From b57860a9670ffc6d7fe182256a24043248c0b93a Mon Sep 17 00:00:00 2001
From: Alex McGrath <amk@amk.ie>
Date: Fri, 5 Feb 2021 18:24:42 +0000
Subject: [PATCH] make podcast audio finding more robust

this will check that if there is no type field in an enclosure that
the url ends with an audio filename, if none of these are true, itll
then check audio:content for an item and do those checks again.
---
 server/podcasts/podcasts.go | 75 +++++++++++++++++++++++++++++--------
 1 file changed, 60 insertions(+), 15 deletions(-)

diff --git a/server/podcasts/podcasts.go b/server/podcasts/podcasts.go
index 37c82cf..8d90909 100644
--- a/server/podcasts/podcasts.go
+++ b/server/podcasts/podcasts.go
@@ -20,6 +20,7 @@ import (
 
 	"go.senan.xyz/gonic/multierr"
 	"go.senan.xyz/gonic/server/db"
+	gmime "go.senan.xyz/gonic/server/mime"
 	"go.senan.xyz/gonic/server/scanner/tags"
 )
 
@@ -170,26 +171,70 @@ func (p *Podcasts) AddEpisode(podcastID int, item *gofeed.Item) error {
 		duration = getSecondsFromString(item.ITunesExt.Duration)
 	}
 
+	if episode, ok := p.findEnclosureAudio(podcastID, duration, item); ok {
+		if err := p.DB.Save(episode).Error; err != nil {
+			return err
+		}
+		return nil
+	}
+	if episode, ok := p.findMediaAudio(podcastID, duration, item); ok {
+		if err := p.DB.Save(episode).Error; err != nil {
+			return err
+		}
+		return nil
+	}
+	// hopefully shouldnt reach here
+	log.Println("failed to find audio in feed item, skipping")
+	return nil
+}
+
+func isAudio(mediaType, url string) bool {
+	if mediaType != "" && strings.HasPrefix(mediaType, "audio") {
+		return true
+	}
+	_, ok := gmime.FromExtension(filepath.Ext(url)[1:])
+	return ok
+}
+
+func itemToEpisode(podcastID, size, duration int, audio string,
+	item *gofeed.Item) *db.PodcastEpisode {
+	return &db.PodcastEpisode{
+		PodcastID:   podcastID,
+		Description: item.Description,
+		Title:       item.Title,
+		Length:      duration,
+		Size:        size,
+		PublishDate: item.PublishedParsed,
+		AudioURL:    audio,
+		Status:      episodeSkipped,
+	}
+}
+
+func (p *Podcasts) findEnclosureAudio(podcastID, duration int,
+	item *gofeed.Item) (*db.PodcastEpisode, bool) {
 	for _, enc := range item.Enclosures {
-		if !strings.HasPrefix(enc.Type, "audio") {
+		if !isAudio(enc.Type, enc.URL) {
 			continue
 		}
 		size, _ := strconv.Atoi(enc.Length)
-		podcastEpisode := db.PodcastEpisode{
-			PodcastID:   podcastID,
-			Description: item.Description,
-			Title:       item.Title,
-			Length:      duration,
-			Size:        size,
-			PublishDate: item.PublishedParsed,
-			AudioURL:    enc.URL,
-			Status:      episodeSkipped,
-		}
-		if err := p.DB.Save(&podcastEpisode).Error; err != nil {
-			return err
-		}
+		return itemToEpisode(podcastID, size, duration, enc.URL, item), true
 	}
-	return nil
+	return nil, false
+}
+
+func (p *Podcasts) findMediaAudio(podcastID, duration int,
+	item *gofeed.Item) (*db.PodcastEpisode, bool) {
+	extensions, ok := item.Extensions["media"]["content"]
+	if !ok {
+		return nil, false
+	}
+	for _, ext := range extensions {
+		if !isAudio(ext.Attrs["type"], ext.Attrs["url"]) {
+			continue
+		}
+		return itemToEpisode(podcastID, 0, duration, ext.Attrs["url"], item), true
+	}
+	return nil, false
 }
 
 func (p *Podcasts) RefreshPodcasts() error {