find-unused-packages

#!/usr/bin/python
# -*- coding: utf-8 -*-
#
# Pisi repository unused files checker
#
# Gets lists of pisi files from filesystem and index and prints the diff
#

import os
import sys
import piksemel
import bz2


indexfile = "pisi-index.xml"

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 parameter or go to that folder"
        sys.exit(1)

def parseXmlData(_index):
    pkglist = []

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

        parent = pkg.getTag("DeltaPackages")
        if parent:
            for node in parent.tags("Delta"):
                pkglist.append(node.getTagData("PackageURI"))

    return pkglist

def getFolderPackages(_path):
    pkglist = []

    for i in os.listdir(_path):
        if i.endswith(".pisi"):
            pkglist.append(i)

    return pkglist

if __name__ == "__main__":
    if len(sys.argv) > 1:
        indexfile = sys.argv[1]

    xmldata = getXmlData(indexfile)
    packagesIndex = parseXmlData(xmldata)

    indexpath = os.path.dirname(indexfile)
    packagesFolder = getFolderPackages(indexpath)

    unusedPackages = list(set(packagesFolder) - set(packagesIndex))
    unusedPackages.sort()

    for i in unusedPackages:
        print i