Determine path of the executing script

Question

I have a script called foo.R that includes another script other.R, which is in the same directory:

#!/usr/bin/env Rscript
message("Hello")
source("other.R")

But I want R to find that other.R no matter what the current working directory.

In other words, foo.R needs to know its own path. How can I do that?

Solution 1

Here there is a simple solution for the problem. This command:

script.dir <- dirname(sys.frame(1)$ofile)

returns the path of the current script file. It works after the script was saved.

Solution 2

You can use the commandArgs function to get all the options that were passed by Rscript to the actual R interpreter and search them for --file=. If your script was launched from the path or if it was launched with a full path, the script.name below will start with a '/'. Otherwise, it must be relative to the cwd and you can concat the two paths to get the full path.

Edit: it sounds like you'd only need the script.name above and to strip off the final component of the path. I've removed the unneeded cwd() sample and cleaned up the main script and posted my other.R. Just save off this script and the other.R script into the same directory, chmod +x them, and run the main script.

main.R:

#!/usr/bin/env Rscript
initial.options <- commandArgs(trailingOnly = FALSE)
file.arg.name <- "--file="
script.name <- sub(file.arg.name, "", initial.options[grep(file.arg.name, initial.options)])
script.basename <- dirname(script.name)
other.name <- file.path(script.basename, "other.R")
print(paste("Sourcing",other.name,"from",script.name))
source(other.name)

other.R:

print("hello")

output:

burner@firefighter:~$ main.R
[1] "Sourcing /home/burner/bin/other.R from /home/burner/bin/main.R"
[1] "hello"
burner@firefighter:~$ bin/main.R
[1] "Sourcing bin/other.R from bin/main.R"
[1] "hello"
burner@firefighter:~$ cd bin
burner@firefighter:~/bin$ main.R
[1] "Sourcing ./other.R from ./main.R"
[1] "hello"

This is what I believe dehmann is looking for.

Solution 3

I couldn't get Suppressingfire's solution to work when 'source'ing from the R console.
I couldn't get hadley's solution to work when using Rscript.

Best of both worlds?

thisFile <- function() {
        cmdArgs <- commandArgs(trailingOnly = FALSE)
        needle <- "--file="
        match <- grep(needle, cmdArgs)
        if (length(match) > 0) {
                # Rscript
                return(normalizePath(sub(needle, "", cmdArgs[match])))
        } else {
                # 'source'd via R console
                return(normalizePath(sys.frames()[[1]]$ofile))
        }
}

Solution 4

frame_files <- lapply(sys.frames(), function(x) x$ofile)
frame_files <- Filter(Negate(is.null), frame_files)
PATH <- dirname(frame_files[[length(frame_files)]])

Don't ask me how it works though, because I've forgotten :/

Solution 5

This works for me

library(rstudioapi)    
rstudioapi::getActiveDocumentContext()$path

View more solutions

Comments

[Frank]

No. :( I haven't seen any solution that actually works. Apart from the workaround to just pass the directory in or use an environment variable.

[Frank]

I hate having to rely on environment variables.

[aL3xa]

Try with system("locate other.R")... but be sure to give your script a unique name... O_o (assuming that you use UNIX system and that locate command is available)

[Etienne Low-Décarie]

This would be amazing to make scripts fully portable and executable by even R neofites!

[Etienne Low-Décarie]

It appears like all the answers require you to input the path at some point (at least to source the file)! It would be great if you could send someone a compressed folder and running any R script file within that folder would read from and save to that folder.

[Giacomo]

this single issue could actually become te reason why I could completely move to Python

[Michael Barton]

@giac_man, I feel R is full of hundreds of tiny problems like this that all add up to making very difficult to work in.

[mjs]

in Matlab there is the magic command 'mfilename' swich gives you the file name of currently running code. It cannot be so difficult to add something like that in R!

[M_Merciless]

Great package from @Andrew , perfect for logging purposes. Many thanks for putting this on CRAN.

[Suppressingfire]

In what context does that work? print(sys.frames()) turns up NULL when I run it.

[Richie Cotton]

@Suppressingfire: sys.frames returns the environments of the call stack, so it only really makes sense when called from a function. Try, e.g., foo <- function() {bar <- function() print(sys.frames()); bar()}; foo(). I can't figure out @hadley's code though because environments don't have an ofile member.

[hadley]

You have to source the file in - i.e. if I save that code then run source("~/code/test.r"), PATH will be set to ~/desktop. If you just evaluate it at the top level, it will return NULL.

[hadley]

I downmodded because your technique doesn't work with source as I thought the OP wanted - but maybe I misread his/her requirement. But I can't un-downmod :( Sorry!

[Suppressingfire]

But actually, it does work fine with source! Just source(other.name) and it works properly.

[Suppressingfire]

I think maybe we're talking at cross purposes. I think we have different understandings of what the dehmann is interested in doing.

[Frank]

This does not answer my question. I need to automatically find the "other.R" file. x$ofile is undefined, so frame_files is empty.

[Etienne Low-Décarie]

This requires you to have the script path. It does not allow you to make a truly portable R script that can run from anywhere.

[Sim]

@hadley, very useful code. I was able to generalize the "reload current script" utility function I add to almost all scripts when they are in active development. RScript reloader

[Jason]

For path concatenation, better to use other.name <- file.path(script.basename, "other.R")

[Ehsan88]

It doesn't work for me. I run R in Windows. Any idea?

[this.is.not.a.nick]

You always have to save the document befor running the script. Maybe is this the problem.

[wch]

I like this because it works with both Rscript and source() within R. I'd suggest doing normalizePath() on both versions, so that it gives the full path in both cases.

[The Unfun Cat]

This did not work recursively; the file I source looks for a data file (but in the wrong directory).

[RalfB]

Got the same error, with a saved scriptt and freshly installed and run R 3.2.0 on windows...

[Murta]

This error happens when you try to execute dirname(sys.frame(1)$ofile) directly from Rstudio. It works ok when the script is executed using source("other.R"), and dirname(sys.frame(1)$ofile) is inside "other.R".

[Ahmed Laatabi]

it's working under Mac.. this should be executed from a script file, nit directly from R terminal

[user9869932]

This option works fine in Mac with R-Studio. Thank-you!

[Mark Adamson]

I got the 'not that many frames on the stack' error when calling as a script with rscript.exe i.e. not using source(). so I had to instead use the solution from Suppressingfire below

[jcarlos]

For me, your solution was the best. Specially because it could be applied to a Shiny app and that one on link not.

[davidski]

Did not work for me as well (Win10, fresh R.3.3.0, ran from cmd line). Answer from @Suppressingfire below worked fine.

[Mark Adamson]

This doesn't work when using debugSource as the relevant property is then fileName rather than oFile. Just needs some simple if statements to get around it as shown in other solutions posted here like stackoverflow.com/a/32016824/538403

[Mark Adamson]

source in Rstudio gave ofile for me, but debugSource gave fileName so your solution works well but the code comments aren't quite right in my case

[O.rka]

This is the only thing that worked. Note, for this to work library(base) took me a while to figure that out lol

[RubenLaguna]

Here the getSrcDirectory is utils::getSrcDirectory

[user1071847]

Nope. That finds the directory of the process, not the file itself.

[Paul]

I gel NULL when this is placed in server.R when using shiny

[Paul]

When I try to run commandArgs(trailingOnly = FALSE) inside server.R in a shiny application, I get [1] "RStudio" "--interactive". No information about the directory it was called from.

[PeterVermont]

I do not see ofile but do see fileName

[geneorama]

This worked for me in windows, in linux, but then it failed when I called it from crontab. SAD!

[ManicMailman]

Doesn't work with interactive R session; I'm getting: ``` > source("csf.R") > csf() Error: RStudio not running ```

[theforestecologist]

what is "ofile"??

[Contango]

This might work nicely under Linux/Mac, but it did not work for me in an interative RStudio session under Windows. sourceDir was blank.

[Contango]

This worked for me in Windows with RStudio in interactive mode.

[John Haberstroh]

@EtienneLow-Décarie It does not require the script path, it gets it from bash. The main issue is that it is not a reliable way to get the path. Something like this is preferred, as in stackoverflow.com/questions/59895/… path_to_script="$( cd "$( dirname "${BASH_SOURCE[0]}" )" && pwd )"

[Michael Barton]

I still get the error "Error in sys.frame(1) : not that many frames on the stack "

[nJGL]

Not from within RStudio, though, except when sourcing

[Pranasas]

After all these years I consider this to be the best and cleanest answer.

[pommedeterresautee]

@Contango on an interactive terminal, there is no path!!! You want the path to a file.

[Ufos]

SOLUTION: user@laptop_model:~$ R -e 'source("/home/user/[path_to_file].R")' -- relative paths work like magic

[bokov]

It always returns NA, even if I create a script that prints its output and then call the script e.g. with R -e "library(getopt); testscript.R"

[Ryan C. Thompson]

As the name of the function implies, you need to run your script using Rscript.

[bokov]

Ah, oops. Thanks.

[Vince W.]

you sir get my vote, because this is the solution that worked for me

[Ista]

This only works from inside RStudio I guess. Trying from the terminal I get Error: RStudio not running.

[quest]

For me this works sometimes while rest of the time I get Error in path.expand(path) : invalid 'path' argument error.

[Andrew]

This solution didn't work for me with Rscript dir/foo.R. It did work with R -e "source('dir/foo.R')".

[Kim]

If this helps anyone, for the original post, that would mean source(file.path(dirname(thisFile()), "other.R")) in foo.R. This works for me.

[Bojan P.]

Package ‘scriptName’ was removed from the CRAN repository. - what now? :o

[Wassadamo]

One issue. Suppose in RStudio I source main.R which sources helper.R which calls thisFile(). It will fetch the path of main.R instead of helper.R. Any tips here?

[abalter]

I'm getting character(0). Suggestions?

[Ron]

For me the here package do exactly the job and seems to be an easy solution

[Boops Boops]

This did not work for me on a Linux machine; instead of returning the path of the file, it returned the directory I was currently located in. I created a test script called TEST.R with one line of code: print(fileSnapshot()$path) I saved it in this folder: /opt/home/boops/Desktop/Testfolder/TEST.R I then navigated to my desktop and tried to run the file: boops@linuxserver:~/Desktop$ Rscript /opt/home/boops/Desktop/Testfolder/TEST.R [1] "/opt/home/boops/Desktop"

[Colombo]

As I just figured what I want on how sys.frames work from previous answer, I have found this solution. sys.frames() will get stack of environment from the most recent to the oldest one. This include source, which will set the ofile atribute, and non sourced environment, such as functions, which do not have ofile atribute. This will simply find the latest environment with the ofile atribute. Brilliant.

[altabq]

I get an error message readLink: illegal option -- e usage: readLink [-FlLnqrsx] [-f format] [-t timefmt] [file ...]

[Kay]

more specifically it works, if run from a R script in R studio. Even on the console in RStudio it will not give the right result "" in my case

[Joe Flack]

Didn't work for me either. Returns the same thing as 'here()' when using the 'here' library. It returned the path to my currently open R project, but not he very file itself being executed.

[Joe Flack]

This is great. Can someone make a package?

[Patrick]

This works while running interactively in Rstudio as long as you don't change the document in focus. If you submit lines to run and then switch to another document while they run, the path to the other document will be returned.

[Patrick]

This works while running interactively in Rstudio as long as you don't change the document in focus. If you submit lines to run and then switch to another document while they run, the path to the other document will be returned.

[Andrew]

I've uploaded a package named "this.path" to CRAN, it should solve this issue!

[johnny]

When running this command on RGui, I get the following message. Any idea on how to get around it? Error in this.path::this.path() : 'this.path' used in an inappropriate fashion * no appropriate 'source' or 'sys.source' call was found up the calling stack * R is being run from RGui which requires a 'source' and 'sys.source' call on the calling stack

[Andrew]

I hadn't realized until you commented that you could run code from within a script from 'RGui', I thought previously that the only way to run code within a script from 'RGui' was to use 'source'. I'm looking into a fix for this issue, hopefully I'll find something soon. For now, you could use 'RStudio' to edit and run your scripts because I know it works from there. Sorry for the lack of an answer, but thank you for pointing out this bug!

[Andrew]

@johnny I believe I found a solution, but it only works on a Windows OS. I'm attempting to find a solution for the macOS version of "RGui" named "AQUA", and then I'll upload the update to the package to CRAN. It'll be about ~10 business days before the update is approved for release by one of the CRAN maintainers, hopefully 'RStudio' is working for you in the meantime!

[Andrew]

@johnny the update was released a few hours ago, much sooner than I was expecting. I've tested on two separate computers now, it seems to work as intended from 'RGui'!

[johnny]

Just tested v.0.2.0 on a script file saved in an RGui session, and it works for me. Thanks!

[Andrew]

@JoeFlack I've made a package for this called "this.path", you can find it on CRAN at CRAN.R-project.org/package=this.path

[Andrew]

This only works when using source or sys.source, and it always grabs the first source on the stack, not the most recent.

[Andrew]

@Wassadamo you could try package "this.path", it handles nested source calls correctly

[Andrew]

fileSnapshot()$path just returns the path of the current working directory, not the path of the executing script. Another thing, on Windows it is unnecessary to substitute the backslashes with forward slashes, but on a Unix-alike OS it is dangerous to do this since filenames can contain backslashes. Last thing, you should not have a path separator at the end of your path since R will no longer recognize the string as a path (as returned by 'file.exists').

[user5359531]

This is the correct answer. Really baffling how many people are wasting time with the other proposed answers to this.

[htlbydgod]

This does not work for me on Mac

[understorey]

In windows/Rstudio it returns character(0)

[understorey]

This does not work for me in RStudio/Windows from sys.frames()[[1]]$ofile I get NULL.