Support for containers

[YuukanOO]

Sometimes the same kind of resource is split in multiple documents tied together by containers. For example, blog posts could live in paths blog/year/month/posts.ttl.

As of now, repositories manage a single resource (dataset). We could extend it to allow source to be a container url (in the example given above, source could be <baseurl>/blog). If it's a container, we can determine the dataset needed to persist a data by asking for a (data: TData) => string to return a relative path given a data to persist. With this function, we can determine the needed path from the source root container.

The repository should then create containers and dataset as needed to save the final stuff. It should also enable data to be moved between dataset by checking if the data.url !== generatedUrl.

For type index, we may determine if it's a container with the instanceContainer ?

[YuukanOO]

For example, if you have blog posts stored by date (http://pod/posts/2019/posts.ttl, http://pod/posts/2020/posts.ttl), you will be able to provide the root (http://pod/posts/) ldp container and an option along the lines:

interface Post {
  permalink: string
  title: string
  date: Date
}

const repo = new Repository<Post>({
  source: "http://pod/posts/",
  // The line below provide informations on how to get a dataset path based on a data to persist
  resolveTo: (post: Post): string => `${post.date.getFullYear()}/posts.ttl`,
  type: "...",
  schema: { ... }
});

And the repository will take care of the rest : creating intermediate containers if it needs to, retrieving posts in nested documents and so on without the developer having to do it itself.