Demo/scripts/ftpstats.py

   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 stdandard 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 regex
  17 import string
  18 import getopt
  19
  20 pat = '^\([a-zA-Z0-9 :]*\)!\(.*\)!\(.*\)!\([<>].*\)!\([0-9]+\)!\([0-9]+\)$'
  21 prog = regex.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, ':', `line`
  64                         continue
  65                 items = prog.group(1, 2, 3, 4, 5, 6)
  66                 logtime, loguser, loghost, logfile, logbytes, logxxx2 = items
  67 ##              print logtime
  68 ##              print '-->', loguser
  69 ##              print '--> -->', loghost
  70 ##              print '--> --> -->', logfile
  71 ##              print '--> --> --> -->', logbytes
  72 ##              print '--> --> --> --> -->', logxxx2
  73 ##              for i in logtime, loghost, logbytes, logxxx2:
  74 ##                      if '!' in i: print '???', i
  75                 add(bydate, logtime[-4:] + ' ' + logtime[:6], items)
  76                 add(bytime, logtime[7:9] + ':00-59', items)
  77                 direction, logfile = logfile[0], logfile[1:]
  78                 # The real path probably starts at the last //...
  79                 while 1:
  80                         i = string.find(logfile, '//')
  81                         if i < 0: break
  82                         logfile = logfile[i+1:]
  83                 add(byfile, logfile + ' ' + direction, items)
  84                 logdir = os.path.dirname(logfile)
  85 ##              logdir = os.path.normpath(logdir) + '/.'
  86                 while 1:
  87                         add(bydir, logdir + ' ' + direction, items)
  88                         dirhead = os.path.dirname(logdir)
  89                         if dirhead == logdir: break
  90                         logdir = dirhead
  91                 add(byhost, loghost, items)
  92                 add(byuser, loguser, items)
  93                 add(bytype, direction, items)
  94         except KeyboardInterrupt:
  95                 print 'Interrupted at line', lineno
  96         show(bytype, 'by transfer direction', maxitems)
  97         show(bydir, 'by directory', maxitems)
  98         show(byfile, 'by file', maxitems)
  99         show(byhost, 'by host', maxitems)
 100         show(byuser, 'by user', maxitems)
 101         showbar(bydate, 'by date')
 102         showbar(bytime, 'by time of day')
 103
 104 def showbar(dict, title):
 105         n = len(title)
 106         print '='*((70-n)/2), title, '='*((71-n)/2)
 107         list = []
 108         keys = dict.keys()
 109         keys.sort()
 110         for key in keys:
 111                 n = len(str(key))
 112                 list.append((len(dict[key]), key))
 113         maxkeylength = 0
 114         maxcount = 0
 115         for count, key in list:
 116                 maxkeylength = max(maxkeylength, len(key))
 117                 maxcount = max(maxcount, count)
 118         maxbarlength = 72 - maxkeylength - 7
 119         for count, key in list:
 120                 barlength = int(round(maxbarlength*float(count)/maxcount))
 121                 bar = '*'*barlength
 122                 print '%5d %-*s %s' % (count, maxkeylength, key, bar)
 123
 124 def show(dict, title, maxitems):
 125         if len(dict) > maxitems:
 126                 title = title + ' (first %d)'%maxitems
 127         n = len(title)
 128         print '='*((70-n)/2), title, '='*((71-n)/2)
 129         list = []
 130         keys = dict.keys()
 131         for key in keys:
 132                 list.append((-len(dict[key]), key))
 133         list.sort()
 134         for count, key in list[:maxitems]:
 135                 print '%5d %s' % (-count, key)
 136
 137 def add(dict, key, item):
 138         if dict.has_key(key):
 139                 dict[key].append(item)
 140         else:
 141                 dict[key] = [item]
 142
 143 main()