具有一对多关系的分层查询

时间:2015-05-11 13:42:45

标签: sql oracle hierarchical-query

我有三张桌子; d_algorithme

   ID_ALGO VERSION_ALGO LIBELLE_ALGO                                     
---------- ------------ --------------------------------------------------
       300 A300V1       Algo_300_V1                                       
       301 A301V1       Algo_301_V1                                       
       302 A302V1       Algo_302_V1                                       

d_algo_ope

NUM_OPERATION    ID_ALGO VERSION_ALGO TYP_OPERATION  NUM_ORDRE
------------- ---------- ------------ ------------- ----------
          300        301 A301V1                   3          1
            1        300 A300V1                   3          1
          301        302 A302V1                   3          1

最后d_algo_maj

NUM_MISE_A_JOUR    ID_ALGO VERSION_ALGO
--------------- ---------- ------------
             11        301 A301V1      

我想创建一个查询结果,如:

id_algo   | version_algo |  has_maj
300       | A300V1       |  0       
301       | A301V1       |  1
302       | A302V1       |  1 

前两列来自d_algorithmehas_maj为0或1,具体取决于d_algo_maj中是否存在直接或间接引用的算法。间接引用是通过一个或多个d_algo_ope记录,它们共同构成一个层次结构。

显示的样本数据:

  • 300:d_algo_ope没有算法或id_algo = 1记录,d_algo_maj没有id_algo = 300记录。
  • 301:d_algo_maj记录id_algo = 301(足以使has_maj列设为1)。
  • 302:d_algo_maj没有id_algo = 302条记录。但是d_algo_openum_operation = 301有一个id_algo = 302记录,这意味着302算法引用301算法(其中包含maj),因此has_maj列应设为1。

这是DDL和DML以及其他细节(简化自我现实中的内容):

-- DDL -----------------------------

-- d_algorithme
CREATE TABLE D_ALGORITHME 
(
  ID_ALGO NUMBER(10, 0) NOT NULL 
, VERSION_ALGO VARCHAR2(6 BYTE) NOT NULL 
, LIBELLE_ALGO VARCHAR2(50 BYTE) NOT NULL 
) ;

ALTER TABLE D_ALGORITHME
ADD CONSTRAINT IX_D_ALGORITHME PRIMARY KEY 
(
  ID_ALGO 
, VERSION_ALGO 
);

-- d_algo_ope
CREATE TABLE D_ALGO_OPE 
(
  NUM_OPERATION NUMBER(10, 0) NOT NULL 
, ID_ALGO NUMBER(10, 0) NOT NULL 
, VERSION_ALGO VARCHAR2(6 BYTE) NOT NULL 
, TYP_OPERATION NUMBER(6, 0) NOT NULL 
, NUM_ORDRE NUMBER(10, 0) NOT NULL 
); 

ALTER TABLE D_ALGO_OPE
ADD CONSTRAINT IX_D_ALGO_OPE PRIMARY KEY 
(
  ID_ALGO 
, VERSION_ALGO 
, NUM_ORDRE 
) ;

-- d_algo_maj
CREATE TABLE D_ALGO_MAJ 
(
  NUM_MISE_A_JOUR NUMBER(10, 0) NOT NULL 
, ID_ALGO NUMBER(10, 0) NOT NULL 
, VERSION_ALGO VARCHAR2(6 BYTE) NOT NULL 
) 
;

ALTER TABLE D_ALGO_MAJ
ADD CONSTRAINT IX_D_ALGO_MAJ PRIMARY KEY 
(
  ID_ALGO 
, VERSION_ALGO 
, NUM_MISE_A_JOUR 
)
;


-- DML ----------------


REM INSERTING into D_ALGORITHME


Insert into D_ALGORITHME (ID_ALGO,VERSION_ALGO,LIBELLE_ALGO) 
    values ('300','A300V1','Algo_300_V1');
Insert into D_ALGORITHME (ID_ALGO,VERSION_ALGO,LIBELLE_ALGO) 
    values ('301','A301V1','Algo_301_V1');
Insert into D_ALGORITHME (ID_ALGO,VERSION_ALGO,LIBELLE_ALGO) 
    values ('302','A302V1','Algo_302_V1');



REM INSERTING into D_ALGO_OPE

Insert into D_ALGO_OPE 
  (NUM_OPERATION,ID_ALGO,VERSION_ALGO,TYP_OPERATION,NUM_ORDRE) 
values ('300','301','A301V1','3','1');
Insert into D_ALGO_OPE (NUM_OPERATION,ID_ALGO,VERSION_ALGO,TYP_OPERATION,NUM_ORDRE) 
    values ('1','300','A300V1','3','1');
Insert into D_ALGO_OPE (NUM_OPERATION,ID_ALGO,VERSION_ALGO,TYP_OPERATION,NUM_ORDRE) 
    values ('301','302','A302V1','3','1');



REM INSERTING into D_ALGO_MAJ

Insert into D_ALGO_MAJ (NUM_MISE_A_JOUR,ID_ALGO,VERSION_ALGO) 
    values ('11','301','A301V1');

1 个答案:

答案 0 :(得分:1)

如果我了解您正在做什么以及您的牌桌之间的联系,那么我认为您可以使用recursive subquery factoring获得所需的结果(假设您已经使用11gR2或更高版本):< / p>

with r (id_algo, version_algo, has_maj, last_id_algo, last_version_algo) as (
  select da.id_algo, da.version_algo, decode(dm.id_algo, null, 0, 1),
    da.id_algo, da.version_algo
  from d_algorithme da
  left join d_algo_maj dm
  on dm.id_algo = da.id_algo
  and dm.version_algo = da.version_algo
  union all
  select dao.id_algo, dao.version_algo, decode(dm.id_algo, null, 0, 1),
    dao.id_algo, dao.version_algo
  from r
  join d_algo_ope dao
  on dao.id_algo = r.last_id_algo
  and dao.version_algo = r.last_version_algo
  left join d_algo_maj dm
  on dm.id_algo = dao.num_operation
)
cycle id_algo, version_algo set is_cycle  to 1 default 0
select id_algo, version_algo, max(has_maj) as has_maj
from r
group by id_algo, version_algo
order by id_algo, version_algo;

   ID_ALGO VERSION_ALGO    HAS_MAJ
---------- ------------ ----------
       300 A300V1                0
       301 A301V1                1
       302 A302V1                1

r CTE有一个锚点成员,它将d_algorithme行外部连接到d_algo_maj,并使用decode在该级别生成一个标记,或者为零或一个。那部分运行自己woud得到:

   ID_ALGO VERSION_ALGO    HAS_MAJ LAST_ID_ALGO LAST_VERSION_ALGO
---------- ------------ ---------- ------------ -----------------
       300 A300V1                0          300 A300V1           
       301 A301V1                1          301 A301V1           
       302 A302V1                0          302 A302V1           

递归成员然后以相同的方式查找任何匹配的d_aldo_ope记录和外部联接 d_algo_maj,获得相同的标志。这个部分本身就会得到:

   ID_ALGO VERSION_ALGO    HAS_MAJ LAST_ID_ALGO LAST_VERSION_ALGO
---------- ------------ ---------- ------------ -----------------
       300 A300V1                0          300 A300V1           
       301 A301V1                0          301 A301V1           
       302 A302V1                1          302 A302V1           

但如果您的样本数据中显示的级别超过您的水平,则递归递送。

通过为每个ID /版本找到聚合max(has_maj)来组合它们意味着任何级别的匹配主要记录都会给出整体标记值1,如果根本没有匹配则只得到0 - 只有带有此数据的ID 300才会发生。