honestly the python version and the default_dataset_id thing look like they're just copied from the v3 docs, but the 11 category slugs are a different story, nobody says where that list comes from. the keyword matching part is fine since you can turn it off, but the guide really should warn about a title like "BMW 320d M Sport" falling out of a "bmw 3 series" search. same goes for a narrow slice coming back empty and tripping the fail-on-empty error, that bit would’ve saved me a lot of debugging
Fair point, thanks. The 11 slugs are the Actor's own allowlist, taken from the category filter links on gumtree.com when it was built in August. Each one is sent as Gumtree's search_category URL parameter. It's a short list of common categories, not Gumtree's full tree, and only cars was benchmarked for this post. The allowlist exists so a typo fails the run instead of maybe returning an unfiltered search that looks like success. I've added this to the post.
OnlineProxy
honestly the python version and the default_dataset_id thing look like they're just copied from the v3 docs, but the 11 category slugs are a different story, nobody says where that list comes from. the keyword matching part is fine since you can turn it off, but the guide really should warn about a title like "BMW 320d M Sport" falling out of a "bmw 3 series" search. same goes for a narrow slice coming back empty and tripping the fail-on-empty error, that bit would’ve saved me a lot of debugging