@Test
public void testComplexChain() throws Exception {
Spark.create("local[4]");
final String rddName =
TestHelper.makeTestDataDirectory(
PersistedInputOutputRDDIntegrateTest.class, "testComplexChain", "graphRDD");
final String rddName2 =
TestHelper.makeTestDataDirectory(
PersistedInputOutputRDDIntegrateTest.class, "testComplexChain", "graphRDD2");
final Configuration configuration = super.getBaseConfiguration();
configuration.setProperty(
Constants.GREMLIN_HADOOP_GRAPH_READER, GryoInputFormat.class.getCanonicalName());
configuration.setProperty(
Constants.GREMLIN_HADOOP_INPUT_LOCATION,
SparkHadoopGraphProvider.PATHS.get("tinkerpop-modern.kryo"));
configuration.setProperty(
Constants.GREMLIN_HADOOP_GRAPH_WRITER, PersistedOutputRDD.class.getCanonicalName());
configuration.setProperty(Constants.GREMLIN_HADOOP_OUTPUT_LOCATION, rddName);
configuration.setProperty(Constants.GREMLIN_SPARK_PERSIST_CONTEXT, true);
assertFalse(Spark.hasRDD(Constants.getGraphLocation(rddName)));
assertEquals(0, Spark.getContext().getPersistentRDDs().size());
Graph graph = GraphFactory.open(configuration);
graph =
graph
.compute(SparkGraphComputer.class)
.persist(GraphComputer.Persist.EDGES)
.program(PageRankVertexProgram.build().iterations(2).create(graph))
.submit()
.get()
.graph();
GraphTraversalSource g = graph.traversal();
assertEquals(6l, g.V().count().next().longValue());
assertEquals(6l, g.E().count().next().longValue());
assertEquals(6l, g.V().values(PageRankVertexProgram.PAGE_RANK).count().next().longValue());
////
assertTrue(Spark.hasRDD(Constants.getGraphLocation(rddName)));
assertEquals(1, Spark.getContext().getPersistentRDDs().size());
////
configuration.setProperty(
Constants.GREMLIN_HADOOP_GRAPH_READER, PersistedInputRDD.class.getCanonicalName());
configuration.setProperty(Constants.GREMLIN_HADOOP_INPUT_LOCATION, rddName);
configuration.setProperty(
Constants.GREMLIN_HADOOP_GRAPH_WRITER, PersistedOutputRDD.class.getCanonicalName());
configuration.setProperty(Constants.GREMLIN_HADOOP_OUTPUT_LOCATION, rddName2);
////
graph = GraphFactory.open(configuration);
graph =
graph
.compute(SparkGraphComputer.class)
.persist(GraphComputer.Persist.EDGES)
.mapReduce(PageRankMapReduce.build().create())
.program(PageRankVertexProgram.build().iterations(2).create(graph))
.submit()
.get()
.graph();
g = graph.traversal();
assertEquals(6l, g.V().count().next().longValue());
assertEquals(6l, g.E().count().next().longValue());
assertEquals(6l, g.V().values(PageRankVertexProgram.PAGE_RANK).count().next().longValue());
////
assertTrue(Spark.hasRDD(Constants.getGraphLocation(rddName)));
assertTrue(Spark.hasRDD(Constants.getGraphLocation(rddName2)));
assertTrue(
Spark.hasRDD(Constants.getMemoryLocation(rddName2, PageRankMapReduce.DEFAULT_MEMORY_KEY)));
assertEquals(3, Spark.getContext().getPersistentRDDs().size());
////
graph = GraphFactory.open(configuration);
graph =
graph
.compute(SparkGraphComputer.class)
.persist(GraphComputer.Persist.VERTEX_PROPERTIES)
.program(PageRankVertexProgram.build().iterations(2).create(graph))
.submit()
.get()
.graph();
g = graph.traversal();
assertEquals(6l, g.V().count().next().longValue());
assertEquals(0l, g.E().count().next().longValue());
assertEquals(6l, g.V().values(PageRankVertexProgram.PAGE_RANK).count().next().longValue());
////
assertTrue(Spark.hasRDD(Constants.getGraphLocation(rddName)));
assertTrue(Spark.hasRDD(Constants.getGraphLocation(rddName2)));
assertFalse(
Spark.hasRDD(Constants.getMemoryLocation(rddName2, PageRankMapReduce.DEFAULT_MEMORY_KEY)));
assertEquals(2, Spark.getContext().getPersistentRDDs().size());
////
graph = GraphFactory.open(configuration);
graph =
graph
.compute(SparkGraphComputer.class)
.persist(GraphComputer.Persist.NOTHING)
.program(PageRankVertexProgram.build().iterations(2).create(graph))
.submit()
.get()
.graph();
assertFalse(Spark.hasRDD(Constants.getGraphLocation(rddName2)));
g = graph.traversal();
assertEquals(0l, g.V().count().next().longValue());
assertEquals(0l, g.E().count().next().longValue());
assertEquals(0l, g.V().values(PageRankVertexProgram.PAGE_RANK).count().next().longValue());
////
assertTrue(Spark.hasRDD(Constants.getGraphLocation(rddName)));
assertFalse(Spark.hasRDD(Constants.getGraphLocation(rddName2)));
assertFalse(
Spark.hasRDD(Constants.getMemoryLocation(rddName2, PageRankMapReduce.DEFAULT_MEMORY_KEY)));
assertEquals(1, Spark.getContext().getPersistentRDDs().size());
Spark.close();
}
@Test
public void testBulkLoaderVertexProgramChainWithInputOutputHelperMapping() throws Exception {
Spark.create("local[4]");
final String rddName =
TestHelper.makeTestDataDirectory(
PersistedInputOutputRDDIntegrateTest.class, UUID.randomUUID().toString());
final Configuration readConfiguration = super.getBaseConfiguration();
readConfiguration.setProperty(
Constants.GREMLIN_HADOOP_GRAPH_READER, GryoInputFormat.class.getCanonicalName());
readConfiguration.setProperty(
Constants.GREMLIN_HADOOP_INPUT_LOCATION,
SparkHadoopGraphProvider.PATHS.get("tinkerpop-modern.kryo"));
readConfiguration.setProperty(
Constants.GREMLIN_HADOOP_GRAPH_WRITER, PersistedOutputRDD.class.getCanonicalName());
readConfiguration.setProperty(Constants.GREMLIN_HADOOP_OUTPUT_LOCATION, rddName);
readConfiguration.setProperty(Constants.GREMLIN_SPARK_PERSIST_CONTEXT, true);
Graph pageRankGraph = GraphFactory.open(readConfiguration);
///////////////
final Configuration writeConfiguration = new BaseConfiguration();
writeConfiguration.setProperty(Graph.GRAPH, TinkerGraph.class.getCanonicalName());
writeConfiguration.setProperty(TinkerGraph.GREMLIN_TINKERGRAPH_GRAPH_FORMAT, "gryo");
writeConfiguration.setProperty(
TinkerGraph.GREMLIN_TINKERGRAPH_GRAPH_LOCATION,
TestHelper.makeTestDataDirectory(PersistedInputOutputRDDIntegrateTest.class)
+ "testBulkLoaderVertexProgramChainWithInputOutputHelperMapping.kryo");
final Graph bulkLoaderGraph =
pageRankGraph
.compute(SparkGraphComputer.class)
.persist(GraphComputer.Persist.EDGES)
.program(PageRankVertexProgram.build().create(pageRankGraph))
.submit()
.get()
.graph();
bulkLoaderGraph
.compute(SparkGraphComputer.class)
.persist(GraphComputer.Persist.NOTHING)
.workers(1)
.program(
BulkLoaderVertexProgram.build()
.userSuppliedIds(true)
.writeGraph(writeConfiguration)
.create(bulkLoaderGraph))
.submit()
.get();
////
Spark.create(readConfiguration);
assertTrue(Spark.hasRDD(Constants.getGraphLocation(rddName)));
assertEquals(1, Spark.getContext().getPersistentRDDs().size());
////
final Graph graph = TinkerGraph.open();
final GraphTraversalSource g = graph.traversal();
graph
.io(IoCore.gryo())
.readGraph(
TestHelper.makeTestDataDirectory(PersistedInputOutputRDDIntegrateTest.class)
+ "testBulkLoaderVertexProgramChainWithInputOutputHelperMapping.kryo");
assertEquals(6l, g.V().count().next().longValue());
assertEquals(6l, g.E().count().next().longValue());
assertEquals("marko", g.V().has("name", "marko").values("name").next());
assertEquals(6l, g.V().values(PageRankVertexProgram.PAGE_RANK).count().next().longValue());
////
Spark.close();
}