find-duplicate-pkgs

#!/usr/bin/python
# -*- coding: utf-8 -*-
#
# Prints duplicate packages that exist in both binary repositories.
#
#
#
# http://paketler.pardus.org.tr/pardus-2009/pisi-index.xml.bz2
# http://paketler.pardus.org.tr/contrib-2009/pisi-index.xml.bz2

import os
import sys
import piksemel
import bz2
import urllib2

def loadUrlData(_url):
    try:
        downloaded_data = urllib2.urlopen(_url).read()
    except urllib2.URLError:
        print "could not download %s" % _url
        sys.exit(1)

    print "Remote file %s has been downloaded." % _url

    if _url.endswith(".bz2"):
        rawData = bz2.decompress(downloaded_data)
        return piksemel.parseString(rawData)
    else:
        return piksemel.parseString(downloaded_data)

def getXmlData(_file):
    if os.path.exists(_file):
        return piksemel.parse(_file)
    elif os.path.exists("%s.bz2" % _file):
        indexdata = bz2.decompress(file("%s.bz2" % _file).read())
        return piksemel.parseString(indexdata)
    else:
        print "please give the index file as a parameter or go to that folder."
        print "    Can it be a remote file?"
        sys.exit(1)

def parseXmlData(_index):
    pkglist = []

    for pkg in _index.tags("Package"):
        pkglist.append(pkg.getTagData("Name"))

    return pkglist

if __name__ == "__main__":
    if len(sys.argv) == 3:
        first_index_file = sys.argv[1]
        second_index_file = sys.argv[2]
    else:
        print "Usage: %s first_index_file second_index_file" % sys.argv[0]
        sys.exit(1)

    try:
        first_xmldata = getXmlData(first_index_file)
    except:
        first_xmldata = loadUrlData(first_index_file)

    first_packagesIndex= parseXmlData(first_xmldata)
    
    try:
        second_xmldata = getXmlData(second_index_file)
    except:
        second_xmldata = loadUrlData(second_index_file)

    second_packagesIndex= parseXmlData(second_xmldata)

    duplicate_pkgs = list(set(first_packagesIndex) & set(second_packagesIndex))

    print "-----------------------------------------------------------------------"
    print "Duplicate Packages between \n%s\n\t\t\tand \n%s are:" % (sys.argv[1], sys.argv[2])
    print "-----------------------------------------------------------------------"
    for i in duplicate_pkgs:
        print i