3 * MediaWiki page data importer.
5 * Copyright © 2003,2005 Brion Vibber <brion@pobox.com>
6 * http://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
45 function __construct( $source ) {
46 $this->reader
= new XMLReader();
48 stream_wrapper_register( 'uploadsource', 'UploadSourceAdapter' );
49 $id = UploadSourceAdapter
::registerSource( $source );
50 if ( defined( 'LIBXML_PARSEHUGE' ) ) {
51 $this->reader
->open( "uploadsource://$id", null, LIBXML_PARSEHUGE
);
53 $this->reader
->open( "uploadsource://$id" );
57 $this->setRevisionCallback( array( $this, "importRevision" ) );
58 $this->setUploadCallback( array( $this, 'importUpload' ) );
59 $this->setLogItemCallback( array( $this, 'importLogItem' ) );
60 $this->setPageOutCallback( array( $this, 'finishImportPage' ) );
63 private function throwXmlError( $err ) {
64 $this->debug( "FAILURE: $err" );
65 wfDebug( "WikiImporter XML error: $err\n" );
68 private function debug( $data ) {
70 wfDebug( "IMPORT: $data\n" );
74 private function warn( $data ) {
75 wfDebug( "IMPORT: $data\n" );
78 private function notice( $msg /*, $param, ...*/ ) {
79 $params = func_get_args();
80 array_shift( $params );
82 if ( is_callable( $this->mNoticeCallback
) ) {
83 call_user_func( $this->mNoticeCallback
, $msg, $params );
84 } else { # No ImportReporter -> CLI
85 echo wfMessage( $msg, $params )->text() . "\n";
93 function setDebug( $debug ) {
94 $this->mDebug
= $debug;
98 * Set 'no updates' mode. In this mode, the link tables will not be updated by the importer
99 * @param $noupdates bool
101 function setNoUpdates( $noupdates ) {
102 $this->mNoUpdates
= $noupdates;
106 * Set a callback that displays notice messages
108 * @param $callback callback
111 public function setNoticeCallback( $callback ) {
112 return wfSetVar( $this->mNoticeCallback
, $callback );
116 * Sets the action to perform as each new page in the stream is reached.
117 * @param $callback callback
120 public function setPageCallback( $callback ) {
121 $previous = $this->mPageCallback
;
122 $this->mPageCallback
= $callback;
127 * Sets the action to perform as each page in the stream is completed.
128 * Callback accepts the page title (as a Title object), a second object
129 * with the original title form (in case it's been overridden into a
130 * local namespace), and a count of revisions.
132 * @param $callback callback
135 public function setPageOutCallback( $callback ) {
136 $previous = $this->mPageOutCallback
;
137 $this->mPageOutCallback
= $callback;
142 * Sets the action to perform as each page revision is reached.
143 * @param $callback callback
146 public function setRevisionCallback( $callback ) {
147 $previous = $this->mRevisionCallback
;
148 $this->mRevisionCallback
= $callback;
153 * Sets the action to perform as each file upload version is reached.
154 * @param $callback callback
157 public function setUploadCallback( $callback ) {
158 $previous = $this->mUploadCallback
;
159 $this->mUploadCallback
= $callback;
164 * Sets the action to perform as each log item reached.
165 * @param $callback callback
168 public function setLogItemCallback( $callback ) {
169 $previous = $this->mLogItemCallback
;
170 $this->mLogItemCallback
= $callback;
175 * Sets the action to perform when site info is encountered
176 * @param $callback callback
179 public function setSiteInfoCallback( $callback ) {
180 $previous = $this->mSiteInfoCallback
;
181 $this->mSiteInfoCallback
= $callback;
186 * Set a target namespace to override the defaults
190 public function setTargetNamespace( $namespace ) {
191 if( is_null( $namespace ) ) {
192 // Don't override namespaces
193 $this->mTargetNamespace
= null;
194 } elseif( $namespace >= 0 ) {
195 // @todo FIXME: Check for validity
196 $this->mTargetNamespace
= intval( $namespace );
203 * Set a target root page under which all pages are imported
205 * @return status object
207 public function setTargetRootPage( $rootpage ) {
208 $status = Status
::newGood();
209 if( is_null( $rootpage ) ) {
211 $this->mTargetRootPage
= null;
212 } elseif( $rootpage !== '' ) {
213 $rootpage = rtrim( $rootpage, '/' ); //avoid double slashes
214 $title = Title
::newFromText( $rootpage, !is_null( $this->mTargetNamespace
) ?
$this->mTargetNamespace
: NS_MAIN
);
215 if( !$title ||
$title->isExternal() ) {
216 $status->fatal( 'import-rootpage-invalid' );
218 if( !MWNamespace
::hasSubpages( $title->getNamespace() ) ) {
221 $displayNSText = $title->getNamespace() == NS_MAIN
222 ?
wfMessage( 'blanknamespace' )->text()
223 : $wgContLang->getNsText( $title->getNamespace() );
224 $status->fatal( 'import-rootpage-nosubpage', $displayNSText );
226 // set namespace to 'all', so the namespace check in processTitle() can passed
227 $this->setTargetNamespace( null );
228 $this->mTargetRootPage
= $title->getPrefixedDBkey();
238 public function setImageBasePath( $dir ) {
239 $this->mImageBasePath
= $dir;
245 public function setImportUploads( $import ) {
246 $this->mImportUploads
= $import;
250 * Default per-revision callback, performs the import.
251 * @param $revision WikiRevision
254 public function importRevision( $revision ) {
256 $dbw = wfGetDB( DB_MASTER
);
257 return $dbw->deadlockLoop( array( $revision, 'importOldRevision' ) );
258 } catch ( MWContentSerializationException
$ex ) {
259 $this->notice( 'import-error-unserialize',
260 $revision->getTitle()->getPrefixedText(),
262 $revision->getModel(),
263 $revision->getFormat() );
268 * Default per-revision callback, performs the import.
269 * @param $rev WikiRevision
272 public function importLogItem( $rev ) {
273 $dbw = wfGetDB( DB_MASTER
);
274 return $dbw->deadlockLoop( array( $rev, 'importLogItem' ) );
282 public function importUpload( $revision ) {
283 $dbw = wfGetDB( DB_MASTER
);
284 return $dbw->deadlockLoop( array( $revision, 'importUpload' ) );
288 * Mostly for hook use
296 public function finishImportPage( $title, $origTitle, $revCount, $sRevCount, $pageInfo ) {
297 $args = func_get_args();
298 return wfRunHooks( 'AfterImportPage', $args );
302 * Alternate per-revision callback, for debugging.
303 * @param $revision WikiRevision
305 public function debugRevisionHandler( &$revision ) {
306 $this->debug( "Got revision:" );
307 if( is_object( $revision->title
) ) {
308 $this->debug( "-- Title: " . $revision->title
->getPrefixedText() );
310 $this->debug( "-- Title: <invalid>" );
312 $this->debug( "-- User: " . $revision->user_text
);
313 $this->debug( "-- Timestamp: " . $revision->timestamp
);
314 $this->debug( "-- Comment: " . $revision->comment
);
315 $this->debug( "-- Text: " . $revision->text
);
319 * Notify the callback function when a new "<page>" is reached.
320 * @param $title Title
322 function pageCallback( $title ) {
323 if( isset( $this->mPageCallback
) ) {
324 call_user_func( $this->mPageCallback
, $title );
329 * Notify the callback function when a "</page>" is closed.
330 * @param $title Title
331 * @param $origTitle Title
332 * @param $revCount Integer
333 * @param int $sucCount number of revisions for which callback returned true
334 * @param array $pageInfo associative array of page information
336 private function pageOutCallback( $title, $origTitle, $revCount, $sucCount, $pageInfo ) {
337 if( isset( $this->mPageOutCallback
) ) {
338 $args = func_get_args();
339 call_user_func_array( $this->mPageOutCallback
, $args );
344 * Notify the callback function of a revision
345 * @param $revision WikiRevision object
348 private function revisionCallback( $revision ) {
349 if ( isset( $this->mRevisionCallback
) ) {
350 return call_user_func_array( $this->mRevisionCallback
,
351 array( $revision, $this ) );
358 * Notify the callback function of a new log item
359 * @param $revision WikiRevision object
362 private function logItemCallback( $revision ) {
363 if ( isset( $this->mLogItemCallback
) ) {
364 return call_user_func_array( $this->mLogItemCallback
,
365 array( $revision, $this ) );
372 * Shouldn't something like this be built-in to XMLReader?
373 * Fetches text contents of the current element, assuming
374 * no sub-elements or such scary things.
378 private function nodeContents() {
379 if( $this->reader
->isEmptyElement
) {
383 while( $this->reader
->read() ) {
384 switch( $this->reader
->nodeType
) {
385 case XmlReader
::TEXT
:
386 case XmlReader
::SIGNIFICANT_WHITESPACE
:
387 $buffer .= $this->reader
->value
;
389 case XmlReader
::END_ELEMENT
:
394 $this->reader
->close();
400 /** Left in for debugging */
401 private function dumpElement() {
402 static $lookup = null;
404 $xmlReaderConstants = array(
419 "SIGNIFICANT_WHITESPACE",
426 foreach( $xmlReaderConstants as $name ) {
427 $lookup[constant("XmlReader::$name")] = $name;
432 $lookup[$this->reader
->nodeType
],
439 * Primary entry point
440 * @throws MWException
443 public function doImport() {
444 $this->reader
->read();
446 if ( $this->reader
->name
!= 'mediawiki' ) {
447 throw new MWException( "Expected <mediawiki> tag, got ".
448 $this->reader
->name
);
450 $this->debug( "<mediawiki> tag is correct." );
452 $this->debug( "Starting primary dump processing loop." );
454 $keepReading = $this->reader
->read();
456 while ( $keepReading ) {
457 $tag = $this->reader
->name
;
458 $type = $this->reader
->nodeType
;
460 if ( !wfRunHooks( 'ImportHandleToplevelXMLTag', $this ) ) {
462 } elseif ( $tag == 'mediawiki' && $type == XmlReader
::END_ELEMENT
) {
464 } elseif ( $tag == 'siteinfo' ) {
465 $this->handleSiteInfo();
466 } elseif ( $tag == 'page' ) {
468 } elseif ( $tag == 'logitem' ) {
469 $this->handleLogItem();
470 } elseif ( $tag != '#text' ) {
471 $this->warn( "Unhandled top-level XML tag $tag" );
477 $keepReading = $this->reader
->next();
479 $this->debug( "Skip" );
481 $keepReading = $this->reader
->read();
490 * @throws MWException
492 private function handleSiteInfo() {
493 // Site info is useful, but not actually used for dump imports.
494 // Includes a quick short-circuit to save performance.
495 if ( ! $this->mSiteInfoCallback
) {
496 $this->reader
->next();
499 throw new MWException( "SiteInfo tag is not yet handled, do not set mSiteInfoCallback" );
502 private function handleLogItem() {
503 $this->debug( "Enter log item handler." );
506 // Fields that can just be stuffed in the pageInfo object
507 $normalFields = array( 'id', 'comment', 'type', 'action', 'timestamp',
508 'logtitle', 'params' );
510 while ( $this->reader
->read() ) {
511 if ( $this->reader
->nodeType
== XmlReader
::END_ELEMENT
&&
512 $this->reader
->name
== 'logitem' ) {
516 $tag = $this->reader
->name
;
518 if ( !wfRunHooks( 'ImportHandleLogItemXMLTag',
519 $this, $logInfo ) ) {
521 } elseif ( in_array( $tag, $normalFields ) ) {
522 $logInfo[$tag] = $this->nodeContents();
523 } elseif ( $tag == 'contributor' ) {
524 $logInfo['contributor'] = $this->handleContributor();
525 } elseif ( $tag != '#text' ) {
526 $this->warn( "Unhandled log-item XML tag $tag" );
530 $this->processLogItem( $logInfo );
537 private function processLogItem( $logInfo ) {
538 $revision = new WikiRevision
;
540 $revision->setID( $logInfo['id'] );
541 $revision->setType( $logInfo['type'] );
542 $revision->setAction( $logInfo['action'] );
543 $revision->setTimestamp( $logInfo['timestamp'] );
544 $revision->setParams( $logInfo['params'] );
545 $revision->setTitle( Title
::newFromText( $logInfo['logtitle'] ) );
546 $revision->setNoUpdates( $this->mNoUpdates
);
548 if ( isset( $logInfo['comment'] ) ) {
549 $revision->setComment( $logInfo['comment'] );
552 if ( isset( $logInfo['contributor']['ip'] ) ) {
553 $revision->setUserIP( $logInfo['contributor']['ip'] );
555 if ( isset( $logInfo['contributor']['username'] ) ) {
556 $revision->setUserName( $logInfo['contributor']['username'] );
559 return $this->logItemCallback( $revision );
562 private function handlePage() {
564 $this->debug( "Enter page handler." );
565 $pageInfo = array( 'revisionCount' => 0, 'successfulRevisionCount' => 0 );
567 // Fields that can just be stuffed in the pageInfo object
568 $normalFields = array( 'title', 'id', 'redirect', 'restrictions' );
573 while ( $skip ?
$this->reader
->next() : $this->reader
->read() ) {
574 if ( $this->reader
->nodeType
== XmlReader
::END_ELEMENT
&&
575 $this->reader
->name
== 'page' ) {
579 $tag = $this->reader
->name
;
582 // The title is invalid, bail out of this page
584 } elseif ( !wfRunHooks( 'ImportHandlePageXMLTag', array( $this,
587 } elseif ( in_array( $tag, $normalFields ) ) {
588 $pageInfo[$tag] = $this->nodeContents();
589 if ( $tag == 'title' ) {
590 $title = $this->processTitle( $pageInfo['title'] );
597 $this->pageCallback( $title );
598 list( $pageInfo['_title'], $origTitle ) = $title;
600 } elseif ( $tag == 'revision' ) {
601 $this->handleRevision( $pageInfo );
602 } elseif ( $tag == 'upload' ) {
603 $this->handleUpload( $pageInfo );
604 } elseif ( $tag != '#text' ) {
605 $this->warn( "Unhandled page XML tag $tag" );
610 $this->pageOutCallback( $pageInfo['_title'], $origTitle,
611 $pageInfo['revisionCount'],
612 $pageInfo['successfulRevisionCount'],
617 * @param $pageInfo array
619 private function handleRevision( &$pageInfo ) {
620 $this->debug( "Enter revision handler" );
621 $revisionInfo = array();
623 $normalFields = array( 'id', 'timestamp', 'comment', 'minor', 'model', 'format', 'text' );
627 while ( $skip ?
$this->reader
->next() : $this->reader
->read() ) {
628 if ( $this->reader
->nodeType
== XmlReader
::END_ELEMENT
&&
629 $this->reader
->name
== 'revision' ) {
633 $tag = $this->reader
->name
;
635 if ( !wfRunHooks( 'ImportHandleRevisionXMLTag', $this,
636 $pageInfo, $revisionInfo ) ) {
638 } elseif ( in_array( $tag, $normalFields ) ) {
639 $revisionInfo[$tag] = $this->nodeContents();
640 } elseif ( $tag == 'contributor' ) {
641 $revisionInfo['contributor'] = $this->handleContributor();
642 } elseif ( $tag != '#text' ) {
643 $this->warn( "Unhandled revision XML tag $tag" );
648 $pageInfo['revisionCount']++
;
649 if ( $this->processRevision( $pageInfo, $revisionInfo ) ) {
650 $pageInfo['successfulRevisionCount']++
;
656 * @param $revisionInfo
659 private function processRevision( $pageInfo, $revisionInfo ) {
660 $revision = new WikiRevision
;
662 if( isset( $revisionInfo['id'] ) ) {
663 $revision->setID( $revisionInfo['id'] );
665 if ( isset( $revisionInfo['text'] ) ) {
666 $revision->setText( $revisionInfo['text'] );
668 if ( isset( $revisionInfo['model'] ) ) {
669 $revision->setModel( $revisionInfo['model'] );
671 if ( isset( $revisionInfo['format'] ) ) {
672 $revision->setFormat( $revisionInfo['format'] );
674 $revision->setTitle( $pageInfo['_title'] );
676 if ( isset( $revisionInfo['timestamp'] ) ) {
677 $revision->setTimestamp( $revisionInfo['timestamp'] );
679 $revision->setTimestamp( wfTimestampNow() );
682 if ( isset( $revisionInfo['comment'] ) ) {
683 $revision->setComment( $revisionInfo['comment'] );
686 if ( isset( $revisionInfo['minor'] ) ) {
687 $revision->setMinor( true );
689 if ( isset( $revisionInfo['contributor']['ip'] ) ) {
690 $revision->setUserIP( $revisionInfo['contributor']['ip'] );
692 if ( isset( $revisionInfo['contributor']['username'] ) ) {
693 $revision->setUserName( $revisionInfo['contributor']['username'] );
695 $revision->setNoUpdates( $this->mNoUpdates
);
697 return $this->revisionCallback( $revision );
704 private function handleUpload( &$pageInfo ) {
705 $this->debug( "Enter upload handler" );
706 $uploadInfo = array();
708 $normalFields = array( 'timestamp', 'comment', 'filename', 'text',
709 'src', 'size', 'sha1base36', 'archivename', 'rel' );
713 while ( $skip ?
$this->reader
->next() : $this->reader
->read() ) {
714 if ( $this->reader
->nodeType
== XmlReader
::END_ELEMENT
&&
715 $this->reader
->name
== 'upload' ) {
719 $tag = $this->reader
->name
;
721 if ( !wfRunHooks( 'ImportHandleUploadXMLTag', $this,
724 } elseif ( in_array( $tag, $normalFields ) ) {
725 $uploadInfo[$tag] = $this->nodeContents();
726 } elseif ( $tag == 'contributor' ) {
727 $uploadInfo['contributor'] = $this->handleContributor();
728 } elseif ( $tag == 'contents' ) {
729 $contents = $this->nodeContents();
730 $encoding = $this->reader
->getAttribute( 'encoding' );
731 if ( $encoding === 'base64' ) {
732 $uploadInfo['fileSrc'] = $this->dumpTemp( base64_decode( $contents ) );
733 $uploadInfo['isTempSrc'] = true;
735 } elseif ( $tag != '#text' ) {
736 $this->warn( "Unhandled upload XML tag $tag" );
741 if ( $this->mImageBasePath
&& isset( $uploadInfo['rel'] ) ) {
742 $path = "{$this->mImageBasePath}/{$uploadInfo['rel']}";
743 if ( file_exists( $path ) ) {
744 $uploadInfo['fileSrc'] = $path;
745 $uploadInfo['isTempSrc'] = false;
749 if ( $this->mImportUploads
) {
750 return $this->processUpload( $pageInfo, $uploadInfo );
758 private function dumpTemp( $contents ) {
759 $filename = tempnam( wfTempDir(), 'importupload' );
760 file_put_contents( $filename, $contents );
769 private function processUpload( $pageInfo, $uploadInfo ) {
770 $revision = new WikiRevision
;
771 $text = isset( $uploadInfo['text'] ) ?
$uploadInfo['text'] : '';
773 $revision->setTitle( $pageInfo['_title'] );
774 $revision->setID( $pageInfo['id'] );
775 $revision->setTimestamp( $uploadInfo['timestamp'] );
776 $revision->setText( $text );
777 $revision->setFilename( $uploadInfo['filename'] );
778 if ( isset( $uploadInfo['archivename'] ) ) {
779 $revision->setArchiveName( $uploadInfo['archivename'] );
781 $revision->setSrc( $uploadInfo['src'] );
782 if ( isset( $uploadInfo['fileSrc'] ) ) {
783 $revision->setFileSrc( $uploadInfo['fileSrc'],
784 !empty( $uploadInfo['isTempSrc'] ) );
786 if ( isset( $uploadInfo['sha1base36'] ) ) {
787 $revision->setSha1Base36( $uploadInfo['sha1base36'] );
789 $revision->setSize( intval( $uploadInfo['size'] ) );
790 $revision->setComment( $uploadInfo['comment'] );
792 if ( isset( $uploadInfo['contributor']['ip'] ) ) {
793 $revision->setUserIP( $uploadInfo['contributor']['ip'] );
795 if ( isset( $uploadInfo['contributor']['username'] ) ) {
796 $revision->setUserName( $uploadInfo['contributor']['username'] );
798 $revision->setNoUpdates( $this->mNoUpdates
);
800 return call_user_func( $this->mUploadCallback
, $revision );
806 private function handleContributor() {
807 $fields = array( 'id', 'ip', 'username' );
810 while ( $this->reader
->read() ) {
811 if ( $this->reader
->nodeType
== XmlReader
::END_ELEMENT
&&
812 $this->reader
->name
== 'contributor' ) {
816 $tag = $this->reader
->name
;
818 if ( in_array( $tag, $fields ) ) {
819 $info[$tag] = $this->nodeContents();
827 * @param $text string
828 * @return Array or false
830 private function processTitle( $text ) {
831 global $wgCommandLineMode;
834 $origTitle = Title
::newFromText( $workTitle );
836 if( !is_null( $this->mTargetNamespace
) && !is_null( $origTitle ) ) {
837 # makeTitleSafe, because $origTitle can have a interwiki (different setting of interwiki map)
838 # and than dbKey can begin with a lowercase char
839 $title = Title
::makeTitleSafe( $this->mTargetNamespace
,
840 $origTitle->getDBkey() );
842 if( !is_null( $this->mTargetRootPage
) ) {
843 $workTitle = $this->mTargetRootPage
. '/' . $workTitle;
845 $title = Title
::newFromText( $workTitle );
848 if( is_null( $title ) ) {
849 # Invalid page title? Ignore the page
850 $this->notice( 'import-error-invalid', $workTitle );
852 } elseif( $title->isExternal() ) {
853 $this->notice( 'import-error-interwiki', $title->getPrefixedText() );
855 } elseif( !$title->canExist() ) {
856 $this->notice( 'import-error-special', $title->getPrefixedText() );
858 } elseif( !$title->userCan( 'edit' ) && !$wgCommandLineMode ) {
859 # Do not import if the importing wiki user cannot edit this page
860 $this->notice( 'import-error-edit', $title->getPrefixedText() );
862 } elseif( !$title->exists() && !$title->userCan( 'create' ) && !$wgCommandLineMode ) {
863 # Do not import if the importing wiki user cannot create this page
864 $this->notice( 'import-error-create', $title->getPrefixedText() );
868 return array( $title, $origTitle );
872 /** This is a horrible hack used to keep source compatibility */
873 class UploadSourceAdapter
{
874 static $sourceRegistrations = array();
884 static function registerSource( $source ) {
885 $id = wfRandomString();
887 self
::$sourceRegistrations[$id] = $source;
896 * @param $opened_path
899 function stream_open( $path, $mode, $options, &$opened_path ) {
900 $url = parse_url( $path );
903 if ( !isset( self
::$sourceRegistrations[$id] ) ) {
907 $this->mSource
= self
::$sourceRegistrations[$id];
916 function stream_read( $count ) {
920 while ( !$leave && !$this->mSource
->atEnd() &&
921 strlen( $this->mBuffer
) < $count ) {
922 $read = $this->mSource
->readChunk();
924 if ( !strlen( $read ) ) {
928 $this->mBuffer
.= $read;
931 if ( strlen( $this->mBuffer
) ) {
932 $return = substr( $this->mBuffer
, 0, $count );
933 $this->mBuffer
= substr( $this->mBuffer
, $count );
936 $this->mPosition +
= strlen( $return );
945 function stream_write( $data ) {
952 function stream_tell() {
953 return $this->mPosition
;
959 function stream_eof() {
960 return $this->mSource
->atEnd();
966 function url_stat() {
969 $result['dev'] = $result[0] = 0;
970 $result['ino'] = $result[1] = 0;
971 $result['mode'] = $result[2] = 0;
972 $result['nlink'] = $result[3] = 0;
973 $result['uid'] = $result[4] = 0;
974 $result['gid'] = $result[5] = 0;
975 $result['rdev'] = $result[6] = 0;
976 $result['size'] = $result[7] = 0;
977 $result['atime'] = $result[8] = 0;
978 $result['mtime'] = $result[9] = 0;
979 $result['ctime'] = $result[10] = 0;
980 $result['blksize'] = $result[11] = 0;
981 $result['blocks'] = $result[12] = 0;
987 class XMLReader2
extends XMLReader
{
990 * @return bool|string
992 function nodeContents() {
993 if( $this->isEmptyElement
) {
997 while( $this->read() ) {
998 switch( $this->nodeType
) {
999 case XmlReader
::TEXT
:
1000 case XmlReader
::SIGNIFICANT_WHITESPACE
:
1001 $buffer .= $this->value
;
1003 case XmlReader
::END_ELEMENT
:
1007 return $this->close();
1012 * @todo document (e.g. one-sentence class description).
1013 * @ingroup SpecialPage
1015 class WikiRevision
{
1016 var $importer = null;
1023 var $timestamp = "20010115000000";
1025 var $user_text = "";
1029 var $content = null;
1036 var $sha1base36 = false;
1037 var $isTemp = false;
1038 var $archiveName = '';
1040 private $mNoUpdates = false;
1044 * @throws MWException
1046 function setTitle( $title ) {
1047 if( is_object( $title ) ) {
1048 $this->title
= $title;
1049 } elseif( is_null( $title ) ) {
1050 throw new MWException( "WikiRevision given a null title in import. You may need to adjust \$wgLegalTitleChars." );
1052 throw new MWException( "WikiRevision given non-object title in import." );
1059 function setID( $id ) {
1066 function setTimestamp( $ts ) {
1067 # 2003-08-05T18:30:02Z
1068 $this->timestamp
= wfTimestamp( TS_MW
, $ts );
1074 function setUsername( $user ) {
1075 $this->user_text
= $user;
1081 function setUserIP( $ip ) {
1082 $this->user_text
= $ip;
1088 function setModel( $model ) {
1089 $this->model
= $model;
1095 function setFormat( $format ) {
1096 $this->format
= $format;
1102 function setText( $text ) {
1103 $this->text
= $text;
1109 function setComment( $text ) {
1110 $this->comment
= $text;
1116 function setMinor( $minor ) {
1117 $this->minor
= (bool)$minor;
1123 function setSrc( $src ) {
1131 function setFileSrc( $src, $isTemp ) {
1132 $this->fileSrc
= $src;
1133 $this->fileIsTemp
= $isTemp;
1137 * @param $sha1base36
1139 function setSha1Base36( $sha1base36 ) {
1140 $this->sha1base36
= $sha1base36;
1146 function setFilename( $filename ) {
1147 $this->filename
= $filename;
1151 * @param $archiveName
1153 function setArchiveName( $archiveName ) {
1154 $this->archiveName
= $archiveName;
1160 function setSize( $size ) {
1161 $this->size
= intval( $size );
1167 function setType( $type ) {
1168 $this->type
= $type;
1174 function setAction( $action ) {
1175 $this->action
= $action;
1181 function setParams( $params ) {
1182 $this->params
= $params;
1188 public function setNoUpdates( $noupdates ) {
1189 $this->mNoUpdates
= $noupdates;
1195 function getTitle() {
1196 return $this->title
;
1209 function getTimestamp() {
1210 return $this->timestamp
;
1216 function getUser() {
1217 return $this->user_text
;
1223 * @deprecated Since 1.21, use getContent() instead.
1225 function getText() {
1226 ContentHandler
::deprecated( __METHOD__
, '1.21' );
1234 function getContent() {
1235 if ( is_null( $this->content
) ) {
1237 ContentHandler
::makeContent(
1245 return $this->content
;
1251 function getModel() {
1252 if ( is_null( $this->model
) ) {
1253 $this->model
= $this->getTitle()->getContentModel();
1256 return $this->model
;
1262 function getFormat() {
1263 if ( is_null( $this->model
) ) {
1264 $this->format
= ContentHandler
::getForTitle( $this->getTitle() )->getDefaultFormat();
1267 return $this->format
;
1273 function getComment() {
1274 return $this->comment
;
1280 function getMinor() {
1281 return $this->minor
;
1292 * @return bool|String
1294 function getSha1() {
1295 if ( $this->sha1base36
) {
1296 return wfBaseConvert( $this->sha1base36
, 36, 16 );
1304 function getFileSrc() {
1305 return $this->fileSrc
;
1311 function isTempSrc() {
1312 return $this->isTemp
;
1318 function getFilename() {
1319 return $this->filename
;
1325 function getArchiveName() {
1326 return $this->archiveName
;
1332 function getSize() {
1339 function getType() {
1346 function getAction() {
1347 return $this->action
;
1353 function getParams() {
1354 return $this->params
;
1360 function importOldRevision() {
1361 $dbw = wfGetDB( DB_MASTER
);
1363 # Sneak a single revision into place
1364 $user = User
::newFromName( $this->getUser() );
1366 $userId = intval( $user->getId() );
1367 $userText = $user->getName();
1371 $userText = $this->getUser();
1372 $userObj = new User
;
1375 // avoid memory leak...?
1376 $linkCache = LinkCache
::singleton();
1377 $linkCache->clear();
1379 $page = WikiPage
::factory( $this->title
);
1380 if( !$page->exists() ) {
1381 # must create the page...
1382 $pageId = $page->insertOn( $dbw );
1384 $oldcountable = null;
1386 $pageId = $page->getId();
1389 $prior = $dbw->selectField( 'revision', '1',
1390 array( 'rev_page' => $pageId,
1391 'rev_timestamp' => $dbw->timestamp( $this->timestamp
),
1392 'rev_user_text' => $userText,
1393 'rev_comment' => $this->getComment() ),
1397 // @todo FIXME: This could fail slightly for multiple matches :P
1398 wfDebug( __METHOD__
. ": skipping existing revision for [[" .
1399 $this->title
->getPrefixedText() . "]], timestamp " . $this->timestamp
. "\n" );
1402 $oldcountable = $page->isCountable();
1405 # @todo FIXME: Use original rev_id optionally (better for backups)
1407 $revision = new Revision( array(
1408 'title' => $this->title
,
1410 'content_model' => $this->getModel(),
1411 'content_format' => $this->getFormat(),
1412 'text' => $this->getContent()->serialize( $this->getFormat() ), //XXX: just set 'content' => $this->getContent()?
1413 'comment' => $this->getComment(),
1415 'user_text' => $userText,
1416 'timestamp' => $this->timestamp
,
1417 'minor_edit' => $this->minor
,
1419 $revision->insertOn( $dbw );
1420 $changed = $page->updateIfNewerOn( $dbw, $revision );
1422 if ( $changed !== false && !$this->mNoUpdates
) {
1423 wfDebug( __METHOD__
. ": running updates\n" );
1424 $page->doEditUpdates( $revision, $userObj, array( 'created' => $created, 'oldcountable' => $oldcountable ) );
1433 function importLogItem() {
1434 $dbw = wfGetDB( DB_MASTER
);
1435 # @todo FIXME: This will not record autoblocks
1436 if( !$this->getTitle() ) {
1437 wfDebug( __METHOD__
. ": skipping invalid {$this->type}/{$this->action} log time, timestamp " .
1438 $this->timestamp
. "\n" );
1441 # Check if it exists already
1442 // @todo FIXME: Use original log ID (better for backups)
1443 $prior = $dbw->selectField( 'logging', '1',
1444 array( 'log_type' => $this->getType(),
1445 'log_action' => $this->getAction(),
1446 'log_timestamp' => $dbw->timestamp( $this->timestamp
),
1447 'log_namespace' => $this->getTitle()->getNamespace(),
1448 'log_title' => $this->getTitle()->getDBkey(),
1449 'log_comment' => $this->getComment(),
1450 #'log_user_text' => $this->user_text,
1451 'log_params' => $this->params
),
1454 // @todo FIXME: This could fail slightly for multiple matches :P
1456 wfDebug( __METHOD__
. ": skipping existing item for Log:{$this->type}/{$this->action}, timestamp " .
1457 $this->timestamp
. "\n" );
1460 $log_id = $dbw->nextSequenceValue( 'logging_log_id_seq' );
1462 'log_id' => $log_id,
1463 'log_type' => $this->type
,
1464 'log_action' => $this->action
,
1465 'log_timestamp' => $dbw->timestamp( $this->timestamp
),
1466 'log_user' => User
::idFromName( $this->user_text
),
1467 #'log_user_text' => $this->user_text,
1468 'log_namespace' => $this->getTitle()->getNamespace(),
1469 'log_title' => $this->getTitle()->getDBkey(),
1470 'log_comment' => $this->getComment(),
1471 'log_params' => $this->params
1473 $dbw->insert( 'logging', $data, __METHOD__
);
1479 function importUpload() {
1481 $archiveName = $this->getArchiveName();
1482 if ( $archiveName ) {
1483 wfDebug( __METHOD__
. "Importing archived file as $archiveName\n" );
1484 $file = OldLocalFile
::newFromArchiveName( $this->getTitle(),
1485 RepoGroup
::singleton()->getLocalRepo(), $archiveName );
1487 $file = wfLocalFile( $this->getTitle() );
1488 wfDebug( __METHOD__
. 'Importing new file as ' . $file->getName() . "\n" );
1489 if ( $file->exists() && $file->getTimestamp() > $this->getTimestamp() ) {
1490 $archiveName = $file->getTimestamp() . '!' . $file->getName();
1491 $file = OldLocalFile
::newFromArchiveName( $this->getTitle(),
1492 RepoGroup
::singleton()->getLocalRepo(), $archiveName );
1493 wfDebug( __METHOD__
. "File already exists; importing as $archiveName\n" );
1497 wfDebug( __METHOD__
. ': Bad file for ' . $this->getTitle() . "\n" );
1501 # Get the file source or download if necessary
1502 $source = $this->getFileSrc();
1503 $flags = $this->isTempSrc() ? File
::DELETE_SOURCE
: 0;
1505 $source = $this->downloadSource();
1506 $flags |
= File
::DELETE_SOURCE
;
1509 wfDebug( __METHOD__
. ": Could not fetch remote file.\n" );
1512 $sha1 = $this->getSha1();
1513 if ( $sha1 && ( $sha1 !== sha1_file( $source ) ) ) {
1514 if ( $flags & File
::DELETE_SOURCE
) {
1515 # Broken file; delete it if it is a temporary file
1518 wfDebug( __METHOD__
. ": Corrupt file $source.\n" );
1522 $user = User
::newFromName( $this->user_text
);
1524 # Do the actual upload
1525 if ( $archiveName ) {
1526 $status = $file->uploadOld( $source, $archiveName,
1527 $this->getTimestamp(), $this->getComment(), $user, $flags );
1529 $status = $file->upload( $source, $this->getComment(), $this->getComment(),
1530 $flags, false, $this->getTimestamp(), $user );
1533 if ( $status->isGood() ) {
1534 wfDebug( __METHOD__
. ": Successful\n" );
1537 wfDebug( __METHOD__
. ': failed: ' . $status->getXml() . "\n" );
1543 * @return bool|string
1545 function downloadSource() {
1546 global $wgEnableUploads;
1547 if( !$wgEnableUploads ) {
1551 $tempo = tempnam( wfTempDir(), 'download' );
1552 $f = fopen( $tempo, 'wb' );
1554 wfDebug( "IMPORT: couldn't write to temp file $tempo\n" );
1559 $src = $this->getSrc();
1560 $data = Http
::get( $src );
1562 wfDebug( "IMPORT: couldn't fetch source $src\n" );
1568 fwrite( $f, $data );
1577 * @todo document (e.g. one-sentence class description).
1578 * @ingroup SpecialPage
1580 class ImportStringSource
{
1581 function __construct( $string ) {
1582 $this->mString
= $string;
1583 $this->mRead
= false;
1590 return $this->mRead
;
1594 * @return bool|string
1596 function readChunk() {
1597 if( $this->atEnd() ) {
1600 $this->mRead
= true;
1601 return $this->mString
;
1606 * @todo document (e.g. one-sentence class description).
1607 * @ingroup SpecialPage
1609 class ImportStreamSource
{
1610 function __construct( $handle ) {
1611 $this->mHandle
= $handle;
1618 return feof( $this->mHandle
);
1624 function readChunk() {
1625 return fread( $this->mHandle
, 32768 );
1629 * @param $filename string
1632 static function newFromFile( $filename ) {
1633 wfSuppressWarnings();
1634 $file = fopen( $filename, 'rt' );
1635 wfRestoreWarnings();
1637 return Status
::newFatal( "importcantopen" );
1639 return Status
::newGood( new ImportStreamSource( $file ) );
1643 * @param $fieldname string
1646 static function newFromUpload( $fieldname = "xmlimport" ) {
1647 $upload =& $_FILES[$fieldname];
1649 if( $upload === null ||
!$upload['name'] ) {
1650 return Status
::newFatal( 'importnofile' );
1652 if( !empty( $upload['error'] ) ) {
1653 switch( $upload['error'] ) {
1654 case 1: # The uploaded file exceeds the upload_max_filesize directive in php.ini.
1655 return Status
::newFatal( 'importuploaderrorsize' );
1656 case 2: # The uploaded file exceeds the MAX_FILE_SIZE directive that was specified in the HTML form.
1657 return Status
::newFatal( 'importuploaderrorsize' );
1658 case 3: # The uploaded file was only partially uploaded
1659 return Status
::newFatal( 'importuploaderrorpartial' );
1660 case 6: #Missing a temporary folder.
1661 return Status
::newFatal( 'importuploaderrortemp' );
1662 # case else: # Currently impossible
1666 $fname = $upload['tmp_name'];
1667 if( is_uploaded_file( $fname ) ) {
1668 return ImportStreamSource
::newFromFile( $fname );
1670 return Status
::newFatal( 'importnofile' );
1676 * @param $method string
1679 static function newFromURL( $url, $method = 'GET' ) {
1680 wfDebug( __METHOD__
. ": opening $url\n" );
1681 # Use the standard HTTP fetch function; it times out
1682 # quicker and sorts out user-agent problems which might
1683 # otherwise prevent importing from large sites, such
1684 # as the Wikimedia cluster, etc.
1685 $data = Http
::request( $method, $url, array( 'followRedirects' => true ) );
1686 if( $data !== false ) {
1688 fwrite( $file, $data );
1691 return Status
::newGood( new ImportStreamSource( $file ) );
1693 return Status
::newFatal( 'importcantopen' );
1700 * @param $history bool
1701 * @param $templates bool
1702 * @param $pageLinkDepth int
1705 public static function newFromInterwiki( $interwiki, $page, $history = false, $templates = false, $pageLinkDepth = 0 ) {
1707 return Status
::newFatal( 'import-noarticle' );
1709 $link = Title
::newFromText( "$interwiki:Special:Export/$page" );
1710 if( is_null( $link ) ||
$link->getInterwiki() == '' ) {
1711 return Status
::newFatal( 'importbadinterwiki' );
1714 if ( $history ) $params['history'] = 1;
1715 if ( $templates ) $params['templates'] = 1;
1716 if ( $pageLinkDepth ) $params['pagelink-depth'] = $pageLinkDepth;
1717 $url = $link->getFullURL( $params );
1718 # For interwikis, use POST to avoid redirects.
1719 return ImportStreamSource
::newFromURL( $url, "POST" );