Splunk Search

json_extend Scalar Bug, or Coercing JSON Scalars into JSON Arrays

tscroggins
Champion

Hi Splunkers!

In the current json_extend documentation <https://help.splunk.com/en/splunk-enterprise/spl-search-reference/10.0/evaluation-functions/json-fun...>, if <path> specifies "a scalar or object value," the value should be "[autowrapped] ... within an array;" however, this doesn't appear to be the case:

| makeresults
| eval obj=json_object("foo", "bar") ``` or obj="{\"foo\":\"bar\"}" ```
| eval arr=json_extend(obj, "foo", json_array("baz")) ``` or arr=json_extend(obj, "foo", "[\"baz\"]") ```

If I understand the documentation correctly, "bar" should be coerced to the array ["bar"] and then extended to ["bar","baz"], but it is not. The original object is returned unmodified. json_extend works correctly when the input value is already an array:

| makeresults
| eval obj=json_object("foo", json_array("bar", "baz"))
| eval arr=json_extend(obj, "foo", json_array("qux"))

The example was contrived to illustrate the issue. In practice, the value of foo may be either an array or a scalar. For example, the lookup eval function returns a scaler on a single match and an array on multiple matches. json_extract is similarly inconsistent when it returns a Splunk native type for scalars and a string for arrays:

| makeresults
| eval obj=json_object("foo", "bar")
| eval val=json_extract(obj, "foo") ``` ==> foo ```
| eval obj=json_object("foo", json_array("bar", "baz"))
| eval val=json_extract(obj, "foo") ``` ==> ["bar","baz"] ```

If scalars can be coerced into arrays, values that may be either scalars or arrays can be parsed by json_array_to_mv or other functions without custom wrappers using if, case, replace, etc.

Have you encountered this issue? How have you solved it?

Labels (2)
0 Karma

PickleRick
SplunkTrust
SplunkTrust

I can confirm that it does indeed seem to work not as it is described in the docs. So it warrants a support case.

A walkaround would require some dirty trick with checking whether an element is a scalar or not (possibly with json_array_to_mv(json_extract(...)) and then do json_delete() and json_append(). Very ugly.

0 Karma

tscroggins
Champion

The "autowrap" functionality was introduced (according to documentation) in 9.2.0 <https://help.splunk.com/en/splunk-enterprise/search/spl-search-reference/9.2/evaluation-functions/js...>.

I just tested in 9.2.0.1, and the behavior is the same as 10.0. I'm not going to test every maintenance release between then and now, but I'd guess it's never worked as documented.

0 Karma

tscroggins
Champion

This is for me, not a customer; I only have a dev license.

0 Karma

PickleRick
SplunkTrust
SplunkTrust

You could post a docs feedback instead.

0 Karma

tscroggins
Champion

So much feedback for the new documentation SPA. 😉 Edit: I've posted feedback. Pre-Cisco, feedback was very effective; I've not submitted anything since the acquisition, so we'll see! Grammar and syntax have taken a downturn over the last year or so.

0 Karma
Got questions? Get answers!

Join the Splunk Community Slack to learn, troubleshoot, and make connections with fellow Splunk practitioners in real time!

Meet up IRL or virtually!

Join Splunk User Groups to connect and learn in-person by region or remotely by topic or industry.

Get Updates on the Splunk Community!

Laser Bananas and Edge Hubs: Exploring Operational Technology (OT) Data Through a ...

  OT is a different environment to traditional IT and can have interesting challenges when interfacing the ...

Event Series: Mastering AI Tokenomics and Splunk Agent Observability

Beyond the Black Box: Correlating AI Performance and Tokenomics with Splunk Agent Observability   As ...

span_metrics: The OpenTelemetry-Idiomatic Way to See Inside Your Services

You open a trace in Splunk Observability Cloud and everything looks fine. One root span, order-pipeline, with ...