Proposed fix. Allow empty group_by([], reducer) ie. GROUPBY 0 (#68)

Co-authored-by: Q <quintonparker@gmail.com>

Author: quintonparker

redisearch/aggregation.py

@@ -86,8 +86,6 @@ class Group(object):
     This object automatically created in the `AggregateRequest.group_by()`
     """
     def __init__(self, fields, reducers):
-        if not fields:
-            raise ValueError('need at least one field')
         if not reducers:
             raise ValueError('Need at least one reducer')
 

test/test_builder.py

@@ -40,7 +40,11 @@ def testGroup(self):
         # Check the group class on its own
         self.assertRaises(ValueError, a.Group, [], [])
         self.assertRaises(ValueError, a.Group, ['foo'], [])
-        self.assertRaises(ValueError, a.Group, [], r.count())
+
+        # Zero fields, single reducer
+        g = a.Group([], r.count())
+        ret = g.build_args()
+        self.assertEqual(['GROUPBY', '0', 'REDUCE', 'COUNT', '0'], ret)
 
         # Single field, single reducer
         g = a.Group('foo', r.count())