wwwbin/clean-dups-archive.py

   1 #!/usr/bin/python3
   2 # arekm, 2008
   3 # remove
   4
   5 from __future__ import print_function
   6
   7 import os
   8 import re
   9 import time
  10 import rpm
  11 import sys
  12
  13 re_rpm = re.compile(r'.*\.rpm$')
  14 re_nvr = re.compile('^(.*)-([^-]*)-([^-]*)\.rpm$')
  15 dir = '/home/pld/admins/th/ftp/.archive/PLD/SRPMS/RPMS'
  16
  17 ts = rpm.TransactionSet("", (rpm.RPMVSF_NOHDRCHK or rpm.RPMVSF_NEEDPAYLOAD))
  18
  19 def compare(f1, f2):
  20         try:
  21                 fd1 = os.open(os.path.join(dir, f1), os.O_RDONLY)
  22         except Exception as e:
  23                 print(e)
  24                 # ignore non-files
  25                 return 0
  26         try:
  27                 fd2 = os.open(os.path.join(dir, f2), os.O_RDONLY)
  28         except Exception as e:
  29                 print(e)
  30                 # ignore non-files
  31                 return 0
  32         try:
  33                 h1 = ts.hdrFromFdno(fd1)
  34         except Exception as e:
  35                 print("hdrFromFdno for %s failed: %s" % (f1, e))
  36                 os.close(fd1)
  37                 os.close(fd2)
  38                 return 0
  39         try:
  40                 h2 = ts.hdrFromFdno(fd2)
  41         except Exception as e:
  42                 print("hdrFromFdno for %s failed: %s" % (f1, e))
  43                 os.close(fd1)
  44                 os.close(fd2)
  45                 return 0
  46         os.close(fd1)
  47         os.close(fd2)
  48
  49         try:
  50                 l1 = rpm.versionCompare(h1, h2)
  51         except ValueError:
  52                 l1 = -1
  53         try:
  54                 l2 = rpm.versionCompare(h2, h1)
  55         except ValueError:
  56                 l2 = -1
  57
  58         if l1 > 0 and l2 > 0:
  59                 return 0
  60
  61         return -l1
  62
  63
  64 def find_old(files):
  65         return sorted(files, compare)
  66
  67 files = {}
  68 dupes = {}
  69
  70 for file in os.listdir(dir):
  71         if not re_rpm.match(file):
  72                 continue
  73
  74         m = re_nvr.match(file)
  75         if not m:
  76                 print("problem with: %s" % file)
  77                 sys.exit(1)
  78
  79         if len(sys.argv) == 0:
  80                 p = os.path.join(dir, file)
  81                 mtime = os.stat(p).st_mtime
  82                 if mtime > time.time() - 3*86400:
  83                         continue
  84
  85         name = m.group(1)
  86
  87         if files.has_key(name):
  88                 if dupes.has_key(name):
  89                         dupes[name].append(file)
  90                 else:
  91                         dupes[name] = [ files[name] ]
  92                         dupes[name].append(file)
  93         else:
  94                 files[name] = file
  95
  96 for i in iter(dupes.keys()):
  97         for old in find_old(dupes[i])[1:]:
  98                 os.system("/home/pld/admins/th/pld-ftp-admin/scripts/remove.py .archive/PLD %s" % old)