7.4 Complicated object initialization and property computations

When working with data in unhelpful formats, it often becomes necessary to build Python objects from source data that has a different structure or different underlying object types. There are two overall ways to treat object creation:

• It’s part of the application as a whole. Data should be decomposed by a parser and recomposed into useful Python objects. This is the approach we’ve taken in previous examples.

• It’s part of the object’s class definition. Source data should be provided more or less in its raw form, and the class definition will perform the necessary conversions.

This distinction is never simple, nor crisp. Pragmatic considerations will identify the best approach for each unique case of building a Pythonic object from source data. The two examples that point to the distinct choices available are the following:

• The Point class: The syntax for geographic points is highly variable...