Skrapi

This is the new plan. Skrapi is a SvelteKit Rest API system in a Node module -- thats it!

1. npm init svelte@next
2. npm i skrapi
3. npx skrapi-init to create your $lib/models, $lib/stores, and routes/api/$/[...slug.js], and also prompt for a DB_CONN and generate a .env
4. Optionally: run npx skrapi-ui to add all the boilerplate stuff including Tailwind, skrapi-ui, etc

Once installed, you can create new crud with npx skrapi-crud to generate basic model/store files. By default, your new models will have a schema with only { name: { type: String } } and the store will include all standard skrapi actions including findMany, findOne, insertOne, updateOne, etc.

Immediately after running npx skrapi-crud (and editing your schema), you'll have access to these actions directly from your templates like this:

<script>
  import { books } from '$lib/stores'

  $books = books.findAll()

  async function handleInsert(newBook) {
    const insertedBook = await books.insertOne(newBook)
    console.log(insertedBook)
  }
</script>

{#await $books}
  {#each $books as book}
    <p>Title: {book.title} Author: {book.author}</p>
  {/each}
{/await}

What's the advantage?

You don't have to think about API endpoints. At all. The "actions" that skrapi makes available take care of all that behind the scenes. In the example above, Skrapi will call the endpoint GET /books. Skrapi will also generate a handler on the server endpoint to listen there, make the appropriate call to the Book model in this case, and return the result.

What about more complex queries like paginated results, filtered results, or protected routes?

You can do that too. For example, what if we only wanted books by 'Kurt Vonnegut'? We could change line 3 above to $books = books.findAll({ query: { author: 'Kurt Vonnegut' }}). The options object passed into findAll() can contain query, auth, and other commonly used parameters.

But now I'm locked in – what if I have something outside the box?

Skrapi makes these handy isomorphic actions available to use when they work. When you need something more complex, just do it how you always have. You can use the same Mongoose models and you can still use client side fetch calls to endpoints you create manually in your api/ directory as usual. Typically, your store file will include mostly Skrapi actions, and a couple of custom methods. Skrapi will 100% stay out of your way when these "shortcuts" aren't enough.