Change expect_json to put parsed json in new attr

[dmitchell]

I didn't change places which aren't using expect_json to use it, but I did fix expect_json and all the unit tests afaik. It's interesting that our existing unit tests were avoiding the json serialization deserialization by never using "application/json" for the ACCEPT but instead using "multipart/form"

@singingwolfboy @cahrens please review

[singingwolfboy]

@dmitchell Jenkins is showing test failures on this PR, but I don't know if those failures are due to your code changes or not. Can you take a look, fix if necessary, rebase onto latest master, and re-run the tests?

[cahrens]

When does one use _replace vs. replace?

[dmitchell]

For locations always use replace since chris created a method which stifles the pylint error of using an internal fn.

[singingwolfboy]

Seems like it would be a good idea to subclass the Django test client and add a new method:

class AjaxEnabledTestClient(Client):
    def ajax_post(self, path, data={}, content_type="application/json", **kwargs):
        if not isinstance(data, basestring):
            data = json.dumps(data)
        kwargs.setdefault("HTTP_X_REQUESTED_WITH", "XMLHttpRequest")
        return self.post(path=path, data=data, content_type=content_type, **kwargs)

Then attach this subclass to the test suite as self.client, and call self.client.ajax_post() where appropriate. That way, this logic isn't duplicated many times, and it's clearer what you're trying to achieve. (You may also want to add an ajax_get method, but that seems less used (if at all).

[dmitchell]

Passes all the tests now.

[cahrens]

👍

[singingwolfboy]

$.postJSON and $["postJSON"] are exactly equivalent in Javascript, and the former is easier to read.

[singingwolfboy]

👍