<?php
/**
- * MediaWiki page data importer
+ * MediaWiki page data importer.
*
* Copyright © 2003,2005 Brion Vibber <brion@pobox.com>
* http://www.mediawiki.org/
class WikiImporter {
private $reader = null;
private $mLogItemCallback, $mUploadCallback, $mRevisionCallback, $mPageCallback;
- private $mSiteInfoCallback, $mTargetNamespace, $mPageOutCallback;
- private $mDebug;
+ private $mSiteInfoCallback, $mTargetNamespace, $mTargetRootPage, $mPageOutCallback;
+ private $mNoticeCallback, $mDebug;
private $mImportUploads, $mImageBasePath;
private $mNoUpdates = false;
/**
* Creates an ImportXMLReader drawing from the source provided
+ * @param $source
*/
function __construct( $source ) {
$this->reader = new XMLReader();
$id = UploadSourceAdapter::registerSource( $source );
if (defined( 'LIBXML_PARSEHUGE' ) ) {
$this->reader->open( "uploadsource://$id", null, LIBXML_PARSEHUGE );
- }
- else {
+ } else {
$this->reader->open( "uploadsource://$id" );
}
wfDebug( "IMPORT: $data\n" );
}
- private function notice( $data ) {
- global $wgCommandLineMode;
- if( $wgCommandLineMode ) {
- print "$data\n";
- } else {
- global $wgOut;
- $wgOut->addHTML( "<li>" . htmlspecialchars( $data ) . "</li>\n" );
+ private function notice( $msg /*, $param, ...*/ ) {
+ $params = func_get_args();
+ array_shift( $params );
+
+ if ( is_callable( $this->mNoticeCallback ) ) {
+ call_user_func( $this->mNoticeCallback, $msg, $params );
+ } else { # No ImportReporter -> CLI
+ echo wfMessage( $msg, $params )->text() . "\n";
}
}
/**
* Set debug mode...
+ * @param $debug bool
*/
function setDebug( $debug ) {
$this->mDebug = $debug;
/**
* Set 'no updates' mode. In this mode, the link tables will not be updated by the importer
+ * @param $noupdates bool
*/
function setNoUpdates( $noupdates ) {
$this->mNoUpdates = $noupdates;
}
+ /**
+ * Set a callback that displays notice messages
+ *
+ * @param $callback callback
+ * @return callback
+ */
+ public function setNoticeCallback( $callback ) {
+ return wfSetVar( $this->mNoticeCallback, $callback );
+ }
+
/**
* Sets the action to perform as each new page in the stream is reached.
* @param $callback callback
/**
* Set a target namespace to override the defaults
+ * @param $namespace
+ * @return bool
*/
public function setTargetNamespace( $namespace ) {
if( is_null( $namespace ) ) {
}
/**
- * @parma $dir
+ * Set a target root page under which all pages are imported
+ * @param $rootpage
+ * @return status object
+ */
+ public function setTargetRootPage( $rootpage ) {
+ $status = Status::newGood();
+ if( is_null( $rootpage ) ) {
+ // No rootpage
+ $this->mTargetRootPage = null;
+ } elseif( $rootpage !== '' ) {
+ $rootpage = rtrim( $rootpage, '/' ); //avoid double slashes
+ $title = Title::newFromText( $rootpage, !is_null( $this->mTargetNamespace ) ? $this->mTargetNamespace : NS_MAIN );
+ if( !$title || $title->isExternal() ) {
+ $status->fatal( 'import-rootpage-invalid' );
+ } else {
+ if( !MWNamespace::hasSubpages( $title->getNamespace() ) ) {
+ global $wgContLang;
+
+ $displayNSText = $title->getNamespace() == NS_MAIN
+ ? wfMessage( 'blanknamespace' )->text()
+ : $wgContLang->getNsText( $title->getNamespace() );
+ $status->fatal( 'import-rootpage-nosubpage', $displayNSText );
+ } else {
+ // set namespace to 'all', so the namespace check in processTitle() can passed
+ $this->setTargetNamespace( null );
+ $this->mTargetRootPage = $title->getPrefixedDBKey();
+ }
+ }
+ }
+ return $status;
+ }
+
+ /**
+ * @param $dir
*/
public function setImageBasePath( $dir ) {
$this->mImageBasePath = $dir;
/**
* Default per-revision callback, performs the import.
* @param $revision WikiRevision
+ * @return bool
*/
public function importRevision( $revision ) {
$dbw = wfGetDB( DB_MASTER );
/**
* Default per-revision callback, performs the import.
* @param $rev WikiRevision
+ * @return bool
*/
public function importLogItem( $rev ) {
$dbw = wfGetDB( DB_MASTER );
/**
* Dummy for now...
+ * @param $revision
+ * @return bool
*/
public function importUpload( $revision ) {
$dbw = wfGetDB( DB_MASTER );
/**
* Mostly for hook use
+ * @param $title
+ * @param $origTitle
+ * @param $revCount
+ * @param $sRevCount
+ * @param $pageInfo
+ * @return
*/
public function finishImportPage( $title, $origTitle, $revCount, $sRevCount, $pageInfo ) {
$args = func_get_args();
}
/**
- * Notify the callback function when a new <page> is reached.
+ * Notify the callback function when a new "<page>" is reached.
* @param $title Title
*/
function pageCallback( $title ) {
}
/**
- * Notify the callback function when a </page> is closed.
+ * Notify the callback function when a "</page>" is closed.
* @param $title Title
* @param $origTitle Title
* @param $revCount Integer
/**
* Notify the callback function of a revision
- * @param $revision A WikiRevision object
+ * @param $revision WikiRevision object
+ * @return bool|mixed
*/
private function revisionCallback( $revision ) {
if ( isset( $this->mRevisionCallback ) ) {
/**
* Notify the callback function of a new log item
- * @param $revision A WikiRevision object
+ * @param $revision WikiRevision object
+ * @return bool|mixed
*/
private function logItemCallback( $revision ) {
if ( isset( $this->mLogItemCallback ) ) {
/**
* Primary entry point
+ * @return bool
*/
public function doImport() {
$this->reader->read();
}
/**
- * @param $pageInfo
+ * @param $pageInfo array
*/
private function handleRevision( &$pageInfo ) {
$this->debug( "Enter revision handler" );
$origTitle = Title::newFromText( $workTitle );
if( !is_null( $this->mTargetNamespace ) && !is_null( $origTitle ) ) {
- $title = Title::makeTitle( $this->mTargetNamespace,
+ # makeTitleSafe, because $origTitle can have a interwiki (different setting of interwiki map)
+ # and than dbKey can begin with a lowercase char
+ $title = Title::makeTitleSafe( $this->mTargetNamespace,
$origTitle->getDBkey() );
} else {
+ if( !is_null( $this->mTargetRootPage ) ) {
+ $workTitle = $this->mTargetRootPage . '/' . $workTitle;
+ }
$title = Title::newFromText( $workTitle );
}
if( is_null( $title ) ) {
# Invalid page title? Ignore the page
- $this->notice( "Skipping invalid page title '$workTitle'" );
+ $this->notice( 'import-error-invalid', $workTitle );
+ return false;
+ } elseif( $title->isExternal() ) {
+ $this->notice( 'import-error-interwiki', $title->getPrefixedText() );
return false;
- } elseif( $title->getInterwiki() != '' ) {
- $this->notice( "Skipping interwiki page title '$workTitle'" );
+ } elseif( !$title->canExist() ) {
+ $this->notice( 'import-error-special', $title->getPrefixedText() );
return false;
} elseif( !$title->userCan( 'edit' ) && !$wgCommandLineMode ) {
# Do not import if the importing wiki user cannot edit this page
- $this->notice( wfMessage( 'import-error-edit', $title->getText() )->text() );
+ $this->notice( 'import-error-edit', $title->getPrefixedText() );
return false;
} elseif( !$title->exists() && !$title->userCan( 'create' ) && !$wgCommandLineMode ) {
# Do not import if the importing wiki user cannot create this page
- $this->notice( wfMessage( 'import-error-create', $title->getText() )->text() );
+ $this->notice( 'import-error-create', $title->getPrefixedText() );
return false;
}
* @return string
*/
static function registerSource( $source ) {
- $id = wfGenerateToken();
+ $id = wfRandomString();
self::$sourceRegistrations[$id] = $source;
*/
class WikiRevision {
var $importer = null;
+
+ /**
+ * @var Title
+ */
var $title = null;
var $id = 0;
var $timestamp = "20010115000000";
var $sha1base36 = false;
var $isTemp = false;
var $archiveName = '';
+ var $fileIsTemp;
private $mNoUpdates = false;
/**
$this->type = $type;
}
+ /**
+ * @param $action
+ */
function setAction( $action ) {
$this->action = $action;
}
$linkCache = LinkCache::singleton();
$linkCache->clear();
- $article = new Article( $this->title );
- $pageId = $article->getId();
- if( $pageId == 0 ) {
+ $page = WikiPage::factory( $this->title );
+ if( !$page->exists() ) {
# must create the page...
- $pageId = $article->insertOn( $dbw );
+ $pageId = $page->insertOn( $dbw );
$created = true;
$oldcountable = null;
} else {
+ $pageId = $page->getId();
$created = false;
$prior = $dbw->selectField( 'revision', '1',
$this->title->getPrefixedText() . "]], timestamp " . $this->timestamp . "\n" );
return false;
}
- $oldcountable = $article->isCountable();
+ $oldcountable = $page->isCountable();
}
# @todo FIXME: Use original rev_id optionally (better for backups)
'minor_edit' => $this->minor,
) );
$revision->insertOn( $dbw );
- $changed = $article->updateIfNewerOn( $dbw, $revision );
+ $changed = $page->updateIfNewerOn( $dbw, $revision );
if ( $changed !== false && !$this->mNoUpdates ) {
wfDebug( __METHOD__ . ": running updates\n" );
- $article->doEditUpdates( $revision, $userObj, array( 'created' => $created, 'oldcountable' => $oldcountable ) );
+ $page->doEditUpdates( $revision, $userObj, array( 'created' => $created, 'oldcountable' => $oldcountable ) );
}
return true;
$this->mRead = false;
}
+ /**
+ * @return bool
+ */
function atEnd() {
return $this->mRead;
}
+ /**
+ * @return bool|string
+ */
function readChunk() {
if( $this->atEnd() ) {
return false;
return feof( $this->mHandle );
}
+ /**
+ * @return string
+ */
function readChunk() {
return fread( $this->mHandle, 32768 );
}
# quicker and sorts out user-agent problems which might
# otherwise prevent importing from large sites, such
# as the Wikimedia cluster, etc.
- $data = Http::request( $method, $url );
+ $data = Http::request( $method, $url, array( 'followRedirects' => true ) );
if( $data !== false ) {
$file = tmpfile();
fwrite( $file, $data );