如何在Oracle

Question

编写一个查询与循环如何编写循环执行以下任务的查询： （代个人成绩的ID）

SELECT * From Table where id= 24 
    Id | next_to_check | next_to_check_2 
    24| 34,23  | 4 

然后我们检查了我们显示的手段34， 23.4

SELECT * From tablle where id = 34 
    SELECT * From tablle where id = 23 
    SELECT * From tablle where id = 4 

则替换结果34,23,4然后与结果的结果，并越陷越深

24-> 34,23,4 
34-> which results in a 
23-> which results in a 
4-> is as a result of 

什么结果 - >结果，结果等......

，当我做手工，它看起来像这样：

Answer 1

的查询是，如果一个表很简单已经适当地正常化了。
正如在评论中提到，有两个问题：

• next_to_check和next_to_check2是存储相同的值
• 两列包含逗号分隔值列表两列，而不是单个值

表应该是这样的：

SELECT * From Table where id= 24 
    Id | next_to_check | 
    24| 34   | 
    24| 23   | 
    24| 4   | 

其中一个next_to_check列的类型必须与id列相同，以避免不必要的转换。

对于上述表的查询可能只是：

：

SELECT * 
FROM "TABLE" 
start with id = 24 
connect by id = prior next_to_check; 

---

如果该表不能正常化，那么你就可以“飞”使用这样的查询数据标准化

WITH normalized_data As (
    SELECT id, trim(regexp_substr(next_to_check, '[^,]+', 1, LEVEL)) next_to_check 
    FROM "TABLE" 
    CONNECT BY LEVEL <= regexp_count(next_to_check, ',')+1 
    UNION ALL 
    SELECT id, trim(regexp_substr(next_to_check_2, '[^,]+', 1, LEVEL)) next_to_check 
    FROM "TABLE" 
    CONNECT BY LEVEL <= regexp_count(next_to_check_2, ',')+1 
) 
SELECT * FROM normalized_data 

再粘上第一查询上述查询：

WITH normalized_data As (
    SELECT id, trim(regexp_substr(next_to_check, '[^,]+', 1, LEVEL)) next_to_check 
    FROM "TABLE" 
    CONNECT BY LEVEL <= regexp_count(next_to_check, ',')+1 
    UNION ALL 
    SELECT id, trim(regexp_substr(next_to_check_2, '[^,]+', 1, LEVEL)) next_to_check 
    FROM "TABLE" 
    CONNECT BY LEVEL <= regexp_count(next_to_check_2, ',')+1 
) 
SELECT * FROM normalized_data 
start with id = 24 
connect by id = prior next_to_check; 

但是这种“解决方法”的表现会很差，它可能适用于100或1000条记录，但在更大的表上需要数年。