# cpu-basic on HF Spaces: lightweight Python image, no CUDA, no llama.cpp build.
# Witness inference runs via HF Inference Providers (huggingface_hub.InferenceClient).
# For local play with llama.cpp, see "Run locally" in README.md.
FROM python:3.12-slim

ENV DEBIAN_FRONTEND=noninteractive \
    PYTHONUNBUFFERED=1 \
    PIP_NO_CACHE_DIR=1 \
    PYTHONPATH=/app \
    WITNESS_BACKEND=providers \
    PORT=7860 \
    HOST=0.0.0.0

WORKDIR /app

COPY requirements.txt ./
RUN pip install --upgrade pip && pip install -r requirements.txt

COPY . .
RUN mkdir -p models data

EXPOSE 7860
CMD ["python3", "-m", "dempsters_court.server"]