#!/usr/bin/python

import os, os.path, gzip, sys

mirrordir = '/mirror'

def get_package_versions_file(f, map):
    currpkg = None
    currversion = None

    for line in gzip.open(f):
        line = line.strip()
        if not line:
            if (not currpkg) or (not currversion):
                print >> sys.stderr, "Error: end of package record without all data available"
                sys.exit(1)
            map.setdefault(currpkg, []).append(currversion)
            currpkg = None
            currversion = None
            continue

        attr = line.split(":", 1)

        if len(attr) < 2:
            continue

        if attr[0] == "Package":
            if currpkg:
                print >> sys.stderr, "Error: read two Packages: lines in a row"
                sys.exit(1)
            currpkg = attr[1].strip()

        if attr[0] == "Version":
            if currversion:
                print >> sys.stderr, "Error: read two Version: lines in a row"
                sys.exit(1)
            currversion = attr[1].strip()
            # remove epochs
            colpos = currversion.find(':')
            if colpos >= 0:
                currversion = currversion[colpos+1:]

def get_package_versions_tree(rootdir):
    """Generate a archive map from a Debian-style package archive directory.

    rootdir: path to archive root path
    return: mapping: sourcepackage -> [versions]
    """

    map = {}

    for release in os.listdir(rootdir + "/dists"):
        for comp in os.listdir(rootdir + "/dists/" + release):
            compdir = rootdir + "/dists/" + release + "/" + comp
            if not os.path.isdir(compdir):
                continue
            get_package_versions_file(compdir + "/source/Sources.gz", map)
            for arch in os.listdir(compdir):
                if arch.startswith('binary-'):
                    get_package_versions_file(compdir + '/' + arch + '/Packages.gz', map)

    return map

known_versions = get_package_versions_tree(mirrordir)

for path, dirs, files in os.walk(os.path.join(mirrordir, 'pool')):
    for f in files:
        (pkg, version) = (f.split('_'))[0:2]
        if f.endswith('.deb'):
            pass
        elif f.endswith('.dsc'):
            version = version[:-4]
        elif f.endswith('.diff.gz'):
            version = version[:-8]
        elif f.endswith('.tar.gz'):
            continue
        else:
            print os.path.join(path, f)
            continue

        if not version in known_versions.get(pkg, ()):
            print os.path.join(path, f)