在三个where语句中加入两个table，count列行

Question

我有一个表lot_bill_of_operations，其中包含行和表lot_bill_of_lots_serial_xref，其中包含序列号和时间戳列。这些表与主键bol_id相关联。

我正在尝试计算逐行传递三个班次的序列号，但每个子查询的结果都相同。

以下是查询：

select lot_bill_of_operations.wc_line_code,

(select  count(part_sn) from lot_bill_of_lots_serial_xref
  where lot_bill_of_lots_serial_xref.time_stamp between '07-MAY-17 06:00:00'
    and '07-MAY-17 14:00:00'
  group by lot_bill_of_operations.wc_line_code)firstshift,

(select  count(part_sn) from lot_bill_of_lots_serial_xref
  where lot_bill_of_lots_serial_xref.time_stamp between '07-MAY-17 14:00:00'
    and '07-MAY-17 22:00:00'
  group by lot_bill_of_operations.wc_line_code)secondshift,

(select  count(part_sn) from lot_bill_of_lots_serial_xref
  where lot_bill_of_lots_serial_xref.time_stamp between '07-MAY-17 22:00:00'
    and '08-MAY-17 06:00:00'
  group by lot_bill_of_operations.wc_line_code)thirdshift

FROM lot_bill_of_operations inner JOIN lot_bill_of_lots_serial_xref
    ON lot_bill_of_operations.bol_id = lot_bill_of_lots_serial_xref.bol_id
group by lot_bill_of_operations.wc_line_code

以下是获得的结果：

我做错了什么？

Answer 1

看起来您需要条件计数，您可以使用案例表达式获取：

PATCH ../tables/{id|name}/columns/{id}

我已切换到时间戳文字，因为您依靠NLS设置将字符串隐式转换为时间戳。 （或者可能是日期，从你所展示的内容中不清楚）。

但是，使用select lbo.wc_line_code, count(case when lblsx.time_stamp between timestamp '2017-05-07 06:00:00' and timestamp '2017-05-07 14:00:00' then lblsx.part_sn end) as firstshift, count(case when lblsx.time_stamp between timestamp '2017-05-07 14:00:00' and timestamp '2017-05-07 22:00:00' then lblsx.part_sn end) as secondshift, count(case when lblsx.time_stamp between timestamp '2017-05-07 22:00:00' and timestamp '2017-05-08 06:00:00' then lblsx.part_sn end) as thirdshift from lot_bill_of_operations lbo inner join lot_bill_of_lots_serial_xref lblsx on lbo.bol_id = lblsx.bol_id group by lbo.wc_line_code; 表示期间略有重叠，因此您可能真的需要：

between

您可能还希望将select lbo.wc_line_code, count(case when lblsx.time_stamp >= timestamp '2017-05-07 06:00:00' and lblsx.time_stamp < timestamp '2017-05-07 14:00:00' then lblsx.part_sn end) as firstshift, count(case when lblsx.time_stamp >= timestamp '2017-05-07 14:00:00' and lblsx.time_stamp < timestamp '2017-05-07 22:00:00' then lblsx.part_sn end) as secondshift, count(case when lblsx.time_stamp >= timestamp '2017-05-07 22:00:00' and lblsx.time_stamp < timestamp '2017-05-08 06:00:00' then lblsx.part_sn end) as thirdshift from lot_bill_of_operations lbo inner join lot_bill_of_lots_serial_xref lblsx on lbo.bol_id = lblsx.bol_id group by lbo.wc_line_code; 范围包含为整体过滤器，以减少检查的数据量：

time_stamp

如果您在当天使用此算法，则可以使用日期算术而不是固定值。

Answer 2

如果没有样本数据，很难理解您的需求，但根据您的查询，我认为您需要以下内容。

• 还要避免比较字符串的日期。使用日期转换功能 比较to_date / to_timestamp，同时比较日期的准确和 没有模糊的结果。
• 在给出范围时使用between。

• 使用 表名的别名，以提高可读性。

  select o.wc_line_code
      ,count(case when 
              x.time_stamp between to_date('07-MAY-17 06:00:00','DD-MON-YY HH24:MI:SS') 
                              and  to_date('07-MAY-17 14:00:00','DD-MON-YY HH24:MI:SS')
              then 1) as firstshift
  -- , similary for other shifts
  
  FROM lot_bill_of_operations o 
      inner JOIN 
       lot_bill_of_lots_serial_xref x
      ON o.bol_id = x.bol_id
  group by o.wc_line_code;
  

Answer 3

尝试以下方法：

select lot_bill_of_operations.wc_line_code
, sum (case when time_stamp between '07-MAY-17 06:00:00' and '07-MAY-17 14:00:00'
 then 1 else 0 end ) firstshift
, sum (case when time_stamp between '07-MAY-17 14:00:00' and '07-MAY-17 22:00:00'
 then 1 else 0 end ) secondshift
, sum (case when time_stamp between '07-MAY-17 22:00:00' and '08-MAY-17 06:00:00'
 then 1 else 0 end ) thirdshift
FROM lot_bill_of_operations inner JOIN lot_bill_of_lots_serial_xref
    ON lot_bill_of_operations.bol_id = lot_bill_of_lots_serial_xref.bol_id
group by lot_bill_of_operations.wc_line_code
;