zulip/tools/total-contributions

#!/usr/bin/env python3
import argparse
import os
import pathlib
import subprocess
import sys
from collections import defaultdict
from datetime import datetime
from typing import Dict, List


def add_log(committer_dict: Dict[str, int], input: List[str]) -> None:
    for dataset in input:
        committer_name = dataset.split("\t")[1]
        commit_count = int(dataset.split("\t")[0])
        committer_dict[committer_name] += commit_count


def retrieve_log(repo: str, lower_version: str, upper_version: str) -> List[str]:
    return subprocess.check_output(
        ["git", "shortlog", "-s", lower_version + ".." + upper_version],
        cwd=find_path(repo),
        text=True,
    ).splitlines()


def find_version(time: str, repository: str) -> str:
    """Find the latest release version for the target repository as of the
    specified time.
    """
    versions_list = subprocess.check_output(
        ["git", "tag", "-l", "--sort=creatordate"], cwd=find_path(repository), text=True
    ).splitlines()

    for version in reversed(versions_list):
        version_time = subprocess.check_output(
            ["git", "log", "-1", "--format=%ci", version],
            cwd=find_path(repository),
            text=True,
        ).split()[0]
        if datetime.strptime(version_time, "%Y-%m-%d") < datetime.strptime(time, "%Y-%m-%d"):
            return version

    # This is a bug, but this tool doesn't support ranges for the
    # beginning of a project's history yet.
    raise AssertionError("Could not find any releases prior to tag.")


def find_path(repository: str) -> str:
    return os.path.dirname((pathlib.Path().resolve()).parents[0]) + "/" + repository


# argparse
parser = argparse.ArgumentParser(
    prog="python3 total-contributions",
    formatter_class=argparse.RawTextHelpFormatter,
    description="""\
Aggregates the total commit contributions to Zulip that should be
attributed to the time window between the two provided
zulip/zulip versions (tags or branches).

The attribution algorithm used by this tool attributes all changes for
a Zulip project between:

* The last release of the target project before the first zulip/zulip version.
* The last release of the target project before the last zulip/zulip version.

This algorithm has the key property that the totals for a given contributor of
2.1.0..4.0 will equal the sum of 2.1.0..3.0 and 3.0..4.0.

Its main downside is that contributions to projects other than
zulip/zulip in the last few weeks before a zulip/zulip release will be
delayed (i.e. counted in the total for the next zulip/zulip release).

Expects that all Zulip repositories repositories are in the current working
directory, which does not need to be the directory this is run from.

# Changes between two major releases.
total-contributions 4.0 5.0

# Changes between a release and the current main branch.
total-contributions 4.0 main
total-contributions 2.1.0
    """,
)
parser.add_argument(
    "version",
    metavar="version",
    nargs="*",
    # TODO: Ideally, we'd replace "1.3.0" with "First commit", to
    # simplify including contributions before the 1.3.0 release.
    default=["1.3.0", "main"],
    help="Git tag or branch in zulip/zulip specifying one end of the commit range to use.",
)

parser.add_argument(
    "-a",
    "--ascending",
    action="store_true",
    help="Sort contributors based on number of commits(ascending order)",
)

args = parser.parse_args()

if len(args.version) > 2:
    parser.error("Expects 0 to 2 version number(s)")

lower_zulip_version = args.version[0]
if len(args.version) == 1:
    upper_zulip_version = "main"
else:
    upper_zulip_version = args.version[1]

subprocess.check_call(["git", "fetch"], cwd=find_path("zulip"))

# Extract git version and time. It's important that we use the commit
# date (%ci), not the author date (%ai), since while those are often
# near identical for release commits, if we pass a branch like `main`,
# it's possible the latest commit on the branch might have a months
# old author date if the last pull request merged was started at that
# time.
try:
    lower_time = subprocess.check_output(
        ["git", "log", "-1", "--format=%ci", lower_zulip_version],
        stderr=subprocess.DEVNULL,
        text=True,
    ).split()[0]
    upper_time = subprocess.check_output(
        ["git", "log", "-1", "--format=%ci", upper_zulip_version],
        stderr=subprocess.DEVNULL,
        text=True,
    ).split()[0]
except subprocess.CalledProcessError:
    print("Specified version(s) don't exist")
    sys.exit(0)

out_dict: Dict[str, int] = defaultdict(int)
subprocess.check_call(["git", "fetch"], cwd=find_path("zulip"))
zulip = retrieve_log("zulip", lower_zulip_version, upper_zulip_version)
add_log(out_dict, zulip)

print(f"Commit range corresponds to {lower_time} to {upper_time}")

# TODO: This should probably include more repositories in the zulip organization.
for repository in [
    "zulip-mobile",
    "zulip-desktop",
    "docker-zulip",
    "python-zulip-api",
    "zulip-terminal",
]:
    subprocess.check_call(["git", "fetch"], cwd=find_path(repository))
    lower_repo_version = find_version(lower_time, repository)
    upper_repo_version = find_version(upper_time, repository)
    repo_log = retrieve_log(repository, lower_repo_version, upper_repo_version)
    print(f"Commit range for {repository} {lower_repo_version}..{upper_repo_version}")
    add_log(out_dict, repo_log)

# Sorting based on number of commits
for commit_count, committer_name in sorted(
    out_dict.items(), key=lambda item: item[1], reverse=not args.ascending
):
    print(str(committer_name) + "\t" + commit_count)

print(
    "Total contributors across all Zulip repos within Zulip versions "
    + lower_zulip_version
    + " - "
    + upper_zulip_version
    + ": "
    + str(len(out_dict))
)