
 453          * @return String
 454          */
 455         function indexEntry( $filename ) {
 456                 return
 457                         "\t<sitemap>\n" .
 458                         "\t\t<loc>{$this->urlpath}$filename</loc>\n" .
 459                         "\t\t<lastmod>{$this->timestamp}</lastmod>\n" .
 460                         "\t</sitemap>\n";
 461         }
 462
 463         /**
 464          * Return the XML required to close a sitemap index file
 465          *
 466          * @return String
 467          */
 468         function closeIndex() {
 469                 return "</sitemapindex>\n";
 470         }
 471
 472         /**
 473          * Return the XML required to open a sitemap file
 474          *
 475          * @return String
 476          */
 477         function openFile() {
 478                 return $this->xmlHead() . '<urlset xmlns="' . $this->xmlSchema() . '">' . "\n";
 479         }
 480
 481         /**
 482          * Return the XML for a single sitemap entry
 483          *
 484          * @param $url String: an RFC 2396 compliant URL
 485          * @param $date String: a ISO 8601 date
 486          * @param $priority String: a priority indicator, 0.0 - 1.0 inclusive with a 0.1 stepsize
 487          * @return String
 488          */
 489         function fileEntry( $url, $date, $priority ) {
 490                 return
 491                         "\t<url>\n" .
 492                         // bug 34666: $url may contain bad characters such as ampersands.
 493                         "\t\t<loc>" . htmlspecialchars( $url ) . "</loc>\n" .
 494                         "\t\t<lastmod>$date</lastmod>\n" .
 495                         "\t\t<priority>$priority</priority>\n" .
 496                         "\t</url>\n";
 497         }
 498
 499         /**
 500          * Return the XML required to close sitemap file
 501          *
 502          * @return String
 503          */
 504         function closeFile() {
 505                 return "</urlset>\n";
 506         }
 507
 508         /**
 509          * Populate $this->limit
 510          */
 511         function generateLimit( $namespace ) {
 512                 // bug 17961: make a title with the longest possible URL in this namespace
 513                 $title = Title::makeTitle( $namespace, str_repeat( "\xf0\xa8\xae\x81", 63 ) . "\xe5\x96\x83" );
 514
 515                 $this->limit = array(
 516                         strlen( $this->openFile() ),
 517                         strlen( $this->fileEntry( $title->getCanonicalURL(), wfTimestamp( TS_ISO_8601, wfTimestamp() ), $this->priority( $namespace ) ) ),
 518                         strlen( $this->closeFile() )
 519                 );
 520         }
 521 }
 522
 523 $maintClass = "GenerateSitemap";
 524 require_once RUN_MAINTENANCE_IF_MAIN;