I pleasantly hacked/drafted a super basic and quick sourcefile enumerator for the new guix 1.5.0 release in python, attached. Contains a Cuirass class for programmatic access fo the build server, although dependencies are scraped from html and better collected from the derivation files as demonstrated a little in the source. Uses joblib to cache http responses if interrupted, outputs store paths on stdout. Does not yet use other servers if the .drv is missing, instead providing other avenues of enumeration. These sourcefiles can be retrieved via the .nar method and their signatures via the .narinfo method. I haven't let this run to troubleshoot edge errors which are presently allowed to propagate if they happen, instead having added resuming via caching. It uses a handful of quick scraping tactics and is delicate in spots, but likely much more robust than before due to derivation walking. Many people have likely already done this.