feat(previews): add file signature check before opening files

Signed-off-by: Benjamin Gaussorgues <benjamin.gaussorgues@nextcloud.com>

Author: Altahrim

lib/private/Preview/Bitmap.php

@@ -37,6 +37,26 @@
  * @package OC\Preview
  */
 abstract class Bitmap extends ProviderV2 {
+	/**
+	 * List of MIME types that this preview provider is allowed to process.
+	 *
+	 * These should correspond to the MIME types *identified* by Imagemagick
+	 * for files to be processed by this provider. These do / will not
+	 * necessarily need to match the MIME types stored in the database
+	 * (which are identified by IMimeTypeDetector).
+	 *
+	 * @return string Regular expression
+	 */
+	abstract protected function getAllowedMimeTypes(): string;
+
+
+	/**
+	 * @return list<string>
+	 */
+	abstract protected function getMagicStrings(): array;
+
+	abstract protected function getImagickFormatHint(): string;
+
 	/**
 	 * {@inheritDoc}
 	 */
@@ -90,8 +110,19 @@ public function getThumbnail(File $file, int $maxX, int $maxY): ?IImage {
 	private function getResizedPreview($tmpPath, $maxX, $maxY) {
 		$bp = new Imagick();
 
+		if (!$this->isMagicStringSupported($tmpPath)) {
+			throw new \Exception('Invalid image type: magic string not recognized');
+		}
+
+		// Validate mime type
+		$bp->pingImage($this->getImagickFormatHint() . ':' . $tmpPath . '[0]');
+		$mimeType = $bp->getImageMimeType();
+		if (!preg_match($this->getAllowedMimeTypes(), $mimeType)) {
+			throw new \Exception('File mime type does not match the preview provider: ' . $mimeType);
+		}
+
 		// Layer 0 contains either the bitmap or a flat representation of all vector layers
-		$bp->readImage($tmpPath . '[0]');
+		$bp->readImage($this->getImagickFormatHint() . ':' . $tmpPath . '[0]');
 
 		$bp = $this->resize($bp, $maxX, $maxY);
 
@@ -100,6 +131,22 @@ private function getResizedPreview($tmpPath, $maxX, $maxY) {
 		return $bp;
 	}
 
+	private function isMagicStringSupported(string $filepath): bool {
+		$signatures = $this->getMagicStrings();
+		if (empty($signatures)) {
+			return true;
+		}
+		$length = array_reduce($signatures, static fn (int $carry, string $signature) => max($carry, strlen($signature)), 0);
+		$firstBytes = file_get_contents($filepath, false, null, 0, $length);
+		foreach ($signatures as $signature) {
+			if (str_starts_with($firstBytes, $signature)) {
+				return true;
+			}
+		}
+
+		return false;
+	}
+
 	/**
 	 * Returns a resized \Imagick object
 	 *

lib/private/Preview/Font.php

@@ -30,4 +30,22 @@ class Font extends Bitmap {
 	public function getMimeType(): string {
 		return '/application\/(?:font-sfnt|x-font$)/';
 	}
+
+	/**
+	 * {@inheritDoc}
+	 */
+	protected function getAllowedMimeTypes(): string {
+		return '/(application|image)\/(?:font-sfnt|x-font|x-otf|x-ttf|x-pfb$)/';
+	}
+
+	protected function getMagicStrings(): array {
+		return [
+			"\x00\x01\x00\x00\x00", // TTF
+			'OTTO', // OTF
+		];
+	}
+
+	protected function getImagickFormatHint(): string {
+		return 'ttf';
+	}
 }

lib/private/Preview/HEIC.php

@@ -112,8 +112,16 @@ public function getThumbnail(File $file, int $maxX, int $maxY): ?IImage {
 	private function getResizedPreview($tmpPath, $maxX, $maxY) {
 		$bp = new \Imagick();
 
+		// Some HEIC files just contain (or at least are identified as) other formats
+		// like JPEG. We just need to check if the image is safe to process.
+		$bp->pingImage('heic:' . $tmpPath . '[0]');
+		$mimeType = $bp->getImageMimeType();
+		if (!preg_match('/^image\/(x-)?(png|jpeg|gif|bmp|tiff|webp|hei(f|c)|avif)$/', $mimeType)) {
+			throw new \Exception('File mime type does not match the preview provider: ' . $mimeType);
+		}
+
 		// Layer 0 contains either the bitmap or a flat representation of all vector layers
-		$bp->readImage($tmpPath . '[0]');
+		$bp->readImage('heic:' . $tmpPath . '[0]');
 
 		// Fix orientation from EXIF
 		$bp->autoOrient();

lib/private/Preview/Illustrator.php

@@ -31,4 +31,19 @@ class Illustrator extends Bitmap {
 	public function getMimeType(): string {
 		return '/application\/illustrator/';
 	}
+
+	/**
+	 * {@inheritDoc}
+	 */
+	protected function getAllowedMimeTypes(): string {
+		return '/application\/(illustrator|pdf)/';
+	}
+
+	protected function getMagicStrings(): array {
+		return ["\x25\x50\x44\x46"];
+	}
+
+	protected function getImagickFormatHint(): string {
+		return 'ai';
+	}
 }

lib/private/Preview/PDF.php

@@ -31,4 +31,19 @@ class PDF extends Bitmap {
 	public function getMimeType(): string {
 		return '/application\/pdf/';
 	}
+
+	/**
+	 * {@inheritDoc}
+	 */
+	protected function getAllowedMimeTypes(): string {
+		return '/application\/pdf/';
+	}
+
+	protected function getMagicStrings(): array {
+		return ['%PDF-'];
+	}
+
+	protected function getImagickFormatHint(): string {
+		return 'pdf';
+	}
 }

lib/private/Preview/Photoshop.php

@@ -31,4 +31,19 @@ class Photoshop extends Bitmap {
 	public function getMimeType(): string {
 		return '/application\/x-photoshop/';
 	}
+
+	/**
+	 * {@inheritDoc}
+	 */
+	protected function getAllowedMimeTypes(): string {
+		return '/(application|image)\/(x-photoshop|x-psd)/';
+	}
+
+	protected function getMagicStrings(): array {
+		return ['8BPS'];
+	}
+
+	protected function getImagickFormatHint(): string {
+		return 'psd';
+	}
 }

lib/private/Preview/Postscript.php

@@ -31,4 +31,19 @@ class Postscript extends Bitmap {
 	public function getMimeType(): string {
 		return '/application\/postscript/';
 	}
+
+	/**
+	 * {@inheritDoc}
+	 */
+	protected function getAllowedMimeTypes(): string {
+		return '/application\/postscript/';
+	}
+
+	protected function getMagicStrings(): array {
+		return ['%!PS'];
+	}
+
+	protected function getImagickFormatHint(): string {
+		return 'ps';
+	}
 }

lib/private/Preview/SGI.php

@@ -30,4 +30,12 @@ class SGI extends Bitmap {
 	public function getMimeType(): string {
 		return '/image\/sgi/';
 	}
+
+	protected function getMagicStrings(): array {
+		return ["\x01\xDA"];
+	}
+
+	protected function getImagickFormatHint(): string {
+		return 'sgi';
+	}
 }

lib/private/Preview/TGA.php

@@ -30,4 +30,12 @@ class TGA extends Bitmap {
 	public function getMimeType(): string {
 		return '/image\/t(ar)?ga/';
 	}
+
+	protected function getMagicStrings(): array {
+		return [];
+	}
+
+	protected function getImagickFormatHint(): string {
+		return 'tga';
+	}
 }

lib/private/Preview/TIFF.php

@@ -31,4 +31,24 @@ class TIFF extends Bitmap {
 	public function getMimeType(): string {
 		return '/image\/tiff/';
 	}
+
+	/**
+	 * {@inheritDoc}
+	 */
+	protected function getAllowedMimeTypes(): string {
+		return '/image\/tiff/';
+	}
+
+	protected function getMagicStrings(): array {
+		return [
+			"II*\x00",
+			"MM\x00*",
+			"II+\x00",
+			"MM\x00+",
+		];
+	}
+
+	protected function getImagickFormatHint(): string {
+		return 'tiff';
+	}
 }