Getting Data In
Highlighted

Help with Advanced Source Type

Explorer

I'm trying to create a custom source type which is reading a TSV log file and the 3 column in the file is a JSON payload wrapped in quotes. I can't figure out how to get the source type to parse out the 3 column in a JSON format on splunk. Here's an example of a line entry below.

6680 "2020-03-06 13:50:13.254" "{"date":"3/6/2020 1:50:13 PM","received":"from FooServer (Unknown [172.20.36.5]) by smtp-dev.foo.com with ESMTP ; Fri, 6 Mar 2020 13:50:13 -0500","message-id":"id@message.com","from":"foo@thisMachine.com","recipients":"John.Smith@example.com","cc":"","subject":"Test Email"}"

Any advice would be helpful, thank you.

0 Karma
Highlighted

Re: Help with Advanced Source Type

Influencer

Add below content to transforms.conf. This will extract key value pair and index them.

transforms.conf.

[extract_json_fields]
REGEX = \"([\w-]+)\":\"([^"]*)\"
FORMAT = $1::$2

Refer extractjsonfields in props.conf under sourcetype stanza.

props.conf

[sourcetype_name]
TRANSFORMS-extract_fields = extract_json_fields

View solution in original post

0 Karma
Highlighted

Re: Help with Advanced Source Type

Explorer

How would I go about this if I am using the UI to create the Source Type?

0 Karma
Highlighted

Re: Help with Advanced Source Type

Influencer

Add new setting EXTRACT-fields and put this below regex. Save and search sourcetype.

"date":"(?<date>[^\"]*)","received":"(?<received>[^\"]*)","message-id":"(?<messageid>[^\"]*)","from":"(?<from>[^\"]*)","recipients":"(?<recipients>[^\"]*)","cc":"(?<cc>[^\"]*)","subject":"(?<subject>[^\"]*)"
0 Karma
Highlighted

Re: Help with Advanced Source Type

Ultra Champion
| makeresults 
| eval _raw="6680 \"2020-03-06 13:50:13.254\" \"{\"date\":\"3/6/2020 1:50:13 PM\",\"received\":\"from FooServer (Unknown [172.20.36.5]) by smtp-dev.foo.com with ESMTP ; Fri, 6 Mar 2020 13:50:13 -0500\",\"message-id\":\" \",\"from\":\" foo@thisMachine.com\",\"recipients\":\" John.Smith@example.com\",\"cc\":\"\",\"subject\":\"Test Email\"}\"" 
| rex mode=sed "s/^.*(?={)(.*)\"/\1/g" 
| spath

this is OK.

props.conf or Create source types >> advanced

[your sourcetype]
DATETIME_CONFIG = %F %T.%3Q
SEDCMD-trim_json = s/^.*(?={)(.*)\"/\1/g
KV_MODE = json
0 Karma
Highlighted

Re: Help with Advanced Source Type

Esteemed Legend

The documentation is EXTREMELY misleading in this regard. There is no such thing as creating a sourcetype. All possible sourcetype values "already exist" in any meaningful sense of the word exist. There is NOTHING that must be done to create one before using it. Just use it and poof it now exists and works.

This is not the right way to do it, however, assuming there is no nesting and no arrays in your JSON, you can get away with doing this:

Settings -> Fields -> Field Transformations -> New Field Transformation then
Set Name to something like <YourSourcetype>_JSONpayload.
Set Type to regex-based.
Set Regular expression to (?<=[,{])"([^"]+)":"([^"]*).
Set Format to $1::$2.
Set Create multivalued fields to checked (e.g. true / yes ).
Click green Save button.

Settings -> Fields -> Field Extractions -> New Field Extraction then:
Set Name to <YourSourcetype>_JSONpayload.
Set Apply to to Sourcetype + named = <YourSourcetype>.
Set Type to Uses transform.
Set Extraction/Transform to <YourSourcetype>_JSONpayload.
Click green Save button.

0 Karma