Source

Dockerizing Python Poetry Applications

Poetry meets Docker

Poetry is one of the new and more elegant dependency management tools Python has at its disposal. In my opinion, its the unofficial successor to pipenv which hasn’t had a major update in a while and is probably dead? Being a relatively new tool, there isn’t much documentation out there regarding dockerizing an application using Poetry. It’s actually a fairly simple process but took me a while to get working due to the lack of documentation. I recently dockerized my steam game recommendations engine which uses embeddings to make recommendations. You can play around with it here. I’ll be using this project’s Docker file as an example:

FROM python:3.7
RUN mkdir /app 
COPY /app /app
COPY pyproject.toml /app 
WORKDIR /app
ENV PYTHONPATH=${PYTHONPATH}:${PWD} 
RUN pip3 install poetry
RUN poetry config virtualenvs.create false
RUN poetry install --no-dev

As you can see its a fairly simple file but there a few lines which are critical for dockerizing this application:

COPY pyproject.toml /app

This line copies the dependencies of my app into the working directory of my container. Make sure this is done otherwise:

1. Poetry will think this is a new project
2. You will have to install the dependencies again individually

Next, the PYTHONPATH was set and poetry installed with pip. The following line is the most important line in this dockerfile:

RUN poetry config virtualenvs.create false

This line ensures when packages are installed with Poetry a virtual environment is NOT created first. You’re already in a virtual environment by using a docker image, there's no need to build another environment unless you really want to inception your application of course.

Finally the last line:

RUN poetry install --no-dev

Installs all the required packages minus the development packages, also don’t forget to add your application’s Entrypoint if it needs one. That’s it! You now have a dockerized Poetry application, hopefully, this has saved you plenty of time compared to the hours of hair pulling it cost me.