Getting Started
Dive Into PyShiny by Appsilon
By Piotr Pasza Storożenko, with Pavel Demin support
Welcome to the Shiny Python Workshop
Original content by Gordon Shotwell & Joe Cheng (Posit).
Assumptions
- You have some basic experience with Python
- You have some basic experience with RShiny
Goals of the workshop
- Understand/refresh the mechanics of a Shiny app (what to do, not why you do it)
- Learn how to build a PyShiny app
- Static Content
- Shiny Core and Shiny Express
- Reactivity
- Custom UI
- Shiny Modules
- What python has that R doesn’t
- Tapyr
Workshop workflow
- (Presentation, exercises)+
- All of the exercises and slides are running live on the website
- You can do them on the website or open the apps in the
apps/problem-setsfolder
Testing your setup
- For the workshop, we advice using the page with all the exercises
- If you want to run the apps locally, setup your local environment and install the packages from
requirements.txt
Hello Shiny
What’s a Shiny app?
Key features of Shiny
- Easy to develop
- Extensible
- Efficient
Why Shiny for Python?
- Shiny is designed to take you from prototype to production
- Easy enough to use for prototypes (especially
shiny.express!) - Efficient enough to handle complexity
- Everything you need to build production quality apps
- You don’t need to throw away your Shiny apps when moving from prototype to production ready
- Comparison with Dash / Streamlit / Jupyter Notebook / Reflex coming later on
Anatomy of a Shiny App
To make a Shiny app work you need to do three things:
- Add inputs and outputs to the UI
- Create a rendering function
- Refer to inputs
Inputs and outputs
Rendering function
Refer to inputs
UI functions
- UI functions are in the
uisubmodule - Inputs start with
ui.input_*()and take an id, and options - Outputs start with
ui.output_*()and usually just take an id - Other functions like
ui.h1()orui.p()add static HTML to the app - We’ll get to layout functions in Session 3
Server function
- Every Shiny app needs a
serverfunction - The server function executes for each user session
- This function contains rendering functions which define how to turn inputs into outputs
Rendering functions
- Functions are defined inside the main
serverfunction - They need to be decorated with
@render.* - The function name should match the output id
- The return value is sent to the Shiny UI
Decorators are just functions
@render.plot(alt="A body mass plot of penguins")
def mass_plot():
df = sample_data(penguins.copy(), input.sample())
df = df.loc[df["body_mass"] < input.mass()]
return dist_plot(df)- Decorators are functions which take other functions
- Shiny uses them to identify reactive functions
- They can take arguments
Simple app example
Common problems
- An output doesn’t render
- An output doesn’t update when an input changes
- There’s some weird error
Connecting inputs to outputs
def server(input: Inputs, output: Outputs, session: Session):
@render.text
def txt():
return f"n*2 is {input.n() * 2}"- Inputs are read by calling them like a function:
input.n(), notinput.n - Referring to an input creates a reactive link between that input and the rendering function
- When the input changes, the rendering function will re-execute
- You can use multiple inputs in the same rendering function
Recap
Most Shiny app development consists of variations of these three things:
- Add UI elements
- Add rendering functions
- Connect inputs and outputs
- …
- Refactoring code into modules and functions!
Getting Started Dive Into PyShiny by Appsilon By Piotr Pasza Storożenko, with Pavel Demin support