Question

I’m trying to write a query that selects from four tables

• campaignSentParent csp
• campaignSentEmail cse
• campaignSentFax csf
• campaignSentSms css

Each of the cse, csf, and css tables are linked to the csp table by csp.id = (cse/csf/css).parentId

The csp table has a column called campaignId,

What I want to do is end up with rows that look like:

| id | dateSent   | emailsSent | faxsSent | smssSent |  
| 1  | 2011-02-04 | 139        | 129      | 140      |  

But instead I end up with a row that looks like:

| 1  | 2011-02-03 | 2510340    | 2510340  | 2510340  |

Here is the query I am trying

SELECT csp.id id, csp.dateSent dateSent,  
       COUNT(cse.parentId) emailsSent,  
       COUNT(csf.parentId) faxsSent,  
       COUNT(css.parentId) smsSent  
FROM   campaignSentParent csp,  
       campaignSentEmail cse,  
       campaignSentFax csf,  
       campaignSentSms css  
WHERE  csp.campaignId = 1  
AND    csf.parentId = csp.id  
AND    cse.parentId = csp.id  
AND    css.parentId = csp.id;  

---

Adding GROUP BY did not help, so I am posting the create statements.

csp

CREATE TABLE `campaignsentparent` (
  `id` int(11) NOT NULL AUTO_INCREMENT,
  `campaignId` int(11) NOT NULL,
  `dateSent` datetime NOT NULL,
  `account` int(11) NOT NULL,
  `status` varchar(15) NOT NULL DEFAULT 'Creating',
  PRIMARY KEY (`id`)
) ENGINE=MyISAM AUTO_INCREMENT=2 DEFAULT CHARSET=latin1

cse/csf (same structure, different names)

CREATE TABLE `campaignsentemail` (
  `id` int(11) NOT NULL AUTO_INCREMENT,
  `parentId` int(11) NOT NULL,
  `contactId` int(11) NOT NULL,
  `content` text,
  `subject` text,
  `status` varchar(15) DEFAULT 'Pending',
  PRIMARY KEY (`id`)
) ENGINE=MyISAM AUTO_INCREMENT=140 DEFAULT CHARSET=latin1

css

CREATE TABLE `campaignsentsms` (
  `id` int(11) NOT NULL AUTO_INCREMENT,
  `parentId` int(11) NOT NULL,
  `contactId` int(11) NOT NULL,
  `content` text,
  `status` varchar(15) DEFAULT 'Pending',
  PRIMARY KEY (`id`)
) ENGINE=MyISAM AUTO_INCREMENT=141 DEFAULT CHARSET=latin1

Answer 1

You need to aggregate the sums separately, not as shown in the question.

SELECT csp.id, csp.dateSent dateSent,  
       e.email_count, f.fax_count, s.sms_count
  FROM campaignSentParent AS csp
  JOIN (SELECT cse.ParentId, COUNT(*) AS email_count
          FROM campaignSentEmail cse
         GROUP BY cse.ParentID) AS e ON e.parentID = csp.id
  JOIN (SELECT csf.ParentId, COUNT(*) AS fax_count
          FROM campaignSentFax csf
         GROUP BY csf.ParentID) AS f ON f.ParentID = csp.id
  JOIN (SELECT css.ParentID, COUNT(*) AS sms_count
          FROM campaignSentSms css
         GROUP BY css.ParentId) AS s ON s.ParentID = csp.id
 WHERE csp.campaignId = 1  

To do this, you pretty much have to use the JOIN notation as shown.

You depending on the quality of your optimizer and the cardinalities of the various tables and the available indexes, you might find it effective to include a join with CampaignSentParent in each of the sub-queries with the csp.CampaignID = 1 condition, so as to limit the data aggregated by the sub-queries.

---

You might notice that the result count you get is 2510340. The prime factorization of 2510340 is 2 × 2 × 3 × 5 × 7 × 43 × 139, and your expected answer is 139, 129, and 140. You can get 3 × 43 = 129; 2 × 2 × 5 × 7 = 140; and 139 = 139. In other words, the original query is generating the Cartesian product of all the rows in the three dependent tables and counting the product, rather than counting the relevant rows from each dependent table separately.