txt file processing fix (#20)

Author: ivadin

safe_s3_storage/file_validator.py

@@ -49,7 +49,9 @@ class FileValidator:
 
     def _validate_mime_type(self, *, file_name: str, file_content: bytes) -> str:
         mime_type_prediction: typing.Final = Magika().identify_bytes(file_content)
-        if mime_type_prediction.dl.extensions:
+        if mime_type_prediction.output.is_text and file_name.endswith(".txt"):
+            mime_type = "text/plain"
+        elif mime_type_prediction.dl.extensions:
             mime_type = mime_type_prediction.dl.mime_type
         else:
             mime_type = mime_type_prediction.output.mime_type

tests/test_file_validator.py

@@ -129,6 +129,19 @@ async def test_ok_image(
             == _IMAGE_CONVERSION_FORMAT_TO_MIME_TYPE_AND_EXTENSION_MAP[image_conversion_format][0]
         )
 
+    @pytest.mark.parametrize("file_content", ["'", "test'", "abracadabra", "python script"])
+    async def test_txt_file_validate(self, file_content: str) -> None:
+        file_name: typing.Final = "file_name.txt"
+
+        validated_file: typing.Final = await FileValidator(allowed_mime_types=["text/plain"]).validate_file(
+            file_name=file_name, file_content=file_content.encode()
+        )
+
+        assert validated_file.file_name == file_name
+        assert validated_file.file_content == file_content.encode()
+        assert validated_file.file_size == len(file_content)
+        assert validated_file.mime_type == "text/plain"
+
     @pytest.mark.parametrize("binary", [True, False])
     async def test_ok_not_image(self, faker: faker.Faker, binary: bool) -> None:
         file_name: typing.Final = faker.file_name()