pyspark.pandas.Series.str.partition#

str.partition(sep=' ', expand=True)#

Not supported.