(where A represents the Atom type that the key will be represented as)

With these restrictions in place, the trie implements a reasonably efficient mechanism for:

• prefix matching
• representing large quantities of data with common prefixes
• finding shortest unique prefix
• finding alternative values
• finding longest common prefixes

For Example:

use trying::trie::TrieVec;
use unicode_segmentation::UnicodeSegmentation;

// Declare a trie which will store &str keys
// with usize values.
let mut trie = TrieVec::<&str, usize>::new();
let s = "a̐éö̲\r\n";
let input = s.graphemes(true);
// Insert our graphemes into the trie
trie.insert(input.clone());
// Anything which implements IntoIterator<Item=&str> can now be used
// to interact with our Trie
assert!(trie.contains(input.clone()));
assert!(trie.remove(input.clone()).is_none());
assert!(!trie.contains(input));

If you don't need prefix matching, then a HashMap is almost always a better choice than a trie...