Implement the CS3APIs connector

[diocas]

In order to have a working implementation of CS3APIs for Jupyterlab, the following tasks need to be completed:

• 1. Write a Python client for the CS3APIs with the basic methods required by Jupyter ContentsManager #2
• 2. Create a Jupyter Contents Manager that uses the client developed in 1. ContentsManager #2
• 3. Create a Jupyter Server Extension to provide the remaining CS3APIs functionality, like Sharing (possibility to share and to retrieve/open what was shared with the user). Server Extensions for extended functionality #3
• 5. Implement the Jupyter Checkpoints Checkpoints #5
• 4. Create a Jupyterlab Extension that gives a web UI for accessing the functionality implemented in 3. FilesViewer for Jupyterlab #4

Resources

• Reva: https://github.com/cs3org/reva
• Wopiserver CS3APIs client implementation (can be used as an example implementation): https://github.com/cs3org/wopiserver
• Jupyter Contents API documentation: https://jupyter-notebook.readthedocs.io/en/stable/extending/contents.html
• Jupyterlab Extensions documentation: https://jupyterlab.readthedocs.io/en/stable/developer/extension_dev.html

[JarCz]

CS3 connector in WopiServer implements basic access method:

File operations:

• readFile
• writeFile
• renameFile
• removeFile

File attribuite:

• stat
• statX
• setAttr
• getAttr
• removeAttr

We probably don't need extra file access methods for Jupyter Lab.

@diocas do you have another opinion?

[diocas]

@JarCz for files, probably no, but we need to list files in directories and we need everything that is related to share (create share, list shares, etc).

[JarCz]

One way is to use API from https://cs3org.github.io/cs3apis/#cs3.sharing.collaboration.v1beta1.CollaborationAPI

and create methods like from reva-client:

• share-create
• share-list
• share-list-received
• share-remove
• share-update
• share-update-received

@labkode is it ok ?

[diocas]

Yes, the idea is to have a reva-client implementation in python that we can use.

For now only focus on the issue #2, which corresponds to the task "Create a Jupyter Contents Manager". This depends on having the CS3 client, but just build the client in parallel.
Let's continue the discussion in #2.