sdsort

Sorts top-level functions and class methods in Python files according to the step-down rule, as described in Robert C. Martin's Clean Code. More concretely, functions and methods are ordered in a depth-first-traversal order of the dependency tree, so higher-level code appears before lower-level code.

pip install sdsort

Or run directly with uv:

uvx sdsort <path>

To target individual files, run the sdsort command, followed by the paths to the files that should be sorted:

sdsort <file_1> <file_2>

To sort all *.py files in a directory, and all of its subdirectories, run the sdsort command followed by the directory path:

sdsort <directory_path>

To check if files are already sorted without modifying them, use the --check flag:

sdsort --check <file_or_directory>

This will exit with code 1 if any files would be re-arranged, making it suitable for CI pipelines and pre-commit hooks.

Larger runs are sorted across parallel worker processes. Use --jobs/-j to control how many:

sdsort --jobs 4 <directory_path>

0 (the default) picks one worker per available CPU; 1 disables parallelism and sorts every file in the current process. Runs of fewer than 50 files always stay serial regardless of this setting, since spawning workers costs more than it saves at that scale.

If sdsort cannot parse a file, it reports the file on stderr and moves on. The rest of the files are still sorted, and the exit code is unaffected.

To prevent sdsort from modifying a particular file, add a # sdsort: skip_file comment somewhere within the file.

sdsort can be used as a pre-commit hook. Add the following to your .pre-commit-config.yaml:

repos:
  - repo: https://github.com/eirikurt/sdsort
    rev: v1.2.0
    hooks:
      - id: sdsort

It's early days. Consider this a beta for now.