clusta 0.0.1 → 0.0.2

data/spec/clusta/serialization/tsv_spec.rb

@@ -0,0 +1,133 @@
+require 'spec_helper'
+
+describe Clusta::Serialization::TSV do
+
+  def tsv_serializable_wrapper_class
+    Class.new.tap do |c|
+      c.send(:include, Clusta::Schema)
+      c.send(:include, Clusta::Serialization)
+      c.send(:include, Clusta::Serialization::TSV)
+    end
+  end
+
+  before do
+    @root = tsv_serializable_wrapper_class
+    @root.key :foo
+
+    @child = tsv_serializable_wrapper_class
+    @child.key :baz
+
+    @child.set_stream_name 'Child'
+  end
+
+  describe "processing inputs" do
+
+    it "should assign declared fields" do
+      instance = @root.new("foovalue")
+      instance.foo.should == "foovalue"
+    end
+
+    it "should parse and assign a declared child element field" do
+      @root.field :child, :type => :geometry
+      instance = @root.new("foovalue", "Child;bazvalue")
+      instance.foo.should == "foovalue"
+      instance.child.class.should == @child
+      instance.child.baz.should == "bazvalue"
+    end
+
+    it "should assign a value to an optional field only if it's present" do
+      @root.field :bar, :optional => true
+      
+      instance = @root.new("foovalue")
+      instance.foo.should == "foovalue"
+      instance.bar.should == nil
+
+      instance = @root.new("foovalue", "barvalue")
+      instance.foo.should == "foovalue"
+      instance.bar.should == "barvalue"
+    end
+
+    it "should stash extra arguments it receives at initialization" do
+      instance = @root.new("foovalue", "barvalue")
+      instance.foo.should == "foovalue"
+      instance.extra_inputs.size.should == 1
+      instance.extra_inputs[0].should == "barvalue"
+    end
+
+    it "should parse and stash structured child elements in the extra arguments it receives at initialization" do
+      instance = @root.new("foovalue", "Child;bazvalue")
+      instance.foo.should == "foovalue"
+      instance.extra_inputs.size.should == 1
+      instance.extra_inputs[0].class.should == @child
+      instance.extra_inputs[0].baz.should == "bazvalue"
+    end
+
+  end
+  
+  describe "serializing" do
+
+    it "returns an array appropriate for Wukong" do
+      output = @root.new("foovalue").to_flat
+      output[0].should == @root.stream_name
+      output[1].should == "foovalue"
+    end
+
+    it "returns an array with an optional field at the end only if it has a value" do
+      @root.field :bar, :optional => true
+      output1 = @root.new("foovalue").to_flat
+      output2 = @root.new("foovalue", "barvalue").to_flat
+
+      output1.size.should == 2
+      output1[0].should == @root.stream_name
+      output1[1].should == 'foovalue'
+      
+      output2.size.should == 3
+      output2[0].should == @root.stream_name
+      output2[1].should == 'foovalue'
+      output2[2].should == 'barvalue'
+    end
+
+    it "returns an array including any extra inputs it received" do
+      output = @root.new("foovalue", "barvalue").to_flat
+      output.size.should == 3
+      output[0].should == @root.stream_name
+      output[1].should == "foovalue"
+      output[2].should == "barvalue"
+    end
+    
+    it "returns an array including any extra inputs it received as well as optional inputs" do
+      @root.field :bar, :optional => true
+      
+      instance = @root.new("foovalue", "barvalue", "extra1", "extra2")
+      instance.foo.should == 'foovalue'
+      instance.bar.should == 'barvalue'
+      instance.extra_inputs.should == ['extra1', 'extra2']
+
+      instance = @root.new("foovalue")
+      instance.foo.should == 'foovalue'
+      instance.bar.should be_nil
+      instance.extra_inputs.should be_empty
+    end
+
+    it "returns an array with properly serialized geometry fields" do
+      @root.field :bar, :type => :geometry
+      output = @root.new("foovalue", "Child;bazvalue").to_flat
+      output.size.should == 3
+      output[0].should == @root.stream_name
+      output[1].should == "foovalue"
+      output[2].should == "Child;bazvalue"
+    end
+
+    it "returns an array with properly serialized geometry fields when they were given as extra inputs" do
+      output = @root.new("foovalue", "Child;bazvalue").to_flat
+      output.size.should == 3
+      output[0].should == @root.stream_name
+      output[1].should == "foovalue"
+      output[2].should == "Child;bazvalue"
+    end
+    
+    
+    
+  end
+
+end

data/spec/clusta/serialization_spec.rb

@@ -0,0 +1,27 @@
+require 'spec_helper'
+
+describe Clusta::Serialization do
+
+  before do
+    @root = Class.new
+    @root.send(:include, Clusta::Serialization)
+  end
+
+  it "defines a stream_name method for its class" do
+    @root.stream_name.class.should == String
+  end
+
+  it "defines a stream_name method for its instances" do
+    @root.new.stream_name.should == @root.stream_name
+  end
+
+  it "allows for abbreviating its stream name" do
+    @root.abbreviate 'r'
+  end
+
+  it "allows for setting the stream name" do
+    @root.set_stream_name 'root'
+    @root.stream_name.should == 'root'
+  end
+  
+end

data/spec/clusta/transforms/degree_pairs_to_assortativities_spec.rb

@@ -0,0 +1,13 @@
+require 'spec_helper'
+
+describe Clusta::Transforms::DegreePairsToAssortativities do
+
+  it "undirected degree pairs" do
+    transforming("degree_pairs/undirected.tsv", :with => "degree_pairs_to_assortativities").should have_output("assortativities/undirected.tsv")
+  end
+  
+  it "handles directed degree pairs" do
+    transforming("degree_pairs/directed.tsv", :with => "degree_pairs_to_assortativities").should have_output("assortativities/directed.tsv")
+  end
+  
+end

data/spec/clusta/transforms/{edges_to_vertex_arrows_spec.rb → edges_to_neighborhoods_spec.rb}

@@ -1,21 +1,21 @@
 require 'spec_helper'
 
-describe Clusta::Transforms::EdgesToVertexArrows do
+describe Clusta::Transforms::EdgesToNeighborhoods do
 
   it "handles undirected, unweighted edges" do
-    transforming("edges/undirected.unweighted.tsv", :with => "edges_to_vertex_arrows").should have_output("vertex_arrows/undirected.unweighted.tsv")
+    transforming("edges/undirected.unweighted.tsv", :with => "edges_to_neighborhoods").should have_output("neighborhoods/undirected.unweighted.tsv")
   end
 
   it "handles undirected, weighted edges" do
-    transforming("edges/undirected.weighted.tsv", :with => "edges_to_vertex_arrows").should have_output("vertex_arrows/undirected.weighted.tsv")
+    transforming("edges/undirected.weighted.tsv", :with => "edges_to_neighborhoods").should have_output("neighborhoods/undirected.weighted.tsv")
   end
   
   it "handles directed, unweighted edges" do
-    transforming("edges/directed.unweighted.tsv", :with => "edges_to_vertex_arrows").should have_output("vertex_arrows/directed.unweighted.tsv")
+    transforming("edges/directed.unweighted.tsv", :with => "edges_to_neighborhoods").should have_output("neighborhoods/directed.unweighted.tsv")
   end
   
   it "handles directed, weighted edges" do
-    transforming("edges/directed.weighted.tsv", :with => "edges_to_vertex_arrows").should have_output("vertex_arrows/directed.weighted.tsv")
+    transforming("edges/directed.weighted.tsv", :with => "edges_to_neighborhoods").should have_output("neighborhoods/directed.weighted.tsv")
   end
   
 end

data/spec/clusta/transforms/import_spec.rb

@@ -0,0 +1,9 @@
+require 'spec_helper'
+
+describe Clusta::Transforms::Import do
+
+  it "handles undirected, unweighted edges" do
+    transforming("external/vertices.tsv", :with => "import", :args => '--as=Vertex').should have_output("imports/vertices.labeled.tsv")
+  end
+  
+end

data/spec/clusta/transforms/neighborhoods_to_degree_pairs_spec.rb

@@ -0,0 +1,21 @@
+require 'spec_helper'
+
+describe Clusta::Transforms::NeighborhoodsToDegreePairs do
+
+  it "handles undirected, unweighted neighborhoods" do
+    transforming("neighborhoods/undirected.unweighted.tsv", :with => "neighborhoods_to_degree_pairs").should have_output("degree_pairs/undirected.tsv")
+  end
+
+  it "handles undirected, weighted neighborhoods" do
+    transforming("neighborhoods/undirected.weighted.tsv", :with => "neighborhoods_to_degree_pairs").should have_output("degree_pairs/undirected.tsv")
+  end
+  
+  it "handles directed, unweighted neighborhoods" do
+    transforming("neighborhoods/directed.unweighted.tsv", :with => "neighborhoods_to_degree_pairs").should have_output("degree_pairs/directed.tsv")
+  end
+  
+  it "handles directed, weighted neighborhoods" do
+    transforming("neighborhoods/directed.weighted.tsv", :with => "neighborhoods_to_degree_pairs").should have_output("degree_pairs/directed.tsv")
+  end
+  
+end

data/spec/clusta/transforms/prune_edges_spec.rb

@@ -0,0 +1,22 @@
+require 'spec_helper'
+
+describe Clusta::Transforms::PruneEdges do
+
+  it "handles undirected, unweighted edges" do
+    transforming("edges/undirected.unweighted.tsv", :with => "prune_edges", :args => '--min_weight=0.5 --max_weight=0.8').should have_output("pruned_edges/undirected.unweighted.tsv")
+  end
+
+  it "handles undirected, weighted edges" do
+    transforming("edges/undirected.weighted.tsv", :with => "prune_edges", :args => '--min_weight=0.5 --max_weight=0.8').should have_output("pruned_edges/undirected.weighted.tsv")
+  end
+  
+  it "handles directed, unweighted edges" do
+    transforming("edges/directed.unweighted.tsv", :with => "prune_edges", :args => '--min_weight=0.5 --max_weight=0.8').should have_output("pruned_edges/directed.unweighted.tsv")
+  end
+  
+  it "handles directed, weighted edges" do
+    transforming("edges/directed.weighted.tsv", :with => "prune_edges", :args => '--min_weight=0.5 --max_weight=0.8').should have_output("pruned_edges/directed.weighted.tsv")
+  end
+  
+  
+end

data/spec/data/assortativities/directed.tsv

@@ -0,0 +1,4 @@
+Assortativity	1	1	1
+Assortativity	1	2	5
+Assortativity	2	1	3
+Assortativity	2	2	1

data/spec/data/assortativities/undirected.tsv

@@ -0,0 +1,7 @@
+Assortativity	1	2	1
+Assortativity	1	3	1
+Assortativity	2	1	1
+Assortativity	2	2	4
+Assortativity	2	3	5
+Assortativity	3	1	1
+Assortativity	3	2	5

data/spec/data/degree_pairs/directed.tsv

@@ -0,0 +1,10 @@
+DirectedDegreePair	1	4	1	0	0	1
+DirectedDegreePair	2	1	2	0	0	1
+DirectedDegreePair	2	3	2	0	0	1
+DirectedDegreePair	4	5	1	0	0	2
+DirectedDegreePair	5	6	2	0	0	1
+DirectedDegreePair	5	7	2	0	0	2
+DirectedDegreePair	6	8	1	0	0	2
+DirectedDegreePair	7	5	1	0	0	2
+DirectedDegreePair	8	7	1	0	0	2
+DirectedDegreePair	9	8	1	0	0	2

data/spec/data/degree_pairs/undirected.tsv

@@ -0,0 +1,18 @@
+DegreePair	1	2	2	2
+DegreePair	1	4	2	2
+DegreePair	2	1	2	2
+DegreePair	2	3	2	1
+DegreePair	3	2	1	2
+DegreePair	4	1	2	2
+DegreePair	4	5	2	3
+DegreePair	5	4	3	2
+DegreePair	5	6	3	2
+DegreePair	5	7	3	2
+DegreePair	6	5	2	3
+DegreePair	6	8	2	3
+DegreePair	7	5	2	3
+DegreePair	7	8	2	3
+DegreePair	8	6	3	2
+DegreePair	8	7	3	2
+DegreePair	8	9	3	1
+DegreePair	9	8	1	3

data/spec/data/external/vertices.tsv

@@ -0,0 +1,9 @@
+1
+2
+3
+4
+5
+6
+7
+8
+9

data/spec/data/imports/vertices.labeled.tsv

@@ -0,0 +1,9 @@
+Vertex	1
+Vertex	2
+Vertex	3
+Vertex	4
+Vertex	5
+Vertex	6
+Vertex	7
+Vertex	8
+Vertex	9

data/spec/data/neighborhoods/directed.unweighted.tsv

@@ -0,0 +1,7 @@
+DirectedNeighborhood	1	DirectedNeighbor;2
+DirectedNeighborhood	3	DirectedNeighbor;2
+DirectedNeighborhood	4	DirectedNeighbor;1
+DirectedNeighborhood	5	DirectedNeighbor;4	DirectedNeighbor;7
+DirectedNeighborhood	6	DirectedNeighbor;5
+DirectedNeighborhood	7	DirectedNeighbor;5	DirectedNeighbor;8
+DirectedNeighborhood	8	DirectedNeighbor;6	DirectedNeighbor;9

data/spec/data/neighborhoods/directed.weighted.tsv

@@ -0,0 +1,7 @@
+DirectedNeighborhood	1	DirectedNeighbor;2;0.5
+DirectedNeighborhood	3	DirectedNeighbor;2;0.1
+DirectedNeighborhood	4	DirectedNeighbor;1;0.8
+DirectedNeighborhood	5	DirectedNeighbor;4;0.9	DirectedNeighbor;7;0.2
+DirectedNeighborhood	6	DirectedNeighbor;5;0.4
+DirectedNeighborhood	7	DirectedNeighbor;5;0.3	DirectedNeighbor;8;0.0
+DirectedNeighborhood	8	DirectedNeighbor;6;0.7	DirectedNeighbor;9;1.0

data/spec/data/neighborhoods/undirected.unweighted.tsv

@@ -0,0 +1,9 @@
+Neighborhood	1	Neighbor;2	Neighbor;4
+Neighborhood	2	Neighbor;1	Neighbor;3
+Neighborhood	3	Neighbor;2
+Neighborhood	4	Neighbor;1	Neighbor;5
+Neighborhood	5	Neighbor;4	Neighbor;6	Neighbor;7
+Neighborhood	6	Neighbor;5	Neighbor;8
+Neighborhood	7	Neighbor;5	Neighbor;8
+Neighborhood	8	Neighbor;6	Neighbor;7	Neighbor;9
+Neighborhood	9	Neighbor;8

data/spec/data/neighborhoods/undirected.weighted.tsv

@@ -0,0 +1,9 @@
+Neighborhood	1	Neighbor;2;0.5	Neighbor;4;0.8
+Neighborhood	2	Neighbor;1;0.5	Neighbor;3;0.1
+Neighborhood	3	Neighbor;2;0.1
+Neighborhood	4	Neighbor;1;0.8	Neighbor;5;0.9
+Neighborhood	5	Neighbor;4;0.9	Neighbor;6;0.4	Neighbor;7;0.3
+Neighborhood	6	Neighbor;5;0.4	Neighbor;8;0.7
+Neighborhood	7	Neighbor;5;0.3	Neighbor;8;0.0
+Neighborhood	8	Neighbor;6;0.7	Neighbor;7;0.0	Neighbor;9;1.0
+Neighborhood	9	Neighbor;8;1.0

data/spec/data/pruned_edges/directed.unweighted.tsv

@@ -0,0 +1 @@
+

data/spec/data/pruned_edges/directed.weighted.tsv

@@ -0,0 +1,3 @@
+DirectedEdge	1	2	0.5
+DirectedEdge	4	1	0.8
+DirectedEdge	8	6	0.7

data/spec/data/pruned_edges/undirected.unweighted.tsv

@@ -0,0 +1 @@
+

data/spec/data/pruned_edges/undirected.weighted.tsv

@@ -0,0 +1,3 @@
+Edge	1	2	0.5
+Edge	1	4	0.8
+Edge	6	8	0.7

data/spec/support/transforms_spec_helper.rb

@@ -35,8 +35,12 @@ module Clusta
         @options[:with] or raise SpecError.new("Must supply a transformation name with the :with option.")
       end
 
+      def args
+        @options[:args] || ''
+      end
+
       def command
-        "#{clusta_bin} --run=local --transform=#{transform_name} --run=local #{@path} -"
+        "#{clusta_bin} --run=local --transform=#{transform_name} #{args} #{@path} -"
       end
 
       def output

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: clusta
 version: !ruby/object:Gem::Version
-  version: 0.0.1
+  version: 0.0.2
   prerelease: 
 platform: ruby
 authors:
@@ -9,11 +9,11 @@ authors:
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2012-03-04 00:00:00.000000000Z
+date: 2012-03-29 00:00:00.000000000Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: rspec
-  requirement: &18809420 !ruby/object:Gem::Requirement
+  requirement: &14646260 !ruby/object:Gem::Requirement
     none: false
     requirements:
     - - ! '>='
@@ -21,10 +21,10 @@ dependencies:
         version: '0'
   type: :development
   prerelease: false
-  version_requirements: *18809420
+  version_requirements: *14646260
 - !ruby/object:Gem::Dependency
   name: diffy
-  requirement: &18808980 !ruby/object:Gem::Requirement
+  requirement: &14645140 !ruby/object:Gem::Requirement
     none: false
     requirements:
     - - ! '>='
@@ -32,10 +32,10 @@ dependencies:
         version: '0'
   type: :development
   prerelease: false
-  version_requirements: *18808980
+  version_requirements: *14645140
 - !ruby/object:Gem::Dependency
   name: wukong
-  requirement: &18808560 !ruby/object:Gem::Requirement
+  requirement: &14643340 !ruby/object:Gem::Requirement
     none: false
     requirements:
     - - ! '>='
@@ -43,7 +43,7 @@ dependencies:
         version: '0'
   type: :runtime
   prerelease: false
-  version_requirements: *18808560
+  version_requirements: *14643340
 description: Clusta is a Ruby library that implements network algorithms using Wukong.  This
   means you can use and extend these algorithms on your laptop and seamlessly lift
   them into a Hadoop cluster when you're ready.
@@ -56,43 +56,67 @@ extra_rdoc_files: []
 files:
 - bin/clusta
 - lib/clusta.rb
+- lib/clusta/geometry/neighborhood.rb
 - lib/clusta/geometry/vertex.rb
 - lib/clusta/geometry/element.rb
-- lib/clusta/geometry/vertex_arrows.rb
-- lib/clusta/geometry/directed/vertex_arrows.rb
+- lib/clusta/geometry/directed/neighborhood.rb
 - lib/clusta/geometry/directed/edge.rb
-- lib/clusta/geometry/directed/arrow.rb
-- lib/clusta/geometry/directed/edge_degree_pair.rb
+- lib/clusta/geometry/directed/degree_pair.rb
+- lib/clusta/geometry/directed/neighbor.rb
 - lib/clusta/geometry/directed/degree.rb
+- lib/clusta/geometry/all.rb
 - lib/clusta/geometry/edge.rb
-- lib/clusta/geometry/arrow.rb
-- lib/clusta/geometry/edge_degree_pair.rb
+- lib/clusta/geometry/degree_pair.rb
 - lib/clusta/geometry/assortativity.rb
+- lib/clusta/geometry/neighbor.rb
 - lib/clusta/geometry/degree.rb
 - lib/clusta/transforms.rb
 - lib/clusta/runner.rb
+- lib/clusta/serialization.rb
+- lib/clusta/schema.rb
 - lib/clusta/transforms/import.rb
-- lib/clusta/transforms/edge_degree_pairs_to_assortativities.rb
+- lib/clusta/transforms/neighborhoods_to_degree_pairs.rb
+- lib/clusta/transforms/prune_edges.rb
+- lib/clusta/transforms/pm3d.rb
+- lib/clusta/transforms/edges_to_neighborhoods.rb
 - lib/clusta/transforms/edges_to_degrees.rb
-- lib/clusta/transforms/vertex_arrows_to_edge_degree_pairs.rb
-- lib/clusta/transforms/edges_to_vertex_arrows.rb
+- lib/clusta/transforms/degree_pairs_to_assortativities.rb
+- lib/clusta/serialization/tsv.rb
+- lib/clusta/serialization/json.rb
 - lib/clusta/geometry.rb
-- spec/clusta/geometry/element_spec.rb
+- spec/clusta/schema_spec.rb
+- spec/clusta/transforms/prune_edges_spec.rb
 - spec/clusta/transforms/edges_to_degrees_spec.rb
-- spec/clusta/transforms/edges_to_vertex_arrows_spec.rb
+- spec/clusta/transforms/neighborhoods_to_degree_pairs_spec.rb
+- spec/clusta/transforms/edges_to_neighborhoods_spec.rb
+- spec/clusta/transforms/import_spec.rb
+- spec/clusta/transforms/degree_pairs_to_assortativities_spec.rb
+- spec/clusta/serialization/json_spec.rb
+- spec/clusta/serialization/tsv_spec.rb
+- spec/clusta/serialization_spec.rb
 - spec/spec_helper.rb
 - spec/support/transforms_spec_helper.rb
+- spec/data/degree_pairs/undirected.tsv
+- spec/data/degree_pairs/directed.tsv
 - spec/data/edges/directed.weighted.tsv
 - spec/data/edges/directed.unweighted.tsv
 - spec/data/edges/undirected.weighted.tsv
 - spec/data/edges/undirected.unweighted.tsv
-- spec/data/vertex_arrows/directed.weighted.tsv
-- spec/data/vertex_arrows/directed.unweighted.tsv
-- spec/data/vertex_arrows/undirected.weighted.tsv
-- spec/data/vertex_arrows/undirected.unweighted.tsv
+- spec/data/external/vertices.tsv
 - spec/data/degrees/undirected.tsv
 - spec/data/degrees/directed.tsv
+- spec/data/assortativities/undirected.tsv
+- spec/data/assortativities/directed.tsv
+- spec/data/neighborhoods/directed.weighted.tsv
+- spec/data/neighborhoods/directed.unweighted.tsv
+- spec/data/neighborhoods/undirected.weighted.tsv
+- spec/data/neighborhoods/undirected.unweighted.tsv
 - spec/data/README.rdoc
+- spec/data/imports/vertices.labeled.tsv
+- spec/data/pruned_edges/directed.weighted.tsv
+- spec/data/pruned_edges/directed.unweighted.tsv
+- spec/data/pruned_edges/undirected.weighted.tsv
+- spec/data/pruned_edges/undirected.unweighted.tsv
 - LICENSE
 - README.rdoc
 - VERSION