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 = [];
  93
  94         /**
  95          * Key => value entries of namespaces and their priorities
  96          *
  97          * @var array
  98          */
  99         public $priorities = [];
 100
 101         /**
 102          * A one-dimensional array of namespaces in the wiki
 103          *
 104          * @var array
 105          */
 106         public $namespaces = [];
 107
 108         /**
 109          * When this sitemap batch was generated
 110          *
 111          * @var string
 112          */
 113         public $timestamp;
 114
 115         /**
 116          * A database replica DB 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         public function __construct() {
 144                 parent::__construct();
 145                 $this->addDescription( 'Creates a sitemap for the site' );
 146                 $this->addOption(
 147                         'fspath',
 148                         'The file system path to save to, e.g. /tmp/sitemap; defaults to current directory',
 149                         false,
 150                         true
 151                 );
 152                 $this->addOption(
 153                         'urlpath',
 154                         'The URL path corresponding to --fspath, prepended to filenames in the index; '
 155                                 . 'defaults to an empty string',
 156                         false,
 157                         true
 158                 );
 159                 $this->addOption(
 160                         'compress',
 161                         'Compress the sitemap files, can take value yes|no, default yes',
 162                         false,
 163                         true
 164                 );
 165                 $this->addOption( 'skip-redirects', 'Do not include redirecting articles in the sitemap' );
 166                 $this->addOption(
 167                         'identifier',
 168                         'What site identifier to use for the wiki, defaults to $wgDBname',
 169                         false,
 170                         true
 171                 );
 172         }
 173
 174         /**
 175          * Execute
 176          */
 177         public function execute() {
 178                 $this->setNamespacePriorities();
 179                 $this->url_limit = 50000;
 180                 $this->size_limit = pow( 2, 20 ) * 10;
 181
 182                 # Create directory if needed
 183                 $fspath = $this->getOption( 'fspath', getcwd() );
 184                 if ( !wfMkdirParents( $fspath, null, __METHOD__ ) ) {
 185                         $this->error( "Can not create directory $fspath.", 1 );
 186                 }
 187
 188                 $this->fspath = realpath( $fspath ) . DIRECTORY_SEPARATOR;
 189                 $this->urlpath = $this->getOption( 'urlpath', "" );
 190                 if ( $this->urlpath !== "" && substr( $this->urlpath, -1 ) !== '/' ) {
 191                         $this->urlpath .= '/';
 192                 }
 193                 $this->identifier = $this->getOption( 'identifier', wfWikiID() );
 194                 $this->compress = $this->getOption( 'compress', 'yes' ) !== 'no';
 195                 $this->skipRedirects = $this->hasOption( 'skip-redirects' );
 196                 $this->dbr = $this->getDB( DB_REPLICA );
 197                 $this->generateNamespaces();
 198                 $this->timestamp = wfTimestamp( TS_ISO_8601, wfTimestampNow() );
 199                 $this->findex = fopen( "{$this->fspath}sitemap-index-{$this->identifier}.xml", 'wb' );
 200                 $this->main();
 201         }
 202
 203         private function setNamespacePriorities() {
 204                 global $wgSitemapNamespacesPriorities;
 205
 206                 // Custom main namespaces
 207                 $this->priorities[self::GS_MAIN] = '0.5';
 208                 // Custom talk namesspaces
 209                 $this->priorities[self::GS_TALK] = '0.1';
 210                 // MediaWiki standard namespaces
 211                 $this->priorities[NS_MAIN] = '1.0';
 212                 $this->priorities[NS_TALK] = '0.1';
 213                 $this->priorities[NS_USER] = '0.5';
 214                 $this->priorities[NS_USER_TALK] = '0.1';
 215                 $this->priorities[NS_PROJECT] = '0.5';
 216                 $this->priorities[NS_PROJECT_TALK] = '0.1';
 217                 $this->priorities[NS_FILE] = '0.5';
 218                 $this->priorities[NS_FILE_TALK] = '0.1';
 219                 $this->priorities[NS_MEDIAWIKI] = '0.0';
 220                 $this->priorities[NS_MEDIAWIKI_TALK] = '0.1';
 221                 $this->priorities[NS_TEMPLATE] = '0.0';
 222                 $this->priorities[NS_TEMPLATE_TALK] = '0.1';
 223                 $this->priorities[NS_HELP] = '0.5';
 224                 $this->priorities[NS_HELP_TALK] = '0.1';
 225                 $this->priorities[NS_CATEGORY] = '0.5';
 226                 $this->priorities[NS_CATEGORY_TALK] = '0.1';
 227
 228                 // Custom priorities
 229                 if ( $wgSitemapNamespacesPriorities !== false ) {
 230                         /**
 231                          * @var $wgSitemapNamespacesPriorities array
 232                          */
 233                         foreach ( $wgSitemapNamespacesPriorities as $namespace => $priority ) {
 234                                 $float = floatval( $priority );
 235                                 if ( $float > 1.0 ) {
 236                                         $priority = '1.0';
 237                                 } elseif ( $float < 0.0 ) {
 238                                         $priority = '0.0';
 239                                 }
 240                                 $this->priorities[$namespace] = $priority;
 241                         }
 242                 }
 243         }
 244
 245         /**
 246          * Generate a one-dimensional array of existing namespaces
 247          */
 248         function generateNamespaces() {
 249                 // Only generate for specific namespaces if $wgSitemapNamespaces is an array.
 250                 global $wgSitemapNamespaces;
 251                 if ( is_array( $wgSitemapNamespaces ) ) {
 252                         $this->namespaces = $wgSitemapNamespaces;
 253
 254                         return;
 255                 }
 256
 257                 $res = $this->dbr->select( 'page',
 258                         [ 'page_namespace' ],
 259                         [],
 260                         __METHOD__,
 261                         [
 262                                 'GROUP BY' => 'page_namespace',
 263                                 'ORDER BY' => 'page_namespace',
 264                         ]
 265                 );
 266
 267                 foreach ( $res as $row ) {
 268                         $this->namespaces[] = $row->page_namespace;
 269                 }
 270         }
 271
 272         /**
 273          * Get the priority of a given namespace
 274          *
 275          * @param int $namespace The namespace to get the priority for
 276          * @return string
 277          */
 278         function priority( $namespace ) {
 279                 return isset( $this->priorities[$namespace] )
 280                         ? $this->priorities[$namespace]
 281                         : $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 MWNamespace::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                 global $wgContLang;
 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 = $wgContLang->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 ( $wgContLang->hasVariants() ) {
 364                                         $variants = $wgContLang->getVariants();
 365                                         foreach ( $variants as $vCode ) {
 366                                                 if ( $vCode == $wgContLang->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 The namespace
 444          * @param int $count The 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( $filename ) {
 487                 return
 488                         "\t<sitemap>\n" .
 489                         "\t\t<loc>{$this->urlpath}$filename</loc>\n" .
 490                         "\t\t<lastmod>{$this->timestamp}</lastmod>\n" .
 491                         "\t</sitemap>\n";
 492         }
 493
 494         /**
 495          * Return the XML required to close a sitemap index file
 496          *
 497          * @return string
 498          */
 499         function closeIndex() {
 500                 return "</sitemapindex>\n";
 501         }
 502
 503         /**
 504          * Return the XML required to open a sitemap file
 505          *
 506          * @return string
 507          */
 508         function openFile() {
 509                 return $this->xmlHead() . '<urlset xmlns="' . $this->xmlSchema() . '">' . "\n";
 510         }
 511
 512         /**
 513          * Return the XML for a single sitemap entry
 514          *
 515          * @param string $url An RFC 2396 compliant URL
 516          * @param string $date A ISO 8601 date
 517          * @param string $priority A priority indicator, 0.0 - 1.0 inclusive with a 0.1 stepsize
 518          * @return string
 519          */
 520         function fileEntry( $url, $date, $priority ) {
 521                 return
 522                         "\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( "\xf0\xa8\xae\x81", 63 ) . "\xe5\x96\x83" );
 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";
 561 require_once RUN_MAINTENANCE_IF_MAIN;