Make a query before spawning cache actor task

core/indexes.py

@@ -1,6 +1,7 @@
 import pymongo
 
 from config import DB
+from config import MEDIA_CACHE
 from core.meta import MetaKey
 from core.meta import _meta
 
@@ -73,3 +74,6 @@ def create_indexes():
             ("meta.deleted", pymongo.ASCENDING),
         ]
     )
+
+    # For the is_actor_icon_cached query
+    MEDIA_CACHE.fs._GridFS__files.create_index([("url", 1), ("kind", 1)])

core/tasks.py

@@ -4,6 +4,7 @@ from datetime import timezone
 
 from poussetaches import PousseTaches
 
+from config import MEDIA_CACHE
 from utils import parse_datetime
 
 p = PousseTaches(
@@ -25,7 +26,10 @@ class Tasks:
         )
 
     @staticmethod
-    def cache_actor_icon(icon_url: str, actor_iri: str):
+    def cache_actor_icon(icon_url: str, actor_iri: str) -> None:
+        if MEDIA_CACHE.is_actor_icon_cached(icon_url):
+            return None
+
         p.push({"icon_url": icon_url, "actor_iri": actor_iri}, "/task/cache_actor_icon")
 
     @staticmethod

utils/media.py

@@ -136,8 +136,11 @@ class MediaCache(object):
                     remote_id=remote_id,
                 )
 
+    def is_actor_icon_cached(self, url: str) -> bool:
+        return bool(self.fs.find_one({"url": url, "kind": Kind.ACTOR_ICON.value}))
+
     def cache_actor_icon(self, url: str) -> None:
-        if self.fs.find_one({"url": url, "kind": Kind.ACTOR_ICON.value}):
+        if self.is_actor_icon_cached(url):
             return
         i = load(url, self.user_agent)
         for size in [50, 80]: