git_oh/git_oh.py

#!/usr/bin/python3
""" Entrouvert test """

import argparse
import calendar
import csv
import datetime
import shutil
import sys
import tempfile
import time
import warnings

import git


def main():
    """ Main function parse args from CLI, count commits in given 
        repository, and display/output them
    """
    args = parse_args()

    repo = TempRemoteRepo(args.url, silent=args.silent)

    commit_stats = {}
    last_progress = time.time()
    for i, commit in enumerate(repo):
        if not args.silent and time.time() - last_progress > 0.1:
            last_progress = time.time()
            print(f"Processing commit {f'#{i}':5s}", end="\r",
                    file=sys.stderr)
        commit_datetime = commit.authored_datetime
        if args.from_date is not None and \
                args.from_date > commit_datetime.date():
            continue
        if args.to_date is not None and\
                args.to_date < commit_datetime.date():
            continue
        in_oh = in_office_hours(commit_datetime,
                    starthour=args.daystart,
                    stophour=args.daystop,
                    weekend=args.weekend)
        commit_by_author(commit.author, commit_datetime, in_oh, commit_stats,
                group=args.group_by)
        if args.verbose and not in_oh:
            print(f"{commit} \
{commit.authored_datetime.strftime('%a %H:%M %Y-%m-%d')} \
{commit.author.name!r}")

    if len(commit_stats) == 0:
        print("No commit in given repository/date range", file=sys.stderr)
        return

    if args.csv_output is None:
        if args.group_by:
            print(result_cli(commit_stats, args.details))
        else:
            print(short_result_cli(commit_stats))
    else:
        result_csv(commit_stats, args.csv_output)
        args.csv_output.close()


def in_office_hours(moment=datetime.datetime.now(),
        starthour=datetime.time(8,0,0),
        stophour=datetime.time(20,0,0),
        weekend=(5,6)):
    """ Indicates if a moment is in office hours.

    Office hours are localized, comparisons is done without
    taking care of tzoffset : in fact, if a working day starts at
    08:00, 07:59:59+0200 is off as 07:59:59+0000 is.

    Arguments :
        - moment : the datetime.datetime to compare with office hours
        - starthour : standard day start datetime.time
        - stophour : standard day stop datetime.time
        - weekend : list of days off (0,1,....,6)

    Returns True if in office hours else False
    """
    for dow in weekend:
        if dow < 0 or dow > 6:
            raise ValueError("Weekend days are integer in [0..6]")

    if starthour.tzinfo is not None or stophour.tzinfo is not None:
        warnings.warn("Start & stop hours should not indicate any \
tzinfo : comparisons are done without taking tzoffset in considaration")

    if moment.weekday() in weekend:
        return False
    localtime = moment.time()
    if stophour == datetime.time(0,0):
        return starthour <= localtime
    if stophour < starthour:
        return localtime >= starthour or localtime <= stophour
    return starthour <= localtime <= stophour


def commit_by_author(author, moment, in_oh, acc, group=None):
    """ Update stats with commits total and commits off office hour by
        authors

    Arguments :
    - author : The commit author (git.util.Author instance)
    - moment : The commit authored datetime.datetime
    - in_oh : boolean indicating if a commit is in office hours
    - acc : the dictionnary accumulating the commits count (modified by
        reference). Keys are author instances and values are dict with
        counters values.
        Counters are dict with group as key. The group 'all' is always
        present.
        When grouping by month group key in format YEAR-MONTH are added.
        When grouping by week group key in format YEAR-WEEK are added.
        Each group is a dict with keys :
        - 'total' for total number of commit
        - 'off_oh' for the number of commit out off office hours
    - group : Commit count can be aggregated by 'month' or 'week'

    Returns None, modifications are done in acc by reference
    """
    group_key = None
    if group == 'week':
        cal = moment.isocalendar()
        group_key = f"{cal.year:04d}-W{cal.week:02d}"
    elif group == 'month':
        group_key = f"{moment.year:04d}-{moment.month:02d}"
    elif group is None:
        group_key = None
    else:
        err = f"Invalid group {group!r}. Valid groups are week and month"
        raise ValueError(err)

    if author not in acc:
        acc[author] = {'all': {'off_oh': 0, 'total': 0}}
    acc[author]['all']['total'] += 1
    if group_key:
        if group_key not in acc[author]:
            acc[author].update({group_key: {'off_oh': 0, 'total': 0}})
        acc[author][group_key]['total'] += 1

    if not in_oh:
        acc[author]['all']['off_oh'] += 1
        if group_key:
            acc[author][group_key]['off_oh'] += 1


def iter_commits(repo):
    """ Generator on all git commits in given repository.

    Recursively iterate on each commit in each branch/ref not yielding
    duplicates commits (based on commit hashes)

    Arguments :
        - repo : The repository instance to fetch commits from
    """
    encountered = set()
    for ref in repo.refs:
        for commit in repo.iter_commits(ref.name):
            if commit.binsha not in encountered:
                encountered.add(commit.binsha)
                yield commit

class TempRemoteRepo(git.repo.base.Repo):
    """ A temporary repository referencing a remote repository

    Allows to iterate on all commits without cloning the remote.
    If repository is local iterate on local commit without using
    a temporary repository
    """

    def __init__(self, remote_url, silent=True, force_remote=False):
        """ Initialize a new empty repository referencing a remote repo in
        order to iterate on its commit

        Arguments :
            - remote_url : The url of the remote repo to reference
            - silent : if False display progress on stderr
            - force_remote : if True create a temporary repo even for
                local file:// url
        """
        self.temppath = None
        if (not force_remote) and remote_url.startswith("file://"):
            repo_path = remote_url[len("file://"):]
        else:
            repo_path = self.temppath = tempfile.mkdtemp(prefix="git_oh_")
            git.Repo.init(self.temppath)
        super().__init__(repo_path)

        if self.temppath is not None:
            remote_progress = None if silent else RemoteFetchProgress()
            origin = self.create_remote("origin", remote_url)
            origin.fetch(progress=remote_progress)

    def __del__(self):
        """ Clean temporary repository if needed """
        if self.temppath is not None:
            shutil.rmtree(self.temppath)
        super().__del__()

    def __iter__(self):
        """ Iterate on commits """
        return iter_commits(self)


class RemoteFetchProgress(git.RemoteProgress):
    """ Report progress for remote fetch """

    def update(self, op_code, cur_count, max_count=None, message=""):
        """ Display progression on a single line of stderr """
        if max_count:
            pct = (cur_count / max_count)*100
            msg = f"{pct:5.1f}% {message}"
            print(f"{msg:<80s}", end="\r", file=sys.stderr)


def result_csv(commit_stats, ofd=sys.stdout):
    """ Output the stats in an open file as CSV

    Output 3 columns by group :
    - GRP-total : the number of commit in the group
    - GRP-off_oh : the number of commit off office hour
    - GRP-prop_off_oh : the ratio off_oh / total

    Arguments :
    - commit_stats : A dict with author as key and stats as values (
        see commit_by_author() )
    - ofd : Output TextIO
    """
    groups = set()
    for author, stats in commit_stats.items():
        groups.update(stats.keys())

    for author in commit_stats:
        for group in groups:
            commit_stats[author].setdefault(group, {'total': 0, 'off_oh': 0})
            gstat = commit_stats[author][group]
            if gstat['total']:
                gstat['prop_off_oh'] = gstat['off_oh']/gstat['total']
            else:
                gstat['prop_off_oh'] = 0.0

    groups = sorted(groups)
    all_keys  = ['author', 'author_email']
    for group in groups:
        all_keys += [
                f"{group}-total",
                f"{group}-off_oh",
                f"{group}-prop_off_oh"]
    writer = csv.DictWriter(ofd, all_keys)
    writer.writeheader()
    for author, stats in commit_stats.items():
        row = {'author': author.name,
               'author_email': author.email}
        for kgroup, values in stats.items():
            row.update({f'{kgroup}-{kval}': val
                        for kval, val in values.items()})
        writer.writerow(row)


def result_cli(commit_stats, details=True):
    """ Format stats for cli output

    Arguments :
    - commit_stats : A dict with author as key and stats as values (
        see commit_by_author() )

    Returns a string representing an ascii array with percentage of commits
    off office hours.
    """

    all_keys = set()
    for stats in commit_stats.values():
        all_keys.update(stats.keys())
    all_keys = sorted(all_keys)
    stats_count = len(all_keys)

    stat_width = 10

    author_title = "author | % off"

    author_width = max((len(author.name) for author in commit_stats))
    author_width = max([len(author_title), author_width])
    author_width += 2

    hsep = '+' + ('-' * author_width) + '+'
    hsep += (('-' * stat_width)+ '+') * stats_count
    hsep += "\n"

    if details:
        header = "| Author |"
        header += f"{'% off':>{author_width - len('Author |') - 2}s} |"
        for key in all_keys:
            header += f"{key:>{stat_width-1}s} |"
        header += "\n"
        header += f"| {'':>{len('Author')}s} |"
        header += f"{'total':>{author_width - len('Author |') - 2}s} |"
        for key in all_keys:
            header += f"{'':>{stat_width-1}s} |"
    else:
        header = f"|{'Author':>{author_width-1}s} |"
        for key in all_keys:
            header += f"{key:>{stat_width-1}s} |"

    result = f"{hsep}{header}\n{hsep}"
    for author, stats in commit_stats.items():
        # First line of stat with percent off
        result += f"|{author.name:>{author_width-1}s} |"
        for key in all_keys:
            if key not in stats:
                result += f"{'none':>{stat_width-1}s} |"
            else:
                pct = (stats[key]['off_oh'] / stats[key]['total'])*100
                result += f"{pct:{stat_width-2}.1f}% |"
        # Second line of stat with total
        if details:
            result += f"\n|{'':>{author_width-1}s} |"
            for key in all_keys:
                if key not in stats:
                    result += f"{0:{stat_width-1}d} |"
                else:
                    result += f"{stats[key]['total']:{stat_width-1}d} |"
        result += f"\n{hsep}"

    return result


def short_result_cli(commit_stats):
    """ Format stats for cli output when no grouping

    Arguments :
    - commit_stats : A dict with author as key and stats as values (
        see commit_by_author() )

    Returns a printable string of author and their details about
    commits off office hours.
    """
    return '\n'.join([f"{author.name:>20s} : \
{stat['all']['off_oh']/stat['all']['total']*100:5.1f}% \
( {stat['all']['off_oh']:4d}/{stat['all']['total']:<5d} off)"
                      for author, stat in commit_stats.items()])


def valid_day(value):
    """ Take a locale dayname and convert it to a day number

    Valid days are calendar.day_name, calendar.day_abbr and integers
    in [0..6]

    Arguments :
        - value : the str to convert

    Returns an int in [0..6]
    """

    value = value.lower()

    try:
        return [day.lower() for day in calendar.day_name].index(value)
    except ValueError:
        pass

    try:
        return [day.lower() for day in calendar.day_abbr].index(value)
    except ValueError:
        pass

    try:
        val = int(value)
        if val not in range(7):
            raise ValueError()
        return val
    except ValueError:
        pass

    days = [str(daynum) for daynum in range(7)]
    days += [day.lower() for day in calendar.day_abbr]
    days += [day.lower() for day in calendar.day_name]
    valid_days = ', '.join(days)
    raise ValueError(f"Invalid day {value!r}. Valid days are : {valid_days}")


def parse_args(argv=None):
    """ Argument parser for CLI
    
    Arguments :
    - argv : A list like sys.argv[1:]

    Print help and sys.exit() if invalid arguments given, else returns 
    arguments in a named tuple.
    """
    parser = argparse.ArgumentParser(description="Count the number of \
commits done in/out of office hours.")

    parser.add_argument("url", help="Git repository URL")
    parser.add_argument("-d", "--daystart", type=datetime.time.fromisoformat,
            default=datetime.time(8,0,0),
            help="Day start time (ISO 8601)")
    parser.add_argument("-D", "--daystop", type=datetime.time.fromisoformat,
            default=datetime.time(20,0,0),
            help="Day stop time (ISO 8601)")
    parser.add_argument("-w", "--weekend", type=str, default="5,6",
            metavar='DAYSOFF',
            help="Indicate days off separated by ','. Locale day names or \
days number (0..6 starting with monday) can be used. By default saturday \
and sunday are off. Use special value -w NUL to indicate no days off.")
    parser.add_argument("-f", "--from", type=datetime.date.fromisoformat,
            dest="from_date",
            help="Exclude commits before given ISO 8601 date")
    parser.add_argument("-t", "--to", type=datetime.date.fromisoformat,
            dest="to_date",
            help="Exclude commits after given ISO 8601 date")
    parser.add_argument("-g", "--group-by", choices=['week', 'month'],
            help="Generate commit stats by week or month")
    parser.add_argument("-s", "--silent", action="store_true", default=False,
            help="Do not display fetch progression on stderr")
    parser.add_argument("-N", "--no-details", action="store_false",
            default=True, dest="details",
            help="Do not display the total number of commit in result array")
    parser.add_argument("-o", "--csv-output", type=argparse.FileType("w"),
            default=None,
            help="Store commit counts in a CSV file instead of printing it \
on stdout (use '-' for stdout)")
    parser.add_argument("-v", "--verbose", action="store_true",
            default=False,
            help="Output all commit off office hours on stdout")

    args = parser.parse_args(argv)

    # parse weekend days list
    if args.weekend == 'NUL':
        args.weekend = []
    else:
        weekend = []
        for day in args.weekend.split(','):
            try:
                weekend.append(valid_day(day.strip()))
            except ValueError as expt:
                print(expt, file=sys.stderr)
                parser.print_help()
                sys.exit(1)
        args.weekend = list(set(weekend))

    return args


if __name__ == "__main__":
    main()