跨列删除NAs并使用牛津逗号的连接

Question

我想在删除NAs的列之间实现串联并观察牛津逗号。

    x <-  data.frame(ID = 1:3,
                 col1 = c("snap", "snap", NA),
                 col2 = c(NA, "crackle", "crackle"),
                 col3 = c(NA, NA, "pop"),
                 col4 = c(NA, "yummy", NA))

使用上面的数据帧，我想连接col1：col4并将结果返回给x $ treats

x$treats[1]
    "snap"

x$treats[2]
"snap, crackle, and yummy" 

x$treats[3]
"crackle and pop"

数据集还有一个不应包含在串联中的ID变量（因此不允许我指定所需列的解决方案不完整）。

Answer 1

这是另一种选择，虽然更加冗长。通过将列表生成包装到函数中，我们还可以添加一个选项来禁用牛津逗号，如果需要：

x <-  data.frame(
  ID = 1:3,
  col1 = c("snap", "snap", NA),
  col2 = c(NA, "crackle", "crackle"),
  col3 = c(NA, NA, "pop"),
  col4 = c(NA, "yummy", NA)
)

language_list <- function(x, oxford_comma = TRUE) {
  x <- x[!is.na(x)]

  if (length(x) < 2) {
    return(x)
  }

  last <- tail(x, 1)
  rest <- head(x, -1)

  if (length(rest) == 1) {
    return(paste(rest, "and", last))
  }

  rest <- paste(rest, collapse = ", ")    
  paste0(rest, if (oxford_comma) ",", " and ", last)
}

cols <- paste0("col", 1:4)
x$treats <- apply(x[, cols], 1, language_list) 

x$treats                                            
#> [1] "snap"                     "snap, crackle, and yummy"
#> [3] "crackle and pop"

Answer 2

> x <-  data.frame(ID = 1:3,
                 col1 = c("snap", "snap", NA),
                  col2 = c(NA, "crackle", "crackle"),
                  col3 = c(NA, NA, "pop"),
                  col4 = c(NA, "yummy", NA),stringsAsFactors = F)

> a=gsub("(\\w)\\s+","\\1, ",trimws(do.call(paste,replace(x[-1],is.na(x[-1]),""))))

(x1=transform(x,treat=gsub(",\\s(\\w+)$",", and \\1",a),stringsAsFactors=F))
  ID col1    col2 col3  col4                    treat
1  1 snap    <NA> <NA>  <NA>                     snap
2  2 snap crackle <NA> yummy snap, crackle, and yummy
3  3 <NA> crackle  pop  <NA>         crackle, and pop

> x1$treat[1]
[1] "snap"

> x1$treat[2]
[1] "snap, crackle, and yummy"

> x1$treat[3]
[1] "crackle, and pop"

你也可以使用collapse包中的glue：

 x$trat=apply(x[-1],1,function(y)glue::collapse(y[!is.na(y)],", ",last = ", and "))
> x$treat[1]
[1] "snap"

> x$treat[2]
[1] "snap, crackle, and yummy"

> x$treat[3]
[1] "crackle, and pop"