How to build a dendrogram from a directory tree?

问题

Given a root absolute directory path. How do I generate a dendrogram object of all path's below it so that I can visualize the directory tree with R?

Suppose the following call returned the following leaf nodes.

list.files(path, full.names = TRUE, recursive = TRUE)

root/a/some/file.R
root/a/another/file.R
root/a/another/cool/file.R
root/b/some/data.csv
root/b/more/data.csv

I'd like to make a plot in R like the output of the unix tree program:

root
├── a
│   ├── another
│   │   ├── cool
│   │   │   └── file.R
│   │   └── file.R
│   └── some
│       └── file.R
└── b
    ├── more
    │   └── data.csv
    └── some
        └── data.csv

It would be especially useful if the solution involved decomposing the file system tree into two data.frame's:

1. a table of nodes (with which I could include attributes such as modification date)
2. and a table of edges (also with attributes)

And then building the dendrogram object from those two data.frames.

回答1:

Here's a possible approach to get what you originally asked for which is a system like tree. This will give a data.tree object that's pretty flexible and could be made to plot like you might want but it's not entirely clear to me what you want:

path <- c(
    "root/a/some/file.R", 
    "root/a/another/file.R", 
    "root/a/another/cool/file.R", 
    "root/b/some/data.csv", 
    "root/b/more/data.csv"
)


library(data.tree); library(plyr)

x <- lapply(strsplit(path, "/"), function(z) as.data.frame(t(z)))
x <- rbind.fill(x)
x$pathString <- apply(x, 1, function(x) paste(trimws(na.omit(x)), collapse="/"))
(mytree <- data.tree::as.Node(x))

1  root                  
2   ¦--a                 
3   ¦   ¦--some          
4   ¦   ¦   °--file.R    
5   ¦   °--another       
6   ¦       ¦--file.R    
7   ¦       °--cool      
8   ¦           °--file.R
9   °--b                 
10      ¦--some          
11      ¦   °--data.csv  
12      °--more          
13          °--data.csv  


plot(mytree)

You can get the parts you want (I think) but it'll require you to do the leg work and figure out conversion between data types in data.tree: https://cran.r-project.org/web/packages/data.tree/vignettes/data.tree.html#tree-conversion

I use this approach in my pathr package's tree function when use.data.tree = TRUE https://github.com/trinker/pathr#tree

EDIT Per@Luke's comment below...data.tree::as.Node takes a path directly:

(mytree <- data.tree::as.Node(data.frame(pathString = path)))

                levelName
1  root2                 
2   ¦--a                 
3   ¦   ¦--some          
4   ¦   ¦   °--file.R    
5   ¦   °--another       
6   ¦       ¦--file.R    
7   ¦       °--cool      
8   ¦           °--file.R
9   °--b                 
10      ¦--some          
11      ¦   °--data.csv  
12      °--more          
13          °--data.csv  

回答2:

If you are on Windows, you can use my package dir2json, by installing it like this:

drat::addRepo("stlarepo")
install.packages("dir2json")

It is also possible to use it on Linux, but there is a DLL linked to the GHC dynamic libraries, which must be installed on the system (while this DLL is standalone on Windows).

> library(dir2json)
> cat(dir2tree("src"))
src
|
`- contrib
   |
   +- PACKAGES.gz
   |
   +- PACKAGES
   |
   +- jsonAccess_0.1.1.tar.gz
   |
   +- expansions_1.2.tar.gz
   |
   `- dir2json_2.1.0.tar.gz
> cat(dir2tree("src", vertical=TRUE))
                                            src                                             
                                             |                                              
                                          contrib                                           
                                             |                                              
      ---------------------------------------------------------------------------           
     /          |                 |                       |                      \          
PACKAGES.gz  PACKAGES  jsonAccess_0.1.1.tar.gz  expansions_1.2.tar.gz  dir2json_2.1.0.tar.gz

The package also contains a Shiny application which generates an interactive Reingold-Tilford tree representation of a folder:

> dir2json::shinyDirTree(".")

回答3:

It's worth adding that excellent fs package offers dir_tree function that delivers this functionality to R in a very convenient manner.

tmp_dir <- tempdir()
# Create some directories
for (i in 1:10) {
    dir.create(path = file.path(tmp_dir,
                                basename(tempfile(pattern = "dir")),
                                basename(tempfile(pattern = "sub_dir"))),
               recursive = TRUE)
}
# Create directory tree
fs::dir_tree(path = tmp_dir, recurse = TRUE)

Results

/tmp/RtmpEhB0ne
├── dir15213121dd5903
│   └── sub_dir1521315a5425ba
├── dir152131227b086f
│   └── sub_dir1521314255d96b
├── dir152131353e6603
│   └── sub_dir1521315b52aeed
├── dir15213136870535
│   └── sub_dir15213127b34f64
├── dir1521313bbf738b
│   └── sub_dir152131473939ea
├── dir152131403f4fd5
│   └── sub_dir152131115296e7
├── dir152131503d0d55
│   └── sub_dir15213114368572
├── dir1521316f0bb0c3
│   └── sub_dir1521314aea266b
├── dir1521317fe305e9
│   └── sub_dir152131bcfe8a
└── dir1521319800dfb
    └── sub_dir15213129defd4a

---

In addition to printing directory tree, discovered paths can be returned to an object.

sink(file = tempfile(fileext = ".log"))
res_fs_tree <- fs::dir_tree(path = tmp_dir, recurse = TRUE)
sink()
res_fs_tree[[1]]
# [1] "/tmp/RtmpEhB0ne/dir15213121dd5903/sub_dir1521315a5425ba"

来源：https://stackoverflow.com/questions/36094183/how-to-build-a-dendrogram-from-a-directory-tree