Dockerfile

# PrimaryDraft: one image, backend serves the built frontend.
#
# Deliberately does NOT ship the annual-report PDF cache (859 MB). The extracted
# records are 620 KB and the pipeline runs entirely from them; PDFs are only
# needed to RE-extract, and live ingest downloads what it needs on demand.

# ---- stage 1: build the UI -------------------------------------------------
FROM node:22-slim AS ui
WORKDIR /ui
COPY app/frontend/package*.json ./
RUN npm ci --no-audit --no-fund
COPY app/frontend/ ./
RUN npm run build

# ---- stage 2: runtime ------------------------------------------------------
FROM python:3.12-slim
ENV PYTHONUNBUFFERED=1 PIP_NO_CACHE_DIR=1 PORT=8000

WORKDIR /app
COPY app/backend/pyproject.toml ./
COPY app/backend/primarydraft ./primarydraft
RUN pip install --no-cache-dir \
      "fastapi>=0.115" "uvicorn[standard]>=0.32" "pydantic>=2.9" "httpx>=0.27" \
      "pdfplumber>=0.11" "python-docx>=1.1" "python-multipart>=0.0.12"

# The BSE scrip master (1.8 MB) so search works on the first request, and
# instantly, even if the exchange is slow or blocks the host.
# Shipped as seed_cache/ because the Hub refuses paths under .cache/, then
# mapped back to the runtime cache location inside the image.
COPY app/backend/seed_cache/bse/scrip_master.json ./.cache/bse/scrip_master.json
COPY app/backend/seed_cache/mca ./.cache/mca

COPY --from=ui /ui/dist ../frontend/dist

# Writable for run snapshots, exports and any live ingest.
RUN mkdir -p .runs .exports .cache/bse && chmod -R 777 .runs .exports .cache

EXPOSE 8000
CMD ["sh","-c","uvicorn primarydraft.main:app --host 0.0.0.0 --port ${PORT:-8000}"]