Changeset 317

Timestamp:

02/18/08 20:36:03 (8 months ago)

Author:

jk

Message:

zwischenstand globale config

Files:

• trunk/demo/app/models/content_base.rb (modified) (1 diff)
• trunk/demo/test/unit/shared_index1_test.rb (modified) (3 diffs)
• trunk/demo/test/unit/shared_index2_test.rb (modified) (1 diff)
• trunk/plugin/acts_as_ferret/init.rb (modified) (1 diff)
• trunk/plugin/acts_as_ferret/lib/act_methods.rb (modified) (6 diffs)
• trunk/plugin/acts_as_ferret/lib/acts_as_ferret.rb (modified) (4 diffs)
• trunk/plugin/acts_as_ferret/lib/blank_slate.rb (modified) (1 diff)
• trunk/plugin/acts_as_ferret/lib/class_methods.rb (modified) (7 diffs)
• trunk/plugin/acts_as_ferret/lib/ferret_result.rb (modified) (2 diffs)
• trunk/plugin/acts_as_ferret/lib/ferret_server.rb (modified) (3 diffs)
• trunk/plugin/acts_as_ferret/lib/index.rb (modified) (1 diff)
• trunk/plugin/acts_as_ferret/lib/instance_methods.rb (modified) (1 diff)
• trunk/plugin/acts_as_ferret/lib/local_index.rb (modified) (2 diffs)
• trunk/plugin/acts_as_ferret/lib/multi_index.rb (modified) (1 diff)
• trunk/plugin/acts_as_ferret/lib/rdig_adapter.rb (added)
• trunk/plugin/acts_as_ferret/lib/without_ar.rb (added)

trunk/demo/app/models/content_base.rb

@@ -11 +11 @@
-  # 'more like this' queries to find other content instances with similar
-  # descriptions
-
+ :index => :untokenized 
-                               :title         => { :boost => :title_boost }, 
-                               :description   => { :boost => 1, :store => :yes },

trunk/demo/test/unit/shared_index1_test.rb

@@ -5 +5 @@
-
-  def setup
-(SharedIndex2)
+
-  end
-
@@ -74 +74 @@
-
-  def test_ferret_destroy
-(SharedIndex2)
+
-    result = SharedIndex1.find_id_by_contents("first OR another", :models => :all)
-    assert_equal 4, result.first
@@ -83 +83 @@
-
-  def test_ferret_destroy_ticket_88
-(SharedIndex2)
+
-    result = SharedIndex1.find_id_by_contents("first OR another", :models => :all)
-    assert_equal 4, result.first

trunk/demo/test/unit/shared_index2_test.rb

@@ -5 +5 @@
-
-  def setup
-(SharedIndex2)
+
-  end
-

trunk/plugin/acts_as_ferret/init.rb

@@ -21 +21 @@
-require 'acts_as_ferret'
-
+ActsAsFerret::logger = RAILS_DEFAULT_LOGGER

trunk/plugin/acts_as_ferret/lib/act_methods.rb

@@ -1 +1 @@
-module ActsAsFerret #:nodoc:
-
+
-  # This module defines the acts_as_ferret method and is included into 
-  # ActiveRecord::Base
@@ -73 +73 @@
-    # For downwards compatibility reasons you can also specify the Ferret options in the 
-    # last Hash argument.
-, ferret_options={}
+
-      # default to DRb mode
-      options[:remote] = true if options[:remote].nil?
@@ -100 +100 @@
-      end
-
-
-      extend ClassMethods
-      extend SharedIndexClassMethods if options[:single_index]
@@ -107 +106 @@
-      include MoreLikeThis::InstanceMethods
-
-
-
-
-
+
+
+
+
+
+
+
+
+
+
+
-
-      cattr_accessor :aaf_configuration
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
+
+
+
+
-      end
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
+
+
+
+
-      end
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
+
+
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
+
+
-
-
-
-
+
+
+
+
+
+
+
+
+
+
+
+
+
-    end
-
@@ -203 +156 @@
-    protected
-    
-    # find the most recent version of an index
-    def find_last_index_version(basedir)
-      # check for versioned index
-      versions = Dir.entries(basedir).select do |f| 
-        dir = File.join(basedir, f)
-        File.directory?(dir) && File.file?(File.join(dir, 'segments')) && f =~ /^\d+(_\d+)?$/
-      end
-      if versions.any?
-        # select latest version
-        versions.sort!
-        File.join basedir, versions.last
-      else
-        basedir
-      end
-    end
-
-
-
+
-    # document instance
-    def define_to_field_method(field, options = {})
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
+
+
+
+
+
+
+
-        rescue
-          logger.warn("Error retrieving value for field #{field}: #{$!}")
-val = 
+
-        end
-        logger.debug("Adding field #{field} with value '#{val}' to index")
@@ -249 +177 @@
-
-    def add_fields(field_config)
+      # TODO
+        #field_config.each do |*args| 
+        #  define_to_field_method *args
+        #end                
-      if field_config.is_a? Hash
-key,val
-(key,val)                  
+field, options
+ field, options
-        end
-      elsif field_config.respond_to?(:each)
-        field_config.each do |field| 
-(field)
+ field
-        end                
-      end

trunk/plugin/acts_as_ferret/lib/acts_as_ferret.rb

@@ -34 +34 @@
-require 'ferret_result'
-require 'instance_methods'
+require 'without_ar'
-
-require 'multi_index'
@@ -45 +46 @@
-require 'ferret_server'
-
+require 'rdig_adapter'
-
-# The Rails ActiveRecord Ferret Mixin.
@@ -70 +72 @@
-#
-module ActsAsFerret
+
+  class ActsAsFerretError < StandardError; end
+  class IndexNotDefined < ActsAsFerretError; end
+  class IndexAlreadyDefined < ActsAsFerretError; end
+
+  # default field list for use with a shared index. Set it globally to
+  # avoid having to specify the same :default_field value in every class using
+  # the shared index.
+  @@shared_index_default_fields = nil
+  mattr_accessor :shared_index_default_fields
+
+  @@logger = nil
+  mattr_accessor :logger
-
-  # global Hash containing all multi indexes created by all classes using the plugin
@@ -82 +97 @@
-  def self.ferret_indexes; @@ferret_indexes end
-
+  # holds per-index configuration, key is the index name
+  @@index_definitions = {}
+  # mapping from class name to index name
+  @@index_using_classes = {}
+  def self.index_definitions; @@index_definitions end
+
+  DEFAULT_FIELD_OPTIONS = {
+    :store       => :no, 
+    :highlight   => :yes, 
+    :index       => :yes, 
+    :term_vector => :with_positions_offsets,
+    :boost       => 1.0
+  }
+
+  def self.field_config_for(fieldname, options = {})
+    config = DEFAULT_FIELD_OPTIONS.merge options
+    config[:term_vector] = :no if config[:index] == :no
+    config.delete :via
+    config.delete :boost if config[:boost].is_a?(Symbol) # dynamic boosts aren't handled here
+    return config
+  end
+
+  def self.build_field_config(fields)
+    field_config = {}
+    case fields
+    when Array
+      fields.each { |name| field_config[name] = field_config_for name }
+    when Hash
+      fields.each { |name, options| field_config[name] = field_config_for name, options }
+    else raise InvalidArgumentError.new(":fields option must be Hash or Array")
+    end if fields
+    return field_config
+  end
+
+  # Globally declares an index.
+  #
+  # Use the index in your model classes with
+  #    acts_as_ferret :index => :index_name
+  #
+  # This method is also used to implicitly declare an index when you use the
+  # acts_as_ferret call without the :index option as usual.
+  def self.define_index(name, options = {})
+    name = name.to_sym
+    raise IndexAlreadyDefined.new(name) if index_definitions.has_key?(name)
+    index_definition = {
+      :index_dir => "#{ActsAsFerret::index_dir}/#{name}",
+      :store_class_name => false,
+      :name => name,
+      :single_index => false,
+      :reindex_batch_size => 1000,
+      :ferret => {},
+      :ferret_fields => {},             # list of indexed fields that will be filled later
+      :enabled => true,                 # used for class-wide disabling of Ferret
+      :mysql_fast_batches => true,      # turn off to disable the faster, id based batching mechanism for MySQL
+      :raise_drb_errors => false        # handle DRb connection errors by default
+    }.update( options )
+
+    index_definition[:registered_models] = []
+    
+    # build ferret configuration
+    index_definition[:ferret] = {
+      :or_default          => false, 
+      :handle_parse_errors => true,
+      :default_field       => nil,              # will be set later on
+      #:max_clauses => 512,
+      #:analyzer => Ferret::Analysis::StandardAnalyzer.new,
+      # :wild_card_downcase => true
+    }.update( options[:ferret] || {} )
+
+    index_definition[:user_default_field] = index_definition[:ferret][:default_field]
+
+    # these properties are somewhat vital to the plugin and shouldn't
+    # be overwritten by the user:
+    index_definition[:ferret].update(
+      :key               => (index_definition[:store_class_name] ? [:id, :class_name] : :id),
+      :path              => index_definition[:index_dir],
+      :auto_flush        => true, # slower but more secure in terms of locking problems TODO disable when running in drb mode?
+      :create_if_missing => true
+    )
+
+
+    unless index_definition[:remote]
+      ActsAsFerret::ensure_directory index_definition[:index_dir] 
+      index_definition[:index_base_dir] = index_definition[:index_dir]
+      index_definition[:index_dir] = find_last_index_version(index_definition[:index_dir])
+      logger.debug "using index in #{index_definition[:index_dir]}"
+    end
+
+    # field config
+    index_definition[:ferret_fields] = build_field_config( options[:fields] )
+    index_definition[:ferret_fields].update build_field_config( options[:additional_fields] )
+
+    index_definitions[name] = index_definition
+    return index_definition
+  end
- 
-
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
-  def self.ensure_directory(dir)
-    FileUtils.mkdir_p dir unless (File.directory?(dir) || File.symlink?(dir))
-  end
+
+
-  
-  # make sure the default index base dir exists. by default, all indexes are created

trunk/plugin/acts_as_ferret/lib/blank_slate.rb

@@ -1 +1 @@
-if defined?(BlankSlate)
-
+it already
-  module ActsAsFerret
-    class BlankSlate < ::BlankSlate

trunk/plugin/acts_as_ferret/lib/class_methods.rb

@@ -21 +21 @@
-    end
-
-
+
+
-    # This is called automatically when no index exists yet.
-    #
-
-
-
-
-
-
-
-
-
+
+
+
+
-    end
-
@@ -44 +40 @@
-      options = Hash === ids.last ? ids.pop : {}
-      ids = ids.first if ids.size == 1 && ids.first.is_a?(Enumerable)
-
+self.name, 
-    end
-
@@ -100 +96 @@
-        offset = 0
-        ids.each_slice(batch_size) do |id_slice|
-          logger.debug "########## slice: #{id_slice.join(',')}"
-          records = find( :all, :conditions => ["id in (?)", id_slice] )
-          logger.debug "########## slice records: #{records.inspect}"
-          #yield records, offset
-          yield find( :all, :conditions => ["id in (?)", id_slice] ), offset
@@ -110 +104 @@
-    end
-
-Switches this class to a new index located in dir.
-Used by the DRb server when switching to a new index version.
+TODO change any references to this method to use
+ActsAsFerret::change_index_dir
-    def index_dir=(dir)
-
-
-
-
-
-
-
-
-
-
-
+
+
-    end
-    
@@ -133 +118 @@
-    # Index object, too.
-    def aaf_index
-ferret_indexes[aaf_configuration[:index_dir]] ||= create_index_instance
+get_index(aaf_configuration[:name])
-    end 
-    
@@ -352 +337 @@
-
-    def lazy_find_by_contents(q, options = {})
+      logger.debug "lazy_find_by_contents: #{q}"
-      result = []
-      rank   = 0
-      total_hits = find_id_by_contents(q, options) do |model, id, score, data|
+        logger.debug "model: #{model}, id: #{id}, data: #{data}"
-        result << FerretResult.new(model, id, score, rank += 1, data)
-      end
@@ -467 +454 @@
-    end
-
-    # creates a new Index instance.
-    def create_index_instance
-      if aaf_configuration[:remote]
-       RemoteIndex
-      elsif aaf_configuration[:single_index]
-        SharedIndex
-      else
-        LocalIndex
-      end.new(aaf_configuration)
-    end
-
-    # combine our conditions with those given by user, if any
-    def combine_conditions(conditions, additional_conditions = [])

trunk/plugin/acts_as_ferret/lib/ferret_result.rb

@@ -21 +21 @@
-      @ferret_rank  = rank
-      @data = data
+      @use_record = false
-    end
-
+
+
+
+
+
-    def method_missing(method, *args, &block)
-
+
+
+
-        to_record.send method, *args, &block
-      else
@@ -40 +47 @@
-        @ar_record.ferret_rank  = ferret_rank
-        @ar_record.ferret_score = ferret_score
+        @use_record = !@ar_record.class.included_modules.include?(ActsAsFerret::RdigAdapter)
-      end
-      @ar_record

trunk/plugin/acts_as_ferret/lib/ferret_server.rb

@@ -145 +145 @@
-
-      # hides LocalIndex#rebuild_index to implement index versioning
-
+
+
+
-        with_class clazz do |clazz|
-          models = models.flatten.uniq.map(&:constantize)
-          models << clazz unless models.include?(clazz)
-          index = new_index_for(clazz, models)
-          reconnect_when_needed(clazz) do
@@ -154 +154 @@
-            index.index_models models
-          end
-clazz.aaf_configura
+defini
-          # create a unique directory name (needed for unit tests where 
-          # multiple rebuilds per second may occur)
@@ -164 +164 @@
-          
-          File.rename index.options[:path], new_version
-clazz.index_dir =
+ActsAsFerret::change_index_dir definition[:name],
-        end
-      end

trunk/plugin/acts_as_ferret/lib/index.rb

@@ -10 +10 @@
-      @logger = Logger.new("#{RAILS_ROOT}/log/ferret_index.log")
-      @logger.level = ActiveRecord::Base.logger.level
+      @index_name = aaf_configuration[:name]
-    end
-    

trunk/plugin/acts_as_ferret/lib/instance_methods.rb

@@ -152 +152 @@
-    end
-
-
-
-
+
+
-      if (dynamic_boost && boost_value = self.send(dynamic_boost))
-        field_data = Ferret::Field.new(field_data)

trunk/plugin/acts_as_ferret/lib/local_index.rb

@@ -46 +46 @@
-    end
-
-
-
-
-
-
-
+
+
+
+
-      logger.debug "rebuild index: #{models.inspect}"
-
-
-
-
+
+
+
+
+
+
+
-      index.logger = logger
-      index.index_models models
-
-
-
-
+
+
+
+
+
-    end
-
@@ -109 +111 @@
-      logger.debug "query: #{ferret_index.process_query query}" if logger.debug?
-      lazy_fields = determine_lazy_fields options
+      logger.debug "lazy_fields: #{lazy_fields}"
-
-      total_hits = index.search_each(query, options) do |hit, score|

trunk/plugin/acts_as_ferret/lib/multi_index.rb

@@ -68 +68 @@
-              @sub_readers = @model_classes.map { |clazz| 
-                begin
-clazz.aaf_configuration
+ActsAsFerret::index_definition(clazz)
-                rescue Exception
-#{clazz.aaf_configuration[:index_dir]
+reader on index for class #{clazz.inspect
-                end
-              }