Optimize the exercise queryset

Author: rolandgeider

wger/exercises/api/views.py

@@ -20,7 +20,10 @@
 # Django
 from django.conf import settings
 from django.contrib.postgres.search import TrigramSimilarity
-from django.db.models import Q
+from django.db.models import (
+    Prefetch,
+    Q,
+)
 from django.utils.decorators import method_decorator
 from django.utils.translation import gettext as _
 from django.views.decorators.cache import cache_page
@@ -328,7 +331,6 @@ class ExerciseInfoViewset(viewsets.ReadOnlyModelViewSet):
     is the recommended way to access the exercise data.
     """
 
-    queryset = Exercise.objects.all()
     serializer_class = ExerciseInfoSerializer
     ordering_fields = '__all__'
     filterset_class = ExerciseFilterSet
@@ -343,6 +345,32 @@ class ExerciseInfoViewset(viewsets.ReadOnlyModelViewSet):
         'license_author',
     )
 
+    def get_queryset(self):
+        """
+        Optimize the queryset with select_related and prefetch_related to avoid
+        n+1 queries
+
+        One improvement is that we access the historical records of the exercise
+        from the django-simple-history package, which are not really prefetchable
+        since they are a manager.
+        """
+
+        return Exercise.objects.select_related(
+            'category',
+            'license',
+            'variations',
+        ).prefetch_related(
+            'muscles',
+            'muscles_secondary',
+            'equipment',
+            'exerciseimage_set',
+            'exercisevideo_set',
+            Prefetch(
+                'translations',
+                queryset=Translation.objects.prefetch_related('alias_set', 'exercisecomment_set'),
+            ),
+        )
+
 
 class ExerciseSubmissionViewSet(CreateAPIView):
     """