capture whatever Readability turns up at the end of each attention event

[hamilton]

Given the future architecture, we could easily capture full page text and send it into indexedDB.

[hamilton]

I looked into this a bit. WebScience has a textContent module that utilizes readability. Some downsides to its current design:

• it is utilizing the isArticle param in Firefox, which is not cross-browser.
• it does not always accurately capture article content. It seems like if the page is not done rendering in time on occasion, it gets mad and doesn't run Readability. I am not 100% sure this is what's happening but it sure seems like it.
• it doesn't have any error messages when Readability does not parse the page.

A few additional constraints:

• we could capture instead the entire DOM per page load, but this gets very expensive very fast. A visit to amazon.com will be ~ .75mb. so 1,000 unique page visits could really, really balloon (and that's really only maybe a few days of browsing). Storing this in IndexedDB feels like a non-starter.
• we could probably just run readability over the article if it exists on the page.