exporting data from Neo4j to csv using java

Question

what is the best methodology to export data from Neo4j to CSV, I had imported data from csv to neo4j using CSV importer in the link https://github.com/sroycode/neo4j-import.

Answer 1

A short snippet in Groovy doing this:

@Grab(group="org.neo4j", module="neo4j-cypher", version="1.9")
@Grab(group='net.sf.opencsv', module='opencsv', version='2.3') 
import org.neo4j.kernel.EmbeddedGraphDatabase
import org.neo4j.cypher.javacompat.ExecutionEngine
import au.com.bytecode.opencsv.CSVWriter

assert args, "specify location of graph.db and cypher statement"

def db = new org.neo4j.kernel.EmbeddedGraphDatabase(args[0])
def ee = new ExecutionEngine(db)
def result = ee.execute(args[1])
def columns = result.columns()

System.out.withWriter { writer ->
    CSVWriter csv = new CSVWriter(writer)
    csv.writeNext(columns as String[])

    for (def row in result) {
        def values = columns.collect {row[it]}
        csv.writeNext(values as String[])
    }
}
db.shutdown()

Of course opencsv can be used in a pure Java environment as well.

Answer 2

As you already have the ids, you can get the ids and their fwd relationships in chunks (of 5) using cypher, the csv separator is '|'

To test use neo4j-shell Syntax: neo4j-shell < infile > outfile

infile for node looks like

START n=node(1,2,3,4,5) return ID(n),n.name?,n.property?;
START n=node(6,7,8,9,10) return ID(n),n.name?,n.property?;
....

infile for reln looks like

START n=node(1,2,3,4,5) MATCH n-[r]->m RETURN ID(n),ID(m),TYPE(r),r.someprop?;
START n=node(6,7,8,9,10) MATCH n-[r]->m RETURN ID(n),ID(m),TYPE(r),r.someprop?;
....

To do the same thing in the Java API , just run the same in a for loop.