topic Re: Field extraction regex issues in Splunk Search

Field extraction regex issues

svarendorff — Thu, 22 Dec 2022 07:30:41 GMT

Having some issue with extraction.

source:

SESSION: Session closed
Client address: 123.CCCCCCC
Client name: CC222C22[123.123.12.123]
User interface: CCCCCCC

https://regex101.com/ shows that ^[^\.\n]*SESSION:(?P<Session>.*) will work.

Splunk when trying returns almost the complete message. Almost like it does not see the new line

Basically I want from SESSION: to the end of line and if Splunk cannot do that to Client.

Re: Field extraction regex issues

gcusello — Thu, 22 Dec 2022 07:38:46 GMT

Hi @svarendorff,

please try this regex :

(?ms)^[^\.\n]*SESSION:(?P<Session>.*)\nClient\s+address

Ciao.

Giuseppe

Re: Field extraction regex issues

svarendorff — Thu, 22 Dec 2022 23:25:22 GMT

Hi Giuseppe,

Looks like that fails. e.g. zero counts for index = XXXXXXX | stats count by Session.

I did forget to put a line in the source so likely that is the issue.

--TIME: 2022-12-23 07:17:09.399
SESSION: Session closed
Client address: 123.CCCCCCC
Client name: CC222C22[123.123.12.123]
User interface: CCCCCCC

Re: Field extraction regex issues

yuanliu — Fri, 23 Dec 2022 01:54:31 GMT

Why struggle with \n when this will do? (New line is one character in SPL's PCRE that is not fully PCRE conformant.)

| rex "SESSION: (?<Session>.+)"

This is my emulation

| makeresults | fields - _time | eval _raw = "--TIME: 2022-12-23 07:17:09.399 SESSION: Session closed Client address: 123.CCCCCCC Client name: CC222C22[123.123.12.123] User interface: CCCCCCC" ``` data emulation above ```

The result is

Session	_raw
Session closed	--TIME: 2022-12-23 07:17:09.399 SESSION: Session closed Client address: 123.CCCCCCC Client name: CC222C22[123.123.12.123] User interface: CCCCCCC

Re: Field extraction regex issues

svarendorff — Fri, 23 Dec 2022 02:28:17 GMT

Thank you Yuanliu. New to this type of items in Splunk so very happy for any advice and assistance.

| rex "SESSION: (?<Session>.+)"

works in a search

However, as an extraction I get nothing. Likely overlap with another similar that looks for Activity. May be best for these to do just in the search hard coded as the events have so many different items.

E,g, the second line has Session, Activity, User, etc that I would like to extract.

Re: Field extraction regex issues

yuanliu — Fri, 23 Dec 2022 06:48:11 GMT

In SPL, newline is often represented by \s (as opposed to \n). I don't know the exact rule to be frank. So, you can try something like this in field extraction

"TIME: \d{4}(-\d\d){2} \d\d:\d\d:\d\d\.\d{3}\sSESSION: (?<Session>.+)\sClient address:"

It works with rex.