groupie 0.1.0 → 0.1.1

data/Rakefile

@@ -19,17 +19,17 @@ rescue LoadError
 end
 
 require 'rake/testtask'
-Rake::TestTask.new(:test) do |test|
-  test.libs << 'lib' << 'test'
-  test.pattern = 'test/**/*_test.rb'
+Rake::TestTask.new(:spec) do |test|
+  test.libs << 'lib' << 'spec'
+  test.pattern = 'spec/**/*_spec.rb'
   test.verbose = true
 end
 
 begin
   require 'rcov/rcovtask'
   Rcov::RcovTask.new do |test|
-    test.libs << 'test'
-    test.pattern = 'test/**/*_test.rb'
+    test.libs << 'spec'
+    test.pattern = 'spec/**/*_spec.rb'
     test.verbose = true
   end
 rescue LoadError

data/VERSION

@@ -1 +1 @@
-0.1.0
+0.1.1

data/groupie.gemspec

@@ -5,7 +5,7 @@
 
 Gem::Specification.new do |s|
   s.name = %q{groupie}
-  s.version = "0.1.0"
+  s.version = "0.1.1"
 
   s.required_rubygems_version = Gem::Requirement.new(">= 0") if s.respond_to? :required_rubygems_version=
   s.authors = ["Wes Oldenbeuving"]
@@ -25,12 +25,15 @@ Gem::Specification.new do |s|
      "lib/groupie/core_ext/string.rb",
      "lib/groupie/group.rb",
      "readme.rdoc",
-     "test/fixtures/ham/spam.la-44116217.txt",
-     "test/fixtures/spam/spam.la-44118014.txt",
-     "test/groupie/core_ext/string_test.rb",
-     "test/groupie/group_test.rb",
-     "test/groupie_test.rb",
-     "test/test_helper.rb"
+     "spec/fixtures/ham/email_ham1.txt",
+     "spec/fixtures/ham/spam.la-44116217.txt",
+     "spec/fixtures/spam/email_spam1.txt",
+     "spec/fixtures/spam/email_spam2.txt",
+     "spec/fixtures/spam/spam.la-44118014.txt",
+     "spec/groupie/core_ext/string_spec.rb",
+     "spec/groupie/group_spec.rb",
+     "spec/groupie_spec.rb",
+     "spec/spec_helper.rb"
   ]
   s.homepage = %q{http://github.com/Narnach/groupie}
   s.rdoc_options = ["--charset=UTF-8"]
@@ -38,10 +41,10 @@ Gem::Specification.new do |s|
   s.rubygems_version = %q{1.3.7}
   s.summary = %q{Group and classify text}
   s.test_files = [
-    "test/groupie/core_ext/string_test.rb",
-     "test/groupie/group_test.rb",
-     "test/groupie_test.rb",
-     "test/test_helper.rb"
+    "spec/groupie/core_ext/string_spec.rb",
+     "spec/groupie/group_spec.rb",
+     "spec/groupie_spec.rb",
+     "spec/spec_helper.rb"
   ]
 
   if s.respond_to? :specification_version then

data/lib/groupie/core_ext/string.rb

@@ -6,7 +6,7 @@ class Groupie
           gsub(/\s/," ").
           gsub(/[$']/,'').
           gsub(/<[^>]+?>|[^\w -.,]/,'').
-          split(" ").map {|str| str.gsub(/[,.]+\Z/,'')}
+          split(" ").map {|str| str.gsub(/\A['"]+|[!,."']+\Z/,'')}
       end
     end
   end

data/readme.rdoc

@@ -9,7 +9,6 @@ The eventual goal is to have Groupie work as a sort of bayesian spam filter, whe
 Groupie is a 'fun' project that has the following goals, in descending order of importance:
 * Have fun playing with code
 * Play with Bayesian-like (spam) filtering
-* Check out the Testy BDD framework. It's pretty good for 60 lines of code!
 
 == Current functionality
 

data/spec/fixtures/ham/email_ham1.txt

@@ -0,0 +1,13 @@
+Re: [ubuntu-art] [Breathe] Network Manager-icons
+Am Sonntag, den 31.05.2009, 17:53 +0200 schrieb Steve Dodier:
+> Hello,
+>
+> I think the notify-osd icons have a completely different style, which
+> is looking great within the notification bubbles, but i doubt it'd
+> look great to have the notify-osd wifi icons in the panel. I think the
+> drawing of the notification- wifi icons should be done afterwards, and
+> if they should be based on those of the icon set, they could be made
+> smoother, and possibly desaturated for some of them, to avoid drawing
+> too much attention from the user when popping up.
+>
+> Cordially, SD.

data/spec/fixtures/spam/email_spam1.txt

@@ -0,0 +1,5 @@
+I noticed your flirt
+If you cannot see the pictures and links below, please click here to view them.
+PHARMACY CLUB | UNSUBSCRIBE | YOUR PRIVACY RIGHTS
+Copyright 2009 Zjfqq, all rights reserved
+Customer Service Dept., 87 Hizq Iveox Street, Isahaylo, VS 25270

data/spec/fixtures/spam/email_spam2.txt

@@ -0,0 +1,7 @@
+Re: Your subscribe #976589
+Tell a friend · Download latest version	See this email as a webpage
+Hello!
+Shipped Privately And Discreetly To Your Door!
+We want to put a great big grin on your face in 2009. You'll be to rejoice all year.
+Unsubscribe · Lost Password · Account Settings · Help · Terms of Service · Privacy
+Ottho Heldringstraat 2, 31719 AZ Amsterdam, The Netherlands

data/spec/groupie/core_ext/string_spec.rb

@@ -0,0 +1,37 @@
+require File.join(File.dirname(__FILE__), %w[.. .. spec_helper])
+
+describe String do
+  context "tokenize" do
+    it 'should split words' do
+      "hello world".tokenize.should == %w[hello world]
+    end
+
+    it 'should downcase words' do
+      "Hello World".tokenize.should == %w[hello world]
+    end
+
+    it 'should strip special characters' do
+      "blah, bla!".tokenize.should == %w[blah bla]
+    end
+
+    it 'should prserve infix hyphens and underscores' do
+      "hyphen-ated under_score".tokenize.should == %w[hyphen-ated under_score]
+    end
+
+    it 'should sanitize html tags' do
+      '<a href="http://example.org">example</a>'.tokenize.should == %w[example]
+    end
+
+    it 'should preserve infix periods' do
+      'example.org rocks. read it...'.tokenize.should == %w[example.org rocks read it]
+    end
+    
+    it "should preserve infix commas" do
+      '$1,000,000.00 or $1.000.000,00'.tokenize.should == %w[1,000,000.00 or 1.000.000,00]
+    end
+    
+    it "should strip quotes around tokens" do
+      '"first last"'.tokenize.should == %w[first last]
+    end
+  end
+end

data/spec/groupie/group_spec.rb

@@ -0,0 +1,12 @@
+require File.join(File.dirname(__FILE__), %w[.. spec_helper])
+require 'yaml'
+
+describe Groupie::Group do
+  it "can be serialized and loaded through YAML" do
+    group = Groupie::Group.new 'group'
+    group.add %w[buy flowers]
+    loaded_group = YAML.load(group.to_yaml)
+    loaded_group.add %w[buy candy]
+    loaded_group.count("candy").should == 1
+  end
+end

data/spec/groupie_spec.rb

@@ -0,0 +1,75 @@
+require File.join(File.dirname(__FILE__), 'spec_helper')
+
+describe Groupie do
+  describe "classify" do
+    it 'should work when 100% certaint' do
+      g = Groupie.new
+      g[:spam].add %w[viagra]
+      g[:ham].add %w[flowers]
+      g.classify('viagra').should == {:spam => 1.0, :ham => 0.0}
+    end
+
+    it 'should work when split 50/50 between two groups' do
+      g = Groupie.new
+      g[:spam].add %w[buy viagra now]
+      g[:ham].add %w[buy flowers for your mom]
+      g.classify('buy').should == {:spam => 0.5, :ham => 0.5}
+    end
+
+    it 'should work when weighed more towards one group' do
+      g = Groupie.new
+      g[:spam].add %w[buy viagra now]
+      g[:spam].add %w[buy cialis now]
+      g[:ham].add %w[buy flowers for your mom]
+      g.classify('buy').should == {:spam => 2 / 3.0, :ham => 1 / 3.0}
+    end
+
+    it 'should work with more than two groups' do
+      g = Groupie.new
+      g[:weight].add 'pound'
+      g[:currency].add 'pound'
+      g[:phone_key].add 'pound'
+      g.classify('pound').should == {:weight => 1/3.0, :currency => 1/3.0, :phone_key => 1/3.0}
+    end
+
+    it 'should tokenize and classify emails' do
+      email = File.read(File.join(File.dirname(__FILE__), %w[fixtures spam email_spam1.txt]))
+      email2 = File.read(File.join(File.dirname(__FILE__), %w[fixtures spam email_spam2.txt]))
+      email3 = File.read(File.join(File.dirname(__FILE__), %w[fixtures ham email_ham1.txt]))
+      g = Groupie.new
+      g[:spam].add email.tokenize
+      g[:spam].add email2.tokenize
+      g[:ham].add email3.tokenize
+      c = g.classify('discreetly')
+      c[:spam].should > c[:ham]
+      c2 = g.classify('user')
+      c2[:ham].should > c2[:spam]
+    end
+  end
+  context "classify_text" do
+    it 'should tokenized html emails' do
+      g = Groupie.new
+      spam_tokens = File.read(File.join(File.dirname(__FILE__), %w[fixtures spam spam.la-44118014.txt])).tokenize
+      ham_tokens = File.read(File.join(File.dirname(__FILE__), %w[fixtures ham spam.la-44116217.txt])).tokenize
+      g[:spam].add spam_tokens
+      g[:ham].add ham_tokens
+
+      c = g.classify 'user'
+      c[:ham].should > c[:spam]
+
+      c = g.classify_text(spam_tokens)
+      c[:spam].should > c[:ham]
+    end
+
+    it 'should classify a text' do
+      g = Groupie.new
+      g[:spam].add %w[buy viagra now to grow fast]
+      g[:spam].add %w[buy cialis on our website]
+      g[:ham].add %w[buy flowers for your mom]
+      result = g.classify_text "Grow flowers to sell on our website".tokenize
+      result[:spam].should > result[:ham]
+      result2 = g.classify_text "Grow flowers to give to your mom".tokenize
+      result2[:ham].should == result2[:spam]
+    end
+  end
+end

data/spec/spec_helper.rb

@@ -0,0 +1 @@
+require 'lib/groupie'

metadata

@@ -1,13 +1,13 @@
 --- !ruby/object:Gem::Specification 
 name: groupie
 version: !ruby/object:Gem::Version 
-  hash: 27
+  hash: 25
   prerelease: false
   segments: 
   - 0
   - 1
-  - 0
-  version: 0.1.0
+  - 1
+  version: 0.1.1
 platform: ruby
 authors: 
 - Wes Oldenbeuving
@@ -50,12 +50,15 @@ files:
 - lib/groupie/core_ext/string.rb
 - lib/groupie/group.rb
 - readme.rdoc
-- test/fixtures/ham/spam.la-44116217.txt
-- test/fixtures/spam/spam.la-44118014.txt
-- test/groupie/core_ext/string_test.rb
-- test/groupie/group_test.rb
-- test/groupie_test.rb
-- test/test_helper.rb
+- spec/fixtures/ham/email_ham1.txt
+- spec/fixtures/ham/spam.la-44116217.txt
+- spec/fixtures/spam/email_spam1.txt
+- spec/fixtures/spam/email_spam2.txt
+- spec/fixtures/spam/spam.la-44118014.txt
+- spec/groupie/core_ext/string_spec.rb
+- spec/groupie/group_spec.rb
+- spec/groupie_spec.rb
+- spec/spec_helper.rb
 has_rdoc: true
 homepage: http://github.com/Narnach/groupie
 licenses: []
@@ -91,7 +94,7 @@ signing_key:
 specification_version: 3
 summary: Group and classify text
 test_files: 
-- test/groupie/core_ext/string_test.rb
-- test/groupie/group_test.rb
-- test/groupie_test.rb
-- test/test_helper.rb
+- spec/groupie/core_ext/string_spec.rb
+- spec/groupie/group_spec.rb
+- spec/groupie_spec.rb
+- spec/spec_helper.rb

data/test/groupie/core_ext/string_test.rb

@@ -1,45 +0,0 @@
-require File.join(File.dirname(__FILE__), %w[.. .. test_helper])
-
-Testy.testing 'String' do
-  context 'tokenize' do
-    test 'split words' do |t|
-      tokens = "hello world".tokenize
-      t.check 'words are split',
-        :expect => %w[hello world],
-        :actual => tokens
-    end
-
-    test 'downcase words' do |t|
-      tokens = "Hello World".tokenize
-      t.check 'words are downcased',
-        :expect => %w[hello world],
-        :actual => tokens
-    end
-
-    test 'most symbols are stripped' do |t|
-      tokens = "hyphen-ated, under_score!".tokenize
-      t.check 'some symbols are left',
-        :expect => %w[hyphen-ated under_score],
-        :actual => tokens
-    end
-
-    test 'html tags are sanitized' do |t|
-      tokens = '<a href="http://example.org">example</a>'.tokenize
-      t.check 'only content of tags is retained',
-        :expect => %w[example],
-        :actual => tokens
-    end
-
-    test 'some dots are ok' do |t|
-      tokens = 'example.org rocks. read it...'.tokenize
-      t.check 'infix dots are kept',
-        :expect => %w[example.org rocks read it],
-        :actual => tokens
-
-      tokens2 = '$1,000,000.00 or $1.000.000,00'.tokenize
-      t.check 'infix commas are kept',
-        :expect => %w[1,000,000.00 or 1.000.000,00],
-        :actual => tokens2
-    end
-  end
-end

data/test/groupie/group_test.rb

@@ -1,15 +0,0 @@
-require File.join(File.dirname(__FILE__), %w[.. test_helper])
-
-Testy.testing 'Groupie::Group' do
-  test 'can be serialized and loaded through YAML' do |t|
-    require 'yaml'
-
-    g = Groupie::Group.new 'group'
-    g.add %w[buy flowers]
-    g2 = YAML.load(g.to_yaml)
-    g2.add %w[buy candy]
-    t.check 'default value works for new entries',
-      :expect => 1,
-      :actual => g2.count('candy')
-  end
-end

data/test/groupie_test.rb

@@ -1,124 +0,0 @@
-require File.join(File.dirname(__FILE__), 'test_helper')
-
-Testy.testing 'Groupie' do
-  test 'classification is certain' do |t|
-    g = Groupie.new
-    g[:spam].add %w[viagra]
-    g[:ham].add %w[flowers]
-    classification = g.classify 'viagra'
-    t.check 'viagra is',
-      :expect => {:spam => 1.0, :ham => 0.0},
-      :actual => classification
-  end
-
-  test 'classification is split between two groups' do |t|
-    g = Groupie.new
-    g[:spam].add %w[buy viagra now]
-    g[:ham].add %w[buy flowers for your mom]
-    classification = g.classify 'buy'
-    t.check 'buy is classified as',
-      :expect => {:spam => 0.5, :ham => 0.5},
-      :actual => classification
-  end
-
-  test 'classification is weighed more heavy in one group' do |t|
-    g = Groupie.new
-    g[:spam].add %w[buy viagra now]
-    g[:spam].add %w[buy cialis now]
-    g[:ham].add %w[buy flowers for your mom]
-    t.check 'buy is classified as',
-      :expect => {:spam => 2 / 3.0, :ham => 1 / 3.0},
-      :actual => g.classify('buy')
-  end
-
-  test 'classification works fine with more than two groups' do |t|
-    g = Groupie.new
-    g[:weight].add 'pound'
-    g[:currency].add 'pound'
-    g[:phone_key].add 'pound'
-    t.check 'pound is classified as',
-      :expect => {:weight => 1/3.0, :currency => 1/3.0, :phone_key => 1/3.0},
-      :actual => g.classify('pound')
-  end
-
-  test 'tokenized emails' do |t|
-    email = <<-EMAIL
-I noticed your flirt
-If you cannot see the pictures and links below, please click here to view them.
-PHARMACY CLUB | UNSUBSCRIBE | YOUR PRIVACY RIGHTS
-Copyright 2009 Zjfqq, all rights reserved
-Customer Service Dept., 87 Hizq Iveox Street, Isahaylo, VS 25270
-    EMAIL
-    email2 = <<-EMAIL
-Re: Your subscribe #976589
-Tell a friend · Download latest version	See this email as a webpage
-Hello!
-Shipped Privately And Discreetly To Your Door!
-We want to put a great big grin on your face in 2009. You'll be to rejoice all year.
-Unsubscribe · Lost Password · Account Settings · Help · Terms of Service · Privacy
-Ottho Heldringstraat 2, 31719 AZ Amsterdam, The Netherlands
-    EMAIL
-    email3 = <<-EMAIL
-Re: [ubuntu-art] [Breathe] Network Manager-icons
-Am Sonntag, den 31.05.2009, 17:53 +0200 schrieb Steve Dodier:
-> Hello,
->
-> I think the notify-osd icons have a completely different style, which
-> is looking great within the notification bubbles, but i doubt it'd
-> look great to have the notify-osd wifi icons in the panel. I think the
-> drawing of the notification- wifi icons should be done afterwards, and
-> if they should be based on those of the icon set, they could be made
-> smoother, and possibly desaturated for some of them, to avoid drawing
-> too much attention from the user when popping up.
->
-> Cordially, SD.
-    EMAIL
-    g = Groupie.new
-    g[:spam].add email.tokenize
-    g[:spam].add email2.tokenize
-    g[:ham].add email3.tokenize
-    c = g.classify('discreetly')
-    t.check 'classification of "discreetly" is spam',
-      :expect => true,
-      :actual => c[:spam] > c[:ham]
-    c2 = g.classify('user')
-    t.check 'classification of "user" is ham',
-      :expect => true,
-      :actual => c2[:ham] > c2[:spam]
-  end
-
-  test 'tokenized html emails' do |t|
-    g = Groupie.new
-    spam_tokens = File.read(File.join(File.dirname(__FILE__),
-      %w[fixtures spam spam.la-44118014.txt])).tokenize
-    ham_tokens = File.read(File.join(File.dirname(__FILE__),
-      %w[fixtures ham spam.la-44116217.txt])).tokenize
-    g[:spam].add spam_tokens
-    g[:ham].add ham_tokens
-
-    c = g.classify 'user'
-    t.check 'classification of the word "user" is ham',
-      :expect => true,
-      :actual => (c[:ham] > c[:spam])
-
-    c = g.classify_text(spam_tokens)
-    t.check 'classification of spam email is spam',
-      :expect => true,
-      :actual => (c[:spam] > c[:ham])
-  end
-
-  test 'classify a text' do |t|
-    g = Groupie.new
-    g[:spam].add %w[buy viagra now to grow fast]
-    g[:spam].add %w[buy cialis on our website]
-    g[:ham].add %w[buy flowers for your mom]
-    result = g.classify_text "Grow flowers to sell on our website".tokenize
-    t.check 'classification of a spammy text is spam',
-      :expect => true,
-      :actual => result[:spam] > result[:ham]
-    result2 = g.classify_text "Grow flowers to give to your mom".tokenize
-    t.check 'classification of a non-spammy text is ham',
-      :expect => true,
-      :actual => result2[:ham] > result2[:spam]
-  end
-end

data/test/test_helper.rb

@@ -1,3 +0,0 @@
-require 'rubygems'
-require 'testy'
-require 'lib/groupie'