DrMark-thinking-sphinx 1.1.6 → 1.1.14

data/lib/thinking_sphinx/search.rb

@@ -1,3 +1,5 @@
+require 'thinking_sphinx/search/facets'
+
 module ThinkingSphinx
   # Once you've got those indexes in and built, this is the stuff that
   # matters - how to search! This class provides a generic search
@@ -13,6 +15,8 @@ module ThinkingSphinx
     }
     
     class << self
+      include ThinkingSphinx::Search::Facets
+      
       # Searches for results that match the parameters provided. Will only
       # return the ids for the matching objects. See #search for syntax
       # examples.
@@ -432,21 +436,6 @@ module ThinkingSphinx
         end
       end
       
-      # Model.facets *args
-      # ThinkingSphinx::Search.facets *args
-      # ThinkingSphinx::Search.facets *args, :all_attributes  => true
-      # ThinkingSphinx::Search.facets *args, :class_facet     => false
-      # 
-      def facets(*args)
-        options = args.extract_options!
-        
-        if options[:class]
-          facets_for_model options[:class], args, options
-        else
-          facets_for_all_models args, options
-        end
-      end
-      
       private
       
       # This method handles the common search functionality, and returns both
@@ -717,64 +706,6 @@ module ThinkingSphinx
         
         string
       end
-      
-      def facets_for_model(klass, args, options)
-        hash    = ThinkingSphinx::FacetCollection.new args + [options]
-        options = options.clone.merge! :group_function => :attr
-        
-        klass.sphinx_facets.inject(hash) do |hash, facet|
-          unless facet.name == :class && !options[:class_facet]
-            options[:group_by] = facet.attribute_name
-            hash.add_from_results facet, search(*(args + [options]))
-          end
-          
-          hash
-        end
-      end
-      
-      def facets_for_all_models(args, options)
-        options = GlobalFacetOptions.merge(options)
-        hash    = ThinkingSphinx::FacetCollection.new args + [options]
-        options = options.merge! :group_function => :attr
-        
-        facet_names(options).inject(hash) do |hash, name|
-          options[:group_by] = name
-          hash.add_from_results name, search(*(args + [options]))
-          hash
-        end
-      end
-      
-      def facet_classes(options)
-        options[:classes] || ThinkingSphinx.indexed_models.collect { |model|
-          model.constantize
-        }
-      end
-      
-      def facet_names(options)
-        classes = facet_classes(options)
-        names   = options[:all_attributes] ?
-          facet_names_for_all_classes(classes) :
-          facet_names_common_to_all_classes(classes)
-        
-        names.delete "class_crc" unless options[:class_facet]
-        names
-      end
-      
-      def facet_names_for_all_classes(classes)
-        classes.collect { |klass|
-          klass.sphinx_facets.collect { |facet| facet.attribute_name }
-        }.flatten.uniq
-      end
-      
-      def facet_names_common_to_all_classes(classes)
-        facet_names_for_all_classes(classes).select { |name|
-          classes.all? { |klass|
-            klass.sphinx_facets.detect { |facet|
-              facet.attribute_name == name
-            }
-          }
-        }
-      end
     end
   end
 end

data/lib/thinking_sphinx/source/internal_properties.rb

@@ -0,0 +1,46 @@
+module ThinkingSphinx
+  class Source
+    module InternalProperties
+      def add_internal_attributes_and_facets
+        add_internal_attribute :sphinx_internal_id, :integer, @model.primary_key.to_sym
+        add_internal_attribute :class_crc,          :integer, crc_column, true
+        add_internal_attribute :subclass_crcs,      :multi,   subclasses_to_s
+        add_internal_attribute :sphinx_deleted,     :integer, "0"
+
+        add_internal_facet :class_crc
+      end
+
+      def add_internal_attribute(name, type, contents, facet = false)
+        return unless attribute_by_alias(name).nil?
+
+        Attribute.new(self,
+          ThinkingSphinx::Index::FauxColumn.new(contents),
+          :type   => type,
+          :as     => name,
+          :facet  => facet,
+          :admin  => true
+        )
+      end
+
+      def add_internal_facet(name)
+        return unless facet_by_alias(name).nil?
+
+        @model.sphinx_facets << ClassFacet.new(attribute_by_alias(name))
+      end
+
+      def attribute_by_alias(attr_alias)
+        @attributes.detect { |attrib| attrib.alias == attr_alias }
+      end
+
+      def facet_by_alias(name)
+        @model.sphinx_facets.detect { |facet| facet.name == name }
+      end
+
+      def subclasses_to_s
+        "'" + (@model.send(:subclasses).collect { |klass|
+          klass.to_crc32.to_s
+        } << @model.to_crc32.to_s).join(",") + "'"
+      end
+    end
+  end
+end

data/lib/thinking_sphinx/source/sql.rb

@@ -0,0 +1,124 @@
+module ThinkingSphinx
+  class Source
+    module SQL
+      # Generates the big SQL statement to get the data back for all the fields
+      # and attributes, using all the relevant association joins. If you want
+      # the version filtered for delta values, send through :delta => true in the
+      # options. Won't do much though if the index isn't set up to support a
+      # delta sibling.
+      # 
+      # Examples:
+      # 
+      #   source.to_sql
+      #   source.to_sql(:delta => true)
+      #
+      def to_sql(options={})
+        sql = <<-SQL
+SELECT #{ sql_select_clause options[:offset] }
+FROM #{ @model.quoted_table_name }
+  #{ all_associations.collect { |assoc| assoc.to_sql }.join(' ') }
+WHERE #{ sql_where_clause(options) }
+GROUP BY #{ sql_group_clause }
+        SQL
+
+        sql += " ORDER BY NULL" if adapter.sphinx_identifier == "mysql"
+        sql
+      end
+
+      # Simple helper method for the query range SQL - which is a statement that
+      # returns minimum and maximum id values. These can be filtered by delta -
+      # so pass in :delta => true to get the delta version of the SQL.
+      # 
+      def to_sql_query_range(options={})
+        min_statement = adapter.convert_nulls(
+          "MIN(#{quote_column(@model.primary_key)})", 1
+        )
+        max_statement = adapter.convert_nulls(
+          "MAX(#{quote_column(@model.primary_key)})", 1
+        )
+
+        sql = "SELECT #{min_statement}, #{max_statement} " +
+              "FROM #{@model.quoted_table_name} "
+        if self.delta? && !@index.delta_object.clause(@model, options[:delta]).blank?
+          sql << "WHERE #{@index.delta_object.clause(@model, options[:delta])}"
+        end
+
+        sql
+      end
+
+      # Simple helper method for the query info SQL - which is a statement that
+      # returns the single row for a corresponding id.
+      # 
+      def to_sql_query_info(offset)
+        "SELECT * FROM #{@model.quoted_table_name} WHERE " +
+        "#{quote_column(@model.primary_key)} = (($id - #{offset}) / #{ThinkingSphinx.indexed_models.size})"
+      end
+
+      def sql_select_clause(offset)
+        unique_id_expr = ThinkingSphinx.unique_id_expression(offset)
+
+        (
+          ["#{@model.quoted_table_name}.#{quote_column(@model.primary_key)} #{unique_id_expr} AS #{quote_column(@model.primary_key)} "] + 
+          @fields.collect     { |field|     field.to_select_sql     } +
+          @attributes.collect { |attribute| attribute.to_select_sql }
+        ).compact.join(", ")
+      end
+
+      def sql_where_clause(options)
+        logic = [
+          "#{@model.quoted_table_name}.#{quote_column(@model.primary_key)} >= $start",
+          "#{@model.quoted_table_name}.#{quote_column(@model.primary_key)} <= $end"
+        ]
+
+        if self.delta? && !@index.delta_object.clause(@model, options[:delta]).blank?
+          logic << "#{@index.delta_object.clause(@model, options[:delta])}"
+        end
+
+        logic += (@conditions || [])
+
+        logic.join(" AND ")
+      end
+
+      def sql_group_clause
+        internal_groupings = []
+        if @model.column_names.include?(@model.inheritance_column)
+           internal_groupings << "#{@model.quoted_table_name}.#{quote_column(@model.inheritance_column)}"
+        end
+
+        (
+          ["#{@model.quoted_table_name}.#{quote_column(@model.primary_key)}"] + 
+          @fields.collect     { |field|     field.to_group_sql     }.compact +
+          @attributes.collect { |attribute| attribute.to_group_sql }.compact +
+          @groupings + internal_groupings
+        ).join(", ")
+      end
+
+      def sql_query_pre_for_core
+        if self.delta? && !@index.delta_object.reset_query(@model).blank?
+          [@index.delta_object.reset_query(@model)]
+        else
+          []
+        end
+      end
+
+      def sql_query_pre_for_delta
+        [""]
+      end
+
+      def quote_column(column)
+        @model.connection.quote_column_name(column)
+      end
+
+      def crc_column
+        if @model.column_names.include?(@model.inheritance_column)
+          adapter.cast_to_unsigned(adapter.convert_nulls(
+            adapter.crc(adapter.quote_with_table(@model.inheritance_column), true),
+            @model.to_crc32
+          ))
+        else
+          @model.to_crc32.to_s
+        end
+      end
+    end
+  end
+end

data/lib/thinking_sphinx/source.rb

@@ -0,0 +1,150 @@
+require 'thinking_sphinx/source/internal_properties'
+require 'thinking_sphinx/source/sql'
+
+module ThinkingSphinx
+  class Source
+    include ThinkingSphinx::Source::InternalProperties
+    include ThinkingSphinx::Source::SQL
+    
+    attr_accessor :model, :fields, :attributes, :conditions, :groupings,
+      :options
+    attr_reader :base
+    
+    def initialize(index, options = {})
+      @index        = index
+      @model        = index.model
+      @fields       = []
+      @attributes   = []
+      @conditions   = []
+      @groupings    = []
+      @options      = options
+      @associations = {}
+      
+      @base = ::ActiveRecord::Associations::ClassMethods::JoinDependency.new(
+        @model, [], nil
+      )
+      
+      unless @model.descends_from_active_record?
+        stored_class = @model.store_full_sti_class ? @model.name : @model.name.demodulize
+        @conditions << "#{@model.quoted_table_name}.#{quote_column(@model.inheritance_column)} = '#{stored_class}'"
+      end
+      
+      add_internal_attributes_and_facets
+    end
+    
+    def name
+      @model.name.underscore.tr(':/\\', '_')
+    end
+    
+    def to_riddle_for_core(offset, index)
+      source = Riddle::Configuration::SQLSource.new(
+        "#{name}_core_#{index}", adapter.sphinx_identifier
+      )
+      
+      set_source_database_settings  source
+      set_source_attributes         source, offset
+      set_source_sql                source, offset
+      set_source_settings           source
+      
+      source
+    end
+    
+    def to_riddle_for_delta(offset, index)
+      source = Riddle::Configuration::SQLSource.new(
+        "#{name}_delta_#{index}", adapter.sphinx_identifier
+      )
+      source.parent = "#{name}_core_#{index}"
+      
+      set_source_database_settings  source
+      set_source_attributes         source, offset
+      set_source_sql                source, offset, true
+      
+      source
+    end
+    
+    def delta?
+      !@index.delta_object.nil?
+    end
+    
+    # Gets the association stack for a specific key.
+    # 
+    def association(key)
+      @associations[key] ||= Association.children(@model, key)
+    end
+    
+    private
+    
+    def adapter
+      @adapter ||= @model.sphinx_database_adapter
+    end
+    
+    def set_source_database_settings(source)
+      config = @model.connection.instance_variable_get(:@config)
+      
+      source.sql_host = config[:host]           || "localhost"
+      source.sql_user = config[:username]       || config[:user] || ""
+      source.sql_pass = (config[:password].to_s || "").gsub('#', '\#')
+      source.sql_db   = config[:database]
+      source.sql_port = config[:port]
+      source.sql_sock = config[:socket]
+    end
+    
+    def set_source_attributes(source, offset)
+      attributes.each do |attrib|
+        source.send(attrib.type_to_config) << attrib.config_value(offset)
+      end
+    end
+    
+    def set_source_sql(source, offset, delta = false)
+      source.sql_query        = to_sql(:offset => offset, :delta => delta).gsub(/\n/, ' ')
+      source.sql_query_range  = to_sql_query_range(:delta => delta)
+      source.sql_query_info   = to_sql_query_info(offset)
+      
+      source.sql_query_pre += send(!delta ? :sql_query_pre_for_core : :sql_query_pre_for_delta)
+      
+      if @options[:group_concat_max_len]
+        source.sql_query_pre << "SET SESSION group_concat_max_len = #{@options[:group_concat_max_len]}"
+      end
+      
+      source.sql_query_pre += [adapter.utf8_query_pre].compact if utf8?
+    end
+    
+    def set_source_settings(source)
+      config = ThinkingSphinx::Configuration.instance
+      config.source_options.each do |key, value|
+        source.send("#{key}=".to_sym, value)
+      end
+      
+      source_options = ThinkingSphinx::Configuration::SourceOptions
+      @options.each do |key, value|
+        if source_options.include?(key.to_s) && !value.nil?
+          source.send("#{key}=".to_sym, value)
+        end
+      end
+    end
+    
+    # Returns all associations used amongst all the fields and attributes.
+    # This includes all associations between the model and what the actual
+    # columns are from.
+    # 
+    def all_associations
+      @all_associations ||= (
+        # field associations
+        @fields.collect { |field|
+          field.associations.values
+        }.flatten +
+        # attribute associations
+        @attributes.collect { |attrib|
+          attrib.associations.values if attrib.include_as_association?
+        }.compact.flatten
+      ).uniq.collect { |assoc|
+        # get ancestors as well as column-level associations
+        assoc.ancestors
+      }.flatten.uniq
+    end
+    
+    def utf8?
+      @index.options[:charset_type] == "utf-8"
+    end
+  end
+end

data/lib/thinking_sphinx/tasks.rb

@@ -30,7 +30,7 @@ namespace :thinking_sphinx do
     if sphinx_running?
       puts "Started successfully (pid #{sphinx_pid})."
     else
-      puts "Failed to start searchd daemon. Check #{config.searchd_log_file}."
+      puts "Failed to start searchd daemon. Check #{config.searchd_log_file}"
     end
   end
   

data/lib/thinking_sphinx.rb

@@ -7,6 +7,7 @@ require 'riddle'
 require 'after_commit'
 
 require 'thinking_sphinx/core/string'
+require 'thinking_sphinx/property'
 require 'thinking_sphinx/active_record'
 require 'thinking_sphinx/association'
 require 'thinking_sphinx/attribute'
@@ -17,6 +18,7 @@ require 'thinking_sphinx/class_facet'
 require 'thinking_sphinx/facet_collection'
 require 'thinking_sphinx/field'
 require 'thinking_sphinx/index'
+require 'thinking_sphinx/source'
 require 'thinking_sphinx/rails_additions'
 require 'thinking_sphinx/search'
 require 'thinking_sphinx/deltas'
@@ -35,7 +37,7 @@ module ThinkingSphinx
   module Version #:nodoc:
     Major = 1
     Minor = 1
-    Tiny  = 6
+    Tiny  = 14
     
     String = [Major, Minor, Tiny].join('.')
   end

data/spec/unit/thinking_sphinx/active_record_spec.rb

@@ -14,7 +14,7 @@ describe "ThinkingSphinx::ActiveRecord" do
       )
       
       @index = ThinkingSphinx::Index.stub_instance(:delta? => false)
-      ThinkingSphinx::Index.stub_method(:new => @index)
+      ThinkingSphinx::Index::Builder.stub_method(:generate => @index)
     end
     
     after :each do
@@ -24,17 +24,17 @@ describe "ThinkingSphinx::ActiveRecord" do
       ThinkingSphinx.indexed_models.delete "TestModule::TestModel"
     end
     
-    it "should return nil and do nothing if indexes are disabled" do
+    it "should do nothing if indexes are disabled" do
       ThinkingSphinx.stub_method(:define_indexes? => false)
       
-      TestModule::TestModel.define_index {}.should be_nil
+      TestModule::TestModel.define_index {}
       ThinkingSphinx::Index.should_not have_received(:new)
       
       ThinkingSphinx.unstub_method(:define_indexes?)
     end
     
     it "should add a new index to the model" do
-      TestModule::TestModel.define_index do; end
+      TestModule::TestModel.define_index {}
       
       TestModule::TestModel.sphinx_indexes.length.should == 1
     end
@@ -64,15 +64,15 @@ describe "ThinkingSphinx::ActiveRecord" do
       
       TestModule::TestModel.define_index do; end
       
-      TestModule::TestModel.should have_received(:before_save)
-      TestModule::TestModel.should have_received(:after_commit)
+      TestModule::TestModel.should have_received(:before_save).with(:toggle_delta)
+      TestModule::TestModel.should have_received(:after_commit).with(:index_delta)
     end
     
     it "should not add before_save and after_commit hooks to the model if delta indexing is disabled" do
       TestModule::TestModel.define_index do; end
       
-      TestModule::TestModel.should_not have_received(:before_save)
-      TestModule::TestModel.should_not have_received(:after_commit)
+      TestModule::TestModel.should_not have_received(:before_save).with(:toggle_delta)
+      TestModule::TestModel.should_not have_received(:after_commit).with(:index_delta)
     end
     
     it "should add an after_destroy hook with delta indexing enabled" do
@@ -144,7 +144,7 @@ describe "ThinkingSphinx::ActiveRecord" do
     end
 
     it "should return the parent if model inherits an index" do
-      Parent.source_of_sphinx_index.should == Person
+      Admin::Person.source_of_sphinx_index.should == Person
     end
   end
   
@@ -289,10 +289,12 @@ describe "ThinkingSphinx::ActiveRecord" do
     end
 
     it "should allow associations to other STI models" do
-      Child.sphinx_indexes.last.link!
-      sql = Child.sphinx_indexes.last.to_riddle_for_core(0, 0).sql_query
+      source = Child.sphinx_indexes.last.sources.first
+      sql = source.to_riddle_for_core(0, 0).sql_query
       sql.gsub!('$start', '0').gsub!('$end', '100')
-      lambda { Child.connection.execute(sql) }.should_not raise_error(ActiveRecord::StatementInvalid)
+      lambda {
+        Child.connection.execute(sql)
+      }.should_not raise_error(ActiveRecord::StatementInvalid)
     end
   end
   

data/spec/unit/thinking_sphinx/attribute_spec.rb

@@ -1,23 +1,28 @@
 require 'spec/spec_helper'
 
 describe ThinkingSphinx::Attribute do
+  before :each do
+    @index  = ThinkingSphinx::Index.new(Person)
+    @source = ThinkingSphinx::Source.new(@index)
+  end
+  
   describe '#initialize' do
     it 'raises if no columns are provided so that configuration errors are easier to track down' do
       lambda {
-        ThinkingSphinx::Attribute.new([])
+        ThinkingSphinx::Attribute.new(@source, [])
       }.should raise_error(RuntimeError)
     end
 
     it 'raises if an element of the columns param is an integer - as happens when you use id instead of :id - so that configuration errors are easier to track down' do
       lambda {
-        ThinkingSphinx::Attribute.new([1234])
+        ThinkingSphinx::Attribute.new(@source, [1234])
       }.should raise_error(RuntimeError)
     end
   end
   
   describe "unique_name method" do
     before :each do
-      @attribute = ThinkingSphinx::Attribute.new [
+      @attribute = ThinkingSphinx::Attribute.new @source, [
         Object.stub_instance(:__stack => [], :__name => "col_name")
       ]
     end
@@ -42,7 +47,7 @@ describe ThinkingSphinx::Attribute do
   
   describe "column_with_prefix method" do
     before :each do
-      @attribute = ThinkingSphinx::Attribute.new [
+      @attribute = ThinkingSphinx::Attribute.new @source, [
         ThinkingSphinx::Index::FauxColumn.new(:col_name)
       ]
       @attribute.columns.each { |col| @attribute.associations[col] = [] }
@@ -88,7 +93,7 @@ describe ThinkingSphinx::Attribute do
       @assoc_c = Object.stub_instance(:is_many? => true)
       
       @attribute = ThinkingSphinx::Attribute.new(
-        [ThinkingSphinx::Index::FauxColumn.new(:col_name)]
+        @source, [ThinkingSphinx::Index::FauxColumn.new(:col_name)]
       )
       @attribute.associations = {
         :a => @assoc_a, :b => @assoc_b, :c => @assoc_c
@@ -122,7 +127,7 @@ describe ThinkingSphinx::Attribute do
       @col_c = ThinkingSphinx::Index::FauxColumn.new("c")
 
       @attribute = ThinkingSphinx::Attribute.new(
-        [@col_a, @col_b, @col_c]
+        @source, [@col_a, @col_b, @col_c]
       )
     end
     
@@ -146,7 +151,7 @@ describe ThinkingSphinx::Attribute do
   describe "type method" do
     before :each do
       @column = ThinkingSphinx::Index::FauxColumn.new(:col_name)
-      @attribute = ThinkingSphinx::Attribute.new([@column])
+      @attribute = ThinkingSphinx::Attribute.new(@source, [@column])
       @attribute.model = Person
       @attribute.stub_method(:is_many? => false)
     end
@@ -177,7 +182,7 @@ describe ThinkingSphinx::Attribute do
   
   describe "all_ints? method" do
     it "should return true if all columns are integers" do
-      attribute = ThinkingSphinx::Attribute.new(
+      attribute = ThinkingSphinx::Attribute.new(@source,
         [ ThinkingSphinx::Index::FauxColumn.new(:id),
           ThinkingSphinx::Index::FauxColumn.new(:team_id) ]
       )
@@ -188,7 +193,7 @@ describe ThinkingSphinx::Attribute do
     end
     
     it "should return false if only some columns are integers" do
-      attribute = ThinkingSphinx::Attribute.new(
+      attribute = ThinkingSphinx::Attribute.new(@source,
         [ ThinkingSphinx::Index::FauxColumn.new(:id),
           ThinkingSphinx::Index::FauxColumn.new(:first_name) ]
       )
@@ -199,7 +204,7 @@ describe ThinkingSphinx::Attribute do
     end
     
     it "should return false if no columns are integers" do
-      attribute = ThinkingSphinx::Attribute.new(
+      attribute = ThinkingSphinx::Attribute.new(@source,
         [ ThinkingSphinx::Index::FauxColumn.new(:first_name),
           ThinkingSphinx::Index::FauxColumn.new(:last_name) ]
       )
@@ -213,7 +218,7 @@ describe ThinkingSphinx::Attribute do
   describe "with custom queries" do
     before :each do
       index = CricketTeam.sphinx_indexes.first
-      @statement = index.to_riddle_for_core(0, 0).sql_attr_multi.first
+      @statement = index.sources.first.to_riddle_for_core(0, 0).sql_attr_multi.last
     end
     
     it "should track the query type accordingly" do

data/spec/unit/thinking_sphinx/facet_collection_spec.rb

@@ -0,0 +1,64 @@
+require 'spec/spec_helper'
+
+describe ThinkingSphinx::FacetCollection do
+  before do
+    @facet_collection = ThinkingSphinx::FacetCollection.new([])
+  end
+
+  # TODO fix nasty hack when we have internet!
+  def mock_results
+    return @results if defined? @results
+    @result = Person.find(:first)
+    @results = [@result]
+    @results.stub!(:each_with_groupby_and_count).and_yield(@result, @result.city.to_crc32, 1)
+    @results
+  end
+
+  describe "#add_from_results" do
+    describe "with empty result set" do
+      before do
+        @facet_collection.add_from_results('attribute_facet', [])
+      end
+
+      it "should add key as attribute" do
+        @facet_collection.should have_key(:attribute)
+      end
+
+      it "should return an empty hash for the facet results" do
+        @facet_collection[:attribute].should be_empty
+      end
+    end
+
+    describe "with non-empty result set" do
+      before do
+        @facet_collection.add_from_results('city_facet', mock_results)
+      end
+
+      it "should return a hash" do
+        @facet_collection.should be_a_kind_of(Hash)
+      end
+
+      it "should add key as attribute" do
+        @facet_collection.keys.should include(:city)
+      end
+
+      it "should return a hash" do
+        @facet_collection[:city].should == {@result.city => 1}
+      end
+    end
+  end
+
+  describe "#for" do
+    before do
+      @facet_collection.add_from_results('city_facet', mock_results)
+    end
+
+    it "should return the search results for the attribute and key pair" do
+      ThinkingSphinx::Search.should_receive(:search) do |options|
+        options[:with].should have_key('city_facet')
+        options[:with]['city_facet'].should == @result.city.to_crc32
+      end
+      @facet_collection.for(:city => @result.city)
+    end
+  end
+end