Add a setting for enabling iterator caching

cachalot/monkey_patch.py

@@ -1,4 +1,5 @@
 import re
+import types
 from collections.abc import Iterable
 from functools import wraps
 from time import time
@@ -62,6 +63,10 @@ def _get_result_or_execute_query(execute_query_func, cache,
                 pass
 
     result = execute_query_func()
+
+    if result.__class__ == types.GeneratorType and not cachalot_settings.CACHALOT_CACHE_ITERATORS:
+        return result
+
     if result.__class__ not in ITERABLES and isinstance(result, Iterable):
         result = list(result)
 

cachalot/settings.py

@@ -49,6 +49,7 @@ class Settings(object):
     CACHALOT_DATABASES = 'supported_only'
     CACHALOT_TIMEOUT = None
     CACHALOT_CACHE_RANDOM = False
+    CACHALOT_CACHE_ITERATORS = False
     CACHALOT_INVALIDATE_RAW = True
     CACHALOT_ONLY_CACHABLE_TABLES = ()
     CACHALOT_ONLY_CACHABLE_APPS = ()

cachalot/tests/read.py

@@ -244,13 +244,20 @@ def test_distinct(self):
         self.assert_query_cached(qs, [self.t1])
 
     def test_iterator(self):
-        with self.assertNumQueries(1):
+        with self.assertNumQueries(2):
             data1 = list(Test.objects.iterator())
-        with self.assertNumQueries(0):
             data2 = list(Test.objects.iterator())
         self.assertListEqual(data2, data1)
         self.assertListEqual(data2, [self.t1, self.t2])
 
+        with self.settings(CACHALOT_CACHE_ITERATORS=True):
+            with self.assertNumQueries(1):
+                data1 = list(Test.objects.iterator())
+            with self.assertNumQueries(0):
+                data2 = list(Test.objects.iterator())
+            self.assertListEqual(data2, data1)
+            self.assertListEqual(data2, [self.t1, self.t2])
+
     def test_in_bulk(self):
         with self.assertNumQueries(1):
             data1 = Test.objects.in_bulk((5432, self.t2.pk, 9200))

docs/quickstart.rst

@@ -106,6 +106,20 @@ Settings
 :Description: If set to ``True``, caches random queries
               (those with ``order_by('?')``).
 
+``CACHALOT_CACHE_ITERATORS``
+~~~~~~~~~~~~~~~~~~~~~~~~~
+
+:Default: ``False``
+:Description:
+   If set to ``True``, cache responses from QuerySets that return
+   generators. This is useful for caching the result sets of QuerySets that
+   use ``.iterator()``.
+
+   .. warnings::
+      ``.iterator()`` is often used for large result sets. Caching these can use large
+      amounts of local memory because django-cachalot has to first convert them to a list to
+      store them in the cache.
+
 .. _CACHALOT_INVALIDATE_RAW:
 
 ``CACHALOT_INVALIDATE_RAW``