Question

I am running the following query which works except when there are duplicate records (duplicate except for the id (primary key, int) field and the modified (timestamp) columns, there are many columns which get duplicated.
The dups are a result of people hitting a submit button multiple times on a form (gotta fix that too, but thats another issue).
Here is the query:

$sql = "SELECT COUNT(id) AS rcount,
    SUM(num_guests) AS gcount,
    DATE_FORMAT(modifed, '%%m-%%Y') AS adate,
    SUM(1st_visit REGEXP '^no') AS repeat_guest
    FROM reservation_stats
    WHERE establishmentid = %i
    AND num_guests > 0
    GROUP BY adate
    ORDER BY added";

I don’t want the SUM to include the duplicates. I am using mysql and php.

Thanks

PS the table structure:

 id     int(10)          
establishmentid     int(11)         
name    varchar(50)     
email   varchar(100)    
phone   varchar(30)     
num_guests  int(11)         

reservation_time    varchar(50)     
reservation_date    date            
1st_visit   varchar(10)     
data    text    
added   date        
modifed     timestamp 

PPS: My query is now after Tom’s suggestion:

 $sql = "SELECT COUNT(*) as rcount, SUM(num_guests) AS gcount, 
DATE_FORMAT(added, '%%m-%%Y') AS adate, 
1 as repeat_guest 
FROM 
( SELECT added, name, email, phone, num_guests, 
  reservation_time, reservation_date 
FROM reservation_stats where establishmentid = %i 
GROUP BY added, name, email, phone, num_guests, 
  reservation_time, reservation_date 
\) dup 
group by adate 
order by added";

Thanks Tom. This works except I lose the repeat_guest data from the original query. How can I preserve it?

PPPS: wait, how about this:

        $sql = "SELECT COUNT(*) as rcount, SUM(num_guests) AS gcount, DATE_FORMAT(added, '%%m-%%Y') AS adate,    SUM(repeat_guest)                       
            FROM
            (
             SELECT  added, name, email, phone, num_guests, reservation_time, reservation_date, SUM(1st_visit REGEXP '^no') AS repeat_guest
             FROM    reservation_stats
             where establishmentid = %i
             GROUP BY added, name, email, phone, num_guests, reservation_time, reservation_date
            ) dup
            group by adate order by added";

seems to work but is it doing what I think it is?

Answer 1

You could do something like this, where FIELD1, FIELD2 and FIELD3 are the fields that may be duplicated (and num_guests):

SELECT COUNT(*) rcount, SUM(num_guests) AS gcount

FROM
(
 SELECT  FIELD1, FIELD2, FIELD3, num_guests
 FROM    YOUR_TABLE
 GROUP BY FIELD1, FIELD2, FIELD3, num_guests
) dup