erichummel-sunspot 1.2.1a

data/spec/api/search/faceting_spec.rb

@@ -0,0 +1,360 @@
+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 facet specified by string' do
+    stub_facet(:title_ss, {})
+    result = session.search Post do
+      facet :title
+    end
+    result.facet('title').field_name.should == :title
+  end
+
+  it 'returns all facets specified by search' do
+    stub_facet(:title_ss, { 'Author 1' => 1 })
+    stub_facet(:blog_id_i, { '1' => 3 })
+    result = session.search(Post) do
+      facet :title
+      facet :blog_id
+    end
+    result.facets.first.field_name.should == :title
+    result.facets.last.field_name.should == :blog_id
+  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_ft, '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_dt,
+      '2009-04-07T20:25:23Z' => 3,
+      '2009-04-07T20:26:19Z' => 2,
+      '2050-04-07T20:27:15Z' => 1
+    )
+    result = session.search Post do
+      facet :published_at
+    end
+    # In JRuby, Time doesn't have 32-bit range constraint, apparently.
+    future_time =
+      begin
+        Time.gm(2050, 4, 7, 20, 27, 15)
+      rescue ArgumentError
+        DateTime.civil(2050, 4, 7, 20, 27, 15)
+      end
+    facet_values(result, :published_at).should ==
+      [Time.gm(2009, 4, 7, 20, 25, 23),
+       Time.gm(2009, 4, 7, 20, 26, 19),
+       future_time]
+  end
+
+  it 'returns 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 'returns trie integer facet' do
+    stub_facet(:size_it, '3' => 2, '1' => 1)
+    result = session.search Photo do
+      facet :size
+    end
+    facet_values(result, :size).should == [3, 1]
+  end
+
+  it 'returns float facet' do
+    stub_facet(:average_rating_ft, '9.3' => 2, '1.1' => 1)
+    result = session.search Photo do
+      facet :average_rating
+    end
+    facet_values(result, :average_rating).should == [9.3, 1.1]
+  end
+
+  it 'returns time facet' do
+    stub_facet(
+      :created_at_dt,
+      '2009-04-07T20:25:23Z' => 3,
+      '2009-04-07T20:26:19Z' => 1
+    )
+    result = session.search Photo do
+      facet :created_at
+    end
+    facet_values(result, :created_at).should ==
+      [Time.gm(2009, 04, 07, 20, 25, 23),
+       Time.gm(2009, 04, 07, 20, 26, 19)]
+  end
+
+  it 'returns boolean facet' do
+    stub_facet(:featured_bs, 'true' => 3, 'false' => 1)
+    result = session.search(Post) { facet(:featured) }
+    facet_values(result, :featured).should == [true, false]
+  end
+
+
+  { 'string' => 'blog', 'symbol' => :blog }.each_pair do |type, name|
+    it "returns field facet with #{type} custom name" do
+      stub_facet(:blog, '2' => 1, '1' => 4)
+      result = session.search(Post) { facet(:blog_id, :name => name) }
+      facet_values(result, :blog).should == [1, 2]
+    end
+
+    it "assigns #{type} custom name to field facet" do
+      stub_facet(:blog, '2' => 1)
+      result = session.search(Post) { facet(:blog_id, :name => name) }
+      result.facet(:blog).name.should == :blog
+    end
+
+    it "retains field name for #{type} custom-named field facet" do
+      stub_facet(:blog, '2' => 1)
+      result = session.search(Post) { facet(:blog_id, :name => name) }
+      result.facet(:blog).field_name.should == :blog_id
+    end
+  end
+
+  it 'returns 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 'returns special :any facet' do
+    stub_query_facet(
+      'category_ids_im:[* TO *]' => 3
+    )
+    search = session.search(Post) { facet(:category_ids, :extra => :any) }
+    row = search.facet(:category_ids).rows.first
+    row.value.should == :any
+    row.count.should == 3
+  end
+
+  it 'returns special :none facet' do
+    stub_query_facet(
+      '-category_ids_im:[* TO *]' => 3
+    )
+    search = session.search(Post) { facet(:category_ids, :extra => :none) }
+    row = search.facet(:category_ids).rows.first
+    row.value.should == :none
+    row.count.should == 3
+  end
+
+  it 'returns date range facet' do
+    stub_date_facet(:published_at_dt, 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 date range facet sorted by count' do
+    stub_date_facet(:published_at_dt, 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, :sort => :count) }
+    facet = result.facet(:published_at)
+    facet.rows.first.value.should == ((start_time+24*60*60)..end_time)
+    facet.rows.last.value.should == (start_time..(start_time+24*60*60))
+  end
+
+  it 'returns query facet' do
+    stub_query_facet(
+      'average_rating_ft:[3\.0 TO 5\.0]' => 3,
+      'average_rating_ft:[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(3) { with(:average_rating, 3.0..4.0) }
+          row(2) { with(:average_rating, 2.0..3.0) }
+          row(4) { with(:average_rating, 4.0..5.0) }
+        end
+      end.facet(:average_rating).rows.map { |row| row.value }
+    end
+
+    before :each do
+      stub_query_facet(
+        'average_rating_ft:[1\.0 TO 2\.0]' => 2,
+        'average_rating_ft:[2\.0 TO 3\.0]' => 3,
+        'average_rating_ft:[3\.0 TO 4\.0]' => 1,
+        'average_rating_ft:[4\.0 TO 5\.0]' => 0
+      )
+    end
+
+    it 'sorts in order of specification if no limit is given' do
+      facet_values_from_options.should == [1, 3, 2]
+    end
+
+    it 'sorts lexically if lexical option is specified' do
+      facet_values_from_options(:sort => :index).should == [1, 2, 3]
+    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, 3]
+    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 'does not limit facets if limit option is negative' do
+      facet_values_from_options(:limit => -2).should == [1, 3, 2]
+    end
+
+    it 'returns all facets if limit greater than number of facets' do
+      facet_values_from_options(:limit => 10).should == [2, 1, 3]
+    end
+
+    it 'allows zero count if specified' do
+      facet_values_from_options(:zeros => true).should == [1, 3, 2, 4]
+    end
+
+    it 'sets minimum count' do
+      facet_values_from_options(:minimum_count => 2).should == [1, 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 all instantiated facet rows, whether or not the instances exist' do
+    blogs = Array.new(2) { Blog.new }
+    blogs.last.destroy
+    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.first, nil]
+  end
+
+  it 'returns only rows with available instances if specified' do
+    blogs = Array.new(2) { Blog.new }
+    blogs.last.destroy
+    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(:verify => true).map { |row| row.instance }.should == blogs[0..0]
+  end
+
+  it 'returns both verified and unverified rows from the same facet' do
+    blogs = Array.new(2) { Blog.new }
+    blogs.last.destroy
+    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(:verify => true).map { |row| row.instance }.should == blogs[0..0]
+    search.facet(:blog_id).rows.map { |row| row.instance }.should == [blogs.first, nil]
+  end
+
+  it 'ignores :verify option if facet not a reference facet' do
+    stub_facet(:category_ids_im, '1' => 2, '2' => 1)
+    search = session.search(Post) { facet(:category_ids) }
+    search.facet(:category_ids).should have(2).rows(:verify => true)
+  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 +1,69 @@
+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 first highlight for a specified field' do
+    @search.hits.first.highlight(:title).format.should == 'one <em>two</em> three'
+  end
+
+  it 'returns an empty array if a given field does not have a highlight' do
+    @search.hits.first.highlights(:body).should == []
+  end
+
+  it 'formats hits with <em> by default' do
+    highlight = @search.hits.first.highlights(:title).first.formatted
+    highlight.should == 'one <em>two</em> three'
+  end
+
+  it 'formats 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 'handles 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 +1,140 @@
+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
+
+  if ENV['USE_WILL_PAGINATE']
+
+    it 'returns search total as attribute of hits' do
+      stub_results(Post.new, 4)
+      session.search(Post) do
+        paginate(:page => 1)
+      end.hits.total_entries.should == 4
+    end
+
+    it 'returns search total as attribute of verified hits' do
+      stub_results(Post.new, 4)
+      session.search(Post) do
+        paginate(:page => 1)
+      end.hits(:verify => true).total_entries.should == 4
+    end
+
+  else
+
+    it 'returns vanilla array of hits if WillPaginate is not available' do
+      stub_results(Post.new)
+      session.search(Post) do
+        paginate(:page => 1)
+      end.hits.should_not respond_to(:total_entries)
+    end
+
+    it 'returns vanilla array of verified hits if WillPaginate is not available' do
+      stub_results(Post.new)
+      session.search(Post) do
+        paginate(:page => 1)
+      end.hits(:verified => true).should_not respond_to(:total_entries)
+    end
+
+  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 provide iterator over hits with instances' do
+    posts = Array.new(2) { Post.new }
+    stub_results(*posts)
+    search = session.search(Post)
+    hits, results = [], []
+    search.each_hit_with_result do |hit, result|
+      hits << hit
+      results << result
+    end
+  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 return only hits whose referenced object exists in the data store if :verify option passed' do
+    posts = Array.new(2) { Post.new }
+    posts.last.destroy
+    stub_results(*posts)
+    search = session.search(Post)
+    search.hits(:verify => true).map { |hit| hit.instance }.should == posts[0..0]
+  end
+
+  it 'should return verified and unverified hits from the same search' do
+    posts = Array.new(2) { Post.new }
+    posts.last.destroy
+    stub_results(*posts)
+    search = session.search(Post)
+    search.hits(:verify => true).map { |hit| hit.instance }.should == posts[0..0]
+    search.hits.map { |hit| hit.instance }.should == [posts.first, nil]
+  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 return stored field values for searches against base type when subtype matches' do
+    class SubclassedPost < Post; end;
+    stub_full_results('instance' => SubclassedPost.new, 'title_ss' => 'Title')
+    session.search(Post).hits.first.stored(:title).should == 'Title'
+  end
+
+  it 'should return stored text fields' do
+    stub_full_results('instance' => Post.new, 'body_textsv' => 'Body')
+    session.search(Post, Namespaced::Comment).hits.first.stored(:body).should == 'Body'
+  end
+
+  it 'should return stored boolean fields' do
+    stub_full_results('instance' => Post.new, 'featured_bs' => true)
+    session.search(Post, Namespaced::Comment).hits.first.stored(:featured).should be_true
+  end
+
+  it 'should return stored dynamic fields' do
+    stub_full_results('instance' => Post.new, 'custom_string:test_ss' => 'Custom')
+    session.search(Post, Namespaced::Comment).hits.first.stored(:custom_string, :test).should == 'Custom'
+  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 stored values for multi-valued fields' do
+    stub_full_results('instance' => User.new, 'role_ids_ims' => %w(1 4 5))
+    session.search(User).hits.first.stored(:role_ids).should == [1, 4, 5]
+  end
+end

data/spec/api/search/results_spec.rb

@@ -0,0 +1,79 @@
+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
+
+  it 'gracefully returns empty results when response is nil' do
+    stub_nil_results
+    session.search(Post).results.should == []
+  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 'returns total' do
+    stub_results(Post.new, Post.new, 4)
+    session.search(Post) { paginate(:page => 1) }.total.should == 4
+  end
+
+  it 'returns total for nil search' do
+    stub_nil_results
+    session.search(Post).total.should == 0
+  end
+
+  it 'returns available results if some results are not available from data store' do
+    posts = [Post.new, Post.new]
+    posts.last.destroy
+    stub_results(*posts)
+    session.search(Post).results.should == posts[0..0]
+  end
+
+  it 'does not attempt to query the data store more than once when results are unavailable' do
+    posts = [Post.new, Post.new]
+    posts.each { |post| post.destroy }
+    stub_results(*posts)
+    search = session.search(Post) do
+      data_accessor_for(Post).should_receive(:load_all).once.and_return([])
+    end
+    search.results.should == []
+  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'))