article_json 0.4.0 → 0.4.1

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

@@ -21,7 +21,7 @@ module ArticleJSON
         private
 
         # Generate an array with the plain text representation of all elements
-        # 
+        #
         # @return [Array]
         def components
           @components ||=

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

@@ -20,7 +20,7 @@ module ArticleJSON
             private
 
             def embed_node
-              type = @element.embed_type.to_s.tr('_','-')
+              type = @element.embed_type.to_s.tr('_', '-')
               create_element(:div, class: "embed #{type}") do |div|
                 div.add_child(embedded_object)
               end
@@ -28,6 +28,7 @@ module ArticleJSON
 
             def embedded_object
               return unavailable_node unless @element.oembed_data
+
               Nokogiri::HTML.fragment(@element.oembed_data[:html])
             end
 

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

@@ -19,7 +19,7 @@ module ArticleJSON
             # @return [Nokogiri::XML::NodeSet]
             def figure_node
               create_element(:figure, node_opts) do |figure|
-                node =  @element&.href ? href_node : image_node
+                node = @element&.href ? href_node : image_node
                 figure.add_child(node)
                 if @element.caption&.any?
                   figure.add_child(caption_node(:figcaption))
@@ -42,6 +42,7 @@ module ArticleJSON
             # @return [Hash]
             def node_opts
               return if floating_class.nil?
+
               { class: floating_class }
             end
           end

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

@@ -11,6 +11,7 @@ module ArticleJSON
               return bold_and_italic_node if @element.bold && @element.italic
               return bold_node if @element.bold
               return italic_node if @element.italic
+
               content_node
             end
 
@@ -38,6 +39,7 @@ module ArticleJSON
             # @return [Nokogiri::XML::NodeSet]
             def content_node
               return create_text_nodes(@element.content) if @element.href.nil?
+
               create_element(:a, href: @element.href) do |a|
                 a.add_child(create_text_nodes(@element.content))
               end

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

@@ -104,6 +104,7 @@ module ArticleJSON
             def find_parser(text)
               text = text.strip.downcase
               return nil if text.empty?
+
               parsers.find { |klass| klass.matches?(text) }
             end
           end

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

@@ -19,11 +19,11 @@ module ArticleJSON
           # @return [Integer]
           def level
             case @node.name
-              when 'h1' then 1
-              when 'h2' then 2
-              when 'h3' then 3
-              when 'h4' then 4
-              when 'h5' then 5
+            when 'h1' then 1
+            when 'h2' then 2
+            when 'h3' then 3
+            when 'h4' then 4
+            when 'h5' then 5
             end
           end
 

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

@@ -54,8 +54,10 @@ module ArticleJSON
           # @return [String]
           def href
             return if @caption_node.nil?
+
             match = @caption_node.content.strip.match(href_regexp)
             return if match.nil?
+
             remove_image_link_tag
             match[:url]
           end

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

@@ -14,8 +14,8 @@ module ArticleJSON
           # @return [Symbol]
           def list_type
             case @node.name
-              when 'ol' then :ordered
-              when 'ul' then :unordered
+            when 'ol' then :ordered
+            when 'ul' then :unordered
             end
           end
 

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

@@ -31,6 +31,7 @@ module ArticleJSON
           # @return [Boolean]
           def empty?
             return @is_empty if defined? @is_empty
+
             @is_empty = node.inner_text.strip.empty? && !image? && !hr? && !br?
           end
 
@@ -38,6 +39,7 @@ module ArticleJSON
           # @return [Boolean]
           def heading?
             return @is_heading if defined? @is_heading
+
             @is_heading =
               !quote? && !text_box? && %w(h1 h2 h3 h4 h5).include?(node.name)
           end
@@ -52,6 +54,7 @@ module ArticleJSON
           # @return [Boolean]
           def paragraph?
             return @is_paragraph if defined? @is_paragraph
+
             @is_paragraph =
               node.name == 'p' &&
                 !empty? &&
@@ -65,7 +68,8 @@ module ArticleJSON
           # @return [Boolean]
           def list?
             return @is_list if defined? @is_list
-            @is_list = %w(ul ol).include?(node.name)
+
+            @is_list = %w[ul ol].include?(node.name)
           end
 
           # Check if the node starts a text box
@@ -73,6 +77,7 @@ module ArticleJSON
           # @return [Boolean]
           def text_box?
             return @is_text_box if defined? @is_text_box
+
             @is_text_box = begins_with?('textbox:') || begins_with?('highlight:')
           end
 
@@ -81,6 +86,7 @@ module ArticleJSON
           # @return [Boolean]
           def quote?
             return @is_quote if defined? @is_quote
+
             @is_quote = has_text?('quote:')
           end
 
@@ -88,6 +94,7 @@ module ArticleJSON
           # @return [Boolean]
           def image?
             return @is_image if defined? @is_image
+
             @is_image = image_url? || node.xpath('.//img').length > 0
           end
 
@@ -105,6 +112,7 @@ module ArticleJSON
           # @return [Boolean]
           def embed?
             return @is_embed if defined? @is_embed
+
             @is_embed = EmbeddedParser.supported?(node)
           end
 
@@ -113,6 +121,7 @@ module ArticleJSON
           # @return [Boolean]
           def br?
             return @is_br if defined? @is_br
+
             @is_br = node.name == 'br' || only_includes_brs?
           end
 
@@ -129,6 +138,7 @@ module ArticleJSON
             return :quote if quote?
             return :image if image?
             return :embed if embed?
+
             :unknown
           end
 
@@ -138,9 +148,11 @@ module ArticleJSON
           # @return [Boolean]
           def only_includes_brs?
             return false unless node.inner_text.strip.empty?
+
             tags = node.children.map(&:name)
             # Check if it only contains <br> and text nodes
-            return false unless tags.all? { |tag| %w(br text).include? tag }
+            return false unless tags.all? { |tag| %w[br text].include? tag }
+
             # Check if at least one is a `<br>` node
             tags.include?('br')
           end

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

@@ -118,6 +118,7 @@ module ArticleJSON
             nodes = []
             until !body_has_more_nodes? ||
                 NodeAnalyzer.new(@body_enumerator.peek).hr?
+
               nodes << @body_enumerator.next
             end
             nodes

data/lib/article_json/import/google_doc/html/shared/caption.rb

@@ -8,6 +8,7 @@ module ArticleJSON
             # @return [Array[ArticleJSON::Elements::Text]]
             def caption
               return [] if no_caption?
+
               ArticleJSON::Import::GoogleDoc::HTML::TextParser.extract(
                 node: @caption_node,
                 css_analyzer: @css_analyzer

data/lib/article_json/import/google_doc/html/shared/float.rb

@@ -8,9 +8,11 @@ module ArticleJSON
             # @return [Symbol]
             def float
               return unless @float_node.has_attribute?('class')
+
               node_class = @float_node.attribute('class').value || ''
               return :right if @css_analyzer.right_aligned?(node_class)
               return :left if @css_analyzer.left_aligned?(node_class)
+
               nil
             end
           end

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

@@ -10,7 +10,7 @@ module ArticleJSON
           #                                         May contain tags, too.
           # @param [Array[Nokogiri::HTML::Node]] nodes
           # @param [ArticleJSON::Import::GoogleDoc::HTML::CSSAnalyzer] css_analyzer
-          def initialize(type_node: ,nodes:, css_analyzer:)
+          def initialize(type_node:, nodes:, css_analyzer:)
             @nodes = nodes.reject { |node| NodeAnalyzer.new(node).empty? }
             @css_analyzer = css_analyzer
 
@@ -32,6 +32,7 @@ module ArticleJSON
             match = /(.*?)[\s\u00A0]+\[(?<tags>.*)\]/
                       .match(@type_node.inner_text)
             return [] unless match
+
             match[:tags].split(' ')
           end
 

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

@@ -43,6 +43,7 @@ module ArticleJSON
             if @node.name == 'span' &&
                 @node.first_element_child&.name == 'a' &&
                 @node.first_element_child&.has_attribute?('href')
+
               strip_google_redirect(
                 @node.first_element_child.attribute('href').value
               )
@@ -68,6 +69,7 @@ module ArticleJSON
             def extract(node:, css_analyzer:)
               node.children.map do |child_node|
                 next if NodeAnalyzer.new(child_node).empty?
+
                 new(node: child_node, css_analyzer: css_analyzer).element
               end.compact
             end

data/lib/article_json/utils/additional_element_placer.rb

@@ -40,6 +40,7 @@ module ArticleJSON
       # @return [Array[ArticleJSON::Elements::Base|Object]]
       def merge_elements
         return @additional_elements if @elements.nil? || @elements.empty?
+
         remaining_elements = @additional_elements.dup
         next_in = insert_next_element_in(0, remaining_elements)
         characters_passed = 0
@@ -48,6 +49,7 @@ module ArticleJSON
           .each_with_object([]) do |(element, next_element), result|
             result << element
             next if remaining_elements.empty?
+
             if element.respond_to?(:length)
               characters_passed += element.length
               next_in -= element.length

data/lib/article_json/utils/o_embed_resolver/base.rb

@@ -23,8 +23,10 @@ module ArticleJSON
         def unavailable_message
           [
             ArticleJSON::Elements::Text.new(content: "The #{name} "),
-            ArticleJSON::Elements::Text.new(content: source_url,
-                                            href: source_url),
+            ArticleJSON::Elements::Text.new(
+              content: source_url,
+              href: source_url
+            ),
             ArticleJSON::Elements::Text.new(content: ' is not available.'),
           ]
         end
@@ -44,19 +46,27 @@ module ArticleJSON
         # @return [Hash|nil]
         def parsed_api_response
           return @api_response if defined? @api_response
+
           @api_response = begin
             uri = URI.parse(oembed_url)
             http = Net::HTTP.new(uri.host, uri.port)
             http.use_ssl = (uri.scheme == 'https')
             response = http.request(Net::HTTP::Get.new(uri, http_headers))
-            if response.kind_of? Net::HTTPSuccess
-              JSON.parse(response.body, symbolize_names: true)
+
+            if response.is_a? Net::HTTPSuccess
+              data = JSON.parse(response.body, symbolize_names: true)
+              transform_api_response(data)
             end
           rescue Net::ProtocolError, JSON::ParserError
             nil
           end
         end
 
+        # @return [Hash]
+        def transform_api_response(data)
+          data
+        end
+
         # @return [Hash]
         def http_headers
           headers = { 'Content-Type' => 'application/json' }

data/lib/article_json/utils/o_embed_resolver/facebook_video.rb

@@ -12,7 +12,7 @@ module ArticleJSON
         # @return [String]
         def oembed_url
           "https://graph.facebook.com/v9.0/oembed_video?url=#{source_url}" \
-          "&access_token=#{access_token}"
+            "&access_token=#{access_token}"
         end
 
         # The video URL of the element

data/lib/article_json/utils/o_embed_resolver/slideshare.rb

@@ -11,8 +11,8 @@ module ArticleJSON
         # The URL for the oembed API call
         # @return [String]
         def oembed_url
-          'https://www.slideshare.net/api/oembed/2?format=json&url='\
-          "#{source_url}"
+          'https://www.slideshare.net/api/oembed/2?format=json&url=' \
+            "#{source_url}"
         end
 
         # The URL of the slideshow

data/lib/article_json/utils/o_embed_resolver/youtube_video.rb

@@ -19,6 +19,19 @@ module ArticleJSON
         def source_url
           "https://www.youtube.com/watch?v=#{@element.embed_id}"
         end
+
+        protected
+
+        # @param [Hash] data
+        # @return [Hash]
+        def transform_api_response(data)
+          return data unless data&.dig(:thumbnail_url)
+          
+          # Replace the default YouTube thumbnail with the max resolution version
+          data.merge(
+            thumbnail_url: data[:thumbnail_url].sub(%r{hqdefault\.jpg$}, 'maxresdefault.jpg')
+          )
+        end
       end
     end
   end

data/lib/article_json/version.rb

@@ -1,3 +1,3 @@
 module ArticleJSON
-  VERSION = '0.4.0'
+  VERSION = '0.4.1'
 end

data/lib/article_json.rb

@@ -100,15 +100,4 @@ require_relative 'article_json/export/amp/elements/embed'
 require_relative 'article_json/export/amp/custom_element_library_resolver'
 require_relative 'article_json/export/amp/exporter'
 
-require_relative 'article_json/export/facebook_instant_article/elements/base'
-require_relative 'article_json/export/facebook_instant_article/elements/text'
-require_relative 'article_json/export/facebook_instant_article/elements/paragraph'
-require_relative 'article_json/export/facebook_instant_article/elements/list'
-require_relative 'article_json/export/facebook_instant_article/elements/heading'
-require_relative 'article_json/export/facebook_instant_article/elements/quote'
-require_relative 'article_json/export/facebook_instant_article/elements/text_box'
-require_relative 'article_json/export/facebook_instant_article/elements/image'
-require_relative 'article_json/export/facebook_instant_article/elements/embed'
-require_relative 'article_json/export/facebook_instant_article/exporter'
-
 require_relative 'article_json/article'

metadata

@@ -1,17 +1,16 @@
 --- !ruby/object:Gem::Specification
 name: article_json
 version: !ruby/object:Gem::Version
-  version: 0.4.0
+  version: 0.4.1
 platform: ruby
 authors:
 - Daniel Sager
 - Manu Campos
 - Nicolas Fricke
 - Damien Dillon
-autorequire:
 bindir: bin
 cert_chain: []
-date: 2024-02-14 00:00:00.000000000 Z
+date: 1980-01-02 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: nokogiri
@@ -105,7 +104,6 @@ description: |2
         - simple HTML format
         - AMP format
         - Apple News Format (ANF)
-        - Facebook Instant Article HTML
         - plain text
         - JSON
 email: info@devex.com
@@ -119,7 +117,6 @@ files:
 - README.md
 - bin/article_json_export_amp.rb
 - bin/article_json_export_apple_news.rb
-- bin/article_json_export_facebook.rb
 - bin/article_json_export_google_doc.rb
 - bin/article_json_export_html.rb
 - bin/article_json_export_plain_text.rb
@@ -173,16 +170,6 @@ files:
 - lib/article_json/export/common/html/elements/text.rb
 - lib/article_json/export/common/html/elements/text_box.rb
 - lib/article_json/export/common/html/exporter.rb
-- lib/article_json/export/facebook_instant_article/elements/base.rb
-- lib/article_json/export/facebook_instant_article/elements/embed.rb
-- lib/article_json/export/facebook_instant_article/elements/heading.rb
-- lib/article_json/export/facebook_instant_article/elements/image.rb
-- lib/article_json/export/facebook_instant_article/elements/list.rb
-- lib/article_json/export/facebook_instant_article/elements/paragraph.rb
-- lib/article_json/export/facebook_instant_article/elements/quote.rb
-- lib/article_json/export/facebook_instant_article/elements/text.rb
-- lib/article_json/export/facebook_instant_article/elements/text_box.rb
-- lib/article_json/export/facebook_instant_article/exporter.rb
 - lib/article_json/export/html/elements/base.rb
 - lib/article_json/export/html/elements/embed.rb
 - lib/article_json/export/html/elements/heading.rb
@@ -236,7 +223,6 @@ homepage: https://github.com/Devex/article_json
 licenses:
 - MIT
 metadata: {}
-post_install_message:
 rdoc_options: []
 require_paths:
 - lib
@@ -251,8 +237,7 @@ required_rubygems_version: !ruby/object:Gem::Requirement
     - !ruby/object:Gem::Version
       version: '0'
 requirements: []
-rubygems_version: 3.4.10
-signing_key:
+rubygems_version: 3.6.9
 specification_version: 4
 summary: JSON Format for News Articles & Ruby Gem
 test_files: []

data/bin/article_json_export_facebook.rb

@@ -1,16 +0,0 @@
-#!/usr/bin/env ruby
-
-# # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # #
-#
-# Simple script to read a JSON document and export it to Facebook Instant
-# Article.
-#
-# Usage:
-#
-#   ./bin/article_json_export_facebook.rb < my_document.json
-#
-# # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # #
-
-require 'bundler/setup'
-require_relative '../lib/article_json'
-puts ArticleJSON::Article.from_json(ARGF.read).to_facebook_instant_article

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

@@ -1,30 +0,0 @@
-module ArticleJSON
-  module Export
-    module FacebookInstantArticle
-      module Elements
-        class Base
-          include ArticleJSON::Export::Common::HTML::Elements::Base
-
-          class << self
-            # Return the module namespace this class and its subclasses are
-            # nested in
-            # @return [Module]
-            def namespace
-              ArticleJSON::Export::FacebookInstantArticle::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
-              :facebook_instant_article
-            end
-          end
-        end
-      end
-    end
-  end
-end

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

@@ -1,44 +0,0 @@
-module ArticleJSON
-  module Export
-    module FacebookInstantArticle
-      module Elements
-        class Embed < Base
-          include ArticleJSON::Export::Common::HTML::Elements::Shared::Caption
-
-          # Generate the embedded element node
-          # @return [Nokogiri::XML::NodeSet]
-          def export
-            create_element(:figure, class: 'op-interactive') do |figure|
-              figure.add_child(embed_node)
-              if @element.caption&.any?
-                figure.add_child(caption_node(:figcaption))
-              end
-            end
-          end
-
-          private
-
-          # Type specific object that should be embedded
-          # @return [Nokogiri::XML::Element]
-          def embed_node
-            if %i(facebook_video tweet).include? @element.embed_type.to_sym
-              iframe_node
-            else
-              embedded_object
-            end
-          end
-
-          def iframe_node
-            create_element(:iframe) do |div|
-              div.add_child(embedded_object)
-            end
-          end
-
-          def embedded_object
-            Nokogiri::HTML.fragment(@element.oembed_data[:html])
-          end
-        end
-      end
-    end
-  end
-end

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

@@ -1,11 +0,0 @@
-module ArticleJSON
-  module Export
-    module FacebookInstantArticle
-      module Elements
-        class Heading < Base
-          include ArticleJSON::Export::Common::HTML::Elements::Heading
-        end
-      end
-    end
-  end
-end

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

@@ -1,11 +0,0 @@
-module ArticleJSON
-  module Export
-    module FacebookInstantArticle
-      module Elements
-        class Image < Base
-          include ArticleJSON::Export::Common::HTML::Elements::Image
-        end
-      end
-    end
-  end
-end

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

@@ -1,11 +0,0 @@
-module ArticleJSON
-  module Export
-    module FacebookInstantArticle
-      module Elements
-        class List < Base
-          include ArticleJSON::Export::Common::HTML::Elements::List
-        end
-      end
-    end
-  end
-end

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

@@ -1,11 +0,0 @@
-module ArticleJSON
-  module Export
-    module FacebookInstantArticle
-      module Elements
-        class Paragraph < Base
-          include ArticleJSON::Export::Common::HTML::Elements::Paragraph
-        end
-      end
-    end
-  end
-end

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

@@ -1,30 +0,0 @@
-module ArticleJSON
-  module Export
-    module FacebookInstantArticle
-      module Elements
-        class Quote < Base
-          include ArticleJSON::Export::Common::HTML::Elements::Quote
-
-          private
-
-          # @return [Hash]
-          def node_opts
-            {}
-          end
-
-          # HTML tag for the wrapping node
-          # @return [Symbol]
-          def quote_tag
-            :aside
-          end
-
-          # HTML tag for the node containing the caption
-          # @return [Symbol]
-          def caption_tag
-            :cite
-          end
-        end
-      end
-    end
-  end
-end

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

@@ -1,11 +0,0 @@
-module ArticleJSON
-  module Export
-    module FacebookInstantArticle
-      module Elements
-        class Text < Base
-          include ArticleJSON::Export::Common::HTML::Elements::Text
-        end
-      end
-    end
-  end
-end