Quick and dirty script that print TODOs and FIXMEs in a git repository sorted by commit date

gitodo.py

#!/usr/bin/python3
import os
import sys
import subprocess
import argparse
import dataclasses
from typing import Iterable, Optional
from datetime import datetime, timezone

@dataclasses.dataclass
class Todo:
    filepath: str
    line: str
    line_number: int
    time: datetime
    type_: str

    def __post_init__(self) -> None:
        if self.time is None:
            self.time = datetime.now(timezone.utc)


def parse_args():
    parser = argparse.ArgumentParser()
    return parser.parse_args()


def main():
    dirpath = '.' # TODO: make this a program arg
    assert_git_directory(dirpath)
    blacklist = load_blacklist()
    todos = search_for_todos(dirpath, blacklist)
    display_todos(todos)


def assert_git_directory(dirpath: str) -> bool:
    is_git =  os.path.exists(os.path.join(dirpath, '.git'))
    if not is_git:
        print(f"{sys.argv[0]} must be run in git directory")
        sys.exit(1)

def load_blacklist() -> Iterable[str]:
    submodules = subprocess.check_output(['git', 'submodule', 'status']).decode().split('\n')
    blacklist = []
    for line in submodules:
        if line == '':
            continue
        blacklist.append(line.strip().split(' ')[1])
    return blacklist

def is_in_blacklist(filepath: str, blacklist: Iterable[str]) -> bool:
    return any(filepath.strip('./').startswith(bad.strip('./')) for bad in blacklist)

def search_for_todos(dirpath: str, blacklist: Iterable[str] = None) -> Iterable[Todo]:
    res = subprocess.check_output(['rg', '-n', 'TODO|FIXME', dirpath])
    return parse_rg_output(res, blacklist)


def parse_rg_output(data: bytes, blacklist: Iterable[str]) -> Iterable[Todo]:
    for line in data.decode().split(os.linesep):
        if line == '':
            continue
        filepath, remaining = line.split(":", 1)
        if is_in_blacklist(filepath, blacklist):
            continue
        line_number, line_content = remaining.split(":", 1)
        date = get_last_edit_date_from_line(filepath, line_number)
        yield Todo(filepath, line_content.strip(), int(line_number), date, type_="FIXME" if "FIXME" in line else "TODO")


def get_last_edit_date_from_line(filepath: str, line_number: int) -> Optional[datetime]:
    sed = subprocess.Popen(['sed', '-n', f"{line_number}p"], stdin=subprocess.PIPE, stdout=subprocess.PIPE)
    blame = subprocess.run(['git', 'blame', '-c', filepath], stdout=sed.stdin, stderr=subprocess.PIPE)
    sed_out, _ = sed.communicate()
    if blame.returncode == 128:
        return None  # This file is probably not versionned yet (or gitignored)
    try:
        raw_date = sed_out.decode().split('\t')[2]
        date_str = make_shit_iso(raw_date)
    except IndexError:
        print(f"Warning: could not get date for todo: {filepath} at line {line_number}", file=sys.stderr)
        return None
    return datetime.fromisoformat(date_str)

def make_shit_iso(rawdate: str) -> str:
    partial = rawdate.replace(' ', 'T', 1).replace(' ', '')
    return partial[:-2] + ':' + partial[-2:]

def display_todos(todos: Iterable[Todo]) -> None:
    if sys.stdout.isatty():
        orange = "\033[93m"
        default_color = "\033[0m"
    else:
        orange = ''
        default_color = ''
    for todo in sorted(todos, key=lambda t: t.time):
        formatted_time = todo.time.astimezone().strftime('%Y-%m-%d %H:%M:%S')
        color = orange if todo.type_ == "FIXME" else default_color
        print(f"{color}{formatted_time}\t{todo.filepath}:{todo.line_number}\t{todo.line}{default_color}")


if __name__ == "__main__":
    args = parse_args()
    main(**args.__dict__)