git-missing-commits

#!/usr/bin/env python3
# Author: Jan Larres <jan@majutsushi.net>
# License: MIT/X11

import argparse
import sys
from collections import defaultdict
from subprocess import STDOUT, CalledProcessError, call, check_output
from typing import Tuple


def parse_args() -> argparse.Namespace:
    parser = argparse.ArgumentParser(
        description="Check for missing commits on a branch compared to another branch"
    )
    parser.add_argument(
        "target", metavar="target-branch", help="The branch to apply commits to"
    )
    parser.add_argument(
        "source", metavar="source-branch", help="The branch to pick commits from"
    )
    parser.add_argument(
        "--limit",
        help="Exclude commits up to (and including) LIMIT on the source branch.",
    )
    return parser.parse_args()


def git(args: list[str]) -> list[str]:
    try:
        return check_output(["git"] + args, stderr=STDOUT).decode("utf-8").splitlines()
    except CalledProcessError as e:
        print("Command '" + " ".join(["git"] + args) + "' failed with output:")
        print(e.output.decode("utf-8"))
        sys.exit(1)


def parse_cherry_output(commits: list[str]) -> list[Tuple[str, str]]:
    result = []

    for commit in commits:
        if commit[0] == "-":
            continue
        # Remove '+ ' at the front
        commit = commit[2:]

        commit_hash = commit[: commit.index(" ")]
        summary = commit[commit.index(" ") + 1 :]

        result.append((commit_hash, summary))

    return result


def show_commit(commit: str) -> None:
    call(
        [
            "git",
            "--no-pager",
            "show",
            "--pretty=tformat:%C(auto,yellow)%h %C(auto,blue)%ai %C(auto,green)%<(15,trunc)%an %Creset%s",
            "--no-patch",
            commit,
        ]
    )


def main(args: argparse.Namespace) -> None:
    missing_in_target_args = ["cherry", "-v", args.target, args.source]
    if args.limit is not None:
        missing_in_target_args += [args.limit]
    missing_in_target = parse_cherry_output(git(missing_in_target_args))
    missing_in_source = parse_cherry_output(
        git(["cherry", "-v", args.source, args.target])
    )

    duplicates = []
    uniques = []

    # Store commits missing in the source branch in a multimap keyed by summary,
    # to be able to quickly look up possible duplicates
    src_by_summary = defaultdict(list)
    for (commit_hash, summary) in missing_in_source:
        src_by_summary[summary].append(commit_hash)

    # If the summary of a missing commit is in the above multimap then we add
    # it to the possible duplicats, otherwise it's clearly unique
    for (commit_hash, summary) in missing_in_target:
        if summary in src_by_summary:
            duplicates.append(commit_hash)
            duplicates += src_by_summary[summary]
        else:
            uniques.append(commit_hash)

    if len(uniques) != 0:
        print("Unique missing commits (oldest first):")
        for commit in uniques:
            show_commit(commit)

    if len(duplicates) != 0:
        if len(uniques) != 0:
            print()
        print("Possible duplicates:")
        for commit in duplicates:
            show_commit(commit)


if __name__ == "__main__":
    main(parse_args())