TSQL多表连接筛选视图，尝试根据筛选结果（而不是一个）返回多个列

Question

我在过滤值模式中有四个表，我无法确定如何扩展现有视图查询以包含基于过滤结果的附加列。

一个表保存默认项目记录，我在下面的示例中称这些ItemKeys。此表还包含每个项目的默认值和默认描述。在示例中，这是dbo.Defaults表。

下一个表是一个表，用于为后续过滤表对这些项的集合进行分组。在示例中称为dbo.FilterGroup。

最后两个表是过滤器，它们指向FilterGroup和Defaults以及它们自己的值和描述。这些表仅包含Items的记录，当使用FilterGroup时，它们将按顺序进行筛选：Filter2> Filter1> Default。正如您将在示例中看到的，这意味着如果Filter1具有项目记录，它将覆盖默认值，并且Filter2将覆盖Filter1（如果存在）。

The problem

当前视图仅寻求恢复过滤器数据的Value部分，并且我必须想出一种方法来恢复生成的过滤项的Value和Description。

让我向您展示我应该更清楚地表明情况的例子：

表和数据设置：

--dbo.Defaults
IF EXISTS (SELECT 1 FROM INFORMATION_SCHEMA.TABLES WHERE TABLE_TYPE='BASE TABLE' 
    AND TABLE_NAME='Defaults') drop table dbo.Defaults; 
CREATE TABLE dbo.Defaults (
    ItemKey varchar(128),
    DefaultValue varchar(128),
    DefaultDesc varchar(512),    
) ON [PRIMARY]
GO
insert into dbo.Defaults VALUES ('Fruit1','Apple','system default value')
insert into dbo.Defaults VALUES ('Fruit2','Orange','system default value')
insert into dbo.Defaults VALUES ('Fruit3','Pear',NULL)
insert into dbo.Defaults VALUES ('Fruit4','Tomato','system default value')

--dbo.Filter group
IF EXISTS (SELECT 1 FROM INFORMATION_SCHEMA.TABLES WHERE TABLE_TYPE='BASE TABLE' 
    AND TABLE_NAME='FilterGroup') drop table dbo.FilterGroup; 
CREATE TABLE dbo.FilterGroup (
    ID integer NOT NULL
        IDENTITY (1,1) CONSTRAINT [PK_ParamFilterSet] PRIMARY KEY CLUSTERED ([ID] ASC) WITH (IGNORE_DUP_KEY = OFF),
    GroupName varchar(128)
) ON [PRIMARY]
GO
insert into FilterGroup values ('Group1')

--dbo.Filter1
IF EXISTS (SELECT 1 FROM INFORMATION_SCHEMA.TABLES WHERE TABLE_TYPE='BASE TABLE' 
    AND TABLE_NAME='Filter1') drop table dbo.Filter1; 
CREATE TABLE dbo.Filter1 (
    ItemKey varchar(128), --Defaults.ItemKey
    GroupID int, --FilterGroup.ID
    FilterValue varchar(128),
    FilterDesc varchar(512),     
) ON [PRIMARY]
GO
insert into dbo.Filter1 VALUES ('Fruit2',1,'Seedless Orange',NULL)

--dbo.Filter2
IF EXISTS (SELECT 1 FROM INFORMATION_SCHEMA.TABLES WHERE TABLE_TYPE='BASE TABLE' 
    AND TABLE_NAME='Filter2') drop table dbo.Filter2; 
CREATE TABLE dbo.Filter2 (
    ItemKey varchar(128), --Defaults.ItemKey
    GroupID int, --FilterGroup.ID
    FilterValue varchar(128),
    FilterDesc varchar(512),     
) ON [PRIMARY]
GO
insert into dbo.Filter1 VALUES ('Fruit3',1,'Pyrus Pear','filter2 value')

我的查询：

SELECT fg.GroupName as [FilterGroup]
    ,d.ItemKey as [ItemKey]
    ,coalesce(f2.FilterValue, f1.FilterValue ,d.DefaultValue) as [Value]
FROM dbo.Defaults d cross join dbo.FilterGroup fg
                    left outer join dbo.Filter1 f1 on f1.ItemKey = d.ItemKey
                                                    and f1.GroupID = fg.ID
                    left outer join dbo.Filter2 f2 on f2.ItemKey = d.ItemKey  
                                                    and f1.GroupID = fg.ID

如上所述，现有查询正确过滤，将Filter2值恢复为Filter1值，Filter1值超过默认值，如果给定项目上没有过滤器，则返回默认值。

我正在努力弄清楚的是我如何根据select coalesce set中的哪个过滤器来包含f1.FilterDesc，f2.FilterDesc或d.DefaultDesc。如何执行此连接并从coalesce中出现的表中提取“value”和“desc”列？

我非常友好地感谢您提供任何帮助！

Answer 1

一个简单的方法是：

SELECT fg.GroupName as [FilterGroup],
       d.ItemKey as [ItemKey],
       coalesce(f2.FilterValue, f1.FilterValue, d.DefaultValue) as [Value],
       (case when f2.FilterValue is not null then 'f2'
             when f1.FilterValue is not null then 'f1'
             else 'default'
        end) as which
FROM dbo.Defaults d cross join
     dbo.FilterGroup fg left outer join
     dbo.Filter1 f1
     on f1.ItemKey = d.ItemKey and f1.GroupID = fg.ID left outer join
     dbo.Filter2 f2
     on f2.ItemKey = d.ItemKey and f1.GroupID = fg.ID;

一种更聪明的方式将逻辑放入from：

SELECT fg.GroupName as [FilterGroup],
       d.ItemKey as [ItemKey],
       coalesce(f2.FilterValue, f1.FilterValue, d.DefaultValue) as [Value],
       coalesce(f2.which, f1.which, 'default') as which
FROM dbo.Defaults d cross join
     dbo.FilterGroup fg left outer join
     (SELECT f1.*, 'f1' as which
      FROM dbo.Filter1 f1
     ) f1
     on f1.ItemKey = d.ItemKey and f1.GroupID = fg.ID left outer join
     (SELECT f2.*, 'f2' as which
      FROM dbo.Filter2 f2
     ) f2
     on f2.ItemKey = d.ItemKey and f1.GroupID = fg.ID;

最后一个条件应该是f1.GroupId = fg.ID还是f2.GroupID = fg.ID？

Answer 2

除了戈登的答案排名方法：

;with allFilters as (
  select f.ItemKey,
         f.GroupID,
         f.FilterValue,
         f.FilterDesc ,
         2 rank --<<<
  from dbo.Filter1 f

  union all

  select f.ItemKey,
         f.GroupID,
         f.FilterValue,
         f.FilterDesc ,
         1 rank --<<<
  from dbo.Filter2 f

  union all

  select f.ItemKey,
         NULL GroupID,
         f.DefaultValue,
         f.DefaultDesc ,
         3 rank --<<<
  from dbo.Defaults f
),
filtersRanked as
(
  select f.ItemKey,
         f.GroupID,
         f.FilterValue,
         f.FilterDesc ,
         -- now computing most preferred per ItemKey
         ROW_NUMBER() OVER(PARTITION BY f.ItemKey ORDER BY f.rank ASC) rn
  from allFilters f
)

SELECT fg.GroupName as [FilterGroup],
    f.ItemKey as [ItemKey],
    f.FilterValue,
    f.FilterDesc
FROM dbo.FilterGroup fg
left join filtersRanked f
  on (f.GroupID = fg.ID or f.GroupID is NULL) -- NULL support for defaults
    and f.rn = 1 --<<< "top 1" by rank

易于操纵哪个源具有更高的优先级。

输出：

FilterGroup ItemKey FilterValue FilterDesc
Group1  Fruit1  Apple           system default value
Group1  Fruit2  Seedless Orange (null)
Group1  Fruit3  Pyrus Pear      filter2 value
Group1  Fruit4  Tomato          system default value

http://sqlfiddle.com/#!18/1511d/1

你也有一些错别字：

• 'Pyrus Pear'我想应该插入Filter2
• 在f1.GroupID的连接谓词中提到Filter2而不是f2.GroupID