Merge pull request apache#104 from sun-rui/add_keys_values

Add keys() and values() for the RDD class.

pkg/NAMESPACE

@@ -14,6 +14,7 @@ exportMethods(
               "filter",
               "flatMap",
               "groupByKey",
+              "keys",
               "length",
               "lapply",
               "lapplyPartition",
@@ -33,7 +34,8 @@ exportMethods(
               "takeSample",
               "unionRDD",
               "unpersist",
-              "value"
+              "value",
+              "values"
              )
 
 # S3 methods exported

pkg/R/RDD.R

@@ -856,6 +856,54 @@ setMethod("takeSample", signature(rdd = "RDD", withReplacement = "logical",
             sample(samples)[1:total]
           })
 
+#' Return an RDD with the keys of each tuple.
+#'
+#' @param rdd The RDD from which the keys of each tuple is returned.
+#' @rdname keys
+#' @export
+#' @examples
+#'\dontrun{
+#' sc <- sparkR.init()
+#' rdd <- parallelize(sc, list(list(1, 2), list(3, 4)))
+#' collect(keys(rdd)) # list(1, 3)
+#'}
+setGeneric("keys", function(rdd) { standardGeneric("keys") })
+
+#' @rdname keys
+#' @aliases keys,RDD
+setMethod("keys",
+          signature(rdd = "RDD"),
+          function(rdd) {
+            func <- function(x) {
+              x[[1]]
+            }
+            lapply(rdd, func)
+          })
+
+#' Return an RDD with the values of each tuple.
+#'
+#' @param rdd The RDD from which the values of each tuple is returned.
+#' @rdname values
+#' @export
+#' @examples
+#'\dontrun{
+#' sc <- sparkR.init()
+#' rdd <- parallelize(sc, list(list(1, 2), list(3, 4)))
+#' collect(values(rdd)) # list(2, 4)
+#'}
+setGeneric("values", function(rdd) { standardGeneric("values") })
+
+#' @rdname values
+#' @aliases values,RDD
+setMethod("values",
+          signature(rdd = "RDD"),
+          function(rdd) {
+            func <- function(x) {
+              x[[2]]
+            }
+            lapply(rdd, func)
+          })
+
 #' Applies a function to all values of the elements, without modifying the keys.
 #'
 #' The same as `mapValues()' in Spark.

pkg/inst/tests/test_rdd.R

@@ -203,3 +203,16 @@ test_that("minimum() on RDDs", {
   min <- minimum(rdd)
   expect_equal(min, 1)
 })
+
+test_that("keys() on RDDs", {
+  keys <- keys(intRdd)
+  actual <- collect(keys)
+  expect_equal(actual, lapply(intPairs, function(x) { x[[1]] }))
+})
+
+test_that("values() on RDDs", {
+  values <- values(intRdd)
+  actual <- collect(values)
+  expect_equal(actual, lapply(intPairs, function(x) { x[[2]] }))
+})
+

pkg/man/keys.Rd

@@ -0,0 +1,26 @@
+% Generated by roxygen2 (4.0.2): do not edit by hand
+\docType{methods}
+\name{keys}
+\alias{keys}
+\alias{keys,RDD}
+\alias{keys,RDD-method}
+\title{Return an RDD with the keys of each tuple.}
+\usage{
+keys(rdd)
+
+\S4method{keys}{RDD}(rdd)
+}
+\arguments{
+\item{rdd}{The RDD from which the keys of each tuple is returned.}
+}
+\description{
+Return an RDD with the keys of each tuple.
+}
+\examples{
+\dontrun{
+sc <- sparkR.init()
+rdd <- parallelize(sc, list(list(1, 2), list(3, 4)))
+collect(keys(rdd)) # list(1, 3)
+}
+}
+

pkg/man/values.Rd

@@ -0,0 +1,26 @@
+% Generated by roxygen2 (4.0.2): do not edit by hand
+\docType{methods}
+\name{values}
+\alias{values}
+\alias{values,RDD}
+\alias{values,RDD-method}
+\title{Return an RDD with the values of each tuple.}
+\usage{
+values(rdd)
+
+\S4method{values}{RDD}(rdd)
+}
+\arguments{
+\item{rdd}{The RDD from which the values of each tuple is returned.}
+}
+\description{
+Return an RDD with the values of each tuple.
+}
+\examples{
+\dontrun{
+sc <- sparkR.init()
+rdd <- parallelize(sc, list(list(1, 2), list(3, 4)))
+collect(values(rdd)) # list(2, 4)
+}
+}
+