perflint

perflint is an extension for pylint for performance anti-patterns.

Installation

$ uv add --dev perflint

Use

… as an independent linter

$ uv run perflint src/

… as pylint plugin

$ uv run perflint src/ --load-plugins=perflint

… as pre-commit hook ~~~~~~~~~~~~~~~~~x~~~

perflint can also be used with the pre-commit framework:

.pre-commit-config.yaml

repos:
  - repo: https://github.com/tonybaloney/perflint
    rev: 0.8.1
    hooks:
    - id: perflint

Rules

W8101: unnecessary-list-cast

Unnecessary use of list() on an already iterable type.

W8102: incorrect-dictionary-iterator

Incorrect iterator method for dict: Python dictionaries store keys and values in two separate tables. They can be iterated separately. Using .items() and discarding either the key or the value with _ is inefficient when .keys() or .values() can be used instead.

W8201: loop-invariant-statement

The loop is examined to determine statements or expressions whose result is constant on each iteration of a loop because they are based on named variables that are not changed during the iteration.

W8202: loop-global-usage

Global name usage in a loop: loading global variables is slower than loading local variables. The difference is marginal, but when passed in a loop, there can be a noticeable speed improvement.

R8203: loop-try-except-usage

Up until Python 3.10, try…except blocks are very computationally intensive compared to if statements.

Avoid using them in a loop as they can cause significant overhead. Refactor your code so that no iteration-specific details are required and put the entire loop in the try block.

W8204: memoryview-over-bytes

Slicing byte objects in loops is inefficient because it creates a copy of the data. Use memoryview() instead.

See also

• Zero-copy interactions

• Memoryview Benchmarks

• Memoryview Benchmarks 2

W8205: dotted-import-in-loop

Direct import of the name %s is more efficient in a loop. In Python, you can import a module and then access submodules as attributes. You can also access functions as attributes of that module. This keeps the import statements to a minimum. However, if you use this method in a loop, it is inefficient because each loop pass loads the global, then the attribute, then the method.

W8301: use-tuple-over-list

Use a tuple instead of a list for an immutable sequence: both the construction and indexing of a tuple is faster than that of a list.

W8401: use-list-comprehension

Use list comprehensions with or without an if statement instead of a for loop.

W8402: use-list-copy

Use list.copy() instead of a for loop.

W8403: use-dict-comprehension

Uses a dictionary comprehension instead of a simple for loop.