[Enhancement] Redownload new media after a delay (#173)

* Added redownload-related columns

* Added methods for fetching re-downloadable media items

* Filled out redownload worker + tests

* Added redownload worker to config.exs cron

* Added to UI and README

lib/pinchflat/downloading/media_download_worker.ex

@@ -35,14 +35,17 @@ defmodule Pinchflat.Downloading.MediaDownloadWorker do
   """
   @impl Oban.Worker
   def perform(%Oban.Job{args: %{"id" => media_item_id} = args}) do
+    should_force = Map.get(args, "force", false)
+    is_redownload = Map.get(args, "redownload?", false)
+
     media_item =
       media_item_id
       |> Media.get_media_item!()
       |> Repo.preload(:source)
 
     # If the source or media item is set to not download media, perform a no-op unless forced
-    if (media_item.source.download_media && !media_item.prevent_download) || args["force"] do
-      download_media_and_schedule_jobs(media_item)
+    if (media_item.source.download_media && !media_item.prevent_download) || should_force do
+      download_media_and_schedule_jobs(media_item, is_redownload)
     else
       :ok
     end
@@ -51,10 +54,13 @@ defmodule Pinchflat.Downloading.MediaDownloadWorker do
     Ecto.StaleEntryError -> Logger.info("#{__MODULE__} discarded: media item #{media_item_id} stale")
   end
 
-  defp download_media_and_schedule_jobs(media_item) do
+  defp download_media_and_schedule_jobs(media_item, is_redownload) do
     case MediaDownloader.download_for_media_item(media_item) do
       {:ok, updated_media_item} ->
-        compute_and_save_media_filesize(updated_media_item)
+        Media.update_media_item(updated_media_item, %{
+          media_size_bytes: compute_media_filesize(updated_media_item),
+          media_redownloaded_at: get_redownloaded_at(is_redownload)
+        })
 
         {:ok, updated_media_item}
 
@@ -66,13 +72,21 @@ defmodule Pinchflat.Downloading.MediaDownloadWorker do
     end
   end
 
-  defp compute_and_save_media_filesize(media_item) do
+  defp compute_media_filesize(media_item) do
     case File.stat(media_item.media_filepath) do
       {:ok, %{size: size}} ->
-        Media.update_media_item(media_item, %{media_size_bytes: size})
+        size
 
       _ ->
-        :ok
+        nil
+    end
+  end
+
+  defp get_redownloaded_at(is_redownload) do
+    if is_redownload do
+      DateTime.utc_now()
+    else
+      nil
     end
   end
 end

lib/pinchflat/downloading/media_redownload_worker.ex

@@ -0,0 +1,31 @@
+defmodule Pinchflat.Downloading.MediaRedownloadWorker do
+  @moduledoc false
+
+  use Oban.Worker,
+    queue: :media_fetching,
+    unique: [period: :infinity, states: [:available, :scheduled, :retryable, :executing]],
+    tags: ["media_item", "media_fetching"]
+
+  require Logger
+
+  alias Pinchflat.Media
+  alias Pinchflat.Downloading.MediaDownloadWorker
+
+  @doc """
+  Redownloads media items that are eligible for redownload.
+
+  This worker is scheduled to run daily via the Oban Cron plugin
+  and it should run _after_ the retention worker.
+
+  Returns :ok
+  """
+  @impl Oban.Worker
+  def perform(%Oban.Job{}) do
+    redownloadable_media = Media.list_redownloadable_media_items()
+    Logger.info("Redownloading #{length(redownloadable_media)} media items")
+
+    Enum.each(redownloadable_media, fn media_item ->
+      MediaDownloadWorker.kickoff_with_task(media_item, %{redownload?: true})
+    end)
+  end
+end

lib/pinchflat/media/media.ex

@@ -31,15 +31,39 @@ defmodule Pinchflat.Media do
   def list_cullable_media_items do
     MediaQuery.new()
     |> MediaQuery.with_media_filepath()
-    |> MediaQuery.with_passed_retention_period()
-    |> MediaQuery.with_no_culling_prevention()
+    |> MediaQuery.where_past_retention_period()
+    |> MediaQuery.where_culling_not_prevented()
+    |> Repo.all()
+  end
+
+  @doc """
+  Returns a list of media_items that are redownloadable based on the redownload delay
+  of the media_profile their source belongs to.
+
+  The logic is that a media_item is past_redownload_delay if the media_item's
+  upload_date is at least redownload_delay_days ago AND
+  `media_downloaded_at` - `redownload_delay_days` is before the media_item's `upload_date`.
+  This logic grabs media that we've recently downloaded AND is recently uploaded, but
+  doesn't grab media that we've recently downloaded and was uploaded a long time ago.
+  This also makes things work as expected when downloading media from a source for the
+  first time.
+
+  Returns [%MediaItem{}, ...]
+  """
+  def list_redownloadable_media_items do
+    MediaQuery.new()
+    |> MediaQuery.with_media_downloaded_at()
+    |> MediaQuery.where_download_not_prevented()
+    |> MediaQuery.where_not_culled()
+    |> MediaQuery.where_media_not_redownloaded()
+    |> MediaQuery.where_past_redownload_delay()
     |> Repo.all()
   end
 
   @doc """
   Returns a list of pending media_items for a given source, where
   pending means the `media_filepath` is `nil` AND the media_item
-  matches satisfies `MediaQuery.with_media_pending_download`. You
+  matches satisfies `MediaQuery.where_pending_download`. You
   should really check out that function if you need to know more
   because it has a lot going on.
 
@@ -48,7 +72,7 @@ defmodule Pinchflat.Media do
   def list_pending_media_items_for(%Source{} = source) do
     MediaQuery.new()
     |> MediaQuery.for_source(source)
-    |> MediaQuery.with_media_pending_download()
+    |> MediaQuery.where_pending_download()
     |> Repo.all()
   end
 
@@ -66,7 +90,7 @@ defmodule Pinchflat.Media do
 
     MediaQuery.new()
     |> MediaQuery.with_id(media_item.id)
-    |> MediaQuery.with_media_pending_download()
+    |> MediaQuery.where_pending_download()
     |> Repo.exists?()
   end
 

lib/pinchflat/media/media_item.ex

@@ -34,7 +34,8 @@ defmodule Pinchflat.Media.MediaItem do
     # These are user or system controlled fields
     :prevent_download,
     :prevent_culling,
-    :culled_at
+    :culled_at,
+    :media_redownloaded_at
   ]
   # Pretty much all the fields captured at index are required.
   @required_fields ~w(
@@ -61,6 +62,7 @@ defmodule Pinchflat.Media.MediaItem do
     field :livestream, :boolean, default: false
     field :short_form_content, :boolean, default: false
     field :media_downloaded_at, :utc_datetime
+    field :media_redownloaded_at, :utc_datetime
     field :upload_date, :date
     field :duration_seconds, :integer
 

lib/pinchflat/media/media_query.ex

@@ -15,7 +15,7 @@ defmodule Pinchflat.Media.MediaQuery do
   # Prefixes:
   # - for_* - belonging to a certain record
   # - join_* - for joining on a certain record
-  # - with_* - for filtering based on full, concrete attributes
+  # - with_*, where_* - for filtering based on full, concrete attributes
   # - matching_* - for filtering based on partial attributes (e.g. LIKE, regex, full-text search)
   #
   # Suffixes:
@@ -33,24 +33,46 @@ defmodule Pinchflat.Media.MediaQuery do
     from(mi in query, join: s in assoc(mi, :source), as: :sources)
   end
 
-  def with_passed_retention_period(query) do
+  def where_past_retention_period(query) do
     query
     |> require_assoc(:source)
     |> where(
       [mi, source],
-      fragment(
-        "IFNULL(?, 0) > 0 AND DATETIME('now', '-' || ? || ' day') > ?",
-        source.retention_period_days,
-        source.retention_period_days,
-        mi.media_downloaded_at
-      )
+      fragment("""
+      IFNULL(retention_period_days, 0) > 0 AND
+      DATETIME('now', '-' || retention_period_days || ' day') > media_downloaded_at
+      """)
     )
   end
 
-  def with_no_culling_prevention(query) do
+  def where_past_redownload_delay(query) do
+    query
+    |> require_assoc(:source)
+    |> require_assoc(:media_profile)
+    |> where(
+      [_mi, _source, _media_profile],
+      # Returns media items where the upload_date is at least redownload_delay_days ago AND
+      # downloaded_at minus the redownload_delay_days is before the upload date
+      fragment("""
+        IFNULL(redownload_delay_days, 0) > 0 AND
+        DATETIME('now', '-' || redownload_delay_days || ' day') > upload_date AND
+        DATETIME(media_downloaded_at, '-' || redownload_delay_days || ' day') < upload_date
+      """)
+    )
+  end
+
+  def where_culling_not_prevented(query) do
     where(query, [mi], mi.prevent_culling == false)
   end
 
+  def where_not_culled(query) do
+    where(query, [mi], is_nil(mi.culled_at))
+  end
+
+  def where_media_not_redownloaded(query) do
+    where(query, [mi], is_nil(mi.media_redownloaded_at))
+  end
+
   def with_id(query, id) do
     where(query, [mi], mi.id == ^id)
   end
@@ -59,6 +81,10 @@ defmodule Pinchflat.Media.MediaQuery do
     where(query, [mi], mi.media_id in ^media_ids)
   end
 
+  def with_media_downloaded_at(query) do
+    where(query, [mi], not is_nil(mi.media_downloaded_at))
+  end
+
   def with_media_filepath(query) do
     where(query, [mi], not is_nil(mi.media_filepath))
   end
@@ -73,7 +99,7 @@ defmodule Pinchflat.Media.MediaQuery do
     |> where([mi, source], is_nil(source.download_cutoff_date) or mi.upload_date >= source.download_cutoff_date)
   end
 
-  def with_no_prevented_download(query) do
+  def where_download_not_prevented(query) do
     where(query, [mi], mi.prevent_download == false)
   end
 
@@ -129,9 +155,9 @@ defmodule Pinchflat.Media.MediaQuery do
     )
   end
 
-  def with_media_pending_download(query) do
+  def where_pending_download(query) do
     query
-    |> with_no_prevented_download()
+    |> where_download_not_prevented()
     |> with_no_media_filepath()
     |> with_upload_date_after_source_cutoff()
     |> with_format_matching_profile_preference()

lib/pinchflat/notifications/source_notifications.ex

@@ -59,7 +59,7 @@ defmodule Pinchflat.Notifications.SourceNotifications do
   defp pending_media_item_count(source) do
     MediaQuery.new()
     |> MediaQuery.for_source(source)
-    |> MediaQuery.with_media_pending_download()
+    |> MediaQuery.where_pending_download()
     |> Repo.aggregate(:count)
   end
 

lib/pinchflat/profiles/media_profile.ex

@@ -26,12 +26,14 @@ defmodule Pinchflat.Profiles.MediaProfile do
     shorts_behaviour
     livestream_behaviour
     preferred_resolution
+    redownload_delay_days
   )a
 
   @required_fields ~w(name output_path_template)a
 
   schema "media_profiles" do
     field :name, :string
+    field :redownload_delay_days, :integer
 
     field :output_path_template, :string,
       default: "/{{ source_custom_name }}/{{ upload_yyyy_mm_dd }} {{ title }}/{{ title }} [{{ id }}].{{ ext }}"
@@ -60,7 +62,6 @@ defmodule Pinchflat.Profiles.MediaProfile do
     # See `build_format_clauses` in the Media context for more.
     field :shorts_behaviour, Ecto.Enum, values: ~w(include exclude only)a, default: :include
     field :livestream_behaviour, Ecto.Enum, values: ~w(include exclude only)a, default: :include
-
     field :preferred_resolution, Ecto.Enum, values: ~w(2160p 1080p 720p 480p 360p audio)a, default: :"1080p"
 
     has_many :sources, Source
@@ -75,6 +76,7 @@ defmodule Pinchflat.Profiles.MediaProfile do
     |> validate_required(@required_fields)
     # Ensures it ends with `.{{ ext }}` or `.%(ext)s` or similar (with a little wiggle room)
     |> validate_format(:output_path_template, ext_regex(), message: "must end with .{{ ext }}")
+    |> validate_number(:redownload_delay_days, greater_than_or_equal_to: 0)
     |> unique_constraint(:name)
   end
 

lib/pinchflat_web/controllers/media_profiles/media_profile_html/media_profile_form.html.heex

@@ -203,6 +203,17 @@
       />
     </section>
 
+    <section x-data="{ presets: { default: null, media_center: 1, audio: null, archiving: 1 } }">
+      <.input
+        field={f[:redownload_delay_days]}
+        type="number"
+        label="Redownload Delay (days)"
+        min="0"
+        help="Delay in days until new media is redownloaded. Redownloading new media can improve its quality or SponsorBlock tags. Leave blank to not redownload"
+        x-init="$watch('selectedPreset', p => p && ($el.value = presets[p]))"
+      />
+    </section>
+
     <h3 class="mt-8 text-2xl text-black dark:text-white">
       Media Center Options
     </h3>

lib/pinchflat_web/controllers/sources/source_controller.ex

@@ -63,7 +63,7 @@ defmodule PinchflatWeb.Sources.SourceController do
     pending_media =
       MediaQuery.new()
       |> MediaQuery.for_source(source)
-      |> MediaQuery.with_media_pending_download()
+      |> MediaQuery.where_pending_download()
       |> order_by(desc: :id)
       |> limit(100)
       |> Repo.all()