response.content_type should not remove charset

[joelpresence]

Hi,

When I run rails and perform a certain request, I get a response with:

Content-Type: application/json; charset=utf-8

However, when performing an Rspec controller test against that same controller, the response.content_type is just application/json which is not correct. Somehow, rspec is splitting off charset into response.charset which comes back as utf-8. This feels odd to do this since the actual header name where the charset is returned is Content-Type.

I'm not sure which code is doing this and obviously not breaking a bunch of existing tests is a concern, but this seems like a mismatch to me ...

Thoughts?

Thanks!

[JonRowe]

As per rspec/rspec-core#1965 this is a Rails issue.

[joelpresence]

OK, sorry, misunderstood, my fault. I'll file against rails.

[cupakromer]

The response object in controller specs is an instance of ActionController::TestResponse. However, content_type is inherited from ActionDispatch::Response. Per ActionDispatch::Response#content_type this is by design:

    # Sets the HTTP response's content MIME type. For example, in the controller
    # you could write this:
    #
    #  response.content_type = "text/plain"
    #
    # If a character set has been defined for this response (see charset=) then
    # the character set information will also be included in the content type
    # information.
    attr_reader   :content_type

    # The charset of the response. HTML wants to know the encoding of the
    # content you're giving them, so we need to send that along.
    attr_accessor :charset

You are asking about the actual header value. You can get that from:

expect(response.headers).to include(
  "Content-Type" => "application/json; charset=utf-8"
)