【发布时间】：2021-04-01 15:04:26
【问题描述】：
我正在使用带有 {dplyr} 1.0.0 的 list-columns 和 data.frames 玩弄，我想知道是否可以在不离开管道的情况下在每个 data.frame 中使用 rename() 和 mutate() 列当嵌套的data.frame被分组rowwise.
为什么我想知道/这样做？据我了解{dplyr} 1.0.0 的理念，它推荐rowwise() 而不是在列上使用{purrr} 的map-family。下面我首先展示我在 {dplyr} 1.0.0 之前所做的事情，然后展示 {dplyr} 1.0.0 的几个示例（大多数都不起作用）。
虽然 {rlang} supports glue strings on the left hand side (LHS) 可以在编写 {dplyr} 自定义函数时使用，但 rowwise tibble 中的 {dplyr} 函数的 LHS 似乎尚不受支持（至少我下面的示例不起作用） .
对于rename，我找到了一种使用rename_with() 的方法，但我不知道如何使用mutate。
我也不明白我收到的大部分错误信息。他们或多或少说我在:= 之前没有在LHS 上使用字符串，但在rowwise 模式下，我引用的列（new）实际上是length == 1 的字符向量。
library(dplyr, quietly = TRUE, warn.conflicts = FALSE)
library(purrr)
myiris <- iris %>%
nest_by(Species, .key = "mydat") %>%
ungroup %>%
mutate(new = letters[1:3])
# our data looks like this
# we want to use the strings in column `new` on the LHS of `rename` and `mutate`
myiris
#> # A tibble: 3 x 3
#> Species mydat new
#> <fct> <list<tbl_df[,4]>> <chr>
#> 1 setosa [50 x 4] a
#> 2 versicolor [50 x 4] b
#> 3 virginica [50 x 4] c
# For reference: under dplyr < 1.0 I did the following:
# rename in pipe
# working
myiris %>%
mutate(mydat = map2(mydat, new,
~ rename_at(.x, "Sepal.Length", function(z) paste(.y)))) %>%
pull(mydat)
#> [[1]]
#> # A tibble: 50 x 4
#> a Sepal.Width Petal.Length Petal.Width
#> <dbl> <dbl> <dbl> <dbl>
#> 1 5.1 3.5 1.4 0.2
#> 2 4.9 3 1.4 0.2
#> 3 4.7 3.2 1.3 0.2
#> 4 4.6 3.1 1.5 0.2
#> # ... with 46 more rows
#>
#> [[2]]
#> # A tibble: 50 x 4
#> b Sepal.Width Petal.Length Petal.Width
#> <dbl> <dbl> <dbl> <dbl>
#> 1 7 3.2 4.7 1.4
#> 2 6.4 3.2 4.5 1.5
#> 3 6.9 3.1 4.9 1.5
#> 4 5.5 2.3 4 1.3
#> # ... with 46 more rows
#>
#> [[3]]
#> # A tibble: 50 x 4
#> c Sepal.Width Petal.Length Petal.Width
#> <dbl> <dbl> <dbl> <dbl>
#> 1 6.3 3.3 6 2.5
#> 2 5.8 2.7 5.1 1.9
#> 3 7.1 3 5.9 2.1
#> 4 6.3 2.9 5.6 1.8
#> # ... with 46 more rows
# mutate in pipe
# was never working even under dplyr < 1.0.0
myiris %>%
mutate(mydat = map2(mydat, new,
~ mutate(.x, eval(.y) := .y))) %>%
pull(mydat)
#> Error: Problem with `mutate()` input `mydat`.
#> x The LHS of `:=` must be a string or a symbol
#> i Input `mydat` is `map2(mydat, new, ~mutate(.x, `:=`(eval(.y), .y)))`.
# mutate with custom function
# working
mymutate <- function(df, y) {
mutate(df, !! y := y)
}
myiris %>%
mutate(mydat = map2(mydat, new,
~ mymutate(.x, .y))) %>%
pull(mydat)
#> [[1]]
#> # A tibble: 50 x 5
#> Sepal.Length Sepal.Width Petal.Length Petal.Width a
#> <dbl> <dbl> <dbl> <dbl> <chr>
#> 1 5.1 3.5 1.4 0.2 a
#> 2 4.9 3 1.4 0.2 a
#> 3 4.7 3.2 1.3 0.2 a
#> 4 4.6 3.1 1.5 0.2 a
#> # ... with 46 more rows
#>
#> [[2]]
#> # A tibble: 50 x 5
#> Sepal.Length Sepal.Width Petal.Length Petal.Width b
#> <dbl> <dbl> <dbl> <dbl> <chr>
#> 1 7 3.2 4.7 1.4 b
#> 2 6.4 3.2 4.5 1.5 b
#> 3 6.9 3.1 4.9 1.5 b
#> 4 5.5 2.3 4 1.3 b
#> # ... with 46 more rows
#>
#> [[3]]
#> # A tibble: 50 x 5
#> Sepal.Length Sepal.Width Petal.Length Petal.Width c
#> <dbl> <dbl> <dbl> <dbl> <chr>
#> 1 6.3 3.3 6 2.5 c
#> 2 5.8 2.7 5.1 1.9 c
#> 3 7.1 3 5.9 2.1 c
#> 4 6.3 2.9 5.6 1.8 c
#> # ... with 46 more rows
# dplyr > 1.0.0
# objective: `rename()` or `mutate()` in pipe on list-column of data.frames
# while using different column names on LHS coming from another
# column (here `new`)
myiris_row <- myiris %>% rowwise
# rename --------
# not working
myiris_row %>%
mutate(mydat = list(mydat %>% rename({{new}} := "Sepal.Length")))
#> Error: Problem with `mutate()` input `mydat`.
#> x The LHS of `:=` must be a string or a symbol
#> i Input `mydat` is `list(...)`.
#> i The error occured in row 1.
# not working
myiris_row %>%
mutate(mydat = list(mydat %>% rename(!! new := "Sepal.Length")))
#> Error: Problem with `mutate()` input `mydat`.
#> x The LHS of `:=` must be a string or a symbol
#> i Input `mydat` is `list(...)`.
#> i The error occured in row 1.
# not working
myiris_row %>%
mutate(mydat = list(mydat %>% rename(!! sym(new) := "Sepal.Length")))
#> Error: Only strings can be converted to symbols
# not working
myiris_row %>%
mutate(mydat = list(mydat %>% rename(all_of(new) := "Sepal.Length")))
#> Error: Problem with `mutate()` input `mydat`.
#> x The LHS of `:=` must be a string or a symbol
#> i Input `mydat` is `list(mydat %>% rename(`:=`(all_of(new), "Sepal.Length")))`.
#> i The error occured in row 1.
# working, but only with `rename_with()`
myiris_row %>%
mutate(mydat = list(mydat %>% rename_with(~ new, "Sepal.Length"))) %>%
pull(mydat)
#> [[1]]
#> # A tibble: 50 x 4
#> a Sepal.Width Petal.Length Petal.Width
#> <dbl> <dbl> <dbl> <dbl>
#> 1 5.1 3.5 1.4 0.2
#> 2 4.9 3 1.4 0.2
#> 3 4.7 3.2 1.3 0.2
#> 4 4.6 3.1 1.5 0.2
#> # ... with 46 more rows
#>
#> [[2]]
#> # A tibble: 50 x 4
#> b Sepal.Width Petal.Length Petal.Width
#> <dbl> <dbl> <dbl> <dbl>
#> 1 7 3.2 4.7 1.4
#> 2 6.4 3.2 4.5 1.5
#> 3 6.9 3.1 4.9 1.5
#> 4 5.5 2.3 4 1.3
#> # ... with 46 more rows
#>
#> [[3]]
#> # A tibble: 50 x 4
#> c Sepal.Width Petal.Length Petal.Width
#> <dbl> <dbl> <dbl> <dbl>
#> 1 6.3 3.3 6 2.5
#> 2 5.8 2.7 5.1 1.9
#> 3 7.1 3 5.9 2.1
#> 4 6.3 2.9 5.6 1.8
#> # ... with 46 more rows
# mutate ------
# the values of the new column don't matter
# here we just use the same input as the name, to show that RHS evaluation is easier.
# not working
myiris_row %>%
mutate(mydat = list(mydat %>% mutate(!! new := new)))
#> Error: Problem with `mutate()` input `mydat`.
#> x The LHS of `:=` must be a string or a symbol
#> i Input `mydat` is `list(...)`.
#> i The error occured in row 1.
# not working
myiris %>%
mutate(mydat = list(mydat %>% mutate(!! sym(new) := new)))
#> Error: Only strings can be converted to symbols
# not working
myiris_row %>%
mutate(mydat = list(mydat %>% mutate(all_of(new) := new)))
#> Error: Problem with `mutate()` input `mydat`.
#> x The LHS of `:=` must be a string or a symbol
#> i Input `mydat` is `list(mydat %>% mutate(`:=`(all_of(new), new)))`.
#> i The error occured in row 1.
# almost working (what's going on in the data[[1]] btw!)
myiris_row %>%
mutate(mydat = list(mydat %>% mutate("{{new}}" := new))) %>%
pull(mydat)
#> [[1]]
#> # A tibble: 50 x 5
#> Sepal.Length Sepal.Width Petal.Length Petal.Width `promise_fn(3L)`
#> <dbl> <dbl> <dbl> <dbl> <chr>
#> 1 5.1 3.5 1.4 0.2 a
#> 2 4.9 3 1.4 0.2 a
#> 3 4.7 3.2 1.3 0.2 a
#> 4 4.6 3.1 1.5 0.2 a
#> # ... with 46 more rows
#>
#> [[2]]
#> # A tibble: 50 x 5
#> Sepal.Length Sepal.Width Petal.Length Petal.Width `"b"`
#> <dbl> <dbl> <dbl> <dbl> <chr>
#> 1 7 3.2 4.7 1.4 b
#> 2 6.4 3.2 4.5 1.5 b
#> 3 6.9 3.1 4.9 1.5 b
#> 4 5.5 2.3 4 1.3 b
#> # ... with 46 more rows
#>
#> [[3]]
#> # A tibble: 50 x 5
#> Sepal.Length Sepal.Width Petal.Length Petal.Width `"c"`
#> <dbl> <dbl> <dbl> <dbl> <chr>
#> 1 6.3 3.3 6 2.5 c
#> 2 5.8 2.7 5.1 1.9 c
#> 3 7.1 3 5.9 2.1 c
#> 4 6.3 2.9 5.6 1.8 c
#> # ... with 46 more rows
由reprex package (v0.3.0) 于 2020 年 12 月 22 日创建
【问题讨论】：
-
这很有趣，你的动态运行的函数不起作用
myiris %>% mutate(mydat = map2(mydat, new, function(df, y) df %>% mutate(!! y := y))) %>% mydat。它可能与在环境中查找对象有关 -
这从来没有真正起作用，因为 bang-bang 运算符
!!仅适用于顶层而不适用于嵌套表达式。如果eval(y) :=或eval_tidy(y) :=可以工作，那也没关系，但它们没有。 -
在他们的 github 页面上提出问题可能会更好
-
我正在考虑，但也许有办法真正做到这一点。我对 LHS 表达式的所有可能符号感到困惑，所以我先在这里发布以确保。