Hive生成日期维度表

1、时间维表(完整版)

1)、 建表

-- 时间维表 完整版
create table if not exists dim.dim_date (
 date_id                   string         comment '日期(yyyymmdd)'
,datestr                   string         comment '日期(yyyy-mm-dd)'
,date_name                 string         comment '日期名称中文'
,weekid                    int            comment '周(0-6,周日~周六)'
,week_cn_name              string         comment '周_名称_中文'
,week_en_name              string         comment '周_名称_英文'
,week_en_nm                string         comment '周_名称_英文缩写'
,yearmonthid               string         comment '月份id(yyyymm)'
,yearmonthstr              string         comment '月份(yyyy-mm)'
,monthid                   int            comment '月份id(1-12)'
,monthstr                  string         comment '月份'
,month_cn_name             string         comment '月份名称_中文'
,month_en_name             string         comment '月份名称_英文'
,month_en_nm               string         comment '月份名称_简写_英文'
,quarterid                 int            comment '季度id(1-4)'
,quarterstr                string         comment '季度名称'
,quarter_cn_name           string         comment '季度名称_中文'
,quarter_en_name           string         comment '季度名称_英文'
,quarter_cn_nm             string         comment '季度名称_简写中文'
,quarter_en_nm             string         comment '季度名称_简写英文'
,yearid                    int            comment '年份id'
,year_cn_name              string         comment '年份名称_中文'
,year_en_name              string         comment '年份名称_英文'
,month_start_date          string         comment '当月1号(yyyy-mm-dd)'
,month_end_date            string         comment '当月最后日期(yyyy-mm-dd)'
,month_timespan            int            comment '月跨天数'
,week_of_year              int            comment '当年第几周'
,workday_flag              string         comment '是否工作日(周一至周五Y,否则:N)'
,weekend_flag              string         comment '是否周末(周六和周日Y,否则:N)'
)comment '日期维度表'
stored as orc;

2)、插入数据

insert overwrite table dim.dim_date

select  date_id
    ,datestr
    ,concat(yearid,'年',monthid,'月',substr(datestr,9,2),'日') as date_name
    ,weekid
    ,case weekid when 0 then '星期日'
        when 1 then '星期一'
        when 2 then '星期二'
        when 3 then '星期三'
        when 4 then '星期四'
        when 5 then '星期五'
        when 6 then '星期六'
    end as week_cn_name
    ,case weekid when 0 then 'Sunday'
        when 1 then 'Monday'
        when 2 then 'Tuesday'
        when 3 then 'Wednesday'
        when 4 then 'Thurday'
        when 5 then 'Friday'
        when 6 then 'Saturday'
    end as week_en_name
    ,case weekid when 0 then 'Sun'
        when 1 then 'Mon'
        when 2 then 'Tues'
        when 3 then 'Wed'
        when 4 then 'Thur'
        when 5 then 'Fri'
        when 6 then 'Sat'
    end as week_en_nm
    ,substr(date_id,1,6) as yearmonthid
    ,substr(datestr,1,7) as yearmonthstr
    ,monthid
    ,concat(yearid,'年',monthid,'月') as monthstr
    ,concat(monthid,'月') as month_cn_name
    ,case monthid when 1 then 'January'
        when 2 then 'February'
        when 3 then 'March'
        when 4 then 'April'
        when 5 then 'May'
        when 6 then 'June'
        when 7 then 'July'
        when 8 then 'August'
        when 9 then 'September'
        when 10 then 'October'
        when 11 then 'November'
        when 12 then 'December'
    end as month_en_name
    ,case monthid when 1 then 'Jan'
        when 2 then 'Feb'
        when 3 then 'Mar'
        when 4 then 'Apr'
        when 5 then 'May'
        when 6 then 'Jun'
        when 7 then 'Jul'
        when 8 then 'Aug'
        when 9 then 'Sept'
        when 10 then 'Oct'
        when 11 then 'Nov'
        when 12 then 'Dec'
    end as month_en_nm
    ,quarterid
    ,concat(yearid,quarterid) as quarterstr
    ,concat(yearid,'年第',quarterid,'季度') as quarter_cn_name
    ,concat(yearid,'Q',quarterid) as quarter_en_name
    ,case quarterid when 1 then '第一季度'
        when 2 then '第二季度'
        when 3 then '第三季度'
        when 4 then '第四季度'
    end as quarter_cn_nm
    ,concat('Q',quarterid) as quarter_en_nm
    ,yearid
    ,concat(yearid,'年') as year_cn_name
    ,yearid as year_en_name
    ,month_start_date
    ,month_end_date
    ,datediff(month_end_date,month_start_date) + 1 as month_timespan
    ,week_of_year
    ,case when weekid in (1,2,3,4,5) then 'Y' else 'N' end as workday_flag
    ,case when weekid in (0,6) then 'Y' else 'N' end as weekend_flag
from
(
    select   from_unixtime(unix_timestamp(datestr,'yyyy-MM-dd'),'yyyyMMdd') as date_id
        ,datestr as datestr
        ,pmod(datediff(datestr, '2012-01-01'), 7)  as weekid
        ,concat(substr(datestr,1,4),substr(datestr,6,2)) as yearmonthid
        ,substr(datestr,1,7) as yearmonthstr
        ,cast(substr(datestr,6,2) as int) as monthid
        ,case when cast(substr(datestr,6,2) as int) <= 3 then 1
            when cast(substr(datestr,6,2) as int) <= 6 then 2
            when cast(substr(datestr,6,2) as int) <= 9 then 3
            when cast(substr(datestr,6,2) as int) <= 12 then 4
        end as quarterid
        ,substr(datestr,1,4) as yearid
        ,date_sub(datestr,dayofmonth(datestr)-1) as  month_start_date  --当月第一天
        ,last_day(date_sub(datestr,dayofmonth(datestr)-1)) month_end_date --当月最后一天
        ,weekofyear(datestr) as week_of_year
    from
    (
        select date_add('2000-01-01',t0.pos) as datestr
            from
            (
                select posexplode(
                        split(
                            repeat('o', datediff(from_unixtime(unix_timestamp('20991231','yyyymmdd'),'yyyy-mm-dd'), '2000-01-01')), 'o'
                        )
                    )
            ) t0
    ) t

) A
;

2、解释一下

使用 posexplode() 函数,获得 [2000-01-01, 2099-12-31] 的日期列表:

select date_add('2000-01-01',t0.pos) as datestr
from (
        select posexplode(
                       split(
                               repeat('o', datediff(from_unixtime(unix_timestamp('20991231','yyyymmdd'),'yyyy-mm-dd'), '2000-01-01')), 'o'
                           )
                   )
    ) t0
  • posexplode(ARRAY a),与 explode 类似,行转列函数,返回的 pos 列用来记录序号,序号从 0 开始,即t0.pos[0,datediff]
    官方解释:将数组分解为多行,并带有 int 类型的附加位置列(原始数组中项的位置,从 0 开始)。返回包含两列 (pos,val) 的行集,数组中的每个元素各占一行。
  • repeat用来构建split分割所需的字符串(以’o’分割的字符串),datediff取日差。

3、熄灯

~

你可能感兴趣的:(大数据,hive,hadoop,数据仓库,时间维表)