firefrag.py

#!/usr/bin/env python

import collections
import os
import re
from subprocess import Popen, PIPE


LSOFItem = collections.namedtuple('LSOFItem', 'command pid user fd type device size node name'.split())
extent_regex = re.compile(r'(\d+) extents? found')

data = [] # (extents, name)

print 'Running lsof...'
lsof = Popen(['lsof'], stdout=PIPE)
lsof_data, _ = lsof.communicate()
print 'Done running lsof...'
for line in lsof_data.splitlines():
    if not line.startswith('firefox'):
        continue
    try:
        item = LSOFItem(*line.strip().split())
    except TypeError:
        print 'Bad line', line.strip()
        continue
    if item.type != 'REG':
        continue
    if not os.path.isfile(item.name):
        continue
    filefrag = Popen(['filefrag', item.name], stdout=PIPE)
    stdout, _ = filefrag.communicate()
    match = extent_regex.search(stdout)
    if not match:
        print 'Bad filefrag result:', stdout.strip()
        continue
    data.append((int(match.group(1)), item.name))

data.sort(reverse=True)
for extents, name in data:
    print '%120s: %d' % (name, extents)