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( dirname( __FILE__ ) . '/Maintenance.php' );
  30
  31 class GenerateSitemap extends Maintenance {
  32         const GS_MAIN = -2;
  33         const GS_TALK = -1;
  34
  35         /**
  36          * The maximum amount of urls in a sitemap file
  37          *
  38          * @link http://www.sitemaps.org/schemas/sitemap/0.9/sitemap.xsd
  39          *
  40          * @var int
  41          */
  42         var $url_limit;
  43
  44         /**
  45          * The maximum size of a sitemap file
  46          *
  47          * @link http://www.sitemaps.org/faq.php#faq_sitemap_size
  48          *
  49          * @var int
  50          */
  51         var $size_limit;
  52
  53         /**
  54          * The path to prepend to the filename
  55          *
  56          * @var string
  57          */
  58         var $fspath;
  59
  60         /**
  61          * The URL path to prepend to filenames in the index; should resolve to the same directory as $fspath
  62          *
  63          * @var string
  64          */
  65         var $urlpath;
  66
  67         /**
  68          * Whether or not to use compression
  69          *
  70          * @var bool
  71          */
  72         var $compress;
  73
  74         /**
  75          * The number of entries to save in each sitemap file
  76          *
  77          * @var array
  78          */
  79         var $limit = array();
  80
  81         /**
  82          * Key => value entries of namespaces and their priorities
  83          *
  84          * @var array
  85          */
  86         var $priorities = array();
  87
  88         /**
  89          * A one-dimensional array of namespaces in the wiki
  90          *
  91          * @var array
  92          */
  93         var $namespaces = array();
  94
  95         /**
  96          * When this sitemap batch was generated
  97          *
  98          * @var string
  99          */
 100         var $timestamp;
 101
 102         /**
 103          * A database slave object
 104          *
 105          * @var object
 106          */
 107         var $dbr;
 108
 109         /**
 110          * A resource pointing to the sitemap index file
 111          *
 112          * @var resource
 113          */
 114         var $findex;
 115
 116
 117         /**
 118          * A resource pointing to a sitemap file
 119          *
 120          * @var resource
 121          */
 122         var $file;
 123
 124         /**
 125          * Identifier to use in filenames, default $wgDBname
 126          *
 127          * @var string
 128          */
 129         private $identifier;
 130
 131         /**
 132          * Constructor
 133          */
 134         public function __construct() {
 135                 parent::__construct();
 136                 $this->mDescription = "Creates a sitemap for the site";
 137                 $this->addOption( 'fspath', 'The file system path to save to, e.g. /tmp/sitemap; defaults to current directory', false, true );
 138                 $this->addOption( 'urlpath', 'The URL path corresponding to --fspath, prepended to filenames in the index; defaults to an empty string', false, true );
 139                 $this->addOption( 'compress', 'Compress the sitemap files, can take value yes|no, default yes', false, true );
 140                 $this->addOption( 'identifier', 'What site identifier to use for the wiki, defaults to $wgDBname', false, true );
 141         }
 142
 143         /**
 144          * Execute
 145          */
 146         public function execute() {
 147                 $this->setNamespacePriorities();
 148                 $this->url_limit = 50000;
 149                 $this->size_limit = pow( 2, 20 ) * 10;
 150                 $this->fspath = self::init_path( $this->getOption( 'fspath', getcwd() ) );
 151                 $this->urlpath = $this->getOption( 'urlpath', "" );
 152                 if ( $this->urlpath !== "" && substr( $this->urlpath, -1 ) !== '/' ) {
 153                         $this->urlpath .= '/';
 154                 }
 155                 $this->identifier = $this->getOption( 'identifier', wfWikiID() );
 156                 $this->compress = $this->getOption( 'compress', 'yes' ) !== 'no';
 157                 $this->dbr = wfGetDB( DB_SLAVE );
 158                 $this->generateNamespaces();
 159                 $this->timestamp = wfTimestamp( TS_ISO_8601, wfTimestampNow() );
 160                 $this->findex = fopen( "{$this->fspath}sitemap-index-{$this->identifier}.xml", 'wb' );
 161                 $this->main();
 162         }
 163
 164         private function setNamespacePriorities() {
 165                 global $wgSitemapNamespacesPriorities;
 166
 167                 // Custom main namespaces
 168                 $this->priorities[self::GS_MAIN] = '0.5';
 169                 // Custom talk namesspaces
 170                 $this->priorities[self::GS_TALK] = '0.1';
 171                 // MediaWiki standard namespaces
 172                 $this->priorities[NS_MAIN] = '1.0';
 173                 $this->priorities[NS_TALK] = '0.1';
 174                 $this->priorities[NS_USER] = '0.5';
 175                 $this->priorities[NS_USER_TALK] = '0.1';
 176                 $this->priorities[NS_PROJECT] = '0.5';
 177                 $this->priorities[NS_PROJECT_TALK] = '0.1';
 178                 $this->priorities[NS_FILE] = '0.5';
 179                 $this->priorities[NS_FILE_TALK] = '0.1';
 180                 $this->priorities[NS_MEDIAWIKI] = '0.0';
 181                 $this->priorities[NS_MEDIAWIKI_TALK] = '0.1';
 182                 $this->priorities[NS_TEMPLATE] = '0.0';
 183                 $this->priorities[NS_TEMPLATE_TALK] = '0.1';
 184                 $this->priorities[NS_HELP] = '0.5';
 185                 $this->priorities[NS_HELP_TALK] = '0.1';
 186                 $this->priorities[NS_CATEGORY] = '0.5';
 187                 $this->priorities[NS_CATEGORY_TALK] = '0.1';
 188
 189                 // Custom priorities
 190                 if ( $wgSitemapNamespacesPriorities !== false ) {
 191                         /**
 192                          * @var $wgSitemapNamespacesPriorities array
 193                          */
 194                         foreach ( $wgSitemapNamespacesPriorities as $namespace => $priority ) {
 195                                 $float = floatval( $priority );
 196                                 if ( $float > 1.0 ) {
 197                                         $priority = '1.0';
 198                                 } elseif ( $float < 0.0 ) {
 199                                         $priority = '0.0';
 200                                 }
 201                                 $this->priorities[$namespace] = $priority;
 202                         }
 203                 }
 204         }
 205
 206         /**
 207          * Create directory if it does not exist and return pathname with a trailing slash
 208          * @param $fspath string
 209          * @return null|string
 210          */
 211         private static function init_path( $fspath ) {
 212                 if ( !isset( $fspath ) ) {
 213                         return null;
 214                 }
 215                 # Create directory if needed
 216                 if ( $fspath && !is_dir( $fspath ) ) {
 217                         wfMkdirParents( $fspath, null, __METHOD__ ) or die( "Can not create directory $fspath.\n" );
 218                 }
 219
 220                 return realpath( $fspath ) . DIRECTORY_SEPARATOR ;
 221         }
 222
 223         /**
 224          * Generate a one-dimensional array of existing namespaces
 225          */
 226         function generateNamespaces() {
 227                 // Only generate for specific namespaces if $wgSitemapNamespaces is an array.
 228                 global $wgSitemapNamespaces;
 229                 if ( is_array( $wgSitemapNamespaces ) ) {
 230                         $this->namespaces = $wgSitemapNamespaces;
 231                         return;
 232                 }
 233
 234                 $res = $this->dbr->select( 'page',
 235                         array( 'page_namespace' ),
 236                         array(),
 237                         __METHOD__,
 238                         array(
 239                                 'GROUP BY' => 'page_namespace',
 240                                 'ORDER BY' => 'page_namespace',
 241                         )
 242                 );
 243
 244                 foreach ( $res as $row )
 245                         $this->namespaces[] = $row->page_namespace;
 246         }
 247
 248         /**
 249          * Get the priority of a given namespace
 250          *
 251          * @param $namespace Integer: the namespace to get the priority for
 252          * @return String
 253          */
 254         function priority( $namespace ) {
 255                 return isset( $this->priorities[$namespace] ) ? $this->priorities[$namespace] : $this->guessPriority( $namespace );
 256         }
 257
 258         /**
 259          * If the namespace isn't listed on the priority list return the
 260          * default priority for the namespace, varies depending on whether it's
 261          * a talkpage or not.
 262          *
 263          * @param $namespace Integer: the namespace to get the priority for
 264          * @return String
 265          */
 266         function guessPriority( $namespace ) {
 267                 return MWNamespace::isSubject( $namespace ) ? $this->priorities[self::GS_MAIN] : $this->priorities[self::GS_TALK];
 268         }
 269
 270         /**
 271          * Return a database resolution of all the pages in a given namespace
 272          *
 273          * @param $namespace Integer: limit the query to this namespace
 274          * @return Resource
 275          */
 276         function getPageRes( $namespace ) {
 277                 return $this->dbr->select( 'page',
 278                         array(
 279                                 'page_namespace',
 280                                 'page_title',
 281                                 'page_touched',
 282                         ),
 283                         array( 'page_namespace' => $namespace ),
 284                         __METHOD__
 285                 );
 286         }
 287
 288         /**
 289          * Main loop
 290          */
 291         public function main() {
 292                 global $wgContLang;
 293
 294                 fwrite( $this->findex, $this->openIndex() );
 295
 296                 foreach ( $this->namespaces as $namespace ) {
 297                         $res = $this->getPageRes( $namespace );
 298                         $this->file = false;
 299                         $this->generateLimit( $namespace );
 300                         $length = $this->limit[0];
 301                         $i = $smcount = 0;
 302
 303                         $fns = $wgContLang->getFormattedNsText( $namespace );
 304                         $this->output( "$namespace ($fns)\n" );
 305                         foreach ( $res as $row ) {
 306                                 if ( $i++ === 0 || $i === $this->url_limit + 1 || $length + $this->limit[1] + $this->limit[2] > $this->size_limit ) {
 307                                         if ( $this->file !== false ) {
 308                                                 $this->write( $this->file, $this->closeFile() );
 309                                                 $this->close( $this->file );
 310                                         }
 311                                         $filename = $this->sitemapFilename( $namespace, $smcount++ );
 312                                         $this->file = $this->open( $this->fspath . $filename, 'wb' );
 313                                         $this->write( $this->file, $this->openFile() );
 314                                         fwrite( $this->findex, $this->indexEntry( $filename ) );
 315                                         $this->output( "\t$this->fspath$filename\n" );
 316                                         $length = $this->limit[0];
 317                                         $i = 1;
 318                                 }
 319                                 $title = Title::makeTitle( $row->page_namespace, $row->page_title );
 320                                 $date = wfTimestamp( TS_ISO_8601, $row->page_touched );
 321                                 $entry = $this->fileEntry( $title->getCanonicalURL(), $date, $this->priority( $namespace ) );
 322                                 $length += strlen( $entry );
 323                                 $this->write( $this->file, $entry );
 324                                 // generate pages for language variants
 325                                 if ( $wgContLang->hasVariants() ) {
 326                                         $variants = $wgContLang->getVariants();
 327                                         foreach ( $variants as $vCode ) {
 328                                                 if ( $vCode == $wgContLang->getCode() ) continue; // we don't want default variant
 329                                                 $entry = $this->fileEntry( $title->getCanonicalURL( '', $vCode ), $date, $this->priority( $namespace ) );
 330                                                 $length += strlen( $entry );
 331                                                 $this->write( $this->file, $entry );
 332                                         }
 333                                 }
 334                         }
 335                         if ( $this->file ) {
 336                                 $this->write( $this->file, $this->closeFile() );
 337                                 $this->close( $this->file );
 338                         }
 339                 }
 340                 fwrite( $this->findex, $this->closeIndex() );
 341                 fclose( $this->findex );
 342         }
 343
 344         /**
 345          * gzopen() / fopen() wrapper
 346          *
 347          * @return Resource
 348          */
 349         function open( $file, $flags ) {
 350                 $resource = $this->compress ? gzopen( $file, $flags ) : fopen( $file, $flags );
 351                 if( $resource === false ) {
 352                         wfDebugDieBacktrace( __METHOD__ . " error opening file $file with flags $flags. Check permissions?" );
 353                 }
 354                 return $resource;
 355         }
 356
 357         /**
 358          * gzwrite() / fwrite() wrapper
 359          */
 360         function write( &$handle, $str ) {
 361                 if( $handle === true || $handle === false ) {
 362                         wfDebugDieBacktrace( __METHOD__ . " was passed a boolean as a file handle.\n" );
 363                 }
 364                 if ( $this->compress )
 365                         gzwrite( $handle, $str );
 366                 else
 367                         fwrite( $handle, $str );
 368         }
 369
 370         /**
 371          * gzclose() / fclose() wrapper
 372          */
 373         function close( &$handle ) {
 374                 if ( $this->compress )
 375                         gzclose( $handle );
 376                 else
 377                         fclose( $handle );
 378         }
 379
 380         /**
 381          * Get a sitemap filename
 382          *
 383          * @param $namespace Integer: the namespace
 384          * @param $count Integer: the count
 385          * @return String
 386          */
 387         function sitemapFilename( $namespace, $count ) {
 388                 $ext = $this->compress ? '.gz' : '';
 389                 return "sitemap-{$this->identifier}-NS_$namespace-$count.xml$ext";
 390         }
 391
 392         /**
 393          * Return the XML required to open an XML file
 394          *
 395          * @return string
 396          */
 397         function xmlHead() {
 398                 return '<?xml version="1.0" encoding="UTF-8"?>' . "\n";
 399         }
 400
 401         /**
 402          * Return the XML schema being used
 403          *
 404          * @return String
 405          */
 406         function xmlSchema() {
 407                 return 'http://www.sitemaps.org/schemas/sitemap/0.9';
 408         }
 409
 410         /**
 411          * Return the XML required to open a sitemap index file
 412          *
 413          * @return String
 414          */
 415         function openIndex() {
 416                 return $this->xmlHead() . '<sitemapindex xmlns="' . $this->xmlSchema() . '">' . "\n";
 417         }
 418
 419         /**
 420          * Return the XML for a single sitemap indexfile entry
 421          *
 422          * @param $filename String: the filename of the sitemap file
 423          * @return String
 424          */
 425         function indexEntry( $filename ) {
 426                 return
 427                         "\t<sitemap>\n" .
 428                         "\t\t<loc>{$this->urlpath}$filename</loc>\n" .
 429                         "\t\t<lastmod>{$this->timestamp}</lastmod>\n" .
 430                         "\t</sitemap>\n";
 431         }
 432
 433         /**
 434          * Return the XML required to close a sitemap index file
 435          *
 436          * @return String
 437          */
 438         function closeIndex() {
 439                 return "</sitemapindex>\n";
 440         }
 441
 442         /**
 443          * Return the XML required to open a sitemap file
 444          *
 445          * @return String
 446          */
 447         function openFile() {
 448                 return $this->xmlHead() . '<urlset xmlns="' . $this->xmlSchema() . '">' . "\n";
 449         }
 450
 451         /**
 452          * Return the XML for a single sitemap entry
 453          *
 454          * @param $url String: an RFC 2396 compliant URL
 455          * @param $date String: a ISO 8601 date
 456          * @param $priority String: a priority indicator, 0.0 - 1.0 inclusive with a 0.1 stepsize
 457          * @return String
 458          */
 459         function fileEntry( $url, $date, $priority ) {
 460                 return
 461                         "\t<url>\n" .
 462                         "\t\t<loc>$url</loc>\n" .
 463                         "\t\t<lastmod>$date</lastmod>\n" .
 464                         "\t\t<priority>$priority</priority>\n" .
 465                         "\t</url>\n";
 466         }
 467
 468         /**
 469          * Return the XML required to close sitemap file
 470          *
 471          * @return String
 472          */
 473         function closeFile() {
 474                 return "</urlset>\n";
 475         }
 476
 477         /**
 478          * Populate $this->limit
 479          */
 480         function generateLimit( $namespace ) {
 481                 // bug 17961: make a title with the longest possible URL in this namespace
 482                 $title = Title::makeTitle( $namespace, str_repeat( "\xf0\xa8\xae\x81", 63 ) . "\xe5\x96\x83" );
 483
 484                 $this->limit = array(
 485                         strlen( $this->openFile() ),
 486                         strlen( $this->fileEntry( $title->getCanonicalURL(), wfTimestamp( TS_ISO_8601, wfTimestamp() ), $this->priority( $namespace ) ) ),
 487                         strlen( $this->closeFile() )
 488                 );
 489         }
 490 }
 491
 492 $maintClass = "GenerateSitemap";
 493 require_once( RUN_MAINTENANCE_IF_MAIN );