Filter candidate annotations by dates

This should limit the Annotation.id that need to be compared with
AnnotationSlim.pubid making the query faster.

The default values here avoid having to sync the deploy with the
h-periodic task.

The since value will pick the first annotations in DB and the until will
provide enough annotations (~2M annos) to work on for a few days.

h/tasks/annotations.py

@@ -6,7 +6,7 @@
 
 
 @celery.task
-def fill_annotation_slim(batch_size=1000):
+def fill_annotation_slim(batch_size=1000, since="2012-01-01", until="2017-12-31"):
     """Task to fill the new AnnotationSlim table in batches."""
     # pylint:disable=no-member
 
@@ -15,7 +15,12 @@ def fill_annotation_slim(batch_size=1000):
     annotations = (
         celery.request.db.query(Annotation)
         .outerjoin(AnnotationSlim)
-        .where(AnnotationSlim.pubid.is_(None), Annotation.deleted.is_(False))
+        .where(
+            AnnotationSlim.pubid.is_(None),
+            Annotation.deleted.is_(False),
+            Annotation.created >= since,
+            Annotation.created <= until,
+        )
         .limit(batch_size)
     )
 

tests/unit/h/tasks/annotations_test.py

@@ -1,3 +1,4 @@
+from datetime import datetime, timedelta
 from unittest.mock import call
 
 import pytest
@@ -13,10 +14,14 @@ class TestFillPKAndUserId:
     USERNAME_2 = "USERNAME_2"
 
     def test_it(self, factories, annotation_write_service):
-        annos = factories.Annotation.create_batch(10)
+        now = datetime.now()
+        annos = factories.Annotation.create_batch(10, created=now)
         factories.Annotation.create_batch(10, deleted=True)
+        factories.Annotation.create_batch(10, created=now + timedelta(days=10))
 
-        fill_annotation_slim(batch_size=10)
+        fill_annotation_slim(
+            batch_size=10, since=now - timedelta(days=1), until=now + timedelta(days=1)
+        )
 
         annotation_write_service.upsert_annotation_slim.assert_has_calls(
             [call(anno) for anno in annos], any_order=True