filter_paths filters results only after running detector

[Yhtiyar]

The problem:
On projects with large amounts of dependencies and especially in foundry projects (if there are lots of tests/scripts), sometimes it takes a lot of time to run the detectors.

Even though the directory/contracts might be specified in filter_paths, the slither first runs the detector on these contracts and filters out only afterward.
https://github.com/crytic/slither/blob/977d7bfeabff6ed90faadbe2ee1423bfe04ff5cb/slither/core/slither_core.py#L413C9-L421C1

It seems like we cannot just filter out the contracts from detector.contracts, because some detectors might need access to these contracts. However, is it possible to build some kind of dependency graph and filter out the contracts that are not depended on, by the contracts which are not in filter_paths?

Or some kind of do_not_run/ignore parameter, which totally removes the contracts from running detectors?

[0xalpharush]

This is being worked on. Instead of implementing a dependency graph, we've created a PR upstream in Foundry that allows skipping building scripts and tests (foundry-rs/foundry#5267). We still need to expose this as a flag/ config and appreciate input on that (crytic/crytic-compile#435)

In the meantime, you can run forge build --build-info --skip '*/test/**' --skip '*/script/**' --force && slither . --ignore-compile to achieve this behavior (this requires upgrading Foundry to include the recently merged PR).

[Yhtiyar]

Thanks!
Also, I have made a PR on crytic-compile: crytic/crytic-compile#466

[0xmichalis]

@0xalpharush I guess for Hardhat you'd need support for NomicFoundation/hardhat#2306 ?