erichummel-sunspot 1.2.1a

data/lib/sunspot/session.rb

@@ -0,0 +1,260 @@
+module Sunspot
+  # 
+  # A Sunspot session encapsulates a connection to Solr and a set of
+  # configuration choices. Though users of Sunspot may manually instantiate
+  # Session objects, in the general case it's easier to use the singleton
+  # stored in the Sunspot module. Since the Sunspot module provides all of
+  # the instance methods of Session as class methods, they are not documented
+  # again here.
+  #
+  class Session
+    class <<self
+      attr_writer :connection_class #:nodoc:
+      
+      # 
+      # For testing purposes
+      #
+      def connection_class #:nodoc:
+        @connection_class ||= RSolr
+      end
+    end
+
+    # 
+    # Sunspot::Configuration object for this session
+    #
+    attr_reader :config
+
+    # 
+    # Sessions are initialized with a Sunspot configuration and a Solr
+    # connection. Usually you will want to stick with the default arguments
+    # when instantiating your own sessions.
+    #
+    def initialize(config = Configuration.build, connection = nil)
+      @config = config
+      yield(@config) if block_given?
+      @connection = connection
+      @deletes = @adds = 0
+    end
+
+    # 
+    # See Sunspot.new_search
+    #
+    def new_search(*types, &block)
+      types.flatten!
+      search = Search::StandardSearch.new(
+        connection,
+        setup_for_types(types),
+        Query::StandardQuery.new(types),
+        @config
+      )
+      search.build(&block) if block
+      search
+    end
+
+    #
+    # See Sunspot.search
+    #
+    def search(*types, &block)
+      search = new_search(*types, &block)
+      search.execute
+    end
+
+    # 
+    # See Sunspot.new_more_like_this
+    #
+    def new_more_like_this(object, *types, &block)
+      types[0] ||= object.class
+      mlt = Search::MoreLikeThisSearch.new(
+        connection,
+        setup_for_types(types),
+        Query::MoreLikeThisQuery.new(object, types),
+        @config
+      )
+      mlt.build(&block) if block
+      mlt
+    end
+
+    #
+    # See Sunspot.more_like_this
+    #
+    def more_like_this(object, *types, &block)
+      mlt = new_more_like_this(object, *types, &block)
+      mlt.execute
+    end
+
+    #
+    # See Sunspot.index
+    #
+    def index(*objects)
+      objects.flatten!
+      @adds += objects.length
+      indexer.add(objects)
+    end
+
+    # 
+    # See Sunspot.index!
+    #
+    def index!(*objects)
+      index(*objects)
+      commit
+    end
+
+    #
+    # See Sunspot.commit
+    #
+    def commit
+      @adds = @deletes = 0
+      connection.commit
+    end
+
+    #
+    # See Sunspot.optimize
+    #
+    def optimize
+      @adds = @deletes = 0
+      connection.optimize
+    end
+
+    # 
+    # See Sunspot.remove
+    #
+    def remove(*objects, &block)
+      if block
+        types = objects
+        conjunction = Query::Connective::Conjunction.new
+        if types.length == 1
+          conjunction.add_positive_restriction(TypeField.instance, Query::Restriction::EqualTo, types.first)
+        else
+          conjunction.add_positive_restriction(TypeField.instance, Query::Restriction::AnyOf, types)
+        end
+        dsl = DSL::Scope.new(conjunction, setup_for_types(types))
+        Util.instance_eval_or_call(dsl, &block)
+        indexer.remove_by_scope(conjunction)
+      else
+        objects.flatten!
+        @deletes += objects.length
+        objects.each do |object|
+          indexer.remove(object)
+        end
+      end
+    end
+
+    # 
+    # See Sunspot.remove!
+    #
+    def remove!(*objects)
+      remove(*objects)
+      commit
+    end
+
+    # 
+    # See Sunspot.remove_by_id
+    #
+    def remove_by_id(clazz, id)
+      class_name =
+        if clazz.is_a?(Class)
+          clazz.name
+        else
+          clazz.to_s
+        end
+      indexer.remove_by_id(class_name, id)
+    end
+
+    # 
+    # See Sunspot.remove_by_id!
+    #
+    def remove_by_id!(clazz, id)
+      remove_by_id(clazz, id)
+      commit
+    end
+
+    #
+    # See Sunspot.remove_all
+    #
+    def remove_all(*classes)
+      classes.flatten!
+      if classes.empty?
+        @deletes += 1
+        indexer.remove_all
+      else
+        @deletes += classes.length
+        classes.each { |clazz| indexer.remove_all(clazz) }
+      end
+    end
+
+    # 
+    # See Sunspot.remove_all!
+    #
+    def remove_all!(*classes)
+      remove_all(*classes)
+      commit
+    end
+
+    # 
+    # See Sunspot.dirty?
+    #
+    def dirty?
+      (@deletes + @adds) > 0
+    end
+
+    # 
+    # See Sunspot.commit_if_dirty
+    #
+    def commit_if_dirty
+      commit if dirty?
+    end
+    
+    # 
+    # See Sunspot.delete_dirty?
+    #
+    def delete_dirty?
+      @deletes > 0
+    end
+
+    # 
+    # See Sunspot.commit_if_delete_dirty
+    #
+    def commit_if_delete_dirty
+      commit if delete_dirty?
+    end
+    
+    # 
+    # See Sunspot.batch
+    #
+    def batch
+      indexer.start_batch
+      yield
+      indexer.flush_batch
+    end
+
+    private
+
+    # 
+    # Retrieve the Solr connection for this session, creating one if it does not
+    # already exist.
+    #
+    # ==== Returns
+    #
+    # RSolr::Connection::Base:: The connection for this session
+    #
+    def connection
+      @connection ||=
+        self.class.connection_class.connect(:url => config.solr.url)
+    end
+
+    def indexer
+      @indexer ||= Indexer.new(connection)
+    end
+
+    def setup_for_types(types)
+      if types.empty?
+        raise(ArgumentError, "You must specify at least one type to search")
+      end
+      if types.length == 1
+        Setup.for(types.first)
+      else
+        CompositeSetup.for(types)
+      end
+    end
+  end
+end

data/lib/sunspot/session_proxy/abstract_session_proxy.rb

@@ -0,0 +1,29 @@
+module Sunspot
+  module SessionProxy
+    class AbstractSessionProxy #:nodoc:
+      class <<self
+        def delegate(*args)
+          options = Util.extract_options_from(args)
+          delegate = options[:to]
+          args.each do |method|
+            module_eval(<<-RUBY, __FILE__, __LINE__ + 1)
+              def #{method}(*args, &block)
+                #{delegate}.#{method}(*args, &block)
+              end
+            RUBY
+          end
+        end
+
+        def not_supported(*methods)
+          methods.each do |method|
+            module_eval(<<-RUBY, __FILE__, __LINE__ + 1)
+              def #{method}(*args, &block)
+                raise NotSupportedError, "#{name} does not support #{method.inspect}"
+              end
+            RUBY
+          end
+        end
+      end
+    end
+  end
+end

data/lib/sunspot/session_proxy/class_sharding_session_proxy.rb

@@ -0,0 +1,66 @@
+module Sunspot
+  module SessionProxy
+    # 
+    # An abstract subclass of ShardingSessionProxy that shards by class.
+    # Concrete subclasses should not override the #session_for method, but
+    # should instead implement the #session_for_class method. They must also
+    # still implement the #all_sessions method.
+    #
+    # Unlike its parent class, ClassShardingSessionProxy implements
+    # #remove_by_id and all flavors of #remove_all.
+    #
+    class ClassShardingSessionProxy < ShardingSessionProxy
+      # 
+      # Remove the Session object pointing at the shard that indexes the given
+      # class.
+      #
+      # <strong>Concrete subclasses must implement this method.</strong>
+      #
+      def session_for_class(clazz)
+        raise NotImplementedError
+      end
+
+      # 
+      # See Sunspot.remove_by_id
+      #
+      def remove_by_id(clazz, id)
+        session_for_class(clazz).remove_by_id(clazz, id)
+      end
+
+      # 
+      # See Sunspot.remove_by_id!
+      #
+      def remove_by_id!(clazz, id)
+        session_for_class(clazz).remove_by_id!(clazz, id)
+      end
+
+      # 
+      # See Sunspot.remove_all
+      #
+      def remove_all(clazz = nil)
+        if clazz
+          session_for_class(clazz).remove_all(clazz)
+        else
+          all_sessions.each { |session| session.remove_all }
+        end
+      end
+
+      # 
+      # See Sunspot.remove_all!
+      #
+      def remove_all!(clazz = nil)
+        if clazz
+          session_for_class(clazz).remove_all!(clazz)
+        else
+          all_sessions.each { |session| session.remove_all! }
+        end
+      end
+
+      private
+
+      def session_for(object)
+        session_for_class(object.class)
+      end
+    end
+  end
+end

data/lib/sunspot/session_proxy/id_sharding_session_proxy.rb

@@ -0,0 +1,89 @@
+module Sunspot
+  module SessionProxy
+    # 
+    # A concrete implementation of ShardingSessionProxy that determines the
+    # shard for a given object based on the hash of its class and ID.
+    #
+    # <strong>If you change the number of shard sessions that this proxy
+    # encapsulates, all objects will point to a different shard.</strong> If you
+    # plan on adding more shards over time, consider your own
+    # ShardingSessionProxy implementation that does not determine the session
+    # using modular arithmetic (e.g., IDs 1-10000 go to shard 1, 10001-20000 go
+    # to shard 2, etc.)
+    #
+    # This implementation will, on average, yield an even distribution of
+    # objects across shards.
+    #
+    # Unlike the abstract ShardingSessionProxy, this proxy supports the
+    # #remove_by_id method.
+    #
+    class IdShardingSessionProxy < ShardingSessionProxy
+      #
+      # The shard sessions encapsulated by this class.
+      #
+      attr_reader :sessions
+      alias_method :all_sessions, :sessions #:nodoc:
+
+      # 
+      # Initialize with a search session (see ShardingSessionProxy.new) and a
+      # collection of one or more shard sessions. See note about changing the
+      # number of shard sessions in the documentation for this class.
+      #
+      def initialize(search_session, shard_sessions)
+        super(search_session)
+        @sessions = shard_sessions
+      end
+
+      # 
+      # Return a session based on the hash of the class and ID, modulo the
+      # number of shard sessions.
+      #
+      def session_for(object) #:nodoc:
+        session_for_index_id(Adapters::InstanceAdapter.adapt(object).index_id)
+      end
+
+      # 
+      # See Sunspot.remove_by_id
+      #
+      def remove_by_id(clazz, id)
+        session_for_index_id(
+          Adapters::InstanceAdapter.index_id_for(clazz, id)
+        ).remove_by_id(clazz, id)
+      end
+
+      # 
+      # See Sunspot.remove_by_id!
+      #
+      def remove_by_id!(clazz, id)
+        session_for_index_id(
+          Adapters::InstanceAdapter.index_id_for(clazz, id)
+        ).remove_by_id!(clazz, id)
+      end
+
+      private
+
+      def session_for_index_id(index_id)
+        @sessions[id_hash(index_id) % @sessions.length]
+      end
+
+      # 
+      # This method is implemented explicitly instead of using String#hash to
+      # give predictable behavior across different Ruby interpreters.
+      #
+      if "".respond_to?(:bytes) # Ruby 1.9
+        def id_hash(id)
+          id.bytes.inject { |hash, byte| hash * 31 + byte }
+        end
+      else
+        def id_hash(id)
+          hash, i, len = 0, 0, id.length
+          while i < len
+            hash = hash * 31 + id[i]
+            i += 1
+          end
+          hash
+        end
+      end
+    end
+  end
+end

data/lib/sunspot/session_proxy/master_slave_session_proxy.rb

@@ -0,0 +1,43 @@
+require File.join(File.dirname(__FILE__), 'abstract_session_proxy')
+
+module Sunspot
+  module SessionProxy
+    # 
+    # This session proxy implementation allows Sunspot to be used with a
+    # master/slave Solr deployment. All write methods are delegated to a master
+    # session, and read methods are delegated to a slave session.
+    #
+    class MasterSlaveSessionProxy < AbstractSessionProxy
+      #
+      # The session that connects to the master Solr instance.
+      #
+      attr_reader :master_session
+      # 
+      # The session that connects to the slave Solr instance.
+      #
+      attr_reader :slave_session
+
+      delegate :batch, :commit, :commit_if_delete_dirty, :commit_if_dirty,
+               :config, :delete_dirty?, :dirty?, :index, :index!, :optimize, :remove,
+               :remove!, :remove_all, :remove_all!, :remove_by_id,
+               :remove_by_id!, :to => :master_session
+      delegate :new_search, :search, :new_more_like_this, :more_like_this, :to => :slave_session
+
+      def initialize(master_session, slave_session)
+        @master_session, @slave_session = master_session, slave_session
+      end
+
+      # 
+      # By default, return the configuration for the master session. If the
+      # +delegate+ param is +:slave+, then return config for the slave session.
+      #
+      def config(delegate = :master)
+        case delegate
+        when :master then @master_session.config
+        when :slave then  @slave_session.config
+        else raise(ArgumentError, "Expected :master or :slave")
+        end
+      end
+    end
+  end
+end

data/lib/sunspot/session_proxy/sharding_session_proxy.rb

@@ -0,0 +1,222 @@
+require File.join(File.dirname(__FILE__), 'abstract_session_proxy')
+
+module Sunspot
+  module SessionProxy
+    # 
+    # This is a generic abstract implementation of a session proxy that allows
+    # Sunspot to be used with a distributed (sharded) Solr deployment. Concrete
+    # subclasses should implement the #session_for method, which takes a
+    # searchable object and returns a Session that points to the appropriate
+    # Solr shard for that object. Subclasses should also implement the
+    # #all_sessions object, which returns the collection of all sharded Session
+    # objects.
+    #
+    # The class is initialized with a session that points to the Solr instance
+    # used to perform searches. Searches will have the +:shards+ param injected,
+    # containing references to the Solr instances returned by #all_sessions.
+    #
+    # For more on distributed search, see:
+    # http://wiki.apache.org/solr/DistributedSearch
+    #
+    # The following methods are not supported (although subclasses may in some
+    # cases be able to support them):
+    #
+    # * batch
+    # * config
+    # * remove_by_id
+    # * remove_by_id!
+    # * remove_all with an argument
+    # * remove_all! with an argument
+    #
+    class ShardingSessionProxy < AbstractSessionProxy
+      not_supported :batch, :config, :remove_by_id, :remove_by_id!
+
+      # 
+      # +search_session+ is the session that should be used for searching.
+      #
+      def initialize(search_session = Sunspot.session.new)
+        @search_session = search_session
+      end
+
+      # 
+      # Return the appropriate shard session for the object.
+      #
+      # <strong>Concrete subclasses must implement this method.</strong>
+      #
+      def session_for(object)
+        raise NotImplementedError
+      end
+
+      # 
+      # Return all shard sessions.
+      #
+      # <strong>Concrete subclasses must implement this method.</strong>
+      #
+      def all_sessions
+        raise NotImplementedError
+      end
+
+      # 
+      # See Sunspot.index
+      #
+      def index(*objects)
+        using_sharded_session(objects) { |session, group| session.index(group) }
+      end
+
+      # 
+      # See Sunspot.index!
+      #
+      def index!(*objects)
+        using_sharded_session(objects) { |session, group| session.index!(group) }
+      end
+
+      # 
+      # See Sunspot.remove
+      #
+      def remove(*objects)
+        using_sharded_session(objects) { |session, group| session.remove(group) }
+      end
+
+      # 
+      # See Sunspot.remove!
+      #
+      def remove!(*objects)
+        using_sharded_session(objects) { |session, group| session.remove!(group) }
+      end
+
+      # 
+      # If no argument is passed, behaves like Sunspot.remove_all
+      #
+      # If an argument is passed, will raise NotSupportedError, as the proxy
+      # does not know which session(s) to which to delegate this operation.
+      #
+      def remove_all(clazz = nil)
+        if clazz
+          raise NotSupportedError, "Sharding session proxy does not support remove_all with an argument."
+        else
+          all_sessions.each { |session| session.remove_all }
+        end
+      end
+
+      # 
+      # If no argument is passed, behaves like Sunspot.remove_all!
+      #
+      # If an argument is passed, will raise NotSupportedError, as the proxy
+      # does not know which session(s) to which to delegate this operation.
+      #
+      def remove_all!(clazz = nil)
+        if clazz
+          raise NotSupportedError, "Sharding session proxy does not support remove_all! with an argument."
+        else
+          all_sessions.each { |session| session.remove_all! }
+        end
+      end
+
+      # 
+      # Commit all shards. See Sunspot.commit
+      #
+      def commit
+        all_sessions.each { |session| session.commit }
+      end
+
+      # 
+      # Optimize all shards. See Sunspot.optimize
+      #
+      def optimize
+        all_sessions.each { |session| session.optimize }
+      end
+
+      # 
+      # Commit all dirty sessions. Only dirty sessions will be committed.
+      #
+      # See Sunspot.commit_if_dirty
+      #
+      def commit_if_dirty
+        all_sessions.each { |session| session.commit_if_dirty }
+      end
+
+      # 
+      # Commit all delete-dirty sessions. Only delete-dirty sessions will be
+      # committed.
+      # 
+      # See Sunspot.commit_if_delete_dirty
+      #
+      def commit_if_delete_dirty
+        all_sessions.each { |session| session.commit_if_delete_dirty }
+      end
+
+      # 
+      # Instantiate a new Search object, but don't execute it. The search will
+      # have an extra :shards param injected into the query, which will tell the
+      # Solr instance referenced by the search session to search across all
+      # shards.
+      #
+      # See Sunspot.new_search
+      #
+      def new_search(*types)
+        shard_urls = all_sessions.map { |session| session.config.solr.url }
+        search = @search_session.new_search(*types)
+        search.build do
+          adjust_solr_params { |params| params[:shards] = shard_urls.join(',') }
+          # I feel a little dirty doing this.
+        end
+        search
+      end
+
+      # 
+      # Build and execute a new Search. The search will have an extra :shards
+      # param injected into the query, which will tell the Solr instance
+      # referenced by the search session to search across all shards.
+      #
+      # See Sunspot.search
+      #
+      def search(*types, &block)
+        new_search(*types).execute
+      end
+
+      def more_like_this(object, &block)
+        #FIXME should use shards
+        new_more_like_this(object, &block).execute
+      end
+
+      def new_more_like_this(object, &block)
+        @search_session.new_more_like_this(object, &block)
+      end
+
+      # 
+      # True if any shard session is dirty. Note that directly using the
+      # #commit_if_dirty method is more efficient if that's what you're
+      # trying to do, since in that case only the dirty sessions are committed.
+      #
+      # See Sunspot.dirty?
+      #
+      def dirty?
+        all_sessions.any? { |session| session.dirty? }
+      end
+
+      # 
+      # True if any shard session is delete-dirty. Note that directly using the
+      # #commit_if_delete_dirty method is more efficient if that's what you're
+      # trying to do, since in that case only the delete-dirty sessions are
+      # committed.
+      #
+      def delete_dirty?
+        all_sessions.any? { |session| session.delete_dirty? }
+      end
+
+      private
+
+      # 
+      # Group the objects by which shard session they correspond to, and yield
+      # each session and is corresponding group of objects.
+      #
+      def using_sharded_session(objects)
+        grouped_objects = Hash.new { |h, k| h[k] = [] }
+        objects.flatten.each { |object| grouped_objects[session_for(object)] << object }
+        grouped_objects.each_pair do |session, group|
+          yield(session, group)
+        end
+      end
+    end
+  end
+end