一分多行列转换（Gbase版）

1、源数据表结构

CREATE TABLE "sf_ref_pd_config" (
  "I_BATCH_NO" decimal(5,0) DEFAULT NULL COMMENT '批次ID',
  "V_ASSET_CLASS_NAME" varchar(200) DEFAULT NULL COMMENT '资产类型',
  "N_EXEC_ID" decimal(5,0) DEFAULT NULL COMMENT '执行排序',
  "N_PD_ID" decimal(10,0) DEFAULT NULL COMMENT 'PD_ID',
  "V_FIELD1_BUSINESS_TYPE" varchar(2000) DEFAULT NULL COMMENT '业务品种'
) ENGINE=EXPRESS REPLICATED  DEFAULT CHARSET=utf8 TABLESPACE='sys_tablespace' COMMENT='PD映射人工配置';

2、插入测试数据

INSERT INTO ifrs10.sf_ref_pd_config (I_BATCH_NO,V_ASSET_CLASS_NAME,N_EXEC_ID,N_PD_ID,V_FIELD1_BUSINESS_TYPE) VALUES 
(4,'信用卡',NULL,2,'1110160')
,(5,'零售信贷',NULL,2,'1110090,1110160,1110161,1110380')
,(5,'信用卡',NULL,3,'1110084,1110086,1110089,1110360')
,(4,'零售信贷',NULL,3,'1110030,1110084,1110086')
,(1,'信用卡',NULL,3,'1110071,1110082,1110087,1110088,1110100,1110110,1110120,1110140,1110370,1110410,1110755,1110760,1110765,1110940,1110950,1110050,1110055,1110090,1110161,1110380')
,(6,'零售信贷',NULL,1,'1110050')
,(6,'信用卡',NULL,3,'1110360')
,(1,'零售信贷',NULL,1,'1110050')
;

3、新建关联表并插入大于逗号分隔符中的数据

create table xx_sequence(id int);

-- 插入数据（列数要大于逗号分隔符中的数据）

insert into xx_sequence 
 select rowid from ex_ods_bond_credit;

4、行列转换查询

SELECT N_PD_ID,
       I_BATCH_NO,
       V_ASSET_CLASS_NAME,
       SUBSTRING_INDEX(SUBSTRING_INDEX(A.V_FIELD1_BUSINESS_TYPE,',',id+1),',',-1) col_new 
 FROM sf_ref_pd_config a 
 inner join xx_sequence tl 
 on t1.id < LENGTH(A.V_FIELD1_BUSINESS_TYPE)-LENGTH(REPLACE(A.V_FIELD1_BUSINESS_TYPE,',',''))+1;

5、后记

逗号分隔字符数据的又一种行列转换方式，脚本是源于Gbase数据库的，应该同样适用于Oracle。

附(原始SQL)：

SELECT 
    I_BATCH_NO,V_ASSET_CLASS_NAME,N_PD_ID,SUBSTRING_INDEX(SUBSTRING_INDEX(A.V_FIELD1_BUSINESS_TYPE,',',id+1),',',-1) AS num 
FROM 
    (select I_BATCH_NO,V_ASSET_CLASS_NAME,N_PD_ID,V_FIELD1_BUSINESS_TYPE from sf_ref_pd_config  ) A join 
    t1  WHERE 
    t1.id < LENGTH(A.V_FIELD1_BUSINESS_TYPE)-LENGTH(REPLACE(A.V_FIELD1_BUSINESS_TYPE,',',''))+1 ;

及内网SQL图片：

关联文章：一分多行列转换（逗号分隔的一列数据转换成多行）-CSDN博客

一分多行列转换-CSDN博客