Merge pull request #18388 from nextcloud/backport/18386/stable15

[stable15] [stable17] Allow to detect mimetype by content

apps/workflowengine/lib/Check/AbstractStringCheck.php

@@ -44,8 +44,9 @@ abstract class AbstractStringCheck implements ICheck {
 	/**
 	 * @param IStorage $storage
 	 * @param string $path
+	 * @param bool $isDir
 	 */
-	public function setFileInfo(IStorage $storage, $path) {
+	public function setFileInfo(IStorage $storage, $path, $isDir = false) {
 		// Nothing changes here with a different path
 	}
 

apps/workflowengine/lib/Check/FileMimeType.php

@@ -58,13 +58,19 @@ class FileMimeType extends AbstractStringCheck {
 	/**
 	 * @param IStorage $storage
 	 * @param string $path
+	 * @param bool $isDir
 	 */
-	public function setFileInfo(IStorage $storage, $path) {
+	public function setFileInfo(IStorage $storage, $path, $isDir = false) {
 		$this->storage = $storage;
 		$this->path = $path;
 		if (!isset($this->mimeType[$this->storage->getId()][$this->path])
 			|| $this->mimeType[$this->storage->getId()][$this->path] === '') {
-			$this->mimeType[$this->storage->getId()][$this->path] = null;
+
+			if ($isDir) {
+				$this->mimeType[$this->storage->getId()][$this->path] = 'httpd/unix-directory';
+			} else {
+				$this->mimeType[$this->storage->getId()][$this->path] = null;
+			}
 		}
 	}
 
@@ -103,93 +109,24 @@ class FileMimeType extends AbstractStringCheck {
 			return $this->cacheAndReturnMimeType($this->storage->getId(), $this->path, 'httpd/unix-directory');
 		}
 
-		if ($this->isWebDAVRequest()) {
-			// Creating a folder
-			if ($this->request->getMethod() === 'MKCOL') {
-				return $this->cacheAndReturnMimeType($this->storage->getId(), $this->path, 'httpd/unix-directory');
-			}
-
-			if ($this->request->getMethod() === 'PUT' || $this->request->getMethod() === 'MOVE') {
-				if ($this->request->getMethod() === 'MOVE') {
-					$mimeType = $this->mimeTypeDetector->detectPath($this->path);
-				} else {
-					$path = $this->request->getPathInfo();
-					$mimeType = $this->mimeTypeDetector->detectPath($path);
-				}
-				return $this->cacheAndReturnMimeType($this->storage->getId(), $this->path, $mimeType);
-			}
-		} else if ($this->isPublicWebDAVRequest()) {
-			if ($this->request->getMethod() === 'PUT') {
-				$path = $this->request->getPathInfo();
-				if (strpos($path, '/webdav/') === 0) {
-					$path = substr($path, strlen('/webdav'));
-				}
-				$path = $this->path . $path;
-				$mimeType = $this->mimeTypeDetector->detectPath($path);
-				return $this->cacheAndReturnMimeType($this->storage->getId(), $path, $mimeType);
-			}
+		if ($this->storage->file_exists($this->path)) {
+			$path = $this->storage->getLocalFile($this->path);
+			$mimeType = $this->mimeTypeDetector->detectContent($path);
+			return $this->cacheAndReturnMimeType($this->storage->getId(), $this->path, $mimeType);
 		}
 
-		if (in_array($this->request->getMethod(), ['POST', 'PUT'])) {
-			$files = $this->request->getUploadedFile('files');
-			if (isset($files['type'][0])) {
-				$mimeType = $files['type'][0];
-				if ($mimeType === 'application/octet-stream') {
-					// Maybe not...
-					$mimeTypeTest = $this->mimeTypeDetector->detectPath($files['name'][0]);
-					if ($mimeTypeTest !== 'application/octet-stream' && $mimeTypeTest !== false) {
-						$mimeType = $mimeTypeTest;
-					} else {
-						$mimeTypeTest = $this->mimeTypeDetector->detect($files['tmp_name'][0]);
-						if ($mimeTypeTest !== 'application/octet-stream' && $mimeTypeTest !== false) {
-							$mimeType = $mimeTypeTest;
-						}
-					}
-				}
-				return $this->cacheAndReturnMimeType($this->storage->getId(), $this->path, $mimeType);
+		if ($this->isWebDAVRequest() || $this->isPublicWebDAVRequest()) {
+			// Creating a folder
+			if ($this->request->getMethod() === 'MKCOL') {
+				return 'httpd/unix-directory';
 			}
 		}
 
-		$mimeType = $this->storage->getMimeType($this->path);
-		if ($mimeType === 'application/octet-stream') {
-			$mimeType = $this->detectMimetypeFromPath();
-		}
-
-		return $this->cacheAndReturnMimeType($this->storage->getId(), $this->path, $mimeType);
-	}
-
-	/**
-	 * @return string
-	 */
-	protected function detectMimetypeFromPath() {
-		$mimeType = $this->mimeTypeDetector->detectPath($this->path);
-		if ($mimeType !== 'application/octet-stream' && $mimeType !== false) {
-			return $mimeType;
-		}
-
-		if ($this->storage->instanceOfStorage('\OC\Files\Storage\Local')
-			|| $this->storage->instanceOfStorage('\OC\Files\Storage\Home')
-			|| $this->storage->instanceOfStorage('\OC\Files\ObjectStore\HomeObjectStoreStorage')) {
-			$localFile = $this->storage->getLocalFile($this->path);
-			if ($localFile !== false) {
-				$mimeType = $this->mimeTypeDetector->detect($localFile);
-				if ($mimeType !== false) {
-					return $mimeType;
-				}
-			}
-
-			return 'application/octet-stream';
-		} else {
-			$handle = $this->storage->fopen($this->path, 'r');
-			$data = fread($handle, 8024);
-			fclose($handle);
-			$mimeType = $this->mimeTypeDetector->detectString($data);
-			if ($mimeType !== false) {
-				return $mimeType;
-			}
-
-			return 'application/octet-stream';
-		}
+		// We do not cache this, as the file did not exist yet.
+		// In case it does in the future, we will check with detectContent()
+		// again to get the real mimetype of the content, rather than
+		// guessing it from the path.
+		return $this->mimeTypeDetector->detectPath($this->path);
 	}
 
 	/**

apps/workflowengine/lib/Check/FileName.php

@@ -49,8 +49,9 @@ class FileName extends AbstractStringCheck {
 	/**
 	 * @param IStorage $storage
 	 * @param string $path
+	 * @param bool $isDir
 	 */
-	public function setFileInfo(IStorage $storage, $path) {
+	public function setFileInfo(IStorage $storage, $path, $isDir = false) {
 		$this->storage = $storage;
 		$this->path = $path;
 	}

apps/workflowengine/lib/Check/FileSize.php

@@ -51,8 +51,9 @@ class FileSize implements ICheck {
 	/**
 	 * @param IStorage $storage
 	 * @param string $path
+	 * @param bool $isDir
 	 */
-	public function setFileInfo(IStorage $storage, $path) {
+	public function setFileInfo(IStorage $storage, $path, $isDir = false) {
 	}
 
 	/**

apps/workflowengine/lib/Check/FileSystemTags.php

@@ -68,8 +68,9 @@ class FileSystemTags implements ICheck {
 	/**
 	 * @param IStorage $storage
 	 * @param string $path
+	 * @param bool $isDir
 	 */
-	public function setFileInfo(IStorage $storage, $path) {
+	public function setFileInfo(IStorage $storage, $path, $isDir = false) {
 		$this->storage = $storage;
 		$this->path = $path;
 	}

apps/workflowengine/lib/Check/RequestRemoteAddress.php

@@ -47,8 +47,9 @@ class RequestRemoteAddress implements ICheck {
 	/**
 	 * @param IStorage $storage
 	 * @param string $path
+	 * @param bool $isDir
 	 */
-	public function setFileInfo(IStorage $storage, $path) {
+	public function setFileInfo(IStorage $storage, $path, $isDir = false) {
 		// A different path doesn't change time, so nothing to do here.
 	}
 

apps/workflowengine/lib/Check/RequestTime.php

@@ -52,8 +52,9 @@ class RequestTime implements ICheck {
 	/**
 	 * @param IStorage $storage
 	 * @param string $path
+	 * @param bool $isDir
 	 */
-	public function setFileInfo(IStorage $storage, $path) {
+	public function setFileInfo(IStorage $storage, $path, $isDir = false) {
 		// A different path doesn't change time, so nothing to do here.
 	}
 

apps/workflowengine/lib/Check/UserGroupMembership.php

@@ -60,8 +60,9 @@ class UserGroupMembership implements ICheck {
 	/**
 	 * @param IStorage $storage
 	 * @param string $path
+	 * @param bool $isDir
 	 */
-	public function setFileInfo(IStorage $storage, $path) {
+	public function setFileInfo(IStorage $storage, $path, $isDir = false) {
 		// A different path doesn't change group memberships, so nothing to do here.
 	}
 

lib/private/Files/Type/Detection.php

@@ -178,11 +178,14 @@ class Detection implements IMimeTypeDetector {
 		if (strpos($fileName, '.') > 0) {
 
 			// remove versioning extension: name.v1508946057 and transfer extension: name.ocTransferId2057600214.part
-			$fileName = preg_replace('!((\.v\d+)|((.ocTransferId\d+)?.part))$!', '', $fileName);
+			$fileName = preg_replace('!((\.v\d+)|((\.ocTransferId\d+)?\.part))$!', '', $fileName);
 
 			//try to guess the type by the file extension
-			$extension = strtolower(strrchr($fileName, '.'));
-			$extension = substr($extension, 1); //remove leading .
+			$extension = strrchr($fileName, '.');
+			if ($extension !== false) {
+				$extension = strtolower($extension);
+				$extension = substr($extension, 1); //remove leading .
+			}
 			return (isset($this->mimetypes[$extension]) && isset($this->mimetypes[$extension][0]))
 				? $this->mimetypes[$extension][0]
 				: 'application/octet-stream';
@@ -192,12 +195,12 @@ class Detection implements IMimeTypeDetector {
 	}
 
 	/**
-	 * detect mimetype based on both filename and content
-	 *
+	 * detect mimetype only based on the content of file
 	 * @param string $path
 	 * @return string
+	 * @since 18.0.0
 	 */
-	public function detect($path) {
+	public function detectContent(string $path): string {
 		$this->loadMappings();
 
 		if (@is_dir($path)) {
@@ -205,41 +208,72 @@ class Detection implements IMimeTypeDetector {
 			return "httpd/unix-directory";
 		}
 
-		$mimeType = $this->detectPath($path);
-
-		if ($mimeType === 'application/octet-stream' and function_exists('finfo_open')
-			and function_exists('finfo_file') and $finfo = finfo_open(FILEINFO_MIME)
-		) {
-			$info = @strtolower(finfo_file($finfo, $path));
+		if (function_exists('finfo_open')
+			&& function_exists('finfo_file')
+			&& $finfo = finfo_open(FILEINFO_MIME)) {
+			$info = @finfo_file($finfo, $path);
 			finfo_close($finfo);
 			if ($info) {
+				$info = strtolower($info);
 				$mimeType = strpos($info, ';') !== false ? substr($info, 0, strpos($info, ';')) : $info;
-				return empty($mimeType) ? 'application/octet-stream' : $mimeType;
+				$mimeType = $this->getSecureMimeType($mimeType);
+				if ($mimeType !== 'application/octet-stream') {
+					return $mimeType;
+				}
 			}
+		}
 
+		if (strpos($path, '://') !== false && strpos($path, 'file://') === 0) {
+			// Is the file wrapped in a stream?
+			return 'application/octet-stream';
 		}
-		$isWrapped = (strpos($path, '://') !== false) and (substr($path, 0, 7) === 'file://');
-		if (!$isWrapped and $mimeType === 'application/octet-stream' && function_exists("mime_content_type")) {
+
+		if (function_exists('mime_content_type')) {
 			// use mime magic extension if available
 			$mimeType = mime_content_type($path);
+			if ($mimeType !== false) {
+				$mimeType = $this->getSecureMimeType($mimeType);
+				if ($mimeType !== 'application/octet-stream') {
+					return $mimeType;
+				}
+			}
 		}
-		if (!$isWrapped and $mimeType === 'application/octet-stream' && \OC_Helper::canExecute("file")) {
+
+		if (\OC_Helper::canExecute('file')) {
 			// it looks like we have a 'file' command,
 			// lets see if it does have mime support
 			$path = escapeshellarg($path);
-			$fp = popen("file -b --mime-type $path 2>/dev/null", "r");
-			$reply = fgets($fp);
+			$fp = popen("test -f $path && file -b --mime-type $path", 'r');
+			$mimeType = fgets($fp);
 			pclose($fp);
 
-			//trim the newline
-			$mimeType = trim($reply);
-
-			if (empty($mimeType)) {
-				$mimeType = 'application/octet-stream';
+			if ($mimeType !== false) {
+				//trim the newline
+				$mimeType = trim($mimeType);
+				$mimeType = $this->getSecureMimeType($mimeType);
+				if ($mimeType !== 'application/octet-stream') {
+					return $mimeType;
+				}
 			}
 
 		}
-		return $mimeType;
+		return 'application/octet-stream';
+	}
+
+	/**
+	 * detect mimetype based on both filename and content
+	 *
+	 * @param string $path
+	 * @return string
+	 */
+	public function detect($path) {
+		$mimeType = $this->detectPath($path);
+
+		if ($mimeType !== 'application/octet-stream') {
+			return $mimeType;
+		}
+
+		return $this->detectContent($path);
 	}
 
 	/**

lib/public/Files/IMimeTypeDetector.php

@@ -39,9 +39,17 @@ interface IMimeTypeDetector {
 	 * @param string $path
 	 * @return string
 	 * @since 8.2.0
-	 **/
+	 */
 	public function detectPath($path);
 
+	/**
+	 * detect mimetype only based on the content of file
+	 * @param string $path
+	 * @return string
+	 * @since 18.0.0
+	 */
+	public function detectContent(string $path): string;
+
 	/**
 	 * detect mimetype based on both filename and content
 	 *

lib/public/WorkflowEngine/ICheck.php

@@ -36,9 +36,10 @@ interface ICheck {
 	/**
 	 * @param IStorage $storage
 	 * @param string $path
+	 * @param bool $isDir
 	 * @since 9.1
 	 */
-	public function setFileInfo(IStorage $storage, $path);
+	public function setFileInfo(IStorage $storage, $path, $isDir = false);
 
 	/**
 	 * @param string $operator

tests/lib/Files/Type/DetectionTest.php

@@ -37,61 +37,96 @@ class DetectionTest extends \Test\TestCase {
 		);
 	}
 
-	public function testDetect() {
-		$dir = \OC::$SERVERROOT.'/tests/data';
+	public function dataDetectPath(): array {
+		return [
+			['foo.txt', 'text/plain'],
+			['foo.png', 'image/png'],
+			['foo.bar.png', 'image/png'],
+			['.hidden.png', 'image/png'],
+			['.hidden.foo.png', 'image/png'],
+			['.hidden/foo.png', 'image/png'],
+			['.hidden/.hidden.png', 'image/png'],
+			['test.jpg/foo.png', 'image/png'],
+			['.png', 'application/octet-stream'],
+			['..hidden', 'application/octet-stream'],
+			['foo', 'application/octet-stream'],
+			['', 'application/octet-stream'],
+			['foo.png.ocTransferId123456789.part', 'image/png'],
+			['foo.png.v1234567890', 'image/png'],
+		];
+	}
 
-		$result = $this->detection->detect($dir."/");
-		$expected = 'httpd/unix-directory';
-		$this->assertEquals($expected, $result);
+	/**
+	 * @dataProvider dataDetectPath
+	 *
+	 * @param string $path
+	 * @param string $expected
+	 */
+	public function testDetectPath(string $path, string $expected) {
+		$this->assertEquals($expected, $this->detection->detectPath($path));
+	}
 
-		$result = $this->detection->detect($dir."/data.tar.gz");
-		$expected = 'application/x-gzip';
-		$this->assertEquals($expected, $result);
+	public function dataDetectContent(): array {
+		return [
+			['/', 'httpd/unix-directory'],
+			['/data.tar.gz', 'application/x-gzip'],
+			['/data.zip', 'application/zip'],
+			['/testimage.mp3', 'audio/mpeg'],
+			['/testimage.png', 'image/png'],
+		];
+	}
 
-		$result = $this->detection->detect($dir."/data.zip");
-		$expected = 'application/zip';
-		$this->assertEquals($expected, $result);
+	/**
+	 * @dataProvider dataDetectContent
+	 *
+	 * @param string $path
+	 * @param string $expected
+	 */
+	public function testDetectContent(string $path, string $expected) {
+		$this->assertEquals($expected, $this->detection->detectContent(\OC::$SERVERROOT . '/tests/data' . $path));
+	}
 
-		$result = $this->detection->detect($dir."/testimagelarge.svg");
-		$expected = 'image/svg+xml';
-		$this->assertEquals($expected, $result);
+	public function dataDetect(): array {
+		return [
+			['/', 'httpd/unix-directory'],
+			['/data.tar.gz', 'application/x-gzip'],
+			['/data.zip', 'application/zip'],
+			['/testimagelarge.svg', 'image/svg+xml'],
+			['/testimage.png', 'image/png'],
+		];
+	}
 
-		$result = $this->detection->detect($dir."/testimage.png");
-		$expected = 'image/png';
-		$this->assertEquals($expected, $result);
+	/**
+	 * @dataProvider dataDetect
+	 *
+	 * @param string $path
+	 * @param string $expected
+	 */
+	public function testDetect(string $path, string $expected) {
+		$this->assertEquals($expected, $this->detection->detect(\OC::$SERVERROOT . '/tests/data' . $path));
 	}
 
-	public function testGetSecureMimeType() {
-		$result = $this->detection->getSecureMimeType('image/svg+xml');
+	public function testDetectString() {
+		$result = $this->detection->detectString('/data/data.tar.gz');
 		$expected = 'text/plain';
 		$this->assertEquals($expected, $result);
-
-		$result = $this->detection->getSecureMimeType('image/png');
-		$expected = 'image/png';
-		$this->assertEquals($expected, $result);
 	}
 
-	public function testDetectPath() {
-		$this->assertEquals('text/plain', $this->detection->detectPath('foo.txt'));
-		$this->assertEquals('image/png', $this->detection->detectPath('foo.png'));
-		$this->assertEquals('image/png', $this->detection->detectPath('foo.bar.png'));
-		$this->assertEquals('image/png', $this->detection->detectPath('.hidden.png'));
-		$this->assertEquals('image/png', $this->detection->detectPath('.hidden.foo.png'));
-		$this->assertEquals('image/png', $this->detection->detectPath('.hidden/foo.png'));
-		$this->assertEquals('image/png', $this->detection->detectPath('.hidden/.hidden.png'));
-		$this->assertEquals('image/png', $this->detection->detectPath('test.jpg/foo.png'));
-		$this->assertEquals('application/octet-stream', $this->detection->detectPath('.png'));
-		$this->assertEquals('application/octet-stream', $this->detection->detectPath('..hidden'));
-		$this->assertEquals('application/octet-stream', $this->detection->detectPath('foo'));
-		$this->assertEquals('application/octet-stream', $this->detection->detectPath(''));
-		$this->assertEquals('image/png', $this->detection->detectPath('foo.png.ocTransferId123456789.part'));
-		$this->assertEquals('image/png', $this->detection->detectPath('foo.png.v1234567890'));
+	public function dataGetSecureMimeType(): array {
+		return [
+			['image/svg+xml', 'text/plain'],
+			['image/png', 'image/png'],
+		];
 	}
 
-	public function testDetectString() {
-		$result = $this->detection->detectString("/data/data.tar.gz");
-		$expected = 'text/plain';
-		$this->assertEquals($expected, $result);
+	/**
+	 * @dataProvider dataGetSecureMimeType
+	 *
+	 * @param string $mimeType
+	 * @param string $expected
+	 */
+	public function testGetSecureMimeType(string $mimeType, string $expected) {
+		$this->assertEquals($expected, $this->detection->getSecureMimeType($mimeType));
 	}
 
 	public function testMimeTypeIcon() {