elastic.py 文件源码

def tokenize(
        string, es_client,
        field_name=None, index_name=None, analyzer_name=None):
    """Tokenize a string based on analyzer of the provided field

    Args:
        string (string): the string to tokenize
        es_client (EsClient): elasticsearch client.
        field_name (string): the field whose analyzer is used to
            tokenize
        index_name (string): name of the index
    Returns:
        tokens (list): a list of tokens

    Raises:
        ElasticsearchClientError: if no field name or index name
            are available.
    """
    if field_name is None:
        field_name = es_client.field_name

    if index_name is None:
        index_name = es_client.index_name

    req = {'body': string, 'index': index_name}
    if analyzer_name is None:
        req['field'] = field_name
    else:
        req['analyzer'] = analyzer_name

    try:
        response = es_client.indices.analyze(**req)
        tokens = [d['token'] for d in response['tokens']]
    except elasticsearch.exceptions.RequestError:
        tokens = []

    return tokens