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
  31 require_once __DIR__ . '/Maintenance.php';
  32
  33 /**
  34  * Maintenance script that generates a sitemap for the site.
  35  *
  36  * @ingroup Maintenance
  37  */
  38 class GenerateSitemap extends Maintenance {
  39         const GS_MAIN = -2;
  40         const GS_TALK = -1;
  41
  42         /**
  43          * The maximum amount of urls in a sitemap file
  44          *
  45          * @link http://www.sitemaps.org/schemas/sitemap/0.9/sitemap.xsd
  46          *
  47          * @var int
  48          */
  49         public $url_limit;
  50
  51         /**
  52          * The maximum size of a sitemap file
  53          *
  54          * @link http://www.sitemaps.org/faq.php#faq_sitemap_size
  55          *
  56          * @var int
  57          */
  58         public $size_limit;
  59
  60         /**
  61          * The path to prepend to the filename
  62          *
  63          * @var string
  64          */
  65         public $fspath;
  66
  67         /**
  68          * The URL path to prepend to filenames in the index;
  69          * should resolve to the same directory as $fspath.
  70          *
  71          * @var string
  72          */
  73         public $urlpath;
  74
  75         /**
  76          * Whether or not to use compression
  77          *
  78          * @var bool
  79          */
  80         public $compress;
  81
  82         /**
  83          * Whether or not to include redirection pages
  84          *
  85          * @var bool
  86          */
  87         public $skipRedirects;
  88
  89         /**
  90          * The number of entries to save in each sitemap file
  91          *
  92          * @var array
  93          */
  94         public $limit = [];
  95
  96         /**
  97          * Key => value entries of namespaces and their priorities
  98          *
  99          * @var array
 100          */
 101         public $priorities = [];
 102
 103         /**
 104          * A one-dimensional array of namespaces in the wiki
 105          *
 106          * @var array
 107          */
 108         public $namespaces = [];
 109
 110         /**
 111          * When this sitemap batch was generated
 112          *
 113          * @var string
 114          */
 115         public $timestamp;
 116
 117         /**
 118          * A database replica DB object
 119          *
 120          * @var object
 121          */
 122         public $dbr;
 123
 124         /**
 125          * A resource pointing to the sitemap index file
 126          *
 127          * @var resource
 128          */
 129         public $findex;
 130
 131         /**
 132          * A resource pointing to a sitemap file
 133          *
 134          * @var resource
 135          */
 136         public $file;
 137
 138         /**
 139          * Identifier to use in filenames, default $wgDBname
 140          *
 141          * @var string
 142          */
 143         private $identifier;
 144
 145         public function __construct() {
 146                 parent::__construct();
 147                 $this->addDescription( 'Creates a sitemap for the site' );
 148                 $this->addOption(
 149                         'fspath',
 150                         'The file system path to save to, e.g. /tmp/sitemap; defaults to current directory',
 151                         false,
 152                         true
 153                 );
 154                 $this->addOption(
 155                         'urlpath',
 156                         'The URL path corresponding to --fspath, prepended to filenames in the index; '
 157                                 . 'defaults to an empty string',
 158                         false,
 159                         true
 160                 );
 161                 $this->addOption(
 162                         'compress',
 163                         'Compress the sitemap files, can take value yes|no, default yes',
 164                         false,
 165                         true
 166                 );
 167                 $this->addOption( 'skip-redirects', 'Do not include redirecting articles in the sitemap' );
 168                 $this->addOption(
 169                         'identifier',
 170                         'What site identifier to use for the wiki, defaults to $wgDBname',
 171                         false,
 172                         true
 173                 );
 174         }
 175
 176         /**
 177          * Execute
 178          */
 179         public function execute() {
 180                 $this->setNamespacePriorities();
 181                 $this->url_limit = 50000;
 182                 $this->size_limit = ( 2 ** 20 ) * 10;
 183
 184                 # Create directory if needed
 185                 $fspath = $this->getOption( 'fspath', getcwd() );
 186                 if ( !wfMkdirParents( $fspath, null, __METHOD__ ) ) {
 187                         $this->fatalError( "Can not create directory $fspath." );
 188                 }
 189
 190                 $this->fspath = realpath( $fspath ) . DIRECTORY_SEPARATOR;
 191                 $this->urlpath = $this->getOption( 'urlpath', "" );
 192                 if ( $this->urlpath !== "" && substr( $this->urlpath, -1 ) !== '/' ) {
 193                         $this->urlpath .= '/';
 194                 }
 195                 $this->identifier = $this->getOption( 'identifier', wfWikiID() );
 196                 $this->compress = $this->getOption( 'compress', 'yes' ) !== 'no';
 197                 $this->skipRedirects = $this->hasOption( 'skip-redirects' );
 198                 $this->dbr = $this->getDB( DB_REPLICA );
 199                 $this->generateNamespaces();
 200                 $this->timestamp = wfTimestamp( TS_ISO_8601, wfTimestampNow() );
 201                 $this->findex = fopen( "{$this->fspath}sitemap-index-{$this->identifier}.xml", 'wb' );
 202                 $this->main();
 203         }
 204
 205         private function setNamespacePriorities() {
 206                 global $wgSitemapNamespacesPriorities;
 207
 208                 // Custom main namespaces
 209                 $this->priorities[self::GS_MAIN] = '0.5';
 210                 // Custom talk namesspaces
 211                 $this->priorities[self::GS_TALK] = '0.1';
 212                 // MediaWiki standard namespaces
 213                 $this->priorities[NS_MAIN] = '1.0';
 214                 $this->priorities[NS_TALK] = '0.1';
 215                 $this->priorities[NS_USER] = '0.5';
 216                 $this->priorities[NS_USER_TALK] = '0.1';
 217                 $this->priorities[NS_PROJECT] = '0.5';
 218                 $this->priorities[NS_PROJECT_TALK] = '0.1';
 219                 $this->priorities[NS_FILE] = '0.5';
 220                 $this->priorities[NS_FILE_TALK] = '0.1';
 221                 $this->priorities[NS_MEDIAWIKI] = '0.0';
 222                 $this->priorities[NS_MEDIAWIKI_TALK] = '0.1';
 223                 $this->priorities[NS_TEMPLATE] = '0.0';
 224                 $this->priorities[NS_TEMPLATE_TALK] = '0.1';
 225                 $this->priorities[NS_HELP] = '0.5';
 226                 $this->priorities[NS_HELP_TALK] = '0.1';
 227                 $this->priorities[NS_CATEGORY] = '0.5';
 228                 $this->priorities[NS_CATEGORY_TALK] = '0.1';
 229
 230                 // Custom priorities
 231                 if ( $wgSitemapNamespacesPriorities !== false ) {
 232                         /**
 233                          * @var array $wgSitemapNamespacesPriorities
 234                          */
 235                         foreach ( $wgSitemapNamespacesPriorities as $namespace => $priority ) {
 236                                 $float = floatval( $priority );
 237                                 if ( $float > 1.0 ) {
 238                                         $priority = '1.0';
 239                                 } elseif ( $float < 0.0 ) {
 240                                         $priority = '0.0';
 241                                 }
 242                                 $this->priorities[$namespace] = $priority;
 243                         }
 244                 }
 245         }
 246
 247         /**
 248          * Generate a one-dimensional array of existing namespaces
 249          */
 250         function generateNamespaces() {
 251                 // Only generate for specific namespaces if $wgSitemapNamespaces is an array.
 252                 global $wgSitemapNamespaces;
 253                 if ( is_array( $wgSitemapNamespaces ) ) {
 254                         $this->namespaces = $wgSitemapNamespaces;
 255
 256                         return;
 257                 }
 258
 259                 $res = $this->dbr->select( 'page',
 260                         [ 'page_namespace' ],
 261                         [],
 262                         __METHOD__,
 263                         [
 264                                 'GROUP BY' => 'page_namespace',
 265                                 'ORDER BY' => 'page_namespace',
 266                         ]
 267                 );
 268
 269                 foreach ( $res as $row ) {
 270                         $this->namespaces[] = $row->page_namespace;
 271                 }
 272         }
 273
 274         /**
 275          * Get the priority of a given namespace
 276          *
 277          * @param int $namespace The namespace to get the priority for
 278          * @return string
 279          */
 280         function priority( $namespace ) {
 281                 return $this->priorities[$namespace] ?? $this->guessPriority( $namespace );
 282         }
 283
 284         /**
 285          * If the namespace isn't listed on the priority list return the
 286          * default priority for the namespace, varies depending on whether it's
 287          * a talkpage or not.
 288          *
 289          * @param int $namespace The namespace to get the priority for
 290          * @return string
 291          */
 292         function guessPriority( $namespace ) {
 293                 return MediaWikiServices::getInstance()->getNamespaceInfo()->isSubject( $namespace )
 294                         ? $this->priorities[self::GS_MAIN]
 295                         : $this->priorities[self::GS_TALK];
 296         }
 297
 298         /**
 299          * Return a database resolution of all the pages in a given namespace
 300          *
 301          * @param int $namespace Limit the query to this namespace
 302          * @return Resource
 303          */
 304         function getPageRes( $namespace ) {
 305                 return $this->dbr->select( 'page',
 306                         [
 307                                 'page_namespace',
 308                                 'page_title',
 309                                 'page_touched',
 310                                 'page_is_redirect'
 311                         ],
 312                         [ 'page_namespace' => $namespace ],
 313                         __METHOD__
 314                 );
 315         }
 316
 317         /**
 318          * Main loop
 319          */
 320         public function main() {
 321                 $contLang = MediaWikiServices::getInstance()->getContentLanguage();
 322
 323                 fwrite( $this->findex, $this->openIndex() );
 324
 325                 foreach ( $this->namespaces as $namespace ) {
 326                         $res = $this->getPageRes( $namespace );
 327                         $this->file = false;
 328                         $this->generateLimit( $namespace );
 329                         $length = $this->limit[0];
 330                         $i = $smcount = 0;
 331
 332                         $fns = $contLang->getFormattedNsText( $namespace );
 333                         $this->output( "$namespace ($fns)\n" );
 334                         $skippedRedirects = 0; // Number of redirects skipped for that namespace
 335                         foreach ( $res as $row ) {
 336                                 if ( $this->skipRedirects && $row->page_is_redirect ) {
 337                                         $skippedRedirects++;
 338                                         continue;
 339                                 }
 340
 341                                 if ( $i++ === 0
 342                                         || $i === $this->url_limit + 1
 343                                         || $length + $this->limit[1] + $this->limit[2] > $this->size_limit
 344                                 ) {
 345                                         if ( $this->file !== false ) {
 346                                                 $this->write( $this->file, $this->closeFile() );
 347                                                 $this->close( $this->file );
 348                                         }
 349                                         $filename = $this->sitemapFilename( $namespace, $smcount++ );
 350                                         $this->file = $this->open( $this->fspath . $filename, 'wb' );
 351                                         $this->write( $this->file, $this->openFile() );
 352                                         fwrite( $this->findex, $this->indexEntry( $filename ) );
 353                                         $this->output( "\t$this->fspath$filename\n" );
 354                                         $length = $this->limit[0];
 355                                         $i = 1;
 356                                 }
 357                                 $title = Title::makeTitle( $row->page_namespace, $row->page_title );
 358                                 $date = wfTimestamp( TS_ISO_8601, $row->page_touched );
 359                                 $entry = $this->fileEntry( $title->getCanonicalURL(), $date, $this->priority( $namespace ) );
 360                                 $length += strlen( $entry );
 361                                 $this->write( $this->file, $entry );
 362                                 // generate pages for language variants
 363                                 if ( $contLang->hasVariants() ) {
 364                                         $variants = $contLang->getVariants();
 365                                         foreach ( $variants as $vCode ) {
 366                                                 if ( $vCode == $contLang->getCode() ) {
 367                                                         continue; // we don't want default variant
 368                                                 }
 369                                                 $entry = $this->fileEntry(
 370                                                         $title->getCanonicalURL( '', $vCode ),
 371                                                         $date,
 372                                                         $this->priority( $namespace )
 373                                                 );
 374                                                 $length += strlen( $entry );
 375                                                 $this->write( $this->file, $entry );
 376                                         }
 377                                 }
 378                         }
 379
 380                         if ( $this->skipRedirects && $skippedRedirects > 0 ) {
 381                                 $this->output( "  skipped $skippedRedirects redirect(s)\n" );
 382                         }
 383
 384                         if ( $this->file ) {
 385                                 $this->write( $this->file, $this->closeFile() );
 386                                 $this->close( $this->file );
 387                         }
 388                 }
 389                 fwrite( $this->findex, $this->closeIndex() );
 390                 fclose( $this->findex );
 391         }
 392
 393         /**
 394          * gzopen() / fopen() wrapper
 395          *
 396          * @param string $file
 397          * @param string $flags
 398          * @return resource
 399          */
 400         function open( $file, $flags ) {
 401                 $resource = $this->compress ? gzopen( $file, $flags ) : fopen( $file, $flags );
 402                 if ( $resource === false ) {
 403                         throw new MWException( __METHOD__
 404                                 . " error opening file $file with flags $flags. Check permissions?" );
 405                 }
 406
 407                 return $resource;
 408         }
 409
 410         /**
 411          * gzwrite() / fwrite() wrapper
 412          *
 413          * @param resource &$handle
 414          * @param string $str
 415          */
 416         function write( &$handle, $str ) {
 417                 if ( $handle === true || $handle === false ) {
 418                         throw new MWException( __METHOD__ . " was passed a boolean as a file handle.\n" );
 419                 }
 420                 if ( $this->compress ) {
 421                         gzwrite( $handle, $str );
 422                 } else {
 423                         fwrite( $handle, $str );
 424                 }
 425         }
 426
 427         /**
 428          * gzclose() / fclose() wrapper
 429          *
 430          * @param resource &$handle
 431          */
 432         function close( &$handle ) {
 433                 if ( $this->compress ) {
 434                         gzclose( $handle );
 435                 } else {
 436                         fclose( $handle );
 437                 }
 438         }
 439
 440         /**
 441          * Get a sitemap filename
 442          *
 443          * @param int $namespace
 444          * @param int $count
 445          * @return string
 446          */
 447         function sitemapFilename( $namespace, $count ) {
 448                 $ext = $this->compress ? '.gz' : '';
 449
 450                 return "sitemap-{$this->identifier}-NS_$namespace-$count.xml$ext";
 451         }
 452
 453         /**
 454          * Return the XML required to open an XML file
 455          *
 456          * @return string
 457          */
 458         function xmlHead() {
 459                 return '<?xml version="1.0" encoding="UTF-8"?>' . "\n";
 460         }
 461
 462         /**
 463          * Return the XML schema being used
 464          *
 465          * @return string
 466          */
 467         function xmlSchema() {
 468                 return 'http://www.sitemaps.org/schemas/sitemap/0.9';
 469         }
 470
 471         /**
 472          * Return the XML required to open a sitemap index file
 473          *
 474          * @return string
 475          */
 476         function openIndex() {
 477                 return $this->xmlHead() . '<sitemapindex xmlns="' . $this->xmlSchema() . '">' . "\n";
 478         }
 479
 480         /**
 481          * Return the XML for a single sitemap indexfile entry
 482          *
 483          * @param string $filename The filename of the sitemap file
 484          * @return string
 485          */
 486         function indexEntry( $filename ) {
 487                 return "\t<sitemap>\n" .
 488                         "\t\t<loc>" . wfGetServerUrl( PROTO_CANONICAL ) .
 489                                 ( substr( $this->urlpath, 0, 1 ) === "/" ? "" : "/" ) .
 490                                 "{$this->urlpath}$filename</loc>\n" .
 491                         "\t\t<lastmod>{$this->timestamp}</lastmod>\n" .
 492                         "\t</sitemap>\n";
 493         }
 494
 495         /**
 496          * Return the XML required to close a sitemap index file
 497          *
 498          * @return string
 499          */
 500         function closeIndex() {
 501                 return "</sitemapindex>\n";
 502         }
 503
 504         /**
 505          * Return the XML required to open a sitemap file
 506          *
 507          * @return string
 508          */
 509         function openFile() {
 510                 return $this->xmlHead() . '<urlset xmlns="' . $this->xmlSchema() . '">' . "\n";
 511         }
 512
 513         /**
 514          * Return the XML for a single sitemap entry
 515          *
 516          * @param string $url An RFC 2396 compliant URL
 517          * @param string $date A ISO 8601 date
 518          * @param string $priority A priority indicator, 0.0 - 1.0 inclusive with a 0.1 stepsize
 519          * @return string
 520          */
 521         function fileEntry( $url, $date, $priority ) {
 522                 return "\t<url>\n" .
 523                         // T36666: $url may contain bad characters such as ampersands.
 524                         "\t\t<loc>" . htmlspecialchars( $url ) . "</loc>\n" .
 525                         "\t\t<lastmod>$date</lastmod>\n" .
 526                         "\t\t<priority>$priority</priority>\n" .
 527                         "\t</url>\n";
 528         }
 529
 530         /**
 531          * Return the XML required to close sitemap file
 532          *
 533          * @return string
 534          */
 535         function closeFile() {
 536                 return "</urlset>\n";
 537         }
 538
 539         /**
 540          * Populate $this->limit
 541          *
 542          * @param int $namespace
 543          */
 544         function generateLimit( $namespace ) {
 545                 // T19961: make a title with the longest possible URL in this namespace
 546                 $title = Title::makeTitle( $namespace, str_repeat( "\u{28B81}", 63 ) . "\u{5583}" );
 547
 548                 $this->limit = [
 549                         strlen( $this->openFile() ),
 550                         strlen( $this->fileEntry(
 551                                 $title->getCanonicalURL(),
 552                                 wfTimestamp( TS_ISO_8601, wfTimestamp() ),
 553                                 $this->priority( $namespace )
 554                         ) ),
 555                         strlen( $this->closeFile() )
 556                 ];
 557         }
 558 }
 559
 560 $maintClass = GenerateSitemap::class;
 561 require_once RUN_MAINTENANCE_IF_MAIN;