Reputation: 1862
Given my two db tables aliases and subscriber have entries like this:
aliases.username = '5551234567'
aliases.contact = 'sip:[email protected]'
subscriber.username = 'a_sip_username'
I'd like to select only the matching rows that have subscriber.username within the aliases.contact field. This was my first attempt but it doesn't return anything:
SELECT
aliases.username as phone_number,
(@B:=subscriber.username) as user_name
FROM aliases,subscriber
WHERE aliases.contact regexp "^sip:@B[.*]"
Is this even possible or should I move the logic to the application?
Upvotes: 1
Views: 1230
Reputation: 425371
SELECT aliases.username AS phone_number,
subscriber.username AS user_name
FROM aliases, subscriber
WHERE aliases.contact REGEXP CONCAT('^sip:', subscriber.user_name, '[.*]')
Note that the following query will be more efficient:
SELECT aliases.username AS phone_number,
subscriber.username AS user_name
FROM aliases, subscriber
WHERE aliases.contact LIKE CONCAT('sip:', subscriber.user_name, '%')
, and this one, though seems complex, is even more efficient:
CREATE FUNCTION fn_get_next_subscriber(initial VARCHAR(200)) RETURNS VARCHAR(200)
NOT DETERMINISTIC
READS SQL DATA
BEGIN
DECLARE _username VARCHAR(200);
DECLARE EXIT HANDLER FOR NOT FOUND RETURN UNHEX('FFFF');
SELECT username
INTO _username
FROM subscribers
WHERE username>= initial
AND username NOT LIKE CONCAT(initial, '%')
ORDER BY
username
LIMIT 1;
RETURN _username;
END
SELECT a.username AS phone_number,
s.username AS user_name
FROM (
SELECT si.*, CONCAT('sip:', username) AS fromcontact
FROM subscriber si
) s, aliases a
WHERE a.contact >= fromcontact
AND a.contact < fn_get_next_subscriber(fromcontact)
This will use an index on aliases (contact)
and avoid full table scan.
See this article in my blog:
JOIN
efficiently on a LIKE
conditionUpvotes: 2