These comprise a small proportion of the sites submitted to HN. As for their share of the web as a whole, not sure how prevalent they are.

As such, I usually approach these as one-offs. I find the endpoint and move on. I just use the endpoint, not the "web app". Before I start scanning the JS files to find the endpoint, I first check for a corresponding Github page as most of the web developers using those frameworks so religiously have one. They often post the web app page to HN instead of the Github page. Obviously one can just run the site in a Javascript enabled browser, observe the traffic, and find the endpoint that way. Once you have the endpoint, you will never need to visit the "blank page" ever again. Of course you should be comfortable with JSON. I have no issue with raw JSON in lieu of HTML. In most cases it is easier to process.

If these JS framework sites were more prevalent, and they actually were sites that had data I cared about, I might automate a solution. Most of them tend to be aimed at collecting data, not supplying it, hence I am not much interested in them usually.