A search engine embedded in your web page with the speed of a local search!

web-bloom is a proof of concept for a lightweight alternative to traditional heavy search engine. Here your request is handled locally within the web browser without going through a third party service. No calls through the Internet, no network latency and no plugin. The power of a local search engine installed directly on the computer of all your visitors!

web-bloom is designed to be fast and memory efficient at the expense of a customizable probability of false positives. It relies on index-bloom to implement a probabilistic search engine and then it is transposed into WebAssembly for execution in a web browser. An index dump can be created easily with cli-bloom and imported as a regular Json file in the browser.

1. Transform a dump file into importable Json object:

$ echo -e "export default $(cat dump.json)" > index_dump.js

2. Copy your dump to the public directory:

$ mv index_dump.js public/

3. Compile the project with wasm-pack:

$ wasm-pack build --target web --out-dir public/pkg

4. Use a local web server (mandatory to bypass CORS restrictions):

$ docker run --rm --name web-bloom-nginx -v ${PWD}/public:/usr/share/nginx/html:ro -d -p 8080:80 nginx:1.19-alpine

5. Go to http://localhost:8080
6. The sample index at public/index_dump.js ingested books of William Shakespeare from the Gutenberg project. You can search words present in:

• Hamlet
• Julius Caesar
• King Lear
• Macbeth
• Othello
• Rome and Juliet The original source represent 968923 bytes and the optimized index is only 263643 bytes, a ~73% win!

web-bloom is released under the MIT license (LICENSE).

• A good reference to start with Rust + WASM: Getting started with WebAssembly and Rust
• Building and working with rust-generated WebAssembly: wasm-pack

https://www.gutenberg.org/ebooks/author/65