Hive提取小时内，分组排名前3的sql

表的结构是这样的，时间的范围我是提取了几个小时内的数据

create table  `alibaba.user_bea` (
user_id  bigint,
    item_id bigint,
    cate_id bigint,
    times     string,
    bea      int
   
)
PARTITIonED BY (dt STRING)
ROW FORMAT DELIMITED FIELDS TERMINATED BY ',' 
STORED AS orc;

提取时间

select item_id,ranks from (Select item_id, from_unixtime(cast(times as bigint),'yyyyMMdd HH')  as  times,row_number() over (partition by times,cate_id  order by count(*)  desc )as  ranks  from user_bea  group by cate_id ,times,item_id)t1 where ranks<=3;

数据下载地址
导入方法如下

#hdfs路径
load data inpath "/flinkworkspace/work/input.har/part-0" overwrite  into table user_bea_tmp;
#本地路径
load data  local   inpath "/home/lin/part-0" overwrite  into table user_bea_tmp;