includes/specials/SpecialExport.php

   1 <?php
   2 /**
   3  * Implements Special:Export
   4  *
   5  * Copyright © 2003-2008 Brion Vibber <brion@pobox.com>
   6  *
   7  * This program is free software; you can redistribute it and/or modify
   8  * it under the terms of the GNU General Public License as published by
   9  * the Free Software Foundation; either version 2 of the License, or
  10  * (at your option) any later version.
  11  *
  12  * This program is distributed in the hope that it will be useful,
  13  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  14  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
  15  * GNU General Public License for more details.
  16  *
  17  * You should have received a copy of the GNU General Public License along
  18  * with this program; if not, write to the Free Software Foundation, Inc.,
  19  * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
  20  * http://www.gnu.org/copyleft/gpl.html
  21  *
  22  * @file
  23  * @ingroup SpecialPage
  24  */
  25
  26 /**
  27  * A special page that allows users to export pages in a XML file
  28  *
  29  * @ingroup SpecialPage
  30  */
  31 class SpecialExport extends SpecialPage {
  32
  33         private $curonly, $doExport, $pageLinkDepth, $templates;
  34         private $images;
  35
  36         public function __construct() {
  37                 parent::__construct( 'Export' );
  38         }
  39
  40         public function execute( $par ) {
  41                 global $wgOut, $wgRequest, $wgSitename, $wgExportAllowListContributors;
  42                 global $wgExportAllowHistory, $wgExportMaxHistory, $wgExportMaxLinkDepth;
  43                 global $wgExportFromNamespaces, $wgUser;
  44
  45                 $this->setHeaders();
  46                 $this->outputHeader();
  47
  48                 // Set some variables
  49                 $this->curonly = true;
  50                 $this->doExport = false;
  51                 $this->templates = $wgRequest->getCheck( 'templates' );
  52                 $this->images = $wgRequest->getCheck( 'images' ); // Doesn't do anything yet
  53                 $this->pageLinkDepth = $this->validateLinkDepth(
  54                         $wgRequest->getIntOrNull( 'pagelink-depth' )
  55                 );
  56                 $nsindex = '';
  57
  58                 if ( $wgRequest->getCheck( 'addcat' ) ) {
  59                         $page = $wgRequest->getText( 'pages' );
  60                         $catname = $wgRequest->getText( 'catname' );
  61
  62                         if ( $catname !== '' && $catname !== null && $catname !== false ) {
  63                                 $t = Title::makeTitleSafe( NS_MAIN, $catname );
  64                                 if ( $t ) {
  65                                         /**
  66                                          * @todo Fixme: this can lead to hitting memory limit for very large
  67                                          * categories. Ideally we would do the lookup synchronously
  68                                          * during the export in a single query.
  69                                          */
  70                                         $catpages = $this->getPagesFromCategory( $t );
  71                                         if ( $catpages ) $page .= "\n" . implode( "\n", $catpages );
  72                                 }
  73                         }
  74                 }
  75                 else if( $wgRequest->getCheck( 'addns' ) && $wgExportFromNamespaces ) {
  76                         $page = $wgRequest->getText( 'pages' );
  77                         $nsindex = $wgRequest->getText( 'nsindex', '' );
  78
  79                         if ( strval( $nsindex ) !== ''  ) {
  80                                 /**
  81                                  * Same implementation as above, so same @todo
  82                                  */
  83                                 $nspages = $this->getPagesFromNamespace( $nsindex );
  84                                 if ( $nspages ) $page .= "\n" . implode( "\n", $nspages );
  85                         }
  86                 }
  87                 else if( $wgRequest->wasPosted() && $par == '' ) {
  88                         $page = $wgRequest->getText( 'pages' );
  89                         $this->curonly = $wgRequest->getCheck( 'curonly' );
  90                         $rawOffset = $wgRequest->getVal( 'offset' );
  91
  92                         if( $rawOffset ) {
  93                                 $offset = wfTimestamp( TS_MW, $rawOffset );
  94                         } else {
  95                                 $offset = null;
  96                         }
  97
  98                         $limit = $wgRequest->getInt( 'limit' );
  99                         $dir = $wgRequest->getVal( 'dir' );
 100                         $history = array(
 101                                 'dir' => 'asc',
 102                                 'offset' => false,
 103                                 'limit' => $wgExportMaxHistory,
 104                         );
 105                         $historyCheck = $wgRequest->getCheck( 'history' );
 106
 107                         if ( $this->curonly ) {
 108                                 $history = WikiExporter::CURRENT;
 109                         } elseif ( !$historyCheck ) {
 110                                 if ( $limit > 0 && ($wgExportMaxHistory == 0 || $limit < $wgExportMaxHistory ) ) {
 111                                         $history['limit'] = $limit;
 112                                 }
 113                                 if ( !is_null( $offset ) ) {
 114                                         $history['offset'] = $offset;
 115                                 }
 116                                 if ( strtolower( $dir ) == 'desc' ) {
 117                                         $history['dir'] = 'desc';
 118                                 }
 119                         }
 120
 121                         if( $page != '' ) $this->doExport = true;
 122                 } else {
 123                         // Default to current-only for GET requests.
 124                         $page = $wgRequest->getText( 'pages', $par );
 125                         $historyCheck = $wgRequest->getCheck( 'history' );
 126
 127                         if( $historyCheck ) {
 128                                 $history = WikiExporter::FULL;
 129                         } else {
 130                                 $history = WikiExporter::CURRENT;
 131                         }
 132
 133                         if( $page != '' ) $this->doExport = true;
 134                 }
 135
 136                 if( !$wgExportAllowHistory ) {
 137                         // Override
 138                         $history = WikiExporter::CURRENT;
 139                 }
 140
 141                 $list_authors = $wgRequest->getCheck( 'listauthors' );
 142                 if ( !$this->curonly || !$wgExportAllowListContributors ) $list_authors = false ;
 143
 144                 if ( $this->doExport ) {
 145                         $wgOut->disable();
 146
 147                         // Cancel output buffering and gzipping if set
 148                         // This should provide safer streaming for pages with history
 149                         wfResetOutputBuffers();
 150                         $wgRequest->response()->header( "Content-type: application/xml; charset=utf-8" );
 151
 152                         if( $wgRequest->getCheck( 'wpDownload' ) ) {
 153                                 // Provide a sane filename suggestion
 154                                 $filename = urlencode( $wgSitename . '-' . wfTimestampNow() . '.xml' );
 155                                 $wgRequest->response()->header( "Content-disposition: attachment;filename={$filename}" );
 156                         }
 157
 158                         $this->doExport( $page, $history, $list_authors );
 159
 160                         return;
 161                 }
 162
 163                 $wgOut->addWikiMsg( 'exporttext' );
 164
 165                 $form = Xml::openElement( 'form', array( 'method' => 'post',
 166                         'action' => $this->getTitle()->getLocalUrl( 'action=submit' ) ) );
 167                 $form .= Xml::inputLabel( wfMsg( 'export-addcattext' )    , 'catname', 'catname', 40 ) . '&#160;';
 168                 $form .= Xml::submitButton( wfMsg( 'export-addcat' ), array( 'name' => 'addcat' ) ) . '<br />';
 169
 170                 if ( $wgExportFromNamespaces ) {
 171                         $form .= Xml::namespaceSelector( $nsindex, null, 'nsindex', wfMsg( 'export-addnstext' ) ) . '&#160;';
 172                         $form .= Xml::submitButton( wfMsg( 'export-addns' ), array( 'name' => 'addns' ) ) . '<br />';
 173                 }
 174
 175                 $form .= Xml::element( 'textarea', array( 'name' => 'pages', 'cols' => 40, 'rows' => 10 ), $page, false );
 176                 $form .= '<br />';
 177
 178                 if( $wgExportAllowHistory ) {
 179                         $form .= Xml::checkLabel( wfMsg( 'exportcuronly' ), 'curonly', 'curonly', true ) . '<br />';
 180                 } else {
 181                         $wgOut->addHTML( wfMsgExt( 'exportnohistory', 'parse' ) );
 182                 }
 183
 184                 $form .= Xml::checkLabel( wfMsg( 'export-templates' ), 'templates', 'wpExportTemplates', false ) . '<br />';
 185
 186                 if( $wgExportMaxLinkDepth || $this->userCanOverrideExportDepth() ) {
 187                         $form .= Xml::inputLabel( wfMsg( 'export-pagelinks' ), 'pagelink-depth', 'pagelink-depth', 20, 0 ) . '<br />';
 188                 }
 189                 // Enable this when we can do something useful exporting/importing image information. :)
 190                 //$form .= Xml::checkLabel( wfMsg( 'export-images' ), 'images', 'wpExportImages', false ) . '<br />';
 191                 $form .= Xml::checkLabel( wfMsg( 'export-download' ), 'wpDownload', 'wpDownload', true ) . '<br />';
 192
 193                 $form .= Xml::submitButton( wfMsg( 'export-submit' ), $wgUser->getSkin()->tooltipAndAccessKeyAttribs( 'export' ) );
 194                 $form .= Xml::closeElement( 'form' );
 195
 196                 $wgOut->addHTML( $form );
 197         }
 198
 199         private function userCanOverrideExportDepth() {
 200                 global $wgUser;
 201                 return $wgUser->isAllowed( 'override-export-depth' );
 202         }
 203
 204         /**
 205          * Do the actual page exporting
 206          *
 207          * @param $page String: user input on what page(s) to export
 208          * @param $history Mixed: one of the WikiExporter history export constants
 209          * @param $list_authors Boolean: Whether to add distinct author list (when
 210          *                      not returning full history)
 211          */
 212         private function doExport( $page, $history, $list_authors ) {
 213                 $pageSet = array(); // Inverted index of all pages to look up
 214
 215                 // Split up and normalize input
 216                 foreach( explode( "\n", $page ) as $pageName ) {
 217                         $pageName = trim( $pageName );
 218                         $title = Title::newFromText( $pageName );
 219                         if( $title && $title->getInterwiki() == '' && $title->getText() !== '' ) {
 220                                 // Only record each page once!
 221                                 $pageSet[$title->getPrefixedText()] = true;
 222                         }
 223                 }
 224
 225                 // Set of original pages to pass on to further manipulation...
 226                 $inputPages = array_keys( $pageSet );
 227
 228                 // Look up any linked pages if asked...
 229                 if( $this->templates ) {
 230                         $pageSet = $this->getTemplates( $inputPages, $pageSet );
 231                 }
 232                 $linkDepth = $this->pageLinkDepth;
 233                 if( $linkDepth ) {
 234                         $pageSet = $this->getPageLinks( $inputPages, $pageSet, $linkDepth );
 235                 }
 236
 237                 /*
 238                  // Enable this when we can do something useful exporting/importing image information. :)
 239                  if( $this->images ) ) {
 240                  $pageSet = $this->getImages( $inputPages, $pageSet );
 241                  }
 242                  */
 243
 244                 $pages = array_keys( $pageSet );
 245
 246                 // Normalize titles to the same format and remove dupes, see bug 17374
 247                 foreach( $pages as $k => $v ) {
 248                         $pages[$k] = str_replace( " ", "_", $v );
 249                 }
 250
 251                 $pages = array_unique( $pages );
 252
 253                 /* Ok, let's get to it... */
 254                 if( $history == WikiExporter::CURRENT ) {
 255                         $lb = false;
 256                         $db = wfGetDB( DB_SLAVE );
 257                         $buffer = WikiExporter::BUFFER;
 258                 } else {
 259                         // Use an unbuffered query; histories may be very long!
 260                         $lb = wfGetLBFactory()->newMainLB();
 261                         $db = $lb->getConnection( DB_SLAVE );
 262                         $buffer = WikiExporter::STREAM;
 263
 264                         // This might take a while... :D
 265                         wfSuppressWarnings();
 266                         set_time_limit(0);
 267                         wfRestoreWarnings();
 268                 }
 269
 270                 $exporter = new WikiExporter( $db, $history, $buffer );
 271                 $exporter->list_authors = $list_authors;
 272                 $exporter->openStream();
 273
 274                 foreach( $pages as $page ) {
 275                         /*
 276                          if( $wgExportMaxHistory && !$this->curonly ) {
 277                          $title = Title::newFromText( $page );
 278                          if( $title ) {
 279                          $count = Revision::countByTitle( $db, $title );
 280                          if( $count > $wgExportMaxHistory ) {
 281                          wfDebug( __FUNCTION__ .
 282                          ": Skipped $page, $count revisions too big\n" );
 283                          continue;
 284                          }
 285                          }
 286                          }*/
 287                         #Bug 8824: Only export pages the user can read
 288                         $title = Title::newFromText( $page );
 289                         if( is_null( $title ) ) continue; #TODO: perhaps output an <error> tag or something.
 290                         if( !$title->userCanRead() ) continue; #TODO: perhaps output an <error> tag or something.
 291
 292                         $exporter->pageByTitle( $title );
 293                 }
 294
 295                 $exporter->closeStream();
 296
 297                 if( $lb ) {
 298                         $lb->closeAll();
 299                 }
 300         }
 301
 302         private function getPagesFromCategory( $title ) {
 303                 global $wgContLang;
 304
 305                 $name = $title->getDBkey();
 306
 307                 $dbr = wfGetDB( DB_SLAVE );
 308                 $res = $dbr->select(
 309                         array( 'page', 'categorylinks' ),
 310                         array( 'page_namespace', 'page_title' ),
 311                         array( 'cl_from=page_id', 'cl_to' => $name ),
 312                         __METHOD__,
 313                         array( 'LIMIT' => '5000' )
 314                 );
 315
 316                 $pages = array();
 317
 318                 foreach ( $res as $row ) {
 319                         $n = $row->page_title;
 320                         if ($row->page_namespace) {
 321                                 $ns = $wgContLang->getNsText( $row->page_namespace );
 322                                 $n = $ns . ':' . $n;
 323                         }
 324
 325                         $pages[] = $n;
 326                 }
 327                 return $pages;
 328         }
 329
 330         private function getPagesFromNamespace( $nsindex ) {
 331                 global $wgContLang;
 332
 333                 $dbr = wfGetDB( DB_SLAVE );
 334                 $res = $dbr->select(
 335                         'page',
 336                         array( 'page_namespace', 'page_title' ),
 337                         array( 'page_namespace' => $nsindex ),
 338                         __METHOD__,
 339                         array( 'LIMIT' => '5000' )
 340                 );
 341
 342                 $pages = array();
 343
 344                 foreach ( $res as $row ) {
 345                         $n = $row->page_title;
 346
 347                         if ( $row->page_namespace ) {
 348                                 $ns = $wgContLang->getNsText( $row->page_namespace );
 349                                 $n = $ns . ':' . $n;
 350                         }
 351
 352                         $pages[] = $n;
 353                 }
 354                 return $pages;
 355         }
 356
 357         /**
 358          * Expand a list of pages to include templates used in those pages.
 359          * @param $inputPages array, list of titles to look up
 360          * @param $pageSet array, associative array indexed by titles for output
 361          * @return array associative array index by titles
 362          */
 363         private function getTemplates( $inputPages, $pageSet ) {
 364                 return $this->getLinks( $inputPages, $pageSet,
 365                         'templatelinks',
 366                         array( 'tl_namespace AS namespace', 'tl_title AS title' ),
 367                         array( 'page_id=tl_from' )
 368                 );
 369         }
 370
 371         /**
 372          * Validate link depth setting, if available.
 373          */
 374         private function validateLinkDepth( $depth ) {
 375                 global $wgExportMaxLinkDepth;
 376
 377                 if( $depth < 0 ) {
 378                         return 0;
 379                 }
 380
 381                 if ( !$this->userCanOverrideExportDepth() ) {
 382                         if( $depth > $wgExportMaxLinkDepth ) {
 383                                 return $wgExportMaxLinkDepth;
 384                         }
 385                 }
 386
 387                 /*
 388                  * There's a HARD CODED limit of 5 levels of recursion here to prevent a
 389                  * crazy-big export from being done by someone setting the depth
 390                  * number too high. In other words, last resort safety net.
 391                  */
 392                 return intval( min( $depth, 5 ) );
 393         }
 394
 395         /** Expand a list of pages to include pages linked to from that page. */
 396         private function getPageLinks( $inputPages, $pageSet, $depth ) {
 397                 for(; $depth > 0; --$depth ) {
 398                         $pageSet = $this->getLinks(
 399                                 $inputPages, $pageSet, 'pagelinks',
 400                                 array( 'pl_namespace AS namespace', 'pl_title AS title' ),
 401                                 array( 'page_id=pl_from' )
 402                         );
 403                         $inputPages = array_keys( $pageSet );
 404                 }
 405
 406                 return $pageSet;
 407         }
 408
 409         /**
 410          * Expand a list of pages to include images used in those pages.
 411          *
 412          * @param $inputPages array, list of titles to look up
 413          * @param $pageSet array, associative array indexed by titles for output
 414          *
 415          * @return array associative array index by titles
 416          */
 417         private function getImages( $inputPages, $pageSet ) {
 418                 return $this->getLinks(
 419                         $inputPages,
 420                         $pageSet,
 421                         'imagelinks',
 422                         array( NS_FILE . ' AS namespace', 'il_to AS title' ),
 423                         array( 'page_id=il_from' )
 424                 );
 425         }
 426
 427         /**
 428          * Expand a list of pages to include items used in those pages.
 429          */
 430         private function getLinks( $inputPages, $pageSet, $table, $fields, $join ) {
 431                 $dbr = wfGetDB( DB_SLAVE );
 432
 433                 foreach( $inputPages as $page ) {
 434                         $title = Title::newFromText( $page );
 435
 436                         if( $title ) {
 437                                 $pageSet[$title->getPrefixedText()] = true;
 438                                 /// @todo Fixme: May or may not be more efficient to batch these
 439                                 ///        by namespace when given multiple input pages.
 440                                 $result = $dbr->select(
 441                                         array( 'page', $table ),
 442                                         $fields,
 443                                         array_merge(
 444                                                 $join,
 445                                                 array(
 446                                                         'page_namespace' => $title->getNamespace(),
 447                                                         'page_title' => $title->getDBkey()
 448                                                 )
 449                                         ),
 450                                         __METHOD__
 451                                 );
 452
 453                                 foreach( $result as $row ) {
 454                                         $template = Title::makeTitle( $row->namespace, $row->title );
 455                                         $pageSet[$template->getPrefixedText()] = true;
 456                                 }
 457                         }
 458                 }
 459
 460                 return $pageSet;
 461         }
 462
 463 }