SQL：仅检索其值与第一行发生变化的记录

Question

对不起，新手。

此处与此类似：SQL: retrieve only the records whose value has changed

我想知道如何也包括第一行？带 * 的行？

ID  description    eaudittimestamp
--  -----------    -----------------------
777012  above       2017-03-27 10:09:59.330 *
777012  above       2017-03-27 10:09:58.550
777012  below       2017-03-27 10:26:03.560 *
777012  below       2017-03-27 10:36:02.423
777012  below       2017-03-27 10:37:15.250
777012  middle      2017-03-27 10:49:11.680 *
777012  middle      2017-03-27 10:49:18.870
777013      something       2017-03-27 11:49:18.870
777013      something       2017-03-27 12:49:18.870
777014      nodescription   2017-03-27 12:49:18.870

抱歉，我是新手。让我改述一下。我有一个表，其中有 ID、描述和时间戳。类似于审计表。如何找到描述已更改的行（包括第一行）。

以及如何排除只有 1 行的那些？即 777013 和 777014？

Answer 1

with ct as
(
    select Code, Date, Rate,
           row_number() over (partition by Code, Rate order by Code, Date) as rn
    from your_table
)
select Code, Date, Rate
from   ct
where  rn = 1
order by Code, Date;

更新

declare @foo table(ID int,  description varchar(100),  eaudittimestamp datetime);
insert into @foo values
(777012, 'above', '2017-03-27 10:09:59.330'),
(777012, 'above', '2017-03-27 10:09:58.550'),
(777012, 'below', '2017-03-27 10:26:03.560'),
(777012, 'below', '2017-03-27 10:36:02.423'),
(777012, 'below', '2017-03-27 10:37:15.250'),
(777012, 'middle', '2017-03-27 10:49:11.680'),
(777012, 'middle', '2017-03-27 10:49:18.870'),
(777013, 'something', '2017-03-27 11:49:18.870'),
(777013, 'something', '2017-03-27 12:49:18.870'),
(777014, 'nodescription', '2017-03-27 12:49:18.870');

with ct as
(
    select ID, description, eaudittimestamp,
           row_number() over (partition by ID, description order by eaudittimestamp) rn,
           count(*) over (partition by ID) as ct
    from   @foo
)
select ID, Description, eaudittimestamp
from ct
where rn = 1 and ct > 1;

GO

身份证 |说明 |审计时间戳 -----: | :------------ | :----------------- 777012 |以上 | 27/03/2017 10:09:58 777012 |下面 | 27/03/2017 10:26:03 777012 |中间 | 27/03/2017 10:49:11 777013 |东西| 27/03/2017 11:49:18

dbfiddle here

Answer 2

对于您提供的数据，row_number() 有效：

select t.*
from (select t.*,
             row_number() over (partition by id, description order by eaudittimestamp) as seqnum
      from t
     ) t
where seqnum = 1;

这会为每个 id/description 对返回一行。 eaudittimestamp 最低的那个。这似乎是你想要的。

编辑：

如果要排除只有一行的组，只需使用另一个窗口函数：

select t.*
from (select t.*,
             row_number() over (partition by id, description order by eaudittimestamp) as seqnum,
             min(description) over (partition by id) as min_d,
             max(description) over (partition by id) as max_d
      from t
     ) t
where seqnum = 1 and min_d <> max_d;