Calculates nntp stats

nntp-stats.py

#!/usr/bin/env python

import nntplib
import re
import sys

usage = """./nntp-stats [nntp-group-name]"""

if len(sys.argv) < 2:
    print usage
    sys.exit(1)

newsgroup = sys.argv[1]

s = nntplib.NNTP('news.mozilla.org')
(resp, count, first, last, name) = s.group(newsgroup)

resp, items = s.xover(first, last)

count = 0
posting_from_mozilla = 0
posting_from_googlegroups = 0
posting_from_other = 0

article_date = []

for id, subject, author, date, message_id, references, size, lines in items:

    # Count number of total messages
    count += 1

    # This is a terrible way of getting this data
    article_date.append(date)

    # if message id has the word googlegroups.com
    # we assume its from groups.google.com
    # otherwise if it has the word lists.mozilla.org and mailman then its from mailman
    # everything else we assume its from nntp
    if re.findall(r'googlegroups.com', message_id):
        posting_from_googlegroups += 1
    elif re.findall(r'lists.mozilla.org', message_id) and re.findall(r'mailman', message_id):
        posting_from_mozilla +=1
    else:
        posting_from_other += 1

percent_mozilla = (float(posting_from_mozilla)*100/float(count))
percent_googlegroups = (float(posting_from_googlegroups)*100/float(count))
percent_other = (float(posting_from_other)*100/float(count))

print "Total number of articles: %s" % count
print "Message from mailman: %s (%d%%)" % (posting_from_mozilla, int(percent_mozilla))
print "Message from google groups %s (%d%%)" % (posting_from_googlegroups, int(percent_googlegroups))
print "Message from other: %s (%d%%)" % (posting_from_other, int(percent_other))
print "First message date: %s" % (article_date[0])
print "Last message date: %s" % (article_date[-1])

s.quit()