Specifying the language in rule

[GoluzinaD]

Hello!

I would like to request a small change to the rule writing logic. You already have linguistic analysis, but we think it can be improved.

Please add the ability to specify the language in which we are searching for hits directly in the rule description. Also, include the most popular extensions (json and yaml) instead of languages. This would allow us to specify that we are searching for secrets in a particular language before applying linguistic analysis, and then apply the linguistic analysis on top of the hit.

[mickgmdb]

Thanks for the suggestion! Just so I understand the intent: is the goal mainly to improve accuracy (reduce false positives by only applying a rule to JSON/YAML or a specific language), to improve speed (skip irrelevant files), or something else?

Kingfisher is already an order of magnitude faster than tools like Gitleaks/Trufflehog because it runs rule regex patterns through Intel’s hardware-accelerated Hyperscan engine, and then layers in Tree-sitter parsing for supported languages. Rules themselves don’t currently carry a language or extension field, and we rely on the scanner to infer file type and apply language-aware parsing automatically.

So before we consider adding language scoping directly in rules, I’d like to understand whether you see the main benefit as accuracy (e.g. better targeting of regexes) or performance (though performance is already very high with Hyperscan), or something else.

[GoluzinaD]

Now we see cases where language analysis does not work because it cannot determine the language (for example, for JSON). And we have false positives. It takes time to sort them out.
For example, when I search secrets only in JAVA and do not expect to see them in JSON, I want to point this out directly.
We are talking about the speed of processing results, not the work of the instrument.

[mickgmdb]

Ok, I think I understand. Let me think about how to best implement this.

[mickgmdb]

@GoluzinaD - Could you please try using the latest version of Kingfisher (v1.55.0)? I have greatly improved the language detection capabilities when parsing source code.