Using marisa.cr for Efficient String Storage in Crystal

Bend — Fri, 11 Apr 2025 15:23:47 +0000

The marisa.cr Crystal shard gives you access to the powerful Marisa Trie data structure, perfect for storing and searching strings efficiently. Let's look at how to use it.

First, install the shard by adding it to your shard.yml:

dependencies:
  marisa:
    git: https://codeberg.org/bendangelo/marisa.cr.git

Basic Usage

Create a trie and add some strings:

require "marisa"

trie = Marisa::Trie.new
trie.add("snow")
trie.add("snow cone")
trie << "ice cream" # same as add

You can search for strings:

trie.search("ice").keys
# => ["ice", "ice cream"]

Check if a string exists:

trie.include?("snow") # => true

Working with Weights

Add strings with weights (useful for prioritization):

trie.add("ice", 1_f32)
trie.get_weight("ice") # => 1.0e-45_f32

Bulk Operations

Add multiple strings at once:

trie.add_many(["icicle", "snowball"])

Iterate through all keys:

trie.each do |key|
  puts key
end

Saving and Loading

Save your trie to disk:

trie.save("winter.trie")

Load it later:

trie = Marisa::Trie.new
trie.load("winter.trie")

Specialized Tries

For binary data:

bytes_trie = Marisa::BytesTrie.new("one" => "1", "two" => "2")
bytes_trie["one"] # => "1"

For integer values:

int_trie = Marisa::IntTrie.new("one" => 1, "two" => 2)
int_trie["one"] # => 1
int_trie.sum("one") # => 4 (sums all matching entries)

Advanced Options

Customize your trie:

trie = Marisa::Trie.new(
  ["test"],
  [1.0_f32],
  binary: true,
  num_tries: 10,
  cache_size: :large,
  order: :weight
)

The marisa.cr shard is a great choice when you need compact, efficient string storage with fast lookup capabilities. Give it a try for your next autocomplete or search feature!