1 #! /usr/bin/env python

     2 

     3 # Extract statistics from ftp daemon log.

     4 

     5 # Usage:

     6 # ftpstats [-m maxitems] [-s search] [file]

     7 # -m maxitems: restrict number of items in "top-N" lists, default 25.

     8 # -s string:   restrict statistics to lines containing this string.

     9 # Default file is /usr/adm/ftpd;  a "-" means read standard input.

    10 

    11 # The script must be run on the host where the ftp daemon runs.

    12 # (At CWI this is currently buizerd.)

    13 

    14 import os

    15 import sys

    16 import re

    17 import string

    18 import getopt

    19 

    20 pat = '^([a-zA-Z0-9 :]*)!(.*)!(.*)!([<>].*)!([0-9]+)!([0-9]+)$'

    21 prog = re.compile(pat)

    22 

    23 def main():

    24     maxitems = 25

    25     search = None

    26     try:

    27         opts, args = getopt.getopt(sys.argv[1:], 'm:s:')

    28     except getopt.error, msg:

    29         print msg

    30         print 'usage: ftpstats [-m maxitems] [file]'

    31         sys.exit(2)

    32     for o, a in opts:

    33         if o == '-m':

    34             maxitems = string.atoi(a)

    35         if o == '-s':

    36             search = a

    37     file = '/usr/adm/ftpd'

    38     if args: file = args[0]

    39     if file == '-':

    40         f = sys.stdin

    41     else:

    42         try:

    43             f = open(file, 'r')

    44         except IOError, msg:

    45             print file, ':', msg

    46             sys.exit(1)

    47     bydate = {}

    48     bytime = {}

    49     byfile = {}

    50     bydir = {}

    51     byhost = {}

    52     byuser = {}

    53     bytype = {}

    54     lineno = 0

    55     try:

    56         while 1:

    57             line = f.readline()

    58             if not line: break

    59             lineno = lineno + 1

    60             if search and string.find(line, search) < 0:

    61                 continue

    62             if prog.match(line) < 0:

    63                 print 'Bad line', lineno, ':', repr(line)

    64                 continue

    65             items = prog.group(1, 2, 3, 4, 5, 6)

    66             (logtime, loguser, loghost, logfile, logbytes,

    67              logxxx2) = items

    68 ##                      print logtime

    69 ##                      print '-->', loguser

    70 ##                      print '--> -->', loghost

    71 ##                      print '--> --> -->', logfile

    72 ##                      print '--> --> --> -->', logbytes

    73 ##                      print '--> --> --> --> -->', logxxx2

    74 ##                      for i in logtime, loghost, logbytes, logxxx2:

    75 ##                              if '!' in i: print '???', i

    76             add(bydate, logtime[-4:] + ' ' + logtime[:6], items)

    77             add(bytime, logtime[7:9] + ':00-59', items)

    78             direction, logfile = logfile[0], logfile[1:]

    79             # The real path probably starts at the last //...

    80             while 1:

    81                 i = string.find(logfile, '//')

    82                 if i < 0: break

    83                 logfile = logfile[i+1:]

    84             add(byfile, logfile + ' ' + direction, items)

    85             logdir = os.path.dirname(logfile)

    86 ##              logdir = os.path.normpath(logdir) + '/.'

    87             while 1:

    88                 add(bydir, logdir + ' ' + direction, items)

    89                 dirhead = os.path.dirname(logdir)

    90                 if dirhead == logdir: break

    91                 logdir = dirhead

    92             add(byhost, loghost, items)

    93             add(byuser, loguser, items)

    94             add(bytype, direction, items)

    95     except KeyboardInterrupt:

    96         print 'Interrupted at line', lineno

    97     show(bytype, 'by transfer direction', maxitems)

    98     show(bydir, 'by directory', maxitems)

    99     show(byfile, 'by file', maxitems)

   100     show(byhost, 'by host', maxitems)

   101     show(byuser, 'by user', maxitems)

   102     showbar(bydate, 'by date')

   103     showbar(bytime, 'by time of day')

   104 

   105 def showbar(dict, title):

   106     n = len(title)

   107     print '='*((70-n)//2), title, '='*((71-n)//2)

   108     list = []

   109     keys = dict.keys()

   110     keys.sort()

   111     for key in keys:

   112         n = len(str(key))

   113         list.append((len(dict[key]), key))

   114     maxkeylength = 0

   115     maxcount = 0

   116     for count, key in list:

   117         maxkeylength = max(maxkeylength, len(key))

   118         maxcount = max(maxcount, count)

   119     maxbarlength = 72 - maxkeylength - 7

   120     for count, key in list:

   121         barlength = int(round(maxbarlength*float(count)/maxcount))

   122         bar = '*'*barlength

   123         print '%5d %-*s %s' % (count, maxkeylength, key, bar)

   124 

   125 def show(dict, title, maxitems):

   126     if len(dict) > maxitems:

   127         title = title + ' (first %d)'%maxitems

   128     n = len(title)

   129     print '='*((70-n)//2), title, '='*((71-n)//2)

   130     list = []

   131     keys = dict.keys()

   132     for key in keys:

   133         list.append((-len(dict[key]), key))

   134     list.sort()

   135     for count, key in list[:maxitems]:

   136         print '%5d %s' % (-count, key)

   137 

   138 def add(dict, key, item):

   139     if dict.has_key(key):

   140         dict[key].append(item)

   141     else:

   142         dict[key] = [item]

   143 

   144 if __name__ == "__main__":

   145     main()