Protocol.load: reload objects if our copy is over 30d old

fixes #628

no clue how much this will impact our outbound request load. we'll see!

Author: snarfed

protocol.py

@@ -1,5 +1,6 @@
 """Base protocol class and common code."""
 import copy
+from datetime import timedelta
 import logging
 from threading import Lock
 from urllib.parse import urljoin
@@ -38,6 +39,8 @@
     'video',
 )
 
+OBJECT_REFRESH_AGE = timedelta(days=30)
+
 # activity ids that we've already handled and can now ignore.
 # used in Protocol.receive
 seen_ids = LRUCache(100000)
@@ -1099,44 +1102,48 @@ def load(cls, id, remote=None, local=True, **kwargs):
           requests.HTTPError: anything that :meth:`fetch` raises
         """
         assert local or remote is not False
-
-        if remote is not True:
-            with objects_cache_lock:
-                cached = objects_cache.get(id)
-                if cached:
-                    # make a copy so that if the client modifies this entity in
-                    # memory, those modifications aren't applied to the cache
-                    # until they explicitly put() the modified entity.
-                    # NOTE: keep in sync with Object._post_put_hook!
-                    return Object(id=cached.key.id(), **cached.to_dict(
-                        # computed properties
-                        exclude=['as1', 'expire', 'object_ids', 'type']))
+        logger.debug(f'Loading Object {id} local={local} remote={remote}')
 
         obj = orig_as1 = None
-        if local:
+        with objects_cache_lock:
+            cached = objects_cache.get(id)
+            if cached:
+                # make a copy so that if the client modifies this entity in
+                # memory, those modifications aren't applied to the cache
+                # until they explicitly put() the modified entity.
+                # NOTE: keep in sync with Object._post_put_hook!
+                logger.debug('  got from cache')
+                obj = Object(id=cached.key.id(), **cached.to_dict(
+                    # computed properties
+                    exclude=['as1', 'expire', 'object_ids', 'type']))
+
+        if local and not obj:
             obj = Object.get_by_id(id)
-            if obj and (obj.as1 or obj.raw or obj.deleted):
-                logger.info('  got from datastore')
+            if not obj:
+                logger.debug(f' not in datastore')
+            elif obj.as1 or obj.raw or obj.deleted:
+                logger.debug('  got from datastore')
                 obj.new = False
-                orig_as1 = obj.as1
                 if remote is not True:
                     with objects_cache_lock:
                         objects_cache[id] = obj
-                    return obj
 
-        if remote is True:
-            logger.debug(f'Loading Object {id} local={local} remote={remote}, forced refresh requested')
-        elif remote is False:
-            logger.debug(f'Loading Object {id} local={local} remote={remote} {"empty" if obj else "not"} in datastore')
+        if remote is False:
             return obj
+        elif remote is None and obj:
+            if obj.updated < util.as_utc(util.now() - OBJECT_REFRESH_AGE):
+                logger.debug(f'  last updated {obj.updated}, refreshing')
+            else:
+                return obj
 
         if obj:
+            orig_as1 = obj.as1
             obj.clear()
             obj.new = False
         else:
             obj = Object(id=id)
             if local:
-                logger.info('  not in datastore')
+                logger.debug('  not in datastore')
                 obj.new = True
                 obj.changed = False
 

tests/test_models.py

@@ -414,7 +414,7 @@ def test_get_by_id(self):
         self.assert_entities_equal(obj, Object.get_by_id('ab^^c'))
 
     def test_get_by_id_uses_cache(self):
-        obj = Object(id='foo', our_as1={'x': 'y'})
+        obj = Object(id='foo', our_as1={'x': 'y'}, updated=util.as_utc(NOW))
         protocol.objects_cache['foo'] = obj
         loaded = Fake.load('foo')
         self.assert_entities_equal(obj, loaded)
@@ -439,7 +439,7 @@ def test_put_cached_makes_copy(self):
         }, Fake.load('foo').our_as1)
 
     def test_get_by_id_cached_makes_copy(self):
-        obj = Object(id='foo', our_as1={'x': 'y'})
+        obj = Object(id='foo', our_as1={'x': 'y'}, updated=util.as_utc(NOW))
         protocol.objects_cache['foo'] = obj
         loaded = Fake.load('foo')
         self.assert_entities_equal(obj, loaded)

tests/test_protocol.py

@@ -1,5 +1,6 @@
 """Unit tests for protocol.py."""
 import copy
+from datetime import timedelta
 import logging
 from unittest import skip
 from unittest.mock import patch
@@ -8,9 +9,9 @@
 from flask import g
 from google.cloud import ndb
 from granary import as2
-from oauth_dropins.webutil import appengine_info
+from oauth_dropins.webutil import appengine_info, util
 from oauth_dropins.webutil.flask_util import CLOUD_TASKS_QUEUE_HEADER, NoContent
-from oauth_dropins.webutil.testutil import requests_response
+from oauth_dropins.webutil.testutil import NOW, requests_response
 import requests
 from werkzeug.exceptions import BadRequest
 
@@ -200,7 +201,7 @@ def test_load_existing_empty_deleted(self):
         self.assertEqual([], Fake.fetched)
 
     def test_load_cached(self):
-        obj = Object(id='foo', our_as1={'x': 'y'})
+        obj = Object(id='foo', our_as1={'x': 'y'}, updated=util.as_utc(NOW))
         protocol.objects_cache['foo'] = obj
         loaded = Fake.load('foo')
         self.assert_entities_equal(obj, loaded)
@@ -346,6 +347,21 @@ def test_load_preserves_fragment(self):
         self.assert_entities_equal(stored, got)
         self.assertEqual([], Fake.fetched)
 
+    def test_load_refresh(self):
+        Fake.fetchable['foo'] = {'fetched': 'x'}
+
+        too_old = (NOW.replace(tzinfo=None)
+                   - protocol.OBJECT_REFRESH_AGE
+                   - timedelta(days=1))
+        with patch('models.Object.updated._now', return_value=too_old):
+            obj = Object(id='foo', our_as1={'orig': 'y'}, status='in progress')
+            obj.put()
+
+        protocol.objects_cache['foo'] = obj
+
+        loaded = Fake.load('foo')
+        self.assertEqual({'fetched': 'x', 'id': 'foo'}, loaded.our_as1)
+
     def test_actor_key(self):
         user = self.make_user(id='fake:a', cls=Fake)
         a_key = user.key