sparklyr pivot数据帧

Question

目前sparklyr (0.5.1)还没有针对Spark的pivot实现。

所以，我想知道如何使用invoke函数来实现这一点。

到目前为止，我已经到了这里。

iris_tbl <- copy_to(sc, iris)
spark_dataframe(iris_tbl) %>% 
  invoke("groupBy", "Species", list()) %>% 
  invoke("pivot", "Sepal.Width", list()) %>% 
  invoke(  ... ) # <-  how to create aggregate expression? 

并纠结于如何编写聚合表达式？

Answer 1

这将带你走完一半的路：

library(stringi)

sc <- spark_connect("local[*]")
df <- data.frame(A = c("a", "b", "c"), B = c(1, 3, 5), C = c(2, 4, 6))
sdf <- copy_to(sc, df, overwrite =TRUE)

助手函数：

#' Given name, return corresponding SQL function
sqlf <- function(f) function(x, ...) {
  invoke_static(sc, "org.apache.spark.sql.functions", f, x, ...)
}

熔化函数：

#' @param df tbl_spark
#' @param sc spark_connection
#' @param id_vars id columns
#'
melt <- function(df, sc, id_vars, value_vars = NULL, 
    var_name = "key", value_name = "value") {
  # Alias for the output view
  alias <- paste(deparse(substitute(df)), stri_rand_strings(1, 10), sep = "_")
  # Get session and JVM object
  spark <- spark_session(sc)
  jdf <- spark_dataframe(df)

  # Convert characters to JVM Columns
  j_id_vars <- lapply(id_vars, sqlf("col"))

  # Combine columns into array<struct<key,value>> and explode
  exploded <- sqlf("explode")(sqlf("array")(lapply(value_vars, function(x) {
    key <- sqlf("lit")(x) %>% invoke("alias", var_name)
    value <- sqlf("col")(x) %>% invoke("alias", value_name)
    sqlf("struct")(list(key, value))
  })))

  # expand struct<..., struct<key, value>> into struct<..., key, value>
  exprs <- lapply(
    c(id_vars, paste("col", c(var_name, value_name), sep = ".")),
    sqlf("col"))

  # Explode and register as temp table
  jdf %>% 
    invoke("withColumn", "col", exploded) %>% 
    invoke("select", exprs) %>% 
    invoke("createOrReplaceTempView", alias)

  dplyr::tbl(sc, alias)
}

示例用法：

melt(sdf, sc, "A", c("B", "C"))

## Source:   query [6 x 3]
## Database: spark connection master=local[*] app=sparklyr local=TRUE
## 
## # A tibble: 6 x 3
##       A   key value
##   <chr> <chr> <dbl>
## 1     a     B     1
## 2     a     C     2
## 3     b     B     3
## 4     b     C     4
## 5     c     B     5
## 6     c     C     6