如何在不更改R中的工作目录的情况下访问子文件夹中的指定文件？

Question

In R, I want to access to some file in subfolder. But I don't want to change working directory then move back. It lost time and long.

For exmaple, I working on /home/phuong folder. Here is the tree structure of phuong.

phuong-> data1, data2, data3.
data1-> abc.csv, def.csv, script1.R
data2-> bond.csv, option.csv, pricing.R
data3->.....

So i want to load data in abc.csv, def.csv and run code in pricing.R.

So if use code setwd , it make me lost many time and look code so stupid, like this:

setwd("/home/phuong/data1" );

read.csv("abc.csv");
read.csv("def.csv");
setwd("/home/phuong/data2" );
source("pricing.R")

I lost a lot of times to move from folder to another folder but all of them in the same folder home/phuong/ . So I need some way to access to any file in subfolder without setwd command. Please help me , thks.

Answer 1

Assuming your working directory is /home/hermie and you want to load a .csv file from a directory below your current WD (let's say /home/hermie/data ), you can simply do this:

setwd('/home/hermie')
myData <- read.csv('./data/myCsvFile.csv')

Of course you could also navigate "upwards" in the directory tree. Let's say you want to load a file in Bob's home directory ( /home/bob ). You can do it as follows:

setwd('/home/hermie')
data_from_bob <- read.csv('../bob/otherDataFile.csv') # Of course, this will work
                                                      # only if you can read
                                                      # files from that directory

Hope this helps.

---

Update

Somehow I think you want someone to write the solution for you... and I propose this:

> setwd('/home/phuong')
> data_abc <- read.csv('./data1/abc.csv')
> data_def <- read.csv('./data1/def.csv')
> source('./data2/pricing.R')

Is it really so dificult to write this? You would have to write much more if you changed your WD on every step of the way.

And, about my sugestion on symlinks, on your bash terminal you could do something like this:

$ cd /home/phuong
$ ln -s ./data1/abc.csv data1_abc.csv
$ ln -s ./data1/def.csv data1_def.csv
$ ln -s ./data2/pricing.R pricing.R

And then, from R:

> setwd('/home/phuong')
> data_abc <- read.csv('data_abc.csv')
> data_def <- read.csv('data_def.csv')
> source('pricing.R')

Answer 2

You could use what Hadley calls a closure in Advanced R if I understand what you're after:

## Make a function that takes a path and another function
## and returns that same function with the path pre-progammed in
pathit <- function(FUN, path){
    function(file, ...){
        FUN(file=file.path(path, file), ...)
    }
}

## generate new functions that have the path pre-programmed in
read.csv2b <- pathit(read.csv, "home/phuong/data1")
source2 <- pathit(source, "home/phuong/data2")


read.csv2b("abc.csv")
read.csv2b("def.csv")
source2("pricing.R")

If you have a lot of stuff to read in this may be worthwhile otherwise why not supply the whole path to the actual functions? If this isn't what you're after it was still a fun learning experience for me :-)

Answer 3

For me, the most intuitive way to learn to navigate folders is by using list.files("../") . You will see how upstream or downstream you need to navigate from your current location :)

Answer 4

I don't know if this is what you were looking for, but I have a library of files in a specific github folder that I source when I initialize a project in a new branch, and I found one solution online like this:

setwd("~/repos/library/all_the_things")
library_files <- list.files()
sapply(library_files, source)

which is great except now I'm starting every file in a directory where I wouldn't want to save plots etc., so I realized all it needed was this:

library_files <- list.files("~/repos/library/all_the_things", full.name=T)
sapply(library_files, source)

and now it runs them without changing directories.

Answer 5

I just discovered the here package. I used it for a project I'm working on. It provides a cleaner method of referencing nested folders than the other options mentioned so far. As an example:

install.packages('here')
library('here')
here::i_am("pricing.R")
read.csv(here("data1/abc.csv"));
read.csv(here("data1/def.csv"));