web.FileResponse overrides content_type

[thelink2012]

Long story short

When returning a web.FileResponse from a route, any changes to the content_type member is not taken into account when the file is served. Instead, the file extension is always used to determine the content type of the response.

Expected behaviour

Content-Type header of the response be that of web.FileResponse.content_type when such has been set.

Actual behaviour

Content-Type header is guessed from the file extension of the served file.

Steps to reproduce

Create a route that responds something along the lines of

    response = web.FileResponse('/path/to/some/image.png')
    response.content_type = 'application/octet-stream'
    return response

then curl -i it and notice the Content-Type is image/png.

Your environment

[~]$ uname -mrs
Linux 4.11.9-1-ARCH x86_64
[~]$ python --version
Python 3.6.2
[~]$ pip show aiohttp | grep Version
Version: 2.2.5

[thelink2012]

I'm inclined to opening a PR to resolve this issue, suggestions are highly appreciated.

[asvetlov]

The only problem is figuring out if content type was explicitly set.
User could do it by resp.content_type = '...' and resp.headers['Content-Type'] modifications.
Default value for property is 'application/octet-stream'.

Well, the following code should work:

if hdrs.CONTENT_TYPE not in self.headers:
        ct, encoding = mimetypes.guess_type(str(filepath))
        if not ct:
            ct = 'application/octet-stream'
        self.content_type = ct
        ....

[thelink2012]

I think setting the content type header based on the path parameter during __init__ is a better idea. This follows the same behaviour as web.Response, and gives us access to the guessed content type from the property. As such, this seems like the least suprising way for web.FileResponse to behave.

Same goes for encoding.

Ok, I see the flaw behind this, nevermind.

[asvetlov]

Fixed by #2319

[lock]

This thread has been automatically locked since there has not been any recent activity after it was closed. Please open a [new issue] for related bugs.
If you feel like there's important points made in this discussion, please include those exceprts into that [new issue].
[new issue]: https:/aio-libs/aiohttp/issues/new