How can I test binary file uploading with django-r

Question

I have a Django application with a view that accepts a file to be uploaded. Using the Django REST framework I'm subclassing APIView and implementing the post() method like this:

class FileUpload(APIView):
    permission_classes = (IsAuthenticated,)

    def post(self, request, *args, **kwargs):
        try:
            image = request.FILES['image']
            # Image processing here.
            return Response(status=status.HTTP_201_CREATED)
        except KeyError:
            return Response(status=status.HTTP_400_BAD_REQUEST, data={'detail' : 'Expected image.'})

Now I'm trying to write a couple of unittests to ensure authentication is required and that an uploaded file is actually processed.

class TestFileUpload(APITestCase):
    def test_that_authentication_is_required(self):
        self.assertEqual(self.client.post('my_url').status_code, status.HTTP_401_UNAUTHORIZED)

    def test_file_is_accepted(self):
        self.client.force_authenticate(self.user)
        image = Image.new('RGB', (100, 100))
        tmp_file = tempfile.NamedTemporaryFile(suffix='.jpg')
        image.save(tmp_file)
        with open(tmp_file.name, 'rb') as data:
            response = self.client.post('my_url', {'image': data}, format='multipart')
            self.assertEqual(status.HTTP_201_CREATED, response.status_code)

But this fails when the REST framework attempts to encode the request

Traceback (most recent call last):
  File "/home/vagrant/.virtualenvs/myapp/lib/python3.3/site-packages/django/utils/encoding.py", line 104, in force_text
    s = six.text_type(s, encoding, errors)
UnicodeDecodeError: 'utf-8' codec can't decode byte 0xff in position 118: invalid start byte

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
  File "/home/vagrant/webapp/myproject/myapp/tests.py", line 31, in test_that_jpeg_image_is_accepted
    response = self.client.post('my_url', { 'image': data}, format='multipart')
  File "/home/vagrant/.virtualenvs/myapp/lib/python3.3/site-    packages/rest_framework/test.py", line 76, in post
    return self.generic('POST', path, data, content_type, **extra)
  File "/home/vagrant/.virtualenvs/myapp/lib/python3.3/site-packages/rest_framework/compat.py", line 470, in generic
    data = force_bytes_or_smart_bytes(data, settings.DEFAULT_CHARSET)
  File "/home/vagrant/.virtualenvs/myapp/lib/python3.3/site-packages/django/utils/encoding.py", line 73, in smart_text
    return force_text(s, encoding, strings_only, errors)
  File "/home/vagrant/.virtualenvs/myapp/lib/python3.3/site-packages/django/utils/encoding.py", line 116, in force_text
    raise DjangoUnicodeDecodeError(s, *e.args)
django.utils.encoding.DjangoUnicodeDecodeError: 'utf-8' codec can't decode byte 0xff in position 118: invalid start byte. You passed in b'--BoUnDaRyStRiNg\r\nContent-Disposition: form-data; name="image"; filename="tmpyz2wac.jpg"\r\nContent-Type: image/jpeg\r\n\r\n\xff\xd8\xff[binary data omitted]' (<class 'bytes'>)

How can I make the test client send the data without attempting to decode it as UTF-8?

Answer 1

Python 3 users: make sure you open the file in mode='rb' (read,binary). Otherwise, when Django calls read on the file the utf-8 codec will immediately start choking. The file should be decoded as binary not utf-8, ascii or any other encoding.

# This won't work in Python 3
with open(tmp_file.name) as fp:
        response = self.client.post('my_url', 
                                   {'image': fp}, 
                                   format='multipart')

# Set the mode to binary and read so it can be decoded as binary
with open(tmp_file.name, 'rb') as fp:
        response = self.client.post('my_url', 
                                   {'image': fp}, 
                                   format='multipart')

Answer 2

It's not so simple to understand how to do it if you want to use the PATCH method, but I found the solution in this question.

from django.test.client import BOUNDARY, MULTIPART_CONTENT, encode_multipart

with open(tmp_file.name, 'rb') as fp:
    response = self.client.patch(
        'my_url', 
        encode_multipart(BOUNDARY, {'image': fp}), 
        content_type=MULTIPART_CONTENT
    )

Answer 3

For those in Windows, the answer is a bit different. I had to do the following:

resp = None
with tempfile.NamedTemporaryFile(suffix='.jpg', delete=False) as tmp_file:
    image = Image.new('RGB', (100, 100), "#ffffd")
    image.save(tmp_file, format="JPEG")
    tmp_file.close()

# create status update
with open(tmp_file.name, 'rb') as photo:
    resp = self.client.post('/api/articles/', {'title': 'title',
                                               'content': 'content',
                                               'photo': photo,
                                               }, format='multipart')
os.remove(tmp_file.name)

The difference, as pointed in this answer (https://stackoverflow.com/a/23212515/72350), the file cannot be used after it was closed in Windows. Under Linux, @Meistro's answer should work.

Answer 4

When testing file uploads, you should pass the stream object into the request, not the data.

This was pointed out in the comments by @arocks

Pass { 'image': file} instead

But that didn't full explain why it was needed (and also didn't match the question). For this specific question, you should be doing

class TestFileUpload(APITestCase):

    def test_file_is_accepted(self):
        self.client.force_authenticate(self.user)

        image = Image.new('RGB', (100, 100))

        tmp_file = tempfile.NamedTemporaryFile(suffix='.jpg')
        image.save(tmp_file)

        response = self.client.post('my_url', {'image': tmp_file}, format='multipart')

       self.assertEqual(status.HTTP_201_CREATED, response.status_code)

This will match a standard Django request, where the file is passed in as a stream object, and Django REST Framework handles it. When you just pass in the file data, Django and Django REST Framework interpret it as a string, which causes issues because it is expecting a stream.

And for those coming here looking to another common error, why file uploads just won't work but normal form data will: make sure to set format="multipart" when creating the request.

This also gives a similar issue, and was pointed out by @RobinElvin in the comments

It was because I was missing format='multipart'