PyNoon Plus Lesson 3 - Tutorial

This tutorial will cover building user interfaces in Jupyter notebooks.

Setup

1. Make a new notebook for this lesson
2. What’s the first thing to do? RENAME IT!
3. Name it pynoon_plus_3.ipynb

We need to install jupyter_bokeh for panel to work in Google Colab:

!pip install jupyter_bokeh

UI Widget Basics

To get started, let’s import the panel module that provides widgets for building user interfaces

• Usually installed with %pip install panel, but it’s pre-installed in Colab.
• We also need to call pn.extension() to finish loading panel.

import panel as pn

pn.extension()

Now let’s make a text field:

text_field = pn.widgets.TextInput()

We can display it:

text_field

And we can get the current contents of the field:

text_field.value

• Try entering some text, then re-run text_field.value.
• See that the value has changed.

Buttons

Now let’s make a button:

submit_button = pn.widgets.Button(name='Submit')
submit_button

• We can use pn.bind() to assign a “callback” function that should be called whenever the button is clicked.

submit_button = pn.widgets.Button(name='Submit')

def on_submit(event):
    if not event:
        return

    return text_field.value

submit_output = pn.bind(on_submit, submit_button)

pn.Column(
    submit_button,
    submit_output,
)

• The value returned by pn.bind() will display outputs returned by the callback.
  • We display the button and output together by wrapping them in a pn.Column.
  • pn.bind() can also bind functions to any widget updates, like text box changes.
• Note that the function we provide must accept a single argument that will contain details of the button click “event”.
  • The callback will be called once initially, but the argument passed to it will only contain a value when the button has been clicked.

Each time you click the button, the current value of the text input is displayed.

Bringing it all together a self-contained form

• Use a consistent structure:
  • Start by defining widgets.
  • Then define and bind event handlers.
  • Then define a layout that includes all of the widgets to be displayed.
• If your interface gets too big, consider splitting it up into functions

text_field = pn.widgets.TextInput()
submit_button = pn.widgets.Button(name='Submit')

def on_submit(event):
    if not event:
        return

    return text_field.value

submit_output = pn.bind(on_submit, submit_button)

layout = pn.Column(
    text_field,
    submit_button,
    submit_output,
)
layout

Deploying our app

• One way we can share a panel application with users is to convert it to a shareable .html file.
• html is the language web pages are written in, so any user will be able to open the file in their web browser.
• Recent developments in web and Python technologies allow us to create html files that can run Python code.

Preparing the app notebook

• Make a new notebook, call it pynoon_plus_3_app.ipynb
• Add from above:

import panel as pn

pn.extension()

...interface...

layout.servable()

• Make sure your layout ends with .servable()
  • This tells panel what components make up the “app”
• Save the file

Converting the app notebook to html

From your original notebook (important to avoid including the !panel convert line in your app notebook):

• Click on the file browser in the left-hand sidebar, and select the button to Mount Drive
• Add a notebook cell to run:

!panel convert "drive/MyDrive/Colab Notebooks/pynoon_plus_3_app.ipynb" --to pyscript --out app_output

• Refresh the file browser
• From the file browser, download app_output/pynoon_plus_3_app.html
• Open pynoon_plus_3_app.html in your web browser to use your app!

Going further with user interfaces

• Look at the panel documentation
  • See what input widgets are available (under Component Gallery -> Widgets in the documentation sidebar)
  • See what rich outputs are available (e.g. Plotly plots, under Component Gallery -> Panes in the documentation sidebar)
• Other user interface libraries exist for:
  • Notebook and web apps
    • ipywidgets
    • Streamlit
    • Plotly Dash
  • Desktop apps (less popular these days)
    • Tkinter
    • PyQt