micromicro 0.1.0 → 1.0.0

data/lib/micro_micro/item.rb

@@ -1,7 +1,9 @@
 module MicroMicro
   class Item
-    attr_accessor :value
+    include Collectible
 
+    # Parse a node for microformats2-encoded data.
+    #
     # @param node [Nokogiri::XML::Element]
     def initialize(node)
       @node = node
@@ -11,11 +13,17 @@ module MicroMicro
       properties << implied_url if implied_url?
     end
 
+    # A collection of child items parsed from the node.
+    #
+    # @see http://microformats.org/wiki/microformats2-parsing#parse_an_element_for_class_microformats
+    #
     # @return [MicroMicro::Collections::ItemsCollection]
     def children
       @children ||= Collections::ItemsCollection.new(Item.items_from(node.element_children))
     end
 
+    # The value of the node's `id` attribute, if present.
+    #
     # @return [String, nil]
     def id
       @id ||= node['id']&.strip
@@ -26,12 +34,22 @@ module MicroMicro
       format(%(#<#{self.class.name}:%#0x types: #{types.inspect}, properties: #{properties.count}, children: #{children.count}>), object_id)
     end
 
+    # A collection of plain text properties parsed from the node.
+    #
+    # @return [MicroMicro::Collections::PropertiesCollection]
+    def plain_text_properties
+      @plain_text_properties ||= Collections::PropertiesCollection.new(properties.select { |property| property.prefix == 'p' })
+    end
+
+    # A collection of properties parsed from the node.
+    #
     # @return [MicroMicro::Collections::PropertiesCollection]
     def properties
       @properties ||= Collections::PropertiesCollection.new(Property.properties_from(node.element_children))
     end
 
-    # @see microformats2 Parsing Specification section 1.2
+    # Return the parsed item as a Hash.
+    #
     # @see http://microformats.org/wiki/microformats2-parsing#parse_an_element_for_class_microformats
     #
     # @return [Hash]
@@ -43,28 +61,42 @@ module MicroMicro
 
       hash[:id] = id if id.present?
       hash[:children] = children.to_a if children.any?
-      hash[:value] = value if value.present?
 
       hash
     end
 
+    # An array of root class names parsed from the node's `class` attribute.
+    #
     # @return [Array<String>]
     def types
       @types ||= self.class.types_from(node)
     end
 
+    # A collection of url properties parsed from the node.
+    #
+    # @return [MicroMicro::Collections::PropertiesCollection]
+    def url_properties
+      @url_properties ||= Collections::PropertiesCollection.new(properties.select { |property| property.prefix == 'u' })
+    end
+
+    # Does this node's `class` attribute contain root class names?
+    #
     # @param node [Nokogiri::XML::Element]
     # @return [Boolean]
     def self.item_node?(node)
       types_from(node).any?
     end
 
+    # Extract items from a context.
+    #
     # @param context [Nokogiri::HTML::Document, Nokogiri::XML::NodeSet, Nokogiri::XML::Element]
     # @return [Array<MicroMicro::Item>]
     def self.items_from(context)
       nodes_from(context).map { |node| new(node) }
     end
 
+    # Extract item nodes from a context.
+    #
     # @param context [Nokogiri::HTML::Document, Nokogiri::XML::NodeSet, Nokogiri::XML::Element]
     # @param node_set [Nokogiri::XML::NodeSet]
     # @return [Nokogiri::XML::NodeSet]
@@ -84,14 +116,15 @@ module MicroMicro
       node_set
     end
 
-    # @param node [Nokogiri::XML::Element]
-    # @return [Array<String>]
+    # Extract root class names from a node.
     #
-    # @example
     #   node = Nokogiri::HTML('<div class="h-card">Jason Garber</div>').at_css('div')
     #   MicroMicro::Item.types_from(node) #=> ['h-card']
+    #
+    # @param node [Nokogiri::XML::Element]
+    # @return [Array<String>]
     def self.types_from(node)
-      node.classes.select { |token| token.match?(/^h(?:\-[0-9a-z]+)?(?:\-[a-z]+)+$/) }.uniq.sort
+      node.classes.select { |token| token.match?(/^h(?:-[0-9a-z]+)?(?:-[a-z]+)+$/) }.uniq.sort
     end
 
     private

data/lib/micro_micro/parsers/base_property_parser.rb

@@ -7,27 +7,30 @@ module MicroMicro
         @node = property.node
       end
 
+      # @see http://microformats.org/wiki/microformats2-parsing#parsing_a_p-_property
+      # @see http://microformats.org/wiki/microformats2-parsing#parsing_an_e-_property
+      #
       # @return [String]
       def value
-        @value ||= serialized_node.text.strip
+        @value ||= begin
+          Document.text_content_from(node) do |context|
+            context.css('img').each { |img| img.content = " #{img['alt'] || img['src']} " }
+          end
+        end
+      end
+
+      # @param node [Nokogiri::XML::Element]
+      # @param attributes_map [Hash{String => Array}]
+      # @return [Array]
+      def self.attribute_value_from(node, attributes_map)
+        attributes_map.map do |attribute, names|
+          node[attribute] if names.include?(node.name) && node[attribute]
+        end.compact.first
       end
 
       private
 
       attr_reader :node, :property
-
-      # @see microformats2 Parsing Specification sections 1.3.1 and 1.3.4
-      # @see http://microformats.org/wiki/microformats2-parsing#parsing_a_p-_property
-      # @see http://microformats.org/wiki/microformats2-parsing#parsing_an_e-_property
-      def serialized_node
-        @serialized_node ||= begin
-          node.css(*Document.ignored_node_names).unlink
-
-          node.css('img').each { |img| img.content = " #{img['alt'] || Absolutely.to_abs(base: node.document.url, relative: img['src'])} " }
-
-          node
-        end
-      end
     end
   end
 end

data/lib/micro_micro/parsers/date_time_parser.rb

@@ -1,7 +1,6 @@
 module MicroMicro
   module Parsers
     class DateTimeParser
-      # @see Value Class Pattern section 4.2
       # @see http://microformats.org/wiki/value-class-pattern#Date_and_time_parsing
       #
       # Regexp pattern matching YYYY-MM-DD and YYY-DDD
@@ -58,12 +57,7 @@ module MicroMicro
 
       # @return [String]
       def value
-        @value ||= "#{normalized_date} #{normalized_time}#{normalized_timezone}".strip
-      end
-
-      # @return [Boolean]
-      def value?
-        value.present?
+        @value ||= "#{normalized_date} #{normalized_time}#{normalized_timezone}".strip if normalized_date || normalized_time || normalized_timezone
       end
 
       # @return [Hash{Symbol => String, nil}]

data/lib/micro_micro/parsers/date_time_property_parser.rb

@@ -1,65 +1,55 @@
 module MicroMicro
   module Parsers
     class DateTimePropertyParser < BasePropertyParser
-      # @see microformats2 Parsing Specification section 1.3.3
-      # @see http://microformats.org/wiki/microformats2-parsing#parsing_a_dt-_property
       HTML_ATTRIBUTES_MAP = {
         'datetime' => %w[del ins time],
         'title'    => %w[abbr],
         'value'    => %w[data input]
       }.freeze
 
+      # @see http://microformats.org/wiki/microformats2-parsing#parsing_a_dt-_property
+      #
       # @return [String]
       def value
-        @value ||= begin
-          return resolved_value if date_time_parser.value?
-          return attribute_values.first if attribute_values.any?
-
-          super
-        end
+        @value ||= resolved_value || attribute_value || super
       end
 
       private
 
+      # @see http://microformats.org/wiki/value-class-pattern#microformats2_parsers_implied_date
+      #
       # @return [MicroMicro::Parsers::DateTimeParser, nil]
-      def adopted_date_time
-        @adopted_date_time ||= begin
-          collections = property.collection.select { |prop| prop.prefix == 'dt' }.split(property)
+      def adopted_date_time_parser
+        @adopted_date_time_parser ||= begin
+          date_time_siblings = (property.prev_all.reverse + property.next_all).select { |prop| prop.prefix == 'dt' }
 
-          (collections.shift.reverse + collections).flatten.map { |prop| DateTimeParser.new(prop.value) }.find(&:normalized_date)
+          date_time_siblings.map { |prop| DateTimeParser.new(prop.value) }.find(&:normalized_date)
         end
       end
 
-      # @return [Array<String>]
-      def attribute_values
-        @attribute_values ||= begin
-          HTML_ATTRIBUTES_MAP.map do |attribute, names|
-            node[attribute] if names.include?(node.name) && node[attribute]
-          end.compact
-        end
+      # @return [String, nil]
+      def attribute_value
+        self.class.attribute_value_from(node, HTML_ATTRIBUTES_MAP)
       end
 
       # @return [MicroMicro::Parsers::DateTimeParser]
       def date_time_parser
-        @date_time_parser ||= DateTimeParser.new(value_class_pattern_parser.value)
+        @date_time_parser ||= DateTimeParser.new(ValueClassPatternParser.new(node, ' ').value)
       end
 
+      # @see http://microformats.org/wiki/value-class-pattern#microformats2_parsers_implied_date
+      #
       # @return [Boolean]
       def imply_date?
-        date_time_parser.normalized_time && !date_time_parser.normalized_date
+        date_time_parser.normalized_time && !date_time_parser.normalized_date && adopted_date_time_parser
       end
 
       # @return [String]
       def resolved_value
-        return "#{adopted_date_time.normalized_date} #{date_time_parser.value}" if imply_date? && adopted_date_time
+        return "#{adopted_date_time_parser.normalized_date} #{date_time_parser.value}" if imply_date?
 
         date_time_parser.value
       end
-
-      # @return [MicroMicro::Parsers::ValueClassPatternParser]
-      def value_class_pattern_parser
-        ValueClassPatternParser.new(node, ' ')
-      end
     end
   end
 end

data/lib/micro_micro/parsers/embedded_markup_property_parser.rb

@@ -1,28 +1,17 @@
 module MicroMicro
   module Parsers
     class EmbeddedMarkupPropertyParser < BasePropertyParser
-      HTML_ATTRIBUTE_NAMES = %w[action cite code codebase data href poster src].freeze
-
+      # @see http://microformats.org/wiki/microformats2-parsing#parsing_an_e-_property
+      #
+      # @return [Hash{Symbol => String}]
       def value
         @value ||= begin
           {
-            html: resolved_node.inner_html.strip,
+            html: node.inner_html.strip,
             value: super
           }
         end
       end
-
-      private
-
-      def resolved_node
-        @resolved_node ||= begin
-          HTML_ATTRIBUTE_NAMES.each do |attribute|
-            node.css("[#{attribute}]").each { |element| element[attribute] = Absolutely.to_abs(base: node.document.url, relative: element[attribute].strip) }
-          end
-
-          node
-        end
-      end
     end
   end
 end

data/lib/micro_micro/parsers/implied_name_property_parser.rb

@@ -1,77 +1,38 @@
 module MicroMicro
   module Parsers
     class ImpliedNamePropertyParser < BasePropertyParser
-      # @see microformats2 Parsing Specification section 1.3.5
-      # @see http://microformats.org/wiki/microformats2-parsing#parsing_for_implied_properties
-      HTML_ELEMENTS_MAP = {
-        'area' => 'alt',
-        'img'  => 'alt',
-        'abbr' => 'title'
+      HTML_ATTRIBUTES_MAP = {
+        'alt'   => %w[area img],
+        'title' => %w[abbr]
       }.freeze
 
+      # @see http://microformats.org/wiki/microformats2-parsing#parsing_for_implied_properties
+      #
       # @return [String]
       def value
-        @value ||= unresolved_value.strip
+        @value ||= attribute_value || text_content
       end
 
       private
 
-      # @return [Array<String>]
-      def attribute_values
-        @attribute_values ||= begin
-          HTML_ELEMENTS_MAP.map do |element, attribute|
-            node[attribute] if node.matches?("#{element}[#{attribute}]")
-          end.compact
-        end
-      end
-
-      # @return [Nokogiri::XML::Element, nil]
-      def child_node
-        @child_node ||= node.at_css('> :only-child')
-      end
-
-      # @return [Array<String>]
-      def child_node_attribute_values
-        @child_node_attribute_values ||= begin
-          HTML_ELEMENTS_MAP.map do |element, attribute|
-            child_node[attribute] if child_node.matches?("#{element}[#{attribute}]")
-          end.compact
-        end
+      # @return [Nokogiri::XML::NodeSet]
+      def candidate_nodes
+        @candidate_nodes ||= Nokogiri::XML::NodeSet.new(node.document, child_nodes.unshift(node))
       end
 
-      # @return [Nokogiri::XML::Element, nil]
-      def grandchild_node
-        @grandchild_node ||= child_node.at_css('> :only-child')
+      # @return [Array]
+      def child_nodes
+        [node.at_css('> :only-child'), node.at_css('> :only-child > :only-child')].compact.reject { |child_node| Item.item_node?(child_node) }
       end
 
-      # @return [Array<String>]
-      def grandchild_node_attribute_values
-        @grandchild_node_attribute_values ||= begin
-          HTML_ELEMENTS_MAP.map do |element, attribute|
-            grandchild_node[attribute] if grandchild_node.matches?("#{element}[#{attribute}]")
-          end.compact
-        end
-      end
-
-      # @return [Boolean]
-      def parse_child_node?
-        child_node && !Item.item_node?(child_node)
-      end
-
-      # @return [Boolean]
-      def parse_grandchild_node?
-        parse_child_node? && grandchild_node && !Item.item_node?(grandchild_node)
+      # @return [String, nil]
+      def attribute_value
+        candidate_nodes.map { |node| self.class.attribute_value_from(node, HTML_ATTRIBUTES_MAP) }.compact.first
       end
 
       # @return [String]
-      def unresolved_value
-        return attribute_values.first if attribute_values.any?
-        return child_node_attribute_values.first if parse_child_node? && child_node_attribute_values.any?
-        return grandchild_node_attribute_values.first if parse_grandchild_node? && grandchild_node_attribute_values.any?
-
-        serialized_node.css('img').each { |img| img.content = img['alt'] }
-
-        serialized_node.text
+      def text_content
+        @text_content ||= Document.text_content_from(node) { |context| context.css('img').each { |img| img.content = img['alt'] } }
       end
     end
   end

data/lib/micro_micro/parsers/implied_photo_property_parser.rb

@@ -1,14 +1,15 @@
 module MicroMicro
   module Parsers
     class ImpliedPhotoPropertyParser < BasePropertyParser
-      # @see microformats2 Parsing Specification section 1.3.5
-      # @see http://microformats.org/wiki/microformats2-parsing#parsing_for_implied_properties
       HTML_ELEMENTS_MAP = {
         'img'    => 'src',
         'object' => 'data'
       }.freeze
 
-      # @return [String, nil]
+      # @see http://microformats.org/wiki/microformats2-parsing#parsing_for_implied_properties
+      # @see http://microformats.org/wiki/microformats2-parsing#parse_an_img_element_for_src_and_alt
+      #
+      # @return [String, Hash{Symbol => String}, nil]
       def value
         @value ||= begin
           return unless resolved_value
@@ -34,12 +35,7 @@ module MicroMicro
 
       # @return [String, nil]
       def resolved_value
-        @resolved_value ||= Absolutely.to_abs(base: node.document.url, relative: unresolved_value.strip) if unresolved_value
-      end
-
-      # @return [String, nil]
-      def unresolved_value
-        @unresolved_value ||= value_node[HTML_ELEMENTS_MAP[value_node.name]] if value_node
+        @resolved_value ||= value_node[HTML_ELEMENTS_MAP[value_node.name]] if value_node
       end
 
       # @return [Nokogiri::XML::Element, nil]

data/lib/micro_micro/parsers/implied_url_property_parser.rb

@@ -1,16 +1,16 @@
 module MicroMicro
   module Parsers
     class ImpliedUrlPropertyParser < BasePropertyParser
-      # @see microformats2 Parsing Specification section 1.3.5
-      # @see http://microformats.org/wiki/microformats2-parsing#parsing_for_implied_properties
       HTML_ELEMENTS_MAP = {
         'a'    => 'href',
         'area' => 'href'
       }.freeze
 
+      # @see http://microformats.org/wiki/microformats2-parsing#parsing_for_implied_properties
+      #
       # @return [String, nil]
       def value
-        @value ||= resolved_value
+        @value ||= value_node[HTML_ELEMENTS_MAP[value_node.name]] if value_node
       end
 
       private
@@ -24,16 +24,6 @@ module MicroMicro
         end
       end
 
-      # @return [String, nil]
-      def resolved_value
-        @resolved_value ||= Absolutely.to_abs(base: node.document.url, relative: unresolved_value.strip) if unresolved_value
-      end
-
-      # @return [String, nil]
-      def unresolved_value
-        @unresolved_value ||= value_node[HTML_ELEMENTS_MAP[value_node.name]] if value_node
-      end
-
       # @return [Nokogiri::XML::Element, nil]
       def value_node
         @value_node ||= begin