Update Dockerfile
Browse files- Dockerfile +16 -6
Dockerfile
CHANGED
|
@@ -1,12 +1,22 @@
|
|
| 1 |
-
#
|
| 2 |
-
#
|
| 3 |
|
| 4 |
-
FROM python:3.
|
| 5 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 6 |
|
| 7 |
WORKDIR /app
|
| 8 |
|
| 9 |
-
|
| 10 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 11 |
|
| 12 |
-
CMD ["python", "
|
|
|
|
| 1 |
+
# HuggingFace Spaces — Common Crawl URL Extractor
|
| 2 |
+
# SDK: Docker | Port: 7860
|
| 3 |
|
| 4 |
+
FROM python:3.11-slim
|
| 5 |
|
| 6 |
+
# HF requires this label for Docker Spaces
|
| 7 |
+
LABEL maintainer="Neon Tech"
|
| 8 |
+
|
| 9 |
+
# Install only what's needed
|
| 10 |
+
RUN pip install --no-cache-dir requests
|
| 11 |
|
| 12 |
WORKDIR /app
|
| 13 |
|
| 14 |
+
# Output lives in /data so you can mount a persistent HF dataset volume
|
| 15 |
+
# If no volume is mounted it falls back to /app/cc_urls
|
| 16 |
+
ENV OUTPUT_DIR=/data/cc_urls
|
| 17 |
+
|
| 18 |
+
COPY cc_get_urls.py .
|
| 19 |
+
|
| 20 |
+
EXPOSE 7860
|
| 21 |
|
| 22 |
+
CMD ["python", "-u", "cc_get_urls.py"]
|