Tweak reply processing

blueprints/tasks.py

@@ -572,12 +572,13 @@ def task_process_reply() -> _Response:
         )
 
         for new_reply in new_replies:
-            if find_one_activity(by_object_id(new_reply.id)) or DB.replies.find_one(
+            if find_one_activity(
-                {"remote_id": root_reply}
+                {**by_object_id(new_reply.id), **by_type(ap.ActivityType.CREATE)}
-            ):
+            ) or DB.replies.find_one(by_remote_id(new_reply.id)):
                 continue
 
             actor = new_reply.get_actor()
+            # Save the reply with the cached actor and the thread flag/ID
             save_reply(
                 new_reply,
                 {

core/activitypub.py

@@ -35,11 +35,11 @@ from core.meta import Box
 from core.meta import MetaKey
 from core.meta import by_object_id
 from core.meta import by_remote_id
+from core.meta import by_type
 from core.meta import flag
 from core.meta import inc
 from core.meta import upsert
 from core.tasks import Tasks
-from core.meta import by_type
 from utils import now
 
 logger = logging.getLogger(__name__)

utils/lookup.py

@@ -35,7 +35,10 @@ def lookup(url: str) -> ap.BaseActivity:
     except requests.RequestException as err:
         raise RemoteServerUnavailableError(f"failed to GET {url}: {err!r}")
 
-    resp.raise_for_status()
+    try:
+        resp.raise_for_status()
+    except Exception:
+        return ap.fetch_remote_activity(url)
 
     # If the page is HTML, maybe it contains an alternate link pointing to an AP object
     for alternate in mf2py.parse(resp.text).get("alternates", []):

utils/nodeinfo.py

@@ -69,7 +69,7 @@ def get_software_name(server: str) -> str:
 
             return SoftwareName.UNKNOWN.value
         except requests.RequestException:
-            pass
+            return SoftwareName.UNKNOWN.value
 
     if _try_mastodon_api(server):
         return SoftwareName.MASTODON.value