spark-reviews mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From davies <...@git.apache.org>
Subject [GitHub] spark pull request: [SPARK-10979] [SparkR] Sparkrmerge: Add merge ...
Date Wed, 21 Oct 2015 23:37:14 GMT
Github user davies commented on a diff in the pull request:

    https://github.com/apache/spark/pull/9012#discussion_r42696855
  
    --- Diff: R/pkg/R/DataFrame.R ---
    @@ -1457,15 +1457,142 @@ setMethod("join",
                 dataFrame(sdf)
               })
     
    -#' @rdname merge
    +#'
     #' @name merge
     #' @aliases join
    +#' @title Merges two data frames
    +#' @param x the first data frame to be joined
    +#' @param y the second data frame to be joined
    +#' @param by a character vector specifying the join columns. If by is not
    +#'   specified, the common column names in \code{x} and \code{y} will be used.
    +#' @param by.x a character vector specifying the joining columns for x.
    +#' @param by.y a character vector specifying the joining columns for y.
    +#' @param all.x a boolean value indicating whether all the rows in x should
    +#'              be including in the join
    +#' @param all.y a boolean value indicating whether all the rows in y should
    +#'              be including in the join
    +#' @param sort a logical argument indicating whether the resulting columns should be
sorted
    +#' @details  If all.x and all.y are set to FALSE, a natural join will be returned. If
    +#'   all.x is set to TRUE and all.y is set to FALSE, a left outer join will
    +#'   be returned. If all.x is set to FALSE and all.y is set to TRUE, a right
    +#'   outer join will be returned. If all.x and all.y are set to TRUE, a full
    +#'   outer join will be returned.
    +#' @rdname merge
    +#' @export
    +#' @examples
    +#'\dontrun{
    +#' sc <- sparkR.init()
    +#' sqlContext <- sparkRSQL.init(sc)
    +#' df1 <- jsonFile(sqlContext, path)
    +#' df2 <- jsonFile(sqlContext, path2)
    +#' merge(df1, df2) # Performs a Cartesian
    +#' merge(df1, df2, by = "col1") # Performs an inner join based on expression
    +#' merge(df1, df2, by.x = "col1", by.y = "col2", all.y = TRUE)
    +#' merge(df1, df2, by.x = "col1", by.y = "col2", all.x = TRUE)
    +#' merge(df1, df2, by.x = "col1", by.y = "col2", all.x = TRUE, all.y = TRUE)
    +#' merge(df1, df2, by.x = "col1", by.y = "col2", all = TRUE, sort = FALSE)
    +#' merge(df1, df2, by = "col1", all = TRUE, suffixes = c("-X", "-Y"))
    +#' }
     setMethod("merge",
               signature(x = "DataFrame", y = "DataFrame"),
    -          function(x, y, joinExpr = NULL, joinType = NULL, ...) {
    -            join(x, y, joinExpr, joinType)
    -          })
    +          function(x, y, by = intersect(names(x), names(y)), by.x = by, by.y = by,
    +                   all = FALSE, all.x = all, all.y = all,
    +                   sort = TRUE, suffixes = c("_x","_y"), ... ) {
    +
    +            if (length(suffixes) != 2) {
    +              stop("suffixes must have length 2")
    +            }
    +
    +            # join type is identified based on the values of all, all.x and all.y
    +            # default join type is inner, according to R it should be natural but since
it
    +            # is not supported in spark inner join is used
    +            joinType <- "inner"
    +            if (all || (all.x && all.y)) {
    +              joinType <- "outer"
    +            } else if (all.x) {
    +              joinType <- "left_outer"
    +            } else if (all.y) {
    +              joinType <- "right_outer"
    +            }
     
    +            # join expression is based on by.x, by.y if both by.x and by.y are not missing
    +            # or on by, if by.x or by.y are missing or have different lengths
    +            if (length(by.x) > 0 && length(by.x) == length(by.y)) {
    +              joinX <- by.x
    +              joinY <- by.y
    +            } else if (length(by) > 0) {
    +              joinX <- by
    +              joinY <- by
    +            } else {
    +              # if by or both by.x and by.y have length 0, use Cartesian Product
    +              joinRes <- join(x, y)
    +              return (joinRes)
    +            }
    +
    +            # sets alias for making colnames unique in dataframes 'x' and 'y'
    +            colsX <- generateAliasesForIntersectedCols(x, by, suffixes[1])
    --- End diff --
    
    This is cool! maybe we should also do this in Scala and Python. cc @rxin 


---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at infrastructure@apache.org or file a JIRA ticket
with INFRA.
---

---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscribe@spark.apache.org
For additional commands, e-mail: reviews-help@spark.apache.org


Mime
View raw message