2005比2000新增了几个函数，分别是row_number()、rank()、dense_rank()、ntile()，下面以实例分别简单讲解一下。

create table gg(sname varchar(10),sort varchar(10),num int)

go

insert into gg

select '白芍','根茎类',55

union all

select '法半夏','根茎类',78

union all

select '柴胡','根茎类',60

union all

select '川芎','根茎类',99

union all

select '天香炉','草类',68

union all

select '灯心草','草类',55

union all

select '龙葵','草类',60

union all

select '石见穿','草类',60

union all

select '猪笼草','草类',70

union all

select '益母草','草类',86

union all

select '扁豆','果实类',86

union all

select '草果','果实类',70

union all

select '金樱子','果实类',55

union all

select '女贞子','果实类',94

union all

select '胖大海','果实类',66

union all

select '桑葚','果实类',78

select sname,sort,num,

row_number() over(order by num) as rownum,

rank() over(order by num) as ranknum,

dense_rank() over(order by num) as dersenum,

ntile(3) over(order by num) as ntilenum

from gg

--结果

--ROW_NUMBER()是按num由小到大逐一排名，不并列，排名连续

--RANK()是按num由小到大逐一排名，并列，排名不连续

--DENSE_RANK()是按num由小到大逐一排名，并列，排名连续

--NTILE()是按num由小到大分成组逐一排名，并列，排名连续

sname      sort       num       rownum        ranknum       dersenum       ntilenum
-------- --------- --------- ------------- ------------- --------------- ---------------
白芍         根茎类       55          1          1             1               1
灯心草       草类         55          2          1             1               1
金樱子       果实类       55          3          1             1               1
龙葵         草类         60          4          4             2               1
石见穿       草类         60          5          4             2               1
柴胡         根茎类       60          6          4             2               1
胖大海       果实类       66          7          7             3               2
天香炉       草类         68          8          8             4               2
草果         果实类       70          9          9             5               2
猪笼草       草类         70          10         9             5               2
法半夏       根茎类       78          11         11            6               2
桑葚         果实类       78          12         11            6               3
益母草       草类         86          13         13            7               3
扁豆         果实类       86          14         13            7               3
女贞子       果实类       94          15         15            8               3
川芎         根茎类       99          16         16            9               3

(16 行受影响)

select sname,sort,num,

row_number() over(partition by sort order by num) as rownum,

rank() over(partition by sort order by num) as ranknum,

dense_rank() over(partition by sort order by num) as dersenum,

ntile(3) over(partition by sort order by num) as ntilenum

from gg

--结果

此时加了partition by sort，就以类别来分类了，ntile(3)意思就是强制分为三组。

sname      sort           num       rownum        ranknum        dersenum       ntilenum
-------- ----------   --------- ------------- --------------- ---------------- -----------
灯心草       草类          55          1            1               1               1
龙葵         草类          60          2            2               2               1
石见穿       草类          60          3            2               2               2
天香炉       草类          68          4            4               3               2
猪笼草       草类          70          5            5               4               3
益母草       草类          86          6            6               5               3
白芍         根茎类        55          1            1               1               1
柴胡         根茎类        60          2            2               2               1
法半夏       根茎类        78          3            3               3               2
川芎         根茎类        99          4            4               4               3
金樱子       果实类        55          1            1               1               1
胖大海       果实类        66          2            2               2               1
草果         果实类        70          3            3               3               2
桑葚         果实类        78          4            4               4               2
扁豆         果实类        86          5            5               5               3
女贞子       果实类        94          6            6               6               3

(16 行受影响)

下面分别用SQL 2000实现，相对比2005要麻烦的多了。

--ROW_NUMBER在sql 2000中的实现

--利用临时表和IDENTITY（函数）

select sname,num,identity(int,1,1) as rownumber

into #tem

from gg

order by num

select sname,num,rownumber

from #tem

drop table #tem

go

--RANK在sql 2000中的实现

select sname,num,

(select count(1)+1 from gg where num<g.num) as ranknum

from gg g

order by num

go

--DENSE_RANK在sql 2000中的实现

select num,identity(int,1,1) as densenum

into #t

from gg

group by num

order by num

select r.sname,r.num,t.densenum

from gg r join #t t

on r.num=t.num

order by num

drop table #t

go