fuzzy_match 1.0.5

data/test/test_cache.rb

@@ -0,0 +1,130 @@
+require 'helper'
+
+require 'active_support/all'
+require 'active_record'
+require 'cohort_scope'
+require 'weighted_average'
+
+ActiveRecord::Base.establish_connection(
+  'adapter' => 'mysql',
+  'database' => 'fuzzy_match_test',
+  'username' => 'root',
+  'password' => 'password'
+)
+
+# ActiveRecord::Base.logger = Logger.new $stderr
+
+ActiveSupport::Inflector.inflections do |inflect|
+  inflect.uncountable 'aircraft'
+end
+
+require 'fuzzy_match/cached_result'
+
+::FuzzyMatch::CachedResult.setup(true)
+::FuzzyMatch::CachedResult.delete_all
+
+class Aircraft < ActiveRecord::Base
+  set_primary_key :icao_code
+  
+  cache_fuzzy_match_matches_with :flight_segments, :primary_key => :aircraft_description, :foreign_key => :aircraft_description
+    
+  def aircraft_description
+    [manufacturer_name, model_name].compact.join(' ')
+  end
+  
+  def self.fuzzy_match
+    @fuzzy_match ||= FuzzyMatch.new all, :read => ::Proc.new { |straw| straw.aircraft_description }
+  end
+  
+  def self.create_table
+    connection.drop_table(:aircraft) rescue nil
+    connection.execute %{
+CREATE TABLE `aircraft` (
+  `icao_code` varchar(255) DEFAULT NULL,
+  `manufacturer_name` varchar(255) DEFAULT NULL,
+  `model_name` varchar(255) DEFAULT NULL,
+  PRIMARY KEY (`icao_code`)
+) ENGINE=InnoDB DEFAULT CHARSET=utf8;
+    }
+    reset_column_information
+  end
+end
+
+class FlightSegment < ActiveRecord::Base
+  set_primary_key :row_hash
+  
+  cache_fuzzy_match_matches_with :aircraft, :primary_key => :aircraft_description, :foreign_key => :aircraft_description
+  
+  extend CohortScope
+  self.minimum_cohort_size = 1
+  
+  def self.create_table
+    connection.drop_table(:flight_segments) rescue nil
+    connection.execute %{
+CREATE TABLE `flight_segments` (
+  `row_hash` varchar(255) NOT NULL DEFAULT '',
+  `aircraft_description` varchar(255) DEFAULT NULL,
+  `passengers` int(11) DEFAULT NULL,
+  `seats` int(11) DEFAULT NULL,
+  PRIMARY KEY (`row_hash`)
+) ENGINE=InnoDB DEFAULT CHARSET=utf8;
+    }
+  end
+end
+
+FlightSegment.create_table
+Aircraft.create_table
+
+a = Aircraft.new
+a.icao_code = 'B742'
+a.manufacturer_name = 'Boeing'
+a.model_name = '747-200'
+a.save!
+
+fs = FlightSegment.new
+fs.row_hash = 'madison to chicago'
+fs.aircraft_description = 'BORING 747200'
+fs.passengers = 10
+fs.seats = 10
+fs.save!
+
+fs = FlightSegment.new
+fs.row_hash = 'madison to minneapolis'
+fs.aircraft_description = 'bing 747'
+fs.passengers = 100
+fs.seats = 5
+fs.save!
+
+FlightSegment.find_each do |fs|
+  fs.cache_aircraft!
+end
+
+class TestCache < Test::Unit::TestCase
+  def test_002_one_degree_of_separation
+    aircraft = Aircraft.find('B742')
+    assert_equal 2, aircraft.flight_segments.count
+  end
+  
+  def test_003_standard_sql_calculations
+    aircraft = Aircraft.find('B742')
+    assert_equal 110, aircraft.flight_segments.sum(:passengers)
+  end
+  
+  def test_004_weighted_average
+    aircraft = Aircraft.find('B742')
+    assert_equal 5.45455, aircraft.flight_segments.weighted_average(:seats, :weighted_by => :passengers)
+  end
+  
+  def test_005_right_way_to_do_cohorts
+    aircraft = Aircraft.find('B742')
+    assert_equal 2, FlightSegment.big_cohort(:aircraft_description => aircraft.flight_segments_foreign_keys).count
+  end
+  
+  def test_006_you_can_get_aircraft_from_flight_segments
+    fs = FlightSegment.first
+    # you need to add an aircraft_description column
+    assert_raises(ActiveRecord::StatementInvalid) do
+      assert_equal 2, fs.aircraft.count
+    end
+  end
+end

data/test/test_fuzzy_match.rb

@@ -0,0 +1,190 @@
+# -*- encoding: utf-8 -*-
+require 'helper'
+
+class TestFuzzyMatch < Test::Unit::TestCase
+  def test_001_find
+    d = FuzzyMatch.new %w{ RATZ CATZ }
+    assert_equal 'RATZ', d.find('RITZ')
+    assert_equal 'RATZ', d.find('RíTZ')
+    
+    d = FuzzyMatch.new [ 'X' ]
+    assert_equal 'X', d.find('X')
+    assert_equal nil, d.find('A')
+  end
+  
+  def test_002_dont_gather_last_result_by_default
+    d = FuzzyMatch.new %w{ NISSAN HONDA }
+    d.find('MISSAM')
+    assert_raises(::RuntimeError, /gather_last_result/) do
+      d.last_result
+    end
+  end
+  
+  def test_003_last_result
+    d = FuzzyMatch.new %w{ NISSAN HONDA }
+    d.find 'MISSAM', :gather_last_result => true
+    assert_equal 0.6, d.last_result.score
+    assert_equal 'NISSAN', d.last_result.record
+  end
+  
+  def test_004_false_positive_without_tightener
+    d = FuzzyMatch.new ['BOEING 737-100/200', 'BOEING 737-900']
+    assert_equal 'BOEING 737-900', d.find('BOEING 737100 number 900')
+  end
+  
+  def test_005_correct_with_tightener
+    tighteners = [
+      %r{(7\d)(7|0)-?(\d{1,3})} # tighten 737-100/200 => 737100, which will cause it to win over 737-900
+    ]
+    d = FuzzyMatch.new ['BOEING 737-100/200', 'BOEING 737-900'], :tighteners => tighteners
+    assert_equal 'BOEING 737-100/200', d.find('BOEING 737100 number 900')
+  end
+  
+  def test_008_false_positive_without_identity
+    d = FuzzyMatch.new %w{ foo bar }
+    assert_equal 'bar', d.find('baz')
+  end
+  
+  def test_008_identify_false_positive
+    d = FuzzyMatch.new %w{ foo bar }, :identities => [ /ba(.)/ ]
+    assert_equal nil, d.find('baz')
+  end
+  
+  # TODO this is not very helpful
+  def test_009_blocking
+    d = FuzzyMatch.new [ 'X' ], :blockings => [ /X/, /Y/ ]
+    assert_equal 'X', d.find('X')
+    assert_equal nil, d.find('A')
+  end
+  
+  # TODO this is not very helpful
+  def test_0095_must_match_blocking
+    d = FuzzyMatch.new [ 'X' ], :blockings => [ /X/, /Y/ ], :must_match_blocking => true
+    assert_equal 'X', d.find('X')
+    assert_equal nil, d.find('A')
+    
+    d = FuzzyMatch.new [ 'X' ], :blockings => [ /X/, /Y/ ]
+    assert_equal 'X', d.find('X', :must_match_blocking => true)
+    assert_equal nil, d.find('A', :must_match_blocking => true)
+  end
+    
+  def test_011_free
+    d = FuzzyMatch.new %w{ NISSAN HONDA }
+    d.free
+    assert_raises(::RuntimeError, /free/) do
+      d.find('foobar')
+    end
+  end
+  
+  def test_012_find_all
+    d = FuzzyMatch.new [ 'X', 'X22', 'Y', 'Y4' ], :blockings => [ /X/, /Y/ ], :must_match_blocking => true
+    assert_equal ['X', 'X22' ], d.find_all('X')
+    assert_equal [], d.find_all('A')
+  end
+  
+  def test_013_first_blocking_decides
+    d = FuzzyMatch.new [ 'Boeing 747', 'Boeing 747SR', 'Boeing ER6' ], :blockings => [ /(boeing \d{3})/i, /boeing/i ]
+    assert_equal [ 'Boeing 747', 'Boeing 747SR', 'Boeing ER6' ], d.find_all('Boeing 747')
+    
+    d = FuzzyMatch.new [ 'Boeing 747', 'Boeing 747SR', 'Boeing ER6' ], :blockings => [ /(boeing \d{3})/i, /boeing/i ], :first_blocking_decides => true
+    assert_equal [ 'Boeing 747', 'Boeing 747SR' ], d.find_all('Boeing 747')
+    
+    # first_blocking_decides refers to the needle
+    d = FuzzyMatch.new [ 'Boeing 747', 'Boeing 747SR', 'Boeing ER6' ], :blockings => [ /(boeing \d{3})/i, /boeing/i ], :first_blocking_decides => true
+    assert_equal [ 'Boeing 747', 'Boeing 747SR', 'Boeing ER6' ], d.find_all('Boeing ER6')
+    
+    d = FuzzyMatch.new [ 'Boeing 747', 'Boeing 747SR', 'Boeing ER6' ], :blockings => [ /(boeing \d{3})/i, /boeing (7|E)/i, /boeing/i ], :first_blocking_decides => true
+    assert_equal [ 'Boeing ER6' ], d.find_all('Boeing ER6')
+  
+    # or equivalently with an identity
+    d = FuzzyMatch.new [ 'Boeing 747', 'Boeing 747SR', 'Boeing ER6' ], :blockings => [ /(boeing \d{3})/i, /boeing/i ], :first_blocking_decides => true, :identities => [ /boeing (7|E)/i ]
+    assert_equal [ 'Boeing ER6' ], d.find_all('Boeing ER6')
+  end
+  
+  MyStruct = Struct.new(:one, :two)
+  def test_014_symbol_read_sends_method
+    ab = MyStruct.new('a', 'b')
+    ba = MyStruct.new('b', 'a')
+    haystack = [ab, ba]
+    by_first = FuzzyMatch.new haystack, :read => :one
+    by_last = FuzzyMatch.new haystack, :read => :two
+    assert_equal ab, by_first.find('a')
+    assert_equal ab, by_last.find('b')
+    assert_equal ba, by_first.find('b')
+    assert_equal ba, by_last.find('a')
+  end
+  
+  def test_015_symbol_read_reads_array
+    ab = ['a', 'b']
+    ba = ['b', 'a']
+    haystack = [ab, ba]
+    by_first = FuzzyMatch.new haystack, :read => 0
+    by_last = FuzzyMatch.new haystack, :read => 1
+    assert_equal ab, by_first.find('a')
+    assert_equal ab, by_last.find('b')
+    assert_equal ba, by_first.find('b')
+    assert_equal ba, by_last.find('a')
+  end
+  
+  def test_016_symbol_read_reads_hash
+    ab = { :one => 'a', :two => 'b' }
+    ba = { :one => 'b', :two => 'a' }
+    haystack = [ab, ba]
+    by_first = FuzzyMatch.new haystack, :read => :one
+    by_last = FuzzyMatch.new haystack, :read => :two
+    assert_equal ab, by_first.find('a')
+    assert_equal ab, by_last.find('b')
+    assert_equal ba, by_first.find('b')
+    assert_equal ba, by_last.find('a')
+  end
+  
+  def test_017_understands_haystack_reader_option
+    ab = ['a', 'b']
+    ba = ['b', 'a']
+    haystack = [ab, ba]
+    by_first = FuzzyMatch.new haystack, :haystack_reader => 0
+    assert_equal ab, by_first.find('a')
+    assert_equal ba, by_first.find('b')
+  end
+  
+  def test_018_no_result_if_best_score_is_zero
+    assert_equal nil, FuzzyMatch.new(['a']).find('b')
+  end
+  
+  def test_019_must_match_at_least_one_word
+    d = FuzzyMatch.new %w{ RATZ CATZ }, :must_match_at_least_one_word => true
+    assert_equal nil, d.find('RITZ')
+  end
+  
+  def test_020_stop_words
+    d = FuzzyMatch.new [ 'A HOTEL', 'B HTL' ]
+    assert_equal 'B HTL', d.find('A HTL', :must_match_at_least_one_word => true)
+    
+    d = FuzzyMatch.new [ 'A HOTEL', 'B HTL' ], :must_match_at_least_one_word => true
+    assert_equal 'B HTL', d.find('A HTL')
+    
+    d = FuzzyMatch.new [ 'A HOTEL', 'B HTL' ], :must_match_at_least_one_word => true, :stop_words => [ %r{HO?TE?L} ]
+    assert_equal 'A HOTEL', d.find('A HTL')
+  end
+  
+  def test_021_explain
+    require 'stringio'
+    capture = StringIO.new
+    begin
+      old_stderr = $stderr
+      $stderr = capture
+      d = FuzzyMatch.new %w{ RATZ CATZ }
+      d.explain('RITZ')
+    ensure
+      $stderr = old_stderr
+    end
+    capture.rewind
+    assert capture.read.include?('CATZ')
+    capture.close
+  end
+  
+  def test_022_compare_words_with_words
+    d = FuzzyMatch.new [ 'PENINSULA HOTELS' ], :must_match_at_least_one_word => true
+    assert_equal nil, d.find('DOLCE LA HULPE BXL FI')
+  end
+end

data/test/test_fuzzy_match_convoluted.rb.disabled

@@ -0,0 +1,268 @@
+require 'helper'
+
+require 'shoulda'
+
+$log = false
+
+class TestFuzzyMatchConvoluted < Test::Unit::TestCase
+  def setup
+    clear_ltd
+    
+    # dh 8 400
+    @a_needle = ['DE HAVILLAND CANADA DHC8400 Dash 8']
+    @a_haystack = ['DEHAVILLAND DEHAVILLAND DHC8-400 DASH-8']
+    # dh 88
+    @b_needle = ['ABCDEFG DH88 HIJKLMNOP']
+    # dh 89
+    @c_haystack = ['ABCDEFG DH89 HIJKLMNOP']
+    # dh 8 200
+    @d_needle = ['DE HAVILLAND CANADA DHC8200 Dash 8']
+    @d_haystack = ['BOMBARDIER DEHAVILLAND DHC8-200Q DASH-8']
+    @d_lookalike = ['ABCD DHC8200 Dash 8']
+    
+    @t_1 = [ '/(dh)c?-?(\d{0,2})-?(\d{0,4})(?:.*?)(dash|\z)/i', 'good tightening for de havilland' ]
+    
+    @r_1 = [ '/(dh)c?-?(\d{0,2})-?(\d{0,4})(?:.*?)(dash|\z)/i', 'good identity for de havilland' ]
+    
+    @needle = [
+      @a_needle,
+      @b_needle,
+      ['DE HAVILLAND DH89 Dragon Rapide'],
+      ['DE HAVILLAND CANADA DHC8100 Dash 8 (E9, CT142, CC142)'],
+      @d_needle,
+      ['DE HAVILLAND CANADA DHC8300 Dash 8'],
+      ['DE HAVILLAND DH90 Dragonfly']
+    ]
+    @haystack = [
+      @a_haystack,
+      @c_haystack,
+      @d_haystack,
+      ['DEHAVILLAND DEHAVILLAND DHC8-100 DASH-8'],
+      ['DEHAVILLAND DEHAVILLAND TWIN OTTER DHC-6']
+    ]
+    @tightenings = []
+    @identities = []
+    @blockings = []
+    @positives = []
+    @negatives = []
+  end
+
+  def clear_ltd
+    @_ltd = nil
+  end
+  
+  def ltd
+    @_ltd ||= FuzzyMatch.new  @haystack,
+                                        :tightenings => @tightenings,
+                                        :identities => @identities,
+                                        :blockings => @blockings,
+                                        :positives => @positives,
+                                        :negatives => @negatives,
+                                        :blocking_only => @blocking_only,
+                                        :log => $log
+  end
+
+  should "optionally only pay attention to things that match blockings" do
+    assert_equal @a_haystack, ltd.improver.match(@a_needle)
+
+    clear_ltd
+    @blocking_only = true
+    assert_equal nil, ltd.improver.match(@a_needle)
+
+    clear_ltd
+    @blocking_only = true
+    @blockings.push ['/dash/i']
+    assert_equal @a_haystack, ltd.improver.match(@a_needle)
+  end
+  
+  # the example from the readme, considerably uglier here
+  should "check a simple table" do
+    @haystack = [ 'seamus', 'andy', 'ben' ]
+    @positives = [ [ 'seamus', 'Mr. Seamus Abshere' ] ]
+    needle = [ 'Mr. Seamus Abshere', 'Sr. Andy Rossmeissl', 'Master BenT' ]
+  
+    assert_nothing_raised do
+      ltd.improver.check needle
+    end
+  end
+  
+  should "treat a String as a full record if passed through" do
+    dash = 'DHC8-400'
+    b747 = 'B747200/300'
+    dc9 = 'DC-9-10'
+    haystack_records = [ dash, b747, dc9 ]
+    simple_ltd = FuzzyMatch.new haystack_records, :log => $log
+    assert_equal dash, simple_ltd.improver.match('DeHavilland Dash-8 DHC-400')
+    assert_equal b747, simple_ltd.improver.match('Boeing 747-300')
+    assert_equal dc9, simple_ltd.improver.match('McDonnell Douglas MD81/DC-9')
+  end
+  
+  should "call it a mismatch if you hit a blank positive" do
+    @positives.push [@a_needle[0], '']
+    assert_raises(FuzzyMatch::Improver::Mismatch) do
+      ltd.improver.match @a_needle
+    end
+  end
+
+  should "call it a false positive if you hit a blank negative" do
+    @negatives.push [@a_needle[0], '']
+    assert_raises(FuzzyMatch::Improver::FalsePositive) do
+      ltd.improver.match @a_needle
+    end
+  end
+  
+  should "have a false match without blocking" do
+    # @d_needle will be our victim
+    @haystack.push @d_lookalike
+    @tightenings.push @t_1
+    
+    assert_equal @d_lookalike, ltd.improver.match(@d_needle)
+  end
+  
+  should "do blocking if the needle matches a block" do
+    # @d_needle will be our victim
+    @haystack.push @d_lookalike
+    @tightenings.push @t_1
+    @blockings.push ['/(bombardier|de ?havilland)/i']
+    
+    assert_equal @d_haystack, ltd.improver.match(@d_needle)
+  end
+  
+  should "treat blocks as exclusive" do
+    @haystack = [ @d_needle ]
+    @tightenings.push @t_1
+    @blockings.push ['/(bombardier|de ?havilland)/i']
+
+    assert_equal nil, ltd.improver.match(@d_lookalike)
+  end
+  
+  should "only use identities if they stem from the same regexp" do
+    @identities.push @r_1
+    @identities.push [ '/(cessna)(?:.*?)(citation)/i' ]
+    @identities.push [ '/(cessna)(?:.*?)(\d\d\d)/i' ]
+    x_needle = [ 'CESSNA D-333 CITATION V']
+    x_haystack = [ 'CESSNA D-333' ]
+    @haystack.push x_haystack
+    
+    assert_equal x_haystack, ltd.improver.match(x_needle)
+  end
+  
+  should "use the best score from all of the tightenings" do
+    x_needle = ["BOEING 737100"]
+    x_haystack = ["BOEING BOEING 737-100/200"]
+    x_haystack_wrong = ["BOEING BOEING 737-900"]
+    @haystack.push x_haystack
+    @haystack.push x_haystack_wrong
+    @tightenings.push ['/(7\d)(7|0)-?\d{1,3}\/(\d\d\d)/i']
+    @tightenings.push ['/(7\d)(7|0)-?(\d{1,3}|[A-Z]{0,3})/i']
+    
+    assert_equal x_haystack, ltd.improver.match(x_needle)
+  end
+  
+  should "compare using prefixes if tightened key is shorter than correct match" do
+    x_needle = ["BOEING 720"]
+    x_haystack = ["BOEING BOEING 720-000"]
+    x_haystack_wrong = ["BOEING BOEING 717-200"]
+    @haystack.push x_haystack
+    @haystack.push x_haystack_wrong
+    @tightenings.push @t_1
+    @tightenings.push ['/(7\d)(7|0)-?\d{1,3}\/(\d\d\d)/i']
+    @tightenings.push ['/(7\d)(7|0)-?(\d{1,3}|[A-Z]{0,3})/i']
+    
+    assert_equal x_haystack, ltd.improver.match(x_needle)
+  end
+  
+  should "use the shortest original input" do
+    x_needle = ['De Havilland DHC8-777 Dash-8 Superstar']
+    x_haystack = ['DEHAVILLAND DEHAVILLAND DHC8-777 DASH-8 Superstar']
+    x_haystack_long = ['DEHAVILLAND DEHAVILLAND DHC8-777 DASH-8 Superstar/Supernova']
+    
+    @haystack.push x_haystack_long
+    @haystack.push x_haystack
+    @tightenings.push @t_1
+    
+    assert_equal x_haystack, ltd.improver.match(x_needle)
+  end
+  
+  should "perform lookups needle to haystack" do
+    assert_equal @a_haystack, ltd.improver.match(@a_needle)
+  end
+
+  should "succeed if there are no checks" do
+    assert_nothing_raised do
+      ltd.improver.check @needle
+    end
+  end
+
+  should "succeed if the positive checks just work" do
+    @positives.push [ @a_needle[0], @a_haystack[0] ]
+  
+    assert_nothing_raised do
+      ltd.improver.check @needle
+    end
+  end
+
+  should "fail if positive checks don't work" do
+    @positives.push [ @d_needle[0], @d_haystack[0] ]
+
+    assert_raises(FuzzyMatch::Improver::Mismatch) do
+      ltd.improver.check @needle
+    end
+  end
+
+  should "succeed if proper tightening is applied" do
+    @positives.push [ @d_needle[0], @d_haystack[0] ]
+    @tightenings.push @t_1
+
+    assert_nothing_raised do
+      ltd.improver.check @needle
+    end
+  end
+
+  should "use a Google Docs spreadsheet as a source of tightenings" do
+    @positives.push [ @d_needle[0], @d_haystack[0] ]
+    @tightenings = RemoteTable.new :url => 'http://spreadsheets.google.com/pub?key=tiS_6CCDDM_drNphpYwE_iw&single=true&gid=0&output=csv', :headers => false
+  
+    # sabshere 9/30/10 this shouldn't raise anything
+    # but the tightenings have been changed... we should be using test-only tightenings, not production ones
+    # assert_nothing_raised do
+    assert_raises(FuzzyMatch::Improver::Mismatch) do
+      ltd.improver.check @needle
+    end
+  end
+  
+  should "fail if negative checks don't work" do
+    @negatives.push [ @b_needle[0], @c_haystack[0] ]
+  
+    assert_raises(FuzzyMatch::Improver::FalsePositive) do
+      ltd.improver.check @needle
+    end
+  end
+  
+  should "do inline checking" do
+    @negatives.push [ @b_needle[0], @c_haystack[0] ]
+  
+    assert_raises(FuzzyMatch::Improver::FalsePositive) do
+      ltd.improver.match @b_needle
+    end
+  end
+
+  should "fail if negative checks don't work, even with tightening" do
+    @negatives.push [ @b_needle[0], @c_haystack[0] ]
+    @tightenings.push @t_1
+  
+    assert_raises(FuzzyMatch::Improver::FalsePositive) do
+      ltd.improver.check @needle
+    end
+  end
+
+  should "succeed if proper identity is applied" do
+    @negatives.push [ @b_needle[0], @c_haystack[0] ]
+    @positives.push [ @d_needle[0], @d_haystack[0] ]
+    @identities.push @r_1
+  
+    assert_nothing_raised do
+      ltd.improver.check @needle
+    end
+  end
+end

data/test/test_identity.rb

@@ -0,0 +1,33 @@
+require 'helper'
+
+class TestIdentity < Test::Unit::TestCase
+  def test_001_identical
+    i = FuzzyMatch::Identity.new %r{(A)[ ]*(\d)}
+    assert_equal true, i.identical?('A1', 'A     1foobar')
+  end
+  
+  def test_002_certainly_different
+    i = FuzzyMatch::Identity.new %r{(A)[ ]*(\d)}
+    assert_equal false, i.identical?('A1', 'A     2foobar')    
+  end
+  
+  def test_003_no_information_ie_possible_identical
+    i = FuzzyMatch::Identity.new %r{(A)[ ]*(\d)}
+    assert_equal nil, i.identical?('B1', 'A     2foobar')    
+  end
+
+  def test_004_regexp
+    i = FuzzyMatch::Identity.new %r{\A\\?/(.*)etc/mysql\$$}
+    assert_equal %r{\A\\?/(.*)etc/mysql\$$}, i.regexp
+  end
+  
+  def test_005_regexp_from_string
+    i = FuzzyMatch::Identity.new '%r{\A\\\?/(.*)etc/mysql\$$}'
+    assert_equal %r{\A\\?/(.*)etc/mysql\$$}, i.regexp
+  end
+  
+  def test_006_regexp_from_string_using_slash_delim
+    i = FuzzyMatch::Identity.new '/\A\\\?\/(.*)etc\/mysql\$$/'
+    assert_equal %r{\A\\?/(.*)etc/mysql\$$}, i.regexp
+  end
+end

data/test/test_tightening.rb

@@ -0,0 +1,10 @@
+require 'helper'
+
+class TestTightener < Test::Unit::TestCase
+  def test_001_apply
+    t = FuzzyMatch::Tightener.new %r{(Ford )[ ]*(F)[\- ]*(\d\d\d)}i
+    assert_equal 'Ford F350', t.apply('Ford F-350')
+    assert_equal 'Ford F150', t.apply('Ford F150')
+    assert_equal 'Ford F350', t.apply('Ford F 350')
+  end
+end