Question

我有两个表 - Test，Child包含版本化数据。关于列的说明：

TEST, CHILD Table
ID - Record id (not unique due to multiple versions)
MODSTAMP - Timestamp at which record is inserted
DELETEDDATE - Timestamp at which record is deleted; NULL value means record is not deleted yet.
COMPOSITE KEY = {ID, MODSTAMP}

CHILD Table
DATA - Foreign key that references Test(ID)
DATA2 - Foreign key that references Test(ID)

我正在研究在两个时间戳之间选择记录的要求 - $ FROMTIME和$ TOTIME。我可以通过运行自联接来选择Test表上的记录。

SELECT v.id, v.modstamp FROM test v
INNER JOIN
    (SELECT Id, MAX(modstamp) AS MaxDateTime
    FROM test where modstamp >= '2017-08-16 15:08:04 +00:00' and modstamp <= '2017-08-16 17:08:04 +00:00' GROUP BY Id) g 
ON v.Id = g.Id 
AND v.modstamp = g.MaxDateTime where v.deleteddate is null

我还需要选择相应子记录的最新版本（没有任何重复的记录ID，因此选择group by子句），其中选择了父记录。我的应用程序保存上述查询的结果，并运行以下查询为每个值['aaa'，'bbb']选择有效的子记录：

 SELECT v.* FROM child v
    INNER JOIN
        (SELECT id, MAX(modstamp) AS MaxDateTime
        FROM child where (data = ? or data is null) and modstamp <= '2017-08-16 17:08:04 +00:00' GROUP BY id
            UNION
        SELECT id, MAX(modstamp) AS MaxDateTime
        FROM child where (data2 = ? or data2 is null) and modstamp <= '2017-08-16 17:08:04 +00:00' GROUP BY id
        ) g 
    ON v.id = g.id 
    AND v.modstamp = g.MaxDateTime

有人可以建议如何优化此解决方案吗？使用当前方法，如果选择了n个父记录，则将有n个内部联接将运行。

这是用于连接的查询，但会抛出“多部分标识符未找到错误”。还有其他任何方法可以重写查询吗？ - 预期查询

SELECT v.* FROM child v
    INNER JOIN (select distinct(id) from DATA) D
    ON v.id = D.id
    INNER JOIN
        (SELECT id, MAX(modstamp) AS MaxDateTime
        FROM child where (data = d.id or data is null) and modstamp <= '2017-08-16 17:08:04 +00:00' GROUP BY id
            UNION
        SELECT id, MAX(modstamp) AS MaxDateTime
        FROM child where (data2 = d.id or data2 is null) and modstamp <= '2017-08-16 17:08:04 +00:00' GROUP BY id
        ) g 
    ON v.id = g.id 
    AND v.modstamp = g.MaxDateTime;

运行脚本的示例信息：

Test
| ID | MODSTAMP | DELETEDDATE |

Child
| ID | DATA | DATA2 | MODSTAMP | DELETEDDATE |

脚本（使用SQL Server）：

drop table test;
drop table child;

CREATE TABLE TEST(ID VARCHAR(20), modstamp DATETIMEOFFSET, deleteddate DATETIMEOFFSET);
insert into test values('aaa', '2017-08-16 15:08:04 +00:00', null);
insert into test values('aaa', '2017-08-16 16:08:04 +00:00', null);
insert into test values('aaa', '2017-08-16 17:08:04 +00:00', null);
insert into test values('aaa', '2017-08-16 18:08:04 +00:00', '2017-08-16 18:08:04 +00:00');
insert into test values('bbb', '2017-08-16 17:08:04 +00:00', null);

CREATE TABLE CHILD(ID VARCHAR(20), DATA VARCHAR(10), DATA2 VARCHAR(10), modstamp DATETIMEOFFSET, deleteddate DATETIMEOFFSET);
insert into CHILD values('1', 'aaa', null, '2017-08-16 15:08:04 +00:00', null);
insert into CHILD values('1', null, 'bbb', '2017-08-16 16:08:04 +00:00', null);
insert into CHILD values('1', null, null, '2017-08-16 17:08:04 +00:00', null);
insert into CHILD values('2', 'aaa', null, '2017-08-16 15:08:04 +00:00', null);
insert into CHILD values('3', null, null, '2017-08-16 15:08:04 +00:00', null);

Answer 1

您的查询看起来不错。我不会调用以下优化的查询，但另一种方法是使用ROW_NUMBER()。

SELECT * FROM (
SELECT *, rownum = ROW_NUMBER() OVER (PARTITION BY id ORDER BY modstamp)
FROM child
WHERE (data = ? OR data IS NULL OR data2 = ? OR data2 IS NULL) AND modstamp <= '2017-08-16 17:08:04 +00:00'
) AS a where rownum = 1

SQL优化内连接（带联合的内连接和子查询的可能使用）

1 个答案: