erichummel-sunspot 1.2.1a

data/lib/sunspot/installer/solrconfig_updater.rb

@@ -0,0 +1,76 @@
+require 'rubygems'
+require 'fileutils'
+require 'nokogiri'
+
+module Sunspot
+  class Installer
+    class SolrconfigUpdater
+      include TaskHelper
+
+      CONFIG_FILES = %w(solrconfig.xml elevate.xml spellings.txt stopwords.txt synonyms.txt) 
+
+      class <<self
+        def execute(solrconfig_path, options = {})
+          new(solrconfig_path, options).execute
+        end
+      end
+
+      def initialize(solrconfig_path, options)
+        @force = !!options[:force]
+        unless @force || File.exist?(solrconfig_path)
+          abort("#{File.expand_path(@solrconfig_path)} doesn't exist." +
+                " Are you sure you're pointing to a SOLR_HOME?")
+        end
+        @solrconfig_path = solrconfig_path
+        @verbose = !!options[:verbose]
+      end
+
+      def execute
+        if @force
+          config_dir = File.dirname(@solrconfig_path)
+          FileUtils.mkdir_p(config_dir)
+          CONFIG_FILES.each do |file|
+            source_path =
+              File.join(File.dirname(__FILE__), '..', '..', '..', 'solr', 'solr', 'conf', file)
+            FileUtils.cp(source_path, config_dir)
+            say("Copied default #{file} to #{config_dir}")
+          end
+        else
+          @document = File.open(@solrconfig_path) do |f|
+            Nokogiri::XML(
+              f, nil, nil, 
+              Nokogiri::XML::ParseOptions::DEFAULT_XML |
+              Nokogiri::XML::ParseOptions::NOBLANKS
+            )
+          end
+          @root = @document.root
+          maybe_add_more_like_this_handler
+          original_path = "#{@solrconfig_path}.orig"
+          FileUtils.cp(@solrconfig_path, original_path)
+          say("Saved backup of original to #{original_path}")
+          File.open(@solrconfig_path, 'w') do |file|
+            @document.write_to(
+              file,
+              :indent => 2
+            )
+          end
+          say("Wrote solrconfig to #{@solrconfig_path}")
+        end
+      end
+
+      private
+
+      def maybe_add_more_like_this_handler
+        unless @root.xpath('requestHandler[@name="/mlt"]').first
+          mlt_node = add_element(
+            @root, 'requestHandler',
+            :name => '/mlt', :class => 'solr.MoreLikeThisHandler'
+          )
+          defaults_node = add_element(mlt_node, 'lst', :name => 'defaults')
+          add_element(defaults_node, 'str', :name => 'mlt.mintf').content = '1'
+          add_element(defaults_node, 'str', :name => 'mlt.mindf').content = '2'
+        end
+      end
+    end
+  end
+end

data/lib/sunspot/installer/task_helper.rb

@@ -0,0 +1,18 @@
+module Sunspot
+  class Installer
+    module TaskHelper
+      def say(message)
+        if @verbose
+          STDOUT.puts(message)
+        end
+      end
+
+      def add_element(node, name, attributes = {})
+        new_node = Nokogiri::XML::Node.new(name, @document)
+        attributes.each_pair { |name, value| new_node[name.to_s] = value }
+        node << new_node 
+        new_node
+      end
+    end
+  end
+end

data/lib/sunspot/installer.rb

@@ -0,0 +1,31 @@
+%w(task_helper library_installer schema_builder solrconfig_updater).each do |file|
+  require File.join(File.dirname(__FILE__), 'installer', file)
+end
+
+module Sunspot
+  class Installer
+    class <<self
+      def execute(solr_home, options = {})
+        new(solr_home, options).execute
+      end
+
+      private :new
+    end
+
+    def initialize(solr_home, options)
+      @solr_home, @options = solr_home, options
+    end
+
+    def execute
+      SchemaBuilder.execute(
+        File.join(@solr_home, 'conf', 'schema.xml'),
+        @options
+      )
+      SolrconfigUpdater.execute(
+        File.join(@solr_home, 'conf', 'solrconfig.xml'),
+        @options
+      )
+      LibraryInstaller.execute(File.join(@solr_home, 'lib'), @options)
+    end
+  end
+end

data/lib/sunspot/query/abstract_field_facet.rb

@@ -0,0 +1,52 @@
+module Sunspot
+  module Query
+    class AbstractFieldFacet
+      include RSolr::Char
+
+      def initialize(field, options)
+        @field, @options = field, options
+      end
+
+      def to_params
+        params = {
+          :facet => 'true',
+        }
+        case @options[:sort]
+        when :count
+          params[qualified_param('sort')] = 'true'
+        when :index
+          params[qualified_param('sort')] = 'false'
+        when nil
+        else
+          raise(
+            ArgumentError,
+            "#{@options[:sort].inspect} is not an allowed value for :sort. Allowed options are :count and :index"
+          )
+        end
+        if @options[:limit]
+          params[qualified_param('limit')] = @options[:limit].to_i
+        end
+        if @options[:prefix]
+          params[qualified_param('prefix')] = escape(@options[:prefix].to_s)
+        end
+        params[qualified_param('mincount')] = 
+          case
+          when @options[:minimum_count] then @options[:minimum_count].to_i
+          when @options[:zeros] then 0
+          else 1
+          end
+        params
+      end
+
+      private
+
+      def qualified_param(param)
+        :"f.#{key}.facet.#{param}"
+      end
+      
+      def key
+        @key ||= @options[:name] || @field.indexed_name
+      end
+    end
+  end
+end

data/lib/sunspot/query/boost_query.rb

@@ -0,0 +1,24 @@
+module Sunspot
+  module Query
+    # 
+    # Representation of a BoostQuery, which allows the searcher to specify a
+    # scope for which matching documents should have an extra boost. This is
+    # essentially a conjunction, with an extra instance variable containing
+    # the boost that should be applied.
+    #
+    class BoostQuery < Connective::Conjunction #:nodoc:
+      def initialize(boost)
+        super(false)
+        @boost = boost
+      end
+
+      def to_boolean_phrase
+        if @boost.is_a?(FunctionQuery)
+          "#{@boost}"
+        else
+          "#{super}^#{@boost}"
+        end
+      end
+    end
+  end
+end

data/lib/sunspot/query/common_query.rb

@@ -0,0 +1,85 @@
+module Sunspot
+  module Query #:nodoc:
+    class CommonQuery
+      def initialize(types)
+        @scope = Scope.new
+        @sort = SortComposite.new
+        @components = [@scope, @sort]
+        if types.length == 1
+          @scope.add_positive_restriction(TypeField.instance, Restriction::EqualTo, types.first)
+        else
+          @scope.add_positive_restriction(TypeField.instance, Restriction::AnyOf, types)
+        end
+      end
+
+      def solr_parameter_adjustment=(block)
+        @parameter_adjustment = block
+      end
+
+      def add_sort(sort)
+        @sort << sort
+      end
+
+      def add_field_facet(facet)
+        @components << facet
+        facet
+      end
+
+      def add_query_facet(facet)
+        @components << facet
+        facet
+      end
+
+      def add_function(function)
+        @components << function
+        function
+      end
+
+      def paginate(page, per_page)
+        if @pagination
+          @pagination.page = page
+          @pagination.per_page = per_page
+        else
+          @components << @pagination = Pagination.new(page, per_page)
+        end
+      end
+
+      def to_params
+        params = {}
+        @components.each do |component|
+          Sunspot::Util.deep_merge!(params, component.to_params)
+        end
+        @parameter_adjustment.call(params) if @parameter_adjustment
+        params[:q] ||= '*:*'
+        params
+      end
+
+      def [](key)
+        to_params[key.to_sym]
+      end
+
+      def page
+        @pagination.page if @pagination
+      end
+
+      def per_page
+        @pagination.per_page if @pagination
+      end
+
+
+      private
+
+      #
+      # If we have a single fulltext query, merge is normally. If there are
+      # multiple nested queries, serialize them as `_query_` subqueries.
+      #
+      def merge_fulltext(params)
+        return nil if @fulltexts.empty?
+        return Sunspot::Util.deep_merge!(params, @fulltexts.first.to_params) if @fulltexts.length == 1
+        subqueries = @fulltexts.map {|fulltext| fulltext.to_subquery }.join(' ')
+        Sunspot::Util.deep_merge!(params, {:q => subqueries})
+      end
+
+    end
+  end
+end

data/lib/sunspot/query/composite_fulltext.rb

@@ -0,0 +1,36 @@
+module Sunspot
+  module Query
+    class CompositeFulltext
+      def initialize
+        @components = []
+      end
+
+      def add(keywords)
+        @components << dismax = Dismax.new(keywords)
+        dismax
+      end
+      
+      def add_location(field, lat, lng, options)
+        @components << location = Geo.new(field, lat, lng, options)
+        location
+      end
+
+      def to_params
+        case @components.length
+        when 0
+          {}
+        when 1
+          @components.first.to_params.merge(:fl => '* score')
+        else
+          to_subqueries.merge(:fl => '* score')
+        end
+      end
+
+      private
+
+      def to_subqueries
+        { :q => @components.map { |dismax| dismax.to_subquery }.join(' ') }
+      end
+    end
+  end
+end

data/lib/sunspot/query/connective.rb

@@ -0,0 +1,206 @@
+module Sunspot
+  module Query
+    module Connective #:nodoc:all
+      # 
+      # Base class for connectives (conjunctions and disjunctions).
+      #
+      class Abstract
+        include Filter
+
+        def initialize(negated = false) #:nodoc:
+          @negated = negated
+          @components = []
+        end
+
+        # 
+        # Add a restriction to the connective.
+        #
+        def add_restriction(negated, field, restriction_type, *value)
+          add_component(restriction_type.new(negated, field, *value))
+        end
+
+        # 
+        # Add a shorthand restriction; the restriction type is determined by
+        # the value.
+        #
+        def add_shorthand_restriction(negated, field, value)
+          restriction_type =
+            case value
+            when Array then Restriction::AnyOf
+            when Range then Restriction::Between
+            else Restriction::EqualTo
+            end
+          add_restriction(negated, field, restriction_type, value)
+        end
+
+        # 
+        # Add a positive restriction. The restriction will match all
+        # documents who match the terms fo the restriction.
+        #
+        def add_positive_restriction(field, restriction_type, value)
+          add_restriction(false, field, restriction_type, value)
+        end
+
+        # 
+        # Add a positive shorthand restriction (see add_shorthand_restriction)
+        #
+        def add_positive_shorthand_restriction(field, value)
+          add_shorthand_restriction(false, field, value)
+        end
+
+        # 
+        # Add a negated restriction. The added restriction will match all
+        # documents who do not match the terms of the restriction.
+        #
+        def add_negated_restriction(field, restriction_type, value)
+          add_restriction(true, field, restriction_type, value)
+        end
+
+        # 
+        # Add a negated shorthand restriction (see add_shorthand_restriction)
+        #
+        def add_negated_shorthand_restriction(field, value)
+          add_shorthand_restriction(true, field, value)
+        end
+
+        # 
+        # Add a new conjunction and return it.
+        #
+        def add_conjunction
+          add_component(Conjunction.new)
+        end
+
+        # 
+        # Add a new disjunction and return it.
+        #
+        def add_disjunction
+          add_component(Disjunction.new)
+        end
+
+        # 
+        # Add an arbitrary component to the conjunction, and return it.
+        # The component must respond to #to_boolean_phrase
+        #
+        def add_component(component)
+          @components << component
+          component
+        end
+
+        # 
+        # Express the connective as a Lucene boolean phrase.
+        #
+        def to_boolean_phrase #:nodoc:
+          unless @components.empty?
+            phrase =
+              if @components.length == 1
+                @components.first.to_boolean_phrase
+              else
+                component_phrases = @components.map do |component|
+                  component.to_boolean_phrase
+                end
+                "(#{component_phrases.join(" #{connector} ")})"
+              end
+            if negated?
+              "-#{phrase}"
+            else
+              phrase
+            end
+          end
+        end
+
+        # 
+        # Connectives can be negated during the process of denormalization that
+        # is performed when a disjunction contains a negated component. This
+        # method conforms to the duck type for all boolean query components.
+        #
+        def negated?
+          @negated
+        end
+
+        # 
+        # Returns a new connective that's a negated version of this one.
+        #
+        def negate
+          negated = self.class.new(!negated?)
+          @components.each do |component|
+            negated.add_component(component)
+          end
+          negated
+        end
+      end
+
+      # 
+      # Disjunctions combine their components with an OR operator.
+      #
+      class Disjunction < Abstract
+        class <<self
+          def inverse
+            Conjunction
+          end
+        end
+
+        # 
+        # Express this disjunction as a Lucene boolean phrase
+        #
+        def to_boolean_phrase
+          if @components.any? { |component| component.negated? }
+            denormalize.to_boolean_phrase
+          else
+            super
+          end
+        end
+
+        # 
+        # No-op - this is already a disjunction
+        #
+        def add_disjunction
+          self
+        end
+
+        private
+
+        def connector
+          'OR'
+        end
+
+        # 
+        # If a disjunction contains negated components, it must be
+        # "denormalized", because the Lucene parser interprets any negated
+        # boolean phrase using AND semantics (this isn't a bug, it's just a
+        # subtlety of how Lucene parses queries). So, per DeMorgan's law we
+        # create a negated conjunction and add to it all of our components,
+        # negated themselves, which creates a query whose Lucene semantics are
+        # in line with our intentions.
+        #
+        def denormalize
+          denormalized = self.class.inverse.new(!negated?)
+          @components.each do |component|
+            denormalized.add_component(component.negate)
+          end
+          denormalized
+        end
+      end
+
+      # 
+      # Conjunctions combine their components with an AND operator.
+      #
+      class Conjunction < Abstract
+        class <<self
+          def inverse
+            Disjunction
+          end
+        end
+
+        def add_conjunction
+          self
+        end
+
+        private
+
+        def connector
+          'AND'
+        end
+      end
+    end
+  end
+end

data/lib/sunspot/query/date_field_facet.rb

@@ -0,0 +1,14 @@
+module Sunspot
+  module Query
+    class DateFieldFacet < AbstractFieldFacet
+      def to_params
+        params = super
+        params[:"facet.date"] = [@field.indexed_name]
+        params[qualified_param('date.start')] = @field.to_indexed(@options[:time_range].first)
+        params[qualified_param('date.end')] = @field.to_indexed(@options[:time_range].last)
+        params[qualified_param('date.gap')] = "+#{@options[:time_interval] || 86400}SECONDS"
+        params
+      end
+    end
+  end
+end

data/lib/sunspot/query/dismax.rb

@@ -0,0 +1,128 @@
+module Sunspot
+  module Query
+
+    #
+    # Solr full-text queries use Solr's DisMaxRequestHandler, a search handler
+    # designed to process user-entered phrases, and search for individual
+    # words across a union of several fields.
+    #
+    class Dismax
+      attr_writer :minimum_match, :phrase_slop, :query_phrase_slop, :tie
+
+      def initialize(keywords)
+        @keywords = keywords
+        @fulltext_fields = {}
+        @boost_queries = []
+        @boost_functions = []
+        @highlights = []
+      end
+
+      #
+      # The query as Solr parameters
+      #
+      def to_params
+        params = { :q => @keywords }
+        params[:fl] = '* score'
+        params[:qf] = @fulltext_fields.values.map { |field| field.to_boosted_field }.join(' ')
+        params[:defType] = 'dismax'
+        if @phrase_fields
+          params[:pf] = @phrase_fields.map { |field| field.to_boosted_field }.join(' ')
+        end
+        unless @boost_queries.empty?
+          params[:bq] = @boost_queries.map do |boost_query|
+            boost_query.to_boolean_phrase
+          end
+        end
+        unless @boost_functions.empty?
+          params[:bf] = @boost_functions.map do |boost_function|
+            boost_function.to_s
+          end
+        end
+        if @minimum_match
+          params[:mm] = @minimum_match
+        end
+        if @phrase_slop
+          params[:ps] = @phrase_slop
+        end
+        if @query_phrase_slop
+          params[:qs] = @query_phrase_slop
+        end
+        if @tie
+          params[:tie] = @tie
+        end
+        @highlights.each do |highlight|
+          Sunspot::Util.deep_merge!(params, highlight.to_params)
+        end
+        params
+      end
+
+      #
+      # Serialize the query as a Solr nested subquery.
+      #
+      def to_subquery
+        params = self.to_params
+        params.delete :defType
+        params.delete :fl
+        keywords = params.delete(:q)
+        options = params.map { |key, value| "#{key}='#{escape_quotes(value)}'"}.join(' ')
+        "_query_:\"{!dismax #{options}}#{escape_quotes(keywords)}\""
+      end
+
+      #
+      # Assign a new boost query and return it.
+      #
+      def create_boost_query(factor)
+        @boost_queries << boost_query = BoostQuery.new(factor)
+        boost_query
+      end
+
+      # 
+      # Add a boost function
+      #
+      def add_boost_function(function_query)
+        @boost_functions << function_query
+      end
+
+      #
+      # Add a fulltext field to be searched, with optional boost.
+      #
+      def add_fulltext_field(field, boost = nil)
+        @fulltext_fields[field.indexed_name] = TextFieldBoost.new(field, boost)
+      end
+
+      #
+      # Add a phrase field for extra boost.
+      #
+      def add_phrase_field(field, boost = nil)
+        @phrase_fields ||= []
+        @phrase_fields << TextFieldBoost.new(field, boost)
+      end
+
+      #
+      # Set highlighting options for the query. If fields is empty, the
+      # Highlighting object won't pass field names at all, which means
+      # the dismax's :qf parameter will be used by Solr.
+      #
+      def add_highlight(fields=[], options={})
+        @highlights << Highlighting.new(fields, options)
+      end
+
+      #
+      # Determine if a given field is being searched. Used by DSL to avoid
+      # overwriting boost parameters when injecting defaults.
+      #
+      def has_fulltext_field?(field)
+        @fulltext_fields.has_key?(field.indexed_name)
+      end
+
+
+      private
+
+      def escape_quotes(value)
+        return value unless value.is_a? String
+        value.gsub(/(['"])/, '\\\\\1')
+      end
+
+    end
+  end
+end