tfwright / spur

Activity tracking for Ecto

Spur

Loosely based on chaps-io/public_activity, a very simple utility for quickly setting up an activity stream in your Elixir/Ecto app.

More detailed examples of configuration and usage are in the tests. See this thread for more information on the idea behind Spur and the changelog.

Installation

Basic steps are

1. Add Spur to your application deps.

2. Tell Spur which Ecto Repo to use:

   config :spur, repo: MyApp.Repo
   

3. Generate and run a migration that adds an "activities" table to your repo (see priv/test/migrations). To use a different table name, set the activities_table_name config.

That's enough to start tracking arbitrary activities:

%Spur.Activity{actor: "caesar", action: "came", object: "your-county", meta: %{also: ["saw", "conquered"]}}

Fields are based on https://www.w3.org/TR/activitystreams-core/#example-1

Getting fancy

"Callbacks"

If you want to make use of automatic tracking of inserts, updates and deletes, make sure your objects implement the required fields as functions:

defmodule Battle do
  defimpl Spur.Trackable, for: __MODULE__ do
    def actor(war), do: "Accounts.User:#{war.general_id}"
    def object(war), do: "war:#{war.id}"
    def target(_war), do: nil
  end
end

Now instead of using Repo to perform your operation, use Spur instead:

%MyApp.Battle{general_id: 5}
|> MyApp.Battle.changeset
|> Spur.insert

In this example, a record for both your Battle and an Activity with action set to insert will be stored in the DB. Of course, the Battle fails validations, neither will be inserted and the changeset will be returned with errors, just as Repo would. Otherwise it will return the Battle. (Note: As of 0.3.0, Spur supports the expose_transactions config, which when sets to true returns the raw Ecto transaction. Us this if you need to access the created Activity struct.)

Each of these callback functions also take a Map of properties that will be added to the Activity record, or a Function that returns a Map. You can use this to set or override default Activity data in the callback itself. If, for example, the actor for a given schema is not stored in the Repo, you may want to use the logged in user instead:

Spur.insert(changeset, %{actor: conn.assigns.current_user.id})

Audience

To automatically associate the Activity with an audience requires a bit of extra configuration:

1. Add audience_module to your app's Spur config: audience_module: MyApp.Accounts.User
2. Add a many_to_many association between your audience module's Ecto schema. By default Spur expects this to be named :activities. If you want to name it something else, add another line to the config: audience_assoc_name: :events.
3. Finally, make sure that your trackable objects implement audience. It should return either an Ecto query or a plain list of the audience structs configured with the above association.

Now when you use one of the callback's above to track an object, the resulting Activity will automatically be associated with the audience records returned for that object:

  # SpurTest.TrackableStruct
  def audience(trackable_struct), do: Ecto.assoc(trackable_struct, :watchers)

  [watcher] = trackable_struct.watchers

  Ecto.Changeset.change(trackable_struct)
  |> Spur.update

  [%Spur.Activity{action: update}] = watcher.activities

---

<<Someone who finds a trace [Spur] certainly also knows that something has existed before and is now left behind. But one does not just take note of this. One begins to search and to ask oneself where it leads.>>

Hans Georg-Gadamer, "Hermeneutik auf dem Spur"