X-Git-Url: https://git.heureux-cyclage.org/?a=blobdiff_plain;f=includes%2FImport.php;h=06355ed11ca4befaabbeaec68516d2599b4cd515;hb=45788085afaa7be7b606c08ccc3cba06d224496b;hp=36028eab9256b88a8750f9701bbcd698ea21a8c6;hpb=e6696ed61129be255ba5e6032c2eadcab85ed3b8;p=lhc%2Fweb%2Fwiklou.git

diff --git a/includes/Import.php b/includes/Import.php
index 36028eab92..06355ed11c 100644
--- a/includes/Import.php
+++ b/includes/Import.php
@@ -42,6 +42,8 @@ class WikiImporter {
 	private $config;
 	/** @var ImportTitleFactory */
 	private $importTitleFactory;
+	/** @var array */
+	private $countableCache = array();
 
 	/**
 	 * Creates an ImportXMLReader drawing from the source provided
@@ -67,6 +69,7 @@ class WikiImporter {
 		}
 
 		// Default callbacks
+		$this->setPageCallback( array( $this, 'beforeImportPage' ) );
 		$this->setRevisionCallback( array( $this, "importRevision" ) );
 		$this->setUploadCallback( array( $this, 'importUpload' ) );
 		$this->setLogItemCallback( array( $this, 'importLogItem' ) );
@@ -288,6 +291,19 @@ class WikiImporter {
 		$this->mImportUploads = $import;
 	}
 
+	/**
+	 * Default per-page callback. Sets up some things related to site statistics
+	 * @param array $titleAndForeignTitle Two-element array, with Title object at
+	 * index 0 and ForeignTitle object at index 1
+	 * @return bool
+	 */
+	public function beforeImportPage( $titleAndForeignTitle ) {
+		$title = $titleAndForeignTitle[0];
+		$page = WikiPage::factory( $title );
+		$this->countableCache['title_' . $title->getPrefixedText()] = $page->isCountable();
+		return true;
+	}
+
 	/**
 	 * Default per-revision callback, performs the import.
 	 * @param WikiRevision $revision
@@ -349,6 +365,31 @@ class WikiImporter {
 	 */
 	public function finishImportPage( $title, $foreignTitle, $revCount,
 			$sRevCount, $pageInfo ) {
+
+		// Update article count statistics (T42009)
+		// The normal counting logic in WikiPage->doEditUpdates() is designed for
+		// one-revision-at-a-time editing, not bulk imports. In this situation it
+		// suffers from issues of slave lag. We let WikiPage handle the total page
+		// and revision count, and we implement our own custom logic for the
+		// article (content page) count.
+		$page = WikiPage::factory( $title );
+		$page->loadPageData( 'fromdbmaster' );
+		$content = $page->getContent();
+		if ( $content === null ) {
+			wfDebug( __METHOD__ . ': Skipping article count adjustment for ' . $title .
+				' because WikiPage::getContent() returned null' );
+		} else {
+			$editInfo = $page->prepareContentForEdit( $content );
+			$countKey = 'title_' . $title->getPrefixedText();
+			$countable = $page->isCountable( $editInfo );
+			if ( array_key_exists( $countKey, $this->countableCache ) &&
+				$countable != $this->countableCache[ $countKey ] ) {
+				DeferredUpdates::addUpdate( SiteStatsUpdate::factory( array(
+					'articles' => ( (int)$countable - (int)$this->countableCache[ $countKey ] )
+				) ) );
+			}
+		}
+
 		$args = func_get_args();
 		return Hooks::run( 'AfterImportPage', $args );
 	}
@@ -441,7 +482,8 @@ class WikiImporter {
 	/**
 	 * Retrieves the contents of the named attribute of the current element.
 	 * @param string $attr The name of the attribute
-	 * @return string The value of the attribute or an empty string if it is not set in the current element.
+	 * @return string The value of the attribute or an empty string if it is not set in the current
+	 * element.
 	 */
 	public function nodeAttribute( $attr ) {
 		return $this->reader->getAttribute( $attr );
@@ -497,36 +539,48 @@ class WikiImporter {
 
 		$keepReading = $this->reader->read();
 		$skip = false;
-		while ( $keepReading ) {
-			$tag = $this->reader->name;
-			$type = $this->reader->nodeType;
-
-			if ( !Hooks::run( 'ImportHandleToplevelXMLTag', array( $this ) ) ) {
-				// 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;
-			}
+		$rethrow = null;
+		try {
+			while ( $keepReading ) {
+				$tag = $this->reader->name;
+				$type = $this->reader->nodeType;
+
+				if ( !Hooks::run( 'ImportHandleToplevelXMLTag', array( $this ) ) ) {
+					// 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();
+				if ( $skip ) {
+					$keepReading = $this->reader->next();
+					$skip = false;
+					$this->debug( "Skip" );
+				} else {
+					$keepReading = $this->reader->read();
+				}
 			}
+		} catch ( Exception $ex ) {
+			$rethrow = $ex;
 		}
 
+		// finally
 		libxml_disable_entity_loader( $oldDisable );
+		$this->reader->close();
+
+		if ( $rethrow ) {
+			throw $rethrow;
+		}
+
 		return true;
 	}
 
@@ -1532,7 +1586,6 @@ class WikiRevision {
 					$this->title->getPrefixedText() . "]], timestamp " . $this->timestamp . "\n" );
 				return false;
 			}
-			$oldcountable = $page->isCountable();
 		}
 
 		# @todo FIXME: Use original rev_id optionally (better for backups)
@@ -1555,10 +1608,11 @@ class WikiRevision {
 
 		if ( $changed !== false && !$this->mNoUpdates ) {
 			wfDebug( __METHOD__ . ": running updates\n" );
+			// countable/oldcountable stuff is handled in WikiImporter::finishImportPage
 			$page->doEditUpdates(
 				$revision,
 				$userObj,
-				array( 'created' => $created, 'oldcountable' => $oldcountable )
+				array( 'created' => $created, 'oldcountable' => 'no-change' )
 			);
 		}
 
@@ -1621,6 +1675,7 @@ class WikiRevision {
 				RepoGroup::singleton()->getLocalRepo(), $archiveName );
 		} else {
 			$file = wfLocalFile( $this->getTitle() );
+			$file->load( File::READ_LATEST );
 			wfDebug( __METHOD__ . 'Importing new file as ' . $file->getName() . "\n" );
 			if ( $file->exists() && $file->getTimestamp() > $this->getTimestamp() ) {
 				$archiveName = $file->getTimestamp() . '!' . $file->getName();
@@ -1692,7 +1747,7 @@ class WikiRevision {
 
 		// @todo FIXME!
 		$src = $this->getSrc();
-		$data = Http::get( $src );
+		$data = Http::get( $src, array(), __METHOD__ );
 		if ( !$data ) {
 			wfDebug( "IMPORT: couldn't fetch source $src\n" );
 			fclose( $f );
@@ -1849,7 +1904,7 @@ class ImportStreamSource implements ImportSource {
 		# 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, array( 'followRedirects' => true ) );
+		$data = Http::request( $method, $url, array( 'followRedirects' => true ), __METHOD__ );
 		if ( $data !== false ) {
 			$file = tmpfile();
 			fwrite( $file, $data );