Question

我正在编写一个查询，显示每周销售两年的销售情况。所以我想比较不同年份的周数。我得到了这样的结果（week是该年内的周数，如第一周，2010年的第二周）：

week|year|price
1   |2010|5
1   |2011|10
2   |2010|7
2   |2011|13

一切都会好的，但是当我尝试在图表中显示结果时，它会计算前几周等等，并将其放在一个字段中。因此，在图表中，我看到第一周的销售额为15周，秒数为20周。所以我接着尝试将周列与另一列结合起来，将不同年份的同一周数分开。

SELECT
  min(so.id) as id,
  date_part('week',  so.date_order) || ' ' ||
  CASE
    WHEN
      so.date_order >='2010-01-01' and so.date_order <= '2010-12-31'
    THEN
      '2010'
    WHEN
      so.date_order >='2010-01-01' and so.date_order <= '2011-12-31'
    THEN
      '2011'
  END as week,
  date_part('year',  so.date_order) as year,
  count(*) as amount,
  sum(sol.price_unit) as price
FROM
  sale_order as so, 
  sale_order_line as sol
WHERE
  sol.order_id=so.id and
  so.date_order >= '2010-01-01' and so.date_order <= '2011-12-31'
GROUP BY
  week,
  year

但是，列周被排序为char类型，我得到的结果是这样的：

week     |year|sales
1 - 2010 |2010|5
1 - 2011 |2011|10
10 - 2010|2010|15
11 - 2011|2011|12
2  - 2010|2010|7
2  - 2011|2011|13

如果我尝试ORDER BY to_number(week, text(99999999))，我会收到week列不存在的错误。我想因为那是临时专栏？

所以我的问题是，如何按临时周列进行正确排序，还是有更好的方法对其进行排序？：）

Answer 1

我可能会像这样重写你的查询：

SELECT
  MIN(id) AS id,
  week || ' ' || year AS week,
  year,
  COUNT(*) AS amount,
  SUM(price_unit) AS price
FROM (
  SELECT
    so.id,
    DATE_PART('week',  so.date_order) AS week,
    DATE_PART('year',  so.date_order) AS year,
    sol.price_unit
  FROM sale_order AS so
    INNER JOIN sale_order_line AS sol ON sol.order_id = so.id
  WHERE so.date_order BETWEEN '2010-01-01' AND '2011-12-31'
) AS x
GROUP BY
  year,
  week
ORDER BY
  week::int,
  year

将char类型的临时列命令为整数

1 个答案: