deep-search

This repository was created to compare golang and javascript implementations of deep search with a json.

Both go and js implementations are tested with the same data.

The results are very similar probably because I don't know go at all (github copilot did all of the work here), I expected go to be way faster.

What it does

• Given an array of objects, it must return all the objects that have the string in any value inside the object.
• Must return the object even if the string is inside a deeply nested object, no matter how many levels deep.

Results

Both implementations are very similar and using the same dataset with 19547 records

language	sensitive case	insensitive case
golang	6ms	32ms
javascript	23ms	33ms

Conclusions

You get the same results in go or js in about the same time. When it's not an insensitive implementation you get different results, go was taking around 6~7ms and js was taking around 23~24ms, in this case go was 3~4 times faster than js.

I learned a little bit of go, so it's a win for me regardless.

Next steps

• Double check both implementations, if you search any genre you get different results, for example Science Fiction returns 1473 results in js and 25 in go, it's probably a js quirk where typeof [] === 'object'.
• Get memory usage and cpu usage of both implementations.
• Create a backend and frontend to query both implementations.
• Test with fuzzy search.
• Test with different data.
• Try a more manual implementation instead of using the language features.
• Get a bigger average time, maybe with some scripts.

Running locally

You need both node and go installed to run this example

js

• cd into apps/js/js-search
• run yarn or npm install
• run yarn dev

go

• cd into apps/go/go-search
• run go run main.go