SeafileContentManager

A custom ContentManger for Jupyter Notebooks based on Seafiles WebAPI following the notes on Custom ContentManagers.

The ContentManger works for both Jupyter Notebook and Lab. Checkpoints for files, folders and notebooks are are based on the Seafile Commit History. Thus, checkpoints are available for a predefined period of time ('retention period', e.g. 30 days, can also be infinite, see here).

Since for Python IO operations (e.g. open('file.txt','r') as file) the underlying file system is used, there is an additional drop-in replacement for those operations

from SeafileContentManager import SeafileFS

fs = SeafileFS()

file = fs.open('/text.txt','r')

file.read()

Ideally, notebooks should be written with this replacement from the start.

Status

Currently under active development.

What works?

• In JupyterLab or Notebook GUI:
  • Folder view
  • Opening and saving notebooks
  • Opening and saving text files (.txt,.md)
  • Creating new folder, file or notebook
  • Download
  • Duplicate
  • Renaming
  • Moving files
• Using the SeafileFS drop-in replacement for io operations:
  • Opening files in all modes (r,a,w,x, or adding b, +)
  • Reading
  • Writing (not yet a+, a+b modes)
  • listdir
  • listdir_attrib (with file size, creation date, etc.)
  • mkdir

What does not work?

• Drag and Drop Upload in JupyterLab
  • Files end up b64 encoded on the Seafile FS
• Relative paths in the SeafileFS
  • All paths are taken from the root, i.e. the at startup selected SeaFile library

Testing

To test the content manager, clone the repository, create a new virtual environment

virtualenv -p python3 env

in the cloned directory, activate it

source env/bin/activate

and install the development version

pip install -U .

Export the following environment variables

• SEAFILE_URL: The url to access your SeaFile instance
• SEAFILE_ACCESS_TOKEN: Your access token
• SEAFILE_LIBRARY: The name of the library, where all

e.g. by having a file env in your testing folder (don't forget to add it to gitignore!) with the following content

#!/bin/bash
export SEAFILE_URL=https://my.seafile.instance
export SEAFILE_LIBRARY=notebooks
export SEAFILE_ACCESS_TOKEN=12341234124124

and then sourcing the file (source testing/env) before running the command below.

Jupyter Notebook

To start Jupyter notebook from the activated environment run

jupyter notebook \
  --NotebookApp.contents_manager_class=SeafileContentManager.SeafileContentManager \
  --ContentsManager.checkpoints_class=SeafileContentManager.SeafileCheckpoints \
  --debug

This should start Jupyter notebooks with the Seafile API contents and checkpoints manager.

JupyterLab

To use the ContentManger with Jupyter Lab, create a settings file with the following content

from SeafileContentManager import SeafileContentManager, SeafileCheckpoints
c = get_config()
c.NotebookApp.contents_manager_class = SeafileContentManager
c.ContentsManager.checkpoints_class = SeafileCheckpoints

and run

jupyter-lab --config /path/to/config/jupyter_notebook_config.py --debug