This project has two implementations, one in Python and one in Rust. See dev-docs.md for an overview over the Python version and tools, and the rust_search README for one on the Rust version and helpers.

Both the python and Rust version contain a service that is able to use a Qdrant vector search engine to do a semantic search in the matter of milliseconds. The python version does a text search below 4 characters, while the Rust version always does a full semantic search.

Both versions use the all-MiniLM-L6-v2 model made available by Hugging Face for embedding, and Qdrant for searching.

The services both have the same GET /api/search endpoint with the following arguments:

• q contains the term that is going to be searched.
• section contains a section that is added by the python version's crawler that can be used to additionally filter the search to a particular section of the page. This argument is optional.

The code in this project powers the Qdrant documentation's search box.