Dockerfile

# Use the official lightweight Python image.
# https://hub.docker.com/_/python
# https://docs.astral.sh/uv/guides/integration/docker/

FROM python:3.12-slim-bookworm

# Allow statements and log messages to immediately appear in the Knative logs
ENV PYTHONUNBUFFERED True

# Download latest listing of available packages:
RUN apt-get -y update
# Upgrade already installed packages:
RUN apt-get -y upgrade
# Install new packages:
RUN apt-get -y install git libpangocairo-1.0-0

# install uv
COPY --from=ghcr.io/astral-sh/uv:latest /uv /uvx /bin/

# Copy local code to the container image.
ENV APP_HOME /app
WORKDIR $APP_HOME
COPY . ./

# Install runtime dependencies plus the docs group (needed to build the
# static doc site below); dev/spectemplates groups are dev-only and are
# excluded to keep the production image lean.
RUN uv sync --frozen --no-default-groups --group docs

# `uv run` (unlike this initial `uv sync`) re-syncs the environment against
# uv's implicit default groups (which include "dev") every time it's
# invoked, silently re-adding ruff/black/ty/pytest to the image. Put the
# venv on PATH instead so later steps call tools directly without
# triggering another sync.
ENV PATH="$APP_HOME/.venv/bin:$PATH"

# Create a temporary directory
RUN mkdir tmp

# Build documentation
RUN cd docs && mkdocs build

# Run the web service on container startup.
# ETC_N_WORKERS sets the ETC engine's ThreadPoolExecutor worker count.
# OMP_NUM_THREADS no longer drives the ETC itself (pfsspecsim v2 is pure
# Python); it is kept to cap the BLAS/OpenMP pools underneath numpy, which
# would otherwise spawn one thread per core inside each ETC worker.
ENV ETC_N_WORKERS=8
ENV OMP_NUM_THREADS=8
CMD panel serve ./app.py --address 0.0.0.0 --port 8080 --allow-websocket-origin="*" --static-dirs doc="./docs/site/"

# TODO
# - Remove temporary files in tmp directory periorically using crontab.
#   https://www.airplane.dev/blog/docker-cron-jobs-how-to-run-cron-inside-containers