Can I define string buckets with regular expressions (regex)?

[manus]

Hello,

Here is the data format:
00:00:01 subject=A.A

00:00:01 subject=B.A

00:00:01 subject=A.A.A

00:00:01 subject=A.B.A

...

I would like to count the events in buckets I would have defined with regular expressions.
For exemple here, I would like to define the following buckets:

A\.A.*
A\.[B-Z].*
B.*
[C-Z].*

and count the event in each bucket.

It looks like rangemap only works with text fields.

Bucketdir doesn't seem to allow to define my buckets with regular expressions.

Second question just in case, is there a smart function which creates clever buckets based on the repartition in the tree defined by the subject string?
By clever, I mean a function which groups a large semantic with few events together (eg: [C-Z].* ), but separate a precise semantic (eg A\.A\.A.*) because it contains more events. So in the end, all buckets are almost equal in size, so it's a very useful visual representation of where the events are in the tree, with some drill down in some parts of the tree.

Just in case, somebody wonders, or for TAG research purpose:
I'm trying to do that to get a good representation of the repartition of TIBCO RV multicast data.

[martin_mueller · Accepted solution]

As for smart clustering, you can always write a Python custom search command that does exactly what you need. Look at etc/apps/search/bin/pyrangemap.py for an outdated but easy to understand example.

As for your regex-based bucketing, you can do that natively roughly like this (pseudosplunk):

your search | eval mybucket = case(match(myfield, "myexpression1"), "mybucket1", match(myfield, "myexpression2"), "mybucket2", etc.) | (event)stats count by mybucket

If you use stats you'll get just the count by mybucket as the result, if you use eventstats you'll get the count field added to each search result according to its value of mybucket.