# Use a lightweight Python base image
FROM python:3.12-slim

# Set environment variables
ENV PYTHONDONTWRITEBYTECODE=1
ENV PYTHONUNBUFFERED=1
ENV PYTHONPATH=/app
ENV PATH="/app/.venv/bin:$PATH"

# Set working directory
WORKDIR /app

# Install system dependencies
RUN apt-get update && apt-get install -y \
    build-essential \
    && rm -rf /var/lib/apt/lists/*

# Install uv (our package manager)
COPY --from=ghcr.io/astral-sh/uv:latest /uv /uvx /bin/

# Copy only requirements first for caching
COPY pyproject.toml uv.lock README.md ./

# Install dependencies using uv
RUN uv sync --frozen --no-dev

# Copy the rest of the application
COPY src/ ./src/
COPY profiles/ ./profiles/

# Download knowledge base and indices from Hugging Face Dataset
RUN uv pip install huggingface_hub
RUN huggingface-cli download JairoDanielMT/prompt-compiler-knowledge-v3 --repo-type dataset --local-dir data/

# Download required spaCy NLP models for English and Spanish
RUN python -m spacy download en_core_web_trf
RUN python -m spacy download en_core_web_sm

# Pre-download Translation models for Always-English pipeline
RUN python -c "from transformers import AutoTokenizer; from optimum.onnxruntime import ORTModelForSeq2SeqLM; \
    AutoTokenizer.from_pretrained('Xenova/opus-mt-es-en'); \
    ORTModelForSeq2SeqLM.from_pretrained('Xenova/opus-mt-es-en')"

# Setup a non-root user for Hugging Face Spaces compatibility
RUN useradd -m -u 1000 user
RUN chown -R user:user /app
USER user

# Expose the API port expected by Hugging Face Spaces
EXPOSE 7860

# Start the API
CMD ["uvicorn", "src.api.app:app", "--host", "0.0.0.0", "--port", "7860"]