Dockerfile

# FROM python:3.11-slim
# Use HuggingFace's pre-built GPU image (includes CUDA, PyTorch, Transformers)
# FROM huggingface/transformers-pytorch-amd-gpu:latest
FROM huggingface/transformers-pytorch-gpu:latest@sha256:4c7317881a534b22e18add49c925096fa902651fb0571c69f3cad58af3ea2c0f


WORKDIR /app

# Install system dependencies
RUN apt-get update && apt-get install -y \
    build-essential \
    curl \
    git \
    && rm -rf /var/lib/apt/lists/*
    
# Upgrade Python to 3.9
# RUN apt-get update && \
#     apt-get install -y python3.9 python3.9-distutils python3.9-dev && \
#     update-alternatives --install /usr/bin/python3 python3 /usr/bin/python3.9 1 && \
#     curl -sS https://bootstrap.pypa.io/get-pip.py | python3.9 && \
#     rm -rf /var/lib/apt/lists/*

# Verify
RUN python3 -V

# Copy requirements first (for better Docker layer caching)
COPY requirements.txt ./

# Install Python dependencies
RUN pip3 install --no-cache-dir -r requirements.txt

# Pre-download Hugging Face models during build
# This caches models in the Docker image for faster container startup
# IMPORTANT: Set cache directory BEFORE downloading to ensure models are in /app/.cache/huggingface
ENV HF_HOME=/app/.cache/huggingface
ENV HF_HUB_CACHE=/app/.cache/huggingface
ENV HF_DATASETS_CACHE=/app/.cache/huggingface
ENV TRANSFORMERS_CACHE=/app/.cache/huggingface
ENV SENTENCE_TRANSFORMERS_HOME=/app/.cache/huggingface

COPY download_models.py ./
COPY src/config/settings.yaml ./src/config/settings.yaml
RUN mkdir -p /app/.cache/huggingface && \
    chmod -R 755 /app/.cache && \
    chmod -R 755 /app/.cache/huggingface && \
    python3 download_models.py

# Copy all application files (excluding .dockerignore patterns)
COPY . .

RUN mkdir -p /app/.streamlit && \      
    mkdir -p /app/.cache/streamlit && \
    mkdir -p /app/.cache/huggingface && \
    mkdir -p /app/conversations && \
    mkdir -p /app/feedback && \
    printf '[server]\nport = 8501\nheadless = true\nenableCORS = false\nenableXsrfProtection = false\n\n[browser]\ngatherUsageStats = false\n' > /app/.streamlit/config.toml && \
    chmod -R 777 /app/.streamlit && \   
    chmod -R 777 /app/.cache/streamlit && \
    chmod -R 777 /app/.cache/huggingface && \
    chmod -R 777 /app/conversations && \
    chmod -R 777 /app/feedback
    
    
ENV STREAMLIT_CONFIG_HOME=/app/.streamlit
ENV STREAMLIT_BROWSER_GATHER_USAGE_STATS=false
ENV STREAMLIT_USER_BASE_PATH=/app/.cache/streamlit

# Expose Streamlit port (HF Spaces maps to 7860 automatically)
EXPOSE 8501

# Health check for Streamlit
HEALTHCHECK --interval=30s --timeout=10s --start-period=40s --retries=3 \
    CMD curl --fail http://localhost:8501/_stcore/health || exit 1

#temp developement commands
RUN python3 -c "import torch; print(torch.cuda.is_available(), torch.cuda.get_device_name(0))" || true
# RUN mkdir /app/conversations && chmod -R 777 conversations
# RUN mkdir /app/feedback && chmod -R 777 feedback #langchain-huggingface==0.1.0
# RUN pip3 install langchain-huggingface

# Run the app
ENTRYPOINT ["streamlit", "run", "app.py", "--server.port=8501", "--server.address=0.0.0.0", "--server.headless", "true"]