update_by_query.py

from elasticsearch_dsl import connections, UpdateByQuery


def bulk_update_docs(ids, instance, fields_changed=[]):
    """
    Arguments:
        ids: {list} -- list of ids of documents we want to update
        instance {Model instance} -- The updated user model object
        fields_changed {list} -- list of the model fields changed
    """

    body = generate_sub_dictionary(instance, fields_changed)
    client = connections.create_connection()
    qry = UpdateByQuery(using=client, index='tweets').params(conflicts='proceed').query('terms', _id=ids)
    
    # update_by_query uses scripts, so we need to generate the formatted source string that matches our body structure
    source = generate_source_string(body)
    qry = qry.script(source=source, params=body)


def generate_source_string(body):
    """
    Recursive algorithm to generate source strings for creating the bulk update call to ES.
    Input:
    {
        'user': {
            'city': {
                'name': 'New Cologne',
                'slug': 'new-cologne',
            }
        }
    }

    Result:
    => ['ctx._source.user.city.name=params.user.city.name', 'ctx._source.user.city.slug=params.user.city.slug']
    """