3 * Base classes for dumps and export
5 * Copyright © 2003, 2005, 2006 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
31 * @ingroup SpecialPage Dump
34 /** @var bool Return distinct author list (when not returning full history) */
35 public $list_authors = false;
38 public $dumpUploads = false;
41 public $dumpUploadFileContents = false;
44 public $author_list = "";
48 const STABLE
= 4; // extension defined
64 /** @var DumpOutput */
68 * Returns the export schema version.
71 public static function schemaVersion() {
76 * If using WikiExporter::STREAM to stream a large amount of data,
77 * provide a database connection which is not managed by
78 * LoadBalancer to read from: some history blob types will
79 * make additional queries to pull source data while the
80 * main query is still running.
82 * @param DatabaseBase $db
83 * @param int|array $history One of WikiExporter::FULL, WikiExporter::CURRENT,
84 * WikiExporter::RANGE or WikiExporter::STABLE, or an associative array:
85 * - offset: non-inclusive offset at which to start the query
86 * - limit: maximum number of rows to return
87 * - dir: "asc" or "desc" timestamp order
88 * @param int $buffer One of WikiExporter::BUFFER or WikiExporter::STREAM
89 * @param int $text One of WikiExporter::TEXT or WikiExporter::STUB
91 function __construct( $db, $history = WikiExporter
::CURRENT
,
92 $buffer = WikiExporter
::BUFFER
, $text = WikiExporter
::TEXT
) {
94 $this->history
= $history;
95 $this->buffer
= $buffer;
96 $this->writer
= new XmlDumpWriter();
97 $this->sink
= new DumpOutput();
102 * Set the DumpOutput or DumpFilter object which will receive
103 * various row objects and XML output for filtering. Filters
104 * can be chained or used as callbacks.
106 * @param DumpOutput $sink
108 public function setOutputSink( &$sink ) {
109 $this->sink
=& $sink;
112 public function openStream() {
113 $output = $this->writer
->openStream();
114 $this->sink
->writeOpenStream( $output );
117 public function closeStream() {
118 $output = $this->writer
->closeStream();
119 $this->sink
->writeCloseStream( $output );
123 * Dumps a series of page and revision records for all pages
124 * in the database, either including complete history or only
125 * the most recent version.
127 public function allPages() {
128 $this->dumpFrom( '' );
132 * Dumps a series of page and revision records for those pages
133 * in the database falling within the page_id range given.
134 * @param int $start Inclusive lower limit (this id is included)
135 * @param int $end Exclusive upper limit (this id is not included)
136 * If 0, no upper limit.
138 public function pagesByRange( $start, $end ) {
139 $condition = 'page_id >= ' . intval( $start );
141 $condition .= ' AND page_id < ' . intval( $end );
143 $this->dumpFrom( $condition );
147 * Dumps a series of page and revision records for those pages
148 * in the database with revisions falling within the rev_id range given.
149 * @param int $start Inclusive lower limit (this id is included)
150 * @param int $end Exclusive upper limit (this id is not included)
151 * If 0, no upper limit.
153 public function revsByRange( $start, $end ) {
154 $condition = 'rev_id >= ' . intval( $start );
156 $condition .= ' AND rev_id < ' . intval( $end );
158 $this->dumpFrom( $condition );
162 * @param Title $title
164 public function pageByTitle( $title ) {
166 'page_namespace=' . $title->getNamespace() .
167 ' AND page_title=' . $this->db
->addQuotes( $title->getDBkey() ) );
171 * @param string $name
172 * @throws MWException
174 public function pageByName( $name ) {
175 $title = Title
::newFromText( $name );
176 if ( is_null( $title ) ) {
177 throw new MWException( "Can't export invalid title" );
179 $this->pageByTitle( $title );
184 * @param array $names
186 public function pagesByName( $names ) {
187 foreach ( $names as $name ) {
188 $this->pageByName( $name );
192 public function allLogs() {
193 $this->dumpFrom( '' );
200 public function logsByRange( $start, $end ) {
201 $condition = 'log_id >= ' . intval( $start );
203 $condition .= ' AND log_id < ' . intval( $end );
205 $this->dumpFrom( $condition );
209 * Generates the distinct list of authors of an article
210 * Not called by default (depends on $this->list_authors)
211 * Can be set by Special:Export when not exporting whole history
215 protected function do_list_authors( $cond ) {
216 wfProfileIn( __METHOD__
);
217 $this->author_list
= "<contributors>";
220 $res = $this->db
->select(
221 array( 'page', 'revision' ),
222 array( 'DISTINCT rev_user_text', 'rev_user' ),
224 $this->db
->bitAnd( 'rev_deleted', Revision
::DELETED_USER
) . ' = 0',
231 foreach ( $res as $row ) {
232 $this->author_list
.= "<contributor>" .
234 htmlentities( $row->rev_user_text
) .
241 $this->author_list
.= "</contributors>";
242 wfProfileOut( __METHOD__
);
246 * @param string $cond
247 * @throws MWException
250 protected function dumpFrom( $cond = '' ) {
251 wfProfileIn( __METHOD__
);
252 # For logging dumps...
253 if ( $this->history
& self
::LOGS
) {
254 $where = array( 'user_id = log_user' );
256 $hideLogs = LogEventsList
::getExcludeClause( $this->db
);
258 $where[] = $hideLogs;
260 # Add on any caller specified conditions
264 # Get logging table name for logging.* clause
265 $logging = $this->db
->tableName( 'logging' );
267 if ( $this->buffer
== WikiExporter
::STREAM
) {
268 $prev = $this->db
->bufferResults( false );
270 $result = null; // Assuring $result is not undefined, if exception occurs early
272 $result = $this->db
->select( array( 'logging', 'user' ),
273 array( "{$logging}.*", 'user_name' ), // grab the user name
276 array( 'ORDER BY' => 'log_id', 'USE INDEX' => array( 'logging' => 'PRIMARY' ) )
278 $this->outputLogStream( $result );
279 if ( $this->buffer
== WikiExporter
::STREAM
) {
280 $this->db
->bufferResults( $prev );
282 } catch ( Exception
$e ) {
283 // Throwing the exception does not reliably free the resultset, and
284 // would also leave the connection in unbuffered mode.
291 } catch ( Exception
$e2 ) {
292 // Already in panic mode -> ignoring $e2 as $e has
296 // Putting database back in previous buffer mode
298 if ( $this->buffer
== WikiExporter
::STREAM
) {
299 $this->db
->bufferResults( $prev );
301 } catch ( Exception
$e2 ) {
302 // Already in panic mode -> ignoring $e2 as $e has
306 // Inform caller about problem
307 wfProfileOut( __METHOD__
);
312 $tables = array( 'page', 'revision' );
313 $opts = array( 'ORDER BY' => 'page_id ASC' );
314 $opts['USE INDEX'] = array();
316 if ( is_array( $this->history
) ) {
317 # Time offset/limit for all pages/history...
318 $revJoin = 'page_id=rev_page';
320 if ( $this->history
['dir'] == 'asc' ) {
322 $opts['ORDER BY'] = 'rev_timestamp ASC';
325 $opts['ORDER BY'] = 'rev_timestamp DESC';
328 if ( !empty( $this->history
['offset'] ) ) {
329 $revJoin .= " AND rev_timestamp $op " .
330 $this->db
->addQuotes( $this->db
->timestamp( $this->history
['offset'] ) );
332 $join['revision'] = array( 'INNER JOIN', $revJoin );
334 if ( !empty( $this->history
['limit'] ) ) {
335 $opts['LIMIT'] = intval( $this->history
['limit'] );
337 } elseif ( $this->history
& WikiExporter
::FULL
) {
338 # Full history dumps...
339 $join['revision'] = array( 'INNER JOIN', 'page_id=rev_page' );
340 } elseif ( $this->history
& WikiExporter
::CURRENT
) {
341 # Latest revision dumps...
342 if ( $this->list_authors
&& $cond != '' ) { // List authors, if so desired
343 $this->do_list_authors( $cond );
345 $join['revision'] = array( 'INNER JOIN', 'page_id=rev_page AND page_latest=rev_id' );
346 } elseif ( $this->history
& WikiExporter
::STABLE
) {
347 # "Stable" revision dumps...
348 # Default JOIN, to be overridden...
349 $join['revision'] = array( 'INNER JOIN', 'page_id=rev_page AND page_latest=rev_id' );
350 # One, and only one hook should set this, and return false
351 if ( wfRunHooks( 'WikiExporter::dumpStableQuery', array( &$tables, &$opts, &$join ) ) ) {
352 wfProfileOut( __METHOD__
);
353 throw new MWException( __METHOD__
. " given invalid history dump type." );
355 } elseif ( $this->history
& WikiExporter
::RANGE
) {
356 # Dump of revisions within a specified range
357 $join['revision'] = array( 'INNER JOIN', 'page_id=rev_page' );
358 $opts['ORDER BY'] = array( 'rev_page ASC', 'rev_id ASC' );
360 # Unknown history specification parameter?
361 wfProfileOut( __METHOD__
);
362 throw new MWException( __METHOD__
. " given invalid history dump type." );
364 # Query optimization hacks
366 $opts[] = 'STRAIGHT_JOIN';
367 $opts['USE INDEX']['page'] = 'PRIMARY';
369 # Build text join options
370 if ( $this->text
!= WikiExporter
::STUB
) { // 1-pass
372 $join['text'] = array( 'INNER JOIN', 'rev_text_id=old_id' );
375 if ( $this->buffer
== WikiExporter
::STREAM
) {
376 $prev = $this->db
->bufferResults( false );
379 $result = null; // Assuring $result is not undefined, if exception occurs early
381 wfRunHooks( 'ModifyExportQuery',
382 array( $this->db
, &$tables, &$cond, &$opts, &$join ) );
385 $result = $this->db
->select( $tables, '*', $cond, __METHOD__
, $opts, $join );
386 # Output dump results
387 $this->outputPageStream( $result );
389 if ( $this->buffer
== WikiExporter
::STREAM
) {
390 $this->db
->bufferResults( $prev );
392 } catch ( Exception
$e ) {
393 // Throwing the exception does not reliably free the resultset, and
394 // would also leave the connection in unbuffered mode.
401 } catch ( Exception
$e2 ) {
402 // Already in panic mode -> ignoring $e2 as $e has
406 // Putting database back in previous buffer mode
408 if ( $this->buffer
== WikiExporter
::STREAM
) {
409 $this->db
->bufferResults( $prev );
411 } catch ( Exception
$e2 ) {
412 // Already in panic mode -> ignoring $e2 as $e has
416 // Inform caller about problem
420 wfProfileOut( __METHOD__
);
424 * Runs through a query result set dumping page and revision records.
425 * The result set should be sorted/grouped by page to avoid duplicate
426 * page records in the output.
429 * streaming (non-buffered) queries, as long as it was made on a
430 * separate database connection not managed by LoadBalancer; some
431 * blob storage types will make queries to pull source data.
433 * @param ResultWrapper $resultset
435 protected function outputPageStream( $resultset ) {
437 foreach ( $resultset as $row ) {
438 if ( $last === null ||
439 $last->page_namespace
!= $row->page_namespace ||
440 $last->page_title
!= $row->page_title
) {
441 if ( $last !== null ) {
443 if ( $this->dumpUploads
) {
444 $output .= $this->writer
->writeUploads( $last, $this->dumpUploadFileContents
);
446 $output .= $this->writer
->closePage();
447 $this->sink
->writeClosePage( $output );
449 $output = $this->writer
->openPage( $row );
450 $this->sink
->writeOpenPage( $row, $output );
453 $output = $this->writer
->writeRevision( $row );
454 $this->sink
->writeRevision( $row, $output );
456 if ( $last !== null ) {
458 if ( $this->dumpUploads
) {
459 $output .= $this->writer
->writeUploads( $last, $this->dumpUploadFileContents
);
461 $output .= $this->author_list
;
462 $output .= $this->writer
->closePage();
463 $this->sink
->writeClosePage( $output );
468 * @param ResultWrapper $resultset
470 protected function outputLogStream( $resultset ) {
471 foreach ( $resultset as $row ) {
472 $output = $this->writer
->writeLogItem( $row );
473 $this->sink
->writeLogItem( $row, $output );
481 class XmlDumpWriter
{
483 * Opens the XML output stream's root "<mediawiki>" element.
484 * This does not include an xml directive, so is safe to include
485 * as a subelement in a larger XML stream. Namespace and XML Schema
486 * references are included.
488 * Output will be encoded in UTF-8.
492 function openStream() {
493 global $wgLanguageCode;
494 $ver = WikiExporter
::schemaVersion();
495 return Xml
::element( 'mediawiki', array(
496 'xmlns' => "http://www.mediawiki.org/xml/export-$ver/",
497 'xmlns:xsi' => "http://www.w3.org/2001/XMLSchema-instance",
499 * When a new version of the schema is created, it needs staging on mediawiki.org.
500 * This requires a change in the operations/mediawiki-config git repo.
502 * Create a changeset like https://gerrit.wikimedia.org/r/#/c/149643/ in which
503 * you copy in the new xsd file.
505 * After it is reviewed, merged and deployed (sync-docroot), the index.html needs purging.
506 * echo "http://www.mediawiki.org/xml/index.html" | mwscript purgeList.php --wiki=aawiki
508 'xsi:schemaLocation' => "http://www.mediawiki.org/xml/export-$ver/ " .
509 "http://www.mediawiki.org/xml/export-$ver.xsd",
511 'xml:lang' => $wgLanguageCode ),
520 function siteInfo() {
526 $this->caseSetting(),
527 $this->namespaces() );
528 return " <siteinfo>\n " .
529 implode( "\n ", $info ) .
536 function sitename() {
538 return Xml
::element( 'sitename', array(), $wgSitename );
546 return Xml
::element( 'dbname', array(), $wgDBname );
552 function generator() {
554 return Xml
::element( 'generator', array(), "MediaWiki $wgVersion" );
560 function homelink() {
561 return Xml
::element( 'base', array(), Title
::newMainPage()->getCanonicalURL() );
567 function caseSetting() {
568 global $wgCapitalLinks;
569 // "case-insensitive" option is reserved for future
570 $sensitivity = $wgCapitalLinks ?
'first-letter' : 'case-sensitive';
571 return Xml
::element( 'case', array(), $sensitivity );
577 function namespaces() {
579 $spaces = "<namespaces>\n";
580 foreach ( $wgContLang->getFormattedNamespaces() as $ns => $title ) {
582 Xml
::element( 'namespace',
585 'case' => MWNamespace
::isCapitalized( $ns ) ?
'first-letter' : 'case-sensitive',
588 $spaces .= " </namespaces>";
593 * Closes the output stream with the closing root element.
594 * Call when finished dumping things.
598 function closeStream() {
599 return "</mediawiki>\n";
603 * Opens a "<page>" section on the output stream, with data
604 * from the given database row.
609 public function openPage( $row ) {
611 $title = Title
::makeTitle( $row->page_namespace
, $row->page_title
);
612 $out .= ' ' . Xml
::elementClean( 'title', array(), self
::canonicalTitle( $title ) ) . "\n";
613 $out .= ' ' . Xml
::element( 'ns', array(), strval( $row->page_namespace
) ) . "\n";
614 $out .= ' ' . Xml
::element( 'id', array(), strval( $row->page_id
) ) . "\n";
615 if ( $row->page_is_redirect
) {
616 $page = WikiPage
::factory( $title );
617 $redirect = $page->getRedirectTarget();
618 if ( $redirect instanceof Title
&& $redirect->isValidRedirectTarget() ) {
620 $out .= Xml
::element( 'redirect', array( 'title' => self
::canonicalTitle( $redirect ) ) );
625 if ( $row->page_restrictions
!= '' ) {
626 $out .= ' ' . Xml
::element( 'restrictions', array(),
627 strval( $row->page_restrictions
) ) . "\n";
630 wfRunHooks( 'XmlDumpWriterOpenPage', array( $this, &$out, $row, $title ) );
636 * Closes a "<page>" section on the output stream.
641 function closePage() {
646 * Dumps a "<revision>" section on the output stream, with
647 * data filled in from the given database row.
653 function writeRevision( $row ) {
654 wfProfileIn( __METHOD__
);
656 $out = " <revision>\n";
657 $out .= " " . Xml
::element( 'id', null, strval( $row->rev_id
) ) . "\n";
658 if ( isset( $row->rev_parent_id
) && $row->rev_parent_id
) {
659 $out .= " " . Xml
::element( 'parentid', null, strval( $row->rev_parent_id
) ) . "\n";
662 $out .= $this->writeTimestamp( $row->rev_timestamp
);
664 if ( isset( $row->rev_deleted
) && ( $row->rev_deleted
& Revision
::DELETED_USER
) ) {
665 $out .= " " . Xml
::element( 'contributor', array( 'deleted' => 'deleted' ) ) . "\n";
667 $out .= $this->writeContributor( $row->rev_user
, $row->rev_user_text
);
670 if ( isset( $row->rev_minor_edit
) && $row->rev_minor_edit
) {
671 $out .= " <minor/>\n";
673 if ( isset( $row->rev_deleted
) && ( $row->rev_deleted
& Revision
::DELETED_COMMENT
) ) {
674 $out .= " " . Xml
::element( 'comment', array( 'deleted' => 'deleted' ) ) . "\n";
675 } elseif ( $row->rev_comment
!= '' ) {
676 $out .= " " . Xml
::elementClean( 'comment', array(), strval( $row->rev_comment
) ) . "\n";
679 if ( isset( $row->rev_content_model
) && !is_null( $row->rev_content_model
) ) {
680 $content_model = strval( $row->rev_content_model
);
682 // probably using $wgContentHandlerUseDB = false;
683 $title = Title
::makeTitle( $row->page_namespace
, $row->page_title
);
684 $content_model = ContentHandler
::getDefaultModelFor( $title );
687 $content_handler = ContentHandler
::getForModelID( $content_model );
689 if ( isset( $row->rev_content_format
) && !is_null( $row->rev_content_format
) ) {
690 $content_format = strval( $row->rev_content_format
);
692 // probably using $wgContentHandlerUseDB = false;
693 $content_format = $content_handler->getDefaultFormat();
697 if ( isset( $row->rev_deleted
) && ( $row->rev_deleted
& Revision
::DELETED_TEXT
) ) {
698 $out .= " " . Xml
::element( 'text', array( 'deleted' => 'deleted' ) ) . "\n";
699 } elseif ( isset( $row->old_text
) ) {
700 // Raw text from the database may have invalid chars
701 $text = strval( Revision
::getRevisionText( $row ) );
702 $text = $content_handler->exportTransform( $text, $content_format );
703 $out .= " " . Xml
::elementClean( 'text',
704 array( 'xml:space' => 'preserve', 'bytes' => intval( $row->rev_len
) ),
705 strval( $text ) ) . "\n";
708 $out .= " " . Xml
::element( 'text',
709 array( 'id' => $row->rev_text_id
, 'bytes' => intval( $row->rev_len
) ),
713 if ( isset( $row->rev_sha1
)
715 && !( $row->rev_deleted
& Revision
::DELETED_TEXT
)
717 $out .= " " . Xml
::element( 'sha1', null, strval( $row->rev_sha1
) ) . "\n";
719 $out .= " <sha1/>\n";
722 $out .= " " . Xml
::element( 'model', null, strval( $content_model ) ) . "\n";
723 $out .= " " . Xml
::element( 'format', null, strval( $content_format ) ) . "\n";
725 wfRunHooks( 'XmlDumpWriterWriteRevision', array( &$this, &$out, $row, $text ) );
727 $out .= " </revision>\n";
729 wfProfileOut( __METHOD__
);
734 * Dumps a "<logitem>" section on the output stream, with
735 * data filled in from the given database row.
741 function writeLogItem( $row ) {
742 wfProfileIn( __METHOD__
);
744 $out = " <logitem>\n";
745 $out .= " " . Xml
::element( 'id', null, strval( $row->log_id
) ) . "\n";
747 $out .= $this->writeTimestamp( $row->log_timestamp
, " " );
749 if ( $row->log_deleted
& LogPage
::DELETED_USER
) {
750 $out .= " " . Xml
::element( 'contributor', array( 'deleted' => 'deleted' ) ) . "\n";
752 $out .= $this->writeContributor( $row->log_user
, $row->user_name
, " " );
755 if ( $row->log_deleted
& LogPage
::DELETED_COMMENT
) {
756 $out .= " " . Xml
::element( 'comment', array( 'deleted' => 'deleted' ) ) . "\n";
757 } elseif ( $row->log_comment
!= '' ) {
758 $out .= " " . Xml
::elementClean( 'comment', null, strval( $row->log_comment
) ) . "\n";
761 $out .= " " . Xml
::element( 'type', null, strval( $row->log_type
) ) . "\n";
762 $out .= " " . Xml
::element( 'action', null, strval( $row->log_action
) ) . "\n";
764 if ( $row->log_deleted
& LogPage
::DELETED_ACTION
) {
765 $out .= " " . Xml
::element( 'text', array( 'deleted' => 'deleted' ) ) . "\n";
767 $title = Title
::makeTitle( $row->log_namespace
, $row->log_title
);
768 $out .= " " . Xml
::elementClean( 'logtitle', null, self
::canonicalTitle( $title ) ) . "\n";
769 $out .= " " . Xml
::elementClean( 'params',
770 array( 'xml:space' => 'preserve' ),
771 strval( $row->log_params
) ) . "\n";
774 $out .= " </logitem>\n";
776 wfProfileOut( __METHOD__
);
781 * @param string $timestamp
782 * @param string $indent Default to six spaces
785 function writeTimestamp( $timestamp, $indent = " " ) {
786 $ts = wfTimestamp( TS_ISO_8601
, $timestamp );
787 return $indent . Xml
::element( 'timestamp', null, $ts ) . "\n";
792 * @param string $text
793 * @param string $indent Default to six spaces
796 function writeContributor( $id, $text, $indent = " " ) {
797 $out = $indent . "<contributor>\n";
798 if ( $id ||
!IP
::isValid( $text ) ) {
799 $out .= $indent . " " . Xml
::elementClean( 'username', null, strval( $text ) ) . "\n";
800 $out .= $indent . " " . Xml
::element( 'id', null, strval( $id ) ) . "\n";
802 $out .= $indent . " " . Xml
::elementClean( 'ip', null, strval( $text ) ) . "\n";
804 $out .= $indent . "</contributor>\n";
809 * Warning! This data is potentially inconsistent. :(
811 * @param bool $dumpContents
814 function writeUploads( $row, $dumpContents = false ) {
815 if ( $row->page_namespace
== NS_FILE
) {
816 $img = wfLocalFile( $row->page_title
);
817 if ( $img && $img->exists() ) {
819 foreach ( array_reverse( $img->getHistory() ) as $ver ) {
820 $out .= $this->writeUpload( $ver, $dumpContents );
822 $out .= $this->writeUpload( $img, $dumpContents );
831 * @param bool $dumpContents
834 function writeUpload( $file, $dumpContents = false ) {
835 if ( $file->isOld() ) {
837 Xml
::element( 'archivename', null, $file->getArchiveName() ) . "\n";
841 if ( $dumpContents ) {
842 $be = $file->getRepo()->getBackend();
843 # Dump file as base64
844 # Uses only XML-safe characters, so does not need escaping
845 # @todo Too bad this loads the contents into memory (script might swap)
846 $contents = ' <contents encoding="base64">' .
847 chunk_split( base64_encode(
848 $be->getFileContents( array( 'src' => $file->getPath() ) ) ) ) .
853 if ( $file->isDeleted( File
::DELETED_COMMENT
) ) {
854 $comment = Xml
::element( 'comment', array( 'deleted' => 'deleted' ) );
856 $comment = Xml
::elementClean( 'comment', null, $file->getDescription() );
858 return " <upload>\n" .
859 $this->writeTimestamp( $file->getTimestamp() ) .
860 $this->writeContributor( $file->getUser( 'id' ), $file->getUser( 'text' ) ) .
861 " " . $comment . "\n" .
862 " " . Xml
::element( 'filename', null, $file->getName() ) . "\n" .
864 " " . Xml
::element( 'src', null, $file->getCanonicalURL() ) . "\n" .
865 " " . Xml
::element( 'size', null, $file->getSize() ) . "\n" .
866 " " . Xml
::element( 'sha1base36', null, $file->getSha1() ) . "\n" .
867 " " . Xml
::element( 'rel', null, $file->getRel() ) . "\n" .
873 * Return prefixed text form of title, but using the content language's
874 * canonical namespace. This skips any special-casing such as gendered
875 * user namespaces -- which while useful, are not yet listed in the
876 * XML "<siteinfo>" data so are unsafe in export.
878 * @param Title $title
882 public static function canonicalTitle( Title
$title ) {
883 if ( $title->isExternal() ) {
884 return $title->getPrefixedText();
888 $prefix = str_replace( '_', ' ', $wgContLang->getNsText( $title->getNamespace() ) );
890 if ( $prefix !== '' ) {
894 return $prefix . $title->getText();
899 * Base class for output stream; prints to stdout or buffer or wherever.
905 * @param string $string
907 function writeOpenStream( $string ) {
908 $this->write( $string );
912 * @param string $string
914 function writeCloseStream( $string ) {
915 $this->write( $string );
919 * @param object $page
920 * @param string $string
922 function writeOpenPage( $page, $string ) {
923 $this->write( $string );
927 * @param string $string
929 function writeClosePage( $string ) {
930 $this->write( $string );
935 * @param string $string
937 function writeRevision( $rev, $string ) {
938 $this->write( $string );
943 * @param string $string
945 function writeLogItem( $rev, $string ) {
946 $this->write( $string );
950 * Override to write to a different stream type.
951 * @param string $string
954 function write( $string ) {
959 * Close the old file, move it to a specified name,
960 * and reopen new file with the old name. Use this
961 * for writing out a file in multiple pieces
962 * at specified checkpoints (e.g. every n hours).
963 * @param string|array $newname File name. May be a string or an array with one element
965 function closeRenameAndReopen( $newname ) {
969 * Close the old file, and move it to a specified name.
970 * Use this for the last piece of a file written out
971 * at specified checkpoints (e.g. every n hours).
972 * @param string|array $newname File name. May be a string or an array with one element
973 * @param bool $open If true, a new file with the old filename will be opened
974 * again for writing (default: false)
976 function closeAndRename( $newname, $open = false ) {
980 * Returns the name of the file or files which are
981 * being written to, if there are any.
984 function getFilenames() {
990 * Stream outputter to send data to a file.
993 class DumpFileOutput
extends DumpOutput
{
994 protected $handle = false, $filename;
997 * @param string $file
999 function __construct( $file ) {
1000 $this->handle
= fopen( $file, "wt" );
1001 $this->filename
= $file;
1005 * @param string $string
1007 function writeCloseStream( $string ) {
1008 parent
::writeCloseStream( $string );
1009 if ( $this->handle
) {
1010 fclose( $this->handle
);
1011 $this->handle
= false;
1016 * @param string $string
1018 function write( $string ) {
1019 fputs( $this->handle
, $string );
1023 * @param string $newname
1025 function closeRenameAndReopen( $newname ) {
1026 $this->closeAndRename( $newname, true );
1030 * @param string $newname
1031 * @throws MWException
1033 function renameOrException( $newname ) {
1034 if ( !rename( $this->filename
, $newname ) ) {
1035 throw new MWException( __METHOD__
. ": rename of file {$this->filename} to $newname failed\n" );
1040 * @param array $newname
1042 * @throws MWException
1044 function checkRenameArgCount( $newname ) {
1045 if ( is_array( $newname ) ) {
1046 if ( count( $newname ) > 1 ) {
1047 throw new MWException( __METHOD__
. ": passed multiple arguments for rename of single file\n" );
1049 $newname = $newname[0];
1056 * @param string $newname
1059 function closeAndRename( $newname, $open = false ) {
1060 $newname = $this->checkRenameArgCount( $newname );
1062 if ( $this->handle
) {
1063 fclose( $this->handle
);
1064 $this->handle
= false;
1066 $this->renameOrException( $newname );
1068 $this->handle
= fopen( $this->filename
, "wt" );
1074 * @return string|null
1076 function getFilenames() {
1077 return $this->filename
;
1082 * Stream outputter to send data to a file via some filter program.
1083 * Even if compression is available in a library, using a separate
1084 * program can allow us to make use of a multi-processor system.
1087 class DumpPipeOutput
extends DumpFileOutput
{
1088 protected $command, $filename;
1089 protected $procOpenResource = false;
1092 * @param string $command
1093 * @param string $file
1095 function __construct( $command, $file = null ) {
1096 if ( !is_null( $file ) ) {
1097 $command .= " > " . wfEscapeShellArg( $file );
1100 $this->startCommand( $command );
1101 $this->command
= $command;
1102 $this->filename
= $file;
1106 * @param string $string
1108 function writeCloseStream( $string ) {
1109 parent
::writeCloseStream( $string );
1110 if ( $this->procOpenResource
) {
1111 proc_close( $this->procOpenResource
);
1112 $this->procOpenResource
= false;
1117 * @param string $command
1119 function startCommand( $command ) {
1121 0 => array( "pipe", "r" ),
1124 $this->procOpenResource
= proc_open( $command, $spec, $pipes );
1125 $this->handle
= $pipes[0];
1129 * @param string $newname
1131 function closeRenameAndReopen( $newname ) {
1132 $this->closeAndRename( $newname, true );
1136 * @param string $newname
1139 function closeAndRename( $newname, $open = false ) {
1140 $newname = $this->checkRenameArgCount( $newname );
1142 if ( $this->handle
) {
1143 fclose( $this->handle
);
1144 $this->handle
= false;
1146 if ( $this->procOpenResource
) {
1147 proc_close( $this->procOpenResource
);
1148 $this->procOpenResource
= false;
1150 $this->renameOrException( $newname );
1152 $command = $this->command
;
1153 $command .= " > " . wfEscapeShellArg( $this->filename
);
1154 $this->startCommand( $command );
1161 * Sends dump output via the gzip compressor.
1164 class DumpGZipOutput
extends DumpPipeOutput
{
1166 * @param string $file
1168 function __construct( $file ) {
1169 parent
::__construct( "gzip", $file );
1174 * Sends dump output via the bgzip2 compressor.
1177 class DumpBZip2Output
extends DumpPipeOutput
{
1179 * @param string $file
1181 function __construct( $file ) {
1182 parent
::__construct( "bzip2", $file );
1187 * Sends dump output via the p7zip compressor.
1190 class Dump7ZipOutput
extends DumpPipeOutput
{
1192 * @param string $file
1194 function __construct( $file ) {
1195 $command = $this->setup7zCommand( $file );
1196 parent
::__construct( $command );
1197 $this->filename
= $file;
1201 * @param string $file
1204 function setup7zCommand( $file ) {
1205 $command = "7za a -bd -si " . wfEscapeShellArg( $file );
1206 // Suppress annoying useless crap from p7zip
1207 // Unfortunately this could suppress real error messages too
1208 $command .= ' >' . wfGetNull() . ' 2>&1';
1213 * @param string $newname
1216 function closeAndRename( $newname, $open = false ) {
1217 $newname = $this->checkRenameArgCount( $newname );
1219 fclose( $this->handle
);
1220 proc_close( $this->procOpenResource
);
1221 $this->renameOrException( $newname );
1223 $command = $this->setup7zCommand( $this->filename
);
1224 $this->startCommand( $command );
1231 * Dump output filter class.
1232 * This just does output filtering and streaming; XML formatting is done
1233 * higher up, so be careful in what you do.
1239 * FIXME will need to be made protected whenever legacy code
1247 protected $sendingThisPage;
1250 * @param DumpOutput $sink
1252 function __construct( &$sink ) {
1253 $this->sink
=& $sink;
1257 * @param string $string
1259 function writeOpenStream( $string ) {
1260 $this->sink
->writeOpenStream( $string );
1264 * @param string $string
1266 function writeCloseStream( $string ) {
1267 $this->sink
->writeCloseStream( $string );
1271 * @param object $page
1272 * @param string $string
1274 function writeOpenPage( $page, $string ) {
1275 $this->sendingThisPage
= $this->pass( $page, $string );
1276 if ( $this->sendingThisPage
) {
1277 $this->sink
->writeOpenPage( $page, $string );
1282 * @param string $string
1284 function writeClosePage( $string ) {
1285 if ( $this->sendingThisPage
) {
1286 $this->sink
->writeClosePage( $string );
1287 $this->sendingThisPage
= false;
1292 * @param object $rev
1293 * @param string $string
1295 function writeRevision( $rev, $string ) {
1296 if ( $this->sendingThisPage
) {
1297 $this->sink
->writeRevision( $rev, $string );
1302 * @param object $rev
1303 * @param string $string
1305 function writeLogItem( $rev, $string ) {
1306 $this->sink
->writeRevision( $rev, $string );
1310 * @param string $newname
1312 function closeRenameAndReopen( $newname ) {
1313 $this->sink
->closeRenameAndReopen( $newname );
1317 * @param string $newname
1320 function closeAndRename( $newname, $open = false ) {
1321 $this->sink
->closeAndRename( $newname, $open );
1327 function getFilenames() {
1328 return $this->sink
->getFilenames();
1332 * Override for page-based filter types.
1333 * @param object $page
1336 function pass( $page ) {
1342 * Simple dump output filter to exclude all talk pages.
1345 class DumpNotalkFilter
extends DumpFilter
{
1347 * @param object $page
1350 function pass( $page ) {
1351 return !MWNamespace
::isTalk( $page->page_namespace
);
1356 * Dump output filter to include or exclude pages in a given set of namespaces.
1359 class DumpNamespaceFilter
extends DumpFilter
{
1361 public $invert = false;
1364 public $namespaces = array();
1367 * @param DumpOutput $sink
1368 * @param array $param
1369 * @throws MWException
1371 function __construct( &$sink, $param ) {
1372 parent
::__construct( $sink );
1375 "NS_MAIN" => NS_MAIN
,
1376 "NS_TALK" => NS_TALK
,
1377 "NS_USER" => NS_USER
,
1378 "NS_USER_TALK" => NS_USER_TALK
,
1379 "NS_PROJECT" => NS_PROJECT
,
1380 "NS_PROJECT_TALK" => NS_PROJECT_TALK
,
1381 "NS_FILE" => NS_FILE
,
1382 "NS_FILE_TALK" => NS_FILE_TALK
,
1383 "NS_IMAGE" => NS_IMAGE
, // NS_IMAGE is an alias for NS_FILE
1384 "NS_IMAGE_TALK" => NS_IMAGE_TALK
,
1385 "NS_MEDIAWIKI" => NS_MEDIAWIKI
,
1386 "NS_MEDIAWIKI_TALK" => NS_MEDIAWIKI_TALK
,
1387 "NS_TEMPLATE" => NS_TEMPLATE
,
1388 "NS_TEMPLATE_TALK" => NS_TEMPLATE_TALK
,
1389 "NS_HELP" => NS_HELP
,
1390 "NS_HELP_TALK" => NS_HELP_TALK
,
1391 "NS_CATEGORY" => NS_CATEGORY
,
1392 "NS_CATEGORY_TALK" => NS_CATEGORY_TALK
);
1394 if ( $param { 0 } == '!' ) {
1395 $this->invert
= true;
1396 $param = substr( $param, 1 );
1399 foreach ( explode( ',', $param ) as $key ) {
1400 $key = trim( $key );
1401 if ( isset( $constants[$key] ) ) {
1402 $ns = $constants[$key];
1403 $this->namespaces
[$ns] = true;
1404 } elseif ( is_numeric( $key ) ) {
1405 $ns = intval( $key );
1406 $this->namespaces
[$ns] = true;
1408 throw new MWException( "Unrecognized namespace key '$key'\n" );
1414 * @param object $page
1417 function pass( $page ) {
1418 $match = isset( $this->namespaces
[$page->page_namespace
] );
1419 return $this->invert
xor $match;
1424 * Dump output filter to include only the last revision in each page sequence.
1427 class DumpLatestFilter
extends DumpFilter
{
1437 * @param object $page
1438 * @param string $string
1440 function writeOpenPage( $page, $string ) {
1441 $this->page
= $page;
1442 $this->pageString
= $string;
1446 * @param string $string
1448 function writeClosePage( $string ) {
1450 $this->sink
->writeOpenPage( $this->page
, $this->pageString
);
1451 $this->sink
->writeRevision( $this->rev
, $this->revString
);
1452 $this->sink
->writeClosePage( $string );
1455 $this->revString
= null;
1457 $this->pageString
= null;
1461 * @param object $rev
1462 * @param string $string
1464 function writeRevision( $rev, $string ) {
1465 if ( $rev->rev_id
== $this->page
->page_latest
) {
1467 $this->revString
= $string;
1473 * Base class for output stream; prints to stdout or buffer or wherever.
1476 class DumpMultiWriter
{
1479 * @param array $sinks
1481 function __construct( $sinks ) {
1482 $this->sinks
= $sinks;
1483 $this->count
= count( $sinks );
1487 * @param string $string
1489 function writeOpenStream( $string ) {
1490 for ( $i = 0; $i < $this->count
; $i++
) {
1491 $this->sinks
[$i]->writeOpenStream( $string );
1496 * @param string $string
1498 function writeCloseStream( $string ) {
1499 for ( $i = 0; $i < $this->count
; $i++
) {
1500 $this->sinks
[$i]->writeCloseStream( $string );
1505 * @param object $page
1506 * @param string $string
1508 function writeOpenPage( $page, $string ) {
1509 for ( $i = 0; $i < $this->count
; $i++
) {
1510 $this->sinks
[$i]->writeOpenPage( $page, $string );
1515 * @param string $string
1517 function writeClosePage( $string ) {
1518 for ( $i = 0; $i < $this->count
; $i++
) {
1519 $this->sinks
[$i]->writeClosePage( $string );
1524 * @param object $rev
1525 * @param string $string
1527 function writeRevision( $rev, $string ) {
1528 for ( $i = 0; $i < $this->count
; $i++
) {
1529 $this->sinks
[$i]->writeRevision( $rev, $string );
1534 * @param array $newnames
1536 function closeRenameAndReopen( $newnames ) {
1537 $this->closeAndRename( $newnames, true );
1541 * @param array $newnames
1544 function closeAndRename( $newnames, $open = false ) {
1545 for ( $i = 0; $i < $this->count
; $i++
) {
1546 $this->sinks
[$i]->closeAndRename( $newnames[$i], $open );
1553 function getFilenames() {
1554 $filenames = array();
1555 for ( $i = 0; $i < $this->count
; $i++
) {
1556 $filenames[] = $this->sinks
[$i]->getFilenames();