如何仅选择 SQL Server 中每个外键的第一条记录？

Question

我在联系人详细信息数据库中有如下所示的行：

contactID - bioID - AddressLine1 - City 
393         1       1 nowhere st   toronto
3921        1       2 somewhere st vancouver
3231        2       1 anywhere rd  barrie
1122        2       2 overthere st halifax      

我目前正在将其加入到包含 firstname, lastname 等列的生物表中，结果如下所示：

bioid    firstname    lastname    addressline1    city
1        some         guy         1 nowhere st    toronto
1        some         guy         2 somewhere st  vancouver
2        that         girl        1 anywhere rd   barrie
2        that         girl        2 overthere st  halifax

所以我基本上为每个简历获得 2 行。无论如何我可以像这样选择这一切作为一行：

bioid    firstname    lastname   addressline1x1   cityx1    addressline1x2    cityx2
1        some         guy        1 nowhere st     toronto   2 somewhere st    vancouver
2        that         girl       1 anywhere rd    barrie    2 overthere st    halifax

感谢任何帮助。

谢谢， 托马斯

编辑：

非常感谢丹尼斯和贾斯汀，我能够解决这个问题。但是，现在我有另一个。

我想实际选择地址作为 1 个字段，例如：

bioid    firstname    lastname    primary address         secondary address
1        some         guy         1 nowhere st, toronto   2 somewhere st, vanvouver
2        that         girl        1 anywhere rd, barrie   2 overthere st, halifax

我知道我可以通过连接如下列来做到这一点：

cd1.addressline1 + ', ' cd1.city AS 'Primary Address' 

但是，记录中的某些字段是空的，例如，如果没有辅助地址 - 那么我怎样才能使辅助地址不输出'，'？这些字段是空字符串，而不是 NULL 值。

再次感谢！

编辑：我有这个工作使用：

STUFF(COALESCE(', ' + NULLIF(C1.[AddressLine1], ''), '') + 
  COALESCE(', ' + NULLIF(C1.[AddressLine2], ''), '') + 
  COALESCE(', ' + NULLIF(C1.[City], ''), '') + 
  COALESCE(', ' + NULLIF(C1.[State], ''), '') + 
  COALESCE(', ' + NULLIF(C1.[Country], ''), '') + 
  COALESCE(', ' + NULLIF(C1.[ZipCode], ''), ''),1, 1, '') AS 'Primary Address'

Answer 1

;with cd as(
select *, rn = dense_rank(partition by bioid order by addressline, city)
from ContactDetails cd
)
select c.bioid, c.firstname, c.lastname, cd1.AddressLine, cd1.City, cd2.AddressLine, cd2.City, cd3.AddressLine, cd3.City
from Contact c 
left join cd cd1 on c.bioid = cd1.bioid and cd1.rn = 1
left join cd cd2 on c.bioid = cd2.bioid and cd2.rn = 2
left join cd cd3 on c.bioid = cd3.bioid and cd3.rn = 3

Answer 2

您可以尝试PIVOT TABLE 或RECURSIVE CTE。如果您知道用户将拥有多少个地址，则数据透视表将起作用（但是，如果您知道您也可以对其进行硬编码:)），CTE 将适用于未知数量的地址，但是，它不会创建多个列，而不是一个逗号分隔的列表。

无论采用哪种方式，您都可能需要使用ROW_NUMBER 来创建唯一性顺序。

这应该是递归 CTE 的一个工作示例：

SELECT *, ROW_NUMBER() OVER(PARTITION BY bioid ORDER BY contactid) AS rownum
INTO #TableToRecurse
FROMContactDetails

;WITH FinalOutput AS
(
--Anchor row of recursion
SELECT bioid, firstname, lastname, addressline, city, rownum
FROM #TableToRecurse 
WHERE rownum = 1
UNION ALL 
--Recursion piece
SELECT tr.bioid, tr.firstname, tr.lastname, 
    FinalOutput.addressline + ', ' + tr.addressline, FinalOutput.city + ', ' + tr.city, tr.rownum
FROM #TableToRecurse AS tr
    JOIN FinalOutput
        ON  FinalOutput.bioid = tr.bioid AND FinalOutput.rownum = tr.rownum +1
)
--Final output only showing the last row (Max)
--Which should have everything concatenated together
SELECT FinalOutput.bioid, FinalOutput.firstname, FinalOutput.lastname, 
    FinalOutput.addressline, FinalOutput.city
FROM FinalOutput 
    JOIN 
    (
        SELECT MAX(rownum) AS MaxRowNum, bioid
        FROM FinalOutput
        GROUP BY bioid
    ) AS MaxForEach
        ON FinalOutput.bioid = MaxForEach.bioid 
            AND FinalOutput.rownum = MaxForEach.MaxRowNum

Answer 3

 SELECT *
    FROM   Table_1 t 

    SELECT *
FROM   Table_1 t 

SELECT af.bioid,
       af.fname,
       af.lastname,
       address1,
       city1,
       address2,
       tbl2,
       rrr
FROM   (
           SELECT tbl1.bioid,
                  tbl1.fname,
                  tbl1.lastname,
                  tbl1.addressline1 AS address1,
                  tbl1.city AS city1,
                  tbl2.addressline1 AS address2,
                  tbl2.city AS tbl2,
                  ROW_NUMBER() OVER(PARTITION BY tbl1.bioid ORDER BY tbl1.bioid) AS 
                  rrr
           FROM   Table_1 tbl1,
                  Table_1 tbl2
           WHERE  tbl1.bioid = tbl2.bioid
                    AND tbl1.addressline1+tbl1.city <> tbl2.addressline1+tbl2.city
       ) af
WHERE  af.rrr = 1