Dockerfile-analyses

# Builds an image for reproducing the efficient-learning-khan analyses.
# Use Dockerfile-experiment to build an environment for running the experiment.
FROM debian:buster-slim

LABEL maintainer="Paxton Fitzpatrick <Paxton.C.Fitzpatrick.GR@Dartmouth.edu>"

ARG notebook_ip=0.0.0.0
ARG notebook_port=8888
ARG notebook_dir="/mnt/code/notebooks"
ARG DEBIAN_FRONTEND=noninteractive

ENV MAMBA_NO_BANNER=1 \
    NOTEBOOK_IP="$notebook_ip" \
    NOTEBOOK_PORT="$notebook_port" \
    NOTEBOOK_DIR="$notebook_dir" \
    PATH="/opt/conda/bin:$PATH"

COPY code/khan_helpers docker/jupyter_notebook_config.py docker/analysis-environment.yml /khan_helpers/

SHELL ["/bin/bash", "-o", "pipefail", "-c"]

RUN apt-get update --fix-missing \
    && apt-get install -y --no-install-recommends eatmydata \
    # temporarily install libraries needed to download & build Python
    # requirements, plus a mini text editor for convenience
    && eatmydata apt-get install -y --no-install-recommends \
        ca-certificates \
        curl \
        gcc \
        git \
        libc6-dev \
        vim-tiny \
    # install mamba (faster C++ conda reimplementation, uses conda-forge)
    && curl --progress-bar \
        -L "https://github.com/conda-forge/miniforge/releases/download/4.11.0-0/Mambaforge-4.11.0-0-Linux-$(uname -m).sh" \
        -o "$HOME/mambaforge.sh" \
    && /bin/bash "$HOME/mambaforge.sh" -bp /opt/conda \
    && rm "$HOME/mambaforge.sh" \
    # install required packages from environment file \
    && mamba env update -f /khan_helpers/analysis-environment.yml --prune \
    # install additional requirements not available via conda
    # wordcloud must be built from source for ARM/Python>3.7
    && pip install --disable-pip-version-check --no-cache-dir \
        hypertools==0.8.0 \
        youtube_transcript_api==0.4.4 \
        git+https://github.com/amueller/word_cloud.git@35ce9b7 \
    # set up jupyter notebook config
    && mkdir "$HOME/.jupyter" \
    && mv "/khan_helpers/jupyter_notebook_config.py" "$HOME/.jupyter/" \
    # download NLTK corpora & models
    && python -uBW ignore -m nltk.downloader \
        stopwords \
        averaged_perceptron_tagger \
        wordnet \
        omw-1.4 \
    # install khan_helpers package (NLTK data must be downloaded first)
    && pip install --disable-pip-version-check --no-cache-dir /khan_helpers \
    # various cleanup to reduce image size
    && rm -rf /khan_helpers \
    && mamba clean -afy \
    && apt-get purge -y --auto-remove  \
        ca-certificates \
        curl \
        gcc \
        git \
        libc6-dev \
    && apt-get clean \
    && rm -rf /var/lib/apt/lists/* \
    && rm -f "$HOME/nltk_data/*/*.zip"

WORKDIR "/mnt"

CMD ["jupyter", "notebook"]