If you bash script then it is a good idea to make sure it does not matter where you put the script and what is your current working directory when you run it.

If you have a large project written using bash then it is even more important to make it directory independent. So developers can have the whole project in any location on their disk and it still functions.

In such case, however, how does a script find other files that come with the project? e.g. Imagaes or data files.

It is easy. Use relative pathes.

Let's say this is the directory structure of your project: 

project/
├── bin
│   └── code.sh
└── data
    └── names.txt

How can the "code.sh" reliably find the "names.txt" file? It needs to find its own directory. Then go one directory up and then enter the "data" directory.

We can use the realpath and dirname command as we saw them in the page about absolute path.

We have a similar case in the code-maven directory tree. It looks like this, just a lot bigger: 

.
├── examples
│   ├── data
│   │   ├── airports.csv
│   │   └── words_to_sort.txt
│   ├── shell
│   │   ├── relative.sh
│   │   ├── relative_one.sh
│   │   ├── relative_project.sh

The point is, however, that getting from the relative.sh we need to go one directory up and then down the "data" direcotry.

This script implements it step by step using realpath and then dirname:

examples/shell/relative.sh

#!/bin/bash

echo $0

full_path=$(realpath $0)
echo $full_path

dir_path=$(dirname $full_path)
echo $dir_path

examples=$(dirname $dir_path )
echo $examples

data_dir="$examples/data"
echo "DATA: $data_dir"





./examples/shell/relative.sh
/home/gabor/work/code-maven.com/examples/shell/relative.sh
/home/gabor/work/code-maven.com/examples/shell
/home/gabor/work/code-maven.com/examples
DATA: /home/gabor/work/code-maven.com/examples/data

In the following example we do it in one step:

examples/shell/relative_one.sh

#!/bin/bash

data_dir="$(dirname $(dirname $(realpath $0)) )/data"
echo "DATA: $data_dir"



DATA: /home/gabor/work/code-maven.com/examples/data

In the last example we split it into two steps and went a bit further.

In the first step we calculate the path to the root of the project, the root of our git repository. (In our case it is the parent directory of the "examples" directory, that is 2 steps above the "shell" directory so we call dirname 3 times. Once to get rid of the filename and then twice to move up the directory tre to the root of the project. Then from there we can add the path of each file or directory relative to the root of the it working directory.

examples/shell/relative_project.sh

#!/bin/bash

project_root=$(dirname $(dirname $(dirname $(realpath $0 ))))
echo $project_root

data_dir="$project_root/examples/data"
echo "DATA: $data_dir"




/home/gabor/work/code-maven.com
DATA: /home/gabor/work/code-maven.com/examples/data

As you can from the results, no matter which version you pick, it can point to the directory where the data is located.