revert previous fix and added normalizer to cache class

3 files changed, 92 insertions, 1 deletions

diff --git a/lib/files/cache/cache.php b/lib/files/cache/cache.php
index 3341fe50525..a7e634c8e41 100644
--- a/lib/files/cache/cache.php
+++ b/lib/files/cache/cache.php
@@ -100,6 +100,9 @@ class Cache {
 	 */
 	public function get($file) {
 		if (is_string($file) or $file == '') {
+			// normalize file
+			$file = $this->normalize($file);
+
 			$where = 'WHERE `storage` = ? AND `path_hash` = ?';
 			$params = array($this->getNumericStorageId(), md5($file));
 		} else { //file id
@@ -177,6 +180,9 @@ class Cache {
 			$this->update($id, $data);
 			return $id;
 		} else {
+			// normalize file
+			$file = $this->normalize($file);
+
 			if (isset($this->partial[$file])) { //add any saved partial data
 				$data = array_merge($this->partial[$file], $data);
 				unset($this->partial[$file]);
@@ -265,6 +271,9 @@ class Cache {
 	 * @return int
 	 */
 	public function getId($file) {
+		// normalize file
+		$file = $this->normalize($file);
+
 		$pathHash = md5($file);
 
 		$query = \OC_DB::prepare('SELECT `fileid` FROM `*PREFIX*filecache` WHERE `storage` = ? AND `path_hash` = ?');
@@ -549,4 +558,19 @@ class Cache {
 			return null;
 		}
 	}
+
+	/**
+	 * normalize the given path
+	 * @param $path
+	 * @return string
+	 */
+	public function normalize($path) {
+
+		//normalize unicode if possible
+		if (class_exists('Normalizer')) {
+			$path = \Normalizer::normalize($path);
+		}
+
+		return $path;
+	}
 }
diff --git a/lib/files/cache/scanner.php b/lib/files/cache/scanner.php
index b73e2e83fcc..a98953b42aa 100644
--- a/lib/files/cache/scanner.php
+++ b/lib/files/cache/scanner.php
@@ -114,7 +114,7 @@ class Scanner {
 		$size = 0;
 		if ($this->storage->is_dir($path) && ($dh = $this->storage->opendir($path))) {
 			\OC_DB::beginTransaction();
-			while ($file = utf8_encode(readdir($dh))) {
+			while ($file = readdir($dh)) {
 				$child = ($path) ? $path . '/' . $file : $file;
 				if (!$this->isIgnoredDir($file)) {
 					$data = $this->scanFile($child, $recursive === self::SCAN_SHALLOW);
diff --git a/tests/lib/files/cache/cache.php b/tests/lib/files/cache/cache.php
index 1612a673838..2b1e5a56212 100644
--- a/tests/lib/files/cache/cache.php
+++ b/tests/lib/files/cache/cache.php
@@ -8,6 +8,8 @@
 
 namespace Test\Files\Cache;
 
+use PHPUnit_Framework_MockObject_MockObject;
+
 class LongId extends \OC\Files\Storage\Temporary {
 	public function getId() {
 		return 'long:' . str_repeat('foo', 50) . parent::getId();
@@ -237,6 +239,71 @@ class Cache extends \PHPUnit_Framework_TestCase {
 		$this->assertEquals(array(md5($storageId), 'foo'), \OC\Files\Cache\Cache::getById($id));
 	}
 
+	/**
+	 * @brief this test show the bug resulting if we have no normalizer installed
+	 */
+	public function testWithoutNormalizer() {
+		// create folder Schön with U+00F6
+		$folderWith00F6 = "\x53\x63\x68\xc3\xb6\x6e";
+
+		// create folder Schön with U+0308
+		$folderWith0308 = "\x53\x63\x68\x6f\xcc\x88\x6e";
+
+		/**
+		 * @var \OC\Files\Cache\Cache | PHPUnit_Framework_MockObject_MockObject $cacheMock
+		 */
+		$cacheMock = $this->getMock('\OC\Files\Cache\Cache', array('normalize'), array($this->storage), '', true);
+
+		$cacheMock->expects($this->any())
+			->method('normalize')
+			->will($this->returnArgument(0));
+
+		$data = array('size' => 100, 'mtime' => 50, 'mimetype' => 'httpd/unix-directory');
+
+		$this->assertFalse($cacheMock->get('folder'));
+		$this->assertGreaterThan(0, $cacheMock->put('folder', $data));
+
+		$this->assertFalse($cacheMock->get('folder/' . $folderWith00F6));
+		$this->assertGreaterThan(0, $cacheMock->put('folder/' .$folderWith00F6, $data));
+
+		$this->assertFalse($cacheMock->get('folder/' .$folderWith0308));
+		$this->assertGreaterThan(0, $cacheMock->put('folder/' .$folderWith0308, $data));
+
+		// this is our bug, we have two different hashes with the same name (Schön)
+		$this->assertEquals(2, count($cacheMock->getFolderContents('folder')));
+	}
+
+	/**
+	 * @brief this test shows that there is no bug if we use the normalizer
+	 */
+	public function testWithNormalizer() {
+
+		if(!class_exists('Normalizer')) {
+			$this->markTestSkipped('The Normalizer extension is not available.');
+			return;
+		}
+
+		// folder name Schön with U+00F6
+		$folderWith00F6 = "\x53\x63\x68\xc3\xb6\x6e";
+
+		// folder name Schön with U+0308
+		$folderWith0308 = "\x53\x63\x68\x6f\xcc\x88\x6e";
+
+		$data = array('size' => 100, 'mtime' => 50, 'mimetype' => 'httpd/unix-directory');
+
+		$this->assertFalse($this->cache->get('folder'));
+		$this->assertGreaterThan(0, $this->cache->put('folder', $data));
+
+		$this->assertFalse($this->cache->get('folder/' . $folderWith00F6));
+		$this->assertGreaterThan(0, $this->cache->put('folder/' .$folderWith00F6, $data));
+
+		$this->assertTrue(is_array($this->cache->get('folder/' .$folderWith0308)));
+		$this->assertGreaterThan(0, $this->cache->put('folder/' .$folderWith0308, $data));
+
+		// at this point we should have only one folder named "Schön"
+		$this->assertEquals(1, count($this->cache->getFolderContents('folder')));
+	}
+
 	public function tearDown() {
 		$this->cache->clear();
 	}