Dockerfile

# 1) Python 3.11
FROM python:3.11-slim

# 2) System deps & git-lfs
RUN pip install --no-cache-dir hf_transfer>=0.1.6 && \
    apt-get update && apt-get install -y \
      build-essential \
      curl \
      git \
      git-lfs \
    && rm -rf /var/lib/apt/lists/*

# 3) Install Ollama
RUN curl -fsSL https://ollama.com/install.sh | sh

# 4) Ollama storage → /tmp (writeable in Spaces)
ENV OLLAMA_HOME=/tmp/ollama
ENV OLLAMA_MODELS=/tmp/ollama
ENV OLLAMA_MODEL=gemma2:9b
RUN install -d -m 777 /tmp/ollama

# 5) App setup
WORKDIR /app
COPY requirements.txt .
RUN pip install --no-cache-dir -r requirements.txt
COPY . /app
RUN git lfs install && git lfs pull || true

# 6) Streamlit & caches
ENV HOME=/app
ENV STREAMLIT_HOME=/app/.streamlit
RUN install -d -m 777 /app/.streamlit

ENV HF_HOME=/tmp/hf-home \
    TRANSFORMERS_CACHE=/tmp/hf-cache \
    HUGGINGFACE_HUB_CACHE=/tmp/hf-cache \
    TORCH_HOME=/tmp/torch-cache \
    XDG_CACHE_HOME=/tmp/xdg-cache
RUN install -d -m 777 /tmp/hf-home /tmp/hf-cache /tmp/torch-cache /tmp/xdg-cache

# 7) Internal Ollama host (same container)
ENV OLLAMA_HOST=http://127.0.0.1:11434
ENV OLLAMA_TIMEOUT=300

EXPOSE 8501

# 8) Runtime: clean /tmp/ollama → serve → healthcheck → pull gemma2:9b → Streamlit
CMD bash -lc '\
  set -euo pipefail; \
  rm -rf /tmp/ollama && install -d -m 777 /tmp/ollama; \
  echo "Starting Ollama with model: ${OLLAMA_MODEL}"; \
  env HOME=/tmp OLLAMA_HOME=/tmp/ollama OLLAMA_MODELS=/tmp/ollama ollama serve & \
  for i in {1..240}; do curl -sf http://127.0.0.1:11434/api/version >/dev/null && break || sleep 1; done; \
  env HOME=/tmp OLLAMA_HOME=/tmp/ollama OLLAMA_MODELS=/tmp/ollama ollama pull "${OLLAMA_MODEL}"; \
  exec streamlit run app.py --server.address=0.0.0.0 --server.port=${PORT:-8501}'