R 数据框到嵌套列表

我想将这种格式的数据帧（tbl）转换为以下嵌套列表（tbllst）：

library(tidyr)

tbl <- tribble(
  ~Col1, ~Col2, ~Col3,
  "Var1", "Var1_1", "Var1_1_1", 
  "Var1", "Var1_1", "Var1_1_2", 
  "Var1", "Var1_2", "Var1_2_1", 
  "Var1", "Var1_2", "Var1_2_2", 
)

tbllst <- list(
  Col1 = list(
    "Var1" = list(
      Col2 = list(
        "Var1_1" = list(
          Col3 = c(
            "Var1_1_1", 
            "Var1_1_2"
          )
        ),
        "Var1_2" = list(
          Col3 = c(
            "Var1_2_1", 
            "Var1_2_2"
          )
        )
      )
    )
  )
)

有没有一种自动化的方法来实现这一目标？

功能rrapply() in the rrapply- 包有一个选项how = "unmelt"它将熔化的 data.frame 转换为嵌套列表，其中 data.frame 中的每一行都成为嵌套列表中的节点路径。

要应用此功能，我们首先需要将tbldata.frame 为所需的输入格式rrapply():

library(purrr)
library(dplyr)
library(rrapply)

## put data.frame in format for rrapply-function
tbl1 <- imap_dfc(tbl, ~bind_cols(.y, .x)) %>%
  group_by(across(num_range(prefix = "...", range = 1:5))) %>%
  summarize(`...6` = list(c(`...6`)))

tbl1
#> # A tibble: 2 x 6
#> # Groups:   ...1, ...2, ...3, ...4 [2]
#>   ...1  ...2  ...3  ...4   ...5  ...6     
#>   <chr> <chr> <chr> <chr>  <chr> <list>   
#> 1 Col1  Var1  Col2  Var1_1 Col3  <chr [2]>
#> 2 Col1  Var1  Col2  Var1_2 Col3  <chr [2]>

## unmelt to nested list
ls_tbl <- rrapply(tbl1, how = "unmelt")

str(ls_tbl)
#> List of 1
#>  $ Col1:List of 1
#>   ..$ Var1:List of 1
#>   .. ..$ Col2:List of 2
#>   .. .. ..$ Var1_1:List of 1
#>   .. .. .. ..$ Col3: chr [1:2] "Var1_1_1" "Var1_1_2"
#>   .. .. ..$ Var1_2:List of 1
#>   .. .. .. ..$ Col3: chr [1:2] "Var1_2_1" "Var1_2_2"

请注意，其目的是group_by() and summarize()操作只是为了获得多个var1_%_%在一个单一的Col3节点。下面的方法要容易得多（但不会产生完全相同的结果）：

ls_tbl <- rrapply(imap_dfc(tbl, ~bind_cols(.y, .x)), how = "unmelt")

str(ls_tbl)
#> List of 1
#>  $ Col1:List of 1
#>   ..$ Var1:List of 1
#>   .. ..$ Col2:List of 2
#>   .. .. ..$ Var1_1:List of 2
#>   .. .. .. ..$ Col3: chr "Var1_1_1"
#>   .. .. .. ..$ Col3: chr "Var1_1_2"
#>   .. .. ..$ Var1_2:List of 2
#>   .. .. .. ..$ Col3: chr "Var1_2_1"
#>   .. .. .. ..$ Col3: chr "Var1_2_2"