How do you set "Content-Type" when saving to S3 using django-storages with S3boto backend? - django

Question

I am using django-storages with s3boto as a backend.
I have one bucket with two folders - one for static and one for media. I achieve this using django-s3-folder-storage.
As well as saving to S3 using a model, I also want to implement an image-resize-and-cache function to save the files to S3. To do this I interact directly with my S3 bucket. The code works, but the Content-Type isn't set on S3.
in iPython:
In [2]: from s3_folder_storage.s3 import DefaultStorage
In [3]: s3media = DefaultStorage()
In [4]: s3media
Out[4]: <s3_folder_storage.s3.DefaultStorage at 0x4788780>
Test we're accessing the right bucket - storage_test is one I created earlier:
In [5]: s3media.exists('storage_test')
Out[5]: True
In [6]: s3media.open("test.txt", "w")
Out[6]: <S3BotoStorageFile: test.txt>
In [7]: test = s3media.open("test.txt", "w")
In [8]: test
Out[8]: <S3BotoStorageFile: test.txt>
In [9]: test.key.content_type = "text/plain"
In [10]: test.write("...")
In [11]: test.close()
In [12]: test = s3media.open("test.txt", "w")
In [13]: test.key.content_type
Out[13]: 'binary/octet-stream'
I've also tried instead of In [9] using test.key.metadata and test.key.set_metadata. None of them do it.
How do I set the correct Content-Type?

Answer 1

If you go through the source code in class S3BotoStorageFile and function write, the header is updated from only 2 places,
upload_headers.update(self._storage.headers) where self._storage.headers is taken from AWS_HEADERS
self._storage.default_acl
And in function _flush_write_buffer only self._storage.headers is considered. Check for the line headers = self._storage.headers.copy()
So updating test.key.content_type will not work.
Instead of test.key.content_type = "text/plain" at In [9]: try using test._storage.headers['Content-Type'] = 'text/plain', it should work.

Answer 2

This is for Boto3 ONLY, not Boto. If you would like to set those headers, you will need to access the object like so, file_ is refereing to a FileField with storage setup to be using Boto3 from django-storages:
file_.storage.object_parameters = { 'ContentType': 'text/plain' }
NOTE: it requires header names to be camelcase, so Content-Type = ContentType, Content-Dispostion = ContentDispostion etc. Hope this helps!

Answer 3

Now you can use django-storages >= 1.4 and it automatically guesses the mime types.

Answer 4

According to this answer, the Content-Type isn't metadata but rather headers that you set up when you upload the file.

Answer 5

I've had a similar issue - I wanted to set my header for all the files uploaded to S3 using django-storages, without relying on the default library approach which is guessing mime type based on the filename.
Please note that you can tweak the way how the header is set and you don't have to have it fixed like I have (my case was specific).
This is what worked for me:
Implement custom file manager:
import os
from storages.backends.s3boto3 import S3Boto3Storage
class ManagedS3BotoS3Storage(S3Boto3Storage):
def _save(self, name, content):
cleaned_name = self._clean_name(name)
name = self._normalize_name(cleaned_name)
params = self._get_write_parameters(name, content)
content_type = "application/octet-stream". # Content-Type that I wanted to have for each file
params["ContentType"] = content_type
encoded_name = self._encode_name(name)
obj = self.bucket.Object(encoded_name)
if self.preload_metadata:
self._entries[encoded_name] = obj
content.seek(0, os.SEEK_SET)
obj.upload_fileobj(content, ExtraArgs=params)
return cleaned_name
Use ManagedS3BotoS3Storage in model:
class SomeCoolModel(models.Model):
file = models.FileField(
storage=ManagedS3BotoS3Storage(bucket="my-awesome-S3-bucket"),
upload_to="my_great_path_to_file",
)
Run python manage.py makemigrations.
That's it, after this all files that I had was uploaded with Content-Type: "application/octet-stream