查找SQL Server 自增ID值不连续记录

在很多的时候，我们会在数据库的表中设置一个字段：ID，这个ID是一个IDENTITY，也就是说这是一个自增ID。当并发量很大并且这个字段不是主键的时候，就有可能会让这个值重复；或者在某些情况（例如插入数据的时候出错，或者是用户使用了Delete删除了记录）下会让ID值不是连续的，比如1,2,3,5,6,7,10，那么在中间就断了几个数据，那么我们希望能在数据中找出这些相关的记录，我希望找出的记录是3,5,7,10，通过这些记录可以查看这些记录的规律来分析或者统计；又或者我需要知道那些ID值是没有的：4,8,9。

解决办法的核心思想是： 获取到当前记录的下一条记录的ID值，再判断这两个ID值是否差值为1，如果不为1那就表示数据不连续了

执行下面的语句生成测试表和测试记录

--生成测试数据
if exists (select * from sysobjects where id = OBJECT_ID('[t_IDNotContinuous]') and OBJECTPROPERTY(id, 'IsUserTable') = 1) 
DROP TABLE [t_IDNotContinuous]

CREATE TABLE [t_IDNotContinuous] (
[ID] [int]  IDENTITY (1, 1)  NOT NULL,
[ValuesString] [nchar]  (10) NULL)

SET IDENTITY_INSERT [t_IDNotContinuous] ON

INSERT [t_IDNotContinuous] ([ID],[ValuesString]) VALUES ( 1,'test')
INSERT [t_IDNotContinuous] ([ID],[ValuesString]) VALUES ( 2,'test')
INSERT [t_IDNotContinuous] ([ID],[ValuesString]) VALUES ( 3,'test')
INSERT [t_IDNotContinuous] ([ID],[ValuesString]) VALUES ( 5,'test')
INSERT [t_IDNotContinuous] ([ID],[ValuesString]) VALUES ( 6,'test')
INSERT [t_IDNotContinuous] ([ID],[ValuesString]) VALUES ( 7,'test')
INSERT [t_IDNotContinuous] ([ID],[ValuesString]) VALUES ( 10,'test')

SET IDENTITY_INSERT [t_IDNotContinuous] OFF

select * from [t_IDNotContinuous]

（图1：测试表）

--拿到当前记录的下一个记录进行连接
select ID,new_ID
into [t_IDNotContinuous_temp]
from (
select ID,new_ID = (
select top 1 ID from [t_IDNotContinuous]
where ID=(select min(ID) from [t_IDNotContinuous] where ID>a.ID)
)
from [t_IDNotContinuous] as a
) as b

select * from [t_IDNotContinuous_temp]

(图2：错位记录)

--不连续的前前后后记录
select * 
from [t_IDNotContinuous_temp]
where ID <> new_ID - 1


--查询原始记录
select a.* from [t_IDNotContinuous] as a
inner join (select * 
from [t_IDNotContinuous_temp]
where ID <> new_ID - 1) as b
on a.ID >= b.ID and a.ID <=b.new_ID
order by a.ID

（图3：效果）

补充1：如果这个ID字段不是主键，那么就会有ID值重复的情况（有可能是一些误操作，之前就有遇到过）那么就需要top 1来处理。但是当前这种情况下可以使用下面的简化语句

select a.id as oid, nid = 
(select min(id) from t_IDNotContinuous b where b.id > a.id) 
from t_IDNotContinuous a

补充2：缺失ID值列表，

--方法一：找出上一条记录+1，再比较大小
select (select max(id)+1 
from [t_IDNotContinuous] 
where id<a.id) as beginId,
(id-1) as endId
from [t_IDNotContinuous] a
where
a.id>(select max(id)+1 from [t_IDNotContinuous] where id<a.id)

（图4：效果）

--方法二：全部+1，再判断在原来记录中找不到
select beginId,
(select min(id)-1 from [t_IDNotContinuous] where id > beginId) as endId 
from (  
select id+1 as beginId from [t_IDNotContinuous]
where id+1 not in 
(select id from [t_IDNotContinuous]) 
and id < (select max(id) from [t_IDNotContinuous])  
) as t