solrizer 0.2.0 → 0.3.0

data/lib/solrizer/repository.rb

@@ -1,54 +0,0 @@
-
-require 'active-fedora'
-
-
-module Solrizer
-class Repository
-
-  #
-  # This method initializes the fedora repository and solr instance
-  #
-  def initialize_repository
-    Fedora::Repository.register( FEDORA_URL )
-    ActiveFedora::SolrService.register( FEDORA_SOLR_URL )
-  end
-
-  #
-  # This method retrieves a comprehensive list of unique ids in the fedora repository
-  #
-  def self.get_pids( num_docs )
-    solr_results = ActiveFedora::SolrService.instance.conn.query( "active_fedora_model_field:Document", { :rows => num_docs } )
-    id_array = []
-    solr_results.hits.each do |hit|
-      id_array << hit[SOLR_DOCUMENT_ID]
-    end
-    return id_array
-  end
-  
-  #
-  # This method retrieves the object associated with the given unique id
-  #
-  def self.get_object( pid )
-    object = ActiveFedora::Base.load_instance( pid )
-  end
-  
-  #
-  # This method retrieves a comprehensive list of datastreams for the given object
-  #
-  def self.get_datastreams( obj )
-    ds_keys = obj.datastreams.keys
-  end
-  
-  #
-  # This method retrieves the datastream for the given object with the given datastream name
-  #
-  def self.get_datastream( obj, ds_name )
-    begin
-      obj.datastreams[ ds_name ]
-    rescue
-      return nil
-    end
-  end
-
-end
-end

data/spec/fixtures/rels_ext_cmodel.xml

@@ -1,8 +0,0 @@
-<rdf:RDF xmlns:rdf="http://www.w3.org/1999/02/22-rdf-syntax-ns#">
-	<rdf:Description rdf:about="info:fedora/demo:multipurpose-objects-model_and_sdef">
-		<hasModel xmlns="info:fedora/fedora-system:def/model#" rdf:resource="info:fedora/fedora-system:ContentModel-3.0"/>
-		<hasModel xmlns="info:fedora/fedora-system:def/model#" rdf:resource="info:fedora/afmodel:SaltDocument"/>
-		<hasModel xmlns="info:fedora/fedora-system:def/model#" rdf:resource="info:fedora/afmodel:JP2Document"/>
-		<hasModel xmlns="info:fedora/fedora-system:def/model#" rdf:resource="info:fedora/afmodel:DCDocument"/>
-	</rdf:Description>
-</rdf:RDF>

data/spec/fixtures/solr_mappings_af_0.1.yml

@@ -1,16 +0,0 @@
-id: id
-date: _date
-string: _field
-text: _field
-symbol: _field
-integer: _field
-long: _field
-boolean: _field
-float: _field
-double: _field
-facet: _facet
-display: _display
-sort: _sort
-unstemmed_search: _unstem_search
-
-

data/spec/integration/indexer_spec.rb

@@ -1,18 +0,0 @@
-require File.expand_path(File.dirname(__FILE__) + '/../spec_helper')
-require 'solrizer'
-
-describe Solrizer::Indexer do
-  
-  before(:each) do
-    @indexer = Solrizer::Indexer.new
-  end
-  
-  describe "index" do
-    it "should update solr with the metadata from the given object" do
-      pending "Got to decide if/how to handle fixtures in this gem. Probably should just mock out Fedora & Solr entirely."
-      obj = Solrizer::Repository.get_object( "druid:sb733gr4073" )
-      @indexer.index( obj )
-    end
-  end
-  
-end

data/spec/units/indexer_spec.rb

@@ -1,127 +0,0 @@
-require File.expand_path(File.dirname(__FILE__) + '/../spec_helper')
-require 'solrizer'
-
-describe Solrizer::Indexer do
-    
-  before(:each) do
-     Solrizer::Indexer.any_instance.stubs(:connect).returns("foo")
-  
-     @extractor = mock("Extractor")
-     @extractor.stubs(:html_content_to_solr).returns(@solr_doc)
-#     @solr_doc = mock('solr_doc')
-#     @solr_doc.stubs(:<<)
-#     @solr_doc.stubs(:[])
-     
-     @solr_doc = Solr::Document.new
-     
-     Solrizer::Extractor.expects(:new).returns(@extractor)
-     @indexer = Solrizer::Indexer.new
-     
-   end
-  
-  describe "#generate_dates" do
-    it "should still give 9999-99-99 date if the solr document does not have a date_t field" do
-    
-    solr_result = @indexer.generate_dates(@solr_doc)
-    solr_result.should be_kind_of Solr::Document
-    solr_result[:date_t].should == "9999-99-99"
-    solr_result[:month_facet].should == "99"
-    solr_result[:day_facet].should == '99'
-    
-    end
-    
-    it "should still give 9999-99-99 date if the solr_doc[:date_t] is not valid date in YYYY-MM-DD format " do
-     
-      @solr_doc << Solr::Field.new(:date_t => "Unknown")
-      solr_result = @indexer.generate_dates(@solr_doc)
-      solr_result.should be_kind_of Solr::Document
-      solr_result[:date_t].should == "Unknown"
-      solr_result[:month_facet].should == "99"
-      solr_result[:day_facet].should == '99'
-  
-    end
-    
-    it "should give month and dates even if the :date_t is not a valid date but is in YYYY-MM-DD format  " do
-      
-       @solr_doc << Solr::Field.new(:date_t => "0000-13-11")
-       solr_result = @indexer.generate_dates(@solr_doc)
-       solr_result.should be_kind_of Solr::Document
-       solr_result[:date_t].should == "0000-13-11"
-       solr_result[:month_facet].should == "99"
-       solr_result[:day_facet].should == '11'
-     end
-     
-     it "should give month and day when in a valid date format" do
-           @solr_doc << Solr::Field.new(:date_t => "1978-04-11")
-            solr_result = @indexer.generate_dates(@solr_doc)
-            solr_result.should be_kind_of Solr::Document
-            solr_result[:date_t].should == "1978-04-11"
-            solr_result[:month_facet].should == "04"
-            solr_result[:day_facet].should == '11'
-     
-     end
-     
-     it "should still give two digit strings even if the month/day is single digit" do
-     
-         @solr_doc << Solr::Field.new(:date_t => "1978-4-1")
-         solr_result = @indexer.generate_dates(@solr_doc)
-         solr_result.should be_kind_of Solr::Document
-         solr_result[:date_t].should == "1978-4-1"
-         solr_result[:month_facet].should == "04"
-         solr_result[:day_facet].should == '01'  
-     
-     end
-     
-  end
-  
-  
-  
-  describe "#solrize" do
-    it "should convert a hash to a solr doc" do
-      example_hash = {"box"=>"Box 51A", "city"=>["Ann Arbor", "Hyderabad", "Palo Alto"], "person"=>["ELLIE ENGELMORE", "Reddy", "EDWARD FEIGENBAUM"], "title"=>"Letter from Ellie Engelmore to Professor K. C. Reddy", "series"=>"eaf7000", "folder"=>"Folder 15", "technology"=>["artificial intelligence"], "year"=>"1985", "organization"=>["Heuristic Programming Project", "Mathematics and Computer/Information Sciences University of Hyderabad Central University P. O. Hyder", "Professor K. C. Reddy School of Mathematics and Computer/Information Sciences"], "collection"=>"e-a-feigenbaum-collection", "state"=>["Michigan", "California"]}
-      
-      example_result = Solrizer::Indexer.solrize( example_hash )
-      example_result.should be_kind_of Solr::Document
-      example_hash.each_pair do |key,values|
-        if values.class == String
-          example_result["#{key}_facet"].should == values
-        else
-          values.each do |v|
-            example_result.inspect.include?("@name=\"#{key}_facet\"").should be_true
-            example_result.inspect.include?("@value=\"#{v}\"").should be_true
-          end
-        end        
-      end
-    end
-    
-    it "should handle hashes with facets listed in a sub-hash" do
-      simple_hash = Hash[:facets => {'technology'=>["t1", "t2"], 'company'=>"c1", "person"=>["p1", "p2"]}]
-      result = Solrizer::Indexer.solrize( simple_hash )
-      result.should be_kind_of Solr::Document
-      result["technology_facet"].should == "t1"
-      result.inspect.include?('@boost=nil').should be_true
-      result.inspect.include?('@name="technology_facet"').should be_true
-      result.inspect.include?('@value="t2"').should be_true
-      result["company_facet"].should == "c1"
-      result["person_facet"].should == "p1"
-      result.inspect.include?('@name="person_facet"').should be_true
-      result.inspect.include?('@value="p2"').should be_true
-      
-    end
-    
-    it "should create symbols from the :symbols subhash" do
-      simple_hash = Hash[:facets => {'technology'=>["t1", "t2"], 'company'=>"c1", "person"=>["p1", "p2"]}, :symbols=>{'technology'=>["t1", "t2"], 'company'=>"c1", "person"=>["p1", "p2"]}]
-      result = Solrizer::Indexer.solrize( simple_hash )
-      result.should be_kind_of Solr::Document
-      result["technology_s"].should == "t1"
-      result.inspect.include?('@name="technology_s"').should be_true
-      result.inspect.include?('@value="t2"').should be_true
-    
-      result["company_s"].should == "c1"
-      result["person_s"].should == "p1"
-      result.inspect.include?('@name="person_s"').should be_true
-      result.inspect.include?('@value="p2"').should be_true
-  
-    end
-  end
-end

data/spec/units/shelver_spec.rb

@@ -1,42 +0,0 @@
-require File.expand_path(File.dirname(__FILE__) + '/../spec_helper')
-
-describe Solrizer::Solrizer do
-  
-  before(:each) do
-    @solrizer = Solrizer::Solrizer.new
-  end
-  
-  describe "solrize" do
-    it "should trigger the indexer for the provided object" do
-      sample_obj = ActiveFedora::Base.new
-      @solrizer.indexer.expects(:index).with( sample_obj )
-      @solrizer.solrize( sample_obj )
-    end
-    it "should work with Fedora::FedoraObject objects" do
-      mock_object = Fedora::FedoraObject.new(:pid=>"my:pid", :label=>"my label")
-      ActiveFedora::Base.expects(:load_instance).with( mock_object.pid ).returns(mock_object)
-      @solrizer.indexer.expects(:index).with( mock_object )
-      @solrizer.solrize( mock_object )
-    end
-    it "should load the object if only a pid is provided" do
-      mock_object = mock("my object")
-      mock_object.stubs(:pid)
-      mock_object.stubs(:label)
-      mock_object.stubs(:datastreams).returns({'descMetadata'=>"foo","location"=>"bar"})
-
-      ActiveFedora::Base.expects(:load_instance).with( "_PID_" ).returns(mock_object)
-      @solrizer.indexer.expects(:index).with(mock_object)
-      @solrizer.solrize("_PID_")
-    end
-
-  end
-  
-  describe "solrize_objects" do
-    it "should call solrize for each object returned by Fedora::Repository.find_objects" do
-      objects = [["pid1"], ["pid2"], ["pid3"]]
-      Fedora::Repository.any_instance.expects(:find_objects).returns(objects)
-      objects.each {|object| @solrizer.expects(:solrize).with( object ) }
-      @solrizer.solrize_objects
-    end
-  end
-end