Finding Your Bash Script’s Location: $0 vs ${BASH_SOURCE[0]}

When a Bash script needs to reference its own location — for sourcing additional files, logging, or managing relative paths — $0 seems like the obvious choice. But it’s a trap: $0 behaves differently depending on whether your script is executed directly or sourced.

The Problem with $0

When you execute a script directly:

./script.sh

$0 contains the script’s path. But when you source it:

source script.sh
# or
. script.sh

$0 no longer contains the sourced script’s path — it contains the path of the calling shell or parent script instead. This breaks any logic that relies on knowing the script’s actual location.

The Solution: ${BASH_SOURCE[0]}

Use ${BASH_SOURCE[0]} instead. It reliably returns the path of the current script regardless of whether it’s sourced or executed directly.

#!/bin/bash

SCRIPT_PATH="${BASH_SOURCE[0]}"
SCRIPT_DIR="$(cd "$(dirname "${SCRIPT_PATH}")" && pwd)"

echo "Script is at: ${SCRIPT_PATH}"
echo "Script directory: ${SCRIPT_DIR}"

Why ${BASH_SOURCE[0]} Works

BASH_SOURCE is an array variable where each element contains the filename of a sourced script or the main script. The index tracks the call stack:

• ${BASH_SOURCE[0]} = current script file
• ${BASH_SOURCE[1]} = script that sourced the current one
• ${BASH_SOURCE[2]} = script that sourced that one, and so on

This works inside functions too. When a function is defined in a sourced script, ${BASH_SOURCE[0]} still points to the file where the function was defined, not where it was called from.

Practical Example

Consider two scripts where b.sh is sourced by a.sh:

a.sh:

#!/bin/bash
echo "a.sh path: ${BASH_SOURCE[0]}"
source ./b.sh

b.sh:

#!/bin/bash
echo "b.sh path: ${BASH_SOURCE[0]}"

get_config() {
    local config_dir="$(dirname "${BASH_SOURCE[0]}")/config"
    echo "Config location: ${config_dir}"
}

get_config

Running ./a.sh produces:

a.sh path: ./a.sh
b.sh path: ./b.sh
Config location: ./config

Sourcing it with . a.sh produces the same output. With $0, you’d get different results depending on how the script was invoked.

Getting the Absolute Path

For scripts that might be called from different directories, get the absolute path:

SCRIPT_PATH="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)/$(basename "${BASH_SOURCE[0]}")"

Or use readlink if symlink resolution matters:

SCRIPT_PATH="$(readlink -f "${BASH_SOURCE[0]}")"

When to Use Each Variable

• ${BASH_SOURCE[0]}: You want the path of the currently executing script file
• $0: You’re documenting program invocation, or you need the command the user actually typed
• ${BASH_SOURCE[@]}: Debugging — prints the entire call stack of sourced files

The bottom line: for any script logic that depends on the script’s own location, reach for ${BASH_SOURCE[0]} and avoid the pitfalls of $0.