coradoc 1.1.2 → 1.1.4

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: b6e77fc4eb6d79071cb1d309530adbd49a4364265b446d139c18eac2da3bce54
-  data.tar.gz: 07a84837196bde12cc9d91ccd5084b8a173a4da5c237eeaea06a6d786c2f876c
+  metadata.gz: e528645f0bdb38f707239ed1862dc74944dc7eba4149cf9f1243458e98591edb
+  data.tar.gz: 268acd80823a507d3a86e83bfeaef55a8145819822aac9903f0bc96cee29d55c
 SHA512:
-  metadata.gz: 69d2d12389a4e254b5ee8f4d67d3ddd406abeabebe1874b343acb4cd7c560c35284c8a58a4d2ee58cc788cfafd68989e88e783aa79fa3c32dc0f9dd389a276ac
-  data.tar.gz: 7f034ef9a649446198793d8d8221a1bba5e22069f2feec85c5d30529170f8a87315540f481bc8b1219c03aaba78e42de67bbf94e25fd3e4ab93b6f221a356ca4
+  metadata.gz: 624f718300d0877f0d610a3fbf953b324d5bfd6a73c68de30db69ddc75b45e1b396eeb5bdfdcfc7d0ecac2d50c8301d7011f118a2d23d4ac02d11c2e54e92e6a
+  data.tar.gz: 849b5851b2ca0b37313d8f7c743defbc168f698284c694c0a467536dbffd9cc10f5ec263fd2c28e1f287a926c6acbe3756c8753f6f3ae303dc61fc9678981ff1

data/lib/coradoc/element/attribute_list.rb

@@ -12,6 +12,16 @@ module Coradoc
         @rejected_named = []
       end
 
+      def inspect
+        "AttributeList: " +
+          [
+            @positional.map(&:inspect).join(", "),
+            @named.map { |k, v| "#{k}: #{v.inspect}" }.join(", "),
+            (@rejected_positional.empty? or "rejected: #{@rejected_positional.inspect}"),
+            (@rejected_positional.empty? or "rejected: #{@rejected_named.inspect}"),
+          ].reject { |i| i == true || i.empty? }.join(", ")
+      end
+
       def add_positional(*attr)
         @positional += attr
       end
@@ -65,7 +75,9 @@ module Coradoc
 
         adoc = +""
         if !@positional.empty?
-          adoc << @positional.map { |p| [nil, ""].include?(p) ? '""' : p }.join(",")
+          adoc << @positional.map do |p|
+            [nil, ""].include?(p) ? '""' : p
+          end.join(",")
         end
         adoc << "," if @positional.any? && @named.any?
         adoc << @named.map do |k, v|

data/lib/coradoc/element/base.rb

@@ -14,6 +14,8 @@ module Coradoc
           when Coradoc::Element::Section
             return content unless i.safe_to_collapse?
 
+            collected_content << i.anchor if i.anchor
+
             simplified = simplify_block_content(i.contents)
 
             if simplified && !simplified.empty?

data/lib/coradoc/element/block/core.rb

@@ -61,12 +61,13 @@ module Coradoc
 
         def type_hash
           @type_hash ||= {
-            "____" => :quote,
-            "****" => :side,
-            "----" => :source,
             "====" => :example,
             "...." => :literal,
+            "--"   => :open,
             "++++" => :pass,
+            "____" => :quote,
+            "****" => :side,
+            "----" => :source,
           }
         end
       end

data/lib/coradoc/element/block/example.rb

@@ -15,7 +15,7 @@ module Coradoc
         end
 
         def to_adoc
-          "\n\n#{gen_anchor}#{gen_title}#{gen_delimiter}\n" << gen_lines << "\n#{gen_delimiter}\n\n"
+          "\n\n#{gen_anchor}#{gen_title}#{gen_attributes}#{gen_delimiter}\n" << gen_lines << "\n#{gen_delimiter}\n\n"
         end
       end
     end

data/lib/coradoc/element/block/listing.rb

@@ -0,0 +1,21 @@
+module Coradoc
+  module Element
+    module Block
+      class Listing < Core
+        def initialize(_title, options = {})
+          @id = options.fetch(:id, nil)
+          @anchor = @id.nil? ? nil : Inline::Anchor.new(@id)
+          @lang = options.fetch(:lang, "")
+          @attributes = options.fetch(:attributes, AttributeList.new)
+          @lines = options.fetch(:lines, [])
+          @delimiter_char = "-"
+          @delimiter_len = options.fetch(:delimiter_len, 4)
+        end
+
+        def to_adoc
+          "\n\n#{gen_anchor}#{gen_attributes}\n#{gen_delimiter}\n" << gen_lines << "\n#{gen_delimiter}\n\n"
+        end
+      end
+    end
+  end
+end

data/lib/coradoc/element/block/literal.rb

@@ -2,16 +2,18 @@ module Coradoc
   module Element
     module Block
       class Literal < Core
-        def initialize(_title, options = {})
+        def initialize(title, options = {})
+          @title = title
           @id = options.fetch(:id, nil)
           @anchor = @id.nil? ? nil : Inline::Anchor.new(@id)
+          @attributes = options.fetch(:attributes, AttributeList.new)
           @lines = options.fetch(:lines, [])
           @delimiter_char = "."
           @delimiter_len = options.fetch(:delimiter_len, 4)
         end
 
         def to_adoc
-          "\n\n#{gen_anchor}#{gen_delimiter}\n" << gen_lines << "\n#{gen_delimiter}\n\n"
+          "\n\n#{gen_anchor}#{gen_title}#{gen_attributes}#{gen_delimiter}\n" << gen_lines << "\n#{gen_delimiter}\n\n"
         end
       end
     end

data/lib/coradoc/element/block/open.rb

@@ -0,0 +1,22 @@
+module Coradoc
+  module Element
+    module Block
+      class Open < Core
+        def initialize(title, options = {})
+          @title = title
+          @id = options.fetch(:id, nil)
+          @anchor = @id.nil? ? nil : Inline::Anchor.new(@id)
+          @lang = options.fetch(:lang, "")
+          @attributes = options.fetch(:attributes, AttributeList.new)
+          @lines = options.fetch(:lines, [])
+          @delimiter_char = "-"
+          @delimiter_len = options.fetch(:delimiter_len, 2)
+        end
+
+        def to_adoc
+          "\n\n#{gen_anchor}#{gen_attributes}#{gen_delimiter}\n" << gen_lines << "\n#{gen_delimiter}\n\n"
+        end
+      end
+    end
+  end
+end

data/lib/coradoc/element/block.rb

@@ -8,8 +8,10 @@ end
 require_relative "block/core"
 require_relative "block/example"
 require_relative "block/literal"
-require_relative "block/quote"
+require_relative "block/listing"
+require_relative "block/open"
 require_relative "block/pass"
+require_relative "block/quote"
 require_relative "block/side"
 require_relative "block/sourcecode"
 require_relative "block/reviewer_comment"

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

@@ -19,8 +19,8 @@ module Coradoc
             m = @items.select do |i|
               i.is_a?(Coradoc::Element::ListItem) &&
                 !i.marker.nil?
-            end.first&.marker
-            @ol_count = m.size if m.is_a?(String)
+            end.first&.marker.to_s
+            @ol_count = m.size
           end
           @ol_count = 1 if @ol_count.nil?
           @attrs = options.fetch(:attrs, AttributeList.new)

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

@@ -7,6 +7,7 @@ module Coradoc
         end
 
         def prefix
+          return @marker if @marker
           "." * [@ol_count, 1].max
         end
       end

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

@@ -7,6 +7,7 @@ module Coradoc
         end
 
         def prefix
+          return @marker if @marker
           "*" * [@ol_count, 1].max
         end
       end

data/lib/coradoc/element/list_item.rb

@@ -1,7 +1,7 @@
 module Coradoc
   module Element
     class ListItem < Base
-      attr_accessor :marker, :id, :anchor, :content, :line_break
+      attr_accessor :marker, :id, :anchor, :content, :subitem, :line_break
 
       declare_children :content, :id, :anchor
 
@@ -10,11 +10,14 @@ module Coradoc
         @id = options.fetch(:id, nil)
         @anchor = @id.nil? ? nil : Inline::Anchor.new(@id)
         @content = content
+        @attached = options.fetch(:attached, [])
+        @nested = options.fetch(:nested, nil)
         @line_break = options.fetch(:line_break, "\n")
       end
 
       def to_adoc
-        anchor = @anchor.nil? ? "" : @anchor.to_adoc.to_s
+        anchor = @anchor.nil? ? "" : " #{@anchor.to_adoc.to_s} "
+        # text = Coradoc::Generator.gen_adoc(@content)
         content = Array(@content).map do |subitem|
           next if subitem.is_a? Inline::HardLineBreak
 
@@ -24,10 +27,15 @@ module Coradoc
           if Coradoc.a_single?(subitem, Coradoc::Element::TextElement)
             subcontent = Coradoc.strip_unicode(subcontent)
           end
-          subcontent.chomp
+          subcontent
         end.compact.join("\n+\n")
-
-        " #{anchor}#{content.chomp}#{@line_break}"
+        # attach = Coradoc::Generator.gen_adoc(@attached)
+        attach = @attached.map do |elem|
+          "+\n" + Coradoc::Generator.gen_adoc(elem)
+        end.join
+        nest = Coradoc::Generator.gen_adoc(@nested)
+        out = " #{anchor}#{content}#{@line_break}"
+        out + attach + nest
       end
     end
   end

data/lib/coradoc/element/section.rb

@@ -1,7 +1,7 @@
 module Coradoc
   module Element
     class Section < Base
-      attr_accessor :id, :title, :attrs, :contents, :sections
+      attr_accessor :id, :title, :attrs, :contents, :sections, :anchor
 
       declare_children :id, :title, :contents, :sections
 
@@ -49,7 +49,7 @@ module Coradoc
       # HTML element and if it happens inside some other block element, can be
       # safely collapsed.
       def safe_to_collapse?
-        @title.nil? && @id.nil? && @sections.empty?
+        @title.nil? && @sections.empty?
       end
 
       private

data/lib/coradoc/element/text_element.rb

@@ -15,6 +15,15 @@ module Coradoc
         end
       end
 
+      def inspect
+        str = "TextElement"
+        str += "(#{@id})" if @id
+        str += ": "
+        str += @content.inspect
+        str += " + #{@line_break.inspect}" unless line_break.to_s.empty?
+        str
+      end
+
       def to_adoc
         Coradoc::Generator.gen_adoc(@content) + @line_break
       end

data/lib/coradoc/input/html/converters/base.rb

@@ -72,14 +72,14 @@ module Coradoc::Input::HTML
         leading_whitespace = $1
         if !leading_whitespace.nil?
           first_text = node.at_xpath("./text()[1]")
-          first_text.replace(first_text.text.lstrip)
+          first_text.replace(first_text.text.lstrip) if first_text
           leading_whitespace = " "
         end
         node.text =~ /(\s+)$/
         trailing_whitespace = $1
         if !trailing_whitespace.nil?
           last_text = node.at_xpath("./text()[last()]")
-          last_text.replace(last_text.text.rstrip)
+          last_text.replace(last_text.text.rstrip) if last_text
           trailing_whitespace = " "
         end
         [leading_whitespace, trailing_whitespace]

data/lib/coradoc/input/html/converters/div.rb

@@ -10,5 +10,6 @@ module Coradoc::Input::HTML
 
     register :div,     Div.new
     register :article, Div.new
+    register :center,  Div.new
   end
 end

data/lib/coradoc/input/html/converters/table.rb

@@ -114,6 +114,12 @@ module Coradoc::Input::HTML
           columns = row.xpath("./td | ./th")
           column_id = 0
 
+          cell_references[i] ||= []
+          cell_matrix[i] ||= []
+
+          # Empty row support: pass row object via an instance variable
+          cell_references[i].instance_variable_set(:@row_obj, row)
+
           columns.each do |cell|
             colspan = cell["colspan"]&.to_i || 1
             rowspan = cell["rowspan"]&.to_i || 1
@@ -179,7 +185,7 @@ module Coradoc::Input::HTML
           min_rows.each do |row|
             break if row.length != cpr_min
 
-            row_obj = row.last.first.parent
+            row_obj = row.last&.first&.parent || row.instance_variable_get(:@row_obj)
             doc = row_obj.document
             added_node = Nokogiri::XML::Node.new("td", doc)
             added_node["x-added"] = "x-added"

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

@@ -4,6 +4,8 @@ module Coradoc::Input::HTML
   # is compatible with what we would get out of Coradoc, if
   # it parsed it directly.
   class Postprocessor
+    Element = Coradoc::Element
+
     def self.process(coradoc)
       new(coradoc).process
     end
@@ -12,17 +14,74 @@ module Coradoc::Input::HTML
       @tree = coradoc
     end
 
+    # Extracts titles from lists. This happens in HTML files
+    # generated from DOCX documents by LibreOffice.
+    #
+    # We are interested in a particular tree:
+    # Element::List::Ordered items:
+    #   Element::List::Ordered items: (any depth)
+    #     Element::ListItem content:
+    #       Element::Title
+    #       (any number of other titles of the same scheme)
+    #
+    # This tree is flattened into:
+    # Element::Title
+    # Element::Title (any number of titles)
+    def extract_titles_from_lists
+      @tree = Element::Base.visit(@tree) do |elem, dir|
+        next elem unless dir == :pre
+        next elem unless elem.is_a?(Element::List::Ordered)
+        next elem if elem.items.length != 1
+
+        anchors = []
+        anchors << elem.anchor if elem.anchor
+
+        # Extract ListItem from any depth of List::Ordered
+        processed = elem
+        while processed.is_a?(Element::List::Ordered)
+          if processed.items.length != 1
+            backtrack = true
+            break
+          end
+          anchors << processed.anchor if processed.anchor
+          processed = processed.items.first
+        end
+
+        # Something went wrong? Anything not matching on the way?
+        next elem if backtrack
+        next elem unless processed.is_a?(Element::ListItem)
+
+        anchors << processed.anchor if processed.anchor
+
+        # Now we must have a title (or titles).
+        titles = processed.content.flatten
+
+        # Don't bother if there's no title in there.
+        next elem unless titles.any? { |i| i.is_a? Element::Title }
+
+        # Ordered is another iteration for our cleanup.
+        next elem unless titles.all? do |i|
+          i.is_a?(Element::Title) || i.is_a?(Element::List::Ordered)
+        end
+
+        # We are done now.
+        titles + anchors
+      end
+    end
+
     # Collapse DIVs that only have a title, or nest another DIV.
     def collapse_meaningless_sections
-      @tree = Coradoc::Element::Base.visit(@tree) do |elem, _dir|
-        if elem.is_a?(Coradoc::Element::Section) && elem.safe_to_collapse?
+      @tree = Element::Base.visit(@tree) do |elem, _dir|
+        if elem.is_a?(Element::Section) && elem.safe_to_collapse?
           children_classes = Array(elem.contents).map(&:class)
           count = children_classes.length
-          safe_classes = [Coradoc::Element::Section, Coradoc::Element::Title]
+          safe_classes = [Element::Section, Element::Title]
 
           # Count > 0 because some documents use <div> as a <br>.
           if count > 0 && children_classes.all? { |i| safe_classes.include?(i) }
-            next elem.contents
+            contents = elem.contents.dup
+            contents.prepend(elem.anchor) if elem.anchor
+            next contents
           end
         end
         elem
@@ -32,12 +91,14 @@ module Coradoc::Input::HTML
     # tree should now be more cleaned up, so we can progress with
     # creating meaningful sections
     def generate_meaningful_sections
-      @tree = Coradoc::Element::Base.visit(@tree) do |elem, dir|
+      @tree = Element::Base.visit(@tree) do |elem, dir|
         # We are searching for an array, that has a title. This
         # will be a candidate for our section array.
         if dir == :post &&
             elem.is_a?(Array) &&
-            !elem.grep(Coradoc::Element::Title).empty?
+            !elem.flatten.grep(Element::Title).empty?
+
+          elem = elem.flatten
 
           new_array = []
           content_array = new_array
@@ -47,12 +108,12 @@ module Coradoc::Input::HTML
           # all descendant sections into those sections. Otherwise, we push
           # an element as content of current section.
           elem.each do |e|
-            if e.is_a? Coradoc::Element::Title
+            if e.is_a? Element::Title
               title = e
               content_array = []
               section_array = []
               level = title.level_int
-              section = Coradoc::Element::Section.new(
+              section = Element::Section.new(
                 title, contents: content_array, sections: section_array
               )
               # Some documents may not be consistent and eg. follow H4 after
@@ -82,11 +143,11 @@ module Coradoc::Input::HTML
       previous_sections = {}
 
       determine_section_id = ->(elem) do
-        if elem.title.style == "appendix"
-          level = "A"
-        else
-          level = 1
-        end
+        level = if elem.title.style == "appendix"
+                  "A"
+                else
+                  1
+                end
 
         section = previous_sections[elem]
         while section
@@ -102,8 +163,8 @@ module Coradoc::Input::HTML
         style
       end
 
-      @tree = Coradoc::Element::Base.visit(@tree) do |elem, dir|
-        title = elem.title if elem.is_a?(Coradoc::Element::Section)
+      @tree = Element::Base.visit(@tree) do |elem, dir|
+        title = elem.title if elem.is_a?(Element::Section)
 
         if title && title.level_int <= max_level
           if dir == :pre
@@ -137,6 +198,7 @@ module Coradoc::Input::HTML
     end
 
     def process
+      extract_titles_from_lists
       collapse_meaningless_sections
       generate_meaningful_sections
       # Do it again to simplify the document further.

data/lib/coradoc/parser/asciidoc/attribute_list.rb

@@ -11,8 +11,13 @@ module Coradoc
           match('[^\],]').repeat(1)
         end
 
+        def named_key
+          (str('reviewer') | 
+            match('[a-zA-Z0-9_-]').repeat(1)).as(:named_key)
+        end
+
         def named_attribute
-          (match('[a-zA-Z0-9_-]').repeat(1).as(:named_key) >>
+          ( named_key >>
             str(' ').maybe >> str("=") >> str(' ').maybe >>
             match['a-zA-Z0-9_\- \"'].repeat(1).as(:named_value) >>
             str(' ').maybe
@@ -51,6 +56,7 @@ module Coradoc
         end
 
         def attribute_list(name = :attribute_list)
+          str('[').present? >>
           str('[') >> str("[").absent? >> 
           ( named_many |
             positional_one_named_many |