INNER JOIN, Duplicate removal

Question

I have two tables

Product table
productid   int
productname varchar(32)

and

productprice table
productid    int
effectiviedt date
inactivedt   date
price        decimal (10,2)

i have a bunch of records some products has duplicate effectivedt what i want to do is for each record find the duplicate effectivedt and only keep the longest running one the inactivedt is also a date so the length would be form the effectivedt to the inactivedt

i have been trying to do it for a while but have had no luck what so ever.

i started with try to find at the earliest dates but realised that doesn't work becuase the duplicates may not be earliest dates

i cant find a way to get the difference between effectivedt and inactivedt as i cant have it in the where or having or on clause

so ye i been trying for a good long whi,e no to no avail any help would make me cry with thankfullness

thanks you in advance

Answer 1

You can calculate the date difference in teh SELECT statement like this:

SELECT ..., DATEDIFF(effectivedt, inactivedt) as active_length, ...

You can then use the result in WHERE, ORDER BY, etc.

Answer 2

Simple delete with exactly the same startdate, but the one that is deleted having an earlier stopdate.

DROP SCHEMA tmp CASCADE;
CREATE SCHEMA tmp ;
SET search_path=tmp;
CREATE TABLE productprice
        ( id INTEGER NOT NULL
        , startdate date
        , stopdate date
        , price        decimal (10,2)
        );

INSERT INTO productprice(id,startdate,stopdate,price) VALUES
 ( 1, '2012-07-01', '2012-07-10', 10.0 )
,( 1, '2012-07-01', '2012-07-01', 20.0 )
,( 2, '2012-07-05', '2012-07-06', 30.0 )
,( 2, '2012-07-05', '2012-07-10', 40.0 )
        ;

DELETE FROM productprice de
WHERE EXISTS ( SELECT *
        FROM productprice ex
        WHERE ex.id=de.id
        AND ex.startdate = de.startdate
        AND ex.stopdate > de.stopdate
        )
        ;

SELECT * FROM productprice;

Result:

CREATE TABLE
INSERT 0 4
DELETE 2
 id | startdate  |  stopdate  | price 
----+------------+------------+-------
  1 | 2012-07-01 | 2012-07-10 | 10.00
  2 | 2012-07-05 | 2012-07-10 | 40.00

Answer 3

Thank you that works very well more or less exactly what i want, the only problem is i only want to keep one record is two are the same for example so if i have

2012-07-05 | 2012-07-10 
2012-07-07 | 2012-07-25 
2012-07-07 | 2012-07-27
2012-07-27 | 2012-07-32

it should keep only one of the 2012-07-07 | 2012-07-27