Question

请参阅下面的DDL：

create table #address (ID int IDENTITY, housenumber varchar(30), street varchar(30), town varchar(30), county varchar(30), postcode varchar(30), primary key (id))
insert into  #address (housenumber,street,town,county,postcode) values ('1', 'The Street', 'Lincoln', null, 'LN21AA')
insert into  #address (housenumber,street,town,county,postcode) values ('1', 'The Street', 'Lincoln', null, 'LN21AA')
insert into  #address (housenumber,street,town,county,postcode) values ('1', 'The Street', 'Lincoln', 'Lincolnshire', 'LN21AA')

和下面的SQL：

   select #address .id as masterid, address2.id as childid  from #address inner join #address as address2 on 
    #address.housenumber=address2.housenumber and #address.street=address2.street
    and #address.town=address2.town
    and #address.county=address2.county
    and #address.postcode=address2.postcode
where #address.id<address2.id

我正在尝试识别重复项。

＆＃39;县＆＃39;有时为null，其他为null。上面的查询不返回任何行。

我试过这个命令：

set ansi_nulls off

然而，它没有任何区别。我意识到我可以这样做：

select #address .id as masterid, address2.id as childid  from #address inner join #address as address2 on 
#address.housenumber=address2.housenumber and #address.street=address2.street
and #address.town=address2.town
and ((#address.county=address2.county) or (#address.county is null and address2.county is null))
and #address.postcode=address2.postcode

但是，我很想知道为什么将ansi nulls设置为off可以让你这样做：

select * from #address where county=null

返回两行。但是，当ANSI NULL关闭时，我的第一个查询不返回任何行。为什么ANSI NULLS对ON子句没有影响。

我花了20分钟谷歌搜索，但我没有找到答案。其中＃address.id

Answer 1

您可以使用group by 识别重复项。当有两个值时，以下内容返回id：

select housenumber, street, town, country postcode, count(*) as cnt,
       min(a.id) as masterid, max(a.id) as childid 
from #address a
group by housenumber, street, town, country postcode
having count(*) >= 2;

获取给定地址的所有ID需要额外的连接或时髦的字符串聚合。

ANSI NULLS和ON子句

1 个答案: