What is fuzzy-string-match

Build Status

  • fuzzy-string-match is a fuzzy string matching library for ruby.
  • It is fast. ( written in C with RubyInline )
  • It supports only Jaro-Winkler distance algorithm.
  • This program was ported by hand from lucene-3.0.2. (lucene is Java product)
  • If you want to add another string distance algorithm, please fork it on github and port by yourself.

The reason why i developed fuzzy-string-match

  • I tried amatch-0.2.5, but it contains some issues.
    1. memory leaks.
    2. I felt difficult to maintain it.
  • So, I decide to create another gem by porting lucene-3.0.x.

Installing

gem install fuzzy-string-match

Features

  • Calculate Jaro-Winkler distance of two strings.
    • Pure ruby version can handle both ASCII and UTF8 strings. (and slow)
    • Native version can only ASCII strings. (but it is fast)

Sample code

Native version

require 'fuzzystringmatch'
jarow = FuzzyStringMatch::JaroWinkler.create( :native )
p jarow.getDistance(  "jones",      "johnson" )

Pure ruby version

require 'fuzzystringmatch'
jarow = FuzzyStringMatch::JaroWinkler.create( :pure )
p jarow.getDistance(  "jones",      "johnson" )
p jarow.getDistance(  "ああ",        "あい"        )

Sample on irb

irb(main):001:0> require 'fuzzystringmatch'
require 'fuzzystringmatch'
=> true

irb(main):002:0> jarow = FuzzyStringMatch::JaroWinkler.create( :native )
jarow = FuzzyStringMatch::JaroWinkler.create( :native )
=> #<FuzzyStringMatch::JaroWinklerNative:0x000001011b0010>

irb(main):003:0> jarow.getDistance( "al",        "al"        )
jarow.getDistance( "al",        "al"        )
=> 1.0

irb(main):004:0> jarow.getDistance( "dixon",     "dicksonx"  )
jarow.getDistance( "dixon",     "dicksonx"  )
=> 0.8133333333333332

Benchmarks

$ rake bench
ruby ./benchmark/vs_amatch.rb
 --- 
 --- Each match functions will be called 1Mega times. --- 
 --- 
[Amatch]
      user     system      total        real
  1.160000   0.050000   1.210000 (  1.218259)
[this Module (pure)]
      user     system      total        real
 39.940000   0.160000  40.100000 ( 40.542448)
[this Module (native)]
      user     system      total        real
  0.480000   0.000000   0.480000 (  0.484187)

Requires

for CRuby

  • RubyInline
  • Ruby 2.0.0 or higher ( includes RubyInstaller.org's CRuby on Windows )

for JRuby

  • JRuby 1.6.6 or higher

Author

  • Copyright (C) Kiyoka Nishiyama [email protected]
  • I ported from java source code of lucene-3.0.2.

See also

License

  • Apache 2.0 LICENSE

ChangeLog

1.0.1 / Jun 25, 2017

  • support JRuby 1.7.26(CRuby 1.9 compatible)

1.0.0 / Mar 10, 2017

  • First stable release

0.9.9 / Mar 9, 2017

  • Supported ruby version is 2.0.0 or higher(for RHEL 7.x)

0.9.8 / Mar 9, 2017

  • Supported ruby version is 2.1.0 or higher
  • Merge pull request #16 from ferdinandrosario/ferdinandrosario-patch-1 (Travis rubies updated)
  • Merge pull request #14 from timsatterfield/master (Reduce calls to strlen() in native jaro winkler)

0.9.7 / Oct 15, 2014

  • Use rspec 3.1 syntax.
  • Fixed: issue #12 Using stack allocated memory.
  • Fixed: remove duplicated dependency of gem package.

0.9.6 / Dec 21, 2013

  • New feature: fuzzy-string-match falls back into pure ruby mode when c-compile fails.
  • fuzzy-string-match_pure is obsolute gem

0.9.5 / Mar 26, 2013

  • Fixed: 'jarowinkler.rb:42: warning: implicit conversion shortens 64-bit value into a 32-bit value' on MacOS X 64bit env.

0.9.4 / July 10, 2012

  • Fixed: undefined method getDistance' error.

0.9.3 / Feb 27, 2012

  • Changed gem dependency of `rspec'. gemspec.dependency( "rspec" ) to gemspec.development_dependency( "rspec" )

0.9.2 / Feb 17, 2012

  • Supported JRuby platform
  • Divided into two gems.
  1. fuzzy-string-match ... native (RubyInline) version.
  2. fuzzy-string-match_pure ... pure ruby version

  • Divided rspec files into several files.

  • Supported testable gem Please install rubygems-test and "gem test".

0.9.1 / Jul 30, 2011

  • Changed gcc compiler's option for RubyInline.

  • Stoped to use obsolute method of RSpec.

0.9.0 / Oct 12, 2010

  • First release.