如何在SQL Server 2014中创建视图以查看缺少某些数据的交叉引用表中的数据？

Question

I am trying to generate a report from a normalized database with tables a, a_xref_b, and b.

CREATE TABLE a (a_rid INT primary key, a1 varchar(20), );

CREATE TABLE b (b_rid INT primary key, b1 varchar(20), b2 varchar(20));

CREATE TABLE a_xref_b (a_rid INT, b_rid INT,xref_type int
CONSTRAINT fk_a FOREIGN KEY (a_rid) REFERENCES a (a_rid),
CONSTRAINT fk_b FOREIGN KEY (b_rid) REFERENCES b (b_rid));

INSERT INTO a VALUES (1,'John'), (2,'Sue')
INSERT INTO b VALUES (1,'Atlanta','GA'), (2,'Macon','GA'), (3,'Opp','AL')
INSERT INTO a_xref_b VALUES (1,1,1), (1,2,2), (2,3,1)

Table A and B are cross-referenced and have the cross reference has a type, which defines what the data in B represents.

Scenario - John is provided a main city and an alternate. Sue only provided a main city.

I am trying to write a view to load page that has the name, main city, and alternate.

Name   b_rid_t1  City_t1   St_1  b_rid_2   City_t2  St_t2
John   1         Atlanta   GA    2         Macon    GA
Sue    3         Opp       AL    NULL      NULL     NULL

I tried this

select
   a.a1,
   b_1.b_rid,
   b_1.b1,
   b_1.b2,
   b_2.b_rid,
   b_2.b1,
   b_2.b2      
from
   a     
left join
   a_xref_b as xf1 
      on a.a_rid = xf1.a_rid      
inner join
   b as b_1 
      on xf1.b_rid = b_1.b_rid 
      and xf1.xref_type = 1     
left join
   a_xref_b xf2 
      on a.a_rid = xf2.a_rid      
inner join
   b as b_2 
      on xf2.b_rid = b_2.b_rid 
      and xf2.xref_type = 2

However it drops Sue's record. If I change it to left joins from the xref to b, then I get duplicate records.

The real world scenario is a contact table cross referenced with an address table and the xref table has the type field. We're trying to build a view to load the page.

Answer 1

Try this, you should add xref_type condition with a_xref_b table join.

select
   a.a1,
   b_1.b_rid,
   b_1.b1,
   b_1.b2,
   b_2.b_rid,
   b_2.b1,
   b_2.b2      
from
   a     
left join a_xref_b as xf1 on a.a_rid = xf1.a_rid      
   and xf1.xref_type = 1     
left join b as b_1 on xf1.b_rid = b_1.b_rid       
left join a_xref_b xf2 on a.a_rid = xf2.a_rid      
   and xf2.xref_type = 2
left join b as b_2 on xf2.b_rid = b_2.b_rid 

Answer 2

You need to use a Pivot table or use a SQL statement that give the same kind of result. For example:

;WITH    Main
          AS ( SELECT   a1 AS Name
                       ,b1 AS City
                       ,b2 AS Code
                       ,xref_type
               FROM     a_xref_b
                        LEFT JOIN a ON a_xref_b.a_rid = a.a_rid
                        LEFT JOIN b ON a_xref_b.b_rid = b.b_rid
             )
    SELECT  Name
           ,MAX(CASE WHEN xref_type = 1 THEN City
                END) FirstCity
           ,MAX(CASE WHEN xref_type = 1 THEN Code
                END) FirstCode
           ,MAX(CASE WHEN xref_type = 2 THEN City
                END) AlternateCity
           ,MAX(CASE WHEN xref_type = 2 THEN Code
                END) AlternateCode
    FROM    Main
    GROUP BY Name

Answer 3

Another approach could be using Union:

select max(name) name, max(id1) id1, max(city1) city1, max(state1) state1, max(id2) id2, max(city2) city2, max(state2) state2
from(
    select a.a1 as name, xf1.b_rid id1, xf1.b1 city1, xf1.b2 state1, null id2, null city2, null state2
    from a left join 
    (select a_rid, b1, b2, b.b_rid from a_xref_b inner join b on a_xref_b.b_rid = b.b_rid 
    where xref_type = 1) xf1 on a.a_rid = xf1.a_rid 
    union
    select a.a1 as name, null id1, null city1, null state1,xf1.b_rid id2, xf1.b1 city2, xf1.b2 state2
    from a left join 
    (select a_rid, b1, b2, b.b_rid from a_xref_b inner join b on a_xref_b.b_rid = b.b_rid 
    where xref_type = 2) xf1 on a.a_rid = xf1.a_rid 
)t group by name