coradoc 1.0.0 → 1.1.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 71bfd2ac723d58ca75b591bceabcdf834abd9cbbf32e54b96352da21b8f15095
-  data.tar.gz: 1a7ea6bf80d64ab4c1349c948aa2aee9ed704dc36ddd07fb7cce78293dca86cd
+  metadata.gz: 5e44c5e565e224487496ecdd7be8c5c88e5f05fd2a42a4f2cbf31746a908aa25
+  data.tar.gz: a1e8fb651b29b516071e91c5c6f7bf72a68c34fd4ae9512a928d3ea3185bc0c6
 SHA512:
-  metadata.gz: 555800621a06ffafc07e03a5c2cdca1210de6369d524a606de51df78b1ee9b19aaba30d32d8a448347237e488362a59677781a02502a1d8891fb89aa1cc267f5
-  data.tar.gz: 22707b18c6ed99fb4c6127885d871875ed1fed2906d2d10dcdd5f4ed9bdd1b1bfe3f19fd092768ccc4d488bdb7e705aba9ba7ab948ceace9cc679f1ee733eddc
+  metadata.gz: 4df0a3edcebe6990006c7ca951b59f717f05d9b25eb6481394fe276dbaa318f6016e7a915d268abc19d61cec0605ab7d44466662d499cfa08a0ffa2f98bd8dcc
+  data.tar.gz: 9a44116b58e7e99a14ae856b1b06e23c26cbe2913f970a6849615282c3f8e33b05c538c39c72a827fa65aef30f56c9eb494d2f48574ea01f86bca858d9692e16

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

@@ -9,7 +9,7 @@ module Coradoc
         end
 
         def to_adoc
-          "\n\n#{gen_delimiter}" << gen_lines << "\n#{gen_delimiter}\n\n"
+          "\n\n#{gen_delimiter}\n" << gen_lines << "\n#{gen_delimiter}\n\n"
         end
       end
     end

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

@@ -12,6 +12,7 @@ module Coradoc
           @anchor = @id.nil? ? nil : Coradoc::Element::Inline::Anchor.new(@id)
           @src = src
           @attributes = options.fetch(:attributes, AttributeList.new)
+          @annotate_missing = options.fetch(:annotate_missing)
           @title = options.fetch(:title, nil)
           if @attributes.any?
             @attributes.validate_positional(VALIDATORS_POSITIONAL)
@@ -20,10 +21,11 @@ module Coradoc
         end
 
         def to_adoc
+          missing = "// FIXME: Missing image: #{@annotate_missing}\n" if @annotate_missing
           anchor = @anchor.nil? ? "" : "#{@anchor.to_adoc}\n"
           title = ".#{@title}\n" unless @title.to_s.empty?
           attrs = @attributes.to_adoc
-          [anchor, title, "image", @colons, @src, attrs].join("")
+          [missing, anchor, title, "image", @colons, @src, attrs].join("")
         end
 
         extend AttributeList::Matchers

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

@@ -25,8 +25,13 @@ module Coradoc
           @items.each do |item|
             c = Coradoc::Generator.gen_adoc(item)
             if !c.empty?
-              content << prefix.to_s
-              content << " " if c[0]!=" "
+              # If there's a list inside a list directly, we want to
+              # skip adding an empty list item.
+              # See: https://github.com/metanorma/coradoc/issues/96
+              unless item.is_a? List::Core
+                content << prefix.to_s
+                content << " " if c[0]!=" "
+              end
               content << c
             end
           end

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

@@ -1,7 +1,7 @@
 module Coradoc
   module Element
     class ListItem < Base
-      attr_accessor :id
+      attr_accessor :id, :content, :anchor
 
       declare_children :content, :id, :anchor
 
@@ -14,8 +14,16 @@ module Coradoc
       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? Coradoc::Element::Inline::HardLineBreak
+
+          subcontent = Coradoc::Generator.gen_adoc(subitem)
+          # Only try to postprocess elements that are text,
+          # otherwise we could strip markup.
+          if Coradoc.is_a_single?(subitem, Coradoc::Element::TextElement)
+            subcontent = Coradoc.strip_unicode(subcontent)
+          end
+          subcontent.chomp
+        end.compact.join("\n+\n")
 
         " #{anchor}#{content.chomp}\n"
       end

data/lib/coradoc/element/paragraph.rb

@@ -24,9 +24,9 @@ module Coradoc
       def to_adoc
         anchor = @anchor.nil? ? "" : "#{@anchor.to_adoc}\n"
         if @tdsinglepara
-          anchor.to_s << Coradoc::Generator.gen_adoc(@content).strip
+          anchor.to_s << Coradoc.strip_unicode(Coradoc::Generator.gen_adoc(@content))
         else
-          "\n\n#{anchor}" << Coradoc::Generator.gen_adoc(@content).strip << "\n\n"
+          "\n\n#{anchor}" << Coradoc.strip_unicode(Coradoc::Generator.gen_adoc(@content)) << "\n\n"
         end
       end
     end

data/lib/coradoc/element/section.rb

@@ -34,6 +34,12 @@ module Coradoc
         # with something.
         content = " #{content}" if content.start_with?(" +\n")
 
+        # Only try to postprocess elements that are text,
+        # otherwise we could strip markup.
+        if Coradoc.is_a_single?(@contents, Coradoc::Element::TextElement)
+          content = Coradoc.strip_unicode(content)
+        end
+
         "\n#{anchor}" << title << content << sections << "\n"
       end
 

data/lib/coradoc/element/table.rb

@@ -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.is_a_single?(@content, Coradoc::Element::TextElement)
+            content = Coradoc.strip_unicode(content)
+          end
           "#{@colrowattr}#{@alignattr}#{@style}| #{anchor}#{content}"
         end
       end

data/lib/coradoc/element/title.rb

@@ -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/reverse_adoc/cleaner.rb

@@ -83,8 +83,7 @@ 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/reverse_adoc/converters/img.rb

@@ -33,7 +33,13 @@ module Coradoc::ReverseAdoc
         # puts "image_dest_path: #{image_dest_path.to_s}"
         # puts "image_src_path: #{image_src_path.to_s}"
 
-        FileUtils.cp(image_src_path, image_dest_path)
+        if File.exist?(image_src_path)
+          FileUtils.cp(image_src_path, image_dest_path)
+        else
+          @annotate_missing = image_src_path
+          Kernel.warn "Image #{image_src_path} does not exist"
+        end
+
         image_number_increment
 
         image_dest_path.relative_path_from(dest_dir)
@@ -88,11 +94,12 @@ module Coradoc::ReverseAdoc
 
         if src
           Coradoc::Element::Image::BlockImage.new(title, id, src,
-                                                  attributes: attributes)
+                                                  attributes: attributes,
+                                                  annotate_missing: @annotate_missing)
         end
       end
     end
 
-    register :img, Img.new
+    register :img, Img
   end
 end

data/lib/coradoc/reverse_adoc/converters/table.rb

@@ -53,13 +53,17 @@ module Coradoc::ReverseAdoc
         rules_attr = rules(node)
         attrs.add_named("rules", rules_attr) if rules_attr
 
-        cols = ensure_row_column_integrity_and_get_column_sizes(node)
-        attrs.add_named("cols", cols)
-
-        # Header first rows can't span multiple riws - drop header if they do.
-        header = node.at_xpath(".//tr")
-        unless header.xpath("./td | ./th").all? { |i| [nil, "1", ""].include? i["rowspan"] }
-          attrs.add_named("options", ["noheader"])
+        # We can't, and shouldn't do those calculation if the table we are
+        # processing is empty.
+        unless empty?(node)
+          cols = ensure_row_column_integrity_and_get_column_sizes(node)
+          attrs.add_named("cols", cols)
+
+          # Header first rows can't span multiple riws - drop header if they do.
+          header = node.at_xpath(".//tr")
+          unless header.xpath("./td | ./th").all? { |i| [nil, "1", ""].include? i["rowspan"] }
+            attrs.add_named("options", ["noheader"])
+          end
         end
 
         # This line should be removed.
@@ -68,6 +72,10 @@ module Coradoc::ReverseAdoc
         attrs
       end
 
+      def empty?(node)
+        !node.at_xpath(".//td | .//th")
+      end
+
       def ensure_row_column_integrity_and_get_column_sizes(node)
         rows = node.xpath(".//tr")
         num_rows = rows.length
@@ -173,7 +181,9 @@ module Coradoc::ReverseAdoc
 
             row_obj = row.last.first.parent
             doc = row_obj.document
-            row_obj.add_child(Nokogiri::XML::Node.new("td", doc))
+            added_node = Nokogiri::XML::Node.new("td", doc)
+            added_node["x-added"] = "x-added"
+            row_obj.add_child(added_node)
 
             modified = true
           end
@@ -194,6 +204,21 @@ module Coradoc::ReverseAdoc
         end
 
         unless cell_matrix_correct
+          # It may be a special case that we need to add virtual cells at the
+          # beginning not the end of a row.
+          needs_recompute = false
+          cell_matrix.each do |row|
+            if row.compact.length != row.length
+              last_cell = row.last
+              if last_cell["x-added"]
+                last_cell.parent.prepend_child(last_cell)
+                needs_recompute = true
+              end
+            end
+          end
+          recompute.() if needs_recompute
+
+          # But otherwise... we've got a really nasty table.
           warn <<~WARNING.gsub("\n", " ")
             **** Couldn't construct a valid image of a table on line
             #{node.line}. We need that to reliably compute column

data/lib/coradoc/reverse_adoc/converters.rb

@@ -10,7 +10,9 @@ module Coradoc::ReverseAdoc
     end
 
     def self.lookup(tag_name)
-      @@converters[tag_name.to_sym] or default_converter(tag_name)
+      converter = @@converters[tag_name.to_sym] || default_converter(tag_name)
+      converter = converter.new if converter.respond_to? :new
+      converter
     end
 
     # Note: process won't run plugin hooks

data/lib/coradoc/reverse_adoc/plugins/plateau.rb

@@ -30,12 +30,20 @@ module Coradoc::ReverseAdoc
         html_tree_change_tag_name_by_css(".pitemdata", "h3")
         html_tree_change_tag_name_by_css(".sitemdata", "h4")
         html_tree_change_tag_name_by_css('td[bgcolor="#D0CECE"]', "th")
+        html_tree_change_tag_name_by_css('td[bgcolor="#d0cece"]', "th")
+        html_tree_change_tag_name_by_css('.framedata, .frame_container_box', 'aside')
+        html_tree_change_tag_name_by_css('.frame2data', 'pre')
+        # Assumption that all code snippets in those documents are XML...
+        html_tree_change_properties_by_css(".frame2data", class: "brush:xml;")
 
         # Remove some CSS ids that are not important to us
         html_tree_change_properties_by_css("#__nuxt", id: nil)
         html_tree_change_properties_by_css("#__layout", id: nil)
         html_tree_change_properties_by_css("#app", id: nil)
 
+        # Handle lists of document 02
+        html_tree_replace_with_children_by_css(".list_num-wrap")
+
         # Convert table/img caption to become a caption
         html_tree.css(".imagedata").each do |e|
           table = e.parent.next&.children&.first
@@ -75,66 +83,82 @@ module Coradoc::ReverseAdoc
           end
         end
 
-        html_tree_add_hook_pre_by_css ".text3data" do |node,|
-          text = html_tree_process_to_adoc(node).strip
-          next "" if text.empty? || text == "\u3000"
-
-          text = text.strip.gsub(/^/, "*** ")
-          "\n\n//-PT3D\n#{text}\n//-ENDPT3D\n\n"
-        end
-
-        html_tree_add_hook_pre_by_css ".text4data" do |node,|
-          text = html_tree_process_to_adoc(node).strip
-          next "" if text.empty? || text == "\u3000"
+        (3..4).each do |i|
+          html_tree_add_hook_pre_by_css ".text#{i}data" do |node,|
+            text = html_tree_process_to_adoc(node).strip
+            next "" if text.empty? || text == "\u3000"
 
-          text = text.strip.gsub(/^/, "**** ")
-          "\n\n//-PT4D\n#{text}\n//-ENDPT4D\n\n"
+            text = text.strip.gsub(/^/, "#{'*' * i} ")
+            "\n\n//-PT#{i}D\n#{text}\n//-ENDPT#{i}D\n\n"
+          end
         end
 
-        html_tree_add_hook_pre_by_css ".text2data_point ul" do |node,|
-          text = html_tree_process_to_adoc(node.children.first.children).strip
+        (2..3).each do |i|
+          html_tree_add_hook_pre_by_css ".text#{i}data_point ul" do |node,|
+            text = html_tree_process_to_adoc(node.children.first.children).strip
 
-          "** #{text}\n"
+            "#{'*' * i} #{text}\n"
+          end
         end
 
-        html_tree_add_hook_pre_by_css ".text3data_point ul" do |node,|
-          text = html_tree_process_to_adoc(node.children.first.children).strip
+        (1..20).each do |i|
+          html_tree_add_hook_pre_by_css ".numtextdata_num .list_num#{i}" do |node,|
+            text = html_tree_process_to_adoc(node).strip
 
-          "*** #{text}\n"
+            "[start=#{i}]\n. #{text}\n"
+          end
         end
 
         # html_tree_preview
       end
 
+      IM = /[A-Z0-9]{1,3}/
+
       def handle_headers(node, coradoc, state)
-        if coradoc.id.start_with?("toc0_")
-          content = coradoc.content.map(&:content).join
+        content = coradoc.content.map(&:content).join
+
+        if %w[toc0 toc_0].any? { |i| coradoc.id&.start_with?(i) }
           # Special content
           case content.strip
           when "はじめに" # Introduction
             coradoc.style = "abstract" # The older version document has ".preface"
+            coradoc.level_int = 1
           when "改定の概要" # Revision overview
             coradoc.style = "abstract" # The older version document has ".preface"
+            coradoc.level_int = 1
           when "参考文献" # Bibliography
             coradoc.style = "bibliography"
+            coradoc.level_int = 1
           when "改訂履歴" # Document history
             coradoc.style = "appendix"
+            coradoc.level_int = 1
+          when "0　概要" # Overview
+            coradoc.style = "abstract" # I'm not sure this is correct
+            coradoc.level_int = 1
+          when "索引" # Index
+            coradoc.style = "index" # I'm not sure this is correct
+            coradoc.level_int = 1
           else
-            warn "Unknown section #{coradoc.content.map(&:content).join.inspect}"
+            warn "Unknown section #{content.inspect}"
           end
+        end
 
-          # Ensure they are generated as level 1
-          coradoc.level_int = 1
+        if node.name == "h1"
+          if content.start_with?("Annex")
+            coradoc.style = "appendix"
+            coradoc.content.first.content.sub!(/\AAnnex [A-Z]/, "")
+          end
         end
 
         # Remove numbers
-        coradoc.content.first.content.sub!(/\A[\d\s.]+/, "")
+        coradoc.content.first.content.sub!(/\A(#{IM}\.)*#{IM}[[:space:]]/, "")
 
         coradoc
       end
 
       def handle_headers_h4(node, coradoc, state)
-        case coradoc.content.first.content
+        title = Coradoc.strip_unicode(coradoc.content.first.content)
+        case title
         when /\A\(\d+\)(.*)/
           coradoc.level_int = 4
           coradoc.content.first.content = $1.strip
@@ -143,8 +167,16 @@ module Coradoc::ReverseAdoc
           coradoc.level_int = 5
           coradoc.content.first.content = $1.strip
           coradoc
+        when /\A#{IM}\.#{IM}\.#{IM}\.#{IM}(.*)/
+          coradoc.level_int = 4
+          coradoc.content.first.content = $1.strip
         else
-          ["// FIXME\n", coradoc]
+          if title.empty?
+            # Strip instances of faulty empty paragraphs
+            nil
+          else
+            ["// FIXME\n", coradoc]
+          end
         end
       end
 

data/lib/coradoc/reverse_adoc/postprocessor.rb

@@ -82,13 +82,18 @@ module Coradoc::ReverseAdoc
       previous_sections = {}
 
       determine_section_id = ->(elem) do
-        level = 0
-        section = elem
+        if elem.title.style == "appendix"
+          level = "A"
+        else
+          level = 1
+        end
+
+        section = previous_sections[elem]
         while section
-          level += 1 if elem.title.style == section.title.style
+          level = level.succ if elem.title.style == section.title.style
           section = previous_sections[section]
         end
-        level
+        level.is_a?(Integer) ? "%02d" % level : level
       end
 
       determine_style = ->(elem) do
@@ -114,8 +119,7 @@ module Coradoc::ReverseAdoc
             # include tag.
             section_file = "sections/"
             section_file += parent_sections[1..title.level_int].map do |parent|
-              style = determine_style.(parent)
-              "%s%02d" % [style, determine_section_id.(parent)]
+              determine_style.(parent) + determine_section_id.(parent)
             end.join("/")
             section_file += ".adoc"
 

data/lib/coradoc/util.rb

@@ -0,0 +1,10 @@
+module Coradoc
+  def self.strip_unicode(str)
+    str.gsub(/\A[[:space:]]+|[[:space:]]+\z/, "")
+  end
+
+  def self.is_a_single?(obj, klass)
+    obj.is_a?(klass) ||
+      (obj.is_a?(Array) && obj.length == 1 && obj.first.is_a?(klass))
+  end
+end

data/lib/coradoc/version.rb

@@ -1,5 +1,5 @@
 # frozen_string_literal: true
 
 module Coradoc
-  VERSION = "1.0.0"
+  VERSION = "1.1.0"
 end

data/lib/coradoc.rb

@@ -4,6 +4,7 @@ require "pathname"
 
 require "parslet"
 require_relative "coradoc/version"
+require_relative "coradoc/util"
 require_relative "coradoc/parser"
 require_relative "coradoc/transformer"
 require_relative "coradoc/generator"

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: coradoc
 version: !ruby/object:Gem::Version
-  version: 1.0.0
+  version: 1.1.0
 platform: ruby
 authors:
 - Ribose Inc.
@@ -9,7 +9,7 @@ authors:
 autorequire: 
 bindir: exe
 cert_chain: []
-date: 2024-06-01 00:00:00.000000000 Z
+date: 2024-06-07 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: marcel
@@ -356,6 +356,7 @@ files:
 - lib/coradoc/reverse_adoc/plugins/plateau.rb
 - lib/coradoc/reverse_adoc/postprocessor.rb
 - lib/coradoc/transformer.rb
+- lib/coradoc/util.rb
 - lib/coradoc/version.rb
 - lib/reverse_adoc.rb
 - todo.md