Text 1.0.0

data/test/data/metaphone.txt

@@ -0,0 +1,51 @@
+#
+# Based on the table at http://aspell.net/metaphone/metaphone-kuhn.txt,
+# with surprising results changed to 'correct' ones (according to my interpretation
+# of the algorithm description), and some more results from around the web:
+#
+ANASTHA: ANS0
+DAVIS-CARTER: TFSKRTR
+ESCARMANT: ESKRMNT
+MCCALL: MKL
+MCCROREY: MKRR
+MERSEAL: MRSL
+PIEURISSAINT: PRSNT
+ROTMAN: RTMN
+SCHEVEL: SXFL
+SCHROM: SXRM
+SEAL: SL
+SPARR: SPR
+STARLEPER: STRLPR
+THRASH: 0RX
+LOGGING: LKNK
+LOGIC: LJK
+JUDGES: JJS
+SHOOS: XS
+SHOES: XS
+CHUTE: XT
+SCHUSS: SXS
+OTTO: OT
+ERIC: ERK
+DAVE: TF
+CATHERINE: K0RN
+KATHERINE: K0RN
+AUBREY: ABR
+BRYAN: BRYN
+BRYCE: BRS
+STEVEN: STFN
+RICHARD: RXRT
+HEIDI: HT
+AUTO: AT
+MAURICE: MRS
+RANDY: RNT
+CAMBRILLO: KMBRL
+BRIAN: BRN
+RAY: R
+GEOFF: JF
+BOB: BB
+AHA: AH
+AAH: A
+PAUL: PL
+BATTLEY: BTL
+WROTE: RT
+THIS: 0S

data/test/data/metaphone_buggy.txt

@@ -0,0 +1,52 @@
+#
+# Based on the table at http://aspell.net/metaphone/metaphone-kuhn.txt,
+# this mimics the behaviour of Lawrence Philips's BASIC implementation,
+# which appears to contain bugs when compared to his description of the 
+# algorithm.
+#
+ANASTHA: ANS0
+DAVIS-CARTER: TFSKRTR
+ESCARMANT: ESKRMNT
+MCCALL: MKKL
+MCCROREY: MKKRR
+MERSEAL: MRSL
+PIEURISSAINT: PRSNT
+ROTMAN: RTMN
+SCHEVEL: SXFL
+SCHROM: SXRM
+SEAL: SL
+SPARR: SPR
+STARLEPER: STRLPR
+THRASH: 0RX
+LOGGING: LKNK
+LOGIC: LJK
+JUDGES: JJS
+SHOOS: XS
+SHOES: XS
+CHUTE: XT
+SCHUSS: SXS
+OTTO: OT
+ERIC: ERK
+DAVE: TF
+CATHERINE: K0RN
+KATHERINE: K0RN
+AUBREY: ABR
+BRYAN: BRYN
+BRYCE: BRS
+STEVEN: STFN
+RICHARD: RXRT
+HEIDI: HT
+AUTO: AT
+MAURICE: MRS
+RANDY: RNT
+CAMBRILLO: KMRL
+BRIAN: BRN
+RAY: R
+GEOFF: JF
+BOB: BB
+AHA: AH
+AAH: A
+PAUL: PL
+BATTLEY: BTL
+WROTE: RT
+THIS: 0S

data/test/double_metaphone_test.rb

@@ -0,0 +1,16 @@
+require File.join(File.dirname(__FILE__), 'preamble')
+require 'rubygems'
+require 'fastercsv'  
+
+class DoubleMetaphoneTest < Test::Unit::TestCase
+
+  def test_cases
+    FasterCSV.read(File.rel('data', 'double_metaphone.csv'), :col_sep => ', ').each_with_index do |row, i|
+      primary, secondary = Text::Metaphone.double_metaphone(row[0])
+
+      assert_equal row[1], primary
+      assert_equal row[2], secondary.nil?? primary : secondary
+    end
+  end
+
+end

data/test/figlet_test.rb

@@ -0,0 +1,17 @@
+require File.join(File.dirname(__FILE__), 'preamble')
+
+class FigletTest < Test::Unit::TestCase
+
+  def test_hello_world
+    font = Text::Figlet::Font.new(File.rel('data', 'big.flf'))
+    figlet = Text::Figlet::Typesetter.new(font)
+    assert_equal File.read(File.rel('data', 'big.txt')), figlet['Hello World']
+  end
+
+  def test_no_smushing
+    font = Text::Figlet::Font.new(File.rel('data', 'chunky.flf'))
+    figlet = Text::Figlet::Typesetter.new(font, :smush => false)
+    assert_equal File.read(File.rel('data', 'chunky.txt')), figlet['Chunky Bacon']
+  end
+
+end

data/test/levenshtein_test.rb

@@ -0,0 +1,80 @@
+require File.join(File.dirname(__FILE__), 'preamble')
+
+class LevenshteinTest < Test::Unit::TestCase
+
+  include Text::Levenshtein
+
+  TEST_CASES = {
+    :easy => [
+      ['test', 'test', 0],
+      ['test', 'tent', 1],
+      ['gumbo', 'gambol', 2],
+      ['kitten', 'sitting', 3]
+    ],
+    :empty => [
+      ['foo', '', 3],
+      ['', '', 0],
+      ['a', '', 1]
+    ],
+    :utf8 => [
+      ["f\303\266o", 'foo', 1],
+      ["fran\303\247ais", 'francais', 1],
+      ["fran\303\247ais", "fran\303\246ais", 1],
+      [
+        "\347\247\201\343\201\256\345\220\215\345\211\215\343\201\257"<<
+        "\343\203\235\343\203\274\343\203\253\343\201\247\343\201\231",
+        "\343\201\274\343\201\217\343\201\256\345\220\215\345\211\215\343\201"<<
+        "\257\343\203\235\343\203\274\343\203\253\343\201\247\343\201\231",
+        2
+      ] # Japanese
+    ],
+    :iso_8859_1 => [
+      ["f\366o", 'foo', 1],
+      ["fran\347ais", 'francais', 1],
+      ["fran\347ais", "fran\346ais", 1]
+    ],
+    :edge => [
+      ['a', 'a', 0],
+      ['0123456789', 'abcdefghijklmnopqrstuvwxyz', 26]
+    ]
+  }
+
+  def assert_set(name)
+    TEST_CASES[name].each do |s, t, x|
+      assert_equal x, distance(s, t)
+      assert_equal x, distance(t, s)
+    end
+  end
+
+  def with_kcode(k)
+    old_kcode = $KCODE
+    $KCODE = k
+    yield
+    $KCODE = old_kcode
+  end
+
+  def test_easy_cases
+    assert_set(:easy)
+  end
+
+  def test_empty_cases
+    assert_set(:empty)
+  end
+
+  def test_edge_cases
+    assert_set(:edge)
+  end
+
+  def test_utf8_cases
+    with_kcode('U') do
+      assert_set(:utf8)
+    end
+  end
+
+  def test_iso_8859_1_cases
+    with_kcode('NONE') do
+      assert_set(:iso_8859_1)
+    end
+  end
+
+end

data/test/metaphone_test.rb

@@ -0,0 +1,39 @@
+require File.join(File.dirname(__FILE__), 'preamble')
+require 'yaml'
+
+class MetaphoneTest < Test::Unit::TestCase
+
+  def test_cases
+    YAML.load(File.read(File.rel('data', 'metaphone.txt'))).each do |input, expected_output|
+      assert_equal expected_output, Text::Metaphone.metaphone(input)
+    end
+  end
+  
+  def test_cases_for_buggy_implementation
+    YAML.load(File.read(File.rel('data', 'metaphone_buggy.txt'))).each do |input, expected_output|
+      assert_equal expected_output, Text::Metaphone.metaphone(input, :buggy=>true)
+    end
+  end
+
+  def test_junk
+    assert_equal Text::Metaphone.metaphone('foobar'),
+      Text::Metaphone.metaphone('%^@#$^f%^&o%^o@b#a@#r%^^&')
+    assert_equal Text::Metaphone.metaphone('foobar', :buggy=>true),
+      Text::Metaphone.metaphone('%^@#$^f%^&o%^o@b#a@#r%^^&', :buggy=>true)
+  end
+
+  def test_caps
+    assert_equal Text::Metaphone.metaphone('foobar'),
+      Text::Metaphone.metaphone('FOOBAR')
+    assert_equal Text::Metaphone.metaphone('foobar', :buggy=>true),
+      Text::Metaphone.metaphone('FOOBAR', :buggy=>true)
+  end
+
+  def test_string
+    assert_equal 'F BR BS', Text::Metaphone.metaphone('foo bar baz')
+    assert_equal 'N WT', Text::Metaphone.metaphone('gnu what')
+    assert_equal 'F BR BS', Text::Metaphone.metaphone('foo bar baz', :buggy=>true)
+    assert_equal 'N WT', Text::Metaphone.metaphone('gnu what', :buggy=>true)
+  end
+
+end

data/test/preamble.rb

@@ -0,0 +1,10 @@
+require 'test/unit'
+
+$:.unshift File.join(File.dirname(__FILE__), '..', 'lib')
+require 'text'
+
+class File
+  def self.rel(*path)
+    join(dirname(__FILE__), *path)
+  end
+end

data/test/soundex_test.rb

@@ -0,0 +1,27 @@
+require File.join(File.dirname(__FILE__), 'preamble')
+require 'yaml'
+
+class SoundexTest < Test::Unit::TestCase
+
+  def test_cases
+    YAML.load(%{
+
+      Euler: E460
+      Ellery: E460
+      Gauss: G200
+      Ghosh: G200
+      Hilbert: H416
+      Heilbronn: H416
+      Knuth: K530
+      Kant: K530
+      Lloyd: L300
+      Ladd: L300
+      Lukasiewicz: L222
+      Lissajous: L222
+
+    }).each do |input, expected_output|
+      assert_equal expected_output, Text::Soundex.soundex(input)
+    end
+  end
+
+end

metadata

@@ -0,0 +1,68 @@
+--- !ruby/object:Gem::Specification 
+rubygems_version: 0.9.0
+specification_version: 1
+name: Text
+version: !ruby/object:Gem::Version 
+  version: 1.0.0
+date: 2006-09-23 00:00:00 +01:00
+summary: A collection of text algorithms
+require_paths: 
+- lib
+email: 
+homepage: http://text.rubyforge.org/
+rubyforge_project: text
+description: 
+autorequire: text
+default_executable: 
+bindir: bin
+has_rdoc: false
+required_ruby_version: !ruby/object:Gem::Version::Requirement 
+  requirements: 
+  - - ">"
+    - !ruby/object:Gem::Version 
+      version: 0.0.0
+  version: 
+platform: ruby
+signing_key: 
+cert_chain: 
+post_install_message: 
+authors: 
+- Paul Battley, Michael Neumann, Tim Fletcher
+files: 
+- lib/text.rb
+- lib/text/double_metaphone.rb
+- lib/text/figlet.rb
+- lib/text/levenshtein.rb
+- lib/text/metaphone.rb
+- lib/text/soundex.rb
+- lib/text/figlet/font.rb
+- lib/text/figlet/smusher.rb
+- lib/text/figlet/typesetter.rb
+- test/double_metaphone_test.rb
+- test/figlet_test.rb
+- test/levenshtein_test.rb
+- test/metaphone_test.rb
+- test/preamble.rb
+- test/soundex_test.rb
+- test/data/big.flf
+- test/data/big.txt
+- test/data/chunky.flf
+- test/data/chunky.txt
+- test/data/double_metaphone.csv
+- test/data/metaphone.txt
+- test/data/metaphone_buggy.txt
+- rakefile.rb
+test_files: []
+
+rdoc_options: []
+
+extra_rdoc_files: []
+
+executables: []
+
+extensions: []
+
+requirements: []
+
+dependencies: []
+