Is it possible to extract nested data make 2 keys?

[batabay]

Hi Splunkers,

I have data like this, 

Primary Key_1:

     subkey_1 : subvalue_1

     subkey_2 : subvalue_2

Primary Key_2:

     subkey_1 : subvalue_1

     subkey_2 : subvalue_2

I extract data like this, But I want to see in splunk like key1.subkey_1 = sub_value_1

Btw, this is one event.

I try in transforms.conf ($1::$2:):$3 , But I failed.

And I want to extract this data, What is the best way extract this for as I want. Or is that possible ? 

[yuanliu · Accepted solution]

If the data is highly formatted as illustrated and assuming primary key names and values do not contain space, I'd turn the string into conformant data using the following, then you can use spath to extract.

| rex mode=sed max_match=0 "s/([^:\s]+)\s*:/\"\1\":/g"
| rex mode=sed max_match=0 "s/: +(\S+)/:\"\1\"/g"
| rex mode=sed "s/^\"/{\"/g"
| rex mode=sed max_match=0 "s/(\s+)\"/\1{\"/g"
| rex mode=sed max_match=0 "s/\"\s\s+{/\", /g"
| rex mode=sed "s/\"\s+{/\"},
 /"
| rex mode=sed "s/\"$/\"}
}/"
| spath

PrimaryKey_1.subkey_1	PrimaryKey_1.subkey_2	PrimaryKey_2.subkey_1	PrimaryKey_2.subkey_2
subvalue_1	subvalue_2	subvalue_1	subvalue_2

In case you have this question in mind: This cannot be implemented in transforms.conf.

[batabay]

Some of pieces of data different but I'll try to change. The answer is correct.

[jdunlea]

Presumably this is a multi line event?

Are there finite limits to how many primary keys will be in the data? Like are there always 2 primary keys? Or could there be more/less depending on the event that comes in?