解码日期并与列进行比较

Question

我有专栏：

1.:delivery_date 列.. 它是 returns 星期几，像这样：

1  --monday
2  --tuesday
3  --wednesday
4  --thursday
5  --friday

2.: date 列.. returns 取件日期 - 这是一个正常的日期列 - 例如：25.03.2019 (format DD.MM.YYYY)

但是， 我们的系统会自动比较并将 date 设置为最接近的 delivery_date。

示例：

合作伙伴每周delivery_date 星期四和星期二.. [所以 4 和 2]。他想在星期五提货。此时我们的系统将 date 更改为星期四。但我想要下一个 date，也就是星期二 - 我想解码它到现在为止（格式 DD.MM.YYYY）

所以输出如下：

--if pick up date: 22.03.2019 (friday)
--partner has delivery_date: 4,2(thursday, tuesday)

I want to get: 26.03.2019 --as output

我可以比较列..这不是问题..但是我如何解码 "actual pick up" 日期？

我该怎么做？

Answer 1

为了解决这个问题，构建一个 "calendar" table 可能是一个优势 - 让它足够大（！），随后可以用来加密你的数据（见注释下面的代码）。为了进行测试，让我们使用以下 tables (Oracle 12c, 18c):

Table CAL_

create table cal_
as
select sysdate + level date_
, to_char( sysdate + level, 'Day' ) dname_
, to_char( sysdate + level, 'D' )   dnumber_
from dual
connect by level <= 31 ;  -- number should be big enough to cover the lifetime of your system

SQL> select * from cal_ ;
DATE_      DNAME_     DNUMBER_  
17-MAR-19  Sunday     7         
18-MAR-19  Monday     1         
19-MAR-19  Tuesday    2         
20-MAR-19  Wednesday  3         
21-MAR-19  Thursday   4  
...
12-APR-19  Friday     5         
13-APR-19  Saturday   6         
14-APR-19  Sunday     7         
15-APR-19  Monday     1         
16-APR-19  Tuesday    2

Table 交货天数

create table deliverydays ( partner, deliveryday, dayofweek )
as
select
  mod( level, 3 ) + 1 
, mod( level, 5 ) + 1
, to_char( sysdate + ( mod( level, 5 ) + 2 ), 'Day' )
from dual
connect by level <= 10 ;

-- partner 4 delivers on Tue and Thu (same days as in the question)
insert into deliverydays ( partner, deliveryday, dayofweek )
values ( 4, 2, 'Tuesday' ) ;
insert into deliverydays ( partner, deliveryday, dayofweek )
values ( 4, 4, 'Thursday' ) ;

SQL> select * from deliverydays order by partner, deliveryday ;
PARTNER  DELIVERYDAY  DAYOFWEEK  
1        2            Tuesday    
1        4            Thursday   
1        5            Friday     
2        1            Monday     
2        2            Tuesday    
2        3            Wednesday  
2        5            Friday     
3        1            Monday     
3        3            Wednesday  
3        4            Thursday   
4        2            Tuesday    
4        4            Thursday   

12 rows selected.

您可以像这样DENSIFY您的数据。请注意，交货日期之间的间隔变得可见。（带有尾随下划线的列名：来自 CAL_ table）。

select 
  partner, deliveryday , dayofweek 
, date_, dname_, dnumber_
from deliverydays D partition by ( partner )
  right join cal_ C on D.deliveryday = C.dnumber_
where partner = 4
order by date_ ;

PARTNER  DELIVERYDAY  DAYOFWEEK  DATE_      DNAME_     DNUMBER_  
4        NULL         NULL       17-MAR-19  Sunday     7         
4        NULL         NULL       18-MAR-19  Monday     1         
4        2            Tuesday    19-MAR-19  Tuesday    2         
4        NULL         NULL       20-MAR-19  Wednesday  3         
4        4            Thursday   21-MAR-19  Thursday   4         
4        NULL         NULL       22-MAR-19  Friday     5         
4        NULL         NULL       23-MAR-19  Saturday   6         
4        NULL         NULL       24-MAR-19  Sunday     7         
4        NULL         NULL       25-MAR-19  Monday     1         
4        2            Tuesday    26-MAR-19  Tuesday    2         
4        NULL         NULL       27-MAR-19  Wednesday  3         
4        4            Thursday   28-MAR-19  Thursday   4  
...
-- 31 rows selected

下一步，您可以使用 LEAD() 函数来查找 "next" 交货日期，然后计算 "picking up" 和交货之间的天数，并根据需要格式化日期。（请参阅查询 here 的 "commented" 版本。）

select pid, dday, next_dday, date_, dname_, dnumber_
, case 
    when dnumber_ > next_dday then ( 7 - dnumber_ + next_dday ) -- next week
    else next_dday - dnumber_
  end ddiff_
, case
    when dnumber_ > next_dday then
      to_char( date_ + ( 7 - dnumber_ + next_dday ), 'DD.MM.YYYY' )
    else 
      to_char( date_ + ( next_dday - dnumber_ ), 'DD.MM.YYYY' )
  end delivered_on 
from (
  select  
    partner as pid, deliveryday as dday
  , lead( deliveryday ) ignore nulls over ( order by date_ ) as next_dday
  , dayofweek 
  , date_, dname_, dnumber_
  from deliverydays D partition by ( partner )
    right join cal_ C on D.deliveryday = C.dnumber_
  where partner = 4 -- partner id
)
where to_char( date_, 'DD.MM.YYYY' ) = '22.03.2019' -- "pick up" date
order by date_  ;

结果

PID  DDAY  NEXT_DDAY  DATE_      DNAME_     DNUMBER_  DDIFF_  DELIVERED_ON  
4    NULL  2          22-MAR-19  Friday     5         4       26.03.2019

不使用外部 WHERE 子句，我们得到 ..

PID  DDAY  NEXT_DDAY  DATE_      DNAME_     DNUMBER_  DDIFF_  DELIVERED_ON  
4    NULL  2          17-MAR-19  Sunday     7         2       19.03.2019    
4    NULL  2          18-MAR-19  Monday     1         1       19.03.2019    
4    2     4          19-MAR-19  Tuesday    2         2       21.03.2019    
4    NULL  4          20-MAR-19  Wednesday  3         1       21.03.2019    
4    4     2          21-MAR-19  Thursday   4         5       26.03.2019    
4    NULL  2          22-MAR-19  Friday     5         4       26.03.2019    
4    NULL  2          23-MAR-19  Saturday   6         3       26.03.2019    
4    NULL  2          24-MAR-19  Sunday     7         2       26.03.2019    
4    NULL  2          25-MAR-19  Monday     1         1       26.03.2019    
4    2     4          26-MAR-19  Tuesday    2         2       28.03.2019    
4    NULL  4          27-MAR-19  Wednesday  3         1       28.03.2019    
4    4     2          28-MAR-19  Thursday   4         5       02.04.2019 
...
-- 31 rows selected

解码日期并与列进行比较

decode the day and compare with a column

sql

oracle

decode