Github user JoshRosen commented on a diff in the pull request:
https://github.com/apache/spark/pull/4205#discussion_r23556968
--- Diff: python/pyspark/graphx/vertex.py ---
@@ -0,0 +1,330 @@
+#
+# 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.
+#
+
+"""
+Python bindings for VertexRDD in GraphX
+"""
+
+import itertools
+import os
+from tempfile import NamedTemporaryFile
+from numpy.numarray.numerictypes import Long
+
+from py4j.java_collections import MapConverter, ListConverter
+import operator
+from pyspark.accumulators import PStatsParam
+from pyspark.rdd import PipelinedRDD
+from pyspark.serializers import CloudPickleSerializer, NoOpSerializer,
AutoBatchedSerializer
+from pyspark import RDD, PickleSerializer, StorageLevel, SparkContext
+from pyspark.traceback_utils import SCCallSiteSync
+
+
+__all__ = ["VertexRDD", "VertexId"]
+
+
+"""
+The default type of vertex id is Long
+A separate VertexId type is defined
+here so that other types can be used
+for vertex ids in future
+"""
+VertexId = Long
+
+
+class VertexRDD(object):
+ """
+ VertexRDD class defines the vertex actions and transformations. The
complete list of
+ transformations and actions for vertices are available at
+ `http://spark.apache.org/docs/latest/graphx-programming-guide.html`
+ These operations are mapped to Scala functions defined
+ in `org.apache.spark.graphx.impl.VertexRDDImpl`
+ """
+
+ def __init__(self, jrdd,
jrdd_deserializer=AutoBatchedSerializer(PickleSerializer())):
+ """
+ Constructor
+ :param jrdd: A JavaRDD reference passed from the
parent
+ RDD object
+ :param jrdd_deserializer: The deserializer used in Python workers
+ created from PythonRDD to execute a
+ serialized Python function and RDD
+ """
+
+ self.name = "VertexRDD"
+ self.jrdd = jrdd
+ self.is_cached = False
+ self.is_checkpointed = False
+ self.ctx = SparkContext._active_spark_context
+ self.jvertex_rdd_deserializer = jrdd_deserializer
+ self.id = jrdd.id()
+ self.partitionFunc = None
+ self.bypass_serializer = False
+ self.preserve_partitioning = False
+
+ self.jvertex_rdd = self.getJavaVertexRDD(jrdd, jrdd_deserializer)
+
+ def __repr__(self):
+ return self.jvertex_rdd.toString()
+
+ def cache(self):
+ """
+ Persist this vertex RDD with the default storage level
(C{MEMORY_ONLY_SER}).
+ """
+ self.is_cached = True
+ self.persist(StorageLevel.MEMORY_ONLY_SER)
+ return self
+
+ def checkpoint(self):
+ self.is_checkpointed = True
+ self.jvertex_rdd.checkpoint()
+
+ def count(self):
+ return self.jvertex_rdd.count()
+
+ def diff(self, other, numPartitions=2):
+ """
+ Hides vertices that are the same between `this` and `other`.
+ For vertices that are different, keeps the values from `other`.
+
+ TODO: give an example
+ """
+ if (isinstance(other, RDD)):
+ vs = self.map(lambda (k, v): (k, (1, v)))
+ ws = other.map(lambda (k, v): (k, (2, v)))
+ return vs.union(ws).groupByKey(numPartitions).mapValues(lambda x:
x.diff(x.__iter__()))
+
+ def isCheckpointed(self):
+ """
+ Return whether this RDD has been checkpointed or not
+ """
+ return self.is_checkpointed
+
+ def mapValues(self, f, preserves_partitioning=False):
+ """
+ Return a new vertex RDD by applying a function to each vertex
attributes,
+ preserving the index
+
+ >>> rdd = sc.parallelize([(1, "b"), (2, "a"), (3, "c")])
+ >>> vertices = VertexRDD(rdd)
+ >>> sorted(vertices.mapValues(lambda x: (x + ":" + x)).collect())
+ [(1, 'a:a'), (2, 'b:b'), (3, 'c:c')]
+ """
+ def func(_, iterator):
+ return itertools.imap(lambda (k, v): (k, f(v)), iterator)
+ return PipelinedVertexRDD(self, func, preserves_partitioning)
+
+ def persist(self, storageLevel=StorageLevel.MEMORY_ONLY_SER):
+ self.is_cached = True
+ java_storage_level = self.ctx._getJavaStorageLevel(storageLevel)
+ self.jvertex_rdd.persist(java_storage_level)
+ return self
+
+ # TODO: This is a hack. take() must call JavaVertexRDD.take()
+ def take(self, num=10):
+ return self.jrdd.take(num)
+
+ def unpersist(self, blocking = False):
+ self.is_cached = False
+ self.jvertex_rdd.unpersist(blocking)
+ return self
+
+ def mapVertexPartitions(self, f, preserve_partitioning=False):
+ def func(s, iterator):
+ return f(iterator)
+ return PipelinedVertexRDD(self, func, preserve_partitioning)
+
+ # TODO
+ def filter(self, f):
+ """
+ Return a new vertex RDD containing only the elements that satisfy
a predicate.
+
+ >>> rdd = sc.parallelize([(1, "b"), (2, "a"), (3, "c")])
+ >>> vertices = VertexRDD(rdd)
+ >>> vertices.filter(lambda x: x._1 % 2 == 0).collect()
+ [2]
+ """
+ return self.jvertex_rdd.filter(f)
+
+ # TODO: The best way to do an innerJoin on vertex RDDs is to use the
optimized inner
+ # TODO: technique defined in VertexRDDImpl. This solution does not
scale
+ def innerJoin(self, other):
+ return self.jrdd.join(other.jrdd)
+
+ def leftJoin(self, other, numPartitions=None):
+ return self.jrdd.leftOuterJoin(other.jrdd, numPartitions)
+
+ def collect(self):
+ """
+ Return a list that contains all of the elements in this RDD.
+ """
+ with SCCallSiteSync(self.ctx) as css:
+ bytesInJava = self.jvertex_rdd.collect().iterator()
+ return list(self._collect_iterator_through_file(bytesInJava))
+
+ def _collect_iterator_through_file(self, iterator):
+ # Transferring lots of data through Py4J can be slow because
+ # socket.readline() is inefficient. Instead, we'll dump the data
to a
+ # file and read it back.
+ tempFile = NamedTemporaryFile(delete=False, dir=self.ctx._temp_dir)
+ tempFile.close()
+ self.ctx._writeToFile(iterator, tempFile.name)
+ # Read the data into Python and deserialize it:
+ with open(tempFile.name, 'rb') as tempFile:
+ for item in
self.jvertex_rdd_deserializer.load_stream(tempFile):
+ yield item
+ os.unlink(tempFile.name)
+
+ def getJavaVertexRDD(self, rdd, rdd_deserializer):
+ if self.bypass_serializer:
+ self.jvertex_rdd_deserializer = NoOpSerializer()
+ rdd_deserializer = NoOpSerializer()
+ enable_profile = self.ctx._conf.get("spark.python.profile",
"false") == "true"
+ profileStats = self.ctx.accumulator(None, PStatsParam) if
enable_profile else None
+ def f(index, iterator):
+ return iterator
+ command = (f, profileStats, rdd_deserializer,
+ rdd_deserializer)
+ # the serialized command will be compressed by broadcast
+ ser = CloudPickleSerializer()
+ pickled_command = ser.dumps(command)
+ if len(pickled_command) > (1 << 20): # 1M
+ self.broadcast = self.ctx.broadcast(pickled_command)
+ pickled_command = ser.dumps(self.broadcast)
+
+ # the serialized command will be compressed by broadcast
+ broadcast_vars = ListConverter().convert(
+ [x._jbroadcast for x in self.ctx._pickled_broadcast_vars],
+ self.ctx._gateway._gateway_client)
+ self.ctx._pickled_broadcast_vars.clear()
+ env = MapConverter().convert(self.ctx.environment,
+ self.ctx._gateway._gateway_client)
+ includes = ListConverter().convert(self.ctx._python_includes,
+
self.ctx._gateway._gateway_client)
+ target_storage_level = StorageLevel.MEMORY_ONLY
+ java_storage_level =
self.ctx._getJavaStorageLevel(target_storage_level)
+ prdd = self.ctx._jvm.PythonVertexRDD(rdd._jrdd,
+
bytearray(pickled_command),
+ env, includes,
self.preserve_partitioning,
+ self.ctx.pythonExec,
+ broadcast_vars,
self.ctx._javaAccumulator,
+ java_storage_level)
+ self.jvertex_rdd = prdd.asJavaVertexRDD()
+ if enable_profile:
+ self.id = self.jvertex_rdd.id()
+ self.ctx._add_profile(self.id, profileStats)
+ return self.jvertex_rdd
+
+
+class PipelinedVertexRDD(VertexRDD):
+
+ """
+ Pipelined mapValues in VertexRDD:
+
+ >>> rdd = sc.parallelize([(1, ("Alice", 29)), (2, ("Bob", 30)), \
+ (3, ("Charlie", 31)), (4, ("Dwayne", 32))])
+ >>> vertices = VertexRDD(rdd)
+ >>> vertices.mapValues(lambda x: x[1] * 2).cache().collect()
+ [(1, ("Alice", 58)), (2, ("Bob", 60)), \
+ (3, ("Charlie", 62)), (4, ("Dwayne", 64))]
+
+ Pipelined reduces in VertexRDD:
+ >>> from operator import add
+ >>> rdd.map(lambda x: 2 * x).reduce(add)
+ 20
+ >>> rdd.flatMap(lambda x: [x, x]).reduce(add)
+ 20
+ """
+
+ def __init__(self, prev, func, preservesPartitioning=False):
+ if not isinstance(prev, PipelinedVertexRDD) or not
prev.is_pipelinable():
+ # This transformation is the first in its stage:
+ self.func = func
+ self.preservesPartitioning = preservesPartitioning
+ self.prev_jvertex_rdd = prev.jvertex_rdd
+ self.prev_jvertex_rdd_deserializer =
prev.jvertex_rdd_deserializer
+ else:
+ prev_func = prev.func
+
+ def pipeline_func(split, iterator):
+ return func(split, prev_func(split, iterator))
+ self.func = pipeline_func
+ self.preservesPartitioning = \
+ prev.preservesPartitioning and preservesPartitioning
+ self.prev_jvertex_rdd = prev.jvertex_rdd
+ self.prev_jvertex_rdd_deserializer =
prev.prev_jvertex_rdd_deserializer
+
+ self.is_cached = False
+ self.is_checkpointed = False
+ self.ctx = prev.ctx
+ self.prev = prev
+ self.jvrdd_val = None
+ self.id = None
+ self.jvertex_rdd_deserializer = self.ctx.serializer
+ self.bypass_serializer = False
+ self.partitionFunc = prev._partitionFunc if
self.preservesPartitioning else None
+ self.broadcast = None
+
+ def __del__(self):
+ if self.broadcast:
+ self.broadcast.unpersist()
+ self.broadcast = None
+
+ @property
+ def jvertex_rdd(self):
--- End diff --
There's a lot of duplication between this method, the one in edge RDD, and
the regular PySpark RDD. I'm worried that this could become a maintenance
problem since lots of common code is duplicated and will fall out of sync. Do
you know if there's a good way to add some abstraction here so that we're not
copying, say, the profiling code into three places?
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at [email protected] or file a JIRA ticket
with INFRA.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
