Dockerfile.render

# Lean single-process image for Render: the embedded `serve` pipeline (no broker), the
# React dashboard, and a model trained at build. momentum + ml_model are the live, graded
# strategies; the LLM strategies stay inert without API credit.

# Stage 1 — build the React dashboard
FROM node:24-slim AS frontend
WORKDIR /app/frontend
COPY frontend/package.json ./
RUN npm install
COPY frontend/ ./
RUN npm run build

# Stage 2 — the Python app (only what embedded serve + ml_model need)
FROM python:3.12-slim
WORKDIR /app
ENV PYTHONUNBUFFERED=1 \
    PIP_NO_CACHE_DIR=1 \
    RTA_FRONTEND_DIST=/app/frontend/dist \
    RTA_SOURCE=rest
# ca-certificates for HTTPS to Binance.US; libgomp1 is LightGBM's OpenMP runtime.
RUN apt-get update \
    && apt-get install -y --no-install-recommends ca-certificates libgomp1 \
    && rm -rf /var/lib/apt/lists/*
COPY pyproject.toml README.md ./
COPY src ./src
RUN pip install -e ".[ingest,serve,ml,lakehouse]"
COPY --from=frontend /app/frontend/dist ./frontend/dist
# Fail the build early if the app can't import; then train the ml_model (best-effort —
# if Binance is unreachable at build time, ml_model simply stays inert).
RUN python -c "from realtime_alpha.serving.app import create_app; create_app(start_pipeline=False)" \
    && (realtime-alpha train --limit 600 --out models/ml_model.pkl || echo "train skipped; ml_model inert")
EXPOSE 8000
# Render injects $PORT; bind to it.
CMD ["sh", "-c", "realtime-alpha serve --host 0.0.0.0 --port ${PORT:-8000}"]