Question

我有一个具有以下结构的表：

表1

f_name           f_content
test1.txt        |0002434299|354534|535345345|05|||BCV RESULT # 174|Test 12%|
test2.txt        |543566677|HTTYE|9w5w RRLL|05|||BBN RESULT # 144|Test 15#%|3

我需要使用管道（|）分隔f_content并放置字符串的适当位置。

输出表是：

f_name          position            value
test1.txt       1                   (null)
test1.txt       2                   0002434299
test1.txt       3                   354534
test1.txt       4                   535345345
test1.txt       5                   05
test1.txt       6                   (null)
test1.txt       7                   (null)
test1.txt       8                   BCV RESULT # 174
test1.txt       9                   Test 12%
test1.txt       10                  (null)

test2.txt       1                   (null)
test2.txt       2                   543566677
test2.txt       3                   HTTYE
test2.txt       4                   9w5w RRLL
test2.txt       5                   05
test2.txt       6                   (null)
test2.txt       7                   (null)
test2.txt       8                   BBN RESULT # 144
test2.txt       9                   Test 15#%
test2.txt       10                  3

表1中有超过50万条记录。每个记录有200多个管道。

有没有一种方法可以编写优化的查询，以便它可以处理200多个管道中的500K条记录而无需填充撤消表空间？

是否可以将SQL查询编写为以块的形式处理并将其继续插入输出表中？

Answer 1

您不需要（慢速）正则表达式，并且可以使用简单的字符串函数来做到这一点：

Oracle设置：

CREATE TABLE table1 ( f_name, f_content ) AS
SELECT 'test1.txt', '|0002434299|354534|535345345|05|||BCV RESULT # 174|Test 12%|' FROM DUAL UNION ALL
SELECT 'test2.txt', '|543566677|HTTYE|9w5w RRLL|05|||BBN RESULT # 144|Test 15#%|3' FROM DUAL

CREATE TABLE output_table (
  f_name   VARCHAR2(20),
  position NUMBER(4,0),
  value    VARCHAR2(50)
);

插入声明：

INSERT INTO output_table ( f_name, position, value )
WITH rsqfc ( f_name, f_content, idx, spos, epos ) AS (
  SELECT f_name, f_content, 1, 1, INSTR( f_content, '|', 1 )
  FROM   table1
UNION ALL
  SELECT f_name, f_content, idx + 1, epos + 1, INSTR( f_content, '|', epos + 1 )
  FROM   rsqfc
  WHERE  epos > 0
)
SELECT f_name,
       idx,
       CASE
         WHEN epos > 0
         THEN SUBSTR( f_content, spos, epos - spos )
         ELSE SUBSTR( f_content, spos )
       END
FROM   rsqfc

输出：

SELECT *
FROM   output_table
ORDER BY f_name, position

F_NAME    | POSITION | VALUE           
:-------- | -------: | :---------------
test1.txt |        1 | null            
test1.txt |        2 | 0002434299      
test1.txt |        3 | 354534          
test1.txt |        4 | 535345345       
test1.txt |        5 | 05              
test1.txt |        6 | null            
test1.txt |        7 | null            
test1.txt |        8 | BCV RESULT # 174
test1.txt |        9 | Test 12%        
test1.txt |       10 | null            
test2.txt |        1 | null            
test2.txt |        2 | 543566677       
test2.txt |        3 | HTTYE           
test2.txt |        4 | 9w5w RRLL       
test2.txt |        5 | 05              
test2.txt |        6 | null            
test2.txt |        7 | null            
test2.txt |        8 | BBN RESULT # 144
test2.txt |        9 | Test 15#%       
test2.txt |       10 | 3

db <>提琴here

Answer 2

您可以结合使用regexp_substr()窗口分析功能和connect by level <= regexp_count(f_content,'\|')

with t(f_name,f_content) as
(
 select 'test1.txt','|0002434299|354534|535345345|05|||BCV RESULT # 174|Test 12%|' 
   from dual
  union all  
 select 'test2.txt','|543566677|HTTYE|9w5w RRLL|05|||BBN RESULT # 144|Test 15#%|3' 
   from dual   
) 
select f_name,
       level as position,          
       replace( regexp_substr(replace(f_content,'|',' |'),
                                      '([^\|])+',
                                       1,
                                       level
                                      ),' ',null) as value                    
  from t
 connect by level <= regexp_count(f_content,'\|') + 1
     and prior f_name = f_name and prior sys_guid() is not null

Demo

分割管道分隔字串-Oracle SQL

2 个答案: