This question may sound silly but I haven't been able to solve perhaps someone could help me.
I have a script called a.sh
# a.sh
source b.sh [ARGUMENT1] [ARGUMENT2]
echo "Running binding script"
echo $b_output
And here is what is on the b.sh script
# b.sh
b_output=$(virsh update-device "$1" "$2" --live --persistent)
export b_output
When you run virsh update-device... command in the terminal, the output of the command if succeded is the following:
>virsh update-device [ARGUMENT1] [ARGUMENT2] --live --persistent
Device updated successfully
However, When I run a.sh this is what I get:
Running binding script
Device
Is there a way I can export the entire output of the command run in the b.sh script, instead of only displaying the first word?
This question already has an answer here:
bash - export doesn't work
(1 answer)
Closed 7 years ago.
I am running the following simple code in a shell script , but it seems like it cant export the variable :
#!/bin/bash
echo -n "Enter AWS_ACCESS_KEY_ID: "
read aws_access_key
export AWS_ACCESS_KEY_ID=$aws_access_key
After that I take the input from the user ,but when I run echo $AWS_ACCESS_KEY_ID I get a blank value .
Run your script in the current shell by using:
source your-script # this runs your-script in the existing shell
...or, if using a POSIX shell...
. your-script # likewise; that space is intentional!
not
./your-script # this starts a new shell just for `your-script`; its variables
# are lost when it exits!
...if you want variables it sets to be available to the shell that calls it.
To be clear, export puts a variable in the current process's environment -- but environment variables are propagated down to child processes, not up to parent processes.
Now, if your goal is to define an interactive command that's easy to call, you might want to consider an entirely different approach altogether -- putting a function in your .bashrc:
awsSetup() {
echo -n "Enter AWS_ACCESS_KEY_ID: "
read && [[ $REPLY ]] && export AWS_ACCESS_KEY_ID=$REPLY
}
...after which the user with this in their .bashrc can run awsSetup, which will run in the current shell.
I am working on a project that is hosted in Heroku. The app is hard coded to use Amazon S3 and looks for the keys in environment variables. This is what I wrote after looking at some examples and I am not sure why its not working.
echo $1
if [ $1 != "unset" ]; then
echo "set"
export AMAZON_ACCESS_KEY_ID=XXXXXXXXXXXX
export AMAZON_SECRET_ACCESS_KEY=XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
export S3_BUCKET_NAME=XXXXXXXXX
else
echo "unset"
export AMAZON_ACCESS_KEY_ID=''
export AMAZON_SECRET_ACCESS_KEY=''
export S3_BUCKET_NAME=''
fi
When running the script it goes to the set section. But when inspecting through echo $AMAZON_ACCESS_KEY_ID # => ''.
I am not sure what is causing the issue. I will be interested in...
A fix for this...
A way to extract and add heroku config variables in the the env in an easier way.
You need to source the script, not run it as a child. If you run the script directly, its environment disappears when it ends. Sourcing the script causes it to be executed in the current environment. help source for more information.
Example:
$ VAR=old_value
$ cat script.sh
#!/bin/bash
export VAR=new_value
$ ./script.sh
$ echo $VAR
old_value
$ source script.sh
$ echo $VAR
new_value
Scripts executed with source don't need to be executable nor do they need the "shebang" line (#!/bin/bash) because they are not run as separate processes. In fact, it is probably a good idea to not make them executable in order to avoid them being run as commands, since that won't work as expected.
I have a bash script which calls another bash script, like so:
#!/bin/bash
echo "Hi"
./script-two.sh
echo "Hello!"
The problem that I have is that it never makes it to printing "Hello!"
I think this is because ./script-two.sh (Which I did not write) is somehow exiting or changing the shell. I have included this script at the end of this post.
Is there a way I can gurentee that my execution will continue after script-two.sh executes?
I have looked into using the trap command, but I don't fully understand its use properly.
Thanks,
Casey
Here is the contents of what would be script-two.sh
#!/bin/sh
# This file is part of the DITA Open Toolkit project hosted on
# Sourceforge.net. See the accompanying license.txt file for
# applicable licenses.
# (c) Copyright IBM Corp. 2006 All Rights Reserved.
export DITA_HOME=cwd
if [ "${DITA_HOME:+1}" != "1" ]; then
echo "DITA_HOME environment variable is empty or not set";
exit 127;
fi
echo $DITA_HOME
cd "$DITA_HOME"
# Get the absolute path of DITAOT's home directory
DITA_DIR="`pwd`"
echo $DITA_DIR
if [ -f "$DITA_DIR"/tools/ant/bin/ant ] && [ ! -x "$DITA_DIR"/tools/ant/bin/ant ]; then
chmod +x "$DITA_DIR"/tools/ant/bin/ant
fi
export ANT_OPTS="-Xmx512m $ANT_OPTS"
export ANT_OPTS="$ANT_OPTS -Djavax.xml.transform.TransformerFactory=net.sf.saxon.TransformerFactoryImpl"
export ANT_HOME="$DITA_DIR"/tools/ant
export PATH="$DITA_DIR"/tools/ant/bin:"$PATH"
NEW_CLASSPATH="$DITA_DIR/lib:$DITA_DIR/lib/dost.jar:$DITA_DIR/lib/commons-codec-1.4.jar:$DITA_DIR/lib/resolver.jar:$DITA_DIR/lib/icu4j.jar"
NEW_CLASSPATH="$DITA_DIR/lib/saxon/saxon9.jar:$DITA_DIR/lib/saxon/saxon9-dom.jar:$NEW_CLASSPATH"
NEW_CLASSPATH="$DITA_DIR/lib/saxon/saxon9-dom4j.jar:$DITA_DIR/lib/saxon/saxon9-jdom.jar:$NEW_CLASSPATH"
NEW_CLASSPATH="$DITA_DIR/lib/saxon/saxon9-s9api.jar:$DITA_DIR/lib/saxon/saxon9-sql.jar:$NEW_CLASSPATH"
NEW_CLASSPATH="$DITA_DIR/lib/saxon/saxon9-xom.jar:$DITA_DIR/lib/saxon/saxon9-xpath.jar:$DITA_DIR/lib/saxon/saxon9-xqj.jar:$NEW_CLASSPATH"
if test -n "$CLASSPATH"
then
export CLASSPATH="$NEW_CLASSPATH":"$CLASSPATH"
else
export CLASSPATH="$NEW_CLASSPATH"
fi
"$SHELL"
It looks like script-two.sh is setting up an ant build environment.
I think the author intended that it sets up the build environment, then you type your build commands in manually, then type exit to leave the build environment.
I say this because the bottom line of script-two.sh is:
"$SHELL"
which starts a new shell.
Try running your script, then type exit. I think you will see it print Hello! after you type exit.
I'm guessing you're trying to do something like:
#!/bin/bash
echo "Hi"
./script-two.sh
ant <some args>
To do that, what you really want to do is source it, by changing:
./script-two.sh
to
. script-two.sh
e.g.
#!/bin/bash
echo "Hi"
. script-two.sh
ant <some args>
But, you will need to edit script-two.sh and change:
"$SHELL"
to:
case $0 in *script-two.sh)
# executed, start a new shell with the new environment
"$SHELL"
;;
*)
# sourced, don't start a new shell
;;
esac
so that it only starts a shell if the script is being run like ./script-two.sh, but not if it is being sourced like . script-two.sh.
Or if you absolutely can't change script-two.sh, then you could do:
#!/bin/bash
echo "Hi"
. script-two.sh </dev/null
ant <some args>
which will trick "$SHELL" into exiting because it has no input.
Also
export DITA_HOME=cwd
doesn't seem right to me.
It should probably be
export DITA_HOME=$(pwd)
or
export DITA_HOME=`pwd`
(both are equivalent)
I had a similar problem today, up on digging I finally found the answer.
The script I was calling (from within my script) actually had an exit 0 in the end. Removing that fixed my issues.
Just leaving this here as someone may find it useful.
Well for starters, you can execute your bash script with the -x switch to see where it is failing:
bash -x script-one.sh
Secondly, if you call the second script like this:
#!/bin/bash
echo "Hi"
var=$(bash script-two.sh)
echo "Hello!"
It will continue, as long as script-two.sh exits cleanly. Again, you can run the -x script against that script find any problems.
And as Mikel mentioned, always make sure to have exit at the bottom of your scripts.
Closed. This question does not meet Stack Overflow guidelines. It is not currently accepting answers.
This question does not appear to be about a specific programming problem, a software algorithm, or software tools primarily used by programmers. If you believe the question would be on-topic on another Stack Exchange site, you can leave a comment to explain where the question may be able to be answered.
Closed 9 years ago.
Improve this question
We have Oracle running on Solaris, and the shell is by default csh. So the login script sets the oracle_home, oracle_sid in csh also. But I don't like csh and want to use bash to do my work. So how to source the csh login script in bash?
e.g, the following is what in the .cshrc file. And when use bash, I'd like use these variables. One way is to copy the variables again and use bash command, such as export ORACLE_SID=TEST. But doing so will require us to maintain two copies of the files. And when we change the database name, or upgrade the database, I need to maintain the bash login file separately. It's nice to just use something like
source .cshr in bash, but it doesn't work.
setenv ORACLE_SID TEST
setenv ORACLE_HOME /oracle/TEST/home/products/10204
setenv EPC_DISABLED TRUE
setenv MANPATH /usr/local/man:/usr/share/man
setenv EDITOR vi
setenv LD_LIBRARY_PATH $ORACLE_HOME/lib:/usr/sfw/lib/64
setenv NLS_LANG AMERICAN_AMERICA.UTF8
setenv NLS_DATE_FORMAT "DD-MON-RR"
In your ~/.bashrc (or the first of ~/.bash_profile, ~/.bash_login, and ~/.profile that exists) source this script using something like . ~/bin/sourcecsh:
#!/bin/bash
# This should be sourced rather than executed
while read cmd var val
do
if [[ $cmd == "setenv" ]]
then
eval "export $var=$val"
fi
done < ~/.cshrc
This version eliminates the evil eval:
#!/bin/bash
# This should be sourced rather than executed
# yes, it will be sourcing within sourcing - what so(u)rcery!
source /dev/stdin < \
<(
while read cmd var val
do
if [[ $cmd == "setenv" ]]
then
echo "export $var=$val"
fi
done < cshrc
)
Edit:
Without sourcing stdin:
while read cmd var val
do
if [[ $cmd == "setenv" ]]
then
declare -x "$var=$val"
fi
done < cshrc
How about just defining a function called setenv, like so
setenv() {
echo setting $1 to $2
export $1=$2
}
and then sourcing the .cshrc file.
When I do this in bash, I get
[dws#oxygen ual-read-only]$ source cshrc
setting ORACLE_SID to TEST
setting ORACLE_HOME to /oracle/TEST/home/products/10204
setting EPC_DISABLED to TRUE
setting MANPATH to /usr/local/man:/usr/share/man
setting EDITOR to vi
setting LD_LIBRARY_PATH to /oracle/TEST/home/products/10204/lib:/usr/sfw/lib/64
setting NLS_LANG to AMERICAN_AMERICA.UTF8
setting NLS_DATE_FORMAT to DD-MON-RR
[dws#oxygen ual-read-only]$ env | grep ORACLE
ORACLE_SID=TEST
ORACLE_HOME=/oracle/TEST/home/products/10204
I'm in the same boat. A coworker showed me the following:
Start off in bash, without the stuff in thwe environment:
bash> echo $$
12632
bash> echo $FOO
Here's the csh file that gets source'd:
bash> cat setup-env.csh
setenv FOO "some csh stuff"
echo FOO=$FOO in csh
Here's the command:
bash> csh -c 'source setup-env.csh;exec bash'
Look at the output from csh
FOO=some csh stuff in csh
And look at the output from the new bash shell
bash> echo $$
13487
bash> echo $FOO
some csh stuff
Now leave, and go back to the original bash shell
bash> exit
exit
bash> echo $$
12632
bash>
Note the echo $$ to see the process IDs, so that we can see they are different shell processes.
My coworker uses this enough that he puts it into an alias, like:
# make csh environment scripts useable (sourceable) from bash function
# from Phil McCoy, Wed Nov 9 2011
source_csh () {
exec csh -c " source $*; setenv ALREADY_SOURCED \"$ALREADY_SOURCED:$*:\"; exec bash"
}
# sounds like a great idea to do source_csh .cshrc or .login
# but naively done is infinitely recursive,
# since the exec'ed bash will run .bashrc
Unfortunately, I have found that I often needed not just environment variable setup, but also aliases setup, as in the modules package http://modules.sourceforge.net/.
I have been able to automate this "csh source script recipes" by using Perl Expect. But I have not been able to be as interactive as I would like, except for the above.
Only way I can think to do it would be to load csh and then call bash from that new shell. That way csh could parse that file, and then the bash that it spawns would inherit that environment as well.
In your bash .profile, you can do the following:
cat .cshrc | sed 's/setenv\s+(\S+)\s+(.*)$/set $1=$2; export $1/' > $HOME/.env_from_csh
source $HOME/.env_from_csh
For something that small, it's common to maintain two setup scripts, one for sh and sh-derived, shells, and one for csh and tcsh. As you mention, that does create the risk of the two scripts getting out of sync -- unless you generate one from the other, or generate both from a common source.
This places the burden on the maintainer of the setup script(s) rather than on each user who needs to use them.
There is a module available by which you can source same file where ever you want in perl script. And you will get all environment paths available in your csh file.
Source::Shell
Go through a little documentation for its usage.
Just having a "#!/bin/tcsh" or similar statement available at the start of the CSH script in conjunction with making the script executable solved the problem for me. I could directly call the CSH script from bash in this case.
As an example, I had to run tools.csh from a bash script called setup.sh. I did something like this in the bash script:
if [ -z \`head -1 tools.csh | grep '^#!'\` ];
then
TCSH=\`which tcsh\`;
echo "'#!'$TCSH" > tools.csh.temp;
cat tools.csh >> tools.csh.temp;
mv tools.csh.temp tools.csh;
fi;
chmod 755 tools.csh;
./tools.csh
# now I have all the setenv commands effective ...
Sourcing a csh file in bash will not work. You can change the default login shell from csh to bash if you are more confortable in bash. You could use chsh or as an admin to change it for you.
chsh -s /usr/local/bin/bash