reduceby defaults to item in no init case

toolz/itertoolz.py

@@ -449,7 +449,7 @@ def frequencies(seq):
     return dict(d)
 
 
-def reduceby(key, binop, seq, init):
+def reduceby(key, binop, seq, init=no_default):
     """ Perform a simultaneous groupby and reduction
 
     The computation:
@@ -468,18 +468,28 @@ def reduceby(key, binop, seq, init):
     operate in much less space.  This makes it suitable for larger datasets
     that do not fit comfortably in memory
 
+    Simple Examples
+    ---------------
+
     >>> from operator import add, mul
-    >>> data = [1, 2, 3, 4, 5]
     >>> iseven = lambda x: x % 2 == 0
-    >>> reduceby(iseven, add, data, 0)
+
+    >>> data = [1, 2, 3, 4, 5]
+
+    >>> reduceby(iseven, add, data)
     {False: 9, True: 6}
-    >>> reduceby(iseven, mul, data, 1)
+
+    >>> reduceby(iseven, mul, data)
     {False: 15, True: 8}
 
+    Complex Example
+    ---------------
+
     >>> projects = [{'name': 'build roads', 'state': 'CA', 'cost': 1000000},
     ...             {'name': 'fight crime', 'state': 'IL', 'cost': 100000},
     ...             {'name': 'help farmers', 'state': 'IL', 'cost': 2000000},
     ...             {'name': 'help farmers', 'state': 'CA', 'cost': 200000}]
+
     >>> reduceby(lambda x: x['state'],              # doctest: +SKIP
     ...          lambda acc, x: acc + x['cost'],
     ...          projects, 0)
@@ -489,7 +499,11 @@ def reduceby(key, binop, seq, init):
     for item in seq:
         k = key(item)
         if k not in d:
-            d[k] = init
+            if init is no_default:
+                d[k] = item
+                continue
+            else:
+                d[k] = init
         d[k] = binop(d[k], item)
     return d
 

toolz/tests/test_itertoolz.py

@@ -208,6 +208,10 @@ def test_reduceby():
                     projects, 0) == {'CA': 1200000, 'IL': 2100000}
 
 
+def test_reduce_by_init():
+    assert reduceby(iseven, add, [1, 2, 3, 4]) == {True: 2 + 4, False: 1 + 3}
+
+
 def test_iterate():
     assert list(itertools.islice(iterate(inc, 0), 0, 5)) == [0, 1, 2, 3, 4]
     assert list(take(4, iterate(double, 1))) == [1, 2, 4, 8]