pinchflat/lib/pinchflat/fast_indexing/youtube_rss.ex
Kieran f6708a327c
[Enhancement] Optionally use the YouTube API for improved fast indexing (#282)
* Started adding youtube API for fast indexing

* Hooked youtube API into fast indexing

* Added youtube_api_key to settings

* Added youtube api key to settings UI

* Added tests

* Refactored the youtube api module

* More refactor

* Changed editing mode name from basic to standard

* [WIP] started on copy changes

* Updated copy
2024-06-10 11:45:41 -07:00

64 lines
1.8 KiB
Elixir

defmodule Pinchflat.FastIndexing.YoutubeRss do
@moduledoc """
Methods for interacting with YouTube RSS feeds for fast indexing
"""
require Logger
alias Pinchflat.Sources.Source
alias Pinchflat.FastIndexing.YoutubeBehaviour
@behaviour YoutubeBehaviour
@doc """
Determines if the YouTube RSS feed is enabled for fast indexing. Used to satisfy
the `YoutubeBehaviour` behaviour.
Returns true
"""
@impl YoutubeBehaviour
def enabled?(), do: true
@doc """
Fetches the recent media IDs from a YouTube RSS feed for a given source.
Returns {:ok, [binary()]} | {:error, binary()}
"""
@impl YoutubeBehaviour
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
{: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))
|> Enum.uniq()
Logger.debug("Media ids fetched from RSS: #{inspect(media_ids)}")
{: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