* Special:Import/importDump fixes: report XML parse errors, accept <minor/>

RELEASE-NOTES

@@ -112,6 +112,7 @@ fully support the editing toolbar, but was found to be too confusing.
 * Correct blob caching to reduce redundant blob loads on backups
 * (bug 3485) Fix bogus warning about filename capitalization when off
 * (bug 2792) Update rebuildrecentchanges.inc for new schema
+* Special:Import/importDump fixes: report XML parse errors, accept <minor/>
 
 
 === Caveats ===

includes/SpecialImport.php

@@ -128,6 +128,7 @@ class WikiRevision {
 	var $user_text = "";
 	var $text = "";
 	var $comment = "";
+	var $minor = false;
 	
 	function setTitle( $text ) {
 		$this->title = Title::newFromText( $text );
@@ -154,6 +155,10 @@ class WikiRevision {
 		$this->comment = $text;
 	}
 	
+	function setMinor( $minor ) {
+		$this->minor = (bool)$minor;
+	}
+	
 	function getTitle() {
 		return $this->title;
 	}
@@ -173,6 +178,10 @@ class WikiRevision {
 	function getComment() {
 		return $this->comment;
 	}
+	
+	function getMinor() {
+		return $this->minor;
+	}
 
 	function importOldRevision() {
 		$fname = "WikiImporter::importOldRevision";
@@ -211,7 +220,7 @@ class WikiRevision {
 			'user'       => $userId,
 			'user_text'  => $userText,
 			'timestamp'  => $this->timestamp,
-			'minor_edit' => 0
+			'minor_edit' => $this->minor,
 			) );
 		$revId = $revision->insertOn( $dbw );
 		$article->updateIfNewerOn( $dbw, $revision );
@@ -239,6 +248,7 @@ class WikiImporter {
 	
 	function throwXmlError( $err ) {
 		$this->debug( "FAILURE: $err" );
+		wfDebug( "WikiImporter XML error: $err\n" );
 	}
 	
 	# --------------
@@ -256,11 +266,14 @@ class WikiImporter {
 		xml_set_object( $parser, &$this );
 		xml_set_element_handler( $parser, "in_start", "" );
 		
+		$offset = 0; // for context extraction on error reporting
 		do {
 			$chunk = $this->mSource->readChunk();
 			if( !xml_parse( $parser, $chunk, $this->mSource->atEnd() ) ) {
-				return new WikiXmlError( $parser );
+				wfDebug( "WikiImporter::doImport encountered XML parsing error\n" );
+				return new WikiXmlError( $parser, 'XML import parse failure', $chunk, $offset );
 			}
+			$offset += strlen( $chunk );
 		} while( $chunk !== false && !$this->mSource->atEnd() );
 		xml_parser_free( $parser );
 		
@@ -380,6 +393,7 @@ class WikiImporter {
 		$this->debug( "in_siteinfo $name" );
 		switch( $name ) {
 		case "sitename":
+		case "base":
 		case "generator":
 		case "case":
 		case "namespaces":
@@ -466,6 +480,9 @@ class WikiImporter {
 		case "comment":
 			$this->workRevision->setComment( $this->appenddata );
 			break;
+		case "minor":
+			$this->workRevision->setMinor( true );
+			break;
 		default:
 			$this->debug( "Bad append: {$this->appendfield}" );
 		}
@@ -479,6 +496,7 @@ class WikiImporter {
 		case "id":
 		case "timestamp":
 		case "comment":
+		case "minor":
 		case "text":
 			$this->parenttag = "revision";
 			$this->appendfield = $name;
@@ -514,6 +532,7 @@ class WikiImporter {
 		switch( $name ) {
 		case "username":
 		case "ip":
+		case "id":
 			$this->parenttag = "contributor";
 			$this->appendfield = $name;
 			xml_set_element_handler( $parser, "in_nothing", "out_append" );

includes/WikiError.php

@@ -89,15 +89,36 @@ class WikiXmlError extends WikiError {
 	 * @param resource $parser
 	 * @param string $message
 	 */
-	function WikiXmlError( $parser, $message = '' ) {
+	function WikiXmlError( $parser, $message = 'XML parsing error', $context = null, $offset = 0 ) {
 		$this->mXmlError = xml_get_error_code( $parser );
+		$this->mColumn = xml_get_current_column_number( $parser );
+		$this->mLine = xml_get_current_line_number( $parser );
+		$this->mByte = xml_get_current_byte_index( $parser );
+		$this->mContext = $this->_extractContext( $context, $offset );
 		$this->mMessage = $message;
 		xml_parser_free( $parser );
+		wfDebug( "WikiXmlError: " . $this->getMessage() . "\n" );
 	}
 
 	/** @return string */	
 	function getMessage() {
-		return $this->mMessage . ': ' . xml_error_string( $this->mXmlError );
+		return sprintf( '%s at line %d, col %d (byte %d%s): %s',
+			$this->mMessage,
+			$this->mLine,
+			$this->mColumn,
+			$this->mByte,
+			$this->mContext,
+			xml_error_string( $this->mXmlError ) );
+	}
+	
+	function _extractContext( $context, $offset ) {
+		if( is_null( $context ) ) {
+			return null;
+		} else {
+			// Hopefully integer overflow will be handled transparently here
+			$inlineOffset = $this->mByte - $offset;
+			return '; "' . substr( $context, $inlineOffset, 16 ) . '"';
+		}
 	}
 }
 

maintenance/importDump.php

@@ -88,12 +88,12 @@ class BackupReader {
 			$filename = 'compress.zlib://' . $filename;
 		}
 		$file = fopen( $filename, 'rt' );
-		$this->importFromHandle( $file );
+		return $this->importFromHandle( $file );
 	}
 
 	function importFromStdin() {
 		$file = fopen( 'php://stdin', 'rt' );
-		$this->importFromHandle( $file );
+		return $this->importFromHandle( $file );
 	}
 
 	function importFromHandle( $handle ) {
@@ -106,7 +106,7 @@ class BackupReader {
 		$this->importCallback =  $importer->setRevisionCallback(
 			array( &$this, 'handleRevision' ) );
 
-		$importer->doImport();
+		return $importer->doImport();
 	}
 }
 
@@ -122,9 +122,15 @@ if( isset( $options['dry-run'] ) ) {
 }
 
 if( isset( $args[0] ) ) {
-	$reader->importFromFile( $args[0] );
+	$result = $reader->importFromFile( $args[0] );
 } else {
-	$reader->importFromStdin();
+	$result = $reader->importFromStdin();
+}
+
+if( WikiError::isError( $result ) ) {
+	echo $result->getMessage() . "\n";
+} else {
+	echo "Done!\n";
 }
 
 ?>