使用pl sql提取子字符串

Question

I have a table T1 with below data

Sno  Ns_NAME       Mode    stat
1    AF_rtf_Nd_1   Manual  2
2    AF_rtf_Nd_2   Manual  3
3    AF_rtf_Nd_2i  Manual  2
4    AF_rtf_Nd_3   Auto    2
5    AF_rtf_Nd_3i  Auto    3

I need to perform below,

check if it is manual, fetch from Ns_NAME upto last "_" and check for duplicates. In this case there is 1 duplicate. Obtain average stat [(2+3)/2] of those two rows and pump into another table T2.

Output:

T2

AF_rtf_Nd      Manual  2.5

I tried using substr function and used etract . But it is not fetching the correct result.

Answer 1

In order to look at 'Manual' records only, use a WHERE clause. Then aggregate over the substring. You get the substring with INSTR plus SUBSTR or with REGEX_REPLACE . Then only keep duplicates by using HAVING COUNT(*) > 1 .

insert into t2
select
  min(sno),
  regexp_replace(ns_name, '_[^_]*$', ''),
  'Manual',
  avg(stat)
from t1
where mode = 'Manual'
group by regexp_replace(ns_name, '_[^_]*$', '') 
having count(*) > 1;

The equivalent to the REGEXP_REPLACE with INSTR and SUBSTR is

substr(ns_name, 1, instr(ns_name, '_', -1) - 1)

(Only difference is when there is no '_' in the string at all.)