Jupyter notebook: protect cells

[profbrady]

I use Jupyter notebooks with Python 3 to teach an engineering programming class. The interface is great for giving students notes and instructions with code cells, however it is very easy for students to accidentally delete or modify cells. Can a method be added to protect cells from inadvertent deletion or modification. I don't know, but maybe a cell toolbar named "Protection" that can be turned on via the View menu with a couple of check boxes; allow deletion, allow modification, allow movement.

This way, an instructor could go into that view mode when creating assignments and turn off allow deletion and allow modification then set the view mode back to None before assigning to students. It would not keep them from switching modes, but it would require an overtly deliberate act.

[williamstein]

Is there anything already like this in the "normal" jupyter ecosystem (for classical jupyter)? If so, can you find it and post a link. The reason is only since it would be easier to make this compatible with that (e.g., use the same description/name).

Actually implementing what you describe in terms of CoCalc Jupyter would I think be very easy.

[haraldschilly]

I found

• https://github.com/ipython-contrib/jupyter_contrib_nbextensions/wiki/Readonly (doesn't mention what it saves in the cell metadata, but we could support it)
• https://github.com/ipython-contrib/jupyter_contrib_nbextensions/tree/master/src/jupyter_contrib_nbextensions/nbextensions/freeze sounds more complex, but covers cases where cells can't be executed (frozen). I think that's even better than just read-only, because the genereated output of such a cell can't be destroyed by accidentally evaluating it. Consequently, "run all cells" should then skip such frozen cells.

[williamstein]

REQUESTED BY: Richard Bilonick, among other people.

[haraldschilly]

Another one, in Jupyter notebook version 5 (i.e. rather recent), there is the metadata field "editable:true|false" and "deletable:true|false". I think we should support these two, and also add a more convenient switch to the menu than having to fiddle around with the metadata json directly.

Steps to see this in action:

1. open the plain classical jupyter server and there an ipynb file.
2. eval a cell with stuff like 2+2.
3. menu: view → cell toolbar → edit metadata.
4. make this like

{
  "trusted": true,
  "editable": false,
  "deletable": false
}

5. try to edit or delete the cell. For me, it only lets you eval it!

[profbrady]

I am all in favor of this. I just opened one of my assignment notebooks in classical mode and edited metadata for the cells I want locked various ways. I love it. Please add and make simple toggles or check boxes. Modifying the metadata for multiple selected cells at one time would also be nice.

[dduque10]

My markdown cell is still editable after editing the metadata as suggested, what could I be doing wrong? I am running jupyter on Pycharm and Python 3

[williamstein]

Instead of directly editing metadata (where you could make a mistake), use the Edit menu with the cell selected:

[dduque10]

I don't have those options :(

[williamstein]

I don't have those options :(

That's not possible if you are using cocalc-jupyter, since they are always visible. Please post a screenshot (showing your entire browser window) so I can better direct you.

[dduque10]

I am not using cocalc-jupyter I thought this was a jupyter forum. I am sorry.

[williamstein]

Oh that makes sense. The same options should be supported in Jupyter classic and JupyterLab (soon, at least), since I made an accepted PR to the nbformat docs:

https://nbformat.readthedocs.io/en/latest/format_description.html#cell-metadata

I'm not sure what the implementation status is though.