Add TimedMediaHandler extension that allows display audio and video files in wiki pages, using the same syntax as for image files

1 files changed, 258 insertions, 0 deletions

diff --git a/extensions/TimedMediaHandler/SpecialOrphanedTimedText.php b/extensions/TimedMediaHandler/SpecialOrphanedTimedText.php
new file mode 100644
index 00000000..52e86bf4
--- /dev/null
+++ b/extensions/TimedMediaHandler/SpecialOrphanedTimedText.php
@@ -0,0 +1,258 @@
+<?php
+/**
+ * Implements Special:OrphanedTimedText
+ *
+ * @author Brian Wolff
+ * @file
+ * @ingroup SpecialPage
+ */
+
+/**
+ * Lists TimedText pages that don't have a corresponding video.
+ *
+ * @ingroup SpecialPage
+ */
+class SpecialOrphanedTimedText extends PageQueryPage {
+
+	/** @var Array with keys being names of valid files */
+	private $existingFiles;
+
+	public function __construct( $name = 'OrphanedTimedText' ) {
+		parent::__construct( $name );
+	}
+
+	/**
+	 * This is alphabetical, so sort ascending.
+	 */
+	public function sortDescending() {
+		return false;
+	}
+
+	/**
+	 * Should this be cached?
+	 *
+	 * This query is actually almost cheap given the current
+	 * number of things in TimedText namespace.
+	 */
+	public function isExpensive() {
+		return true;
+	}
+
+	/**
+	 * Main execution function
+	 *
+	 * @param $par String subpage
+	 */
+	public function execute( $par ) {
+		global $wgEnableLocalTimedText;
+
+		if ( !$wgEnableLocalTimedText ) {
+			$this->setHeaders();
+			$this->getOutput()->addWikiMsg( 'orphanedtimedtext-notimedtext' );
+			return;
+		} elseif ( !$this->canExecuteQuery() ) {
+			$this->setHeaders();
+			$this->outputHeader();
+			$this->getOutput()->addWikiMsg( 'orphanedtimedtext-unsupported' );
+			return;
+		}
+		return parent::execute( $par );
+	}
+
+	/**
+	 * Can we cache the results of this query?
+	 *
+	 * Only if we support the query.
+	 * @return bool
+	 */
+	public function isCacheable() {
+		return $this->canExecute();
+	}
+
+	/**
+	 * List in Special:SpecialPages?
+	 *
+	 * @return bool
+	 */
+	public function isListed() {
+		return $this->canExecute();
+	}
+
+	/**
+	 * Can we execute this special page?
+	 *
+	 * The query uses a mysql specific feature (substring_index), so disable on non mysql dbs.
+	 *
+	 * @return bool
+	 */
+	private function canExecuteQuery() {
+		$dbr = wfGetDB( DB_SLAVE );
+		return $dbr->getType() === 'mysql';
+	}
+
+	/**
+	 * Can we execute this special page
+	 *
+	 * That is, db is mysql, and TimedText namespace enabled.
+	 */
+	private function canExecute() {
+		global $wgEnableLocalTimedText;
+
+		return $this->canExecuteQuery() && $wgEnableLocalTimedText;
+	}
+
+	/**
+	 * Get query info
+	 *
+	 * The query here is meant to retrieve all pages in the TimedText namespace,
+	 * such that if you strip the last two extensions (e.g. Foo.bar.baz.en.srt -> Foo.bar.baz)
+	 * there is no corresponding img_name in image table. So if there is a page in TimedText
+	 * namespace named TimedText:My.Dog.webm.ceb.srt, it will include it in the list provided
+	 * that File:My.Dog.webm is not uploaded.
+	 *
+	 * TimedText does not support file redirects or foreign files, so we don't have
+	 * to worry about those.
+	 *
+	 * Potentially this should maybe also include pages not ending in
+	 * .<valid lang code>.srt . However, determining what a valid lang code
+	 * is, is pretty hard (although perhaps it could check if its [a-z]{2,3}
+	 * however then we've got things like roa-tara, cbk-zam, etc)
+	 * and TimedText throws away the final .srt extension and will work with
+	 * any extension, so things not ending in .srt arguably aren't oprhaned.
+	 *
+	 * @note This uses "substring_index" which is a mysql extension.
+	 * @return Array Standard query info values.
+	 */
+	function getQueryInfo() {
+		$tables = array( 'page', 'image' );
+		$fields = array(
+			'namespace' => 'page_namespace',
+			'title' => 'page_title',
+			'value' => 0,
+		);
+		$conds = array(
+			'img_name' => null,
+			'page_namespace' => NS_TIMEDTEXT,
+		);
+
+		// Now for the complicated bit
+		// Note: This bit is mysql specific. Probably could do something
+		// equivalent in postgress via split_part or regex substr,
+		// but my sql-fu is not good enough to figure out how to do
+		// this in standard sql, or in sqlite.
+		$baseCond = 'substr( page_title, 1, length( page_title ) - '
+			. "length( substring_index( page_title, '.' ,-2 ) ) - 1 )";
+		$joinConds = array(
+			'image' => array(
+				'LEFT OUTER JOIN',
+				 $baseCond . ' = img_name'
+			)
+		);
+		return array(
+			'tables' => $tables,
+			'fields' => $fields,
+			'conds' => $conds,
+			'join_conds' => $joinConds
+		);
+	}
+
+	public function getOrderFields() {
+		return array( 'namespace', 'title' );
+	}
+
+	/**
+	 * Is the TimedText page really orphaned?
+	 *
+	 * Given a title like "TimedText:Some bit here.webm.en.srt"
+	 * check to see if "File:Some bit here.webm" really exists (locally).
+	 * @return bool True if we should cross out the line.
+	 */
+	protected function existenceCheck( Title $title ) {
+		$fileTitle = $this->getCorrespondingFile( $title );
+		if ( !$fileTitle ) {
+			return $title && !$title->isKnown();
+		}
+		return !$title->isKnown() ||
+			( isset( $this->existingFiles[ $fileTitle->getDBKey() ] )
+			&& $this->existingFiles[$fileTitle->getDBKey()]->getHandler()
+			&& $this->existingFiles[$fileTitle->getDBKey()]->getHandler() instanceof TimedMediaHandler );
+	}
+
+	/**
+	 * Given a TimedText title, get the File title
+	 *
+	 * @return Title|null Title in File namespace. null on error.
+	 */
+	private function getCorrespondingFile( Title $timedText ) {
+		if ( !$timedText ) {
+			return false;
+		}
+		$titleParts = explode( '.', $timedText->getDBkey() );
+		array_pop( $titleParts );
+		array_pop( $titleParts );
+		$fileTitle = Title::makeTitleSafe( NS_FILE, implode( '.', $titleParts ) );
+		return $fileTitle;
+	}
+
+	/**
+	 * What group to include this page in on Special:SpecialPages
+	 * @return String
+	 */
+	protected function getGroupName() {
+		return 'media';
+	}
+
+	/**
+	 * Preprocess result to do existence checks all at once.
+	 *
+	 * @param $db Database
+	 * @param $res ResultWraper
+	 */
+	public function preprocessResults( $db, $res ) {
+		parent::preprocessResults( $db, $res );
+
+		if ( !$res->numRows() ) {
+			return;
+		}
+
+		$filesToLookFor = array();
+		foreach( $res as $row ) {
+			$title = Title::makeTitle( $row->namespace, $row->title );
+			$fileTitle = $this->getCorrespondingFile( $title );
+			if ( !$fileTitle ) {
+				continue;
+			}
+			$filesToLookFor[] = array( 'title' => $fileTitle, 'ignoreRedirect' => true );
+		}
+		$this->existingFiles = RepoGroup::singleton()->getLocalRepo()->findFiles( $filesToLookFor );
+		$res->seek( 0 );
+	}
+
+	/**
+	 * Format the result as a simple link to the page
+	 *
+	 * Based on parent class but with an existence check added.
+	 *
+	 * @param Skin $skin
+	 * @param object $row Result row
+	 * @return string
+	 */
+	public function formatResult( $skin, $row ) {
+		global $wgContLang;
+
+		$title = Title::makeTitleSafe( $row->namespace, $row->title );
+
+		if ( $title instanceof Title ) {
+			$text = $wgContLang->convert( $title->getPrefixedText() );
+			$link = Linker::link( $title, htmlspecialchars( $text ) );
+			if ( $this->existenceCheck( $title ) ) {
+				// File got uploaded since this page was cached
+				$link = '<del>' . $link . '</del>';
+			}
+			return $link;
+		} else {
+			return Html::element( 'span', array( 'class' => 'mw-invalidtitle' ),
+				Linker::getInvalidTitleDescription( $this->getContext(), $row->namespace, $row->title ) );
+		}
+	}
+}