coradoc 1.0.0 → 1.1.1

data/lib/coradoc/element/list/definition.rb

@@ -1,9 +1,11 @@
 module Coradoc
   module Element
     module List
-      class Definition < Core
+      class Definition < Base
         attr_accessor :items, :delimiter
 
+        declare_children :items
+
         def initialize(items, options = {})
           @items = items
           @delimiter = options.fetch(:delimiter, "::")

data/lib/coradoc/element/list_item.rb

@@ -1,23 +1,33 @@
 module Coradoc
   module Element
     class ListItem < Base
-      attr_accessor :id
+      attr_accessor :marker, :id, :anchor, :content, :line_break
 
       declare_children :content, :id, :anchor
 
       def initialize(content, options = {})
-        @content = content
+        @marker = options.fetch(:marker, nil)
         @id = options.fetch(:id, nil)
         @anchor = @id.nil? ? nil : Inline::Anchor.new(@id)
+        @content = content
+        @line_break = options.fetch(:line_break, "\n")
       end
 
       def to_adoc
         anchor = @anchor.nil? ? "" : @anchor.to_adoc.to_s
         content = Array(@content).map do |subitem|
-          Coradoc::Generator.gen_adoc(subitem).chomp
-        end.join("\n+\n")
+          next if subitem.is_a? Inline::HardLineBreak
+
+          subcontent = Coradoc::Generator.gen_adoc(subitem)
+          # Only try to postprocess elements that are text,
+          # otherwise we could strip markup.
+          if Coradoc.a_single?(subitem, Coradoc::Element::TextElement)
+            subcontent = Coradoc.strip_unicode(subcontent)
+          end
+          subcontent.chomp
+        end.compact.join("\n+\n")
 
-        " #{anchor}#{content.chomp}\n"
+        " #{anchor}#{content.chomp}#{@line_break}"
       end
     end
   end

data/lib/coradoc/element/paragraph.rb

@@ -7,9 +7,10 @@ module Coradoc
 
       def initialize(content, options = {})
         @content = content
-        @meta = options.fetch(:meta, nil)
-        id = options.fetch(:id, nil)
-        @anchor = Inline::Anchor.new(id) if id
+        @id = options.fetch(:id, nil)
+        @anchor = Inline::Anchor.new(@id) if @id
+        @title = options.fetch(:title, nil)
+        @attributes = options.fetch(:attributes, nil)
         @tdsinglepara = options.fetch(:tdsinglepara, nil)
       end
 
@@ -22,11 +23,13 @@ module Coradoc
       end
 
       def to_adoc
+        title = @title.nil? ? "" : ".#{Coradoc::Generator.gen_adoc(@title)}\n"
         anchor = @anchor.nil? ? "" : "#{@anchor.to_adoc}\n"
+        attrs = @attributes.nil? ? "" : "#{@attributes.to_adoc}\n"
         if @tdsinglepara
-          anchor.to_s << Coradoc::Generator.gen_adoc(@content).strip
+          "#{title}#{anchor}" << Coradoc.strip_unicode(Coradoc::Generator.gen_adoc(@content))
         else
-          "\n\n#{anchor}" << Coradoc::Generator.gen_adoc(@content).strip << "\n\n"
+          "\n\n#{title}#{anchor}#{attrs}" << Coradoc.strip_unicode(Coradoc::Generator.gen_adoc(@content)) << "\n\n"
         end
       end
     end

data/lib/coradoc/element/revision.rb

@@ -19,7 +19,7 @@ module Coradoc
         elsif @date.nil?
           "#{@number}: #{@remark}\n"
         else
-          "#{@number}, #{@date}: #{@revision}\n"
+          "#{@number}, #{@date}: #{@remark}\n"
         end
       end
     end

data/lib/coradoc/element/section.rb

@@ -1,7 +1,7 @@
 module Coradoc
   module Element
     class Section < Base
-      attr_accessor :id, :title, :contents, :sections
+      attr_accessor :id, :title, :attrs, :contents, :sections
 
       declare_children :id, :title, :contents, :sections
 
@@ -9,9 +9,10 @@ module Coradoc
         @title = title
         @id = options.fetch(:id, nil)
         @id = nil if @id == ""
+        @anchor = @id.nil? ? nil : Inline::Anchor.new(@id)
+        @attrs = options.fetch(:attribute_list, "")
         @contents = options.fetch(:contents, [])
         @sections = options.fetch(:sections, [])
-        @anchor = @id.nil? ? nil : Inline::Anchor.new(@id)
       end
 
       def glossaries
@@ -27,6 +28,7 @@ module Coradoc
       def to_adoc
         anchor = @anchor.nil? ? "" : "#{@anchor.to_adoc}\n"
         title = Coradoc::Generator.gen_adoc(@title)
+        attrs = @attrs.to_s.empty? ? "" : "#{@attrs.to_adoc}\n"
         content = Coradoc::Generator.gen_adoc(@contents)
         sections = Coradoc::Generator.gen_adoc(@sections)
 
@@ -34,7 +36,13 @@ module Coradoc
         # with something.
         content = "&nbsp;#{content}" if content.start_with?(" +\n")
 
-        "\n#{anchor}" << title << content << sections << "\n"
+        # Only try to postprocess elements that are text,
+        # otherwise we could strip markup.
+        if Coradoc.a_single?(@contents, Coradoc::Element::TextElement)
+          content = Coradoc.strip_unicode(content)
+        end
+
+        "\n#{anchor}" << attrs << title << content << sections << "\n"
       end
 
       # Check for cases when Section is simply an equivalent of an empty <DIV>

data/lib/coradoc/element/table.rb

@@ -10,7 +10,7 @@ module Coradoc
         @title = title
         @id = options.fetch(:id, nil)
         @anchor = @id.nil? ? nil : Inline::Anchor.new(@id)
-        @attrs = options.fetch(:attrs, "")
+        @attrs = options.fetch(:attributes, nil)
       end
 
       def to_adoc
@@ -76,6 +76,11 @@ module Coradoc
           anchor = @anchor.nil? ? "" : @anchor.to_adoc.to_s
           content = simplify_block_content(@content)
           content = Coradoc::Generator.gen_adoc(content)
+          # Only try to postprocess elements that are text,
+          # otherwise we could strip markup.
+          if Coradoc.a_single?(@content, Coradoc::Element::TextElement)
+            content = Coradoc.strip_unicode(content)
+          end
           "#{@colrowattr}#{@alignattr}#{@style}| #{anchor}#{content}"
         end
       end

data/lib/coradoc/element/tag.rb

@@ -0,0 +1,19 @@
+module Coradoc
+  module Element
+    class Tag < Base
+      attr_accessor :name, :prefix, :attrs, :line_break
+
+      def initialize(name, options = {})
+        @name = name
+        @prefix = options.fetch(:prefix, "tag")
+        @attrs = options.fetch(:attribute_list, AttributeList.new)
+        @line_break = options.fetch(:line_break, "\n")
+      end
+
+      def to_adoc
+        attrs = @attrs.to_adoc
+        "// #{@prefix}::#{@name}#{attrs}#{@line_break}"
+      end
+    end
+  end
+end

data/lib/coradoc/element/term.rb

@@ -0,0 +1,21 @@
+module Coradoc
+  module Element
+    class Term < Base
+      attr_accessor :term, :options
+
+      declare_children :term, :options
+
+      def initialize(term, options = {})
+        @term = term
+        @type = options.fetch(:type, nil)
+        @lang = options.fetch(:lang, :en)
+        @line_break = options.fetch(:line_break, "")
+      end
+
+      def to_adoc
+        return "#{@type}:[#{@term}]#{@line_break}" if @lang == :en
+        "[#{@type}]##{@term}##{@line_break}"
+      end
+    end
+  end
+end

data/lib/coradoc/element/text_element.rb

@@ -9,10 +9,14 @@ module Coradoc
         @content = content # .to_s
         @id = options.fetch(:id, nil)
         @line_break = options.fetch(:line_break, "")
+        @html_cleanup = options.fetch(:html_cleanup, false)
+        if @html_cleanup
+          @content = treat_text_to_adoc(@content)
+        end
       end
 
       def to_adoc
-        Coradoc::Generator.gen_adoc(treat_text_to_adoc(@content))
+        Coradoc::Generator.gen_adoc(@content) + @line_break
       end
 
       def treat_text_to_adoc(text)
@@ -64,6 +68,10 @@ module Coradoc
       def initialize(line_break)
         @line_break = line_break
       end
+
+      def to_adoc
+        @line_break
+      end
     end
 
     class Highlight < Element::TextElement

data/lib/coradoc/element/title.rb

@@ -7,7 +7,7 @@ module Coradoc
 
       def initialize(content, level, options = {})
         @level_int = level
-        @level_int = level.length if level.is_a?(String)
+        # @level_int = level.length - 1 if level.is_a?(String)
         @content = content
         @id = options.fetch(:id, nil)
         @anchor = @id.nil? ? nil : Inline::Anchor.new(@id)
@@ -21,7 +21,7 @@ module Coradoc
 
       def to_adoc
         anchor = @anchor.nil? ? "" : "#{@anchor.to_adoc}\n"
-        content = Coradoc::Generator.gen_adoc(@content)
+        content = Coradoc.strip_unicode(Coradoc::Generator.gen_adoc(@content))
         <<~HERE
 
         #{anchor}#{style_str}#{level_str} #{content}

data/lib/coradoc/generator.rb

@@ -11,6 +11,8 @@ module Coradoc
         content
       elsif content.nil?
         ""
+      elsif content.is_a?(Parslet::Slice)
+        content.to_s
       end
     end
   end

data/lib/coradoc/input/adoc.rb

@@ -0,0 +1,28 @@
+require "coradoc/input"
+
+module Coradoc
+  module Input::Adoc
+    def self.processor_id
+      :adoc
+    end
+
+    def self.processor_match?(filename)
+      %w[.adoc].any? { |i| filename.downcase.end_with?(i) }
+    end
+
+    def self.processor_execute(input, _options = {})
+      ast = Coradoc::Parser::Base.new.parse(input)
+      Coradoc::Transformer.transform(ast[:document])
+    end
+
+    def self.processor_postprocess(input, options)
+      if options[:output_processor] == :adoc
+        Coradoc::Input::HTML::Cleaner.new.tidy(input)
+      else
+        input
+      end
+    end
+
+    Coradoc::Input.define(self)
+  end
+end

data/lib/coradoc/input/docx.rb

@@ -0,0 +1,35 @@
+require "word-to-markdown"
+require "coradoc/input/html"
+require "fileutils"
+
+module Coradoc
+  module Input::Docx
+    def self.processor_id
+      :docx
+    end
+
+    def self.processor_match?(filename)
+      %w[.docx .doc].any? { |i| filename.downcase.end_with?(i) }
+    end
+
+    def self.processor_execute(input, options = {})
+      image_dir = Dir.mktmpdir
+      options = options.merge(sourcedir: image_dir)
+      doc = WordToMarkdown.new(input, image_dir)
+      doc = Coradoc::Input::HTML.cleaner.preprocess_word_html(doc.document.html)
+      options = WordToMarkdown::REVERSE_MARKDOWN_OPTIONS.merge(options)
+      Coradoc::Input::HTML.to_coradoc(doc, options)
+    ensure
+      FileUtils.rm_rf(image_dir)
+    end
+
+    def self.processor_postprocess(data, options)
+      Coradoc::Input::HTML.processor_postprocess(data, options)
+    end
+
+    # This processor prefers to work on original files.
+    def self.processor_wants_filenames; true; end
+
+    Coradoc::Input.define(self)
+  end
+end

data/lib/coradoc/{reverse_adoc → input/html}/README.adoc

@@ -1,4 +1,4 @@
-= AsciiDoc from HTML and Microsoft Word: reverse_adoc
+= AsciiDoc from HTML and Microsoft Word: formerly reverse_adoc
 
 == Purpose
 
@@ -253,9 +253,9 @@ Simple to use.
 
 [source,ruby]
 ----
-require 'coradoc/reverse_adoc'
+require 'coradoc/input/html'
 
-result = Coradoc::ReverseAdoc.convert input
+result = Coradoc::Input::HTML.convert input
 result.inspect # " *feelings* "
 ----
 
@@ -265,9 +265,9 @@ Just pass your chosen configuration options in after the input. The given option
 
 [source,ruby]
 ----
-require 'coradoc/reverse_adoc'
+require 'coradoc/input/html'
 
-Coradoc::ReverseAdoc.convert(input, unknown_tags: :raise, mathml2asciimath: true)
+Coradoc::Input::HTML.convert(input, unknown_tags: :raise, mathml2asciimath: true)
 ----
 
 
@@ -277,9 +277,9 @@ Or configure it block style on a initializer level. These configurations will la
 
 [source,ruby]
 ----
-require 'coradoc/reverse_adoc'
+require 'coradoc/input/html'
 
-Coradoc::ReverseAdoc.config do |config|
+Coradoc::Input::HTML.config do |config|
   config.unknown_tags      = :bypass
   config.mathml2asciimath  = true
   config.tag_border  = ''
@@ -290,10 +290,10 @@ end
 
 [source,ruby]
 ----
-require 'coradoc/reverse_adoc'
+require 'coradoc/input/html'
 
 # Options can be supplied as keyword arguments
-Coradoc::ReverseAdoc::HtmlConverter.to_coradoc("<b><i>Some input</i></b>")
+Coradoc::Input::HTML::HtmlConverter.to_coradoc("<b><i>Some input</i></b>")
 ----
 
 

data/lib/coradoc/{reverse_adoc → input/html}/cleaner.rb

@@ -1,6 +1,10 @@
-module Coradoc::ReverseAdoc
+module Coradoc::Input::HTML
   class Cleaner
     def tidy(string)
+      if string.is_a? Hash
+        return string.transform_values { |i| tidy(i) }
+      end
+
       result = HtmlConverter.track_time "Removing inner whitespace" do
         remove_inner_whitespaces(String.new(string))
       end
@@ -16,6 +20,16 @@ module Coradoc::ReverseAdoc
       result = HtmlConverter.track_time "Cleaning punctuation characters" do
         clean_punctuation_characters(result)
       end
+      result = remove_block_leading_newlines(result)
+      result = remove_section_attribute_newlines(result)
+    end
+
+    def remove_block_leading_newlines(string)
+      string.gsub("]\n****\n\n", "]\n****\n")
+    end
+
+    def remove_section_attribute_newlines(string)
+      string.gsub("]\n\n==", "]\n==")
     end
 
     def remove_newlines(string)
@@ -47,20 +61,20 @@ module Coradoc::ReverseAdoc
     # Same for underscores and brackets.
     def clean_tag_borders(string)
       # result = string.gsub(/\s?\*{2,}.*?\*{2,}\s?/) do |match|
-      # preserve_border_whitespaces(match, default_border: Coradoc::ReverseAdoc.config.tag_border) do
+      # preserve_border_whitespaces(match, default_border: Coradoc::Input::HTML.config.tag_border) do
       #   match.strip.sub("** ", "**").sub(" **", "**")
       # end
       # end
 
       # result = string.gsub(/\s?_{2,}.*?_{2,}\s?/) do |match|
-      #   preserve_border_whitespaces(match, default_border: Coradoc::ReverseAdoc.config.tag_border) do
+      #   preserve_border_whitespaces(match, default_border: Coradoc::Input::HTML.config.tag_border) do
       #     match.strip.sub("__ ", "__").sub(" __", "__")
       #   end
       # end
 
       result = string.gsub(/\s?~{2,}.*?~{2,}\s?/) do |match|
         preserve_border_whitespaces(match,
-                                    default_border: Coradoc::ReverseAdoc.config.tag_border) do
+                                    default_border: Coradoc::Input::HTML.config.tag_border) do
           match.strip.sub("~~ ", "~~").sub(" ~~", "~~")
         end
       end
@@ -83,8 +97,7 @@ module Coradoc::ReverseAdoc
 
     def scrub_whitespace(string)
       string.gsub!(/ | |\u00a0/i, " ") # HTML encoded spaces
-      string.sub!(/^\A[[:space:]]+/m, "") # document leading whitespace
-      string.sub!(/[[:space:]]+\z$/m, "") # document trailing whitespace
+      string = Coradoc.strip_unicode(string) # Strip document-level leading and trailing whitespace
       string.gsub!(/( +)$/, " ") # line trailing whitespace
       string.gsub!(/\n\n\n\n/, "\n\n") # Quadruple line breaks
       # string.delete!('?| ')               # Unicode non-breaking spaces, injected as tabs

data/lib/coradoc/input/html/config.rb

@@ -0,0 +1,75 @@
+require "tmpdir"
+
+module Coradoc
+  module Input::HTML
+    class Config
+      def initialize
+        @unknown_tags     = :pass_through
+        @input_format     = :html
+        @mathml2asciimath = false
+        @external_images  = false
+
+        # Destination to save file and images
+        @destination      = nil
+
+        # Source of HTML
+        # @sourcedir        = nil
+
+        # Image counter, assuming there are max 999 images
+        @image_counter = 1
+        # pad with 0s
+        @image_counter_pattern = "%03d"
+
+        @em_delimiter     = "_".freeze
+        @strong_delimiter = "*".freeze
+        @inline_options   = {}
+        @tag_border       = " ".freeze
+
+        @split_sections   = nil
+
+        # Document width - used to compute table sizes.
+        # This is an assumption for screen size in input document.
+        # If column widths are specified in absolute values, then we
+        # have to convert them to relative values, as AsciiDoc only
+        # supports those.
+        @doc_width        = 1000
+
+        # Plugin system
+        @plugins          = []
+
+        # Debugging options
+        @track_time       = false
+      end
+
+      def with(options = {})
+        old_options = @inline_options
+        @inline_options = options
+        result = yield
+        @inline_options = old_options
+        result
+      end
+
+      def self.declare_option(option)
+        define_method(option) do
+          @inline_options[option] || instance_variable_get(:"@#{option}")
+        end
+
+        attr_writer option
+      end
+
+      declare_option :unknown_tags
+      declare_option :tag_border
+      declare_option :mathml2asciimath
+      declare_option :external_images
+      declare_option :destination
+      declare_option :sourcedir
+      declare_option :image_counter
+      declare_option :image_counter_pattern
+      declare_option :input_format
+      declare_option :split_sections
+      declare_option :doc_width
+      declare_option :plugins
+      declare_option :track_time
+    end
+  end
+end

data/lib/coradoc/{reverse_adoc → input/html}/converters/a.rb

@@ -1,6 +1,6 @@
 require "coradoc"
 
-module Coradoc::ReverseAdoc
+module Coradoc::Input::HTML
   module Converters
     class A < Base
       def to_coradoc(node, state = {})

data/lib/coradoc/{reverse_adoc → input/html}/converters/aside.rb

@@ -1,4 +1,4 @@
-module Coradoc::ReverseAdoc
+module Coradoc::Input::HTML
   module Converters
     class Aside < Base
       def to_coradoc(node, state = {})

data/lib/coradoc/{reverse_adoc → input/html}/converters/audio.rb

@@ -1,4 +1,4 @@
-module Coradoc::ReverseAdoc
+module Coradoc::Input::HTML
   module Converters
     class Audio < Base
       def to_coradoc(node, _state = {})

data/lib/coradoc/{reverse_adoc → input/html}/converters/base.rb

@@ -1,4 +1,4 @@
-module Coradoc::ReverseAdoc
+module Coradoc::Input::HTML
   module Converters
     class Base
       # Default implementation to convert a given Nokogiri node

data/lib/coradoc/{reverse_adoc → input/html}/converters/blockquote.rb

@@ -1,4 +1,4 @@
-module Coradoc::ReverseAdoc
+module Coradoc::Input::HTML
   module Converters
     class Blockquote < Base
       def to_coradoc(node, state = {})
@@ -7,7 +7,7 @@ module Coradoc::ReverseAdoc
         attributes = Coradoc::Element::AttributeList.new
         attributes.add_positional("quote", cite) if !cite.nil?
         content = treat_children(node, state).strip
-        content = Coradoc::ReverseAdoc.cleaner.remove_newlines(content)
+        content = Coradoc::Input::HTML.cleaner.remove_newlines(content)
         Coradoc::Element::Block::Quote.new(nil, lines: content,
                                                 attributes: attributes)
       end

data/lib/coradoc/{reverse_adoc → input/html}/converters/br.rb

@@ -1,4 +1,4 @@
-module Coradoc::ReverseAdoc
+module Coradoc::Input::HTML
   module Converters
     class Br < Base
       def to_coradoc(_node, _state = {})

data/lib/coradoc/{reverse_adoc → input/html}/converters/bypass.rb

@@ -1,4 +1,4 @@
-module Coradoc::ReverseAdoc
+module Coradoc::Input::HTML
   module Converters
     class Bypass < Base
       def to_coradoc(node, state = {})

data/lib/coradoc/{reverse_adoc → input/html}/converters/code.rb

@@ -1,4 +1,4 @@
-module Coradoc::ReverseAdoc
+module Coradoc::Input::HTML
   module Converters
     class Code < Markup
       def coradoc_class

data/lib/coradoc/{reverse_adoc → input/html}/converters/div.rb

@@ -1,4 +1,4 @@
-module Coradoc::ReverseAdoc
+module Coradoc::Input::HTML
   module Converters
     class Div < Base
       def to_coradoc(node, state = {})

data/lib/coradoc/{reverse_adoc → input/html}/converters/dl.rb

@@ -1,4 +1,4 @@
-module Coradoc::ReverseAdoc
+module Coradoc::Input::HTML
   module Converters
     class Dl < Base
       def to_coradoc(node, state = {})

data/lib/coradoc/{reverse_adoc → input/html}/converters/drop.rb

@@ -1,4 +1,4 @@
-module Coradoc::ReverseAdoc
+module Coradoc::Input::HTML
   module Converters
     class Drop < Base
       def to_coradoc(node, state = {})

data/lib/coradoc/{reverse_adoc → input/html}/converters/em.rb

@@ -1,4 +1,4 @@
-module Coradoc::ReverseAdoc
+module Coradoc::Input::HTML
   module Converters
     class Em < Markup
       def coradoc_class

data/lib/coradoc/{reverse_adoc → input/html}/converters/figure.rb

@@ -1,4 +1,4 @@
-module Coradoc::ReverseAdoc
+module Coradoc::Input::HTML
   module Converters
     class Figure < Base
       def to_coradoc(node, state = {})

data/lib/coradoc/{reverse_adoc → input/html}/converters/h.rb

@@ -1,4 +1,4 @@
-module Coradoc::ReverseAdoc
+module Coradoc::Input::HTML
   module Converters
     class H < Base
       def to_coradoc(node, state = {})

data/lib/coradoc/{reverse_adoc → input/html}/converters/head.rb

@@ -1,4 +1,4 @@
-module Coradoc::ReverseAdoc
+module Coradoc::Input::HTML
   module Converters
     class Head < Base
       def to_coradoc(node, _state = {})

data/lib/coradoc/{reverse_adoc → input/html}/converters/hr.rb

@@ -1,4 +1,4 @@
-module Coradoc::ReverseAdoc
+module Coradoc::Input::HTML
   module Converters
     class Hr < Base
       def to_coradoc(_node, _state = {})

data/lib/coradoc/{reverse_adoc → input/html}/converters/ignore.rb

@@ -1,4 +1,4 @@
-module Coradoc::ReverseAdoc
+module Coradoc::Input::HTML
   module Converters
     class Ignore < Base
       def to_coradoc(node, state = {})