在 GraphX 中从 RDD[Long， Long， ED] 构造一个 RDD[Edge[ED]]

Question

我现在正在尝试从RDD[(Long,Long,String)]构造一个RDD[Edge[String]]我看过Edge的src代码，它的参数为Edge(VertexId,VertexId,ED)，所以我想如果我有RDD[(Long,Long,String)]可以把它转换成RDD[Edge[String]]

我的代码是

val c:RDD[(String, String)] = something
val s:RDD[(String, String)] = something
val edgeRDD: RDD[Edge[String]] = c
.join(s).map({
case (num: String, (src: String, dst: String)) 
=> (src.toLong, dst.toLong, expRelation)
}
) 

但最后我得到了一个type mismatch错误

[error]  found   : (Long, Long, String)
[error]  required: org.apache.spark.graphx.Edge[String]
[error]                     => (src.toLong, dst.toLong, expRelation)
[error]                        ^

所以我想知道是否有办法直接从RDD[(Long,Long,ED)]构建RDD[Edge[ED]]

Answer 1

您的函数未返回 Edge[String]。将返回类型更改为

Edge(src.toLong, dst.toLong, expRelation)

法典：

val c:RDD[(String, String)] = something
val s:RDD[(String, String)] = something
val edgeRDD: RDD[Edge[String]] = c
.join(s).map({
case (num: String, (src: String, dst: String)) 
=> Edge(src.toLong, dst.toLong, expRelation)
}
)