Dockerfile

# Based on OpenAI's mujoco-py Dockerfile

ARG USE_MPI=True

# base stage contains just binary dependencies.
# This is used in the CI build.
FROM nvidia/cuda:10.0-runtime-ubuntu18.04 AS base
ARG USE_MPI
ARG DEBIAN_FRONTEND=noninteractive

RUN echo ttf-mscorefonts-installer msttcorefonts/accepted-mscorefonts-eula select true | debconf-set-selections \
    && apt-get update -q \
    && apt-get install -y --no-install-recommends \
    build-essential \
    curl \
    ffmpeg \
    git \
    libgl1-mesa-dev \
    libgl1-mesa-glx \
    libglew-dev \
    libosmesa6-dev \
    net-tools \
    parallel \
    patchelf \
    python3.7 \
    python3.7-dev \
    python3-pip \
    rsync \
    software-properties-common \
    unzip \
    vim \
    virtualenv \
    xpra \
    xserver-xorg-dev \
    ttf-mscorefonts-installer \
    && apt-get clean \
    && rm -rf /var/lib/apt/lists/*

ENV LANG C.UTF-8

RUN    mkdir -p /root/.mujoco \
    && curl -o mjpro150.zip https://www.roboti.us/download/mjpro150_linux.zip \
    && unzip mjpro150.zip -d /root/.mujoco \
    && rm mjpro150.zip \
    && curl -o mujoco131.zip https://www.roboti.us/download/mjpro131_linux.zip \
    && unzip mujoco131.zip -d /root/.mujoco \
    && rm mujoco131.zip \
    && curl -o /root/.mujoco/mjkey.txt https://www.roboti.us/file/mjkey.txt

COPY vendor/Xdummy /usr/local/bin/Xdummy
RUN chmod +x /usr/local/bin/Xdummy

RUN if [ $USE_MPI = "True" ]; then \
    add-apt-repository --yes ppa:marmistrz/openmpi \
    && apt-get update -q \
    && apt-get install -y libopenmpi3 libopenmpi-dev \
    && apt-get clean \
    && rm -rf /var/lib/apt/lists/*; \
    fi

# Set the PATH to the venv before we create the venv, so it's visible in base.
# This is since we may create the venv outside of Docker, e.g. in CI
# or by binding it in for local development.
ENV PATH="/venv/bin:$PATH"
ENV LD_LIBRARY_PATH /root/.mujoco/mjpro150/bin:${LD_LIBRARY_PATH}

# python-req stage contains Python venv, but not code.
# It is useful for development purposes: you can mount
# code from outside the Docker container.
FROM base as python-req
ARG USE_MPI

WORKDIR /adversarial-policies
# Copy over just requirements.txt at first. That way, the Docker cache doesn't
# expire until we actually change the requirements.
COPY ./requirements-build.txt /adversarial-policies/
COPY ./requirements.txt /adversarial-policies/
COPY ./requirements-dev.txt /adversarial-policies/
COPY ./ci/build_venv.sh /adversarial-policies/ci/build_venv.sh
RUN  ci/build_venv.sh /venv && rm -rf $HOME/.cache/pip

# full stage contains everything.
# Can be used for deployment and local testing.
FROM python-req as full

# Delay copying (and installing) the code until the very end
COPY . /adversarial-policies
# Build a wheel then install to avoid copying whole directory (pip issue #2195)
RUN python3 setup.py sdist bdist_wheel
RUN pip install --upgrade dist/aprl-*.whl

# Default entrypoints
ENTRYPOINT ["/adversarial-policies/vendor/Xdummy-entrypoint"]
CMD ["ci/run_tests.sh"]