It is possible to assign the google colab notebook name to a python variable. In Jupyter you can use javascript but this does not work in colab. I have found examples using %javascript to access the html but these do not appear to work with colab notebooks. So I want to copy the notebook to my google drive and rename it but then assign the new name to a python variable.
Add a comment
|
2 Answers
Here's the code to get the notebook name.
from requests import get
filename = get('http://172.28.0.2:9000/api/sessions').json()[0]['name']
UPDATE - Jan 2023
The URL has changed from "172.28.0.2" to "172.28.0.12". See the other post for details.
New code:
from requests import get
filename = get("http://172.28.0.12:9000/api/sessions").json()[0]["name"]
Or, to adapt if ip is changed again
from requests import get
from socket import gethostname, gethostbyname
ip = gethostbyname(gethostname()) # 172.28.0.12
filename = get(f"http://{ip}:9000/api/sessions").json()[0]["name"]
1 Comment
itismoej
This workaround doesn't work anymore.
As of Jan 2023, korakot's solution needs to be updated to the new URL from "172.28.0.2" to "172.28.0.12". The working code is below:
import requests
filename = requests.get("http://172.28.0.12:9000/api/sessions").json()[0]["name"]
# Output will be of the form "xyz.ipynb"
Alternative: A solution robust to IP changes
colab_ip = %system hostname -I # uses colab magic to get list from bash
colab_ip = colab_ip[0].strip() # returns "172.28.0.12"
colab_port = 9000 # could use 6000, 8080, or 9000
import requests
filename = requests.get(f"http://{colab_ip}:{colab_port}/api/sessions").json()[0]["name"]
# Output will be of the form "xyz.ipynb"
Explanation
In case the URL changes again the future, you can find the latest IP address using code from this post: https://stackoverflow.com/a/66709565/2879686. The relevant code is below:
# Gives IP address
!hostname -I
print() # empty line
# Gives IP addresses with port numbers
!sudo lsof -i -P -n | grep LISTEN
The current output for me is:
172.28.0.12
node 7 root 21u IPv6 19966 0t0 TCP *:8080 (LISTEN)
kernel_ma 35 root 7u IPv4 19643 0t0 TCP 172.28.0.12:6000 (LISTEN)
colab-fil 65 root 3u IPv4 19927 0t0 TCP *:3453 (LISTEN)
colab-fil 65 root 4u IPv6 19929 0t0 TCP *:3453 (LISTEN)
jupyter-n 91 root 4u IPv4 20144 0t0 TCP 172.28.0.12:9000 (LISTEN)
python3 138 root 23u IPv4 20719 0t0 TCP 127.0.0.1:45899 (LISTEN)
python3 174 root 3u IPv4 19440 0t0 TCP 127.0.0.1:18901 (LISTEN)
python3 174 root 5u IPv4 20935 0t0 TCP 127.0.0.1:40271 (LISTEN)
Interestingly, there are multiple ports that work:
- 172.28.0.12:6000
- 172.28.0.12:8080
- 172.28.0.12:9000
You can use colab magic to create a solution that is robust to IP changes.
Reference: https://ipython.readthedocs.io/en/stable/interactive/magics.html#cell-magics
colab_ip = %system hostname -I # uses colab magic to get list from bash
colab_ip = colab_ip[0].strip() # returns "172.28.0.12"
colab_port = 9000 # could use 6000, 8080, or 9000
import requests
filename = requests.get(f"http://{colab_ip}:{colab_port}/api/sessions").json()[0]["name"]
# Output will be of the form "xyz.ipynb"
