check-repo-index-consistency

#!/usr/bin/python
# -*- coding: utf-8 -*-
#
# tool to find dependency inconsistencies in repositories
# can work on both source and binary index
# can work with local or remote files
# supports multiple inputs
#

import os
import sys

import urllib2
import bz2
import lzma

import pisi
import piksemel


def getIndex(uri):
    try:
        if "://" in uri:
            rawdata = urllib2.urlopen(uri).read()
        else:
            rawdata = open(uri, "r").read()
    except IOError:
        print "could not fetch %s" % uri
        return None

    if uri.endswith("bz2"):
        data = bz2.decompress(rawdata)
    elif uri.endswith("xz") or uri.endswith("lzma"):
        data = lzma.decompress(rawdata)
    else:
        data = rawdata

    return data.replace("<PISI>", "").replace("</PISI>", "")

def convertToRepoDB(ix):
    doc = piksemel.parseString(ix)
    dbobj = pisi.index.Index()
    dbobj.decode(doc, [])

    return dbobj

def printMissing(pkgname, pkgprop, latest, isruntime):
    depstr = "runtime dep" if isruntime else "build dep"

    if latest != pkgname:
        print "\n    %s" % pkgname
    print "        %s %s" % (depstr, pkgprop)

def checkRepoDB(repodb):
    if len(repodb.specs):
        print "found source index"

        repodict = {}
        for spec in repodb.specs:
            for package in spec.packages:
                package.version = spec.history[0].version
                package.release = spec.history[0].release
                repodict[package.name] = package

        latest = ""
        for spec in repodb.specs:
            for d in spec.source.buildDependencies:
                if not d.satisfied_by_dict_repo(repodict):
                    printMissing(spec.source.name, str(d), latest, False)
                    latest = spec.source.name

            for package in spec.packages:
                for d in package.runtimeDependencies():
                    if not d.satisfied_by_dict_repo(repodict):
                        printMissing(package.name, str(d), latest, True)
                        latest = package.name

    else:
        print "found binary index"

        repodict = dict((pkg.name, pkg) for pkg in repodb.packages)
        latest = ""
        for package in repodb.packages:
            for d in package.runtimeDependencies():
                if not d.satisfied_by_dict_repo(repodict):
                    printMissing(package.name, str(d), latest, True)
                    latest = package.name


if __name__ == "__main__":
    print

    if len(sys.argv) > 1:
        targets = sys.argv[1:]
    else:
        targets = ["pisi-index.xml"]

    print "* working on"
    for i in targets:
        print "    %s" % i

    # FIXME: find a way to update iksemel object with another one
    srcstring = "<PISI>"
    for indexfile in targets:
        srcstring += getIndex(indexfile)

    srcstring += "</PISI>"
    if srcstring != "":
        repo = convertToRepoDB(srcstring)
        checkRepoDB(repo)
    print