问题
I have a set of 4 tables that I want to search across. Each has a full text index. Can a query make use of every index?
CREATE TABLE `categories` (
`id` int(5) unsigned NOT NULL auto_increment,
`display_order` int(5) unsigned default NULL,
`name` varchar(64) default NULL,
`last_modified` timestamp NOT NULL default CURRENT_TIMESTAMP on update CURRENT_TIMESTAMP,
PRIMARY KEY (`id`),
UNIQUE KEY `order` (`display_order`),
FULLTEXT KEY `full_name` (`name`)
) ENGINE=MyISAM AUTO_INCREMENT=6 DEFAULT CHARSET=latin1;
CREATE TABLE `host_types` (
`id` int(5) unsigned NOT NULL auto_increment,
`category_id` int(5) unsigned default NULL,
`display_order` int(5) unsigned default NULL,
`name` varchar(64) default NULL,
`last_modified` timestamp NOT NULL default CURRENT_TIMESTAMP on update CURRENT_TIMESTAMP,
PRIMARY KEY (`id`),
UNIQUE KEY `order` (`category_id`,`display_order`),
FULLTEXT KEY `full_name` (`name`)
) ENGINE=MyISAM AUTO_INCREMENT=13 DEFAULT CHARSET=latin1;
CREATE TABLE `hosts` (
`id` int(5) unsigned NOT NULL auto_increment,
`host_id` int(5) unsigned default NULL,
`display_order` int(5) unsigned default NULL,
`name` varchar(64) default NULL,
`last_modified` timestamp NOT NULL default CURRENT_TIMESTAMP on update CURRENT_TIMESTAMP,
PRIMARY KEY (`id`),
UNIQUE KEY `order` (`host_id`,`display_order`),
FULLTEXT KEY `full_name` (`name`)
) ENGINE=MyISAM AUTO_INCREMENT=4 DEFAULT CHARSET=latin1;
CREATE TABLE `products` (
`id` int(11) unsigned NOT NULL auto_increment,
`host_id` int(5) unsigned default NULL,
`display_order` int(5) unsigned default NULL,
`uid` varchar(10) default NULL,
`name` varchar(128) default NULL,
`keywords` text,
`description` text,
`price` decimal(10,2) default NULL,
`quantity` int(11) unsigned default NULL,
`last_modified` timestamp NOT NULL default CURRENT_TIMESTAMP on update CURRENT_TIMESTAMP,
PRIMARY KEY (`id`),
FULLTEXT KEY `full_name` (`name`,`keywords`,`description`,`uid`)
) ENGINE=MyISAM AUTO_INCREMENT=14 DEFAULT CHARSET=latin1;
Here is my query;
SELECT categories.name AS category,
categories.id AS category_id,
host_types.name AS host_type,
host_types.id AS host_type_id,
hosts.name AS host,
hosts.id AS host_id,
products.name as name,
products.id AS product_id,
products.keywords as keywords,
products.description AS description,
products.quantity AS quantity,
products.price AS price,
products.uid as catalogue,
MATCH(categories.name, host_types.name, hosts.name, products.name,
products.keywords, products.description, products.uid)
AGAINST('search term') as score
FROM products
LEFT JOIN hosts ON products.host_id = hosts.id
LEFT JOIN host_types ON hosts.host_id = host_types.id
LEFT JOIN categories ON host_types.category_id = categories.id
WHERE MATCH(categories.name, host_types.name, hosts.name, products.name,
products.keywords, products.description, products.uid)
AGAINST('search term')
ORDER BY score DESC;
- categories.name == FULLTEXT - 1
- host_types.name == FULLTEXT - 2
- hosts.name == FULLTEXT - 3
- products.name, products.keywords, products.description, products.uid == FULLTEXT - 4
Here is my SQL structure, and I used the above Query.
SELECT
categories.name AS category,
categories.id AS category_id,
host_types.name AS host_type,
host_types.id AS host_type_id,
hosts.name AS host,
hosts.id AS host_id,
products.name as name,
products.id AS product_id,
products.keywords as keywords,
products.description AS description,
products.quantity AS quantity,
products.price AS price,
products.uid as catalgue
MATCH(categories.name) AGAINST('search term') as cscore,
MATCH(host_types.name) AGAINST('search term') as htscore,
MATCH(hosts.name) AGAINST('search term') as hscore,
MATCH(products.name, products.keywords, products.description, products.uid)
AGAINST('search term') as score
FROM products
LEFT JOIN hosts ON products.host_id = hosts.id
LEFT JOIN host_types ON hosts.host_id = host_types.id
LEFT JOIN categories ON host_types.category_id = categories.id
WHERE
MATCH(categories.name) AGAINST('search term') OR
MATCH(host_types.name) AGAINST('search term') OR
MATCH(hosts.name) AGAINST('search term') OR
MATCH(products.name, products.keywords, products.description, products.uid)
AGAINST('search term')
ORDER BY score DESC
CREATE TABLE `categories` (
`id` int(5) unsigned NOT NULL auto_increment,
`display_order` int(5) unsigned default NULL,
`name` varchar(64) default NULL,
`last_modified` timestamp NOT NULL default CURRENT_TIMESTAMP on update CURRENT_TIMESTAMP,
PRIMARY KEY (`id`),
UNIQUE KEY `order` (`display_order`),
FULLTEXT KEY `full_name` (`name`)
) ENGINE=MyISAM AUTO_INCREMENT=6 DEFAULT CHARSET=latin1;
CREATE TABLE `host_types` (
`id` int(5) unsigned NOT NULL auto_increment,
`category_id` int(5) unsigned default NULL,
`display_order` int(5) unsigned default NULL,
`name` varchar(64) default NULL,
`last_modified` timestamp NOT NULL default CURRENT_TIMESTAMP on update CURRENT_TIMESTAMP,
PRIMARY KEY (`id`),
UNIQUE KEY `order` (`category_id`,`display_order`),
FULLTEXT KEY `full_name` (`name`)
) ENGINE=MyISAM AUTO_INCREMENT=13 DEFAULT CHARSET=latin1;
CREATE TABLE `hosts` (
`id` int(5) unsigned NOT NULL auto_increment,
`host_id` int(5) unsigned default NULL,
`display_order` int(5) unsigned default NULL,
`name` varchar(64) default NULL,
`last_modified` timestamp NOT NULL default CURRENT_TIMESTAMP on update CURRENT_TIMESTAMP,
PRIMARY KEY (`id`),
UNIQUE KEY `order` (`host_id`,`display_order`),
FULLTEXT KEY `full_name` (`name`)
) ENGINE=MyISAM AUTO_INCREMENT=4 DEFAULT CHARSET=latin1;
CREATE TABLE `products` (
`id` int(11) unsigned NOT NULL auto_increment,
`host_id` int(5) unsigned default NULL,
`display_order` int(5) unsigned default NULL,
`uid` varchar(10) default NULL,
`name` varchar(128) default NULL,
`keywords` text,
`description` text,
`price` decimal(10,2) default NULL,
`quantity` int(11) unsigned default NULL,
`last_modified` timestamp NOT NULL default CURRENT_TIMESTAMP on update CURRENT_TIMESTAMP,
PRIMARY KEY (`id`),
FULLTEXT KEY `full_name` (`name`,`keywords`,`description`,`uid`)
) ENGINE=MyISAM AUTO_INCREMENT=14 DEFAULT CHARSET=latin1;
回答1:
You can't define fulltext indexes (or any kind of index) across multiple tables in MySQL. Each index definition references exactly one table. All columns in a given fulltext index must be from the same table.
The columns named as arguments to the
MATCH()
function must be part of a single fulltext index. You can't use a single call toMATCH()
to search all columns that are part of all fulltext indexes in your database.Fulltext indexes only index columns defined with
CHAR
,VARCHAR
, andTEXT
datatypes.You can define a fulltext index in each table.
Example:
CREATE TABLE categories (
id SERIAL PRIMARY KEY,
name VARCHAR(100),
FULLTEXT INDEX ftcat (name)
);
CREATE TABLE host_types (
id SERIAL PRIMARY KEY,
category_id BIGINT UNSIGNED,
name VARCHAR(100),
FULLTEXT INDEX ftht (name)
);
CREATE TABLE hosts (
id SERIAL PRIMARY KEY,
host_id BIGINT UNSIGNED,
category_id BIGINT UNSIGNED,
name VARCHAR(100),
FULLTEXT INDEX fthost (name)
);
CREATE TABLE products (
id SERIAL PRIMARY KEY,
name VARCHAR(100),
keywords VARCHAR(100),
uid VARCHAR(100),
description VARCHAR(100),
quantity INTEGER,
price NUMERIC(9,2),
host_id BIGINT UNSIGNED,
FULLTEXT INDEX ftprod (name, keywords, description, uid)
);
And then you can write a query that uses each respective fulltext index:
SELECT ...
MATCH(categories.name) AGAINST('search term') as cscore,
MATCH(host_types.name) AGAINST('search term') as htscore,
MATCH(hosts.name) AGAINST('search term') as hscore,
MATCH(products.name, products.keywords, products.description, products.uid)
AGAINST('search term') as score
FROM products
LEFT JOIN hosts ON products.host_id = hosts.id
LEFT JOIN host_types ON hosts.host_id = host_types.id
LEFT JOIN categories ON host_types.category_id = categories.id
WHERE
MATCH(categories.name) AGAINST('search term') OR
MATCH(host_types.name) AGAINST('search term') OR
MATCH(hosts.name) AGAINST('search term') OR
MATCH(products.name, products.keywords, products.description, products.uid)
AGAINST('search term')
ORDER BY score DESC;
回答2:
Just to expand on the answer above. There are some things to consider. If you left join multiple rows and use GROUP to group those rows, you need to get the MAX(). Also, the above only orders by relevance of one of the tables, therefore disregarding the relevance of the other tables.
SELECT ...
MAX(MATCH(categories.name) AGAINST('search term'))+
MAX(MATCH(host_types.name) AGAINST('search term'))+
MAX(MATCH(hosts.name) AGAINST('search term'))+
MAX(MATCH(products.name, products.keywords, products.description, products.uid) AGAINST('search term')) as score
FROM products
LEFT JOIN hosts ON products.host_id = hosts.id
LEFT JOIN host_types ON hosts.host_id = host_types.id
LEFT JOIN categories ON host_types.category_id = categories.id
WHERE
MATCH(categories.name) AGAINST('search term') OR
MATCH(host_types.name) AGAINST('search term') OR
MATCH(hosts.name) AGAINST('search term') OR
MATCH(products.name, products.keywords, products.description, products.uid)
AGAINST('search term')
ORDER BY score DESC;
This way you add the relevance from all the tables. The MAX() allows you to group your data. It also helps when you are LEFT joining multiple rows.
来源:https://stackoverflow.com/questions/1241602/mysql-match-across-multiple-tables