拉链表的展开算法
在做数据仓库项目的过程中，有时候可能也会根据历史拉链表，展开为每天全量表；相当于一个还原的过程，即构建拉链表的反过程。
1、 建表及插入测试数据语句
--建表语句 --生成EDW_T00_H表(历史拉链表) -- Create table create table EDW_T00_H ( ID VARCHAR2(2) not null, START_DATE DATE not null, STATUS VARCHAR2(2), END_DATE DATE ); -- Add comments to the table comment on table EDW_T00_H is '历史拉链表'; -- Add comments to the columns comment on column EDW_T00_H.ID is 'ID'; comment on column EDW_T00_H.START_DATE is '开始日期'; comment on column EDW_T00_H.STATUS is '状态'; comment on column EDW_T00_H.END_DATE is '结束日期'; -- Create/Recreate primary, unique and foreign key constraints alter table EDW_T00_H add constraint EDW_T00_ID primary key (ID, START_DATE); --生成EDW_T00_ALL表(历史拉链展开后的表) -- Create table create table EDW_T00_ALL ( ID VARCHAR2(2) not null, DATA_DATE DATE not null, STATUS VARCHAR2(2) not null ); -- Add comments to the table comment on table EDW_T00_ALL is '历史拉链展开后的表'; -- Add comments to the columns comment on column EDW_T00_ALL.ID is 'ID'; comment on column EDW_T00_ALL.DATA_DATE is '数据日期'; comment on column EDW_T00_ALL.STATUS is '状态'; --插入历史表测试数据 insert into edw_t00_H (ID, START_DATE, STATUS, END_DATE) values ('3', to_date('01-12-2007', 'dd-mm-yyyy'), 'N', to_date('01-12-2009', 'dd-mm-yyyy')); insert into edw_t00_H (ID, START_DATE, STATUS, END_DATE) values ('3', to_date('01-12-2009', 'dd-mm-yyyy'), 'Y', to_date('16-12-2010', 'dd-mm-yyyy')); insert into edw_t00_H (ID, START_DATE, STATUS, END_DATE) values ('1', to_date('01-12-2010', 'dd-mm-yyyy'), 'Y', to_date('16-12-2010', 'dd-mm-yyyy')); insert into edw_t00_H (ID, START_DATE, STATUS, END_DATE) values ('2', to_date('01-12-2010', 'dd-mm-yyyy'), 'N', to_date('12-12-2010', 'dd-mm-yyyy')); COMMIT;
2、 插入系统维护日期表
主要用来取ROWNUM供第3步操作
--插入系统维护时间 CREATE TABLE SYS_MATIAN_DATE(DATE_COL DATE); DECLARE V_DATE DATE := TO_DATE('20080101', 'YYYYMMDD'); V_DATE_MAX DATE; BEGIN EXECUTE IMMEDIATE 'TRUNCATE TABLE SYS_MATIAN_DATE'; SELECT SYSDATE INTO V_DATE_MAX FROM DUAL; WHILE V_DATE <= V_DATE_MAX LOOP INSERT INTO SYS_MATIAN_DATE SELECT V_DATE FROM DUAL; V_DATE := V_DATE + 1; COMMIT; END LOOP; END; SELECT * FROM SYS_MATIAN_DATE;
3、 实现拉链表展开的代码
DECLARE CURSOR LSLL_ID IS SELECT ID, START_DATE, STATUS, DECODE(END_DATE, TO_DATE('30001231', 'YYYYMMDD'), SYSDATE, --TO_DATE(P_ETLDATE, 'YYYYMMDD'), END_DATE) END_DATE FROM EDW_T00_H; V_ID VARCHAR2(1000); V_START_DATE DATE; V_END_DATE DATE; V_DATA_DATE DATE; V_STATUS VARCHAR2(2); BEGIN BEGIN EXECUTE IMMEDIATE 'TRUNCATE TABLE EDW_T00_ALL'; OPEN LSLL_ID; LOOP FETCH LSLL_ID INTO V_ID, V_START_DATE, V_STATUS, V_END_DATE; EXIT WHEN LSLL_ID%NOTFOUND; INSERT INTO EDW_T00_ALL (ID, DATA_DATE, STATUS) SELECT V_ID, V_START_DATE + ROWNUM - 1 AS DATA_DATE, V_STATUS FROM SYS_MATIAN_DATE WHERE V_START_DATE + ROWNUM - 1 < V_END_DATE; COMMIT; END LOOP; CLOSE LSLL_ID; END; END;
4、 总结说明
第3步中，游标的START_DATE、END_DATE要根据实际情况，灵活变通；代码的EXECUTE IMMEDIATE 'TRUNCATE TABLE EDW_T00_ALL';需要不需要，根据实际情况来定；第2步中生成数据日期维护表，主要用来第3步
SELECT V_ID, V_START_DATE + ROWNUM - 1 AS DATA_DATE, V_STATUS
FROM SYS_MATIAN_DATE
WHERE V_START_DATE + ROWNUM - 1 < V_END_DATE;
取数用，目的是保证SYS_MATIAN_DATE表有足够的记录，以使对于每一条历史拉链表中的记录都能取得到；本文给出了我们进行拉链表展开时的一个思路。