From 687c5b557778885a58902abc0643db7017e2f082 Mon Sep 17 00:00:00 2001
From: Aaron Schulz <aschulz@wikimedia.org>
Date: Wed, 27 Nov 2013 10:39:50 -0800
Subject: [PATCH] Optimized LocalRepo::findFiles()

Change-Id: Icea9c416c973a6b83c53e5bc201aed4528e9dd34
---
 includes/filerepo/FileRepo.php  |  4 +-
 includes/filerepo/LocalRepo.php | 84 +++++++++++++++++++++++++++++++++
 2 files changed, 87 insertions(+), 1 deletion(-)

diff --git a/includes/filerepo/FileRepo.php b/includes/filerepo/FileRepo.php
index 8611238d8e..466abe88df 100644
--- a/includes/filerepo/FileRepo.php
+++ b/includes/filerepo/FileRepo.php
@@ -473,7 +473,9 @@ class FileRepo {
 	 *     $findItem = array( 'title' => $title, 'private' => true );
 	 *     $findBatch = array( $findItem );
 	 *     $repo->findFiles( $findBatch );
-	 * @return array
+	 *
+	 *    No title should appear in $items twice, as the result use titles as keys
+	 * @return array (Map of file names => File objects) for matches
 	 */
 	public function findFiles( array $items ) {
 		$result = array();
diff --git a/includes/filerepo/LocalRepo.php b/includes/filerepo/LocalRepo.php
index 49c2b8fa0e..4bbcc7d84f 100644
--- a/includes/filerepo/LocalRepo.php
+++ b/includes/filerepo/LocalRepo.php
@@ -229,6 +229,90 @@ class LocalRepo extends FileRepo {
 		return $id;
 	}
 
+	public function findFiles( array $items ) {
+		$finalFiles = array(); // map of (DB key => corresponding File) for matches
+
+		$searchSet = array(); // map of (DB key => normalized search params)
+		foreach ( $items as $item ) {
+			$title = is_array( $item )
+				? File::normalizeTitle( $item['title'] )
+				: File::normalizeTitle( $item );
+			if ( $title ) { // valid title
+				$searchSet[$title->getDbKey()] = ( is_array( $item ) ? $item : array() );
+			}
+		}
+
+		$fileMatchesSearch = function( File $file, array $search ) {
+			// Note: file name comparison done elsewhere (to handle redirects)
+			return (
+				$file->exists() &&
+				(
+					( empty( $search['time'] ) && !$file->isOld() ) ||
+					( !empty( $search['time'] ) && $search['time'] === $file->getTimestamp() )
+				) &&
+				( !empty( $search['private'] ) || !$file->isDeleted( File::DELETED_FILE ) ) &&
+				$file->userCan( File::DELETED_FILE )
+			);
+		};
+
+		$repo = $this;
+		$applyMatchingFiles = function( ResultWrapper $res, &$searchSet, &$finalFiles )
+			use ( $repo, $fileMatchesSearch )
+		{
+			foreach ( $res as $row ) {
+				$possFile = $repo->newFileFromRow( $row );
+				$dbKey = $possFile->getName();
+				// There must have been a search for this DB Key
+				if ( $fileMatchesSearch( $possFile, $searchSet[$dbKey] ) ) {
+					$finalFiles[$dbKey] = $possFile;
+					unset( $searchSet[$dbKey] );
+				}
+			}
+		};
+
+		$dbr = $this->getSlaveDB();
+
+		// Query image table
+		$imgNames = array_keys( $searchSet );
+		if ( count( $imgNames ) ) {
+			$res = $dbr->select( 'image',
+				LocalFile::selectFields(), array( 'img_name' => $imgNames ), __METHOD__ );
+			$applyMatchingFiles( $res, $searchSet, $finalFiles );
+		}
+
+		// Query old image table
+		$oiConds = array(); // WHERE clause array for each file
+		foreach ( $searchSet as $dbKey => $search ) {
+			if ( isset( $search['params']['time'] ) ) {
+				$oiConds[] = $dbr->makeList( array( 'oi_name' => $dbKey,
+					'oi_timestamp' => $dbr->timestamp( $search['params']['time'] ) ), LIST_AND );
+			}
+		}
+		if ( count( $oiConds ) ) {
+			$res = $dbr->select( 'oldimage',
+				OldLocalFile::selectFields(), $dbr->makeList( $oiConds, LIST_OR ), __METHOD__ );
+			$applyMatchingFiles( $res, $searchSet, $finalFiles );
+		}
+
+		// Check for redirects...
+		foreach ( $searchSet as $dbKey => $search ) {
+			if ( !empty( $search['ignoreRedirect'] ) ) {
+				continue;
+			}
+			$title = File::normalizeTitle( $dbKey );
+			$redir = $this->checkRedirect( $title ); // hopefully hits memcached
+			if ( $redir && $redir->getNamespace() == NS_FILE ) {
+				$possFile = $this->newFile( $redir );
+				if ( $possFile && $fileMatchesSearch( $possFile, $search ) ) {
+					$possFile->redirectedFrom( $title->getDBkey() );
+					$finalFiles[$dbKey] = $possFile;
+				}
+			}
+		}
+
+		return $finalFiles;
+	}
+
 	/**
 	 * Get an array or iterator of file objects for files that have a given
 	 * SHA-1 content hash.
-- 
2.20.1