[Enhancement] Media from before a source's cutoff date is now automatically deleted (#329)

* Changed the purpose of the 'culled_at' flag. renamed some methods

* implemented new culling behaviour for source cutoff dates

* Removed culled_at flag if a media item gets redownloaded

* Updated source form

* Removed unused method

lib/pinchflat/downloading/downloading_helpers.ex

@@ -91,8 +91,7 @@ defmodule Pinchflat.Downloading.DownloadingHelpers do
         [m, s, mp],
         ^MediaQuery.for_source(source) and
           ^MediaQuery.downloaded() and
-          not (^MediaQuery.download_prevented()) and
-          not (^MediaQuery.culled())
+          not (^MediaQuery.download_prevented())
       )
     )
     |> Repo.all()

lib/pinchflat/downloading/media_downloader.ex

@@ -79,6 +79,7 @@ defmodule Pinchflat.Downloading.MediaDownloader do
       |> MetadataParser.parse_for_media_item()
       |> Map.merge(%{
         media_downloaded_at: DateTime.utc_now(),
+        culled_at: nil,
         nfo_filepath: determine_nfo_filepath(media_with_preloads, parsed_json),
         metadata: %{
           # IDEA: might be worth kicking off a job for this since thumbnail fetching

lib/pinchflat/downloading/media_quality_upgrade_worker.ex

@@ -23,10 +23,10 @@ defmodule Pinchflat.Downloading.MediaQualityUpgradeWorker do
   """
   @impl Oban.Worker
   def perform(%Oban.Job{}) do
-    redownloadable_media = Media.list_redownloadable_media_items()
-    Logger.info("Redownloading #{length(redownloadable_media)} media items")
+    upgradable_media = Media.list_upgradeable_media_items()
+    Logger.info("Redownloading #{length(upgradable_media)} media items")
 
-    Enum.each(redownloadable_media, fn media_item ->
+    Enum.each(upgradable_media, fn media_item ->
       MediaDownloadWorker.kickoff_with_task(media_item, %{quality_upgrade?: true})
     end)
   end

lib/pinchflat/downloading/media_retention_worker.ex

@@ -6,8 +6,11 @@ defmodule Pinchflat.Downloading.MediaRetentionWorker do
     unique: [period: :infinity, states: [:available, :scheduled, :retryable, :executing]],
     tags: ["media_item", "local_metadata"]
 
+  use Pinchflat.Media.MediaQuery
+
   require Logger
 
+  alias Pinchflat.Repo
   alias Pinchflat.Media
 
   @doc """
@@ -20,14 +23,51 @@ defmodule Pinchflat.Downloading.MediaRetentionWorker do
   """
   @impl Oban.Worker
   def perform(%Oban.Job{}) do
-    cullable_media = Media.list_cullable_media_items()
+    cull_cullable_media_items()
+    delete_media_items_from_before_cutoff()
+
+    :ok
+  end
+
+  defp cull_cullable_media_items do
+    cullable_media =
+      MediaQuery.new()
+      |> MediaQuery.require_assoc(:source)
+      |> where(^MediaQuery.cullable())
+      |> Repo.all()
+
     Logger.info("Culling #{length(cullable_media)} media items past their retention date")
 
     Enum.each(cullable_media, fn media_item ->
+      # Setting `prevent_download` does what it says on the tin, but `culled_at` is purely informational.
+      # We don't actually do anything with that in terms of queries and it gets set to nil if the media item
+      # gets re-downloaded.
       Media.delete_media_files(media_item, %{
         prevent_download: true,
         culled_at: DateTime.utc_now()
       })
     end)
   end
+
+  defp delete_media_items_from_before_cutoff do
+    deletable_media =
+      MediaQuery.new()
+      |> MediaQuery.require_assoc(:source)
+      |> where(^MediaQuery.deletable_based_on_source_cutoff())
+      |> Repo.all()
+
+    Logger.info("Deleting #{length(deletable_media)} media items that are from before the source cutoff")
+
+    Enum.each(deletable_media, fn media_item ->
+      # Note that I'm not setting `prevent_download` on the media_item here.
+      # That's because cutoff_date can easily change and it's a valid behavior to re-download older
+      # media items if the cutoff_date changes.
+      # Download is ultimately prevented because `MediaQuery.pending()` only returns media items
+      # from after the cutoff date (among other things), so it's not like the media will just immediately
+      # be re-downloaded.
+      Media.delete_media_files(media_item, %{
+        culled_at: DateTime.utc_now()
+      })
+    end)
+  end
 end

lib/pinchflat/media/media.ex

@@ -25,21 +25,10 @@ defmodule Pinchflat.Media do
   end
 
   @doc """
-  Returns a list of media_items that are cullable based on the retention period
-  of the source they belong to.
-
-  Returns [%MediaItem{}, ...]
-  """
-  def list_cullable_media_items do
-    MediaQuery.new()
-    |> MediaQuery.require_assoc(:source)
-    |> where(^MediaQuery.cullable())
-    |> 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.
+  Returns a list of media_items that are upgradeable based on the redownload delay
+  of the media_profile their source belongs to. In this context, upgradeable means
+  that it's been long enough since upload that the video may be in a higher quality
+  or have better sponsorblock segments (or similar).
 
   The logic is that a media_item is past_redownload_delay if the media_item's uploaded_at is
   at least redownload_delay_days ago AND `media_downloaded_at` - `redownload_delay_days`
@@ -52,10 +41,10 @@ defmodule Pinchflat.Media do
 
   Returns [%MediaItem{}, ...]
   """
-  def list_redownloadable_media_items do
+  def list_upgradeable_media_items do
     MediaQuery.new()
     |> MediaQuery.require_assoc(:media_profile)
-    |> where(^MediaQuery.redownloadable())
+    |> where(^MediaQuery.upgradeable())
     |> Repo.all()
   end
 

lib/pinchflat/media/media_query.ex

@@ -33,7 +33,6 @@ defmodule Pinchflat.Media.MediaQuery do
   def downloaded, do: dynamic([mi], not is_nil(mi.media_filepath))
   def download_prevented, do: dynamic([mi], mi.prevent_download == true)
   def culling_prevented, do: dynamic([mi], mi.prevent_culling == true)
-  def culled, do: dynamic([mi], not is_nil(mi.culled_at))
   def redownloaded, do: dynamic([mi], not is_nil(mi.media_redownloaded_at))
   def upload_date_matches(other_date), do: dynamic([mi], fragment("date(?) = date(?)", mi.uploaded_at, ^other_date))
 
@@ -108,6 +107,15 @@ defmodule Pinchflat.Media.MediaQuery do
     )
   end
 
+  def deletable_based_on_source_cutoff do
+    dynamic(
+      [mi, source],
+      ^downloaded() and
+        not (^upload_date_after_source_cutoff()) and
+        not (^culling_prevented())
+    )
+  end
+
   def pending do
     dynamic(
       [mi],
@@ -119,12 +127,11 @@ defmodule Pinchflat.Media.MediaQuery do
     )
   end
 
-  def redownloadable do
+  def upgradeable do
     dynamic(
       [mi, source],
       ^downloaded() and
         not (^download_prevented()) and
-        not (^culled()) and
         not (^redownloaded()) and
         ^past_redownload_delay()
     )

lib/pinchflat_web/controllers/sources/source_html/source_form.html.heex

@@ -94,7 +94,7 @@
       maxlength="10"
       pattern="((?:19|20)[0-9][0-9])-(0[1-9]|1[012])-(0[1-9]|[12][0-9]|3[01])"
       title="YYYY-MM-DD"
-      help="Only download media uploaded after this date. Leave blank to download all media. Must be in YYYY-MM-DD format"
+      help="Only download media uploaded after this date. Leave blank to download all media. Must be in YYYY-MM-DD format. Old media may be deleted or downloaded if you change this date"
     />
 
     <.input