将字符串拆分为行 Oracle SQL

Question

在搜索论坛后，我想出了以下内容，但它不起作用：/

我有一张桌子，上面有以下内容；

ID |   Strings     
123|   abc fgh dwd   
243|   dfs dfd dfg  
353|   dfs  
424|   dfd dfw  
523|    
.  
.  
. 

请不要说大约有 20,000 行，我的另一个选择是编写一个存储过程来执行此操作...基本上我需要将字符串拆分，这样每个字符串都有一行

ID |  Strings  
123| abc  
123| fgh  
123| dwd  
243| dfs  

等等……

这就是我所拥有的。

create table Temp AS   
SELECT ID, strings   
From mytable;  

SELECT DISTINCT ID, trim(regexp_substr(str, '[^ ]+', 1, level)) str  
FROM (SELECT ID, strings str FROM temp) t  
CONNECT BY instr(str, ' ', 1, level -1) >0  
ORDER BY ID;  

感谢任何帮助

Answer 1

这应该可以解决问题：

SELECT DISTINCT ID, regexp_substr("Strings", '[^ ]+', 1, LEVEL)
FROM T
CONNECT BY regexp_substr("Strings", '[^ ]+', 1, LEVEL) IS NOT NULL
ORDER BY ID;

请注意我是如何在 connect by 子句中使用 regexp_substr 的。这是为了处理多个空格的情况。

---

如果您对每行的项目数有一个可预测的上限，则可能值得将上述递归查询的性能与简单的CROSS JOIN 进行比较：

WITH N as (SELECT LEVEL POS FROM DUAL CONNECT BY LEVEL < 10)
--                                                       ^^
--                                                 up to 10 substrings
SELECT ID, regexp_substr("Strings", '[^ ]+', 1, POS)
FROM T CROSS JOIN N
WHERE regexp_substr("Strings", '[^ ]+', 1, POS) IS NOT NULL
ORDER BY ID;

查看http://sqlfiddle.com/#!4/444e3/1 进行现场演示

Answer 2

更灵活和更好的解决方案：

• 不依赖于每行项目数的可预测性。
• 不依赖于 ID 列，无论列数如何，该解决方案都会给出正确的结果。
• 甚至不依赖于 DISTINCT 关键字。

还有其他使用XMLTABLE和MODEL子句的例子，请阅读Split comma delimited strings in a table。

例如，

没有 ID 列：

SQL> WITH T AS
  2    (SELECT 'abc fgh dwd' AS text FROM dual
  3    UNION
  4    SELECT 'dfs dfd dfg' AS text FROM dual
  5    UNION
  6    SELECT 'dfs' AS text FROM Dual
  7    UNION
  8    SELECT 'dfd dfw' AS text FROM dual
  9    )
 10  SELECT trim(regexp_substr(t.text, '[^ ]+', 1, lines.column_value)) text
 11  FROM t,
 12    TABLE (CAST (MULTISET
 13    (SELECT LEVEL FROM dual CONNECT BY instr(t.text, ' ', 1, LEVEL - 1) > 0
 14    ) AS sys.odciNumberList )) lines
 15  /

TEXT
-----------
abc
fgh
dwd
dfd
dfw
dfs
dfs
dfd
dfg

9 rows selected.

带 ID 列：

SQL> WITH T AS
  2    (SELECT 123 AS id, 'abc fgh dwd' AS text FROM dual
  3    UNION
  4    SELECT 243 AS id, 'dfs dfd dfg' AS text FROM dual
  5    UNION
  6    SELECT 353 AS Id, 'dfs' AS text FROM Dual
  7    UNION
  8    SELECT 424 AS id, 'dfd dfw' AS text FROM dual
  9    )
 10  SELECT id, trim(regexp_substr(t.text, '[^ ]+', 1, lines.column_value)) text
 11  FROM t,
 12    TABLE (CAST (MULTISET
 13    (SELECT LEVEL FROM dual CONNECT BY instr(t.text, ' ', 1, LEVEL - 1) > 0
 14    ) AS sys.odciNumberList )) lines
 15   ORDER BY id
 16   /

        ID TEXT
---------- -----------
       123 abc
       123 fgh
       123 dwd
       243 dfs
       243 dfd
       243 dfg
       353 dfs
       424 dfd
       424 dfw

9 rows selected.

SQL>

Answer 3

With T As 
      (select 123 as id, 'abc fgh dwd' as strings from dual

      union

      select 243 as id, 'dfs dfd dfg' as strings from dual

      union 

      Select 353 As Id, 'dfs' As Strings From Dual

      union 

      select 424 as id, 'dfd dfw' as strings from dual


      )


select distinct id, REGEXP_SUBSTR (Replace(Strings, ' ', ','), '[^,]+', 1, level) as Strings
from t
Connect By Level <= Length(Regexp_Replace(Replace(Strings, ' ', ','),'[^,]*'))+1
order by id, strings;


**********OUTPUT*************
        ID STRINGS   
---------- -----------
   123 abc     
   123 dwd   
   123 fgh     
   243 dfd     
   243 dfg     
   243 dfs      
   353 dfs      
   424 dfd     
   424 dfw    

 9 rows selected