I have a list of folders. In each folder there's an R identical script that have to run on files into the folder. I wrote the script one time and copied the script in each folder. The problem is that I have a list of around 100 folders so it is impossible to me to setwd() in the current working dir by hand. I would like to know if it is possible to set current working dir with for example a "." in this way:
setwd("/User/myname/./")
or in another easy way that tells R the current working dir instead of typing every time the folder name.
how about this?
# set the working directory to the main folder containing all the directories
setwd( "/user/yourdir/" )
# pull all files and folders (including subfolders) into a character vector
# keep ONLY the files that END with ".R" or ".r"
r.scripts <- list.files( pattern=".*\\.[rR]$" , recursive = TRUE )
# look at the contents.. now you've got just the R scripts..
# i think that's what you want?
r.scripts
# and you can loop through and source() each one
for ( i in r.scripts ) source( i )
As far as I understand, you want to trigger a batch of R scripts, where the scripts are distributed across a number of folders.
Personally, I would probably write a shell script (or OS equivilent) to do this, rather than doing it in R.
for dir in /directoriesLocation/*/
do
cat $dir/scriptName.R | R --slave --args $arg1 $arg2
done
where $dir is the the location of all directories containing the R script scriptName.R
In addition to the other great answers the source function has a chdir argument that will temporarily change the working directory to the one that the sourced file is in.
One option would be to create a vector with the filenames (including paths) for each of your script files, using list.files and/or other tools. Then source each of those files and letting source with chdir handle setting the working directory for you.
Related
I'm working on an R script where I create multiple sub-folders according to some parameters. After everything I have to do I then have to copy one file into all of these new sub-folders.
Is there a way in R to do this easily? I would like to avoid for loops if I can. I've read about file.copy but that doesn't seem to do the trick (from what I've read, it works the opposite way: it copies everything from a folder -including subfolders and their contents if recursive=TRUE- to a new folder, whereas I want one specific file copied to all subdirectories of a folder)
I've seen questions like Copy files from folders and sub-folders to another folder and saving the structure of the folders, How to copy specific file from sub-folder to another Folder in R? or Copy folder recursive in R but that's just the opposite direction of what I want.
file.copy() has vectorized from and to arguments. If you have a vector of directories you want to copy the file to you can simply do:
filename <- "my_file.txt"
to_directories <- c("dir 1", "dir 2", "dir 3")
file.copy(filename, file.path(to_directories, filename))
[1] TRUE TRUE TRUE
I would like to import an external dataset using read.table() (or any other function for reading files) and then randomize or sample over it. The file is stored in a subfolder within the parent folder that contains the exercises *.rmd. I am working within a RStudio project. I tried placing the dataset in different levels of the folder structure. Using relative path did not work, but absolute paths did.
My folder structure is:
$home/project_name/exercises # It contains the RMD files
$home/project_name/exercises/data # It contains data files that I want to process
$home/project_name/datasets # this folder could eventually contain the dataset I want to process
To make this code more portable, I would like to know o the manage relative paths within *.Rmd for the knitting process.
The exercises are copied to a temporary directory and processed there. Hence, the easiest option is to copy these files to the temporary directory using include_supplement("file.csv"). By default this assumes that the file.csv is in the same directory that the exercise itself resides in. If it is in a subdirectory you can use include_supplement("file.csv", recursive = TRUE) and then subdirectories are searched recursively for file.csv.
After using include_supplement(), the copied file is available locally and can be processed with read.table() or also included in the exercise as a supplementary file. See http://www.R-exams.org/templates/Rlogo/ for a worked example. However, note that the Rlogo template explicitly specifies the directory from which the file should be copied. This is not necessary if that directory is the same as the exercise or a subdirectory.
I am trying to provide a way of running unattended R scripts through Rscript. The problem I am having is that the default output for graphics is a PDF file in the current directory. I would like to redirect this output to a separate folder but without having to change the script too much.
If my script is something simple as:
args = commandArgs(trailingOnly = TRUE)
mydata <- get some data frame somehow
plot(tayside)
And I execute the following commandline:
Rscript.exe --vanilla --default-packages=RODBC,graphics,grDevices sample.R > C:\temp\sample.Rout
I get a Rplots.pdf in the current folder and the sample.Rout file in the C:\temp\ folder.
Is there a way to specify an output folder and have Rscript put all output there? I have tried playing with the pdf.options(...) to pre-pend a default folder to the file parameter but no can do.
Ok, apparently it was easier than I thought, no need to use pdf.options() but simply pdf() at the top of the file (after getting the arguments):
pdf(paste0(args[1], "MyFile.pdf"))
or, for multiple files:
pdf(paste0(args[1], "MyFile%03d.pdf"), onefile=FALSE)
Can I work with parallel working directories in R, or can I change the working directory in a loop to access the files from different folders?
I find it easier to have a single working directory. You find out what that is using the
getwd()
function. Typically, my working directory is something like:
~/colin/project1/R
You can change your working directory using
setwd()
You can easily access other files using the full path. In particular, I find
##List files in current directory
list.files()
##Give full path
list.files(full.names=TRUE)
##list files in the species1 directory
list.files("species1/", full.names=TRUE)
very handy.
Don't change the working directory in a loop, loop over the directories and use file.path to get to the file you want. Something like:
for(path in c("data1","data2","data3")){
for(file in c("file1.txt","file2.txt")){
fullPath = file.path(path,file)
doSomethingWith(fullPath)
}
}
That will loop over data1/file1.txt, data1/file2.txt and so on. Note it will also handle differences between path separators in different operating systems - don't try and paste file path components together with paste because you'll get it wrong.
I want to be able to source() a file which includes a different file in its same directory, but I don't want to have to set the working directory from the R-prompt before running this file:
> getwd()
[1] "/Users/myser"
> source("/Users/myuser/workspace/myproject/myfile.r")
Inside /Users/myuser/workspace/myproject, there would be myfile.r and my-utils.r. myfile.r calls source('my-utils.r') from within it.
Other programming languages can determine the current file's path. Does R have something similar? Example:
cur_dir <- sys.get_current_file_path()
source(file.path(cur_dir, "my-utils.r"))
source("/Users/myuser/workspace/myproject/my-utils.r", chdir=TRUE)
When chdir option is set to true and the source file parameter is a full path, the directory of file will be used as the working directory while sourcing the file.
NOTE: cur_dir <- sys.get_current_file_path() doesn't make much sense because pathnames are not unique.