Update Dockerfile

Dockerfile

@@ -1,13 +1,11 @@
 FROM python:3.11-slim-bookworm
 
-# Metadata as per https://github.com/opencontainers/image-spec/blob/master/annotations.md
+COPY --from=ghcr.io/astral-sh/uv:0.6.16 /uv /uvx /bin/
+
 LABEL org.opencontainers.image.authors="Bram Vanroy"
 LABEL org.opencontainers.image.title="MAchine Translation Evaluation Online - Demo"
 
-# Avoid prompts from apt
 ENV DEBIAN_FRONTEND=noninteractive
-
-# Install dependencies in a single RUN command to reduce image layers
 RUN apt-get update && apt-get install -y --no-install-recommends \
     build-essential \
     curl \
@@ -15,43 +13,65 @@ RUN apt-get update && apt-get install -y --no-install-recommends \
     && apt-get clean \
     && rm -rf /var/lib/apt/lists/*
 
-# Create a non-root user
+# --- Application User ---
 RUN useradd -m -u 1000 mateo_user
 USER mateo_user
 ENV HOME="/home/mateo_user"
+WORKDIR $HOME
+
+# --- Application Code & Dependencies ---
+ARG REPO_BRANCH=v1.6.0
+RUN git clone --depth 1 --branch ${REPO_BRANCH} https://github.com/BramVanroy/mateo-demo.git $HOME/mateo-demo
+
+WORKDIR $HOME/mateo-demo
 
-# Environment variables
+ARG USE_CUDA=false
+RUN uv venv .venv --python 3.11 \
+    && . .venv/bin/activate \
+    && uv pip install --no-cache-dir --upgrade pip wheel setuptools \
+        # Conditionally install PyTorch (CPU or CUDA 12.1)
+    && if [ "$USE_CUDA" = "true" ]; then \
+        echo "Installing PyTorch with CUDA 12.1 support"; \
+        uv pip install --no-cache-dir torch==2.2.1 --index-url https://download.pytorch.org/whl/cu121; \
+    else \
+        echo "Installing PyTorch with CPU support"; \
+        uv pip install --no-cache-dir torch==2.2.1+cpu --index-url https://download.pytorch.org/whl/cpu; \
+    fi \
+    && uv pip install --no-cache-dir --upgrade $HOME/mateo-demo[patch] \
+    && uv cache clean
+
+# --- Runtime Configuration ---
+# Set runtime environment variables
+# NOTE: for Hugging Face spaces we set XSRF to false and demo mode to false
 ENV PORT=7860 \
     SERVER="localhost" \
+    BASE="" \
+    DEMO_MODE=false \
     HF_HUB_ENABLE_HF_TRANSFER=1 \
-    PATH="${HOME}/.local/bin:${PATH}" \
-    USE_CUDA=false
-
-WORKDIR ${HOME}/mateo
+    PRELOAD_METRICS=true \
+    PATH="$HOME/.local/bin:$HOME/mateo-demo/.venv/bin:${PATH}" \
+    ENABLE_XSRF_PROTECTION=false
 
-# Clone the repository
-RUN git clone --depth 1 --branch v1.5.0 https://github.com/BramVanroy/mateo-demo.git
-WORKDIR mateo-demo
+# --- Set default cache dir for the user ---
+RUN mkdir -p "$HOME/.cache" \
+    && chown -R mateo_user:mateo_user "$HOME/.cache"
 
-# Install Python dependencies with conditional torch installation
-RUN python -m pip install --no-cache-dir --upgrade pip wheel setuptools \
-    && python -m pip install --no-cache-dir torch==2.6.0+cpu -f https://download.pytorch.org/whl/torch \
-    && python -m pip install --no-cache-dir --upgrade .
-
-# Pre-download default models
-RUN huggingface-cli download bert-base-multilingual-cased model.safetensors tokenizer.json vocab.txt; \
-    huggingface-cli download facebook/nllb-200-distilled-600M pytorch_model.bin sentencepiece.bpe.model tokenizer.json; \
-    python -c "import comet; from comet import download_model; download_model('Unbabel/wmt22-comet-da')"; \
-    python -c "import evaluate; evaluate.load('bleurt', 'BLEURT-20')"
+# --- Entrypoint & CMD ---
+COPY --chown=mateo_user:mateo_user entrypoint.sh /usr/local/bin/entrypoint.sh
+RUN chmod +x /usr/local/bin/entrypoint.sh
 
 # Expose the port the app runs on
 EXPOSE $PORT
 
-# Healthcheck to ensure the service is running
-HEALTHCHECK CMD curl --fail http://$SERVER:$PORT/_stcore/health || exit 1
+# Healthcheck (adjust start-period if runtime loading takes longer)
+# Note: SERVER env var needs to be passed at runtime if not localhost
+HEALTHCHECK --interval=30s --timeout=5s --start-period=60s \
+    CMD curl --fail http://${SERVER:-localhost}:${PORT}${BASE}/_stcore/health || exit 1
+
+# Set the working directory for the application
+WORKDIR $HOME/mateo-demo/src/mateo_st
 
-# Set the working directory to the Streamlit app
-WORKDIR src/mateo_st
+ENTRYPOINT ["/usr/local/bin/entrypoint.sh"]
 
-# Launch app
-CMD streamlit run 01_🎈_MATEO.py --server.port $PORT --server.enableXsrfProtection false
+# CMD is now empty as the entrypoint handles the final command execution
+CMD []