Fetch email address from NVARCHAR2 DATATYPE - sql

Question

I have a table in which there is a column of NVARCHAR2 datatype which holds a string.
The string contains some Email Ids which I require to fetch in a comma separated manner.
Below is the test data --
create table nvarchar2_email (email_reject nvarchar2(1000));
insert into nvarchar2_email values ('com.wm.app.b2b.server.ServiceException: javax.mail.SendFailedException: Invalid Addresses; nested exception is:
com.sun.mail.smtp.SMTPAddressFailedException: 550 5.1.1 <manoj.dalai#gmail.com>: Recipient address rejected: User unknown in virtual alias table;
nested exception is:
com.sun.mail.smtp.SMTPAddressFailedException: 550 5.1.1 <santoshi.k#gmail.com>: Recipient address rejected: User unknown in virtual alias table
nested exception is:
com.sun.mail.smtp.SMTPAddressFailedException: 550 5.1.1 <biswajit-kumar.p#gmail.com>: Recipient address rejected: User unknown in virtual alias table');
insert into nvarchar2_email values ('com.wm.app.b2b.server.ServiceException: javax.mail.SendFailedException: Invalid Addresses; nested exception is:
com.sun.mail.smtp.SMTPAddressFailedException: 550 5.1.1 <manoj.dalai#gmail.com>: Recipient address rejected: User unknown in virtual alias table;
nested exception is:
com.sun.mail.smtp.SMTPAddressFailedException: 550 5.1.1 <santoshi.k#gmail.com>: Recipient address rejected: User unknown in virtual alias table');
I am trying to use the below SQL but it is repeating the Email Ids !!
select email_rejetc, listagg(REGEXP_substr (email_rejetc,'[A-Za-z0-9._%+-]+#[A-Za-z0-9.-]+\.[A-Za-z]{2,4}', 1,level), ',') within group (order by email_rejetc) invalid_email
from nvarchar2_email
connect by level <= REGEXP_count (email_rejetc,'[A-Za-z0-9._%+-]+#[A-Za-z0-9.-]+\.[A-Za-z]{2,4}')
group by EMAIL_REJETC
Here the required output is like
manoj.dalai#gmail.com,santosh.k#gmail.com,biswajit-kumar#gmail.com
Number of emails can VARY in different rows of the table;
My DB is :
Oracle Database 11g Enterprise Edition Release 11.2.0.3.0 - 64bit Production

Answer 1

select (select listagg (regexp_substr(cast(e.email_reject as varchar2(1000)),'<(.*?#.*?)>',1,level,'',1),',')
within group (order by e.email_reject)
from dual
connect by level <= regexp_count (e.email_reject,'<.*?#.*?>')
) as emails
from nvarchar2_email e
;
P.s.
There seem to be an issue with regexp_substr and nvarchar that causes each character in the result to be preceded by \0.
Tested on Oracle Database 11g Express Edition Release 11.2.0.2.0 - 64bit Production

Answer 2

According to your example, it would appear that the e-mail address is always presented as <aaaa#bbbb>, meaning a <, a string with a # in the middle, and a > sign.
You could try something like this (cannot check syntax, so you might need to do some tests):
SUBSTR(<input string> ,
INSTR(<input string>,'<') + 1 ,
(INSTR(<input string>,'>') - INSTR(<input string>,'<') - 2
) ;
This will yield the FIRST e-mail address within the string. You may use the same concept (providing a string without the first section that contains the first e-mail address) in a loop to extract additional addresses within the same string.
I can't see a way to do this through a single "SELECT" statement because each string may have several (and not all string the same number of) addresses.
One option to investigate is to implement a recursive select (Oracle supports this), but it will be much more complex.
Personally, I would go with the approach suggested above.