picky 0.12.1 → 0.12.2

data/lib/picky/sources/wrappers/location.rb

@@ -8,19 +8,17 @@ module Sources
       
       # TODO Save min and grid!
       #
-      def initialize backend, options = {}
-        super backend
+      def initialize category, options = {}
+        super category
         
-        @user_grid = extract_user_grid options
-        @precision = extract_precision options
-        
-        @grid      = @user_grid / (@precision + 0.5)
+        @precision   = extract_precision options
+        @calculation = Calculations::Location.new extract_user_grid(options), @precision
       end
       
       #
       #
       def extract_user_grid options
-        options[:grid] || raise # TODO
+        options[:grid] || raise("Option :grid needs to be passed to a location.")
       end
       # Extracts an amount of grids that this 
       # Precision is given in a value.
@@ -41,7 +39,7 @@ module Sources
       # Yield the data (id, text for id) for the given type and category.
       #
       def harvest type, category
-        reset
+        minimum = 1.0/0
         
         # Cache. TODO Make option?
         #
@@ -51,36 +49,32 @@ module Sources
         #
         backend.harvest type, category do |indexed_id, location|
           location = location.to_f
-          @min = location if location < @min
+          minimum = location if location < minimum
           locations << [indexed_id, location]
         end
         
-        # Add a margin.
-        #
-        marginize
+        @calculation.minimum = minimum
         
         # Recalculate locations.
         #
         locations.each do |indexed_id, location|
-          locations_for(location).each do |new_location|
+          locations_for(@calculation.recalculate(location)).each do |new_location|
             yield indexed_id, new_location.to_s
           end
         end
-      end
-      
-      def marginize
-        @min -= @user_grid
+        
+        # TODO Move to the right place.
+        #
+        category.exact[:location_minimum] = minimum
       end
       
       # Put location onto multiple places on a grid.
       #
       # Note: Always returns an integer.
       #
-      def locations_for location
-        new_location = ((location - @min) / grid).floor
-        
-        min_location = new_location - precision
-        max_location = new_location + precision
+      def locations_for repositioned_location
+        min_location = repositioned_location - precision
+        max_location = repositioned_location + precision
         
         (min_location..max_location).to_a
       end

data/lib/tasks/try.rake

@@ -8,14 +8,14 @@ namespace :try do
     
     tokenizer = index_and_category ? Indexes.find(*index_and_category.split(':')).tokenizer : Tokenizers::Index.default
     
-    puts "\"#{text}\" is index tokenized as #{tokenizer.tokenize(text.dup).to_a}"
+    puts "\"#{text}\" is saved in the index as             #{tokenizer.tokenize(text.dup).to_a}"
   end
   
   # desc "Try how a given word would be tokenized when querying."
   task :query, [:text] => :application do |_, options|
     text = options.text
     
-    puts "\"#{text}\" is query tokenized as #{Tokenizers::Query.default.tokenize(text.dup).to_a.map(&:to_s).map(&:to_sym)}"
+    puts "\"#{text}\" as a query will be preprocessed into #{Tokenizers::Query.default.tokenize(text.dup).to_a.map(&:to_s).map(&:to_sym)}"
   end
   
   # desc "Try the given text with both the index and the query (type:category optional)."

data/spec/lib/calculations/location_spec.rb

@@ -0,0 +1,35 @@
+require 'spec_helper'
+
+describe Calculations::Location do
+  
+  context 'without margin' do
+    before(:each) do
+      @calculation = Calculations::Location.new 3, 1
+      @calculation.minimum = 5
+    end
+    describe 'reposition' do
+      it 'calculates correctly' do
+        @calculation.recalculate(13).should == 5
+      end
+      it 'calculates correctly' do
+        @calculation.recalculate(5).should == 1
+      end
+    end
+  end
+  
+  context 'without margin' do
+    before(:each) do
+      @calculation = Calculations::Location.new 3, 3
+      @calculation.minimum = 5
+    end
+    describe 'reposition' do
+      it 'calculates correctly' do
+        @calculation.recalculate(13).should == 12
+      end
+      it 'calculates correctly' do
+        @calculation.recalculate(5).should == 3
+      end
+    end
+  end
+  
+end

data/spec/lib/extensions/array_spec.rb

@@ -7,16 +7,6 @@ describe Array do
       ['fish', 'flash', 'flush', 'smooch'].sort_by_levenshtein!(:fush).should == ['fish', 'flush', 'flash', 'smooch']
     end
   end
-
-  describe 'random' do
-    it 'should choose one element from the array' do
-      left = [1,2,3]
-      100.times do
-        left.delete [1,2,3].random
-      end
-      left.should be_empty
-    end
-  end
   
   describe "clustered_uniq_fast" do
     it "should generate a new array" do

data/spec/lib/extensions/symbol_spec.rb

@@ -7,17 +7,10 @@ describe Symbol do
     before(:each) do
       @token = (((0..9).to_a)*10).to_s.to_sym
     end
-    # Note: They influence each other. each_subtoken is faster though.
-    #
-    it 'should be fast' do
-      timed do
-        @token.subtokens
-      end.should < 0.0009
-    end
     it "should be fast" do
       timed do
         @token.each_subtoken do |subtoken| end
-      end.should < 0.0005
+      end.should < 0.0006
     end
   end
   
@@ -232,65 +225,4 @@ describe Symbol do
     end
   end
   
-  describe "subtokens" do
-    context 'normal symbol' do
-      before(:each) do
-        @sym = :reinke
-      end
-      context 'no downto' do
-        it "should return an array of pieces of the original token, each 1 smaller than the other" do
-          @sym.subtokens.should == [:reinke, :reink, :rein, :rei, :re, :r]
-        end
-      end
-      context 'downto is larger than the symbol' do
-        before(:each) do
-          @downto = 8
-        end
-        it "should return an array of pieces of the original token, each 1 smaller than the other" do
-          @sym.subtokens(@downto).should == [:reinke]
-        end
-      end
-      context 'downto is exactly the same as symbol' do
-        before(:each) do
-          @downto = 6
-        end
-        it "should return an array of pieces of the original token, each 1 smaller than the other" do
-          @sym.subtokens(@downto).should == [:reinke]
-        end
-      end
-      context 'downto is smaller than the length of the symbol' do
-        before(:each) do
-          @downto = 4
-        end
-        it "should return an array of pieces of the original token, each 1 smaller than the other" do
-          @sym.subtokens(@downto).should == [:reinke, :reink, :rein]
-        end
-      end
-      context 'downto is 1' do
-        before(:each) do
-          @downto = 1
-        end
-        it "should return an array of pieces of the original token, each 1 smaller than the other" do
-          @sym.subtokens(@downto).should == [:reinke, :reink, :rein, :rei, :re, :r]
-        end
-      end
-      context 'downto is 0' do
-        before(:each) do
-          @downto = 0
-        end
-        it "should return an array of pieces of the original token, each 1 smaller than the other" do
-          @sym.subtokens(@downto).should == [:reinke, :reink, :rein, :rei, :re, :r, :""]
-        end
-      end
-      context 'downto is less than zero' do
-        before(:each) do
-          @downto = -2
-        end
-        it "should return an array of pieces of the original token, each 1 smaller than the other" do
-          @sym.subtokens(@downto).should == [:reinke, :reink, :rein]
-        end
-      end
-    end
-  end
-
 end

data/spec/lib/index/files_spec.rb

@@ -3,17 +3,18 @@ require 'spec_helper'
 describe Index::Files do
 
   before(:each) do
-    @index         = stub :index, :name => :some_index
-    @category      = stub :category, :name => :some_category
-    @configuration = Configuration::Index.new @index, @category
+    index         = stub :index, :name => :some_index
+    category      = stub :category, :name => :some_category
+    configuration = Configuration::Index.new index, category
     
-    @files         = Index::Files.new :some_name, @configuration
+    @files         = Index::Files.new :some_name, configuration
     
     @prepared      = @files.prepared
     
     @index         = @files.index
-    @similarity    = @files.similarity
     @weights       = @files.weights
+    @similarity    = @files.similarity
+    @configuration = @files.configuration
   end
   
   describe "retrieve" do
@@ -32,6 +33,13 @@ describe Index::Files do
         @files.dump_index :some_hash
       end
     end
+    describe "dump_weights" do
+      it "uses the right file" do
+        @weights.should_receive(:dump).once.with :some_hash
+        
+        @files.dump_weights :some_hash
+      end
+    end
     describe "dump_similarity" do
       it "uses the right file" do
         @similarity.should_receive(:dump).once.with :some_hash
@@ -39,11 +47,11 @@ describe Index::Files do
         @files.dump_similarity :some_hash
       end
     end
-    describe "dump_weights" do
+    describe "dump_configuration" do
       it "uses the right file" do
-        @weights.should_receive(:dump).once.with :some_hash
+        @configuration.should_receive(:dump).once.with :some_hash
         
-        @files.dump_weights :some_hash
+        @files.dump_configuration :some_hash
       end
     end
   end
@@ -61,6 +69,15 @@ describe Index::Files do
         @files.load_index
       end
     end
+    describe "load_weights" do
+      it "uses the right file" do
+        Yajl::Parser.stub! :parse
+        
+        File.should_receive(:open).once.with 'some/search/root/index/test/some_index/some_category_some_name_weights.json', 'r'
+        
+        @files.load_weights
+      end
+    end
     describe "load_similarity" do
       it "uses the right file" do
         Marshal.stub! :load
@@ -70,61 +87,61 @@ describe Index::Files do
         @files.load_similarity
       end
     end
-    describe "load_weights" do
+    describe "load_configuration" do
       it "uses the right file" do
         Yajl::Parser.stub! :parse
         
-        File.should_receive(:open).once.with 'some/search/root/index/test/some_index/some_category_some_name_weights.json', 'r'
+        File.should_receive(:open).once.with 'some/search/root/index/test/some_index/some_category_some_name_configuration.json', 'r'
         
-        @files.load_weights
+        @files.load_configuration
       end
     end
   end
   
   describe "dump indexes" do
     describe "index_cache_ok?" do
-      it "uses the right method" do
+      it 'uses the right method' do
         @index.should_receive(:cache_ok?).once.with
         
         @files.index_cache_ok?
       end
     end
-    describe "similarity_cache_ok?" do
-      it "uses the right method" do
-        @similarity.should_receive(:cache_ok?).once.with
-        
-        @files.similarity_cache_ok?
-      end
-    end
     describe "weights_cache_ok?" do
-      it "uses the right method" do
+      it 'uses the right method' do
         @weights.should_receive(:cache_ok?).once.with
         
         @files.weights_cache_ok?
       end
     end
+    describe "similarity_cache_ok?" do
+      it 'uses the right method' do
+        @similarity.should_receive(:cache_ok?).once.with
+        
+        @files.similarity_cache_ok?
+      end
+    end
   end
   
-  describe "dump indexes" do
-    describe "index_cache_small?" do
-      it "uses the right method" do
+  describe 'dump indexes' do
+    describe 'index_cache_small?' do
+      it 'uses the right method' do
         @index.should_receive(:cache_small?).once.with
         
         @files.index_cache_small?
       end
     end
-    describe "similarity_cache_small?" do
-      it "uses the right method" do
-        @similarity.should_receive(:cache_small?).once.with
+    describe 'weights_cache_small?' do
+      it 'uses the right method' do
+        @weights.should_receive(:cache_small?).once.with
         
-        @files.similarity_cache_small?
+        @files.weights_cache_small?
       end
     end
-    describe "weights_cache_small?" do
-      it "uses the right method" do
-        @weights.should_receive(:cache_small?).once.with
+    describe 'similarity_cache_small?' do
+      it 'uses the right method' do
+        @similarity.should_receive(:cache_small?).once.with
         
-        @files.weights_cache_small?
+        @files.similarity_cache_small?
       end
     end
   end
@@ -132,8 +149,9 @@ describe Index::Files do
   describe 'backup' do
     it 'should call backup on all' do
       @index.should_receive(:backup).once.with
-      @similarity.should_receive(:backup).once.with
       @weights.should_receive(:backup).once.with
+      @similarity.should_receive(:backup).once.with
+      @configuration.should_receive(:backup).once.with
       
       @files.backup
     end
@@ -141,8 +159,9 @@ describe Index::Files do
   describe 'restore' do
     it 'should call delete on all' do
       @index.should_receive(:restore).once.with
-      @similarity.should_receive(:restore).once.with
       @weights.should_receive(:restore).once.with
+      @similarity.should_receive(:restore).once.with
+      @configuration.should_receive(:restore).once.with
       
       @files.restore
     end
@@ -150,8 +169,9 @@ describe Index::Files do
   describe 'delete' do
     it 'should call delete on all' do
       @index.should_receive(:delete).once.with
-      @similarity.should_receive(:delete).once.with
       @weights.should_receive(:delete).once.with
+      @similarity.should_receive(:delete).once.with
+      @configuration.should_receive(:delete).once.with
       
       @files.delete
     end

data/spec/lib/indexed/bundle_spec.rb

@@ -18,16 +18,6 @@ describe Indexed::Bundle do
     end
   end
 
-  describe 'load_from_index_file' do
-    it 'should call two methods in order' do
-      @bundle.should_receive(:load_from_index_generation_message).once.ordered
-      @bundle.should_receive(:clear).once.ordered
-      @bundle.should_receive(:retrieve).once.ordered
-      
-      @bundle.load_from_index_file
-    end
-  end
-
   describe 'ids' do
     before(:each) do
       @bundle.instance_variable_set :@index, { :existing => :some_ids }
@@ -55,8 +45,9 @@ describe Indexed::Bundle do
   describe 'load' do
     it 'should trigger loads' do
       @bundle.should_receive(:load_index).once.with
-      @bundle.should_receive(:load_similarity).once.with
       @bundle.should_receive(:load_weights).once.with
+      @bundle.should_receive(:load_similarity).once.with
+      @bundle.should_receive(:load_configuration).once.with
       
       @bundle.load
     end
@@ -74,6 +65,15 @@ describe Indexed::Bundle do
         @bundle.load_index
       end
     end
+    describe "load_weights" do
+      it "uses the right file" do
+        Yajl::Parser.stub! :parse
+        
+        File.should_receive(:open).once.with 'some/search/root/index/test/some_index/some_category_some_name_weights.json', 'r'
+        
+        @bundle.load_weights
+      end
+    end
     describe "load_similarity" do
       it "uses the right file" do
         Marshal.stub! :load
@@ -83,13 +83,13 @@ describe Indexed::Bundle do
         @bundle.load_similarity
       end
     end
-    describe "load_weights" do
+    describe "load_configuration" do
       it "uses the right file" do
         Yajl::Parser.stub! :parse
         
-        File.should_receive(:open).once.with 'some/search/root/index/test/some_index/some_category_some_name_weights.json', 'r'
+        File.should_receive(:open).once.with 'some/search/root/index/test/some_index/some_category_some_name_configuration.json', 'r'
         
-        @bundle.load_weights
+        @bundle.load_configuration
       end
     end
   end
@@ -111,6 +111,9 @@ describe Indexed::Bundle do
     it 'should initialize the similarity index correctly' do
       @bundle.similarity.should == {}
     end
+    it 'should initialize the configuration correctly' do
+      @bundle.configuration.should == {}
+    end
     it 'should initialize the similarity strategy correctly' do
       @bundle.similarity_strategy.should == :similarity
     end

data/spec/lib/indexed/wrappers/bundle/calculation_spec.rb

@@ -0,0 +1,37 @@
+require 'spec_helper'
+
+describe Indexed::Wrappers::Bundle::Calculation do
+  
+  before(:each) do
+    @bundle       = stub :bundle
+    
+    @calculation = Indexed::Wrappers::Bundle::Calculation.new @bundle
+  end
+  
+  describe 'ids' do
+    it 'calls bundle#ids correctly' do
+      @bundle.should_receive(:ids).once.with :'0.0'
+      
+      @calculation.ids :some_sym
+    end
+    it 'calls bundle#ids correctly' do
+      @bundle.should_receive(:ids).once.with :'6.28'
+      
+      @calculation.ids :'6.28'
+    end
+  end
+  
+  describe 'weight' do
+    it 'calls bundle#ids correctly' do
+      @bundle.should_receive(:weight).once.with :'0.0'
+      
+      @calculation.weight :some_sym
+    end
+    it 'calls bundle#ids correctly' do
+      @bundle.should_receive(:weight).once.with :'6.28'
+      
+      @calculation.weight :'6.28'
+    end
+  end
+  
+end

data/spec/lib/indexed/wrappers/bundle/wrapper_spec.rb

@@ -0,0 +1,27 @@
+require 'spec_helper'
+
+describe Indexed::Wrappers::Bundle::Wrapper do
+  
+  before(:each) do
+    @bundle       = stub :bundle
+    
+    @calculation = Indexed::Wrappers::Bundle::Wrapper.new @bundle
+  end
+  
+  describe 'ids' do
+    it 'calls bundle#ids correctly' do
+      @bundle.should_receive(:ids).once.with :some_sym
+      
+      @calculation.ids :some_sym
+    end
+  end
+  
+  describe 'weight' do
+    it 'calls bundle#ids correctly' do
+      @bundle.should_receive(:weight).once.with :some_sym
+      
+      @calculation.weight :some_sym
+    end
+  end
+  
+end

data/spec/lib/indexing/bundle_spec.rb

@@ -94,8 +94,9 @@ describe Indexing::Bundle do
   describe 'dump' do
     it 'should trigger dumps' do
       @index.should_receive(:dump_index).once.with
-      @index.should_receive(:dump_similarity).once.with
       @index.should_receive(:dump_weights).once.with
+      @index.should_receive(:dump_similarity).once.with
+      @index.should_receive(:dump_configuration).once.with
       
       @index.dump
     end
@@ -222,6 +223,9 @@ describe Indexing::Bundle do
     it 'should initialize the similarity index correctly' do
       @index.similarity.should == {}
     end
+    it 'should initialize the configuration correctly' do
+      @index.configuration.should == {}
+    end
     it 'should initialize the partial strategy correctly' do
       @index.partial_strategy.should == @partial
     end

data/spec/lib/sources/couch_spec.rb

@@ -16,7 +16,7 @@ describe Sources::Couch do
     
     describe "harvest" do
       it "yields the right data" do
-        field = stub :b, :name => :b
+        field = stub :b, :from => :b
         @source.harvest :anything, field do |id, token|
           id.should    eql(7) 
           token.should eql('b data')

data/spec/lib/sources/csv_spec.rb

@@ -1,7 +1,33 @@
 require 'spec_helper'
+require 'csv'
 
 describe Sources::CSV do
   
+  describe 'without separator' do
+    before(:each) do
+      @source = Sources::CSV.new :a, :b, :c, :file => :some_file
+    end
+    it 'calls foreach correctly' do
+      block = lambda { |*args| }
+      
+      ::CSV.should_receive(:foreach).once.with :some_file, {}, &block
+      
+      @source.get_data &block
+    end
+  end
+  describe 'with separator' do
+    before(:each) do
+      @source = Sources::CSV.new :a, :b, :c, :file => :some_file, :col_sep => 'some_separator'
+    end
+    it 'calls foreach correctly' do
+      block = lambda { |*args| }
+      
+      ::CSV.should_receive(:foreach).once.with :some_file, :col_sep => 'some_separator', &block
+      
+      @source.get_data &block
+    end
+  end
+  
   context "without file" do
     it "should fail correctly" do
       lambda { @source = Sources::CSV.new(:a, :b, :c) }.should raise_error(Sources::NoCSVFileGiven)
@@ -9,21 +35,25 @@ describe Sources::CSV do
   end
   context "with file" do
     before(:each) do
-      @source = Sources::CSV.new :a, :b, :c, :file => :some_file
       ::CSV.should_receive(:foreach).any_number_of_times.and_yield ['7', 'a data', 'b data', 'c data']
     end
-    describe "harvest" do
-      it "should yield the right data" do
-        field = stub :b, :name => :b
-        @source.harvest :anything, field do |id, token|
-          [id, token].should == [7, 'b data']
+    context 'without separator' do
+      before(:each) do
+        @source = Sources::CSV.new :a, :b, :c, :file => :some_file
+      end
+      describe "harvest" do
+        it "should yield the right data" do
+          field = stub :b, :from => :b
+          @source.harvest :anything, field do |id, token|
+            [id, token].should == [7, 'b data']
+          end
         end
       end
-    end
-    describe "get_data" do
-      it "should yield each line" do
-        @source.get_data do |data|
-          data.should == ['7', 'a data', 'b data', 'c data']
+      describe "get_data" do
+        it "should yield each line" do
+          @source.get_data do |data|
+            data.should == ['7', 'a data', 'b data', 'c data']
+          end
         end
       end
     end