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 use MediaWiki\MediaWikiServices;
  30 use Wikimedia\Rdbms\IResultWrapper;
  31
  32 require_once __DIR__ . '/Maintenance.php';
  33
  34 /**
  35  * Maintenance script that generates a sitemap for the site.
  36  *
  37  * @ingroup Maintenance
  38  */
  39 class GenerateSitemap extends Maintenance {
  40         const GS_MAIN = -2;
  41         const GS_TALK = -1;
  42
  43         /**
  44          * The maximum amount of urls in a sitemap file
  45          *
  46          * @link http://www.sitemaps.org/schemas/sitemap/0.9/sitemap.xsd
  47          *
  48          * @var int
  49          */
  50         public $url_limit;
  51
  52         /**
  53          * The maximum size of a sitemap file
  54          *
  55          * @link http://www.sitemaps.org/faq.php#faq_sitemap_size
  56          *
  57          * @var int
  58          */
  59         public $size_limit;
  60
  61         /**
  62          * The path to prepend to the filename
  63          *
  64          * @var string
  65          */
  66         public $fspath;
  67
  68         /**
  69          * The URL path to prepend to filenames in the index;
  70          * should resolve to the same directory as $fspath.
  71          *
  72          * @var string
  73          */
  74         public $urlpath;
  75
  76         /**
  77          * Whether or not to use compression
  78          *
  79          * @var bool
  80          */
  81         public $compress;
  82
  83         /**
  84          * Whether or not to include redirection pages
  85          *
  86          * @var bool
  87          */
  88         public $skipRedirects;
  89
  90         /**
  91          * The number of entries to save in each sitemap file
  92          *
  93          * @var array
  94          */
  95         public $limit = [];
  96
  97         /**
  98          * Key => value entries of namespaces and their priorities
  99          *
 100          * @var array
 101          */
 102         public $priorities = [];
 103
 104         /**
 105          * A one-dimensional array of namespaces in the wiki
 106          *
 107          * @var array
 108          */
 109         public $namespaces = [];
 110
 111         /**
 112          * When this sitemap batch was generated
 113          *
 114          * @var string
 115          */
 116         public $timestamp;
 117
 118         /**
 119          * A database replica DB object
 120          *
 121          * @var object
 122          */
 123         public $dbr;
 124
 125         /**
 126          * A resource pointing to the sitemap index file
 127          *
 128          * @var resource
 129          */
 130         public $findex;
 131
 132         /**
 133          * A resource pointing to a sitemap file
 134          *
 135          * @var resource|false
 136          */
 137         public $file;
 138
 139         /**
 140          * Identifier to use in filenames, default $wgDBname
 141          *
 142          * @var string
 143          */
 144         private $identifier;
 145
 146         public function __construct() {
 147                 parent::__construct();
 148                 $this->addDescription( '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 = ( 2 ** 20 ) * 10;
 184
 185                 # Create directory if needed
 186                 $fspath = $this->getOption( 'fspath', getcwd() );
 187                 if ( !wfMkdirParents( $fspath, null, __METHOD__ ) ) {
 188                         $this->fatalError( "Can not create directory $fspath." );
 189                 }
 190
 191                 $dbDomain = WikiMap::getCurrentWikiDbDomain()->getId();
 192                 $this->fspath = realpath( $fspath ) . DIRECTORY_SEPARATOR;
 193                 $this->urlpath = $this->getOption( 'urlpath', "" );
 194                 if ( $this->urlpath !== "" && substr( $this->urlpath, -1 ) !== '/' ) {
 195                         $this->urlpath .= '/';
 196                 }
 197                 $this->identifier = $this->getOption( 'identifier', $dbDomain );
 198                 $this->compress = $this->getOption( 'compress', 'yes' ) !== 'no';
 199                 $this->skipRedirects = $this->hasOption( 'skip-redirects' );
 200                 $this->dbr = $this->getDB( DB_REPLICA );
 201                 $this->generateNamespaces();
 202                 $this->timestamp = wfTimestamp( TS_ISO_8601, wfTimestampNow() );
 203                 $encIdentifier = rawurlencode( $this->identifier );
 204                 $this->findex = fopen( "{$this->fspath}sitemap-index-{$encIdentifier}.xml", 'wb' );
 205                 $this->main();
 206         }
 207
 208         private function setNamespacePriorities() {
 209                 global $wgSitemapNamespacesPriorities;
 210
 211                 // Custom main namespaces
 212                 $this->priorities[self::GS_MAIN] = '0.5';
 213                 // Custom talk namesspaces
 214                 $this->priorities[self::GS_TALK] = '0.1';
 215                 // MediaWiki standard namespaces
 216                 $this->priorities[NS_MAIN] = '1.0';
 217                 $this->priorities[NS_TALK] = '0.1';
 218                 $this->priorities[NS_USER] = '0.5';
 219                 $this->priorities[NS_USER_TALK] = '0.1';
 220                 $this->priorities[NS_PROJECT] = '0.5';
 221                 $this->priorities[NS_PROJECT_TALK] = '0.1';
 222                 $this->priorities[NS_FILE] = '0.5';
 223                 $this->priorities[NS_FILE_TALK] = '0.1';
 224                 $this->priorities[NS_MEDIAWIKI] = '0.0';
 225                 $this->priorities[NS_MEDIAWIKI_TALK] = '0.1';
 226                 $this->priorities[NS_TEMPLATE] = '0.0';
 227                 $this->priorities[NS_TEMPLATE_TALK] = '0.1';
 228                 $this->priorities[NS_HELP] = '0.5';
 229                 $this->priorities[NS_HELP_TALK] = '0.1';
 230                 $this->priorities[NS_CATEGORY] = '0.5';
 231                 $this->priorities[NS_CATEGORY_TALK] = '0.1';
 232
 233                 // Custom priorities
 234                 if ( $wgSitemapNamespacesPriorities !== false ) {
 235                         /**
 236                          * @var array $wgSitemapNamespacesPriorities
 237                          */
 238                         foreach ( $wgSitemapNamespacesPriorities as $namespace => $priority ) {
 239                                 $float = floatval( $priority );
 240                                 if ( $float > 1.0 ) {
 241                                         $priority = '1.0';
 242                                 } elseif ( $float < 0.0 ) {
 243                                         $priority = '0.0';
 244                                 }
 245                                 $this->priorities[$namespace] = $priority;
 246                         }
 247                 }
 248         }
 249
 250         /**
 251          * Generate a one-dimensional array of existing namespaces
 252          */
 253         function generateNamespaces() {
 254                 // Only generate for specific namespaces if $wgSitemapNamespaces is an array.
 255                 global $wgSitemapNamespaces;
 256                 if ( is_array( $wgSitemapNamespaces ) ) {
 257                         $this->namespaces = $wgSitemapNamespaces;
 258
 259                         return;
 260                 }
 261
 262                 $res = $this->dbr->select( 'page',
 263                         [ 'page_namespace' ],
 264                         [],
 265                         __METHOD__,
 266                         [
 267                                 'GROUP BY' => 'page_namespace',
 268                                 'ORDER BY' => 'page_namespace',
 269                         ]
 270                 );
 271
 272                 foreach ( $res as $row ) {
 273                         $this->namespaces[] = $row->page_namespace;
 274                 }
 275         }
 276
 277         /**
 278          * Get the priority of a given namespace
 279          *
 280          * @param int $namespace The namespace to get the priority for
 281          * @return string
 282          */
 283         function priority( $namespace ) {
 284                 return $this->priorities[$namespace] ?? $this->guessPriority( $namespace );
 285         }
 286
 287         /**
 288          * If the namespace isn't listed on the priority list return the
 289          * default priority for the namespace, varies depending on whether it's
 290          * a talkpage or not.
 291          *
 292          * @param int $namespace The namespace to get the priority for
 293          * @return string
 294          */
 295         function guessPriority( $namespace ) {
 296                 return MediaWikiServices::getInstance()->getNamespaceInfo()->isSubject( $namespace )
 297                         ? $this->priorities[self::GS_MAIN]
 298                         : $this->priorities[self::GS_TALK];
 299         }
 300
 301         /**
 302          * Return a database resolution of all the pages in a given namespace
 303          *
 304          * @param int $namespace Limit the query to this namespace
 305          * @return IResultWrapper
 306          */
 307         function getPageRes( $namespace ) {
 308                 return $this->dbr->select( 'page',
 309                         [
 310                                 'page_namespace',
 311                                 'page_title',
 312                                 'page_touched',
 313                                 'page_is_redirect'
 314                         ],
 315                         [ 'page_namespace' => $namespace ],
 316                         __METHOD__
 317                 );
 318         }
 319
 320         /**
 321          * Main loop
 322          */
 323         public function main() {
 324                 $contLang = MediaWikiServices::getInstance()->getContentLanguage();
 325
 326                 fwrite( $this->findex, $this->openIndex() );
 327
 328                 foreach ( $this->namespaces as $namespace ) {
 329                         $res = $this->getPageRes( $namespace );
 330                         $this->file = false;
 331                         $this->generateLimit( $namespace );
 332                         $length = $this->limit[0];
 333                         $i = $smcount = 0;
 334
 335                         $fns = $contLang->getFormattedNsText( $namespace );
 336                         $this->output( "$namespace ($fns)\n" );
 337                         $skippedRedirects = 0; // Number of redirects skipped for that namespace
 338                         foreach ( $res as $row ) {
 339                                 if ( $this->skipRedirects && $row->page_is_redirect ) {
 340                                         $skippedRedirects++;
 341                                         continue;
 342                                 }
 343
 344                                 if ( $i++ === 0
 345                                         || $i === $this->url_limit + 1
 346                                         || $length + $this->limit[1] + $this->limit[2] > $this->size_limit
 347                                 ) {
 348                                         if ( $this->file !== false ) {
 349                                                 $this->write( $this->file, $this->closeFile() );
 350                                                 $this->close( $this->file );
 351                                         }
 352                                         $filename = $this->sitemapFilename( $namespace, $smcount++ );
 353                                         $this->file = $this->open( $this->fspath . $filename, 'wb' );
 354                                         $this->write( $this->file, $this->openFile() );
 355                                         fwrite( $this->findex, $this->indexEntry( $filename ) );
 356                                         $this->output( "\t$this->fspath$filename\n" );
 357                                         $length = $this->limit[0];
 358                                         $i = 1;
 359                                 }
 360                                 $title = Title::makeTitle( $row->page_namespace, $row->page_title );
 361                                 $date = wfTimestamp( TS_ISO_8601, $row->page_touched );
 362                                 $entry = $this->fileEntry( $title->getCanonicalURL(), $date, $this->priority( $namespace ) );
 363                                 $length += strlen( $entry );
 364                                 $this->write( $this->file, $entry );
 365                                 // generate pages for language variants
 366                                 if ( $contLang->hasVariants() ) {
 367                                         $variants = $contLang->getVariants();
 368                                         foreach ( $variants as $vCode ) {
 369                                                 if ( $vCode == $contLang->getCode() ) {
 370                                                         continue; // we don't want default variant
 371                                                 }
 372                                                 $entry = $this->fileEntry(
 373                                                         $title->getCanonicalURL( '', $vCode ),
 374                                                         $date,
 375                                                         $this->priority( $namespace )
 376                                                 );
 377                                                 $length += strlen( $entry );
 378                                                 $this->write( $this->file, $entry );
 379                                         }
 380                                 }
 381                         }
 382
 383                         if ( $this->skipRedirects && $skippedRedirects > 0 ) {
 384                                 $this->output( "  skipped $skippedRedirects redirect(s)\n" );
 385                         }
 386
 387                         if ( $this->file ) {
 388                                 $this->write( $this->file, $this->closeFile() );
 389                                 $this->close( $this->file );
 390                         }
 391                 }
 392                 fwrite( $this->findex, $this->closeIndex() );
 393                 fclose( $this->findex );
 394         }
 395
 396         /**
 397          * gzopen() / fopen() wrapper
 398          *
 399          * @param string $file
 400          * @param string $flags
 401          * @return resource
 402          */
 403         function open( $file, $flags ) {
 404                 $resource = $this->compress ? gzopen( $file, $flags ) : fopen( $file, $flags );
 405                 if ( $resource === false ) {
 406                         throw new MWException( __METHOD__
 407                                 . " error opening file $file with flags $flags. Check permissions?" );
 408                 }
 409
 410                 return $resource;
 411         }
 412
 413         /**
 414          * gzwrite() / fwrite() wrapper
 415          *
 416          * @param resource &$handle
 417          * @param string $str
 418          */
 419         function write( &$handle, $str ) {
 420                 if ( $handle === true || $handle === false ) {
 421                         throw new MWException( __METHOD__ . " was passed a boolean as a file handle.\n" );
 422                 }
 423                 if ( $this->compress ) {
 424                         gzwrite( $handle, $str );
 425                 } else {
 426                         fwrite( $handle, $str );
 427                 }
 428         }
 429
 430         /**
 431          * gzclose() / fclose() wrapper
 432          *
 433          * @param resource &$handle
 434          */
 435         function close( &$handle ) {
 436                 if ( $this->compress ) {
 437                         gzclose( $handle );
 438                 } else {
 439                         fclose( $handle );
 440                 }
 441         }
 442
 443         /**
 444          * Get a sitemap filename
 445          *
 446          * @param int $namespace
 447          * @param int $count
 448          * @return string
 449          */
 450         function sitemapFilename( $namespace, $count ) {
 451                 $ext = $this->compress ? '.gz' : '';
 452
 453                 return "sitemap-{$this->identifier}-NS_$namespace-$count.xml$ext";
 454         }
 455
 456         /**
 457          * Return the XML required to open an XML file
 458          *
 459          * @return string
 460          */
 461         function xmlHead() {
 462                 return '<?xml version="1.0" encoding="UTF-8"?>' . "\n";
 463         }
 464
 465         /**
 466          * Return the XML schema being used
 467          *
 468          * @return string
 469          */
 470         function xmlSchema() {
 471                 return 'http://www.sitemaps.org/schemas/sitemap/0.9';
 472         }
 473
 474         /**
 475          * Return the XML required to open a sitemap index file
 476          *
 477          * @return string
 478          */
 479         function openIndex() {
 480                 return $this->xmlHead() . '<sitemapindex xmlns="' . $this->xmlSchema() . '">' . "\n";
 481         }
 482
 483         /**
 484          * Return the XML for a single sitemap indexfile entry
 485          *
 486          * @param string $filename The filename of the sitemap file
 487          * @return string
 488          */
 489         function indexEntry( $filename ) {
 490                 return "\t<sitemap>\n" .
 491                         "\t\t<loc>" . wfGetServerUrl( PROTO_CANONICAL ) .
 492                                 ( substr( $this->urlpath, 0, 1 ) === "/" ? "" : "/" ) .
 493                                 "{$this->urlpath}$filename</loc>\n" .
 494                         "\t\t<lastmod>{$this->timestamp}</lastmod>\n" .
 495                         "\t</sitemap>\n";
 496         }
 497
 498         /**
 499          * Return the XML required to close a sitemap index file
 500          *
 501          * @return string
 502          */
 503         function closeIndex() {
 504                 return "</sitemapindex>\n";
 505         }
 506
 507         /**
 508          * Return the XML required to open a sitemap file
 509          *
 510          * @return string
 511          */
 512         function openFile() {
 513                 return $this->xmlHead() . '<urlset xmlns="' . $this->xmlSchema() . '">' . "\n";
 514         }
 515
 516         /**
 517          * Return the XML for a single sitemap entry
 518          *
 519          * @param string $url An RFC 2396 compliant URL
 520          * @param string $date A ISO 8601 date
 521          * @param string $priority A priority indicator, 0.0 - 1.0 inclusive with a 0.1 stepsize
 522          * @return string
 523          */
 524         function fileEntry( $url, $date, $priority ) {
 525                 return "\t<url>\n" .
 526                         // T36666: $url may contain bad characters such as ampersands.
 527                         "\t\t<loc>" . htmlspecialchars( $url ) . "</loc>\n" .
 528                         "\t\t<lastmod>$date</lastmod>\n" .
 529                         "\t\t<priority>$priority</priority>\n" .
 530                         "\t</url>\n";
 531         }
 532
 533         /**
 534          * Return the XML required to close sitemap file
 535          *
 536          * @return string
 537          */
 538         function closeFile() {
 539                 return "</urlset>\n";
 540         }
 541
 542         /**
 543          * Populate $this->limit
 544          *
 545          * @param int $namespace
 546          */
 547         function generateLimit( $namespace ) {
 548                 // T19961: make a title with the longest possible URL in this namespace
 549                 $title = Title::makeTitle( $namespace, str_repeat( "\u{28B81}", 63 ) . "\u{5583}" );
 550
 551                 $this->limit = [
 552                         strlen( $this->openFile() ),
 553                         strlen( $this->fileEntry(
 554                                 $title->getCanonicalURL(),
 555                                 wfTimestamp( TS_ISO_8601, wfTimestamp() ),
 556                                 $this->priority( $namespace )
 557                         ) ),
 558                         strlen( $this->closeFile() )
 559                 ];
 560         }
 561 }
 562
 563 $maintClass = GenerateSitemap::class;
 564 require_once RUN_MAINTENANCE_IF_MAIN;