Question

我正在尝试运行此查询，其中包含Hive中的sum(...) over(...)函数，但显示错误。我试图在子查询中使用select distinct，但它仍然不起作用。

join函数中是否有错误？

这是我的SQL代码

    select
c.driver_city_name,
c.driver_car_brand,
c.year,
count(distinct c.driver_id)over(PARTITION BY c.driver_car_brand,c.driver_city_name,c.year),
sum(c.upply_hr)over(PARTITION BY c.driver_car_brand,c.driver_city_name,c.year),
sum(c.work_hr)over(PARTITION BY c.driver_car_brand,c.driver_city_name,c.year)
from (
    select
    a.driver_city_name,
    a.driver_car_brand,
    a.driver_id,
    year(b.reg_date_cheling) as year,
    d.onlinetime as supply_hr,
    d.charge_time_length/60/60 as work_hr
    from gulfstream_dw.dw_v_driver_base a
    join (
        select 
        driver_id,
        reg_date_cheling
        from g_bi.t_brx_xinzheng_driver_diaodu_1
        where driver_city_id in (17,18,2,3,1,5,10,4,24,23)
        and concat_ws('-',year,month,day) BETWEEN '2016-11-19' and '2017-02-21'
    ) b on a.driver_id = b.driver_id
    join (
        select
        e.driver_id,
        sum(e.onlinetime) as onlinetime, --在线时长（单位：小时）
        sum(e.charge_time_length) as charge_time_length --计费总时长（单位：秒）
        from(
            select distinct
            concat_ws('-',year,month,day) date1,
            driver_id,
            onlinetime, --在线时长（单位：小时）
            charge_time_length --计费总时长（单位：秒）
            from gulfstream_dw.dw_m_driver_strategy
            where concat_ws('-',year,month,day) between '2016-11-19' and '2017-02-21'
            and onlinetime > 0
         ) e
         group by e.driver_id
    ) d on a.driver_id = d.driver_id
    where driver_city_id in (17,18,2,3,1,5,10,4,24,23)
    and to_date(max_success_strive_time) BETWEEN '2016-11-20' and '2017-02-20' 
    and concat_ws('-',year,month,day) BETWEEN '2016-11-19' and '2017-02-21'
    and a.driver_car_brand in (
    "奇瑞-E3",
    "吉利-远景",
    )
    group by a.driver_city_name,a.driver_car_brand,a.driver_id,year(b.reg_date_cheling),d.onlinetime,d.charge_time_length/60/60
)c

group by c.driver_city_name,c.driver_car_brand,c.year

但是有错误：

错误类型：SEMANTIC_FAILED编译语句时出错：FAILED： SemanticException无法将窗口调用分解为组。至少有一个组必须仅依赖于输入列。还要检查循环依赖。潜在的错误： org.apache.hadoop.hive.ql.parse.SemanticException：第6行：6无效列引用'supply_hr'

如果我遗漏sum over函数并保留count over函数，则会成功运行。

使用sum（）超过函数错误的Hive - 无法将窗口调用分解为组

0 个答案: