Skip to main content

pyrefact

Pyrefact does advanced python refactoring, with the goal of simplifying complicated code, deleting dead code, and improving performance. Pyrefact is entirely rule-based.

Features

Readability

  • Move common code in if/else blocks before or after the if/else.
  • De-indent code with early continue and return statements.
  • Simplify boolean expressions.
  • Replace for loops with immediate if conditions with filter.
  • Remove commented code.
  • Move code into primitive functions.
  • Replace loops that only fill up lists, dicts or sets with comprehensions.
  • Invert if/else to put the smaller block first.
  • Rename variables, functions and classes with conventions.
  • Rewrite defaultdict-like use of dictionaries with collections.defaultdict().
  • Formats lines that are longer than 100 characters with black.
  • Put overused constant expressions in variables.
  • Remove redundancies and improve definitions of lists, sets, dicts.
  • Use is instead of == for comparisons to None, True and False.

Performance

  • Replace sum comprehensions and for loops with constant expressions. The symbolic algebra tool Sympy is used under the hood.
  • Replace hardcoded inlined collections and comprehensions with set or generator equivalents in places where that would improve performance.
  • Replace sorted()[:n] with heapq.nsmallest, replace sorted()[0] with min.
  • Replace matrix operation comprehensions with equivalent np.matmul() and np.dot() calls, for code that already depends on numpy.
  • Replace pandas .loc[], .iloc[] and .iterrows() with .at[], .iat[] and .itertuples()/.index where appropriate.
  • Move constant code in loops before the loops.
  • De-interpolate interpolated logging calls.

Removing dead and useless code

  • Delete unused functions, classes and variables.
  • Remove most pointless simple statements.
  • Remove branches of code that obviously do nothing useful.
  • Remove unreachable code.
  • Rename unused variables to _.
  • Delete variables named _, unless where that would cause a syntax error.
  • Remove redundant chained calls involving sorted(), set(), tuple(), reversed(), iter() and list().
  • Remove duplicate function definitions.
  • Remove redundant elif and else.
  • Remove unused self and cls function arguments, and add @staticmethod or @classmethod.
  • Move functions decorated with @staticmethod outside of their class namespaces.
  • Simplify deterministic if, elif and else statements.

Imports

  • Delete unused imports.
  • Refactor star imports (e.g. from pathlib import *) to normal imports (e.g. from pathlib import Path)
  • Move builtin and otherwise safe imports to toplevel.
  • Replace indirect imports with direct imports, in cases where a name is imported from a file that also imports that name. Exceptions exists for __init__.py files, and files that define __all__.
  • Merge or remove duplicate or partially duplicate imports
  • Break out stacked plain imports to individual lines
  • Add missing imports by guessing what you probably wanted.
    • For example, if Sequence is used but never defined, it will insert from typing import Sequence at the top of the file.

Pattern-matching

Pyrefact supports pattern-matching analogous to Python's builtin re library. The functions finditer, findall, sub, subn, search, match, fullmatch and compile are implemented:

>>> from pyrefact import pattern_matching
>>> source = """
... x = 1
... y = "asdf"
... """
>>> pattern = "x = 1"
>>> list(pattern_matching.finditer(pattern, source))
[Match(span=Range(start=1, end=6), source='\nx = 1\ny = "asdf"\n', groups=(<ast.Assign object at 0x1015f38e0>,))]
>>> pattern_matching.findall(pattern, source)
['x = 1']
>>> pattern_matching.sub(pattern, "x = 33 - x", source)
'\nx = 33 - x\ny = "asdf"\n'
>>> pattern_matching.search(pattern, source)
Match(span=Range(start=1, end=6), source='\nx = 1\ny = "asdf"\n', groups=(<ast.Assign object at 0x103acaf20>,))

Pattern-matching can also be used from the command-line:

python -m pyrefact.pattern_matching find "x = {{value}}" /path/to/filename.py
python -m pyrefact.pattern_matching replace "x = {{value}}" "x = 1 - {{value}} ** 3" /path/to/filename.py

Installation

Pyrefact can be installed with pip, and works on Python 3.8 or newer:

pip install pyrefact

Usage

The --preserve flag lets you define places where code is used. When this is set, pyrefact will try to keep these usages intact. The --safe flag will entirely prevent pyrefact from renaming or removing code. The --from-stdin flag will format code recieved from stdin, and output the result to stdout.

pip install pyrefact
pyrefact /path/to/filename.py --preserve /path/to/module/where/filename/is/used
pyrefact /path/to/filename.py --safe
cat /path/to/filename.py | pyrefact --from-stdin

It is possible to disable pyrefact for a given file by adding a comment with pyrefact: skip_file anywhere in the file, as done here.

Contributing

To contribute to Pyrefact, please view CONTRIBUTING.md

VS Code Extension

Pyrefact is also available as a VS Code extension, simply named Pyrefact. The extension allows you to use pyrefact as your formatter, similar to how other formatting extensions work.

Pyrefact always runs with the --safe flag when used through the VS Code extension.

The extension is published through the VS Code Marketplace, and the source code is available at pyrefact-vscode-extension.

Prerequisites

Pyrefact is tested on CPython 3.8, 3.9, 3.10, 3.11 and 3.12, and on Windows, MacOS and Linux. Pyrefact is also tested on PyPy3.10.

PyPy tests may be removed in the future, see https://github.com/OlleLindgren/pyrefact/issues/25. I will add tests for new CPython versions when they enter alpha, and remove tests when they become EOL, along with any special logic in place to support those versions.

Metadata

Release files for pyrefact 99

For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.

Source distribution (sdist)

Source distribution for pyrefact 99
File Size Uploaded Provenance Actions
pyrefact-99.tar.gz 85.7 kB Without provenance Details for pyrefact-99.tar.gz

Built distribution (wheel)

Table of built distributions (wheels) for pyrefact 99
File Interpreter ABI Platform Actions
pyrefact-99-py3-none-any.whl Python 3 none any Details for pyrefact-99-py3-none-any.whl

Total release size: 175.0 kB

Release files / pyrefact-99.tar.gz

Download URL pyrefact-99.tar.gz
Size 85.7 kB
Tags Source
SHA-256 checksum
How to use checksums
1c09e9d7f035d23630a62c14731b154bc4d94265806e1e8ce0dc30e151b06c4c
BLAKE2b-256 checksum
How to use checksums
95cb19ffac70226afede954b3fa448158951e395c35bc867ed974a5e4963f133
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
No
Uploaded via twine/5.0.0 CPython/3.12.2

Release files / pyrefact-99-py3-none-any.whl

Download URL pyrefact-99-py3-none-any.whl
Size 89.3 kB
Tags Python 3
SHA-256 checksum
How to use checksums
45ffa8657ba6ed946c7b28eba09ee60ac84d010a45f9ecf356e83b530a383f38
BLAKE2b-256 checksum
How to use checksums
dfdeffc5136126ed3d5d62390c5a17fb9c6ce087e77f3e1bcc8b58b730e4930c
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
No
Uploaded via twine/5.0.0 CPython/3.12.2

Security for pyrefact 99

Provenance

Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.

No provenance found

None of the files in this release were uploaded with provenance metadata.

A note on package security

PyPI is a repository for distributing packages, not a service for auditing them. While third-party security researchers and our staff do their best to detect and remove malicious packages, PyPI cannot guarantee the quality, security or suitability of any given package.

To learn more about protecting yourself from malicious actors, visit our package security documentation.

Release history Release notifications | RSS feed

This release

99 This release

2 release files No provenance

Anthropic, PBC Visionary sponsor Bloomberg Visionary sponsor Hudson River Trading Visionary sponsor Meta Visionary sponsor NVIDIA Visionary sponsor Microsoft Sustainability sponsor Depot Continuous Integration AWS Cloud computing and Security Sponsor Datadog Monitoring Fastly CDN Google Download Analytics Sentry Error logging StatusPage Status page