Install python package in docker file

Question

In my docker file, I want to install med2image python package (https://github.com/FNNDSC/med2image). I use the following code:

FROM ubuntu:16.04
RUN apt-get update && apt-get install -y --no-install-recommends \
    python3.5 \
    python3-pip \
    && \
apt-get clean && \
rm -rf /var/lib/apt/lists/*
RUN pip install nibabel pydicom matplotlib pillow
RUN pip install med2image

But I get the following error when I want to build the image:

Downloading https://files.pythonhosted.org/packages/6f/e5/948b023c7feb72adf7dfb26d90a13c838737bbf52be704f5ddd0878e3264/med2image-1.1.2.tar.gz
Complete output from command python setup.py egg_info:
Sorry, only Python 3.5+ is supported.

----------------------------------------
Command "python setup.py egg_info" failed with error code 1 in  /tmp/pip-install-FnNb_S/med2image/
The command '/bin/sh -c pip install med2image' returned a non-zero code: 1

What should I do?!

Answer 1

Recommended base image

As suggested in my comment, you could write a Dockerfile that looks like:

FROM python:3

RUN pip install --no-cache-dir --upgrade pip && \
    pip install --no-cache-dir nibabel pydicom matplotlib pillow med2image
    # Note: we had to merge the two "pip install" package lists here, otherwise
    # the last "pip install" command in the OP may break dependency resolution…

CMD ["cat", "/etc/os-release"]

And the command example above could confirm at runtime (docker build --pull -t test . && docker run --rm -it test) that this image is based on the GNU/Linux distribution "Debian stable".

Generic Dockerfile template

Finally to give a comprehensive answer, note that a good practice regarding Python dependencies consists in specifying them in a declarative way in a dedicated text file (in alphabetical order, to ease review and update) so that for your example, you may want to write the following file:

requirements.txt

matplotlib
med2image
nibabel
pillow
pydicom

and use the following generic Dockerfile

FROM python:3

WORKDIR /usr/src/app

COPY requirements.txt ./

RUN pip install --no-cache-dir --upgrade pip \
  && pip install --no-cache-dir -r requirements.txt

COPY . .

CMD ["python", "./your-daemon-or-script.py"]

To be more precise, this is the approach suggested in the documentation of the Docker official image python, §. How to use this image

Answer 2

Some of the other answers/comments are suggesting to change your base image but if you want to keep your ubuntu 16.04 you can also simply specify your version of pip/python to use pip3 or pip3.5 like shown below.

FROM ubuntu:16.04

RUN apt-get update && apt-get install -y --no-install-recommends \
    python3.5 \
    python3-pip \
    && \
    apt-get clean && \
    rm -rf /var/lib/apt/lists/*

RUN pip3 install nibabel pydicom matplotlib pillow
RUN pip3 install med2image

Answer 3

Try pip3

RUN pip3 install nibabel pydicom matplotlib pillow
RUN pip3 install med2image

Answer 4

If you have Jupyter Notebook in your docker container, you can install any python package by running a new Terminal in Jupyter by clicking the button shown here:

and running: pip install <package-name>

The package stays in the docker container even after you exit the container.