article_json 0.3.7 → 0.4.0

data/lib/article_json/export/apple_news/elements/base.rb

@@ -0,0 +1,53 @@
+module ArticleJSON
+  module Export
+    module AppleNews
+      module Elements
+        class Base
+          include ArticleJSON::Export::Common::Elements::Base
+
+          # Export the given element. Dynamically looks up the right
+          # export-element-class, instantiates it and then calls the `#export`
+          # method.
+          # Defaults to nil, e.g. if no exporter is specified for the given
+          # type.
+          # @return [String]
+          def export
+            super || nil
+          end
+
+          class << self
+            # Return the module namespace this class and its subclasses are
+            # nested within.
+            # @return [Module]
+            def namespace
+              ArticleJSON::Export::AppleNews::Elements
+            end
+
+            private
+
+            # The format this exporter is returning. This is used to determine
+            # which custom element exporters should be applied from the
+            # configuration.
+            # @return [Symbol]
+            def export_format
+              :apple_news
+            end
+
+            def default_exporter_mapping
+              {
+                text: namespace::Text,
+                paragraph: namespace::Paragraph,
+                heading: namespace::Heading,
+                quote: namespace::Quote,
+                list: namespace::List,
+                image: namespace::Image,
+                embed: namespace::Embed,
+                text_box: namespace::TextBox,
+              }
+            end
+          end
+        end
+      end
+    end
+  end
+end

data/lib/article_json/export/apple_news/elements/embed.rb

@@ -0,0 +1,130 @@
+module ArticleJSON
+  module Export
+    module AppleNews
+      module Elements
+        class Embed < Base
+          # Embed| Embed, Caption
+          # @return [Hash, Array<Hash>]
+          def export
+            caption_text.nil? ? embed : [embed, caption]
+          end
+
+          private
+
+          # Embed
+          # @return [Hash]
+          def embed
+            {
+              role: role,
+              URL: source_url,
+              caption: caption_text,
+            }.compact
+          end
+
+          # Caption
+          # @return [Hash]
+          def caption
+            {
+              role: 'caption',
+              text: caption_text,
+              format: 'html',
+              layout: 'captionLayout',
+              textStyle: 'captionStyle',
+            }
+          end
+
+          # Get the exporter class for text elements
+          # @return [ArticleJSON::Export::Common::HTML::Elements::Base]
+          def text_exporter
+            self.class.exporter_by_type(:text)
+          end
+
+          # Caption Text
+          # @return [String]
+          def caption_text
+            return nil if role.nil? # Do not show captions for unsupported components
+
+            text.empty? ? nil : text
+          end
+
+          # @return [String]
+          def text
+            @element.caption.map do |child_element|
+              text_exporter.new(child_element)
+                .export
+            end.join
+          end
+
+          def role
+            @role ||=
+              case embed_type
+              when :youtube_video, :vimeo_video, :dailymotion_video
+                :embedwebvideo
+              when :facebook_video
+                :facebook_post
+              when :tweet
+                :tweet
+              when :slideshare
+                nil
+              when :soundcloud
+                nil
+              else
+                nil
+              end
+          end
+
+          def source_url
+            case embed_type
+            when :youtube_video
+              build_embeded_youtube_url
+            when :vimeo_video
+              build_embeded_vimeo_url
+            when :dailymotion_video
+              build_embeded_vimeo_url
+            when :facebook_video
+              build_facebook_video_url
+            when :tweet
+              build_twitter_url
+            when :slideshare
+              nil
+            when :soundcloud
+              nil
+            else
+              nil
+            end
+          end
+
+          def build_embeded_youtube_url
+            "https://www.youtube.com/embed/#{embed_id}"
+          end
+
+          def build_embeded_vimeo_url
+            "https://player.vimeo.com/video/#{embed_id}"
+          end
+
+          def build_embeded_dailymotion_url
+            "https://geo.dailymotion.com/player.html?video=#{embed_id}"
+          end
+
+          def build_facebook_video_url
+            username, id = embed_id.to_s.split("/", 2)
+            "https://www.facebook.com/#{username}/videos/#{id}"
+          end
+
+          def build_twitter_url
+            username, id = embed_id.to_s.split("/", 2)
+            "https://twitter.com/#{username}/status/#{id}"
+          end
+
+          def embed_type
+            @embed_type ||= @element.embed_type.to_sym
+          end
+
+          def embed_id
+            @embed_id ||= @element.embed_id.to_sym
+          end
+        end
+      end
+    end
+  end
+end

data/lib/article_json/export/apple_news/elements/heading.rb

@@ -0,0 +1,32 @@
+module ArticleJSON
+  module Export
+    module AppleNews
+      module Elements
+        class Heading < Base
+          # Headline
+          # @return [Hash]
+          def export
+            {
+              role: role,
+              text: @element.content,
+              layout: 'titleLayout',
+              textStyle: 'defaultTitle',
+            }
+          end
+
+          private
+
+          # The role of text component for adding a heading. (Required) Always
+          # one of these roles for this component: heading, heading1, heading2,
+          # heading3, heading4, heading5, or heading6.
+          # @return [String]
+          def role
+            return 'heading' if @element.level.nil?
+
+            "heading#{@element.level}"
+          end
+        end
+      end
+    end
+  end
+end

data/lib/article_json/export/apple_news/elements/image.rb

@@ -0,0 +1,58 @@
+module ArticleJSON
+  module Export
+    module AppleNews
+      module Elements
+        class Image < Base
+          # Image | Image, Caption
+          # @return [Hash, Array<Hash>]
+          def export
+            caption_text.nil? ? image : [image, caption]
+          end
+
+          private
+          # Image
+          # @return [Hash]
+          def image
+            {
+              role: 'image',
+              URL: @element.source_url,
+              caption: caption_text,
+            }.compact
+          end
+
+          # Caption
+          # @return [Hash]
+          def caption
+            {
+              role: 'caption',
+              text: caption_text,
+              format: 'html',
+              layout: 'captionLayout',
+              textStyle: 'captionStyle',
+            }
+          end
+
+          # Get the exporter class for text elements
+          # @return [ArticleJSON::Export::Common::HTML::Elements::Base]
+          def text_exporter
+            self.class.exporter_by_type(:text)
+          end
+
+          # Caption Text
+          # @return [String]
+          def caption_text
+            text.empty? ? nil : text
+          end
+
+          # @return [String]
+          def text
+            @element.caption.map do |child_element|
+              text_exporter.new(child_element)
+                .export
+            end.join
+          end
+        end
+      end
+    end
+  end
+end

data/lib/article_json/export/apple_news/elements/list.rb

@@ -0,0 +1,67 @@
+module ArticleJSON
+  module Export
+    module AppleNews
+      module Elements
+        class List < Base
+          # List
+          # @return [Hash]
+          def export
+            {
+              role: 'body',
+              text: list_text,
+              format: 'html',
+              layout: 'bodyLayout',
+              textStyle: 'bodyStyle',
+            }
+          end
+
+          private
+
+          # Get the exporter class for text elements
+          #
+          # @return [ArticleJSON::Export::Common::HTML::Elements::<Class>]
+          def text_exporter
+            self.class.exporter_by_type(:text)
+          end
+
+          # When it is an unordered list wrap it in <ul></ul>
+          # When it is an ordered list wrap it in <ol></ol>
+          #
+          # List Text
+          # @return [String]
+          def list_text
+            prepend_list_tag + list + append_list_tag
+          end
+
+          # Each list item should be wrapped in <li></li>
+          #
+          # @return [String]
+          def list
+            @element.content.map do |paragraph_element|
+              line_item = paragraph_element.content.map do |text_element|
+                text_exporter.new(text_element).export
+              end.join
+
+              "<li>#{line_item}</li>"
+            end.join
+          end
+
+          # @return [String]
+          def prepend_list_tag
+            ordered_list? ? '<ol>' : '<ul>'
+          end
+
+          # @return [String]
+          def append_list_tag
+            ordered_list? ? '</ol>' : '</ul>'
+          end
+
+          # @return [Boolean]
+          def ordered_list?
+            @element.list_type == :ordered
+          end
+        end
+      end
+    end
+  end
+end

data/lib/article_json/export/apple_news/elements/paragraph.rb

@@ -0,0 +1,36 @@
+module ArticleJSON
+  module Export
+    module AppleNews
+      module Elements
+        class Paragraph < Base
+          # Generate the paragraph node with its containing text elements
+          # @return [Hash]
+          def export
+            {
+              role: 'body',
+              text: text,
+              format: 'html',
+              layout: 'bodyLayout',
+              textStyle: 'bodyStyle',
+            }
+          end
+
+          private
+
+          # Get the exporter class for text elements
+          # @return [ArticleJSON::Export::Common::HTML::Elements::Base]
+          def text_exporter
+            self.class.exporter_by_type(:text)
+          end
+
+          def text
+            @element.content.map do |child_element|
+              text_exporter.new(child_element)
+                .export
+            end.join
+          end
+        end
+      end
+    end
+  end
+end

data/lib/article_json/export/apple_news/elements/quote.rb

@@ -0,0 +1,60 @@
+module ArticleJSON
+  module Export
+    module AppleNews
+      module Elements
+        class Quote < Base
+          include ArticleJSON::Export::Common::HTML::Elements::Base
+          include ArticleJSON::Export::Common::HTML::Elements::Text
+
+          def export
+            [quote, author]
+          end
+
+          private
+
+          # Quote
+          # @return [Hash]
+          def quote
+            {
+              role: 'pullquote',
+              text: quote_text,
+              format: 'html',
+              layout: 'pullquoteLayout',
+              textStyle: 'pullquoteStyle',
+            }
+          end
+
+          # Author
+          # @return [Hash]
+          def author
+            {
+              role: 'author',
+              text: author_text,
+              format: 'html',
+              layout: 'pullquoteAttributeLayout',
+              textStyle: 'quoteAttributeStyle',
+            }
+          end
+
+          def text_exporter
+            self.class.exporter_by_type(:text)
+          end
+
+          # Quote Text
+          # @return [String]
+          def quote_text
+            element = @element.content.first&.content.first
+            text_exporter.new(element).export
+          end
+
+          # Author Text
+          # @return [String]
+          def author_text
+            element = @element.caption.first
+            text_exporter.new(element).export
+          end
+        end
+      end
+    end
+  end
+end

data/lib/article_json/export/apple_news/elements/text.rb

@@ -0,0 +1,42 @@
+module ArticleJSON
+  module Export
+    module AppleNews
+      module Elements
+        class Text < Base
+          include ArticleJSON::Export::Common::HTML::Elements::Base
+          include ArticleJSON::Export::Common::HTML::Elements::Text
+
+          UNSUPPORTED_HTML_TAGS = %w[title meta script noscript style link applet object iframe
+            noframes form select option optgroup
+          ].freeze
+
+          # A Nokogiri object is returned with`super`, which is is then
+          # returned as a either a string or as HTML (when not plain text),
+          # both of which are compatible with Apple News format. Takes into
+          # account bold, italic and href.
+          # @return [String]
+          def export
+            super.to_s
+          end
+
+          # @param [String] text
+          def create_text_nodes(text)
+            Nokogiri::HTML.fragment(sanitize_text(text).gsub(/\n/, '<br>')).children
+          end
+
+          # Removes UNSUPPORTED_TAGS from text
+          #
+          # @param [String] text
+          # @return [String]
+          def sanitize_text(text)
+            doc = Nokogiri::HTML.fragment(text)
+            UNSUPPORTED_HTML_TAGS.each do |tag|
+              doc.search(tag).each(&:remove)
+            end
+            doc.inner_html
+          end
+        end
+      end
+    end
+  end
+end

data/lib/article_json/export/apple_news/elements/text_box.rb

@@ -0,0 +1,51 @@
+module ArticleJSON
+  module Export
+    module AppleNews
+      module Elements
+        class TextBox < Base
+          include ArticleJSON::Export::Common::HTML::Elements::TextBox
+          # List
+          # @return [Hash]
+          def export
+            {
+              role: 'container',
+              layout: 'textBoxLayout',
+              style: 'textBoxStyle',
+              components: map_styles(elements),
+            }
+          end
+
+          private
+
+          # @return [Array]
+          def elements
+            @element.content.map do |child_element|
+              case child_element
+              when ArticleJSON::Elements::Heading
+                namespace::Heading.new(child_element).export
+              when ArticleJSON::Elements::Paragraph
+                namespace::Paragraph.new(child_element).export
+              when ArticleJSON::Elements::List
+                namespace::List.new(child_element).export
+              else
+                namespace::Text.new(child_element).export
+              end
+            end
+          end
+
+          # @return [Module]
+          def namespace
+            ArticleJSON::Export::AppleNews::Elements
+          end
+
+          # @return [Array]
+          def map_styles(elements)
+            elements.map do |child_element|
+              child_element.merge(layout: 'textBox' +  child_element[:layout].sub(/\S/, &:upcase))
+            end
+          end
+        end
+      end
+    end
+  end
+end

data/lib/article_json/export/apple_news/exporter.rb

@@ -0,0 +1,37 @@
+module ArticleJSON
+  module Export
+    module AppleNews
+      class Exporter
+        # @param [Array[ArticleJSON::Elements::Base]] elements
+        def initialize(elements)
+          @elements = elements
+        end
+
+        # Return the components section of an Apple News Article as JSON
+        #
+        # Images and EmbededVideos are nested in an array with the components
+        # array when they contain captions. As Apple News skips over these
+        # nested arrays, we must flatten the array.
+        #
+        # @return [String]
+        def to_json
+          { components: components.flatten }.to_json
+        end
+
+        private
+
+        # Generate an array with the plain text representation of all elements
+        # 
+        # @return [Array]
+        def components
+          @components ||=
+            @elements.map do |element|
+              ArticleJSON::Export::AppleNews::Elements::Base
+                .build(element)
+                &.export
+            end.reject { |hash| hash.nil? || hash.empty? }
+        end
+      end
+    end
+  end
+end

data/lib/article_json/export/common/html/elements/image.rb

@@ -29,7 +29,7 @@ module ArticleJSON
 
             # @return [Nokogiri::XML::NodeSet]
             def image_node
-              create_element(:img, src: @element.source_url)
+              create_element(:img, src: @element.source_url, alt: @element.alt)
             end
 
             # @return [Nokogiri::XML::NodeSet]

data/lib/article_json/import/google_doc/html/image_parser.rb

@@ -19,22 +19,34 @@ module ArticleJSON
             @float_node = @node
           end
 
+          # The value of the image's `alt` attribute
+          # @return [String]
+          def alt
+            return '' if image_url?
+
+            image_node.attribute('alt')&.value || ''
+          end
+
           # The value of the image's `src` attribute
           # @return [String]
           def source_url
+            return @node.inner_text.strip if image_url?
+
             image_node.attribute('src').value
           end
 
           # The node of the actual image
           # @return [Nokogiri::HTML::Node]
           def image_node
-            @node.xpath('.//img').first
+            return @image_node if defined? @image_node
+
+            @image_node = @node.xpath('.//img').first
           end
 
           # Check if the image is floating (left, right or not at all)
           # @return [Symbol]
           def float
-            super if floatable_size?
+            super if image_url? || floatable_size?
           end
 
           # Extracts an href from the tag [image-link-to: url]) if present
@@ -54,7 +66,8 @@ module ArticleJSON
               source_url: source_url,
               float: float,
               caption: caption,
-              href: @href
+              href: @href,
+              alt: alt
             )
           end
 
@@ -73,6 +86,7 @@ module ArticleJSON
           def href_regexp
             %r{\[image-link-to:\s+(?<url>.*?)\]}
           end
+
           # Check if the image's width can be determined and is less than 500px
           # This is about 3/4 of the google document width...
           # @return [Boolean]
@@ -94,6 +108,13 @@ module ArticleJSON
                 match['px'].to_i if match && match['px']
               end
           end
+
+          # When the current node doesn't contain an actual image tag,
+          # we're dealing with an image URL
+          # @return [Boolean]
+          def image_url?
+            image_node.nil?
+          end
         end
       end
     end

data/lib/article_json/import/google_doc/html/node_analyzer.rb

@@ -88,7 +88,17 @@ module ArticleJSON
           # @return [Boolean]
           def image?
             return @is_image if defined? @is_image
-            @is_image = node.xpath('.//img').length > 0
+            @is_image = image_url? || node.xpath('.//img').length > 0
+          end
+
+          # Check if the node contains an image URL
+          # @return [Boolean]
+          def image_url?
+            return @is_image_url if defined? @is_image_url
+
+            text = node.inner_text.strip
+            url_regexp = %r{https?:\/\/\S+\.(?:jpg|jpeg|png|gif)}i
+            @is_image_url = !!(url_regexp =~ text)
           end
 
           # Check if the node contains an embedded element

data/lib/article_json/import/google_doc/html/parser.rb

@@ -6,7 +6,12 @@ module ArticleJSON
           # @param [String] html
           def initialize(html)
             doc = Nokogiri::HTML(html)
-            @body_enumerator = doc.xpath('//body').last.children.to_enum
+            selection = if doc.xpath('//body/div').empty?
+                          doc.xpath('//body')
+                        else
+                          doc.xpath('//body/div')
+                        end
+            @body_enumerator = selection.last.children.to_enum
 
             css_node = doc.xpath('//head/style').last
             @css_analyzer = CSSAnalyzer.new(css_node&.inner_text)