RSS feed for sources (#110)

* Add media streaming (#108)

* [WIP] set up streaming route

* Added UUID to sources and media items

* Added media preview to MI show page

* Added plug to strip file extensions

* [VERY WIP] basic podcast RSS setup

* [WIP] got basic podcast RSS working

* [WIP] more expanding on RSS

* Comment

* [WIP] Working on refactoring feed

* Added UUID backfill to a migration

* [WIP] Moar refactoring

* [WIP] Adding UI for getting RSS feed

* Many tests

* Added conditional routing for feed URLs

* Removed the need for url_base to be set

* Updated preset name

* Rendered certain fields HTML-safe; Added logging to confirm range request support

* Fixed incorrect scheme issue

* Updated env var

* Updated other UI to use dropdown

* removed commented code

* Generate rss feeds (#123)

* Added plug to strip file extensions

* [VERY WIP] basic podcast RSS setup

* [WIP] got basic podcast RSS working

* [WIP] more expanding on RSS

* [WIP] Working on refactoring feed

* Added UUID backfill to a migration

* [WIP] Moar refactoring

* [WIP] Adding UI for getting RSS feed

* Many tests

* Added conditional routing for feed URLs

* Removed the need for url_base to be set

* Updated preset name

* Rendered certain fields HTML-safe; Added logging to confirm range request support

* Fixed incorrect scheme issue

* Updated env var

* Updated other UI to use dropdown

* removed commented code

* docs

* Added unique index to UUID fields

lib/pinchflat/media/media_item.ex

@@ -5,6 +5,7 @@ defmodule Pinchflat.Media.MediaItem do
 
   use Ecto.Schema
   import Ecto.Changeset
+  import Pinchflat.Utils.ChangesetUtils
 
   alias Pinchflat.Tasks.Task
   alias Pinchflat.Sources.Source
@@ -32,6 +33,7 @@ defmodule Pinchflat.Media.MediaItem do
   ]
   # Pretty much all the fields captured at index are required.
   @required_fields ~w(
+    uuid
     title
     original_url
     livestream
@@ -42,6 +44,11 @@ defmodule Pinchflat.Media.MediaItem do
     )a
 
   schema "media_items" do
+    # This is _not_ used as the primary key or internally in the database
+    # relations. This is only used to prevent an enumeration attack on the streaming
+    # and RSS feed endpoints since those _must_ be public (ie: no basic auth)
+    field :uuid, Ecto.UUID
+
     field :title, :string
     field :media_id, :string
     field :description, :string
@@ -78,6 +85,7 @@ defmodule Pinchflat.Media.MediaItem do
     media_item
     |> cast(attrs, @allowed_fields)
     |> cast_assoc(:metadata, with: &MediaMetadata.changeset/2, required: false)
+    |> dynamic_default(:uuid, fn _ -> Ecto.UUID.generate() end)
     |> validate_required(@required_fields)
     |> unique_constraint([:media_id, :source_id])
   end

lib/pinchflat/podcasts/podcast_helpers.ex

@@ -0,0 +1,75 @@
+defmodule Pinchflat.Podcasts.PodcastHelpers do
+  @moduledoc """
+  Methods for fetching postcast-related data from a source
+  or its media items
+  """
+
+  alias Pinchflat.Repo
+  alias Pinchflat.Media
+  alias Pinchflat.Metadata.MediaMetadata
+  alias Pinchflat.Metadata.SourceMetadata
+
+  @doc """
+  Returns a list of media items that have been downloaded to disk
+  and have been proven to still exist there.
+
+  Useful for podcasts since we don't want to serve media that
+  has been deleted or moved, but it's also fairly generally useful
+  so I could see this being moved in the future.
+
+  Options:
+    - limit: integer - the maximum number of media items to return
+
+  Returns: [%MediaItem{}]
+  """
+  def persisted_media_items_for(source, opts \\ []) do
+    limit = Keyword.get(opts, :limit, 500)
+
+    source
+    |> Media.list_downloaded_media_items_for(limit: limit)
+    |> Enum.filter(fn media_item -> File.exists?(media_item.media_filepath) end)
+  end
+
+  @doc """
+  Selects a cover image for a source based on the source's metadata
+  and the metadata of the media items associated with the source. Also
+  ensures images exist on disk.
+
+  Only one media item should need to be returned since this is using the
+  internal metadata which, so long as the media_item was _downloaded_, should
+  be guaranteed to exist.
+
+  Prefers the source's poster, then fanart, then the media item's thumbnail.
+
+  Returns: {:ok, filepath} | {:error, :no_suitable_image}
+  """
+  def select_cover_image(source, media_items) do
+    source_with_preloads = Repo.preload(source, :metadata)
+
+    source_with_preloads
+    |> get_images_by_preference(media_items)
+    |> Enum.reject(&is_nil(&1))
+    |> Enum.find(&File.exists?/1)
+    |> case do
+      nil -> {:error, :no_suitable_image}
+      filepath -> {:ok, filepath}
+    end
+  end
+
+  defp get_images_by_preference(source_with_preloads, []) do
+    source_metadata = source_with_preloads.metadata || %SourceMetadata{}
+
+    [
+      source_metadata.poster_filepath,
+      source_metadata.fanart_filepath
+    ]
+  end
+
+  defp get_images_by_preference(source_with_preloads, [media_item | _]) do
+    media_item_with_preloads = Repo.preload(media_item, :metadata)
+    media_item_metadata = media_item_with_preloads.metadata || %MediaMetadata{}
+    source_images = get_images_by_preference(source_with_preloads, [])
+
+    source_images ++ [media_item_metadata.thumbnail_filepath]
+  end
+end

lib/pinchflat/podcasts/rss_feed_builder.ex

@@ -0,0 +1,138 @@
+defmodule Pinchflat.Podcasts.RssFeedBuilder do
+  @moduledoc """
+  Methods for building an RSS feed for a source and its media items.
+  """
+
+  @datetime_format "%a, %d %b %Y %H:%M:%S %z"
+
+  alias Pinchflat.Utils.DatetimeUtils
+  alias Pinchflat.Podcasts.PodcastHelpers
+  alias PinchflatWeb.Router.Helpers, as: Routes
+
+  @doc """
+  Builds an RSS feed for a given source and its media items.
+  Only MediaItems that have been persisted will be included in the feed.
+
+  ## Options:
+    - `:limit` - The maximum number of media items to include in the feed. Defaults to 300.
+
+  Returns an XML document as a string.
+  """
+  def build(source, opts \\ []) do
+    limit = Keyword.get(opts, :limit, 300)
+    url_base = Keyword.get(opts, :url_base, PinchflatWeb.Endpoint.url())
+
+    media_items = PodcastHelpers.persisted_media_items_for(source, limit: limit)
+    build_source_xml(source, media_items, url_base)
+  end
+
+  defp build_source_xml(source, media_items, url_base) do
+    media_item_xml = Enum.map(media_items, &build_media_item_xml(source, &1, url_base))
+    # "caching" the image path since it requires some DB calls and is used twice
+    feed_image_path = feed_image_path(url_base, source, media_items)
+
+    # Useful: resources:
+    #   - https://validator.w3.org/feed/#validate_by_input
+    #   - https://github.com/Podcastindex-org/podcast-namespace/blob/main/docs/1.0.md
+    #   - https://podba.se/validate
+    """
+    <?xml version="1.0" encoding="UTF-8"?>
+    <rss version="2.0"
+      xmlns:itunes="http://www.itunes.com/dtds/podcast-1.0.dtd"
+      xmlns:podcast="https://podcastindex.org/namespace/1.0"
+      xmlns:atom="http://www.w3.org/2005/Atom">
+      <channel>
+        <title>#{safe(source.custom_name)}</title>
+        <link>#{source.original_url}</link>
+        <description>#{safe(source.description)}</description>
+        <category>TV &amp; Film</category>
+        <generator>Generated by Pinchflat</generator>
+        <language>en-us</language>
+        <lastBuildDate>#{Calendar.strftime(source.updated_at, @datetime_format)}</lastBuildDate>
+        <pubDate>#{Calendar.strftime(source.inserted_at, @datetime_format)}</pubDate>
+        <atom:link href="#{generate_self_link(url_base, source)}" rel="self" type="application/rss+xml" />
+        <podcast:locked>yes</podcast:locked>
+        <podcast:guid>#{source.uuid}</podcast:guid>
+        <image>
+          <url>#{feed_image_path}</url>
+          <title>#{safe(source.custom_name)}</title>
+          <link>#{source.original_url}</link>
+        </image>
+        <itunes:author>#{safe(source.custom_name)}</itunes:author>
+        <itunes:subtitle>#{safe(source.custom_name)}</itunes:subtitle>
+        <itunes:block>yes</itunes:block>
+        <itunes:image href="#{feed_image_path}"></itunes:image>
+        <itunes:explicit>false</itunes:explicit>
+        <itunes:category text="TV &amp; Film"></itunes:category>
+
+        #{Enum.join(media_item_xml, "\n")}
+
+      </channel>
+    </rss>
+    """
+  end
+
+  defp build_media_item_xml(source, media_item, url_base) do
+    """
+    <item>
+      <guid isPermaLink="false">#{media_item.uuid}</guid>
+      <title>#{safe(media_item.title)}</title>
+      <link>#{media_item.original_url}</link>
+      <description>#{safe(media_item.description)}</description>
+      <pubDate>#{generate_upload_date(media_item)}</pubDate>
+      <enclosure
+        url="#{media_stream_path(url_base, media_item)}"
+        length="#{media_item.media_size_bytes}"
+        type="#{MIME.from_path(media_item.media_filepath)}"
+      />
+      <itunes:author>#{safe(source.custom_name)}</itunes:author>
+      <itunes:subtitle>#{safe(media_item.title)}</itunes:subtitle>
+      <itunes:summary><![CDATA[#{media_item.description}]]></itunes:summary>
+      <itunes:explicit>false</itunes:explicit>
+    </item>
+    """
+  end
+
+  defp safe(nil), do: ""
+
+  defp safe(value) do
+    value
+    |> Phoenix.HTML.html_escape()
+    |> Phoenix.HTML.safe_to_string()
+  end
+
+  defp generate_self_link(url_base, source) do
+    Path.join(url_base, "#{podcast_route(:rss_feed, source.uuid)}.xml")
+  end
+
+  defp media_stream_path(url_base, media_item) do
+    extension = Path.extname(media_item.media_filepath)
+
+    Path.join(url_base, "#{media_route(:stream, media_item.uuid)}#{extension}")
+  end
+
+  defp feed_image_path(url_base, source, media_items) do
+    case PodcastHelpers.select_cover_image(source, media_items) do
+      {:error, _} ->
+        ""
+
+      {:ok, filepath} ->
+        extension = Path.extname(filepath)
+        Path.join(url_base, "#{podcast_route(:feed_image, source.uuid)}#{extension}")
+    end
+  end
+
+  defp generate_upload_date(media_item) do
+    media_item.upload_date
+    |> DatetimeUtils.date_to_datetime()
+    |> Calendar.strftime(@datetime_format)
+  end
+
+  defp podcast_route(action, params) do
+    Routes.podcast_path(PinchflatWeb.Endpoint, action, params)
+  end
+
+  defp media_route(action, params) do
+    Routes.media_item_path(PinchflatWeb.Endpoint, action, params)
+  end
+end

lib/pinchflat/sources/source.ex

@@ -48,6 +48,7 @@ defmodule Pinchflat.Sources.Source do
 
   @pre_insert_required_fields @initially_required_fields ++
                                 ~w(
+                                  uuid
                                   custom_name
                                   collection_name
                                   collection_id
@@ -55,6 +56,11 @@ defmodule Pinchflat.Sources.Source do
                                 )a
 
   schema "sources" do
+    # This is _not_ used as the primary key or internally in the database
+    # relations. This is only used to prevent an enumeration attack on the streaming
+    # and RSS feed endpoints since those _must_ be public (ie: no basic auth)
+    field :uuid, Ecto.UUID
+
     field :custom_name, :string
     field :description, :string
     field :collection_name, :string
@@ -98,6 +104,7 @@ defmodule Pinchflat.Sources.Source do
     source
     |> cast(attrs, @allowed_fields)
     |> dynamic_default(:custom_name, fn cs -> get_field(cs, :collection_name) end)
+    |> dynamic_default(:uuid, fn _ -> Ecto.UUID.generate() end)
     |> validate_required(required_fields)
     |> cast_assoc(:metadata, with: &SourceMetadata.changeset/2, required: false)
   end

lib/pinchflat/utils/datetime_utils.ex

@@ -0,0 +1,17 @@
+defmodule Pinchflat.Utils.DatetimeUtils do
+  @moduledoc """
+  Utility methods for working with dates and datetimes
+  """
+
+  @doc """
+  Converts a Date to a DateTime
+
+  Returns %DateTime{}
+  """
+  def date_to_datetime(date) do
+    date
+    |> Date.to_gregorian_days()
+    |> Kernel.*(86_400)
+    |> DateTime.from_gregorian_seconds()
+  end
+end