UDF返回数据的例子

spark = SparkSession.builder. 
    appName("wk_stg_10650018_cs"). 
    enableHiveSupport(). 
    config("spark.debug.maxToStringFields", "100"). 
    getOrCreate()

df = spark.createDataFrame([('a', [1, 2, 3],['x','y']), ('b', [5, 2, 3],['x','y'])], ['id', 'c1','c2'])
df.createOrReplaceTempView('t1')

def gen(begin,end):
    return list(range(begin,end))
#
spark.udf.register('gen', gen, ArrayType(IntegerType()))

spark.sql('select id, gen(1,3) as arr from t1').createOrReplaceTempView('t1')
spark.sql(
    "select id,c"
    " from t1"
    " lateral view explode(arr) A as c"
).show()