Dockerfile-backend

FROM python:3.10-bullseye

# Geneplexus Job Runner
# this is the Dockerfile to build an image that will run the model and save html results
# it uses the same Python code as the application
# but is a different Python and set of libs for speed
# the main difference is that this image/container does not run a webserver, 
# it only runs the "runner.sh" shell script that uses environment variables to 
# the parameters for the model 
# those environment variables are set when the docker container is run 
# ARG CONDA_VERSION=4.9.2
ARG PYTHON_VERSION=3.10
ARG APP_POST_URL_ARG="https://geneplexusdev1.azurewebsites.net/jobs"
ENV USERNAME=dockeruser

RUN apt-get update --fix-missing && \
    apt-get install -y apt-utils wget bzip2 curl zip && \
    apt-get install -y fuse && \
    apt-get clean -y && \
    rm -rf /var/lib/apt/lists/*

RUN useradd --create-home $USERNAME && \
    echo PATH=$PATH:/home/$USERNAME/.local/bin >> /home/$USERNAME/.bashrc

WORKDIR /home/$USERNAME
USER $USERNAME

COPY mljob/requirements.txt requirements.txt

# install packages using conda, one at a time
# the requirements file is in pip format, not conda, so need to install one a time 
RUN pip install -r requirements.txt

COPY mljob mljob
COPY runner.py .
COPY runner.sh .

# this needs to be set for the runner to know how to POST status updates.  
ENV APP_POST_URL=$APP_POST_URL_ARG  

# note about back-end data files
# the data files needed by the model are not copied into this container
# but the scripts assume the files are mounted and available to the container
# based on the DATA_PATH env variable, set when the the container is run

# RUN CMD
# this command launches the python model runner shell script 
# which reads environment variables 
# invoking this container requires setting env variables on the command line
# see runner.sh for those
CMD cd /home/dockeruser && ./runner.sh

# no entry point is needed since the container runs a single script ('job')
ENTRYPOINT []