1
0
git-time-tracker/git_time_tracker/__init__.py

173 lines
5.0 KiB
Python
Raw Normal View History

2024-06-03 20:42:05 +00:00
import argparse
2024-06-03 21:01:45 +00:00
import dataclasses
import datetime
2024-06-08 12:43:44 +00:00
import logging
import sys
import time
2024-06-03 21:01:45 +00:00
from collections.abc import Iterator, Sequence
2024-06-03 21:13:16 +00:00
from pathlib import Path
2024-06-03 20:42:05 +00:00
2024-06-08 12:43:44 +00:00
import git
2024-06-03 22:15:18 +00:00
logger = logging.getLogger(__name__)
2024-06-08 12:43:44 +00:00
2024-06-03 21:49:42 +00:00
@dataclasses.dataclass(frozen=True, order=True)
2024-06-03 21:01:45 +00:00
class WorkSample:
registered_at: datetime.datetime
labels: Sequence[str]
2024-06-03 22:02:49 +00:00
def determine_default(repo: git.Repo):
try:
repo.commit('main')
return 'main'
except:
return 'master'
2024-06-08 12:43:44 +00:00
2024-06-03 22:02:49 +00:00
HIDDEN_LABEL_PREFIX = '__'
HIDDEN_LABEL_TOTAL = HIDDEN_LABEL_PREFIX + 'TOTAL'
2024-06-08 12:43:44 +00:00
2024-06-03 22:15:18 +00:00
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
2024-06-08 12:43:44 +00:00
2024-06-03 21:01:45 +00:00
def get_samples_from_project(repo: git.Repo) -> Iterator[WorkSample]:
2024-06-03 22:30:55 +00:00
project_name = determine_project_name(repo)
assert project_name is not None
2024-06-03 21:01:45 +00:00
2024-06-03 21:13:16 +00:00
# TODO: Branch on main or master or default
2024-06-03 22:02:49 +00:00
repo.commit()
for commit in repo.iter_commits(determine_default(repo)):
2024-06-03 22:30:55 +00:00
labels = [HIDDEN_LABEL_TOTAL]
labels.append('project:' + project_name)
2024-06-08 12:43:44 +00:00
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),
)
2024-06-03 22:30:55 +00:00
del labels
2024-06-03 20:42:05 +00:00
2024-06-03 21:49:42 +00:00
2024-06-08 12:43:44 +00:00
DEFAULT_EST_TIME = datetime.timedelta(hours=1)
ZERO_DURATION = datetime.timedelta(seconds=0)
HOUR = datetime.timedelta(hours=1)
2024-06-08 12:49:26 +00:00
MINUTE = datetime.timedelta(minutes=1)
2024-06-08 12:43:44 +00:00
2024-06-03 21:49:42 +00:00
2024-06-08 13:09:57 +00:00
def fmt_year_ranges_internal(years: list[int]) -> Iterator[str]:
years = sorted(years)
for idx, year in enumerate(years):
at_end = idx == len(years) - 1
range_before = idx > 0 and years[idx-1] == year - 1
range_after = not at_end and years[idx+1] == year + 1
if not range_before or not range_after:
yield str(year)
if not at_end:
if not range_before and range_after:
yield '-'
elif not range_after:
yield ','
def fmt_year_ranges(years: list[int]) -> str:
return ''.join(list(fmt_year_ranges_internal(years)))
2024-06-08 12:49:26 +00:00
def fmt_line(label_type: str, label: str, total_time: datetime.timedelta) -> str:
hours = int(total_time / HOUR)
minutes = int((total_time - hours*HOUR)/MINUTE)
return f' {label_type:10} {label:40} {hours:-4d}h {minutes:-2d}m'
2024-06-08 12:49:26 +00:00
def generate_report(samples: list[WorkSample], sample_filter = frozenset()) -> Iterator[str]:
2024-06-03 22:46:15 +00:00
LABEL_FILTER = {}
2024-06-03 21:49:42 +00:00
# Time spent per label
time_per_label: dict[str, datetime.timedelta] = {}
2024-06-08 13:09:57 +00:00
years_per_label: dict[str, set[int]] = {}
2024-06-03 21:49:42 +00:00
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:
2024-06-03 22:30:55 +00:00
pass
elif not set(sample.labels).intersection(sample_filter):
2024-06-03 22:30:55 +00:00
continue
2024-06-03 21:49:42 +00:00
for label in sample.labels:
2024-06-08 12:43:44 +00:00
time_per_label.setdefault(label, ZERO_DURATION)
2024-06-03 21:49:42 +00:00
time_per_label[label] += est_time
2024-06-08 13:09:57 +00:00
years_per_label.setdefault(label,set()).add(sample.registered_at.year)
2024-06-03 21:49:42 +00:00
prev_time = sample.registered_at
del sample, est_time
2024-06-08 12:43:44 +00:00
time_and_label = [(duration, label) for label, duration in time_per_label.items()]
2024-06-03 22:02:49 +00:00
time_and_label.sort(reverse=True)
2024-06-03 21:49:42 +00:00
#
2024-06-03 22:30:55 +00:00
yield '-' * 66
2024-06-03 22:02:49 +00:00
yield '\n'
2024-06-08 12:43:44 +00:00
for total_time, label_and_type in time_and_label:
2024-06-03 22:30:55 +00:00
if label_and_type.startswith(HIDDEN_LABEL_PREFIX):
2024-06-03 22:02:49 +00:00
continue
2024-06-03 22:30:55 +00:00
label_type, label = label_and_type.split(':', 1)
2024-06-03 22:46:15 +00:00
if len(LABEL_FILTER) > 0 and label_type not in LABEL_FILTER:
2024-06-03 22:30:55 +00:00
continue
2024-06-08 12:49:26 +00:00
yield fmt_line(label_type, label, total_time)
2024-06-08 13:09:57 +00:00
yield ' ('
yield fmt_year_ranges(years_per_label.get(label_and_type,[]))
yield ')'
yield '\n'
2024-06-03 21:49:42 +00:00
del label, total_time
2024-06-03 22:30:55 +00:00
yield '-' * 66
2024-06-03 21:49:42 +00:00
yield '\n'
2024-06-08 12:49:26 +00:00
yield fmt_line('', 'TOTAL', time_per_label.get(HIDDEN_LABEL_TOTAL, ZERO_DURATION))
2024-06-08 13:09:57 +00:00
yield '\n'
2024-06-03 22:02:49 +00:00
def parse_arguments():
parser = argparse.ArgumentParser()
parser.add_argument('repositories', action='extend', nargs='+', type=Path)
parser.add_argument('--filter', action='extend', nargs='+', type=str, dest='sample_filter', default=[])
return parser.parse_args()
2024-06-03 20:42:05 +00:00
def main():
2024-06-03 22:15:18 +00:00
logging.basicConfig()
2024-06-03 20:42:05 +00:00
args = parse_arguments()
2024-06-03 21:13:16 +00:00
shared_time_stamps: set[WorkSample] = set()
for repo_path in args.repositories:
2024-06-03 22:15:18 +00:00
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)
2024-06-03 21:13:16 +00:00
shared_time_stamps |= set(get_samples_from_project(repo))
2024-06-03 21:49:42 +00:00
shared_time_stamps = sorted(shared_time_stamps)
for t in generate_report(shared_time_stamps, sample_filter=args.sample_filter):
2024-06-03 21:49:42 +00:00
sys.stdout.write(t)