weighted_trie
Quickstart
To use weigthed-trie, add the following to your Cargo.toml file:
[dependencies]
weighted_trie = "0.1.0" # NOTE: Replace to latest minor version.
Usage overview
use weighted_trie::WeightedTrie;
fn main() {
let mut trie = WeightedTrie::new();
// build trie with words and assoicated weights
trie.insert("pie".to_owned(), 5);
trie.insert("pita".to_owned(), 2);
trie.insert("pi".to_owned(), 1);
trie.insert("pizza".to_owned(), 10);
// get prefix based suggestions sorted by weight
let suggestions = trie.search("pi");
assert_eq!(suggestions, vec!["pizza", "pie", "pita", "pi"]);
let suggestions = trie.search("piz");
assert_eq!(suggestions, vec!["pizza"]);
// out of vocabulary
let suggestions = trie.search("apple");
assert_eq!(suggestions.len(), 0);
}
Alternatively you can use .build
method
use weighted_trie::{WeightedString, WeightedTrie};
fn main() {
let weighted_strings = vec![
WeightedString {
word: "pie".to_owned(),
weight: 5,
},
WeightedString {
word: "pita".to_owned(),
weight: 2,
},
WeightedString {
word: "pi".to_owned(),
weight: 1,
},
WeightedString {
word: "pizza".to_owned(),
weight: 10,
},
];
let trie = WeightedTrie::build(weighted_strings);
}
Benchmarks
Using 100k weighted strings
weighted_trie/insert time: [342.41 ms 343.86 ms 345.56 ms]
weighted_trie/lookup time: [1.8608 ms 1.9351 ms 2.0834 ms]
weighted_trie/build time: [326.27 ms 330.74 ms 337.03 ms]
Guidelines
README.md
is generated from cargo readme
command. Do not manually update README.md
instead edit src/lib.rs
and then run cargo readme > README.md
.
TODO:
Add testsBenchmark lookup speedBenchmark insert speed- Measure memory footprint
- Try low hanging fruit optimizations like usage of
hashbrown
instead of standart HashMap
License
License: Apache-2.0