MySQL with multiple LIKE options. What's the alternative option? - php

Question

I am writing a PDO prepare, but I want to know if there is another good way to write it so it runs faster.
Here is the code:
function comp_post_code($cat, $comp_post_code){
global $DBH;
$STH = $DBH->prepare("SELECT * from uk_data where
cat10 like :comp_post_code AND (
cat1 like :cat OR
cat2 like :cat OR
cat3 like :cat OR
cat4 like :cat OR
cat5 like :cat OR
cat6 like :cat OR
cat7 like :cat
)") ;
$STH->bindValue(':cat', "%$cat%", PDO::PARAM_STR);
$STH->bindValue(':comp_post_code', "$comp_post_code%", PDO::PARAM_STR);
$STH->execute();
$STH->setFetchMode(PDO::FETCH_ASSOC);
return $STH;
$DBH = Null;
}
I want full data from table so I am using Select. * ....Thx
Edited:- cat10 is a postcode, cat 1 to cat7 are categories. I need to search the categories in a given postcode.
Here is the table format:
CREATE TABLE IF NOT EXISTS `uk_data` (
`slno` int(10) NOT NULL AUTO_INCREMENT,
`comp_name` varchar(150) DEFAULT NULL,
`comp_no` varchar(50) DEFAULT NULL,
`comp_street` varchar(100) DEFAULT NULL,
`comp_area` varchar(100) DEFAULT NULL,
`comp_post_code` varchar(15) DEFAULT NULL,
`comp_phone` varchar(100) DEFAULT NULL,
`comp_phone1` varchar(100) DEFAULT NULL,
`cat1` varchar(100) DEFAULT NULL,
`cat2` varchar(100) DEFAULT NULL,
`cat3` varchar(100) DEFAULT NULL,
`cat4` varchar(100) DEFAULT NULL,
`cat5` varchar(100) DEFAULT NULL,
`cat6` varchar(100) DEFAULT NULL,
`cat7` varchar(100) DEFAULT NULL,
`cat8` decimal(9,6) DEFAULT NULL,
`cat9` decimal(9,6) DEFAULT NULL,
`cat10` varchar(15) DEFAULT NULL,
PRIMARY KEY (`slno`),
UNIQUE KEY `Phone` (`comp_phone`),
KEY `cat10` (`cat10`)
) ENGINE=MyISAM DEFAULT CHARSET=latin1 AUTO_INCREMENT=74504 ;
The output I am using:
$uk_data_radious = comp_post_code($q,$pc_o);
while (($row = $uk_data_radious->fetch()) !== false) {
$comp_name[] = $row['comp_name'];
$comp_phone[] = $row['comp_phone'];
$comp_phone1[] = $row['comp_phone1'];
$post_code[] = $row['cat10'];
$post_code1[] = $row['comp_post_code'];
$comp_no[] = $row['comp_no'];
$comp_street[] = $row['comp_street'];
$comp_area[] = $row['comp_area'];
$cat1[] = $row['cat1'];
$cat2[] = $row['cat2'];
$cat3[] = $row['cat3'];
$cat4[] = $row['cat4'];
$cat5[] = $row['cat5'];
$cat6[] = $row['cat6'];
$cat7[] = $row['cat7'];
$distance_m[] = distance($Latitude[0],$Longitude[0],$row['cat8'],$row['cat9'],"M");
}

Answer 1

The Full Text search works better on long string for searches anywhere inside. But it requires to create a special index, and change the queries (and is not compatible with InnoDB tables).
This is quite some work. But if the like queries are really slow, FTI (Full text index) is a fast alternative.
--
Keeping the like queries, there is no much you can do - unless of course change the way you organize your data within the columns - i.e. in order to avoid the like.
To optimize somehow the like queries you could maybe merge all your cats in a single column (i.e. add a new column catx), separated with a special char that doesn't occur in your cats, like :, and do only one "like" (that should speed up a bit). E.g.
cat1: alpha
cat2: beta
cat3: gamma
...
gives
catx is :alpha:beta:gamma:
and do the like on catx
catx like :cat
to search a specific cat, search ":mycat:"
to search for a part of cat, starting with a word, search for ":start" or
to search for a ending word, search for "end:"
The like search algorithm is more efficient on a long string, than running several times on smaller strings. MySQL uses the very efficient Turbo Boyer-Moore algorithm when the search string has more than 3 characters.
But I have to warn you : there are several constraints linked to this strategy
the special separator char that shouldn't appear within the cats
any update to cat[1-7] requires the catx adjustment, so if cats are likely to change a lot, maybe this isn't a good solution
this solution works usually for well known data - like identifiers - which format changes rarely
Empirically, I don't think you can expect more than 25% gain with this strategy.