[Enhancement] Add rate limiting to yt-dlp requests; prevent saving Media Items when throttled by YouTube (#559)

* Added sleep interval to settings

* Added new sleep setting to yt-dlp runner and added tests

* Added setting for form; updated setting name

* Updated form label

* Prevented saving/updating of media items if being throttled by youtube

* Added the bot message to the list of non-retryable errors

* Fixed typo

lib/pinchflat/downloading/media_download_worker.ex

@@ -115,7 +115,7 @@ defmodule Pinchflat.Downloading.MediaDownloadWorker do
   defp action_on_error(message) do
     # This will attempt re-download at the next indexing, but it won't be retried
     # immediately as part of job failure logic
-    non_retryable_errors = ["Video unavailable"]
+    non_retryable_errors = ["Video unavailable", "Sign in to confirm"]
 
     if String.contains?(to_string(message), non_retryable_errors) do
       Logger.error("yt-dlp download will not be retried: #{inspect(message)}")

lib/pinchflat/media/media_item.ex

@@ -112,6 +112,9 @@ defmodule Pinchflat.Media.MediaItem do
     |> dynamic_default(:uuid, fn _ -> Ecto.UUID.generate() end)
     |> update_upload_date_index()
     |> validate_required(@required_fields)
+    # Validate that the title does NOT start with "youtube video #" since that indicates a restriction by YouTube.
+    # See issue #549 for more information.
+    |> validate_format(:title, ~r/^(?!youtube video #)/)
     |> unique_constraint([:media_id, :source_id])
   end
 

lib/pinchflat/settings/setting.ex

@@ -14,15 +14,17 @@ defmodule Pinchflat.Settings.Setting do
     :apprise_server,
     :video_codec_preference,
     :audio_codec_preference,
-    :youtube_api_key
+    :youtube_api_key,
+    :extractor_sleep_interval_seconds
   ]
 
-  @required_fields ~w(
-    onboarding
-    pro_enabled
-    video_codec_preference
-    audio_codec_preference
-  )a
+  @required_fields [
+    :onboarding,
+    :pro_enabled,
+    :video_codec_preference,
+    :audio_codec_preference,
+    :extractor_sleep_interval_seconds
+  ]
 
   schema "settings" do
     field :onboarding, :boolean, default: true
@@ -32,6 +34,7 @@ defmodule Pinchflat.Settings.Setting do
     field :apprise_server, :string
     field :youtube_api_key, :string
     field :route_token, :string
+    field :extractor_sleep_interval_seconds, :integer, default: 0
 
     field :video_codec_preference, :string
     field :audio_codec_preference, :string
@@ -42,5 +45,6 @@ defmodule Pinchflat.Settings.Setting do
     setting
     |> cast(attrs, @allowed_fields)
     |> validate_required(@required_fields)
+    |> validate_number(:extractor_sleep_interval_seconds, greater_than: 0)
   end
 end

lib/pinchflat/sources/sources.ex

@@ -180,9 +180,12 @@ defmodule Pinchflat.Sources do
   end
 
   defp add_source_details_to_changeset(source, changeset) do
+    original_url = changeset.changes.original_url
     use_cookies = Ecto.Changeset.get_field(changeset, :use_cookies)
+    # Skipping sleep interval since this is UI blocking and we want to keep this as fast as possible
+    addl_opts = [use_cookies: use_cookies, skip_sleep_interval: true]
 
-    case MediaCollection.get_source_details(changeset.changes.original_url, [], use_cookies: use_cookies) do
+    case MediaCollection.get_source_details(original_url, [], addl_opts) do
       {:ok, source_details} ->
         add_source_details_by_collection_type(source, changeset, source_details)
 

lib/pinchflat/utils/number_utils.ex

@@ -36,4 +36,18 @@ defmodule Pinchflat.Utils.NumberUtils do
       end
     end)
   end
+
+  @doc """
+  Adds jitter to a number based on a percentage. Returns 0 if the number is less than or equal to 0.
+
+  Returns integer()
+  """
+  def add_jitter(num, jitter_percentage \\ 0.5)
+  def add_jitter(num, _jitter_percentage) when num <= 0, do: 0
+
+  def add_jitter(num, jitter_percentage) do
+    jitter = :rand.uniform(round(num * jitter_percentage))
+
+    round(num + jitter)
+  end
 end

lib/pinchflat/yt_dlp/command_runner.ex

@@ -5,7 +5,9 @@ defmodule Pinchflat.YtDlp.CommandRunner do
 
   require Logger
 
+  alias Pinchflat.Settings
   alias Pinchflat.Utils.CliUtils
+  alias Pinchflat.Utils.NumberUtils
   alias Pinchflat.YtDlp.YtDlpCommandRunner
   alias Pinchflat.Utils.FilesystemUtils, as: FSUtils
 
@@ -22,23 +24,23 @@ defmodule Pinchflat.YtDlp.CommandRunner do
       for a file watcher.
     - :use_cookies - if true, will add a cookie file to the command options. Will not
       attach a cookie file if the user hasn't set one up.
+    - :skip_sleep_interval - if true, will not add the sleep interval options to the command.
+      Usually only used for commands that would be UI-blocking
 
   Returns {:ok, binary()} | {:error, output, status}.
   """
   @impl YtDlpCommandRunner
   def run(url, action_name, command_opts, output_template, addl_opts \\ []) do
     Logger.debug("Running yt-dlp command for action: #{action_name}")
-    # This approach lets us mock the command for testing
-    command = backend_executable()
 
     output_filepath = generate_output_filepath(addl_opts)
     print_to_file_opts = [{:print_to_file, output_template}, output_filepath]
-    user_configured_opts = cookie_file_options(addl_opts)
+    user_configured_opts = cookie_file_options(addl_opts) ++ sleep_interval_opts(addl_opts)
     # These must stay in exactly this order, hence why I'm giving it its own variable.
     all_opts = command_opts ++ print_to_file_opts ++ user_configured_opts ++ global_options()
     formatted_command_opts = [url] ++ CliUtils.parse_options(all_opts)
 
-    case CliUtils.wrap_cmd(command, formatted_command_opts, stderr_to_stdout: true) do
+    case CliUtils.wrap_cmd(backend_executable(), formatted_command_opts, stderr_to_stdout: true) do
       # yt-dlp exit codes:
       #   0 = Everything is successful
       #   100 = yt-dlp must restart for update to complete
@@ -96,6 +98,20 @@ defmodule Pinchflat.YtDlp.CommandRunner do
     end
   end
 
+  defp sleep_interval_opts(addl_opts) do
+    sleep_interval = Settings.get!(:extractor_sleep_interval_seconds)
+
+    if sleep_interval <= 0 || Keyword.get(addl_opts, :skip_sleep_interval) do
+      []
+    else
+      [
+        sleep_requests: NumberUtils.add_jitter(sleep_interval),
+        sleep_interval: NumberUtils.add_jitter(sleep_interval),
+        sleep_subtitles: NumberUtils.add_jitter(sleep_interval)
+      ]
+    end
+  end
+
   defp add_cookie_file do
     base_dir = Application.get_env(:pinchflat, :extras_directory)
     filename_options_map = %{cookies: "cookies.txt"}

lib/pinchflat_web/controllers/settings/setting_html/setting_form.html.heex

@@ -29,7 +29,7 @@
   <section class="mt-8">
     <section>
       <h3 class="text-2xl text-black dark:text-white">
-        Indexing Settings
+        Extractor Settings
       </h3>
 
       <.input
@@ -41,6 +41,14 @@
         html_help={true}
         inputclass="font-mono text-sm mr-4"
       />
+
+      <.input
+        field={f[:extractor_sleep_interval_seconds]}
+        placeholder="0"
+        type="number"
+        label="Sleep Interval (seconds)"
+        help="Sleep interval in seconds between each extractor request. Must be a positive whole number (or set to 0 to disable)"
+      />
     </section>
   </section>