Fluent Python

Data Structures

The Python Data Model

• Using special methods
  • Users don’t have to memorize arbitrary method names.
  • Benefit from rich python std lib and avoid reinventing the wheel.
  • Special methods are meant to be called by the Python interpreter, not by you.
  • Dunder method = double-underscored method
  • Pythonic
• # doctest: +ELLIPSIS directive to elide the output
• __repr__
  • Called by interactive console and debugger
  • Similar to %r place holder and !r conversion field (for standard representation, without quote)
  • This method should be capable of recreating the object
• __str__
  • Is meant for end users.
  • This falls back to __repr__
  • If only one is implemented, implement __repr__
• __bool__
  • bool invokes __bool__ and uses the result
  • If not implemented, invoke __len__, if 0, return False
  • Otherwise truthy.
• ABC
  • Sequence, Mapping, and Set
  • Collection should implement: Iterable, Sized, and Container
  • Since Python 3.7, dict is officially “ordered”, i.e. key insertion order is preserved
• Reversed operator = when the special method on the first operand can’t be used
• len() is not a method
  • When x is a built-in type, len(x) simply read from a field in the CPython C struct.
  • “Practicality is better than purity”
  • len may be considered as an unary operator
• The Art of the Metaobject Protocol (AMOP)
  • metaobject = the objects that are the building blocks of the language itself
  • an API for core language constructs
  • A rich metaobject protocol enables extending a language to support new programming paradigms