requested by management that judges productivity by PRs

git-metrics.py

#!/usr/bin/env python

import subprocess
import collections
import re

from prettytable import PrettyTable
from datetime import timedelta, date
from dateutil.parser import parse
from subprocess import PIPE, STDOUT
from itertools import izip

# def daterange(start_date, end_date):
#     for n in range(int((end_date - start_date).days)):
#         yield start_date + timedelta(n)

def format_date(date):
    return date.strftime("%Y-%m-%d")

def chomp(str, suffix):
    index = len(suffix)
    return str[:-index] if str.endswith(suffix) else str

def as_dict(array):
    array = [reversed(x) for x in array]
    metrics = {}

    for k, v in array:
        key = chomp(k, "s")
        metrics.setdefault(key, int(v))
    return metrics

def pairwise(iterable):
    a = iter(iterable)
    return izip(a, a)

def update(collection, author, metrics):
    current_stat = collection.get(author)
    metrics = as_dict(metrics)

    if current_stat:
        for key, value in metrics.iteritems():
            current_stat[key] = current_stat.setdefault(key, 0) + value
    else:
        current_stat = metrics
    collection[author] = current_stat

# constants
GIT_STAT_REGEX = re.compile(r"(\d+) (\w+)")
AUTHOR_REGEX = re.compile(r"(\w+\.\w+)(?:@your-company.com)")
START_DATE = parse("2016-03-02")
# END_DATE = parse("2016-03-01")
END_DATE = date.today()

#  git log --after=2016-03-01 --shortstat --pretty="%cE" --no-merges
GIT_COMMAND = [
    'git',
    'log',
    '--shortstat',
    '--before=%s' % format_date(END_DATE),
    '--after=%s' % format_date(START_DATE),
    '--pretty="%cE"',
    '--no-merges']

# script
proc = subprocess.Popen(GIT_COMMAND, shell=True, stdout=PIPE, stderr=STDOUT)
authors = {}
lines = filter(lambda x: len(x), (line.strip() for line in proc.stdout))

for author, stat in pairwise(lines):
    metrics = GIT_STAT_REGEX.findall(stat)
    usernames = AUTHOR_REGEX.findall(author)
    for user in usernames:
        update(authors, user, metrics)

table = PrettyTable(["author", "files", "add", "deleted", "net"])
authors = collections.OrderedDict(sorted(authors.items()))

print START_DATE.strftime("%Y-%m-%d"), END_DATE.strftime("%Y-%m-%d")
for author, stat in authors.iteritems():
    files = stat["file"]
    added = stat["insertion"]
    deleted = stat["deletion"]
    net = added - deleted
    table.add_row([author, files, added, deleted, net])
print table