使用PostgreSQL中的常见表表达式(CTE)构建表时，如何获得唯一的字段

Question

我正在尝试生成一系列时间戳，计算每个时间戳的一周一天，并仅选择周一至周六的时间戳，使用PostgreSQL。

。

这是我到目前为止的尝试：

WITH
candidates AS (SELECT * FROM generate_series(
  date_trunc('hour', CURRENT_TIMESTAMP - INTERVAL '2 hours'),
  date_trunc('hour', CURRENT_TIMESTAMP - INTERVAL '2 hours')+INTERVAL '1 week',
  INTERVAL '1 hour')),
candidows AS (SELECT EXTRACT(DOW FROM generate_series) FROM candidates),
candidatable AS (SELECT b.date_part AS dow, a.generate_series as start_booking FROM candidates a, candidows b)
SELECT * FROM candidatable WHERE dow != 0;

返回的视图包含一个陶列列，该列的行为是我所期望的，但是start_booking列只是重复相同的值：

 dow |     start_booking
-----+------------------------
   2 | 2017-09-19 11:00:00+01
   2 | 2017-09-19 11:00:00+01
   2 | 2017-09-19 11:00:00+01
   2 | 2017-09-19 11:00:00+01
   2 | 2017-09-19 11:00:00+01
   2 | 2017-09-19 11:00:00+01
   2 | 2017-09-19 11:00:00+01
   2 | 2017-09-19 11:00:00+01
   2 | 2017-09-19 11:00:00+01
   2 | 2017-09-19 11:00:00+01
   2 | 2017-09-19 11:00:00+01
   2 | 2017-09-19 11:00:00+01
   2 | 2017-09-19 11:00:00+01
   3 | 2017-09-19 11:00:00+01
   3 | 2017-09-19 11:00:00+01
   3 | 2017-09-19 11:00:00+01

如果我将最后一个选择语句替换为：

SELECT * FROM candidates;

然后，我试图将上述启动图视图查看的数据是顺序的，正如我所期望的：

    generate_series
------------------------
 2017-09-19 11:00:00+01
 2017-09-19 12:00:00+01
 2017-09-19 13:00:00+01
 2017-09-19 14:00:00+01
 2017-09-19 15:00:00+01
 2017-09-19 16:00:00+01

我该怎么做才能从candidates.generate_series出现为candidatables.start_booking？

Answer 1

一个疯狂的猜测 - 您是否想：

WITH
candidates AS (SELECT generate_series start_booking, EXTRACT(DOW FROM generate_series) dow FROM generate_series(
  date_trunc('hour', CURRENT_TIMESTAMP - INTERVAL '2 hours'),
  date_trunc('hour', CURRENT_TIMESTAMP - INTERVAL '2 hours')+INTERVAL '1 week',
  INTERVAL '1 hour'))
SELECT * FROM candidates WHERE dow != 0;

？..