MySQL - alternative for like - regexp?

Question

i have got over one milion record in my database table. When I use like is very slowly, when i use match against they lost some records.

I create help table:

tag_list

tag_id
tag_name

tag_rel_message

tag_id
messag_id

messages

message_id
message_text

in tag_list i add all words of $message_text - explode(" ", $message_text);

My new query is:

SELECT m.* 
FROM tag_rel_messages trm 
INNER JOIN messages m ON (trm.message_id = m.message_id) 
INNER JOIN tag_list tl ON (trm.tag_id=tl.tag_id 
WHERE tl.tag_name REGEXP 'pionas' AND tl.tag_name REGEXP 'website' 
GROUP By trm.message_id

But not display any records.

What's wrong with this query? Maybe i should use something other than REGEXP?

Thanks for help

Answer 1

I'm not sure how one column could match two things at the same time! You can use OR in place of AND , or you can use a single call to REGEXP .

SELECT m.* 
FROM tag_rel_messages trm 
INNER JOIN messages m ON (trm.message_id = m.message_id) 
INNER JOIN tag_list tl ON (trm.tag_id=tl.tag_id 
WHERE tl.tag_name REGEXP 'pionas|website' 
GROUP By trm.message_id

Answer 2

You need to join with the tag_list and tag_rel_message tables twice to match two different tags.

SELECT m.*
FROM messages AS m
JOIN tag_rel_messages AS trm1 ON m.message_id = trm1.message_id
JOIN tag_list AS t1 ON t1.tag_id = trm1.tag_id
JOIN tag_rel_messages AS trm2 ON m.message_id = trm2.message_id
JOIN tag_list AS t2 on t2.tag_id = trm2.tag_id
WHERE t1.tag_name = 'pionas' AND t2.tag_name = 'website'

Another way to do it is to join with either tag, and count the number of results per message

SELECT m.*
FROM messages AS m
JOIN tag_rel_messages AS trm ON m.message_id = trm.message_id
JOIN tag_list AS t ON t1.tag_id = trm.tag_id
WHERE t.tag_name IN ('pionas', 'website')
GROUP BY m.message_id
HAVING COUNT(*) = 2

This second form is a little easier to generalize to any number of tags. Put all the tags in the IN clause, and change the HAVING clause to match the number of tags.

Answer 3

$words = array('pionas', 'website');
$tags_id = array();
foreach ($words as $key => $val) {
    $sql = "SELECT tag_id FROM tag_list WHERE tag_name LIKE '%".$val."%'";
    $records = $db->query($sql);
    $tags_id[$key] = array_column($records, "tag_id");
}

$inner_array = array();
foreach ($tags_id as $k => $v) {
    $short_name = "trm_".$k;
    $inner_array[] = " INNER JOIN tag_rel_message as ".$short_name." ON ( ".$short_name.".message_id = m.message_id AND ".$short_name.".tag_id IN (".implode(", ", $v).")) ";
}

$sql = "SELECT DISTINCT m.* FROM messages m".implode(" ", $inner_array);

I think this is the same:

SELECT * FROM messages WHERE message_text like '%pionas%' AND message_text like '%website%'