pduey-sunspot 1.2.1.1

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

data/lib/sunspot/session_proxy/silent_fail_session_proxy.rb

@@ -0,0 +1,42 @@
+require File.join(File.dirname(__FILE__), 'abstract_session_proxy')
+
+module Sunspot
+  module SessionProxy
+    class SilentFailSessionProxy < AbstractSessionProxy
+      
+      attr_reader :search_session
+      
+      delegate :new_search, :search, :config,
+                :new_more_like_this, :more_like_this,
+                :delete_dirty, :delete_dirty?,
+                :to => :search_session
+      
+      def initialize(search_session = Sunspot.session)
+        @search_session = search_session
+      end
+      
+      def rescued_exception(method, e)
+        $stderr.puts("Exception in #{method}: #{e.message}")
+      end
+
+      SUPPORTED_METHODS = [
+        :batch, :commit, :commit_if_dirty, :commit_if_delete_dirty, :dirty?,
+        :index!, :index, :optimize, :remove!, :remove, :remove_all!, :remove_all,
+        :remove_by_id!, :remove_by_id
+      ]
+
+      SUPPORTED_METHODS.each do |method|
+        module_eval(<<-RUBY)
+          def #{method}(*args, &block)
+            begin
+              search_session.#{method}(*args, &block)
+            rescue => e
+              self.rescued_exception(:#{method}, e)
+            end
+          end
+        RUBY
+      end
+      
+    end
+  end
+end

data/lib/sunspot/session_proxy/thread_local_session_proxy.rb

@@ -0,0 +1,37 @@
+require 'monitor'
+require File.join(File.dirname(__FILE__), 'abstract_session_proxy')
+
+module Sunspot
+  module SessionProxy
+    # 
+    # This class implements a session proxy that creates a different Session
+    # object for each thread. Any multithreaded application should use this
+    # proxy.
+    #
+    class ThreadLocalSessionProxy < AbstractSessionProxy
+      FINALIZER = Proc.new do |object_id|
+        Thread.current[:"sunspot_session_#{object_id}"] = nil
+      end
+
+      # The configuration with which the thread-local sessions are initialized.
+      attr_reader :config
+      @@next_id = 0
+
+      delegate :batch, :commit, :commit_if_delete_dirty, :commit_if_dirty, :delete_dirty?, :dirty?, :index, :index!, :new_search, :optimize, :remove, :remove!, :remove_all, :remove_all!, :remove_by_id, :remove_by_id!, :search, :more_like_this, :new_more_like_this, :to => :session
+
+      # 
+      # Optionally pass an existing Sunspot::Configuration object. If none is
+      # passed, a default configuration is used; it can then be modified using
+      # the #config attribute.
+      #
+      def initialize(config = Sunspot::Configuration.new)
+        @config = config
+        ObjectSpace.define_finalizer(self, FINALIZER)
+      end
+
+      def session #:nodoc:
+        Thread.current[:"sunspot_session_#{object_id}"] ||= Session.new(config)
+      end
+    end
+  end
+end

data/lib/sunspot/setup.rb

@@ -0,0 +1,350 @@
+module Sunspot
+  # 
+  # This class encapsulates the search/indexing setup for a given class. Its
+  # contents are built using the Sunspot.setup method.
+  #
+  class Setup #:nodoc:
+    attr_reader :class_object_id
+    def initialize(clazz)
+      @class_object_id = clazz.object_id
+      @class_name = clazz.name
+      @field_factories, @text_field_factories, @dynamic_field_factories,
+        @field_factories_cache, @text_field_factories_cache,
+        @dynamic_field_factories_cache = *Array.new(6) { Hash.new }
+      @stored_field_factories_cache = Hash.new { |h, k| h[k] = [] }
+      @more_like_this_field_factories_cache = Hash.new { |h, k| h[k] = [] }
+      @dsl = DSL::Fields.new(self)
+      add_field_factory(:class, Type::ClassType.instance)
+    end
+
+    def type_names
+      [@class_name]
+    end
+
+    # 
+    # Add field factory for scope/ordering
+    #
+    def add_field_factory(name, type, options = {}, &block)
+      stored, more_like_this = options[:stored], options[:more_like_this]
+      field_factory = FieldFactory::Static.new(name, type, options, &block)
+      @field_factories[field_factory.signature] = field_factory
+      @field_factories_cache[field_factory.name] = field_factory
+      if stored
+        @stored_field_factories_cache[field_factory.name] << field_factory
+      end
+      if more_like_this
+        @more_like_this_field_factories_cache[field_factory.name] << field_factory
+      end
+    end
+
+    # 
+    # Add field_factories for fulltext search
+    #
+    # ==== Parameters
+    #
+    # field_factories<Array>:: Array of Sunspot::Field objects
+    #
+    def add_text_field_factory(name, options = {}, &block)
+      stored, more_like_this = options[:stored], options[:more_like_this]
+      field_factory = FieldFactory::Static.new(name, Type::TextType.instance, options, &block)
+      @text_field_factories[name] = field_factory
+      @text_field_factories_cache[field_factory.name] = field_factory
+      if stored
+        @stored_field_factories_cache[field_factory.name] << field_factory
+      end
+      if more_like_this
+        @more_like_this_field_factories_cache[field_factory.name] << field_factory
+      end
+    end
+
+    #
+    # Add dynamic field_factories
+    #
+    # ==== Parameters
+    # 
+    # field_factories<Array>:: Array of dynamic field objects
+    # 
+    def add_dynamic_field_factory(name, type, options = {}, &block)
+      stored, more_like_this = options[:stored], options[:more_like_this]
+      field_factory = FieldFactory::Dynamic.new(name, type, options, &block)
+      @dynamic_field_factories[field_factory.signature] = field_factory
+      @dynamic_field_factories_cache[field_factory.name] = field_factory
+      if stored
+        @stored_field_factories_cache[field_factory.name] << field_factory
+      end
+      if more_like_this
+        @more_like_this_field_factories_cache[field_factory.name] << field_factory
+      end
+    end
+
+    # 
+    # Add a document boost to documents at index time. Document boost can be
+    # static (the same for all documents of this class), or extracted on a per-
+    # document basis using either attribute or block extraction as per usual.
+    #
+    def add_document_boost(attr_name, &block)
+      @document_boost_extractor =
+        if attr_name
+          if attr_name.respond_to?(:to_f)
+            DataExtractor::Constant.new(attr_name)
+          else
+            DataExtractor::AttributeExtractor.new(attr_name)
+          end
+        else
+          DataExtractor::BlockExtractor.new(&block)
+        end
+    end
+
+    # 
+    # Builder method for evaluating the setup DSL
+    #
+    def setup(&block)
+      Util.instance_eval_or_call(@dsl, &block)
+    end
+
+    # 
+    # Return the Field with the given (public-facing) name
+    #
+    def field(field_name)
+      if field_factory = @field_factories_cache[field_name.to_sym]
+        field_factory.build
+      else
+        raise(
+          UnrecognizedFieldError,
+          "No field configured for #{@class_name} with name '#{field_name}'"
+        )
+      end
+    end
+
+    # 
+    # Return one or more text fields with the given public-facing name. This
+    # implementation will always return a single field (in an array), but
+    # CompositeSetup objects might return more than one.
+    #
+    def text_fields(field_name)
+      text_field = 
+        if field_factory = @text_field_factories_cache[field_name.to_sym]
+          field_factory.build
+        else
+          raise(
+            UnrecognizedFieldError,
+            "No text field configured for #{@class_name} with name '#{field_name}'"
+          )
+        end
+      [text_field]
+    end
+
+    #
+    # Return one or more stored fields (can be either attribute or text fields)
+    # for the given name.
+    #
+    def stored_fields(field_name, dynamic_field_name = nil)
+      @stored_field_factories_cache[field_name.to_sym].map do |field_factory|
+        if dynamic_field_name
+          field_factory.build(dynamic_field_name)
+        else
+          field_factory.build
+        end
+      end
+    end
+
+    #
+    # Return one or more more_like_this fields (can be either attribute or text fields)
+    # for the given name.
+    #
+    def more_like_this_fields(field_name)
+      @more_like_this_field_factories_cache[field_name.to_sym].map do |field_factory|
+        field_factory.build
+      end
+    end
+
+    # 
+    # Return the DynamicFieldFactory with the given base name
+    #
+    def dynamic_field_factory(field_name)
+      @dynamic_field_factories_cache[field_name.to_sym] || raise(
+        UnrecognizedFieldError,
+        "No dynamic field configured for #{@class_name} with name '#{field_name}'"
+      )
+    end
+
+    # 
+    # Return all attribute fields
+    #
+    def fields
+      field_factories.map { |field_factory| field_factory.build }
+    end
+
+    # 
+    # Return all text fields
+    #
+    def all_text_fields
+      text_field_factories.map { |text_field_factory| text_field_factory.build }
+    end
+
+    # 
+    # Return all more_like_this fields
+    #
+    def all_more_like_this_fields
+      @more_like_this_field_factories_cache.values.map do |field_factories| 
+        field_factories.map { |field_factory| field_factory.build }
+      end.flatten
+    end
+
+    # 
+    # Get the field_factories associated with this setup as well as all inherited field_factories
+    #
+    # ==== Returns
+    #
+    # Array:: Collection of all field_factories associated with this setup
+    #
+    def field_factories
+      collection_from_inheritable_hash(:field_factories)
+    end
+
+    # 
+    # Get the text field_factories associated with this setup as well as all inherited
+    # text field_factories
+    #
+    # ==== Returns
+    #
+    # Array:: Collection of all text field_factories associated with this setup
+    #
+    def text_field_factories
+      collection_from_inheritable_hash(:text_field_factories)
+    end
+
+    # 
+    # Get all static, dynamic, and text field_factories associated with this setup as
+    # well as all inherited field_factories
+    #
+    # ==== Returns
+    #
+    # Array:: Collection of all text and scope field_factories associated with this setup
+    #
+    def all_field_factories
+      all_field_factories = []
+      all_field_factories.concat(field_factories).concat(text_field_factories).concat(dynamic_field_factories)
+      all_field_factories
+    end
+
+    # 
+    # Get all dynamic field_factories for this and parent setups
+    # 
+    # ==== Returns
+    #
+    # Array:: Dynamic field_factories
+    #
+    def dynamic_field_factories
+      collection_from_inheritable_hash(:dynamic_field_factories)
+    end
+
+    # 
+    # Return the class associated with this setup.
+    #
+    # ==== Returns
+    #
+    # clazz<Class>:: Class setup is configured for
+    #
+    def clazz
+      Util.full_const_get(@class_name)
+    end
+
+    # 
+    # Get the document boost for a given model
+    #
+    def document_boost_for(model)
+      if @document_boost_extractor
+        @document_boost_extractor.value_for(model)
+      end
+    end
+
+    protected
+
+    # 
+    # Get the nearest inherited setup, if any
+    #
+    # ==== Returns
+    # 
+    # Sunspot::Setup:: Setup for the nearest ancestor of this setup's class
+    #
+    def parent
+      Setup.for(clazz.superclass)
+    end
+
+    def get_inheritable_hash(name)
+      hash = instance_variable_get(:"@#{name}")
+      parent.get_inheritable_hash(name).each_pair do |key, value|
+        hash[key] = value unless hash.has_key?(key)
+      end if parent
+      hash
+    end
+
+    private
+
+    def collection_from_inheritable_hash(name)
+      get_inheritable_hash(name).values
+    end
+
+    class <<self
+      # 
+      # Retrieve or create the Setup instance for the given class, evaluating
+      # the given block to add to the setup's configuration
+      #
+      def setup(clazz, &block) #:nodoc:
+        self.for!(clazz).setup(&block)
+      end
+
+      # 
+      # Retrieve the setup instance for the given class, or for the nearest
+      # ancestor that has a setup, if any.
+      #
+      # ==== Parameters
+      #
+      # clazz<Class>:: Class for which to retrieve a setup
+      #
+      # ==== Returns
+      #
+      # Sunspot::Setup::
+      #   Setup instance associated with the given class or its nearest ancestor
+      #   
+      def for(clazz) #:nodoc:
+        setups[clazz.name.to_sym] || self.for(clazz.superclass) if clazz
+      end
+
+      protected
+
+      # 
+      # Retrieve or create a Setup instance for this class
+      #
+      # ==== Parameters
+      #
+      # clazz<Class>:: Class for which to retrieve a setup
+      #
+      # ==== Returns
+      #
+      # Sunspot::Setup:: New or existing setup for this class
+      #
+      def for!(clazz) #:nodoc:
+        setup = setups[clazz.name.to_sym]
+        if setup && setup.class_object_id == clazz.object_id
+          setup
+        else
+          setups[clazz.name.to_sym] = new(clazz)
+        end
+      end
+
+      private
+
+      # Singleton hash of class names to Setup instances
+      #
+      # ==== Returns
+      #
+      # Hash:: Class names keyed to Setup instances
+      #
+      def setups
+        @setups ||= {}
+      end
+    end
+  end
+end