load-edges-with-properties-and-string-id.scala

import org.apache.spark._
import org.apache.spark.graphx._

val triplets = sc.textFile(path).flatMap { line =>
  if (!line.isEmpty && line(0) != '#') {
    val lineArray = line.split("\\s+")
    if (lineArray.length < 2) {
      None
    } else {
      val t = new EdgeTriplet[String, String]
      t.srcId = lineArray(0).hashCode
      t.srcAttr = lineArray(0)
      t.attr = lineArray(1)
      t.dstId = lineArray(2).hashCode
      t.dstAttr = lineArray(2)
      Some(t)
    }
  } else {
    None
  }
}

val vertices = triplets.flatMap(t => Array((t.srcId, t.srcAttr), (t.dstId, t.dstAttr)))
val edges = triplets.map(t => t: Edge[String])

Graph(vertices, edges)

load-edges-with-properties.scala

import org.apache.spark._
import org.apache.spark.graphx._

// Load one edge per line in whitespace-delimited format: srcVertexId edgeAttr dstVertexId
val edges = sc.textFile(path).flatMap { line =>
  if (!line.isEmpty && line(0) != '#') {
    val lineArray = line.split("\\s+")
    if (lineArray.length < 2) {
      None
    } else {
      val srcId = lineArray(0).toLong
      val attr = // parse lineArray(1) as appropriate
      val dstId = lineArray(2).toLong
      Some(Edge(srcId, dstId, attr))
    }
  } else {
    None
  }
}

Graph.fromEdges(edges, 1)

Hi Ankur,
The input that I have is n1 P1 n2 where n1 and n2 are strings and not numbers(int or long). How can I make a graph in this case?

Thanks
Soumitra

Would hash collisions cause wrong graph construction?

I am also worried about hash collision part