前言

在一个项目中有两张表，一张是商品码表，一张是记录出库单明细的出库记录表，记录表中有一个字段保存了以逗号隔开的商品码表的id字符串，需要根据出库明细id查找到对应出库的商品码。

表结构

goods_detail 商品码表结构

CREATE TABLE `goods_detail` (
  `id` bigint(20) NOT NULL COMMENT '主键',
  `goods_id` bigint(20) NOT NULL COMMENT '商品id',
  `bar_code` varchar(20) DEFAULT NULL COMMENT '箱码',
  `bottle_code` varchar(20) DEFAULT NULL COMMENT '瓶码'
) ENGINE=InnoDB DEFAULT CHARSET=utf8mb4 ROW_FORMAT=DYNAMIC COMMENT='商品码表';

out_goods_record 出库记录表结构

CREATE TABLE `out_goods_record` (
  `id` bigint(20) NOT NULL COMMENT 'ID',
  `order_detail_id` bigint(20) NOT NULL COMMENT '出库单明细id',
  `goods_detail_id` longtext COMMENT '商品码id,多个用'',''分隔'
) ENGINE=InnoDB DEFAULT CHARSET=utf8mb4 ROW_FORMAT=DYNAMIC COMMENT='出库记录';

sql语句

• 使用查询函数FIND_IN_SET和函数GROUP_CONCAT

SELECT
	id,
	bar_code,
	bottle_code
FROM
	goods_detail
WHERE
	FIND_IN_SET(
		id,(
		SELECT
			GROUP_CONCAT( goods_detail_id )
		FROM
			out_good_record
		WHERE
			del_flag = '0'
			AND order_detail_id = 1734492741712220161
		GROUP BY
			order_detail_id
		))>0;

该sql查询效率非常低，需要147秒，前端会提示请求接口超时，不推荐。

• 将逗号隔开的商品码id转成多行再查询
  将逗号隔开的商品码id转成多行的sql

SELECT
			substring_index( substring_index( r.goods_detail_id, ',', s.id + 1 ), ',',- 1 ) AS goods_detail_id
			FROM
			out_good_record r
			CROSS JOIN auto_add_seq s ON s.id <= LENGTH( r.goods_detail_id ) - LENGTH(
			REPLACE ( r.goods_detail_id, ',', '' ))
			WHERE
			r.order_detail_id = 1734492741712220161

auto_add_seq 这张表为自定义表，参考 mysql中字符串截取与拆分链接中的 二、分割成多行

• 使用EXISTS

SELECT
	id,
	bar_code,
	bottle_code 
FROM
	goods_detail d 
WHERE
	EXISTS (
	SELECT
		1 
	FROM
		(
		SELECT
			substring_index( substring_index( r.goods_detail_id, ',', s.id + 1 ), ',',- 1 ) AS goods_detail_id 
		FROM
			out_good_record r
			CROSS JOIN incr_sequence s ON s.id <= LENGTH( r.goods_detail_id ) - LENGTH(
			REPLACE ( r.goods_detail_id, ',', '' )) 
		WHERE
			r.order_detail_id = 1734492741712220161 
		) t 
	WHERE
		t.goods_detail_id = d.id 
	);

由于数据量太大，使用EXISTS花费的时间太多了，EXISTS比较适合内部是大表的情况，不推荐使用

• 使用IN查询

SELECT
	id,
	bar_code,
	bottle_code 
FROM
	goods_detail d 
WHERE
	d.id IN (
	SELECT
		substring_index( substring_index( r.goods_detail_id, ',', s.id + 1 ), ',',- 1 ) AS goods_detail_id 
	FROM
		out_good_record r
		CROSS JOIN incr_sequence s ON s.id < LENGTH( r.goods_detail_id ) - LENGTH( REPLACE ( r.goods_detail_id, ',', '' ) + 1 ) 
	WHERE
		r.order_detail_id = 1734492741712220161 
	);

使用IN查询，要效率要高多了

• 关联查询

SELECT
	id,
	bar_code,
	bottle_code
FROM
	goods_detail d,
	(
	SELECT
	substring_index( substring_index( r.goods_detail_id, ',', s.id + 1 ), ',',- 1 ) AS goods_detail_id
	FROM
	out_good_record r
	CROSS JOIN incr_sequence s ON s.id <= LENGTH( r.goods_detail_id ) - LENGTH(
	REPLACE ( r.goods_detail_id, ',', '' ))
	WHERE
	r.order_detail_id = 1734492741712220161
	) t
WHERE
	d.id = t.goods_detail_id

关联查询效率最高，推荐使用