patmcnally-amatch 0.2.4

data/tests/test_longest_substring.rb

@@ -0,0 +1,61 @@
+require 'test/unit'
+require 'amatch'
+
+class TestLongestSubstring < Test::Unit::TestCase
+  include Amatch
+
+  D = 0.000001
+
+  def setup
+    @small   = LongestSubstring.new('test')
+    @empty   = LongestSubstring.new('')
+    @long    = LongestSubstring.new('A' * 160)
+  end
+
+  def test_empty_substring
+    assert_in_delta 0, @empty.match(''), D
+    assert_in_delta 0, @empty.match('a'), D
+    assert_in_delta 0, @small.match(''), D
+    assert_in_delta 0, @empty.match('not empty'), D
+  end
+
+  def test_small_substring
+    assert_in_delta 4, @small.match('test'), D
+    assert_in_delta 4, @small.match('testa'), D
+    assert_in_delta 4, @small.match('atest'), D
+    assert_in_delta 2, @small.match('teast'), D
+    assert_in_delta 3, @small.match('est'), D
+    assert_in_delta 3, @small.match('tes'), D
+    assert_in_delta 2, @small.match('tst'), D
+    assert_in_delta 3, @small.match('best'), D
+    assert_in_delta 2, @small.match('tost'), D
+    assert_in_delta 3, @small.match('tesa'), D
+    assert_in_delta 1, @small.match('taex'), D
+    assert_in_delta 1, @small.match('aaatbbb'), D
+    assert_in_delta 1, @small.match('aaasbbb'), D
+    assert_in_delta 4, @small.match('aaatestbbb'), D
+  end
+
+  def test_similar
+    assert_in_delta 1, @empty.similar(''), D
+    assert_in_delta 0, @empty.similar('not empty'), D
+    assert_in_delta 0.0, @small.similar(''), D
+    assert_in_delta 1.0, @small.similar('test'), D
+    assert_in_delta 0.8, @small.similar('testa'), D
+    assert_in_delta 0.8, @small.similar('atest'), D
+    assert_in_delta 0.4, @small.similar('teast'), D
+    assert_in_delta 0.75, @small.similar('est'), D
+    assert_in_delta 0.75, @small.similar('tes'), D
+    assert_in_delta 0.5, @small.similar('tst'), D
+    assert_in_delta 0.75, @small.similar('best'), D
+    assert_in_delta 0.5, @small.similar('tost'), D
+    assert_in_delta 0.75, @small.similar('tesa'), D
+    assert_in_delta 0.25, @small.similar('taex'), D
+    assert_in_delta 0.4, @small.similar('aaatestbbb'), D
+    assert_in_delta 0.75, @small.pattern.longest_substring_similar('est'), D
+  end
+
+  def test_long
+    assert_in_delta 1.0, @long.similar(@long.pattern), D
+  end
+end

data/tests/test_pair_distance.rb

@@ -0,0 +1,86 @@
+require 'test/unit'
+require 'amatch'
+
+class TestPairDistance < Test::Unit::TestCase
+  include Amatch
+
+  D = 0.000001
+
+  def setup
+    @single   = PairDistance.new('test')
+    @empty    = PairDistance.new('')
+    @france   = PairDistance.new('republic of france')
+    @germany  = PairDistance.new('federal republic of germany')
+    @csv      = PairDistance.new('foo,bar,baz')
+    @long     = PairDistance.new('A' * 160)
+  end
+
+  def test_empty
+    assert_in_delta 1, @empty.match(''), D
+    assert_in_delta 0, @empty.match('not empty'), D
+    assert_in_delta 1, @empty.similar(''), D
+    assert_in_delta 0, @empty.similar('not empty'), D
+  end
+
+  def test_countries
+    assert_in_delta 0.5555555,  @france.match('france'), D
+    assert_in_delta 0.1052631,  @france.match('germany'), D
+    assert_in_delta 0.4615384,  @germany.match('germany'), D
+    assert_in_delta 0.16,       @germany.match('france'), D
+    assert_in_delta 0.6829268,
+      @germany.match('german democratic republic'), D
+    assert_in_delta 0.72,
+      @france.match('french republic'), D
+    assert_in_delta 0.4375,
+      @germany.match('french republic'), D
+    assert_in_delta 0.5294117,
+      @france.match('german democratic republic'), D
+  end
+
+  def test_single
+    assert_in_delta 0,          @single.match(''), D
+    assert_in_delta 1,          @single.match('test'), D
+    assert_in_delta 0.8571428,  @single.match('testa'), D
+    assert_in_delta 0.8571428,  @single.match('atest'), D
+    assert_in_delta 0.5714285,  @single.match('teast'), D
+    assert_in_delta 0.8,        @single.match('est'), D
+    assert_in_delta 0.8,        @single.match('tes'), D
+    assert_in_delta 0.4,        @single.match('tst'), D
+    assert_in_delta 0.6666666,  @single.match('best'), D
+    assert_in_delta 0.3333333,  @single.match('tost'), D
+    assert_in_delta 0.6666666,  @single.match('tesa'), D
+    assert_in_delta 0.0,        @single.match('taex'), D
+    assert_in_delta 0.5,        @single.match('aaatestbbb'), D
+    assert_in_delta 0.6,        @single.match('aaa test bbb'), D
+    assert_in_delta 0.6,        @single.match('test aaa bbb'), D
+    assert_in_delta 0.6,        @single.match('bbb aaa test'), D
+    assert_in_delta 0.8571428,  @single.pattern.pair_distance_similar('atest'), D
+  end
+
+  def test_csv
+    assert_in_delta 0,          @csv.match('', /,/), D
+    assert_in_delta 0.5,        @csv.match('foo', /,/), D
+    assert_in_delta 0.5,        @csv.match('bar', /,/), D
+    assert_in_delta 0.5,        @csv.match('baz', /,/), D
+    assert_in_delta 0.8,        @csv.match('foo,bar', /,/), D
+    assert_in_delta 0.8,        @csv.match('bar,foo', /,/), D
+    assert_in_delta 0.8,        @csv.match('bar,baz', /,/), D
+    assert_in_delta 0.8,        @csv.match('baz,bar', /,/), D
+    assert_in_delta 0.8,        @csv.match('foo,baz', /,/), D
+    assert_in_delta 0.8,        @csv.match('baz,foo', /,/), D
+    assert_in_delta 1,          @csv.match('foo,bar,baz', /,/), D
+    assert_in_delta 1,          @csv.match('foo,baz,bar', /,/), D
+    assert_in_delta 1,          @csv.match('baz,foo,bar', /,/), D
+    assert_in_delta 1,          @csv.match('baz,bar,foo', /,/), D
+    assert_in_delta 1,          @csv.match('bar,foo,baz', /,/), D
+    assert_in_delta 1,          @csv.match('bar,baz,foo', /,/), D
+    assert_in_delta 1,          @csv.match('foo,bar,baz', nil), D
+    assert_in_delta 0.9,        @csv.match('foo,baz,bar', nil), D
+    assert_in_delta 0.9,        @csv.match('foo,baz,bar'), D
+    assert_in_delta 0.9,        @csv.similar('foo,baz,bar'), D
+  end
+
+  def test_long
+    assert_in_delta 1.0, @long.similar(@long.pattern), D
+  end
+end

data/tests/test_sellers.rb

@@ -0,0 +1,96 @@
+require 'test/unit'
+require 'amatch'
+require 'test_levenshtein'
+
+class TestSellers < TestLevenshtein
+  include Amatch
+  
+  def setup
+    @d        = 0.000001
+    @empty    = Sellers.new('')
+    @simple   = Sellers.new('test')
+    @long     = Sellers.new('A' * 160)
+  end
+
+  def test_weights
+    assert_in_delta 1, @simple.substitution, @d
+    assert_in_delta 1, @simple.insertion, @d
+    assert_in_delta 1, @simple.deletion, @d
+    @simple.insertion = 1
+    @simple.substitution = @simple.deletion = 1000
+    assert_in_delta 1, @simple.match('tst'), @d
+    assert_in_delta 1, @simple.search('bbbtstccc'), @d
+    @simple.deletion = 1
+    @simple.substitution = @simple.insertion = 1000
+    assert_in_delta 1, @simple.match('tedst'), @d
+    assert_in_delta 1, @simple.search('bbbtedstccc'), @d
+    @simple.substitution = 1
+    @simple.deletion = @simple.insertion = 1000
+    assert_in_delta 1, @simple.match('tast'), @d
+    assert_in_delta 1, @simple.search('bbbtastccc'), @d
+    @simple.insertion = 0.5
+    @simple.substitution = @simple.deletion = 1000
+    assert_in_delta 0.5, @simple.match('tst'), @d
+    assert_in_delta 0.5, @simple.search('bbbtstccc'), @d
+    @simple.deletion = 0.5
+    @simple.substitution = @simple.insertion = 1000
+    assert_in_delta 0.5, @simple.match('tedst'), @d
+    assert_in_delta 0.5, @simple.search('bbbtedstccc'), @d
+    @simple.substitution = 0.5
+    @simple.deletion = @simple.insertion = 1000
+    assert_in_delta 0.5, @simple.match('tast'), @d
+    assert_in_delta 0.5, @simple.search('bbbtastccc'), @d
+    @simple.reset_weights
+    assert_in_delta 1, @simple.substitution, @d
+    assert_in_delta 1, @simple.insertion, @d
+    assert_in_delta 1, @simple.deletion, @d
+  end
+
+  def test_weight_exceptions
+    assert_raises(TypeError) { @simple.substitution = :something }
+    assert_raises(TypeError) { @simple.insertion = :something }
+    assert_raises(TypeError) { @simple.deletion = :something }
+  end
+
+  def test_similar
+    assert_in_delta 0.0, @simple.similar(''), @d
+    assert_in_delta 1.0, @simple.similar('test'), @d
+    assert_in_delta 0.8, @simple.similar('testa'), @d
+    assert_in_delta 0.8, @simple.similar('atest'), @d
+    assert_in_delta 0.8, @simple.similar('teast'), @d
+    assert_in_delta 0.75, @simple.similar('est'), @d
+    assert_in_delta 0.75, @simple.similar('tes'), @d
+    assert_in_delta 0.75, @simple.similar('tst'), @d
+    assert_in_delta 0.75, @simple.similar('best'), @d
+    assert_in_delta 0.75, @simple.similar('tost'), @d
+    assert_in_delta 0.75, @simple.similar('tesa'), @d
+    assert_in_delta 0.25, @simple.similar('taex'), @d
+    assert_in_delta 0.4, @simple.similar('aaatestbbb'), @d
+    assert_in_delta 0.75, @simple.pattern.levenshtein_similar('est'), @d
+  end
+
+  def test_similar
+    assert_in_delta 1, @empty.similar(''), @d
+    assert_in_delta 0, @empty.similar('not empty'), @d
+    assert_in_delta 0.0, @simple.similar(''), @d
+    assert_in_delta 1.0, @simple.similar('test'), @d
+    assert_in_delta 0.8, @simple.similar('testa'), @d
+    assert_in_delta 0.8, @simple.similar('atest'), @d
+    assert_in_delta 0.8, @simple.similar('teast'), @d
+    assert_in_delta 0.75, @simple.similar('est'), @d
+    assert_in_delta 0.75, @simple.similar('tes'), @d
+    assert_in_delta 0.75, @simple.similar('tst'), @d
+    assert_in_delta 0.75, @simple.similar('best'), @d
+    assert_in_delta 0.75, @simple.similar('tost'), @d
+    assert_in_delta 0.75, @simple.similar('tesa'), @d
+    assert_in_delta 0.25, @simple.similar('taex'), @d
+    assert_in_delta 0.4, @simple.similar('aaatestbbb'), @d
+    @simple.insertion = 1
+    @simple.substitution = @simple.deletion = 2
+    assert_in_delta 0.875, @simple.similar('tst'), @d
+  end
+
+  def test_long
+    assert_in_delta 1.0, @long.similar(@long.pattern), @d
+  end
+end

metadata

@@ -0,0 +1,85 @@
+--- !ruby/object:Gem::Specification 
+name: patmcnally-amatch
+version: !ruby/object:Gem::Version 
+  version: 0.2.4
+platform: ruby
+authors: 
+- Florian Frank
+autorequire: 
+bindir: bin
+cert_chain: []
+
+date: 2009-05-16 00:00:00 -07:00
+default_executable: agrep.rb
+dependencies: []
+
+description: Amatch is a library for approximate string matching and searching in strings. Several algorithms can be used to do this, and it's also possible to compute a similarity metric number between 0.0 and 1.0 for two given strings.
+email: flori@ping.de
+executables: 
+- agrep.rb
+extensions: 
+- ext/extconf.rb
+extra_rdoc_files: 
+- ext/amatch.c
+- lib/amatch/version.rb
+- doc-main.txt
+files: 
+- amatch.gemspec
+- bin
+- bin/agrep.rb
+- CHANGES
+- COPYING
+- ext
+- ext/amatch.c
+- ext/common.h
+- ext/extconf.rb
+- ext/pair.c
+- ext/pair.h
+- install.rb
+- lib
+- lib/amatch
+- lib/amatch/version.rb
+- Rakefile
+- README
+- tests
+- tests/test_hamming.rb
+- tests/test_jaro.rb
+- tests/test_jaro_winkler.rb
+- tests/test_levenshtein.rb
+- tests/test_longest_subsequence.rb
+- tests/test_longest_substring.rb
+- tests/test_pair_distance.rb
+- tests/test_sellers.rb
+- VERSION
+- doc-main.txt
+has_rdoc: true
+homepage: http://amatch.rubyforge.org
+post_install_message: 
+rdoc_options: 
+- --main
+- doc-main.txt
+- --title
+- amatch - Approximate Matching
+require_paths: 
+- ext
+required_ruby_version: !ruby/object:Gem::Requirement 
+  requirements: 
+  - - ">="
+    - !ruby/object:Gem::Version 
+      version: "0"
+  version: 
+required_rubygems_version: !ruby/object:Gem::Requirement 
+  requirements: 
+  - - ">="
+    - !ruby/object:Gem::Version 
+      version: "0"
+  version: 
+requirements: []
+
+rubyforge_project: amatch
+rubygems_version: 1.2.0
+signing_key: 
+specification_version: 2
+summary: Approximate String Matching library
+test_files: []
+