Empact-hierclust 0.2.0

data/lib/hierclust/point.rb

@@ -0,0 +1,53 @@
+module Hierclust
+  # A Point represents a single point in n-dimensional space.
+  class Point
+    # x-coordinate
+    attr_accessor :coordinates
+    
+    # Create a new Point with the given coordinates.
+    def initialize(*coordinates)
+      @coordinates = coordinates
+    end
+    
+    # Returns this distance from this Point to an +other+ Point.
+    def distance_to(other)
+      sum_of_squares = coordinates.zip(other.coordinates).map do |point, other_point|
+        (other_point - point) ** 2
+      end.inject(0) {|sum, distance| sum + distance }
+      Math.sqrt(sum_of_squares)
+    end
+    
+    # Simplifies code by letting us treat Clusters and Points interchangeably
+    def size #:nodoc:
+      1
+    end
+    
+    # Simplifies code by letting us treat Clusters and Points interchangeably
+    def radius #:nodoc:
+      0
+    end
+    
+    # Simplifies code by letting us treat Clusters and Points interchangeably
+    def points #:nodoc:
+      [self]
+    end
+    
+    # Returns a legible representation of this Point.
+    def to_s
+      "(#{coordinates.join(', ')})"
+    end
+    
+    # Sorts points relative to each other on the x-axis.
+    # 
+    # Uses y-axis as a tie-breaker, so that sorting is stable even if
+    # multiple points have the same x-coordinate.
+    # 
+    # Uses object_id as a final tie-breaker, so sorts are guaranteed to
+    # be stable even when multiple points have the same coordinates.
+    def <=>(other)
+      cmp = coordinates <=> other.coordinates
+      cmp = object_id <=> other.object_id if cmp == 0
+      cmp
+    end
+  end
+end

data/script/destroy

@@ -0,0 +1,14 @@
+#!/usr/bin/env ruby
+APP_ROOT = File.join(File.dirname(__FILE__), '..')
+
+begin
+  require 'rubigen'
+rescue LoadError
+  require 'rubygems'
+  require 'rubigen'
+end
+require 'rubigen/scripts/destroy'
+
+ARGV.shift if ['--help', '-h'].include?(ARGV[0])
+RubiGen::Base.use_component_sources! [:rubygems, :newgem, :newgem_theme, :test_unit]
+RubiGen::Scripts::Destroy.new.run(ARGV)

data/script/generate

@@ -0,0 +1,14 @@
+#!/usr/bin/env ruby
+APP_ROOT = File.join(File.dirname(__FILE__), '..')
+
+begin
+  require 'rubigen'
+rescue LoadError
+  require 'rubygems'
+  require 'rubigen'
+end
+require 'rubigen/scripts/generate'
+
+ARGV.shift if ['--help', '-h'].include?(ARGV[0])
+RubiGen::Base.use_component_sources! [:rubygems, :newgem, :newgem_theme, :test_unit]
+RubiGen::Scripts::Generate.new.run(ARGV)

data/script/txt2html

@@ -0,0 +1,74 @@
+#!/usr/bin/env ruby
+
+require 'rubygems'
+begin
+  require 'newgem'
+rescue LoadError
+  puts "\n\nGenerating the website requires the newgem RubyGem"
+  puts "Install: gem install newgem\n\n"
+  exit(1)
+end
+require 'redcloth'
+require 'syntax/convertors/html'
+require 'erb'
+require File.dirname(__FILE__) + '/../lib/hierclust/version.rb'
+
+version  = File.exist?('../VERSION') ? File.read('../VERSION') : ""
+download = 'http://rubyforge.org/projects/hierclust'
+
+class Fixnum
+  def ordinal
+    # teens
+    return 'th' if (10..19).include?(self % 100)
+    # others
+    case self % 10
+    when 1: return 'st'
+    when 2: return 'nd'
+    when 3: return 'rd'
+    else    return 'th'
+    end
+  end
+end
+
+class Time
+  def pretty
+    return "#{mday}#{mday.ordinal} #{strftime('%B')} #{year}"
+  end
+end
+
+def convert_syntax(syntax, source)
+  return Syntax::Convertors::HTML.for_syntax(syntax).convert(source).gsub(%r!^<pre>|</pre>$!,'')
+end
+
+if ARGV.length >= 1
+  src, template = ARGV
+  template ||= File.join(File.dirname(__FILE__), '/../website/template.rhtml')
+  
+else
+  puts("Usage: #{File.split($0).last} source.txt [template.rhtml] > output.html")
+  exit!
+end
+
+template = ERB.new(File.open(template).read)
+
+title = nil
+body = nil
+File.open(src) do |fsrc|
+  title_text = fsrc.readline
+  body_text = fsrc.read
+  syntax_items = []
+  body_text.gsub!(%r!<(pre|code)[^>]*?syntax=['"]([^'"]+)[^>]*>(.*?)</\1>!m){
+    ident = syntax_items.length
+    element, syntax, source = $1, $2, $3
+    syntax_items << "<#{element} class='syntax'>#{convert_syntax(syntax, source)}</#{element}>"
+    "syntax-temp-#{ident}"
+  }
+  title = RedCloth.new(title_text).to_html.gsub(%r!<.*?>!,'').strip
+  body = RedCloth.new(body_text).to_html
+  body.gsub!(%r!(?:<pre><code>)?syntax-temp-(\d+)(?:</code></pre>)?!){ syntax_items[$1.to_i] }
+end
+stat = File.stat(src)
+created = stat.ctime
+modified = stat.mtime
+
+$stdout << template.result(binding)

data/spec/hierclust/cluster_spec.rb

@@ -0,0 +1,90 @@
+require File.join(File.dirname(__FILE__), '..', 'spec_helper.rb')
+
+module Hierclust
+  describe Cluster, " with no points" do
+    before do
+      @c = Cluster.new([])
+    end
+
+    it "should have nil x-coordinate" do
+      @c.coordinates.should be_nil
+    end
+
+    it "should have nil radius" do
+      @c.radius.should be_nil
+    end
+  end
+
+  describe Cluster, " with one point" do
+    before do
+      @x = 123
+      @y = 234
+      @p = Point.new(@x, @y)
+      @c = Cluster.new([@p])
+    end
+
+    it "should have the same coordinates as the point" do
+      @c.coordinates.should == @p.coordinates
+    end
+
+    it "should have the same coordinates as used to create the point" do
+      @c.coordinates.should == [@x, @y]
+    end
+
+    it "should have 0 radius" do
+      @c.radius.should == 0
+    end
+  end
+
+  describe Cluster, " with two points" do
+    before do
+      @x_1, @x_2 = 5, 15
+      @y_1, @y_2 = 4, 8
+      @p_1 = Point.new(@x_1, @y_1)
+      @p_2 = Point.new(@x_2, @y_2)
+      @c = Cluster.new([@p_1, @p_2])
+      @points = @c.points
+    end
+
+    it "should have coordinates at the average of points' coordinates" do
+      @c.coordinates.should == [10, 6]
+    end
+
+    it "should have two points" do
+      @points.size.should == 2
+    end
+
+    it "should include both points" do
+      @points.should include(@p_1, @p_2)
+    end
+
+    it "should have correct radius" do
+      radius = Math.sqrt((@x_1 - @x_2) ** 2 + (@y_1 - @y_2) ** 2) / 2.0
+      @c.radius.should == radius
+    end
+  end
+
+  describe Cluster, " with one point and one cluster" do
+    before do
+      @x_1, @x_2, @x_3 = 1, 2, 3
+      @y_1, @y_2, @y_3 = 2, 2, 5
+      @p_1 = Point.new(@x_1, @y_1)
+      @p_2 = Point.new(@x_2, @y_2)
+      @p_3 = Point.new(@x_3, @y_3)
+      @c_1 = Cluster.new([@p_1, @p_2])
+      @c_2 = Cluster.new([@p_3, @c_1])
+    end
+
+    it "should have two items" do
+      @c_2.items.size.should == 2
+    end
+
+    it "should have three points" do
+      @c_2.points.size.should == 3
+    end
+
+    it "should have coordinates at the average of points' coordinates" do
+      @c_2.coordinates.should == [2, 3]
+    end
+  end
+end

data/spec/hierclust/clusterer_spec.rb

@@ -0,0 +1,208 @@
+require File.join(File.dirname(__FILE__), '..', 'spec_helper.rb')
+
+module Hierclust
+  describe Clusterer do
+    describe "with no data" do
+      before do
+        @c = Clusterer.new([])
+      end
+
+      it "should return no clusters" do
+        @c.clusters.should == []
+      end
+    end
+
+    describe "with one point" do
+      before do
+        @x = 1
+        @y = 2
+        @p = Point.new(@x, @y)
+        @c = Clusterer.new([@p])
+      end
+
+      it "should return the point" do
+        @c.clusters.should == [@p]
+      end
+    end
+
+    describe "with two points" do
+      before do
+        @x_1, @x_2 = 1, 5
+        @y_1, @y_2 = 2, 8
+        @p_1 = Point.new(@x_1, @y_1)
+        @p_2 = Point.new(@x_2, @y_2)
+        @c = Clusterer.new([@p_1, @p_2])
+      end
+
+      it "should return one cluster" do
+        @c.clusters.size.should == 1
+      end
+
+      it "should have two points in the cluster" do
+        @c.clusters.first.size.should == 2
+      end
+
+      it "should have the first point in the cluster" do
+        @c.clusters.first.should include(@p_1)
+      end
+
+      it "should have the second point in the cluster" do
+        @c.clusters.first.should include(@p_2)
+      end
+    end
+
+    describe "with three points" do
+      before do
+        @x_1, @x_2, @x_3 = 1, 5, 2
+        @y_1, @y_2, @y_3 = 2, 6, 3
+        @p_1 = Point.new(@x_1, @y_1)
+        @p_2 = Point.new(@x_2, @y_2)
+        @p_3 = Point.new(@x_3, @y_3)
+        @c = Clusterer.new([@p_1, @p_2, @p_3])
+        @cluster = @c.clusters.first
+        @clusters = @cluster.items.sort_by{|c|c.size}
+      end
+
+      it "should return one cluster" do
+        @c.clusters.size.should == 1
+      end
+
+      it "containing two items" do
+        @cluster.items.size.should == 2
+      end
+
+      it "should have one Cluster" do
+        @clusters[1].class.should == Cluster
+      end
+
+      it "and one Point" do
+        @clusters[0].class.should == Point
+      end
+
+      it "should have the first and third points in the bigger cluster" do
+        @clusters[1].should include(@p_1, @p_3)
+      end
+
+      it "should have the second point in the smaller cluster" do
+        @clusters[0].should == @p_2
+      end
+    end
+
+    describe "with four points" do
+      before do
+        @points = [
+          Point.new(0, 1),
+          Point.new(1, 0),
+          Point.new(3, 4),
+          Point.new(4, 3),
+        ]
+      end
+
+      describe "and no separation" do
+        before do
+          @c = Clusterer.new(@points)
+        end
+
+        it "should return one cluster" do
+          @c.clusters.size.should == 1
+        end
+      end
+
+      describe "and separation 1" do
+        before do
+          @c = Clusterer.new(@points, 1)
+        end
+
+        it "should return all four individual points" do
+          @c.clusters.size.should == 4
+        end
+      end
+
+      describe "and separation 2" do
+        before do
+          @c = Clusterer.new(@points, 2)
+        end
+
+        it "should return two clusters" do
+          @c.clusters.size.should == 2
+        end
+      end
+    end
+
+    describe "with eight points" do
+      before do
+        @points = [
+          Point.new(0, 1),
+          Point.new(1, 0),
+          Point.new(3, 4),
+          Point.new(4, 3),
+          Point.new(7, 8),
+          Point.new(8, 7),
+          Point.new(8, 9),
+          Point.new(9, 8),
+        ]
+      end
+  
+      describe "and no separation" do
+        before do
+          @clusters = Clusterer.new(@points).clusters.sort
+        end
+
+        it "should return one cluster when no minimum separation is given" do
+          @clusters.size.should == 1
+        end
+      end
+
+      describe "and separation 1" do
+        before do
+          @clusters = Clusterer.new(@points, 1).clusters.sort
+        end
+
+        it "should have all eight points in individual clusters" do
+          @clusters.size.should == 8
+        end
+      end
+
+      describe "and separation 3" do
+        describe "with no resolution limit" do
+          before do
+            @clusters = Clusterer.new(@points, 3).clusters.sort
+          end
+
+          it "should have three clusters" do
+            @clusters.size.should == 3
+          end
+
+          it "should have clusters size 2, 2, and 4 " do
+            @clusters[0].points.size.should == 2
+            @clusters[1].points.size.should == 2
+            @clusters[2].points.size.should == 4
+          end
+
+          it "should have 2 items in large cluster" do
+            @clusters[2].items.size.should == 2
+          end
+        end
+
+        describe "with coarse resolution" do
+          before do
+            @clusters = Clusterer.new(@points, 3, 5).clusters.sort
+          end
+
+          it "should have three clusters" do
+            @clusters.size.should == 2
+          end
+
+          it "should have clusters size 2, 2, and 4 " do
+            @clusters[0].points.size.should == 4
+            @clusters[1].points.size.should == 4
+          end
+
+          it "should have 4 items in large cluster" do
+            @clusters[1].items.size.should == 4
+          end
+        end
+      end
+    end
+  end
+end