[go: up one dir, main page]

Skip to content

Multi-Cell Scopes for Jupyter Notebooks

License

Notifications You must be signed in to change notification settings

erawn/pagebreaks

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Pagebreaks : Multi-Cell Scopes for Jupyter Notebooks

Pagebreaks is a Jupyter Notebooks extension (with a supporting IPython plugin) which creates scope boundaries between groups of cells, allowing cells within a pagebreak to share state as usual, but keeping state isolated to that group. To use variables between Pagebreaks, they can be "exported" at the footer of the pagebreak in a read-only format to be used in all later cells.

The goal is to make it easier to keep variables organized in notebooks while changing as little as possible about how notebook programmers like to work. To do that, Pagebreaks allows users to organize their notebook state by organizing their cells within the notebook.

Install

To install the extension, execute:

python -m pip install pagebreaks

Then, open Jupyter Lab jupyter lab and open a Notebook. You should see a warning banner telling you to run %load_ext pagebreaksip, at which point the warning will disappear, and you'll be good to go. (You might have to restart the kernel and reload the webpage, just to be safe)

Overview

Each "pagebreak" keeps top-level variables isolated, so that within a pagebreak you can reference variables between cells normally, but in the rest of the notebook those variables will be inaccessible: showscopebound
To reference a variable defined in an earlier Pagebreak, add the variable to the “export” list at the bottom of the Pagebreak in which it was defined. Once it’s exported, later cells can read its value, but they can’t overwrite it. Screenshot 2024-07-01 at 2 37 02 PM
Exported variables continue to be readable and writeable in their own Pagebreak, as usual. After that Pagebreak, later cells can read but not write the exported variable. Before that Pagebreak, cells can neither read nor write the exported variable. Screenshot 2024-07-01 at 2 45 00 PM
To check out the current state of the notebook, you can use the %who_pb IPython magic. If the magic is run in a Pagebreak in which an exported variable is available to be called (i.e. the variable is exported from a previous pagebreak, but not a later one), it will list under Export Exist? as True. who_pb
Modules remain global, so you only have to import them once: packages_are_global

*Because Python doesn't have a built-in way to ensure read-only variables, we check for redefinitions at the AST level and dynamically after each cell run, checking to see if the value has changed.

Pagebreak Actions

Making a New Pagebreak

New pagebreaks are made by pressing this button: Screenshot 2024-07-01 at 1 50 30 PM on the "Export" cell of a pagebreak at the bottom.

makenewpb

Merging Pagebreaks

Instead of deleting Pagebreaks, you can merge the cells of a pagebreak into the one above it with:Screenshot 2024-07-01 at 1 50 56 PM

mergepb

For example:

The bottom Pagebreak ----> Merges with the top
Screenshot 2024-07-01 at 1 52 36 PM ----> Screenshot 2024-07-01 at 1 52 44 PM

%who_pb

We've added the IPython magic %who_pb", which is a pagebreaks-specific version of %who_ls. %who_pb" prints out your notebook state by its pagebreak, listing whether each variable is currently being exported. Pagebreaks only generates the export variables it needs for each cell, so you won't see variables that are exported in later pagebreaks, because those are currently out of scope!

Autoloading the IPython Plugin

To avoid having to run %load_ext pagebreaksip each time you start your kernel, you can start it automatically by adding: c.InteractiveShellApp.extensions.append('pagebreaksip') to your .ipython/profile/ipython_config.py file.

How it works

You shouldn't need to know what's going on under the hood to use Pagebreaks, but if you're curious, read on!

Rather than dynamically storing and reloading different global variables in your kernel, Pagebreaks manipulates the programs you write before they go to the interpreter, changing the names of variables under the hood.

For example, the variable a is actually stored as pb_0_a in the global state, because it is in Pagebreak 0. Screenshot 2024-07-01 at 2 47 14 PM

When a variable is exported to be used between pagebreaks, a new variable pb_export_b is generated for each cell run (as a user, you don't have to worry about any of this, you can just use a and b as normal!). Because Python doesn't have a way to enforce that variables are read-only at compile time, Pagebreaks will check after your cell has run that the pb_export_b variable still matches the original pb_0_b variable. If it doesn't, Pagebreaks will revert the variables in your current pagebreak back to what they were before you ran the cell.

Because b is accessible in the second pagebreak because it's been exported, a pb_export_b varaible is generated for later cells to reference, preventing those cells from modifying our real b variable, which is pb_0_b Screenshot 2024-07-01 at 2 49 40 PM

Requirements

  • JupyterLab >= 4.0.0
  • Pagebreaks is currently only available for IPython notebooks in Jupyter.

Uninstall

To remove the extension, execute:

pip uninstall pagebreaks

About

Multi-Cell Scopes for Jupyter Notebooks

Resources

License

Stars

Watchers

Forks