SuggestionTrie

A Radix trie for suggestion search, it allows searching for data indexed by a set of keywords fast.

• Compressed trie by default, more memory efficient for sparse tries
• It supports a simple fuzzy to fix typos in the search.
• Customizable suggestions data, you can add anything associated with a word or a set of words.
• Non-fuzzy suggestion retrieval in µs to ns for even millions of entries.

Note

This crate uses a nightly feature #![feature(round_char_boundary)]
it only works on nightly.

Examples

A common usage for suggestion tries is to find data like words, documents, executables and any other data that is associated with a set of words;

Search for a word (in this example, an password) in a list with +256M entries in µs.

You can add custom data to the indexed suggestions, like file paths, dates, hashes, etc.

Usage

Cargo.toml

suggestion_trie = "^0.1"

Code example

use suggestion_trie::{TrieRoot, TrieInputData};
use suggestion_trie::Suggestion;
let mut trie_root: TrieRoot<i32> = TrieRoot::<i32>::new(5, Some(100));
// Get the data you want to insert into the trie
let entries = vec![
// Use lowercase keywords and query if you want to do case insensitive searches
TrieInputData {
    suggestion: Suggestion::new("Rat".to_string(), Some(4)),
    keywords: vec!["Mice".to_string(), "Rat".to_string(), "Mouse".to_string(), "Rodent".to_string()],
},
TrieInputData {
    suggestion: Suggestion::new("Cat".to_string(), Some(8000)),
    keywords: vec!["Cat".to_string(), "Kitten".to_string(), "Kitty".to_string()],
}
];
// Build the trie
trie_root.build(&entries);

// Search and get results
let results = trie_root.get_suggestions("Rodent");
assert_eq!(results.unwrap()[0].title, "Rat");

Docs

check the docs here