Implemented MwDump::touchModified() to touch timestamp for altered re…

…vision texts via their base36sha1.
Dmitri-Sintsov · Aug 11, 2014 · 079cc19 · 079cc19
1 parent 7759f43
commit 079cc19
Show file tree

Hide file tree

Showing 4 changed files with 42 additions and 11 deletions.
diff --git a/MwDumpProcessor.php b/MwDumpProcessor.php
@@ -62,14 +62,27 @@ public function filterFiles( PageModel $pageModel ) {
 		return $pageModel->getSubProp( 'pageProps', 'ns' ) === PageModel::NS_FILE;
 	}
 
+	public function touchModified( PageModel $pageModel ) {
+		if ( $pageModel->getSubProp( 'revProps', 'base36sha1' ) !==
+			$pageModel->getSubProp( 'textProps', 'base36sha1' ) ) {
+			$pageModel->setTimestamp( time() );
+			$pageModel->setRevBase36Sha1( $pageModel->getSubProp( 'textProps', 'base36sha1' ) );
+			# Q\Dbg\log('modifiedPage',$pageModel);
+			return true;
+		} else {
+			return false;
+		}
+	}
+
 	public function process( $inFileName, $outFileName ) {
 		$dumpParser = MwDumpParser::newFromUri( $inFileName );
 		$dumpParser->setOutputFileName( $outFileName );
 		# $dumpParser->addPageHook( array( $this, 'noopHook' ) );
 		# $dumpParser->addPageHook( array( $this, 'addArchiveCategory' ) );
 		# $dumpParser->addPageHook( array( $this, 'renameCategory' ) );
 		# $dumpParser->addPageHook( array( $this, 'forceRenameCategory' ) );
-		$dumpParser->addPageHook( array( $this, 'filterFiles' ) );
+		# $dumpParser->addPageHook( array( $this, 'filterFiles' ) );
+		$dumpParser->addPageHook( array( $this, 'touchModified' ) );
 		$dumpParser->parse();
 	}
 

diff --git a/model/PageModel.php b/model/PageModel.php
@@ -9,10 +9,12 @@ class PageModel extends Q\FeaturesObject {
 
 	protected $pageProps;
 	protected $revProps;
+	protected $textProps;
 
 	function __construct() {
 		$this->pageProps = new \stdClass();
 		$this->revProps = new \stdClass();
+		$this->textProps = new \stdClass();
 	}
 
 	public function setTitle( $title ) {
@@ -43,13 +45,17 @@ public function setParentId( $parentId ) {
 	}
 
 	public function setTimestamp( $ts ) {
-		$ts = Q\Gl::timeStamp( Q\Gl::TS_UNIX, $ts );
+		$ts = is_numeric( $ts ) ? $ts : Q\Gl::timeStamp( Q\Gl::TS_UNIX, $ts );
 		if ( $ts === false ) {
 			SdvException::throwRecoverable( 'Invalid timestamp of revision', __METHOD__, $ts );
 		}
 		$this->revProps->ts = $ts;
 	}
 
+	public function setUserIP( $userIP ) {
+		$this->revProps->userIP = $userIP;
+	}
+
 	public function setUserName( $userName ) {
 		$this->revProps->userName = $userName;
 	}
@@ -72,8 +78,8 @@ public function setTextAttrs( $textAttrs ) {
 
 	public function setText( $text ) {
 		$this->revProps->text = $text;
-		$this->revProps->base36sha1 = Q\Gl::baseConvert( sha1( $text ), 16, 36, 31 );
-		$this->revProps->textAttrs->bytes = strlen( $text );
+		$this->textProps->base36sha1 = Q\Gl::baseConvert( sha1( $text ), 16, 36, 31 );
+		$this->textProps->bytes = strlen( $text );
 	}
 
 	public function setRevBase36Sha1( $base36sha1 ) {
@@ -104,17 +110,28 @@ public function getTagArray() {
 		if ( property_exists( $this->revProps, 'parentId' ) ) {
 			$revision[] = array( '@tag' => 'parentid', $this->revProps->parentId );
 		}
+		if ( !property_exists( $this->revProps, 'textAttrs' ) ) {
+			Q\SdvException::throwError( 'Revision property textAttrs does not exists', __METHOD__, $this->revProps );
+		}
 		$textTagArray = (array) $this->revProps->textAttrs;
 		$textTagArray['@tag'] = 'text';
 		if ( $this->revProps->text !== '' ) {
 			$textTagArray[] = $this->revProps->text;
 		}
+		$contributor = array( '@tag' => 'contributor' );
+		if ( property_exists( $this->revProps, 'userIP' ) ) {
+			array_push( $contributor,
+				array( '@tag' => 'ip', $this->revProps->userIP )
+			);
+		} else {
+			array_push( $contributor,
+				array( '@tag' => 'username', $this->revProps->userName ),
+				array( '@tag' => 'id', $this->revProps->uid )
+			);
+		}
 		array_push( $revision,
 			array( '@tag' => 'timestamp', Q\Gl::timeStamp( Q\Gl::TS_ISO_8601, $this->revProps->ts ) ),
-			array( '@tag' => 'contributor',
-				array( '@tag' => 'username', $this->revProps->userName ),
-				array( '@tag' => 'id', $this->revProps->uid ),
-			)
+			$contributor
 		);
 		if ( property_exists( $this->revProps, 'isMinor' ) &&
 				$this->revProps->isMinor ) {

diff --git a/parser/PageParser.php b/parser/PageParser.php
@@ -15,8 +15,9 @@ class PageParser extends Q\XmlOneLevelParser {
 		'revision/id' => array( self::TAG_MANDATORY, 'setRevisionId' ),
 		'revision/parentid' => array( self::TAG_OPTIONAL, 'setParentId' ),
 		'revision/timestamp' => array( self::TAG_MANDATORY, 'setTimestamp' ),
-		'revision/contributor/username' => array( self::TAG_MANDATORY, 'setUserName' ),
-		'revision/contributor/id' => array( self::TAG_MANDATORY, 'setUserId' ),
+		'revision/contributor/ip' => array( self::TAG_OPTIONAL, 'setUserIP' ),
+		'revision/contributor/username' => array( self::TAG_OPTIONAL, 'setUserName' ),
+		'revision/contributor/id' => array( self::TAG_OPTIONAL, 'setUserId' ),
 		'revision/comment' => array( self::TAG_OPTIONAL, 'setRevComment' ),
 		'revision/minor' => array( self::TAG_OPTIONAL, 'setRevMinor' ),
 		'revision/text' => array( self::TAG_MANDATORY, array( 'parsetag_text' ), 'logNameAttrVal' ),

diff --git a/sdv_engine.txt b/sdv_engine.txt
@@ -1 +1 @@
-e:\www\questpc
+/var/www/questpc