在未知数量的EAV属性上管理JOIN

Question

给定一个名为item的表，其中包含＆＃34; master＆＃34;其中的记录，例如：

item_id    item_type    item_name    item_description
1          clothing     T-Shirt      Get your t-shirt here
2          clothing     Polo Shirt   Another kind of short-sleeve shirt
3          computer     Macbook      2016 Macbook Pro with retina

另一个名为custom_fields的表格，我相信它将被称为EAV结构（实体 - 属性 - 值）：

id       item_id     attribute    value
1        1           size         large
2        1           color         purple
3        2           size          medium
4        2           color         green
5        3           memory        16GB
6        3           hard drive    512GB SSD

对包含所有指定/已配置属性的查询进行编码的最佳方法是什么？还支持 ORDER BY？

事实：
1.数据结构不能改变 2.属性未知（用户配置）。 3.技术上有无限数量的＆＃34;属性＆＃34; （例如，上面的size或color）可能针对特定类型的项目设置（实际上，我无法想象超过50个）。
4.查询中包含哪些属性未知（用户配置）。

假设指定的属性存在于数组中，例如：

$attributes = ['size', 'color'];

似乎某些PHP做了类似的事情：

$counter = 1;
$from = 'item AS i';
$fields = 'name, description';
$order_by = 'size';

foreach( $attributes AS $attribute ) {
    $table_alias = "a{$counter}";
    $fields.= ",{$table_alias}.value AS {$attribute}";
    $from.= " INNER JOIN custom_fields AS {$table_alias} ON i.item_id = {$table_alias}.item_id AND {$table_alias}.attribute = '{$attribute}'";
    $counter++;
}   

$query = "SELECT {$fields} FROM {$from} {$where} ORDER BY {$order_by}";

构造一个可用于包含数据的查询，以及允许ORDER BY。该查询看起来像：

SELECT name, description, a1.value AS size, a2.value AS color 
    FROM item AS i 
    INNER JOIN custom_fields AS a1 ON i.item_id = a1.item_id 
        AND a1.attribute = 'size'
    INNER JOIN custom_fields AS a2 ON i.item_id = a2.item_id 
        AND a2.attribute = 'color'

但是，如果有20多列，我担心这个查询似乎很讨厌（20 JOINS）并且可能会出现性能问题。

有没有更好的方法来构建它？是否有一些方法，比如使用临时表，可能更有用或更高效？

更新
MySQL的版本不能确保任何特定的，所以这应该与任何版本兼容，所以请假设版本5.0 。

Answer 1

你也可以只用一个** JOIN *这样做：

SELECT m.*
    ,GROUP_CONCAT(IF(cf.attribute = 'color', cf.`value`, NULL)) AS 'color'
    ,GROUP_CONCAT(IF(cf.attribute = 'size', cf.`value`, NULL)) AS 'size'
    ,GROUP_CONCAT(IF(cf.attribute = 'memory', cf.`value`, NULL)) AS 'memory'
    ,GROUP_CONCAT(IF(cf.attribute = 'hard drive', cf.`value`, NULL)) AS 'hard drive'
FROM master m
LEFT JOIN custom_fields cf ON m.item_id = cf.item_id
GROUP BY m.item_id;

要为产品（或更多）生成查询，您可以使用以下命令： 对于所有产品，删除（1）;

中的 WHERE item_id行

SELECT m。*

select DISTINCT
 CONCAT("SELECT m.*\n"
 , GROUP_CONCAT(",GROUP_CONCAT(IF(cf.attribute = '",attribute,"', cf.`value`, NULL)) AS '",attribute,"'" SEPARATOR '\n')
 , "FROM master m\nLEFT JOIN custom_fields cf ON m.item_id = cf.item_id\nGROUP BY m.item_id") as myquery
FROM custom_fields
WHERE item_id in (1);

<强>样品

mysql> select DISTINCT
    ->  CONCAT("SELECT m.*\n"
    ->  , GROUP_CONCAT(",GROUP_CONCAT(IF(cf.attribute = '",attribute,"', cf.`value`, NULL)) AS '",attribute,"'" SEPARATOR '\n')
    ->  , "FROM master m\nLEFT JOIN custom_fields cf ON m.item_id = cf.item_id\nGROUP BY m.item_id") as myquery
    -> FROM custom_fields;

<强>结果

    SELECT m.*
,GROUP_CONCAT(IF(cf.attribute = 'size', cf.`value`, NULL)) AS 'size'
,GROUP_CONCAT(IF(cf.attribute = 'color', cf.`value`, NULL)) AS 'color'
,GROUP_CONCAT(IF(cf.attribute = 'size', cf.`value`, NULL)) AS 'size'
,GROUP_CONCAT(IF(cf.attribute = 'color', cf.`value`, NULL)) AS 'color'
,GROUP_CONCAT(IF(cf.attribute = 'memory', cf.`value`, NULL)) AS 'memory'
,GROUP_CONCAT(IF(cf.attribute = 'hard drive', cf.`value`, NULL)) AS 'hard drive'FROM master m
LEFT JOIN custom_fields cf ON m.item_id = cf.item_id
GROUP BY m.item_id 

1 row in set (0,00 sec)

mysql>

并执行此查询

mysql>  SELECT m.*
    -> ,GROUP_CONCAT(IF(cf.attribute = 'size', cf.`value`, NULL)) AS 'size'
    -> ,GROUP_CONCAT(IF(cf.attribute = 'color', cf.`value`, NULL)) AS 'color'
    -> ,GROUP_CONCAT(IF(cf.attribute = 'size', cf.`value`, NULL)) AS 'size'
    -> ,GROUP_CONCAT(IF(cf.attribute = 'color', cf.`value`, NULL)) AS 'color'
    -> ,GROUP_CONCAT(IF(cf.attribute = 'memory', cf.`value`, NULL)) AS 'memory'
    -> ,GROUP_CONCAT(IF(cf.attribute = 'hard drive', cf.`value`, NULL)) AS 'hard drive'FROM master m
    -> LEFT JOIN custom_fields cf ON m.item_id = cf.item_id
    -> GROUP BY m.item_id ;
+---------+-----------+------------+------------------------------------+--------+--------+--------+--------+--------+------------+
| item_id | item_type | item_name  | item_description                   | size   | color  | size   | color  | memory | hard drive |
+---------+-----------+------------+------------------------------------+--------+--------+--------+--------+--------+------------+
|       1 | clothing  | T-Shirt    | Get your t-shirt here              | large  | purple | large  | purple | NULL   | NULL       |
|       2 | clothing  | Polo Shirt | Another kind of short-sleeve shirt | medium | green  | medium | green  | NULL   | NULL       |
|       3 | computer  | macbook    | 2016 Macbook Pro with retina       | NULL   | NULL   | NULL   | NULL   | 16GB   | 512GB SSD  |
+---------+-----------+------------+------------------------------------+--------+--------+--------+--------+--------+------------+
3 rows in set (0,00 sec)

mysql>

具有预付款声明的样本

mysql> select DISTINCT
    ->  CONCAT("SELECT m.*\n"
    ->  , GROUP_CONCAT(",GROUP_CONCAT(IF(cf.attribute = '",attribute,"', cf.`value`, NULL)) AS '",attribute,"'" SEPARATOR '\n')
    ->  , "FROM master m\nLEFT JOIN custom_fields cf ON m.item_id = cf.item_id\nGROUP BY m.item_id") as myquery INTO @sql
    -> FROM custom_fields;
Query OK, 1 row affected (0,00 sec)

mysql> PREPARE stmt FROM @sql;
Query OK, 0 rows affected (0,01 sec)
Statement prepared

mysql> execute stmt;
+---------+-----------+------------+------------------------------------+--------+--------+--------+--------+--------+------------+
| item_id | item_type | item_name  | item_description                   | size   | color  | size   | color  | memory | hard drive |
+---------+-----------+------------+------------------------------------+--------+--------+--------+--------+--------+------------+
|       1 | clothing  | T-Shirt    | Get your t-shirt here              | large  | purple | large  | purple | NULL   | NULL       |
|       2 | clothing  | Polo Shirt | Another kind of short-sleeve shirt | medium | green  | medium | green  | NULL   | NULL       |
|       3 | computer  | macbook    | 2016 Macbook Pro with retina       | NULL   | NULL   | NULL   | NULL   | 16GB   | 512GB SSD  |
+---------+-----------+------------+------------------------------------+--------+--------+--------+--------+--------+------------+
3 rows in set (0,00 sec)

mysql> DEALLOCATE PREPARE stmt;
Query OK, 0 rows affected (0,00 sec)

mysql>

Answer 2

使用mysql 5.7，您可以考虑将属性检索为json：

SELECT 
    i.item_id,
    i.item_name,
    i.item_description,
    CONCAT('{',
            GROUP_CONCAT(CONCAT(JSON_QUOTE(a1.attribute),
                        ':',
                        JSON_QUOTE(a1.value))
                SEPARATOR ','),
            '}') AS att
FROM
    item AS i
        INNER JOIN
    custom_fields AS a1 ON i.item_id = a1.item_id
        INNER JOIN
    custom_fields AS a2 ON i.item_id = a2.item_id
        AND a2.attribute = 'size'

WHERE
    a1.attribute IN ('color' , 'size')
GROUP BY 1
ORDER BY a2.value DESC

结果：

+---------+------------+------------------------------------+-----------------------------------+
| item_id | item_name  | item_description                   | att                               |
+---------+------------+------------------------------------+-----------------------------------+
|       2 | Polo Shirt | Another kind of short-sleeve shirt | {"size":"medium","color":"green"} |
|       1 | T-Shirt    | Get your t-shirt here              | {"size":"large","color":"purple"} |
+---------+------------+------------------------------------+-----------------------------------+

假设custom_fields具有唯一索引（item_id，属性）。

AND a2.attribute = 'size'是动态部分，您可以在其中定义要按顺序排列的属性，以及定义要提取的属性的a1.attribute IN ('color' , 'size')。

它与原始多连接查询略有不同。如果缺少一个或多个属性，它仍会返回该项目。