wwwbin/clean-dups.py

   1 #!/usr/bin/python
   2 # arekm, 2008
   3 # remove
   4
   5 import os
   6 import re
   7 import time
   8 import rpm
   9 import sys
  10
  11 re_rpm = re.compile(r'.*\.rpm$')
  12 re_nvr = re.compile('^(.*)-([^-]*)-([^-]*)\.rpm$')
  13 dir = '/home/pld/admins/th/ftp/test/SRPMS/RPMS'
  14
  15 ignore = re.compile('^(kernel-.*|\
  16         dahdi-linux-.*|\
  17         e1000e-.*|\
  18         igb-.*|\
  19         ipset-.*|\
  20         ixgbe-.*|\
  21         linuxrdac-.*|\
  22         lirc-.*|\
  23         madwifi-ng-.*|\
  24         open-vm-tools-.*|\
  25         r8168-.*|\
  26         VirtualBox-.*|\
  27         xorg-driver-video-fglrx-.*|\
  28         xorg-driver-video-fglrx-legacy-12.x-.*|\
  29         xorg-driver-video-nvidia-.*|\
  30         xtables-addons-.*|\
  31         iscsitarget-.*|\
  32         openvswitch-.*|\
  33         xorg-driver-video-nvidia-legacy3-.*)$')
  34
  35 ts = rpm.TransactionSet("", (rpm._RPMVSF_NOSIGNATURES or rpm.RPMVSF_NOHDRCHK or rpm._RPMVSF_NODIGESTS or rpm.RPMVSF_NEEDPAYLOAD))
  36
  37 def compare(f1, f2):
  38         try:
  39                 fd1 = os.open(os.path.join(dir, f1), os.O_RDONLY)
  40         except Exception, e:
  41                 print e
  42                 # ignore non-files
  43                 return 0
  44         try:
  45                 fd2 = os.open(os.path.join(dir, f2), os.O_RDONLY)
  46         except Exception, e:
  47                 print e
  48                 # ignore non-files
  49                 return 0
  50         try:
  51                 h1 = ts.hdrFromFdno(fd1)
  52         except Exception, e:
  53                 print "hdrFromFdno for %s failed: %s" % (f1, e)
  54                 os.close(fd1)
  55                 os.close(fd2)
  56                 return 0
  57
  58         try:
  59                 h2 = ts.hdrFromFdno(fd2)
  60         except Exception, e:
  61                 print "hdrFromFdno for %s failed: %s" % (f2, e)
  62                 os.close(fd1)
  63                 os.close(fd2)
  64                 return 0
  65
  66         os.close(fd1)
  67         os.close(fd2)
  68
  69         l1 = rpm.versionCompare(h1, h2)
  70         l2 = rpm.versionCompare(h2, h1)
  71
  72         if l1 > 0 and l2 > 0:
  73                 return 0
  74
  75         return -l1
  76
  77
  78 def find_old(files):
  79         return sorted(files, compare)
  80
  81 files = {}
  82 dupes = {}
  83
  84 for file in os.listdir(dir):
  85         if not re_rpm.match(file):
  86                 continue
  87
  88         if ignore.match(file):
  89                 continue
  90
  91         m = re_nvr.match(file)
  92         if not m:
  93                 print "problem with: %s" % file
  94                 sys.exit(1)
  95
  96         if len(sys.argv) == 0:
  97                 p = os.path.join(dir, file)
  98                 mtime = os.stat(p).st_mtime
  99                 if mtime > time.time() - 3*86400:
 100                         continue
 101
 102         name = m.group(1)
 103
 104         if files.has_key(name):
 105                 if dupes.has_key(name):
 106                         dupes[name].append(file)
 107                 else:
 108                         dupes[name] = [ files[name] ]
 109                         dupes[name].append(file)
 110         else:
 111                 files[name] = file
 112
 113 for i in dupes.iterkeys():
 114         for old in find_old(dupes[i])[1:]:
 115                 os.system("/home/pld/admins/th/pld-ftp-admin/scripts/remove.py test %s" % old)