From 906bbb34bbb61984d9bfc9116cfbfe80e1523760 Mon Sep 17 00:00:00 2001
From: Tony Wasserka <918-neobrain@users.noreply.dev.funkwhale.audio>
Date: Sun, 11 Apr 2021 14:47:59 +0200
Subject: [PATCH] Display RSS episode descriptions based on the
 <content:encoded> RSS tags

The previously used <description> tag often only contains plain text,
whereas <content:encoded> is typically HTML-encoded and hence better
suited for display.

Closes #1405
---
 api/funkwhale_api/audio/serializers.py | 22 +++++++++++++++++++++-
 changes/changelog.d/1405.bugfix        |  1 +
 2 files changed, 22 insertions(+), 1 deletion(-)
 create mode 100644 changes/changelog.d/1405.bugfix

diff --git a/api/funkwhale_api/audio/serializers.py b/api/funkwhale_api/audio/serializers.py
index e15325d8c..0bd8e6c4a 100644
--- a/api/funkwhale_api/audio/serializers.py
+++ b/api/funkwhale_api/audio/serializers.py
@@ -636,6 +636,7 @@ class RssFeedItemSerializer(serializers.Serializer):
     links = serializers.ListField()
     tags = serializers.ListField(required=False)
     summary_detail = serializers.DictField(required=False)
+    content = serializers.ListField(required=False)
     published_parsed = DummyField(required=False)
     image = serializers.DictField(required=False)
 
@@ -648,6 +649,16 @@ class RssFeedItemSerializer(serializers.Serializer):
             "text": content,
         }
 
+    def validate_content(self, v):
+        # TODO: Are there RSS feeds that use more than one content item?
+        content = v[0].get("value")
+        if not content:
+            return
+        return {
+            "content_type": v[0].get("type", "text/plain"),
+            "text": content,
+        }
+
     def validate_image(self, v):
         url = v.get("href")
         if url:
@@ -782,7 +793,16 @@ class RssFeedItemSerializer(serializers.Serializer):
         if tags:
             tags_models.set_tags(track, *tags)
 
-        summary = validated_data.get("summary_detail")
+        # "content" refers to the <content:encoded> node in the RSS feed,
+        # whereas "summary_detail" refers to the <description> node.
+        # <description> is intended to be used as a short summary and is often
+        # encoded merely as plain text, whereas <content:encoded> contains
+        # the full episode description and is generally encoded as HTML.
+        #
+        # For details, see https://www.rssboard.org/rss-profile#element-channel-item-description
+        summary = validated_data.get("content")
+        if not summary:
+            summary = validated_data.get("summary_detail")
         if summary:
             common_utils.attach_content(track, "description", summary)
 
diff --git a/changes/changelog.d/1405.bugfix b/changes/changelog.d/1405.bugfix
new file mode 100644
index 000000000..6c3ed5752
--- /dev/null
+++ b/changes/changelog.d/1405.bugfix
@@ -0,0 +1 @@
+Improve formatting of RSS episode descriptions (#1405)