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;
  67          * should resolve to the same directory as $fspath.
  68          *
  69          * @var string
  70          */
  71         public $urlpath;
  72
  73         /**
  74          * Whether or not to use compression
  75          *
  76          * @var bool
  77          */
  78         public $compress;
  79
  80         /**
  81          * Whether or not to include redirection pages
  82          *
  83          * @var bool
  84          */
  85         public $skipRedirects;
  86
  87         /**
  88          * The number of entries to save in each sitemap file
  89          *
  90          * @var array
  91          */
  92         public $limit = array();
  93
  94         /**
  95          * Key => value entries of namespaces and their priorities
  96          *
  97          * @var array
  98          */
  99         public $priorities = array();
 100
 101         /**
 102          * A one-dimensional array of namespaces in the wiki
 103          *
 104          * @var array
 105          */
 106         public $namespaces = array();
 107
 108         /**
 109          * When this sitemap batch was generated
 110          *
 111          * @var string
 112          */
 113         public $timestamp;
 114
 115         /**
 116          * A database slave object
 117          *
 118          * @var object
 119          */
 120         public $dbr;
 121
 122         /**
 123          * A resource pointing to the sitemap index file
 124          *
 125          * @var resource
 126          */
 127         public $findex;
 128
 129         /**
 130          * A resource pointing to a sitemap file
 131          *
 132          * @var resource
 133          */
 134         public $file;
 135
 136         /**
 137          * Identifier to use in filenames, default $wgDBname
 138          *
 139          * @var string
 140          */
 141         private $identifier;
 142
 143         /**
 144          * Constructor
 145          */
 146         public function __construct() {
 147                 parent::__construct();
 148                 $this->mDescription = "Creates a sitemap for the site";
 149                 $this->addOption(
 150                         'fspath',
 151                         'The file system path to save to, e.g. /tmp/sitemap; defaults to current directory',
 152                         false,
 153                         true
 154                 );
 155                 $this->addOption(
 156                         'urlpath',
 157                         'The URL path corresponding to --fspath, prepended to filenames in the index; '
 158                                 . 'defaults to an empty string',
 159                         false,
 160                         true
 161                 );
 162                 $this->addOption(
 163                         'compress',
 164                         'Compress the sitemap files, can take value yes|no, default yes',
 165                         false,
 166                         true
 167                 );
 168                 $this->addOption( 'skip-redirects', 'Do not include redirecting articles in the sitemap' );
 169                 $this->addOption(
 170                         'identifier',
 171                         'What site identifier to use for the wiki, defaults to $wgDBname',
 172                         false,
 173                         true
 174                 );
 175         }
 176
 177         /**
 178          * Execute
 179          */
 180         public function execute() {
 181                 $this->setNamespacePriorities();
 182                 $this->url_limit = 50000;
 183                 $this->size_limit = pow( 2, 20 ) * 10;
 184                 $this->fspath = self::init_path( $this->getOption( 'fspath', getcwd() ) );
 185                 $this->urlpath = $this->getOption( 'urlpath', "" );
 186                 if ( $this->urlpath !== "" && substr( $this->urlpath, -1 ) !== '/' ) {
 187                         $this->urlpath .= '/';
 188                 }
 189                 $this->identifier = $this->getOption( 'identifier', wfWikiID() );
 190                 $this->compress = $this->getOption( 'compress', 'yes' ) !== 'no';
 191                 $this->skipRedirects = $this->getOption( 'skip-redirects', false ) !== false;
 192                 $this->dbr = wfGetDB( DB_SLAVE );
 193                 $this->generateNamespaces();
 194                 $this->timestamp = wfTimestamp( TS_ISO_8601, wfTimestampNow() );
 195                 $this->findex = fopen( "{$this->fspath}sitemap-index-{$this->identifier}.xml", 'wb' );
 196                 $this->main();
 197         }
 198
 199         private function setNamespacePriorities() {
 200                 global $wgSitemapNamespacesPriorities;
 201
 202                 // Custom main namespaces
 203                 $this->priorities[self::GS_MAIN] = '0.5';
 204                 // Custom talk namesspaces
 205                 $this->priorities[self::GS_TALK] = '0.1';
 206                 // MediaWiki standard namespaces
 207                 $this->priorities[NS_MAIN] = '1.0';
 208                 $this->priorities[NS_TALK] = '0.1';
 209                 $this->priorities[NS_USER] = '0.5';
 210                 $this->priorities[NS_USER_TALK] = '0.1';
 211                 $this->priorities[NS_PROJECT] = '0.5';
 212                 $this->priorities[NS_PROJECT_TALK] = '0.1';
 213                 $this->priorities[NS_FILE] = '0.5';
 214                 $this->priorities[NS_FILE_TALK] = '0.1';
 215                 $this->priorities[NS_MEDIAWIKI] = '0.0';
 216                 $this->priorities[NS_MEDIAWIKI_TALK] = '0.1';
 217                 $this->priorities[NS_TEMPLATE] = '0.0';
 218                 $this->priorities[NS_TEMPLATE_TALK] = '0.1';
 219                 $this->priorities[NS_HELP] = '0.5';
 220                 $this->priorities[NS_HELP_TALK] = '0.1';
 221                 $this->priorities[NS_CATEGORY] = '0.5';
 222                 $this->priorities[NS_CATEGORY_TALK] = '0.1';
 223
 224                 // Custom priorities
 225                 if ( $wgSitemapNamespacesPriorities !== false ) {
 226                         /**
 227                          * @var $wgSitemapNamespacesPriorities array
 228                          */
 229                         foreach ( $wgSitemapNamespacesPriorities as $namespace => $priority ) {
 230                                 $float = floatval( $priority );
 231                                 if ( $float > 1.0 ) {
 232                                         $priority = '1.0';
 233                                 } elseif ( $float < 0.0 ) {
 234                                         $priority = '0.0';
 235                                 }
 236                                 $this->priorities[$namespace] = $priority;
 237                         }
 238                 }
 239         }
 240
 241         /**
 242          * Create directory if it does not exist and return pathname with a trailing slash
 243          * @param string $fspath
 244          * @return null|string
 245          */
 246         private static function init_path( $fspath ) {
 247                 if ( !isset( $fspath ) ) {
 248                         return null;
 249                 }
 250                 # Create directory if needed
 251                 if ( $fspath && !is_dir( $fspath ) ) {
 252                         wfMkdirParents( $fspath, null, __METHOD__ ) or die( "Can not create directory $fspath.\n" );
 253                 }
 254
 255                 return realpath( $fspath ) . DIRECTORY_SEPARATOR;
 256         }
 257
 258         /**
 259          * Generate a one-dimensional array of existing namespaces
 260          */
 261         function generateNamespaces() {
 262                 // Only generate for specific namespaces if $wgSitemapNamespaces is an array.
 263                 global $wgSitemapNamespaces;
 264                 if ( is_array( $wgSitemapNamespaces ) ) {
 265                         $this->namespaces = $wgSitemapNamespaces;
 266                         return;
 267                 }
 268
 269                 $res = $this->dbr->select( 'page',
 270                         array( 'page_namespace' ),
 271                         array(),
 272                         __METHOD__,
 273                         array(
 274                                 'GROUP BY' => 'page_namespace',
 275                                 'ORDER BY' => 'page_namespace',
 276                         )
 277                 );
 278
 279                 foreach ( $res as $row ) {
 280                         $this->namespaces[] = $row->page_namespace;
 281                 }
 282         }
 283
 284         /**
 285          * Get the priority of a given namespace
 286          *
 287          * @param int $namespace The namespace to get the priority for
 288          * @return string
 289          */
 290         function priority( $namespace ) {
 291                 return isset( $this->priorities[$namespace] )
 292                         ? $this->priorities[$namespace]
 293                         : $this->guessPriority( $namespace );
 294         }
 295
 296         /**
 297          * If the namespace isn't listed on the priority list return the
 298          * default priority for the namespace, varies depending on whether it's
 299          * a talkpage or not.
 300          *
 301          * @param int $namespace The namespace to get the priority for
 302          * @return string
 303          */
 304         function guessPriority( $namespace ) {
 305                 return MWNamespace::isSubject( $namespace )
 306                         ? $this->priorities[self::GS_MAIN]
 307                         : $this->priorities[self::GS_TALK];
 308         }
 309
 310         /**
 311          * Return a database resolution of all the pages in a given namespace
 312          *
 313          * @param int $namespace Limit the query to this namespace
 314          * @return Resource
 315          */
 316         function getPageRes( $namespace ) {
 317                 return $this->dbr->select( 'page',
 318                         array(
 319                                 'page_namespace',
 320                                 'page_title',
 321                                 'page_touched',
 322                                 'page_is_redirect'
 323                         ),
 324                         array( 'page_namespace' => $namespace ),
 325                         __METHOD__
 326                 );
 327         }
 328
 329         /**
 330          * Main loop
 331          */
 332         public function main() {
 333                 global $wgContLang;
 334
 335                 fwrite( $this->findex, $this->openIndex() );
 336
 337                 foreach ( $this->namespaces as $namespace ) {
 338                         $res = $this->getPageRes( $namespace );
 339                         $this->file = false;
 340                         $this->generateLimit( $namespace );
 341                         $length = $this->limit[0];
 342                         $i = $smcount = 0;
 343
 344                         $fns = $wgContLang->getFormattedNsText( $namespace );
 345                         $this->output( "$namespace ($fns)\n" );
 346                         $skippedRedirects = 0;  // Number of redirects skipped for that namespace
 347                         foreach ( $res as $row ) {
 348                                 if ( $this->skipRedirects && $row->page_is_redirect ) {
 349                                         $skippedRedirects++;
 350                                         continue;
 351                                 }
 352
 353                                 if ( $i++ === 0
 354                                         || $i === $this->url_limit + 1
 355                                         || $length + $this->limit[1] + $this->limit[2] > $this->size_limit
 356                                 ) {
 357                                         if ( $this->file !== false ) {
 358                                                 $this->write( $this->file, $this->closeFile() );
 359                                                 $this->close( $this->file );
 360                                         }
 361                                         $filename = $this->sitemapFilename( $namespace, $smcount++ );
 362                                         $this->file = $this->open( $this->fspath . $filename, 'wb' );
 363                                         $this->write( $this->file, $this->openFile() );
 364                                         fwrite( $this->findex, $this->indexEntry( $filename ) );
 365                                         $this->output( "\t$this->fspath$filename\n" );
 366                                         $length = $this->limit[0];
 367                                         $i = 1;
 368                                 }
 369                                 $title = Title::makeTitle( $row->page_namespace, $row->page_title );
 370                                 $date = wfTimestamp( TS_ISO_8601, $row->page_touched );
 371                                 $entry = $this->fileEntry( $title->getCanonicalURL(), $date, $this->priority( $namespace ) );
 372                                 $length += strlen( $entry );
 373                                 $this->write( $this->file, $entry );
 374                                 // generate pages for language variants
 375                                 if ( $wgContLang->hasVariants() ) {
 376                                         $variants = $wgContLang->getVariants();
 377                                         foreach ( $variants as $vCode ) {
 378                                                 if ( $vCode == $wgContLang->getCode() ) {
 379                                                         continue; // we don't want default variant
 380                                                 }
 381                                                 $entry = $this->fileEntry(
 382                                                         $title->getCanonicalURL( '', $vCode ),
 383                                                         $date,
 384                                                         $this->priority( $namespace )
 385                                                 );
 386                                                 $length += strlen( $entry );
 387                                                 $this->write( $this->file, $entry );
 388                                         }
 389                                 }
 390                         }
 391
 392                         if ( $this->skipRedirects && $skippedRedirects > 0 ) {
 393                                 $this->output( "  skipped $skippedRedirects redirect(s)\n" );
 394                         }
 395
 396                         if ( $this->file ) {
 397                                 $this->write( $this->file, $this->closeFile() );
 398                                 $this->close( $this->file );
 399                         }
 400                 }
 401                 fwrite( $this->findex, $this->closeIndex() );
 402                 fclose( $this->findex );
 403         }
 404
 405         /**
 406          * gzopen() / fopen() wrapper
 407          *
 408          * @param string $file
 409          * @param string $flags
 410          * @return resource
 411          */
 412         function open( $file, $flags ) {
 413                 $resource = $this->compress ? gzopen( $file, $flags ) : fopen( $file, $flags );
 414                 if ( $resource === false ) {
 415                         throw new MWException( __METHOD__
 416                                 . " error opening file $file with flags $flags. Check permissions?" );
 417                 }
 418                 return $resource;
 419         }
 420
 421         /**
 422          * gzwrite() / fwrite() wrapper
 423          *
 424          * @param resource $handle
 425          * @param string $str
 426          */
 427         function write( &$handle, $str ) {
 428                 if ( $handle === true || $handle === false ) {
 429                         throw new MWException( __METHOD__ . " was passed a boolean as a file handle.\n" );
 430                 }
 431                 if ( $this->compress ) {
 432                         gzwrite( $handle, $str );
 433                 } else {
 434                         fwrite( $handle, $str );
 435                 }
 436         }
 437
 438         /**
 439          * gzclose() / fclose() wrapper
 440          *
 441          * @param resource $handle
 442          */
 443         function close( &$handle ) {
 444                 if ( $this->compress ) {
 445                         gzclose( $handle );
 446                 } else {
 447                         fclose( $handle );
 448                 }
 449         }
 450
 451         /**
 452          * Get a sitemap filename
 453          *
 454          * @param int $namespace The namespace
 455          * @param int $count The count
 456          * @return string
 457          */
 458         function sitemapFilename( $namespace, $count ) {
 459                 $ext = $this->compress ? '.gz' : '';
 460                 return "sitemap-{$this->identifier}-NS_$namespace-$count.xml$ext";
 461         }
 462
 463         /**
 464          * Return the XML required to open an XML file
 465          *
 466          * @return string
 467          */
 468         function xmlHead() {
 469                 return '<?xml version="1.0" encoding="UTF-8"?>' . "\n";
 470         }
 471
 472         /**
 473          * Return the XML schema being used
 474          *
 475          * @return string
 476          */
 477         function xmlSchema() {
 478                 return 'http://www.sitemaps.org/schemas/sitemap/0.9';
 479         }
 480
 481         /**
 482          * Return the XML required to open a sitemap index file
 483          *
 484          * @return string
 485          */
 486         function openIndex() {
 487                 return $this->xmlHead() . '<sitemapindex xmlns="' . $this->xmlSchema() . '">' . "\n";
 488         }
 489
 490         /**
 491          * Return the XML for a single sitemap indexfile entry
 492          *
 493          * @param string $filename The filename of the sitemap file
 494          * @return string
 495          */
 496         function indexEntry( $filename ) {
 497                 return
 498                         "\t<sitemap>\n" .
 499                         "\t\t<loc>{$this->urlpath}$filename</loc>\n" .
 500                         "\t\t<lastmod>{$this->timestamp}</lastmod>\n" .
 501                         "\t</sitemap>\n";
 502         }
 503
 504         /**
 505          * Return the XML required to close a sitemap index file
 506          *
 507          * @return string
 508          */
 509         function closeIndex() {
 510                 return "</sitemapindex>\n";
 511         }
 512
 513         /**
 514          * Return the XML required to open a sitemap file
 515          *
 516          * @return string
 517          */
 518         function openFile() {
 519                 return $this->xmlHead() . '<urlset xmlns="' . $this->xmlSchema() . '">' . "\n";
 520         }
 521
 522         /**
 523          * Return the XML for a single sitemap entry
 524          *
 525          * @param string $url An RFC 2396 compliant URL
 526          * @param string $date A ISO 8601 date
 527          * @param string $priority A priority indicator, 0.0 - 1.0 inclusive with a 0.1 stepsize
 528          * @return string
 529          */
 530         function fileEntry( $url, $date, $priority ) {
 531                 return
 532                         "\t<url>\n" .
 533                         // bug 34666: $url may contain bad characters such as ampersands.
 534                         "\t\t<loc>" . htmlspecialchars( $url ) . "</loc>\n" .
 535                         "\t\t<lastmod>$date</lastmod>\n" .
 536                         "\t\t<priority>$priority</priority>\n" .
 537                         "\t</url>\n";
 538         }
 539
 540         /**
 541          * Return the XML required to close sitemap file
 542          *
 543          * @return string
 544          */
 545         function closeFile() {
 546                 return "</urlset>\n";
 547         }
 548
 549         /**
 550          * Populate $this->limit
 551          *
 552          * @param int $namespace
 553          */
 554         function generateLimit( $namespace ) {
 555                 // bug 17961: make a title with the longest possible URL in this namespace
 556                 $title = Title::makeTitle( $namespace, str_repeat( "\xf0\xa8\xae\x81", 63 ) . "\xe5\x96\x83" );
 557
 558                 $this->limit = array(
 559                         strlen( $this->openFile() ),
 560                         strlen( $this->fileEntry(
 561                                 $title->getCanonicalURL(),
 562                                 wfTimestamp( TS_ISO_8601, wfTimestamp() ),
 563                                 $this->priority( $namespace )
 564                         ) ),
 565                         strlen( $this->closeFile() )
 566                 );
 567         }
 568 }
 569
 570 $maintClass = "GenerateSitemap";
 571 require_once RUN_MAINTENANCE_IF_MAIN;