Overview

When developping scientific software, from the simplest script to the most complex application, one systematically needs to manipulate data sets (e.g. parameters for a data processing feature). These data sets may consist of various data types: real numbers (e.g. physical quantities), integers (e.g. array indexes), strings (e.g. filenames), booleans (e.g. enable/disable an option), and so on.

Most of the time, the programmer will need the following features:

  • allow the user to enter each parameter through a graphical user interface, using widgets which are adapted to data types (e.g. a single combo box or check boxes are suitable for presenting an option selection among multiple choices)

  • entered values have to be stored by the program with a convention which is again adapted to data types (e.g. when storing a combo box selection value, should we store the option string, the list index or an associated key?)

  • showing the stored values in a dialog box or within a graphical user interface layout, again with widgets adapted to data types

  • using the stored values easily (e.g. for data processing) by regrouping parameters in data structures

  • using those data structures to easily construct application data models (e.g. for storing application settings or data processing parameters) and to serialize and deserialize them (i.e. save and load them to/from HDF5, JSON or INI files)

  • update and restore a data set to/from a dictionary

  • generate a data set from a function signature (i.e. a function prototype) and use it to automatically generate a graphical user interface for calling the function

This library aims to provide these features thanks to automatic graphical user interface generation for data set editing and display. Widgets inside GUIs are automatically generated depending on each data item type.

Jupyter Notebook and Interactive Python Support

guidata provides enhanced support for interactive Python environments:

Rich HTML display in Jupyter notebooks:

DataSet objects and LabeledEnum values automatically render as styled HTML tables when displayed in Jupyter notebooks. This is powered by the _repr_html_() method, which Jupyter calls automatically when displaying objects as cell output.

import guidata.dataset as gds

class Parameters(gds.DataSet):
    """Processing parameters"""
    threshold = gds.FloatItem("Threshold", default=0.5)
    enabled = gds.BoolItem("Enable processing", default=True)

params = Parameters()
params  # Displays as a styled HTML table in Jupyter

Interactive attribute discovery:

When working in a Python shell or notebook, repr(dataset) (or simply typing the variable name) shows attribute names instead of labels, making it easy to discover the programmatic names needed to access or modify dataset items:

>>> params = Parameters()
>>> params  # Shows attribute names for programmatic access
Processing parameters:
    threshold: 0.5 (FloatItem)
    enabled: True (BoolItem)
>>> print(params)  # Shows user-friendly labels
Processing parameters:
    Threshold: 0.5
    Enable processing: ☑

The guidata library provides the following modules: