source: CMTManagement/utils/uptime.py @ 421

#!/usr/bin/env python
"""This displays uptime information using uptime.
This is redundant perhaps, but it demonstrates expecting for a
regular expression that uses subgroups.
"""

import pexpect
import re

# There are many different styles of uptime results.
# I try to parse them all. Yeee!
#
# [x86] Linux 2.4 (Redhat 7.3)
#  2:06pm  up 63 days, 18 min,  3 users,  load average: 0.32, 0.08, 0.02
# [x86] Linux 2.4.18-14 (Redhat 8.0)
#  3:07pm  up 29 min,  1 user,  load average: 2.44, 2.51, 1.57
# [PPC - G4] MacOS X 10.1 SERVER Edition
# 2:11PM  up 3 days, 13:50, 3 users, load averages: 0.01, 0.00, 0.00
# [Sparc - R220] Sun Solaris (8)
#  2:13pm  up 22 min(s),  1 user,  load average: 0.02, 0.01, 0.01
# [x86] Linux 2.4.18-14 (Redhat 8)
# 11:36pm  up 4 days, 17:58,  1 user,  load average: 0.03, 0.01, 0.00
# AIX jwdir 2 5 0001DBFA4C00
#  09:43AM   up  23:27,  1 user,  load average: 0.49, 0.32, 0.23

# This parses uptime output into the major groups using
# regex group matching.
p = pexpect.spawn ('uptime')
p.expect('up\s+(.*?),\s+([0-9]+) users?,\s+load averages?: ([0-9]+\.[0-9][0-9]), ([0-9]+\.[0-9][0-9]), ([0-9]+\.[0-9][0-9])')
match = p.match
duration, users, av1, av5, av15 = match.groups()

# The duration is a little harder to parse because of all the different
# styles of uptime. I'm sure there is a way to do this all at once with
# one single regex, but I bet it would be hard to read and maintain.
# If anyone wants to send me a version using a single regex I'd be
# happy to see it.
days = '0'
if 'day' in duration:
    match = re.search('([0-9]+)\s+day',duration)
    days = match.group(1)
hours = '0:0'
if ':' in duration:
    match = re.search('([0-9]+:[0-9]+)',duration)
    hours = match.group(1)
mins = '0'
if 'min' in duration:
    match = re.search('([0-9]+)\s+min',duration)
    mins = match.group(1)

# Print the parsed fields in CSV format.
print 'days, hours, minutes, users, cpu avg 1 min, cpu avg 5 min, cpu avg 15 min'
print '%s, %s, %s, %s, %s, %s, %s' % (days, hours, mins, users, av1, av5, av15)