coradoc 1.1.2 → 1.1.3

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: b6e77fc4eb6d79071cb1d309530adbd49a4364265b446d139c18eac2da3bce54
-  data.tar.gz: 07a84837196bde12cc9d91ccd5084b8a173a4da5c237eeaea06a6d786c2f876c
+  metadata.gz: 4f515fbc05baa87f58f84a59737c9818603c6e9f0fc8835cdeb9bd6be9eb39e4
+  data.tar.gz: ce51ff395a3dfb4bf77c37f6b54fe745a14a7262911df50479088d43392d4927
 SHA512:
-  metadata.gz: 69d2d12389a4e254b5ee8f4d67d3ddd406abeabebe1874b343acb4cd7c560c35284c8a58a4d2ee58cc788cfafd68989e88e783aa79fa3c32dc0f9dd389a276ac
-  data.tar.gz: 7f034ef9a649446198793d8d8221a1bba5e22069f2feec85c5d30529170f8a87315540f481bc8b1219c03aaba78e42de67bbf94e25fd3e4ab93b6f221a356ca4
+  metadata.gz: 3269512745aea59b9780e5df2d872af7c3f19851666868335bc4876a534e50c64df410c5257a335324e9a9e0ad573dacf00058752ee00fd4b830f2f150cadeb2
+  data.tar.gz: 00e4122ec5e234e8e7e54d3e5769e03df15ad0f9851059576dadde7de58c0e2dd87f3c18ab14283574347bda6c95536cf31ee3ce6c202ea8b4c3f0da96f3f399

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/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.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/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/version.rb

@@ -1,5 +1,5 @@
 # frozen_string_literal: true
 
 module Coradoc
-  VERSION = "1.1.2"
+  VERSION = "1.1.3"
 end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: coradoc
 version: !ruby/object:Gem::Version
-  version: 1.1.2
+  version: 1.1.3
 platform: ruby
 authors:
 - Ribose Inc.
@@ -9,7 +9,7 @@ authors:
 autorequire: 
 bindir: exe
 cert_chain: []
-date: 2024-09-18 00:00:00.000000000 Z
+date: 2024-11-13 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: marcel