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          * Whether or not to include redirection pages
  76          *
  77          * @var bool
  78          */
  79         var $skipRedirects;
  80
  81         /**
  82          * The number of entries to save in each sitemap file
  83          *
  84          * @var array
  85          */
  86         var $limit = array();
  87
  88         /**
  89          * Key => value entries of namespaces and their priorities
  90          *
  91          * @var array
  92          */
  93         var $priorities = array();
  94
  95         /**
  96          * A one-dimensional array of namespaces in the wiki
  97          *
  98          * @var array
  99          */
 100         var $namespaces = array();
 101
 102         /**
 103          * When this sitemap batch was generated
 104          *
 105          * @var string
 106          */
 107         var $timestamp;
 108
 109         /**
 110          * A database slave object
 111          *
 112          * @var object
 113          */
 114         var $dbr;
 115
 116         /**
 117          * A resource pointing to the sitemap index file
 118          *
 119          * @var resource
 120          */
 121         var $findex;
 122
 123
 124         /**
 125          * A resource pointing to a sitemap file
 126          *
 127          * @var resource
 128          */
 129         var $file;
 130
 131         /**
 132          * Identifier to use in filenames, default $wgDBname
 133          *
 134          * @var string
 135          */
 136         private $identifier;
 137
 138         /**
 139          * Constructor
 140          */
 141         public function __construct() {
 142                 parent::__construct();
 143                 $this->mDescription = "Creates a sitemap for the site";
 144                 $this->addOption( 'fspath', 'The file system path to save to, e.g. /tmp/sitemap; defaults to current directory', false, true );
 145                 $this->addOption( 'urlpath', 'The URL path corresponding to --fspath, prepended to filenames in the index; defaults to an empty string', false, true );
 146                 $this->addOption( 'compress', 'Compress the sitemap files, can take value yes|no, default yes', false, true );
 147                 $this->addOption( 'skip-redirects', 'Do not include redirecting articles in the sitemap' );
 148                 $this->addOption( 'identifier', 'What site identifier to use for the wiki, defaults to $wgDBname', false, true );
 149         }
 150
 151         /**
 152          * Execute
 153          */
 154         public function execute() {
 155                 $this->setNamespacePriorities();
 156                 $this->url_limit = 50000;
 157                 $this->size_limit = pow( 2, 20 ) * 10;
 158                 $this->fspath = self::init_path( $this->getOption( 'fspath', getcwd() ) );
 159                 $this->urlpath = $this->getOption( 'urlpath', "" );
 160                 if ( $this->urlpath !== "" && substr( $this->urlpath, -1 ) !== '/' ) {
 161                         $this->urlpath .= '/';
 162                 }
 163                 $this->identifier = $this->getOption( 'identifier', wfWikiID() );
 164                 $this->compress = $this->getOption( 'compress', 'yes' ) !== 'no';
 165                 $this->skipRedirects = $this->getOption( 'skip-redirects', false ) !== false ;
 166                 $this->dbr = wfGetDB( DB_SLAVE );
 167                 $this->generateNamespaces();
 168                 $this->timestamp = wfTimestamp( TS_ISO_8601, wfTimestampNow() );
 169                 $this->findex = fopen( "{$this->fspath}sitemap-index-{$this->identifier}.xml", 'wb' );
 170                 $this->main();
 171         }
 172
 173         private function setNamespacePriorities() {
 174                 global $wgSitemapNamespacesPriorities;
 175
 176                 // Custom main namespaces
 177                 $this->priorities[self::GS_MAIN] = '0.5';
 178                 // Custom talk namesspaces
 179                 $this->priorities[self::GS_TALK] = '0.1';
 180                 // MediaWiki standard namespaces
 181                 $this->priorities[NS_MAIN] = '1.0';
 182                 $this->priorities[NS_TALK] = '0.1';
 183                 $this->priorities[NS_USER] = '0.5';
 184                 $this->priorities[NS_USER_TALK] = '0.1';
 185                 $this->priorities[NS_PROJECT] = '0.5';
 186                 $this->priorities[NS_PROJECT_TALK] = '0.1';
 187                 $this->priorities[NS_FILE] = '0.5';
 188                 $this->priorities[NS_FILE_TALK] = '0.1';
 189                 $this->priorities[NS_MEDIAWIKI] = '0.0';
 190                 $this->priorities[NS_MEDIAWIKI_TALK] = '0.1';
 191                 $this->priorities[NS_TEMPLATE] = '0.0';
 192                 $this->priorities[NS_TEMPLATE_TALK] = '0.1';
 193                 $this->priorities[NS_HELP] = '0.5';
 194                 $this->priorities[NS_HELP_TALK] = '0.1';
 195                 $this->priorities[NS_CATEGORY] = '0.5';
 196                 $this->priorities[NS_CATEGORY_TALK] = '0.1';
 197
 198                 // Custom priorities
 199                 if ( $wgSitemapNamespacesPriorities !== false ) {
 200                         /**
 201                          * @var $wgSitemapNamespacesPriorities array
 202                          */
 203                         foreach ( $wgSitemapNamespacesPriorities as $namespace => $priority ) {
 204                                 $float = floatval( $priority );
 205                                 if ( $float > 1.0 ) {
 206                                         $priority = '1.0';
 207                                 } elseif ( $float < 0.0 ) {
 208                                         $priority = '0.0';
 209                                 }
 210                                 $this->priorities[$namespace] = $priority;
 211                         }
 212                 }
 213         }
 214
 215         /**
 216          * Create directory if it does not exist and return pathname with a trailing slash
 217          * @param $fspath string
 218          * @return null|string
 219          */
 220         private static function init_path( $fspath ) {
 221                 if ( !isset( $fspath ) ) {
 222                         return null;
 223                 }
 224                 # Create directory if needed
 225                 if ( $fspath && !is_dir( $fspath ) ) {
 226                         wfMkdirParents( $fspath, null, __METHOD__ ) or die( "Can not create directory $fspath.\n" );
 227                 }
 228
 229                 return realpath( $fspath ) . DIRECTORY_SEPARATOR ;
 230         }
 231
 232         /**
 233          * Generate a one-dimensional array of existing namespaces
 234          */
 235         function generateNamespaces() {
 236                 // Only generate for specific namespaces if $wgSitemapNamespaces is an array.
 237                 global $wgSitemapNamespaces;
 238                 if ( is_array( $wgSitemapNamespaces ) ) {
 239                         $this->namespaces = $wgSitemapNamespaces;
 240                         return;
 241                 }
 242
 243                 $res = $this->dbr->select( 'page',
 244                         array( 'page_namespace' ),
 245                         array(),
 246                         __METHOD__,
 247                         array(
 248                                 'GROUP BY' => 'page_namespace',
 249                                 'ORDER BY' => 'page_namespace',
 250                         )
 251                 );
 252
 253                 foreach ( $res as $row )
 254                         $this->namespaces[] = $row->page_namespace;
 255         }
 256
 257         /**
 258          * Get the priority of a given namespace
 259          *
 260          * @param $namespace Integer: the namespace to get the priority for
 261          * @return String
 262          */
 263         function priority( $namespace ) {
 264                 return isset( $this->priorities[$namespace] ) ? $this->priorities[$namespace] : $this->guessPriority( $namespace );
 265         }
 266
 267         /**
 268          * If the namespace isn't listed on the priority list return the
 269          * default priority for the namespace, varies depending on whether it's
 270          * a talkpage or not.
 271          *
 272          * @param $namespace Integer: the namespace to get the priority for
 273          * @return String
 274          */
 275         function guessPriority( $namespace ) {
 276                 return MWNamespace::isSubject( $namespace ) ? $this->priorities[self::GS_MAIN] : $this->priorities[self::GS_TALK];
 277         }
 278
 279         /**
 280          * Return a database resolution of all the pages in a given namespace
 281          *
 282          * @param $namespace Integer: limit the query to this namespace
 283          * @return Resource
 284          */
 285         function getPageRes( $namespace ) {
 286                 return $this->dbr->select( 'page',
 287                         array(
 288                                 'page_namespace',
 289                                 'page_title',
 290                                 'page_touched',
 291                                 'page_is_redirect'
 292                         ),
 293                         array( 'page_namespace' => $namespace ),
 294                         __METHOD__
 295                 );
 296         }
 297
 298         /**
 299          * Main loop
 300          */
 301         public function main() {
 302                 global $wgContLang;
 303
 304                 fwrite( $this->findex, $this->openIndex() );
 305
 306                 foreach ( $this->namespaces as $namespace ) {
 307                         $res = $this->getPageRes( $namespace );
 308                         $this->file = false;
 309                         $this->generateLimit( $namespace );
 310                         $length = $this->limit[0];
 311                         $i = $smcount = 0;
 312
 313                         $fns = $wgContLang->getFormattedNsText( $namespace );
 314                         $this->output( "$namespace ($fns)\n" );
 315                         $skippedRedirects = 0;  // Number of redirects skipped for that namespace
 316                         foreach ( $res as $row ) {
 317                                 if ($this->skipRedirects && $row->page_is_redirect ) {
 318                                         $skippedRedirects++;
 319                                         continue;
 320                                 }
 321
 322                                 if ( $i++ === 0 || $i === $this->url_limit + 1 || $length + $this->limit[1] + $this->limit[2] > $this->size_limit ) {
 323                                         if ( $this->file !== false ) {
 324                                                 $this->write( $this->file, $this->closeFile() );
 325                                                 $this->close( $this->file );
 326                                         }
 327                                         $filename = $this->sitemapFilename( $namespace, $smcount++ );
 328                                         $this->file = $this->open( $this->fspath . $filename, 'wb' );
 329                                         $this->write( $this->file, $this->openFile() );
 330                                         fwrite( $this->findex, $this->indexEntry( $filename ) );
 331                                         $this->output( "\t$this->fspath$filename\n" );
 332                                         $length = $this->limit[0];
 333                                         $i = 1;
 334                                 }
 335                                 $title = Title::makeTitle( $row->page_namespace, $row->page_title );
 336                                 $date = wfTimestamp( TS_ISO_8601, $row->page_touched );
 337                                 $entry = $this->fileEntry( $title->getCanonicalURL(), $date, $this->priority( $namespace ) );
 338                                 $length += strlen( $entry );
 339                                 $this->write( $this->file, $entry );
 340                                 // generate pages for language variants
 341                                 if ( $wgContLang->hasVariants() ) {
 342                                         $variants = $wgContLang->getVariants();
 343                                         foreach ( $variants as $vCode ) {
 344                                                 if ( $vCode == $wgContLang->getCode() ) continue; // we don't want default variant
 345                                                 $entry = $this->fileEntry( $title->getCanonicalURL( '', $vCode ), $date, $this->priority( $namespace ) );
 346                                                 $length += strlen( $entry );
 347                                                 $this->write( $this->file, $entry );
 348                                         }
 349                                 }
 350                         }
 351
 352                         if ($this->skipRedirects && $skippedRedirects > 0) {
 353                                 $this->output( "  skipped $skippedRedirects redirect(s)\n" );
 354                         }
 355
 356                         if ( $this->file ) {
 357                                 $this->write( $this->file, $this->closeFile() );
 358                                 $this->close( $this->file );
 359                         }
 360                 }
 361                 fwrite( $this->findex, $this->closeIndex() );
 362                 fclose( $this->findex );
 363         }
 364
 365         /**
 366          * gzopen() / fopen() wrapper
 367          *
 368          * @return Resource
 369          */
 370         function open( $file, $flags ) {
 371                 $resource = $this->compress ? gzopen( $file, $flags ) : fopen( $file, $flags );
 372                 if( $resource === false ) {
 373                         wfDebugDieBacktrace( __METHOD__ . " error opening file $file with flags $flags. Check permissions?" );
 374                 }
 375                 return $resource;
 376         }
 377
 378         /**
 379          * gzwrite() / fwrite() wrapper
 380          */
 381         function write( &$handle, $str ) {
 382                 if( $handle === true || $handle === false ) {
 383                         wfDebugDieBacktrace( __METHOD__ . " was passed a boolean as a file handle.\n" );
 384                 }
 385                 if ( $this->compress )
 386                         gzwrite( $handle, $str );
 387                 else
 388                         fwrite( $handle, $str );
 389         }
 390
 391         /**
 392          * gzclose() / fclose() wrapper
 393          */
 394         function close( &$handle ) {
 395                 if ( $this->compress )
 396                         gzclose( $handle );
 397                 else
 398                         fclose( $handle );
 399         }
 400
 401         /**
 402          * Get a sitemap filename
 403          *
 404          * @param $namespace Integer: the namespace
 405          * @param $count Integer: the count
 406          * @return String
 407          */
 408         function sitemapFilename( $namespace, $count ) {
 409                 $ext = $this->compress ? '.gz' : '';
 410                 return "sitemap-{$this->identifier}-NS_$namespace-$count.xml$ext";
 411         }
 412
 413         /**
 414          * Return the XML required to open an XML file
 415          *
 416          * @return string
 417          */
 418         function xmlHead() {
 419                 return '<?xml version="1.0" encoding="UTF-8"?>' . "\n";
 420         }
 421
 422         /**
 423          * Return the XML schema being used
 424          *
 425          * @return String
 426          */
 427         function xmlSchema() {
 428                 return 'http://www.sitemaps.org/schemas/sitemap/0.9';
 429         }
 430
 431         /**
 432          * Return the XML required to open a sitemap index file
 433          *
 434          * @return String
 435          */
 436         function openIndex() {
 437                 return $this->xmlHead() . '<sitemapindex xmlns="' . $this->xmlSchema() . '">' . "\n";
 438         }
 439
 440         /**
 441          * Return the XML for a single sitemap indexfile entry
 442          *
 443          * @param $filename String: the filename of the sitemap file
 444          * @return String
 445          */
 446         function indexEntry( $filename ) {
 447                 return
 448                         "\t<sitemap>\n" .
 449                         "\t\t<loc>{$this->urlpath}$filename</loc>\n" .
 450                         "\t\t<lastmod>{$this->timestamp}</lastmod>\n" .
 451                         "\t</sitemap>\n";
 452         }
 453
 454         /**
 455          * Return the XML required to close a sitemap index file
 456          *
 457          * @return String
 458          */
 459         function closeIndex() {
 460                 return "</sitemapindex>\n";
 461         }
 462
 463         /**
 464          * Return the XML required to open a sitemap file
 465          *
 466          * @return String
 467          */
 468         function openFile() {
 469                 return $this->xmlHead() . '<urlset xmlns="' . $this->xmlSchema() . '">' . "\n";
 470         }
 471
 472         /**
 473          * Return the XML for a single sitemap entry
 474          *
 475          * @param $url String: an RFC 2396 compliant URL
 476          * @param $date String: a ISO 8601 date
 477          * @param $priority String: a priority indicator, 0.0 - 1.0 inclusive with a 0.1 stepsize
 478          * @return String
 479          */
 480         function fileEntry( $url, $date, $priority ) {
 481                 return
 482                         "\t<url>\n" .
 483                         "\t\t<loc>$url</loc>\n" .
 484                         "\t\t<lastmod>$date</lastmod>\n" .
 485                         "\t\t<priority>$priority</priority>\n" .
 486                         "\t</url>\n";
 487         }
 488
 489         /**
 490          * Return the XML required to close sitemap file
 491          *
 492          * @return String
 493          */
 494         function closeFile() {
 495                 return "</urlset>\n";
 496         }
 497
 498         /**
 499          * Populate $this->limit
 500          */
 501         function generateLimit( $namespace ) {
 502                 // bug 17961: make a title with the longest possible URL in this namespace
 503                 $title = Title::makeTitle( $namespace, str_repeat( "\xf0\xa8\xae\x81", 63 ) . "\xe5\x96\x83" );
 504
 505                 $this->limit = array(
 506                         strlen( $this->openFile() ),
 507                         strlen( $this->fileEntry( $title->getCanonicalURL(), wfTimestamp( TS_ISO_8601, wfTimestamp() ), $this->priority( $namespace ) ) ),
 508                         strlen( $this->closeFile() )
 509                 );
 510         }
 511 }
 512
 513 $maintClass = "GenerateSitemap";
 514 require_once( RUN_MAINTENANCE_IF_MAIN );