wwwbin/clean-dups-archive.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/.archive/PLD/SRPMS/RPMS'
  14
  15 ts = rpm.TransactionSet("", (rpm.RPMVSF_NOHDRCHK or rpm.RPMVSF_NEEDPAYLOAD))
  16
  17 def compare(f1, f2):
  18         try:
  19                 fd1 = os.open(os.path.join(dir, f1), os.O_RDONLY)
  20         except Exception, e:
  21                 print e
  22                 # ignore non-files
  23                 return 0
  24         try:
  25                 fd2 = os.open(os.path.join(dir, f2), os.O_RDONLY)
  26         except Exception, e:
  27                 print e
  28                 # ignore non-files
  29                 return 0
  30         h1 = ts.hdrFromFdno(fd1)
  31         h2 = ts.hdrFromFdno(fd2)
  32         os.close(fd1)
  33         os.close(fd2)
  34
  35         l1 = rpm.versionCompare(h1, h2)
  36         l2 = rpm.versionCompare(h2, h1)
  37
  38         if l1 > 0 and l2 > 0:
  39                 return 0
  40
  41         return -l1
  42
  43
  44 def find_old(files):
  45         return sorted(files, compare)
  46
  47 files = {}
  48 dupes = {}
  49
  50 for file in os.listdir(dir):
  51         if not re_rpm.match(file):
  52                 continue
  53
  54         m = re_nvr.match(file)
  55         if not m:
  56                 print "problem with: %s" % file
  57                 sys.exit(1)
  58
  59         if len(sys.argv) == 0:
  60                 p = os.path.join(dir, file)
  61                 mtime = os.stat(p).st_mtime
  62                 if mtime > time.time() - 3*86400:
  63                         continue
  64
  65         name = m.group(1)
  66
  67         if files.has_key(name):
  68                 if dupes.has_key(name):
  69                         dupes[name].append(file)
  70                 else:
  71                         dupes[name] = [ files[name] ]
  72                         dupes[name].append(file)
  73         else:
  74                 files[name] = file
  75
  76 for i in dupes.iterkeys():
  77         for old in find_old(dupes[i])[1:]:
  78                 os.system("/home/pld/admins/th/pld-ftp-admin/scripts/remove.py .archive/PLD %s" % old)