[SPARK-8807] [SPARKR] Add between operator in SparkR

JIRA: https://issues.apache.org/jira/browse/SPARK-8807

Add between operator in SparkR.

Author: Liang-Chi Hsieh <[email protected]>

Closes apache#7356 from viirya/add_r_between and squashes the following commits:

7f51b44 [Liang-Chi Hsieh] Add test for non-numeric column.
c6a25c5 [Liang-Chi Hsieh] Add between function.

Author: viirya

R/pkg/NAMESPACE

@@ -77,6 +77,7 @@ exportMethods("abs",
               "atan",
               "atan2",
               "avg",
+              "between",
               "cast",
               "cbrt",
               "ceiling",

R/pkg/R/column.R

@@ -187,6 +187,23 @@ setMethod("substr", signature(x = "Column"),
             column(jc)
           })
 
+#' between
+#'
+#' Test if the column is between the lower bound and upper bound, inclusive.
+#'
+#' @rdname column
+#'
+#' @param bounds lower and upper bounds
+setMethod("between", signature(x = "Column"),
+          function(x, bounds) {
+            if (is.vector(bounds) && length(bounds) == 2) {
+              jc <- callJMethod(x@jc, "between", bounds[1], bounds[2])
+              column(jc)
+            } else {
+              stop("bounds should be a vector of lower and upper bounds")
+            }
+          })
+
 #' Casts the column to a different data type.
 #'
 #' @rdname column

R/pkg/R/generics.R

@@ -567,6 +567,10 @@ setGeneric("asc", function(x) { standardGeneric("asc") })
 #' @export
 setGeneric("avg", function(x, ...) { standardGeneric("avg") })
 
+#' @rdname column
+#' @export
+setGeneric("between", function(x, bounds) { standardGeneric("between") })
+
 #' @rdname column
 #' @export
 setGeneric("cast", function(x, dataType) { standardGeneric("cast") })

R/pkg/inst/tests/test_sparkSQL.R

@@ -638,6 +638,18 @@ test_that("column functions", {
   c7 <- floor(c) + log(c) + log10(c) + log1p(c) + rint(c)
   c8 <- sign(c) + sin(c) + sinh(c) + tan(c) + tanh(c)
   c9 <- toDegrees(c) + toRadians(c)
+
+  df <- jsonFile(sqlContext, jsonPath)
+  df2 <- select(df, between(df$age, c(20, 30)), between(df$age, c(10, 20)))
+  expect_equal(collect(df2)[[2, 1]], TRUE)
+  expect_equal(collect(df2)[[2, 2]], FALSE)
+  expect_equal(collect(df2)[[3, 1]], FALSE)
+  expect_equal(collect(df2)[[3, 2]], TRUE)
+
+  df3 <- select(df, between(df$name, c("Apache", "Spark")))
+  expect_equal(collect(df3)[[1, 1]], TRUE)
+  expect_equal(collect(df3)[[2, 1]], FALSE)
+  expect_equal(collect(df3)[[3, 1]], TRUE)
 })
 
 test_that("column binary mathfunctions", {