检索唯一记录SQL

Question

I have a table UNIQUES like :

ID        DESC      DATE         NUMBER        Amount
100       TEST1     01-01-18     1009674       10.20 
100       TEST2     01-02-18     1009674       245.10
100       TEST3     01-03-18     1009942       156.000  
100       TEST4     02-14-18     00042EX       154.6
100       TEST5     04-15-18     00042EX       25.10
100       TEST6     05-20-18     1011055       1564.0

And I would like to get the unique records where the Number field has not duplicate or is not repeating.

Result expected:

ID      DESC      DATE          NUMBER         AMOUNT
100     TEST3     01-03-18      1009942        156.000 
100     TEST6     05-20-18      1011055        1564.0 

Query I'm using:

SELECT * FROM UNIQUES 
WHERE NUMBER NOT IN (SELECT NUMBER FROM 
UNIQUES GROUP BY NUMBER HAVING COUNT(NUMBER)=1)

Any assistance or help would be really appreciated.

Answer 1

I think you are pretty close

SELECT * FROM UNIQUES 
WHERE NUMBER IN (SELECT NUMBER FROM 
UNIQUES GROUP BY NUMBER HAVING COUNT(NUMBER)=1)

And you want:

ID      DESC      DATE          NUMBER         AMOUNT
100     TEST3     01-03-18      1009942        156.000 
100     TEST6     05-20-18      1011055        1564.0 

And my answer gives:

There are several good ways to do this, I'm just going with the simplest based upon the question by just changing it from NOT IN to IN

Answer 2

You can use subquery :

select u.*
from uniques u
where not exists (select 1 from uniques u1 where u1.number = u.number and u1.desc <> u.desc);

Answer 3

I would just use window functions:

select u.*
from (select u.*, count(*) over (partition by number) as cnt
      from uniques u
     ) u
where cnt = 1;

Answer 4

SELECT SRC.ID, SRC.DESC, SRC.DATE, SRC.NUMBER, SRC.Amount
FROM UNIQUES AS SRC

INNER JOIN (SELECT NUMBER FROM UNIQUES GROUP BY NUMBER HAVING COUNT(*)=1) AS NonDupe 
ON SRC.Number = NonDupe.Number

I would avoid using column names such as DESC, DATE and NUMBER as they occur in the list of SQL and ODBC reserved words.

Answer 5

Aside from the not exists method, you can outer join the table to itself, and check for nulls in the "duplicate" table.

select *
from
    uniques u
    left outer join uniques dup
        on (u.NUMBER=dup.NUMBER)
where
    dup.NUMBER is null

This works because the left outer join will return everything from the "left" table (aliased to u above), and it will return null values from the "right" table (aliased to dup above) if there is no row found with the join on criteria.