finder: Find feed from YouTube playlist

The feed from a YouTube playlist page is derived in practically the same way as a feed from a YouTube channel page.

2 files changed, 53 insertions, 5 deletions

diff --git a/internal/reader/subscription/finder.go b/internal/reader/subscription/finder.go
index a086136a..1b359a82 100644
--- a/internal/reader/subscription/finder.go
+++ b/internal/reader/subscription/finder.go
@@ -23,8 +23,9 @@ import (
 )
 
 var (
-	youtubeChannelRegex = regexp.MustCompile(`youtube\.com/channel/(.*)$`)
-	youtubeVideoRegex   = regexp.MustCompile(`youtube\.com/watch\?v=(.*)$`)
+	youtubeChannelRegex  = regexp.MustCompile(`youtube\.com/channel/(.*)$`)
+	youtubeVideoRegex    = regexp.MustCompile(`youtube\.com/watch\?v=(.*)$`)
+	youtubePlaylistRegex = regexp.MustCompile(`youtube\.com/playlist\?list=(.*)$`)
 )
 
 type SubscriptionFinder struct {
@@ -98,7 +99,19 @@ func (f *SubscriptionFinder) FindSubscriptions(websiteURL, rssBridgeURL string)
 		return subscriptions, nil
 	}
 
-	// Step 4) Parse web page to find feeds from HTML meta tags.
+	// Step 4) Check if the website URL is a YouTube playlist.
+	slog.Debug("Try to detect feeds from YouTube playlist page", slog.String("website_url", websiteURL))
+	subscriptions, localizedError = f.FindSubscriptionsFromYouTubePlaylistPage(websiteURL)
+	if localizedError != nil {
+		return nil, localizedError
+	}
+
+	if len(subscriptions) > 0 {
+		slog.Debug("Subscriptions found from YouTube playlist page", slog.String("website_url", websiteURL), slog.Any("subscriptions", subscriptions))
+		return subscriptions, nil
+	}
+
+	// Step 5) Parse web page to find feeds from HTML meta tags.
 	slog.Debug("Try to detect feeds from HTML meta tags",
 		slog.String("website_url", websiteURL),
 		slog.String("content_type", responseHandler.ContentType()),
@@ -113,7 +126,7 @@ func (f *SubscriptionFinder) FindSubscriptions(websiteURL, rssBridgeURL string)
 		return subscriptions, nil
 	}
 
-	// Step 5) Check if the website URL can use RSS-Bridge.
+	// Step 6) Check if the website URL can use RSS-Bridge.
 	if rssBridgeURL != "" {
 		slog.Debug("Try to detect feeds with RSS-Bridge", slog.String("website_url", websiteURL))
 		subscriptions, localizedError := f.FindSubscriptionsFromRSSBridge(websiteURL, rssBridgeURL)
@@ -127,7 +140,7 @@ func (f *SubscriptionFinder) FindSubscriptions(websiteURL, rssBridgeURL string)
 		}
 	}
 
-	// Step 6) Check if the website has a known feed URL.
+	// Step 7) Check if the website has a known feed URL.
 	slog.Debug("Try to detect feeds from well-known URLs", slog.String("website_url", websiteURL))
 	subscriptions, localizedError = f.FindSubscriptionsFromWellKnownURLs(websiteURL)
 	if localizedError != nil {
@@ -322,3 +335,16 @@ func (f *SubscriptionFinder) FindSubscriptionsFromYouTubeVideoPage(websiteURL st
 
 	return nil, nil
 }
+
+func (f *SubscriptionFinder) FindSubscriptionsFromYouTubePlaylistPage(websiteURL string) (Subscriptions, *locale.LocalizedErrorWrapper) {
+	matches := youtubePlaylistRegex.FindStringSubmatch(websiteURL)
+
+	if len(matches) == 2 {
+		feedURL := fmt.Sprintf(`https://www.youtube.com/feeds/videos.xml?playlist_id=%s`, matches[1])
+		return Subscriptions{NewSubscription(websiteURL, feedURL, parser.FormatAtom)}, nil
+	}
+
+	slog.Debug("This website is not a YouTube playlist page, the regex doesn't match", slog.String("website_url", websiteURL))
+
+	return nil, nil
+}
diff --git a/internal/reader/subscription/finder_test.go b/internal/reader/subscription/finder_test.go
index 160b52cb..216d81cb 100644
--- a/internal/reader/subscription/finder_test.go
+++ b/internal/reader/subscription/finder_test.go
@@ -8,6 +8,28 @@ import (
 	"testing"
 )
 
+func TestFindYoutubePlaylistFeed(t *testing.T) {
+	scenarios := map[string]string{
+		"https://www.youtube.com/playlist?list=PLOOwEPgFWm_NHcQd9aCi5JXWASHO_n5uR": "https://www.youtube.com/feeds/videos.xml?playlist_id=PLOOwEPgFWm_NHcQd9aCi5JXWASHO_n5uR",
+		"https://www.youtube.com/playlist?list=PLOOwEPgFWm_N42HlCLhqyJ0ZBWr5K1QDM": "https://www.youtube.com/feeds/videos.xml?playlist_id=PLOOwEPgFWm_N42HlCLhqyJ0ZBWr5K1QDM",
+	}
+
+	for websiteURL, expectedFeedURL := range scenarios {
+		subscriptions, localizedError := NewSubscriptionFinder(nil).FindSubscriptionsFromYouTubePlaylistPage(websiteURL)
+		if localizedError != nil {
+			t.Fatalf(`Parsing a correctly formatted YouTube playlist page should not return any error: %v`, localizedError)
+		}
+
+		if len(subscriptions) != 1 {
+			t.Fatal(`Incorrect number of subscriptions returned`)
+		}
+
+		if subscriptions[0].URL != expectedFeedURL {
+			t.Errorf(`Unexpected Feed, got %s, instead of %s`, subscriptions[0].URL, expectedFeedURL)
+		}
+	}
+}
+
 func TestFindYoutubeChannelFeed(t *testing.T) {
 	scenarios := map[string]string{
 		"https://www.youtube.com/channel/UC-Qj80avWItNRjkZ41rzHyw": "https://www.youtube.com/feeds/videos.xml?channel_id=UC-Qj80avWItNRjkZ41rzHyw",