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
28 * XML file reader for the page data importer
30 * implements Special:Import
31 * @ingroup SpecialPage
34 private $reader = null;
35 private $mLogItemCallback, $mUploadCallback, $mRevisionCallback, $mPageCallback;
36 private $mSiteInfoCallback, $mTargetNamespace, $mTargetRootPage, $mPageOutCallback;
37 private $mNoticeCallback, $mDebug;
38 private $mImportUploads, $mImageBasePath;
39 private $mNoUpdates = false;
42 * Creates an ImportXMLReader drawing from the source provided
43 * @param string $source
45 function __construct( $source ) {
46 $this->reader
= new XMLReader();
48 if ( !in_array( 'uploadsource', stream_get_wrappers() ) ) {
49 stream_wrapper_register( 'uploadsource', 'UploadSourceAdapter' );
51 $id = UploadSourceAdapter
::registerSource( $source );
52 if ( defined( 'LIBXML_PARSEHUGE' ) ) {
53 $this->reader
->open( "uploadsource://$id", null, LIBXML_PARSEHUGE
);
55 $this->reader
->open( "uploadsource://$id" );
59 $this->setRevisionCallback( array( $this, "importRevision" ) );
60 $this->setUploadCallback( array( $this, 'importUpload' ) );
61 $this->setLogItemCallback( array( $this, 'importLogItem' ) );
62 $this->setPageOutCallback( array( $this, 'finishImportPage' ) );
66 * @return null|XMLReader
68 public function getReader() {
72 public function throwXmlError( $err ) {
73 $this->debug( "FAILURE: $err" );
74 wfDebug( "WikiImporter XML error: $err\n" );
77 public function debug( $data ) {
78 if ( $this->mDebug
) {
79 wfDebug( "IMPORT: $data\n" );
83 public function warn( $data ) {
84 wfDebug( "IMPORT: $data\n" );
87 public function notice( $msg /*, $param, ...*/ ) {
88 $params = func_get_args();
89 array_shift( $params );
91 if ( is_callable( $this->mNoticeCallback
) ) {
92 call_user_func( $this->mNoticeCallback
, $msg, $params );
93 } else { # No ImportReporter -> CLI
94 echo wfMessage( $msg, $params )->text() . "\n";
102 function setDebug( $debug ) {
103 $this->mDebug
= $debug;
107 * Set 'no updates' mode. In this mode, the link tables will not be updated by the importer
108 * @param bool $noupdates
110 function setNoUpdates( $noupdates ) {
111 $this->mNoUpdates
= $noupdates;
115 * Set a callback that displays notice messages
117 * @param callable $callback
120 public function setNoticeCallback( $callback ) {
121 return wfSetVar( $this->mNoticeCallback
, $callback );
125 * Sets the action to perform as each new page in the stream is reached.
126 * @param callable $callback
129 public function setPageCallback( $callback ) {
130 $previous = $this->mPageCallback
;
131 $this->mPageCallback
= $callback;
136 * Sets the action to perform as each page in the stream is completed.
137 * Callback accepts the page title (as a Title object), a second object
138 * with the original title form (in case it's been overridden into a
139 * local namespace), and a count of revisions.
141 * @param callable $callback
144 public function setPageOutCallback( $callback ) {
145 $previous = $this->mPageOutCallback
;
146 $this->mPageOutCallback
= $callback;
151 * Sets the action to perform as each page revision is reached.
152 * @param callable $callback
155 public function setRevisionCallback( $callback ) {
156 $previous = $this->mRevisionCallback
;
157 $this->mRevisionCallback
= $callback;
162 * Sets the action to perform as each file upload version is reached.
163 * @param callable $callback
166 public function setUploadCallback( $callback ) {
167 $previous = $this->mUploadCallback
;
168 $this->mUploadCallback
= $callback;
173 * Sets the action to perform as each log item reached.
174 * @param callable $callback
177 public function setLogItemCallback( $callback ) {
178 $previous = $this->mLogItemCallback
;
179 $this->mLogItemCallback
= $callback;
184 * Sets the action to perform when site info is encountered
185 * @param callable $callback
188 public function setSiteInfoCallback( $callback ) {
189 $previous = $this->mSiteInfoCallback
;
190 $this->mSiteInfoCallback
= $callback;
195 * Set a target namespace to override the defaults
196 * @param null|int $namespace
199 public function setTargetNamespace( $namespace ) {
200 if ( is_null( $namespace ) ) {
201 // Don't override namespaces
202 $this->mTargetNamespace
= null;
203 } elseif ( $namespace >= 0 ) {
204 // @todo FIXME: Check for validity
205 $this->mTargetNamespace
= intval( $namespace );
212 * Set a target root page under which all pages are imported
213 * @param null|string $rootpage
216 public function setTargetRootPage( $rootpage ) {
217 $status = Status
::newGood();
218 if ( is_null( $rootpage ) ) {
220 $this->mTargetRootPage
= null;
221 } elseif ( $rootpage !== '' ) {
222 $rootpage = rtrim( $rootpage, '/' ); //avoid double slashes
223 $title = Title
::newFromText( $rootpage, !is_null( $this->mTargetNamespace
)
224 ?
$this->mTargetNamespace
228 if ( !$title ||
$title->isExternal() ) {
229 $status->fatal( 'import-rootpage-invalid' );
231 if ( !MWNamespace
::hasSubpages( $title->getNamespace() ) ) {
234 $displayNSText = $title->getNamespace() == NS_MAIN
235 ?
wfMessage( 'blanknamespace' )->text()
236 : $wgContLang->getNsText( $title->getNamespace() );
237 $status->fatal( 'import-rootpage-nosubpage', $displayNSText );
239 // set namespace to 'all', so the namespace check in processTitle() can passed
240 $this->setTargetNamespace( null );
241 $this->mTargetRootPage
= $title->getPrefixedDBkey();
251 public function setImageBasePath( $dir ) {
252 $this->mImageBasePath
= $dir;
256 * @param bool $import
258 public function setImportUploads( $import ) {
259 $this->mImportUploads
= $import;
263 * Default per-revision callback, performs the import.
264 * @param WikiRevision $revision
267 public function importRevision( $revision ) {
268 if ( !$revision->getContent()->getContentHandler()->canBeUsedOn( $revision->getTitle() ) ) {
269 $this->notice( 'import-error-bad-location',
270 $revision->getTitle()->getPrefixedText(),
272 $revision->getModel(),
273 $revision->getFormat() );
279 $dbw = wfGetDB( DB_MASTER
);
280 return $dbw->deadlockLoop( array( $revision, 'importOldRevision' ) );
281 } catch ( MWContentSerializationException
$ex ) {
282 $this->notice( 'import-error-unserialize',
283 $revision->getTitle()->getPrefixedText(),
285 $revision->getModel(),
286 $revision->getFormat() );
293 * Default per-revision callback, performs the import.
294 * @param WikiRevision $revision
297 public function importLogItem( $revision ) {
298 $dbw = wfGetDB( DB_MASTER
);
299 return $dbw->deadlockLoop( array( $revision, 'importLogItem' ) );
304 * @param WikiRevision $revision
307 public function importUpload( $revision ) {
308 $dbw = wfGetDB( DB_MASTER
);
309 return $dbw->deadlockLoop( array( $revision, 'importUpload' ) );
313 * Mostly for hook use
314 * @param Title $title
315 * @param string $origTitle
316 * @param int $revCount
317 * @param int $sRevCount
318 * @param array $pageInfo
321 public function finishImportPage( $title, $origTitle, $revCount, $sRevCount, $pageInfo ) {
322 $args = func_get_args();
323 return wfRunHooks( 'AfterImportPage', $args );
327 * Alternate per-revision callback, for debugging.
328 * @param WikiRevision $revision
330 public function debugRevisionHandler( &$revision ) {
331 $this->debug( "Got revision:" );
332 if ( is_object( $revision->title
) ) {
333 $this->debug( "-- Title: " . $revision->title
->getPrefixedText() );
335 $this->debug( "-- Title: <invalid>" );
337 $this->debug( "-- User: " . $revision->user_text
);
338 $this->debug( "-- Timestamp: " . $revision->timestamp
);
339 $this->debug( "-- Comment: " . $revision->comment
);
340 $this->debug( "-- Text: " . $revision->text
);
344 * Notify the callback function when a new "<page>" is reached.
345 * @param Title $title
347 function pageCallback( $title ) {
348 if ( isset( $this->mPageCallback
) ) {
349 call_user_func( $this->mPageCallback
, $title );
354 * Notify the callback function when a "</page>" is closed.
355 * @param Title $title
356 * @param Title $origTitle
357 * @param int $revCount
358 * @param int $sucCount Number of revisions for which callback returned true
359 * @param array $pageInfo Associative array of page information
361 private function pageOutCallback( $title, $origTitle, $revCount, $sucCount, $pageInfo ) {
362 if ( isset( $this->mPageOutCallback
) ) {
363 $args = func_get_args();
364 call_user_func_array( $this->mPageOutCallback
, $args );
369 * Notify the callback function of a revision
370 * @param WikiRevision $revision
373 private function revisionCallback( $revision ) {
374 if ( isset( $this->mRevisionCallback
) ) {
375 return call_user_func_array( $this->mRevisionCallback
,
376 array( $revision, $this ) );
383 * Notify the callback function of a new log item
384 * @param WikiRevision $revision
387 private function logItemCallback( $revision ) {
388 if ( isset( $this->mLogItemCallback
) ) {
389 return call_user_func_array( $this->mLogItemCallback
,
390 array( $revision, $this ) );
397 * Retrieves the contents of the named attribute of the current element.
398 * @param string $attr the name of the attribute
399 * @return string the value of the attribute or an empty string if it is not set in the current element.
401 public function nodeAttribute( $attr ) {
402 return $this->reader
->getAttribute( $attr );
406 * Shouldn't something like this be built-in to XMLReader?
407 * Fetches text contents of the current element, assuming
408 * no sub-elements or such scary things.
412 public function nodeContents() {
413 if ( $this->reader
->isEmptyElement
) {
417 while ( $this->reader
->read() ) {
418 switch ( $this->reader
->nodeType
) {
419 case XmlReader
::TEXT
:
420 case XmlReader
::SIGNIFICANT_WHITESPACE
:
421 $buffer .= $this->reader
->value
;
423 case XmlReader
::END_ELEMENT
:
428 $this->reader
->close();
434 /** Left in for debugging */
435 private function dumpElement() {
436 static $lookup = null;
438 $xmlReaderConstants = array(
453 "SIGNIFICANT_WHITESPACE",
460 foreach ( $xmlReaderConstants as $name ) {
461 $lookup[constant( "XmlReader::$name" )] = $name;
466 $lookup[$this->reader
->nodeType
],
473 * Primary entry point
474 * @throws MWException
477 public function doImport() {
479 // Calls to reader->read need to be wrapped in calls to
480 // libxml_disable_entity_loader() to avoid local file
481 // inclusion attacks (bug 46932).
482 $oldDisable = libxml_disable_entity_loader( true );
483 $this->reader
->read();
485 if ( $this->reader
->name
!= 'mediawiki' ) {
486 libxml_disable_entity_loader( $oldDisable );
487 throw new MWException( "Expected <mediawiki> tag, got " .
488 $this->reader
->name
);
490 $this->debug( "<mediawiki> tag is correct." );
492 $this->debug( "Starting primary dump processing loop." );
494 $keepReading = $this->reader
->read();
496 while ( $keepReading ) {
497 $tag = $this->reader
->name
;
498 $type = $this->reader
->nodeType
;
500 if ( !wfRunHooks( 'ImportHandleToplevelXMLTag', array( $this ) ) ) {
502 } elseif ( $tag == 'mediawiki' && $type == XmlReader
::END_ELEMENT
) {
504 } elseif ( $tag == 'siteinfo' ) {
505 $this->handleSiteInfo();
506 } elseif ( $tag == 'page' ) {
508 } elseif ( $tag == 'logitem' ) {
509 $this->handleLogItem();
510 } elseif ( $tag != '#text' ) {
511 $this->warn( "Unhandled top-level XML tag $tag" );
517 $keepReading = $this->reader
->next();
519 $this->debug( "Skip" );
521 $keepReading = $this->reader
->read();
525 libxml_disable_entity_loader( $oldDisable );
531 * @throws MWException
533 private function handleSiteInfo() {
534 // Site info is useful, but not actually used for dump imports.
535 // Includes a quick short-circuit to save performance.
536 if ( ! $this->mSiteInfoCallback
) {
537 $this->reader
->next();
540 throw new MWException( "SiteInfo tag is not yet handled, do not set mSiteInfoCallback" );
543 private function handleLogItem() {
544 $this->debug( "Enter log item handler." );
547 // Fields that can just be stuffed in the pageInfo object
548 $normalFields = array( 'id', 'comment', 'type', 'action', 'timestamp',
549 'logtitle', 'params' );
551 while ( $this->reader
->read() ) {
552 if ( $this->reader
->nodeType
== XmlReader
::END_ELEMENT
&&
553 $this->reader
->name
== 'logitem' ) {
557 $tag = $this->reader
->name
;
559 if ( !wfRunHooks( 'ImportHandleLogItemXMLTag', array(
563 } elseif ( in_array( $tag, $normalFields ) ) {
564 $logInfo[$tag] = $this->nodeContents();
565 } elseif ( $tag == 'contributor' ) {
566 $logInfo['contributor'] = $this->handleContributor();
567 } elseif ( $tag != '#text' ) {
568 $this->warn( "Unhandled log-item XML tag $tag" );
572 $this->processLogItem( $logInfo );
576 * @param array $logInfo
579 private function processLogItem( $logInfo ) {
580 $revision = new WikiRevision
;
582 $revision->setID( $logInfo['id'] );
583 $revision->setType( $logInfo['type'] );
584 $revision->setAction( $logInfo['action'] );
585 $revision->setTimestamp( $logInfo['timestamp'] );
586 $revision->setParams( $logInfo['params'] );
587 $revision->setTitle( Title
::newFromText( $logInfo['logtitle'] ) );
588 $revision->setNoUpdates( $this->mNoUpdates
);
590 if ( isset( $logInfo['comment'] ) ) {
591 $revision->setComment( $logInfo['comment'] );
594 if ( isset( $logInfo['contributor']['ip'] ) ) {
595 $revision->setUserIP( $logInfo['contributor']['ip'] );
597 if ( isset( $logInfo['contributor']['username'] ) ) {
598 $revision->setUserName( $logInfo['contributor']['username'] );
601 return $this->logItemCallback( $revision );
604 private function handlePage() {
606 $this->debug( "Enter page handler." );
607 $pageInfo = array( 'revisionCount' => 0, 'successfulRevisionCount' => 0 );
609 // Fields that can just be stuffed in the pageInfo object
610 $normalFields = array( 'title', 'id', 'redirect', 'restrictions' );
615 while ( $skip ?
$this->reader
->next() : $this->reader
->read() ) {
616 if ( $this->reader
->nodeType
== XmlReader
::END_ELEMENT
&&
617 $this->reader
->name
== 'page' ) {
621 $tag = $this->reader
->name
;
624 // The title is invalid, bail out of this page
626 } elseif ( !wfRunHooks( 'ImportHandlePageXMLTag', array( $this,
629 } elseif ( in_array( $tag, $normalFields ) ) {
633 // <title>Page</title>
634 // <redirect title="NewTitle"/>
636 // Because the redirect tag is built differently, we need special handling for that case.
637 if ( $tag == 'redirect' ) {
638 $pageInfo[$tag] = $this->nodeAttribute( 'title' );
640 $pageInfo[$tag] = $this->nodeContents();
641 if ( $tag == 'title' ) {
642 $title = $this->processTitle( $pageInfo['title'] );
649 $this->pageCallback( $title );
650 list( $pageInfo['_title'], $origTitle ) = $title;
653 } elseif ( $tag == 'revision' ) {
654 $this->handleRevision( $pageInfo );
655 } elseif ( $tag == 'upload' ) {
656 $this->handleUpload( $pageInfo );
657 } elseif ( $tag != '#text' ) {
658 $this->warn( "Unhandled page XML tag $tag" );
663 $this->pageOutCallback( $pageInfo['_title'], $origTitle,
664 $pageInfo['revisionCount'],
665 $pageInfo['successfulRevisionCount'],
670 * @param array $pageInfo
672 private function handleRevision( &$pageInfo ) {
673 $this->debug( "Enter revision handler" );
674 $revisionInfo = array();
676 $normalFields = array( 'id', 'timestamp', 'comment', 'minor', 'model', 'format', 'text' );
680 while ( $skip ?
$this->reader
->next() : $this->reader
->read() ) {
681 if ( $this->reader
->nodeType
== XmlReader
::END_ELEMENT
&&
682 $this->reader
->name
== 'revision' ) {
686 $tag = $this->reader
->name
;
688 if ( !wfRunHooks( 'ImportHandleRevisionXMLTag', array(
689 $this, $pageInfo, $revisionInfo
692 } elseif ( in_array( $tag, $normalFields ) ) {
693 $revisionInfo[$tag] = $this->nodeContents();
694 } elseif ( $tag == 'contributor' ) {
695 $revisionInfo['contributor'] = $this->handleContributor();
696 } elseif ( $tag != '#text' ) {
697 $this->warn( "Unhandled revision XML tag $tag" );
702 $pageInfo['revisionCount']++
;
703 if ( $this->processRevision( $pageInfo, $revisionInfo ) ) {
704 $pageInfo['successfulRevisionCount']++
;
709 * @param array $pageInfo
710 * @param array $revisionInfo
713 private function processRevision( $pageInfo, $revisionInfo ) {
714 $revision = new WikiRevision
;
716 if ( isset( $revisionInfo['id'] ) ) {
717 $revision->setID( $revisionInfo['id'] );
719 if ( isset( $revisionInfo['text'] ) ) {
720 $revision->setText( $revisionInfo['text'] );
722 if ( isset( $revisionInfo['model'] ) ) {
723 $revision->setModel( $revisionInfo['model'] );
725 if ( isset( $revisionInfo['format'] ) ) {
726 $revision->setFormat( $revisionInfo['format'] );
728 $revision->setTitle( $pageInfo['_title'] );
730 if ( isset( $revisionInfo['timestamp'] ) ) {
731 $revision->setTimestamp( $revisionInfo['timestamp'] );
733 $revision->setTimestamp( wfTimestampNow() );
736 if ( isset( $revisionInfo['comment'] ) ) {
737 $revision->setComment( $revisionInfo['comment'] );
740 if ( isset( $revisionInfo['minor'] ) ) {
741 $revision->setMinor( true );
743 if ( isset( $revisionInfo['contributor']['ip'] ) ) {
744 $revision->setUserIP( $revisionInfo['contributor']['ip'] );
746 if ( isset( $revisionInfo['contributor']['username'] ) ) {
747 $revision->setUserName( $revisionInfo['contributor']['username'] );
749 $revision->setNoUpdates( $this->mNoUpdates
);
751 return $this->revisionCallback( $revision );
755 * @param array $pageInfo
758 private function handleUpload( &$pageInfo ) {
759 $this->debug( "Enter upload handler" );
760 $uploadInfo = array();
762 $normalFields = array( 'timestamp', 'comment', 'filename', 'text',
763 'src', 'size', 'sha1base36', 'archivename', 'rel' );
767 while ( $skip ?
$this->reader
->next() : $this->reader
->read() ) {
768 if ( $this->reader
->nodeType
== XmlReader
::END_ELEMENT
&&
769 $this->reader
->name
== 'upload' ) {
773 $tag = $this->reader
->name
;
775 if ( !wfRunHooks( 'ImportHandleUploadXMLTag', array(
779 } elseif ( in_array( $tag, $normalFields ) ) {
780 $uploadInfo[$tag] = $this->nodeContents();
781 } elseif ( $tag == 'contributor' ) {
782 $uploadInfo['contributor'] = $this->handleContributor();
783 } elseif ( $tag == 'contents' ) {
784 $contents = $this->nodeContents();
785 $encoding = $this->reader
->getAttribute( 'encoding' );
786 if ( $encoding === 'base64' ) {
787 $uploadInfo['fileSrc'] = $this->dumpTemp( base64_decode( $contents ) );
788 $uploadInfo['isTempSrc'] = true;
790 } elseif ( $tag != '#text' ) {
791 $this->warn( "Unhandled upload XML tag $tag" );
796 if ( $this->mImageBasePath
&& isset( $uploadInfo['rel'] ) ) {
797 $path = "{$this->mImageBasePath}/{$uploadInfo['rel']}";
798 if ( file_exists( $path ) ) {
799 $uploadInfo['fileSrc'] = $path;
800 $uploadInfo['isTempSrc'] = false;
804 if ( $this->mImportUploads
) {
805 return $this->processUpload( $pageInfo, $uploadInfo );
810 * @param string $contents
813 private function dumpTemp( $contents ) {
814 $filename = tempnam( wfTempDir(), 'importupload' );
815 file_put_contents( $filename, $contents );
820 * @param array $pageInfo
821 * @param array $uploadInfo
824 private function processUpload( $pageInfo, $uploadInfo ) {
825 $revision = new WikiRevision
;
826 $text = isset( $uploadInfo['text'] ) ?
$uploadInfo['text'] : '';
828 $revision->setTitle( $pageInfo['_title'] );
829 $revision->setID( $pageInfo['id'] );
830 $revision->setTimestamp( $uploadInfo['timestamp'] );
831 $revision->setText( $text );
832 $revision->setFilename( $uploadInfo['filename'] );
833 if ( isset( $uploadInfo['archivename'] ) ) {
834 $revision->setArchiveName( $uploadInfo['archivename'] );
836 $revision->setSrc( $uploadInfo['src'] );
837 if ( isset( $uploadInfo['fileSrc'] ) ) {
838 $revision->setFileSrc( $uploadInfo['fileSrc'],
839 !empty( $uploadInfo['isTempSrc'] ) );
841 if ( isset( $uploadInfo['sha1base36'] ) ) {
842 $revision->setSha1Base36( $uploadInfo['sha1base36'] );
844 $revision->setSize( intval( $uploadInfo['size'] ) );
845 $revision->setComment( $uploadInfo['comment'] );
847 if ( isset( $uploadInfo['contributor']['ip'] ) ) {
848 $revision->setUserIP( $uploadInfo['contributor']['ip'] );
850 if ( isset( $uploadInfo['contributor']['username'] ) ) {
851 $revision->setUserName( $uploadInfo['contributor']['username'] );
853 $revision->setNoUpdates( $this->mNoUpdates
);
855 return call_user_func( $this->mUploadCallback
, $revision );
861 private function handleContributor() {
862 $fields = array( 'id', 'ip', 'username' );
865 while ( $this->reader
->read() ) {
866 if ( $this->reader
->nodeType
== XmlReader
::END_ELEMENT
&&
867 $this->reader
->name
== 'contributor' ) {
871 $tag = $this->reader
->name
;
873 if ( in_array( $tag, $fields ) ) {
874 $info[$tag] = $this->nodeContents();
882 * @param string $text
885 private function processTitle( $text ) {
886 global $wgCommandLineMode;
889 $origTitle = Title
::newFromText( $workTitle );
891 if ( !is_null( $this->mTargetNamespace
) && !is_null( $origTitle ) ) {
892 # makeTitleSafe, because $origTitle can have a interwiki (different setting of interwiki map)
893 # and than dbKey can begin with a lowercase char
894 $title = Title
::makeTitleSafe( $this->mTargetNamespace
,
895 $origTitle->getDBkey() );
897 if ( !is_null( $this->mTargetRootPage
) ) {
898 $workTitle = $this->mTargetRootPage
. '/' . $workTitle;
900 $title = Title
::newFromText( $workTitle );
903 if ( is_null( $title ) ) {
904 # Invalid page title? Ignore the page
905 $this->notice( 'import-error-invalid', $workTitle );
907 } elseif ( $title->isExternal() ) {
908 $this->notice( 'import-error-interwiki', $title->getPrefixedText() );
910 } elseif ( !$title->canExist() ) {
911 $this->notice( 'import-error-special', $title->getPrefixedText() );
913 } elseif ( !$title->userCan( 'edit' ) && !$wgCommandLineMode ) {
914 # Do not import if the importing wiki user cannot edit this page
915 $this->notice( 'import-error-edit', $title->getPrefixedText() );
917 } elseif ( !$title->exists() && !$title->userCan( 'create' ) && !$wgCommandLineMode ) {
918 # Do not import if the importing wiki user cannot create this page
919 $this->notice( 'import-error-create', $title->getPrefixedText() );
923 return array( $title, $origTitle );
927 /** This is a horrible hack used to keep source compatibility */
928 class UploadSourceAdapter
{
930 private static $sourceRegistrations = array();
942 * @param string $source
945 static function registerSource( $source ) {
946 $id = wfRandomString();
948 self
::$sourceRegistrations[$id] = $source;
954 * @param string $path
955 * @param string $mode
956 * @param array $options
957 * @param string $opened_path
960 function stream_open( $path, $mode, $options, &$opened_path ) {
961 $url = parse_url( $path );
964 if ( !isset( self
::$sourceRegistrations[$id] ) ) {
968 $this->mSource
= self
::$sourceRegistrations[$id];
977 function stream_read( $count ) {
981 while ( !$leave && !$this->mSource
->atEnd() &&
982 strlen( $this->mBuffer
) < $count ) {
983 $read = $this->mSource
->readChunk();
985 if ( !strlen( $read ) ) {
989 $this->mBuffer
.= $read;
992 if ( strlen( $this->mBuffer
) ) {
993 $return = substr( $this->mBuffer
, 0, $count );
994 $this->mBuffer
= substr( $this->mBuffer
, $count );
997 $this->mPosition +
= strlen( $return );
1003 * @param string $data
1006 function stream_write( $data ) {
1013 function stream_tell() {
1014 return $this->mPosition
;
1020 function stream_eof() {
1021 return $this->mSource
->atEnd();
1027 function url_stat() {
1030 $result['dev'] = $result[0] = 0;
1031 $result['ino'] = $result[1] = 0;
1032 $result['mode'] = $result[2] = 0;
1033 $result['nlink'] = $result[3] = 0;
1034 $result['uid'] = $result[4] = 0;
1035 $result['gid'] = $result[5] = 0;
1036 $result['rdev'] = $result[6] = 0;
1037 $result['size'] = $result[7] = 0;
1038 $result['atime'] = $result[8] = 0;
1039 $result['mtime'] = $result[9] = 0;
1040 $result['ctime'] = $result[10] = 0;
1041 $result['blksize'] = $result[11] = 0;
1042 $result['blocks'] = $result[12] = 0;
1048 class XMLReader2
extends XMLReader
{
1051 * @return bool|string
1053 function nodeContents() {
1054 if ( $this->isEmptyElement
) {
1058 while ( $this->read() ) {
1059 switch ( $this->nodeType
) {
1060 case XmlReader
::TEXT
:
1061 case XmlReader
::SIGNIFICANT_WHITESPACE
:
1062 $buffer .= $this->value
;
1064 case XmlReader
::END_ELEMENT
:
1068 return $this->close();
1073 * @todo document (e.g. one-sentence class description).
1074 * @ingroup SpecialPage
1076 class WikiRevision
{
1077 /** @todo Unused? */
1078 private $importer = null;
1081 public $title = null;
1087 public $timestamp = "20010115000000";
1091 * @todo Can't find any uses. Public, because that's suspicious. Get clarity. */
1095 public $user_text = "";
1098 protected $model = null;
1101 protected $format = null;
1110 protected $content = null;
1113 public $comment = "";
1116 protected $minor = false;
1119 protected $type = "";
1122 protected $action = "";
1125 protected $params = "";
1128 protected $fileSrc = '';
1130 /** @var bool|string */
1131 protected $sha1base36 = false;
1137 private $isTemp = false;
1140 protected $archiveName = '';
1142 protected $filename;
1147 /** @todo Unused? */
1148 private $fileIsTemp;
1151 private $mNoUpdates = false;
1154 * @param Title $title
1155 * @throws MWException
1157 function setTitle( $title ) {
1158 if ( is_object( $title ) ) {
1159 $this->title
= $title;
1160 } elseif ( is_null( $title ) ) {
1161 throw new MWException( "WikiRevision given a null title in import. "
1162 . "You may need to adjust \$wgLegalTitleChars." );
1164 throw new MWException( "WikiRevision given non-object title in import." );
1171 function setID( $id ) {
1178 function setTimestamp( $ts ) {
1179 # 2003-08-05T18:30:02Z
1180 $this->timestamp
= wfTimestamp( TS_MW
, $ts );
1184 * @param string $user
1186 function setUsername( $user ) {
1187 $this->user_text
= $user;
1193 function setUserIP( $ip ) {
1194 $this->user_text
= $ip;
1198 * @param string $model
1200 function setModel( $model ) {
1201 $this->model
= $model;
1205 * @param string $format
1207 function setFormat( $format ) {
1208 $this->format
= $format;
1212 * @param string $text
1214 function setText( $text ) {
1215 $this->text
= $text;
1219 * @param string $text
1221 function setComment( $text ) {
1222 $this->comment
= $text;
1226 * @param bool $minor
1228 function setMinor( $minor ) {
1229 $this->minor
= (bool)$minor;
1235 function setSrc( $src ) {
1240 * @param string $src
1241 * @param bool $isTemp
1243 function setFileSrc( $src, $isTemp ) {
1244 $this->fileSrc
= $src;
1245 $this->fileIsTemp
= $isTemp;
1249 * @param string $sha1base36
1251 function setSha1Base36( $sha1base36 ) {
1252 $this->sha1base36
= $sha1base36;
1256 * @param string $filename
1258 function setFilename( $filename ) {
1259 $this->filename
= $filename;
1263 * @param string $archiveName
1265 function setArchiveName( $archiveName ) {
1266 $this->archiveName
= $archiveName;
1272 function setSize( $size ) {
1273 $this->size
= intval( $size );
1277 * @param string $type
1279 function setType( $type ) {
1280 $this->type
= $type;
1284 * @param string $action
1286 function setAction( $action ) {
1287 $this->action
= $action;
1291 * @param array $params
1293 function setParams( $params ) {
1294 $this->params
= $params;
1298 * @param bool $noupdates
1300 public function setNoUpdates( $noupdates ) {
1301 $this->mNoUpdates
= $noupdates;
1307 function getTitle() {
1308 return $this->title
;
1321 function getTimestamp() {
1322 return $this->timestamp
;
1328 function getUser() {
1329 return $this->user_text
;
1335 * @deprecated Since 1.21, use getContent() instead.
1337 function getText() {
1338 ContentHandler
::deprecated( __METHOD__
, '1.21' );
1346 function getContent() {
1347 if ( is_null( $this->content
) ) {
1349 ContentHandler
::makeContent(
1357 return $this->content
;
1363 function getModel() {
1364 if ( is_null( $this->model
) ) {
1365 $this->model
= $this->getTitle()->getContentModel();
1368 return $this->model
;
1374 function getFormat() {
1375 if ( is_null( $this->model
) ) {
1376 $this->format
= ContentHandler
::getForTitle( $this->getTitle() )->getDefaultFormat();
1379 return $this->format
;
1385 function getComment() {
1386 return $this->comment
;
1392 function getMinor() {
1393 return $this->minor
;
1404 * @return bool|string
1406 function getSha1() {
1407 if ( $this->sha1base36
) {
1408 return wfBaseConvert( $this->sha1base36
, 36, 16 );
1416 function getFileSrc() {
1417 return $this->fileSrc
;
1423 function isTempSrc() {
1424 return $this->isTemp
;
1430 function getFilename() {
1431 return $this->filename
;
1437 function getArchiveName() {
1438 return $this->archiveName
;
1444 function getSize() {
1451 function getType() {
1458 function getAction() {
1459 return $this->action
;
1465 function getParams() {
1466 return $this->params
;
1472 function importOldRevision() {
1473 $dbw = wfGetDB( DB_MASTER
);
1475 # Sneak a single revision into place
1476 $user = User
::newFromName( $this->getUser() );
1478 $userId = intval( $user->getId() );
1479 $userText = $user->getName();
1483 $userText = $this->getUser();
1484 $userObj = new User
;
1487 // avoid memory leak...?
1488 $linkCache = LinkCache
::singleton();
1489 $linkCache->clear();
1491 $page = WikiPage
::factory( $this->title
);
1492 if ( !$page->exists() ) {
1493 # must create the page...
1494 $pageId = $page->insertOn( $dbw );
1496 $oldcountable = null;
1498 $pageId = $page->getId();
1501 $prior = $dbw->selectField( 'revision', '1',
1502 array( 'rev_page' => $pageId,
1503 'rev_timestamp' => $dbw->timestamp( $this->timestamp
),
1504 'rev_user_text' => $userText,
1505 'rev_comment' => $this->getComment() ),
1509 // @todo FIXME: This could fail slightly for multiple matches :P
1510 wfDebug( __METHOD__
. ": skipping existing revision for [[" .
1511 $this->title
->getPrefixedText() . "]], timestamp " . $this->timestamp
. "\n" );
1514 $oldcountable = $page->isCountable();
1517 # @todo FIXME: Use original rev_id optionally (better for backups)
1519 $revision = new Revision( array(
1520 'title' => $this->title
,
1522 'content_model' => $this->getModel(),
1523 'content_format' => $this->getFormat(),
1524 //XXX: just set 'content' => $this->getContent()?
1525 'text' => $this->getContent()->serialize( $this->getFormat() ),
1526 'comment' => $this->getComment(),
1528 'user_text' => $userText,
1529 'timestamp' => $this->timestamp
,
1530 'minor_edit' => $this->minor
,
1532 $revision->insertOn( $dbw );
1533 $changed = $page->updateIfNewerOn( $dbw, $revision );
1535 if ( $changed !== false && !$this->mNoUpdates
) {
1536 wfDebug( __METHOD__
. ": running updates\n" );
1537 $page->doEditUpdates(
1540 array( 'created' => $created, 'oldcountable' => $oldcountable )
1550 function importLogItem() {
1551 $dbw = wfGetDB( DB_MASTER
);
1552 # @todo FIXME: This will not record autoblocks
1553 if ( !$this->getTitle() ) {
1554 wfDebug( __METHOD__
. ": skipping invalid {$this->type}/{$this->action} log time, timestamp " .
1555 $this->timestamp
. "\n" );
1558 # Check if it exists already
1559 // @todo FIXME: Use original log ID (better for backups)
1560 $prior = $dbw->selectField( 'logging', '1',
1561 array( 'log_type' => $this->getType(),
1562 'log_action' => $this->getAction(),
1563 'log_timestamp' => $dbw->timestamp( $this->timestamp
),
1564 'log_namespace' => $this->getTitle()->getNamespace(),
1565 'log_title' => $this->getTitle()->getDBkey(),
1566 'log_comment' => $this->getComment(),
1567 #'log_user_text' => $this->user_text,
1568 'log_params' => $this->params
),
1571 // @todo FIXME: This could fail slightly for multiple matches :P
1574 . ": skipping existing item for Log:{$this->type}/{$this->action}, timestamp "
1575 . $this->timestamp
. "\n" );
1578 $log_id = $dbw->nextSequenceValue( 'logging_log_id_seq' );
1580 'log_id' => $log_id,
1581 'log_type' => $this->type
,
1582 'log_action' => $this->action
,
1583 'log_timestamp' => $dbw->timestamp( $this->timestamp
),
1584 'log_user' => User
::idFromName( $this->user_text
),
1585 #'log_user_text' => $this->user_text,
1586 'log_namespace' => $this->getTitle()->getNamespace(),
1587 'log_title' => $this->getTitle()->getDBkey(),
1588 'log_comment' => $this->getComment(),
1589 'log_params' => $this->params
1591 $dbw->insert( 'logging', $data, __METHOD__
);
1597 function importUpload() {
1599 $archiveName = $this->getArchiveName();
1600 if ( $archiveName ) {
1601 wfDebug( __METHOD__
. "Importing archived file as $archiveName\n" );
1602 $file = OldLocalFile
::newFromArchiveName( $this->getTitle(),
1603 RepoGroup
::singleton()->getLocalRepo(), $archiveName );
1605 $file = wfLocalFile( $this->getTitle() );
1606 wfDebug( __METHOD__
. 'Importing new file as ' . $file->getName() . "\n" );
1607 if ( $file->exists() && $file->getTimestamp() > $this->getTimestamp() ) {
1608 $archiveName = $file->getTimestamp() . '!' . $file->getName();
1609 $file = OldLocalFile
::newFromArchiveName( $this->getTitle(),
1610 RepoGroup
::singleton()->getLocalRepo(), $archiveName );
1611 wfDebug( __METHOD__
. "File already exists; importing as $archiveName\n" );
1615 wfDebug( __METHOD__
. ': Bad file for ' . $this->getTitle() . "\n" );
1619 # Get the file source or download if necessary
1620 $source = $this->getFileSrc();
1621 $flags = $this->isTempSrc() ? File
::DELETE_SOURCE
: 0;
1623 $source = $this->downloadSource();
1624 $flags |
= File
::DELETE_SOURCE
;
1627 wfDebug( __METHOD__
. ": Could not fetch remote file.\n" );
1630 $sha1 = $this->getSha1();
1631 if ( $sha1 && ( $sha1 !== sha1_file( $source ) ) ) {
1632 if ( $flags & File
::DELETE_SOURCE
) {
1633 # Broken file; delete it if it is a temporary file
1636 wfDebug( __METHOD__
. ": Corrupt file $source.\n" );
1640 $user = User
::newFromName( $this->user_text
);
1642 # Do the actual upload
1643 if ( $archiveName ) {
1644 $status = $file->uploadOld( $source, $archiveName,
1645 $this->getTimestamp(), $this->getComment(), $user, $flags );
1647 $status = $file->upload( $source, $this->getComment(), $this->getComment(),
1648 $flags, false, $this->getTimestamp(), $user );
1651 if ( $status->isGood() ) {
1652 wfDebug( __METHOD__
. ": Successful\n" );
1655 wfDebug( __METHOD__
. ': failed: ' . $status->getXml() . "\n" );
1661 * @return bool|string
1663 function downloadSource() {
1664 global $wgEnableUploads;
1665 if ( !$wgEnableUploads ) {
1669 $tempo = tempnam( wfTempDir(), 'download' );
1670 $f = fopen( $tempo, 'wb' );
1672 wfDebug( "IMPORT: couldn't write to temp file $tempo\n" );
1677 $src = $this->getSrc();
1678 $data = Http
::get( $src );
1680 wfDebug( "IMPORT: couldn't fetch source $src\n" );
1686 fwrite( $f, $data );
1695 * @todo document (e.g. one-sentence class description).
1696 * @ingroup SpecialPage
1698 class ImportStringSource
{
1699 function __construct( $string ) {
1700 $this->mString
= $string;
1701 $this->mRead
= false;
1708 return $this->mRead
;
1712 * @return bool|string
1714 function readChunk() {
1715 if ( $this->atEnd() ) {
1718 $this->mRead
= true;
1719 return $this->mString
;
1724 * @todo document (e.g. one-sentence class description).
1725 * @ingroup SpecialPage
1727 class ImportStreamSource
{
1728 function __construct( $handle ) {
1729 $this->mHandle
= $handle;
1736 return feof( $this->mHandle
);
1742 function readChunk() {
1743 return fread( $this->mHandle
, 32768 );
1747 * @param string $filename
1750 static function newFromFile( $filename ) {
1751 wfSuppressWarnings();
1752 $file = fopen( $filename, 'rt' );
1753 wfRestoreWarnings();
1755 return Status
::newFatal( "importcantopen" );
1757 return Status
::newGood( new ImportStreamSource( $file ) );
1761 * @param string $fieldname
1764 static function newFromUpload( $fieldname = "xmlimport" ) {
1765 $upload =& $_FILES[$fieldname];
1767 if ( $upload === null ||
!$upload['name'] ) {
1768 return Status
::newFatal( 'importnofile' );
1770 if ( !empty( $upload['error'] ) ) {
1771 switch ( $upload['error'] ) {
1773 # The uploaded file exceeds the upload_max_filesize directive in php.ini.
1774 return Status
::newFatal( 'importuploaderrorsize' );
1776 # The uploaded file exceeds the MAX_FILE_SIZE directive that
1777 # was specified in the HTML form.
1778 return Status
::newFatal( 'importuploaderrorsize' );
1780 # The uploaded file was only partially uploaded
1781 return Status
::newFatal( 'importuploaderrorpartial' );
1783 # Missing a temporary folder.
1784 return Status
::newFatal( 'importuploaderrortemp' );
1785 # case else: # Currently impossible
1789 $fname = $upload['tmp_name'];
1790 if ( is_uploaded_file( $fname ) ) {
1791 return ImportStreamSource
::newFromFile( $fname );
1793 return Status
::newFatal( 'importnofile' );
1798 * @param string $url
1799 * @param string $method
1802 static function newFromURL( $url, $method = 'GET' ) {
1803 wfDebug( __METHOD__
. ": opening $url\n" );
1804 # Use the standard HTTP fetch function; it times out
1805 # quicker and sorts out user-agent problems which might
1806 # otherwise prevent importing from large sites, such
1807 # as the Wikimedia cluster, etc.
1808 $data = Http
::request( $method, $url, array( 'followRedirects' => true ) );
1809 if ( $data !== false ) {
1811 fwrite( $file, $data );
1814 return Status
::newGood( new ImportStreamSource( $file ) );
1816 return Status
::newFatal( 'importcantopen' );
1821 * @param string $interwiki
1822 * @param string $page
1823 * @param bool $history
1824 * @param bool $templates
1825 * @param int $pageLinkDepth
1828 public static function newFromInterwiki( $interwiki, $page, $history = false,
1829 $templates = false, $pageLinkDepth = 0
1831 if ( $page == '' ) {
1832 return Status
::newFatal( 'import-noarticle' );
1834 $link = Title
::newFromText( "$interwiki:Special:Export/$page" );
1835 if ( is_null( $link ) ||
!$link->isExternal() ) {
1836 return Status
::newFatal( 'importbadinterwiki' );
1840 $params['history'] = 1;
1843 $params['templates'] = 1;
1845 if ( $pageLinkDepth ) {
1846 $params['pagelink-depth'] = $pageLinkDepth;
1848 $url = $link->getFullURL( $params );
1849 # For interwikis, use POST to avoid redirects.
1850 return ImportStreamSource
::newFromURL( $url, "POST" );