From 46ebee74a6d4e193e46ec0e4c817d6dd09211f26 Mon Sep 17 00:00:00 2001
From: rillke <rillke@wikipedia.de>
Date: Fri, 13 Feb 2015 19:10:36 +0100
Subject: [PATCH] Detect duplicate archived files by SHA1 search on upload

- As of now, we detect duplicate files by a combination of sha1
  and file extension.
- Since multiple file extensions for the same MIME are permitted
  this method does not reliably work.

Bug: T74070
Change-Id: If13059441097799227f23ece36a96c8375f17aab
---
 RELEASE-NOTES-1.25                      |  2 ++
 includes/filerepo/file/ArchivedFile.php | 12 ++++++++++--
 includes/upload/UploadBase.php          |  8 ++++----
 3 files changed, 16 insertions(+), 6 deletions(-)

diff --git a/RELEASE-NOTES-1.25 b/RELEASE-NOTES-1.25
index 2c4288501e..8992ce0ce9 100644
--- a/RELEASE-NOTES-1.25
+++ b/RELEASE-NOTES-1.25
@@ -152,6 +152,8 @@ production.
 ** 'extend' method added to QuickTemplate to append additional values to any field of data array
 * (T86974) Several Title methods now load from the database when necessary
   (instead of returning incorrect results) even when the page ID is known.
+* (T74070) Duplicate search for archived files on file upload now omits the extension.
+  This requires the fa_sha1 field being populated.
 
 === Action API changes in 1.25 ===
 * (T67403) XML tag highlighting is now only performed for formats
diff --git a/includes/filerepo/file/ArchivedFile.php b/includes/filerepo/file/ArchivedFile.php
index 5b0d8e2baa..1d45428310 100644
--- a/includes/filerepo/file/ArchivedFile.php
+++ b/includes/filerepo/file/ArchivedFile.php
@@ -100,8 +100,9 @@ class ArchivedFile {
 	 * @param Title $title
 	 * @param int $id
 	 * @param string $key
+	 * @param string $sha1
 	 */
-	function __construct( $title, $id = 0, $key = '' ) {
+	function __construct( $title, $id = 0, $key = '', $sha1 = '' ) {
 		$this->id = -1;
 		$this->title = false;
 		$this->name = false;
@@ -136,7 +137,11 @@ class ArchivedFile {
 			$this->key = $key;
 		}
 
-		if ( !$id && !$key && !( $title instanceof Title ) ) {
+		if ( $sha1 ) {
+			$this->sha1 = $sha1;
+		}
+
+		if ( !$id && !$key && !( $title instanceof Title ) && !$sha1 ) {
 			throw new MWException( "No specifications provided to ArchivedFile constructor." );
 		}
 	}
@@ -162,6 +167,9 @@ class ArchivedFile {
 		if ( $this->title ) {
 			$conds['fa_name'] = $this->title->getDBkey();
 		}
+		if ( $this->sha1 ) {
+			$conds['fa_sha1'] = $this->sha1;
+		}
 
 		if ( !count( $conds ) ) {
 			throw new MWException( "No specific information for retrieving archived file" );
diff --git a/includes/upload/UploadBase.php b/includes/upload/UploadBase.php
index a8a38c70cb..fccb5e1a3a 100644
--- a/includes/upload/UploadBase.php
+++ b/includes/upload/UploadBase.php
@@ -677,10 +677,10 @@ abstract class UploadBase {
 		}
 
 		// Check dupes against archives
-		$archivedImage = new ArchivedFile( null, 0, "{$hash}.{$this->mFinalExtension}" );
-		if ( $archivedImage->getID() > 0 ) {
-			if ( $archivedImage->userCan( File::DELETED_FILE ) ) {
-				$warnings['duplicate-archive'] = $archivedImage->getName();
+		$archivedFile = new ArchivedFile( null, 0, '', $hash );
+		if ( $archivedFile->getID() > 0 ) {
+			if ( $archivedFile->userCan( File::DELETED_FILE ) ) {
+				$warnings['duplicate-archive'] = $archivedFile->getName();
 			} else {
 				$warnings['duplicate-archive'] = '';
 			}
-- 
2.20.1