wikipedia-vandalism_detection 0.0.1

data/spec/vandalism_detection/features/compressibility_spec.rb

@@ -0,0 +1,42 @@
+require 'spec_helper'
+require 'zlib'
+
+describe Wikipedia::VandalismDetection::Features::Compressibility do
+
+  before do
+    @feature = Wikipedia::VandalismDetection::Features::Compressibility.new
+  end
+
+  it { should be_a Wikipedia::VandalismDetection::Features::Base }
+
+  describe "#calculate" do
+
+    it "returns the ratio of compressed text size to uncompressed text size" do
+      old_text = 'text'
+      new_text = 'text [[If this is a quite long textpart]] of normal words then it might be less possible to be a vandalism.'
+
+      old_revision = build(:old_revision, text: Wikipedia::VandalismDetection::Text.new(old_text))
+      new_revision = build(:new_revision, text: Wikipedia::VandalismDetection::Text.new(new_text))
+      edit = build(:edit, old_revision: old_revision, new_revision: new_revision)
+
+      bytesize = 10.0
+
+      Zlib::Deflate.stub(deflate: Wikipedia::VandalismDetection::Text.new)
+      Wikipedia::VandalismDetection::Text.any_instance.stub(bytesize: bytesize)
+      ratio = bytesize / (bytesize + bytesize)
+
+      expect(@feature.calculate(edit)).to eq ratio
+    end
+
+    it "returns 0.5 on emtpy inserted text" do
+      old_text = Wikipedia::VandalismDetection::Text.new("deletion text")
+      new_text = Wikipedia::VandalismDetection::Text.new(" text")
+
+      old_revision = build(:old_revision, text: old_text)
+      new_revision = build(:new_revision, text: new_text)
+      edit = build(:edit, new_revision: new_revision, old_revision: old_revision)
+
+      expect(@feature.calculate(edit)).to eq 0.5
+    end
+  end
+end

data/spec/vandalism_detection/features/contains_base_spec.rb

@@ -0,0 +1,33 @@
+require 'spec_helper'
+
+describe Wikipedia::VandalismDetection::Features::ContainsBase do
+
+  before do
+    @base = Wikipedia::VandalismDetection::Features::ContainsBase.new
+  end
+
+  it { should be_a Wikipedia::VandalismDetection::Features::Base }
+
+  describe "#contains" do
+
+    it "returns 1 if a given text contains the given terms array" do
+      text = "Content including text"
+      expect(@base.contains(text, ['content', 'anything'])).to eq 1
+    end
+
+    it "returns 1 if a given text contains the given string" do
+      text = "Content including text"
+      expect(@base.contains(text, 'content')).to eq 1
+    end
+
+    it "returns 0 if a given text does not contain the given string" do
+      text = "not containing anything con tent"
+      expect(@base.contains(text, 'content')).to eq 0
+    end
+
+    it "returns 0 if a given text does not contain any of the given terms" do
+      text = "not containing anything con tent"
+      expect(@base.contains(text, ['content', 'text'])).to eq 0
+    end
+  end
+end

data/spec/vandalism_detection/features/copyedit_spec.rb

@@ -0,0 +1,33 @@
+require 'spec_helper'
+
+describe Wikipedia::VandalismDetection::Features::Copyedit do
+
+  before do
+    @feature = Wikipedia::VandalismDetection::Features::Copyedit.new
+  end
+
+  it { should be_a Wikipedia::VandalismDetection::Features::ContainsBase }
+
+  describe "#calculate" do
+
+    it "returns 1 if the edit comment includes 'copyedit'" do
+      comment = Wikipedia::VandalismDetection::Text.new "copyediting content"
+      edit = build :edit, new_revision: build(:new_revision, comment: comment)
+
+      expect(@feature.calculate(edit)).to eq 1
+    end
+
+    it "returns 1 if the edit comment includes 'copy edit'" do
+      comment = Wikipedia::VandalismDetection::Text.new "copy editing content"
+      edit = build :edit, new_revision: build(:new_revision, comment: comment)
+
+      expect(@feature.calculate(edit)).to eq 1
+    end
+
+    it "returns 0 on emtpy comment" do
+      edit = build :edit, new_revision: build(:new_revision, comment: "")
+
+      expect(@feature.calculate(edit)).to eq 0
+    end
+  end
+end

data/spec/vandalism_detection/features/digit_ratio_spec.rb

@@ -0,0 +1,35 @@
+require 'spec_helper'
+
+describe Wikipedia::VandalismDetection::Features::DigitRatio do
+
+  before do
+    @feature = Wikipedia::VandalismDetection::Features::DigitRatio.new
+  end
+
+  it { should be_a Wikipedia::VandalismDetection::Features::Base }
+
+  describe "#calculate" do
+
+    it "returns the digit to all letters ratio of the edit's new revision inserted text" do
+      old_text = Wikipedia::VandalismDetection::Text.new('text1')
+      new_text = Wikipedia::VandalismDetection::Text.new 'text1 [[1A4 B6 8Cd]]' # 3 digit letters of total 8 letters
+
+      old_revision = build(:old_revision, text: old_text)
+      new_revision = build(:new_revision, text: new_text)
+      edit = build(:edit, old_revision: old_revision, new_revision: new_revision)
+
+      expect(@feature.calculate(edit)).to eq (1.0 + 4) / (1.0 + 8)
+    end
+
+    it "returns 0.0 if no text inserted" do
+      old_text = Wikipedia::VandalismDetection::Text.new("deletion text")
+      new_text = Wikipedia::VandalismDetection::Text.new("text")
+
+      old_revision = build(:old_revision, text: old_text)
+      new_revision = build(:new_revision, text: new_text)
+      edit = build(:edit, new_revision: new_revision, old_revision: old_revision)
+
+      expect(@feature.calculate(edit)).to eq 0.0
+    end
+  end
+end

data/spec/vandalism_detection/features/edits_per_user_spec.rb

@@ -0,0 +1,49 @@
+require 'spec_helper'
+
+describe Wikipedia::VandalismDetection::Features::EditsPerUser do
+
+  before do
+    @feature = Wikipedia::VandalismDetection::Features::EditsPerUser.new
+  end
+
+  it { should be_a Wikipedia::VandalismDetection::Features::Base }
+
+  describe "#calculate" do
+
+    describe "online" do
+      it "returns the number of previously submitted edit from the same IP or ID" do
+        #http://en.wikipedia.org/w/api.php?action=query&format=json&list=usercontribs&ucuser=<name or ip>&ucprop=ids
+        old_revision = build(:old_revision, id: '527136737')
+        new_revision = build(:new_revision, id: '527137015', parent_id: '527136737',
+                             contributor: '142.11.81.219', timestamp: '2012-12-09T05:30:07Z' )
+
+        edit = build(:edit, old_revision: old_revision, new_revision: new_revision)
+
+        expect(@feature.calculate(edit)).to eq 1
+      end
+    end
+
+    describe "offline" do
+      before do
+        page = build(:page)
+        page.id = '1234'
+        page.title = 'Page Title'
+
+        # contributor: see factories/page.rb !
+        old_revision = build(:new_revision, contributor: 'User')
+        new_revision = build(:even_newer_revision, contributor: 'User')
+
+        @edit = build(:edit, old_revision: old_revision, new_revision: new_revision, page: page)
+      end
+
+      it "does not use an API call if the edit has a page reference" do
+        expect(Wikipedia).to_not receive :api_request
+        @feature.calculate(@edit)
+      end
+
+      it "returns the number of previously submitted edit from the same IP or ID" do
+        expect(@feature.calculate(@edit)).to eq 1
+      end
+    end
+  end
+end

data/spec/vandalism_detection/features/emoticons_frequency_spec.rb

@@ -0,0 +1,36 @@
+require 'spec_helper'
+
+describe Wikipedia::VandalismDetection::Features::EmoticonsFrequency do
+
+  before do
+    @feature = Wikipedia::VandalismDetection::Features::EmoticonsFrequency.new
+  end
+
+  it { should be_a Wikipedia::VandalismDetection::Features::Base }
+
+  describe "#calculate" do
+
+    it "returns the number of emoticons relative to all words count" do
+      # total 10 words, 3 emoticons
+      old_text = Wikipedia::VandalismDetection::Text.new 'Old :-).'
+      new_text = Wikipedia::VandalismDetection::Text.new "Old :-). ;) you do love icons and emoticons? :D :P, yeah."
+
+      old_revision = build(:old_revision, text: old_text)
+      new_revision = build(:new_revision, text: new_text)
+      edit = build(:edit, old_revision: old_revision, new_revision: new_revision)
+
+      expect(@feature.calculate(edit)).to eq 3.0 / 10.0
+    end
+
+    it "returns 0.0 on emtpy clean text revisions" do
+      old_text = Wikipedia::VandalismDetection::Text.new 'Old :-).'
+      new_text = Wikipedia::VandalismDetection::Text.new "Old :-). {{speedy deletion}}"
+
+      old_revision = build(:old_revision, text: old_text)
+      new_revision = build(:new_revision, text: new_text)
+      edit = build(:edit, old_revision: old_revision, new_revision: new_revision)
+
+      expect(@feature.calculate(edit)).to eq 0.0
+    end
+  end
+end

data/spec/vandalism_detection/features/emoticons_impact_spec.rb

@@ -0,0 +1,51 @@
+require 'spec_helper'
+
+describe Wikipedia::VandalismDetection::Features::EmoticonsImpact do
+
+  before do
+    @feature = Wikipedia::VandalismDetection::Features::EmoticonsImpact.new
+  end
+
+  it { should be_a Wikipedia::VandalismDetection::Features::Base }
+
+  describe "#calculate" do
+
+    it "returns the impact of emoticon words of the edit's new revision text" do
+      # 3 emoticons
+      old_text = ":) Hi you I got some :-X, you know ;)"
+
+      # 4 emoticons
+      new_text = ":) Hi (=you) I added another :-X you know ;)? (="
+
+      old_revision = build(:old_revision, text: old_text)
+      new_revision = build(:new_revision, text: new_text)
+      edit = build(:edit, old_revision: old_revision, new_revision: new_revision)
+
+      expect(@feature.calculate(edit)).to eq 3.0 / (3.0 + 4.0)
+    end
+
+    it "returns 0.5 on both no terms in text revisions" do
+      old_revision = build(:old_revision, text: "")
+      new_revision = build(:new_revision, text: "")
+      edit = build(:edit, new_revision: new_revision, old_revision: old_revision)
+
+      expect(@feature.calculate(edit)).to eq 0.5
+    end
+
+    it "returns 0.0 on emtpy text of old revision" do
+      old_revision = build(:old_revision, text: "")
+      new_revision = build(:new_revision, text: ":)")
+      edit = build(:edit, new_revision: new_revision, old_revision: old_revision)
+
+      expect(@feature.calculate(edit)).to eq 0.0
+    end
+
+    it "returns 1.0 on emtpy text of new revision" do
+      old_revision = build(:old_revision, text: ":)")
+      new_revision = build(:new_revision, text: "")
+      edit = build(:edit, new_revision: new_revision, old_revision: old_revision)
+
+      expect(@feature.calculate(edit)).to eq 1.0
+    end
+  end
+end

data/spec/vandalism_detection/features/frequency_base_spec.rb

@@ -0,0 +1,26 @@
+require 'spec_helper'
+require 'wikipedia/vandalism_detection/features/frequency_base'
+
+describe Wikipedia::VandalismDetection::Features::FrequencyBase do
+
+  before do
+    @feature = Wikipedia::VandalismDetection::Features::FrequencyBase.new
+    @terms = Wikipedia::VandalismDetection::WordLists::PRONOUNS
+  end
+
+  it { should be_a Wikipedia::VandalismDetection::Features::Base }
+
+  describe "#frequency" do
+
+    it { should respond_to :frequency }
+
+    it "returns the frequency in percentage of given word counts" do
+      text = "I am, i like you."
+      expect(@feature.frequency(text , @terms)).to eq 3.0/5.0
+    end
+
+    it "returns 0.0 if total word count is zero" do
+      expect(@feature.frequency("", @terms)).to eq 0.0
+    end
+  end
+end

data/spec/vandalism_detection/features/impact_base_spec.rb

@@ -0,0 +1,41 @@
+require 'spec_helper'
+require 'wikipedia/vandalism_detection/features/impact_base'
+
+describe Wikipedia::VandalismDetection::Features::ImpactBase do
+
+  before do
+    @feature = Wikipedia::VandalismDetection::Features::ImpactBase.new
+    @pronouns = Wikipedia::VandalismDetection::WordLists::PRONOUNS
+  end
+
+  it { should be_a Wikipedia::VandalismDetection::Features::Base }
+
+  describe "#impact" do
+
+    it { should respond_to :impact }
+
+    it "returns the impact in percentage of given terms in old realitve to new text" do
+      # 3 pronouns
+      old_text = "Your old text will be mine or Your's"
+      # 4 pronouns
+      new_text = "My new text and your old text will be ours and mine"
+
+      text = "I am, i like you."
+      expect(@feature.impact(old_text , new_text, @pronouns)).to eq 3.0 / (3.0 + 4.0)
+    end
+
+    it "returns 0.0 if old terms word count is zero" do
+      new_text = "My new text and your old text will be ours and mine"
+      expect(@feature.impact("", new_text, @pronouns)).to eq 0.0
+    end
+
+    it "returns 1.0 if new terms word count is zero" do
+      old_text = "My new text and your old text will be ours and mine"
+      expect(@feature.impact(old_text, "", @pronouns)).to eq 1.0
+    end
+
+    it "returns 0.5 if both terms word count is zero" do
+      expect(@feature.impact("", "", @pronouns)).to eq 0.5
+    end
+  end
+end

data/spec/vandalism_detection/features/inserted_character_distribution_spec.rb

@@ -0,0 +1,46 @@
+require 'spec_helper'
+
+describe Wikipedia::VandalismDetection::Features::InsertedCharacterDistribution do
+
+  before do
+    @feature = Wikipedia::VandalismDetection::Features::InsertedCharacterDistribution.new
+  end
+
+  it { should be_a Wikipedia::VandalismDetection::Features::Base }
+
+  describe "#calculate" do
+
+    it "returns the Kullback-Leibler Divergence of the inserted text's character distribution" do
+      old_text = Wikipedia::VandalismDetection::Text.new('old text')
+      new_text = Wikipedia::VandalismDetection::Text.new 'old text [[new inserted text]] given dero 9'
+
+      old_revision = build(:old_revision, text: old_text)
+      new_revision = build(:new_revision, text: new_text)
+      edit = build(:edit, old_revision: old_revision, new_revision: new_revision)
+
+      expect(@feature.calculate(edit)).to eq 1.6609633564650683
+    end
+
+    it "returns missing value if no alphanumeric characters are inserted" do
+      old_text = Wikipedia::VandalismDetection::Text.new("old text")
+      new_text = Wikipedia::VandalismDetection::Text.new("old text !* [[?]]")
+
+      old_revision = build(:old_revision, text: old_text)
+      new_revision = build(:new_revision, text: new_text)
+      edit = build(:edit, new_revision: new_revision, old_revision: old_revision)
+
+      expect(@feature.calculate(edit)).to eq Wikipedia::VandalismDetection::Features::MISSING_VALUE
+    end
+
+    it "returns missing value if no text inserted" do
+      old_text = Wikipedia::VandalismDetection::Text.new("deletion text")
+      new_text = Wikipedia::VandalismDetection::Text.new("text")
+
+      old_revision = build(:old_revision, text: old_text)
+      new_revision = build(:new_revision, text: new_text)
+      edit = build(:edit, new_revision: new_revision, old_revision: old_revision)
+
+      expect(@feature.calculate(edit)).to eq Wikipedia::VandalismDetection::Features::MISSING_VALUE
+    end
+  end
+end

data/spec/vandalism_detection/features/inserted_external_links_spec.rb

@@ -0,0 +1,35 @@
+require 'spec_helper'
+
+describe Wikipedia::VandalismDetection::Features::InsertedExternalLinks do
+
+  before do
+    @feature = Wikipedia::VandalismDetection::Features::InsertedExternalLinks.new
+  end
+
+  it { should be_a Wikipedia::VandalismDetection::Features::Base }
+
+  describe "#calculate" do
+
+    it "returns the number of inserted external links" do
+      old_revision_text = Wikipedia::VandalismDetection::Text.new '123'
+      new_revision_text = Wikipedia::VandalismDetection::Text.new '123 [http://google.com Google] https://example.com'
+
+      old_revision = build(:old_revision, text: old_revision_text)
+      new_revision = build(:new_revision, text: new_revision_text)
+      edit = build(:edit, old_revision: old_revision, new_revision: new_revision)
+
+      expect(@feature.calculate(edit)).to eq 2
+    end
+
+    it "returns 0 if no inserted text" do
+      old_revision_text = Wikipedia::VandalismDetection::Text.new '123 456789'
+      new_revision_text = Wikipedia::VandalismDetection::Text.new '123' # 0 inserted
+
+      old_revision = build(:old_revision, text: old_revision_text)
+      new_revision = build(:new_revision, text: new_revision_text)
+      edit = build(:edit, old_revision: old_revision, new_revision: new_revision)
+
+      expect(@feature.calculate(edit)).to eq 0
+    end
+  end
+end

data/spec/vandalism_detection/features/inserted_internal_links_spec.rb

@@ -0,0 +1,35 @@
+require 'spec_helper'
+
+describe Wikipedia::VandalismDetection::Features::InsertedInternalLinks do
+
+  before do
+    @feature = Wikipedia::VandalismDetection::Features::InsertedInternalLinks.new
+  end
+
+  it { should be_a Wikipedia::VandalismDetection::Features::Base }
+
+  describe "#calculate" do
+
+    it "returns the number of inserted internal links" do
+      old_revision_text = Wikipedia::VandalismDetection::Text.new '123'
+      new_revision_text = Wikipedia::VandalismDetection::Text.new '123 [[link]] [[linkname|link]]'
+
+      old_revision = build(:old_revision, text: old_revision_text)
+      new_revision = build(:new_revision, text: new_revision_text)
+      edit = build(:edit, old_revision: old_revision, new_revision: new_revision)
+
+      expect(@feature.calculate(edit)).to eq 2
+    end
+
+    it "returns 0 if no inserted text" do
+      old_revision_text = Wikipedia::VandalismDetection::Text.new '123 456789'
+      new_revision_text = Wikipedia::VandalismDetection::Text.new '123' # 0 inserted
+
+      old_revision = build(:old_revision, text: old_revision_text)
+      new_revision = build(:new_revision, text: new_revision_text)
+      edit = build(:edit, old_revision: old_revision, new_revision: new_revision)
+
+      expect(@feature.calculate(edit)).to eq 0
+    end
+  end
+end