Utility for grouping files by codeower and generating commands

codemapper.py

"""
Utility for grouping files by codeower and generating commands
"""

import argparse
import re
import shlex
from collections import defaultdict
from glob import fnmatch
from typing import DefaultDict, FrozenSet, Iterator, Optional, Set, Tuple


def main():
    parser = argparse.ArgumentParser(
        description="""
        Designed for use with massive editing tools. E.g.

        black $(git ls-files '*.py')
        git diff --name-only | codemapper --from-file - --template 'git checkout -b black-{owners}; git commit {filenames} -m "autoformat"; git checkout master;'
        """
    )
    parser.add_argument("filenames", nargs="*")
    parser.add_argument(
        "--from-file",
        dest="filenames_file",
        type=argparse.FileType("r"),
        help="read file names from file or pipe",
    )

    parser.add_argument(
        "--codeowners", type=argparse.FileType("r"), default=".github/CODEOWNERS"
    )
    parser.add_argument(
        "--template",
        help="command template with optional variables like {filenames} and {owners}",
        default="{owners}: {filenames}",
    )

    args = parser.parse_args()
    if args.filenames_file:
        args.filenames.extend(
            [filename.rstrip('\n') for filename in args.filenames_file.readlines()]
        )

    for owners, filenames in assign(args.codeowners, set(args.filenames)).items():
        print(
            args.template.format(
                owners="-".join(owners),
                filenames=" ".join(map(shlex.quote, filenames)))
        )


def assign(codeowners, filenames: Set[str]):
    """ assign each of the filenames to their codeowners """
    unassigned_filenames = filenames

    by_owner: DefaultDict[FrozenSet[str], Set[str]] = defaultdict(set)
    for file_pattern, owners in list(parse_codeowners(codeowners))[::-1]:
        matches = fnmatch.filter(unassigned_filenames, file_pattern)
        if matches:
            by_owner[owners].update(matches)
            unassigned_filenames.difference_update(matches)

    if unassigned_filenames:
        by_owner[frozenset()] = unassigned_filenames
    return by_owner


def parse_codeowners(codeowners) -> Iterator[Tuple[str, FrozenSet[str]]]:
    for line in codeowners.readlines():
        line = re.sub("#.*", "", line).strip()
        if line:
            file_pattern, *owners = re.split(r"\s+", line)
            file_pattern = file_pattern.lstrip('/')  # / is root of repo

            yield file_pattern, frozenset(owners)

            if not file_pattern.endswith("*"):
                # can't tell file_pattern will match files or directories
                # so yield a file_pattern which treats it as a directory
                yield file_pattern + "*", frozenset(owners)


if __name__ == "__main__":
    main()