Dockerfile

# Vector DB API — build and run in Docker or Podman; uv for dependencies.
# All dependency installation (uv sync) happens inside Docker during build.
# Use: docker build -t vector-db-api .  OR  podman build -t vector-db-api .

FROM ghcr.io/astral-sh/uv:python3.12-bookworm-slim AS builder

WORKDIR /app

# Set to 1 to add Image embedder (CLIP/ViT). Default 0 = no image embedder.
ARG INSTALL_EMBEDDING_EXTRAS=0

# Install deps: pdf (ingest-pdf) + embedding-transformer (Sentence Transformers) so UI works with Cohere or Transformer.
COPY pyproject.toml uv.lock ./
COPY app ./app
RUN uv sync --frozen --no-dev --extra pdf --extra embedding-transformer

# Optional: Image embedder (CLIP/ViT)
RUN if [ "$INSTALL_EMBEDDING_EXTRAS" = "1" ]; then uv sync --no-dev --extra embedding-image; fi

# ---
# Test stage: dev deps + embedding extras so all tests (including Transformer/Image) run inside container.
# Build: docker build --target test -t vector-db-api-test .
# Run:   docker run --rm vector-db-api-test
# ---
FROM builder AS test
RUN uv sync --frozen
# Install optional embedding deps inside Docker so embedding tests run (Transformer, Image).
RUN uv sync --extra embedding-transformer --extra embedding-image
COPY app ./app
COPY tests ./tests
CMD ["uv", "run", "pytest"]

# ---
# Runtime stage: smaller image, no build tools.
# ---
FROM ghcr.io/astral-sh/uv:python3.12-bookworm-slim AS runtime

WORKDIR /app

# Copy venv and project from builder.
COPY --from=builder /app/.venv /app/.venv
COPY pyproject.toml uv.lock ./
COPY app ./app

ENV PATH="/app/.venv/bin:$PATH"

EXPOSE 8000

# Run the API; override with docker run ... <cmd> or compose command.
CMD ["uvicorn", "app.main:app", "--host", "0.0.0.0", "--port", "8000"]