multi-engine search with scoring

I was missing a proper multi-engine search tool (searx instances tend to lag, time out, or just die) plus something with scoring (closest thing out there is Kagi's personalized results system).

I also like coming back to my old search results (archive-type sites don't really offer that) and always had a soft spot for that late-2000s google look.

So I wrote a simple bash script that generates a local html page with search results, following this logic:

• hits the 4get api multiple times (you need to find a server with api enabled yourself)
• calls marginalia's api directly once
• stores intermediate results in a JSON file
• if a result appears in multiple engines, it gets extra points
• if the result comes from a domain in my language, it gets a bonus point (you can tweak the scoring by editing the adjust_points function)
• results from reddit get a bonus point
• results from facebook lose one point
• once everything is done, the generated html opens automatically in your default browser (I use /usr/bin/linkhandler, but you can replace that with open, xdg-open, etc.)

It sometimes takes up to half a minute to run, but that's still faster than manually checking several search engines.

You need curl, jq, perl and iconv on your OS installed.

Script link here.

Remember to check and fill in the variables before using it.