#!/usr/bin/python
import operator
import webkit
counts = {}
for date, author in webkit.parse_log():
author = webkit.canonicalize_email(author)
counts[author] = counts.get(author, 0) + 1
companies = {}
unknown = {}
for email, count in counts.iteritems():
company = webkit.classify_email(email)
companies[company] = companies.get(company, 0) + count
if company == "unknown":
unknown[email] = count
elif company == "misc":
unknown["*" + email] = count
if unknown:
print ('unclassified (star denotes "commits examined, and their ' 'backer is a minor contributor"):')
for email, count in sorted(unknown.iteritems(), key=operator.itemgetter(1), reverse=True):
print " %s (%d)" % (email, count)
for company, count in sorted(companies.iteritems(), key=operator.itemgetter(1), reverse=True):
print "%s: %d" % (company, count)
#!/usr/bin/python
import webkit
import datetime
from collections import defaultdict
start = None
end = None
data = defaultdict(lambda: defaultdict(lambda: 0))
for date, author in webkit.parse_log(since='3 years ago'):
author = webkit.canonicalize_email(author)
company = webkit.classify_email(author)
date = datetime.date(*map(int, date.split('-')))
if start is None or date < start:
start = date
if end is None or date > end:
end = date
data[date][company] += 1
show_companies = set(['google', 'apple', 'nokia', 'rim'])
print "Date," + ','.join(show_companies) + ",other"
date = start
while date <= end:
row = [date.strftime("%Y%m%d")]
for company in show_companies:
row.append(data[date][company])
misc = 0
for company in data[date].keys():
if company not in show_companies:
misc += data[date][company]
row.append(misc)
