Dynamic LDP containers

[srosset81]

• Use "catch-all" routes which points to a new LdpApiService, which itself redirects to LdpResourceService, LdpContainerService or ActivityPubCollectionService depending on the @type.

Method	LDP resource	LDP container	ActivityStreams collection
GET *	ldp.resource.get	ldp.container.get	activitypub.collection.get
POST *	ldp.resource.post	ldp.container.post	(If defined by ControlledCollectionMixin)
PUT *	ldp.resource.put	-	-
PATCH *	ldp.resource.patch	ldp.container.patch	-
DELETE *	ldp.resource.delete	ldp.container.delete	-

• Allow to create and delete LDP containers via API (with an optional title and description). New tests have been added for that.
• New resourcesWithContainerPath option for the LdpService. If true, the URI of all new resources will include the container path, otherwise it will use the root container path (default: true)
• The auto-generated slug of ressource use the uuid/v4 library instead of MongoDB-like ObjectID. The previous method generated unique but easily guessable slugs.
• Use node-fetch instead of supertest for all API tests.

Breaking changes ⚠️

• The dereference option has been removed. Blank nodes are now automatically found (on 4 levels of depth).
• The VoID endpoint doesn't return blank nodes anymore (the data provider will be updated to automatically find blank nodes)
• The disassembly option has been removed. Use the new DisassemblyMixin, along with the ControlledContainerMixin, if you still need this feature (which will later be moved to the frontend)
• The undocumented ResourceWatcher bot has been removed. Use ControlledContainerMixin instead.
• The ldp.resource.get action now always return the semantic representation, even for binaries. The forceSemantic param is thus not necessary anymore. Binaries are handled by the LdpApiService. New tests have been added for that.
• API routes with catchAll: true are always handled last. This requires a modification of Moleculer's ApiGateway (see this Moleculer issue: Improve routes priorization moleculerjs/moleculer-web#335). If you don't use SemApps CoreService, you must add this method to the service using the ApiGateway mixin or the catch-all route of the LDP service will be handled before other endpoints like VOID, SPARQL, etc.

  methods: {
    optimizeRouteOrder() {
      // Overwrite optimization method to put catchAll routes at the end
      this.routes.sort((a, b) => (a.opts.catchAll ? 1 : -1));
      this.aliases.sort((a, b) => (a.route.opts.catchAll ? 1 : -1));
    }
  }

[simonLouvet]

hello @srosset81,
considering
The ldp.resource.get action now always return the semantic representation, even for binaries. The forceSemantic param is thus not necessary anymore. Binaries are handled by the LdpApiService. New tests have been added for that.
does the client have to request semantic data to obtain binary url?
does the location header return after post (201) is binary or semantic data url?

[srosset81]

When the frontend GET the binary through the API, it is returned as a binary. The change is only if you call ldp.resource.get from another Moleculer service.
No changes have been made about the Location header.