A Wadler–Lindig ✨pretty-printer✨ for Python

This library is for you if you need:

• Something like the built-in pprint.pprint, but which consumes less horizontal space. For example in error messages.
• If you have complicated custom types that you'd like to create pretty well-formatted reprs for. For example nested trees of dataclasses / PyTorch modules / etc.

Main features:

• Absolutely tiny implementation (96 lines of code for the main Wadler–Lindig algorithm, 342 more for teaching it how to handle all Python types).
• Simpler than the original algorithm by Wadler & Lindig (removes some dead code).
• Supports multi-line unbroken text strings.
• Supports ANSI escape codes and colours.
• Zero dependencies.

Installation

pip install wadler_lindig

Documentation

Available at https://docs.kidger.site/wadler_lindig.

Example

import dataclasses
import numpy as np
import wadler_lindig as wl

@dataclasses.dataclass
class MyDataclass:
    x: list[str]
    y: np.ndarray

obj = MyDataclass(["lorem", "ipsum", "dolor sit amet"], np.zeros((2, 3)))

wl.pprint(obj, width=30, indent=4)
# MyDataclass(
#     x=[
#         'lorem',
#         'ipsum',
#         'dolor sit amet'
#     ],
#     y=f64[2,3](numpy)
# )

API at a glance

For day-to-day pretty-printing objects: pprint (to stdout), pformat (as a string), pdiff (between two objects).

For creating custom pretty-printed representations:

• The core Wadler–Lindig document types: AbstractDoc, BreakDoc, ConcatDoc, GroupDoc, NestDoc, TextDoc.
• pdoc will convert any Python object to a Wadler–Lindig document, with the __pdoc__ method called on custom types if it is available.
• ansi_format (to add ANSI colour codes), ansi_strip (to remove ANSI codes).
• Several common helpers for creating Wadler–Lindig documents: array_summary, bracketed, comma, join, named_objs.

FAQ

MyDataclass(
  x=SomeNestedClass(
    y=[1, 2, 3]
  )
)

MyDataclass(x=SomeNestedClass(y=[1,
                                 2,
                                 3]))