如何将DataFrame列直接通过管道传入函数，整合到管道链末尾？

现有代码可以从DataFrame的两列生成拼接后的语句：

library(dplyr); library(tibble); library(magrittr)

mtcars %>% 
  rownames_to_column(var = "car") %>%
  sample_n(5) -> 
  df

paste0(df$car, " (", df$mpg, ")", collapse = ", ")

# 输出示例："Mazda RX4 Wag (21), Hornet Sportabout (18.7), Merc 280 (19.2), Dodge Challenger (15.5), Merc 450SLC (15.2)"

但想把拼接语句paste0(df$car, " (", df$mpg, ")", collapse = ", ")直接放到管道末尾，直接运行下面的代码会报错，调整方法如下：

方案1：逐行拼接后合并 #

先给每行生成拼接好的文本，再提取所有行的结果合并成字符串：

mtcars %>% 
  rownames_to_column(var = "car") %>%
  sample_n(5) %>%
  mutate(text = paste0(car, " (", mpg, ")")) %>%
  pull(text) %>%
  paste(collapse = ", ")

方案2：用magrittr的%$%管道 #

%$%是magrittr提供的特殊管道，会把前一步数据框的列名直接暴露出来，不用再用$引用：

mtcars %>% 
  rownames_to_column(var = "car") %>%
  sample_n(5) %$%
  paste0(car, " (", mpg, ")", collapse = ", ")

方案3：用with()包裹拼接逻辑 #

在管道里用with()直接引用数据框的列：

mtcars %>% 
  rownames_to_column(var = "car") %>%
  sample_n(5) %>%
  with(paste0(car, " (", mpg, ")", collapse = ", "))

方案4：用glue包简化写法（更易读） #

如果安装了glue包，用glue_data()可以更直观地拼接文本：

library(glue)

mtcars %>% 
  rownames_to_column(var = "car") %>%
  sample_n(5) %>%
  glue_data("{car} ({mpg})") %>%
  paste(collapse = ", ")

内容的提问来源于stack exchange，提问作者luciano