#!/usr/bin/env python2
import re
import sys

# COMMAND     PID   TID     USER   FD      TYPE DEVICE SIZE/OFF     NODE NAME
# systemd       1              0  cwd   unknown                          / (stat: No such file or directory)
# systemd       1              0  rtd   unknown                          / (stat: No such file or directory)
# systemd       1              0  txt   unknown                          /usr/lib/systemd/systemd (stat: Resource temporarily unavailable)
# systemd       1              0  mem       REG  254,1          19534244 /usr/lib/systemd/systemd (stat: Resource temporarily unavailable)
# systemd       1              0  mem       REG  254,1          19530388 /usr/lib64/libuuid.so.1.3.0 (stat: Resource temporarily unavailable)
# systemd-j  1291              0  mem       REG  254,1           1572958  /lib64/libselinux.so.1 (stat: Resource temporarily unavailable)
# ms_pipe_w  1465   330      201   98   unknown                          socket lstat: No such file or directory) (stat: No such file or directory)

expr = re.compile("(?P<command>.{9}) (?P<pid>.{5}) (?P<tid>.{5}) (?P<user>.{8}) (?P<fd>.{4}) (?P<type>.{9}) (?P<device>.{6}) (?P<rest>.*)")

entries = []
by_key = {}
by_user = {}

for line in open(sys.argv[1]):
    line = line.strip()
    if (not line or
            line.startswith("#") or
            "avoiding stat" in line or
            line.startswith("lsof:") or
            "Output information" in line):
        continue
    found = expr.search(line)
    if not found:
        continue
    entry = found.groupdict()
    for key in ("pid", "tid", "user"):
        entry[key] = entry[key].strip()
    # entries.append(entry)
    key = (entry["user"], entry["rest"])
    by_key[key] = by_key.get(key, 0) + 1

by_sum = {}

for (user, rest), count in by_key.iteritems():
    by_sum[user] = by_sum.get(user, 0) + count
    by_user.setdefault(user, []).append(rest)

import pdb; pdb.set_trace()