stepmod-utils 0.2.0 → 0.2.3

data/lib/stepmod/utils/converters/a.rb

@@ -0,0 +1,47 @@
+# frozen_string_literal: true
+
+require "uri"
+
+module Stepmod
+  module Utils
+    module Converters
+      class A < ReverseAdoc::Converters::Base
+        def convert(node, state = {})
+          name  = treat_children(node, state)
+          href  = node['href']
+          title = extract_title(node)
+          id = node['id'] || node['name']
+
+          id = id&.gsub(/\s/, "")&.gsub(/__+/, "_")
+
+          if /^_Toc\d+$|^_GoBack$/.match id
+            ""
+          elsif !id.nil? && !id.empty?
+            "[[#{id}]]"
+          elsif href.to_s.start_with?('#')
+            href = href.sub(/^#/, "").gsub(/\s/, "").gsub(/__+/, "_")
+            if name.empty?
+              "<<#{href}>>"
+            else
+              "<<#{href},#{name}>>"
+            end
+          elsif href.to_s.empty?
+            name
+          else
+            name = title if name.empty?
+            href = "link:#{href}" unless href.to_s =~ URI::DEFAULT_PARSER.make_regexp
+            link = "#{href}[#{name}]"
+            # link.prepend(' ')
+            link = " #{link}"
+            link
+          end
+        end
+
+        private
+
+      end
+
+      ReverseAdoc::Converters.register :a, A.new
+    end
+  end
+end

data/lib/stepmod/utils/converters/blockquote.rb

@@ -0,0 +1,22 @@
+# frozen_string_literal: true
+
+module Stepmod
+  module Utils
+    module Converters
+      class Blockquote < ReverseAdoc::Converters::Base
+        def convert(node, state = {})
+          id = node['id']
+          anchor = id ? "[[#{id}]]\n" : ""
+          cite = node['cite']
+          attrs = cite ? "[quote, #{cite}]\n" : ""
+          content = treat_children(node, state).strip
+          content = ReverseAdoc.cleaner.remove_newlines(content)
+          #"\n\n> " << content.lines.to_a.join('> ') << "\n\n"
+          "\n\n#{attrs}____\n" << content.lines.to_a.join('') << "\n____\n\n"
+        end
+      end
+
+      ReverseAdoc::Converters.register :blockquote, Blockquote.new
+    end
+  end
+end

data/lib/stepmod/utils/converters/br.rb

@@ -0,0 +1,15 @@
+# frozen_string_literal: true
+
+module Stepmod
+  module Utils
+    module Converters
+      class Br < ReverseAdoc::Converters::Base
+        def convert(node, state = {})
+          " +\n"
+        end
+      end
+
+      ReverseAdoc::Converters.register :br, Br.new
+    end
+  end
+end

data/lib/stepmod/utils/converters/bypass.rb

@@ -0,0 +1,81 @@
+# frozen_string_literal: true
+
+module Stepmod
+  module Utils
+    module Converters
+      class Bypass < ReverseAdoc::Converters::Base
+        def convert(node, state = {})
+          treat_children(node, state)
+        end
+      end
+
+      ReverseAdoc::Converters.register :document, Bypass.new
+      ReverseAdoc::Converters.register :html,     Bypass.new
+      ReverseAdoc::Converters.register :body,     Bypass.new
+      ReverseAdoc::Converters.register :span,     Bypass.new
+      ReverseAdoc::Converters.register :thead,    Bypass.new
+      ReverseAdoc::Converters.register :tbody,    Bypass.new
+      ReverseAdoc::Converters.register :tfoot,    Bypass.new
+      ReverseAdoc::Converters.register :abbr,    Bypass.new
+      ReverseAdoc::Converters.register :acronym,    Bypass.new
+      ReverseAdoc::Converters.register :address,    Bypass.new
+      ReverseAdoc::Converters.register :applet,    Bypass.new
+      ReverseAdoc::Converters.register :map,    Bypass.new
+      ReverseAdoc::Converters.register :area,    Bypass.new
+      ReverseAdoc::Converters.register :bdi,    Bypass.new
+      ReverseAdoc::Converters.register :bdo,    Bypass.new
+      ReverseAdoc::Converters.register :big,    Bypass.new
+      ReverseAdoc::Converters.register :button,    Bypass.new
+      ReverseAdoc::Converters.register :canvas,    Bypass.new
+      ReverseAdoc::Converters.register :data,    Bypass.new
+      ReverseAdoc::Converters.register :datalist,    Bypass.new
+      ReverseAdoc::Converters.register :del,    Bypass.new
+      ReverseAdoc::Converters.register :ins,    Bypass.new
+      ReverseAdoc::Converters.register :dfn,    Bypass.new
+      ReverseAdoc::Converters.register :dialog,    Bypass.new
+      ReverseAdoc::Converters.register :embed,    Bypass.new
+      ReverseAdoc::Converters.register :fieldset,    Bypass.new
+      ReverseAdoc::Converters.register :font,    Bypass.new
+      ReverseAdoc::Converters.register :footer,    Bypass.new
+      ReverseAdoc::Converters.register :form,    Bypass.new
+      ReverseAdoc::Converters.register :frame,    Bypass.new
+      ReverseAdoc::Converters.register :frameset,    Bypass.new
+      ReverseAdoc::Converters.register :header,    Bypass.new
+      ReverseAdoc::Converters.register :iframe,    Bypass.new
+      ReverseAdoc::Converters.register :input,    Bypass.new
+      ReverseAdoc::Converters.register :label,    Bypass.new
+      ReverseAdoc::Converters.register :legend,    Bypass.new
+      ReverseAdoc::Converters.register :main,    Bypass.new
+      ReverseAdoc::Converters.register :menu,    Bypass.new
+      ReverseAdoc::Converters.register :menulist,    Bypass.new
+      ReverseAdoc::Converters.register :meter,    Bypass.new
+      ReverseAdoc::Converters.register :nav,    Bypass.new
+      ReverseAdoc::Converters.register :noframes,    Bypass.new
+      ReverseAdoc::Converters.register :noscript,    Bypass.new
+      ReverseAdoc::Converters.register :object,    Bypass.new
+      ReverseAdoc::Converters.register :optgroup,    Bypass.new
+      ReverseAdoc::Converters.register :option,    Bypass.new
+      ReverseAdoc::Converters.register :output,    Bypass.new
+      ReverseAdoc::Converters.register :param,    Bypass.new
+      ReverseAdoc::Converters.register :picture,    Bypass.new
+      ReverseAdoc::Converters.register :progress,    Bypass.new
+      ReverseAdoc::Converters.register :ruby,    Bypass.new
+      ReverseAdoc::Converters.register :rt,    Bypass.new
+      ReverseAdoc::Converters.register :rp,    Bypass.new
+      ReverseAdoc::Converters.register :s,    Bypass.new
+      ReverseAdoc::Converters.register :select,    Bypass.new
+      ReverseAdoc::Converters.register :small,    Bypass.new
+      ReverseAdoc::Converters.register :span,    Bypass.new
+      ReverseAdoc::Converters.register :strike,    Bypass.new
+      ReverseAdoc::Converters.register :details,    Bypass.new
+      ReverseAdoc::Converters.register :section,    Bypass.new
+      ReverseAdoc::Converters.register :summary,    Bypass.new
+      ReverseAdoc::Converters.register :svg,    Bypass.new
+      ReverseAdoc::Converters.register :template,    Bypass.new
+      ReverseAdoc::Converters.register :textarea,    Bypass.new
+      ReverseAdoc::Converters.register :track,    Bypass.new
+      ReverseAdoc::Converters.register :u,    Bypass.new
+      ReverseAdoc::Converters.register :wbr,    Bypass.new
+    end
+  end
+end

data/lib/stepmod/utils/converters/code.rb

@@ -0,0 +1,19 @@
+# frozen_string_literal: true
+
+module Stepmod
+  module Utils
+    module Converters
+      class Code < ReverseAdoc::Converters::Base
+        def convert(node, state = {})
+          "`#{node.text}`"
+        end
+      end
+
+      ReverseAdoc::Converters.register :code, Code.new
+      ReverseAdoc::Converters.register :tt, Code.new
+      ReverseAdoc::Converters.register :kbd, Code.new
+      ReverseAdoc::Converters.register :samp, Code.new
+      ReverseAdoc::Converters.register :var, Code.new
+    end
+  end
+end

data/lib/stepmod/utils/converters/comment.rb

@@ -0,0 +1,16 @@
+# frozen_string_literal: true
+
+module Stepmod
+  module Utils
+    module Converters
+      class Comment < ReverseAdoc::Converters::Base
+        def convert(node, state = {})
+          comment = node.text.strip.split("\n").map{|line| "// #{line}"}.join("\n")
+          "#{comment}\n"
+        end
+      end
+
+      ReverseAdoc::Converters.register :comment, Comment.new
+    end
+  end
+end

data/lib/stepmod/utils/converters/dd.rb

@@ -0,0 +1,15 @@
+# frozen_string_literal: true
+
+module Stepmod
+  module Utils
+    module Converters
+      class Dd < ReverseAdoc::Converters::Base
+        def convert(node, state = {})
+          "#{node.text}\n"
+        end
+
+        ReverseAdoc::Converters.register :dd, Dd.new
+      end
+    end
+  end
+end

data/lib/stepmod/utils/converters/def.rb

@@ -20,11 +20,14 @@ module Stepmod
           previous = nil
           result = ''
           converted.each.with_index do |(child, content), i|
-            if i == 0 || !block_tag?(child, previous)
-              result += " #{content}"
+            if block_tag?(child, previous)
+              result += "\n\n"
+            elsif comment_tag?(child, previous)
+              result += "\n"
             else
-              result += "\n\n#{content}"
+              result += ' '
             end
+            result += content
             previous = child
           end
           result.strip
@@ -32,7 +35,11 @@ module Stepmod
 
         def block_tag?(child, previous)
           %w[ul example note alt p].include?(child.name) ||
-            (child.previous && %w[ul example note alt p].include?(child.previous.name))
+            (previous && %w[ul example note alt p].include?(previous.name))
+        end
+
+        def comment_tag?(child, previous)
+          child.name == 'comment' || (previous && previous.name === 'comment')
         end
 
         def additional_block(node)

data/lib/stepmod/utils/converters/dl.rb

@@ -0,0 +1,31 @@
+# frozen_string_literal: true
+
+module Stepmod
+  module Utils
+    module Converters
+      class Dl < ReverseAdoc::Converters::Base
+        def convert(node, state = {})
+          cleaned_node = cleanup_trash_tags(node.clone)
+          treat_children(cleaned_node, state)
+        end
+
+        private
+
+        # https://github.com/metanorma/stepmod-utils/issues/48#issuecomment-784000377
+        # For simplicity reasons and so that we don't depend on the CVS repository's updates, we directly converting:
+        # <dt></dt><dd>a3ma &#160; : &#160; annotated 3d model assembly</dd> into a3ma:: annotated 3d model assembly
+        def cleanup_trash_tags(node)
+          inner_content = node.inner_html
+          inner_content
+            .gsub!(/<dt><\/dt>\s*?<dd>(.+?) &#xA0; : &#xA0; (.+?)<\/dd>/) do
+              "<dt>#{$1}</dt><dd>#{$2}</dd>"
+            end
+          node.inner_html = inner_content
+          node
+        end
+
+        ReverseAdoc::Converters.register :dl, Dl.new
+      end
+    end
+  end
+end

data/lib/stepmod/utils/converters/drop.rb

@@ -0,0 +1,22 @@
+# frozen_string_literal: true
+
+module Stepmod
+  module Utils
+    module Converters
+      class Drop < ReverseAdoc::Converters::Base
+        def convert(node, state = {})
+          ''
+        end
+      end
+
+      ReverseAdoc::Converters.register :caption, Drop.new
+      ReverseAdoc::Converters.register :figcaption, Drop.new
+      ReverseAdoc::Converters.register :title, Drop.new
+      ReverseAdoc::Converters.register :link, Drop.new
+      ReverseAdoc::Converters.register :style, Drop.new
+      ReverseAdoc::Converters.register :meta, Drop.new
+      ReverseAdoc::Converters.register :script, Drop.new
+      ReverseAdoc::Converters.register :comment, Drop.new
+    end
+  end
+end

data/lib/stepmod/utils/converters/dt.rb

@@ -0,0 +1,17 @@
+# frozen_string_literal: true
+
+module Stepmod
+  module Utils
+    module Converters
+      class Dt < ReverseAdoc::Converters::Base
+        def convert(node, state = {})
+          return "\n\n{blank}::" if node.text.strip.length.zero?
+
+          "\n\n#{node.text}:: "
+        end
+
+        ReverseAdoc::Converters.register :dt, Dt.new
+      end
+    end
+  end
+end

data/lib/stepmod/utils/converters/em_express_description.rb

@@ -0,0 +1,22 @@
+# frozen_string_literal: true
+
+module Stepmod
+  module Utils
+    module Converters
+      class Em < ReverseAdoc::Converters::Base
+        def convert(node, state = {})
+          content = treat_children(node, state.merge(already_italic: true))
+          if content.strip.empty? || state[:already_italic]
+            content
+          else
+            "#{content[/^\s*/]}_#{content.strip}_#{content[/\s*$/]}"
+          end
+        end
+      end
+
+      ReverseAdoc::Converters.register :em, Em.new
+      ReverseAdoc::Converters.register :i,  Em.new
+      ReverseAdoc::Converters.register :cite,  Em.new
+    end
+  end
+end

data/lib/stepmod/utils/converters/eqn.rb

@@ -0,0 +1,96 @@
+# frozen_string_literal: true
+
+require 'stepmod/utils/html_to_asciimath'
+
+module Stepmod
+  module Utils
+    module Converters
+      class Eqn < ReverseAdoc::Converters::Base
+        TAGS_NOT_IN_CONTEXT = %w[b i].freeze
+
+        def convert(node, state = {})
+          cloned_node = node.clone
+          if definition_node?(cloned_node)
+            return definition_converted(cloned_node, state)
+          end
+
+          stem_converted(cloned_node, state)
+        end
+
+        private
+
+        def definition_node?(node)
+          first_strong_node = node
+                                .children
+                                .find do |n|
+                                  return false if !n.text? && n.name != 'b'
+
+                                  n.name == 'b'
+                                end
+          first_strong_node &&
+            first_strong_node.next &&
+            first_strong_node.next.text? &&
+            first_strong_node.next.content =~ /\s+:/
+        end
+
+        def definition_converted(cloned_node, state)
+          first_strong_node = cloned_node
+                                .children
+                                .find do |n|
+                                  return false if !n.text? && n.name != 'b'
+
+                                  n.name == 'b'
+                                end
+          first_strong_node.next.content = first_strong_node.next.content.gsub(/\s?:/, '')
+          term = first_strong_node.text.strip
+          first_strong_node.remove
+          "#{term}:: #{treat_children(cloned_node, state)}"
+        end
+
+        def stem_converted(cloned_node, state)
+          remove_tags_not_in_context(cloned_node)
+          internal_content = treat_children(cloned_node, state)
+          content = Stepmod::Utils::HtmlToAsciimath.new.call(internal_content)
+          res = <<~TEMPLATE
+
+                [stem]
+                ++++
+                #{remove_trash_symbols(content.strip)}
+                ++++
+
+                TEMPLATE
+          res = "[[#{cloned_node['id']}]]\n#{res}" if cloned_node['id'] && cloned_node['id'].length > 0
+          res
+        end
+
+        def remove_trash_symbols(content)
+          content
+            .gsub(/ /, '')
+            .strip
+            .gsub(/\(\d\)$/, '')
+            .gsub(/\b(\w*?[_]+\w+)\b/, '"\1"')
+            .strip
+        end
+
+        # Remove all tags that make no sense in equations, eg: strong, italic
+        # Search for such tags, move their children into the root
+        # context and remove them
+        def remove_tags_not_in_context(node)
+          TAGS_NOT_IN_CONTEXT.each do |tag_name|
+            node
+              .children
+              .xpath("./#{tag_name}")
+              .map do |n|
+                n.tap{ |n| n.add_previous_sibling(n.children) }.remove
+              end
+          end
+          node.traverse do |descendant|
+            descendant.content = descendant.content.gsub('#8195;', '')
+          end
+        end
+      end
+
+      ReverseAdoc::Converters.register :eqn, Eqn.new
+    end
+  end
+end