在SQL中选择DISTINCT对COLUMN

Question

Consider

create table pairs ( number a, number b )

Where the data is

1,4
4,1
2,4
2,4
3,2
3,2
2,3
5,1
Etc.

I'm Getting this

1,4
4,1
2,4
3,2
2,3
5,1

What query gives me the distinct values the number column b has So I can see

1,4
5,1
2,4
3,2

only

I don't want the value in column a should be present in column b. Please help.

what i need is to select distinct values on both sides of column. for eg. if (1,2) is present then (2,1) must not be present

Answer 1

You didn't state your DBMS, but this works on many DBMS ( least and greatest aren't part of the SQL standard unfortunately)

select distinct least(a,b), greatest(a,b)
from pairs

Answer 2

select distinct p1.a from pairs
inner join pairs p2 on p1.a=p2.b

Answer 3

You can massage your data and put the smaller value first:

SELECT CASE WHEN a < b THEN a ELSE b END a,
       CASE WHEN a < b THEN b ELSE a END b
FROM pairs

Then you can select distinct from this:

SELECT DISTINCT a, b
FROM
(
    SELECT CASE WHEN a < b THEN a ELSE b END a,
           CASE WHEN a < b THEN b ELSE a END b
    FROM pairs
) x

---

Alternatively if there is no difference in meaning between a and b , you can enforce a constraint that forces a to always be smaller or equal to b , controlling the data on the way in. Then you can just do a simple SELECT DISTINCT .

Answer 4

您可以尝试此查询以获取列a中的所有不同值

SELECT distinct(a) FROM pairs WHERE not exists(select b from pairs where a=b)

Answer 5

Sounds like you need you're own extended version of DISTINCT.

You can do this (in SQL Server)... via a CURSOR.

Load up the cursor from your table. Loop through the cursor, pattern match and keep what you require by populating (INSERT) into a working (or temp) table in each cursor iteration.

Selecting from this final table would present your results. Drop your working table, or the temp table will be dropped at the end of your session.

Also you could put this logic in a stored procedure.

Note, cursors are memory intensive, especially when not used properly. But very handy for non-trivial or complex logic.