0

我有查询，但现在我需要将查询更改为数据库，以便只为表中的每个公司获取一个结果Price。

为此，我添加GROUP by p.id_firm了行，然后得到下一个查询：

SELECT TOP 20 
  p.id_price as p_id_price, 
  p.id_service as p_id_service, 
  p.name as p_name, 
  p.name_original as p_name_original, 
  p.id_producer_country as p_id_producer_country, 
  p.id_firm as p_id_firm, 
  f.name as f_name, 
  f.address as f_address, 
  f.phone as f_phone, 
  city.name as city_name, 
  pc.name as pc_name 
FROM Price p 
left join Firm f 
  on f.id_service=p.id_service 
  AND f.id_city=p.id_city 
  AND f.id_firm=p.id_firm 
left join City city 
  on city.id_city = p.id_city 
left join Producer_country pc 
  on pc.id_producer_country = p.id_producer_country 
WHERE p.id_city='73041' 
  AND p.include='1' 
  AND p.blocked='0' 
  AND f.blocked='0' 
  AND ( f.name LIKE 'Окно%' COLLATE SQL_Latin1_General_Cp1251_CI_AS OR p.name LIKE 'Окно%' COLLATE SQL_Latin1_General_Cp1251_CI_AS ) 
GROUP by p.id_firm ORDER BY p.name ASC

但是如果我使用它，我会得到错误：

消息 8120，级别 16，状态 1，第 2 行列“Price.id_price”在选择列表中无效，因为它既不包含在聚合函数中，也不包含在 GROUP BY 子句中。

请告诉我更改此查询或进行其他查询的正确方法？

结构化您可以在此处看到的所有表格

PS：抱歉不准确。按公司名称和产品名称搜索。如果搜索词包含在产品名称或公司名称中，并且有一些结果具有相同的 id_firm，则需要选择一个最合适的值，即结果搜索中的 id_firm 应该是唯一的。

4

2 回答 2

1

使用ROW_NUMBER()而不是分组。

分别编号每个公司的行。对 的行进行排序时ROW_NUMBER()，使用p.name LIKE 'Окно%'条件将具有匹配产品的行放在其他行之前。

这是一个示例：

;
WITH ranked AS (
  SELECT
    p.id_price as p_id_price, 
    p.id_service as p_id_service, 
    p.name as p_name, 
    p.name_original as p_name_original, 
    p.id_producer_country as p_id_producer_country, 
    p.id_firm as p_id_firm, 
    f.name as f_name, 
    f.address as f_address, 
    f.phone as f_phone, 
    city.name as city_name, 
    pc.name as pc_name,
    ROW_NUMBER() OVER (
      PARTITION BY p.id_firm
      ORDER BY
        CASE  -- this criterion puts matching products before non-matching ones
          WHEN p.name LIKE 'Окно%' COLLATE SQL_Latin1_General_Cp1251_CI_AS
          THEN 1 ELSE 2
        END,
        p.id_price  -- you may use any sorting criteria at this point,
                    -- just ensure it makes the results predictable
    ) AS rnk
  FROM Price p 
  left join Firm f 
    on f.id_service=p.id_service 
    AND f.id_city=p.id_city 
    AND f.id_firm=p.id_firm 
  left join City city 
    on city.id_city = p.id_city 
  left join Producer_country pc 
    on pc.id_producer_country = p.id_producer_country 
  WHERE p.id_city='73041' 
    AND p.include='1' 
    AND p.blocked='0' 
    AND f.blocked='0' 
    AND ( f.name LIKE 'Окно%' COLLATE SQL_Latin1_General_Cp1251_CI_AS
       OR p.name LIKE 'Окно%' COLLATE SQL_Latin1_General_Cp1251_CI_AS ) 
)
SELECT TOP 20
  p_id_price, 
  p_id_service, 
  p_name, 
  p_name_original, 
  p_id_producer_country, 
  p_id_firm, 
  f_name, 
  f_address, 
  f_phone, 
  city_name, 
  pc_name
FROM ranked
WHERE rnk = 1
-- the absence of ORDER BY makes your TOP 20 results indeterminate
;

基本上，这对每个公司的行进行排名，然后仅从1所有公司中提取具有排名的行，最终使每个公司一行。

于 2013-03-01T23:11:25.903 回答
0

您可以尝试选择 Distinct 以获得可能的快速修复 http://www.w3schools.com/sql/sql_distinct.asp

于 2013-03-02T01:30:40.830 回答