In a Windows command script, one can determine the directory path of the currently executing script using %~dp0. For example:
@echo Running from %~dp0
What would be the equivalent in a Bash script?
Add a comment
|
5 Answers
For the relative path (i.e. the direct equivalent of Windows' %~dp0):
MY_PATH="$(dirname -- "${BASH_SOURCE[0]}")"
echo "$MY_PATH"
For the absolute, normalized path:
MY_PATH="$(dirname -- "${BASH_SOURCE[0]}")" # relative
MY_PATH="$(cd -- "$MY_PATH" && pwd)" # absolutized and normalized
if [[ -z "$MY_PATH" ]] ; then
# error; for some reason, the path is not accessible
# to the script (e.g. permissions re-evaled after suid)
exit 1 # fail
fi
echo "$MY_PATH"
14 Comments
vladr
what is the problem on your mac bash? works just fine over here on Cygwin, Linux, Solaris, etc., and it must also work on mac
JamesThomasMoon
Within bash you can also use
$BASH_SOURCE. From this post.
Dmitri Sologoubenko
$0 would be OK in most cases, some exceptions are, for instance, when the script you're executing is aliased (through alias in .bash_profile). You should really use $BASH_SOURCE variable, instead of $0.
shaunc
Doesn't work for me when I execute a script via a symlink. Is there a variant which will determine the fully-resolved script directory? EDIT:
$(dirname $(readlink $0)) seems to work
Christopher Schultz
Using
$0 does not work when the script is run using source script or . script; the name of the script is not available. |
Assuming you type in the full path to the bash script, use $0 and dirname, e.g.:
#!/bin/bash
echo "$0"
dirname "$0"
Example output:
$ /a/b/c/myScript.bash
/a/b/c/myScript.bash
/a/b/c
If necessary, append the results of the $PWD variable to a relative path.
EDIT: Added quotation marks to handle space characters.
answered Mar 10, 2009 at 14:17
2 Comments
vladr
@Rothko, your solution will fail if $0 contains blanks.
JamesThomasMoon
Within bash you can also use
$BASH_SOURCE. From this post.Contributed by Stephane CHAZELAS on c.u.s. Assuming POSIX shell:
prg=$0
if [ ! -e "$prg" ]; then
case $prg in
(*/*) exit 1;;
(*) prg=$(command -v -- "$prg") || exit;;
esac
fi
dir=$(
cd -P -- "$(dirname -- "$prg")" && pwd -P
) || exit
prg=$dir/$(basename -- "$prg") || exit
printf '%s\n' "$prg"
answered Mar 10, 2009 at 14:58
5 Comments
ams
brilliant works for me on Mac OSX the other approaches in this question did not.
vladr
+1 for dealing with script invocation via
$PATH search; however, the above is NOT POSIX. It will only work with bash. Use which instead of command and backticks instead of $(...) is this has to run under other, older shells.
Dimitre Radoulov
Hi @vladr, the above code is POSIX! If I understand correctly, what you mean is that the code won't run with the old Bourne shell (/bin/sh on Solaris < 11 for instance), but that has nothing to do with POSIX.
vladr
@DimitreRadoulov sorry yes I meant the lowest common-denominator POSIX, for most portability, i.e. POSIX.1. Technically yes, any POSIX.2 shell as well as old
ksh etc., not just bash, will run the above, but quite a few people out there are still on Solaris 10 (end-of-support is 2018, quite a few more years to go), and some even run old AIX and HP-UX.Dimitre Radoulov
@vladr, Bourne shell is not POSIX compliant and it predates the POSIX standards. The standardized user command line and scripting interface of POSIX.1 were based on the Korn Shell. As far as I know all commercial Unices offer a POSIX compliant shell (it's /usr/xpg4/bin/sh on Solaris for example).
echo Running from $(dirname "$0")
Comments
Vlad's code is overquoted. Should be:
MY_PATH=`dirname "$0"`
MY_PATH=`( cd "$MY_PATH" && pwd )`
3 Comments
Eric Seppanen
Your second line has an extra double-quote character.
vladr
The overquoting was needed on older Solaris versions, which were still being used at the time of the original post.
tripleee
Surely the parentheses are unnecessary even on Solaris?