From 582eb53698bc1e88d76646dc99f9969876f71364 Mon Sep 17 00:00:00 2001
From: Kieran <kieran.eglin@gmail.com>
Date: Thu, 6 Jun 2024 11:54:39 -0700
Subject: [PATCH] Add regex for validating source URL (#285)

---
 lib/pinchflat/sources/source.ex           |  8 ++++
 test/pinchflat/sources_test.exs           | 55 +++++++++++++++++++++++
 test/support/fixtures/sources_fixtures.ex |  2 +-
 3 files changed, 64 insertions(+), 1 deletion(-)

diff --git a/lib/pinchflat/sources/source.ex b/lib/pinchflat/sources/source.ex
index 02af5bd..e409725 100644
--- a/lib/pinchflat/sources/source.ex
+++ b/lib/pinchflat/sources/source.ex
@@ -115,6 +115,7 @@ defmodule Pinchflat.Sources.Source do
     |> validate_number(:retention_period_days, greater_than_or_equal_to: 0)
     # Ensures it ends with `.{{ ext }}` or `.%(ext)s` or similar (with a little wiggle room)
     |> validate_format(:output_path_template_override, MediaProfile.ext_regex(), message: "must end with .{{ ext }}")
+    |> validate_format(:original_url, youtube_channel_or_playlist_regex(), message: "must be a channel or playlist URL")
     |> cast_assoc(:metadata, with: &SourceMetadata.changeset/2, required: false)
     |> unique_constraint([:collection_id, :media_profile_id, :title_filter_regex], error_key: :original_url)
   end
@@ -141,6 +142,13 @@ defmodule Pinchflat.Sources.Source do
     ~w(__meta__ __struct__ metadata tasks media_items)a
   end
 
+  def youtube_channel_or_playlist_regex do
+    # Validate that the original URL is not a video URL
+    # Also matches if the string does NOT contain youtube.com or youtu.be. This preserves my tenuous support
+    # for non-youtube sources.
+    ~r<^(?:(?!youtube\.com/(watch|shorts|embed)|youtu\.be).)*$>
+  end
+
   defimpl Jason.Encoder, for: Source do
     def encode(value, opts) do
       value
diff --git a/test/pinchflat/sources_test.exs b/test/pinchflat/sources_test.exs
index a63c049..22e4fc7 100644
--- a/test/pinchflat/sources_test.exs
+++ b/test/pinchflat/sources_test.exs
@@ -659,6 +659,61 @@ defmodule Pinchflat.SourcesTest do
     end
   end
 
+  describe "change_source/3 when testing original_url validation" do
+    test "succeeds when an original URL is valid" do
+      source = source_fixture()
+
+      valid_urls = [
+        "https://www.youtube.com/channel/UCkRfArvrzheW2E7b6SVT7vQ",
+        "https://www.youtube.com/channel/UCkRfArvrzheW2E7b6SVT7vQ/videos",
+        "https://www.youtube.com/@youtubecreators/featured",
+        "https://www.youtube.com/@youtubecreators",
+        "https://www.youtube.com/c/YouTubeCreators",
+        "https://www.youtube.com/user/YouTubeCreators",
+        "https://www.youtube.com/YouTubeCreators",
+        "https://www.youtube.com/playlist?list=PLpjK416fmKwRtq-9-O_NbZlkW0k6zu2Wn",
+        "https://www.youtube.com/playlist?list=UUkRfArvrzheW2E7b6SVT7vQ"
+      ]
+
+      Enum.each(valid_urls, fn url ->
+        assert %{errors: []} = Sources.change_source(source, %{original_url: url})
+      end)
+    end
+
+    test "fails when an original URL points to a video" do
+      source = source_fixture()
+
+      invalid_urls = [
+        "https://www.youtube.com/watch?v=72maj9FLQZI",
+        "https://youtu.be/72maj9FLQZI",
+        "https://www.youtube.com/watch?v=1FwGFhMAmBo&list=PLpjK416fmKwRtq-9-O_NbZlkW0k6zu2Wn",
+        "https://www.youtube.com/shorts/Dq0eH-ZhQTU",
+        "https://www.youtube.com/embed/X64LHlfx4qg"
+      ]
+
+      Enum.each(invalid_urls, fn url ->
+        assert %{errors: [_]} = Sources.change_source(source, %{original_url: url})
+      end)
+    end
+
+    test "passes when a non-youtube link is provided" do
+      source = source_fixture()
+
+      valid_urls = [
+        "https://www.example.com",
+        "https://www.example.com/playlist",
+        "https://www.example.com/channel",
+        "https://www.example.com/user",
+        "https://www.example.com/watch?v=72maj9FLQZI",
+        "https://www.example.com/embed/X64LHlfx4qg"
+      ]
+
+      Enum.each(valid_urls, fn url ->
+        assert %{errors: []} = Sources.change_source(source, %{original_url: url})
+      end)
+    end
+  end
+
   defp playlist_mock(_url, _opts, _ot) do
     {
       :ok,
diff --git a/test/support/fixtures/sources_fixtures.ex b/test/support/fixtures/sources_fixtures.ex
index bb768db..a06329a 100644
--- a/test/support/fixtures/sources_fixtures.ex
+++ b/test/support/fixtures/sources_fixtures.ex
@@ -25,7 +25,7 @@ defmodule Pinchflat.SourcesFixtures do
             collection_type: "channel",
             custom_name: "Cool and good internal name!",
             description: "This is a description",
-            original_url: "https://www.youtube.com/channel/#{Faker.String.base64(12)}",
+            original_url: "https://www.youtube.com/@#{Faker.String.base64(12)}",
             media_profile_id: ProfilesFixtures.media_profile_fixture().id,
             index_frequency_minutes: 60
           }