如何在R的dplyr嵌套函数中传递列名完成CSV数据连接？

尝试用tidyverse编写嵌套函数：

• 外层函数接收CSV文件名，读取为数据框，提取首列作为键列名后调用内层函数
• 内层函数接收两个数据框和键列名，通过inner_join完成连接

硬编码列名时内层函数可正常运行，但外层传递变量（如df1_key）或表达式（如colnames(df1)[1]）形式的列名时报错：

• 传递变量时的报错：

Error in inner_join():
! Join columns in x must be present in the data.
✖ Problem with df1_key.
Run rlang::last_trace() to see where the error occurred.

• 传递表达式时的报错：

Error in join_by():
! Expressions can't contain computed columns, and can only reference columns by name or by explicitly specifying
a side, like x$col or y$col.
ℹ Expression 1 contains colnames(df1)[1].
Run rlang::last_trace() to see where the error occurred.

可复现代码：

library(tidyverse)

# 创建示例CSV
df1 = tibble(
  key = LETTERS,
  value = sample.int(100, 26)
)
write_csv(df1, "df1.csv")

df2 = tibble(
  key = LETTERS[1:13],
  value = sample.int(100, 13)
)
write_csv(df2, "df2.csv")

# 连接两个数据框的函数
join_dfs = function(x, y, x_key = "key", y_key = "key") {
  
  df = x |>
    inner_join(y, by = join_by({{ x_key }} == {{ y_key }} ))
  
}

# 硬编码列名可正常运行
df3 = join_dfs(df1, df2, "key", "key")

# 加载CSV并调用join_dfs的外层函数
join_csvs = function(x, y, file) {
  df1 = read_csv(x)
  df1_key = colnames(df1)[1]
  print(df1_key)
  
  df2 = read_csv(y)
  df2_key = colnames(df2)[1]
  print(df2_key)
  
  df3 = join_dfs(df1, df2, df1_key, df2_key)
  write_csv(df3, file)
  
}

# 运行报错
join_csvs("df1.csv", "df2.csv", "df3.csv")

问题出在join_by({{ x_key }} == {{ y_key }})的用法上：{{}}（大括号插值）是用于直接引用符号形式的列名，但外层函数传递的是字符串类型的列名变量，两者不匹配。可以通过两种方式解决：

方法1：将字符串转为符号后插值 #

用rlang::sym()把字符串列名转为符号，再用!!（强制求值）在join_by中使用：

join_dfs = function(x, y, x_key = "key", y_key = "key") {
  # 把字符串转为符号
  x_sym = sym(x_key)
  y_sym = sym(y_key)
  
  df = x |>
    inner_join(y, by = join_by(!!x_sym == !!y_sym))
}

方法2：使用命名向量指定连接列 #

如果不需要join_by的复杂逻辑，直接用inner_join的by参数传递命名向量，更简洁：

join_dfs = function(x, y, x_key = "key", y_key = "key") {
  # 构造命名向量：x的列名作为名字，y的列名作为值
  join_cols = setNames(y_key, x_key)
  
  df = x |>
    inner_join(y, by = join_cols)
}

修改后，运行join_csvs("df1.csv", "df2.csv", "df3.csv")即可正常完成连接并写入文件。

内容的提问来源于stack exchange，提问作者Lee Hachadoorian