Started adding youtube API for fast indexing

docker-compose.yml

@@ -12,3 +12,5 @@ services:
       - ./docker-run.dev.sh
     stdin_open: true
     tty: true
+    env_file:
+      - .env

lib/pinchflat/fast_indexing/fast_indexing_helpers.ex

@@ -26,7 +26,7 @@ defmodule Pinchflat.FastIndexing.FastIndexingHelpers do
   downloaded_.
   """
   def kickoff_download_tasks_from_youtube_rss_feed(%Source{} = source) do
-    {:ok, media_ids} = YoutubeRss.get_recent_media_ids_from_rss(source)
+    {:ok, media_ids} = YoutubeRss.get_recent_media_ids(source)
     existing_media_items = list_media_items_by_media_id_for(source, media_ids)
     new_media_ids = media_ids -- Enum.map(existing_media_items, & &1.media_id)
 

lib/pinchflat/fast_indexing/youtube_api.ex

@@ -0,0 +1,66 @@
+defmodule Pinchflat.FastIndexing.YoutubeApi do
+  @moduledoc """
+  TODO
+  """
+
+  require Logger
+
+  alias Pinchflat.Sources.Source
+  alias Pinchflat.Utils.FunctionUtils
+
+  # TODO: test
+  # TODO: make this a behaviour
+  def get_recent_media_ids(%Source{} = source) do
+    api_response =
+      source
+      |> determine_playlist_id()
+      |> do_api_request()
+
+    case api_response do
+      {:ok, parsed_json} -> get_media_ids_from_response(parsed_json)
+      {:error, reason} -> {:error, reason}
+    end
+  end
+
+  # The UC prefix is for channels which won't work with this API endpoint. Swapping
+  # the prefix to UU will get us the playlist that represents the channel's uploads
+  defp determine_playlist_id(%{collection_id: c_id}) do
+    String.replace_prefix(c_id, "UC", "UU")
+  end
+
+  defp do_api_request(playlist_id) do
+    Logger.debug("Fetching recent media IDs from YouTube API for playlist: #{playlist_id}")
+
+    api_base = "https://youtube.googleapis.com/youtube/v3/playlistItems"
+    request_url = "#{api_base}?part=contentDetails&maxResults=50&playlistId=#{playlist_id}&key=#{api_key()}"
+
+    case http_client().get(request_url, accept: "application/json") do
+      {:ok, response} ->
+        Phoenix.json_library().decode(response)
+
+      {:error, reason} ->
+        {:error, reason}
+    end
+  end
+
+  defp get_media_ids_from_response(parsed_json) do
+    parsed_json
+    |> Map.get("items", [])
+    |> Enum.map(fn item ->
+      item
+      |> Map.get("contentDetails", %{})
+      |> Map.get("videoId", nil)
+    end)
+    |> Enum.reject(&is_nil/1)
+    |> FunctionUtils.wrap_ok()
+  end
+
+  # TODO: replace this with a user setting
+  defp api_key do
+    System.get_env("YOUTUBE_API_KEY")
+  end
+
+  defp http_client do
+    Application.get_env(:pinchflat, :http_client, Pinchflat.HTTP.HTTPClient)
+  end
+end

lib/pinchflat/fast_indexing/youtube_behaviour.ex

@@ -0,0 +1,3 @@
+defmodule Pinchflat.FastIndexing.YoutubeBehaviour do
+  # TODO
+end

lib/pinchflat/fast_indexing/youtube_rss.ex

@@ -12,7 +12,7 @@ defmodule Pinchflat.FastIndexing.YoutubeRss do
 
   Returns {:ok, [binary()]} | {:error, binary()}
   """
-  def get_recent_media_ids_from_rss(%Source{} = source) do
+  def get_recent_media_ids(%Source{} = source) do
     Logger.debug("Fetching recent media IDs from YouTube RSS feed for source: #{source.collection_id}")
 
     case http_client().get(rss_url_for_source(source)) do

lib/pinchflat/http/http_client.ex

@@ -21,9 +21,11 @@ defmodule Pinchflat.HTTP.HTTPClient do
   """
   @impl HTTPBehaviour
   def get(url, headers \\ [], opts \\ []) do
+    headers = parse_headers(headers)
+
     case :httpc.request(:get, {url, headers}, [], opts) do
       {:ok, {{_version, 200, _reason_phrase}, _headers, body}} ->
-        {:ok, body}
+        {:ok, to_string(body)}
 
       {:ok, {{_version, status_code, reason_phrase}, _headers, _body}} ->
         {:error, "HTTP request failed with status code #{status_code}: #{reason_phrase}"}
@@ -32,4 +34,8 @@ defmodule Pinchflat.HTTP.HTTPClient do
         {:error, "HTTP request failed: #{reason}"}
     end
   end
+
+  defp parse_headers(headers) do
+    Enum.map(headers, fn {k, v} -> {to_charlist(k), to_charlist(v)} end)
+  end
 end

test/pinchflat/fast_indexing/youtube_rss_test.exs

@@ -11,7 +11,7 @@ defmodule Pinchflat.FastIndexing.YoutubeRssTest do
     {:ok, source: source}
   end
 
-  describe "get_recent_media_ids_from_rss/1" do
+  describe "get_recent_media_ids/1" do
     test "calls the expected URL for channel sources" do
       source = source_fixture(collection_type: :channel, collection_id: "channel_id")
 
@@ -21,7 +21,7 @@ defmodule Pinchflat.FastIndexing.YoutubeRssTest do
         {:ok, ""}
       end)
 
-      assert {:ok, _} = YoutubeRss.get_recent_media_ids_from_rss(source)
+      assert {:ok, _} = YoutubeRss.get_recent_media_ids(source)
     end
 
     test "calls the expected URL for playlist sources" do
@@ -33,13 +33,13 @@ defmodule Pinchflat.FastIndexing.YoutubeRssTest do
         {:ok, ""}
       end)
 
-      assert {:ok, _} = YoutubeRss.get_recent_media_ids_from_rss(source)
+      assert {:ok, _} = YoutubeRss.get_recent_media_ids(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)
+      assert {:error, "Failed to fetch RSS feed"} = YoutubeRss.get_recent_media_ids(source)
     end
 
     test "returns the media IDs from the RSS feed", %{source: source} do
@@ -47,7 +47,7 @@ defmodule Pinchflat.FastIndexing.YoutubeRssTest do
         {: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)
+      assert {:ok, ["test_1", "test_2"]} = YoutubeRss.get_recent_media_ids(source)
     end
 
     test "strips whitespace from media IDs", %{source: source} do
@@ -55,7 +55,7 @@ defmodule Pinchflat.FastIndexing.YoutubeRssTest do
         {: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)
+      assert {:ok, ["test_1", "test_2"]} = YoutubeRss.get_recent_media_ids(source)
     end
 
     test "removes empty media IDs", %{source: source} do
@@ -63,7 +63,7 @@ defmodule Pinchflat.FastIndexing.YoutubeRssTest do
         {:ok, "<yt:videoId>test_1</yt:videoId><yt:videoId></yt:videoId>"}
       end)
 
-      assert {:ok, ["test_1"]} = YoutubeRss.get_recent_media_ids_from_rss(source)
+      assert {:ok, ["test_1"]} = YoutubeRss.get_recent_media_ids(source)
     end
 
     test "removes duplicate media IDs", %{source: source} do
@@ -71,7 +71,7 @@ defmodule Pinchflat.FastIndexing.YoutubeRssTest do
         {:ok, "<yt:videoId>test_1</yt:videoId><yt:videoId>test_1</yt:videoId>"}
       end)
 
-      assert {:ok, ["test_1"]} = YoutubeRss.get_recent_media_ids_from_rss(source)
+      assert {:ok, ["test_1"]} = YoutubeRss.get_recent_media_ids(source)
     end
   end
 end