[SPARK-11812][PYSPARK] invFunc=None works properly with python's reduceByKeyAndWindow

invFunc is optional and can be None. Instead of invFunc (the parameter) invReduceFunc (a local function) was checked for trueness (that is, not None, in this context). A local function is never None,
thus the case of invFunc=None (a common one when inverse reduction is not defined) was treated incorrectly, resulting in loss of data.

In addition, the docstring used wrong parameter names, also fixed.

Author: David Tolpin <[email protected]>

Closes #9775 from dtolpin/master.

(cherry picked from commit 599a8c6)
Signed-off-by: Tathagata Das <[email protected]>

Author: dtolpin

python/pyspark/streaming/dstream.py

@@ -524,8 +524,8 @@ def reduceByKeyAndWindow(self, func, invFunc, windowDuration, slideDuration=None
         `invFunc` can be None, then it will reduce all the RDDs in window, could be slower
         than having `invFunc`.
 
-        @param reduceFunc:     associative reduce function
-        @param invReduceFunc:  inverse function of `reduceFunc`
+        @param func:           associative reduce function
+        @param invFunc:        inverse function of `reduceFunc`
         @param windowDuration: width of the window; must be a multiple of this DStream's
                               batching interval
         @param slideDuration:  sliding interval of the window (i.e., the interval after which
@@ -556,7 +556,7 @@ def invReduceFunc(t, a, b):
                                     if kv[1] is not None else kv[0])
 
         jreduceFunc = TransformFunction(self._sc, reduceFunc, reduced._jrdd_deserializer)
-        if invReduceFunc:
+        if invFunc:
             jinvReduceFunc = TransformFunction(self._sc, invReduceFunc, reduced._jrdd_deserializer)
         else:
             jinvReduceFunc = None

python/pyspark/streaming/tests.py

@@ -448,6 +448,17 @@ def test_reduce_by_invalid_window(self):
         self.assertRaises(ValueError, lambda: d1.reduceByKeyAndWindow(None, None, 0.1, 0.1))
         self.assertRaises(ValueError, lambda: d1.reduceByKeyAndWindow(None, None, 1, 0.1))
 
+    def test_reduce_by_key_and_window_with_none_invFunc(self):
+        input = [range(1), range(2), range(3), range(4), range(5), range(6)]
+
+        def func(dstream):
+            return dstream.map(lambda x: (x, 1))\
+                .reduceByKeyAndWindow(operator.add, None, 5, 1)\
+                .filter(lambda kv: kv[1] > 0).count()
+
+        expected = [[2], [4], [6], [6], [6], [6]]
+        self._test_func(input, func, expected)
+
 
 class StreamingContextTests(PySparkStreamingTestCase):