From bc25bf67e46340f8d2c1c9844bbcf28f4b04fc27 Mon Sep 17 00:00:00 2001
From: Kieran Eglin <kieran.eglin@gmail.com>
Date: Sun, 10 Mar 2024 15:10:46 -0700
Subject: [PATCH] Adds backfill job to startup tasks

---
 lib/pinchflat/application.ex                  |  4 +--
 lib/pinchflat/startup_tasks.ex                | 15 ++++++++-
 lib/pinchflat/workers/data_backfill_worker.ex | 18 +++++++++-
 .../workers/data_backfill_worker_test.exs     | 33 +++++++++++++++++++
 4 files changed, 66 insertions(+), 4 deletions(-)

diff --git a/lib/pinchflat/application.ex b/lib/pinchflat/application.ex
index 1527d7d..6f838bc 100644
--- a/lib/pinchflat/application.ex
+++ b/lib/pinchflat/application.ex
@@ -10,9 +10,9 @@ defmodule Pinchflat.Application do
     children = [
       PinchflatWeb.Telemetry,
       Pinchflat.Repo,
-      # {Task, &run_startup_tasks/0},
-      Pinchflat.StartupTasks,
+      # Must be before startup tasks
       {Oban, Application.fetch_env!(:pinchflat, Oban)},
+      Pinchflat.StartupTasks,
       {DNSCluster, query: Application.get_env(:pinchflat, :dns_cluster_query) || :ignore},
       {Phoenix.PubSub, name: Pinchflat.PubSub},
       # Start the Finch HTTP client for sending emails
diff --git a/lib/pinchflat/startup_tasks.ex b/lib/pinchflat/startup_tasks.ex
index e4b1aec..8e4be23 100644
--- a/lib/pinchflat/startup_tasks.ex
+++ b/lib/pinchflat/startup_tasks.ex
@@ -8,8 +8,11 @@ defmodule Pinchflat.StartupTasks do
 
   # restart: :temporary means that this process will never be restarted (ie: will run once and then die)
   use GenServer, restart: :temporary
+  import Ecto.Query, warn: false
 
+  alias Pinchflat.Repo
   alias Pinchflat.Settings
+  alias Pinchflat.Workers.DataBackfillWorker
 
   def start_link(opts \\ []) do
     GenServer.start_link(__MODULE__, %{}, opts)
@@ -21,11 +24,13 @@ defmodule Pinchflat.StartupTasks do
   Any code defined here will run every time the application starts. You must
   make sure that the code is idempotent and safe to run multiple times.
 
-  This is a good place to set up default settings, create initial records, stuff like that
+  This is a good place to set up default settings, create initial records, stuff like that.
+  Should be fast - anything with the potential to be slow should be kicked off as a job instead.
   """
   @impl true
   def init(state) do
     apply_default_settings()
+    enqueue_backfill_worker()
 
     {:ok, state}
   end
@@ -34,4 +39,12 @@ defmodule Pinchflat.StartupTasks do
     Settings.fetch!(:onboarding, true)
     Settings.fetch!(:pro_enabled, false)
   end
+
+  defp enqueue_backfill_worker do
+    DataBackfillWorker.cancel_pending_backfill_jobs()
+
+    %{}
+    |> DataBackfillWorker.new()
+    |> Repo.insert_unique_job()
+  end
 end
diff --git a/lib/pinchflat/workers/data_backfill_worker.ex b/lib/pinchflat/workers/data_backfill_worker.ex
index bd4d949..15a7041 100644
--- a/lib/pinchflat/workers/data_backfill_worker.ex
+++ b/lib/pinchflat/workers/data_backfill_worker.ex
@@ -14,11 +14,24 @@ defmodule Pinchflat.Workers.DataBackfillWorker do
   # I'm just trying out that pattern and seeing if I like it better
   # so this may change.
   import Ecto.Query, warn: false
+  require Logger
 
   alias __MODULE__
   alias Pinchflat.Repo
   alias Pinchflat.Media.MediaItem
 
+  @doc """
+  Cancels all pending backfill jobs. Useful for ensuring worker runs immediately
+  on app boot.
+
+  Returns {:ok, integer()}
+  """
+  def cancel_pending_backfill_jobs do
+    Oban.Job
+    |> where(worker: "Pinchflat.Workers.DataBackfillWorker")
+    |> Oban.cancel_all_jobs()
+  end
+
   @impl Oban.Worker
   @doc """
   Performs one-off tasks to get data in the right shape.
@@ -30,6 +43,7 @@ defmodule Pinchflat.Workers.DataBackfillWorker do
   Returns :ok
   """
   def perform(%Oban.Job{}) do
+    Logger.info("Running data backfill worker")
     backfill_shorts_data()
 
     reschedule_backfill()
@@ -45,7 +59,9 @@ defmodule Pinchflat.Workers.DataBackfillWorker do
         where: m.short_form_content == false
       )
 
-    Repo.update_all(query, set: [short_form_content: true])
+    {count, _} = Repo.update_all(query, set: [short_form_content: true])
+
+    Logger.info("Backfill worker set short_form_content to true for #{count} media items.")
   end
 
   defp reschedule_backfill do
diff --git a/test/pinchflat/workers/data_backfill_worker_test.exs b/test/pinchflat/workers/data_backfill_worker_test.exs
index 470035e..856bfb6 100644
--- a/test/pinchflat/workers/data_backfill_worker_test.exs
+++ b/test/pinchflat/workers/data_backfill_worker_test.exs
@@ -4,8 +4,41 @@ defmodule Pinchflat.Workers.DataBackfillWorkerTest do
   import Pinchflat.MediaFixtures
 
   alias Pinchflat.Workers.DataBackfillWorker
+  alias Pinchflat.Workers.FilesystemDataWorker
+
+  describe "cancel_pending_backfill_jobs/0" do
+    test "cancels all pending backfill jobs" do
+      %{}
+      |> DataBackfillWorker.new()
+      |> Repo.insert_unique_job()
+
+      assert_enqueued(worker: DataBackfillWorker)
+
+      DataBackfillWorker.cancel_pending_backfill_jobs()
+
+      refute_enqueued(worker: DataBackfillWorker)
+    end
+
+    test "does not cancel jobs for other workers" do
+      %{id: 0}
+      |> FilesystemDataWorker.new()
+      |> Repo.insert_unique_job()
+
+      assert_enqueued(worker: FilesystemDataWorker)
+
+      DataBackfillWorker.cancel_pending_backfill_jobs()
+
+      assert_enqueued(worker: FilesystemDataWorker)
+    end
+  end
 
   describe "perform/1" do
+    setup do
+      DataBackfillWorker.cancel_pending_backfill_jobs()
+
+      :ok
+    end
+
     test "reschedules itself once complete" do
       perform_job(DataBackfillWorker, %{})