dpkg-origins

#!/usr/bin/env python
"""
(c) 2008 Cory Dodt

Released under the terms of the MIT license.

dpkg-origins v0.9

Print out every system package, grouped by origin, for system recovery.
Requires 'python-apt' and 'python-twisted' packages.

    dpkg-origins > selections.txt

To recover the system packages, take the output (which you have previously
saved off or emailed to yourself) and do:

    cat selections.txt | sudo dpkg --set-selections && sudo apt-get -u dselect-upgrade
"""
import warnings
warnings.simplefilter('ignore', FutureWarning)

import sys, os
import subprocess
import itertools

from twisted.python import usage

import apt
import apt_pkg
apt_pkg.init()



class Options(usage.Options):
    synopsis = "dpkg-origins | mail ..."

    def postOptions(self):
        origins = sorted(self.packageOrigins())

        sourceTypes = [((lambda s: s==''), 'blank'),
                ((lambda s: 'ubuntu' in s), 'ubuntu'),
                ((lambda s: 'ubuntu' not in s), 'thirdParty')
                ]
        def getType(site):
            for fn, key in sourceTypes:
                if fn(site): return key
            assert 0, "This site doesn't match any known type"

        for key, grouper in itertools.groupby(origins, lambda o: o[0]):
            type = getType(key)
            printer = getattr(self, 'printGroup_%s' % type)
            printer(grouper)
            print ''

    def printGroup_blank(self, grouper):
        print '#' * 78
        print '## No apt archive - packages were installed manually from .deb'
        print '## or other dodgy source (including preinstalled by ISP).'
        print '## Note - this list may contain packages from third-party archives if the'
        print '## PPA is configured incorrectly.'
        print ''
        for package in grouper:
            print "# %s == %s" % (package[1], package[2])

    def printGroup_thirdParty(self, grouper):
        print '#' * 78
        print '''## Third-party apt archive packages
## Add the following archives to your sources list, then uncomment all the
## following packages and pipe into dpkg --set-selections'''
        print ''

        sources = apt_pkg.GetPkgSourceList()
        sources.ReadMainList()
        for meta in sources.List:
            if 'ubuntu.com' not in meta.URI:
                print '## deb %s %s main' % (meta.URI, meta.Dist)
        print ''

        for package in grouper:
            print "# %s install" % (package[1],)

    def printGroup_ubuntu(self, grouper):
        print '#' * 78
        print '## Canonical packages'
        print '## Pipe the following into dpkg --set-selections'''
        print ''
        for package in grouper:
            print '%s install' % (package[1],)
        
    def packageOrigins(self):
        dpkg = subprocess.Popen(r"dpkg-query -W -f'${Status}\t${Package}\t${Version}\n'",
                stdout=subprocess.PIPE, shell=True)
        cache = apt.Cache()

        for n, line in enumerate(dpkg.stdout.readlines()):
            status, package, version = line.strip().split('\t')
            if not status.startswith('install ok'):
                continue
                
            try:
                cpkg = cache[package]
                origin = cpkg.candidateOrigin[0] # FIXME - is it always the first??
                yield (origin.site, package, version)
            except KeyError:
                yield ("manual", package, version )


def run(argv=None):
    if argv is None:
        argv = sys.argv
    o = Options()
    try:
        o.parseOptions(argv[1:])
    except usage.UsageError, e:
        print str(o)
        print str(e)
        return 1

    return 0


if __name__ == '__main__': sys.exit(run())