#trie #radix #matching #compression #rules #supporting #radix-trie

ab-radix-trie

A compressed radix trie implementation supporting matching rules

3 unstable releases

0.2.1 Mar 27, 2024
0.2.0 Dec 26, 2023
0.1.0 Dec 25, 2023

#802 in Text processing

21 downloads per month
Used in icann-rdap-srv

Custom license

36KB
683 lines

Latest Version

Radix-Trie

Radix-trie implementation i.e. compressed prefix-trie

https://en.wikipedia.org/wiki/Radix_tree

Some nice features:

  1. Compressed nodes
  2. Fuzzy matching - match on whitespace, replacing characters, etc.
  3. Supports all unicode characters
  4. Arbitrarily associate values to text (i.e. map strings to values)
  5. Serializable with serde

Performance

Approximately:

  1. insertion O(depth of trie)
  2. retrieval O(depth of trie)
  3. deletion O(depth of trie)
  4. Space - I don't know really, but it will behave according to ~O(entropy of text) - Similar texts are compressed together - i.e. "ABC", "ABCD" will occupy O("ABCD") space split into "ABC" and "D"

Usage:

I suggest checking out the examples and the tests for some patterns

The basic usage is along these lines

let mut trie: Trie<i32> = Trie::new();
trie.insert("romanus", None);
trie.insert("romulus", Some(10));
trie.insert("rubens", None);
trie.insert("ruber", None);
trie.insert("rubicon", None);
trie.insert("rubicundus", None);

Dependencies

~4–6MB
~108K SLoC