Merge pull request apache#189 from palantir/rk/upstream

Merge from upstream

R/pkg/NAMESPACE

@@ -84,6 +84,7 @@ exportClasses("SparkDataFrame")
 exportMethods("arrange",
               "as.data.frame",
               "attach",
+              "broadcast",
               "cache",
               "checkpoint",
               "coalesce",

R/pkg/R/DataFrame.R

@@ -3769,3 +3769,32 @@ setMethod("alias",
             sdf <- callJMethod(object@sdf, "alias", data)
             dataFrame(sdf)
           })
+
+#' broadcast
+#'
+#' Return a new SparkDataFrame marked as small enough for use in broadcast joins.
+#'
+#' Equivalent to \code{hint(x, "broadcast")}.
+#'
+#' @param x a SparkDataFrame.
+#' @return a SparkDataFrame.
+#'
+#' @aliases broadcast,SparkDataFrame-method
+#' @family SparkDataFrame functions
+#' @rdname broadcast
+#' @name broadcast
+#' @export
+#' @examples
+#' \dontrun{
+#' df <- createDataFrame(mtcars)
+#' avg_mpg <- mean(groupBy(createDataFrame(mtcars), "cyl"), "mpg")
+#'
+#' head(join(df, broadcast(avg_mpg), df$cyl == avg_mpg$cyl))
+#' }
+#' @note broadcast since 2.3.0
+setMethod("broadcast",
+          signature(x = "SparkDataFrame"),
+          function(x) {
+            sdf <- callJStatic("org.apache.spark.sql.functions", "broadcast", x@sdf)
+            dataFrame(sdf)
+          })

R/pkg/R/context.R

@@ -258,15 +258,15 @@ includePackage <- function(sc, pkg) {
 #'
 #' # Large Matrix object that we want to broadcast
 #' randomMat <- matrix(nrow=100, ncol=10, data=rnorm(1000))
-#' randomMatBr <- broadcast(sc, randomMat)
+#' randomMatBr <- broadcastRDD(sc, randomMat)
 #'
 #' # Use the broadcast variable inside the function
 #' useBroadcast <- function(x) {
 #'   sum(value(randomMatBr) * x)
 #' }
 #' sumRDD <- lapply(rdd, useBroadcast)
 #'}
-broadcast <- function(sc, object) {
+broadcastRDD <- function(sc, object) {
   objName <- as.character(substitute(object))
   serializedObj <- serialize(object, connection = NULL)