oracle拆分字符串并查询

样例sql:

select gro.*,
       regexp_substr(group_name, '[^/]+', 1, 1) as one_group,
       regexp_substr(group_name, '[^/]+', 1, 2) as two_group,
       regexp_substr(group_name, '[^/]+', 1, 3) as three_group,
       cfg.address as address
  from app_wam_cfg_group gro
  left join (select dbms_lob.substr(wm_concat(office.address)) address,
                    grop.id
               from app_wam_cfg_group grop, app_wam_cfg_office office
              where office.id in
                    (select regexp_substr(grop.work_address_id,
                                          '[^,]+',
                                          1,
                                          rownum)
                       from dual
                     connect by rownum <=
                                length(grop.work_address_id) -
                                length(replace(grop.work_address_id, ',')) + 1)
              group by grop.id) cfg on cfg.id = gro.id

最内层的sql作用为根据将组合id拆分成id的一个集合，可以使用in关键字进行处理；

首先以正则进行字符串截取，因为regexp_substr的第4个参数为分隔组的第几位，由于是不确定的，所以后面通过connect by 来限制分隔组有几个，通过字符串的长度，去掉分隔符，获取长度大小。

倒数第二层则通过wm_concat函数进行合并，dbms_lob.substr函数将clob转换成varchar.

最外一层则为关联