From 47b728b45b9f8a980f7cb8ca28f6317476756f75 Mon Sep 17 00:00:00 2001
From: Tim Starling <tstarling@users.mediawiki.org>
Date: Mon, 31 Jan 2011 07:07:43 +0000
Subject: [PATCH] Cleanup for r66268, r66267: merge WikiImporter back into
 Import.php, where it started. It doesn't really make sense to have a class
 called WikiImporter in a file called ImportXMLReader.php, and a few little
 helper classes for it in a file called Import.php.

---
 includes/AutoLoader.php      |   2 +-
 includes/Import.php          | 733 ++++++++++++++++++++++++++++++++++
 includes/ImportXMLReader.php | 737 -----------------------------------
 3 files changed, 734 insertions(+), 738 deletions(-)
 delete mode 100644 includes/ImportXMLReader.php

diff --git a/includes/AutoLoader.php b/includes/AutoLoader.php
index 6001b17b1d..d99a0978b0 100644
--- a/includes/AutoLoader.php
+++ b/includes/AutoLoader.php
@@ -681,7 +681,7 @@ $wgAutoloadLocalClasses = array(
 	'WantedFilesPage' => 'includes/specials/SpecialWantedfiles.php',
 	'WantedPagesPage' => 'includes/specials/SpecialWantedpages.php',
 	'WantedTemplatesPage' => 'includes/specials/SpecialWantedtemplates.php',
-	'WikiImporter' => 'includes/ImportXMLReader.php',
+	'WikiImporter' => 'includes/Import.php',
 	'WikiRevision' => 'includes/Import.php',
 	'WithoutInterwikiPage' => 'includes/specials/SpecialWithoutinterwiki.php',
 
diff --git a/includes/Import.php b/includes/Import.php
index 533d538192..15fd15164c 100644
--- a/includes/Import.php
+++ b/includes/Import.php
@@ -24,6 +24,739 @@
  * @ingroup SpecialPage
  */
 
+/**
+ * XML file reader for the page data importer
+ *
+ * implements Special:Import
+ * @ingroup SpecialPage
+ */
+class WikiImporter {
+	private $reader = null;
+	private $mLogItemCallback, $mUploadCallback, $mRevisionCallback, $mPageCallback;
+	private $mSiteInfoCallback, $mTargetNamespace, $mPageOutCallback;
+	private $mDebug;
+
+	/**
+	 * Creates an ImportXMLReader drawing from the source provided
+	*/
+	function __construct( $source ) {
+		$this->reader = new XMLReader2();
+
+		stream_wrapper_register( 'uploadsource', 'UploadSourceAdapter' );
+		$id = UploadSourceAdapter::registerSource( $source );
+		$this->reader->open( "uploadsource://$id" );
+
+		// Default callbacks
+		$this->setRevisionCallback( array( $this, "importRevision" ) );
+		$this->setUploadCallback( array( $this, 'importUpload' ) );
+		$this->setLogItemCallback( array( $this, 'importLogItem' ) );
+		$this->setPageOutCallback( array( $this, 'finishImportPage' ) );
+	}
+
+	private function throwXmlError( $err ) {
+		$this->debug( "FAILURE: $err" );
+		wfDebug( "WikiImporter XML error: $err\n" );
+	}
+
+	private function debug( $data ) {
+		if( $this->mDebug ) {
+			wfDebug( "IMPORT: $data\n" );
+		}
+	}
+
+	private function warn( $data ) {
+		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" );
+		}
+	}
+
+	/**
+	 * Set debug mode...
+	 */
+	function setDebug( $debug ) {
+		$this->mDebug = $debug;
+	}
+
+	/**
+	 * Sets the action to perform as each new page in the stream is reached.
+	 * @param $callback callback
+	 * @return callback
+	 */
+	public function setPageCallback( $callback ) {
+		$previous = $this->mPageCallback;
+		$this->mPageCallback = $callback;
+		return $previous;
+	}
+
+	/**
+	 * Sets the action to perform as each page in the stream is completed.
+	 * Callback accepts the page title (as a Title object), a second object
+	 * with the original title form (in case it's been overridden into a
+	 * local namespace), and a count of revisions.
+	 *
+	 * @param $callback callback
+	 * @return callback
+	 */
+	public function setPageOutCallback( $callback ) {
+		$previous = $this->mPageOutCallback;
+		$this->mPageOutCallback = $callback;
+		return $previous;
+	}
+
+	/**
+	 * Sets the action to perform as each page revision is reached.
+	 * @param $callback callback
+	 * @return callback
+	 */
+	public function setRevisionCallback( $callback ) {
+		$previous = $this->mRevisionCallback;
+		$this->mRevisionCallback = $callback;
+		return $previous;
+	}
+
+	/**
+	 * Sets the action to perform as each file upload version is reached.
+	 * @param $callback callback
+	 * @return callback
+	 */
+	public function setUploadCallback( $callback ) {
+		$previous = $this->mUploadCallback;
+		$this->mUploadCallback = $callback;
+		return $previous;
+	}
+
+	/**
+	 * Sets the action to perform as each log item reached.
+	 * @param $callback callback
+	 * @return callback
+	 */
+	public function setLogItemCallback( $callback ) {
+		$previous = $this->mLogItemCallback;
+		$this->mLogItemCallback = $callback;
+		return $previous;
+	}
+
+	/**
+	 * Sets the action to perform when site info is encountered
+	 * @param $callback callback
+	 * @return callback
+	 */
+	public function setSiteInfoCallback( $callback ) {
+		$previous = $this->mSiteInfoCallback;
+		$this->mSiteInfoCallback = $callback;
+		return $previous;
+	}
+
+	/**
+	 * Set a target namespace to override the defaults
+	 */
+	public function setTargetNamespace( $namespace ) {
+		if( is_null( $namespace ) ) {
+			// Don't override namespaces
+			$this->mTargetNamespace = null;
+		} elseif( $namespace >= 0 ) {
+			// FIXME: Check for validity
+			$this->mTargetNamespace = intval( $namespace );
+		} else {
+			return false;
+		}
+	}
+
+	/**
+	 * Default per-revision callback, performs the import.
+	 * @param $revision WikiRevision
+	 */
+	public function importRevision( $revision ) {
+		$dbw = wfGetDB( DB_MASTER );
+		return $dbw->deadlockLoop( array( $revision, 'importOldRevision' ) );
+	}
+
+	/**
+	 * Default per-revision callback, performs the import.
+	 * @param $rev WikiRevision
+	 */
+	public function importLogItem( $rev ) {
+		$dbw = wfGetDB( DB_MASTER );
+		return $dbw->deadlockLoop( array( $rev, 'importLogItem' ) );
+	}
+
+	/**
+	 * Dummy for now...
+	 */
+	public function importUpload( $revision ) {
+		//$dbw = wfGetDB( DB_MASTER );
+		//return $dbw->deadlockLoop( array( $revision, 'importUpload' ) );
+		return false;
+	}
+
+	/**
+	 * Mostly for hook use
+	 */
+	public function finishImportPage( $title, $origTitle, $revCount, $sRevCount, $pageInfo ) {
+		$args = func_get_args();
+		return wfRunHooks( 'AfterImportPage', $args );
+	}
+
+	/**
+	 * Alternate per-revision callback, for debugging.
+	 * @param $revision WikiRevision
+	 */
+	public function debugRevisionHandler( &$revision ) {
+		$this->debug( "Got revision:" );
+		if( is_object( $revision->title ) ) {
+			$this->debug( "-- Title: " . $revision->title->getPrefixedText() );
+		} else {
+			$this->debug( "-- Title: <invalid>" );
+		}
+		$this->debug( "-- User: " . $revision->user_text );
+		$this->debug( "-- Timestamp: " . $revision->timestamp );
+		$this->debug( "-- Comment: " . $revision->comment );
+		$this->debug( "-- Text: " . $revision->text );
+	}
+
+	/**
+	 * Notify the callback function when a new <page> is reached.
+	 * @param $title Title
+	 */
+	function pageCallback( $title ) {
+		if( isset( $this->mPageCallback ) ) {
+			call_user_func( $this->mPageCallback, $title );
+		}
+	}
+
+	/**
+	 * Notify the callback function when a </page> is closed.
+	 * @param $title Title
+	 * @param $origTitle Title
+	 * @param $revCount Integer
+	 * @param $sucCount Int: number of revisions for which callback returned true
+	 * @param $pageInfo Array: associative array of page information
+	 */
+	private function pageOutCallback( $title, $origTitle, $revCount, $sucCount, $pageInfo ) {
+		if( isset( $this->mPageOutCallback ) ) {
+			$args = func_get_args();
+			call_user_func_array( $this->mPageOutCallback, $args );
+		}
+	}
+
+	/**
+	 * Notify the callback function of a revision
+	 * @param $revision A WikiRevision object
+	 */
+	private function revisionCallback( $revision ) {
+		if ( isset( $this->mRevisionCallback ) ) {
+			return call_user_func_array( $this->mRevisionCallback,
+					array( $revision, $this ) );
+		} else {
+			return false;
+		}
+	}
+
+	/**
+	 * Notify the callback function of a new log item
+	 * @param $revision A WikiRevision object
+	 */
+	private function logItemCallback( $revision ) {
+		if ( isset( $this->mLogItemCallback ) ) {
+			return call_user_func_array( $this->mLogItemCallback,
+					array( $revision, $this ) );
+		} else {
+			return false;
+		}
+	}
+
+	/**
+	 * Shouldn't something like this be built-in to XMLReader?
+	 * Fetches text contents of the current element, assuming
+	 * no sub-elements or such scary things.
+	 * @return string
+	 * @access private
+	 */
+	private function nodeContents() {
+		return $this->reader->nodeContents();
+	}
+
+	# --------------
+
+	/** Left in for debugging */
+	private function dumpElement() {
+		static $lookup = null;
+		if (!$lookup) {
+			$xmlReaderConstants = array(
+				"NONE",
+				"ELEMENT",
+				"ATTRIBUTE",
+				"TEXT",
+				"CDATA",
+				"ENTITY_REF",
+				"ENTITY",
+				"PI",
+				"COMMENT",
+				"DOC",
+				"DOC_TYPE",
+				"DOC_FRAGMENT",
+				"NOTATION",
+				"WHITESPACE",
+				"SIGNIFICANT_WHITESPACE",
+				"END_ELEMENT",
+				"END_ENTITY",
+				"XML_DECLARATION",
+				);
+			$lookup = array();
+
+			foreach( $xmlReaderConstants as $name ) {
+				$lookup[constant("XmlReader::$name")] = $name;
+			}
+		}
+
+		print( var_dump(
+			$lookup[$this->reader->nodeType],
+			$this->reader->name,
+			$this->reader->value
+		)."\n\n" );
+	}
+
+	/**
+	 * Primary entry point
+	 */
+	public function doImport() {
+		$this->reader->read();
+
+		if ( $this->reader->name != 'mediawiki' ) {
+			throw new MWException( "Expected <mediawiki> tag, got ".
+				$this->reader->name );
+		}
+		$this->debug( "<mediawiki> tag is correct." );
+
+		$this->debug( "Starting primary dump processing loop." );
+
+		$keepReading = $this->reader->read();
+		$skip = false;
+		while ( $keepReading ) {
+			$tag = $this->reader->name;
+			$type = $this->reader->nodeType;
+
+			if ( !wfRunHooks( 'ImportHandleToplevelXMLTag', $this->reader ) ) {
+				// Do nothing
+			} elseif ( $tag == 'mediawiki' && $type == XmlReader::END_ELEMENT ) {
+				break;
+			} elseif ( $tag == 'siteinfo' ) {
+				$this->handleSiteInfo();
+			} elseif ( $tag == 'page' ) {
+				$this->handlePage();
+			} elseif ( $tag == 'logitem' ) {
+				$this->handleLogItem();
+			} elseif ( $tag != '#text' ) {
+				$this->warn( "Unhandled top-level XML tag $tag" );
+
+				$skip = true;
+			}
+
+			if ($skip) {
+				$keepReading = $this->reader->next();
+				$skip = false;
+				$this->debug( "Skip" );
+			} else {
+				$keepReading = $this->reader->read();
+			}
+		}
+
+		return true;
+	}
+
+	private function handleSiteInfo() {
+		// Site info is useful, but not actually used for dump imports.
+		// Includes a quick short-circuit to save performance.
+		if ( ! $this->mSiteInfoCallback ) {
+			$this->reader->next();
+			return true;
+		}
+		throw new MWException( "SiteInfo tag is not yet handled, do not set mSiteInfoCallback" );
+	}
+
+	private function handleLogItem() {
+		$this->debug( "Enter log item handler." );
+		$logInfo = array();
+
+		// Fields that can just be stuffed in the pageInfo object
+		$normalFields = array( 'id', 'comment', 'type', 'action', 'timestamp',
+					'logtitle', 'params' );
+
+		while ( $this->reader->read() ) {
+			if ( $this->reader->nodeType == XmlReader::END_ELEMENT &&
+					$this->reader->name == 'logitem') {
+				break;
+			}
+
+			$tag = $this->reader->name;
+
+			if ( !wfRunHooks( 'ImportHandleLogItemXMLTag',
+						$this->reader, $logInfo ) ) {
+				// Do nothing
+			} elseif ( in_array( $tag, $normalFields ) ) {
+				$logInfo[$tag] = $this->nodeContents();
+			} elseif ( $tag == 'contributor' ) {
+				$logInfo['contributor'] = $this->handleContributor();
+			} elseif ( $tag != '#text' ) {
+				$this->warn( "Unhandled log-item XML tag $tag" );
+			}
+		}
+
+		$this->processLogItem( $logInfo );
+	}
+
+	private function processLogItem( $logInfo ) {
+		$revision = new WikiRevision;
+
+		$revision->setID( $logInfo['id'] );
+		$revision->setType( $logInfo['type'] );
+		$revision->setAction( $logInfo['action'] );
+		$revision->setTimestamp( $logInfo['timestamp'] );
+		$revision->setParams( $logInfo['params'] );
+		$revision->setTitle( Title::newFromText( $logInfo['logtitle'] ) );
+
+		if ( isset( $logInfo['comment'] ) ) {
+			$revision->setComment( $logInfo['comment'] );
+		}
+
+		if ( isset( $logInfo['contributor']['ip'] ) ) {
+			$revision->setUserIP( $logInfo['contributor']['ip'] );
+		}
+		if ( isset( $logInfo['contributor']['username'] ) ) {
+			$revision->setUserName( $logInfo['contributor']['username'] );
+		}
+
+		return $this->logItemCallback( $revision );
+	}
+
+	private function handlePage() {
+		// Handle page data.
+		$this->debug( "Enter page handler." );
+		$pageInfo = array( 'revisionCount' => 0, 'successfulRevisionCount' => 0 );
+
+		// Fields that can just be stuffed in the pageInfo object
+		$normalFields = array( 'title', 'id', 'redirect', 'restrictions' );
+
+		$skip = false;
+		$badTitle = false;
+
+		while ( $skip ? $this->reader->next() : $this->reader->read() ) {
+			if ( $this->reader->nodeType == XmlReader::END_ELEMENT &&
+					$this->reader->name == 'page') {
+				break;
+			}
+
+			$tag = $this->reader->name;
+
+			if ( $badTitle ) {
+				// The title is invalid, bail out of this page
+				$skip = true;
+			} elseif ( !wfRunHooks( 'ImportHandlePageXMLTag', array( $this->reader,
+						&$pageInfo ) ) ) {
+				// Do nothing
+			} elseif ( in_array( $tag, $normalFields ) ) {
+				$pageInfo[$tag] = $this->nodeContents();
+				if ( $tag == 'title' ) {
+					$title = $this->processTitle( $pageInfo['title'] );
+
+					if ( !$title ) {
+						$badTitle = true;
+						$skip = true;
+					}
+
+					$this->pageCallback( $title );
+					list( $pageInfo['_title'], $origTitle ) = $title;
+				}
+			} elseif ( $tag == 'revision' ) {
+				$this->handleRevision( $pageInfo );
+			} elseif ( $tag == 'upload' ) {
+				$this->handleUpload( $pageInfo );
+			} elseif ( $tag != '#text' ) {
+				$this->warn( "Unhandled page XML tag $tag" );
+				$skip = true;
+			}
+		}
+
+		$this->pageOutCallback( $pageInfo['_title'], $origTitle,
+					$pageInfo['revisionCount'],
+					$pageInfo['successfulRevisionCount'],
+					$pageInfo );
+	}
+
+	private function handleRevision( &$pageInfo ) {
+		$this->debug( "Enter revision handler" );
+		$revisionInfo = array();
+
+		$normalFields = array( 'id', 'timestamp', 'comment', 'minor', 'text' );
+
+		$skip = false;
+
+		while ( $skip ? $this->reader->next() : $this->reader->read() ) {
+			if ( $this->reader->nodeType == XmlReader::END_ELEMENT &&
+					$this->reader->name == 'revision') {
+				break;
+			}
+
+			$tag = $this->reader->name;
+
+			if ( !wfRunHooks( 'ImportHandleRevisionXMLTag', $this->reader,
+						$pageInfo, $revisionInfo ) ) {
+				// Do nothing
+			} elseif ( in_array( $tag, $normalFields ) ) {
+				$revisionInfo[$tag] = $this->nodeContents();
+			} elseif ( $tag == 'contributor' ) {
+				$revisionInfo['contributor'] = $this->handleContributor();
+			} elseif ( $tag != '#text' ) {
+				$this->warn( "Unhandled revision XML tag $tag" );
+				$skip = true;
+			}
+		}
+
+		$pageInfo['revisionCount']++;
+		if ( $this->processRevision( $pageInfo, $revisionInfo ) ) {
+			$pageInfo['successfulRevisionCount']++;
+		}
+	}
+
+	private function processRevision( $pageInfo, $revisionInfo ) {
+		$revision = new WikiRevision;
+
+		$revision->setID( $revisionInfo['id'] );
+		$revision->setText( $revisionInfo['text'] );
+		$revision->setTitle( $pageInfo['_title'] );
+		$revision->setTimestamp( $revisionInfo['timestamp'] );
+
+		if ( isset( $revisionInfo['comment'] ) ) {
+			$revision->setComment( $revisionInfo['comment'] );
+		}
+
+		if ( isset( $revisionInfo['minor'] ) )
+			$revision->setMinor( true );
+
+		if ( isset( $revisionInfo['contributor']['ip'] ) ) {
+			$revision->setUserIP( $revisionInfo['contributor']['ip'] );
+		}
+		if ( isset( $revisionInfo['contributor']['username'] ) ) {
+			$revision->setUserName( $revisionInfo['contributor']['username'] );
+		}
+
+		return $this->revisionCallback( $revision );
+	}
+
+	private function handleUpload( &$pageInfo ) {
+		$this->debug( "Enter upload handler" );
+		$uploadInfo = array();
+
+		$normalFields = array( 'timestamp', 'comment', 'filename', 'text',
+					'src', 'size' );
+
+		$skip = false;
+
+		while ( $skip ? $this->reader->next() : $this->reader->read() ) {
+			if ( $this->reader->nodeType == XmlReader::END_ELEMENT &&
+					$this->reader->name == 'upload') {
+				break;
+			}
+
+			$tag = $this->reader->name;
+
+			if ( !wfRunHooks( 'ImportHandleUploadXMLTag', $this->reader,
+						$pageInfo ) ) {
+				// Do nothing
+			} elseif ( in_array( $tag, $normalFields ) ) {
+				$uploadInfo[$tag] = $this->nodeContents();
+			} elseif ( $tag == 'contributor' ) {
+				$uploadInfo['contributor'] = $this->handleContributor();
+			} elseif ( $tag != '#text' ) {
+				$this->warn( "Unhandled upload XML tag $tag" );
+				$skip = true;
+			}
+		}
+
+		return $this->processUpload( $pageInfo, $uploadInfo );
+	}
+
+	private function processUpload( $pageInfo, $uploadInfo ) {
+		$revision = new WikiRevision;
+
+		$revision->setTitle( $pageInfo['_title'] );
+		$revision->setID( $uploadInfo['id'] );
+		$revision->setTimestamp( $uploadInfo['timestamp'] );
+		$revision->setText( $uploadInfo['text'] );
+		$revision->setFilename( $uploadInfo['filename'] );
+		$revision->setSrc( $uploadInfo['src'] );
+		$revision->setSize( intval( $uploadInfo['size'] ) );
+		$revision->setComment( $uploadInfo['comment'] );
+
+		if ( isset( $uploadInfo['contributor']['ip'] ) ) {
+			$revision->setUserIP( $uploadInfo['contributor']['ip'] );
+		}
+		if ( isset( $uploadInfo['contributor']['username'] ) ) {
+			$revision->setUserName( $uploadInfo['contributor']['username'] );
+		}
+
+		return $this->uploadCallback( $revision );
+	}
+
+	private function handleContributor() {
+		$fields = array( 'id', 'ip', 'username' );
+		$info = array();
+
+		while ( $this->reader->read() ) {
+			if ( $this->reader->nodeType == XmlReader::END_ELEMENT &&
+					$this->reader->name == 'contributor') {
+				break;
+			}
+
+			$tag = $this->reader->name;
+
+			if ( in_array( $tag, $fields ) ) {
+				$info[$tag] = $this->nodeContents();
+			}
+		}
+
+		return $info;
+	}
+
+	private function processTitle( $text ) {
+		$workTitle = $text;
+		$origTitle = Title::newFromText( $workTitle );
+
+		if( !is_null( $this->mTargetNamespace ) && !is_null( $origTitle ) ) {
+			$title = Title::makeTitle( $this->mTargetNamespace,
+				$origTitle->getDBkey() );
+		} else {
+			$title = Title::newFromText( $workTitle );
+		}
+
+		if( is_null( $title ) ) {
+			// Invalid page title? Ignore the page
+			$this->notice( "Skipping invalid page title '$workTitle'" );
+			return false;
+		} elseif( $title->getInterwiki() != '' ) {
+			$this->notice( "Skipping interwiki page title '$workTitle'" );
+			return false;
+		}
+
+		return array( $origTitle, $title );
+	}
+}
+
+/** This is a horrible hack used to keep source compatibility */
+class UploadSourceAdapter {
+	static $sourceRegistrations = array();
+
+	private $mSource;
+	private $mBuffer;
+	private $mPosition;
+
+	static function registerSource( $source ) {
+		$id = wfGenerateToken();
+
+		self::$sourceRegistrations[$id] = $source;
+
+		return $id;
+	}
+
+	function stream_open( $path, $mode, $options, &$opened_path ) {
+		$url = parse_url($path);
+		$id = $url['host'];
+
+		if ( !isset( self::$sourceRegistrations[$id] ) ) {
+			return false;
+		}
+
+		$this->mSource = self::$sourceRegistrations[$id];
+
+		return true;
+	}
+
+	function stream_read( $count ) {
+		$return = '';
+		$leave = false;
+
+		while ( !$leave && !$this->mSource->atEnd() &&
+				strlen($this->mBuffer) < $count ) {
+			$read = $this->mSource->readChunk();
+
+			if ( !strlen($read) ) {
+				$leave = true;
+			}
+
+			$this->mBuffer .= $read;
+		}
+
+		if ( strlen($this->mBuffer) ) {
+			$return = substr( $this->mBuffer, 0, $count );
+			$this->mBuffer = substr( $this->mBuffer, $count );
+		}
+
+		$this->mPosition += strlen($return);
+
+		return $return;
+	}
+
+	function stream_write( $data ) {
+		return false;
+	}
+
+	function stream_tell() {
+		return $this->mPosition;
+	}
+
+	function stream_eof() {
+		return $this->mSource->atEnd();
+	}
+
+	function url_stat() {
+		$result = array();
+
+		$result['dev'] = $result[0] = 0;
+		$result['ino'] = $result[1] = 0;
+		$result['mode'] = $result[2] = 0;
+		$result['nlink'] = $result[3] = 0;
+		$result['uid'] = $result[4] = 0;
+		$result['gid'] = $result[5] = 0;
+		$result['rdev'] = $result[6] = 0;
+		$result['size'] = $result[7] = 0;
+		$result['atime'] = $result[8] = 0;
+		$result['mtime'] = $result[9] = 0;
+		$result['ctime'] = $result[10] = 0;
+		$result['blksize'] = $result[11] = 0;
+		$result['blocks'] = $result[12] = 0;
+
+		return $result;
+	}
+}
+
+class XMLReader2 extends XMLReader {
+	function nodeContents() {
+		if( $this->isEmptyElement ) {
+			return "";
+		}
+		$buffer = "";
+		while( $this->read() ) {
+			switch( $this->nodeType ) {
+			case XmlReader::TEXT:
+			case XmlReader::SIGNIFICANT_WHITESPACE:
+				$buffer .= $this->value;
+				break;
+			case XmlReader::END_ELEMENT:
+				return $buffer;
+			}
+		}
+		return $this->close();
+	}
+}
+
 /**
  * @todo document (e.g. one-sentence class description).
  * @ingroup SpecialPage
diff --git a/includes/ImportXMLReader.php b/includes/ImportXMLReader.php
deleted file mode 100644
index bd49fc0a3e..0000000000
--- a/includes/ImportXMLReader.php
+++ /dev/null
@@ -1,737 +0,0 @@
-<?php
-/**
- * XML file reader for the page data importer
- *
- * @file
- */
-
-/**
- * implements Special:Import
- * @ingroup SpecialPage
- */
-class WikiImporter {
-	private $reader = null;
-	private $mLogItemCallback, $mUploadCallback, $mRevisionCallback, $mPageCallback;
-	private $mSiteInfoCallback, $mTargetNamespace, $mPageOutCallback;
-	private $mDebug;
-
-	/**
-	 * Creates an ImportXMLReader drawing from the source provided
-	*/
-	function __construct( $source ) {
-		$this->reader = new XMLReader2();
-
-		stream_wrapper_register( 'uploadsource', 'UploadSourceAdapter' );
-		$id = UploadSourceAdapter::registerSource( $source );
-		$this->reader->open( "uploadsource://$id" );
-
-		// Default callbacks
-		$this->setRevisionCallback( array( $this, "importRevision" ) );
-		$this->setUploadCallback( array( $this, 'importUpload' ) );
-		$this->setLogItemCallback( array( $this, 'importLogItem' ) );
-		$this->setPageOutCallback( array( $this, 'finishImportPage' ) );
-	}
-
-	private function throwXmlError( $err ) {
-		$this->debug( "FAILURE: $err" );
-		wfDebug( "WikiImporter XML error: $err\n" );
-	}
-
-	private function debug( $data ) {
-		if( $this->mDebug ) {
-			wfDebug( "IMPORT: $data\n" );
-		}
-	}
-
-	private function warn( $data ) {
-		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" );
-		}
-	}
-
-	/**
-	 * Set debug mode...
-	 */
-	function setDebug( $debug ) {
-		$this->mDebug = $debug;
-	}
-
-	/**
-	 * Sets the action to perform as each new page in the stream is reached.
-	 * @param $callback callback
-	 * @return callback
-	 */
-	public function setPageCallback( $callback ) {
-		$previous = $this->mPageCallback;
-		$this->mPageCallback = $callback;
-		return $previous;
-	}
-
-	/**
-	 * Sets the action to perform as each page in the stream is completed.
-	 * Callback accepts the page title (as a Title object), a second object
-	 * with the original title form (in case it's been overridden into a
-	 * local namespace), and a count of revisions.
-	 *
-	 * @param $callback callback
-	 * @return callback
-	 */
-	public function setPageOutCallback( $callback ) {
-		$previous = $this->mPageOutCallback;
-		$this->mPageOutCallback = $callback;
-		return $previous;
-	}
-
-	/**
-	 * Sets the action to perform as each page revision is reached.
-	 * @param $callback callback
-	 * @return callback
-	 */
-	public function setRevisionCallback( $callback ) {
-		$previous = $this->mRevisionCallback;
-		$this->mRevisionCallback = $callback;
-		return $previous;
-	}
-
-	/**
-	 * Sets the action to perform as each file upload version is reached.
-	 * @param $callback callback
-	 * @return callback
-	 */
-	public function setUploadCallback( $callback ) {
-		$previous = $this->mUploadCallback;
-		$this->mUploadCallback = $callback;
-		return $previous;
-	}
-
-	/**
-	 * Sets the action to perform as each log item reached.
-	 * @param $callback callback
-	 * @return callback
-	 */
-	public function setLogItemCallback( $callback ) {
-		$previous = $this->mLogItemCallback;
-		$this->mLogItemCallback = $callback;
-		return $previous;
-	}
-
-	/**
-	 * Sets the action to perform when site info is encountered
-	 * @param $callback callback
-	 * @return callback
-	 */
-	public function setSiteInfoCallback( $callback ) {
-		$previous = $this->mSiteInfoCallback;
-		$this->mSiteInfoCallback = $callback;
-		return $previous;
-	}
-
-	/**
-	 * Set a target namespace to override the defaults
-	 */
-	public function setTargetNamespace( $namespace ) {
-		if( is_null( $namespace ) ) {
-			// Don't override namespaces
-			$this->mTargetNamespace = null;
-		} elseif( $namespace >= 0 ) {
-			// FIXME: Check for validity
-			$this->mTargetNamespace = intval( $namespace );
-		} else {
-			return false;
-		}
-	}
-
-	/**
-	 * Default per-revision callback, performs the import.
-	 * @param $revision WikiRevision
-	 */
-	public function importRevision( $revision ) {
-		$dbw = wfGetDB( DB_MASTER );
-		return $dbw->deadlockLoop( array( $revision, 'importOldRevision' ) );
-	}
-
-	/**
-	 * Default per-revision callback, performs the import.
-	 * @param $rev WikiRevision
-	 */
-	public function importLogItem( $rev ) {
-		$dbw = wfGetDB( DB_MASTER );
-		return $dbw->deadlockLoop( array( $rev, 'importLogItem' ) );
-	}
-
-	/**
-	 * Dummy for now...
-	 */
-	public function importUpload( $revision ) {
-		//$dbw = wfGetDB( DB_MASTER );
-		//return $dbw->deadlockLoop( array( $revision, 'importUpload' ) );
-		return false;
-	}
-
-	/**
-	 * Mostly for hook use
-	 */
-	public function finishImportPage( $title, $origTitle, $revCount, $sRevCount, $pageInfo ) {
-		$args = func_get_args();
-		return wfRunHooks( 'AfterImportPage', $args );
-	}
-
-	/**
-	 * Alternate per-revision callback, for debugging.
-	 * @param $revision WikiRevision
-	 */
-	public function debugRevisionHandler( &$revision ) {
-		$this->debug( "Got revision:" );
-		if( is_object( $revision->title ) ) {
-			$this->debug( "-- Title: " . $revision->title->getPrefixedText() );
-		} else {
-			$this->debug( "-- Title: <invalid>" );
-		}
-		$this->debug( "-- User: " . $revision->user_text );
-		$this->debug( "-- Timestamp: " . $revision->timestamp );
-		$this->debug( "-- Comment: " . $revision->comment );
-		$this->debug( "-- Text: " . $revision->text );
-	}
-
-	/**
-	 * Notify the callback function when a new <page> is reached.
-	 * @param $title Title
-	 */
-	function pageCallback( $title ) {
-		if( isset( $this->mPageCallback ) ) {
-			call_user_func( $this->mPageCallback, $title );
-		}
-	}
-
-	/**
-	 * Notify the callback function when a </page> is closed.
-	 * @param $title Title
-	 * @param $origTitle Title
-	 * @param $revCount Integer
-	 * @param $sucCount Int: number of revisions for which callback returned true
-	 * @param $pageInfo Array: associative array of page information
-	 */
-	private function pageOutCallback( $title, $origTitle, $revCount, $sucCount, $pageInfo ) {
-		if( isset( $this->mPageOutCallback ) ) {
-			$args = func_get_args();
-			call_user_func_array( $this->mPageOutCallback, $args );
-		}
-	}
-
-	/**
-	 * Notify the callback function of a revision
-	 * @param $revision A WikiRevision object
-	 */
-	private function revisionCallback( $revision ) {
-		if ( isset( $this->mRevisionCallback ) ) {
-			return call_user_func_array( $this->mRevisionCallback,
-					array( $revision, $this ) );
-		} else {
-			return false;
-		}
-	}
-
-	/**
-	 * Notify the callback function of a new log item
-	 * @param $revision A WikiRevision object
-	 */
-	private function logItemCallback( $revision ) {
-		if ( isset( $this->mLogItemCallback ) ) {
-			return call_user_func_array( $this->mLogItemCallback,
-					array( $revision, $this ) );
-		} else {
-			return false;
-		}
-	}
-
-	/**
-	 * Shouldn't something like this be built-in to XMLReader?
-	 * Fetches text contents of the current element, assuming
-	 * no sub-elements or such scary things.
-	 * @return string
-	 * @access private
-	 */
-	private function nodeContents() {
-		return $this->reader->nodeContents();
-	}
-
-	# --------------
-
-	/** Left in for debugging */
-	private function dumpElement() {
-		static $lookup = null;
-		if (!$lookup) {
-			$xmlReaderConstants = array(
-				"NONE",
-				"ELEMENT",
-				"ATTRIBUTE",
-				"TEXT",
-				"CDATA",
-				"ENTITY_REF",
-				"ENTITY",
-				"PI",
-				"COMMENT",
-				"DOC",
-				"DOC_TYPE",
-				"DOC_FRAGMENT",
-				"NOTATION",
-				"WHITESPACE",
-				"SIGNIFICANT_WHITESPACE",
-				"END_ELEMENT",
-				"END_ENTITY",
-				"XML_DECLARATION",
-				);
-			$lookup = array();
-
-			foreach( $xmlReaderConstants as $name ) {
-				$lookup[constant("XmlReader::$name")] = $name;
-			}
-		}
-
-		print( var_dump(
-			$lookup[$this->reader->nodeType],
-			$this->reader->name,
-			$this->reader->value
-		)."\n\n" );
-	}
-
-	/**
-	 * Primary entry point
-	 */
-	public function doImport() {
-		$this->reader->read();
-
-		if ( $this->reader->name != 'mediawiki' ) {
-			throw new MWException( "Expected <mediawiki> tag, got ".
-				$this->reader->name );
-		}
-		$this->debug( "<mediawiki> tag is correct." );
-
-		$this->debug( "Starting primary dump processing loop." );
-
-		$keepReading = $this->reader->read();
-		$skip = false;
-		while ( $keepReading ) {
-			$tag = $this->reader->name;
-			$type = $this->reader->nodeType;
-
-			if ( !wfRunHooks( 'ImportHandleToplevelXMLTag', $this->reader ) ) {
-				// Do nothing
-			} elseif ( $tag == 'mediawiki' && $type == XmlReader::END_ELEMENT ) {
-				break;
-			} elseif ( $tag == 'siteinfo' ) {
-				$this->handleSiteInfo();
-			} elseif ( $tag == 'page' ) {
-				$this->handlePage();
-			} elseif ( $tag == 'logitem' ) {
-				$this->handleLogItem();
-			} elseif ( $tag != '#text' ) {
-				$this->warn( "Unhandled top-level XML tag $tag" );
-
-				$skip = true;
-			}
-
-			if ($skip) {
-				$keepReading = $this->reader->next();
-				$skip = false;
-				$this->debug( "Skip" );
-			} else {
-				$keepReading = $this->reader->read();
-			}
-		}
-
-		return true;
-	}
-
-	private function handleSiteInfo() {
-		// Site info is useful, but not actually used for dump imports.
-		// Includes a quick short-circuit to save performance.
-		if ( ! $this->mSiteInfoCallback ) {
-			$this->reader->next();
-			return true;
-		}
-		throw new MWException( "SiteInfo tag is not yet handled, do not set mSiteInfoCallback" );
-	}
-
-	private function handleLogItem() {
-		$this->debug( "Enter log item handler." );
-		$logInfo = array();
-
-		// Fields that can just be stuffed in the pageInfo object
-		$normalFields = array( 'id', 'comment', 'type', 'action', 'timestamp',
-					'logtitle', 'params' );
-
-		while ( $this->reader->read() ) {
-			if ( $this->reader->nodeType == XmlReader::END_ELEMENT &&
-					$this->reader->name == 'logitem') {
-				break;
-			}
-
-			$tag = $this->reader->name;
-
-			if ( !wfRunHooks( 'ImportHandleLogItemXMLTag',
-						$this->reader, $logInfo ) ) {
-				// Do nothing
-			} elseif ( in_array( $tag, $normalFields ) ) {
-				$logInfo[$tag] = $this->nodeContents();
-			} elseif ( $tag == 'contributor' ) {
-				$logInfo['contributor'] = $this->handleContributor();
-			} elseif ( $tag != '#text' ) {
-				$this->warn( "Unhandled log-item XML tag $tag" );
-			}
-		}
-
-		$this->processLogItem( $logInfo );
-	}
-
-	private function processLogItem( $logInfo ) {
-		$revision = new WikiRevision;
-
-		$revision->setID( $logInfo['id'] );
-		$revision->setType( $logInfo['type'] );
-		$revision->setAction( $logInfo['action'] );
-		$revision->setTimestamp( $logInfo['timestamp'] );
-		$revision->setParams( $logInfo['params'] );
-		$revision->setTitle( Title::newFromText( $logInfo['logtitle'] ) );
-
-		if ( isset( $logInfo['comment'] ) ) {
-			$revision->setComment( $logInfo['comment'] );
-		}
-
-		if ( isset( $logInfo['contributor']['ip'] ) ) {
-			$revision->setUserIP( $logInfo['contributor']['ip'] );
-		}
-		if ( isset( $logInfo['contributor']['username'] ) ) {
-			$revision->setUserName( $logInfo['contributor']['username'] );
-		}
-
-		return $this->logItemCallback( $revision );
-	}
-
-	private function handlePage() {
-		// Handle page data.
-		$this->debug( "Enter page handler." );
-		$pageInfo = array( 'revisionCount' => 0, 'successfulRevisionCount' => 0 );
-
-		// Fields that can just be stuffed in the pageInfo object
-		$normalFields = array( 'title', 'id', 'redirect', 'restrictions' );
-
-		$skip = false;
-		$badTitle = false;
-
-		while ( $skip ? $this->reader->next() : $this->reader->read() ) {
-			if ( $this->reader->nodeType == XmlReader::END_ELEMENT &&
-					$this->reader->name == 'page') {
-				break;
-			}
-
-			$tag = $this->reader->name;
-
-			if ( $badTitle ) {
-				// The title is invalid, bail out of this page
-				$skip = true;
-			} elseif ( !wfRunHooks( 'ImportHandlePageXMLTag', array( $this->reader,
-						&$pageInfo ) ) ) {
-				// Do nothing
-			} elseif ( in_array( $tag, $normalFields ) ) {
-				$pageInfo[$tag] = $this->nodeContents();
-				if ( $tag == 'title' ) {
-					$title = $this->processTitle( $pageInfo['title'] );
-
-					if ( !$title ) {
-						$badTitle = true;
-						$skip = true;
-					}
-
-					$this->pageCallback( $title );
-					list( $pageInfo['_title'], $origTitle ) = $title;
-				}
-			} elseif ( $tag == 'revision' ) {
-				$this->handleRevision( $pageInfo );
-			} elseif ( $tag == 'upload' ) {
-				$this->handleUpload( $pageInfo );
-			} elseif ( $tag != '#text' ) {
-				$this->warn( "Unhandled page XML tag $tag" );
-				$skip = true;
-			}
-		}
-
-		$this->pageOutCallback( $pageInfo['_title'], $origTitle,
-					$pageInfo['revisionCount'],
-					$pageInfo['successfulRevisionCount'],
-					$pageInfo );
-	}
-
-	private function handleRevision( &$pageInfo ) {
-		$this->debug( "Enter revision handler" );
-		$revisionInfo = array();
-
-		$normalFields = array( 'id', 'timestamp', 'comment', 'minor', 'text' );
-
-		$skip = false;
-
-		while ( $skip ? $this->reader->next() : $this->reader->read() ) {
-			if ( $this->reader->nodeType == XmlReader::END_ELEMENT &&
-					$this->reader->name == 'revision') {
-				break;
-			}
-
-			$tag = $this->reader->name;
-
-			if ( !wfRunHooks( 'ImportHandleRevisionXMLTag', $this->reader,
-						$pageInfo, $revisionInfo ) ) {
-				// Do nothing
-			} elseif ( in_array( $tag, $normalFields ) ) {
-				$revisionInfo[$tag] = $this->nodeContents();
-			} elseif ( $tag == 'contributor' ) {
-				$revisionInfo['contributor'] = $this->handleContributor();
-			} elseif ( $tag != '#text' ) {
-				$this->warn( "Unhandled revision XML tag $tag" );
-				$skip = true;
-			}
-		}
-
-		$pageInfo['revisionCount']++;
-		if ( $this->processRevision( $pageInfo, $revisionInfo ) ) {
-			$pageInfo['successfulRevisionCount']++;
-		}
-	}
-
-	private function processRevision( $pageInfo, $revisionInfo ) {
-		$revision = new WikiRevision;
-
-		$revision->setID( $revisionInfo['id'] );
-		$revision->setText( $revisionInfo['text'] );
-		$revision->setTitle( $pageInfo['_title'] );
-		$revision->setTimestamp( $revisionInfo['timestamp'] );
-
-		if ( isset( $revisionInfo['comment'] ) ) {
-			$revision->setComment( $revisionInfo['comment'] );
-		}
-
-		if ( isset( $revisionInfo['minor'] ) )
-			$revision->setMinor( true );
-
-		if ( isset( $revisionInfo['contributor']['ip'] ) ) {
-			$revision->setUserIP( $revisionInfo['contributor']['ip'] );
-		}
-		if ( isset( $revisionInfo['contributor']['username'] ) ) {
-			$revision->setUserName( $revisionInfo['contributor']['username'] );
-		}
-
-		return $this->revisionCallback( $revision );
-	}
-
-	private function handleUpload( &$pageInfo ) {
-		$this->debug( "Enter upload handler" );
-		$uploadInfo = array();
-
-		$normalFields = array( 'timestamp', 'comment', 'filename', 'text',
-					'src', 'size' );
-
-		$skip = false;
-
-		while ( $skip ? $this->reader->next() : $this->reader->read() ) {
-			if ( $this->reader->nodeType == XmlReader::END_ELEMENT &&
-					$this->reader->name == 'upload') {
-				break;
-			}
-
-			$tag = $this->reader->name;
-
-			if ( !wfRunHooks( 'ImportHandleUploadXMLTag', $this->reader,
-						$pageInfo ) ) {
-				// Do nothing
-			} elseif ( in_array( $tag, $normalFields ) ) {
-				$uploadInfo[$tag] = $this->nodeContents();
-			} elseif ( $tag == 'contributor' ) {
-				$uploadInfo['contributor'] = $this->handleContributor();
-			} elseif ( $tag != '#text' ) {
-				$this->warn( "Unhandled upload XML tag $tag" );
-				$skip = true;
-			}
-		}
-
-		return $this->processUpload( $pageInfo, $uploadInfo );
-	}
-
-	private function processUpload( $pageInfo, $uploadInfo ) {
-		$revision = new WikiRevision;
-
-		$revision->setTitle( $pageInfo['_title'] );
-		$revision->setID( $uploadInfo['id'] );
-		$revision->setTimestamp( $uploadInfo['timestamp'] );
-		$revision->setText( $uploadInfo['text'] );
-		$revision->setFilename( $uploadInfo['filename'] );
-		$revision->setSrc( $uploadInfo['src'] );
-		$revision->setSize( intval( $uploadInfo['size'] ) );
-		$revision->setComment( $uploadInfo['comment'] );
-
-		if ( isset( $uploadInfo['contributor']['ip'] ) ) {
-			$revision->setUserIP( $uploadInfo['contributor']['ip'] );
-		}
-		if ( isset( $uploadInfo['contributor']['username'] ) ) {
-			$revision->setUserName( $uploadInfo['contributor']['username'] );
-		}
-
-		return $this->uploadCallback( $revision );
-	}
-
-	private function handleContributor() {
-		$fields = array( 'id', 'ip', 'username' );
-		$info = array();
-
-		while ( $this->reader->read() ) {
-			if ( $this->reader->nodeType == XmlReader::END_ELEMENT &&
-					$this->reader->name == 'contributor') {
-				break;
-			}
-
-			$tag = $this->reader->name;
-
-			if ( in_array( $tag, $fields ) ) {
-				$info[$tag] = $this->nodeContents();
-			}
-		}
-
-		return $info;
-	}
-
-	private function processTitle( $text ) {
-		$workTitle = $text;
-		$origTitle = Title::newFromText( $workTitle );
-
-		if( !is_null( $this->mTargetNamespace ) && !is_null( $origTitle ) ) {
-			$title = Title::makeTitle( $this->mTargetNamespace,
-				$origTitle->getDBkey() );
-		} else {
-			$title = Title::newFromText( $workTitle );
-		}
-
-		if( is_null( $title ) ) {
-			// Invalid page title? Ignore the page
-			$this->notice( "Skipping invalid page title '$workTitle'" );
-			return false;
-		} elseif( $title->getInterwiki() != '' ) {
-			$this->notice( "Skipping interwiki page title '$workTitle'" );
-			return false;
-		}
-
-		return array( $origTitle, $title );
-	}
-}
-
-/** This is a horrible hack used to keep source compatibility */
-class UploadSourceAdapter {
-	static $sourceRegistrations = array();
-
-	private $mSource;
-	private $mBuffer;
-	private $mPosition;
-
-	static function registerSource( $source ) {
-		$id = wfGenerateToken();
-
-		self::$sourceRegistrations[$id] = $source;
-
-		return $id;
-	}
-
-	function stream_open( $path, $mode, $options, &$opened_path ) {
-		$url = parse_url($path);
-		$id = $url['host'];
-
-		if ( !isset( self::$sourceRegistrations[$id] ) ) {
-			return false;
-		}
-
-		$this->mSource = self::$sourceRegistrations[$id];
-
-		return true;
-	}
-
-	function stream_read( $count ) {
-		$return = '';
-		$leave = false;
-
-		while ( !$leave && !$this->mSource->atEnd() &&
-				strlen($this->mBuffer) < $count ) {
-			$read = $this->mSource->readChunk();
-
-			if ( !strlen($read) ) {
-				$leave = true;
-			}
-
-			$this->mBuffer .= $read;
-		}
-
-		if ( strlen($this->mBuffer) ) {
-			$return = substr( $this->mBuffer, 0, $count );
-			$this->mBuffer = substr( $this->mBuffer, $count );
-		}
-
-		$this->mPosition += strlen($return);
-
-		return $return;
-	}
-
-	function stream_write( $data ) {
-		return false;
-	}
-
-	function stream_tell() {
-		return $this->mPosition;
-	}
-
-	function stream_eof() {
-		return $this->mSource->atEnd();
-	}
-
-	function url_stat() {
-		$result = array();
-
-		$result['dev'] = $result[0] = 0;
-		$result['ino'] = $result[1] = 0;
-		$result['mode'] = $result[2] = 0;
-		$result['nlink'] = $result[3] = 0;
-		$result['uid'] = $result[4] = 0;
-		$result['gid'] = $result[5] = 0;
-		$result['rdev'] = $result[6] = 0;
-		$result['size'] = $result[7] = 0;
-		$result['atime'] = $result[8] = 0;
-		$result['mtime'] = $result[9] = 0;
-		$result['ctime'] = $result[10] = 0;
-		$result['blksize'] = $result[11] = 0;
-		$result['blocks'] = $result[12] = 0;
-
-		return $result;
-	}
-}
-
-class XMLReader2 extends XMLReader {
-	function nodeContents() {
-		if( $this->isEmptyElement ) {
-			return "";
-		}
-		$buffer = "";
-		while( $this->read() ) {
-			switch( $this->nodeType ) {
-			case XmlReader::TEXT:
-			case XmlReader::SIGNIFICANT_WHITESPACE:
-				$buffer .= $this->value;
-				break;
-			case XmlReader::END_ELEMENT:
-				return $buffer;
-			}
-		}
-		return $this->close();
-	}
-}
-- 
2.20.1