3 * MediaWiki page data importer.
5 * Copyright © 2003,2005 Brion Vibber <brion@pobox.com>
6 * https://www.mediawiki.org/
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.
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.
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
24 * @ingroup SpecialPage
27 use MediaWiki\MediaWikiServices
;
30 * XML file reader for the page data importer.
32 * implements Special:Import
33 * @ingroup SpecialPage
36 private $reader = null;
37 private $foreignNamespaces = null;
38 private $mLogItemCallback, $mUploadCallback, $mRevisionCallback, $mPageCallback;
39 private $mSiteInfoCallback, $mPageOutCallback;
40 private $mNoticeCallback, $mDebug;
41 private $mImportUploads, $mImageBasePath;
42 private $mNoUpdates = false;
43 private $pageOffset = 0;
46 /** @var ImportTitleFactory */
47 private $importTitleFactory;
49 private $countableCache = [];
51 private $disableStatisticsUpdate = false;
52 /** @var ExternalUserNames */
53 private $externalUserNames;
56 * Creates an ImportXMLReader drawing from the source provided
57 * @param ImportSource $source
58 * @param Config $config
61 function __construct( ImportSource
$source, Config
$config ) {
62 if ( !class_exists( 'XMLReader' ) ) {
63 throw new Exception( 'Import requires PHP to have been compiled with libxml support' );
66 $this->reader
= new XMLReader();
67 $this->config
= $config;
69 if ( !in_array( 'uploadsource', stream_get_wrappers() ) ) {
70 stream_wrapper_register( 'uploadsource', UploadSourceAdapter
::class );
72 $id = UploadSourceAdapter
::registerSource( $source );
74 // Enable the entity loader, as it is needed for loading external URLs via
75 // XMLReader::open (T86036)
76 $oldDisable = libxml_disable_entity_loader( false );
77 if ( defined( 'LIBXML_PARSEHUGE' ) ) {
78 $status = $this->reader
->open( "uploadsource://$id", null, LIBXML_PARSEHUGE
);
80 $status = $this->reader
->open( "uploadsource://$id" );
83 $error = libxml_get_last_error();
84 libxml_disable_entity_loader( $oldDisable );
85 throw new MWException( 'Encountered an internal error while initializing WikiImporter object: ' .
88 libxml_disable_entity_loader( $oldDisable );
91 $this->setPageCallback( [ $this, 'beforeImportPage' ] );
92 $this->setRevisionCallback( [ $this, "importRevision" ] );
93 $this->setUploadCallback( [ $this, 'importUpload' ] );
94 $this->setLogItemCallback( [ $this, 'importLogItem' ] );
95 $this->setPageOutCallback( [ $this, 'finishImportPage' ] );
97 $this->importTitleFactory
= new NaiveImportTitleFactory();
98 $this->externalUserNames
= new ExternalUserNames( 'imported', false );
102 * @return null|XMLReader
104 public function getReader() {
105 return $this->reader
;
108 public function throwXmlError( $err ) {
109 $this->debug( "FAILURE: $err" );
110 wfDebug( "WikiImporter XML error: $err\n" );
113 public function debug( $data ) {
114 if ( $this->mDebug
) {
115 wfDebug( "IMPORT: $data\n" );
119 public function warn( $data ) {
120 wfDebug( "IMPORT: $data\n" );
123 public function notice( $msg /*, $param, ...*/ ) {
124 $params = func_get_args();
125 array_shift( $params );
127 if ( is_callable( $this->mNoticeCallback
) ) {
128 call_user_func( $this->mNoticeCallback
, $msg, $params );
129 } else { # No ImportReporter -> CLI
130 // T177997: the command line importers should call setNoticeCallback()
131 // for their own custom callback to echo the notice
132 wfDebug( wfMessage( $msg, $params )->text() . "\n" );
140 function setDebug( $debug ) {
141 $this->mDebug
= $debug;
145 * Set 'no updates' mode. In this mode, the link tables will not be updated by the importer
146 * @param bool $noupdates
148 function setNoUpdates( $noupdates ) {
149 $this->mNoUpdates
= $noupdates;
153 * Sets 'pageOffset' value. So it will skip the first n-1 pages
154 * and start from the nth page. It's 1-based indexing.
155 * @param int $nthPage
158 function setPageOffset( $nthPage ) {
159 $this->pageOffset
= $nthPage;
163 * Set a callback that displays notice messages
165 * @param callable $callback
168 public function setNoticeCallback( $callback ) {
169 return wfSetVar( $this->mNoticeCallback
, $callback );
173 * Sets the action to perform as each new page in the stream is reached.
174 * @param callable $callback
177 public function setPageCallback( $callback ) {
178 $previous = $this->mPageCallback
;
179 $this->mPageCallback
= $callback;
184 * Sets the action to perform as each page in the stream is completed.
185 * Callback accepts the page title (as a Title object), a second object
186 * with the original title form (in case it's been overridden into a
187 * local namespace), and a count of revisions.
189 * @param callable $callback
192 public function setPageOutCallback( $callback ) {
193 $previous = $this->mPageOutCallback
;
194 $this->mPageOutCallback
= $callback;
199 * Sets the action to perform as each page revision is reached.
200 * @param callable $callback
203 public function setRevisionCallback( $callback ) {
204 $previous = $this->mRevisionCallback
;
205 $this->mRevisionCallback
= $callback;
210 * Sets the action to perform as each file upload version is reached.
211 * @param callable $callback
214 public function setUploadCallback( $callback ) {
215 $previous = $this->mUploadCallback
;
216 $this->mUploadCallback
= $callback;
221 * Sets the action to perform as each log item reached.
222 * @param callable $callback
225 public function setLogItemCallback( $callback ) {
226 $previous = $this->mLogItemCallback
;
227 $this->mLogItemCallback
= $callback;
232 * Sets the action to perform when site info is encountered
233 * @param callable $callback
236 public function setSiteInfoCallback( $callback ) {
237 $previous = $this->mSiteInfoCallback
;
238 $this->mSiteInfoCallback
= $callback;
243 * Sets the factory object to use to convert ForeignTitle objects into local
245 * @param ImportTitleFactory $factory
247 public function setImportTitleFactory( $factory ) {
248 $this->importTitleFactory
= $factory;
252 * Set a target namespace to override the defaults
253 * @param null|int $namespace
256 public function setTargetNamespace( $namespace ) {
257 if ( is_null( $namespace ) ) {
258 // Don't override namespaces
259 $this->setImportTitleFactory( new NaiveImportTitleFactory() );
263 MWNamespace
::exists( intval( $namespace ) )
265 $namespace = intval( $namespace );
266 $this->setImportTitleFactory( new NamespaceImportTitleFactory( $namespace ) );
274 * Set a target root page under which all pages are imported
275 * @param null|string $rootpage
278 public function setTargetRootPage( $rootpage ) {
279 $status = Status
::newGood();
280 if ( is_null( $rootpage ) ) {
282 $this->setImportTitleFactory( new NaiveImportTitleFactory() );
283 } elseif ( $rootpage !== '' ) {
284 $rootpage = rtrim( $rootpage, '/' ); // avoid double slashes
285 $title = Title
::newFromText( $rootpage );
287 if ( !$title ||
$title->isExternal() ) {
288 $status->fatal( 'import-rootpage-invalid' );
289 } elseif ( !MWNamespace
::hasSubpages( $title->getNamespace() ) ) {
290 $displayNSText = $title->getNamespace() == NS_MAIN
291 ?
wfMessage( 'blanknamespace' )->text()
292 : MediaWikiServices
::getInstance()->getContentLanguage()->
293 getNsText( $title->getNamespace() );
294 $status->fatal( 'import-rootpage-nosubpage', $displayNSText );
296 // set namespace to 'all', so the namespace check in processTitle() can pass
297 $this->setTargetNamespace( null );
298 $this->setImportTitleFactory( new SubpageImportTitleFactory( $title ) );
307 public function setImageBasePath( $dir ) {
308 $this->mImageBasePath
= $dir;
312 * @param bool $import
314 public function setImportUploads( $import ) {
315 $this->mImportUploads
= $import;
320 * @param string $usernamePrefix Prefix to apply to unknown (and possibly also known) usernames
321 * @param bool $assignKnownUsers Whether to apply the prefix to usernames that exist locally
323 public function setUsernamePrefix( $usernamePrefix, $assignKnownUsers ) {
324 $this->externalUserNames
= new ExternalUserNames( $usernamePrefix, $assignKnownUsers );
328 * Statistics update can cause a lot of time
331 public function disableStatisticsUpdate() {
332 $this->disableStatisticsUpdate
= true;
336 * Default per-page callback. Sets up some things related to site statistics
337 * @param array $titleAndForeignTitle Two-element array, with Title object at
338 * index 0 and ForeignTitle object at index 1
341 public function beforeImportPage( $titleAndForeignTitle ) {
342 $title = $titleAndForeignTitle[0];
343 $page = WikiPage
::factory( $title );
344 $this->countableCache
['title_' . $title->getPrefixedText()] = $page->isCountable();
349 * Default per-revision callback, performs the import.
350 * @param WikiRevision $revision
353 public function importRevision( $revision ) {
354 if ( !$revision->getContentHandler()->canBeUsedOn( $revision->getTitle() ) ) {
355 $this->notice( 'import-error-bad-location',
356 $revision->getTitle()->getPrefixedText(),
358 $revision->getModel(),
359 $revision->getFormat() );
365 return $revision->importOldRevision();
366 } catch ( MWContentSerializationException
$ex ) {
367 $this->notice( 'import-error-unserialize',
368 $revision->getTitle()->getPrefixedText(),
370 $revision->getModel(),
371 $revision->getFormat() );
378 * Default per-revision callback, performs the import.
379 * @param WikiRevision $revision
382 public function importLogItem( $revision ) {
383 return $revision->importLogItem();
388 * @param WikiRevision $revision
391 public function importUpload( $revision ) {
392 return $revision->importUpload();
396 * Mostly for hook use
397 * @param Title $title
398 * @param ForeignTitle $foreignTitle
399 * @param int $revCount
400 * @param int $sRevCount
401 * @param array $pageInfo
404 public function finishImportPage( $title, $foreignTitle, $revCount,
405 $sRevCount, $pageInfo
407 // Update article count statistics (T42009)
408 // The normal counting logic in WikiPage->doEditUpdates() is designed for
409 // one-revision-at-a-time editing, not bulk imports. In this situation it
410 // suffers from issues of replica DB lag. We let WikiPage handle the total page
411 // and revision count, and we implement our own custom logic for the
412 // article (content page) count.
413 if ( !$this->disableStatisticsUpdate
) {
414 $page = WikiPage
::factory( $title );
415 $page->loadPageData( 'fromdbmaster' );
416 $content = $page->getContent();
417 if ( $content === null ) {
418 wfDebug( __METHOD__
. ': Skipping article count adjustment for ' . $title .
419 ' because WikiPage::getContent() returned null' );
421 $editInfo = $page->prepareContentForEdit( $content );
422 $countKey = 'title_' . $title->getPrefixedText();
423 $countable = $page->isCountable( $editInfo );
424 if ( array_key_exists( $countKey, $this->countableCache
) &&
425 $countable != $this->countableCache
[$countKey] ) {
426 DeferredUpdates
::addUpdate( SiteStatsUpdate
::factory( [
427 'articles' => ( (int)$countable - (int)$this->countableCache
[$countKey] )
433 $args = func_get_args();
434 return Hooks
::run( 'AfterImportPage', $args );
438 * Alternate per-revision callback, for debugging.
439 * @param WikiRevision &$revision
441 public function debugRevisionHandler( &$revision ) {
442 $this->debug( "Got revision:" );
443 if ( is_object( $revision->title
) ) {
444 $this->debug( "-- Title: " . $revision->title
->getPrefixedText() );
446 $this->debug( "-- Title: <invalid>" );
448 $this->debug( "-- User: " . $revision->user_text
);
449 $this->debug( "-- Timestamp: " . $revision->timestamp
);
450 $this->debug( "-- Comment: " . $revision->comment
);
451 $this->debug( "-- Text: " . $revision->text
);
455 * Notify the callback function of site info
456 * @param array $siteInfo
459 private function siteInfoCallback( $siteInfo ) {
460 if ( isset( $this->mSiteInfoCallback
) ) {
461 return call_user_func_array( $this->mSiteInfoCallback
,
462 [ $siteInfo, $this ] );
469 * Notify the callback function when a new "<page>" is reached.
470 * @param Title $title
472 function pageCallback( $title ) {
473 if ( isset( $this->mPageCallback
) ) {
474 call_user_func( $this->mPageCallback
, $title );
479 * Notify the callback function when a "</page>" is closed.
480 * @param Title $title
481 * @param ForeignTitle $foreignTitle
482 * @param int $revCount
483 * @param int $sucCount Number of revisions for which callback returned true
484 * @param array $pageInfo Associative array of page information
486 private function pageOutCallback( $title, $foreignTitle, $revCount,
487 $sucCount, $pageInfo ) {
488 if ( isset( $this->mPageOutCallback
) ) {
489 $args = func_get_args();
490 call_user_func_array( $this->mPageOutCallback
, $args );
495 * Notify the callback function of a revision
496 * @param WikiRevision $revision
499 private function revisionCallback( $revision ) {
500 if ( isset( $this->mRevisionCallback
) ) {
501 return call_user_func_array( $this->mRevisionCallback
,
502 [ $revision, $this ] );
509 * Notify the callback function of a new log item
510 * @param WikiRevision $revision
513 private function logItemCallback( $revision ) {
514 if ( isset( $this->mLogItemCallback
) ) {
515 return call_user_func_array( $this->mLogItemCallback
,
516 [ $revision, $this ] );
523 * Retrieves the contents of the named attribute of the current element.
524 * @param string $attr The name of the attribute
525 * @return string The value of the attribute or an empty string if it is not set in the current
528 public function nodeAttribute( $attr ) {
529 return $this->reader
->getAttribute( $attr );
533 * Shouldn't something like this be built-in to XMLReader?
534 * Fetches text contents of the current element, assuming
535 * no sub-elements or such scary things.
539 public function nodeContents() {
540 if ( $this->reader
->isEmptyElement
) {
544 while ( $this->reader
->read() ) {
545 switch ( $this->reader
->nodeType
) {
546 case XMLReader
::TEXT
:
547 case XMLReader
::CDATA
:
548 case XMLReader
::SIGNIFICANT_WHITESPACE
:
549 $buffer .= $this->reader
->value
;
551 case XMLReader
::END_ELEMENT
:
556 $this->reader
->close();
561 * Primary entry point
563 * @throws MWException
566 public function doImport() {
567 // Calls to reader->read need to be wrapped in calls to
568 // libxml_disable_entity_loader() to avoid local file
569 // inclusion attacks (T48932).
570 $oldDisable = libxml_disable_entity_loader( true );
571 $this->reader
->read();
573 if ( $this->reader
->localName
!= 'mediawiki' ) {
574 libxml_disable_entity_loader( $oldDisable );
575 throw new MWException( "Expected <mediawiki> tag, got " .
576 $this->reader
->localName
);
578 $this->debug( "<mediawiki> tag is correct." );
580 $this->debug( "Starting primary dump processing loop." );
582 $keepReading = $this->reader
->read();
587 while ( $keepReading ) {
588 $tag = $this->reader
->localName
;
589 if ( $this->pageOffset
) {
590 if ( $tag === 'page' ) {
593 if ( $pageCount < $this->pageOffset
) {
594 $keepReading = $this->reader
->next();
598 $type = $this->reader
->nodeType
;
600 if ( !Hooks
::run( 'ImportHandleToplevelXMLTag', [ $this ] ) ) {
602 } elseif ( $tag == 'mediawiki' && $type == XMLReader
::END_ELEMENT
) {
604 } elseif ( $tag == 'siteinfo' ) {
605 $this->handleSiteInfo();
606 } elseif ( $tag == 'page' ) {
608 } elseif ( $tag == 'logitem' ) {
609 $this->handleLogItem();
610 } elseif ( $tag != '#text' ) {
611 $this->warn( "Unhandled top-level XML tag $tag" );
617 $keepReading = $this->reader
->next();
619 $this->debug( "Skip" );
621 $keepReading = $this->reader
->read();
624 } catch ( Exception
$ex ) {
629 libxml_disable_entity_loader( $oldDisable );
630 $this->reader
->close();
639 private function handleSiteInfo() {
640 $this->debug( "Enter site info handler." );
643 // Fields that can just be stuffed in the siteInfo object
644 $normalFields = [ 'sitename', 'base', 'generator', 'case' ];
646 while ( $this->reader
->read() ) {
647 if ( $this->reader
->nodeType
== XMLReader
::END_ELEMENT
&&
648 $this->reader
->localName
== 'siteinfo' ) {
652 $tag = $this->reader
->localName
;
654 if ( $tag == 'namespace' ) {
655 $this->foreignNamespaces
[$this->nodeAttribute( 'key' )] =
656 $this->nodeContents();
657 } elseif ( in_array( $tag, $normalFields ) ) {
658 $siteInfo[$tag] = $this->nodeContents();
662 $siteInfo['_namespaces'] = $this->foreignNamespaces
;
663 $this->siteInfoCallback( $siteInfo );
666 private function handleLogItem() {
667 $this->debug( "Enter log item handler." );
670 // Fields that can just be stuffed in the pageInfo object
671 $normalFields = [ 'id', 'comment', 'type', 'action', 'timestamp',
672 'logtitle', 'params' ];
674 while ( $this->reader
->read() ) {
675 if ( $this->reader
->nodeType
== XMLReader
::END_ELEMENT
&&
676 $this->reader
->localName
== 'logitem' ) {
680 $tag = $this->reader
->localName
;
682 if ( !Hooks
::run( 'ImportHandleLogItemXMLTag', [
686 } elseif ( in_array( $tag, $normalFields ) ) {
687 $logInfo[$tag] = $this->nodeContents();
688 } elseif ( $tag == 'contributor' ) {
689 $logInfo['contributor'] = $this->handleContributor();
690 } elseif ( $tag != '#text' ) {
691 $this->warn( "Unhandled log-item XML tag $tag" );
695 $this->processLogItem( $logInfo );
699 * @param array $logInfo
702 private function processLogItem( $logInfo ) {
703 $revision = new WikiRevision( $this->config
);
705 if ( isset( $logInfo['id'] ) ) {
706 $revision->setID( $logInfo['id'] );
708 $revision->setType( $logInfo['type'] );
709 $revision->setAction( $logInfo['action'] );
710 if ( isset( $logInfo['timestamp'] ) ) {
711 $revision->setTimestamp( $logInfo['timestamp'] );
713 if ( isset( $logInfo['params'] ) ) {
714 $revision->setParams( $logInfo['params'] );
716 if ( isset( $logInfo['logtitle'] ) ) {
717 // @todo Using Title for non-local titles is a recipe for disaster.
718 // We should use ForeignTitle here instead.
719 $revision->setTitle( Title
::newFromText( $logInfo['logtitle'] ) );
722 $revision->setNoUpdates( $this->mNoUpdates
);
724 if ( isset( $logInfo['comment'] ) ) {
725 $revision->setComment( $logInfo['comment'] );
728 if ( isset( $logInfo['contributor']['ip'] ) ) {
729 $revision->setUserIP( $logInfo['contributor']['ip'] );
732 if ( !isset( $logInfo['contributor']['username'] ) ) {
733 $revision->setUsername( $this->externalUserNames
->addPrefix( 'Unknown user' ) );
735 $revision->setUsername(
736 $this->externalUserNames
->applyPrefix( $logInfo['contributor']['username'] )
740 return $this->logItemCallback( $revision );
743 private function handlePage() {
745 $this->debug( "Enter page handler." );
746 $pageInfo = [ 'revisionCount' => 0, 'successfulRevisionCount' => 0 ];
748 // Fields that can just be stuffed in the pageInfo object
749 $normalFields = [ 'title', 'ns', 'id', 'redirect', 'restrictions' ];
754 while ( $skip ?
$this->reader
->next() : $this->reader
->read() ) {
755 if ( $this->reader
->nodeType
== XMLReader
::END_ELEMENT
&&
756 $this->reader
->localName
== 'page' ) {
762 $tag = $this->reader
->localName
;
765 // The title is invalid, bail out of this page
767 } elseif ( !Hooks
::run( 'ImportHandlePageXMLTag', [ $this,
770 } elseif ( in_array( $tag, $normalFields ) ) {
774 // <title>Page</title>
775 // <redirect title="NewTitle"/>
777 // Because the redirect tag is built differently, we need special handling for that case.
778 if ( $tag == 'redirect' ) {
779 $pageInfo[$tag] = $this->nodeAttribute( 'title' );
781 $pageInfo[$tag] = $this->nodeContents();
783 } elseif ( $tag == 'revision' ||
$tag == 'upload' ) {
784 if ( !isset( $title ) ) {
785 $title = $this->processTitle( $pageInfo['title'],
786 $pageInfo['ns'] ??
null );
788 // $title is either an array of two titles or false.
789 if ( is_array( $title ) ) {
790 $this->pageCallback( $title );
791 list( $pageInfo['_title'], $foreignTitle ) = $title;
799 if ( $tag == 'revision' ) {
800 $this->handleRevision( $pageInfo );
802 $this->handleUpload( $pageInfo );
805 } elseif ( $tag != '#text' ) {
806 $this->warn( "Unhandled page XML tag $tag" );
811 // @note $pageInfo is only set if a valid $title is processed above with
812 // no error. If we have a valid $title, then pageCallback is called
813 // above, $pageInfo['title'] is set and we do pageOutCallback here.
814 // If $pageInfo['_title'] is not set, then $foreignTitle is also not
815 // set since they both come from $title above.
816 if ( array_key_exists( '_title', $pageInfo ) ) {
817 $this->pageOutCallback( $pageInfo['_title'], $foreignTitle,
818 $pageInfo['revisionCount'],
819 $pageInfo['successfulRevisionCount'],
825 * @param array $pageInfo
827 private function handleRevision( &$pageInfo ) {
828 $this->debug( "Enter revision handler" );
831 $normalFields = [ 'id', 'timestamp', 'comment', 'minor', 'model', 'format', 'text', 'sha1' ];
835 while ( $skip ?
$this->reader
->next() : $this->reader
->read() ) {
836 if ( $this->reader
->nodeType
== XMLReader
::END_ELEMENT
&&
837 $this->reader
->localName
== 'revision' ) {
841 $tag = $this->reader
->localName
;
843 if ( !Hooks
::run( 'ImportHandleRevisionXMLTag', [
844 $this, $pageInfo, $revisionInfo
847 } elseif ( in_array( $tag, $normalFields ) ) {
848 $revisionInfo[$tag] = $this->nodeContents();
849 } elseif ( $tag == 'contributor' ) {
850 $revisionInfo['contributor'] = $this->handleContributor();
851 } elseif ( $tag != '#text' ) {
852 $this->warn( "Unhandled revision XML tag $tag" );
857 $pageInfo['revisionCount']++
;
858 if ( $this->processRevision( $pageInfo, $revisionInfo ) ) {
859 $pageInfo['successfulRevisionCount']++
;
864 * @param array $pageInfo
865 * @param array $revisionInfo
866 * @throws MWException
869 private function processRevision( $pageInfo, $revisionInfo ) {
870 global $wgMaxArticleSize;
872 // Make sure revisions won't violate $wgMaxArticleSize, which could lead to
873 // database errors and instability. Testing for revisions with only listed
874 // content models, as other content models might use serialization formats
875 // which aren't checked against $wgMaxArticleSize.
876 if ( ( !isset( $revisionInfo['model'] ) ||
877 in_array( $revisionInfo['model'], [
885 strlen( $revisionInfo['text'] ) > $wgMaxArticleSize * 1024
887 throw new MWException( 'The text of ' .
888 ( isset( $revisionInfo['id'] ) ?
889 "the revision with ID $revisionInfo[id]" :
891 ) . " exceeds the maximum allowable size ($wgMaxArticleSize KB)" );
894 // FIXME: process schema version 11!
895 $revision = new WikiRevision( $this->config
);
897 if ( isset( $revisionInfo['id'] ) ) {
898 $revision->setID( $revisionInfo['id'] );
900 if ( isset( $revisionInfo['model'] ) ) {
901 $revision->setModel( $revisionInfo['model'] );
903 if ( isset( $revisionInfo['format'] ) ) {
904 $revision->setFormat( $revisionInfo['format'] );
906 $revision->setTitle( $pageInfo['_title'] );
908 if ( isset( $revisionInfo['text'] ) ) {
909 $handler = $revision->getContentHandler();
910 $text = $handler->importTransform(
911 $revisionInfo['text'],
912 $revision->getFormat() );
914 $revision->setText( $text );
916 $revision->setTimestamp( $revisionInfo['timestamp'] ??
wfTimestampNow() );
918 if ( isset( $revisionInfo['comment'] ) ) {
919 $revision->setComment( $revisionInfo['comment'] );
922 if ( isset( $revisionInfo['minor'] ) ) {
923 $revision->setMinor( true );
925 if ( isset( $revisionInfo['contributor']['ip'] ) ) {
926 $revision->setUserIP( $revisionInfo['contributor']['ip'] );
927 } elseif ( isset( $revisionInfo['contributor']['username'] ) ) {
928 $revision->setUsername(
929 $this->externalUserNames
->applyPrefix( $revisionInfo['contributor']['username'] )
932 $revision->setUsername( $this->externalUserNames
->addPrefix( 'Unknown user' ) );
934 if ( isset( $revisionInfo['sha1'] ) ) {
935 $revision->setSha1Base36( $revisionInfo['sha1'] );
937 $revision->setNoUpdates( $this->mNoUpdates
);
939 return $this->revisionCallback( $revision );
943 * @param array $pageInfo
946 private function handleUpload( &$pageInfo ) {
947 $this->debug( "Enter upload handler" );
950 $normalFields = [ 'timestamp', 'comment', 'filename', 'text',
951 'src', 'size', 'sha1base36', 'archivename', 'rel' ];
955 while ( $skip ?
$this->reader
->next() : $this->reader
->read() ) {
956 if ( $this->reader
->nodeType
== XMLReader
::END_ELEMENT
&&
957 $this->reader
->localName
== 'upload' ) {
961 $tag = $this->reader
->localName
;
963 if ( !Hooks
::run( 'ImportHandleUploadXMLTag', [
967 } elseif ( in_array( $tag, $normalFields ) ) {
968 $uploadInfo[$tag] = $this->nodeContents();
969 } elseif ( $tag == 'contributor' ) {
970 $uploadInfo['contributor'] = $this->handleContributor();
971 } elseif ( $tag == 'contents' ) {
972 $contents = $this->nodeContents();
973 $encoding = $this->reader
->getAttribute( 'encoding' );
974 if ( $encoding === 'base64' ) {
975 $uploadInfo['fileSrc'] = $this->dumpTemp( base64_decode( $contents ) );
976 $uploadInfo['isTempSrc'] = true;
978 } elseif ( $tag != '#text' ) {
979 $this->warn( "Unhandled upload XML tag $tag" );
984 if ( $this->mImageBasePath
&& isset( $uploadInfo['rel'] ) ) {
985 $path = "{$this->mImageBasePath}/{$uploadInfo['rel']}";
986 if ( file_exists( $path ) ) {
987 $uploadInfo['fileSrc'] = $path;
988 $uploadInfo['isTempSrc'] = false;
992 if ( $this->mImportUploads
) {
993 return $this->processUpload( $pageInfo, $uploadInfo );
998 * @param string $contents
1001 private function dumpTemp( $contents ) {
1002 $filename = tempnam( wfTempDir(), 'importupload' );
1003 file_put_contents( $filename, $contents );
1008 * @param array $pageInfo
1009 * @param array $uploadInfo
1012 private function processUpload( $pageInfo, $uploadInfo ) {
1013 $revision = new WikiRevision( $this->config
);
1014 $text = $uploadInfo['text'] ??
'';
1016 $revision->setTitle( $pageInfo['_title'] );
1017 $revision->setID( $pageInfo['id'] );
1018 $revision->setTimestamp( $uploadInfo['timestamp'] );
1019 $revision->setText( $text );
1020 $revision->setFilename( $uploadInfo['filename'] );
1021 if ( isset( $uploadInfo['archivename'] ) ) {
1022 $revision->setArchiveName( $uploadInfo['archivename'] );
1024 $revision->setSrc( $uploadInfo['src'] );
1025 if ( isset( $uploadInfo['fileSrc'] ) ) {
1026 $revision->setFileSrc( $uploadInfo['fileSrc'],
1027 !empty( $uploadInfo['isTempSrc'] ) );
1029 if ( isset( $uploadInfo['sha1base36'] ) ) {
1030 $revision->setSha1Base36( $uploadInfo['sha1base36'] );
1032 $revision->setSize( intval( $uploadInfo['size'] ) );
1033 $revision->setComment( $uploadInfo['comment'] );
1035 if ( isset( $uploadInfo['contributor']['ip'] ) ) {
1036 $revision->setUserIP( $uploadInfo['contributor']['ip'] );
1038 if ( isset( $uploadInfo['contributor']['username'] ) ) {
1039 $revision->setUsername(
1040 $this->externalUserNames
->applyPrefix( $uploadInfo['contributor']['username'] )
1043 $revision->setNoUpdates( $this->mNoUpdates
);
1045 return call_user_func( $this->mUploadCallback
, $revision );
1051 private function handleContributor() {
1052 $fields = [ 'id', 'ip', 'username' ];
1055 if ( $this->reader
->isEmptyElement
) {
1058 while ( $this->reader
->read() ) {
1059 if ( $this->reader
->nodeType
== XMLReader
::END_ELEMENT
&&
1060 $this->reader
->localName
== 'contributor' ) {
1064 $tag = $this->reader
->localName
;
1066 if ( in_array( $tag, $fields ) ) {
1067 $info[$tag] = $this->nodeContents();
1075 * @param string $text
1076 * @param string|null $ns
1077 * @return array|bool
1079 private function processTitle( $text, $ns = null ) {
1080 if ( is_null( $this->foreignNamespaces
) ) {
1081 $foreignTitleFactory = new NaiveForeignTitleFactory();
1083 $foreignTitleFactory = new NamespaceAwareForeignTitleFactory(
1084 $this->foreignNamespaces
);
1087 $foreignTitle = $foreignTitleFactory->createForeignTitle( $text,
1090 $title = $this->importTitleFactory
->createTitleFromForeignTitle(
1093 $commandLineMode = $this->config
->get( 'CommandLineMode' );
1094 if ( is_null( $title ) ) {
1095 # Invalid page title? Ignore the page
1096 $this->notice( 'import-error-invalid', $foreignTitle->getFullText() );
1098 } elseif ( $title->isExternal() ) {
1099 $this->notice( 'import-error-interwiki', $title->getPrefixedText() );
1101 } elseif ( !$title->canExist() ) {
1102 $this->notice( 'import-error-special', $title->getPrefixedText() );
1104 } elseif ( !$title->userCan( 'edit' ) && !$commandLineMode ) {
1105 # Do not import if the importing wiki user cannot edit this page
1106 $this->notice( 'import-error-edit', $title->getPrefixedText() );
1108 } elseif ( !$title->exists() && !$title->userCan( 'create' ) && !$commandLineMode ) {
1109 # Do not import if the importing wiki user cannot create this page
1110 $this->notice( 'import-error-create', $title->getPrefixedText() );
1114 return [ $title, $foreignTitle ];