This is a new project that is still a work in progress. Please do not attempt to use it in production yet. Contributors welcome!
In this GIF, Alice logs in, right-clicks a notebook and chooses "Copy Shareable Link". A dialog box appears, saying:
For the next hour, any other user on this JupyterHub who has this link will be able to fetch a copy of your latest saved version of dask-examples/array.ipynb.
She copies the link and gives it to Bob. Then, on the right, Bob logs in and pastes the link into his browser. He is given a copy of Alice's notebook.
This is for low-effort, short-term sharing between users who are on the same Hub.
The sender right-clicks a notebook (or any file) and clicks "Copy Shareable Link." The sender gives that link to any other user on the same Hub. When another user clicks the share link, the last saved version of the file is copied from the sender's notebook server to the recipient's. If the sender changes the file, the recipient can click the link again to make another copy reflecting the changes. After a given time interval, the link expires.
On Hubs that provide the user with options at spawn time, such as a container-based spawner, the share link encodes both the notebook and the options (e.g. the container image) that the sender was running that notebook in. The recipient will automatically be directed to a server spawned with the same options: the service finds a suitable existing one or spawns one if necessary. Thus the recipient has some assurance that they will be running the notebook in a compatible software environment.
This approach is not suitable for persistent sharing, such as galleries or lists of links to be maintained long term. For those use cases, it is better to encode software dependencies (as in a Binder repo) rather than relying on the availability of a specific image.
This approach should be compatible with any spawner. Two examples are given here, a local process spawner and a container-based spawner.
Install JupyterHub.
pip install jupyterhub
The usual prequisites for installing JupyterHub apply. One expedient way to install node is:
pip install nodeenv
nodeenv -p
Install using pip.
pip install jupyterhub-share-link
Install DockerSpawner.
pip install dockerspawner
Generate a key pair that will be used to sign and verify share links.
# creates private.pem and public.pem in the current directory
python -m jupyterhub_share_link.generate_keys
Start JupyterHub using an example configuration provided in this repo.
jupyterhub -f example_config_dockerspawner.py
Log in with any username and password---for example, alice
.
(The DummyAuthenticator
is used by this demo configuration.)
Spawn a server using the default image,
danielballan/base-notebook-with-image-spec-extension
.
Create and save a notebook Untitled.ipynb
to share.
Find Untitled.ipynb
in the file browser and right-click it.
A dialog box will appear. Click the button to copy the link.
Log in as a different user and paste the shared link.
The user will have a new server started running the same image as alice
,
and the notebook will be copied and opened.
Install JupyterHub.
pip install jupyterhub
The usual prequisites for installing JupyterHub apply. One expedient way to install node is:
pip install nodeenv
nodeenv -p
Install using pip.
pip install jupyterhub-share-link
Generate a key pair that will be used to sign and verify share links.
# creates private.pem and public.pem in the current directory
python -m jupyterhub_share_link.generate_keys
Install the labextension into the user environment.
# Disable the default share-file extension and register our custom one.
jupyter labextension disable @jupyterlab/filebrowser-extension:share-file
jupyter labextension install jupyterhub-share-link-labextension
Start JupyterHub using an example configuration provided in this repo. (In order to be able to log in as multiple users, you will likely need to run this as root.)
jupyterhub -f example_config_no_containers.py
Log in as a system user and start the user's server.
Create and save a notebook Untitled.ipynb
to share.
Find Untitled.ipynb
in the file browser and right-click it.
A dialog box will appear. Click the button to copy the link.
Log in as a different user and paste the shared link.
The notebook will be copied to that user's server and opened.
This involves:
A stateless Hub Service (in this repository) with the routes:
POST /create # issue a shareable link
GET /open # open a shared link
GET /inspect # inspect a shared link
GET / # verion info
The file-copying occurs via the notebook's ContentsManager, so there is no need for users to be on the same filesystem. They only have to be on the same Hub.