merge 'master' and resolve conflicts

Author: WeichenXu123

R/pkg/NAMESPACE

@@ -232,6 +232,7 @@ exportMethods("%<=>%",
               "date_sub",
               "datediff",
               "dayofmonth",
+              "dayofweek",
               "dayofyear",
               "decode",
               "dense_rank",

R/pkg/R/DataFrame.R

@@ -58,14 +58,23 @@ setMethod("initialize", "SparkDataFrame", function(.Object, sdf, isCached) {
 #' Set options/mode and then return the write object
 #' @noRd
 setWriteOptions <- function(write, path = NULL, mode = "error", ...) {
-    options <- varargsToStrEnv(...)
-    if (!is.null(path)) {
-      options[["path"]] <- path
-    }
-    jmode <- convertToJSaveMode(mode)
-    write <- callJMethod(write, "mode", jmode)
-    write <- callJMethod(write, "options", options)
-    write
+  options <- varargsToStrEnv(...)
+  if (!is.null(path)) {
+    options[["path"]] <- path
+  }
+  write <- setWriteMode(write, mode)
+  write <- callJMethod(write, "options", options)
+  write
+}
+
+#' Set mode and then return the write object
+#' @noRd
+setWriteMode <- function(write, mode) {
+  if (!is.character(mode)) {
+    stop("mode should be character or omitted. It is 'error' by default.")
+  }
+  write <- handledCallJMethod(write, "mode", mode)
+  write
 }
 
 #' @export
@@ -556,9 +565,8 @@ setMethod("registerTempTable",
 setMethod("insertInto",
           signature(x = "SparkDataFrame", tableName = "character"),
           function(x, tableName, overwrite = FALSE) {
-            jmode <- convertToJSaveMode(ifelse(overwrite, "overwrite", "append"))
             write <- callJMethod(x@sdf, "write")
-            write <- callJMethod(write, "mode", jmode)
+            write <- setWriteMode(write, ifelse(overwrite, "overwrite", "append"))
             invisible(callJMethod(write, "insertInto", tableName))
           })
 
@@ -810,7 +818,8 @@ setMethod("toJSON",
 #'
 #' @param x A SparkDataFrame
 #' @param path The directory where the file is saved
-#' @param mode one of 'append', 'overwrite', 'error', 'ignore' save mode (it is 'error' by default)
+#' @param mode one of 'append', 'overwrite', 'error', 'errorifexists', 'ignore'
+#'             save mode (it is 'error' by default)
 #' @param ... additional argument(s) passed to the method.
 #'
 #' @family SparkDataFrame functions
@@ -841,7 +850,8 @@ setMethod("write.json",
 #'
 #' @param x A SparkDataFrame
 #' @param path The directory where the file is saved
-#' @param mode one of 'append', 'overwrite', 'error', 'ignore' save mode (it is 'error' by default)
+#' @param mode one of 'append', 'overwrite', 'error', 'errorifexists', 'ignore'
+#'             save mode (it is 'error' by default)
 #' @param ... additional argument(s) passed to the method.
 #'
 #' @family SparkDataFrame functions
@@ -872,7 +882,8 @@ setMethod("write.orc",
 #'
 #' @param x A SparkDataFrame
 #' @param path The directory where the file is saved
-#' @param mode one of 'append', 'overwrite', 'error', 'ignore' save mode (it is 'error' by default)
+#' @param mode one of 'append', 'overwrite', 'error', 'errorifexists', 'ignore'
+#'             save mode (it is 'error' by default)
 #' @param ... additional argument(s) passed to the method.
 #'
 #' @family SparkDataFrame functions
@@ -917,7 +928,8 @@ setMethod("saveAsParquetFile",
 #'
 #' @param x A SparkDataFrame
 #' @param path The directory where the file is saved
-#' @param mode one of 'append', 'overwrite', 'error', 'ignore' save mode (it is 'error' by default)
+#' @param mode one of 'append', 'overwrite', 'error', 'errorifexists', 'ignore'
+#'             save mode (it is 'error' by default)
 #' @param ... additional argument(s) passed to the method.
 #'
 #' @family SparkDataFrame functions
@@ -2871,18 +2883,19 @@ setMethod("except",
 #' Additionally, mode is used to specify the behavior of the save operation when data already
 #' exists in the data source. There are four modes:
 #' \itemize{
-#'   \item append: Contents of this SparkDataFrame are expected to be appended to existing data.
-#'   \item overwrite: Existing data is expected to be overwritten by the contents of this
+#'   \item 'append': Contents of this SparkDataFrame are expected to be appended to existing data.
+#'   \item 'overwrite': Existing data is expected to be overwritten by the contents of this
 #'         SparkDataFrame.
-#'   \item error: An exception is expected to be thrown.
-#'   \item ignore: The save operation is expected to not save the contents of the SparkDataFrame
+#'   \item 'error' or 'errorifexists': An exception is expected to be thrown.
+#'   \item 'ignore': The save operation is expected to not save the contents of the SparkDataFrame
 #'         and to not change the existing data.
 #' }
 #'
 #' @param df a SparkDataFrame.
 #' @param path a name for the table.
 #' @param source a name for external data source.
-#' @param mode one of 'append', 'overwrite', 'error', 'ignore' save mode (it is 'error' by default)
+#' @param mode one of 'append', 'overwrite', 'error', 'errorifexists', 'ignore'
+#'             save mode (it is 'error' by default)
 #' @param ... additional argument(s) passed to the method.
 #'
 #' @family SparkDataFrame functions
@@ -2940,17 +2953,18 @@ setMethod("saveDF",
 #'
 #' Additionally, mode is used to specify the behavior of the save operation when
 #' data already exists in the data source. There are four modes: \cr
-#'  append: Contents of this SparkDataFrame are expected to be appended to existing data. \cr
-#'  overwrite: Existing data is expected to be overwritten by the contents of this
+#'  'append': Contents of this SparkDataFrame are expected to be appended to existing data. \cr
+#'  'overwrite': Existing data is expected to be overwritten by the contents of this
 #'     SparkDataFrame. \cr
-#'  error: An exception is expected to be thrown. \cr
-#'  ignore: The save operation is expected to not save the contents of the SparkDataFrame
+#'  'error' or 'errorifexists': An exception is expected to be thrown. \cr
+#'  'ignore': The save operation is expected to not save the contents of the SparkDataFrame
 #'     and to not change the existing data. \cr
 #'
 #' @param df a SparkDataFrame.
 #' @param tableName a name for the table.
 #' @param source a name for external data source.
-#' @param mode one of 'append', 'overwrite', 'error', 'ignore' save mode (it is 'error' by default).
+#' @param mode one of 'append', 'overwrite', 'error', 'errorifexists', 'ignore'
+#'             save mode (it is 'error' by default)
 #' @param ... additional option(s) passed to the method.
 #'
 #' @family SparkDataFrame functions
@@ -2972,12 +2986,11 @@ setMethod("saveAsTable",
             if (is.null(source)) {
               source <- getDefaultSqlSource()
             }
-            jmode <- convertToJSaveMode(mode)
             options <- varargsToStrEnv(...)
 
             write <- callJMethod(df@sdf, "write")
             write <- callJMethod(write, "format", source)
-            write <- callJMethod(write, "mode", jmode)
+            write <- setWriteMode(write, mode)
             write <- callJMethod(write, "options", options)
             invisible(callJMethod(write, "saveAsTable", tableName))
           })
@@ -3236,7 +3249,7 @@ setMethod("as.data.frame",
 #'
 #' @family SparkDataFrame functions
 #' @rdname attach
-#' @aliases attach,SparkDataFrame-method
+#' @aliases attach attach,SparkDataFrame-method
 #' @param what (SparkDataFrame) The SparkDataFrame to attach
 #' @param pos (integer) Specify position in search() where to attach.
 #' @param name (character) Name to use for the attached SparkDataFrame. Names
@@ -3252,9 +3265,12 @@ setMethod("as.data.frame",
 #' @note attach since 1.6.0
 setMethod("attach",
           signature(what = "SparkDataFrame"),
-          function(what, pos = 2, name = deparse(substitute(what)), warn.conflicts = TRUE) {
-            newEnv <- assignNewEnv(what)
-            attach(newEnv, pos = pos, name = name, warn.conflicts = warn.conflicts)
+          function(what, pos = 2L, name = deparse(substitute(what), backtick = FALSE),
+                   warn.conflicts = TRUE) {
+            args <- as.list(environment()) # capture all parameters - this must be the first line
+            newEnv <- assignNewEnv(args$what)
+            args$what <- newEnv
+            do.call(attach, args)
           })
 
 #' Evaluate a R expression in an environment constructed from a SparkDataFrame
@@ -3541,18 +3557,19 @@ setMethod("histogram",
 #' Also, mode is used to specify the behavior of the save operation when
 #' data already exists in the data source. There are four modes:
 #' \itemize{
-#'   \item append: Contents of this SparkDataFrame are expected to be appended to existing data.
-#'   \item overwrite: Existing data is expected to be overwritten by the contents of this
+#'   \item 'append': Contents of this SparkDataFrame are expected to be appended to existing data.
+#'   \item 'overwrite': Existing data is expected to be overwritten by the contents of this
 #'         SparkDataFrame.
-#'   \item error: An exception is expected to be thrown.
-#'   \item ignore: The save operation is expected to not save the contents of the SparkDataFrame
+#'   \item 'error' or 'errorifexists': An exception is expected to be thrown.
+#'   \item 'ignore': The save operation is expected to not save the contents of the SparkDataFrame
 #'         and to not change the existing data.
 #' }
 #'
 #' @param x a SparkDataFrame.
 #' @param url JDBC database url of the form \code{jdbc:subprotocol:subname}.
 #' @param tableName yhe name of the table in the external database.
-#' @param mode one of 'append', 'overwrite', 'error', 'ignore' save mode (it is 'error' by default).
+#' @param mode one of 'append', 'overwrite', 'error', 'errorifexists', 'ignore'
+#'             save mode (it is 'error' by default)
 #' @param ... additional JDBC database connection properties.
 #' @family SparkDataFrame functions
 #' @rdname write.jdbc
@@ -3569,10 +3586,9 @@ setMethod("histogram",
 setMethod("write.jdbc",
           signature(x = "SparkDataFrame", url = "character", tableName = "character"),
           function(x, url, tableName, mode = "error", ...) {
-            jmode <- convertToJSaveMode(mode)
             jprops <- varargsToJProperties(...)
             write <- callJMethod(x@sdf, "write")
-            write <- callJMethod(write, "mode", jmode)
+            write <- setWriteMode(write, mode)
             invisible(handledCallJMethod(write, "jdbc", url, tableName, jprops))
           })
 

R/pkg/R/functions.R

@@ -696,7 +696,7 @@ setMethod("hash",
 #'
 #' \dontrun{
 #' head(select(df, df$time, year(df$time), quarter(df$time), month(df$time),
-#'            dayofmonth(df$time), dayofyear(df$time), weekofyear(df$time)))
+#'             dayofmonth(df$time), dayofweek(df$time), dayofyear(df$time), weekofyear(df$time)))
 #' head(agg(groupBy(df, year(df$time)), count(df$y), avg(df$y)))
 #' head(agg(groupBy(df, month(df$time)), avg(df$y)))}
 #' @note dayofmonth since 1.5.0
@@ -707,6 +707,21 @@ setMethod("dayofmonth",
             column(jc)
           })
 
+#' @details
+#' \code{dayofweek}: Extracts the day of the week as an integer from a
+#' given date/timestamp/string.
+#'
+#' @rdname column_datetime_functions
+#' @aliases dayofweek dayofweek,Column-method
+#' @export
+#' @note dayofweek since 2.3.0
+setMethod("dayofweek",
+          signature(x = "Column"),
+          function(x) {
+            jc <- callJStatic("org.apache.spark.sql.functions", "dayofweek", x@jc)
+            column(jc)
+          })
+
 #' @details
 #' \code{dayofyear}: Extracts the day of the year as an integer from a
 #' given date/timestamp/string.

R/pkg/R/generics.R

@@ -409,7 +409,8 @@ setGeneric("as.data.frame",
              standardGeneric("as.data.frame")
            })
 
-#' @rdname attach
+# Do not document the generic because of signature changes across R versions
+#' @noRd
 #' @export
 setGeneric("attach")
 
@@ -1047,6 +1048,11 @@ setGeneric("date_sub", function(y, x) { standardGeneric("date_sub") })
 #' @name NULL
 setGeneric("dayofmonth", function(x) { standardGeneric("dayofmonth") })
 
+#' @rdname column_datetime_functions
+#' @export
+#' @name NULL
+setGeneric("dayofweek", function(x) { standardGeneric("dayofweek") })
+
 #' @rdname column_datetime_functions
 #' @export
 #' @name NULL
@@ -1569,12 +1575,9 @@ setGeneric("year", function(x) { standardGeneric("year") })
 #' @export
 setGeneric("fitted")
 
-#' @param x,y For \code{glm}: logical values indicating whether the response vector
-#'          and model matrix used in the fitting process should be returned as
-#'          components of the returned value.
-#' @inheritParams stats::glm
-#' @rdname glm
+# Do not carry stats::glm usage and param here, and do not document the generic
 #' @export
+#' @noRd
 setGeneric("glm")
 
 #' @param object a fitted ML model object.

R/pkg/R/install.R

@@ -152,6 +152,11 @@ install.spark <- function(hadoopVersion = "2.7", mirrorUrl = NULL,
                      })
   if (!tarExists || overwrite || !success) {
     unlink(packageLocalPath)
+    if (success) {
+      # if tar file was not there before (or it was, but we are told to overwrite it),
+      # and untar is successful - set a flag that we have downloaded (and untar) Spark package.
+      assign(".sparkDownloaded", TRUE, envir = .sparkREnv)
+    }
   }
   if (!success) stop("Extract archive failed.")
   message("DONE.")
@@ -266,6 +271,7 @@ hadoopVersionName <- function(hadoopVersion) {
 
 # The implementation refers to appdirs package: https://pypi.python.org/pypi/appdirs and
 # adapt to Spark context
+# see also sparkCacheRelPathLength()
 sparkCachePath <- function() {
   if (is_windows()) {
     winAppPath <- Sys.getenv("LOCALAPPDATA", unset = NA)
@@ -282,7 +288,7 @@ sparkCachePath <- function() {
     }
   } else if (.Platform$OS.type == "unix") {
     if (Sys.info()["sysname"] == "Darwin") {
-      path <- file.path(Sys.getenv("HOME"), "Library/Caches", "spark")
+      path <- file.path(Sys.getenv("HOME"), "Library", "Caches", "spark")
     } else {
       path <- file.path(
         Sys.getenv("XDG_CACHE_HOME", file.path(Sys.getenv("HOME"), ".cache")), "spark")
@@ -293,6 +299,16 @@ sparkCachePath <- function() {
   normalizePath(path, mustWork = FALSE)
 }
 
+# Length of the Spark cache specific relative path segments for each platform
+# eg. "Apache\Spark\Cache" is 3 in Windows, or "spark" is 1 in unix
+# Must match sparkCachePath() exactly.
+sparkCacheRelPathLength <- function() {
+  if (is_windows()) {
+    3
+  } else {
+    1
+  }
+}
 
 installInstruction <- function(mode) {
   if (mode == "remote") {
@@ -310,3 +326,22 @@ installInstruction <- function(mode) {
     stop(paste0("No instruction found for ", mode, " mode."))
   }
 }
+
+uninstallDownloadedSpark <- function() {
+  # clean up if Spark was downloaded
+  sparkDownloaded <- getOne(".sparkDownloaded",
+                            envir = .sparkREnv,
+                            inherits = TRUE,
+                            ifnotfound = FALSE)
+  sparkDownloadedDir <- Sys.getenv("SPARK_HOME")
+  if (sparkDownloaded && nchar(sparkDownloadedDir) > 0) {
+    unlink(sparkDownloadedDir, recursive = TRUE, force = TRUE)
+
+    dirs <- traverseParentDirs(sparkCachePath(), sparkCacheRelPathLength())
+    lapply(dirs, function(d) {
+      if (length(list.files(d, all.files = TRUE, include.dirs = TRUE, no.. = TRUE)) == 0) {
+        unlink(d, recursive = TRUE, force = TRUE)
+      }
+    })
+  }
+}

R/pkg/R/mllib_regression.R

@@ -210,6 +210,7 @@ setMethod("spark.glm", signature(data = "SparkDataFrame", formula = "formula"),
 #'                  1.0.
 #' @return \code{glm} returns a fitted generalized linear model.
 #' @rdname glm
+#' @aliases glm
 #' @export
 #' @examples
 #' \dontrun{

R/pkg/R/sparkR.R

@@ -420,6 +420,18 @@ sparkR.session <- function(
                                 enableHiveSupport)
     assign(".sparkRsession", sparkSession, envir = .sparkREnv)
   }
+
+  # Check if version number of SparkSession matches version number of SparkR package
+  jvmVersion <- callJMethod(sparkSession, "version")
+  # Remove -SNAPSHOT from jvm versions
+  jvmVersionStrip <- gsub("-SNAPSHOT", "", jvmVersion)
+  rPackageVersion <- paste0(packageVersion("SparkR"))
+
+  if (jvmVersionStrip != rPackageVersion) {
+    warning(paste("Version mismatch between Spark JVM and SparkR package. JVM version was",
+                  jvmVersion, ", while R package version was", rPackageVersion))
+  }
+
   sparkSession
 }