geoff-maddock / ai-onion Goto Github PK

In order to simulate how this would work w/o having to wait for the API responses:

• From the POST endpoint, return an array of example data that I'd get back from the OPENAPI call
• API/chat/server.ts
• format of chat responses - this is a series of events, so not entirely sure how to mock that event stream
• extract the code that's currently in onMount that makes the API calls into it's own function, and mock the response there
• create the format of the response that will be the FULL response used in the UI

article
{
  title: string = "Title of the onion article",
  body: string = "Body of the onion article,
  seed: int = size of the seed used to generate article - to differentiate the short and long version
  createdAt: datetime = <time created>
  image: string = "url for generated image"
}

Mock returning one article, or 12 articles, and build the layout using that.
First article is the "main" and the rest will be displayed as thumbnails or just titles.

Find the layout for The Onion that I want to copy exactly and copy it with dummy data.
Then work on replacing that.

How much do I want to get from the AI? Can I make a single query and get a number of headlines or articles?
Maybe ask for 10 headlines and then one full article?

Same with images - can I get 10 thumbnails and then one large?

If I want to make this a public site, I'll have to prevent people from maxing out my account.
Should I refresh and then store in a DB rather than in local storage?

Consider some options about how frequently to allow refresh of the content?

• On every page load
• Every time a "get more news" button is clicked
• Once every hour via cron?
• Once every day