Question

Not the best title, I apologise.

Table structure:

messages

• id – int
• message – varchar
• user – varchar
• date – varchar
• from – varchar

contacts

• contactId – int
• contactUser – varchar
• contactName – varchar
• contactFrom – varchar

My query currently is:

 SELECT
     messages1.`from`,
     messages1.`message`,
     messages1.`date` AS d,
     messages2.`count`,
     contacts.`contactId`,
     contacts.`contactName`,
     contacts.`contactFrom`
 FROM messages AS messages1,
   (
          SELECT
              MAX(`date`) AS maxdate,
             `from`,
             COUNT(*) AS `c`
          FROM messages
          WHERE
             `user` = 'MYUSER'
          GROUP BY `from`
   ) AS messages2
LEFT JOIN contacts ON
   (
       contacts.`contactUser` = 'MYUSER' AND
       contacts.`contactsFrom` = messages2.`from`
   )
WHERE
   messages1.`from`= messages2.`from` AND
   messages1.`date` = messages2.`date`
ORDER BY `date` DESC;

However, when a user has the same contact ‘from’ for two different users, they will get the same ‘conversation’ twice.

Eg:

If the following data exists:

contacts

| contactId | contactUser |  contactName | contactFrom |
|-----------|-------------|--------------|-------------|
|     1     |   giggsey   |  MyNameOne   | [phone] |
|     2     |   giggsey   | MySecondName | [phone] |

messages

|  id  |  message  |   user   |    date    |     from     |
|------|-----------|----------|------------|--------------|
|   1  |     h1    | giggsey  | 111111111  | [phone]  |
|   2  |     h2    | giggsey  | 111111113  | [phone]  |
|   3  |   random  | giggsey  | 111111116  | [phone]  |
|   4  |     h3    | giggsey  | 111111119  | [phone]  |

Then the query returns:

|     from    | message |      d    | count | contactId |  contactName | contactFrom |
|-------------|---------|-----------|-------|-----------|--------------|-------------|
| [phone] |    h3   | 111111119 |   3   |     1     |   MyNameOne  | [phone] |
| [phone] |    h3   | 111111119 |   3   |     2     | MySecondName | [phone] |
| [phone] |  random | 111111116 |   3   |  (NULL)   |     (NULL)   |   (NULL)    |

As you can see above, it is returning the same ‘conversation’ twice, but once for each duplicate of the contact.

I’m pretty sure it’s because of the JOIN, but I can’t seem to work out why.

Answer 1

Fixed it by using a subquery instead of just reading from the table.

 SELECT
     messages1.`from`,
     messages1.`message`,
     messages1.`date` AS d,
     messages2.`count`,
     contacts.`contactId`,
     contacts.`contactName`,
     contacts.`contactFrom`
 FROM messages AS messages1,
   (
          SELECT
              MAX(`date`) AS maxdate,
             `from`,
             COUNT(*) AS `c`
          FROM messages
          WHERE
             `user` = 'MYUSER'
          GROUP BY `from`
   ) AS messages2
LEFT JOIN
  (
      SELECT
          *
      FROM
         `contacts`
      GROUP BY `contactsFrom`
  ) contacts ON
   (
       contacts.`contactUser` = 'MYUSER' AND
       contacts.`contactsFrom` = messages2.`from`
   )
WHERE
   messages1.`from`= messages2.`from` AND
   messages1.`date` = messages2.`date`
ORDER BY `date` DESC;