# Dockerfile for "conda-pack-style" sample images.
# Image only contains the activated conda env + system toolchain.
# Sample code + pipeline are mounted at runtime via -v.
#
# Build args:
#   PACK_FILE     e.g. 001_rm_bench_test.tar.gz   (file in build context)
#   ENV_DIR_NAME  e.g. py-001                      (folder under /opt/conda/envs)
#
# Build (on remote, from /home/dockerfiles/):
#   docker buildx build \
#     --build-arg PACK_FILE=001_rm_bench_test.tar.gz \
#     --build-arg ENV_DIR_NAME=py-001 \
#     -t mmsci-py-001 -f Dockerfile.conda-env .
#
# Run:
#   docker run --rm -it --gpus all \
#     -v /home/code:/workspace/code:ro \
#     -v /home/code/data:/workspace/data:ro \
#     mmsci-py-001 bash -lc 'python -V'

FROM nvidia/cuda:11.8.0-cudnn8-runtime-ubuntu22.04

ARG PACK_FILE
ARG ENV_DIR_NAME

ENV DEBIAN_FRONTEND=noninteractive \
    LANG=C.UTF-8 LC_ALL=C.UTF-8 \
    TZ=Etc/UTC

RUN apt-get update && apt-get install -y --no-install-recommends \
        ca-certificates curl wget git build-essential gcc g++ make \
        libgl1 libglib2.0-0 libsm6 libxext6 libxrender1 libgomp1 \
        libstdc++6 procps less vim-tiny \
    && rm -rf /var/lib/apt/lists/*

RUN mkdir -p /opt/conda/envs/${ENV_DIR_NAME}

COPY ${PACK_FILE} /tmp/env.tar.gz
RUN tar -xzf /tmp/env.tar.gz -C /opt/conda/envs/${ENV_DIR_NAME} \
 && rm /tmp/env.tar.gz \
 && /opt/conda/envs/${ENV_DIR_NAME}/bin/conda-unpack \
 && echo "env unpacked: ${ENV_DIR_NAME}"

WORKDIR /workspace

ENV PATH=/opt/conda/envs/${ENV_DIR_NAME}/bin:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin
ENV CONDA_PREFIX=/opt/conda/envs/${ENV_DIR_NAME}
ENV LD_LIBRARY_PATH=/opt/conda/envs/${ENV_DIR_NAME}/lib

CMD ["bash", "-lc", "python -V && which python && echo READY: $CONDA_PREFIX"]