meteor 0.9.12 → 0.9.14

data/lib/meteor/element_factory.rb

@@ -0,0 +1,68 @@
+# -* coding: UTF-8 -*-
+# frozen_string_literal: true
+
+module Meteor
+  #
+  # Element Factory Class (要素ファクトリ クラス)
+  #
+  class ElementFactory
+    @@pf = Meteor::ParserFactory.new
+
+    #
+    # set options (オプションをセットする)
+    # @param [Hash] opts option (オプション)
+    # @option opts [String] :root root directory (基準ディレクトリ)
+    # @option @deprecated opts [String] :base_dir root directory (基準ディレクトリ)
+    # @option opts [String] :enc default character encoding (デフォルト文字エンコーディング)
+    # @option @deprecated opts [String] :base_enc default character encoding (デフォルト文字エンコーディング)
+    # @option opts [FixNum,Symbol] :type default type of parser (デフォルトのパーサ・タイプ)
+    # @option @deprecated opts [FixNum | Symbol] :base_type default type of parser (デフォルトのパーサ・タイプ)
+    #
+    def self.options=(opts)
+      @@pf.options = opts
+    end
+
+    #
+    #@overload link(type,relative_path,enc)
+    # generate parser (パーサを作成する)
+    # @param [Fixnum] type type of parser (パーサ・タイプ)
+    # @param [String] relative_path relative file path (相対ファイルパス)
+    # @param [String] enc character encoding (エンコーディング)
+    # @return [Meteor::Parser] parser (パーサ)
+    #@overload link(type,relative_path)
+    # generate parser (パーサを作成する)
+    # @param [Fixnum] type type of parser (パーサ・タイプ)
+    # @param [String] relative_path relative file path (相対ファイルパス)
+    # @return [Meteor::Parser] parser (パーサ)
+    #
+    def self.link(*args)
+      @@pf.link(*args)
+    end
+
+    #
+    # @overload link_str(type, relative_url, doc)
+    #  generate parser (パーサを作成する)
+    #  @param [Fixnum] type type of parser (パーサ・タイプ)
+    #  @param [String] relative_url relative URL (相対URL)
+    #  @param [String] doc document (ドキュメント)
+    #  @return [Meteor::Parser] parser (パーサ)
+    # @overload link_str(relative_url, doc)
+    #  generate parser (パーサを作成する)
+    #  @param [String] relative_url relative URL (相対URL)
+    #  @param [String] doc document (ドキュメント)
+    #  @return [Meteor::Parser] parser (パーサ)
+    #
+    def self.link_str(*args)
+      @@pf.link_str(args)
+    end
+
+    #
+    # get root element (ルート要素を取得する)
+    # @param [String,Symbol] key identifier (キー)
+    # @return [Meteor::RootElement] root element (ルート要素)
+    #
+    def self.element(key)
+      @@pf.element(key)
+    end
+  end
+end

data/lib/meteor/exception/no_such_element_exception.rb

@@ -0,0 +1,84 @@
+# -* coding: UTF-8 -*-
+# frozen_string_literal: true
+
+module Meteor
+  module Exception
+    #
+    # Element Search Exception (要素検索例外)
+    #
+    # @!attribute [rw] message
+    #  @return [String] message (メッセージ)
+    #
+    class NoSuchElementException
+      attr_accessor :message
+
+      #
+      # initializer (イニシャライザ)
+      # @overload initialize(name)
+      #  @param [String,Symbol] name tag name (タグ名)
+      # @overload initialize(attr_name,attr_value)
+      #  @param [String,Symbol] attr_name attribute name (属性名)
+      #  @param [String] attr_value attribute value (属性値)
+      # @overload initialize(name,attr_name,attr_value)
+      #  @param [String,Symbol] name tag name (タグ名)
+      #  @param [String,Symbol] attr_name attribute name (属性名)
+      #  @param [String] attr_value attribute value (属性値)
+      # @overload initialize(attr_name1,attr_value1,attr_name2,attr_value2)
+      #  @param [String,Symbol] attr_name1 attribute name1 (属性名１)
+      #  @param [String] attr_value1 attribute value1 (属性値１)
+      #  @param [String,Symbol] attr_name2 attribute name2 (属性名２)
+      #  @param [String] attr_value2 attribute value2 (属性値２)
+      # @overload initialize(name,attr_name1,attr_value1,attr_name2,attr_value2)
+      #  @param [String,Symbol] name tag name (タグ名)
+      #  @param [String,Symbol] attr_name1 attribute name1 (属性名１)
+      #  @param [String] attr_value1 attribute value1 (属性値１)
+      #  @param [String,Symbol] attr_name2 attribute name2 (属性名２)
+      #  @param [String] attr_value2 attribute value2 (属性値２)
+      #
+      def initialize(*args)
+        case args.length
+        when ONE
+          initialize_1(args[0])
+        when TWO
+          initialize_2(args[0], args[1])
+        when THREE
+          initialize_3(args[0], args[1], args[2])
+        when FOUR
+          initialize_4(args[0], args[1], args[2], args[3])
+        when FIVE
+          initialize_5(args[0], args[1], args[2], args[3], args[4])
+        end
+      end
+
+      def initialize_1(name)
+        self.message = "element not found : #{name}"
+      end
+
+      private :initialize_1
+
+      def initialize_2(attr_name, attr_value)
+        self.message = "element not found : [#{attr_name}=#{attr_value}]"
+      end
+
+      private :initialize_2
+
+      def initialize_3(name, attr_name, attr_value)
+        self.message = "element not found : #{name}[#{attr_name}=#{attr_value}]"
+      end
+
+      private :initialize_3
+
+      def initialize_4(attr_name1, attr_value1, attr_name2, attr_value2)
+        self.message = "element not found : [#{attr_name1}=#{attr_value1}][#{attr_name2}=#{attr_value2}]"
+      end
+
+      private :initialize_4
+
+      def initialize_5(name, attr_name1, attr_value1, attr_name2, attr_value2)
+        self.message = "element not found : #{name}[#{attr_name1}=#{attr_value1}][#{attr_name2}=#{attr_value2}]"
+      end
+
+      private :initialize_5
+    end
+  end
+end

data/lib/meteor/ml/html/parser_impl.rb

@@ -0,0 +1,167 @@
+# -* coding: UTF-8 -*-
+# frozen_string_literal: true
+
+module Meteor
+  module Ml
+    module Html
+      #
+      # HTML parser (HTMLパーサ)
+      #
+      class ParserImpl < Meteor::Ml::Html4::ParserImpl
+        #[Array] void elements (空要素)
+        MATCH_TAG = ["br", "hr", "img", "input", "meta", "base", "embed", "command", "keygen"]
+
+        #[Array] non-nestable elements (入れ子にできない要素)
+        MATCH_TAG_SNG = [
+          "texarea",
+          "select",
+          "option",
+          "form",
+          "fieldset",
+          "figure",
+          "figcaption",
+          "video",
+          "audio",
+          "progress",
+          "meter",
+          "time",
+          "ruby",
+          "rt",
+          "rp",
+          "datalist",
+          "output"
+        ]
+
+        #[Array] boolean attributes (論理値で指定する属性)
+        ATTR_LOGIC = ["disabled", "readonly", "checked", "selected", "multiple", "required"]
+
+        #[Array] elements with the disabled attribute (disabled属性のある要素)
+        DISABLE_ELEMENT = ["input", "textarea", "select", "optgroup", "fieldset"]
+
+        #[Array] elements with the required attribute (required属性のある要素)
+        REQUIRE_ELEMENT = ["input", "textarea"]
+
+        REQUIRED_M = "\\srequired\\s|\\srequired$|\\sREQUIRED\\s|\\sREQUIRED$"
+        # REQUIRED_M = [' required ',' required',' REQUIRED ',' REQUIRED']
+        REQUIRED_R = "required\\s|required$|REQUIRED\\s|REQUIRED$"
+
+        @@pattern_required_m = Regexp.new(REQUIRED_M)
+        @@pattern_required_r = Regexp.new(REQUIRED_R)
+
+        #
+        # initializer (イニシャライザ)
+        # @overload initialize
+        # @overload initialize(ps)
+        #  @param [Meteor::Parser] ps paser (パーサ)
+        #
+        def initialize(*args)
+          super()
+          @@match_tag = MATCH_TAG
+          @@match_tag_sng = MATCH_TAG_SNG
+          @@attr_logic = ATTR_LOGIC
+          @doc_type = Parser::HTML
+          case args.length
+          when ZERO
+            # initialize_0
+          when ONE
+            initialize_1(args[0])
+          else
+            raise ArgumentError
+          end
+        end
+
+        #
+        # initializer (イニシャライザ)
+        #
+        # def initialize_0
+        # end
+        #
+        # private :initialize_0
+
+        #
+        # initializer (イニシャライザ)
+        # @param [Meteor::Parser] ps parser (パーサ)
+        #
+        def initialize_1(ps)
+          @root.document = String.new(ps.document)
+          self.document_hook = String.new(ps.document_hook)
+          @root.content_type = String.new(ps.root_element.content_type)
+          @root.charset = ps.root_element.charset
+          @root.kaigyo_code = ps.root_element.kaigyo_code
+        end
+
+        private :initialize_1
+
+        #
+        # analyze document , set content type (ドキュメントをパースし、コンテントタイプをセットする)
+        #
+        def analyze_content_type
+          @error_check = false
+
+          element_3("meta", "charset", "[a-zA-Z-]+", false)
+
+          if !@elm_
+            element_3("meta", "charset", "[a-zA-Z-]+", false)
+          end
+
+          @error_check = true
+
+          if @elm_
+            @root.charset = @elm_.attr("charset")
+            if !@root.charset
+              @root.charset = "utf-8"
+            end
+          else
+            @root.charset = "utf-8"
+          end
+
+          @root.content_type = "text/html"
+        end
+
+        private :analyze_content_type
+
+        def edit_attrs_(elm, attr_name, attr_value)
+          if is_match("selected", attr_name) && is_match("option", elm.name)
+            edit_attrs_5(elm, attr_name, attr_value, @@pattern_selected_m, @@pattern_selected_r)
+          elsif is_match("multiple", attr_name) && is_match("select", elm.name)
+            edit_attrs_5(elm, attr_name, attr_value, @@pattern_multiple_m, @@pattern_multiple_r)
+          elsif is_match("disabled", attr_name) && is_match(DISABLE_ELEMENT, elm.name)
+            edit_attrs_5(elm, attr_name, attr_value, @@pattern_disabled_m, @@pattern_disabled_r)
+          elsif is_match("checked", attr_name) && is_match("input", elm.name) && is_match("radio", get_type(elm))
+            edit_attrs_5(elm, attr_name, attr_value, @@pattern_checked_m, @@pattern_checked_r)
+          elsif is_match("readonly", attr_name) &&
+              (is_match("textarea", elm.name) || (is_match("input", elm.name) && is_match(READONLY_TYPE, get_type(elm))))
+            edit_attrs_5(elm, attr_name, attr_value, @@pattern_readonly_m, @@pattern_readonly_r)
+          elsif is_match("required", attr_name) && is_match(REQUIRE_ELEMENT, elm.name)
+            edit_attrs_5(elm, attr_name, attr_value, @@pattern_required_m, @@pattern_required_r)
+          else
+            super(elm, attr_name, attr_value)
+          end
+        end
+
+        private :edit_attrs_
+
+        def get_attr_value_(elm, attr_name)
+          if is_match("selected", attr_name) && is_match("option", elm.name)
+            get_attr_value_r(elm, @@pattern_selected_m)
+          elsif is_match("multiple", attr_name) && is_match("select", elm.name)
+            get_attr_value_r(elm, @@pattern_multiple_m)
+          elsif is_match("disabled", attr_name) && is_match(DISABLE_ELEMENT, elm.name)
+            get_attr_value_r(elm, @@pattern_disabled_m)
+          elsif is_match("checked", attr_name) && is_match("input", elm.name) && is_match("radio", get_type(elm))
+            get_attr_value_r(elm, @@pattern_checked_m)
+          elsif is_match("readonly", attr_name) &&
+              (is_match("textarea", elm.name) || (is_match("input", elm.name) && is_match(READONLY_TYPE, get_type(elm))))
+            get_attr_value_r(elm, @@pattern_readonly_m)
+          elsif is_match("required", attr_name) && is_match(REQUIRE_ELEMENT, elm.name)
+            get_attr_value_r(elm, @@pattern_required_m)
+          else
+            super(elm, attr_name)
+          end
+        end
+
+        private :get_attr_value_
+      end
+    end
+  end
+end