git-time-tracker/git_time_tracker/__init__.py

import argparse
import sys
import time
import dataclasses
import git
import datetime
from collections.abc import Iterator, Sequence
from pathlib import Path

import logging

logger = logging.getLogger(__name__)

def parse_arguments():
    parser = argparse.ArgumentParser()
    parser.add_argument('repositories', action='extend', nargs='+', type=Path)
    return parser.parse_args()

@dataclasses.dataclass(frozen=True, order=True)
class WorkSample:
    registered_at: datetime.datetime
    labels: Sequence[str]


def determine_default(repo: git.Repo):
    try:
        repo.commit('main')
        return 'main'
    except:
        return 'master'

HIDDEN_LABEL_PREFIX = '__'
HIDDEN_LABEL_TOTAL = HIDDEN_LABEL_PREFIX + 'TOTAL'

def determine_project_name(repo: git.Repo) -> str:
    remotes = repo.remotes
    if len(remotes) > 0:
        return remotes.origin.url.removeprefix('git@gitfub.space:')
    return Path(repo.working_tree_dir).name

def get_samples_from_project(repo: git.Repo) -> Iterator[WorkSample]:
    project_name = determine_project_name(repo)
    assert project_name is not None

    # TODO: Branch on main or master or default

    repo.commit()

    for commit in repo.iter_commits(determine_default(repo)):
        labels = [HIDDEN_LABEL_TOTAL]
        labels.append('project:' + project_name)
        labels.append('author:' +  commit.author.email)
        yield WorkSample(datetime.datetime.fromtimestamp(commit.authored_date, tz=datetime.UTC), tuple(labels))
        yield WorkSample(datetime.datetime.fromtimestamp(commit.committed_date, tz=datetime.UTC), tuple(labels))
        del labels

DEFAULT_EST_TIME=datetime.timedelta(hours=1)

ZERO_DURATION = datetime.timedelta(seconds = 0)
HOUR = datetime.timedelta(hours = 1)

def generate_report(samples: list[WorkSample]) -> Iterator[str]:
    #SAMPLE_FILTER = {}
    SAMPLE_FILTER = {'author:jonjmaa@gmail.com', 'author:jon.michael.aanes@secata.com'}
    LABEL_FILTER = {'project'}

    # Time spent per label
    time_per_label: dict[str, datetime.timedelta] = {}
    prev_time = datetime.datetime.fromtimestamp(0, datetime.UTC)
    for sample in samples:
        est_time: datetime.timedelta = DEFAULT_EST_TIME
        est_time = min(sample.registered_at - prev_time, est_time)

        if len(SAMPLE_FILTER) == 0:
            pass
        elif not set(sample.labels).intersection(SAMPLE_FILTER):
            continue

        for label in sample.labels:
            time_per_label.setdefault(label,ZERO_DURATION)
            time_per_label[label] += est_time

        prev_time = sample.registered_at
        del sample, est_time

    time_and_label = [(duration, label) for label,duration in time_per_label.items()]
    time_and_label.sort(reverse=True)

    #
    yield '-' * 66
    yield '\n'
    for (total_time, label_and_type) in time_and_label:
        if label_and_type.startswith(HIDDEN_LABEL_PREFIX):
            continue

        label_type, label = label_and_type.split(':', 1)

        if label_type not in LABEL_FILTER:
            continue

        label_type = '' # TODO

        yield f'  {label_type:8} {label:40}  {total_time / HOUR:-4.0f} hours\n'
        del label, total_time

    yield '-' * 66
    yield '\n'

    yield '  {label_type:8} {label:40}  {hours:-4.0f} hours\n'.format(label_type='', label='TOTAL', hours = time_per_label[HIDDEN_LABEL_TOTAL] / HOUR)

def main():
    logging.basicConfig()

    args = parse_arguments()

    shared_time_stamps: set[WorkSample] = set()
    for repo_path in args.repositories:
        try:
            repo = git.Repo(repo_path)
        except git.exc.InvalidGitRepositoryError:
            logger.warning('Ignoring non-repo %s', repo_path)
            continue
        logger.warning('Visit %s', repo_path)
        shared_time_stamps |= set(get_samples_from_project(repo))

    shared_time_stamps = sorted(shared_time_stamps)

    for t in generate_report(shared_time_stamps):
        sys.stdout.write(t)