RichMedia backfill processing through Oban

lib/pleroma/web/mastodon_api/views/status_view.ex

@@ -8,7 +8,6 @@ defmodule Pleroma.Web.MastodonAPI.StatusView do
   require Pleroma.Constants
 
   alias Pleroma.Activity
-  alias Pleroma.Config
   alias Pleroma.HTML
   alias Pleroma.Maps
   alias Pleroma.Object
@@ -31,13 +30,7 @@ defmodule Pleroma.Web.MastodonAPI.StatusView do
   # pagination is restricted to 40 activities at a time
   defp fetch_rich_media_for_activities(activities) do
     Enum.each(activities, fn activity ->
-      fun = fn -> Card.get_by_activity(activity) end
-
-      if Config.get([__MODULE__, :sync_fetching], false) do
-        fun.()
-      else
-        spawn(fun)
-      end
+      Card.get_by_activity(activity)
     end)
   end
 

lib/pleroma/web/rich_media/backfill.ex

@@ -10,31 +10,21 @@ defmodule Pleroma.Web.RichMedia.Backfill do
 
   require Logger
 
-  @backfiller Pleroma.Config.get([__MODULE__, :provider], Pleroma.Web.RichMedia.Backfill.Task)
   @cachex Pleroma.Config.get([:cachex, :provider], Cachex)
-  @max_attempts 3
-  @retry 5_000
 
-  def start(%{url: url} = args) when is_binary(url) do
+  @spec run(map()) ::
+          :ok | {:error, {:invalid_metadata, any()} | :body_too_large | {:content, any()} | any()}
+  def run(%{"url" => url} = args) do
     url_hash = Card.url_to_hash(url)
 
-    args =
-      args
-      |> Map.put(:attempt, 1)
-      |> Map.put(:url_hash, url_hash)
-
-    @backfiller.run(args)
-  end
-
-  def run(%{url: url, url_hash: url_hash, attempt: attempt} = args)
-      when attempt <= @max_attempts do
     case Parser.parse(url) do
       {:ok, fields} ->
         {:ok, card} = Card.create(url, fields)
 
         maybe_schedule_expiration(url, fields)
 
-        if Map.has_key?(args, :activity_id) do
+        with %{"activity_id" => activity_id} <- args,
+             false <- is_nil(activity_id) do
           stream_update(args)
         end
 
@@ -54,19 +44,10 @@ defmodule Pleroma.Web.RichMedia.Backfill do
 
       e ->
         Logger.debug("Rich media error for #{url}: #{inspect(e)}")
-
-        :timer.sleep(@retry * attempt)
-
-        run(%{args | attempt: attempt + 1})
+        {:error, e}
     end
   end
 
-  def run(%{url: url, url_hash: url_hash}) do
-    Logger.debug("Rich media failure for #{url}")
-
-    negative_cache(url_hash, :timer.minutes(15))
-  end
-
   defp maybe_schedule_expiration(url, fields) do
     case TTL.process(fields, url) do
       {:ok, ttl} when is_number(ttl) ->
@@ -80,22 +61,14 @@ defmodule Pleroma.Web.RichMedia.Backfill do
     end
   end
 
-  defp stream_update(%{activity_id: activity_id}) do
+  defp stream_update(%{"activity_id" => activity_id}) do
     Pleroma.Activity.get_by_id(activity_id)
     |> Pleroma.Activity.normalize()
     |> Pleroma.Web.ActivityPub.ActivityPub.stream_out()
   end
 
   defp warm_cache(key, val), do: @cachex.put(:rich_media_cache, key, val)
-  defp negative_cache(key, ttl \\ nil), do: @cachex.put(:rich_media_cache, key, nil, ttl: ttl)
-end
 
-defmodule Pleroma.Web.RichMedia.Backfill.Task do
-  alias Pleroma.Web.RichMedia.Backfill
-
-  def run(args) do
-    Task.Supervisor.start_child(Pleroma.TaskSupervisor, Backfill, :run, [args],
-      name: {:global, {:rich_media, args.url_hash}}
-    )
-  end
+  defp negative_cache(key, ttl \\ :timer.minutes(15)),
+    do: @cachex.put(:rich_media_cache, key, nil, ttl: ttl)
 end

lib/pleroma/web/rich_media/card.ex

@@ -7,8 +7,8 @@ defmodule Pleroma.Web.RichMedia.Card do
   alias Pleroma.HTML
   alias Pleroma.Object
   alias Pleroma.Repo
-  alias Pleroma.Web.RichMedia.Backfill
   alias Pleroma.Web.RichMedia.Parser
+  alias Pleroma.Workers.RichMediaWorker
 
   @cachex Pleroma.Config.get([:cachex, :provider], Cachex)
   @config_impl Application.compile_env(:pleroma, [__MODULE__, :config_impl], Pleroma.Config)
@@ -75,17 +75,18 @@ defmodule Pleroma.Web.RichMedia.Card do
 
   def get_by_url(nil), do: nil
 
-  @spec get_or_backfill_by_url(String.t(), map()) :: t() | nil
-  def get_or_backfill_by_url(url, backfill_opts \\ %{}) do
+  @spec get_or_backfill_by_url(String.t(), keyword()) :: t() | nil
+  def get_or_backfill_by_url(url, opts \\ []) do
     if @config_impl.get([:rich_media, :enabled]) do
       case get_by_url(url) do
         %__MODULE__{} = card ->
           card
 
         nil ->
-          backfill_opts = Map.put(backfill_opts, :url, url)
+          activity_id = Keyword.get(opts, :activity, nil)
 
-          Backfill.start(backfill_opts)
+          RichMediaWorker.new(%{"op" => "backfill", "url" => url, "activity_id" => activity_id})
+          |> Oban.insert()
 
           nil
 
@@ -137,7 +138,7 @@ defmodule Pleroma.Web.RichMedia.Card do
       nil
     else
       {:cached, url} ->
-        get_or_backfill_by_url(url, %{activity_id: activity.id})
+        get_or_backfill_by_url(url, activity_id: activity.id)
 
       _ ->
         :error

lib/pleroma/workers/rich_media_worker.ex

@@ -3,13 +3,17 @@
 # SPDX-License-Identifier: AGPL-3.0-only
 
 defmodule Pleroma.Workers.RichMediaWorker do
+  alias Pleroma.Web.RichMedia.Backfill
   alias Pleroma.Web.RichMedia.Card
 
-  use Oban.Worker,
-    queue: :background
+  use Oban.Worker, queue: :background, max_attempts: 3, unique: [period: 300]
 
   @impl Oban.Worker
   def perform(%Job{args: %{"op" => "expire", "url" => url} = _args}) do
     Card.delete(url)
   end
+
+  def perform(%Job{args: %{"op" => "backfill", "url" => _url} = args}) do
+    Backfill.run(args)
+  end
 end