A Python script to categorise a list of filepaths by code owners

codeowner-breakdown

#!/usr/bin/env python
"""
Print a breakdown of the passed filepaths by CODEOWNER team.

This script requires the `codeowners` CLI tool to be installed and available on the $PATH.
https://github.com/hmarr/codeowners

Intended usage is to pipe filepaths into this script from the root of a repo:

    cat filepaths.txt | codeowner-breakdown

On an M3 MBP, breaking down 5000 lines of input takes ~1 minute.
"""
import sys
import itertools
import collections
import subprocess


def _group_by_codeowner(filepaths: list[str]) -> str:
    """
    Return a codeowner summary of the passed filepaths.
    """
    # Group the filepaths to minimise the number of codeowner lookups.
    grouped = {filepath: len(list(group)) for filepath, group in itertools.groupby(sorted(filepaths))}
    codeowner_map = collections.defaultdict(int)
    for filepath, count in grouped.items():
        for codeowner in _determine_codeowners(filepath):
            codeowner_map[codeowner] += count

    # Sort results with most frequent team first. 
    ranked = sorted(codeowner_map.items(), key=lambda x: x[1], reverse=True)

    # Create a string summary
    output_lines = [f"{codeowner}: {count}" for codeowner, count in ranked]
    return "\n".join(output_lines)


def _determine_codeowners(filepath: str) -> str:
    """
    Return the codeowner for the passed filepath.
    """
    command = ["codeowners", filepath]
    output = subprocess.check_output(command, shell=False).decode('utf-8')

    # The output has form:
    # <filepath>    @team1 @team2 ...
    # We return all the teams.
    return output.split()[1:]


if __name__ == "__main__":
    # Convert STDIN into a list of filepaths.
    filepaths = [line.strip() for line in sys.stdin.readlines()]
    print(_group_by_codeowner(filepaths))