wukong 3.0.0.pre → 3.0.0.pre2

data/spec/examples/munging/airline_flights/identifiers_spec.rb

@@ -0,0 +1,16 @@
+# require 'spec_helper'
+# require 'wukong'
+
+# load Pathname.path_to(:examples, 'munging/airline_flights/identifiers.rb')
+
+# describe Airport::IdMapping, :only do
+#   it 'loads and reconciles' do
+#     described_class.load(Pathname.path_to(:data, 'airline_flights'))
+#     #
+#     Airport::IdMapping::ID_MAPPINGS.each do |identifier, hsh|
+#       hsh.each do |id, id_mapping|
+#         # puts [identifier, id, id_mapping.to_tsv].join("\t")
+#       end
+#     end
+#   end
+# end

data/spec/examples/munging/airline_flights_spec.rb

@@ -0,0 +1,202 @@
+# require 'spec_helper'
+# require 'wukong'
+# require 'gorillib/datetime/parse'
+
+# load Pathname.path_to(:examples, 'munging/airline_flights/models.rb')
+
+# describe 'Airline Flight Delays Dataset' do
+#   let(:example_tuple    ){ ["2007", "1",  "1", "1", "1232", "1225", "1341", "1340", "WN", "2891",   "N351", "69",  "75", "54",  "1",  "7", "SMF", "ONT", "389", "4", "11", "0", "",  "0", "0", "0", "0", "0", "0"] }
+#   let(:cancelled_tuple_a){ ["2007", "1",  "1", "1",   "NA", "2030",   "NA", "2135", "WN", "2734",      "0", "NA",  "65", "NA", "NA", "NA", "SNA", "LAS", "226", "0",  "0", "1", "A", "0", "0", "0", "0", "0", "0"] }
+#   let(:cancelled_tuple_c){ ["2007", "1",  "4", "4",   "NA", "2120",   "NA", "2125", "WN", "1631",      "0", "NA",  "65", "NA", "NA", "NA", "PHX", "SAN", "304", "0",  "0", "1", "C", "0", "0", "0", "0", "0", "0"] }
+#   let(:diverted_tuple   ){ ["2007", "1", "12", "5", "1054", "1054",   "NA", "1209", "EV", "4351", "N857AS", "NA", "135", "NA", "NA",  "0", "ATL", "TUL", "674", "0", "11", "0", "",  "1", "0", "0", "0", "0", "0"] }
+
+#   let(:raw_flight         ){ RawAirlineFlight.from_tuple(*example_tuple)     }
+#   let(:raw_cancelled      ){ RawAirlineFlight.from_tuple(*cancelled_tuple_a) }
+#   let(:raw_diverted       ){ RawAirlineFlight.from_tuple(*diverted_tuple)    }
+#   let(:example_flight     ){ raw_flight.to_airline_flight                    }
+#   let(:cancelled_flight   ){ raw_cancelled.to_airline_flight              }
+#   let(:diverted_flight    ){ raw_diverted.to_airline_flight               }
+
+#   let(:de_airports_filename   ){ Pathname.path_to(:data, 'airline_flights/dataexpo_airports-raw.csv') }
+
+#   let(:raw_airports_filename  ){ Pathname.path_to(:data, 'airline_flights/openflights_airports-raw.csv') }
+#   let(:raw_airlines_filename  ){ Pathname.path_to(:data, 'airline_flights/openflights_airlines-raw-sample.csv') }
+
+#   let(:example_flight_attrs) { {
+#       flight_datestr: '20070101', unique_carrier: "WN", flight_num: 2891,
+#       from_airport: "SMF", into_airport: "ONT", tail_num: "N351", distance_km: 626, day_of_week: 1,
+#       crs_dep_itime:  1167654300, crs_arr_itime: 1167658800,
+#       act_dep_itime:  1167654720, act_arr_itime: 1167658860,
+#       crs_dep_tod:    "1225",   crs_arr_tod: "1340",
+#       act_dep_tod:    "1232",   act_arr_tod: "1341",
+#       crs_duration:   75, act_duration: 69, air_duration: 54, taxi_in_duration: 4, taxi_out_duration: 11,
+#       is_diverted: false, is_cancelled: false, cancellation_code: "Z",
+#       dep_delay: 7, arr_delay: 1, carrier_delay: 0, weather_delay: 0, nas_delay: 0, security_delay: 0, late_aircraft_delay: 0,
+#     } }
+
+#   describe RawAirlineFlight do
+#     subject{ raw_flight }
+  
+#     it 'loads from a hash' do
+#       p subject.compact_attributes
+#       subject.compact_attributes.should == {
+#         date_year:  2007, date_month: 1, date_day: 1, day_of_week: 1,
+#         act_arr_tod: "1341",         act_dep_tod: "1232",
+#         crs_arr_tod: "1340",         crs_dep_tod: "1225",
+#         # act_arr_itime: 1167658860, act_dep_itime: 1167654720,
+#         # crs_arr_itime: 1167658800, crs_dep_itime: 1167654300,
+#         unique_carrier: "WN", flight_num: 2891, tail_num: "N351",
+#         act_duration: 69, crs_duration: 75, air_duration: 54, arr_delay: 1, dep_delay: 7,
+#         from_airport: "SMF", into_airport: "ONT", distance_mi: 389, taxi_in_duration: 4, taxi_out_duration: 11,
+#         is_cancelled: false, cancellation_code: "Z", is_diverted: false,
+#         carrier_delay: 0, weather_delay: 0, nas_delay: 0, security_delay: 0, late_aircraft_delay: 0,
+#       }
+#     end
+  
+#     it 'loads cancelled flights OK' do
+#       # ff = RawAirlineFlight.fields[:act_dep_itime].type
+#       flight = described_class.from_tuple(*cancelled_tuple_a)
+#       p flight.compact_attributes
+#       flight.compact_attributes.should == {
+#         date_year:  2007, date_month: 1, date_day: 1, day_of_week: 1,
+#         act_arr_tod: nil,            act_dep_tod: nil,
+#         crs_arr_tod: "2135",         crs_dep_tod: "2030",
+#         # act_arr_itime: nil,        act_dep_itime: nil,
+#         # crs_arr_itime: 1167687300, crs_dep_itime: 1167683400,
+#         unique_carrier: "WN", flight_num: 2734,  tail_num: nil,
+#         act_duration: nil, crs_duration: 65, air_duration: nil, arr_delay: nil, dep_delay: nil,
+#         from_airport: "SNA", into_airport: "LAS", distance_mi: 226, taxi_in_duration: 0, taxi_out_duration: 0,
+#         is_cancelled: true, cancellation_code: "A", is_diverted: false,
+#         carrier_delay: 0, weather_delay: 0, nas_delay: 0, security_delay: 0, late_aircraft_delay: 0,
+#       }
+#     end
+  
+#     it 'loads diverted flights OK' do
+#       # ff = RawAirlineFlight.fields[:act_dep_itime].type
+#       flight = described_class.from_tuple(*diverted_tuple)
+#       p flight.compact_attributes
+#       flight.compact_attributes.should == {
+#         date_year:  2007, date_month: 1, date_day: 12, day_of_week: 5,
+#         act_arr_tod: nil,            act_dep_tod: "1054",
+#         crs_arr_tod: "1209",         crs_dep_tod: "1054",
+#         # act_arr_itime: nil,        act_dep_itime: 1168599240,
+#         # crs_arr_itime: 1168603740, crs_dep_itime: 1168599240,
+#         unique_carrier: "EV", flight_num: 4351, tail_num: "N857AS",
+#         act_duration: nil, crs_duration: 135, air_duration: nil, arr_delay: nil, dep_delay: 0,
+#         from_airport: "ATL", into_airport: "TUL", distance_mi: 674, taxi_in_duration: 0, taxi_out_duration: 11,
+#         is_cancelled: false, cancellation_code: "Z", is_diverted: true,
+#         carrier_delay: 0, weather_delay: 0, nas_delay: 0, security_delay: 0, late_aircraft_delay: 0,
+#       }
+#     end
+  
+#     it 'does dates right' do
+#       { normal:    [example_tuple,     raw_flight],
+#         cancelled: [cancelled_tuple_a, raw_cancelled],
+#         diverted:  [diverted_tuple,    raw_diverted],
+#       }.each do |label, (raw_values, raw_flight)|
+#         [ [raw_flight.act_dep_itime, raw_values[4] ],
+#           [raw_flight.crs_dep_itime, raw_values[5] ],
+#           [raw_flight.act_arr_itime, raw_values[6] ],
+#           [raw_flight.crs_arr_itime, raw_values[7] ],
+#         ].each do |itime, hhmm|
+#           next unless itime
+#           tm = Time.at(itime).utc
+#           (tm.hour * 100 + tm.min).to_s.should == hhmm
+#         end
+#       end
+#     end
+  
+#     it 'receives idempotently' do
+#       subject.should == RawAirlineFlight.receive(subject.compact_attributes)
+#     end
+  
+#     it '#to_airline_flight' do
+#       flight = subject.to_airline_flight
+#       flight.should be_a(AirlineFlight)
+#       flight.compact_attributes.should == example_flight_attrs
+#     end
+#   end
+  
+#   describe AirlineFlight do
+#     subject{ example_flight }
+  
+#     it "makes sense" do
+#       { normal: example_flight, cancelled: cancelled_flight, diverted: diverted_flight
+#       }.each do |label, flight|
+#         linted = subject.lint
+#         p [label, linted, flight] unless linted.values.all?
+#         linted.values.should be_all
+#       end
+#     end
+  
+#     it 'has correct field alignment' do
+#       described_class.field_names.should == example_flight_attrs.keys
+#       described_class.fields.values.map(&:position).should == (0..30).to_a
+#     end
+
+#     it 'calculates local times correctly' do
+#       Airport.load(raw_airports_filename)
+#       Airport::AIRPORTS.each{|id,airport| puts airport.to_tsv }
+#     end
+  
+#   end
+  
+#   describe 'parsing raw' do
+#     it 'works' do
+#       raw_file = File.open(raw_airlines_filename)
+#       raw_file.readline
+#       puts AirlineFlight.field_names.map{|fn| fn[0..6] }.join("\t")
+#       raw_file.each do |line|
+#         tuple = line.split(',')
+#         # next unless tuple[23] == "1"
+#         raw_flight = RawAirlineFlight.from_tuple(*tuple)
+#         flight = raw_flight.to_airline_flight
+#         # if not flight.lint.values.all?
+#           # p flight.lint.values
+#           puts flight.to_tsv
+#           # p [
+#           #   [raw_flight.crs_dep_itime, tuple[5] ],
+#           #   [raw_flight.crs_arr_itime, tuple[7] ],
+#           #   (raw_flight.crs_arr_itime - raw_flight.crs_arr_itime),
+#           #   (raw_flight.crs_arr_itime - raw_flight.crs_dep_itime)/60.0,
+#           #   raw_flight.crs_duration
+#           # ]
+#         # end
+#       end
+#     end
+#   end
+
+#   describe RawDataexpoAirport do
+#     it 'works' do
+#       puts described_class.field_names.map{|fn| fn[0..6] }.join("\t")
+#       raw_airports = RawDataexpoAirport.load_csv(de_airports_filename)
+#       raw_airports.each do |airport|
+#         puts airport.to_tsv
+#       end
+#     end
+#   end
+
+#   describe RawOpenflightAirport do
+#     it 'works' do
+#       puts described_class.field_names.join("\t") # .map{|fn| fn[0..6] }.join("\t")
+#       raw_airports = described_class.load_csv(raw_airports_filename)
+#       raw_airports.each do |airport|
+#         # puts airport.to_tsv
+#         linted = airport.lint
+#         puts [airport.iata, airport.icao, linted.inspect, airport.to_tsv, ].join("\t") if linted.present?
+#       end
+#     end
+#   end
+  
+#   describe Airport do
+#     it 'loads and reconciles' do
+#       Airport.load(raw_airports_filename, de_airports_filename)
+#       Airport::AIRPORTS.each{|id,airport|
+#         #puts airport.to_tsv
+#         linted = airport.lint
+#         warn [airport.iata, airport.icao, airport.de_iata, "%-25s" % airport.name, linted.inspect].join("\t") if linted.present?
+#       }
+#     end
+#   end
+
+# end

data/spec/examples/text/pig_latin_spec.rb

@@ -1,21 +1,18 @@
 require 'spec_helper'
-require 'wukong'
-require 'wukong/local_runner'
+# require 'wukong'
 
-load Pathname.path_to(:examples, 'text/pig_latin.rb')
+# describe_example_script(:pig_latin, 'text/pig_latin.rb') do
 
-describe 'Pig Latin', :examples_spec => true, :helpers => true do
+#   context 'processor' do
+#     subject{ Wukong::Widget::PigLatinize.new }
+#     it 'breaks text into pig latin' do
+#       subject.should_receive(:emit).with("Iway indfay ethay astramipay otay ebay ethay ostmay ensualsay ofway allway ethay altedsay uredcay eatsmay.")
+#       subject.process("I find the pastrami to be the most sensual of all the salted cured meats.")
+#     end
+#   end
 
-  context 'processor' do
-    subject{ Wukong::Widget::PigLatinize.new }
-    it 'breaks text into pig latin' do
-      subject.should_receive(:emit).with("Iway indfay ethay astramipay otay ebay ethay ostmay ensualsay ofway allway ethay altedsay uredcay eatsmay.")
-      subject.process("I find the pastrami to be the most sensual of all the salted cured meats.")
-    end
-  end
+#   it 'runs' do
+#     Wukong::LocalRunner.run(ExampleUniverse.dataflow(:pig_latin), :default)
+#   end
 
-  it 'runs' do
-    Wukong::LocalRunner.run(ExampleUniverse.dataflow(:pig_latin), :default)
-  end
-
-end
+# end

data/spec/examples/workflow/cherry_pie_spec.rb

@@ -1,6 +1,36 @@
 require 'spec_helper'
-require 'wukong'
-require 'wukong/local_runner'
+# require 'wukong'
 
-require 'hanuman/graphvizzer'
-require 'hanuman/graphviz'
+# require 'hanuman/graphvizzer/gv_presenter'
+
+# describe 'cherry_pie' do
+
+#   # it 'fiddles' do
+#   #   load(Pathname.path_to(:examples, 'workflow/cherry_pie.rb'))
+#   #   Wukong.workflow(:cherry_pie) do
+#   #     Log.dump self
+#   #
+#   #     graph(:crust) do
+#   #       Log.dump self
+#   #       action(:bob)
+#   #     end
+#   #
+#   #     Log.dump graph(:crust)
+#   #
+#   #   end
+#   # end
+
+
+#   describe 'Graphviz View', :if => GRAPHVIZ do
+#     it 'makes a png' do
+#       require Pathname.path_to(:examples, 'workflow/cherry_pie.rb')
+
+#       gv = Wukong.to_graphviz
+
+#       basename = Pathname.path_to(:tmp, 'cherry_pie')
+#       gv.save(basename, 'png')
+#       puts File.read("#{basename}.dot")
+#     end
+#   end
+
+# end

data/spec/hanuman/graph_spec.rb

@@ -1,7 +1,6 @@
 require 'spec_helper'
-require 'hanuman'
 
-describe Hanuman::Graph, :helpers => true do
+describe Hanuman::Graph, :hanuman => true do
 
   # it 'makes a tree' do
   #   example_graph.tree.should == {
@@ -15,3 +14,29 @@ describe Hanuman::Graph, :helpers => true do
   # end
 
 end
+
+describe Hanuman::GraphBuilder, :hanuman => true do
+
+  it_behaves_like 'a Stage::Builder'
+  
+  its(:namespace){ should be(Hanuman::Graph) }
+  
+  context '#define' do
+    let(:block_arg){ ->(){ def say() "I'm the Juggernaut!" ; end } }
+    
+    it 'evalutes a supplied block itself' do
+      subject.label = :juggernaut
+      subject.should_receive(:instance_eval).with(&block_arg)
+      subject.define(&block_arg)
+    end
+  end
+  
+  context '#build' do
+  end
+  
+  context '#serialize' do
+    it 'serializes into a Hash with stages and links' do
+      subject.serialize.should include(:stages, :links)
+    end  
+  end
+end

data/spec/hanuman/hanuman_spec.rb

@@ -0,0 +1,10 @@
+require 'spec_helper'
+
+describe Hanuman do
+  
+  it_behaves_like Hanuman::Shortcuts
+
+  it{ should respond_to(:stage) }
+  it{ should respond_to(:graph) }
+
+end

data/spec/hanuman/registry_spec.rb

@@ -0,0 +1,123 @@
+require 'spec_helper'
+
+describe Hanuman::Registry, :hanuman => true do
+
+  before(:each) do
+    @orig_reg = Hanuman.registry.show
+    Hanuman.registry.clear!
+  end
+
+  after(:each) do
+    Hanuman.registry.clear!
+    Hanuman.registry.merge!(@orig_reg)
+  end
+  
+  let(:definition){ { :universe => 'Marvel' }                 }
+  let(:builder)   { Hanuman::StageBuilder.receive(definition) }
+  
+  context '#create' do
+    context 'non-existing entry' do
+      it 'creates a new entry for supplied definition' do
+        subject.create(:storm, builder).should be true
+        subject.should be_registered(:storm)
+        subject.retrieve(:storm).should == builder
+      end
+    end
+
+    context 'existing' do
+      it 'does not create a new entry' do
+        subject.create(:storm, builder)
+        subject.create(:storm, builder).should be false
+      end
+    end
+  end
+
+  context '#update' do
+    context 'non-existing entry' do
+      it 'does not update the entry' do
+        subject.update(:rogue, definition).should be false
+        subject.should_not be_registered(:rogue)
+      end
+    end
+
+    context 'existing' do
+      let(:new_definition){ { :universe => 'alternate' } }
+      
+      it 'updates the entry' do
+        subject.create(:rogue, builder)
+        subject.update(:rogue, new_definition).should be true
+        subject.retrieve(:rogue).serialize.should include(new_definition)
+      end
+    end    
+  end
+
+  context '#create_or_update' do
+    context 'non-existing entry' do
+      it 'creates the entry' do
+        subject.should_not be_registered(:cyclops)
+        subject.create_or_update(:cyclops, definition).should be true
+        subject.retrieve(:cyclops).should == definition
+      end
+    end
+
+    context 'existing' do
+      let(:new_definition){ { :universe => 'alternate' } }
+
+      it 'updates the entry' do
+        subject.create(:cyclops, builder)
+        subject.create_or_update(:cyclops, new_definition).should be true
+        subject.retrieve(:cyclops).serialize.should include(new_definition)
+      end
+    end
+  end
+  
+  context '#decorate_with_registry' do
+    let(:graph_builder){ Hanuman::GraphBuilder.receive(label: :beastman) }
+    let(:definition)   { Hanuman::StageBuilder.receive(label: :mystique) }
+    
+    before(:each) do
+      subject.create(:mystique, definition)
+      subject.decorate_with_registry(graph_builder)            
+    end
+
+    it 'decorates a builder with the registry entries as methods' do
+      graph_builder.should respond_to(:mystique)
+    end
+
+    it 'decorates using singleton method definitions' do
+      graph_builder.dup.should_not respond_to(:mystique)
+    end
+    
+    it 'returns an instance of Hanuman::StageBuilder for chaining' do
+      graph_builder.send(:mystique).should be_instance_of(Hanuman::StageBuilder)
+    end
+    
+    it 'adds the instance of Hanuman::StageBuilder to its :stages attribute when used' do
+      graph_builder.stages.should_not include(:mystique)
+      graph_builder.send(:mystique)
+      graph_builder.stages.should include(:mystique)
+    end
+
+    it 'retrieves the Hanuman::StageBuilder from the registry' do
+      Hanuman.registry.should_receive(:retrieve).with(:mystique).and_return(definition)
+      graph_builder.send(:mystique)
+    end
+
+    it 'saves updated definitions in its :stages attribute' do
+      graph_builder.send(:mystique, for_class: String)
+      graph_builder.stages[:mystique].for_class.should == String
+    end
+    
+    it 'allows for relabelling of Hanuman::StageBuilders' do
+      graph_builder.send(:mystique, label: :colossus)
+      graph_builder.stages.should include(:colossus)
+      graph_builder.stages.should_not include(:mystique)
+    end
+  end
+  
+  context '#show' do
+    it 'shows what is currently registered' do
+      subject.show.should be_a(Hash)
+    end
+  end  
+end