From 97c8de184262d9107c78232563626d7ae24f8750 Mon Sep 17 00:00:00 2001
From: idradm <robakadam@gmail.com>
Date: Tue, 8 Sep 2015 10:25:10 +0200
Subject: [PATCH 1/2] DAT-3207 add mw namespaces aliases to image filename
 sanitizer

---
 services/Helpers/ImageFilenameSanitizer.php | 30 ++++++++++++++-------
 tests/ImageFilenameSanitizerTest.php        |  8 +++++-
 2 files changed, 28 insertions(+), 10 deletions(-)

diff --git a/services/Helpers/ImageFilenameSanitizer.php b/services/Helpers/ImageFilenameSanitizer.php
index 3536cee..f98aaba 100644
--- a/services/Helpers/ImageFilenameSanitizer.php
+++ b/services/Helpers/ImageFilenameSanitizer.php
@@ -20,6 +20,7 @@ class ImageFilenameSanitizer {
 		if ( is_null( self::$instance ) ) {
 			self::$instance = new self;
 		}
+
 		return self::$instance;
 	}
 
@@ -28,23 +29,30 @@ class ImageFilenameSanitizer {
 	 * Used as local cache for getting string to remove
 	 */
 	private function getFilePrefixRegex( $contLang ) {
+		global $wgNamespaceAliases;
 		$langCode = $contLang->getCode();
-		if ( empty( $this->filePrefixRegex[$langCode] ) ) {
-			$fileNamespaces = [ \MWNamespace::getCanonicalName( NS_FILE ), $contLang->getNamespaces()[NS_FILE] ];
+		if ( empty( $this->filePrefixRegex[ $langCode ] ) ) {
+			$fileNamespaces = [
+				\MWNamespace::getCanonicalName( NS_FILE ),
+				$contLang->getNamespaces()[ NS_FILE ],
+			];
 
-			$aliases = $contLang->getNamespaceAliases();
+			$aliases = array_merge( $contLang->getNamespaceAliases(), $wgNamespaceAliases );
 			foreach ( $aliases as $alias => $namespaceId ) {
-				if ( $namespaceId == NS_FILE )
+				if ( $namespaceId == NS_FILE ) {
 					$fileNamespaces [] = $alias;
+				}
 			}
-			$this->filePrefixRegex[$langCode] = '^(' . implode( '|', $fileNamespaces ) . '):';
+			$this->filePrefixRegex[ $langCode ] = '^(' . implode( '|', $fileNamespaces ) . '):';
 		}
-		return $this->filePrefixRegex[$langCode];
+
+		return $this->filePrefixRegex[ $langCode ];
 	}
 
 	/**
 	 * @param $filename string
 	 * @param $contLang \Language
+	 *
 	 * @return mixed
 	 */
 	public function sanitizeImageFileName( $filename, $contLang ) {
@@ -54,16 +62,18 @@ class ImageFilenameSanitizer {
 
 		foreach ( $textLines as $potentialFilename ) {
 			$filename = $this->extractFilename( $potentialFilename, $filePrefixRegex );
-			if ($filename) {
+			if ( $filename ) {
 				return $filename;
 			}
 
 		}
+
 		return $plainText;
 	}
 
 	/**
 	 * @param $filename
+	 *
 	 * @return string
 	 */
 	private function convertToPlainText( $filename ) {
@@ -71,20 +81,22 @@ class ImageFilenameSanitizer {
 		$filename = strip_tags( $filename );
 		// replace the surrounding whitespace
 		$filename = trim( $filename );
+
 		return $filename;
 	}
 
 	/**
 	 * @param $potentialFilename
 	 * @param $filePrefixRegex
+	 *
 	 * @return string|null
 	 */
 	private function extractFilename( $potentialFilename, $filePrefixRegex ) {
 		$trimmedFilename = trim( $potentialFilename, "[]" );
 		$unprefixedFilename = mb_ereg_replace( $filePrefixRegex, "", $trimmedFilename );
 		$filenameParts = explode( '|', $unprefixedFilename );
-		if ( !empty( $filenameParts[0] ) ) {
-			return $filenameParts[0];
+		if ( !empty( $filenameParts[ 0 ] ) ) {
+			return $filenameParts[ 0 ];
 		}
 
 		return null;
diff --git a/tests/ImageFilenameSanitizerTest.php b/tests/ImageFilenameSanitizerTest.php
index 9961808..7917df6 100644
--- a/tests/ImageFilenameSanitizerTest.php
+++ b/tests/ImageFilenameSanitizerTest.php
@@ -175,7 +175,13 @@ class ImageFilenameSanitizerTest extends WikiaBaseTest {
 				'en',
 				'文件名óśłżźćńę?.jpg',
 				'Content of gallery with diacritics and UTF characters'
-			]
+			],
+			[
+				'[[Image:image.jpg|300px|lorem ipsum]]',
+				'es',
+				'image.jpg',
+				'Link to filename with canonical namespace, width and caption on a non-EN wiki '
+			],
 		];
 	}
 }

From c5adee4ff706d4a91bf7855702513e9975e54323 Mon Sep 17 00:00:00 2001
From: idradm <robakadam@gmail.com>
Date: Tue, 8 Sep 2015 10:26:48 +0200
Subject: [PATCH 2/2] DAT-3207 tests update

---
 tests/ImageFilenameSanitizerTest.php | 8 ++++----
 1 file changed, 4 insertions(+), 4 deletions(-)

diff --git a/tests/ImageFilenameSanitizerTest.php b/tests/ImageFilenameSanitizerTest.php
index 7917df6..b51e424 100644
--- a/tests/ImageFilenameSanitizerTest.php
+++ b/tests/ImageFilenameSanitizerTest.php
@@ -177,10 +177,10 @@ class ImageFilenameSanitizerTest extends WikiaBaseTest {
 				'Content of gallery with diacritics and UTF characters'
 			],
 			[
-				'[[Image:image.jpg|300px|lorem ipsum]]',
-				'es',
-				'image.jpg',
-				'Link to filename with canonical namespace, width and caption on a non-EN wiki '
+				'Image:filename.jpg',
+				'en',
+				'filename.jpg',
+				'Filename with alias to namespace'
 			],
 		];
 	}