apt-mirror-verify-cache

#!/usr/bin/python

import hashlib
import os
import subprocess
import sys

from debian_bundle import deb822

DIR = os.path.abspath(os.path.dirname(__file__))

DISTS_DIR = os.path.join(DIR, "dists")
CACHE_DIR = os.path.join(DIR, "cache")

def cat_blob(obj):
    try:
        subprocess.check_call(['git', 'cat-file', '-e', obj],
                              cwd=DISTS_DIR)
        return subprocess.Popen(['git', 'cat-file',
                                 'blob',
                                 obj],
                                cwd=DISTS_DIR,
                                stdout=subprocess.PIPE)
    except subprocess.CalledProcessError:
        raise Exception('Unable to cat file %s' % obj)

def walk(ts):
    p = subprocess.Popen(['git', 'rev-list',
                          '--min-age=%s' % ts.encode('utf-8'),
                          '-1',
                          'HEAD'],
                         stdout=subprocess.PIPE,
                         cwd=DISTS_DIR)
    commit, _ = p.communicate()
    commit = commit.strip()

    if not commit:
        raise Exception('Invalid timestamp')

    p = subprocess.Popen(['git', 'ls-tree',
                          '--name-only',
                          commit],
                         stdout=subprocess.PIPE,
                         cwd=DISTS_DIR)
    suites, _ = p.communicate()
    suites = suites.strip().split()
    suites.remove('.gitignore')

    for suite in suites:
        release = deb822.Release(cat_blob('%s:%s/Release' % (commit, suite)).stdout)

        components = release['Components'].split()
        arches = release['Architectures'].split()

        for component in components:
            for arch in arches:
                print 'Checking %s:%s/%s/%s' % (commit, suite, component, arch)
                try:
                    packages = cat_blob('%s:%s/%s/binary-%s/Packages' % (commit, suite, component, arch)).stdout
                except:
                    print 'Packages file unavailable'
                    continue
                for p in deb822.Packages.iter_paragraphs(packages):
                    f = hashlib.sha1(os.path.basename(p['Filename'])).hexdigest()
                    path = os.path.join(CACHE_DIR, f[0], f[1], f[2:])
                    if os.path.exists(path) and os.stat(path).st_size != p['Size']:
                        print 'Purging %s' % p['Filename']
                        os.unlink(path)

def main(args):
    if not args:
        print >>sys.stderr, 'Usage: %s <timestamp> [<timestamp> ..]' % sys.argv[0]
        return

    for arg in args:
        walk(arg)

if __name__ == '__main__':
    main(sys.argv[1:])