backend/Dockerfile

# syntax=docker/dockerfile:1
# ---------------------------------------------------------------------------
# Production image for the Document-Agent FastAPI backend.
#
# It bundles Docling + (CPU) torch + the layout/table/OCR models so the
# container is self-contained: no model downloads at runtime, fast cold start,
# works offline. On Linux the HuggingFace symlink cache works normally, so the
# Windows WinError-1314 issue does not apply here.
# ---------------------------------------------------------------------------
FROM python:3.12-slim

ENV PYTHONUNBUFFERED=1 \
    PYTHONDONTWRITEBYTECODE=1 \
    PIP_NO_CACHE_DIR=1 \
    PIP_DISABLE_PIP_VERSION_CHECK=1 \
    HF_HUB_DISABLE_SYMLINKS_WARNING=1 \
    # Fixed, shared cache locations so the non-root runtime user finds the
    # models baked in during build.
    HF_HOME=/opt/models/hf \
    EASYOCR_MODULE_PATH=/opt/models/easyocr \
    TORCH_HOME=/opt/models/torch \
    DATA_DIR=/data

# Native libs needed by docling / opencv / easyocr.
RUN apt-get update && apt-get install -y --no-install-recommends \
        libgl1 \
        libglib2.0-0 \
        libgomp1 \
    && rm -rf /var/lib/apt/lists/*

WORKDIR /app

# Install CPU-only torch/torchvision FIRST so docling's torch dependency
# resolves to the lightweight CPU wheels instead of multi-GB CUDA builds.
RUN pip install --index-url https://download.pytorch.org/whl/cpu \
        torch torchvision

COPY requirements.txt .
RUN pip install -r requirements.txt

# Bake the models into the image.
COPY scripts/prefetch_models.py scripts/prefetch_models.py
RUN mkdir -p /opt/models && python scripts/prefetch_models.py

# App source.
COPY . .

# Run as a non-root user; /data holds uploads, sqlite db and rendered pages and
# is expected to be a mounted persistent volume in production.
RUN useradd --create-home app \
    && mkdir -p /data \
    && chown -R app:app /data /app /opt/models
USER app

EXPOSE 8000

# Honour $PORT (Render/most PaaS inject it); default 8000 for local/compose.
HEALTHCHECK --interval=30s --timeout=5s --start-period=60s --retries=3 \
    CMD python -c "import os,urllib.request,sys; sys.exit(0 if urllib.request.urlopen('http://localhost:%s/api/health' % os.environ.get('PORT','8000')).status==200 else 1)"

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