clean up codes

Author: Ken Takagiwa

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

@@ -11,20 +11,15 @@
         exit(-1)
     conf = SparkConf()
     conf.setAppName("PythonStreamingNetworkWordCount")
-    conf.set("spark.default.parallelism", 1)
     ssc = StreamingContext(conf=conf, duration=Seconds(1))
 
     lines = ssc.socketTextStream(sys.argv[1], int(sys.argv[2]))
     fm_lines = lines.flatMap(lambda x: x.split(" "))
-    filtered_lines = fm_lines.filter(lambda line: "Spark" in line)
     mapped_lines = fm_lines.map(lambda x: (x, 1))
-    reduced_lines = mapped_lines.reduce(add)
-    counted_lines = reduced_lines.count()
+    reduced_lines = mapped_lines.reduceByKey(add)
 
     fm_lines.pyprint()
-    filtered_lines.pyprint()
     mapped_lines.pyprint()
     reduced_lines.pyprint()
-    counted_lines.pyprint()
     ssc.start()
     ssc.awaitTermination()

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

@@ -21,7 +21,7 @@
     fm_lines = lines.flatMap(lambda x: x.split(" "))
     filtered_lines = fm_lines.filter(lambda line: "Spark" in line)
     mapped_lines = fm_lines.map(lambda x: (x, 1))
-    reduced_lines = mapped_lines.reduce(add)
+    reduced_lines = mapped_lines.reduceByKey(add)
 
     fm_lines.pyprint()
     filtered_lines.pyprint()

python/pyspark/streaming/dstream.py

@@ -22,13 +22,15 @@ def count(self):
         """
 
         """
-        #TODO make sure count implementation, thiis different from what pyspark does
-        return self._mapPartitions(lambda i: [sum(1 for _ in i)]).map(lambda x: (None, 1))
+        pass
+        #TODO: make sure count implementation, thiis different from what pyspark does
+        #return self._mapPartitions(lambda i: [sum(1 for _ in i)]).map(lambda x: (None, 1))
 
     def _sum(self):
         """
         """
-        return self._mapPartitions(lambda x: [sum(x)]).reduce(operator.add)
+        pass
+        #return self._mapPartitions(lambda x: [sum(x)]).reduce(operator.add)
 
     def print_(self):
         """
@@ -85,7 +87,6 @@ def _mapPartitionsWithIndex(self, f, preservesPartitioning=False):
         """
         return PipelinedDStream(self, f, preservesPartitioning)
 
-
     def reduceByKey(self, func, numPartitions=None):
         """
         Merge the value for each key using an associative reduce function.
@@ -121,7 +122,7 @@ def combineLocally(iterator):
                 else:
                     combiners[k] = mergeValue(combiners[k], v)
             return combiners.iteritems()
-        locally_combined = self.mapPartitions(combineLocally)
+        locally_combined = self._mapPartitions(combineLocally)
         shuffled = locally_combined.partitionBy(numPartitions)
         def _mergeCombiners(iterator):
             combiners = {}
@@ -131,12 +132,11 @@ def _mergeCombiners(iterator):
                 else:
                     combiners[k] = mergeCombiners(combiners[k], v)
             return combiners.iteritems()
-        return shuffled.mapPartitions(_mergeCombiners) 
+        return shuffled._mapPartitions(_mergeCombiners)
 
     def partitionBy(self, numPartitions, partitionFunc=None):
         """
         Return a copy of the DStream partitioned using the specified partitioner.
-
         """
         if numPartitions is None:
             numPartitions = self.ctx._defaultReducePartitions()

streaming/src/main/scala/org/apache/spark/streaming/dstream/DStream.scala

@@ -623,7 +623,7 @@ abstract class DStream[T: ClassTag] (
     new ForEachDStream(this, context.sparkContext.clean(foreachFunc)).register()
   }
 
-//TODO move pyprint to PythonDStream and executed by py4j call back function
+//TODO: move pyprint to PythonDStream and executed by py4j call back function
   /**
    * Print the first ten elements of each PythonRDD generated in this PythonDStream. This is an output
    * operator, so this PythonDStream will be registered as an output stream and there materialized.
@@ -647,6 +647,7 @@ abstract class DStream[T: ClassTag] (
 
       // pythonExec should be passed from python. Move pyprint to PythonDStream
       val pythonExec = new ProcessBuilder().environment().get("PYSPARK_PYTHON")
+
       val sparkHome = new ProcessBuilder().environment().get("SPARK_HOME")
       // Call python script to deserialize and print result in stdout
       val pb = new ProcessBuilder(pythonExec, sparkHome + "/python/pyspark/streaming/pyprint.py", tempFile.getAbsolutePath)