bin/md5check.py

   1 #!/usr/bin/python -O
   2 # Copyright 1999-2006 Gentoo Foundation
   3 # Distributed under the terms of the GNU General Public License v2
   4 # $Id: /var/cvsroot/gentoo-src/portage/bin/md5check.py,v 1.4 2004/10/10 10:07:20 carpaski Exp $
   5
   6 import os,sys,string
   7 os.environ["PORTAGE_CALLER"]="mirror"
   8 os.environ["FEATURES"]="mirror cvs"
   9 sys.path = ["/usr/lib/portage/pym"]+sys.path
  10
  11 import portage
  12 from threading import *
  13 from output import red,green,blue,bold
  14 from random import shuffle
  15 from time import sleep
  16
  17
  18 def cstrip(mystr,mychars):
  19         newstr = ""
  20         for x in mystr:
  21                 if x not in mychars:
  22                         newstr += x
  23         return newstr
  24
  25 md5_list = {}
  26 bn_list  = []
  27 col_list = []
  28
  29 hugelist = []
  30 for mycp in portage.db["/"]["porttree"].dbapi.cp_all():
  31         hugelist += portage.db["/"]["porttree"].dbapi.cp_list(mycp)
  32 hugelist.sort()
  33
  34 for mycpv in hugelist:
  35         pv = string.split(mycpv, "/")[-1]
  36
  37         newuri = portage.db["/"]["porttree"].dbapi.aux_get(mycpv,["SRC_URI"])[0]
  38         newuri = string.split(newuri)
  39
  40         digestpath = portage.db["/"]["porttree"].dbapi.findname(mycpv)
  41         digestpath = os.path.dirname(digestpath)+"/files/digest-"+pv
  42         md5sums    = portage.digestParseFile(digestpath)
  43
  44         if md5sums == None:
  45                 portage.writemsg("Missing digest: %s\n" % mycpv)
  46                 md5sums = {}
  47
  48         for x in md5sums.keys():
  49                 if x[0] == '/':
  50                         del md5sums[x]
  51
  52         #portage.writemsg("\n\ndigestpath: %s\n" % digestpath)
  53         #portage.writemsg("md5sums: %s\n" % md5sums)
  54         #portage.writemsg("newuri: %s\n" % newuri)
  55
  56         bn_list = []
  57         for x in newuri:
  58                 if not x:
  59                         continue
  60                 if (x in [")","(",":","||"]) or (x[-1] == "?"):
  61                         # ignore it. :)
  62                         continue
  63                 x = cstrip(x,"()|?")
  64                 if not x:
  65                         continue
  66
  67                 mybn = os.path.basename(x)
  68                 if mybn not in bn_list:
  69                         bn_list += [mybn]
  70                 else:
  71                         continue
  72
  73                 if mybn not in md5sums.keys():
  74                         portage_util.writemsg("Missing md5sum: %s in %s\n" % (mybn, mycpv))
  75                 else:
  76                         if mybn in md5_list.keys():
  77                                 if (md5_list[mybn]["MD5"]  != md5sums[mybn]["MD5"]) or \
  78                                    (md5_list[mybn]["size"] != md5sums[mybn]["size"]):
  79
  80                                         # This associates teh md5 with each file. [md5/size]
  81                                         md5joins = string.split(md5_list[mybn][2],",")
  82                                         md5joins = string.join(md5joins," ["+md5_list[mybn][0]+"/"+md5_list[mybn][1]+"],")
  83                                         md5joins += " ["+md5_list[mybn][0]+"/"+md5_list[mybn][1]+"]"
  84
  85                                         portage.writemsg("Colliding md5: %s of %s [%s/%s] and %s\n" % (mybn,mycpv,md5sums[mybn][0],md5sums[mybn][1],md5joins))
  86                                         col_list += [mybn]
  87                                 else:
  88                                         md5_list[mybn][2] += ","+mycpv
  89                         else:
  90                                 md5_list[mybn] = md5sums[mybn]+[mycpv]
  91                         del md5sums[mybn]
  92
  93         #portage.writemsg(str(bn_list)+"\n")
  94         for x in md5sums.keys():
  95                 if x not in bn_list:
  96                         portage.writemsg("Extra md5sum: %s in %s\n" % (x, mycpv))
  97
  98
  99 print col_list
 100 print
 101 print str(len(md5_list.keys()))+" unique distfile md5s."
 102 print str(len(bn_list))+" unique distfile names."