Fast fuzzy string searching/matching for Rails
Ruby
Clone or download
mezis Merge pull request #49 from savef/do-not-raise-on-missing-objects
Fix AR::RecordNotFound being raised with missing objects.
Latest commit 026fd3d Oct 18, 2017
Permalink
Failed to load latest commit information.
gemfiles Bumps version Feb 1, 2014
lib
rebund @ 03a688a Adds rebund submodule Jan 4, 2014
spec
.gitignore
.gitmodules
.rspec First working version, specs not complete Oct 25, 2012
.ruby-version Bumps version Mar 20, 2015
.travis.yml Drops 1.9.2, adds 2.1.0 to the build matrix Feb 1, 2014
Appraisals Adds Rails 4 compatibility. Fixes #12. Aug 31, 2013
Gemfile
Gemfile.lock Bumps version Mar 20, 2015
LICENSE.txt Fixes gemspec and license Aug 23, 2013
README.md Correct sentence in readme. Oct 6, 2015
Rakefile
fuzzily.gemspec Fixes gemspec and license Aug 23, 2013

README.md

Fuzzily - fuzzy string matching for ActiveRecord

Gem Version Build Status Dependency Status Code Climate Coverage Status

Show me photos of Marakech !

Here aresome photos of Marrakesh, Morroco. Did you mean Martanesh, Albania, Marakkanam, India, or Marasheshty, Romania?

Fuzzily finds misspelled, prefix, or partial needles in a haystack of strings. It's a fast, trigram-based, database-backed fuzzy string search/match engine for Rails. Loosely inspired from an old blog post.

Tested with ActiveRecord (2.3, 3.0, 3.1, 3.2, 4.0) on various Rubies (1.8.7, 1.9.3, 2.0.0, 2.1.0) and the most common adapters (SQLite3, MySQL, and PostgreSQL).

If your dateset is big, if you need yet more speed, or do not use ActiveRecord, check out blurrily, another gem (backed with a C extension) with the same intent.

Installation

Add this line to your application's Gemfile:

gem 'fuzzily'

And then execute:

$ bundle

Or install it yourself as:

$ gem install fuzzily

Usage

You'll need to setup 2 things:

  • a trigram model (your search index) and its migration
  • the model you want to search for

Create an ActiveRecord model in your app (this will be used to store a "fuzzy index" of all the models and fields you will be indexing):

class Trigram < ActiveRecord::Base
  include Fuzzily::Model
end

Create a migration for it:

class AddTrigramsModel < ActiveRecord::Migration
  extend Fuzzily::Migration
end

Instrument your model:

class MyStuff < ActiveRecord::Base
  # assuming my_stuffs has a 'name' attribute
  fuzzily_searchable :name
end

Note: The name part in the following method calls refers to the :name field. Replace it to match your searchable attribute.

Index your model (will happen automatically for new/updated records):

MyStuff.bulk_update_fuzzy_name

Search!

MyStuff.find_by_fuzzy_name('Some Name', :limit => 10)
# => records

You can force an update on a specific record with

MyStuff.find(123).update_fuzzy_name!

Indexing more than one field

Just list all the field you want to index, or call fuzzily_searchable more than once:

class MyStuff < ActiveRecord::Base
  fuzzily_searchable :name_fr, :name_en
  fuzzily_searchable :name_de
end

Custom name for the index model

If you want or need to name your index model differently (e.g. because you already have a class called Trigram):

class CustomTrigram < ActiveRecord::Base
  include Fuzzily::Model
end

class AddTrigramsModel < ActiveRecord::Migration
  extend Fuzzily::Migration
  self.trigrams_table_name = :custom_trigrams
end

class MyStuff < ActiveRecord::Base
  fuzzily_searchable :name, :class_name => 'CustomTrigram'
end

Speeding things up

For large data sets (millions of rows to index), the "compatible" storage used by default will typically no longer be enough to keep the index small enough.

Users have reported major improvements (2 order of magnitude) when turning the owner_type and fuzzy_field columns of the trigrams table from VARCHAR (the default) into ENUM. This is particularly efficient with MySQL and pgSQL.

This is not the default in the gem as ActiveRecord does not suport ENUM columns in any version.

UUID's

When using Rails 4 with UUID's, you will need to change the owner_id column type to UUID.

class AddTrigramsModel < ActiveRecord::Migration
  extend Fuzzily::Migration
  trigrams_owner_id_column_type = :uuid
end

Model primary key (id) is VARCHAR

If you set your Model primary key (id) AS VARCHAR instead of INT, you will need to change the owner_id column type from INT to VARCHAR in the trigrams table.

Searching virtual attributes

Your searchable fields do not have to be stored, they can be dynamic methods too. Just remember to add a virtual change method as well. For instance, if you model has first_name and last_name attributes, and you want to index a compound name dynamic attribute:

class Employee < ActiveRecord::Base
  fuzzily_searchable :name
  def name
    "#{first_name} #{last_name}"
  end

  def name_changed?
    first_name_changed? || last_name_changed?
  end
end

Update Trigram index using sidekiq-delay

For larger text, it takes time to build the index. Thus it can be moved into delay task using sidekiq + sidekiq-delay or delayed_job gem, both of them provide the method delay to move the execution to background thread by adding option async:

class Employee < ActiveRecord::Base
  fuzzily_searchable :name, async: true

end

License

MIT licence. Quite permissive if you ask me.

Copyright (c) 2013 HouseTrip Ltd.

Contributing

  1. Fork it
  2. Create your feature branch (git checkout -b my-new-feature)
  3. Commit your changes (git commit -am 'Add some feature')
  4. Push to the branch (git push origin my-new-feature)
  5. Create a new Pull Request

Thanks to @bclennox, @fdegiuli, @nickbender, @Shanison, @rickbutton for pointing out and/or helping on various issues.