MySQL搜索优化以获得更好的性能

Question

I have an array with a list of keywords,

$arr = array('london','england','football',...);

The array could have N ^th keywords.

I need to use this on a sql query but i dont want to do too many repetitions like this.

select * from t1 where title LIKE('%$arr[0]%') OR title LIKE('%$arr[1]%') [...]

just need a better solution as each keyword filters out the records.

thnx

Answer 1

One solution would be to create a lookup table that is populated with the rowid and keywordid whenever a row is created or changed.

Then you can skip all the LIKE and get much more performance.

A trigger on insert and update should work but it will cost a little more performance on insert and update.

You would just use

SELECT * FROM t1 WHERE id in (SELECT id FROM lookup WHERE keywordid IN ())

And if the keywords are many you could use a sub query more to get the keyword id's or if fewer use a cached lookup structure directly in code, dictionary or likewise.

Answer 2

http://www.webdevelopersnotes.com/tutorials/sql/tutorial_mysql_in_and_between.php3

Answer 3

I'm unsure whether you wanted to optimize the actual query itself or the means to insert keywords into the query. If it's the latter, then what comes to mind quickly is:

$query = "SELECT * FROM t1 WHERE title";

foreach($arr as $keyword)
{
    $stack[] = " LIKE '%$keyword%' ";
}

$query .= implode(' OR title ', $stack);

If you want to avoid full table scans based on keywords, that requires a different topic in itself with more explanation at your end.

Answer 4

Depending on the kind (and amount) of searches you'll be doing, you might need to consider using something else than pure-MySQL for that.

MySQL by itself is not quite well-suited when it comes to fulltext search :

• Using like '%...%' will result in bad performances (scanning all the lines of your table)
• Using a FULLTEXT index means using MyISAM as storage engine.

If you need to do a lot of searches, it might become more interesting to invest in a solution dedicated to indexing/searching through text, like Solr , or Sphinx .