[SPARK-20726][SPARKR] wrapper for SQL broadcast

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)
 

R/pkg/R/generics.R

@@ -799,6 +799,10 @@ setGeneric("write.df", function(df, path = NULL, ...) { standardGeneric("write.d
 #' @export
 setGeneric("randomSplit", function(x, weights, seed) { standardGeneric("randomSplit") })
 
+#' @rdname broadcast
+#' @export
+setGeneric("broadcast", function(x) { standardGeneric("broadcast") })
+
 ###################### Column Methods ##########################
 
 #' @rdname columnfunctions

R/pkg/inst/tests/testthat/test_broadcast.R

@@ -29,7 +29,7 @@ test_that("using broadcast variable", {
   skip_on_cran()
 
   randomMat <- matrix(nrow = 10, ncol = 10, data = rnorm(100))
-  randomMatBr <- broadcast(sc, randomMat)
+  randomMatBr <- broadcastRDD(sc, randomMat)
 
   useBroadcast <- function(x) {
     sum(SparkR:::value(randomMatBr) * x)

R/pkg/inst/tests/testthat/test_sparkSQL.R

@@ -2216,6 +2216,11 @@ test_that("join(), crossJoin() and merge() on a DataFrame", {
     explain(join(df1, hint(df2, "broadcast"), df1$id == df2$id))
   )
   expect_true(any(grepl("BroadcastHashJoin", execution_plan_hint)))
+
+  execution_plan_broadcast <- capture.output(
+    explain(join(df1, broadcast(df2), df1$id == df2$id))
+  )
+  expect_true(any(grepl("BroadcastHashJoin", execution_plan_broadcast)))
 })
 
 test_that("toJSON() on DataFrame", {

R/pkg/inst/tests/testthat/test_utils.R

@@ -136,7 +136,7 @@ test_that("cleanClosure on R functions", {
 
   # Test for broadcast variables.
   a <- matrix(nrow = 10, ncol = 10, data = rnorm(100))
-  aBroadcast <- broadcast(sc, a)
+  aBroadcast <- broadcastRDD(sc, a)
   normMultiply <- function(x) { norm(aBroadcast$value) * x }
   newnormMultiply <- SparkR:::cleanClosure(normMultiply)
   env <- environment(newnormMultiply)