adress comments

examples/src/main/python/streaming/mqtt_wordcount.py

@@ -44,11 +44,10 @@
     sc = SparkContext(appName="PythonStreamingMQTTWordCount")
     ssc = StreamingContext(sc, 1)
 
-    broker_url = sys.argv[1]
+    brokerUrl = sys.argv[1]
     topic = sys.argv[2]
 
-    data = MQTTUtils.createStream(ssc, topic, broker_url)
-    lines = data.map(lambda x: x[1])
+    lines = MQTTUtils.createStream(ssc, brokerUrl, topic)
     counts = lines.flatMap(lambda line: line.split(" ")) \
         .map(lambda word: (word, 1)) \
         .reduceByKey(lambda a, b: a+b)

python/pyspark/streaming/mqtt.py

@@ -19,7 +19,7 @@
 from py4j.java_gateway import java_import, Py4JError
 
 from pyspark.storagelevel import StorageLevel
-from pyspark.serializers import PairDeserializer, NoOpSerializer
+from pyspark.serializers import UTF8Deserializer
 from pyspark.streaming import DStream
 
 __all__ = ['MQTTUtils']
@@ -28,7 +28,7 @@
 class MQTTUtils(object):
 
     @staticmethod
-    def createStream(ssc, brokerUrl, topic
+    def createStream(ssc, brokerUrl, topic,
                      storageLevel=StorageLevel.MEMORY_AND_DISK_SER_2):
         """
         Create an input stream that pulls messages from a Mqtt Broker.
@@ -52,6 +52,4 @@ def createStream(ssc, brokerUrl, topic
                 print " $ bin/spark-submit --driver-class-path external/mqtt-assembly/target/" + \
                       "scala-*/spark-streaming-mqtt-assembly-*.jar"
             raise e
-        ser = PairDeserializer(NoOpSerializer(), NoOpSerializer())
-        stream = DStream(jstream, ssc, ser)
-        return stream
+        return DStream(jstream, ssc, UTF8Deserializer())