Dockerfile.cuda

# ─── Stage 1: Builder ────────────────────────────────────────────────────────
FROM nvidia/cuda:12.4.1-devel-ubuntu22.04 AS builder

# Install Rust and build dependencies
RUN apt-get update && apt-get install -y --no-install-recommends \
        curl \
        g++ \
        pkg-config \
        git \
        ca-certificates \
    && rm -rf /var/lib/apt/lists/*

# Install Rust
RUN curl --proto '=https' --tlsv1.2 -sSf https://sh.rustup.rs | sh -s -- -y
ENV PATH="/root/.cargo/bin:${PATH}"

WORKDIR /build

# Copy manifests first to cache dependency compilation
COPY Cargo.toml Cargo.lock ./

# Create a dummy binary so cargo can resolve and compile all dependencies
# before the real source is copied in (improves layer caching).
RUN mkdir -p src \
    && echo 'fn main() {}' > src/main.rs \
    && cargo build --release --features cuda 2>/dev/null || true \
    && rm -rf src

# Copy real source and build with CUDA support
# CUDA_COMPUTE_CAP is needed since GPU isn't available during build
COPY src ./src
ARG CUDA_COMPUTE_CAP=86
ENV CUDA_COMPUTE_CAP=${CUDA_COMPUTE_CAP}
RUN touch src/main.rs \
    && cargo build --release --features cuda

# ─── Stage 2: Runtime ────────────────────────────────────────────────────────
FROM nvidia/cuda:12.4.1-runtime-ubuntu22.04

RUN apt-get update && apt-get install -y --no-install-recommends \
        ca-certificates \
    && rm -rf /var/lib/apt/lists/*

COPY --from=builder /build/target/release/open-stt-server /usr/local/bin/open-stt-server

# HuggingFace model cache — mount a volume here to persist downloads
VOLUME ["/root/.cache/huggingface"]

ENV OPEN_STT_PORT=8080 \
    OPEN_STT_MODELS=whisper-base \
    OPEN_STT_DOWNLOAD=true \
    RUST_LOG=info

EXPOSE 8080

ENTRYPOINT ["open-stt-server"]