sunspot 0.9.8 → 0.10.0

data/spec/api/search/faceting_spec.rb

@@ -0,0 +1,206 @@
+require File.join(File.dirname(__FILE__), 'spec_helper')
+
+describe 'faceting', :type => :search do
+  it 'returns field name for facet' do
+    stub_facet(:title_ss, {})
+    result = session.search Post do
+      facet :title
+    end
+    result.facet(:title).field_name.should == :title
+  end
+
+  it 'returns string facet' do
+    stub_facet(:title_ss, 'Author 1' => 2, 'Author 2' => 1)
+    result = session.search Post do
+      facet :title
+    end
+    facet_values(result, :title).should == ['Author 1', 'Author 2']
+  end
+
+  it 'returns counts for facet' do
+    stub_facet(:title_ss, 'Author 1' => 2, 'Author 2' => 1)
+    result = session.search Post do
+      facet :title
+    end
+    facet_counts(result, :title).should == [2, 1]
+  end
+
+  it 'returns integer facet' do
+    stub_facet(:blog_id_i, '3' => 2, '1' => 1)
+    result = session.search Post do
+      facet :blog_id
+    end
+    facet_values(result, :blog_id).should == [3, 1]
+  end
+
+  it 'returns float facet' do
+    stub_facet(:average_rating_f, '9.3' => 2, '1.1' => 1)
+    result = session.search Post do
+      facet :average_rating
+    end
+    facet_values(result, :average_rating).should == [9.3, 1.1]
+  end
+
+  it 'returns time facet' do
+    stub_facet(
+      :published_at_d,
+      '2009-04-07T20:25:23Z' => 3,
+      '2009-04-07T20:26:19Z' => 1
+    )
+    result = session.search Post do
+      facet :published_at
+    end
+    facet_values(result, :published_at).should ==
+      [Time.gm(2009, 04, 07, 20, 25, 23),
+       Time.gm(2009, 04, 07, 20, 26, 19)]
+  end
+
+  it 'should return date facet' do
+    stub_facet(
+      :expire_date_d,
+      '2009-07-13T00:00:00Z' => 3,
+      '2009-04-01T00:00:00Z' => 1
+    )
+    result = session.search(Post) do
+      facet :expire_date
+    end
+    facet_values(result, :expire_date).should ==
+      [Date.new(2009, 07, 13),
+       Date.new(2009, 04, 01)]
+  end
+
+  it 'should return boolean facet' do
+    stub_facet(:featured_b, 'true' => 3, 'false' => 1)
+    result = session.search(Post) { facet(:featured) }
+    facet_values(result, :featured).should == [true, false]
+  end
+
+  it 'should return class facet' do
+    stub_facet(:class_name, 'Post' => 3, 'Namespaced::Comment' => 1)
+    result = session.search(Post) { facet(:class) }
+    facet_values(result, :class).should == [Post, Namespaced::Comment]
+  end
+
+  it 'should return date range facet' do
+    stub_date_facet(:published_at_d, 60*60*24, '2009-07-08T04:00:00Z' => 2, '2009-07-07T04:00:00Z' => 1)
+    start_time = Time.utc(2009, 7, 7, 4)
+    end_time = start_time + 2*24*60*60
+    result = session.search(Post) { facet(:published_at, :time_range => start_time..end_time) }
+    facet = result.facet(:published_at)
+    facet.rows.first.value.should == (start_time..(start_time+24*60*60))
+    facet.rows.last.value.should == ((start_time+24*60*60)..end_time)
+  end
+
+  it 'returns query facet' do
+    stub_query_facet(
+      'average_rating_f:[3\.0 TO 5\.0]' => 3,
+      'average_rating_f:[1\.0 TO 3\.0]' => 1
+    )
+    search = session.search(Post) do
+      facet :average_rating do
+        row 3.0..5.0 do
+          with :average_rating, 3.0..5.0
+        end
+        row 1.0..3.0 do
+          with :average_rating, 1.0..3.0
+        end
+      end
+    end
+    facet = search.facet(:average_rating)
+    facet.rows.first.value.should == (3.0..5.0)
+    facet.rows.first.count.should == 3
+    facet.rows.last.value.should == (1.0..3.0)
+    facet.rows.last.count.should == 1
+  end
+
+  describe 'query facet option handling' do
+    def facet_values_from_options(options = {})
+      session.search(Post) do
+        facet :average_rating, options do
+          row(1) { with(:average_rating, 1.0..2.0) }
+          row(2) { with(:average_rating, 2.0..3.0) }
+          row(3) { with(:average_rating, 3.0..4.0) }
+        end
+      end.facet(:average_rating).rows.map { |row| row.value }
+    end
+
+    before :each do
+      stub_query_facet(
+        'average_rating_f:[1\.0 TO 2\.0]' => 1,
+        'average_rating_f:[2\.0 TO 3\.0]' => 2,
+        'average_rating_f:[3\.0 TO 4\.0]' => 0
+      )
+    end
+
+    it 'sorts lexically by default if no limit is given' do
+      facet_values_from_options.should == [1, 2]
+    end
+
+    it 'sorts by count by default if limit is given' do
+      facet_values_from_options(:limit => 2).should == [2, 1]
+    end
+
+    it 'sorts by count if count option is specified' do
+      facet_values_from_options(:sort => :count).should == [2, 1]
+    end
+
+    it 'sorts lexically if lexical option is specified even if limit is given' do
+      facet_values_from_options(:sort => :index, :limit => 2).should == [1, 2]
+    end
+
+    it 'limits facets if limit option is given' do
+      facet_values_from_options(:limit => 1).should == [2]
+    end
+
+    it 'allows zero count if specified' do
+      facet_values_from_options(:zeros => true).should == [1, 2, 3]
+    end
+
+    it 'sets minimum count' do
+      facet_values_from_options(:minimum_count => 2).should == [2]
+    end
+  end
+
+  it 'returns limited field facet' do
+    stub_query_facet(
+      'category_ids_im:1' => 3,
+      'category_ids_im:3' => 1
+    )
+    search = session.search(Post) do
+      facet :category_ids, :only => [1, 3, 5]
+    end
+    facet = search.facet(:category_ids)
+    facet.rows.first.value.should == 1
+    facet.rows.first.count.should == 3
+    facet.rows.last.value.should == 3
+    facet.rows.last.count.should == 1
+  end
+
+  it 'returns instantiated facet values' do
+    blogs = Array.new(2) { Blog.new }
+    stub_facet(:blog_id_i, blogs[0].id.to_s => 2, blogs[1].id.to_s => 1)
+    search = session.search(Post) { facet(:blog_id) }
+    search.facet(:blog_id).rows.map { |row| row.instance }.should == blogs
+  end
+
+  it 'returns instantiated facet values for limited field facet' do
+    blogs = Array.new(2) { Blog.new }
+    stub_query_facet(
+      "blog_id_i:#{blogs[0].id}" => 3,
+      "blog_id_i:#{blogs[1].id}" => 1
+    )
+    search = session.search(Post) do
+      facet(:blog_id, :only => blogs.map { |blog| blog.id })
+    end
+    search.facet(:blog_id).rows.map { |row| row.instance }.should == blogs
+  end
+
+  it 'only queries the persistent store once for an instantiated facet' do
+    query_count = Blog.query_count
+    blogs = Array.new(2) { Blog.new }
+    stub_facet(:blog_id_i, blogs[0].id.to_s => 2, blogs[1].id.to_s => 1)
+    result = session.search(Post) { facet(:blog_id) }
+    result.facet(:blog_id).rows.each { |row| row.instance }
+    (Blog.query_count - query_count).should == 1
+  end
+end

data/spec/api/search/highlighting_spec.rb

@@ -0,0 +1,65 @@
+require File.join(File.dirname(__FILE__), 'spec_helper')
+
+describe 'search with highlighting results', :type => :search do
+  before :each do
+    @posts = Array.new(2) { Post.new }
+    stub_results_with_highlighting(
+      @posts[0],
+      { 'title_text' => ['one @@@hl@@@two@@@endhl@@@ three'] },
+      @posts[1],
+      { 'title_text' => ['three four @@@hl@@@five@@@endhl@@@'],
+        'body_text' => ['@@@hl@@@five@@@ six seven', '@@@hl@@@eight@@@endhl@@@ nine @@@hl@@@ten@@@endhl@@@'] }
+    )
+    @search = session.search(Post)
+  end
+
+  it 'returns all highlights' do
+    @search.hits.last.should have(3).highlights
+  end
+
+  it 'returns all highlights for a specified field' do
+    @search.hits.last.should have(2).highlights(:body)
+  end
+
+  it 'returns nil if a given field does not have a highlight' do
+    @search.hits.first.highlights(:body).should be_nil
+  end
+
+  it 'should format hits with <em> by default' do
+    highlight = @search.hits.first.highlights(:title).first.formatted
+    highlight.should == 'one <em>two</em> three'
+  end
+
+  it 'should format hits with provided block' do
+    highlight = @search.hits.first.highlights(:title).first.format do |word|
+      "<i>#{word}</i>"
+    end
+    highlight.should == 'one <i>two</i> three'
+  end
+
+  it 'should handle multiple highlighted words' do
+    highlight = @search.hits.last.highlights(:body).last.format do |word|
+      "<b>#{word}</b>"
+    end
+    highlight.should == '<b>eight</b> nine <b>ten</b>'
+  end
+
+  private
+
+  def stub_results_with_highlighting(*instances_and_highlights)
+    docs, highlights = [], []
+    instances_and_highlights.each_slice(2) do |doc, highlight|
+      docs << doc
+      highlights << highlight
+    end
+    response = stub_full_results(*docs.map { |doc| { 'instance' => doc }})
+    highlighting = response['highlighting'] = {}
+    highlights.each_with_index do |highlight, i|
+      if highlight
+        instance = docs[i]
+        highlighting["#{instance.class.name} #{instance.id}"] = highlight
+      end
+    end
+    response
+  end
+end

data/spec/api/search/hits_spec.rb

@@ -0,0 +1,62 @@
+require File.join(File.dirname(__FILE__), 'spec_helper')
+
+describe 'hits', :type => :search do
+  it 'should return hits without loading instances' do
+    post_1, post_2 = Array.new(2) { Post.new }
+    stub_results(post_1, post_2)
+    %w(load load_all).each do |message|
+      MockAdapter::DataAccessor.should_not_receive(message)
+    end
+    session.search(Post).hits.map do |hit|
+      [hit.class_name, hit.primary_key]
+    end.should == [['Post', post_1.id.to_s], ['Post', post_2.id.to_s]]
+  end
+
+  it 'should return instance from hit' do
+    posts = Array.new(2) { Post.new }
+    stub_results(*posts)
+    session.search(Post).hits.first.instance.should == posts.first
+  end
+
+  it 'should hydrate all hits when an instance is requested from a hit' do
+    posts = Array.new(2) { Post.new }
+    stub_results(*posts)
+    search = session.search(Post)
+    search.hits.first.instance
+    %w(load load_all).each do |message|
+      MockAdapter::DataAccessor.should_not_receive(message)
+    end
+    search.hits.last.instance.should == posts.last
+  end
+
+  it 'should attach score to hits' do
+    stub_full_results('instance' => Post.new, 'score' => 1.23)
+    session.search(Post).hits.first.score.should == 1.23
+  end
+
+  it 'should return stored field values in hits' do
+    stub_full_results('instance' => Post.new, 'title_ss' => 'Title')
+    session.search(Post).hits.first.stored(:title).should == 'Title'
+  end
+
+  it 'should return stored field values for searches against multiple types' do
+    stub_full_results('instance' => Post.new, 'title_ss' => 'Title')
+    session.search(Post, Namespaced::Comment).hits.first.stored(:title).should == 'Title'
+  end
+
+  it 'should typecast stored field values in hits' do
+    time = Time.utc(2008, 7, 8, 2, 45)
+    stub_full_results('instance' => Post.new, 'last_indexed_at_ds' => time.xmlschema)
+    session.search(Post).hits.first.stored(:last_indexed_at).should == time
+  end
+
+  it 'should return geo distance' do
+    stub_full_results('instance' => Post.new, 'geo_distance' => '1.23')
+    session.search(Post).hits.first.distance.should == 1.23
+  end
+
+  it 'should return nil if no geo distance' do
+    stub_results(Post.new)
+    session.search(Post).hits.first.distance.should be_nil
+  end
+end

data/spec/api/search/results_spec.rb

@@ -0,0 +1,52 @@
+require File.join(File.dirname(__FILE__), 'spec_helper')
+
+describe 'search results', :type => :search do
+  it 'loads single result' do
+    post = Post.new
+    stub_results(post)
+    session.search(Post).results.should == [post]
+  end
+
+  it 'loads multiple results in order' do
+    post_1, post_2 = Post.new, Post.new
+    stub_results(post_1, post_2)
+    session.search(Post).results.should == [post_1, post_2]
+    stub_results(post_2, post_1)
+    session.search(Post).results.should == [post_2, post_1]
+  end
+
+  # This is a reduction of a crazy bug I found in production where some hits
+  # were inexplicably not being populated.
+  it 'properly loads results of multiple classes that have the same primary key' do
+    Post.reset!
+    Namespaced::Comment.reset!
+    results = [Post.new, Namespaced::Comment.new]
+    stub_results(*results)
+    session.search(Post, Namespaced::Comment).results.should == results
+  end
+
+  if ENV['USE_WILL_PAGINATE']
+
+    it 'returns search total as attribute of results' do
+      stub_results(Post.new, 4)
+      session.search(Post) do
+        paginate(:page => 1)
+      end.results.total_entries.should == 4
+    end
+
+  else
+
+    it 'returns vanilla array if WillPaginate is not available' do
+      stub_results(Post.new)
+      session.search(Post) do
+        paginate(:page => 1)
+      end.results.should_not respond_to(:total_entries)
+    end
+
+  end
+
+  it 'should return total' do
+    stub_results(Post.new, Post.new, 4)
+    session.search(Post) { paginate(:page => 1) }.total.should == 4
+  end
+end

data/spec/api/search/search_spec.rb

@@ -0,0 +1,23 @@
+require File.join(File.dirname(__FILE__), 'spec_helper')
+
+describe Sunspot::Search do
+  it 'should allow access to the data accessor' do
+    stub_results(posts = Post.new)
+    search = session.search Post do
+      data_accessor_for(Post).custom_title = 'custom title'
+    end
+    search.results.first.title.should == 'custom title'
+  end
+  
+  it 'should re-execute search' do
+    post_1, post_2 = Post.new, Post.new
+    
+    stub_results(post_1)
+    search = session.search Post
+    search.results.should == [post_1]
+    
+    stub_results(post_2)
+    search.execute!
+    search.results.should == [post_2]
+  end
+end

data/spec/api/search/spec_helper.rb

@@ -0,0 +1 @@
+require File.expand_path(File.join(File.dirname(__FILE__), '..', 'spec_helper'))

data/spec/api/spec_helper.rb

@@ -1 +1 @@
-require File.join(File.dirname(__FILE__), '..', 'spec_helper')
+require File.expand_path(File.join(File.dirname(__FILE__), '..', 'spec_helper'))

data/spec/helpers/indexer_helper.rb

@@ -0,0 +1,29 @@
+module IndexerHelper
+  def config
+    Sunspot::Configuration.build
+  end
+
+  def connection
+    @connection ||= Mock::Connection.new
+  end
+
+  def session
+    @session ||= Sunspot::Session.new(config, connection)
+  end
+
+  def post(attrs = {})
+    @post ||= Post.new(attrs)
+  end
+
+  def last_add
+    @connection.adds.last
+  end
+
+  def value_in_last_document_for(field_name)
+    @connection.adds.last.last.field_by_name(field_name).value
+  end
+
+  def values_in_last_document_for(field_name)
+    @connection.adds.last.last.fields_by_name(field_name).map { |field| field.value }
+  end
+end

data/spec/helpers/query_helper.rb

@@ -0,0 +1,13 @@
+module QueryHelper
+  def config
+    @config ||= Sunspot::Configuration.build
+  end
+
+  def connection
+    @connection ||= Mock::Connection.new
+  end
+
+  def session
+    @session ||= Sunspot::Session.new(config, connection)
+  end
+end

data/spec/helpers/search_helper.rb

@@ -0,0 +1,78 @@
+module SearchHelper
+  def stub_full_results(*results)
+    count =
+      if results.last.is_a?(Integer) then results.pop
+      else results.length
+      end
+    docs = results.map do |result|
+      instance = result.delete('instance')
+      result.merge('id' => "#{instance.class.name} #{instance.id}")
+    end
+    response = {
+      'response' => {
+        'docs' => docs,
+        'numFound' => count
+      }
+    }
+    connection.stub!(:select).and_return(response)
+    response
+  end
+
+  def stub_results(*results)
+    stub_full_results(
+      *results.map do |result|
+        if result.is_a?(Integer)
+          result
+        else
+          { 'instance' => result }
+        end
+      end
+    )
+  end
+
+  def stub_facet(name, values)
+    connection.stub!(:select).and_return(
+      'facet_counts' => {
+        'facet_fields' => {
+          name.to_s => values.to_a.sort_by { |value, count| -count }.flatten
+        }
+      }
+    )
+  end
+
+  def stub_date_facet(name, gap, values)
+    connection.stub!(:select).and_return(
+      'facet_counts' => {
+        'facet_dates' => {
+          name.to_s => { 'gap' => "+#{gap}SECONDS" }.merge(values)
+        }
+      }
+    )
+  end
+
+  def stub_query_facet(values)
+    connection.stub!(:select).and_return(
+      'facet_counts' => { 'facet_queries' => values }
+    )
+  end
+
+  def facet_values(result, field_name)
+    result.facet(field_name).rows.map { |row| row.value }
+  end
+
+  def facet_counts(result, field_name)
+    result.facet(field_name).rows.map { |row| row.count }
+  end
+
+  def config
+    @config ||= Sunspot::Configuration.build
+  end
+
+  def connection
+    @connection ||= Mock::Connection.new
+  end
+
+  def session
+    @session ||= Sunspot::Session.new(config, connection)
+  end
+end

data/spec/integration/faceting_spec.rb

@@ -148,7 +148,7 @@ describe 'search faceting' do
 
     it 'should return specified facets' do
       search = Sunspot.search(Post) do
-        facet :rating_range 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

data/spec/integration/highlighting_spec.rb

@@ -0,0 +1,22 @@
+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/keyword_search_spec.rb

@@ -80,4 +80,69 @@ describe 'keyword search' do
       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', :featured => false)
+        ]
+      )
+    end
+
+    it 'should assign a higher boost to the document matching the boost query' do
+      search = Sunspot.search(Post) do
+        keywords('rhino') do
+          boost(2.0) do
+            with(:featured, true)
+          end
+        end
+      end
+      search.results.should == @posts
+      search.hits[0].score.should > search.hits[1].score
+    end
+  end
 end