dashlab 0.3.0

A Python package for dashboard and data visualization tools.

DashLab

An enhanced dashboarding library based on ipywidgets' interaction functionality that lets you observe any trait of widgets, observe multiple functions and build beautiful dashboards which can be turned into full screen.

Installation

You can install dashlab using pip:

pip install dashlab

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

• DashboardBase: Create interactive dashboard applications with minimal code by extending the dashlab.DashboardBase class and defining methods with the @callback decorator.
• Dashboard: A ready-to-use dashboard class that allows quick setup of interactive dashboards without subclassing and supports callbacks after initialization.
• 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.
  • Tuple pattern (widget, 'trait') for trait observation where widget is accessible via params and trait value goes to callback. This is useful to have widget and trait in a single parameter, such as x = (fig, 'selected') for plotly FigureWidget. Other traits of same widget can be observed by separate parameters with y = 'x.trait' pattern.
  • You can use '.fullscreen' to detect fullscreen change and do actions based on that.
  • Use .params to acess widgets built with given parameters.
  • Use var to observe any python variable which is not a widget and trigger callbacks when var.value changes.
  • 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
• Markdown support:
  • Convert markdown to HTML widget using dashlab.markdown function.
  • hstack and vstack functions support markdown strings to automatically convert to HTML and place in stack.
• 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 DashboardBase 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 dashlab as dl

dash = dl.Dashboard(
    fig = dl.patched_plotly(go.FigureWidget()), 
    html = dl.markdown('**Select Box/Lesso on figure traces**'),
    A = (1,10), omega = (0,20), phi = (0,10),
    sdata = 'fig.selected', cdata = 'fig.clicked', fs = '.isfullscreen',
)
@dash.callback('out-click', throttle = 200) # limit click rate by 200 ms
def on_click(cdata,html):
    display(pd.DataFrame(cdata or {}))

@dash.callback('out-select')
def on_select(sdata, html):
    plt.scatter(sdata.get('xs',[]),sdata.get('ys',[]))
    plt.show()

@dash.callback('out-fs')
def detect_fs(fig, fs):
    print("isfullscreen = ",fs)
    fig.layout.autosize = False # double trigger
    fig.layout.autosize = True

@dash.callback
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'))

dash.set_css({
    '.left-sidebar':{'background':'whitesmoke'},
    ':fullscreen': {'height': '100vh'}}
)
dash.set_layout(
    left_sidebar=['A','omega','phi','html', 'out-select','out-main'], 
    center=['fig','out-click'], 
    pane_widths=[3,7,0],
)

dash

Comprehensive Examples

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