erichummel-sunspot_rails 1.2.1a

data/lib/sunspot/rails/searchable.rb

@@ -0,0 +1,412 @@
+module Sunspot #:nodoc:
+  module Rails #:nodoc:
+    # 
+    # This module adds Sunspot functionality to ActiveRecord models. As well as
+    # providing class and instance methods, it optionally adds lifecycle hooks
+    # to automatically add and remove models from the Solr index as they are
+    # created and destroyed.
+    #
+    module Searchable
+      class <<self
+        def included(base) #:nodoc:
+          base.module_eval do
+            extend(ActsAsMethods)
+          end
+        end
+      end
+
+      module ActsAsMethods
+        # 
+        # Makes a class searchable if it is not already, or adds search
+        # configuration if it is. Note that the options passed in are only used
+        # the first time this method is called for a particular class; so,
+        # search should be defined before activating any mixins that extend
+        # search configuration.
+        #
+        # The block passed into this method is evaluated by the
+        # <code>Sunspot.setup</code> method. See the Sunspot documentation for
+        # complete information on the functionality provided by that method.
+        #
+        # ==== Options (+options+)
+        # 
+        # :auto_index<Boolean>::
+        #   Automatically index models in Solr when they are saved.
+        #   Default: true
+        # :auto_remove<Boolean>::
+        #   Automatically remove models from the Solr index when they are
+        #   destroyed. <b>Setting this option to +false+ is not recommended
+        #   </b>(see the README).
+        # :ignore_attribute_changes_of<Array>::
+        #   Define attributes, that should not trigger a reindex of that
+        #   object. Usual suspects are updated_at or counters.
+        # :include<Mixed>::
+        #   Define default ActiveRecord includes, set this to allow ActiveRecord
+        #   to load required associations when indexing. See ActiveRecord's 
+        #   documentation on eager-loading for examples on how to set this
+        #   Default: [] 
+        #
+        # ==== Example
+        #
+        #   class Post < ActiveRecord::Base
+        #     searchable do
+        #       text :title, :body
+        #       string :sort_title do
+        #         title.downcase.sub(/^(an?|the)/, '')
+        #       end
+        #       integer :blog_id
+        #       time :updated_at
+        #     end
+        #   end
+        #
+        def searchable(options = {}, &block)
+          Sunspot.setup(self, &block)
+
+          if searchable?
+            sunspot_options[:include].concat(Util::Array(options[:include]))
+          else
+            extend ClassMethods
+            include InstanceMethods
+
+            class_attribute :sunspot_options
+
+            unless options[:auto_index] == false
+              before_save :maybe_mark_for_auto_indexing
+              after_save :maybe_auto_index
+            end
+
+            unless options[:auto_remove] == false
+              after_destroy do |searchable|
+                searchable.remove_from_index
+              end
+            end
+            options[:include] = Util::Array(options[:include])
+            
+            self.sunspot_options = options
+          end
+        end
+
+        # 
+        # This method is defined on all ActiveRecord::Base subclasses. It
+        # is false for classes on which #searchable has not been called, and
+        # true for classes on which #searchable has been called.
+        #
+        # ==== Returns
+        #
+        # +false+
+        #
+        def searchable?
+          false
+        end
+      end
+
+      module ClassMethods
+        def self.extended(base) #:nodoc:
+          class <<base
+            alias_method :search, :solr_search unless method_defined? :search
+            alias_method :search_ids, :solr_search_ids unless method_defined? :search_ids
+            alias_method :remove_all_from_index, :solr_remove_all_from_index unless method_defined? :remove_all_from_index
+            alias_method :remove_all_from_index!, :solr_remove_all_from_index! unless method_defined? :remove_all_from_index!
+            alias_method :reindex, :solr_reindex unless method_defined? :reindex
+            alias_method :index, :solr_index unless method_defined? :index
+            alias_method :index_orphans, :solr_index_orphans unless method_defined? :index_orphans
+            alias_method :clean_index_orphans, :solr_clean_index_orphans unless method_defined? :clean_index_orphans
+          end
+        end
+        # 
+        # Search for instances of this class in Solr. The block is delegated to
+        # the Sunspot.search method - see the Sunspot documentation for the full
+        # API.
+        #
+        # ==== Example
+        #
+        #   Post.search(:include => [:blog]) do
+        #     keywords 'best pizza'
+        #     with :blog_id, 1
+        #     order :updated_at, :desc
+        #     facet :category_ids
+        #   end
+        #
+        # ==== Options
+        #
+        # :include:: Specify associations to eager load
+        # :select:: Specify columns to select from database when loading results
+        #
+        # ==== Returns
+        #
+        # Sunspot::Search:: Object containing results, totals, facets, etc.
+        #
+        def solr_search(options = {}, &block)
+          solr_execute_search(options) do
+            Sunspot.new_search(self, &block)
+          end
+        end
+
+        # 
+        # Get IDs of matching results without loading the result objects from
+        # the database. This method may be useful if search is used as an
+        # intermediate step in a larger find operation. The block is the same
+        # as the block provided to the #search method.
+        #
+        # ==== Returns
+        #
+        # Array:: Array of IDs, in the order returned by the search
+        #
+        def solr_search_ids(&block)
+          solr_execute_search_ids do
+            solr_search(&block)
+          end
+        end
+
+        # 
+        # Remove instances of this class from the Solr index.
+        #
+        def solr_remove_all_from_index
+          Sunspot.remove_all(self)
+        end
+
+        # 
+        # Remove all instances of this class from the Solr index and immediately
+        # commit.
+        #
+        #
+        def solr_remove_all_from_index!
+          Sunspot.remove_all!(self)
+        end
+
+        # 
+        # Completely rebuild the index for this class. First removes all
+        # instances from the index, then loads records and indexes them.
+        #
+        # See #index for information on options, etc.
+        #
+        def solr_reindex(options = {})
+          solr_remove_all_from_index
+          solr_index(options)
+        end
+
+        #
+        # Add/update all existing records in the Solr index. The
+        # +batch_size+ argument specifies how many records to load out of the
+        # database at a time. The default batch size is 50; if nil is passed,
+        # records will not be indexed in batches. By default, a commit is issued
+        # after each batch; passing +false+ for +batch_commit+ will disable
+        # this, and only issue a commit at the end of the process. If associated
+        # objects need to indexed also, you can specify +include+ in format 
+        # accepted by ActiveRecord to improve your sql select performance
+        #
+        # ==== Options (passed as a hash)
+        #
+        # batch_size<Integer>:: Batch size with which to load records. Passing
+        #                       'nil' will skip batches.  Default is 50.
+        # batch_commit<Boolean>:: Flag signalling if a commit should be done after
+        #                         after each batch is indexed, default is 'true'
+        # include<Mixed>:: include option to be passed to the ActiveRecord find,
+        #                  used for including associated objects that need to be
+        #                  indexed with the parent object, accepts all formats
+        #                  ActiveRecord::Base.find does
+        # first_id:: The lowest possible ID for this class. Defaults to 0, which
+        #            is fine for integer IDs; string primary keys will need to
+        #            specify something reasonable here.
+        #
+        # ==== Examples
+        #   
+        #   # index in batches of 50, commit after each
+        #   Post.index 
+        #
+        #   # index all rows at once, then commit
+        #   Post.index(:batch_size => nil) 
+        #
+        #   # index in batches of 50, commit when all batches complete
+        #   Post.index(:batch_commit => false) 
+        #
+        #   # include the associated +author+ object when loading to index
+        #   Post.index(:include => :author) 
+        #
+        def solr_index(opts={})
+          options = {
+            :batch_size => 50,
+            :batch_commit => true,
+            :include => self.sunspot_options[:include],
+            :first_id => 0
+          }.merge(opts)
+
+          if options[:batch_size]
+            counter = 0
+            find_in_batches(:include => options[:include], :batch_size => options[:batch_size]) do |records|
+              solr_benchmark options[:batch_size], counter do
+                Sunspot.index(records)
+              end
+              Sunspot.commit if options[:batch_commit]
+              counter += 1
+            end
+            Sunspot.commit unless options[:batch_commit]
+          else
+            Sunspot.index!(all(:include => options[:include]))
+          end
+        end
+
+        # 
+        # Return the IDs of records of this class that are indexed in Solr but
+        # do not exist in the database. Under normal circumstances, this should
+        # never happen, but this method is provided in case something goes
+        # wrong. Usually you will want to rectify the situation by calling
+        # #clean_index_orphans or #reindex
+        # 
+        # ==== Returns
+        #
+        # Array:: Collection of IDs that exist in Solr but not in the database
+        def solr_index_orphans
+          count = self.count
+          indexed_ids = solr_search_ids { paginate(:page => 1, :per_page => count) }.to_set
+          all(:select => 'id').each do |object|
+            indexed_ids.delete(object.id)
+          end
+          indexed_ids.to_a
+        end
+
+        # 
+        # Find IDs of records of this class that are indexed in Solr but do not
+        # exist in the database, and remove them from Solr. Under normal
+        # circumstances, this should not be necessary; this method is provided
+        # in case something goes wrong.
+        #
+        def solr_clean_index_orphans
+          solr_index_orphans.each do |id|
+            new do |fake_instance|
+              fake_instance.id = id
+            end.solr_remove_from_index
+          end
+        end
+
+        # 
+        # Classes that have been defined as searchable return +true+ for this
+        # method.
+        #
+        # ==== Returns
+        #
+        # +true+
+        #
+        def searchable?
+          true
+        end
+        
+        def solr_execute_search(options = {})
+          options.assert_valid_keys(:include, :select)
+          search = yield
+          unless options.empty?
+            search.build do |query|
+              if options[:include]
+                query.data_accessor_for(self).include = options[:include]
+              end
+              if options[:select]
+                query.data_accessor_for(self).select = options[:select]
+              end
+            end
+          end
+          search.execute
+        end
+
+        def solr_execute_search_ids(options = {})
+          search = yield
+          search.raw_results.map { |raw_result| raw_result.primary_key.to_i }
+        end
+        
+        protected
+        
+        # 
+        # Does some logging for benchmarking indexing performance
+        #
+        def solr_benchmark(batch_size, counter,  &block)
+          start = Time.now
+          logger.info("[#{Time.now}] Start Indexing")
+          yield
+          elapsed = Time.now-start
+          logger.info("[#{Time.now}] Completed Indexing. Rows indexed #{counter * batch_size}. Rows/sec: #{batch_size/elapsed.to_f} (Elapsed: #{elapsed} sec.)")
+        end
+
+      end
+
+      module InstanceMethods
+        def self.included(base) #:nodoc:
+          base.module_eval do
+            alias_method :index, :solr_index unless method_defined? :index
+            alias_method :index!, :solr_index! unless method_defined? :index!
+            alias_method :remove_from_index, :solr_remove_from_index unless method_defined? :remove_from_index
+            alias_method :remove_from_index!, :solr_remove_from_index! unless method_defined? :remove_from_index!
+            alias_method :more_like_this, :solr_more_like_this unless method_defined? :more_like_this
+            alias_method :more_like_this_ids, :solr_more_like_this_ids unless method_defined? :more_like_this_ids
+          end
+        end
+        # 
+        # Index the model in Solr. If the model is already indexed, it will be
+        # updated. Using the defaults, you will usually not need to call this
+        # method, as models are indexed automatically when they are created or
+        # updated. If you have disabled automatic indexing (see
+        # ClassMethods#searchable), this method allows you to manage indexing
+        # manually.
+        #
+        def solr_index
+          Sunspot.index(self)
+        end
+
+        # 
+        # Index the model in Solr and immediately commit. See #index
+        #
+        def solr_index!
+          Sunspot.index!(self)
+        end
+        
+        # 
+        # Remove the model from the Solr index. Using the defaults, this should
+        # not be necessary, as models will automatically be removed from the
+        # index when they are destroyed. If you disable automatic removal
+        # (which is not recommended!), you can use this method to manage removal
+        # manually.
+        #
+        def solr_remove_from_index
+          Sunspot.remove(self)
+        end
+
+        # 
+        # Remove the model from the Solr index and commit immediately. See
+        # #remove_from_index
+        #
+        def solr_remove_from_index!
+          Sunspot.remove!(self)
+        end
+
+        def solr_more_like_this(*args, &block)
+          options = args.extract_options!
+          self.class.solr_execute_search(options) do
+            Sunspot.new_more_like_this(self, *args, &block)
+          end
+        end
+
+        def solr_more_like_this_ids(&block)
+          self.class.solr_execute_search_ids do
+            solr_more_like_this(&block)
+          end
+        end
+
+        private
+
+        def maybe_mark_for_auto_indexing
+          @marked_for_auto_indexing =
+            if !new_record? && ignore_attributes = self.class.sunspot_options[:ignore_attribute_changes_of]
+              @marked_for_auto_indexing = !(changed.map { |attr| attr.to_sym } - ignore_attributes).blank?
+            else
+              true
+            end
+          true
+        end
+
+        def maybe_auto_index
+          if @marked_for_auto_indexing
+            solr_index
+            remove_instance_variable(:@marked_for_auto_indexing)
+          end
+        end
+      end
+    end
+  end
+end

data/lib/sunspot/rails/server.rb

@@ -0,0 +1,173 @@
+module Sunspot
+  module Rails
+    class Server < Sunspot::Server
+      # ActiveSupport log levels are integers; this array maps them to the
+      # appropriate java.util.logging.Level constant
+      LOG_LEVELS = %w(FINE INFO WARNING SEVERE SEVERE INFO)
+
+      def start
+        bootstrap
+        super
+      end
+
+      def run
+        bootstrap
+        super
+      end
+
+      #
+      # Bootstrap a new solr_home by creating all required
+      # directories. 
+      #
+      # ==== Returns
+      #
+      # Boolean:: success
+      #
+      def bootstrap
+        unless @bootstrapped
+          install_solr_home
+          @bootstrapped = true
+        end
+      end
+
+      # 
+      # Directory to store custom libraries for solr
+      #
+      def lib_path
+        File.join( solr_home, 'lib' )
+      end
+
+      # 
+      # Directory in which to store PID files
+      #
+      def pid_dir
+        configuration.pid_dir || File.join(::Rails.root, 'tmp', 'pids')
+      end
+
+      # 
+      # Name of the PID file
+      #
+      def pid_file
+        "sunspot-solr-#{::Rails.env}.pid"
+      end
+
+      # 
+      # Directory to store lucene index data files
+      #
+      # ==== Returns
+      #
+      # String:: data_path
+      #
+      def solr_data_dir
+        File.join(solr_home, 'data', ::Rails.env)
+      end
+
+      # 
+      # Directory to use for Solr home.
+      #
+      def solr_home
+        File.join(configuration.solr_home)
+      end
+
+      #
+      # Solr start jar
+      #
+      def solr_jar
+        configuration.solr_jar || super
+      end
+
+      # 
+      # Port on which to run Solr
+      #
+      def port
+        configuration.port
+      end
+
+      #
+      # Severity level for logging. This is based on the severity level for the
+      # Rails logger.
+      #
+      def log_level
+        LOG_LEVELS[::Rails.logger.level]
+      end
+
+      # 
+      # Log file for Solr. File is in the rails log/ directory.
+      #
+      def log_file
+        File.join(::Rails.root, 'log', "sunspot-solr-#{::Rails.env}.log")
+      end
+
+      # 
+      # Minimum Java heap size for Solr
+      #
+      def min_memory
+        configuration.min_memory
+      end
+
+      # 
+      # Maximum Java heap size for Solr
+      #
+      def max_memory
+        configuration.max_memory
+      end
+
+      private
+
+      #
+      # access to the Sunspot::Rails::Configuration, defined in
+      # sunspot.yml. Use Sunspot::Rails.configuration if you want
+      # to access the configuration directly.
+      #
+      # ==== returns
+      #
+      # Sunspot::Rails::Configuration:: configuration
+      #
+      def configuration
+        Sunspot::Rails.configuration
+      end
+
+      # 
+      # Directory to store solr config files
+      #
+      # ==== Returns
+      #
+      # String:: config_path
+      #
+      def config_path
+        File.join(solr_home, 'conf')
+      end
+
+      #
+      # Copy default solr configuration files from sunspot
+      # gem to the new solr_home/config directory
+      #
+      # ==== Returns
+      #
+      # Boolean:: success
+      #
+      def install_solr_home
+        unless File.exists?(solr_home)
+          Sunspot::Installer.execute(
+            solr_home,
+            :force => true,
+            :verbose => true
+          )
+        end
+      end
+
+      # 
+      # Create new solr_home, config, log and pid directories
+      #
+      # ==== Returns
+      #
+      # Boolean:: success
+      #
+      def create_solr_directories
+        [solr_data_dir, pid_dir].each do |path|
+          FileUtils.mkdir_p( path )
+        end
+      end
+    end
+  end
+end

data/lib/sunspot/rails/solr_instrumentation.rb

@@ -0,0 +1,21 @@
+module Sunspot
+  module Rails
+    module SolrInstrumentation
+      extend ActiveSupport::Concern
+
+      included do
+        alias_method :request_without_as_instrumentation, :request
+        alias_method :request, :request_with_as_instrumentation
+      end
+
+      module InstanceMethods
+        def request_with_as_instrumentation(path, params={}, *extra)
+          ActiveSupport::Notifications.instrument("request.rsolr",
+                                                  {:path => path, :parameters => params}) do
+            request_without_as_instrumentation(path, params, *extra)
+          end
+        end
+      end
+    end
+  end
+end

data/lib/sunspot/rails/solr_logging.rb

@@ -0,0 +1,63 @@
+module Sunspot
+  module Rails
+    module SolrLogging
+      class <<self
+        def included(base)
+          base.module_eval { alias_method_chain(:request, :rails_logging) }
+        end
+      end
+
+      def request_with_rails_logging(path, params={}, *extra)
+
+        # Set up logging text.
+        body = (params.nil? || params.empty?) ? extra.first : params.inspect
+        action = path[1..-1].capitalize
+        if body == "<commit/>"
+          action = 'Commit'
+          body = ''
+        end
+        body = body[0, 800] + '...' if body.length > 800
+
+        # Make request and log.
+        response = nil
+        begin
+          ms = Benchmark.ms do
+            response = request_without_rails_logging(path, params, *extra)
+          end
+          log_name = 'Solr %s (%.1fms)' % [action, ms]
+          ::Rails.logger.debug(format_log_entry(log_name, body))
+        rescue Exception => e
+          log_name = 'Solr %s (Error)' % [action]
+          ::Rails.logger.error(format_log_entry(log_name, body))
+          raise e
+        end
+
+        response
+      end
+
+      private
+
+      def format_log_entry(message, dump = nil)
+        @colorize_logging ||= begin
+            ::Rails.application.config.colorize_logging # Rails 3
+          rescue NoMethodError
+            ActiveRecord::Base.colorize_logging # Rails 2
+          end
+        if @colorize_logging
+          message_color, dump_color = "4;32;1", "0;1"
+          log_entry = "  \e[#{message_color}m#{message}\e[0m   "
+          log_entry << "\e[#{dump_color}m%#{String === dump ? 's' : 'p'}\e[0m" % dump if dump
+          log_entry
+        else
+          "%s  %s" % [message, dump]
+        end
+      end
+    end
+  end
+end
+
+begin
+  RSolr::Client.module_eval { include(Sunspot::Rails::SolrLogging) }
+rescue NameError # RSolr 0.9.6
+  RSolr::Connection::Base.module_eval { include(Sunspot::Rails::SolrLogging) }
+end

data/lib/sunspot/rails/spec_helper.rb

@@ -0,0 +1,26 @@
+module Sunspot
+  module Rails
+    module SpecHelper
+      def disconnect_sunspot
+        before(:each) do
+          Sunspot.session = StubSessionProxy.new(Sunspot.session)
+        end
+
+        after(:each) do
+          Sunspot.session = Sunspot.session.original_session
+        end
+      end
+    end
+  end
+end
+
+rspec =
+  begin
+    RSpec
+  rescue NameError, ArgumentError
+    Spec::Runner
+  end
+
+rspec.configure do |config|
+  config.extend(Sunspot::Rails::SpecHelper)
+end