einteract 1.3.7

An enhanced version of ipywidgets's interactive widgets.

einteract

An enhanced interactive widget that lets you observe any trait of widgets, observe multiple functions and build beautiful dashboards which can be turned into full screen. This is a wrapper library around interact functionality in ipyslides which also provides rich content representations.

Installation

You can install einteract using pip:

pip install einteract

Or if you prefer to install from source, clone the repository and in its top folder, run:

pip install -e .

Interactive Playground

✨ Try it in your browser ✨

Jupyter Lab	Notebook	Binder

Features

• InteractBase: Create interactive dashboard applications with minimal code by extending the InteractBase class and defining methods with the @callback decorator.
• Custom Widgets:
  • Included custom built widgets for enhanced interaction.
  • Pass any DOMWidget as a parameter to interact/interactive functions unlike default ipywidgets.interactive behavior.
  • Observe any trait of the widget by 'widget_name.trait_name' where 'widget_name' is assigned to a widget/fixed(widget) in control parameters, OR '.trait_name' if trait_name exists on instance of interactive.
  • You can use '.fullscreen' to detect fullscreen change and do actions based on that.
  • Add ipywidgets.Button to hold callbacks which use it as paramter for a click
• Plotly Integration: Modified plotly support with additional traits like selected and clicked
• HTML support:
  • Convert matplotlib plots to HTML format using plt2html.
  • Any (supported) python object can be converted to html using html, hstack and vstack function.
• JupyTimer: A non-blocking widget timer for Jupyter Notebooks without threading/asyncio.
• Event Callbacks: Easy widget event handling with the @callback decorator inside the subclass of InteractBase or multiple functions in interact/interactive functions.
• Full Screen Mode: Transform your dashboards into full-screen applications by added button.

Usage Example

import numpy as np
import matplotlib.pyplot as plt
import ipywidgets as ipw
import pandas as pd
import plotly.graph_objects as go
import einteract as ei

@ei.callback('out-click', throttle = 200) # limit click rate by 200 ms
def on_click(cdata,html):
    display(pd.DataFrame(cdata or {}))

def on_select(sdata, html):
    plt.scatter(sdata.get('xs',[]),sdata.get('ys',[]))
    html.value = ei.plt2html().value

def detect_fs(fig, fs):
    print("isfullscreen = ",fs)
    fig.layout.autosize = False # double trigger
    fig.layout.autosize = True

@ei.interact(
    on_select,
    on_click,
    detect_fs,
    fig = ei.patched_plotly(go.FigureWidget()), 
    html = ipw.HTML('**Select Box/Lesso on figure traces**'),
    A = (1,10), omega = (0,20), phi = (0,10),
    sdata = 'fig.selected', cdata = 'fig.clicked', fs = '.isfullscreen',
    post_init = lambda dash: (
        dash.set_css({
            '.left-sidebar':{'background':'whitesmoke'},
            ':fullscreen': {'height': '100vh'}}
        ),
        dash.set_layout(
            left_sidebar=['A','omega','phi','html','out-main'], 
            center=['fig','out-click'], 
            pane_widths=[3,7,0],
        ),
    ),
)
def plot(fig:go.FigureWidget, A, omega,phi): # adding type hint allows auto-completion inside function
    fig.data = []
    x = np.linspace(0,10,100)
    fig.add_trace(go.Scatter(x=x, y=A*np.sin(omega*x + phi), mode='lines+markers'))

Comprehensive Examples

• Check out the einteract-demo.ipynb which demonstates subclassing InteractBase, using custom widgets, and observing multiple functions through the @callback decorator.
• See Bandstructure Visualizer and KPath Builder as comprehensive dashboards.