视图中的动态列名 (Postgres)

Question

我目前正在编写一个 SQL 视图，它应该提供特定月份的填充字段的计数。

这就是我希望构建视图的方式：

Country   | (Current Month - 12) Eg Feb 2011 | (Current Month - 11) | (Current Month - 10)
----------|----------------------------------|----------------------|---------------------
UK        | 10                               | 11                   | 23

月份下的数字应该是特定国家/地区所有填充字段的计数。该字段名为 eldate，是格式为 10-12-2011 的日期（转换为字符）。我希望计数只计算与月份匹配的日期。

因此，“当前月份 - 12”列应仅包含当前 12 个月之前的月份内的日期计数。例如，当前月份 - 英国的 12 应该包括在 2011 年 2 月之间的日期计数。

我希望列标题实际反映它所查看的月份：

Country | Feb 2011 | March 2011 | April 2011
--------|----------|------------|------------
UK      | 4        | 12         | 0

比如：

SELECT c.country_name,
        (SELECT COUNT("C1".eldate) FROM "C1" WHERE "C1".eldate LIKE %NOW()-12 Months% AS NOW() - 12 Months
        (SELECT COUNT("C1".eldate) FROM "C1" WHERE "C1".eldate LIKE %NOW()-11 Months% AS NOW() - 11 Months
FROM country AS c
INNER JOIN "site" AS s using (country_id)
INNER JOIN "subject_C1" AS "C1" ON "s"."site_id" = "C1"."site_id"

显然这不起作用，只是为了让您了解我的意思。

有什么想法吗？

感谢您的帮助，如有更多疑问请咨询。

Answer 1

我的第一个想法是制作这张桌子：

+---------+-------+--------+
| Country | Month | Amount |
+---------+-------+--------+
| UK      | Jan   | 4      |
+---------+-------+--------+
| UK      | Feb   | 12     |
+---------+-------+--------+

等等。并转动它。所以你会从（例如）开始：

SELECT 
  c.country, 
  EXTRACT(MONTH FROM s.eldate) AS month, 
  COUNT(*) AS amount
FROM country AS c
JOIN site AS s ON s.country_id = c.id
WHERE 
  s.eldate > NOW() - INTERVAL '1 year'
GROUP BY c.country, EXTRACT(MONTH FROM s.eldate);

然后您可以将其插入 one the crosstab functions from the tablefunc module 以实现枢轴，执行以下操作：

SELECT * 
FROM crosstab('<query from above goes here>') 
  AS ct(country varchar, january integer, february integer, ... december integer);

Answer 2

您可以截断日期以进行比较：

WHERE date_trunc('month', eldate) = date_trunc('month', now()) - interval '12 months'

更新

您的查询的这种替换：

(SELECT COUNT("C1".eldate) FROM "C1" WHERE date_trunc('month', "C1".eldate) =
    date_trunc('month', now()) - interval '12 months') AS TWELVE_MONTHS_AGO

但这将涉及每个月对表的扫描，因此您可以进行单次扫描，并按照以下方式进行更多操作：

SELECT SUM( CASE WHEN date_trunc('month', "C1".eldate) =  date_trunc('month', now()) - interval '12 months' THEN 1 ELSE 0 END ) AS TWELVE_MONTHS_AGO
      ,SUM( CASE WHEN date_trunc('month', "C1".eldate) =  date_trunc('month', now()) - interval '11 months' THEN 1 ELSE 0 END ) AS ELEVEN_MONTHS_AGO
...

或与其他人显示的月份表进行联接。

更新2

关于修复从一月到十二月的列的评论，我在想这样的事情：过滤过去几年的记录，然后在适当的月份求和。可能是这样的：

SELECT SUM( CASE WHEN EXTRACT(MONTH FROM "C1".eldate) = 1 THEN 1 ELSE 0 END ) AS JAN
      ,SUM( CASE WHEN EXTRACT(MONTH FROM "C1".eldate) = 2 THEN 1 ELSE 0 END ) AS FEB
      ...

  WHERE date_trunc('month', "C1".eldate) <  date_trunc('month', now())
    AND date_trunc('month', "C1".eldate) >= date_trunc('month', now()) - interval '12 months'