coradoc 0.3.0 → 1.1.0

data/lib/coradoc/reverse_adoc/postprocessor.rb

@@ -0,0 +1,152 @@
+module Coradoc::ReverseAdoc
+  # Postprocessor's aim is to convert a Coradoc tree from
+  # a mess that has been created from HTML into a tree that
+  # is compatible with what we would get out of Coradoc, if
+  # it parsed it directly.
+  class Postprocessor
+    def self.process(coradoc)
+      new(coradoc).process
+    end
+
+    def initialize(coradoc)
+      @tree = coradoc
+    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?
+          children_classes = Array(elem.contents).map(&:class)
+          count = children_classes.length
+          safe_classes = [Coradoc::Element::Section, Coradoc::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
+          end
+        end
+        elem
+      end
+    end
+
+    # 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|
+        # 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?
+
+          new_array = []
+          content_array = new_array
+          section_arrays_by_level = [new_array] * 8
+
+          # For each title element, we create a new section. Then we push
+          # 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
+              title = e
+              content_array = []
+              section_array = []
+              level = title.level_int
+              section = Coradoc::Element::Section.new(
+                title, contents: content_array, sections: section_array
+              )
+              # Some documents may not be consistent and eg. follow H4 after
+              # H2. Let's ensure that proceeding sections will land in a
+              # correct place.
+              (8 - level).times do |j|
+                section_arrays_by_level[level + j] = section_array
+              end
+              section_arrays_by_level[level - 1] << section
+            else
+              content_array << e
+            end
+          end
+          next new_array
+        end
+        elem
+      end
+    end
+
+    def split_sections
+      max_level = Coradoc::ReverseAdoc.config.split_sections
+
+      return unless max_level
+
+      sections = {}
+      parent_sections = []
+      previous_sections = {}
+
+      determine_section_id = ->(elem) do
+        if elem.title.style == "appendix"
+          level = "A"
+        else
+          level = 1
+        end
+
+        section = previous_sections[elem]
+        while section
+          level = level.succ if elem.title.style == section.title.style
+          section = previous_sections[section]
+        end
+        level.is_a?(Integer) ? "%02d" % level : level
+      end
+
+      determine_style = ->(elem) do
+        style = elem.title.style || "section"
+        style += "-"
+        style
+      end
+
+      @tree = Coradoc::Element::Base.visit(@tree) do |elem, dir|
+        title = elem.title if elem.is_a?(Coradoc::Element::Section)
+
+        if title && title.level_int <= max_level
+          if dir == :pre
+            # In the PRE pass, we build a tree of sections, so that
+            # we can compute numbers
+            previous_sections[elem] = parent_sections[title.level_int]
+            parent_sections[title.level_int] = elem
+            parent_sections[(title.level_int+1)..nil] = nil
+
+            elem
+          else
+            # In the POST pass, we replace the sections with their
+            # include tag.
+            section_file = "sections/"
+            section_file += parent_sections[1..title.level_int].map do |parent|
+              determine_style.(parent) + determine_section_id.(parent)
+            end.join("/")
+            section_file += ".adoc"
+
+            sections[section_file] = elem
+            up = "../" * (title.level_int - 1)
+            "\ninclude::#{up}#{section_file}[]\n"
+          end
+        else
+          elem
+        end
+      end
+
+      sections[nil] = @tree
+      @tree = sections
+    end
+
+    def process
+      collapse_meaningless_sections
+      generate_meaningful_sections
+      # Do it again to simplify the document further.
+      # Since the structure is changed, we may have new meaningful
+      # sections as only children of some meaningless sections.
+      collapse_meaningless_sections
+
+      split_sections
+
+      @tree
+    end
+  end
+end

data/lib/coradoc/reverse_adoc.rb

@@ -9,6 +9,9 @@ require_relative "reverse_adoc/config"
 require_relative "reverse_adoc/converters"
 require_relative "reverse_adoc/converters/base"
 require_relative "reverse_adoc/html_converter"
+require_relative "reverse_adoc/plugin"
+require_relative "reverse_adoc/postprocessor"
+
 
 module Coradoc::ReverseAdoc
   def self.convert(input, options = {})

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 = "0.3.0"
+  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"

data/lib/reverse_adoc.rb

@@ -8,7 +8,7 @@ warn <<~END
   | You are referencing an old require here:
   |   #{caller.join("\n|   ")}
   |
-  | You should also replace 'coradoc' with 'reverse_adoc' in your gem dependencies.
+  | You should also replace 'reverse_adoc' with 'coradoc' in your gem dependencies.
   | reverse_adoc 2.0.0 will be kept with 'coradoc' as the only dependency.
   |
   | Please also ensure that you replace all references to ReverseAdoc in your code

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: coradoc
 version: !ruby/object:Gem::Version
-  version: 0.3.0
+  version: 1.1.0
 platform: ruby
 authors:
 - Ribose Inc.
@@ -9,7 +9,7 @@ authors:
 autorequire: 
 bindir: exe
 cert_chain: []
-date: 2024-05-21 00:00:00.000000000 Z
+date: 2024-06-07 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: marcel
@@ -278,7 +278,6 @@ files:
 - lib/coradoc/element/inline/cross_reference.rb
 - lib/coradoc/element/inline/hard_line_break.rb
 - lib/coradoc/element/inline/highlight.rb
-- lib/coradoc/element/inline/image.rb
 - lib/coradoc/element/inline/italic.rb
 - lib/coradoc/element/inline/link.rb
 - lib/coradoc/element/inline/monospace.rb
@@ -291,6 +290,7 @@ files:
 - lib/coradoc/element/list/ordered.rb
 - lib/coradoc/element/list/unordered.rb
 - lib/coradoc/element/list_item.rb
+- lib/coradoc/element/list_item_definition.rb
 - lib/coradoc/element/paragraph.rb
 - lib/coradoc/element/revision.rb
 - lib/coradoc/element/section.rb
@@ -323,6 +323,7 @@ files:
 - lib/coradoc/reverse_adoc/converters/bypass.rb
 - lib/coradoc/reverse_adoc/converters/code.rb
 - lib/coradoc/reverse_adoc/converters/div.rb
+- lib/coradoc/reverse_adoc/converters/dl.rb
 - lib/coradoc/reverse_adoc/converters/drop.rb
 - lib/coradoc/reverse_adoc/converters/em.rb
 - lib/coradoc/reverse_adoc/converters/figure.rb
@@ -333,6 +334,7 @@ files:
 - lib/coradoc/reverse_adoc/converters/img.rb
 - lib/coradoc/reverse_adoc/converters/li.rb
 - lib/coradoc/reverse_adoc/converters/mark.rb
+- lib/coradoc/reverse_adoc/converters/markup.rb
 - lib/coradoc/reverse_adoc/converters/math.rb
 - lib/coradoc/reverse_adoc/converters/ol.rb
 - lib/coradoc/reverse_adoc/converters/p.rb
@@ -350,7 +352,11 @@ files:
 - lib/coradoc/reverse_adoc/converters/video.rb
 - lib/coradoc/reverse_adoc/errors.rb
 - lib/coradoc/reverse_adoc/html_converter.rb
+- lib/coradoc/reverse_adoc/plugin.rb
+- 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

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

@@ -1,26 +0,0 @@
-require_relative "../image/"
-module Coradoc
-  module Element
-    module Inline
-      class Image
-        attr_reader :title, :id, :src, :attributes
-
-        def initialize(title, id, src, options = ())
-          @title = title
-          @id = id
-          @anchor = @id.nil? ? nil : Inline::Anchor.new(@id)
-          @src = src
-          @attributes = options.fetch(:attributes, [])
-          @title = options.fetch(:title, nil)
-        end
-
-        def to_adoc
-          anchor = @anchor.nil? ? "" : "#{@anchor.to_adoc}\n"
-          title = ".#{@title}\n" unless @title.to_s.empty?
-          attrs = @attributes.empty? ? "\[\]" : @attributes.to_adoc
-          [anchor, title, "image::", @src, attrs].join("")
-        end
-      end
-    end
-  end
-end