R语言中：如何替代list.dirs返回命名嵌套列表而非字符向量？

场景说明 #

我们有如下层级的文件夹结构（可通过以下R代码创建，实际为已存在的目录）：

c("Level1/Level1A/Level1AA", 
  "Level1/Level1A/Level1AB", 
  "Level1/Level1B/Level1BA", 
  "Level1/Level1B/Level1BB",
  "Level2/Level2A/Level2AA", 
  "Level2/Level2A/Level2AB", 
  "Level2/Level2B/Level2BA", 
  "Level2/Level2B/Level2BB") %>% 
  purrr::map(dir.create, recursive = TRUE)

在父目录执行list.dirs()会返回一个扁平的字符向量：

[1] "."                         "./Level1"                  "./Level1/Level1A"          "./Level1/Level1A/Level1AA" "./Level1/Level1A/Level1AB"
 [6] "./Level1/Level1B"          "./Level1/Level1B/Level1BA" "./Level1/Level1B/Level1BB" "./Level2"                  "./Level2/Level2A"         
[11] "./Level2/Level2A/Level2AA" "./Level2/Level2A/Level2AB" "./Level2/Level2B"          "./Level2/Level2B/Level2BA" "./Level2/Level2B/Level2BB"

但我们需要将其转换为命名嵌套列表，格式如下：

list(Level1 = list(Level1A = list("Level1AA",
                                  "Level1AB"),
                   Level1B = list("Level1BA",
                                  "Level1BB")),
     Level2 = list(Level2A = list("Level2AA",
                                  "Level2AB"),
                   Level2B = list("Level2BA",
                                  "Level2BB")))

最简解决方案 #

使用data.tree包可以快速实现需求，该包专门用于树形结构的处理，代码简洁直观：

1. 先安装并加载包：

install.packages("data.tree")
library(data.tree)

2. 执行转换代码：

# 获取所有目录路径，排除当前目录（"."）
dir_paths <- list.dirs(recursive = TRUE, full.names = FALSE)[-1]

# 将路径转换为树形结构，再导出为嵌套列表
dir_tree <- as.Node(data.frame(path = paste0("root/", dir_paths), stringsAsFactors = FALSE))
result <- dir_tree$ToListExcludeRoot()

执行后result就是目标格式的命名嵌套列表。

替代方案（仅用tidyverse工具） #

如果不想额外安装包，也可以用tidyverse的组合函数实现：

library(tidyverse)

dir_paths <- list.dirs(recursive = TRUE, full.names = FALSE)[-1]

result <- dir_paths %>%
  # 拆分每个路径的层级
  str_split("/", simplify = TRUE) %>%
  as.data.frame(stringsAsFactors = FALSE) %>%
  # 转长格式并过滤空值
  pivot_longer(cols = everything(), names_to = "depth", values_to = "name", values_drop_na = TRUE) %>%
  mutate(depth = as.integer(depth)) %>%
  # 按层级去重并嵌套
  group_by(depth) %>%
  distinct(name) %>%
  nest() %>%
  # 逐层构建嵌套列表
  mutate(data = map(data, ~ set_names(map(.$name, list), .$name))) %>%
  pull(data) %>%
  reduce(function(prev, curr) map(prev, ~ c(., curr[names(.)])))

内容的提问来源于stack exchange，提问作者J. Doe