创建根据枚举值插入行的触发器？

Question

我对触发器很陌生，我遇到了一般问题。我想实现这一目标：

CREATE TABLE `searcharticles` (
 `articleID` int(11) unsigned NOT NULL,
 `ean` char(13) COLLATE utf8_unicode_ci DEFAULT NULL,
 `manufacturerNumber` varchar(20) COLLATE utf8_unicode_ci NOT NULL,  
 PRIMARY KEY (`articleID`)
) ENGINE=InnoDB DEFAULT CHARSET=utf8 COLLATE=utf8_unicode_ci


CREATE TABLE `addresses` (
 `articleID` int(11) unsigned NOT NULL,
 `shop` enum('shop1','shop2') COLLATE utf8_unicode_ci NOT NULL,
 `url` varchar(255) COLLATE utf8_unicode_ci NULL,
 `status` tinyint(3) NOT NULL DEFAULT '0',
 `createTime` datetime NOT NULL,
 `updateTime` timestamp NULL DEFAULT NULL ON UPDATE CURRENT_TIMESTAMP,
 PRIMARY KEY (`articleID`,`shop`)
) ENGINE=InnoDB DEFAULT CHARSET=utf8 COLLATE=utf8_unicode_ci 

在searcharticles 中插入新行时，应填充表地址，但不能只填充一次。我想为enum 中的每个商店做一个记录shop（我知道这设计得不好）。我被困在这个：

CREATE TRIGGER ins_addresses AFTER INSERT on searcharticles
FOR EACH ROW
BEGIN
  INSERT INTO addresses(articleID, shop, createtime) VALUES(NEW.articleID, ?, CURRENT_TIMESTAMP);
END

如何遍历可能的枚举值以触发带有“shop1”的插入和带有“shop2”的另一个插入，而无需在插入中写入可能的商店名称？

Answer 1

这不是很漂亮，但这从INFORMATION_SCHEMA.COLUMNS.

DELIMITER $$

DROP TRIGGER ins_addresses $$

CREATE TRIGGER ins_addresses AFTER INSERT on searcharticles
FOR EACH ROW
BEGIN

DECLARE column_list char(200);
DECLARE current_value char(200);
DECLARE counter INT DEFAULT 0;
DECLARE num_enums INT;

  /* Format goes from "enum('shop1','shop2')" to "shop1,shop2" */

  SELECT REPLACE(SUBSTR(COLUMN_TYPE,6,LENGTH(COLUMN_TYPE)-6),'''','')
  FROM INFORMATION_SCHEMA.COLUMNS 
  WHERE TABLE_SCHEMA = 'stackex' 
  AND TABLE_NAME = 'addresses' 
  AND COLUMN_NAME = 'shop'
  INTO column_list;

  SET num_enums = LENGTH(column_list)- LENGTH(REPLACE(column_list, ',', ''))+1;

  WHILE counter < num_enums DO
    SET current_value = (SELECT REPLACE(SUBSTRING(SUBSTRING_INDEX(column_list, ',', counter+1),
          LENGTH(SUBSTRING_INDEX(column_list, ',', counter )) + 1), ',', ''));

    INSERT INTO addresses(articleID, shop, createtime)
    VALUES (NEW.articleID, current_value, CURRENT_TIMESTAMP);

    SET counter = counter + 1;  
  END WHILE;
END$$
DELIMITER ;

测试：

mysql> use stackex;
Database changed
mysql> insert into searcharticles values ( 1, 'foo','bar' );
Query OK, 1 row affected (0.00 sec)

mysql> select * from addresses;
+-----------+-------+------+--------+---------------------+------------+
| articleID | shop  | url  | status | createTime          | updateTime |
+-----------+-------+------+--------+---------------------+------------+
|         1 | shop1 | NULL |      0 | 2012-08-17 09:32:33 |       NULL |
|         1 | shop2 | NULL |      0 | 2012-08-17 09:32:33 |       NULL |
+-----------+-------+------+--------+---------------------+------------+
2 rows in set (0.00 sec)

mysql>

我真的很讨厌自己回答这个问题！更改数据模型以使用查找表shops。

如果商店枚举名称中有逗号或引号（可能），它会非常糟糕。