[Bugfix] Ensure livestreams aren't downloaded until they're finished processing (#485)

* Added logic to ignore downloads that aren't in the right live state

* Added tests for get_downloadable_status method

* Added tests for media downloader module

* Added tests to download worker modeule

lib/pinchflat/downloading/media_download_worker.ex

@@ -94,6 +94,9 @@ defmodule Pinchflat.Downloading.MediaDownloadWorker do
       {:recovered, _} ->
         {:error, :retry}
 
+      {:error, :unsuitable_for_download} ->
+        {:ok, :non_retry}
+
       {:error, message} ->
         action_on_error(message)
     end

lib/pinchflat/downloading/media_downloader.ex

@@ -37,6 +37,13 @@ defmodule Pinchflat.Downloading.MediaDownloader do
       {:ok, parsed_json} ->
         update_media_item_from_parsed_json(media_with_preloads, parsed_json)
 
+      {:error, :unsuitable_for_download} ->
+        Logger.warning(
+          "Media item ##{media_with_preloads.id} isn't suitable for download yet. May be an active or processing live stream"
+        )
+
+        {:error, :unsuitable_for_download}
+
       {:error, message, _exit_code} ->
         Logger.error("yt-dlp download error for media item ##{media_with_preloads.id}: #{inspect(message)}")
 
@@ -108,7 +115,11 @@ defmodule Pinchflat.Downloading.MediaDownloader do
     {:ok, options} = DownloadOptionBuilder.build(item_with_preloads, override_opts)
     runner_opts = [output_filepath: output_filepath, use_cookies: item_with_preloads.source.use_cookies]
 
-    YtDlpMedia.download(url, options, runner_opts)
+    case YtDlpMedia.get_downloadable_status(url) do
+      {:ok, :downloadable} -> YtDlpMedia.download(url, options, runner_opts)
+      {:ok, :ignorable} -> {:error, :unsuitable_for_download}
+      err -> err
+    end
   end
 
   defp recoverable_errors do

lib/pinchflat/yt_dlp/media.ex

@@ -49,6 +49,24 @@ defmodule Pinchflat.YtDlp.Media do
     end
   end
 
+  @doc """
+  Determines if the media at the given URL is ready to be downloaded.
+  Common examples of non-downloadable media are upcoming or in-progress live streams.
+
+  Returns {:ok, :downloadable | :ignorable} | {:error, any}
+  """
+  def get_downloadable_status(url) do
+    case backend_runner().run(url, [:simulate, :skip_download], "%(.{live_status})j") do
+      {:ok, output} ->
+        output
+        |> Phoenix.json_library().decode!()
+        |> parse_downloadable_status()
+
+      err ->
+        err
+    end
+  end
+
   @doc """
   Downloads a thumbnail for a single piece of media. Usually used for
   downloading thumbnails for internal use
@@ -71,11 +89,10 @@ defmodule Pinchflat.YtDlp.Media do
   Returns {:ok, %Media{}} | {:error, any, ...}.
   """
   def get_media_attributes(url, command_opts \\ [], addl_opts \\ []) do
-    runner = Application.get_env(:pinchflat, :yt_dlp_runner)
     all_command_opts = [:simulate, :skip_download] ++ command_opts
     output_template = indexing_output_template()
 
-    case runner.run(url, all_command_opts, output_template, addl_opts) do
+    case backend_runner().run(url, all_command_opts, output_template, addl_opts) do
       {:ok, output} ->
         output
         |> Phoenix.json_library().decode!()
@@ -147,6 +164,16 @@ defmodule Pinchflat.YtDlp.Media do
   defp parse_uploaded_at(%{"upload_date" => nil}), do: nil
   defp parse_uploaded_at(response), do: MetadataFileHelpers.parse_upload_date(response["upload_date"])
 
+  defp parse_downloadable_status(response) do
+    case response["live_status"] do
+      status when status in ["is_live", "is_upcoming", "post_live"] -> {:ok, :ignorable}
+      status when status in ["was_live", "not_live"] -> {:ok, :downloadable}
+      # This preserves my tenuous support for non-youtube sources.
+      nil -> {:ok, :downloadable}
+      _ -> {:error, "Unknown live status: #{response["live_status"]}"}
+    end
+  end
+
   defp backend_runner do
     # This approach lets us mock the command for testing
     Application.get_env(:pinchflat, :yt_dlp_runner)