What regex flavor does Grafana use?

This source from the Grafana forum said that Grafana uses the golang regex flavor.

I'm trying to extract a key value pair from the format:

(Foobar -- 48)

After following the Grafana doc on Extracting fields, it works with:

/\((?<key>Foobar) -- (?<value>\d+)\)/

But it doesn't work without the leading and trailing / when I try it on Regex101 (using the Golang flavor). It works if I drop /.../, or change it to JavaScript flavor.

So my question is what flavor of regex does Grafana really use? If it is using RE2, why the /.../ is required?

Grafana uses two regex engines in different places.

JS / TS / ECMAScript

For everything happening on the client side JS engine is used. Almost everything that is related to Dashboard/Panel editing is done in client. So following actions

• Dashboard variables
• Transformations
• Value mappings
• Overrides (selection of fields) all use JS engine.

Please, note that in most cases Grafana doesn't evaluate queries and only passes them to the data source. So specifically for queries flavour of regex will depend on the data source itself.

RE2 / Golang

For actions taking place on the backend Golang/RE2 engine is used. Places where you might encounter this flavour:

• Alerting
  • Templating functions (like reReplaceAll, match)
  • Silence management
• Configuration
  In theory, if any of the configuration options accept regex, it will probably be of RE2 flavour. But I personally don't know any such options.

Additionally, I would like to note that in some input parameters might look like regex, but aren't one. For example, permitted_provisioning_paths accepts |-delimited lists, but those are just manually split into list.