[SPARK-2377] Python API for Streaming

This patch brings Python API for Streaming.

This patch is based on work from @giwa

Author: giwa <ugw.gi.world@gmail.com>
Author: Ken Takagiwa <ken@Kens-MacBook-Pro.local>
Author: Davies Liu <davies.liu@gmail.com>
Author: Ken Takagiwa <ken@kens-mbp.gateway.sonic.net>
Author: Tathagata Das <tathagata.das1565@gmail.com>
Author: Ken <ugw.gi.world@gmail.com>
Author: Ken Takagiwa <ugw.gi.world@gmail.com>
Author: Matthew Farrellee <matt@redhat.com>

Closes #2538 from davies/streaming and squashes the following commits:

64561e4 [Davies Liu] fix tests
331ecce [Davies Liu] fix example
3e2492b [Davies Liu] change updateStateByKey() to easy API
182be73 [Davies Liu] Merge branch 'master' of github.com:apache/spark into streaming
02d0575 [Davies Liu] add wrapper for foreachRDD()
bebeb4a [Davies Liu] address all comments
6db00da [Davies Liu] Merge branch 'master' of github.com:apache/spark into streaming
8380064 [Davies Liu] Merge branch 'master' of github.com:apache/spark into streaming
52c535b [Davies Liu] remove fix for sum()
e108ec1 [Davies Liu]  address comments
37fe06f [Davies Liu] use random port for callback server
d05871e [Davies Liu] remove reuse of PythonRDD
be5e5ff [Davies Liu] merge branch of env, make tests stable.
8071541 [Davies Liu] Merge branch 'env' into streaming
c7bbbce [Davies Liu] fix sphinx docs
6bb9d91 [Davies Liu] Merge branch 'master' of github.com:apache/spark into streaming
4d0ea8b [Davies Liu] clear reference of SparkEnv after stop
54bd92b [Davies Liu] improve tests
c2b31cb [Davies Liu] Merge branch 'master' of github.com:apache/spark into streaming
7a88f9f [Davies Liu] rollback RDD.setContext(), use textFileStream() to test checkpointing
bd8a4c2 [Davies Liu] fix scala style
7797c70 [Davies Liu] refactor
ff88bec [Davies Liu] rename RDDFunction to TransformFunction
d328aca [Davies Liu] fix serializer in queueStream
6f0da2f [Davies Liu] recover from checkpoint
fa7261b [Davies Liu] refactor
a13ff34 [Davies Liu] address comments
8466916 [Davies Liu] support checkpoint
9a16bd1 [Davies Liu] change number of partitions during tests
b98d63f [Davies Liu] change private[spark] to private[python]
eed6e2a [Davies Liu] rollback not needed changes
e00136b [Davies Liu] address comments
069a94c [Davies Liu] fix the number of partitions during window()
338580a [Davies Liu] change _first(), _take(), _collect() as private API
19797f9 [Davies Liu] clean up
6ebceca [Davies Liu] add more tests
c40c52d [Davies Liu] change first(), take(n) to has the same behavior as RDD
98ac6c2 [Davies Liu] support ssc.transform()
b983f0f [Davies Liu] address comments
847f9b9 [Davies Liu] add more docs, add first(), take()
e059ca2 [Davies Liu] move check of window into Python
fce0ef5 [Davies Liu] rafactor of foreachRDD()
7001b51 [Davies Liu] refactor of queueStream()
26ea396 [Davies Liu] refactor
74df565 [Davies Liu] fix print and docs
b32774c [Davies Liu] move java_import into streaming
604323f [Davies Liu] enable streaming tests
c499ba0 [Davies Liu] remove Time and Duration
3f0fb4b [Davies Liu] refactor fix tests
c28f520 [Davies Liu] support updateStateByKey
d357b70 [Davies Liu] support windowed dstream
bd13026 [Davies Liu] fix examples
eec401e [Davies Liu] refactor, combine TransformedRDD, fix reuse PythonRDD, fix union
9a57685 [Davies Liu] fix python style
bd27874 [Davies Liu] fix scala style
7339be0 [Davies Liu] delete tests
7f53086 [Davies Liu] support transform(), refactor and cleanup
df098fc [Davies Liu] Merge branch 'master' into giwa
550dfd9 [giwa] WIP fixing 1.1 merge
5cdb6fa [giwa] changed for SCCallSiteSync
e685853 [giwa] meged with rebased 1.1 branch
2d32a74 [giwa] added some StreamingContextTestSuite
4a59e1e [giwa] WIP:added more test for StreamingContext
8ffdbf1 [giwa] added atexit to handle callback server
d5f5fcb [giwa] added comment for StreamingContext.sparkContext
63c881a [giwa] added StreamingContext.sparkContext
d39f102 [giwa] added StreamingContext.remember
d542743 [giwa] clean up code
2fdf0de [Matthew Farrellee] Fix scalastyle errors
c0a06bc [giwa] delete not implemented functions
f385976 [giwa] delete inproper comments
b0f2015 [giwa] added comment in dstream._test_output
bebb3f3 [giwa] remove the last brank line
fbed8da [giwa] revert pom.xml
8ed93af [giwa] fixed explanaiton
066ba90 [giwa] revert pom.xml
fa4af88 [giwa] remove duplicated import
6ae3caa [giwa] revert pom.xml
7dc7391 [giwa] fixed typo
62dc7a3 [giwa] clean up exmples
f04882c [giwa] clen up examples
b171ec3 [giwa] fixed pep8 violation
f198d14 [giwa] clean up code
3166d31 [giwa] clean up
c00e091 [giwa] change test case not to use awaitTermination
e80647e [giwa] adopted the latest compression way of python command
58e41ff [giwa] merge with master
455e5af [giwa] removed wasted print in DStream
af336b7 [giwa] add comments
ddd4ee1 [giwa] added TODO coments
99ce042 [giwa] added saveAsTextFiles and saveAsPickledFiles
2a06cdb [giwa] remove waste duplicated code
c5ecfc1 [giwa] basic function test cases are passed
8dcda84 [giwa] all tests are passed if numSlice is 2 and the numver of each input is over 4
795b2cd [giwa] broke something
1e126bf [giwa] WIP: solved partitioned and None is not recognized
f67cf57 [giwa] added mapValues and flatMapVaules WIP for glom and mapPartitions test
953deb0 [giwa] edited the comment to add more precise description
af610d3 [giwa] removed unnesessary changes
c1d546e [giwa] fixed PEP-008 violation
99410be [giwa] delete waste file
b3b0362 [giwa] added basic operation test cases
9cde7c9 [giwa] WIP added test case
bd3ba53 [giwa] WIP
5c04a5f [giwa] WIP: added PythonTestInputStream
019ef38 [giwa] WIP
1934726 [giwa] update comment
376e3ac [giwa] WIP
932372a [giwa] clean up dstream.py
0b09cff [giwa] added stop in StreamingContext
92e333e [giwa] implemented reduce and count function in Dstream
1b83354 [giwa] Removed the waste line
88f7506 [Ken Takagiwa] Kill py4j callback server properly
54b5358 [Ken Takagiwa] tried to restart callback server
4f07163 [Tathagata Das] Implemented DStream.foreachRDD in the Python API using Py4J callback server.
fe02547 [Ken Takagiwa] remove waste file
2ad7bd3 [Ken Takagiwa] clean up codes
6197a11 [Ken Takagiwa] clean up code
eb4bf48 [Ken Takagiwa] fix map function
98c2a00 [Ken Takagiwa] added count operation but this implementation need double check
58591d2 [Ken Takagiwa] reduceByKey is working
0df7111 [Ken Takagiwa] delete old file
f485b1d [Ken Takagiwa] fied input of socketTextDStream
dd6de81 [Ken Takagiwa] initial commit for socketTextStream
247fd74 [Ken Takagiwa] modified the code base on comment in https://github.com/tdas/spark/pull/10
4bcb318 [Ken Takagiwa] implementing transform function in Python
38adf95 [Ken Takagiwa] added reducedByKey not working yet
66fcfff [Ken Takagiwa] modify dstream.py to fix indent error
41886c2 [Ken Takagiwa] comment PythonDStream.PairwiseDStream
0b99bec [Ken] initial commit for pySparkStreaming
c214199 [giwa] added testcase for combineByKey
5625bdc [giwa] added gorupByKey testcase
10ab87b [giwa] added sparkContext as input parameter in StreamingContext
10b5b04 [giwa] removed wasted print in DStream
e54f986 [giwa] add comments
16aa64f [giwa] added TODO coments
74535d4 [giwa] added saveAsTextFiles and saveAsPickledFiles
f76c182 [giwa] remove waste duplicated code
18c8723 [giwa] modified streaming test case to add coment
13fb44c [giwa] basic function test cases are passed
3000b2b [giwa] all tests are passed if numSlice is 2 and the numver of each input is over 4
ff14070 [giwa] broke something
bcdec33 [giwa] WIP: solved partitioned and None is not recognized
270a9e1 [giwa] added mapValues and flatMapVaules WIP for glom and mapPartitions test
bb10956 [giwa] edited the comment to add more precise description
253a863 [giwa] removed unnesessary changes
3d37822 [giwa] fixed PEP-008 violation
f21cab3 [giwa] delete waste file
878bad7 [giwa] added basic operation test cases
ce2acd2 [giwa] WIP added test case
9ad6855 [giwa] WIP
1df77f5 [giwa] WIP: added PythonTestInputStream
1523b66 [giwa] WIP
8a0fbbc [giwa] update comment
fe648e3 [giwa] WIP
29c2bc5 [giwa] initial commit for testcase
4d40d63 [giwa] clean up dstream.py
c462bb3 [giwa] added stop in StreamingContext
d2c01ba [giwa] clean up examples
3c45cd2 [giwa] implemented reduce and count function in Dstream
b349649 [giwa] Removed the waste line
3b498e1 [Ken Takagiwa] Kill py4j callback server properly
84a9668 [Ken Takagiwa] tried to restart callback server
9ab8952 [Tathagata Das] Added extra line.
05e991b [Tathagata Das] Added missing file
b1d2a30 [Tathagata Das] Implemented DStream.foreachRDD in the Python API using Py4J callback server.
678e854 [Ken Takagiwa] remove waste file
0a8bbbb [Ken Takagiwa] clean up codes
bab31c1 [Ken Takagiwa] clean up code
72b9738 [Ken Takagiwa] fix map function
d3ee86a [Ken Takagiwa] added count operation but this implementation need double check
15feea9 [Ken Takagiwa] edit python sparkstreaming example
6f98e50 [Ken Takagiwa] reduceByKey is working
c455c8d [Ken Takagiwa] added reducedByKey not working yet
dc6995d [Ken Takagiwa] delete old file
b31446a [Ken Takagiwa] fixed typo of network_workdcount.py
ccfd214 [Ken Takagiwa] added doctest for pyspark.streaming.duration
0d1b954 [Ken Takagiwa] fied input of socketTextDStream
f746109 [Ken Takagiwa] initial commit for socketTextStream
bb7ccf3 [Ken Takagiwa] remove unused import in python
224fc5e [Ken Takagiwa] add empty line
d2099d8 [Ken Takagiwa] sorted the import following Spark coding convention
5bac7ec [Ken Takagiwa] revert streaming/pom.xml
e1df940 [Ken Takagiwa] revert pom.xml
494cae5 [Ken Takagiwa] remove not implemented DStream functions in python
17a74c6 [Ken Takagiwa] modified the code base on comment in https://github.com/tdas/spark/pull/10
1a0f065 [Ken Takagiwa] implementing transform function in Python
d7b4d6f [Ken Takagiwa] added reducedByKey not working yet
87438e2 [Ken Takagiwa] modify dstream.py to fix indent error
b406252 [Ken Takagiwa] comment PythonDStream.PairwiseDStream
454981d [Ken] initial commit for pySparkStreaming
150b94c [giwa] added some StreamingContextTestSuite
f7bc8f9 [giwa] WIP:added more test for StreamingContext
ee50c5a [giwa] added atexit to handle callback server
fdc9125 [giwa] added comment for StreamingContext.sparkContext
f5bfb70 [giwa] added StreamingContext.sparkContext
da09768 [giwa] added StreamingContext.remember
d68b568 [giwa] clean up code
4afa390 [giwa] clean up code
1fd6bc7 [Ken Takagiwa] Merge pull request #2 from mattf/giwa-master
d9d59fe [Matthew Farrellee] Fix scalastyle errors
67473a9 [giwa] delete not implemented functions
c97377c [giwa] delete inproper comments
2ea769e [giwa] added comment in dstream._test_output
3b27bd4 [giwa] remove the last brank line
acfcaeb [giwa] revert pom.xml
93f7637 [giwa] fixed explanaiton
50fd6f9 [giwa] revert pom.xml
4f82c89 [giwa] remove duplicated import
9d1de23 [giwa] revert pom.xml
7339df2 [giwa] fixed typo
9c85e48 [giwa] clean up exmples
24f95db [giwa] clen up examples
0d30109 [giwa] fixed pep8 violation
b7dab85 [giwa] improve test case
583e66d [giwa] move tests for streaming inside streaming directory
1d84142 [giwa] remove unimplement test
f0ea311 [giwa] clean up code
171edeb [giwa] clean up
4dedd2d [giwa] change test case not to use awaitTermination
268a6a5 [giwa] Changed awaitTermination not to call awaitTermincation in Scala. Just use time.sleep instread
09a28bf [giwa] improve testcases
58150f5 [giwa] Changed the test case to focus the test operation
199e37f [giwa] adopted the latest compression way of python command
185fdbf [giwa] merge with master
f1798c4 [giwa] merge with master
e70f706 [giwa] added testcase for combineByKey
e162822 [giwa] added gorupByKey testcase
97742fe [giwa] added sparkContext as input parameter in StreamingContext
14d4c0e [giwa] removed wasted print in DStream
6d8190a [giwa] add comments
4aa99e4 [giwa] added TODO coments
e9fab72 [giwa] added saveAsTextFiles and saveAsPickledFiles
94f2b65 [giwa] remove waste duplicated code
580fbc2 [giwa] modified streaming test case to add coment
99e4bb3 [giwa] basic function test cases are passed
7051a84 [giwa] all tests are passed if numSlice is 2 and the numver of each input is over 4
35933e1 [giwa] broke something
9767712 [giwa] WIP: solved partitioned and None is not recognized
4f2d7e6 [giwa] added mapValues and flatMapVaules WIP for glom and mapPartitions test
33c0f94d [giwa] edited the comment to add more precise description
774f18d [giwa] removed unnesessary changes
3a671cc [giwa] remove export PYSPARK_PYTHON in spark submit
8efa266 [giwa] fixed PEP-008 violation
fa75d71 [giwa] delete waste file
7f96294 [giwa] added basic operation test cases
3dda31a [giwa] WIP added test case
1f68b78 [giwa] WIP
c05922c [giwa] WIP: added PythonTestInputStream
1fd12ae [giwa] WIP
c880a33 [giwa] update comment
5d22c92 [giwa] WIP
ea4b06b [giwa] initial commit for testcase
5a9b525 [giwa] clean up dstream.py
79c5809 [giwa] added stop in StreamingContext
189dcea [giwa] clean up examples
b8d7d24 [giwa] implemented reduce and count function in Dstream
b6468e6 [giwa] Removed the waste line
b47b5fd [Ken Takagiwa] Kill py4j callback server properly
19ddcdd [Ken Takagiwa] tried to restart callback server
c9fc124 [Tathagata Das] Added extra line.
4caae3f [Tathagata Das] Added missing file
4eff053 [Tathagata Das] Implemented DStream.foreachRDD in the Python API using Py4J callback server.
5e822d4 [Ken Takagiwa] remove waste file
aeaf8a5 [Ken Takagiwa] clean up codes
9fa249b [Ken Takagiwa] clean up code
05459c6 [Ken Takagiwa] fix map function
a9f4ecb [Ken Takagiwa] added count operation but this implementation need double check
d1ee6ca [Ken Takagiwa] edit python sparkstreaming example
0b8b7d0 [Ken Takagiwa] reduceByKey is working
d25d5cf [Ken Takagiwa] added reducedByKey not working yet
7f7c5d1 [Ken Takagiwa] delete old file
967dc26 [Ken Takagiwa] fixed typo of network_workdcount.py
57fb740 [Ken Takagiwa] added doctest for pyspark.streaming.duration
4b69fb1 [Ken Takagiwa] fied input of socketTextDStream
02f618a [Ken Takagiwa] initial commit for socketTextStream
4ce4058 [Ken Takagiwa] remove unused import in python
856d98e [Ken Takagiwa] add empty line
490e338 [Ken Takagiwa] sorted the import following Spark coding convention
5594bd4 [Ken Takagiwa] revert pom.xml
2adca84 [Ken Takagiwa] remove not implemented DStream functions in python
e551e13 [Ken Takagiwa] add coment for hack why PYSPARK_PYTHON is needed in spark-submit
3758175 [Ken Takagiwa] add coment for hack why PYSPARK_PYTHON is needed in spark-submit
c5518b4 [Ken Takagiwa] modified the code base on comment in https://github.com/tdas/spark/pull/10
dcf243f [Ken Takagiwa] implementing transform function in Python
9af03f4 [Ken Takagiwa] added reducedByKey not working yet
6e0d9c7 [Ken Takagiwa] modify dstream.py to fix indent error
e497b9b [Ken Takagiwa] comment PythonDStream.PairwiseDStream
5c3a683 [Ken] initial commit for pySparkStreaming
665bfdb [giwa] added testcase for combineByKey
a3d2379 [giwa] added gorupByKey testcase
636090a [giwa] added sparkContext as input parameter in StreamingContext
e7ebb08 [giwa] removed wasted print in DStream
d8b593b [giwa] add comments
ea9c873 [giwa] added TODO coments
89ae38a [giwa] added saveAsTextFiles and saveAsPickledFiles
e3033fc [giwa] remove waste duplicated code
a14c7e1 [giwa] modified streaming test case to add coment
536def4 [giwa] basic function test cases are passed
2112638 [giwa] all tests are passed if numSlice is 2 and the numver of each input is over 4
080541a [giwa] broke something
0704b86 [giwa] WIP: solved partitioned and None is not recognized
90a6484 [giwa] added mapValues and flatMapVaules WIP for glom and mapPartitions test
a65f302 [giwa] edited the comment to add more precise description
bdde697 [giwa] removed unnesessary changes
e8c7bfc [giwa] remove export PYSPARK_PYTHON in spark submit
3334169 [giwa] fixed PEP-008 violation
db0a303 [giwa] delete waste file
2cfd3a0 [giwa] added basic operation test cases
90ae568 [giwa] WIP added test case
a120d07 [giwa] WIP
f671cdb [giwa] WIP: added PythonTestInputStream
56fae45 [giwa] WIP
e35e101 [giwa] Merge branch 'master' into testcase
ba5112d [giwa] update comment
28aa56d [giwa] WIP
fb08559 [giwa] initial commit for testcase
a613b85 [giwa] clean up dstream.py
c40c0ef [giwa] added stop in StreamingContext
31e4260 [giwa] clean up examples
d2127d6 [giwa] implemented reduce and count function in Dstream
48f7746 [giwa] Removed the waste line
0f83eaa [Ken Takagiwa] delete py4j 0.8.1
1679808 [Ken Takagiwa] Kill py4j callback server properly
f96cd4e [Ken Takagiwa] tried to restart callback server
fe86198 [Ken Takagiwa] add py4j 0.8.2.1 but server is not launched
1064fe0 [Ken Takagiwa] Merge branch 'master' of https://github.com/giwa/spark
28c6620 [Ken Takagiwa] Implemented DStream.foreachRDD in the Python API using Py4J callback server
85b0fe1 [Ken Takagiwa] Merge pull request #1 from tdas/python-foreach
54e2e8c [Tathagata Das] Added extra line.
e185338 [Tathagata Das] Added missing file
a778d4b [Tathagata Das] Implemented DStream.foreachRDD in the Python API using Py4J callback server.
cc2092b [Ken Takagiwa] remove waste file
d042ac6 [Ken Takagiwa] clean up codes
84a021f [Ken Takagiwa] clean up code
bd20e17 [Ken Takagiwa] fix map function
d01a125 [Ken Takagiwa] added count operation but this implementation need double check
7d05109 [Ken Takagiwa] merge with remote branch
ae464e0 [Ken Takagiwa] edit python sparkstreaming example
04af046 [Ken Takagiwa] reduceByKey is working
3b6d7b0 [Ken Takagiwa] implementing transform function in Python
571d52d [Ken Takagiwa] added reducedByKey not working yet
5720979 [Ken Takagiwa] delete old file
e604fcb [Ken Takagiwa] fixed typo of network_workdcount.py
4b7c08b [Ken Takagiwa] Merge branch 'master' of https://github.com/giwa/spark
ce7d426 [Ken Takagiwa] added doctest for pyspark.streaming.duration
a8c9fd5 [Ken Takagiwa] fixed for socketTextStream
a61fa9e [Ken Takagiwa] fied input of socketTextDStream
1e84f41 [Ken Takagiwa] initial commit for socketTextStream
6d012f7 [Ken Takagiwa] remove unused import in python
25d30d5 [Ken Takagiwa] add empty line
6e0a64a [Ken Takagiwa] sorted the import following Spark coding convention
fa4a7fc [Ken Takagiwa] revert streaming/pom.xml
8f8202b [Ken Takagiwa] revert streaming pom.xml
c9d79dd [Ken Takagiwa] revert pom.xml
57e3e52 [Ken Takagiwa] remove not implemented DStream functions in python
0a516f5 [Ken Takagiwa] add coment for hack why PYSPARK_PYTHON is needed in spark-submit
a7a0b5c [Ken Takagiwa] add coment for hack why PYSPARK_PYTHON is needed in spark-submit
72bfc66 [Ken Takagiwa] modified the code base on comment in https://github.com/tdas/spark/pull/10
69e9cd3 [Ken Takagiwa] implementing transform function in Python
94a0787 [Ken Takagiwa] added reducedByKey not working yet
88068cf [Ken Takagiwa] modify dstream.py to fix indent error
1367be5 [Ken Takagiwa] comment PythonDStream.PairwiseDStream
eb2b3ba [Ken] Merge remote-tracking branch 'upstream/master'
d8e51f9 [Ken] initial commit for pySparkStreaming

1 files changed, 325 insertions, 0 deletions

diff --git a/python/pyspark/streaming/context.py b/python/pyspark/streaming/context.py
new file mode 100644
index 0000000000..dc9dc41121
--- /dev/null
+++ b/python/pyspark/streaming/context.py
@@ -0,0 +1,325 @@
+#
+# Licensed to the Apache Software Foundation (ASF) under one or more
+# contributor license agreements.  See the NOTICE file distributed with
+# this work for additional information regarding copyright ownership.
+# The ASF licenses this file to You under the Apache License, Version 2.0
+# (the "License"); you may not use this file except in compliance with
+# the License.  You may obtain a copy of the License at
+#
+#    http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+#
+import os
+import sys
+
+from py4j.java_collections import ListConverter
+from py4j.java_gateway import java_import, JavaObject
+
+from pyspark import RDD, SparkConf
+from pyspark.serializers import UTF8Deserializer, CloudPickleSerializer
+from pyspark.context import SparkContext
+from pyspark.storagelevel import StorageLevel
+from pyspark.streaming.dstream import DStream
+from pyspark.streaming.util import TransformFunction, TransformFunctionSerializer
+
+__all__ = ["StreamingContext"]
+
+
+def _daemonize_callback_server():
+    """
+    Hack Py4J to daemonize callback server
+
+    The thread of callback server has daemon=False, it will block the driver
+    from exiting if it's not shutdown. The following code replace `start()`
+    of CallbackServer with a new version, which set daemon=True for this
+    thread.
+
+    Also, it will update the port number (0) with real port
+    """
+    # TODO: create a patch for Py4J
+    import socket
+    import py4j.java_gateway
+    logger = py4j.java_gateway.logger
+    from py4j.java_gateway import Py4JNetworkError
+    from threading import Thread
+
+    def start(self):
+        """Starts the CallbackServer. This method should be called by the
+        client instead of run()."""
+        self.server_socket = socket.socket(socket.AF_INET, socket.SOCK_STREAM)
+        self.server_socket.setsockopt(socket.SOL_SOCKET, socket.SO_REUSEADDR,
+                                      1)
+        try:
+            self.server_socket.bind((self.address, self.port))
+            if not self.port:
+                # update port with real port
+                self.port = self.server_socket.getsockname()[1]
+        except Exception as e:
+            msg = 'An error occurred while trying to start the callback server: %s' % e
+            logger.exception(msg)
+            raise Py4JNetworkError(msg)
+
+        # Maybe thread needs to be cleanup up?
+        self.thread = Thread(target=self.run)
+        self.thread.daemon = True
+        self.thread.start()
+
+    py4j.java_gateway.CallbackServer.start = start
+
+
+class StreamingContext(object):
+    """
+    Main entry point for Spark Streaming functionality. A StreamingContext
+    represents the connection to a Spark cluster, and can be used to create
+    L{DStream} various input sources. It can be from an existing L{SparkContext}.
+    After creating and transforming DStreams, the streaming computation can
+    be started and stopped using `context.start()` and `context.stop()`,
+    respectively. `context.awaitTransformation()` allows the current thread
+    to wait for the termination of the context by `stop()` or by an exception.
+    """
+    _transformerSerializer = None
+
+    def __init__(self, sparkContext, batchDuration=None, jssc=None):
+        """
+        Create a new StreamingContext.
+
+        @param sparkContext: L{SparkContext} object.
+        @param batchDuration: the time interval (in seconds) at which streaming
+                              data will be divided into batches
+        """
+
+        self._sc = sparkContext
+        self._jvm = self._sc._jvm
+        self._jssc = jssc or self._initialize_context(self._sc, batchDuration)
+
+    def _initialize_context(self, sc, duration):
+        self._ensure_initialized()
+        return self._jvm.JavaStreamingContext(sc._jsc, self._jduration(duration))
+
+    def _jduration(self, seconds):
+        """
+        Create Duration object given number of seconds
+        """
+        return self._jvm.Duration(int(seconds * 1000))
+
+    @classmethod
+    def _ensure_initialized(cls):
+        SparkContext._ensure_initialized()
+        gw = SparkContext._gateway
+
+        java_import(gw.jvm, "org.apache.spark.streaming.*")
+        java_import(gw.jvm, "org.apache.spark.streaming.api.java.*")
+        java_import(gw.jvm, "org.apache.spark.streaming.api.python.*")
+
+        # start callback server
+        # getattr will fallback to JVM, so we cannot test by hasattr()
+        if "_callback_server" not in gw.__dict__:
+            _daemonize_callback_server()
+            # use random port
+            gw._start_callback_server(0)
+            # gateway with real port
+            gw._python_proxy_port = gw._callback_server.port
+            # get the GatewayServer object in JVM by ID
+            jgws = JavaObject("GATEWAY_SERVER", gw._gateway_client)
+            # update the port of CallbackClient with real port
+            gw.jvm.PythonDStream.updatePythonGatewayPort(jgws, gw._python_proxy_port)
+
+        # register serializer for TransformFunction
+        # it happens before creating SparkContext when loading from checkpointing
+        cls._transformerSerializer = TransformFunctionSerializer(
+            SparkContext._active_spark_context, CloudPickleSerializer(), gw)
+
+    @classmethod
+    def getOrCreate(cls, checkpointPath, setupFunc):
+        """
+        Either recreate a StreamingContext from checkpoint data or create a new StreamingContext.
+        If checkpoint data exists in the provided `checkpointPath`, then StreamingContext will be
+        recreated from the checkpoint data. If the data does not exist, then the provided setupFunc
+        will be used to create a JavaStreamingContext.
+
+        @param checkpointPath Checkpoint directory used in an earlier JavaStreamingContext program
+        @param setupFunc      Function to create a new JavaStreamingContext and setup DStreams
+        """
+        # TODO: support checkpoint in HDFS
+        if not os.path.exists(checkpointPath) or not os.listdir(checkpointPath):
+            ssc = setupFunc()
+            ssc.checkpoint(checkpointPath)
+            return ssc
+
+        cls._ensure_initialized()
+        gw = SparkContext._gateway
+
+        try:
+            jssc = gw.jvm.JavaStreamingContext(checkpointPath)
+        except Exception:
+            print >>sys.stderr, "failed to load StreamingContext from checkpoint"
+            raise
+
+        jsc = jssc.sparkContext()
+        conf = SparkConf(_jconf=jsc.getConf())
+        sc = SparkContext(conf=conf, gateway=gw, jsc=jsc)
+        # update ctx in serializer
+        SparkContext._active_spark_context = sc
+        cls._transformerSerializer.ctx = sc
+        return StreamingContext(sc, None, jssc)
+
+    @property
+    def sparkContext(self):
+        """
+        Return SparkContext which is associated with this StreamingContext.
+        """
+        return self._sc
+
+    def start(self):
+        """
+        Start the execution of the streams.
+        """
+        self._jssc.start()
+
+    def awaitTermination(self, timeout=None):
+        """
+        Wait for the execution to stop.
+        @param timeout: time to wait in seconds
+        """
+        if timeout is None:
+            self._jssc.awaitTermination()
+        else:
+            self._jssc.awaitTermination(int(timeout * 1000))
+
+    def stop(self, stopSparkContext=True, stopGraceFully=False):
+        """
+        Stop the execution of the streams, with option of ensuring all
+        received data has been processed.
+
+        @param stopSparkContext: Stop the associated SparkContext or not
+        @param stopGracefully: Stop gracefully by waiting for the processing
+                              of all received data to be completed
+        """
+        self._jssc.stop(stopSparkContext, stopGraceFully)
+        if stopSparkContext:
+            self._sc.stop()
+
+    def remember(self, duration):
+        """
+        Set each DStreams in this context to remember RDDs it generated
+        in the last given duration. DStreams remember RDDs only for a
+        limited duration of time and releases them for garbage collection.
+        This method allows the developer to specify how to long to remember
+        the RDDs (if the developer wishes to query old data outside the
+        DStream computation).
+
+        @param duration: Minimum duration (in seconds) that each DStream
+                        should remember its RDDs
+        """
+        self._jssc.remember(self._jduration(duration))
+
+    def checkpoint(self, directory):
+        """
+        Sets the context to periodically checkpoint the DStream operations for master
+        fault-tolerance. The graph will be checkpointed every batch interval.
+
+        @param directory: HDFS-compatible directory where the checkpoint data
+                         will be reliably stored
+        """
+        self._jssc.checkpoint(directory)
+
+    def socketTextStream(self, hostname, port, storageLevel=StorageLevel.MEMORY_AND_DISK_SER_2):
+        """
+        Create an input from TCP source hostname:port. Data is received using
+        a TCP socket and receive byte is interpreted as UTF8 encoded ``\\n`` delimited
+        lines.
+
+        @param hostname:      Hostname to connect to for receiving data
+        @param port:          Port to connect to for receiving data
+        @param storageLevel:  Storage level to use for storing the received objects
+        """
+        jlevel = self._sc._getJavaStorageLevel(storageLevel)
+        return DStream(self._jssc.socketTextStream(hostname, port, jlevel), self,
+                       UTF8Deserializer())
+
+    def textFileStream(self, directory):
+        """
+        Create an input stream that monitors a Hadoop-compatible file system
+        for new files and reads them as text files. Files must be wrriten to the
+        monitored directory by "moving" them from another location within the same
+        file system. File names starting with . are ignored.
+        """
+        return DStream(self._jssc.textFileStream(directory), self, UTF8Deserializer())
+
+    def _check_serializers(self, rdds):
+        # make sure they have same serializer
+        if len(set(rdd._jrdd_deserializer for rdd in rdds)) > 1:
+            for i in range(len(rdds)):
+                # reset them to sc.serializer
+                rdds[i] = rdds[i]._reserialize()
+
+    def queueStream(self, rdds, oneAtATime=True, default=None):
+        """
+        Create an input stream from an queue of RDDs or list. In each batch,
+        it will process either one or all of the RDDs returned by the queue.
+
+        NOTE: changes to the queue after the stream is created will not be recognized.
+
+        @param rdds:       Queue of RDDs
+        @param oneAtATime: pick one rdd each time or pick all of them once.
+        @param default:    The default rdd if no more in rdds
+        """
+        if default and not isinstance(default, RDD):
+            default = self._sc.parallelize(default)
+
+        if not rdds and default:
+            rdds = [rdds]
+
+        if rdds and not isinstance(rdds[0], RDD):
+            rdds = [self._sc.parallelize(input) for input in rdds]
+        self._check_serializers(rdds)
+
+        jrdds = ListConverter().convert([r._jrdd for r in rdds],
+                                        SparkContext._gateway._gateway_client)
+        queue = self._jvm.PythonDStream.toRDDQueue(jrdds)
+        if default:
+            default = default._reserialize(rdds[0]._jrdd_deserializer)
+            jdstream = self._jssc.queueStream(queue, oneAtATime, default._jrdd)
+        else:
+            jdstream = self._jssc.queueStream(queue, oneAtATime)
+        return DStream(jdstream, self, rdds[0]._jrdd_deserializer)
+
+    def transform(self, dstreams, transformFunc):
+        """
+        Create a new DStream in which each RDD is generated by applying
+        a function on RDDs of the DStreams. The order of the JavaRDDs in
+        the transform function parameter will be the same as the order
+        of corresponding DStreams in the list.
+        """
+        jdstreams = ListConverter().convert([d._jdstream for d in dstreams],
+                                            SparkContext._gateway._gateway_client)
+        # change the final serializer to sc.serializer
+        func = TransformFunction(self._sc,
+                                 lambda t, *rdds: transformFunc(rdds).map(lambda x: x),
+                                 *[d._jrdd_deserializer for d in dstreams])
+        jfunc = self._jvm.TransformFunction(func)
+        jdstream = self._jssc.transform(jdstreams, jfunc)
+        return DStream(jdstream, self, self._sc.serializer)
+
+    def union(self, *dstreams):
+        """
+        Create a unified DStream from multiple DStreams of the same
+        type and same slide duration.
+        """
+        if not dstreams:
+            raise ValueError("should have at least one DStream to union")
+        if len(dstreams) == 1:
+            return dstreams[0]
+        if len(set(s._jrdd_deserializer for s in dstreams)) > 1:
+            raise ValueError("All DStreams should have same serializer")
+        if len(set(s._slideDuration for s in dstreams)) > 1:
+            raise ValueError("All DStreams should have same slide duration")
+        first = dstreams[0]
+        jrest = ListConverter().convert([d._jdstream for d in dstreams[1:]],
+                                        SparkContext._gateway._gateway_client)
+        return DStream(self._jssc.union(first._jdstream, jrest), self, first._jrdd_deserializer)