I am trying to call a bash script in perl, by doing this -
my $which_mpi = "/sw/tools/tacc/builds/carter/site/salt_which_mpi";
$mpi_stack = system("$which_mpi",-n);
the problem is that I want the script to execute when it is called by system command in line
$mpi_stack = system("$which_mpi",-n);
but the problem with this is while assigning the path in $which_mpi it automatically executes the script. So instead of me getting this value
WHICH_MPI : /sw/tools/salt/builds/carter/altd/bin/site/salt_which_mpi
I am getting the path with the output of the bash script like
WHICH_MPI : /sw/tools/salt/builds/carter/altd/bin/site/salt_which_mpiopenmpi 1.6.1
where openmpi 1.6.1 is the output of my salt_which_mpi bash script.
You can use perl's backtick operator
`bash -c "$WHICH_MPI"`
take a look at this question:
another stackoverflow question
Related
Running Lubuntu -
Beginner Perl programmer
Script is XXX.pl located at ~/projects/XXX/XXX.pl
First line is the shebang
#!/usr/bin/perl
Permission to run is set to Anyone.
In directory ~/projects/XXX, the command
~/projects/XXX$ perl XXX.pl
works as desired, but the command
~/projects/XXX$ XXX.pl
Fails with XXX.pl: command not found
What am I missing ?
The two usual options to execute your Perl script are:
perl XXX.pl
or
./XXX.pl
Both ways assume that your current working directory contains the script XXX.pl, otherwise it won't work.
As already pointed out by jm666 in the comments, you can usually not execute a program or script from your current working directory without prepending ./, primarily because of security reasons. Now, you may wonder why it's necessary.
Explanation:
Your shell uses the contents of an environment variable called $PATH to find out about where external commands (non-builtin programs) are located in your filesystem. If you want to see what's in $PATH, just type the following in your shell:
echo $PATH
Now you can see that the $PATH variable does NOT contain your current working directory. The consequence is that your shell is not able to find the program XXX.pl. By prepending ./ you instruct the shell to execute the program which comes after.
But there are two requirements if you want to execute your Perl script with ./script.pl:
The script has to be executable (check with ls -l)
The first line (shebang line) has to be #!/path/to/your/perl because your shell needs that information to find the perl interpreter in order to run your script
However, #1 and #2 are NOT required when you execute your script with
perl XXX.pl
because it invokes the perl interpreter directly with your script.
See how to make Perl scripts executable on Linux and make the script itself directly executable with chmod for some more details.
Can the script be found?
Is . in your path? If it's not, add it to your path, or use ./XXX.pl instead of XXX.pl.
Can the script be executed?
Do you have execute permission to the file? Fix using chmod u+x XXX.pl.
Is the interpreter correct?
which perl will tell you which interpreter is used when you use perl XXX.pl. That's the path that should be on your shebang (#!) line.
I am trying to run a perl script from within a *.csh script, but I get the following error:
Can't open perl script "checkLength.perl": No such file or directory
I think it has something to do with the path of the perl script. I can run a perl script fine when it's not being called within the shell script, but a perl script won't run when I have it being called in the cshell script.
The shebang I use for the perl script is : #!/usr/bin/perl. I checked, and perl is located there (not in /usr/bin/env perl). I made sure it's executable.
Do I need to edit the $PATH to get this to work? I'm confused because if I run a perl script without calling it within another csh script, it runs just fine.
Of course one obvious solution to your problem would be to simply use the full path to the script. But this is often not preferable, as the script would brake on installations that do not mirror exactly the original structure.
So there are two possibilities:
Either you put your perl script into a well defined location that is contained in your PATH variable so any program that runs under the respective users environment finds it.
Or you have some well defined location relative to the location of your calling script. Say your bash script is located in somedir and you perl script is in somedir/subdir, then your call would be subdir/perlscript.pl. Then, to become independent of possible changes in the working directory of the caller, you could determine the current full path of the perl script to be called on start of the bash script.
A template for this would be:
#!/bin/bash
FULLPATH=$(pwd)/subdir/somescript.pl
# do something else, cd ...
$FULLPATH
I am assuming your Perl script is kept in a fixed path relative to your bash script. Assuming your directory structure to be like so:
(BASEDIR)/bin/example.sh
(BASEDIR)/perl/example.pl
To allow you to run your bash script from anywhere in your system you must specify the relative path to your Perl script by getting the BASEDIR.
#!/bin/bash
BASEDIR=$(dirname $0)
perl $BASEDIR/../perl/example.pl
What we are doing above is finding the location of your bash script(BASEDIR/bin) that you call and then finding the relative location of the Perl script using the location of the bash script as reference. Now you will be able to call the bash script from anywhere and run your Perl script normally.
I have a series of perl scripts that I want to run one after another on a unix system. What type of file would this be / could I reference it as in documentation? BASH, BATCH, Shell Script File?
Any help would be appreciated.
Simply put the commands you would use to run them manually in a file (say, perlScripts.sh):
#!/bin/sh
perl script1.pl
perl script2.pl
perl script3.pl
Then from the command line:
$ sh perlScripts.sh
Consider using Perl itself to run all of the scripts. If the scripts don't take command line arguments, you can simply use:
do 'script1.pl';
do 'script2.pl';
etc.
do 'file_name' basically copies the file's code into the current script and executes it. It gives each file its own scope, however, so variables won't clash.
This approach is more efficient, because it starts only one instance of the Perl interpreter. It will also avoid repeated loading of modules.
If you do need to pass arguments or capture the output, you can still do it in a Perl file with backquotes or system:
my $output = `script3.pl file1.txt`; #If the output is needed.
system("script3.pl","file1.txt"); #If the output is not needed.
This is similar to using a shell script. However, it is cross-platform compatible. It means your scripts only rely on Perl being present, and no other external programs. And it allows you to easily add functionality to the calling script.
I am looking for a nice way to get the following done:
So I have a script that I need to run in Python in Unix by calling from a Perl script that was, in turn, called from my Excel VBA macro in Windows using Plink. The Python script, due to dependency issues, has to run in either csh or bash, and I will need to use export/setenv to add a few libraries before running the script. However by default, perl runs in sh shell and as such, there is no way I can add in all the dependencies and have the Python script to run.
So, I am just wondering if there is EITHER: 1. a way for me to add dependencies to sh shell in the perl script, OR 2. force my perl script to run in csh (preferred, since for some reason .bashrc for the account runs into permission issues).
Thanks a lot!
How about "3. Set the appropriate environment variable in the Perl or Python scripts"?
$ENV{'PATH'} = ...
...
os.environ['PATH'] = os.pathsep.join(newpaths + os.environ['PATH'].split(os.pathsep))
(dunno how to get the path separator in Perl, sorz)
To force the shell to csh, try the following in Perl :
`/bin/csh -c "command_name"`;
Edit:
You can use ENV variable, like this. Try that :
$s = `/bin/bash -c 'VAR_FOO=753; echo \$VAR_FOO'`;
print $s;
I ended up just change the .cshrc script, apparently the addition to PATH, for some reason, did not work for me. After that, everything runs smoothly by putting all into one line
so basically it looks something like this
/path/to/.cshrc && /python/path/to/python
Hope that helps!
I have a Korn shell script at a location like /opt/apps/abc/folder/properties.env. I can execute it from Unix bash using the dot command:
. /opt/apps/abc/folder/properties.env
This works.
I have a Perl script abc.pl from which I am calling the script properties.env. I tried the following different:
system('/usr/bin/ksh','-c', '. /opt/apps/abc/folder/properties.env');
/usr/bin/ksh -c /opt/apps/abc/folder/properties.env;
system('. /opt/apps/abc/folder/properties.env');
None of the above work. I don't want to use exec because I want to return to the Perl script. What am I doing wrong?
The environment changes will only last as long as the life of the ksh session spawned by the system command. If you want the environment changes to affect the Perl script, then you have to source that file before you launch the Perl program.
If you need those environment variables in your perl code, (not in the environment where you called perl), you can also read and parse that properties.env and set the environment in the %ENV variable.
e.g
$ENV{'ENV_VAR1'}=VALUE_OF_ENV_VAR1
using system() spawn another process, as the other poster said. changing environment in the child does not affect the parent.