Redesign register_output_renderer callback

[simonw]

In building https://github.com/simonw/datasette-atom it became clear that the callback function (which currently accepts just args, data and view_name) would also benefit from access to a mechanism to render templates and a datasette instance so it can execute SQL.

To maintain backwards compatibility with existing plugins, we can introspect the callback function to see if it wants those new arguments or not.

At a minimum I want to make datasette and ASGI scope available.

[simonw]

The request object should be made available too.

[simonw]

To solve simonw/datasette-atom#6 the name of the current canned query should be made available somehow.

That way the plugin configuration could specify that the title for browse/feed should be X.

[simonw]

Rather than pass a request object (hence promoting that object into part of the documented, stable API) I think I'll pass the ASGI scope - that's already a stable, documented standard.

UPDATE: changed my mind since request is used by other plugins too, see #706.

[simonw]

Here's the code that calls the renderers - this needs to be expanded to check for those extra optional arguments:

datasette/datasette/views/base.py

Lines 387 to 398 in 2d099ad

	if _format in self.ds.renderers.keys():
	# Dispatch request to the correct output format renderer
	# (CSV is not handled here due to streaming)
	result = self.ds.renderers[_format](request.args, data, self.name)
	if result is None:
	raise NotFound("No data")
	r = Response(
	body=result.get("body"),
	status=result.get("status_code", 200),
	content_type=result.get("content_type", "text/plain"),
	)

[simonw]

While I'm doing this, another feature I would like is the ability for renderers to opt-in / opt-out of being displayed as options on the page.

https://www.niche-museums.com/browse/museums for example shows a atom link because the datasette-atom plugin is installed... but clicking it will give you a 400 error because the correct columns are not present:

[simonw]

Here's the code that passes a list of renderers to the template:

datasette/datasette/views/base.py

Lines 411 to 423 in 2d099ad

	renderers = {
	key: path_with_format(request, key, {**url_labels_extra})
	for key in self.ds.renderers.keys()
	}
	url_csv_args = {"_size": "max", **url_labels_extra}
	url_csv = path_with_format(request, "csv", url_csv_args)
	url_csv_path = url_csv.split("?")[0]
	context = {
	**data,
	**extras,
	**{
	"renderers": renderers,
	"url_csv": url_csv,

A renderer is currently defined as a two-key dictionary:

@hookimpl
def register_output_renderer(datasette):
    return {
        'extension': 'test',
        'callback': render_test
    }

I can add a third key, "should_suggest" which is a function that returns True or False for a given query. If that key is missing it is assumed to return True.

One catch: what arguments should be passed to the should_suggest(...) function?

[simonw]

I'll use #770 for the should_suggest mechanism - this issue is for the extra arguments passed to the rendering callback.