以防万一，如果您也对配置单元解决方案感兴趣，则横向视图在连接多个阵列结果集时会产生笛卡尔积。使用posexplode可以获得相同的结果，如下所示：

select Userid,Name,phone,address
from source_table
lateral view posexplode(array(Phone1,Phone2,Phone3))  valphone as x,phone
lateral view posexplode(array(Address1,Address2,Address3)) valaddress as t,address
where x=t
;

hive> set hive.cli.print.header=true;

userid  name    phone   address
123     Jitu    123456  DELHI
123     Jitu    987654  GURGAON
123     Jitu    111111  NOIDA
234     Mark    123456  UK
234     Mark    987654  USA
234     Mark    111111  IND
Time taken: 2.759 seconds, Fetched: 6 row(s)

可以为每个列索引多次选择原始Dataframe，然后通过“union”将选定的Dataframe组合成一个：

val df = Seq(
  (123, "Jitu", "123456", "987654", "111111", "DELHI", "GURGAON", "NOIDA"),
  (234, "Mark", "123456", "987654", "111111", "UK", "USA", "IND")
).toDF(
  "Userid", "Name", "Phone1", "Phone2", "Phone3", "Address1", "Address2", "Address3"
)

val columnIndexes = Seq(1, 2, 3)
val onlyOneIndexDfs = columnIndexes.map(idx =>
  df.select(
    $"Userid",
    $"Name",
    col(s"Phone$idx").alias("Phone_no"),
    col(s"Address$idx").alias("Address")))

val result = onlyOneIndexDfs.reduce(_ union _)

输出：

+------+----+--------+-------+
|Userid|Name|Phone_no|Address|
+------+----+--------+-------+
|123   |Jitu|123456  |DELHI  |
|123   |Jitu|111111  |NOIDA  |
|123   |Jitu|987654  |GURGAON|
|234   |Mark|123456  |UK     |
|234   |Mark|987654  |USA    |
|234   |Mark|111111  |IND    |
+------+----+--------+-------+