Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
| Total | |
0.00% |
0 / 318 |
|
0.00% |
0 / 12 |
CRAP | |
0.00% |
0 / 1 |
| SpecialExport | |
0.00% |
0 / 317 |
|
0.00% |
0 / 12 |
6320 | |
0.00% |
0 / 1 |
| __construct | |
0.00% |
0 / 5 |
|
0.00% |
0 / 1 |
2 | |||
| execute | |
0.00% |
0 / 200 |
|
0.00% |
0 / 1 |
2070 | |||
| userCanOverrideExportDepth | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| doExport | |
0.00% |
0 / 33 |
|
0.00% |
0 / 1 |
182 | |||
| getPagesFromCategory | |
0.00% |
0 / 14 |
|
0.00% |
0 / 1 |
6 | |||
| getPagesFromNamespace | |
0.00% |
0 / 12 |
|
0.00% |
0 / 1 |
6 | |||
| getTemplates | |
0.00% |
0 / 11 |
|
0.00% |
0 / 1 |
2 | |||
| getExtraPages | |
0.00% |
0 / 5 |
|
0.00% |
0 / 1 |
6 | |||
| validateLinkDepth | |
0.00% |
0 / 7 |
|
0.00% |
0 / 1 |
30 | |||
| getPageLinks | |
0.00% |
0 / 14 |
|
0.00% |
0 / 1 |
6 | |||
| getLinks | |
0.00% |
0 / 14 |
|
0.00% |
0 / 1 |
20 | |||
| getGroupName | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| 1 | <?php |
| 2 | /** |
| 3 | * Copyright © 2003-2008 Brooke Vibber <bvibber@wikimedia.org> |
| 4 | * |
| 5 | * This program is free software; you can redistribute it and/or modify |
| 6 | * it under the terms of the GNU General Public License as published by |
| 7 | * the Free Software Foundation; either version 2 of the License, or |
| 8 | * (at your option) any later version. |
| 9 | * |
| 10 | * This program is distributed in the hope that it will be useful, |
| 11 | * but WITHOUT ANY WARRANTY; without even the implied warranty of |
| 12 | * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the |
| 13 | * GNU General Public License for more details. |
| 14 | * |
| 15 | * You should have received a copy of the GNU General Public License along |
| 16 | * with this program; if not, write to the Free Software Foundation, Inc., |
| 17 | * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA. |
| 18 | * http://www.gnu.org/copyleft/gpl.html |
| 19 | * |
| 20 | * @file |
| 21 | */ |
| 22 | |
| 23 | namespace MediaWiki\Specials; |
| 24 | |
| 25 | use MediaWiki\Export\WikiExporterFactory; |
| 26 | use MediaWiki\HTMLForm\Field\HTMLTextAreaField; |
| 27 | use MediaWiki\HTMLForm\HTMLForm; |
| 28 | use MediaWiki\Linker\LinksMigration; |
| 29 | use MediaWiki\Logger\LoggerFactory; |
| 30 | use MediaWiki\MainConfigNames; |
| 31 | use MediaWiki\Page\PageIdentity; |
| 32 | use MediaWiki\SpecialPage\SpecialPage; |
| 33 | use MediaWiki\Title\Title; |
| 34 | use MediaWiki\Title\TitleFormatter; |
| 35 | use WikiExporter; |
| 36 | use Wikimedia\Rdbms\IConnectionProvider; |
| 37 | use Wikimedia\Rdbms\SelectQueryBuilder; |
| 38 | |
| 39 | /** |
| 40 | * A special page that allows users to export pages in a XML file |
| 41 | * |
| 42 | * @ingroup SpecialPage |
| 43 | * @ingroup Dump |
| 44 | */ |
| 45 | class SpecialExport extends SpecialPage { |
| 46 | protected bool $curonly; |
| 47 | protected bool $doExport; |
| 48 | protected int $pageLinkDepth; |
| 49 | protected bool $templates; |
| 50 | |
| 51 | private IConnectionProvider $dbProvider; |
| 52 | private WikiExporterFactory $wikiExporterFactory; |
| 53 | private TitleFormatter $titleFormatter; |
| 54 | private LinksMigration $linksMigration; |
| 55 | |
| 56 | /** |
| 57 | * @param IConnectionProvider $dbProvider |
| 58 | * @param WikiExporterFactory $wikiExporterFactory |
| 59 | * @param TitleFormatter $titleFormatter |
| 60 | * @param LinksMigration $linksMigration |
| 61 | */ |
| 62 | public function __construct( |
| 63 | IConnectionProvider $dbProvider, |
| 64 | WikiExporterFactory $wikiExporterFactory, |
| 65 | TitleFormatter $titleFormatter, |
| 66 | LinksMigration $linksMigration |
| 67 | ) { |
| 68 | parent::__construct( 'Export' ); |
| 69 | $this->dbProvider = $dbProvider; |
| 70 | $this->wikiExporterFactory = $wikiExporterFactory; |
| 71 | $this->titleFormatter = $titleFormatter; |
| 72 | $this->linksMigration = $linksMigration; |
| 73 | } |
| 74 | |
| 75 | public function execute( $par ) { |
| 76 | $this->setHeaders(); |
| 77 | $this->outputHeader(); |
| 78 | $config = $this->getConfig(); |
| 79 | |
| 80 | $this->curonly = true; |
| 81 | $this->doExport = false; |
| 82 | $request = $this->getRequest(); |
| 83 | $this->templates = $request->getCheck( 'templates' ); |
| 84 | $this->pageLinkDepth = $this->validateLinkDepth( |
| 85 | $request->getIntOrNull( 'pagelink-depth' ) |
| 86 | ); |
| 87 | $nsindex = ''; |
| 88 | $exportall = false; |
| 89 | |
| 90 | if ( $request->getCheck( 'addcat' ) ) { |
| 91 | $page = $request->getText( 'pages' ); |
| 92 | $catname = $request->getText( 'catname' ); |
| 93 | |
| 94 | if ( $catname !== '' && $catname !== null && $catname !== false ) { |
| 95 | $t = Title::makeTitleSafe( NS_MAIN, $catname ); |
| 96 | if ( $t ) { |
| 97 | /** |
| 98 | * @todo FIXME: This can lead to hitting memory limit for very large |
| 99 | * categories. Ideally we would do the lookup synchronously |
| 100 | * during the export in a single query. |
| 101 | */ |
| 102 | $catpages = $this->getPagesFromCategory( $t ); |
| 103 | if ( $catpages ) { |
| 104 | if ( $page !== '' ) { |
| 105 | $page .= "\n"; |
| 106 | } |
| 107 | $page .= implode( "\n", $catpages ); |
| 108 | } |
| 109 | } |
| 110 | } |
| 111 | } elseif ( $request->getCheck( 'addns' ) && $config->get( MainConfigNames::ExportFromNamespaces ) ) { |
| 112 | $page = $request->getText( 'pages' ); |
| 113 | $nsindex = $request->getText( 'nsindex', '' ); |
| 114 | |
| 115 | if ( strval( $nsindex ) !== '' ) { |
| 116 | /** |
| 117 | * Same implementation as above, so same @todo |
| 118 | */ |
| 119 | $nspages = $this->getPagesFromNamespace( (int)$nsindex ); |
| 120 | if ( $nspages ) { |
| 121 | $page .= "\n" . implode( "\n", $nspages ); |
| 122 | } |
| 123 | } |
| 124 | } elseif ( $request->getCheck( 'exportall' ) && $config->get( MainConfigNames::ExportAllowAll ) ) { |
| 125 | $this->doExport = true; |
| 126 | $exportall = true; |
| 127 | |
| 128 | /* Although $page and $history are not used later on, we |
| 129 | nevertheless set them to avoid that PHP notices about using |
| 130 | undefined variables foul up our XML output (see call to |
| 131 | doExport(...) further down) */ |
| 132 | $page = ''; |
| 133 | $history = ''; |
| 134 | } elseif ( $request->wasPosted() && $par == '' ) { |
| 135 | // Log to see if certain parameters are actually used. |
| 136 | // If not, we could deprecate them and do some cleanup, here and in WikiExporter. |
| 137 | LoggerFactory::getInstance( 'export' )->debug( |
| 138 | 'Special:Export POST, dir: [{dir}], offset: [{offset}], limit: [{limit}]', [ |
| 139 | 'dir' => $request->getRawVal( 'dir' ), |
| 140 | 'offset' => $request->getRawVal( 'offset' ), |
| 141 | 'limit' => $request->getRawVal( 'limit' ), |
| 142 | ] ); |
| 143 | |
| 144 | $page = $request->getText( 'pages' ); |
| 145 | $this->curonly = $request->getCheck( 'curonly' ); |
| 146 | $rawOffset = $request->getVal( 'offset' ); |
| 147 | |
| 148 | if ( $rawOffset ) { |
| 149 | $offset = wfTimestamp( TS_MW, $rawOffset ); |
| 150 | } else { |
| 151 | $offset = null; |
| 152 | } |
| 153 | |
| 154 | $maxHistory = $config->get( MainConfigNames::ExportMaxHistory ); |
| 155 | $limit = $request->getInt( 'limit' ); |
| 156 | $dir = $request->getVal( 'dir' ); |
| 157 | $history = [ |
| 158 | 'dir' => 'asc', |
| 159 | 'offset' => false, |
| 160 | 'limit' => $maxHistory, |
| 161 | ]; |
| 162 | $historyCheck = $request->getCheck( 'history' ); |
| 163 | |
| 164 | if ( $this->curonly ) { |
| 165 | $history = WikiExporter::CURRENT; |
| 166 | } elseif ( !$historyCheck ) { |
| 167 | if ( $limit > 0 && ( $maxHistory == 0 || $limit < $maxHistory ) ) { |
| 168 | $history['limit'] = $limit; |
| 169 | } |
| 170 | |
| 171 | if ( $offset !== null ) { |
| 172 | $history['offset'] = $offset; |
| 173 | } |
| 174 | |
| 175 | if ( strtolower( $dir ?? '' ) == 'desc' ) { |
| 176 | $history['dir'] = 'desc'; |
| 177 | } |
| 178 | } |
| 179 | |
| 180 | if ( $page != '' ) { |
| 181 | $this->doExport = true; |
| 182 | } |
| 183 | } else { |
| 184 | // Default to current-only for GET requests. |
| 185 | $page = $request->getText( 'pages', $par ?? '' ); |
| 186 | $historyCheck = $request->getCheck( 'history' ); |
| 187 | |
| 188 | if ( $historyCheck ) { |
| 189 | $history = WikiExporter::FULL; |
| 190 | } else { |
| 191 | $history = WikiExporter::CURRENT; |
| 192 | } |
| 193 | |
| 194 | if ( $page != '' ) { |
| 195 | $this->doExport = true; |
| 196 | } |
| 197 | } |
| 198 | |
| 199 | if ( !$config->get( MainConfigNames::ExportAllowHistory ) ) { |
| 200 | // Override |
| 201 | $history = WikiExporter::CURRENT; |
| 202 | } |
| 203 | |
| 204 | $list_authors = $request->getCheck( 'listauthors' ); |
| 205 | if ( !$this->curonly || !$config->get( MainConfigNames::ExportAllowListContributors ) ) { |
| 206 | $list_authors = false; |
| 207 | } |
| 208 | |
| 209 | if ( $this->doExport ) { |
| 210 | $this->getOutput()->disable(); |
| 211 | |
| 212 | // Cancel output buffering and gzipping if set |
| 213 | // This should provide safer streaming for pages with history |
| 214 | wfResetOutputBuffers(); |
| 215 | $request->response()->header( 'Content-type: application/xml; charset=utf-8' ); |
| 216 | $request->response()->header( 'X-Robots-Tag: noindex,nofollow' ); |
| 217 | |
| 218 | if ( $request->getCheck( 'wpDownload' ) ) { |
| 219 | // Provide a sensible filename suggestion |
| 220 | $filename = urlencode( $config->get( MainConfigNames::Sitename ) . '-' . |
| 221 | wfTimestampNow() . '.xml' ); |
| 222 | $request->response()->header( "Content-disposition: attachment;filename={$filename}" ); |
| 223 | } |
| 224 | |
| 225 | // @phan-suppress-next-next-line PhanPossiblyUndeclaredVariable |
| 226 | // @phan-suppress-next-line PhanTypeMismatchArgumentNullable history is set when used |
| 227 | $this->doExport( $page, $history, $list_authors, $exportall ); |
| 228 | |
| 229 | return; |
| 230 | } |
| 231 | |
| 232 | $out = $this->getOutput(); |
| 233 | $out->addWikiMsg( 'exporttext' ); |
| 234 | |
| 235 | if ( $page == '' ) { |
| 236 | $categoryName = $request->getText( 'catname' ); |
| 237 | } else { |
| 238 | $categoryName = ''; |
| 239 | } |
| 240 | $canExportAll = $config->get( MainConfigNames::ExportAllowAll ); |
| 241 | $hideIf = $canExportAll ? [ 'hide-if' => [ '===', 'exportall', '1' ] ] : []; |
| 242 | |
| 243 | $formDescriptor = [ |
| 244 | 'catname' => [ |
| 245 | 'type' => 'textwithbutton', |
| 246 | 'name' => 'catname', |
| 247 | 'horizontal-label' => true, |
| 248 | 'label-message' => 'export-addcattext', |
| 249 | 'default' => $categoryName, |
| 250 | 'size' => 40, |
| 251 | 'buttontype' => 'submit', |
| 252 | 'buttonname' => 'addcat', |
| 253 | 'buttondefault' => $this->msg( 'export-addcat' )->text(), |
| 254 | ] + $hideIf, |
| 255 | ]; |
| 256 | if ( $config->get( MainConfigNames::ExportFromNamespaces ) ) { |
| 257 | $formDescriptor += [ |
| 258 | 'nsindex' => [ |
| 259 | 'type' => 'namespaceselectwithbutton', |
| 260 | 'default' => $nsindex, |
| 261 | 'label-message' => 'export-addnstext', |
| 262 | 'horizontal-label' => true, |
| 263 | 'name' => 'nsindex', |
| 264 | 'id' => 'namespace', |
| 265 | 'cssclass' => 'namespaceselector', |
| 266 | 'buttontype' => 'submit', |
| 267 | 'buttonname' => 'addns', |
| 268 | 'buttondefault' => $this->msg( 'export-addns' )->text(), |
| 269 | ] + $hideIf, |
| 270 | ]; |
| 271 | } |
| 272 | |
| 273 | if ( $canExportAll ) { |
| 274 | $formDescriptor += [ |
| 275 | 'exportall' => [ |
| 276 | 'type' => 'check', |
| 277 | 'label-message' => 'exportall', |
| 278 | 'name' => 'exportall', |
| 279 | 'id' => 'exportall', |
| 280 | 'default' => $request->wasPosted() && $request->getCheck( 'exportall' ), |
| 281 | ], |
| 282 | ]; |
| 283 | } |
| 284 | |
| 285 | $formDescriptor += [ |
| 286 | 'textarea' => [ |
| 287 | 'class' => HTMLTextAreaField::class, |
| 288 | 'name' => 'pages', |
| 289 | 'label-message' => 'export-manual', |
| 290 | 'nodata' => true, |
| 291 | 'rows' => 10, |
| 292 | 'default' => $page, |
| 293 | ] + $hideIf, |
| 294 | ]; |
| 295 | |
| 296 | if ( $config->get( MainConfigNames::ExportAllowHistory ) ) { |
| 297 | $formDescriptor += [ |
| 298 | 'curonly' => [ |
| 299 | 'type' => 'check', |
| 300 | 'label-message' => 'exportcuronly', |
| 301 | 'name' => 'curonly', |
| 302 | 'id' => 'curonly', |
| 303 | 'default' => !$request->wasPosted() || $request->getCheck( 'curonly' ), |
| 304 | ], |
| 305 | ]; |
| 306 | } else { |
| 307 | $out->addWikiMsg( 'exportnohistory' ); |
| 308 | } |
| 309 | |
| 310 | $formDescriptor += [ |
| 311 | 'templates' => [ |
| 312 | 'type' => 'check', |
| 313 | 'label-message' => 'export-templates', |
| 314 | 'name' => 'templates', |
| 315 | 'id' => 'wpExportTemplates', |
| 316 | 'default' => $request->wasPosted() && $request->getCheck( 'templates' ), |
| 317 | ], |
| 318 | ]; |
| 319 | |
| 320 | if ( $config->get( MainConfigNames::ExportMaxLinkDepth ) || $this->userCanOverrideExportDepth() ) { |
| 321 | $formDescriptor += [ |
| 322 | 'pagelink-depth' => [ |
| 323 | 'type' => 'text', |
| 324 | 'name' => 'pagelink-depth', |
| 325 | 'id' => 'pagelink-depth', |
| 326 | 'label-message' => 'export-pagelinks', |
| 327 | 'default' => '0', |
| 328 | 'size' => 20, |
| 329 | ], |
| 330 | ]; |
| 331 | } |
| 332 | |
| 333 | $formDescriptor += [ |
| 334 | 'wpDownload' => [ |
| 335 | 'type' => 'check', |
| 336 | 'name' => 'wpDownload', |
| 337 | 'id' => 'wpDownload', |
| 338 | 'default' => !$request->wasPosted() || $request->getCheck( 'wpDownload' ), |
| 339 | 'label-message' => 'export-download', |
| 340 | ], |
| 341 | ]; |
| 342 | |
| 343 | if ( $config->get( MainConfigNames::ExportAllowListContributors ) ) { |
| 344 | $formDescriptor += [ |
| 345 | 'listauthors' => [ |
| 346 | 'type' => 'check', |
| 347 | 'label-message' => 'exportlistauthors', |
| 348 | 'default' => $request->wasPosted() && $request->getCheck( 'listauthors' ), |
| 349 | 'name' => 'listauthors', |
| 350 | 'id' => 'listauthors', |
| 351 | ], |
| 352 | ]; |
| 353 | } |
| 354 | |
| 355 | $htmlForm = HTMLForm::factory( 'ooui', $formDescriptor, $this->getContext() ); |
| 356 | $htmlForm->setSubmitTextMsg( 'export-submit' ); |
| 357 | $htmlForm->prepareForm()->displayForm( false ); |
| 358 | $this->addHelpLink( 'Help:Export' ); |
| 359 | } |
| 360 | |
| 361 | /** |
| 362 | * @return bool |
| 363 | */ |
| 364 | protected function userCanOverrideExportDepth() { |
| 365 | return $this->getAuthority()->isAllowed( 'override-export-depth' ); |
| 366 | } |
| 367 | |
| 368 | /** |
| 369 | * Do the actual page exporting |
| 370 | * |
| 371 | * @param string $page User input on what page(s) to export |
| 372 | * @param int $history One of the WikiExporter history export constants |
| 373 | * @param bool $list_authors Whether to add distinct author list (when |
| 374 | * not returning full history) |
| 375 | * @param bool $exportall Whether to export everything |
| 376 | */ |
| 377 | protected function doExport( $page, $history, $list_authors, $exportall ) { |
| 378 | // If we are grabbing everything, enable full history and ignore the rest |
| 379 | if ( $exportall ) { |
| 380 | $history = WikiExporter::FULL; |
| 381 | } else { |
| 382 | $pageSet = []; // Inverted index of all pages to look up |
| 383 | |
| 384 | // Split up and normalize input |
| 385 | foreach ( explode( "\n", $page ) as $pageName ) { |
| 386 | $pageName = trim( $pageName ); |
| 387 | $title = Title::newFromText( $pageName ); |
| 388 | if ( $title && !$title->isExternal() && $title->getText() !== '' ) { |
| 389 | // Only record each page once! |
| 390 | $pageSet[$title->getPrefixedText()] = true; |
| 391 | } |
| 392 | } |
| 393 | |
| 394 | // Set of original pages to pass on to further manipulation... |
| 395 | $inputPages = array_keys( $pageSet ); |
| 396 | |
| 397 | // Look up any linked pages if asked... |
| 398 | if ( $this->templates ) { |
| 399 | $pageSet = $this->getTemplates( $inputPages, $pageSet ); |
| 400 | } |
| 401 | $pageSet = $this->getExtraPages( $inputPages, $pageSet ); |
| 402 | $linkDepth = $this->pageLinkDepth; |
| 403 | if ( $linkDepth ) { |
| 404 | $pageSet = $this->getPageLinks( $inputPages, $pageSet, $linkDepth ); |
| 405 | } |
| 406 | |
| 407 | $pages = array_keys( $pageSet ); |
| 408 | |
| 409 | // Normalize titles to the same format and remove dupes, see T19374 |
| 410 | foreach ( $pages as $k => $v ) { |
| 411 | $pages[$k] = str_replace( ' ', '_', $v ); |
| 412 | } |
| 413 | |
| 414 | $pages = array_unique( $pages ); |
| 415 | } |
| 416 | |
| 417 | /* Ok, let's get to it... */ |
| 418 | $db = $this->dbProvider->getReplicaDatabase(); |
| 419 | |
| 420 | $exporter = $this->wikiExporterFactory->getWikiExporter( $db, $history ); |
| 421 | $exporter->list_authors = $list_authors; |
| 422 | $exporter->openStream(); |
| 423 | |
| 424 | if ( $exportall ) { |
| 425 | $exporter->allPages(); |
| 426 | } else { |
| 427 | // @phan-suppress-next-line PhanPossiblyUndeclaredVariable |
| 428 | foreach ( $pages as $page ) { |
| 429 | # T10824: Only export pages the user can read |
| 430 | $title = Title::newFromText( $page ); |
| 431 | if ( $title === null ) { |
| 432 | // @todo Perhaps output an <error> tag or something. |
| 433 | continue; |
| 434 | } |
| 435 | |
| 436 | if ( !$this->getAuthority()->authorizeRead( 'read', $title ) ) { |
| 437 | // @todo Perhaps output an <error> tag or something. |
| 438 | continue; |
| 439 | } |
| 440 | |
| 441 | $exporter->pageByTitle( $title ); |
| 442 | } |
| 443 | } |
| 444 | |
| 445 | $exporter->closeStream(); |
| 446 | } |
| 447 | |
| 448 | /** |
| 449 | * @param PageIdentity $page |
| 450 | * @return string[] |
| 451 | */ |
| 452 | protected function getPagesFromCategory( PageIdentity $page ) { |
| 453 | $maxPages = $this->getConfig()->get( MainConfigNames::ExportPagelistLimit ); |
| 454 | |
| 455 | $name = $page->getDBkey(); |
| 456 | |
| 457 | $dbr = $this->dbProvider->getReplicaDatabase(); |
| 458 | $res = $dbr->newSelectQueryBuilder() |
| 459 | ->select( [ 'page_namespace', 'page_title' ] ) |
| 460 | ->from( 'page' ) |
| 461 | ->join( 'categorylinks', null, 'cl_from=page_id' ) |
| 462 | ->where( [ 'cl_to' => $name ] ) |
| 463 | ->limit( $maxPages ) |
| 464 | ->caller( __METHOD__ )->fetchResultSet(); |
| 465 | |
| 466 | $pages = []; |
| 467 | |
| 468 | foreach ( $res as $row ) { |
| 469 | $pages[] = Title::makeName( $row->page_namespace, $row->page_title ); |
| 470 | } |
| 471 | |
| 472 | return $pages; |
| 473 | } |
| 474 | |
| 475 | /** |
| 476 | * @param int $nsindex |
| 477 | * @return string[] |
| 478 | */ |
| 479 | protected function getPagesFromNamespace( $nsindex ) { |
| 480 | $maxPages = $this->getConfig()->get( MainConfigNames::ExportPagelistLimit ); |
| 481 | |
| 482 | $dbr = $this->dbProvider->getReplicaDatabase(); |
| 483 | $res = $dbr->newSelectQueryBuilder() |
| 484 | ->select( [ 'page_namespace', 'page_title' ] ) |
| 485 | ->from( 'page' ) |
| 486 | ->where( [ 'page_namespace' => $nsindex ] ) |
| 487 | ->limit( $maxPages ) |
| 488 | ->caller( __METHOD__ )->fetchResultSet(); |
| 489 | |
| 490 | $pages = []; |
| 491 | |
| 492 | foreach ( $res as $row ) { |
| 493 | $pages[] = Title::makeName( $row->page_namespace, $row->page_title ); |
| 494 | } |
| 495 | |
| 496 | return $pages; |
| 497 | } |
| 498 | |
| 499 | /** |
| 500 | * Expand a list of pages to include templates used in those pages. |
| 501 | * @param array $inputPages List of titles to look up |
| 502 | * @param array $pageSet Associative array indexed by titles for output |
| 503 | * @return array Associative array index by titles |
| 504 | */ |
| 505 | protected function getTemplates( $inputPages, $pageSet ) { |
| 506 | [ $nsField, $titleField ] = $this->linksMigration->getTitleFields( 'templatelinks' ); |
| 507 | $queryInfo = $this->linksMigration->getQueryInfo( 'templatelinks' ); |
| 508 | $dbr = $this->dbProvider->getReplicaDatabase(); |
| 509 | $queryBuilder = $dbr->newSelectQueryBuilder() |
| 510 | ->caller( __METHOD__ ) |
| 511 | ->select( [ 'namespace' => $nsField, 'title' => $titleField ] ) |
| 512 | ->from( 'page' ) |
| 513 | ->join( 'templatelinks', null, 'page_id=tl_from' ) |
| 514 | ->tables( array_diff( $queryInfo['tables'], [ 'templatelinks' ] ) ) |
| 515 | ->joinConds( $queryInfo['joins'] ); |
| 516 | return $this->getLinks( $inputPages, $pageSet, $queryBuilder ); |
| 517 | } |
| 518 | |
| 519 | /** |
| 520 | * Add extra pages to the list of pages to export. |
| 521 | * @param string[] $inputPages List of page titles to export |
| 522 | * @param bool[] $pageSet Initial associative array indexed by string page titles |
| 523 | * @return bool[] Associative array indexed by string page titles including extra pages |
| 524 | */ |
| 525 | private function getExtraPages( $inputPages, $pageSet ) { |
| 526 | $extraPages = []; |
| 527 | $this->getHookRunner()->onSpecialExportGetExtraPages( $inputPages, $extraPages ); |
| 528 | foreach ( $extraPages as $extraPage ) { |
| 529 | $pageSet[$this->titleFormatter->getPrefixedText( $extraPage )] = true; |
| 530 | } |
| 531 | return $pageSet; |
| 532 | } |
| 533 | |
| 534 | /** |
| 535 | * Validate link depth setting, if available. |
| 536 | * @param int|null $depth |
| 537 | * @return int |
| 538 | */ |
| 539 | protected function validateLinkDepth( $depth ) { |
| 540 | if ( $depth === null || $depth < 0 ) { |
| 541 | return 0; |
| 542 | } |
| 543 | |
| 544 | if ( !$this->userCanOverrideExportDepth() ) { |
| 545 | $maxLinkDepth = $this->getConfig()->get( MainConfigNames::ExportMaxLinkDepth ); |
| 546 | if ( $depth > $maxLinkDepth ) { |
| 547 | return $maxLinkDepth; |
| 548 | } |
| 549 | } |
| 550 | |
| 551 | /* |
| 552 | * There's a HARD CODED limit of 5 levels of recursion here to prevent a |
| 553 | * crazy-big export from being done by someone setting the depth |
| 554 | * number too high. In other words, last resort safety net. |
| 555 | */ |
| 556 | |
| 557 | return intval( min( $depth, 5 ) ); |
| 558 | } |
| 559 | |
| 560 | /** |
| 561 | * Expand a list of pages to include pages linked to from that page. |
| 562 | * @param array $inputPages |
| 563 | * @param array $pageSet |
| 564 | * @param int $depth |
| 565 | * @return array |
| 566 | */ |
| 567 | protected function getPageLinks( $inputPages, $pageSet, $depth ) { |
| 568 | for ( ; $depth > 0; --$depth ) { |
| 569 | [ $nsField, $titleField ] = $this->linksMigration->getTitleFields( 'pagelinks' ); |
| 570 | $queryInfo = $this->linksMigration->getQueryInfo( 'pagelinks' ); |
| 571 | $dbr = $this->dbProvider->getReplicaDatabase(); |
| 572 | $queryBuilder = $dbr->newSelectQueryBuilder() |
| 573 | ->caller( __METHOD__ ) |
| 574 | ->select( [ 'namespace' => $nsField, 'title' => $titleField ] ) |
| 575 | ->from( 'page' ) |
| 576 | ->join( 'pagelinks', null, 'page_id=pl_from' ) |
| 577 | ->tables( array_diff( $queryInfo['tables'], [ 'pagelinks' ] ) ) |
| 578 | ->joinConds( $queryInfo['joins'] ); |
| 579 | $pageSet = $this->getLinks( $inputPages, $pageSet, $queryBuilder ); |
| 580 | $inputPages = array_keys( $pageSet ); |
| 581 | } |
| 582 | |
| 583 | return $pageSet; |
| 584 | } |
| 585 | |
| 586 | /** |
| 587 | * Expand a list of pages to include items used in those pages. |
| 588 | * @param array $inputPages Array of page titles |
| 589 | * @param array $pageSet |
| 590 | * @param SelectQueryBuilder $queryBuilder |
| 591 | * @return array |
| 592 | */ |
| 593 | protected function getLinks( $inputPages, $pageSet, SelectQueryBuilder $queryBuilder ) { |
| 594 | foreach ( $inputPages as $page ) { |
| 595 | $title = Title::newFromText( $page ); |
| 596 | if ( $title ) { |
| 597 | $pageSet[$title->getPrefixedText()] = true; |
| 598 | /// @todo FIXME: May or may not be more efficient to batch these |
| 599 | /// by namespace when given multiple input pages. |
| 600 | $result = ( clone $queryBuilder ) |
| 601 | ->where( [ |
| 602 | 'page_namespace' => $title->getNamespace(), |
| 603 | 'page_title' => $title->getDBkey() |
| 604 | ] ) |
| 605 | ->fetchResultSet(); |
| 606 | |
| 607 | foreach ( $result as $row ) { |
| 608 | $template = Title::makeTitle( $row->namespace, $row->title ); |
| 609 | $pageSet[$template->getPrefixedText()] = true; |
| 610 | } |
| 611 | } |
| 612 | } |
| 613 | |
| 614 | return $pageSet; |
| 615 | } |
| 616 | |
| 617 | protected function getGroupName() { |
| 618 | return 'pagetools'; |
| 619 | } |
| 620 | } |
| 621 | |
| 622 | /** @deprecated class alias since 1.41 */ |
| 623 | class_alias( SpecialExport::class, 'SpecialExport' ); |