coradoc-html 1.1.21 → 1.1.23

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 78a6409c37e529e753bea5a39adc627dc5c2c059970feac19f0026fd3faf640b
-  data.tar.gz: e3edeb7c93d81f268f52919a36960061c9ea3c56c8aa31f0095d3c18708a6585
+  metadata.gz: 13fc1658ebea3a1a812dbe92977e3e42dcf47e3586caaf7faf9f195f3c06070e
+  data.tar.gz: 7e39f8f780d457d4b002517737663ef93ba282f3ee56fc3cc9a465bceed141d7
 SHA512:
-  metadata.gz: a522853052a688ed9342752416eecb09118250e3c175191a8e3a88baa0ae1e4976f17446a5daddf82e849a9b34c6313258a6926e1f8efc2aa93725dcd086d275
-  data.tar.gz: addbcb3a1a19879f4dc4f83b68ae2f34cd5687488fa017d31a1367d847f2c693c7dcc639bdeb3445d4437ccfa9d958cf21d67428c8ec8992d4d8ab9aed854d31
+  metadata.gz: fbb70e244e35b6e3961e5238db99f4303e0ad798d2cdfb5b562d5ae2ee2919e91f3791b255d78a36867480b09ad7067b82e1e8509640deb93af9dde53ed93005
+  data.tar.gz: 4a50767b69167bda8f232249cf99b371139d55a7860d1e475a35a0bc24cd6e36fd95cf34408e6804b09b73a3545b55ef147113567e2aac8894b969dc77aa7e34

data/lib/coradoc/html/asset_resolver.rb

@@ -1,7 +1,5 @@
 # frozen_string_literal: true
 
-require 'nokogiri'
-
 module Coradoc
   module Html
     module AssetResolver
@@ -55,11 +53,7 @@ module Coradoc
 
         def css_link_tag(options = {})
           href = stylesheet_path(options)
-          doc = Nokogiri::HTML::Document.new
-          node = Nokogiri::XML::Node.new('link', doc)
-          node['rel'] = 'stylesheet'
-          node['href'] = href
-          node.to_html
+          Builder.new { |doc| doc.link(rel: 'stylesheet', href: href) }.to_html
         end
 
         def css_style_tag(options = {})
@@ -122,11 +116,7 @@ module Coradoc
 
         def js_link_tag(options = {})
           src = javascript_path(options)
-          doc = Nokogiri::HTML::Document.new
-          node = Nokogiri::XML::Node.new('script', doc)
-          node['src'] = src
-          node['defer'] = ''
-          node.to_html
+          Builder.new { |doc| doc.script(src: src, defer: '') }.to_html
         end
 
         def js_script_tag(options = {})
@@ -164,28 +154,19 @@ module Coradoc
 
         def highlightjs_tags(options = {})
           theme = options[:highlightjs_theme] || Config::DEFAULT_OPTIONS[:highlightjs_theme]
-          doc = Nokogiri::HTML::Document.new
-
-          link_node = Nokogiri::XML::Node.new('link', doc)
-          link_node['rel'] = 'stylesheet'
-          link_node['href'] = "https://cdnjs.cloudflare.com/ajax/libs/highlight.js/11.9.0/styles/#{theme}.min.css"
-
-          script_node = Nokogiri::XML::Node.new('script', doc)
-          script_node['src'] = 'https://cdnjs.cloudflare.com/ajax/libs/highlight.js/11.9.0/highlight.min.js'
-
-          init_node = Nokogiri::XML::Node.new('script', doc)
-          init_node.content = 'hljs.highlightAll();'
 
-          [link_node.to_html, script_node.to_html, init_node.to_html].join("\n")
+          Builder.new do |doc|
+            doc.link(rel: 'stylesheet',
+                     href: "https://cdnjs.cloudflare.com/ajax/libs/highlight.js/11.9.0/styles/#{theme}.min.css")
+            doc.script(src: 'https://cdnjs.cloudflare.com/ajax/libs/highlight.js/11.9.0/highlight.min.js')
+            doc.script('hljs.highlightAll();')
+          end.to_html
         end
 
         private
 
         def build_text_element(tag_name, content)
-          doc = Nokogiri::HTML::Document.new
-          node = Nokogiri::XML::Node.new(tag_name, doc)
-          node.content = content
-          node.to_html
+          Builder.new { |doc| doc.send(tag_name, content) }.to_html
         end
       end
     end

data/lib/coradoc/html/builder.rb

@@ -0,0 +1,94 @@
+# frozen_string_literal: true
+
+module Coradoc
+  module Html
+    # Builds HTML trees over the Leptris DOM using a
+    # Nokogiri::HTML::Builder-style DSL. Single source of truth for
+    # programmatic HTML construction in this gem (layout fallbacks,
+    # asset tag emitters) — no raw HTML string concatenation anywhere.
+    #
+    #   Builder.new do |doc|
+    #     doc.html(lang: 'en') do
+    #       doc.head { doc.meta(charset: 'UTF-8') }
+    #       doc.body { doc << body_html }
+    #     end
+    #   end.to_html
+    #
+    # Serialization uses leptris-ruby's HTML mode (#309): script/style
+    # text is emitted unescaped and void elements are unclosed.
+    class Builder
+      # Methods that are builder API, not tag names.
+      RESERVED = %i[text to_html to_s document parent].freeze
+
+      attr_reader :document
+
+      def initialize(document = Leptris::XML::Document.create, &block)
+        @document = document
+        @roots = []
+        @parents = []
+        instance_eval(&block) if block
+      end
+
+      # Tag DSL. A positional String argument becomes the element's text
+      # content; the block receives the builder with the new element as
+      # insertion parent; keyword arguments become attributes.
+      def method_missing(name, *args, **attrs, &block)
+        element = @document.create_element(name.to_s)
+        attrs.each { |key, value| element[key.to_s] = value.to_s }
+
+        args.each do |arg|
+          next unless arg.is_a?(String)
+
+          element.add_child(@document.create_text_node(arg))
+        end
+
+        if block
+          @parents.push(element)
+          begin
+            yield self
+          ensure
+            @parents.pop
+          end
+        end
+
+        if current
+          current.add_child(element)
+        else
+          @roots << element
+        end
+        element
+      end
+
+      def respond_to_missing?(name, include_private = false)
+        !RESERVED.include?(name) || super
+      end
+
+      # Raw markup insertion into the current element (parsed, never
+      # concatenated as text).
+      def <<(markup)
+        current&.add_child(markup)
+        self
+      end
+
+      # Text node insertion into the current element.
+      def text(content)
+        current&.add_child(@document.create_text_node(content.to_s))
+        self
+      end
+
+      # Serializes with HTML semantics: script/style text unescaped,
+      # void elements unclosed. Multiple top-level elements are joined
+      # with newlines.
+      def to_html
+        @roots.map(&:to_html).join("\n")
+      end
+      alias to_s to_html
+
+      private
+
+      def current
+        @parents.last
+      end
+    end
+  end
+end

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

@@ -23,12 +23,14 @@ module Coradoc
         end
 
         def node_has_ancestor?(node, name)
-          case name
-          when String
-            node.ancestors(name).any?
-          when Array
-            name.any? { |n| node.ancestors(n).any? }
+          names = Array(name)
+          parent = node.parent
+          while parent
+            return true if parent.element? && names.include?(parent.name)
+
+            parent = parent.parent
           end
+          false
         end
 
         def textnode_before_end_with?(node, str)
@@ -39,7 +41,7 @@ module Coradoc
           str = /(?:#{str})\z/
 
           node2 = node.at_xpath('preceding-sibling::node()[1]')
-          node2.is_a?(Nokogiri::XML::Text) && node2.text.match?(str)
+          node2.is_a?(Leptris::XML::Text) && node2.content.match?(str)
         end
 
         def textnode_after_start_with?(node, str)
@@ -50,7 +52,7 @@ module Coradoc
           str = /\A(?:#{str})/
 
           node2 = node.at_xpath('following-sibling::node()[1]')
-          node2.is_a?(Nokogiri::XML::Text) && node2.text.match?(str)
+          node2.is_a?(Leptris::XML::Text) && node2.content.match?(str)
         end
 
         def extract_leading_trailing_whitespace(node)
@@ -58,14 +60,14 @@ module Coradoc
           leading_whitespace = ::Regexp.last_match(1)
           unless leading_whitespace.nil?
             first_text = node.at_xpath('./text()[1]')
-            first_text&.replace(first_text.text.lstrip)
+            first_text&.content = first_text.content.lstrip
             leading_whitespace = ' '
           end
           node.text =~ /(\s+)$/
           trailing_whitespace = ::Regexp.last_match(1)
           unless trailing_whitespace.nil?
             last_text = node.at_xpath('./text()[last()]')
-            last_text&.replace(last_text.text.rstrip)
+            last_text&.content = last_text.content.rstrip
             trailing_whitespace = ' '
           end
           [leading_whitespace, trailing_whitespace]

data/lib/coradoc/html/converters/pass_through.rb

@@ -7,7 +7,7 @@ module Coradoc
         INSTANCE = new
 
         def to_coradoc(node, _state = {})
-          node.to_s
+          node.is_a?(Leptris::XML::Node) ? node.to_xml(no_decl: true) : node.to_s
         end
       end
     end

data/lib/coradoc/html/converters.rb

@@ -71,7 +71,7 @@ module Coradoc
       end
 
       def self.process_coradoc(node, state)
-        node = node.to_a if node.is_a? Nokogiri::XML::NodeSet
+        node = node.to_a if node.is_a? Leptris::XML::NodeSet
         return node.map { |i| process_coradoc(i, state) } if node.is_a? Array
 
         plugins = state[:plugin_instances] || {}

data/lib/coradoc/html/frontmatter_meta.rb

@@ -1,7 +1,5 @@
 # frozen_string_literal: true
 
-require 'nokogiri'
-
 module Coradoc
   module Html
     # Single source of truth for FrontmatterBlock -> HTML `<meta>` and
@@ -9,7 +7,7 @@ module Coradoc
     # HTML gem; CoreModel has no knowledge of HTML.)
     #
     # The module produces a small data structure that layout templates
-    # and Nokogiri::HTML::Builder fallbacks both consume, so we never
+    # and Builder fallbacks both consume, so we never
     # duplicate the mapping rule across output paths (DRY).
     #
     # Mapping table (single source of truth — extend here only):
@@ -50,7 +48,7 @@ module Coradoc
           }
         end
 
-        # Emit meta + link tags into a Nokogiri head builder context.
+        # Emit meta + link tags into a Builder head context.
         def emit_into_builder(builder_doc, block)
           data = extract(block)
           Array(data[:metas]).each do |meta|

data/lib/coradoc/html/html_converter.rb

@@ -10,10 +10,12 @@ module Coradoc
           root = track_time 'Loading input HTML document' do
             case input
             when String
-              Nokogiri::HTML(input).root
-            when Nokogiri::XML::Document
+              next nil if input.strip.empty?
+
+              html_parse(input).root
+            when Leptris::XML::Document
               input.root
-            when Nokogiri::XML::Node
+            when Leptris::XML::Node
               input
             end
           end
@@ -57,6 +59,23 @@ module Coradoc
         options[:plugin_instances] || Html.input_config.plugins.map(&:new)
       end
 
+      # Single dispatch point for the HTML parsing lane: :html4 keeps
+      # Nokogiri::HTML byte-parity; :html5 selects the WHATWG-conformant
+      # engine (implied tbody/tr, foster parenting, full implied head).
+      # A synthesized empty <head/> is stripped before conversion — an
+      # authored empty head (html4 lane, explicit markup) still produces
+      # a DocumentElement with the placeholder title.
+      def self.html_parse(input)
+        if Html.input_config.html_version == :html5
+          doc = Leptris::HTML5.parse(input)
+          head = doc.at_css('head')
+          head.parent&.remove_child(head) if head && head.children.empty?
+          doc
+        else
+          Leptris::HTML.parse(input)
+        end
+      end
+
       @track_time_indentation = 0
       def self.track_time(task)
         if Html.input_config.track_time

data/lib/coradoc/html/input_config.rb

@@ -8,6 +8,7 @@ module Coradoc
       def initialize
         @unknown_tags     = :pass_through
         @input_format     = :html
+        @html_version     = :html4
         @mathml2asciimath = false
         @external_images  = false
 
@@ -57,6 +58,7 @@ module Coradoc
       declare_option :image_counter
       declare_option :image_counter_pattern
       declare_option :input_format
+      declare_option :html_version
       declare_option :split_sections
       declare_option :doc_width
       declare_option :plugins

data/lib/coradoc/html/layout_renderer.rb

@@ -1,6 +1,5 @@
 # frozen_string_literal: true
 
-require 'nokogiri'
 require 'liquid'
 
 module Coradoc
@@ -90,7 +89,7 @@ module Coradoc
 
       def build_static_fallback(document, body_html, opts)
         frontmatter = extract_frontmatter(document)
-        Nokogiri::HTML::Builder.new do |doc|
+        Builder.new do |doc|
           doc.html(lang: resolve_lang(opts)) do
             doc.head do
               doc.meta(charset: 'UTF-8')
@@ -152,7 +151,7 @@ module Coradoc
 
       def build_spa_fallback(document, opts, assets, safe_json)
         frontmatter = extract_frontmatter(document)
-        Nokogiri::HTML::Builder.new do |doc|
+        Builder.new do |doc|
           doc.html(lang: resolve_lang(opts)) do
             doc.head do
               doc.meta(charset: 'UTF-8')

data/lib/coradoc/html/plugin.rb

@@ -41,12 +41,15 @@ module Coradoc
       end
 
       def html_tree_remove_by_css(css)
-        html_tree.css(css).each(&:remove)
+        html_tree.css(css).each do |e|
+          e.parent&.remove_child(e)
+        end
       end
 
       def html_tree_replace_with_children_by_css(css)
         html_tree.css(css).each do |e|
-          e.replace(e.children)
+          e.children.each { |child| e.add_previous_sibling(child) }
+          e.parent&.remove_child(e)
         end
       end
 

data/lib/coradoc/html/renderer.rb

@@ -1,7 +1,7 @@
 # frozen_string_literal: true
 
 require 'liquid'
-require 'nokogiri'
+require 'leptris'
 
 module Coradoc
   module Html
@@ -118,12 +118,9 @@ module Coradoc
         resolved = TitleText.resolve(drop.model)
         text = resolved ? Escape.escape_html(resolved) : ''
 
-        fragment = Nokogiri::HTML.fragment
-        div = Nokogiri::XML::Node.new('div', fragment.document)
-        div['class'] = "element element-#{type}"
-        div.content = text
-        fragment.add_child(div)
-        fragment.to_html
+        Builder.new do |doc|
+          doc.div(class: "element element-#{type}") { doc.text text }
+        end.to_html
       end
 
       def normalize_dirs(dirs)

data/lib/coradoc/html/transform/to_core_model.rb

@@ -1,28 +1,28 @@
 # frozen_string_literal: true
 
-require 'nokogiri'
+require 'leptris'
 require 'coradoc'
 
 module Coradoc
   module Html
     module Transform
-      # Transforms Nokogiri HTML nodes to CoreModel
+      # Transforms Leptris HTML nodes to CoreModel
       #
-      # Nokogiri serves as the HTML model layer. This transformer converts
-      # Nokogiri::XML::Document or Nokogiri::XML::Node objects into CoreModel
+      # Leptris serves as the HTML model layer. This transformer converts
+      # Leptris::XML::Document or Leptris::XML::Node objects into CoreModel
       # by delegating to the existing input converter pipeline.
       class ToCoreModel
         class << self
-          # Transform an HTML model (Nokogiri node) to CoreModel
+          # Transform an HTML model (Leptris node) to CoreModel
           #
-          # @param model [Nokogiri::XML::Document, Nokogiri::XML::Node, Coradoc::CoreModel::Base]
+          # @param model [Leptris::XML::Document, Leptris::XML::Node, Coradoc::CoreModel::Base]
           #   HTML input model to transform
           # @return [Coradoc::CoreModel::Base] CoreModel equivalent
           def transform(model)
             case model
             when Coradoc::CoreModel::Base
               model
-            when Nokogiri::XML::Document, Nokogiri::XML::Node
+            when Leptris::XML::Document, Leptris::XML::Node
               ::Coradoc::Html::HtmlConverter.to_core_model(model)
             when Array
               model.map { |item| transform(item) }

data/lib/coradoc/html/version.rb

@@ -2,6 +2,6 @@
 
 module Coradoc
   module Html
-    VERSION = '1.1.21'
+    VERSION = '1.1.23'
   end
 end

data/lib/coradoc/html.rb

@@ -73,6 +73,7 @@ module Coradoc
     require_relative 'html/template_helpers'
     autoload :Renderer, 'coradoc/html/renderer'
     autoload :LayoutRenderer, 'coradoc/html/layout_renderer'
+    autoload :Builder, 'coradoc/html/builder'
     autoload :RenderOptions, 'coradoc/html/render_options'
     autoload :TocSerializer, 'coradoc/html/toc_serializer'
 
@@ -208,14 +209,14 @@ module Coradoc
 
     # Check if this format can transform the given model to CoreModel
     #
-    # HTML uses Nokogiri as its model layer. Accepts Nokogiri nodes
+    # HTML uses Leptris::XML as its model layer. Accepts Leptris nodes
     # and CoreModel objects (pass-through).
     #
     # @param model [Object] The model to check
-    # @return [Boolean] true if the model is a Nokogiri node or CoreModel
+    # @return [Boolean] true if the model is a Leptris node or CoreModel
     def self.handles_model?(model)
-      model.is_a?(Nokogiri::XML::Node) ||
-        model.is_a?(Nokogiri::XML::Document) ||
+      model.is_a?(Leptris::XML::Node) ||
+        model.is_a?(Leptris::XML::Document) ||
         model.is_a?(Coradoc::CoreModel::Base)
     end
 

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: coradoc-html
 version: !ruby/object:Gem::Version
-  version: 1.1.21
+  version: 1.1.23
 platform: ruby
 authors:
 - Ribose Inc.
@@ -52,19 +52,19 @@ dependencies:
       - !ruby/object:Gem::Version
         version: '1.0'
 - !ruby/object:Gem::Dependency
-  name: nokogiri
+  name: leptris
   requirement: !ruby/object:Gem::Requirement
     requirements:
-    - - "~>"
+    - - ">="
       - !ruby/object:Gem::Version
-        version: '1.0'
+        version: '0'
   type: :runtime
   prerelease: false
   version_requirements: !ruby/object:Gem::Requirement
     requirements:
-    - - "~>"
+    - - ">="
       - !ruby/object:Gem::Version
-        version: '1.0'
+        version: '0'
 - !ruby/object:Gem::Dependency
   name: rspec
   requirement: !ruby/object:Gem::Requirement
@@ -118,6 +118,7 @@ files:
 - LICENSE.txt
 - lib/coradoc/html.rb
 - lib/coradoc/html/asset_resolver.rb
+- lib/coradoc/html/builder.rb
 - lib/coradoc/html/cleaner.rb
 - lib/coradoc/html/config.rb
 - lib/coradoc/html/converter_base.rb