As you develop more and more Python code for data science tasks, you may find yourself using the same code (functions) over and over again (e.g., a customized eda function). Instead of copying and pasting code snippets to your current Jupyter notebook, there are slicker ways to call Python functions from anywhere on your computer without having to specify the path, which I’ll describe in this blog.
Note: Methods described here are applicable to UNIX systems (Mac or Linux).
Save the function(s) in a .py file.
This can be done using any text editors, such as Atom or gedit; or IDEs, such as Spyder or PyCharm.
In the terminal, go to your home directory (cd and press “enter”).
Run ls -a to see all the files in your home directory, including hidden files. There should be a file named .bashrc.
Open it with an text editor of your choosing. At the end of the document, add the following line:
export PYTHONPATH="<path to directory containing your Python script>:$PYTHONPATH"
For example, mine is :
export PYTHONPATH="/home/fay/code/data_science_general:$PYTHONPATH"
Putting :$PYTHONPATH at the end makes sure that you append to PYTHONPATH, and not overwrite it.
For this to take effect, run source ~/.bashrc or simply log out and then log in again.
Note: .bashrc is read at log-in, that’s why you need to source it if you don’t want to log in again.
Now, to use the function(s) in your .py file(s), simply run import <script> in your Jupyter notebook.
For example, I have my functions saved in eda.py, all I need to do is to add import eda or from eda import * to my Jupyter notebook.
If you are ready to package your script(s) and share it with the world, you can register it with PyPI, which you (or anyone else) can pip install in the system (or a virtual environment), and then import it like you would any other Python libraries. You can find more details at the resources below: