when_exe 0.4.2 → 0.5.0

data/lib/when_exe/events.rb

@@ -0,0 +1,1888 @@
+# -*- coding: utf-8 -*-
+=begin
+    Copyright (C) 2015-2021 Takashi SUGA
+
+    You may use and/or modify this file according to the license
+    described in the LICENSE.txt file included in this archive.
+=end
+
+require 'when_exe/namespace'
+require 'when_exe/tmduration'
+
+module When
+
+  #
+  # 時間座標を持つイベント記録の管理
+  #
+  module Events
+
+    #
+    # イベント管理用範囲オブジェクト
+    #
+    class Range < ::Range
+
+      # 実数のための exclude_end? 判定用マージン
+      #
+      Delta = When::TM::Duration::SECOND / 4096.0 / When::TM::Duration::DAY
+
+      # 小さい方の境界
+      #
+      # @return [Object]
+      #
+      attr_reader :start
+
+      # 大きい方の境界
+      #
+      # @return [Object]
+      #
+      attr_reader :until
+
+      # オリジナル文字列
+      #
+      # @return [String]
+      #
+      attr_accessor :original
+      private :original=
+
+      # オブジェクトを When::Events::Range 型に変換する
+      #
+      # @param [Object] source 変換元のオブジェクト
+      #
+      # @return [When::Events::Range] 変換結果
+      #
+      def self.convert_from(source)
+        case
+        when source.kind_of?(self)             ; source
+        when source.respond_to?(:exclude_end?) ; new(source, source.last, source.exclude_end?)
+        when source.respond_to?(:succ)         ; new(source, source.succ, true)
+        else raise ArgumentError, "Can't convert #{source} to #{self}"
+        end
+      end
+
+      # 範囲の重なりの判断が複雑になるか？
+      #
+      # @return [Boolean] true - 複雑, false - 単純
+      #
+      def is_complex?
+        case @start
+        when Array, Enumerator ; true
+        else                   ; false
+        end
+      end
+
+      # 指定オブジェクトが範囲内か？
+      #
+      # @param [Object] target 判定するオブジェクト
+      #
+      # @return [Boolean] true - 範囲内のオブジェクトあり, false - 範囲内のオブジェクトなし
+      #
+      def include?(target)
+        !include(target).empty?
+      end
+
+      # 含む対象の抽出
+      #
+      # @param [Object] target 判定するオブジェクト
+      #
+      # @return [Array<Object>] 実際に包含した範囲オブジェクトの Array
+      #
+      def include(target)
+        return [] if (exclude_end? && @until <= target) || (!exclude_end? && @until < target)
+
+        case @start
+        when ::Range
+          return @start.include?(target) ? [@start] : []
+
+        when Array
+          list = []
+          @start.each do |range|
+            break if _range_exceeded?(range, target)
+            list << range if _target_included?(range, target)
+          end
+          return list
+
+        when Enumerator
+          begin
+            list = []
+            while (range = @start.succ)
+              break if _range_exceeded?(range, target)
+              list << range if _target_included?(range, target)
+            end
+            return list
+          ensure
+            @start._rewind
+          end
+ 
+        else
+          return @start <= target ? [@start] : []
+        end
+      end
+
+      #
+      # 範囲の重なりの判断
+      #
+      # @param [Range] range 確認対象の単純範囲オブジェクト
+      #
+      # @return [Boolean] true - 重なる, false - 重ならない
+      #
+      def is_overlaped?(range)
+        !overlaped(range).empty?
+      end
+
+      #
+      # 範囲が重なる対象の抽出
+      #
+      # @param [Range] target 確認対象の単純範囲オブジェクト
+      #
+      # @return [Array<Object>] 実際に重なった範囲オブジェクトの Array
+      #
+      def overlaped(target)
+        first = target.first
+        last  = target.last
+        last  = last.respond_to?(:prev) ? last.prev : last - Delta if target.exclude_end?
+        return [] if (exclude_end? && @until <= first) || (!exclude_end? && @until < first)
+
+        case @start
+        when ::Range
+          return _target_exceeded?(@start, first) || _range_exceeded?(@start, last) ? [] : [@start]
+
+        when Array
+          list = []
+          @start.each do |range|
+            break if _range_exceeded?(range, last)
+            list << range unless _target_exceeded?(range, first)
+          end
+          return list
+
+        when Enumerator
+          begin
+            list = []
+            while (range = @start.succ)
+              break if _range_exceeded?(range, last)
+              list << range unless _target_exceeded?(range, first)
+            end
+            return list
+          ensure
+            @start._rewind
+          end
+ 
+        else
+          return @start <= last ? [@start] : []
+        end
+      end
+
+      #
+      # イベント管理用範囲オブジェクトの生成
+      #
+      # @param [Object] first 小さい方の境界
+      # @param [Object] last 大きい方の境界
+      # @param [Boolean] exclude_end 大きい方の境界を含まないか？
+      #
+      def initialize(first, last, exclude_end=false)
+        @start = first
+        @until = last
+        range  = [first, last].map {|edge| edge.respond_to?(:first) ? edge.first : edge}
+        rase ArgumentError, "#{range.last} is less than #{range.first}" if range.last < range.first
+        super(range.first, range.last, exclude_end)
+      end
+
+      private
+
+      # target が小さすぎるか？
+      def _range_exceeded?(range, target)
+        focused = range.respond_to?(:first) ? range.first : range
+        return true if  exclude_end? && @until <= focused
+        return true if !exclude_end? && @until <  focused
+        target < focused
+      end
+
+      # target が大きすぎるか？
+      def _target_exceeded?(range, target)
+        return false if target == (range.respond_to?(:first) ? range.first : range)
+        return target >= range.succ unless range.respond_to?(:last)
+        return target >= range.last if range.exclude_end?
+        return target >  range.last
+      end
+
+      # target が含まれるか？
+      def _target_included?(range, target)
+        range.respond_to?(:include?) ? range.include?(target) : (range == target)
+      end
+    end
+
+    if When.const_defined?(:Parts)
+
+      #
+      # 多言語対応データセット群
+      #
+      #   言語ごとに異なるデータセットを保持する
+      #
+      class DataSets
+
+        include When::Parts::Resource
+
+        # 名前
+        #
+        # @return [When::BasicTypes::M17n]
+        #
+        attr_reader :label
+
+        # 各言語用のデータセットの Hash
+        #
+        # @return [Hash<String=>When::Events::DataSet>]
+        #
+        attr_reader :datasets
+
+        # 単言語データセットオブジェクトの取得
+        #
+        # @param [String] language 言語コード
+        # @param [Numeric] limit オブジェクト生成待ち時間 / 秒
+        # @param [Boolean] limit false - 待たない, true - 無限に待つ
+        #
+        # @return [When::Events::DataSet]
+        #
+        def dataset(language='', limit=true)
+          if When.multi_thread
+            joined =
+              case limit
+              when Numeric   ; @thread.join(limit)
+              when nil,false ; true
+              else           ; @thread.join
+              end
+            return nil unless joined
+          end
+          return nil unless @datasets
+          When::Locale._hash_value(@datasets, language)
+        end
+
+        # 多言語対応 namespace の取得
+        #
+        # @param [String] prefix プレフィクス
+        # @param [String] language 言語コード
+        #
+        # @return [Array<String(namespace), String(説明)>]
+        #
+        def namespace(prefix, language)
+          When::Locale._hash_value(@prefixes[prefix], language)
+        end
+
+        #
+        # 指定の Event を主語とする Statement からなる RDF:Repository を生成する
+        #
+        # @param [Array<When::Events::Event>] events 登録する Event の Array
+        #
+        # @return [Hash<String(GraphURI)=>RDF:Repository>] 生成した Repository の Hash
+        #
+        def repository(events=nil)
+          repositories = {}
+          @datasets.each_pair do |language, dataset|
+            repositories[language] = dataset.repository(events)
+          end
+          _merge_each_graph(repositories)
+        end
+
+        #
+        # 指定の URI を主語とする Statement からなる RDF:Repository を生成する
+        #
+        # @param [String] uri 主語の URI
+        # @param [Integer] uri 主語のイベントの通し番号
+        # @param [String] graph 検索対象のグラフ(ダミー)
+        #
+        # @return [Hash<String(GraphURI)=>RDF:Repository>] 生成した Repository の Hash
+        #
+        def event(uri, graph=nil)
+          repositories = {}
+          @datasets.each_pair do |language, dataset|
+            repositories[language] = dataset.event(uri, graph)
+          end
+          _merge_each_graph(repositories)
+        end
+
+        #
+        # RDF::URI リソースで使用された prefix - namespace 対
+        #
+        # @return [Hash<String(prefix)=>String(namespace)>]
+        #
+        def used_ns
+          pair = {}
+          @datasets.each_value do |dataset|
+            pair.update(dataset.used_ns)
+          end
+          pair
+        end
+
+        #
+        # graph ごとに repository を merge する
+        #
+        def _merge_each_graph(repositories)
+          merged = {}
+          repositories.keys.inject([]) {|graphs,language| (graphs | repositories[language].keys) }.each do |graph|
+            repository_for_graph = ::RDF::Repository.new
+            repositories.each_value do |repository|
+              repository_for_graph << repository[graph] if repository.key?(graph)
+            end
+            merged[graph] = repository_for_graph
+          end
+          merged
+        end
+        private :_merge_each_graph
+
+        # 多言語対応データセットオブジェクトの生成
+        #
+        # @param [String] uri データセット定義の所在
+        # @param [Array<Array<String>>] rows 多言語対応データセット定義
+        # @param [Block] block キャッシュされたファイルパスの読み替え処理
+        #
+        def initialize(uri, rows, &block)
+
+          # 定義行を言語ごとに仕分けする
+          definitions = {}
+          @prefixes   = {}
+          labels   = nil
+          datasets = nil
+          (0...rows.size).to_a.reverse.each do |index|
+            items = rows[index].map {|item| item.strip}
+            prefix, namespace, word = items
+            if /\A(.+?):@\*\z/ =~ prefix
+              rows[index..index] = prefix_wildcard($1, namespace, word)
+            else
+              rows[index] = items
+            end
+          end
+          rows.each do |row|
+            next if row.first =~ /\A\s*#/
+            name, language= row.shift.split('@', 2)
+            language    ||= ''
+            language.sub!('_', '-')
+            if definitions.key?(name)
+              if language == ''
+                definitions[name].values.each do |definition|
+                  definition << row
+                end
+              else
+                definitions[name][language] << row
+              end
+            else
+              definitions[name] = Hash.new {|hash,key| hash[key]=[]}.merge({''=>[row]})
+              definitions[name][language] << row unless language == ''
+            end
+            case extract(name)
+            when LABEL       ; labels        = filter_for_dataset(definitions[name])
+            when REFERENCE   ; datasets      = filter_for_dataset(definitions[name])
+            when /\A(.+?):\z/; @prefixes[$1] = filter_for_dataset(definitions[name], true)
+            end
+          end
+
+          # 名前オブジェクトを生成する
+          @label = When::BasicTypes::M17n.new({
+            'label'=>labels[''].first,
+            'names'=>Hash[*(labels.keys.map   {|label|   [label,   labels[label].first    ]}).flatten],
+            'link' =>Hash[*(datasets.keys.map {|dataset| [dataset, datasets[dataset].first]}).flatten]
+          })
+          @_pool = {'..'=>uri, '.'=>rows, @label.to_s=>@label}
+          @child             = [@label]
+          @label._pool['..'] = self
+
+          # 各言語用のデータセットオブジェクトを生成する
+          if When.multi_thread
+            @thread = Thread.new do
+              begin
+                @datasets = create_datasets(datasets, uri, definitions, &block)
+              rescue => exception
+                puts exception
+              # puts exception.backtrace
+                raise exception
+              end
+            end
+          else
+            @datasets = create_datasets(datasets, uri, definitions, &block)
+          end
+        end
+
+        private
+
+        # 言語ごとのデータセットを登録する
+        def create_datasets(datasets, uri, definitions, &block)
+          Hash[*(datasets.keys.map {|language|
+            lines = []
+            definitions.each_pair do |name, definition|
+              lang = When::Locale._hash_key(definition, language)
+              definition[lang].each do |defs|
+                lines << [[name] + defs, lang]
+              end
+            end
+            [language, DataSet.new(lines, language, uri, self, &block)]
+          }).flatten]
+        end
+
+        # 名前空間のワイルドカードを展開する
+        def prefix_wildcard(prefix, namespace, word)
+          hash = When.Wikipedia(word).to_h
+          hash[:names].keys.map {|loc|
+            next nil unless /./ =~ loc
+            ["#{prefix}_#{loc}:@#{loc}", namespace.sub('*',loc), hash[:names][loc]]
+          }.compact
+        end
+
+        # データセット用の設定を選択する
+        def filter_for_dataset(target, filter=false)
+          hash = {}
+          target.each_pair do |language, definitions|
+            definitions.each do |definition|
+              if filter || DataSet.for_dataset?(definition.first)
+                hash[language] = definition
+                break
+              end
+            end
+          end
+          hash
+        end
+
+        # プレフィクスを名前空間に展開する
+        def extract(predicate)
+          return predicate unless /\A\{(.+?):(.+)\}\z/ =~ predicate && @prefixes.key?($1)
+          prefix, body = $~[1..2]
+          namespace = @prefixes[prefix].values.first.first
+          namespace = namespace.sub(/([a-z0-9])\z/i, '\1#')
+          namespace + body
+        end
+      end
+
+      # When.exe Time Schema
+      TS = When::Parts::Resource.base_uri.sub(/When\/$/, 'ts#')
+
+    else
+
+      # When.exe Time Schema - TS may not equal When::Resource.base_uri.sub(/When\/$/, 'ts#')
+      TS = 'http://hosi.org/ts#'
+
+    end
+
+    # Time Schema
+    TT          = TS.sub(/#\z/,'/')
+    RANGE       = TT + Range.class.to_s.gsub('::','/')
+    ID          = TS + 'id'
+    IRI         = TS + 'IRI'
+    REFERENCE   = TS + 'reference'
+    GROUP       = TS + 'group'
+    WHAT_DAY    = TS + 'whatDay'
+    START       = TS + 'start'
+    UNTIL       = TS + 'until'
+    WEST        = TS + 'west'
+    EAST        = TS + 'east'
+    SOUTH       = TS + 'south'
+    NORTH       = TS + 'north'
+    BOTTOM      = TS + 'bottom'
+    TOP         = TS + 'top'
+
+    # XML Schema Definition
+    INTEGER     = Namespace::XSD  + '#integer'
+    FLOAT       = Namespace::XSD  + '#float'
+    DOUBLE      = Namespace::XSD  + '#double'
+    STRING      = Namespace::XSD  + '#string'
+    DATE        = Namespace::XSD  + '#date'
+    TIME        = Namespace::XSD  + '#time'
+    DATETIME    = Namespace::XSD  + '#dateTime'
+
+    # Resource Description Framework
+    TYPE        = Namespace::RDF  + 'type'
+    SUBJECT     = Namespace::RDF  + 'subject'
+    LABEL       = Namespace::RDFS + 'label'
+    GRAPH       = Namespace::RDFC + 'section-rdf-graph'
+
+    # Dublin Core
+    SOURCE      = Namespace::DC   + 'source'
+    CONTRIBUTOR = Namespace::DC   + 'contributor'
+    LICENSE     = Namespace::DCQ  + 'license'
+    VALID       = Namespace::DCQ  + 'valid'
+    ABSTRACT    = Namespace::DCQ  + 'abstract'
+    HAS_PART    = Namespace::DCQ  + 'hasPart'
+    SPATIAL     = Namespace::DCQ  + 'spatial'
+    URI         = Namespace::DCQ  + 'URI'
+
+    # For Dataset
+    ForDataset  = [LABEL, REFERENCE, CONTRIBUTOR, LICENSE]
+
+    # Index
+    EqualIndex  = [SUBJECT, GRAPH, GROUP, CONTRIBUTOR, SPATIAL]
+    FirstEdge   = {'valid'=>START, 'longitude'=>WEST, 'latitude'=>SOUTH, 'altitude'=>BOTTOM}
+    LastEdge    = {'valid'=>UNTIL, 'longitude'=>EAST, 'latitude'=>NORTH, 'altitude'=>TOP   }
+
+    # 同値インデクスの判断
+    #
+    # @param [String] predicate 述語
+    #
+    # @return [Symbol] :equal 作成する, nil 作成しない
+    #
+    def self.equal(predicate)
+      EqualIndex.include?(predicate) ?
+        :equal :
+        nil
+    end
+
+    # 境界インデクスの判断
+    #
+    # @param [String] predicate 述語
+    #
+    # @return [Symbol] :first 下限境界, :last 上限境界, nil どちらでもなし
+    #
+    def self.edge(predicate)
+      FirstEdge.values.include?(predicate) ?
+        :first :
+      LastEdge.values.include?(predicate) ?
+        :last  :
+        nil
+    end
+
+    # 出現回数の限定
+    #
+    # @param [String] predicate 述語
+    #
+    # @return [Boolean] true 1回限定, false 限定なし
+    #
+    def self.cardinality1(predicate)
+      predicate != HAS_PART
+    end
+
+    #
+    # 一言語対応データセット
+    #
+    #   特定の言語用データセットを保持する
+    #
+    class DataSet
+
+      #
+      # オブジェクト操作定義
+      #
+
+      # 何もしない
+      nooperation  = proc {|event, obj|
+        obj
+      }
+
+      # 名前空間の展開
+      to_uri       = proc {|event, obj|
+        event.dataset.extract(obj)
+      }
+
+      # 整数化
+      to_i         = proc {|event, obj|
+        obj.to_i
+      }
+
+      # 実数化
+      to_f         = proc {|event, obj|
+        obj.to_f
+      }
+
+      # 下限整数化
+      first_to_i   = proc {|event, obj|
+        obj.respond_to?(:first) ?
+          obj.first.to_i :
+          obj.to_i
+      }
+
+      # 下限実数化
+      first_to_f   = proc {|event, obj|
+        obj.respond_to?(:first) ?
+          obj.first.to_f :
+          obj.to_f
+      }
+
+      # 上限整数化
+      last_to_i    = proc {|event, obj|
+        !obj.respond_to?(:last) ?
+          obj.succ.to_i - 1 :
+        obj.exclude_end? ?
+          obj.last.to_i - 1 :
+          obj.last.succ.to_i - 1
+      }
+
+      # 上限実数化
+      last_to_f    =  proc {|event, obj|
+        !obj.respond_to?(:last) ?
+          obj.to_f :
+        obj.exclude_end? ?
+          obj.last.to_f - Range::Delta :
+        obj.last.respond_to?(:succ) ?
+          obj.last.succ.to_f - Range::Delta :
+          obj.last.to_f
+      }
+
+      # 日時範囲化
+      to_range     = proc {|event, date|
+        target = When.when?(date, :parse=>When::Locale::EasternParser)
+        target = Range.convert_from(target) if target.kind_of?(Array) || target.kind_of?(Enumerator)
+        target.send(:original=, date) if target.kind_of?(Range)
+        target
+      }
+
+      # 日付化
+      to_date      = proc {|event, date|
+        Date.parse(date)
+      }
+
+      # 時刻化
+      to_time      = proc {|event, time|
+        Time.parse(time)
+      }
+
+      # 日時化
+      to_date_time = proc {|event, date_time|
+        DateTime.parse(date_time)
+      }
+
+      # 経度
+      to_long      = proc {|event, location|
+        location.kind_of?(Numeric) ? location :
+        When.where?(location).long
+      }
+
+      # 緯度
+      to_lat       = proc {|event, location|
+        location.kind_of?(Numeric) ? location :
+        When.where?(location).lat
+      }
+
+      # 高度
+      to_alt       = proc {|event, location|
+        location.kind_of?(Numeric) ? location :
+        When.where?(location).alt
+      }
+
+      #
+      # オブジェクト操作対応付け
+      #
+      Operations = Hash.new(nooperation).merge({
+        URI               => to_uri,
+        IRI               => to_uri,
+        INTEGER           => to_i,
+        FLOAT             => to_f,
+        DOUBLE            => to_f,
+        [:first, INTEGER] => first_to_i,
+        [:first, FLOAT  ] => first_to_f,
+        [:first, DOUBLE ] => first_to_f,
+        [:last,  INTEGER] => last_to_i,
+        [:last,  FLOAT  ] => last_to_f,
+        [:last,  DOUBLE ] => last_to_f,
+        RANGE             => to_range,
+        DATE              => to_date,
+        TIME              => to_time,
+        DATETIME          => to_date_time,
+
+        WEST              => to_long,
+        EAST              => to_long,
+        SOUTH             => to_lat,
+        NORTH             => to_lat,
+        BOTTOM            => to_alt,
+        TOP               => to_alt
+      })
+
+      #
+      # 自身の言語
+      #
+      # @return [String]
+      #
+      attr_reader :language
+
+      #
+      # 所属する多言語対応データセット
+      #
+      # @return [When::Events::DataSets]
+      #
+      attr_reader :parent
+
+      #
+      # 定義行の元情報
+      #
+      # @return [Array<String>]
+      #
+      attr_reader :definitions
+
+      #
+      # 時間座標を持つイベント
+      #
+      # @return [When::Events::Event]
+      #
+      attr_reader :events
+
+      #
+      # デフォルト Graph の URI
+      #
+      # @return [String(URI)]
+      #
+      attr_reader :default_graph
+
+      #
+      # 名前空間
+      #
+      # @return [Hash<String(prefix)=>String(namespace)>]
+      #
+      attr_reader :prefix
+
+      #
+      # 名前空間の説明
+      #
+      # @return [Array<Array<String(namespace), String(description)>>]
+      #
+      attr_reader :prefix_description
+
+      #
+      # ロール変数の定義
+      #
+      # @return [Hash<String(prefix:name)=>Hash>]
+      #
+      attr_reader :role
+
+      #
+      # RDF変数の定義
+      #
+      # @return [Hash<String(prefix:name)=>Hash>]
+      #
+      attr_reader :rdf
+
+      #
+      # CSV変数の定義
+      #
+      # @return [Hash<String(prefix:name)=>Hash>]
+      #
+      attr_reader :csv
+
+      #
+      # 一致キーのためのインデクス
+      #
+      # @return [Hash<String(prefix:name)=>Hash<String(key)=>Array>>]
+      #
+      attr_reader :index
+
+      #
+      # 順序キーのためのインデクス
+      #
+      # @return [Hash<String(prefix:name)=>Array>]
+      #
+      attr_reader :order
+
+      #
+      # データセットで使用しているRDF::URI リソース
+      #
+      # @return [Hash<String(iri)=>::RDF::URI>]
+      #
+      attr_reader :resource
+
+      #
+      # RDF::URI リソースで使用された prefix - namespace 対
+      #
+      # @return [Hash<String(prefix)=>String(namespace)>]
+      #
+      attr_reader :used_ns
+
+      #
+      # 指定の Event を主語とする Statement からなる RDF:Repository を生成する
+      #
+      # @param [Array<When::Events::Event>] events 登録する Event の Array
+      #
+      # @return [Hash<String(GraphURI)=>RDF:Repository>] 生成した Repository の Hash
+      #
+      def repository(events=nil)
+        if events
+          rep = Hash.new {|hash,key| hash[key] = ::RDF::Repository.new}
+          events.each do |event|
+            rep[''].insert(*event.statements)
+            rep[event.role[GRAPH]].insert(*event.statements) if @role.key?(GRAPH)
+          end
+          rep
+        else
+          @repository ||= repository(@events)
+        end
+      end
+
+      #
+      # 指定の URIまたはイベントの通し番号を主語とする Statement からなる RDF:Repository を生成する
+      #
+      # @param [String] uri 主語の URI
+      # @param [Integer] uri 主語のイベントの通し番号
+      # @param [String] graph 検索対象のグラフ(ダミー)
+      #
+      # @return [Hash<String(GraphURI)=>RDF:Repository>] 生成した Repository の Hash
+      #
+      def event(uri, graph=nil)
+        rep   = Hash.new {|hash,key| hash[key] = ::RDF::Repository.new}
+        list  = uri.kind_of?(String) ? @index[SUBJECT][uri] : [uri]
+        unless list.empty?
+          event = @events[list.first-1]
+          rep[''].insert(*event.statements)
+          rep[event.role[GRAPH]].insert(*event.statements) if @role.key?(GRAPH)
+        end
+        rep
+      end
+
+      #
+      # 外部 RDF を読み込んだ場合に動作を置き換える
+      #
+      module ExternalRepository
+        #
+        # 指定の Event を主語とする Statement からなる RDF:Repository を生成する
+        #
+        # @param [Array<When::Events::Event>] events 登録する Event の Array
+        #
+        # @return [Hash<String(GraphURI)=>RDF:Repository>] 生成した Repository の Hash
+        #
+        def repository(events=nil)
+          if events
+            rep = Hash.new {|hash,key| hash[key] = ::RDF::Repository.new}
+            events.each do |event|
+              @repository[''].query({:subject=>@resource[event.role[SUBJECT]]}) do |statement|
+                rep[''].insert(statement)
+                rep[event.role[GRAPH]].insert(statement) if @role.key?(GRAPH)
+              end
+            end
+            rep
+          else
+            @repository
+          end
+        end
+
+        #
+        # 指定の URI を主語とする Statement からなる RDF:Repository を生成する
+        #
+        # @param [String] uri 主語の URI
+        # @param [String] graph 検索対象のグラフ(ダミー)
+        #
+        # @return [Hash<String(GraphURI)=>RDF:Repository>] 生成した Repository の Hash
+        #
+        def event(uri, graph=nil)
+          rep = Hash.new {|hash,key| hash[key] = ::RDF::Repository.new}
+          @repository[''].query({:subject=>@resource[uri]}) do |statement|
+            rep[''].insert(statement)
+            rep[event.role[GRAPH]].insert(statement) if @role.key?(GRAPH)
+          end
+          rep
+        end
+
+        private
+
+        # リポジトリを準備する
+        def initialize_repository(source)
+          @used_predicate = {}
+          @repository = Hash.new {|hash,key| hash[key] = ::RDF::Repository.new}
+          @repository.update({''=>::RDF::Repository.load(source)})
+          valid = @resource[@role[VALID] && /<(.+?)>/ =~ @role[VALID][:target] ?
+                    $1 : VALID]
+          @repository[''].query({:predicate=>valid}) do |statement|
+            yield(rdf_to_hash(statement.subject))
+          end
+          @events.each do |event|
+            statements = event.statements.reject {|statement|
+              predicate = statement.predicate.to_s
+              @used_predicate.key?(predicate) || @role[predicate][:copied]
+            }
+            @repository[''].insert(*statements)
+            @repository[event.role[GRAPH]].insert(*statements) if @role.key?(GRAPH)
+          end
+        end
+
+        # 主語 subject を指定してトリプルをハッシュ化する
+        def rdf_to_hash(subject)
+          hash = {SUBJECT=>subject.to_s}
+          @repository[''].query({:subject=>subject}) do |statement|
+            key   = statement.predicate.to_s
+            value = statement.object
+            value = value.to_s unless value.kind_of?(Numeric)
+            case hash[key]
+            when Array ; hash[key] << value
+            when nil   ; hash[key] = When::Events.cardinality1(key) ? value : [value]
+            else       ; hash[key] = [hash[key], value]
+            end
+            @used_predicate[key] ||= statement.predicate
+          end
+          hash
+        end
+      end
+
+      #
+      # 外部 SPARQL サーバーを利用する場合に動作を置き換える
+      #
+      module SparqlRepository
+
+        #
+        # 指定の URI を主語とする Statement からなる RDF:Repository を生成する
+        #
+        # @param [String] uri 主語の URI
+        # @param [String] graph 検索対象のグラフ
+        #
+        # @return [Hash<String(GraphURI)=>RDF:Repository>] 生成した Repository の Hash
+        #
+        def event(uri, graph=nil)
+          rep     = Hash.new {|hash,key| hash[key] = ::RDF::Repository.new}
+          subject = @resource[uri]
+
+          # 問い合わせ文字列を準備する
+          query_string  = "SELECT DISTINCT ?predicate ?object \n"
+          query_string << "FROM <#{graph}> \n" if /./ =~ graph
+          query_string << "WHERE { \n <#{uri}> ?predicate ?object . \n} \n"
+        # return query_string
+
+          # 問い合わせを実行する
+          client = ::SPARQL::Client.new(@endpoint)
+          client.query(query_string).each do |solution|
+            statement = ::RDF::Statement.new(subject, solution[:predicate], solution[:object])
+            rep[''].insert(statement)
+            rep[graph].insert(statement) if graph
+          end
+          rep
+        end
+
+        #
+        # 指定の条件を満たす Event の Array を返す
+        #
+        # @param [Hash] options 以下の通り
+        # @option options [String]             'date'        日付範囲
+        # @option options [String]             'contributor' 情報提供者
+        # @option options [String]             'graph'       グラフ
+        # @option options [String]             'group'       グループ
+        # @option options [String]             'location'    空間位置
+        # @option options [String]             'keyword'     キーワード
+        # @option options [String or Integer]  'limit'       最大取得数
+        # @option options [String or Integer]  'offset'      取得開始レコード
+        # @option options [Boolean]            'count'       イベント数のみの確認か?
+        #
+        # @return [Array<When::Events::Event>] Event の Array
+        #
+        def intersection_events(options)
+
+          # 問い合わせ文字列を準備する
+          rdf_keys   = Hash.new {|hash,key| hash[key]=[]}
+          role_keys  = @role.keys - [LABEL, REFERENCE, WHAT_DAY]
+          role_keys.each do |role_key|
+            @role[role_key][:target].scan(/<(.+?)>/) {rdf_keys[$1] << role_key}
+          end
+          keywords   = (/./ =~ options['keyword']) ? options['keyword'].split('*') : []
+          loc_edges  = @order.key?(WEST) || @order.key?(SOUTH) || @order.key?(BOTTOM)
+          triples    = []
+          predicates = {}
+          rdf_keys.each_pair do |rdf_key, role_key|
+            if role_key.include?(SUBJECT)
+              predicates['?s'] = rdf_key
+            elsif role_key.include?(HAS_PART)
+              keywords.each do |keyword|
+                triples << "?s <#{rdf_key}> \"#{keyword}\" . "
+              end
+            elsif role_key.include?(CONTRIBUTOR) && /\A[^!]/ =~ options['contributor']
+              triples << "?s <#{rdf_key}> \"#{options['contributor']}\" . "
+            elsif role_key.include?(GROUP) && /./ =~ options['group']
+              triples << "?s <#{rdf_key}> \"#{options['group']}\" . "
+            elsif role_key.include?(SPATIAL) && /./ =~ options['location'] && !loc_edges
+              triples << "?s <#{rdf_key}> \"#{options['location']}\" . "
+            else
+              object = "?o#{predicates.size}"
+              predicates[object] = rdf_key
+              triples << "?s <#{rdf_key}> #{object} . "
+            end
+          end
+          select  = options['count'] ?
+                      "SELECT DISTINCT (COUNT(*) AS ?count) " :
+                      "SELECT DISTINCT #{predicates.keys.join(' ')} "
+          filters = []
+          filters.concat(range_filter(When.date_or_era(options['date']), predicates, rdf_keys, 'valid', options['date'])) if /./ =~ options['date']
+          filters.concat(regex_filter(options['contributor'], predicates, rdf_keys, 'contributor'))
+          if /./ =~ options['location'] && loc_edges
+            location = When.where?(options['location'])
+            filters.concat(range_filter(location.long, predicates, rdf_keys, 'longitude')) if @order.key?(WEST)
+            filters.concat(range_filter(location.lat,  predicates, rdf_keys, 'latitude' )) if @order.key?(SOUTH)
+            filters.concat(range_filter(location.alt,  predicates, rdf_keys, 'altitude' )) if @order.key?(BOTTOM)
+          end
+          query_string  = "#{select}\n"
+          query_string << "FROM <#{options['graph']}> \n" if @role.key?(GRAPH) && /./ =~ options['graph']
+          query_string << "WHERE { \n #{triples.map {|t| t+"\n "}.join('')}"
+          query_string << " FILTER ( \n  #{filters.map {|f| ' '+f}.join(" && \n  ")}\n  ) \n" unless filters.empty?
+          query_string << "} \n"
+          query_string << "ORDER BY #{attr2var(@order.key?(START) ? 'start' : 'valid', predicates, rdf_keys)} \n" unless options['count']
+          query_string << "LIMIT  #{options['limit']} \n"  if options['limit' ].to_i > 1
+          query_string << "OFFSET #{options['offset']} \n" if options['offset'].to_i > 1
+          return query_string if options['debug']
+
+          # 問い合わせを実行する
+          client = ::SPARQL::Client.new(@endpoint)
+          if options['count']
+            # イベント数のみ確認する
+            client.query(query_string).each do |solution|
+              return solution[:count].to_i
+            end
+            return 0
+
+          else
+            # イベントの内容を取り出す
+            client.query(query_string).map {|solution|
+              event = Hash[*predicates.keys.map {|predicate|
+                        object = solution[predicate[1..-1].to_sym]
+                        [predicates[predicate],
+                         case object
+                         when ::RDF::Literal::Integer ; object.to_i
+                         when ::RDF::Literal::Numeric ; object.to_f
+                         else object.to_s
+                         end]
+                      }.flatten]
+              rdf_keys.each_pair do |rdf_key, role_key|
+                next if event.key?(rdf_key)
+                if role_key.include?(HAS_PART)
+                  event[rdf_key] = keywords
+                elsif role_key.include?(CONTRIBUTOR)
+                  event[rdf_key] = options['contributor']
+                elsif role_key.include?(GRAPH)
+                  event[rdf_key] = (/./ =~ options['graph'] ? options['graph'] : @default_graph)
+                elsif role_key.include?(SPATIAL)
+                  event[rdf_key] = options['location']
+                end
+              end
+              uri = event[predicates['?s']]
+              id  = uri[@default_graph.length..-1] if @default_graph && uri.index(@default_graph)
+              Event.new(self, id || uri, event)
+            }
+          end
+        end
+
+        private
+
+        # リポジトリを準備する
+        def initialize_repository(source)
+          @endpoint = source
+        end
+
+        def range_filter(target, predicates, rdf_keys, attr, altanative=nil)
+          first, last, exclude =
+            case @role[FirstEdge[attr]] && @role[FirstEdge[attr]][:type]
+            when INTEGER
+              range_to_args(target, :to_i)
+            when DOUBLE, FLOAT
+              range_to_args(target, :to_f)
+            else
+              case altanative
+              when /\A(.+?)(\.{2,3})(.+?)(\^{1,2}.+)?\z/
+                ["\"#{$1}#{$4}\"", "\"#{$3}#{$4}\"", $2=='...']
+              else
+                [altanative]
+              end
+            end
+          lower  = @order.key?(LastEdge[ attr]) ? LastEdge[ attr][/[a-z]+\z/i] :
+                   @order.key?(FirstEdge[attr]) ? FirstEdge[attr][/[a-z]+\z/i] : attr
+          upper  = @order.key?(FirstEdge[attr]) ? FirstEdge[attr][/[a-z]+\z/i] : attr
+          last ||= first
+          lower_var = attr2var(lower, predicates, rdf_keys)
+          upper_var = attr2var(upper, predicates, rdf_keys)
+          [lower,first] == [upper,last] ?
+            ["#{lower_var} = #{first}"] : ["#{lower_var} >= #{first}", "#{upper_var} #{exclude ? '<' : '<='} #{last}"]
+        end
+
+        def regex_filter(target, predicates, rdf_keys, attr='contributor')
+          return [] unless target && /\A!(.+)\z/ =~ target.strip
+          ["!regex(#{attr2var(attr, predicates, rdf_keys)}, \"^#{$1}$\") "]
+        end
+
+        def range_to_args(target, method)
+          if target.respond_to?(:first)
+            return [target.first.send(method), target.last.succ.send(method), true] if !target.exclude_end? && target.last.respond_to?(:succ)
+            return [target.first.send(method), target.last.send(method), target.exclude_end?]
+          elsif target.respond_to?(:succ)
+            return [target.send(method), target.succ.send(method), true]
+          else
+            return [target.send(method)]
+          end
+        end
+
+        def attr2var(attr, predicates, rdf_keys)
+          predicates.each_pair do |var, rdf_key|
+            rdf_keys[rdf_key].each do |role|
+              return var if role.index(attr)
+            end
+          end
+          nil
+        end
+      end
+
+      #
+      # 指定の条件を満たす Event の Array を返す
+      #
+      # @param [Hash] options 以下の通り
+      # @option options [String] 'date'        日付範囲
+      # @option options [String] 'contributor' 情報提供者
+      # @option options [String] 'graph'       グラフ
+      # @option options [String] 'group'       グループ
+      # @option options [String] 'location'    空間位置
+      # @option options [String] 'day'         今日は何の日用の日付(共通)
+      # @option options [String] 'lsday'       今日は何の日用の日付(エスニック)
+      # @option options [String] 'keyword'     キーワード
+      #
+      # @return [Array<When::Events::Event>] Event の Array
+      #
+      def intersection_events(options)
+        list = []
+
+        # 日付範囲
+        if @order.key?(START) && /./ =~ options['date']
+          range = Range.convert_from(When.date_or_era(options['date']))
+          list << (@order.key?(UNTIL) ?
+                     range_overlaped(START, UNTIL, range) :
+                     edge_included(START, range))
+        end
+
+        # キーワード
+        if @index.key?(HAS_PART) && /./ =~ options['keyword']
+          options['keyword'].split('*').each {|key| list << @index[HAS_PART][key]}
+        end
+
+        # 情報提供者
+        if @index.key?(CONTRIBUTOR) && /\A(!)?(.+)/ =~ options['contributor']
+          reverse, contributor = $1, $2
+          sublist = @index[CONTRIBUTOR][contributor]
+          if sublist
+            sublist = (@order[START] || (1..@events.size).to_a) - sublist if reverse
+            list << sublist
+          end
+        end
+
+        # グループ
+        if @index.key?(GROUP) && /./ =~ options['group']
+          list << @index[GROUP][options['group']]
+        end
+
+        # 空間位置
+        if /./ =~ options['location']
+          if @order.key?(WEST) || @order.key?(SOUTH) || @order.key?(BOTTOM)
+            range = When.where?(options['location'])
+            range = When::Coordinates::Spatial::Range.new(range,range) unless range.kind_of?(When::Coordinates::Spatial::Range)
+            [[WEST,   EAST,  :long],
+             [SOUTH,  NORTH, :lat ],
+             [BOTTOM, TOP,   :alt ]].each do |pattern|
+               first, last, method = pattern
+               if @order.key?(first)
+                 list << (@order.key?(last) ?
+                            range_overlaped(first, last, range.send(method), :to_f) :
+                            edge_included(first, range.send(method), :to_f))
+               end
+             end
+          elsif @index.key?(SPATIAL)
+            list << @index[SPATIAL][options['location']]
+          end
+        end
+
+        # グラフ
+        if @index.key?(GRAPH) && /./ =~ options['graph']
+          list << @index[GRAPH][options['graph']]
+        end
+
+        # 今日は何の日
+        if @index.key?(WHAT_DAY)
+          list << @index[WHAT_DAY][[true,  $1.to_i,$2.to_i]] if /\A(\d{1,2})[-\/]?(\d{1,2})\z/ =~ options['day']
+          list << @index[WHAT_DAY][[false, $1.to_i,$2.to_i]] if /\A(\d{1,2})[-\/]?(\d{1,2})\z/ =~ options['lsday']
+        end
+
+        # 共通集合
+        list.compact!
+        return list if list.empty?
+        intersection = list.shift
+        list.each {|s| intersection &= s}
+        events = intersection.map {|id| @events[id-1]}
+        return events unless range
+
+        # 個別の絞込み
+        narrowed_events = []
+        sort_required = false
+        events.each do |event|
+          event_range = event.role[VALID]
+          if event_range.kind_of?(Range) && event_range.is_complex?
+            event_range.overlaped(range).each do |focused_date|
+              sort_required = true
+              focused_event = event.deep_copy
+              focused_event.send(:date=, focused_date)
+              narrowed_events << focused_event
+            end
+          else
+            narrowed_events << event
+          end
+        end
+        narrowed_events.sort_by! {|event| event.date} if sort_required
+        narrowed_events
+      end
+
+      #
+      # 順序キーによる検索
+      #
+      # @param [String] edge 順序キー(ts:start, ts:until, ts:west, ts:east, ts:south, ts:north, ts:bottom, ts:top)
+      # @param [Range] range 絞り込む範囲
+      # @param [Symbol] method スカラー化メソッドの指定 :to_i - 整数変数, :to_f - 実数変数
+      #
+      # @return [Array<Integer>] イベントのIDの配列
+      #
+      def edge_included(edge, range, method=:to_i)
+        _edge_included(edge, Range.new(*range_args(range, method)))
+      end
+
+      #
+      # 範囲の重なり
+      #
+      # @param [String] first_edge 順序キー(ts:start, ts:west, ts:south, ts:bottom)
+      # @param [String] last_edge  順序キー(ts:until, ts:east, ts:north, ts:top)
+      # @param [Range] range 絞り込む範囲
+      # @param [Symbol] method スカラー化メソッドの指定 :to_i - 整数変数, :to_f - 実数変数
+      #
+      # @return [Array<Integer>] イベントのIDの配列
+      #
+      def range_overlaped(first_edge, last_edge, range, method=:to_i)
+        args  = range_args(range, method)
+        upper = ::Range.new(args[0], +Float::MAX/4, false)
+        lower = ::Range.new(-Float::MAX/4, args[1], args[2])
+        ([first_edge, last_edge].map {|edge| [@order[edge].first, @order[edge].last]}.flatten.uniq.select {|id|
+          lower.include?(@events[id-1].role[first_edge]) && upper.include?(@events[id-1].role[last_edge])
+        } | (
+          @order[last_edge][_start_edge(last_edge, upper, 0, @events.size-1)..-1] &
+          @order[first_edge][0..._ended_edge(first_edge, lower, 0, @events.size-1)]
+        )).sort_by {|id|
+          @events[id-1].role[first_edge]
+        }
+      end
+
+      #
+      # 範囲生成用引数の準備
+      #
+      def range_args(range, method)
+        range.exclude_end? ?
+          [range.first.send(method), range.last.send(method), true] :
+        range.last.respond_to?(:succ) ?
+          [range.first.send(method), range.last.succ.send(method), true] :
+          [range.first.send(method), range.last.send(method), false]
+      end
+      private :range_args
+
+      # プレフィクスを名前空間に展開する
+      #
+      # @param [String] predicate プレフィクス付きの URI
+      #
+      # @return [String] プレフィクスを名前空間に展開した URI
+      #
+      def extract(predicate)
+        return predicate unless /\A(.+?):(.+)\z/ =~ predicate && @prefix.key?($1)
+        prefix, body = $~[1..2]
+        namespace = @prefix[prefix].first
+        namespace = namespace.sub(/([a-z0-9])\z/i, '\1#')
+        namespace + body
+      end
+
+      #
+      # 一言語対応データセットを生成する
+      #
+      # @param [Array<Array<Array<String>, String>>] definitions 定義行の情報
+      # @param [String] language 言語コード
+      # @param [String] uri 定義の所在のルート(ts:referenceが相対位置の場合に使用)
+      # @param [When::Events::DataSets] parent 所属する多言語対応データセット
+      # @param [Block] block キャッシュされたファイルパスの読み替え処理
+      #
+      def initialize(definitions, language='', uri='', parent=nil, &block)
+        @language = language
+        @parent   = parent
+        @prefix   = {}
+        @role     = {}
+        @rdf      = {}
+        @csv      = {}
+        @l_to_i   = {}
+        @i_to_l   = {}
+        @role_for_dataset = {}
+        @definitions = definitions
+        @definitions.each do |definition, lang|
+          parameters = definition[1..2].map {|item| item.strip}
+          case definition.first
+          when /\A(.+):\z/
+            @prefix[$1] = parameters
+          when /\A\[(.+)\]\z/
+            key=extract($1)
+            @csv[key ] = operation(key, parameters, lang)
+          when /\A<(.+)>\z/
+            key=extract($1)
+            @rdf[key ] = operation(key, parameters, lang)
+          when /\A\{(.+)\}\z/
+            key=extract($1)
+            (DataSet.for_dataset?(parameters.first, key) ? @role_for_dataset : @role)[key] = operation(key, parameters, lang)
+          end
+        end
+        @prefix_description  = @prefix.values.reject {|value| value.size < 2}.sort_by {|value| -value.first.length}
+        @namespace_to_prefix = @prefix.invert.sort.reverse
+        @used_ns  = {}
+        @resource = Hash.new {|hash,key|
+          @namespace_to_prefix.each do |prefix|
+            index = key.index(prefix.first.first)
+            if index && index == 0
+              @used_ns[prefix.last] = prefix.first.first
+              break
+            end
+          end
+          hash[key] = ::RDF::URI.new(key)
+        }
+
+        @events = []
+        @index  = {}
+        (@role.keys + @rdf.keys + @csv.keys).each {|item| @index[item] = Hash.new {|hash,key| hash[key]=[]}}
+        target = @role_for_dataset[REFERENCE][:target]
+        unless target =~ /:/ # Relative path
+          path     = uri.split('/')
+          path[-1] = target
+          target   = path.join('/')
+        end
+        operation = @role_for_dataset[REFERENCE][:original]
+        source    = extract(target)
+        source    = yield(source) if block_given? && operation !~ /SPARQL|CalendarEra/i
+        raise IOError, target + ': not ready' unless source
+        for_each_record(source, operation) do |row|
+          begin
+            event = Event.new(self, @events.size+1, row)
+            @events << event
+  
+            @role.keys.each do |item|
+              case item
+              when LABEL, REFERENCE
+              when HAS_PART
+                if event.role[HAS_PART].kind_of?(Array)
+                  event.role[HAS_PART].each do |word|
+                    @index[HAS_PART][word] << @events.size
+                  end
+                else
+                  event.each_word do |word|
+                    @index[HAS_PART][word] << @events.size
+                  end
+                end
+              when WHAT_DAY
+                date = event.role[WHAT_DAY]
+                key = [date.class.to_s !~ /\AWhen/ ||
+                       date.frame.kind_of?(When::CalendarTypes::Christian),
+                       date.month * 1, date.day]
+                @index[WHAT_DAY][key] << @events.size
+              else
+                add_index(:role, item)
+              end
+            end
+  
+            [:rdf, :csv].each do |method|
+              send(method).keys.each do |item|
+                add_index(method, item)
+              end
+            end
+          rescue => exception
+            puts "#{source}##{@events.size+1}: #{exception}"
+          # puts exception.backtrace
+          end
+        end
+
+        @order = {}
+        [:role, :rdf, :csv].each do |method|
+          send(method).each_pair do |item, definition|
+            next unless definition[:index] == :order
+            @order[item] = (1..@events.size).to_a.sort_by {|id| @events[id-1].send(method)[item]}
+            @order[item].each_with_index do |id, index|
+              @events[id-1].order[item] = index + 1
+            end
+          end
+        end
+        @index.each_value do |hash|
+          hash.each_value do |value|
+            value.uniq!
+            value.sort_by! {|id| @events[id-1].order[START]} if @order.key?(START)
+          end
+        end
+
+        @default_graph   = $1 if @role.key?(SUBJECT) && /\A(.+)<(.+)>\z/ =~ extract(@role[SUBJECT][:target])
+        @default_graph ||= parent.iri if parent
+      end
+
+      private
+
+      # 定義行一行を解釈する
+      def operation(key, definition, lang='')
+        edge  = When::Events.edge(key)
+        if definition[1]
+          type = nil
+          definition[1].split(/\s+/).each do |predicate|
+            if /\A(.+?):(.*)\z/ =~ predicate && @prefix.key?($1)
+              body      = $2
+              namespace = @prefix[$1].first
+              namespace = namespace.sub(/([a-z0-9])\z/i, '\1#')
+              type      = namespace + body
+            end
+            break
+          end
+        end
+        type      ||= (When.const_defined?(:Parts) ? RANGE : ::Date) if key == VALID
+        operation   = if /\A\s*strptime\((.+)\)\s*\z/ =~ definition[1]
+                        fmt = $1
+                        proc {|date| When.strptime(date, fmt)}
+                      else
+                        Operations[edge ? [edge, type] : type]
+                      end
+        {:target    => definition[0].gsub(/<(.+?)>|\{(.+?)\}/) {|match|
+             match[1..-2] = extract($1 || $2)
+             match
+           },
+         :original  => definition[1],
+         :index     => When::Events.equal(key) || (edge ? :order : ''),
+         :type      => type,
+         :operation => operation,
+         :copied    => /\A<[^>]+>\z/ =~ definition[0] && operation.equal?(Operations.default),
+         :lang      => lang
+        }
+      end
+
+      # データセット用の定義かイベント用の定義かの判断
+      def self.for_dataset?(target, key=nil)
+        return false if key && !ForDataset.include?(key)
+        /\[.+?\]|<.+?>|\{.+?\}/ !~ target
+      end
+
+      # イベントを生成・登録
+      def for_each_record(source, operation='csv with header', &block)
+        case operation
+        when /(out|no)\s*header\s*(\((\d+)\))?/i
+          @limit = $3.to_i if $3
+          csv_row_vs_label([])
+          ::URI.send(:open, source,'r') do |io|
+            CSV.parse(io.read) do |row|
+              yield(row)
+              break if @limit && @events.size >= @limit
+            end
+          end
+        when /header\s*(\((\d+)\))?/i
+          @limit = $2.to_i if $2
+          ::URI.send(:open, source,'r') do |io|
+            CSV.parse(io.read) do |row|
+              if @row_vs_label
+                yield(row)
+              else
+                csv_row_vs_label(row)
+              end
+              break if @limit && @events.size >= @limit
+            end
+          end
+        when /SPARQL/i
+          extend SparqlRepository
+          initialize_repository(source, &block)
+        when /RDF/i
+          extend ExternalRepository
+          initialize_repository(source, &block)
+        when /CalendarEra/i
+          era = When.Resource(source)
+          era = era.child.first unless era.child.empty?
+          while (era)
+            yield({VALID     => When::Parts::GeometricComplex.new(era.first, era.last.indeterminated_position ?
+                                                                   When.today+When::P6W : era.last),
+                   LABEL     => era.label.translate(@language),
+                   REFERENCE => era.label.reference(@language)})
+            era = era.succ
+          end
+        when /\A\/(.+)\/(\((\d+)\))?\z/
+          @limit = $3.to_i if $3
+          rexp = Regexp.compile($1)
+          csv_row_vs_label([])
+          ::URI.send(:open, source, 'r') do |file|
+            file.read.gsub(/[\r\n]/,'').scan(rexp) do
+              yield((1...$~.size).to_a.map {|i| $~[i]})
+            end
+            break if @limit && @events.size >= @limit
+          end
+        else
+          raise ArgumentError, "Undefined operation: #{operation}"
+        end
+      end
+
+      # CSVのフィールド番号とフィールド名の対応を管理する
+      def csv_row_vs_label(row)
+        row.each_with_index do |label, index|
+          @l_to_i[label]     = index + 1
+          @i_to_l[index + 1] = label
+        end
+
+        [@rdf, @role, @csv].each do |element|
+          element.each_value do |values|
+            values[:target].scan(/\[(.+?)\]/) {verify_index_and_label($1)} if values[:target]
+          end
+        end
+        @csv.keys.each do |key|
+          verify_index_and_label(key)
+        end
+
+        [@rdf, @role, @csv].each do |element|
+          element.each_value do |values|
+            values[:target].gsub!(/\[(.+?)\]/) {label_to_index($1)} if values[:target]
+          end
+        end
+        csv_verified = {}
+        @csv.each_pair do |item, value|
+          next if value.empty?
+          key = label_to_index(item)
+          if @csv_varified.key?(key)
+            raise ArgumentError, "Duplicated index and label: #{item}" unless value == csv_verified[key]
+          else
+            csv_verified[key] = value
+          end
+        end
+        @csv = csv_verified
+
+        @i_to_l.each_pair do |index, label|
+          next unless /:/ =~ label
+          @rdf[label] ||= {:target    => "[#{index}]",
+                           :original  => '',
+                           :index     => '',
+                           :operation => Operations.default}
+        end
+
+        @row_vs_label = true
+      end
+
+      # CSVのフィールド名→フィールド番号
+      def label_to_index(item)
+        case item
+        when /\A\d+\z/
+          "[#{item.to_i}]"
+        when /\A[^:]+\z/, /\A[^\d:].*?:/
+          raise ArgumentError, "Label undefined: #{item}" unless @l_to_i.key?(item)
+          "[#{@l_to_i[item]}]"
+        else
+          index = verify_index_and_label(item)
+          raise ArgumentError, "Irregal index format: #{item}" unless index
+          "[#{index}]"
+        end
+      end
+
+      # CSVのフィールド番号とフィールド名を照合する
+      def verify_index_and_label(item)
+        return nil unless /\A(\d+?):(.+)\z/ =~ item
+        index ,label = $~[1..2]
+        index = index.to_i
+        label = extact(label)
+        if @i_to_l.key?(index) && @i_to_l[index] != label ||
+           @l_to_i.key?(label) && @l_to_i[label] != index
+          raise ArgumentError, "Duplicated index and label: #{item}"
+        else
+           @i_to_l[index] = label
+           @l_to_i[label] = index
+        end
+        index
+      end
+
+      # ロールを一致インデクス管理対象として登録する
+      def add_index(method, item)
+        return unless send(method)[item][:index] == :equal
+        key = @events.last.send(method)[item]
+        return unless key
+        @index[item][key] << @events.size
+      end
+
+      # 二分検索 - メイン
+      def _edge_included(edge, range, from=0, to=@events.size-1)
+        middle = (from + to) / 2
+        if range.include?(@events[@order[edge][middle]-1].role[edge])
+          if range.first <= @events[@order[edge].first-1].role[edge]
+            start = 0
+          else
+            start = _start_edge(edge, range, 0, middle)
+          end
+          if range.exclude_end? && range.last >= @events[@order[edge].last-1].role[edge] ||
+            !range.exclude_end? && range.last >  @events[@order[edge].last-1].role[edge]
+            ended = @events.size
+          else
+            ended = _ended_edge(edge, range, middle, @events.size-1)
+          end
+          return (start...ended).to_a.map {|i| @order[edge][i]}
+        end
+        return [] if from == to
+        @events[@order[edge][middle]-1].role[edge] < range.first ?
+          _edge_included(edge, range, middle+1, to) :
+          _edge_included(edge, range, from,  middle)
+      end
+
+      # 二分検索 - 小さい方の境界
+      def _start_edge(edge, range, from, to)
+        return to if to - from <= 1
+        middle = (from + to) / 2
+        range.include?(@events[@order[edge][middle]-1].role[edge]) ?
+          _start_edge(edge, range, from,   middle) :
+          _start_edge(edge, range, middle, to)
+      end
+
+      # 二分検索 - 大きい方の境界
+      def _ended_edge(edge, range, from, to)
+        return to if to - from <= 1
+        middle = (from + to) / 2
+        range.include?(@events[@order[edge][middle]-1].role[edge]) ?
+          _ended_edge(edge, range, middle, to) :
+          _ended_edge(edge, range, from,   middle)
+      end
+    end
+
+    #
+    # 時間座標を持つイベント
+    #
+    class Event
+
+      #
+      # 所属する一言語対応データセット
+      #
+      # @return [When::Events::DataSet]
+      #
+      attr_reader :dataset
+
+      #
+      # 通し番号
+      #
+      # @return [Integer]
+      #
+      attr_reader :id
+
+      #
+      # CSVの一行分
+      #
+      # @return [Array<String>]
+      #
+      attr_reader :row
+
+      #
+      # ロール変数
+      #
+      # @return [Hash<String(prefix:name)=>Object>]
+      #
+      attr_accessor :role
+      protected :role=
+
+      #
+      # RDF変数
+      #
+      # @return [Hash<String(prefix:name)=>Object>]
+      #
+      attr_accessor :rdf
+      protected :rdf=
+
+      #
+      # CSV変数
+      #
+      # @return [Hash<String(番号)=>Object>]
+      #
+      attr_accessor :csv
+      protected :csv=
+
+      #
+      # 当該順序キーで何番目のイベントか
+      #
+      # @return [Hash<String(prefix:name)=>Array(Integer(id))>]
+      #
+      attr_reader :order
+
+      #
+      # deep copy 時の元イベント
+      #
+      # @return [When::Events::Event]
+      #
+      attr_accessor :original
+      protected :original=
+
+      #
+      # 日付または日付範囲
+      #
+      # @return [When::TM::TemporalPosition or When::Events::Range]
+      #
+      attr_accessor :date
+      private :date=
+
+      #
+      # HAS_PART対象の文字列中の{}で囲まれた語に対して yield で指定された処理を行う
+      #
+      def each_word
+        @role[HAS_PART].scan(/(\{+)(.*?)(\}+)/) do
+          bra, word, cket = $~[1..3]
+          next unless bra.length.odd? && cket.length.odd?
+          yield(word)
+        end
+      end
+
+      #
+      # 指定の情報を{}部分のマークアップ処理など行って整形して返す
+      #
+      # @param [String] item 返す情報の名称
+      # @param [Symbol] method 返す情報が属するグループ(:role, :rdf, :csv)
+      #
+      # @return [String]
+      #
+      # @note ブロックを渡された場合、そのブロックに{}部分のマークアップを依頼する
+      #
+      def abstract(item=ABSTRACT, method=:role)
+        send(method)[item].gsub(/(\{+)(.*?)(\}+)/) {
+          bra, word, cket = $~[1..3]
+          '{'*(bra.length/2) + (block_given? ? yield(word) : word) + '}'*(cket.length/2)
+        }
+      end
+
+      #
+      # 指定の情報のIRIとその説明を返す
+      #
+      # @param [String] item 返す情報の名称
+      # @param [Symbol] method 返す情報が属するグループ(:role, :rdf, :csv)
+      #
+      # @return [Array<String(IRI), String(説明)>]
+      #
+      def source(item=SOURCE, method=:role)
+        iri = send(method)[item]
+        return [nil, iri] unless /:\/\// =~ iri
+        @dataset.prefix_description.each do |description|
+          index = iri.index(description[0])
+          return [iri, description[1]] if index && index == 0
+        end
+        [iri]
+      end
+
+      #
+      # イベントが属するグループ(@role[When::Events::GROUP])を返す
+      #
+      # @return [String]
+      #
+      def group
+        @role[GROUP]
+      end
+
+      #
+      # 自身を主語とする RDF::Statement の Array を返す
+      #
+      # @return [Array<RDF::Statement>]
+      #
+      def statements
+        unless @statements
+          @statements = []
+          raise ArgumentError, 'Role for rdf:subject not defined' unless @role.key?(SUBJECT)
+          subject = @dataset.resource[@role[SUBJECT]]
+          @role.each_pair do |predicate, object|
+            case predicate
+            when SUBJECT, ID, GRAPH, WHAT_DAY
+              # Do nothing
+            when HAS_PART
+              if @role[HAS_PART].kind_of?(Array)
+                words = @role[HAS_PART]
+              else
+                words = []
+                each_word do |word|
+                  words << word
+                end
+                words.uniq!
+              end
+              words.each do |word|
+                @statements << ::RDF::Statement(subject, @dataset.resource[predicate], word)
+              end
+            when VALID
+              if object.kind_of?(Range)
+                object = object.original
+              elsif object.respond_to?(:to_uri)
+                object = object.to_uri
+              end
+              @statements << ::RDF::Statement(subject, @dataset.resource[predicate], object)
+          # when ABSTRACT
+          #   @statements << ::RDF::Statement(subject, @dataset.resource[predicate], abstract)
+            else
+              if [URI, IRI].include?(@dataset.role[predicate][:type])
+                object = @dataset.resource[object]
+              elsif object.respond_to?(:to_uri)
+                object = object.to_uri
+              elsif @dataset.role[predicate][:lang] != '' && object.kind_of?(String)
+                object = ::RDF::Literal.new(object, {:language=>@dataset.role[predicate][:lang]})
+              end
+              @statements << ::RDF::Statement(subject, @dataset.resource[predicate], object)
+            end
+          end
+        end
+        @statements
+      end
+
+      #
+      # イベントの複製
+      #
+      # @return [When::Events::Event] コピー結果
+      #
+      def deep_copy
+        result          = self.dup
+        result.csv      = @csv.dup
+        result.rdf      = @rdf.dup
+        result.role     = @role.dup
+        result.original = self
+        result
+      end
+
+      #
+      # イベントの生成
+      #
+      # @param [When::Events::DataSet] dataset 所属する一言語対応データセット
+      # @param [Integer] id 通し番号
+      # @param [Array<String>] row CSVの一行分
+      # @param [Hash<String=>Object>] row RDFの1クエリ分
+      #
+      def initialize(dataset, id, row)
+        @dataset = dataset
+        @id      = id
+        @row     = row
+        @role    = {}
+        @rdf     = {}
+        @csv     = {}
+        @order   = {}
+
+        case row
+        when Hash
+          @rdf = row
+        when Array
+          row.each_with_index do |item, index|
+            @csv[(index+1).to_s] = item
+          end
+
+          dataset.rdf.each_pair do |item, definition|
+            fields = []
+            format = definition[:target].gsub(/\[(\d+)\]/) {
+              field  =  row[$1.to_i-1]
+              field  =  escape(field, definition[:type])
+              fields << field
+              '%s'
+            }
+            @rdf[item] = definition[:operation].call(self, format == '%s' ? fields.first : format % fields)
+          end
+        end
+
+        @role[ID] = @rdf[ID] = id
+        dataset.role.each_pair do |item, definition|
+          fields = []
+          format = definition[:target].gsub('%','%%').gsub(/\[(\d+)\]|<(.+?)>|\{(.+?)\}/) {|match|
+            key  = $1 || $2 || $3
+            field = case match
+                    when /\A\[/ ; row[key.to_i-1]
+                    when /\A</  ; @rdf[key]
+                    when /\A\{/ ; @role[key]
+                    end
+            field  =  escape(field, definition[:type])
+            fields << field
+            '%s'
+          }
+          pre_operation = DataSet::Operations[item]
+          @role[item] = 
+            if pre_operation
+              definition[:operation].call(self, pre_operation.call(self, format == '%s' ? fields.first : format % fields))
+            else
+              definition[:operation].call(self, format == '%s' ? fields.first : format % fields)
+            end
+        end
+
+        @date = @role[VALID]
+      end
+
+      private
+
+      # URI 仕様外の文字をエスケープする
+      def escape(part, type)
+        return part unless part.kind_of?(String)
+        part = part.strip
+        return part if /:\/\/|%/ =~ part
+        case type
+        when URI; CGI.escape(part.gsub(' ', '_')).gsub(/%2F/i,'/')
+        when IRI; part.gsub(' ', '_')
+        else part
+        end
+      end
+    end
+  end
+end