check-integrity-archives

#!/usr/bin/python
# -*- coding: utf-8 -*-
#
# Pisi archives integrity checker
#
# Parses source repo pisi-index.xml or pisi-index.xml.bz2 file and checks sha1sum of archives
#
# can be given package names with components as command line parameters
#

import os
import sys
import piksemel
import bz2
import hashlib

indexfile = "pisi-index.xml"
archiveDir = "/var/cache/pisi/archives"


def printColorize(p, c, s):
        print "\x1b[1;33m%s \x1b[%sm %s\x1b[0m" % (p, c, s)

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 "\x1b[1;31mcould not find pisi index file\x1b[0m"
        sys.exit(1)

def getPkgValues(_node):
    pkgUri = _node.getTagData("SourceURI")
    archiveUri = _node.getTagData("Archive")
    archiveSum = _node.getTag("Archive").getAttribute("sha1sum")

    return pkgUri.replace("/pspec.xml", ""), archiveUri, archiveSum

def parseXmlData(_index):
    tmpDict = {}

    for pkg in _index.tags("SpecFile"):
        tmpUri, tmpSrcUri, tmpHash = getPkgValues(pkg.getTag("Source"))
        tmpDict[tmpUri] = [tmpSrcUri, tmpHash]

    return tmpDict

def checkHash(pkguri, pkginfo):
    foundhash = ""

    pkg = os.path.basename(pkginfo[0])
    hash = pkginfo[1]

    try:
        h = hashlib.sha1()
        f = open(os.path.join(archiveDir, pkg))
        h.update(f.read())
        foundhash = h.hexdigest()
        f.close()

        if hash == foundhash:
            printColorize(pkg, "1;32", "OK")
        else:
            printColorize(pkg, "0;31", "expected %s got %s" % (hash, foundhash))
            errorlist.append(pkguri)

    except IOError:
        if debug:
            printColorize(pkg, "0;31", "not found !")
        missinglist.append(pkguri)


if __name__ == "__main__":
    missinglist = []
    errorlist = []
    searchPkgs = []
    debug = False

    args = sys.argv

    if len(args) > 1:
        if args[1] == "-d":
            args.remove("-d")
            debug = True

    if len(args) > 1:
        searchPkgs.extend(args[1:])

    xmldata = getXmlData(indexfile)
    pkgDict = parseXmlData(xmldata)

    if not len(searchPkgs):
        searchPkgs = pkgDict.keys()

    searchPkgs.sort()
    for package in searchPkgs:
        if package in pkgDict:
            checkHash(package, pkgDict[package])
        else:
            printColorize(package, "0;31", "is not in index")
            missinglist.append(package)

    if len(errorlist):
        print
        print "\x1b[1;30mFound errors in ...\x1b[0m"
        print
        for e in errorlist:
            print e

    if debug and len(missinglist):
        print
        print "\x1b[1;30mCould not find ...\x1b[0m"
        print
        for e in missinglist:
            print e