在oracle中使用regexp_replace将定义的模式替换为字符

Question

我想用oracle中的函数regexp_replace替换字符串中的单个字符。字符串中的替换应该以定义的模式开始。

实施例：

在字符串 “Heyho || HeyheyHo” 我将取代图案后面所有的 “Y” 字 “||”与字符“我”。在模式之前出现的字符应该被忽略。

字符串：置换后

Heyho || HeyheyHo 

字符串：

Heyho || HeiheiHo 

当然很容易的吗？

Answer 1

这是使用regexp_replace的解决方案。第四个参数是开始的位置。经过一番思考，我决定不跳过'+2'。不要懒惰和浪费周期测试你知道不是目标角色的角色。

SQL> with tbl(str) as (
    select 'Heyho || HeyheyHo' from dual 
    ) 
    select str before, 
      regexp_replace(str, 'y', 'i', instr(str, '||')+2) after 
    from tbl; 

BEFORE   AFTER 
----------------- ----------------- 
Heyho || HeyheyHo Heyho || HeiheiHo 

SQL> 

Answer 2

您不需要正则表达式;你可以使用INSTR，SUBSTR和REPLACE，让您的需求：

with test(s) as (
     select 'Heyho || HeyheyHo' from dual 
    ) 
    /* the query */ 
    select s as input, 
      substr(s, 1, instr(s, '||')+1) || 
      replace(substr(s, instr(s, '||')+2), 'y', 'i') as result 
    from test 

给出：

INPUT    RESULT 
----------------- -------------------- 
Heyho || HeyheyHo Heyho || HeiheiHo 

工作原理：

select s as input, 
     substr(s, 1, instr(s, '||')+1) beforeDelimiter, 
     substr(s, instr(s, '||')+2) afterDelimiter, 
     replace(substr(s, instr(s, '||')+2), 'y', 'i') afterDelimiterEdited, 
     substr(s, 1, instr(s, '||')+1) || 
     replace(substr(s, instr(s, '||')+2), 'y', 'i') as result 
from test 

给出：

INPUT    BEFOREDELI AFTERDELIM AFTERDELIM RESULT 
----------------- ---------- ---------- ---------- -------------------- 
Heyho || HeyheyHo Heyho || HeyheyHo HeiheiHo Heyho || HeiheiHo 

如果字符串中出现多个||，则replace将在第一次发生后修改字符。

根据Mathguy的评论，我不能说这个解决方案比正则表达式更快。

与正则表达式中的溶液可以是：

select regexp_replace(s, 'y', 'i', instr(s, '||')) as result 

这里是2代表小的性能测试创建的相同方式用相同的数据（5个百万行）：

SQL> create table testA3(s) as 
    2 select regexp_replace(s, 'y', 'i', instr(s, '||')) as result 
    3 from testA; 

Table created. 

Elapsed: 00:00:30.75 
SQL> create table testB3(s) as 
    2 select substr(s, 1, instr(s, '||')+1) || 
    3    replace(substr(s, instr(s, '||')+2), 'y', 'i') as result 
    4 from testB; 

Table created. 

Elapsed: 00:00:14.82 

标准这里的做法似乎更快;使用3M行的相同测试需要18秒的正则表达式方法和7秒的标准方法。

该测试当然不是详尽无遗，结果可能会根据许多事情而改变，但它是一个点，要考虑标准方式是一种很好的正则表达式替代，即使在这种情况下，需要许多标准操作来获得正则表达式的结果相同。

这是3M行的完整测试;我做了一个CREATE和2 INSERT s，以避免内存问题与CONNECT BY非常高的水平。

此外，在3M和5M行测试之间，我删除了表并重新创建它们，以确保缓存不会影响结果。

SQL> create table testA(s) as 
    2 select 'Heyho || HeyheyHo' || level || 'HeyheyHo' 
    3 from dual 
    4 connect by level <= 1000000; 

Table created. 

SQL> create table testB(s) as 
    2 select 'Heyho || HeyheyHo' || level || 'HeyheyHo' 
    3 from dual 
    4 connect by level <= 1000000; 

Table created. 

SQL> insert into testB(s) 
    2 select 'Heyho || HeyheyHo' || to_char(level + 1000000) || 'HeyheyHo' 
    3 from dual 
    4 connect by level <= 1000000; 

1000000 rows created. 

SQL> insert into testA(s) 
    2 select 'Heyho || HeyheyHo' || to_char(level + 1000000) || 'HeyheyHo' 
    3 from dual 
    4 connect by level <= 1000000; 

1000000 rows created. 

SQL> insert into testB(s) 
    2 select 'Heyho || HeyheyHo' || to_char(level + 2000000) || 'HeyheyHo' 
    3 from dual 
    4 connect by level <= 1000000; 

1000000 rows created. 

SQL> insert into testA(s) 
    2 select 'Heyho || HeyheyHo' || to_char(level + 2000000) || 'HeyheyHo' 
    3 from dual 
    4 connect by level <= 1000000; 

1000000 rows created. 

SQL> select count(1), count(distinct s) from testA; 

    COUNT(1) COUNT(DISTINCTS) 
---------- ---------------- 
    3000000   3000000 

SQL> select count(1), count(distinct s) from testB; 

    COUNT(1) COUNT(DISTINCTS) 
---------- ---------------- 
    3000000   3000000 

SQL> set timing on 
SQL> create table testA2(s) as 
    2 select regexp_replace(s, 'y', 'i', instr(s, '||')+2) as result 
    3 from testA; 

Table created. 

Elapsed: 00:00:17.66 
SQL> create table testB2(s) as 
    2 select substr(s, 1, instr(s, '||')+1) || 
    3    replace(substr(s, instr(s, '||')+2), 'y', 'i') as result 
    4 from testB; 

Table created. 

Elapsed: 00:00:06.96 
SQL>