api.py 文件源码

def search_for_field(search_obj, field_name, page_size=DEFAULT_ES_LOOP_PAGE_SIZE):
    """
    Retrieves all unique instances of a field for documents that match an ES query

    Args:
        search_obj (Search): Search object
        field_name (str): The name of the field for the value to get
        page_size (int): Number of docs per page of results

    Returns:
        set: Set of unique values
    """
    results = set()
    # Maintaining a consistent sort on '_doc' will help prevent bugs where the
    # index is altered during the loop.
    # This also limits the query to only return the field value.
    search_obj = search_obj.sort('_doc').fields(field_name)
    loop = 0
    all_results_returned = False
    while not all_results_returned:
        from_index = loop * page_size
        to_index = from_index + page_size
        search_results = execute_search(search_obj[from_index: to_index])
        # add the field value for every search result hit to the set
        for hit in search_results.hits:
            results.add(getattr(hit, field_name)[0])
        all_results_returned = to_index >= search_results.hits.total
        loop += 1
    return results