#!/usr/bin/python3

# Checks the master manifest and ensures the local manifest is up-to-date

import codecs
import contextlib
import gzip
import json
import time
import urllib.request


@contextlib.contextmanager
def local_manifest(manifile):
    reader = codecs.getreader("utf-8")
    with gzip.open(manifile) as f:
        manifest = json.load(reader(f))
        yield manifest


def remote_manifest(url):
    reader = codecs.getreader("utf-8")
    tries = 3
    for x in range(tries):
        try:
            remote = urllib.request.urlopen(url)
            with gzip.open(remote) as f:
                return json.load(reader(f))
        except:
            if x + 1 == tries:
                raise
            time.sleep(x+1)


def _handle_deletes(lrepos, rrepos):
    # we don't know when repos are deleted, so we use a local temp file to
    # track deletes we've missed, we allow one miss before issuing a warning
    try:
        with open('/tmp/grok-check-manifest-deletes.json') as f:
            to_del_local = json.load(f)
    except:
        to_del_local = []
    deletes = lrepos - rrepos

    with open('/tmp/grok-check-manifest-deletes.json', 'w') as f:
        json.dump(list(deletes), f)

    to_del = []
    for r in deletes:
        if r in to_del_local:
            to_del.append(r)  # we should have deleted by now
    return to_del


def compare_manifests(local, remote):
    # ensure they have the same repos defined
    lrepos = set(sorted(local.keys()))
    rrepos = set(sorted(remote.keys()))

    to_del = _handle_deletes(lrepos, rrepos)

    to_add = []
    for r in rrepos - lrepos:
        # give newly added repos a little slack for cloning over
        elapsed = time.time() - remote[r]['modified']
        if elapsed > 120:
            to_add.append(r)

    # test fingerprints match
    bad = []
    for repo in lrepos & rrepos:
        rem_fp = remote[repo]['fingerprint']
        local_fp = local[repo]['fingerprint']

        if rem_fp != local_fp:
            elapsed = time.time() - remote[repo]['modified']
            threshold = 90  # normally 90 seconds is enough
            if 'kernel.git' in repo or 'linux.git' in repo:
                threshold = 240  # give more time for kernel syncs
            if elapsed > threshold:
                bad.append('%s rem=%s us=%s' % (repo, rem_fp, local_fp))

    return to_add, to_del, bad


if __name__ == '__main__':
    import argparse
    parser = argparse.ArgumentParser(
        description='''Verify local repositories are in sync by checking
                    remote server's manifest''')
    parser.add_argument('--local', default='/srv/repositories/manifest.js.gz',
                        help='Local manifest to analyze. default=%(default)s')
    parser.add_argument('manifest_url', help='''Manifest url. eg
                        http://git-us.linaro.org/manifest.js.gz''')
    args = parser.parse_args()

    remote = remote_manifest(args.manifest_url)
    with local_manifest(args.local) as local:
        to_add, to_del, to_sync = compare_manifests(local, remote)

        if to_add:
            print('Missing repos:\n %s' % '\n '.join(to_add))
        if to_del:
            print('Delete repos:\n  %s' % '\n  '.join(to_del))
        if to_sync:
            print('Repos out sync:')
            print(' ' + '\n '.join(to_sync))