pduey-sunspot 1.2.1.1

data/spec/integration/dynamic_fields_spec.rb

@@ -0,0 +1,57 @@
+require File.expand_path('spec_helper', File.dirname(__FILE__))
+
+describe 'dynamic fields' do
+  before :each do
+    Sunspot.remove_all
+    @posts = Post.new(:custom_string => { :cuisine => 'Pizza' }),
+             Post.new(:custom_string => { :cuisine => 'Greek' }),
+             Post.new(:custom_string => { :cuisine => 'Greek' })
+    Sunspot.index!(@posts)
+  end
+
+  it 'should search for dynamic string field' do
+    Sunspot.search(Post) do
+      dynamic(:custom_string) do
+        with(:cuisine, 'Pizza')
+      end
+    end.results.should == [@posts.first]
+  end
+
+  describe 'faceting' do
+    before :each do
+      @search = Sunspot.search(Post) do
+        dynamic :custom_string do
+          facet :cuisine
+        end
+      end
+    end
+
+    it 'should return value "value" with count 2' do
+      row = @search.dynamic_facet(:custom_string, :cuisine).rows[0]
+      row.value.should == 'Greek'
+      row.count.should == 2
+    end
+
+    it 'should return value "other" with count 1' do
+      row = @search.dynamic_facet(:custom_string, :cuisine).rows[1]
+      row.value.should == 'Pizza'
+      row.count.should == 1
+    end
+  end
+
+  it 'should order by dynamic string field ascending' do
+    Sunspot.search(Post) do
+      dynamic :custom_string do
+        order_by :cuisine, :asc
+      end
+    end.results.last.should == @posts.first
+  end
+
+  it 'should order by dynamic string field descending' do
+    Sunspot.search(Post) do
+      dynamic :custom_string do
+        order_by :cuisine, :desc
+      end
+    end.results.first.should == @posts.first
+  end
+end

data/spec/integration/faceting_spec.rb

@@ -0,0 +1,238 @@
+require File.expand_path('spec_helper', File.dirname(__FILE__))
+
+describe 'search faceting' do
+  def self.test_field_type(name, attribute, field, *args)
+    clazz, value1, value2 =
+      if args.length == 2
+        [Post, args.first, args.last]
+      else
+        args
+      end
+
+    context "with field of type #{name}" do
+      before :all do
+        Sunspot.remove_all
+        2.times do
+          Sunspot.index(clazz.new(attribute => value1))
+        end
+        Sunspot.index(clazz.new(attribute => value2))
+        Sunspot.commit
+      end
+
+      before :each do
+        @search = Sunspot.search(clazz) do
+          facet(field)
+        end
+      end
+
+      it "should return value #{value1.inspect} with count 2" do
+        row = @search.facet(field).rows[0]
+        row.value.should == value1
+        row.count.should == 2
+      end
+
+      it "should return value #{value2.inspect} with count 1" do
+        row = @search.facet(field).rows[1]
+        row.value.should == value2
+        row.count.should == 1
+      end
+    end
+  end
+
+  test_field_type('String', :title, :title, 'Title 1', 'Title 2')
+  test_field_type('Integer', :blog_id, :blog_id, 3, 4)
+  test_field_type('Float', :ratings_average, :average_rating, 2.2, 1.1)
+  test_field_type('Time', :published_at, :published_at, Time.mktime(2008, 02, 17, 17, 45, 04),
+                                                        Time.mktime(2008, 07, 02, 03, 56, 22))
+  test_field_type('Trie Integer', :size, :size, Photo, 3, 4)
+  test_field_type('Float', :average_rating, :average_rating, Photo, 2.2, 1.1)
+  test_field_type('Time', :created_at, :created_at, Photo, Time.mktime(2008, 02, 17, 17, 45, 04),
+                                                           Time.mktime(2008, 07, 02, 03, 56, 22))
+  test_field_type('Boolean', :featured, :featured, true, false)
+
+  context 'facet options' do
+    before :all do
+      Sunspot.remove_all
+      facet_values = %w(zero one two three four)
+      facet_values.each_with_index do |value, i|
+        i.times { Sunspot.index(Post.new(:title => value, :blog_id => 1)) }
+      end
+      Sunspot.index(Post.new(:blog_id => 1))
+      Sunspot.index(Post.new(:title => 'zero', :blog_id => 2))
+      Sunspot.commit
+    end
+
+    it 'should limit the number of facet rows' do
+      search = Sunspot.search(Post) do
+        facet :title, :limit => 3
+      end
+      search.facet(:title).should have(3).rows
+    end
+
+    it 'should not return zeros by default' do
+      search = Sunspot.search(Post) do
+        with :blog_id, 1
+        facet :title
+      end
+      search.facet(:title).rows.map { |row| row.value }.should_not include('zero')
+    end
+
+    it 'should return zeros when specified' do
+      search = Sunspot.search(Post) do
+        with :blog_id, 1
+        facet :title, :zeros => true
+      end
+      search.facet(:title).rows.map { |row| row.value }.should include('zero')
+    end
+
+    it 'should return a specified minimum count' do
+      search = Sunspot.search(Post) do
+        with :blog_id, 1
+        facet :title, :minimum_count => 2
+      end
+      search.facet(:title).rows.map { |row| row.value }.should == %w(four three two)
+    end
+
+    it 'should order facets lexically' do
+      search = Sunspot.search(Post) do
+        with :blog_id, 1
+        facet :title, :sort => :index
+      end
+      search.facet(:title).rows.map { |row| row.value }.should == %w(four one three two)
+    end
+
+    it 'should order facets by count' do
+      search = Sunspot.search(Post) do
+        with :blog_id, 1
+        facet :title, :sort => :count
+      end
+      search.facet(:title).rows.map { |row| row.value }.should == %w(four three two one)
+    end
+
+    it 'should limit facet values by prefix' do
+      search = Sunspot.search(Post) do
+        with :blog_id, 1
+        facet :title, :prefix => 't'
+      end
+      search.facet(:title).rows.map { |row| row.value }.sort.should == %w(three two)
+    end
+
+    it 'should return :all facet' do
+      search = Sunspot.search(Post) do
+        with :blog_id, 1
+        facet :title, :extra => :any
+      end
+      search.facet(:title).rows.first.value.should == :any
+      search.facet(:title).rows.first.count.should == 10
+    end
+
+    it 'should return :none facet' do
+      search = Sunspot.search(Post) do
+        with :blog_id, 1
+        facet :title, :extra => :none
+      end
+      search.facet(:title).rows.first.value.should == :none
+      search.facet(:title).rows.first.count.should == 1
+    end
+  end
+
+  context 'multiselect faceting' do
+    before do
+      Sunspot.remove_all
+      Sunspot.index!(
+        Post.new(:blog_id => 1, :category_ids => [1]),
+        Post.new(:blog_id => 1, :category_ids => [2]),
+        Post.new(:blog_id => 3, :category_ids => [3])
+      )
+    end
+
+    it 'should exclude filter from faceting' do
+      search = Sunspot.search(Post) do
+        with(:blog_id, 1)
+        category_filter = with(:category_ids, 1)
+        facet(:category_ids, :exclude => category_filter)
+      end
+      search.facet(:category_ids).rows.map { |row| row.value }.to_set.should == Set[1, 2]
+    end
+
+    it 'should use facet keys to facet more than once with different exclusions' do
+      search = Sunspot.search(Post) do
+        with(:blog_id, 1)
+        category_filter = with(:category_ids, 1)
+        facet(:category_ids)
+        facet(:category_ids, :exclude => category_filter, :name => :all_category_ids)
+      end
+      search.facet(:category_ids).rows.map { |row| row.value }.should == [1]
+      search.facet(:all_category_ids).rows.map { |row| row.value }.to_set.should == Set[1, 2]
+    end
+  end
+
+  context 'date facets' do
+    before :all do
+      Sunspot.remove_all
+      time = Time.utc(2009, 7, 8)
+      Sunspot.index!(
+        (0..2).map { |i| Post.new(:published_at => time + i*60*60*16) }
+      )
+    end
+
+    it 'should return time ranges' do
+      time = Time.utc(2009, 7, 8)
+      search = Sunspot.search(Post) do
+        facet :published_at, :time_range => time..(time + 60*60*24*2), :sort => :count
+      end
+      search.facet(:published_at).rows.first.value.should == (time..(time + 60*60*24))
+      search.facet(:published_at).rows.first.count.should == 2
+      search.facet(:published_at).rows.last.value.should == ((time + 60*60*24)..(time + 60*60*24*2))
+      search.facet(:published_at).rows.last.count.should == 1
+    end
+  end
+
+  context 'class facets' do
+    before :all do
+      Sunspot.remove_all
+      Sunspot.index!(Post.new, Post.new, Namespaced::Comment.new)
+    end
+
+    it 'should return classes' do
+      search = Sunspot.search(Post, Namespaced::Comment) do
+        facet(:class, :sort => :count)
+      end
+      search.facet(:class).rows.first.value.should == Post
+      search.facet(:class).rows.first.count.should == 2
+      search.facet(:class).rows.last.value.should == Namespaced::Comment
+      search.facet(:class).rows.last.count.should == 1
+    end
+  end
+
+  context 'query facets' do
+    before :all do
+      Sunspot.remove_all
+      Sunspot.index!(
+        [1.1, 1.2, 3.2, 3.4, 3.9, 4.1].map do |rating|
+          Post.new(:ratings_average => rating)
+        end
+      )
+    end
+
+    it 'should return specified facets' do
+      search = Sunspot.search(Post) do
+        facet :rating_range, :sort => :count do
+          for rating in [1.0, 2.0, 3.0, 4.0]
+            range = rating..(rating + 1.0)
+            row range do
+              with :average_rating, rating..(rating + 1.0)
+            end
+          end
+        end
+      end
+      facet = search.facet(:rating_range)
+      facet.rows[0].value.should == (3.0..4.0)
+      facet.rows[0].count.should == 3
+      facet.rows[1].value.should == (1.0..2.0)
+      facet.rows[1].count.should == 2
+      facet.rows[2].value.should == (4.0..5.0)
+      facet.rows[2].count.should == 1
+    end
+  end
+end

data/spec/integration/highlighting_spec.rb

@@ -0,0 +1,24 @@
+require File.expand_path('spec_helper', File.dirname(__FILE__))
+
+describe 'keyword highlighting' do
+  before :all do
+    @posts = []
+    @posts << Post.new(:body => 'And the fox laughed')
+    @posts << Post.new(:body => 'Lorem ipsum dolor sit amet, consectetur adipiscing elit', :blog_id => 1)
+    Sunspot.index!(*@posts)
+    @search_result = Sunspot.search(Post) { keywords 'fox', :highlight => true }
+  end
+  
+  it 'should include highlights in the results' do
+    @search_result.hits.first.highlights.length.should == 1
+  end
+  
+  it 'should return formatted highlight fragments' do
+    @search_result.hits.first.highlights(:body).first.format.should == 'And the <em>fox</em> laughed'
+  end
+  
+  it 'should be empty for non-keyword searches' do
+    search_result = Sunspot.search(Post){ with :blog_id, 1 }
+    search_result.hits.first.highlights.should be_empty
+  end
+end

data/spec/integration/indexing_spec.rb

@@ -0,0 +1,33 @@
+require File.expand_path('spec_helper', File.dirname(__FILE__))
+
+describe 'indexing' do
+  it 'should index non-multivalued field with newlines' do
+    lambda do
+      Sunspot.index!(Post.new(:title => "A\nTitle"))
+    end.should_not raise_error
+  end
+
+  it 'should correctly remove by model instance' do
+    post = Post.new(:title => 'test post')
+    Sunspot.index!(post)
+    Sunspot.remove!(post)
+    Sunspot.search(Post) { with(:title, 'test post') }.results.should be_empty
+  end
+
+  it 'should correctly delete by ID' do
+    post = Post.new(:title => 'test post')
+    Sunspot.index!(post)
+    Sunspot.remove_by_id!(Post, post.id)
+    Sunspot.search(Post) { with(:title, 'test post') }.results.should be_empty
+  end
+
+  it 'removes documents by query' do
+    Sunspot.remove_all!
+    posts = [Post.new(:title => 'birds'), Post.new(:title => 'monkeys')]
+    Sunspot.index!(posts)
+    Sunspot.remove! do
+      with(:title, 'birds')
+    end
+    Sunspot.search(Post).should have(2).results
+  end
+end

data/spec/integration/keyword_search_spec.rb

@@ -0,0 +1,317 @@
+require File.expand_path('spec_helper', File.dirname(__FILE__))
+
+describe 'keyword search' do
+  describe 'generally' do
+    before :all do
+      Sunspot.remove_all
+      @posts = []
+      @posts << Post.new(:title => 'The toast elects the insufficient spirit',
+                         :body => 'Does the wind write?')
+      @posts << Post.new(:title => 'A nail abbreviates the recovering insight outside the moron',
+                         :body => 'The interpreted strain scans the buffer around the upper temper')
+      @posts << Post.new(:title => 'The toast abbreviates the recovering spirit',
+                         :body => 'Does the host\'s wind interpret the buffer, moron?')
+      Sunspot.index!(*@posts)
+      @comment = Namespaced::Comment.new(:body => 'Hey there where ya goin, not exactly knowin, who says you have to call just one place toast.')
+      Sunspot.index!(@comment)
+    end
+
+    it 'matches a single keyword out of a single field' do
+      results = Sunspot.search(Post) { keywords 'toast' }.results
+      [0, 2].each { |i| results.should include(@posts[i]) }
+      [1].each { |i| results.should_not include(@posts[i]) }
+    end
+
+    it 'matches multiple words out of a single field' do
+      results = Sunspot.search(Post) { keywords 'elects toast' }.results
+      results.should == [@posts[0]]
+    end
+
+    it 'matches multiple words in multiple fields' do
+      results = Sunspot.search(Post) { keywords 'toast wind' }.results
+      [0, 2].each { |i| results.should include(@posts[i]) }
+      [1].each { |i| results.should_not include(@posts[i]) }
+    end
+
+    it 'matches multiple types' do
+      results = Sunspot.search(Post, Namespaced::Comment) do
+        keywords 'toast'
+      end.results
+      [@posts[0], @posts[2], @comment].each  { |obj| results.should include(obj) }
+      results.should_not include(@posts[1])
+    end
+
+    it 'matches keywords from only the fields specified' do
+      results = Sunspot.search(Post) do
+        keywords 'moron', :fields => [:title]
+      end.results
+      results.should == [@posts[1]]
+    end
+
+    it 'matches multiple keywords on different fields using subqueries' do
+      search = Sunspot.search(Post) do
+        keywords 'moron', :fields => [:title]
+        keywords 'wind',  :fields => [:body]
+      end
+      search.results.should == []
+
+      search = Sunspot.search(Post) do
+        keywords 'moron',   :fields => [:title]
+        keywords 'buffer',  :fields => [:body]
+      end
+      search.results.should == [@posts[1]]
+    end
+
+    it 'matches multiple keywords with escaped characters' do
+      search = Sunspot.search(Post) do
+        keywords 'spirit',   :fields => [:title]
+        keywords 'host\'s',  :fields => [:body]
+      end
+      search.results.should == [@posts[2]]
+    end
+
+    it 'matches multiple keywords with phrase-based search' do
+      search = Sunspot.search(Post) do
+        keywords 'spirit', :fields => [:title]
+        keywords '"interpret the buffer"', :fields => [:body]
+        keywords '"does the"', :fields => [:body]
+      end
+      search.results.should == [@posts[2]]
+    end
+
+    it 'matches multiple keywords different options' do
+      search = Sunspot.search(Post) do
+        keywords 'insufficient nonexistent', :fields => [:title], :minimum_match => 1
+        keywords 'wind does', :fields => [:body], :minimum_match => 2
+      end
+      search.results.should == [@posts[0]]
+    end
+  end
+
+  describe 'with field boost' do
+    before :all do
+      Sunspot.remove_all
+      @posts = [:title, :body].map { |field| Post.new(field => 'rhinoceros') }
+      Sunspot.index!(*@posts)
+    end
+
+    it 'should assign a higher score to the result matching the higher-boosted field' do
+      search = Sunspot.search(Post) { keywords 'rhinoceros' }
+      search.hits.map { |hit| hit.primary_key }.should ==
+        @posts.map { |post| post.id.to_s }
+      search.hits.first.score.should > search.hits.last.score
+    end
+  end
+
+  describe 'with document boost' do
+    before :all do
+      Sunspot.remove_all
+      @posts = [4.0, 2.0].map do |rating|
+        Post.new(:title => 'Test', :ratings_average => rating)
+      end
+      Sunspot.index!(*@posts)
+    end
+
+    it 'should assign a higher score to the higher-boosted document' do
+      search = Sunspot.search(Post) { keywords 'test' }
+      search.hits.map { |hit| hit.primary_key }.should ==
+        @posts.map { |post| post.id.to_s }
+      search.hits.first.score.should > search.hits.last.score
+    end
+  end
+
+  describe 'with search-time boost' do
+    before :each do
+      Sunspot.remove_all
+      @comments = [
+        Namespaced::Comment.new(:body => 'test text'),
+        Namespaced::Comment.new(:author_name => 'test text')
+      ]
+      Sunspot.index!(@comments)
+    end
+
+    it 'assigns a higher score to documents in which all words appear in the phrase field' do
+      hits = Sunspot.search(Namespaced::Comment) do
+        keywords 'test text' do
+          phrase_fields :body => 2.0
+        end
+      end.hits
+      hits.first.instance.should == @comments.first
+      hits.first.score.should > hits.last.score
+    end
+
+    it 'assigns a higher score to documents in which the search terms appear in a boosted field' do
+      hits = Sunspot.search(Namespaced::Comment) do
+        keywords 'test' do
+          fields :body => 2.0, :author_name => 0.75
+        end
+      end.hits
+      hits.first.instance.should == @comments.first
+      hits.first.score.should > hits.last.score
+    end
+
+    it 'assigns a higher score to documents in which the search terms appear in a higher boosted phrase field' do
+      hits = Sunspot.search(Namespaced::Comment) do
+        keywords 'test text' do
+          phrase_fields :body => 2.0, :author_name => 0.75
+        end
+      end.hits
+      hits.first.instance.should == @comments.first
+      hits.first.score.should > hits.last.score
+    end
+  end
+
+  describe 'boost query' do
+    before :all do
+      Sunspot.remove_all
+      Sunspot.index!(
+        @posts = [
+          Post.new(:title => 'Rhino', :featured => true),
+          Post.new(:title => 'Rhino', :ratings_average => 3.3),
+          Post.new(:title => 'Rhino')
+        ]
+      )
+    end
+
+    it 'should assign a higher score to the document matching the boost query' do
+      search = Sunspot.search(Post) do |query|
+        query.keywords('rhino') do
+          boost(2.0) do
+            with(:featured, true)
+          end
+        end
+        query.without(@posts[1])
+      end
+      search.results.should == [@posts[0], @posts[2]]
+      search.hits[0].score.should > search.hits[1].score
+    end
+
+    it 'should assign scores in order of multiple boost query match' do
+      search = Sunspot.search(Post) do
+        keywords 'rhino' do
+          boost(2.0) { with(:featured, true) }
+          boost(1.5) { with(:average_rating).greater_than(3.0) }
+        end
+      end
+      search.results.should == @posts
+      search.hits[0].score.should > search.hits[1].score
+      search.hits[1].score.should > search.hits[2].score
+    end
+  end
+
+  describe 'minimum match' do
+    before do
+      Sunspot.remove_all
+      @posts = [
+        Post.new(:title => 'Pepperoni Sausage Anchovies'),
+        Post.new(:title => 'Pepperoni Tomatoes Mushrooms')
+      ]
+      Sunspot.index!(@posts)
+      @search = Sunspot.search(Post) do
+        keywords 'pepperoni sausage extra cheese', :minimum_match => 2
+      end
+    end
+
+    it 'should match documents that contain the minimum_match number of search terms' do
+      @search.results.should include(@posts[0])
+    end
+
+    it 'should not match documents that do not contain the minimum_match number of search terms' do
+      @search.results.should_not include(@posts[1])
+    end
+  end
+
+  describe 'query phrase slop' do
+    before do
+      Sunspot.remove_all
+      @posts = [
+        Post.new(:title => 'One four'),
+        Post.new(:title => 'One three four'),
+        Post.new(:title => 'One two three four')
+      ]
+      Sunspot.index!(@posts)
+      @search = Sunspot.search(Post) do
+        keywords '"one four"', :query_phrase_slop => 1
+      end
+    end
+
+    it 'should match exact phrase' do
+      @search.results.should include(@posts[0])
+    end
+
+    it 'should match phrase divided by query phrase slop terms' do
+      @search.results.should include(@posts[1])
+    end
+
+    it 'should not match phrase divided by more than query phrase slop terms' do
+      @search.results.should_not include(@posts[2])
+    end
+  end
+
+  describe 'phrase field slop' do
+    before do
+      Sunspot.remove_all
+      @comments = [
+        Namespaced::Comment.new(:author_name => 'one four'),
+        Namespaced::Comment.new(:body => 'one four'),
+        Namespaced::Comment.new(:author_name => 'one three four'),
+        Namespaced::Comment.new(:body => 'one three four'),
+        Namespaced::Comment.new(:author_name => 'one two three four'),
+        Namespaced::Comment.new(:body => 'one two three four')
+      ]
+      Sunspot.index!(@comments)
+      @search = Sunspot.search(Namespaced::Comment) do
+        keywords 'one four' do
+          phrase_fields :author_name => 3.0
+          phrase_slop 1
+        end
+      end
+      @sorted_hits = @search.hits.sort_by { |hit| @comments.index(hit.instance) }
+    end
+
+    it 'should give phrase field boost to exact match' do
+      @sorted_hits[0].score.should > @sorted_hits[1].score
+    end
+
+    it 'should give phrase field boost to match within slop' do
+      @sorted_hits[2].score.should > @sorted_hits[3].score
+    end
+
+    it 'should not give phrase field boost to match beyond slop' do
+      @sorted_hits[4].score.should == @sorted_hits[5].score
+    end
+  end
+
+  describe 'with function queries' do
+    before :each do
+      Sunspot.remove_all
+    end
+
+    after :each do
+      @search.results.should == @posts
+      @search.hits.first.score.should > @search.hits.last.score
+    end
+
+    it 'boosts via function query with float' do
+      @posts = [Post.new(:title => 'test', :ratings_average => 4.0),
+                Post.new(:title => 'test', :ratings_average => 2.0)]
+      Sunspot.index!(@posts)
+      @search = Sunspot.search(Post) do
+        keywords('test') do
+          boost function { :average_rating }
+        end
+      end
+    end
+
+    it 'boosts via function query with date' do
+      @posts = [Post.new(:title => 'test', :published_at => Time.now),
+                Post.new(:title => 'test', :published_at => Time.now - 60*60*24*31*6)] # roughly six months ago
+      Sunspot.index!(@posts)
+      @search = Sunspot.search(Post) do
+        keywords('test') do
+          boost function { recip(ms(Time.now, :published_at), 3.16e-11, 1, 1) }
+        end
+      end
+    end
+  end
+end