使用dplyr和lazyeval与'...'

Question

简化编辑：

归结为：

df = data.frame(a = 1:10)

#example function that takes optional arguments
mymean <- function(x, w = 1, s = 0) { s + mean(w * x) }

summarize_(df, m = "mean(a)")
#>     m
#> 1 5.5

summarize_(df, m = "mymean(a)")
#> Error: could not find function "mymean"

根据`vignette（“nse”），在使用非标准汇总函数时，必须给出摘要_公式语法。

最终，我希望能够将summarize_包装在这样的函数中：

my_summary <- function(df, x, ...) {
  summarize_(df, 
             m = "mean(a)",
             wm = "mymean(a, ...)" #gives error 
}

#Partial working function
my_summary <- function(df, x, ...) {
  summarize_(df, 
             m = "mean(a)", #works
             wm1 = interp(mymean(a), a = as.name(x) #works but doesn't  allow ... 
             wm2 = interp(mymean(a, b), 
                          .values=list(a = as.name(x),
                                       b = quote(...)),  #doesn't work
             wm3 = interp(mymean(a, ...), a = as.name(x) #doesn't work
}

工作功能可以让我打电话：

my_summary(df, a)
#> 5.5

my_summary(df, a, w=5, s=2)
#> 29.5

Answer 1

这个非常hacky，可怕的函数可以解决这个问题，但如果mymean包含许多可选参数会怎么样呢？

mymean <- function(x, w=1, s = 0) { s + mean(w * x) }

my_summarize <- function(df, x, ...) {
  vlist = list(...)
  vlist_names = names(vlist)
  if ("w" %in% vlist_names & "s" %in% vlist_names) {
    res = summarize_(df, m = interp(~mymean(a, w=b, s=c), 
                                    .values = list(a = as.name(x),
                                                   b = vlist$w,
                                                   c = vlist$s)))
  }
  else if ("w" %in% vlist_names) {
    res = summarize_(df, m = interp(~mymean(a, w=b), 
                                    .values = list(a = as.name(x),
                                                   b = vlist$w)))
  }
  else if ("s" %in% vlist_names) {
    res = summarize_(df, m = interp(~mymean(a, s=c), 
                                    .values = list(a = as.name(x),
                                                   c = vlist$s)))
  }
  else {
    res = summarize_(df, m = interp(~mymean(a), a = as.name(x)))
  }
  res    
}


df = data.frame(a = 1:10)

my_summarize(df, "a")
#>     m
#> 1 5.5
my_summarize(df, "a", s=5)
#>      m
#> 1 10.5
my_summarize(df, "a", w=2)
#>    m
#> 1 11
my_summarize(df, "a", w=2, s=5)
#>    m
#> 1 16