SQL-如何选择列中具有value1但同一列中没有value2的所有行？

Question

I have a sql table like this:

------------------------------------------
ID    | SKU      | Name        |  Type
-------------------------------------------
2     | ABC      | Pasta       |  2
3     | XYZ      | Maggi       |  5
2     | ABC      | Pasta       |  5
6     | MNO      | Macroni     |  2
3     | XYZ      | Maggi       |  0
3     | XYZ      | Maggi       |  2

I need to find rows which have type 2 but not have an entry for type 5 or 0

For example: It should result in

------------------------------------------
ID    | SKU      | Name        |  Type
-------------------------------------------
6     | MNO      | Macroni     |  2

Is it possible to result so using this one table only? Of course it is. This table has 1.4M rows and I used this query (not sure if it is correct)

SELECT e1.* FROM reportmal e1, reportmal e2 
WHERE e1.id= e2.id
AND e1.type=2 AND e2.type=5

The query never returned anything, because it is still running. Can you help?

Answer 1

I like to do this with having and group by :

select e.id
from reportmal e
group by e.id
having sum(e.type = 2) > 0 and
       sum(e.type in (0, 5)) = 0;

MySQL treats boolean expressions as integers in a numeric context, with "1" for true and "0" for false. Hence sum(e.type = 2) counts the number of records in each group where type = 2 . The > 0 means that at least one such row exists. The = 0 means that no such rows exist.

If you want the original rows you can join this result back to the table.

Answer 2

Sounds like a typical "not exists"-query:

select * from atable result
where result.type = 2 and not exists
 (select 1 from atable no50
  where no50.type in (5,0)
        and no50.sku = result.sku
 )

or:

select * from atable result
where result.type = 2 and result.sku not in
 (select sku from atable no50
  where no50.type in (5,0)
 )

Answer 3

Try this using group by to find the ID which has 2 and not 0 or 5 using conditional count and then using IN to get the relevant rows.

select *
from your_table
where id in (
    select
        id
    from your_table t
    group by id
    having count(case when type = 2 then 1 end) > 0
    and count(case when type in (0,5) then 1 end) = 0
);

Demo

Answer 4

select  *
from    reportmal e
where   e.type = 2
    and not exists 
        (
            select  null
            from    reportmal e2
            where   e2.id = e.id
                and e2.type in (0,5)
        ) 

As for the performance issue -
The minimum you need is an index on reportmal (id).
Index on reportmal (id,type) might be better.

Answer 5

I think you should use this:

SELECT a.*
  FROM reportmal a
 WHERE a.type = 2
   AND NOT EXISTS (SELECT 1
                     FROM reportmal b
                    WHERE b.name = a.name
                      AND b.type IN (5,0));

It's like: "All rows with TYPE 2, and NOT EXISTS a row with same NAME and TYPE 5 or 0"

I don't know what's your parameter to define an "entry". Maybe you should add SKU on subquery (it all depends your data model).

Answer 6

Your query was close, but you were joining on Id which will give you a Cartesian product.

This is how I'd do it.

Select p1.*
    From pasta p1
        Left outer join pasta p2 on p1.sku=p2.sku
            And p2.type in (0,5)
        Where p2.type is null

Hope that helps you.