Ejemplo n.º 1
0
def fetch_user_events(login):
    logger = logging.getLogger('events')
    logger.debug('fetching events for %s' % (login, ))

    db = get_db()
    user = db.users.find_one({'login': login})

    # calculating time of last saved event
    last_item = list(
        db.received_events.find({
            'gitorama.login': user['login']
        }).sort([('created_at', -1)])[:1])
    if last_item:
        last_item = last_item[0]['id']  # this is a GitHub's event id
    else:
        last_item = None

    gh = net.GitHub(token=user['gitorama']['token'])
    for event in gh.get_iter('/users/{0}/received_events'.format(
            user['login']),
                             per_page=30):
        if event['id'] == last_item:
            # don't fetch more than needed
            # this item already saved
            break

        event['created_at'] = times.to_universal(event['created_at'])
        event['gitorama'] = {'login': user['login']}
        db.received_events.save(event)
Ejemplo n.º 2
0
def update_user(login):
    g.db = core.get_db()
    now = times.now()

    stats_to_save = ('followers', 'following', 'disk_usage', 'public_repos')

    user = g.db.users.find_one({'login': login})

    gh = net.GitHub(token=user['gitorama']['token'])

    # update user's data
    new_user_data = gh.get('/user')
    user.update(new_user_data)
    user['gitorama']['update_at'] = now + app.config['USER_UPDATE_INTERVAL']
    g.db.users.save(user)

    # update users's repositories
    repositories = gh.get('/user/repos')

    for rep in repositories:
        rep_from_db = g.db.user_reps.find_one(
            {
                'owner.login': rep['owner']['login'],
                'name': rep['name']
            }) or {}
        rep_from_db.update(rep)
        g.db.user_reps.save(rep_from_db)

    today = datetime.datetime(now.year, now.month, now.day)
    key = dict(login=user['login'], date=today)
    stats = g.db.user_stats.find_one(key) or key

    stats.update((key, value) for key, value in user.iteritems()
                 if key in stats_to_save)
    g.db.user_stats.save(stats)
Ejemplo n.º 3
0
def cluster_reps_into_networks():
    db = core.get_db()
    gh = net.GitHub()

    query = {'gitorama.net_id': {'$exists': False}}

    rep = db.user_reps.find(query).limit(1)
    while rep:
        forks = gh.get('/repos/{0[owner][login]}/{0[name]}/forks'.format(rep))
        import pdb;pdb.set_trace()
Ejemplo n.º 4
0
def update_relations(login):
    from gitorama import app
    db = core.get_db()
    gh = net.GitHub()
    now = times.now()

    db.relations.ensure_index('login', unique=True)
    doc = db.relations.find_one({'login': login})

    if 'following' not in doc and 'followers' not in doc:
        doc['following'] = []
        doc['followers'] = []
        # don't create events for the first run
        create_events = False
    else:
        create_events = True

    def process(handle, new_event, missing_event):
        """Process new users in following/followed lists.
        """
        old = set(doc[handle])

        new = gh.get('/users/{login}/{handle}'.format(**locals()))
        new = set(f['login'] for f in new)

        new_items = new - old
        absent_items = old - new

        if create_events:
            if new_items:
                db.events.insert(
                    (
                        {'login': login, 'e': new_event, 'who': who, 'date': now}
                        for who in new_items
                    ),
                    manipulate = False,
                )

            if absent_items:
                db.events.insert(
                    (
                        {'login': login, 'e': missing_event, 'who': who, 'date': now}
                        for who in absent_items
                    ),
                    manipulate = False,
                )

    process('following', 'follow', 'unfollow')
    process('followers', 'followed', 'unfollowed')
    doc['update_at'] = now + app.config['USER_UPDATE_INTERVAL']
    db.relations.save(doc, manipulate=False, safe=True)
Ejemplo n.º 5
0
def update_reps_data():
    db = core.get_db()
    gh = net.GitHub()

    timestamp = int(time.time())

    query = {
        '$or': [
            {'gitorama.updated_at': {'$exists': False}},
            {'gitorama.updated_at': {'$lt': timestamp - 24 * 3600}},
        ]
    }

    for rep in db.user_reps.find(query):
        # update rep's data
        new_rep_data = gh.get('/repos/{0[owner][login]}/{0[name]}'.format(rep))
        rep.update(new_rep_data)
        rep.setdefault('gitorama', {})
        rep['gitorama']['updated_at'] = timestamp
        db.user_reps.save(rep)