demo.scala

def dumpCypher() {
  val createConstraint = """create index on :Node(nid)"""

  val dumpNodes =
    """USING PERIODIC COMMIT 5000
      |LOAD CSV WITH HEADERS FROM 'file:///home/roman/toDump.csv' AS line
      |CREATE (c:Node {title: line.title, nid: toInt(line.nid)})""".stripMargin

  val db     = new GraphDatabaseFactory().newEmbeddedDatabase("/tmp/skip")
  val cypher = new ExecutionEngine(db)

  def withTx(b: => Unit) = {
    val tx = db.beginTx()
    b
    tx.success()
    tx.close()
  }

  withTx {
    cypher.execute("""CREATE (r:Root:Node {title: "", treeName: "tree", nid: 0})""")
  }

  cypher.execute(createConstraint)

  cypher.execute(dumpNodes)

  for (i <- 0 to (550000 / 20000)) {
    cypher.execute(
      s"""USING PERIODIC COMMIT 5000
         |LOAD CSV WITH HEADERS FROM 'file:///home/roman/toDump.csv' AS line
         |WITH line SKIP ${i * 20000} limit ${(i + 1) * 20000}
         |MATCH
         |  (parent:Node {nid: toInt(line.pid)}),
         |  (child: Node {nid: toInt(line.nid)})
         |CREATE
         |  (child)-[:CHILD_OF]->(parent)""".stripMargin
    )
  }

  db.shutdown()
}