Moving the workload from PHP to MySQL - php

Question

I currently have two queries that are executed in order. The first query pulls a few rows from a database and the 2nd query is executed using one value from those rows. I am only interested in the rows from the 1st query where I know the 2nd query is going to return a value higher than 0. I could potentially have thousands of rows in my first table, so I rather do the filtering in MySQL and not in PHP, I would assume it is faster.
The first query is as following
SELECT
*
FROM
wp_invoices
WHERE
deleted = 0
The 2nd query is as following:
SELECT COUNT(*) FROM wp_users A
INNER JOIN wp_usermeta B ON (A.ID = B.user_id)
LEFT JOIN wp_invoices_records C ON (A.ID = C.uid
AND C.invoice_id = %d
AND C.status != 100)
WHERE (B.meta_key = 'wp_capabilities' AND CAST(B.meta_value AS CHAR) LIKE '%%\"subscriber\"%%')
AND (C.uid IS NOT NULL)
Both queries work separately, this is not the problem. Currently, the %d in the 2nd query is replaced by PHP, which inserts the wp_invoices.id value for each row.
Ideally, I would like a query that returns only the rows from wp_invoices, where the 2nd query would return a value higher than 0. At this point I am not interested in the rows of the 2nd query, only the amount of rows found.
My table structures are as following: http://sqlfiddle.com/#!9/466cb
This is what I do in PHP, currently, to handle all this. Please keep in mind that this code was written to clarify this post, I am using deprecated MySQL functions because I don't write PHP that often anymore and haven't really looked into mysqli and I haven't checked if this code actually runs. It should give you a better indication of what I am currently doing.
<?php
/* Example code. I am not using mysql_ functions in my own solution, I just
know these functions by heart. Assume a database connection is already
established */
/* Refer to this as QUERY 1 */
$result = mysql_query("SELECT * FROM wp_invoices WHERE deleted = 0");
/* Loop over all invoices (this is not an invoice 'record'. These are
the invoice definitions, not the actual instances that people received */
while($row = mysql_fetch_assoc($result)) {
/* Refer to this as QUERY 2 */
$count_result = mysql_query(sprintf("
SELECT
COUNT(*) AS count
FROM wp_users A
INNER JOIN wp_usermeta B ON (A.ID = B.user_id)
LEFT JOIN wp_invoices_records C ON (A.ID = C.uid
AND C.invoice_id = %d
AND C.status != 100)
WHERE (B.meta_key = 'wp_capabilities'
AND CAST(B.meta_value AS CHAR) LIKE '%%\"subscriber\"%%')
AND (C.uid IS NOT NULL)", $row['id']));
if(mysql_result($count_result, 0) > 0) {
/* Here I would pull all data I would need for this invoice. $row
contains all the information about the invoice, $invoice_record
will contain all the information for this particular user about
this invoice*/
$invoice_records = mysql_query(sprintf("
SELECT
A.ID AS id,
A.display_name,
C.id AS received,
DATE_FORMAT(C.received_on, '%%d-%%m-%%Y') AS received_on,
C.status,
A.user_email,
C.price,
(SELECT meta_value FROM wp_usermeta WHERE meta_key = 'first_name'
AND user_id = A.ID) AS first_name,
(SELECT meta_value FROM wp_usermeta WHERE meta_key = 'tssnvgsl'
AND user_id = A.ID) AS insertion,
(SELECT meta_value FROM wp_usermeta WHERE meta_key = 'last_name'
AND user_id = A.ID) AS last_name
FROM wp_users A
INNER JOIN wp_usermeta B ON (A.ID = B.user_id)
LEFT JOIN wp_invoices_records C ON (A.ID = C.uid
AND C.invoice_id = %d
AND C.status != 100)
WHERE (B.meta_key = 'wp_capabilities'
AND CAST(B.meta_value AS CHAR) LIKE '%%\"subscriber\"%%')
AND (C.uid IS NOT NULL)", $row['id']));
while($invoice_record = mysql_fetch_assoc($invoice_records)) {
/* Perform logic to see if this person should receive a reminder */
}
}
}
?>
What I attempt to achieve, though:
<?php
$result = mysql_query(/* here a query that pulls all rows from QUERY 1
where QUERY 2 would have returned a count higher than 2 */);
while($row = mysql_fetch_assoc($result)) {
/* Now here I should be 100% sure that every $row has at least one
"pending" invoice record, that has not yet been paid and has a
status of < 100 */
}
?>

Answer 1

Alright. I have been racking my brain staring at your process and I think it is starting to make sense. If I am not mistaken you should be able to eliminate both query1 and query2 and move the logic into the LEFT JOIN of your third query like such.
SELECT
A.ID AS id,
A.display_name,
C.id AS received,
DATE_FORMAT(C.received_on, '%%d-%%m-%%Y') AS received_on,
C.status,
A.user_email,
C.price,
(SELECT meta_value FROM wp_usermeta WHERE meta_key = 'first_name'
AND user_id = A.ID) AS first_name,
(SELECT meta_value FROM wp_usermeta WHERE meta_key = 'tssnvgsl'
AND user_id = A.ID) AS insertion,
(SELECT meta_value FROM wp_usermeta WHERE meta_key = 'last_name'
AND user_id = A.ID) AS last_name
FROM wp_users A
INNER JOIN wp_usermeta B ON (A.ID = B.user_id)
LEFT JOIN wp_invoices_records C ON (A.ID = C.uid
AND C.invoice_id IN (SELECT id FROM wp_invoices WHERE deleted = 0)
AND C.status != 100)
WHERE (B.meta_key = 'wp_capabilities'
AND CAST(B.meta_value AS CHAR) LIKE '%%\"subscriber\"%%')
AND (C.uid IS NOT NULL)"
This query, instead of having to be ran every time for every record and the $row['id'] added via php within your LEFT JOIN I am using MySQL IN functionality with a subselect. This gives you the ability to join only on records from wp_invoices which where deleted.