我想知道我是否可以在SQLite中运行以下过程:
set nocount on
select T.ID, max(T.SerialNo) as SerialNo
into #Tmp_Ticket_ID
from Ticket as T, Ticket as inserted
where t.ID = inserted.ID
group by T.id having count(*) > 1
declare zeiger cursor for
select SerialNo
from #Tmp_Ticket_ID
declare @SerialNo int
OPEN Zeiger
FETCH NEXT FROM zeiger INTO @SerialNo
WHILE (@@fetch_status <> -1)
BEGIN
IF (@@fetch_status <> -2)
BEGIN
update T
set ID = (select max(id) + 1 from Ticket)
from ticket AS T, #Tmp_Ticket_ID as I
where t.serialNo = i.serialno
and I.Serialno = @SerialNo
END
FETCH NEXT FROM zeiger INTO @SerialNo
END
CLOSE Zeiger
DEALLOCATE Zeiger
DROP TABLE #Tmp_Ticket_ID
这是ms-sql2000中的一个小程序,它清除了以下结构的给定表Ticket中的Ticket_id的两倍:
create table Ticket (serialNo int identity(1,1) not null
, ID as int not null
, Ticket_issue as varchar(50)
, some_more_field varchar(500))
由于来自不同数据库的简单合并,ticket_id变得不唯一。为了通过重新编号来修复我开发了这个过程,但现在我们在SQLite-db上遇到了类似的问题。
答案 0 :(得分:4)
delete from Ticket
where exists
(select rowid from Ticket t2
where t2.ID = Ticket.ID and t2.rowid < Ticket.rowid)
rowid
是始终存在的SQLite btree索引列。
感谢Martin Engelschalk在2009-08-17的SQLite邮件列表上。