# HuggingFace Spaces — Common Crawl URL Extractor
# SDK: Docker  |  Port: 7860

FROM python:3.11-slim

# HF requires this label for Docker Spaces
LABEL maintainer="Neon Tech"

# Install only what's needed
RUN pip install --no-cache-dir requests

WORKDIR /app

# Output lives in /data so you can mount a persistent HF dataset volume
# If no volume is mounted it falls back to /app/cc_urls
ENV OUTPUT_DIR=/data/cc_urls

COPY app.py .

EXPOSE 7860

CMD ["python", "-u", "app.py"]