import boto.emr


ACTIVE_STATES = [
    'BOOTSTRAPING',
    'RUNNING',
    'STARTING',
    'WAITING']


def find_active_emr_cluster_id_by_tag(tag_name):
    """Find an active emr cluster id from a tag name.

    Args:
        tag_name (str): the tag to look for.
    Return:
        str: the emr cluster id.
    """

    connector = boto.emr.connection.EmrConnection()
    cluster_list = connector.list_clusters()
    if not cluster_list:
        return None

    for cluster in cluster_list.clusters:
        if cluster.status.state not in ACTIVE_STATES:
            continue

        # List Cluster returns the list of all the clusters but some
        # informations such as the tag list is missing. Calling cluster.tag
        # returns constantly an empty list. For all active clusters, we ask
        # a more complete description – which containst he tags.
        tags = connector.describe_cluster(cluster.id).tags

        if not tags:
            continue

        for tag in tags:
            if tag_name == tag.key:
                return cluster.id

    return None