gistfile1.groovy

conf = new BaseConfiguration()
conf.setProperty("storage.backend","cassandra")
//conf.setProperty("storage.hostname","localhost")
conf.setProperty("storage.hostname","10.172.137.237,10.174.89.108,10.172.111.134")
conf.setProperty("storage.batch-loading","true")
g = TitanFactory.open(conf)
bg = new BatchGraph(g, VertexIDType.NUMBER, 2000);
bg.setVertexIdKey("uid")

def getOrCreate(BatchGraph bg, long id) {
  v = bg.getVertex(id)
  if(null == v) v = bg.addVertex(id)
  return v;
}


counter = 0;
new File('/graph/friendster/').eachFile { File f ->
  println "Processing ${f}"
  f.eachLine { String line ->
   parts = line.trim().split(':')
   if(parts.length == 1) {
    v = getOrCreate(bg, Long.valueOf(parts[0].trim()))
    v.setProperty('private',false)
   } else if(!parts[1].equals('notfound')) {
     v = getOrCreate(bg, Long.valueOf(parts[0].trim()))
     if(!parts[1].equals('private')) {
       v.setProperty('private',false)
       parts[1].trim().split(',').each {
         u = getOrCreate(bg, Long.valueOf(it))
         bg.addEdge(null, v, u, 'friend')
       }
     } else {
       v.setProperty('private',true)     
     }
   }
   if(++counter % 1000 == 0) println("Processed ${counter} vertices.")
  }
}