Implement Enhanced Error Handling for Disallowed MIME Types

.gitignore

@@ -6,4 +6,8 @@
 _trial_temp
 __pycache__
 /dist
-config.yaml
+config.yaml
+/build
+/.vscode
+mcs_pickle.txt
+pickle

docs/api.md

@@ -30,6 +30,7 @@ status code of the response for each scenario:
 | 404         | `M_NOT_FOUND`                 | No route could be found at the given path.                                                                                                                                        |
 | 404         | `M_NOT_FOUND`                 | The requested media was not present in the media repo.                                                                                                                            |
 | 403         | `MCS_MEDIA_NOT_CLEAN`         | The server scanned the downloaded media but the antivirus script returned a non-zero exit code.                                                                                   |
+| 403         | `MCS_MIME_TYPE_FORBIDDEN`     | The Mime type is not in the allowed list of Mime types.                                                                                                                           |
 | 403         | `MCS_BAD_DECRYPTION`          | The provided `encrypted_body` could not be decrypted, or the encrypted file could not be decrypted. The client should request the public key of the server and then retry (once). |
 | 500         | `M_UNKNOWN`                   | The server experienced an unexpected error.                                                                                                                                       |
 | 502         | `MCS_MEDIA_REQUEST_FAILED`    | The server failed to request media from the media repo.                                                                                                                           |

src/matrix_content_scanner/scanner/scanner.py

@@ -30,7 +30,11 @@
 from multidict import MultiMapping
 
 from matrix_content_scanner.utils.constants import ErrCode
-from matrix_content_scanner.utils.errors import ContentScannerRestError, FileDirtyError
+from matrix_content_scanner.utils.errors import (
+    ContentScannerRestError,
+    FileDirtyError,
+    FileMimeTypeForbiddenError,
+)
 from matrix_content_scanner.utils.types import JsonDict, MediaDescription
 
 if TYPE_CHECKING:
@@ -544,4 +548,6 @@ def _check_mimetype(
                 "MIME type for file is forbidden: %s",
                 detected_mimetype,
             )
-            raise FileDirtyError("File type not supported")
+            raise FileMimeTypeForbiddenError(
+                f"File type: {detected_mimetype} not allowed"
+            )

src/matrix_content_scanner/utils/constants.py

@@ -30,3 +30,5 @@ class ErrCode(str, Enum):
     FAILED_TO_DECRYPT = "MCS_MEDIA_FAILED_TO_DECRYPT"
     # The request body isn't valid JSON, or is missing a required parameter.
     MALFORMED_JSON = "MCS_MALFORMED_JSON"
+    # The Mime type is not in the allowed list of Mime types.
+    MIME_TYPE_FORBIDDEN = "MCS_MIME_TYPE_FORBIDDEN"

src/matrix_content_scanner/utils/errors.py

@@ -49,6 +49,17 @@ def __init__(
         self.cacheable = cacheable
 
 
+class FileMimeTypeForbiddenError(ContentScannerRestError):
+    """An error indicating that the file's MIME type is forbidden."""
+
+    def __init__(self, info: Optional[str]) -> None:
+        super(FileMimeTypeForbiddenError, self).__init__(
+            http_status=403,
+            reason=ErrCode.MIME_TYPE_FORBIDDEN,
+            info=info,
+        )
+
+
 class ConfigError(Exception):
     """An error indicating an issue with the configuration file."""
 

tests/scanner/test_scanner.py

@@ -21,7 +21,11 @@
 
 from matrix_content_scanner.scanner.scanner import CacheEntry
 from matrix_content_scanner.utils.constants import ErrCode
-from matrix_content_scanner.utils.errors import ContentScannerRestError, FileDirtyError
+from matrix_content_scanner.utils.errors import (
+    ContentScannerRestError,
+    FileDirtyError,
+    FileMimeTypeForbiddenError,
+)
 from matrix_content_scanner.utils.types import MediaDescription
 from tests.testutils import (
     ENCRYPTED_FILE_METADATA,
@@ -231,7 +235,7 @@ async def test_mimetype(self) -> None:
         self.scanner._allowed_mimetypes = ["image/jpeg"]
 
         # Check that the scan fails since the file is a PNG.
-        with self.assertRaises(FileDirtyError):
+        with self.assertRaises(FileMimeTypeForbiddenError):
             await self.scanner.scan_file(MEDIA_PATH)
 
     async def test_mimetype_encrypted(self) -> None:
@@ -244,7 +248,7 @@ async def test_mimetype_encrypted(self) -> None:
         self.scanner._allowed_mimetypes = ["image/jpeg"]
 
         # Check that the scan fails since the file is a PNG.
-        with self.assertRaises(FileDirtyError):
+        with self.assertRaises(FileMimeTypeForbiddenError):
             await self.scanner.scan_file(MEDIA_PATH, ENCRYPTED_FILE_METADATA)
 
     async def test_mimetype_content_type_mismatch(self) -> None: