Added module for working with youtube RSS feed

.iex.exs

@@ -1,10 +1,10 @@
 alias Pinchflat.Repo
 
 alias Pinchflat.Tasks.Task
+alias Pinchflat.Sources.Source
 alias Pinchflat.Media.MediaItem
 alias Pinchflat.Tasks.SourceTasks
 alias Pinchflat.Media.MediaMetadata
-alias Pinchflat.Sources.Source
 alias Pinchflat.Profiles.MediaProfile
 
 alias Pinchflat.Tasks
@@ -13,7 +13,11 @@ alias Pinchflat.Profiles
 alias Pinchflat.Sources
 alias Pinchflat.Settings
 
-alias Pinchflat.MediaClient.{SourceDetails, MediaDownloader}
+alias Pinchflat.MediaClient.MediaDownloader
+alias Pinchflat.YtDlp.Backend.Media, as: YtDlpMedia
+alias Pinchflat.YtDlp.Backend.MediaCollection, as: YtDlpCollection
+
+alias Pinchflat.Api.YoutubeRss
 alias Pinchflat.Metadata.MetadataFileHelpers
 
 alias Pinchflat.Utils.FilesystemUtils.FileFollowerServer
@@ -38,7 +42,7 @@ defmodule IexHelpers do
         :channel -> channel_url()
       end
 
-    SourceDetails.get_source_details(source)
+    YtDlpCollection.get_source_details(source)
   end
 
   def ids(type) do
@@ -48,7 +52,7 @@ defmodule IexHelpers do
         :channel -> channel_url()
       end
 
-    SourceDetails.get_media_attributes_for_collection(source)
+    YtDlpCollection.get_media_attributes_for_collection(source)
   end
 end
 

lib/pinchflat/api/youtube_rss.ex

@@ -0,0 +1,44 @@
+defmodule Pinchflat.Api.YoutubeRss do
+  @moduledoc """
+  Methods for interacting with YouTube RSS feeds
+  """
+
+  alias Pinchflat.Sources.Source
+
+  @doc """
+  Fetches the recent media IDs from a YouTube RSS feed for a given source.
+
+  Returns {:ok, [binary()]} | {:error, binary()}
+  """
+  def get_recent_media_ids_from_rss(%Source{} = source) do
+    case http_client().get(rss_url_for_source(source)) do
+      {:ok, response} ->
+        response = to_string(response)
+        media_id_regex = ~r/<yt:videoId>(.*?)<\/yt:videoId>/
+
+        # Don't get on me about using regex to search XML.
+        # The content is known, well-formed, and simple.
+        media_ids =
+          media_id_regex
+          |> Regex.scan(response)
+          |> Enum.map(fn [_, id] -> String.trim(id) end)
+          |> Enum.filter(&(String.length(&1) > 0))
+
+        {:ok, media_ids}
+
+      {:error, _reason} ->
+        {:error, "Failed to fetch RSS feed"}
+    end
+  end
+
+  defp rss_url_for_source(source) do
+    case source.collection_type do
+      :channel -> "https://www.youtube.com/feeds/videos.xml?channel_id=#{source.collection_id}"
+      :playlist -> "https://www.youtube.com/feeds/videos.xml?playlist_id=#{source.collection_id}"
+    end
+  end
+
+  defp http_client do
+    Application.get_env(:pinchflat, :http_client, Pinchflat.HTTP.HTTPClient)
+  end
+end

lib/pinchflat/http/http_behaviour.ex

@@ -4,5 +4,7 @@ defmodule Pinchflat.HTTP.HTTPBehaviour do
   so I can use Mox to create an HTTP mock
   """
 
+  @callback get(String.t()) :: {:ok, String.t()} | {:error, String.t()}
+  @callback get(String.t(), Keyword.t()) :: {:ok, String.t()} | {:error, String.t()}
   @callback get(String.t(), Keyword.t(), Keyword.t()) :: {:ok, String.t()} | {:error, String.t()}
 end

test/pinchflat/api/youtube_rss_test.exs

@@ -0,0 +1,71 @@
+defmodule Pinchflat.Api.YoutubeRssTest do
+  use Pinchflat.DataCase
+  import Mox
+  import Pinchflat.SourcesFixtures
+
+  alias Pinchflat.Api.YoutubeRss
+
+  setup :verify_on_exit!
+
+  setup do
+    source = source_fixture()
+
+    {:ok, source: source}
+  end
+
+  describe "get_recent_media_ids_from_rss/1" do
+    test "calls the expected URL for channel sources" do
+      source = source_fixture(collection_type: :channel, collection_id: "channel_id")
+
+      expect(HTTPClientMock, :get, fn url ->
+        assert url =~ "https://www.youtube.com/feeds/videos.xml?channel_id=#{source.collection_id}"
+
+        {:ok, ""}
+      end)
+
+      assert {:ok, _} = YoutubeRss.get_recent_media_ids_from_rss(source)
+    end
+
+    test "calls the expected URL for playlist sources" do
+      source = source_fixture(collection_type: :playlist, collection_id: "playlist_id")
+
+      expect(HTTPClientMock, :get, fn url ->
+        assert url =~ "https://www.youtube.com/feeds/videos.xml?playlist_id=#{source.collection_id}"
+
+        {:ok, ""}
+      end)
+
+      assert {:ok, _} = YoutubeRss.get_recent_media_ids_from_rss(source)
+    end
+
+    test "returns an error if the HTTP request fails", %{source: source} do
+      expect(HTTPClientMock, :get, fn _url -> {:error, ""} end)
+
+      assert {:error, "Failed to fetch RSS feed"} = YoutubeRss.get_recent_media_ids_from_rss(source)
+    end
+
+    test "returns the media IDs from the RSS feed", %{source: source} do
+      expect(HTTPClientMock, :get, fn _url ->
+        {:ok, "<yt:videoId>test_1</yt:videoId><yt:videoId>test_2</yt:videoId>"}
+      end)
+
+      assert {:ok, ["test_1", "test_2"]} = YoutubeRss.get_recent_media_ids_from_rss(source)
+    end
+
+    test "strips whitespace from media IDs", %{source: source} do
+      expect(HTTPClientMock, :get, fn _url ->
+        {:ok, "<yt:videoId> test_1 </yt:videoId><yt:videoId> test_2 </yt:videoId>"}
+      end)
+
+      assert {:ok, ["test_1", "test_2"]} = YoutubeRss.get_recent_media_ids_from_rss(source)
+    end
+
+    test "removes empty media IDs", %{source: source} do
+      expect(HTTPClientMock, :get, fn _url ->
+        {:ok, "<yt:videoId>test_1</yt:videoId><yt:videoId></yt:videoId>"}
+      end)
+
+      assert {:ok, ["test_1"]} = YoutubeRss.get_recent_media_ids_from_rss(source)
+    end
+  end
+end