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 ts = rpm.TransactionSet("", (rpm._RPMVSF_NOSIGNATURES or rpm.RPMVSF_NOHDRCHK or rpm._RPMVSF_NODIGESTS 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         try:
  31                 h1 = ts.hdrFromFdno(fd1)
  32         except Exception, e:
  33                 print "hdrFromFdno for %s failed: %s" % (f1, e)
  34                 os.close(fd1)
  35                 os.close(fd2)
  36                 return 0
  37
  38         try:
  39                 h2 = ts.hdrFromFdno(fd2)
  40         except Exception, e:
  41                 print "hdrFromFdno for %s failed: %s" % (f2, e)
  42                 os.close(fd1)
  43                 os.close(fd2)
  44                 return 0
  45
  46         os.close(fd1)
  47         os.close(fd2)
  48
  49         l1 = rpm.versionCompare(h1, h2)
  50         l2 = rpm.versionCompare(h2, h1)
  51
  52         if l1 > 0 and l2 > 0:
  53                 return 0
  54
  55         return -l1
  56
  57
  58 def find_old(files):
  59         return sorted(files, compare)
  60
  61 files = {}
  62 dupes = {}
  63
  64 for file in os.listdir(dir):
  65         if not re_rpm.match(file):
  66                 continue
  67
  68         m = re_nvr.match(file)
  69         if not m:
  70                 print "problem with: %s" % file
  71                 sys.exit(1)
  72
  73         if len(sys.argv) == 0:
  74                 p = os.path.join(dir, file)
  75                 mtime = os.stat(p).st_mtime
  76                 if mtime > time.time() - 3*86400:
  77                         continue
  78
  79         name = m.group(1)
  80
  81         if files.has_key(name):
  82                 if dupes.has_key(name):
  83                         dupes[name].append(file)
  84                 else:
  85                         dupes[name] = [ files[name] ]
  86                         dupes[name].append(file)
  87         else:
  88                 files[name] = file
  89
  90 for i in dupes.iterkeys():
  91         for old in find_old(dupes[i])[1:]:
  92                 os.system("/home/pld/admins/th/pld-ftp-admin/scripts/remove.py test %s" % old)