由于很多业务表因为历史原因或者性能原因，都使用了违反第一范式的设计模式。即同一个列中存储了多个属性值（具体结构见下表）。

这种模式下，应用常常需要将这个列依据分隔符进行分割，并得到列转行的结果。

表数据：

期望得到结果：

#需要处理的表create table tbl_name (ID int ,mSize varchar(100));insert into tbl_name values (1,'tiny,small,big');insert into tbl_name values (2,'small,medium');insert into tbl_name values (3,'tiny,big');#用于循环的自增表create table incre_table (AutoIncreID int);insert into incre_table values (1);insert into incre_table values (2);insert into incre_table values (3);

select a.ID,substring_index(substring_index(a.mSize,',',b.AutoIncreID),',',-1) from tbl_name ajoinincre_table bon b.AutoIncreID <= (length(a.mSize) - length(replace(a.mSize,',',''))+1)order by a.ID;

原理分析：

这个join最基本原理是笛卡尔积。通过这个方式来实现循环。

以下是具体问题分析：

length(a.Size) - length(replace(a.mSize,',',''))+1 表示了，按照逗号分割后，改列拥有的数值数量，下面简称n

select a.ID,substring_index(substring_index(a.mSize,',',b.AutoIncreID),',',-1) from tbl_name ajoinincre_table bon b.AutoIncreID <= (length(a.mSize) - length(replace(a.mSize,',',''))+1)order by a.ID;

原理分析：

这个join最基本原理是笛卡尔积。通过这个方式来实现循环。

以下是具体问题分析：

length(a.Size) - length(replace(a.mSize,',',''))+1 表示了，按照逗号分割后，改列拥有的数值数量，下面简称n

join过程的伪代码：

根据ID进行循环

{

判断：i 是否 <= n

{

获取最靠近第 i 个逗号之前的数据， 即 substring_index(substring_index(a.mSize,',',b.ID),',',-1)

i = i +1

}

ID = ID +1

}

总结：

这种方法的缺点在于，我们需要一个拥有连续数列的独立表（这里是incre_table)。并且连续数列的最大值一定要大于符合分割的值的个数。

例如有一行的mSize 有100个逗号分割的值，那么我们的incre_table 就需要有至少100个连续行。

当然，mysql内部也有现成的连续数列表可用。如mysql.help_topic： help_topic_id 共有504个数值，一般能满足于大部分需求了。

改写后如下:

select a.ID,substring_index(substring_index(a.mSize,',',b.help_topic_id+1),',',-1) from tbl_name ajoinmysql.help_topic bon b.help_topic_id < (length(a.mSize) - length(replace(a.mSize,',',''))+1)order by a.ID;

-- SELECT  help_topic_id  FROM mysql.help_topic-- eg.把一个字段用“，”分隔开组合select group_concat(user_id ORDER BY user_id ASC) as nids from admin_userSELECT b.did,GROUP_ConCAT(b.sid ORDER BY adjustment DESC,similar DESC) FROM test b GROUP BY b.did -- 1.如果多个导购同１张单的先分解-- 加时间段select a.DJBH,a.je,substring_index(substring_index(a.dgy_list_id,',',b.help_topic_id+1),',',-1) from ipos_qtlsd  ajoinmysql.help_topic bon b.help_topic_id < (length(a.dgy_list_id) - length(replace(a.dgy_list_id,',',''))+1) and a.djbh='BP0102_qtsy000070'order by a.DJBH;-- 2.取平均值-- SELECT  help_topic_id  FROM mysql.help_topic-- 1.如果多个导购同１张单的先分解-- @zddm-- @ rqselect a.DJBH,substring_index(substring_index(a.dgy_list_id,',',b.help_topic_id+1),',',-1) AS FJID,substring_index(substring_index(a.dgy_list_mc,',',b.help_topic_id+1),',',-1) AS FJMC,FORMAT(a.je/(length(a.dgy_list_id) - length(replace(a.dgy_list_id,',',''))+1),2) AS FJJE,je from ipos_qtlsd  ajoinmysql.help_topic bon b.help_topic_id < (length(a.dgy_list_id) - length(replace(a.dgy_list_id,',',''))+1) and a.rq BETWEEN UNIX_TIMESTAMP('2016-04-01') and UNIX_TIMESTAMP('2016-05-01')and a.djbh='gd_151125000001'order by a.DJBH;-- gd_151125000001 --3.分解后的指标-- SELECT  help_topic_id  FROM mysql.help_topic-- 1.如果多个导购同１张单的先分解-- @khdm_change 终端代码-- @start_time　开始时间-- @end_time　　结束时间-- SELECT * FROM ipos_qtlsd WHERE djbh='gd_151125000001'set @khdm_change ='BP0102';set @start_time=UNIX_TIMESTAMP('2016-04-01');set @end_time=UNIX_TIMESTAMP('2016-05-01');SELECT FJID,FJMC,SUM(FJJE) FROM(select a.zddm,a.zdmc,a.DJBH, substring_index(substring_index(a.dgy_list_id,',',b.help_topic_id+1),',',-1) AS FJID, substring_index(substring_index(a.dgy_list_mc,',',b.help_topic_id+1),',',-1) AS FJMC, FORMAT(a.je/(length(a.dgy_list_id) - length(replace(a.dgy_list_id,',',''))+1),2) AS FJJE, je from ipos_qtlsd  a join mysql.help_topic b on b.help_topic_id < (length(a.dgy_list_id) - length(replace(a.dgy_list_id,',',''))+1) and a.rq BETWEEN  @start_time and @end_timeand a.zd_id=(SELECT id from com_base_kehu where khdm=@khdm_change)) AAGROUP BY FJID,FJMC-- and a.djbh='gd_151125000001'-- order by a.DJBH;