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
 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                 $this->fspath = realpath( $fspath ) . DIRECTORY_SEPARATOR;
 192                 $this->urlpath = $this->getOption( 'urlpath', "" );
 193                 if ( $this->urlpath !== "" && substr( $this->urlpath, -1 ) !== '/' ) {
 194                         $this->urlpath .= '/';
 195                 }
 196                 $this->identifier = $this->getOption( 'identifier', wfWikiID() );
 197                 $this->compress = $this->getOption( 'compress', 'yes' ) !== 'no';
 198                 $this->skipRedirects = $this->hasOption( 'skip-redirects' );
 199                 $this->dbr = $this->getDB( DB_REPLICA );
 200                 $this->generateNamespaces();
 201                 $this->timestamp = wfTimestamp( TS_ISO_8601, wfTimestampNow() );
 202                 $this->findex = fopen( "{$this->fspath}sitemap-index-{$this->identifier}.xml", 'wb' );
 203                 $this->main();
 204         }
 205
 206         private function setNamespacePriorities() {
 207                 global $wgSitemapNamespacesPriorities;
 208
 209                 // Custom main namespaces
 210                 $this->priorities[self::GS_MAIN] = '0.5';
 211                 // Custom talk namesspaces
 212                 $this->priorities[self::GS_TALK] = '0.1';
 213                 // MediaWiki standard namespaces
 214                 $this->priorities[NS_MAIN] = '1.0';
 215                 $this->priorities[NS_TALK] = '0.1';
 216                 $this->priorities[NS_USER] = '0.5';
 217                 $this->priorities[NS_USER_TALK] = '0.1';
 218                 $this->priorities[NS_PROJECT] = '0.5';
 219                 $this->priorities[NS_PROJECT_TALK] = '0.1';
 220                 $this->priorities[NS_FILE] = '0.5';
 221                 $this->priorities[NS_FILE_TALK] = '0.1';
 222                 $this->priorities[NS_MEDIAWIKI] = '0.0';
 223                 $this->priorities[NS_MEDIAWIKI_TALK] = '0.1';
 224                 $this->priorities[NS_TEMPLATE] = '0.0';
 225                 $this->priorities[NS_TEMPLATE_TALK] = '0.1';
 226                 $this->priorities[NS_HELP] = '0.5';
 227                 $this->priorities[NS_HELP_TALK] = '0.1';
 228                 $this->priorities[NS_CATEGORY] = '0.5';
 229                 $this->priorities[NS_CATEGORY_TALK] = '0.1';
 230
 231                 // Custom priorities
 232                 if ( $wgSitemapNamespacesPriorities !== false ) {
 233                         /**
 234                          * @var array $wgSitemapNamespacesPriorities
 235                          */
 236                         foreach ( $wgSitemapNamespacesPriorities as $namespace => $priority ) {
 237                                 $float = floatval( $priority );
 238                                 if ( $float > 1.0 ) {
 239                                         $priority = '1.0';
 240                                 } elseif ( $float < 0.0 ) {
 241                                         $priority = '0.0';
 242                                 }
 243                                 $this->priorities[$namespace] = $priority;
 244                         }
 245                 }
 246         }
 247
 248         /**
 249          * Generate a one-dimensional array of existing namespaces
 250          */
 251         function generateNamespaces() {
 252                 // Only generate for specific namespaces if $wgSitemapNamespaces is an array.
 253                 global $wgSitemapNamespaces;
 254                 if ( is_array( $wgSitemapNamespaces ) ) {
 255                         $this->namespaces = $wgSitemapNamespaces;
 256
 257                         return;
 258                 }
 259
 260                 $res = $this->dbr->select( 'page',
 261                         [ 'page_namespace' ],
 262                         [],
 263                         __METHOD__,
 264                         [
 265                                 'GROUP BY' => 'page_namespace',
 266                                 'ORDER BY' => 'page_namespace',
 267                         ]
 268                 );
 269
 270                 foreach ( $res as $row ) {
 271                         $this->namespaces[] = $row->page_namespace;
 272                 }
 273         }
 274
 275         /**
 276          * Get the priority of a given namespace
 277          *
 278          * @param int $namespace The namespace to get the priority for
 279          * @return string
 280          */
 281         function priority( $namespace ) {
 282                 return $this->priorities[$namespace] ?? $this->guessPriority( $namespace );
 283         }
 284
 285         /**
 286          * If the namespace isn't listed on the priority list return the
 287          * default priority for the namespace, varies depending on whether it's
 288          * a talkpage or not.
 289          *
 290          * @param int $namespace The namespace to get the priority for
 291          * @return string
 292          */
 293         function guessPriority( $namespace ) {
 294                 return MediaWikiServices::getInstance()->getNamespaceInfo()->isSubject( $namespace )
 295                         ? $this->priorities[self::GS_MAIN]
 296                         : $this->priorities[self::GS_TALK];
 297         }
 298
 299         /**
 300          * Return a database resolution of all the pages in a given namespace
 301          *
 302          * @param int $namespace Limit the query to this namespace
 303          * @return IResultWrapper
 304          */
 305         function getPageRes( $namespace ) {
 306                 return $this->dbr->select( 'page',
 307                         [
 308                                 'page_namespace',
 309                                 'page_title',
 310                                 'page_touched',
 311                                 'page_is_redirect'
 312                         ],
 313                         [ 'page_namespace' => $namespace ],
 314                         __METHOD__
 315                 );
 316         }
 317
 318         /**
 319          * Main loop
 320          */
 321         public function main() {
 322                 $contLang = MediaWikiServices::getInstance()->getContentLanguage();
 323
 324                 fwrite( $this->findex, $this->openIndex() );
 325
 326                 foreach ( $this->namespaces as $namespace ) {
 327                         $res = $this->getPageRes( $namespace );
 328                         $this->file = false;
 329                         $this->generateLimit( $namespace );
 330                         $length = $this->limit[0];
 331                         $i = $smcount = 0;
 332
 333                         $fns = $contLang->getFormattedNsText( $namespace );
 334                         $this->output( "$namespace ($fns)\n" );
 335                         $skippedRedirects = 0; // Number of redirects skipped for that namespace
 336                         foreach ( $res as $row ) {
 337                                 if ( $this->skipRedirects && $row->page_is_redirect ) {
 338                                         $skippedRedirects++;
 339                                         continue;
 340                                 }
 341
 342                                 if ( $i++ === 0
 343                                         || $i === $this->url_limit + 1
 344                                         || $length + $this->limit[1] + $this->limit[2] > $this->size_limit
 345                                 ) {
 346                                         if ( $this->file !== false ) {
 347                                                 $this->write( $this->file, $this->closeFile() );
 348                                                 $this->close( $this->file );
 349                                         }
 350                                         $filename = $this->sitemapFilename( $namespace, $smcount++ );
 351                                         $this->file = $this->open( $this->fspath . $filename, 'wb' );
 352                                         $this->write( $this->file, $this->openFile() );
 353                                         fwrite( $this->findex, $this->indexEntry( $filename ) );
 354                                         $this->output( "\t$this->fspath$filename\n" );
 355                                         $length = $this->limit[0];
 356                                         $i = 1;
 357                                 }
 358                                 $title = Title::makeTitle( $row->page_namespace, $row->page_title );
 359                                 $date = wfTimestamp( TS_ISO_8601, $row->page_touched );
 360                                 $entry = $this->fileEntry( $title->getCanonicalURL(), $date, $this->priority( $namespace ) );
 361                                 $length += strlen( $entry );
 362                                 $this->write( $this->file, $entry );
 363                                 // generate pages for language variants
 364                                 if ( $contLang->hasVariants() ) {
 365                                         $variants = $contLang->getVariants();
 366                                         foreach ( $variants as $vCode ) {
 367                                                 if ( $vCode == $contLang->getCode() ) {
 368                                                         continue; // we don't want default variant
 369                                                 }
 370                                                 $entry = $this->fileEntry(
 371                                                         $title->getCanonicalURL( '', $vCode ),
 372                                                         $date,
 373                                                         $this->priority( $namespace )
 374                                                 );
 375                                                 $length += strlen( $entry );
 376                                                 $this->write( $this->file, $entry );
 377                                         }
 378                                 }
 379                         }
 380
 381                         if ( $this->skipRedirects && $skippedRedirects > 0 ) {
 382                                 $this->output( "  skipped $skippedRedirects redirect(s)\n" );
 383                         }
 384
 385                         if ( $this->file ) {
 386                                 $this->write( $this->file, $this->closeFile() );
 387                                 $this->close( $this->file );
 388                         }
 389                 }
 390                 fwrite( $this->findex, $this->closeIndex() );
 391                 fclose( $this->findex );
 392         }
 393
 394         /**
 395          * gzopen() / fopen() wrapper
 396          *
 397          * @param string $file
 398          * @param string $flags
 399          * @return resource
 400          */
 401         function open( $file, $flags ) {
 402                 $resource = $this->compress ? gzopen( $file, $flags ) : fopen( $file, $flags );
 403                 if ( $resource === false ) {
 404                         throw new MWException( __METHOD__
 405                                 . " error opening file $file with flags $flags. Check permissions?" );
 406                 }
 407
 408                 return $resource;
 409         }
 410
 411         /**
 412          * gzwrite() / fwrite() wrapper
 413          *
 414          * @param resource &$handle
 415          * @param string $str
 416          */
 417         function write( &$handle, $str ) {
 418                 if ( $handle === true || $handle === false ) {
 419                         throw new MWException( __METHOD__ . " was passed a boolean as a file handle.\n" );
 420                 }
 421                 if ( $this->compress ) {
 422                         gzwrite( $handle, $str );
 423                 } else {
 424                         fwrite( $handle, $str );
 425                 }
 426         }
 427
 428         /**
 429          * gzclose() / fclose() wrapper
 430          *
 431          * @param resource &$handle
 432          */
 433         function close( &$handle ) {
 434                 if ( $this->compress ) {
 435                         gzclose( $handle );
 436                 } else {
 437                         fclose( $handle );
 438                 }
 439         }
 440
 441         /**
 442          * Get a sitemap filename
 443          *
 444          * @param int $namespace
 445          * @param int $count
 446          * @return string
 447          */
 448         function sitemapFilename( $namespace, $count ) {
 449                 $ext = $this->compress ? '.gz' : '';
 450
 451                 return "sitemap-{$this->identifier}-NS_$namespace-$count.xml$ext";
 452         }
 453
 454         /**
 455          * Return the XML required to open an XML file
 456          *
 457          * @return string
 458          */
 459         function xmlHead() {
 460                 return '<?xml version="1.0" encoding="UTF-8"?>' . "\n";
 461         }
 462
 463         /**
 464          * Return the XML schema being used
 465          *
 466          * @return string
 467          */
 468         function xmlSchema() {
 469                 return 'http://www.sitemaps.org/schemas/sitemap/0.9';
 470         }
 471
 472         /**
 473          * Return the XML required to open a sitemap index file
 474          *
 475          * @return string
 476          */
 477         function openIndex() {
 478                 return $this->xmlHead() . '<sitemapindex xmlns="' . $this->xmlSchema() . '">' . "\n";
 479         }
 480
 481         /**
 482          * Return the XML for a single sitemap indexfile entry
 483          *
 484          * @param string $filename The filename of the sitemap file
 485          * @return string
 486          */
 487         function indexEntry( $filename ) {
 488                 return "\t<sitemap>\n" .
 489                         "\t\t<loc>" . wfGetServerUrl( PROTO_CANONICAL ) .
 490                                 ( substr( $this->urlpath, 0, 1 ) === "/" ? "" : "/" ) .
 491                                 "{$this->urlpath}$filename</loc>\n" .
 492                         "\t\t<lastmod>{$this->timestamp}</lastmod>\n" .
 493                         "\t</sitemap>\n";
 494         }
 495
 496         /**
 497          * Return the XML required to close a sitemap index file
 498          *
 499          * @return string
 500          */
 501         function closeIndex() {
 502                 return "</sitemapindex>\n";
 503         }
 504
 505         /**
 506          * Return the XML required to open a sitemap file
 507          *
 508          * @return string
 509          */
 510         function openFile() {
 511                 return $this->xmlHead() . '<urlset xmlns="' . $this->xmlSchema() . '">' . "\n";
 512         }
 513
 514         /**
 515          * Return the XML for a single sitemap entry
 516          *
 517          * @param string $url An RFC 2396 compliant URL
 518          * @param string $date A ISO 8601 date
 519          * @param string $priority A priority indicator, 0.0 - 1.0 inclusive with a 0.1 stepsize
 520          * @return string
 521          */
 522         function fileEntry( $url, $date, $priority ) {
 523                 return "\t<url>\n" .
 524                         // T36666: $url may contain bad characters such as ampersands.
 525                         "\t\t<loc>" . htmlspecialchars( $url ) . "</loc>\n" .
 526                         "\t\t<lastmod>$date</lastmod>\n" .
 527                         "\t\t<priority>$priority</priority>\n" .
 528                         "\t</url>\n";
 529         }
 530
 531         /**
 532          * Return the XML required to close sitemap file
 533          *
 534          * @return string
 535          */
 536         function closeFile() {
 537                 return "</urlset>\n";
 538         }
 539
 540         /**
 541          * Populate $this->limit
 542          *
 543          * @param int $namespace
 544          */
 545         function generateLimit( $namespace ) {
 546                 // T19961: make a title with the longest possible URL in this namespace
 547                 $title = Title::makeTitle( $namespace, str_repeat( "\u{28B81}", 63 ) . "\u{5583}" );
 548
 549                 $this->limit = [
 550                         strlen( $this->openFile() ),
 551                         strlen( $this->fileEntry(
 552                                 $title->getCanonicalURL(),
 553                                 wfTimestamp( TS_ISO_8601, wfTimestamp() ),
 554                                 $this->priority( $namespace )
 555                         ) ),
 556                         strlen( $this->closeFile() )
 557                 ];
 558         }
 559 }
 560
 561 $maintClass = GenerateSitemap::class;
 562 require_once RUN_MAINTENANCE_IF_MAIN;