合并 SQL 查询以获得相似的结果

Question

我现在有一个系统来存储与每个产品相关的产品和标签。

例子

产品：麦克风

标签：音乐、电子产品、音频

有 Tag 表、Product 表和 TagProductMapping 表。第三个表显然将产品映射到标签以实现一对多关系。当我使用 LEFT JOIN 查询麦克风产品时，我得到 3 条几乎重复的记录，除了“TagName”列显然它具有所有 3 个不同的标签。我怎样才能合并这个结果？令人沮丧的是，如果我尝试准确查询 10 个产品，它只会限制为 10 个结果，而实际上并不是 10 个产品。

有人对此有好的想法吗？

编辑：

这是我查询的结果，请注意 3 个 JobId 之间唯一不同的是 CategoryName，它们是标签。

这是我的桌子的样子

-标签映射表：

-标签表

-“产品表”（在这种情况下，它是我的工作表）

这是我的存储过程：

ALTER PROCEDURE [dbo].[JobPostingSelectAll]                       
(                      
    @StartRowIndex     INT,                      
    @PageSize          INT,                      
    @OrderBy           VARCHAR(50),                      
    @OrderByDirection  VARCHAR(4),                      
    @CurrentUserId     INT,            
    @CategoryId     INT                      
)                      
AS                      
 SET NOCOUNT ON                                            



 SELECT         
  JobId,        
  Title,                      
  Answers,          
  UserId,                  
        UserName,        
        ProfileImageName,        
        CategoryId,        
        CategoryName,        
        Fees,            
        DESCRIPTION,                
        DateCreated,      
        UniqueTitle,    
        IsSecured                                
 FROM   (                      
            SELECT J.JobId,        
       J.Title,        
       (SELECT COUNT(ja2.JobId) FROM JobApplication ja2 left join Deliverable d2 ON d2.DeliverableId = ja2.DeliverableId
       WHERE ja2.JobId=j.JobId and (d2.PurchaseCount>0 OR d2.IsFrozen=0)) AS Answers,            
                   J.UserId,        
                   U.UserName,        
                   U.ImageName as ProfileImageName,        
                   J.CategoryId,        
                   C.CategoryName,        
                   J.Fees,        
                   J.Description,                                              
                   J.DateCreated,      
                   J.UniqueTitle,    
                   J.IsSecured,                                                
                   ROW_NUMBER() OVER(                      
                       ORDER BY                  
                       CASE                       
                            WHEN @OrderByDirection = 'asc' AND @OrderBy = 'Answers'             
                            THEN (SELECT COUNT(ja2.JobId) FROM JobApplication ja2 left join Deliverable d2 ON d2.DeliverableId = ja2.DeliverableId
       WHERE ja2.JobId=j.JobId and (d2.PurchaseCount>0 OR d2.IsFrozen=0)) END ASC,            
                       CASE                       
                            WHEN @OrderByDirection = 'asc' AND @OrderBy = 'Answers'             
                            THEN J.DateCreated END DESC,            
                       CASE                       
                            WHEN @OrderByDirection = 'asc' AND @OrderBy = 'Answers'             
                            THEN J.Title END ASC,            
                       CASE                       
                            WHEN @OrderByDirection = 'desc' AND @OrderBy = 'Answers'             
                            THEN (SELECT COUNT(ja2.JobId) FROM JobApplication ja2 left join Deliverable d2 ON d2.DeliverableId = ja2.DeliverableId
       WHERE ja2.JobId=j.JobId and (d2.PurchaseCount>0 OR d2.IsFrozen=0)) END DESC,                                             
      CASE                       
                            WHEN @OrderByDirection = 'desc' AND @OrderBy = 'Answers'             
                            THEN J.DateCreated END DESC,            
                         CASE                       
                            WHEN @OrderByDirection = 'desc' AND @OrderBy = 'Answers'             
                            THEN J.Title END ASC,            
                       CASE WHEN @OrderByDirection = 'asc' AND @OrderBy = 'Fees'             
       THEN J.Fees END ASC,            
       CASE WHEN @OrderByDirection = 'asc' AND @OrderBy = 'Fees'             
       THEN J.DateCreated END DESC,            
        CASE WHEN @OrderByDirection = 'desc' AND @OrderBy = 'Fees'             
       THEN J.Fees END DESC,            
                       CASE WHEN @OrderByDirection = 'desc' AND @OrderBy = 'Fees'             
       THEN J.DateCreated END DESC,            
                       CASE WHEN @OrderByDirection = 'asc' AND @OrderBy = 'DateCreated'             
       THEN J.DateCreated END ASC,                      
                       CASE WHEN @OrderByDirection = 'desc' AND @OrderBy = 'DateCreated'             
       THEN J.DateCreated END DESC            
                   ) AS RowIndex           
            FROM   [JobPosting] J    
                LEFT JOIN TagMapping TM ON J.JobId = TM.QuestionId      
                        LEFT JOIN Categories C ON TM.TagId = C.CategoryID            
            Left Join [User] U ON J.UserId = U.UserID            
            WHERE  J.IsLocked = 0  AND j.IsDeleted = 0                                           
       AND (@CategoryId IS NULL OR  J.CategoryId = @CategoryId)            

        ) AS JobPostingList                      
 WHERE  RowIndex BETWEEN @StartRowIndex AND (@StartRowIndex + @PageSize) - 1             


 SELECT COUNT(J.JobID) AS TotalRecords                      
 FROM   JobPosting J                                      
 WHERE  J.IsLocked = 0 AND J.IsDeleted = 0                        
        AND (@CategoryId IS NULL OR  J.CategoryId = @CategoryId)          


-- select all filecount grouped by Type Of File for specific Job                
SELECT J.JobId, F.MimeType, COUNT(F.FileId) AS FileCount                
FROM                   
JobPosting J  
Left Join Files F ON  F.JobPostingId = J.JobId  
WHERE  J.IsLocked = 0 AND J.IsDeleted = 0                        
        AND (@CategoryId IS NULL OR  J.CategoryId = @CategoryId)          
GROUP BY                                
       F.MimeType,J.JobId  
Having COUNT(F.FileId) > 0

Answer 1

您遇到的问题是因为您的数据库结构标准化了每个产品的标签数据（我发现this page 是一个很好的参考）。

当您从 Products 表中 SELECT 和 JOIN 到您的 Tags 表时，请务必记住您没有获得产品列表。相反，您将获得产品标签组合的列表。

如果您想获得前 10 种产品的列表以及它们的标签信息，我建议使用子查询：

select * from 
    (select top 10 * from ProductsTable) TopProducts
    inner join Tagmapping on TopProducts.ProductID = Tagmapping.ProductID
    inner join Tags on Tagmapping.TagID = Tags.TagID

即使这解决了您最初的选择问题，这仍然会产生您在上面显示的那种多重列表，其中只有标签信息因行而异。

可以将输出格式化为具有多个标记条目（可能用逗号分隔）as described here，但这将是您在将数据呈现给用户之前的最后一步，或者通过 SQL或您用作中间人的任何软件层。

Answer 2

我相信如果您对标签名称不感兴趣，您可以选择除标签名称之外的所有列，并使用“选择不同”来仅显示没有标签的产品。不过我可能是错的 =(